{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "7a951476f7d84de7d49723ba", "attempt_id": "09fdad8b6b8d", "run_group_id": "E4_core:A4|closed|qwen3b:h1", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "d83aa60f20d2879d02898cb9d9d8bd1983ed2d8984cae017c9153cafee3a779a", "git_commit": null, "timestamp_utc": "2026-10-30T06:18:24.218554Z", "task_family": "synthetic", "document_id": "syn_03000_2000", "document_hash": "c9cc27c20b05b1f6e592846e84990ee9b69c42f9720a8e338ec2f6fca777fdbc", "question_id": "q00_single_fact", "split": "test", "cluster_id": "syn_03000_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "laptop-m5max-48g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 1, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1904, "query_tokens": 592, "message_tokens": 511, "message_utf8_bytes": 1090, "cache_tensor_bytes": null, "serialized_payload_bytes": 1090, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 22814.76537499111, "hops": {"1": 22814.76537499111}, "receiver_prefill_ms": 517.9990419419482, "ttft_ms": 38.97520899772644, "decode_ms": 22.368042031303048}, "total_wall_ms": 23394.779458991252, "ttft_ms": 38.97520899772644, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 29, "retention_rate": 0.4461538461538462, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5, "status": 1.0, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1090}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 29, "retention_rate": 0.4461538461538462, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5, "status": 1.0, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1090}}}, "reference_key_id": "keys/synthetic/syn_03000_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "562010df7e3ceb9bcd9a843b326413164596adc2f7ca548f1a2eb3ede9dfbb44", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "400b8a1fb6e5600c20af66cb", "attempt_id": "2a8573c22805", "run_group_id": "E4_core:A4|closed|qwen3b:h1", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "d83aa60f20d2879d02898cb9d9d8bd1983ed2d8984cae017c9153cafee3a779a", "git_commit": null, "timestamp_utc": "2026-10-30T06:18:24.817910Z", "task_family": "synthetic", "document_id": "syn_03000_2000", "document_hash": "c9cc27c20b05b1f6e592846e84990ee9b69c42f9720a8e338ec2f6fca777fdbc", "question_id": "q01_single_fact", "split": "test", "cluster_id": "syn_03000_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "laptop-m5max-48g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 1, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1904, "query_tokens": 597, "message_tokens": 511, "message_utf8_bytes": 1090, "cache_tensor_bytes": null, "serialized_payload_bytes": 1090, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 22814.76537499111, "hops": {"1": 22814.76537499111}, "receiver_prefill_ms": 398.0731660267338, "ttft_ms": 46.948457951657474, "decode_ms": 275.96624998841435}, "total_wall_ms": 23536.37066599913, "ttft_ms": 46.948457951657474, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 9}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "Lot 19-N, Gaspé", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 29, "retention_rate": 0.4461538461538462, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5, "status": 1.0, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1090}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 29, "retention_rate": 0.4461538461538462, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5, "status": 1.0, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1090}}}, "reference_key_id": "keys/synthetic/syn_03000_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 0.8571428571428571}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "562010df7e3ceb9bcd9a843b326413164596adc2f7ca548f1a2eb3ede9dfbb44", "output_hash": "2e8cf782bd118a1c6d4e23417785d72fb92e930806b5752cd1ec405a0c9b95e5"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "66bf35b9d57b3c67d2bd115d", "attempt_id": "f838f0866fca", "run_group_id": "E4_core:A4|closed|qwen3b:h1", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "d83aa60f20d2879d02898cb9d9d8bd1983ed2d8984cae017c9153cafee3a779a", "git_commit": null, "timestamp_utc": "2026-10-30T06:18:25.555550Z", "task_family": "synthetic", "document_id": "syn_03000_2000", "document_hash": "c9cc27c20b05b1f6e592846e84990ee9b69c42f9720a8e338ec2f6fca777fdbc", "question_id": "q02_exact_copy", "split": "test", "cluster_id": "syn_03000_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "laptop-m5max-48g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 1, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1904, "query_tokens": 596, "message_tokens": 511, "message_utf8_bytes": 1090, "cache_tensor_bytes": null, "serialized_payload_bytes": 1090, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 22814.76537499111, "hops": {"1": 22814.76537499111}, "receiver_prefill_ms": 382.4166670674458, "ttft_ms": 51.02800007443875, "decode_ms": 41.260291007347405}, "total_wall_ms": 23290.292000048794, "ttft_ms": 51.02800007443875, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 29, "retention_rate": 0.4461538461538462, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5, "status": 1.0, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1090}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 29, "retention_rate": 0.4461538461538462, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5, "status": 1.0, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1090}}}, "reference_key_id": "keys/synthetic/syn_03000_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "562010df7e3ceb9bcd9a843b326413164596adc2f7ca548f1a2eb3ede9dfbb44", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "6a438a472b96101145b12fde", "attempt_id": "eedc1b0a63d3", "run_group_id": "E4_core:A4|closed|qwen3b:h1", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "d83aa60f20d2879d02898cb9d9d8bd1983ed2d8984cae017c9153cafee3a779a", "git_commit": null, "timestamp_utc": "2026-10-30T06:18:26.047620Z", "task_family": "synthetic", "document_id": "syn_03000_2000", "document_hash": "c9cc27c20b05b1f6e592846e84990ee9b69c42f9720a8e338ec2f6fca777fdbc", "question_id": "q03_exact_copy", "split": "test", "cluster_id": "syn_03000_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "laptop-m5max-48g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 1, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1904, "query_tokens": 595, "message_tokens": 511, "message_utf8_bytes": 1090, "cache_tensor_bytes": null, "serialized_payload_bytes": 1090, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 22814.76537499111, "hops": {"1": 22814.76537499111}, "receiver_prefill_ms": 568.8447080319747, "ttft_ms": 22.362500079907477, "decode_ms": 230.89674999937415}, "total_wall_ms": 23637.692291056737, "ttft_ms": 22.362500079907477, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 7}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "Kwame Haldorsen", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 29, "retention_rate": 0.4461538461538462, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5, "status": 1.0, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1090}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 29, "retention_rate": 0.4461538461538462, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5, "status": 1.0, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1090}}}, "reference_key_id": "keys/synthetic/syn_03000_2000.json", "task_scores": {"correct": 0.0, "abstained": 0.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "562010df7e3ceb9bcd9a843b326413164596adc2f7ca548f1a2eb3ede9dfbb44", "output_hash": "175253e7cf463934ec4d54e361a781f323cd87290f01306ed368933bcbd2e25d"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "4a55d2643c8cb90e989dd5f0", "attempt_id": "0084309e8b15", "run_group_id": "E4_core:A4|closed|qwen3b:h1", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "d83aa60f20d2879d02898cb9d9d8bd1983ed2d8984cae017c9153cafee3a779a", "git_commit": null, "timestamp_utc": "2026-10-30T06:18:26.887482Z", "task_family": "synthetic", "document_id": "syn_03000_2000", "document_hash": "c9cc27c20b05b1f6e592846e84990ee9b69c42f9720a8e338ec2f6fca777fdbc", "question_id": "q04_multi_hop", "split": "test", "cluster_id": "syn_03000_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "laptop-m5max-48g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 1, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1904, "query_tokens": 601, "message_tokens": 511, "message_utf8_bytes": 1090, "cache_tensor_bytes": null, "serialized_payload_bytes": 1090, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 22814.76537499111, "hops": {"1": 22814.76537499111}, "receiver_prefill_ms": 393.20120797492564, "ttft_ms": 31.520708929747343, "decode_ms": 354.74533401429653}, "total_wall_ms": 23595.150999957696, "ttft_ms": 31.520708929747343, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 9}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "9,220,745", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 29, "retention_rate": 0.4461538461538462, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5, "status": 1.0, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1090}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 29, "retention_rate": 0.4461538461538462, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5, "status": 1.0, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1090}}}, "reference_key_id": "keys/synthetic/syn_03000_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "562010df7e3ceb9bcd9a843b326413164596adc2f7ca548f1a2eb3ede9dfbb44", "output_hash": "ae55ba88b2cb1b0f1a4e0960ae312e0a1c177c79a6d14b67bff1677cec19b649"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "bb52497feccc7c7f81924ef0", "attempt_id": "8e4db573b5db", "run_group_id": "E4_core:A4|closed|qwen3b:h1", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "d83aa60f20d2879d02898cb9d9d8bd1983ed2d8984cae017c9153cafee3a779a", "git_commit": null, "timestamp_utc": "2026-10-30T06:18:27.684537Z", "task_family": "synthetic", "document_id": "syn_03000_2000", "document_hash": "c9cc27c20b05b1f6e592846e84990ee9b69c42f9720a8e338ec2f6fca777fdbc", "question_id": "q05_multi_hop", "split": "test", "cluster_id": "syn_03000_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "laptop-m5max-48g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 1, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1904, "query_tokens": 603, "message_tokens": 511, "message_utf8_bytes": 1090, "cache_tensor_bytes": null, "serialized_payload_bytes": 1090, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 22814.76537499111, "hops": {"1": 22814.76537499111}, "receiver_prefill_ms": 553.3657920313999, "ttft_ms": 32.090791035443544, "decode_ms": 397.4217919167131}, "total_wall_ms": 23798.368375049904, "ttft_ms": 32.090791035443544, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 10}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "LY-51E-3199", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 29, "retention_rate": 0.4461538461538462, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5, "status": 1.0, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1090}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 29, "retention_rate": 0.4461538461538462, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5, "status": 1.0, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1090}}}, "reference_key_id": "keys/synthetic/syn_03000_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "562010df7e3ceb9bcd9a843b326413164596adc2f7ca548f1a2eb3ede9dfbb44", "output_hash": "a1245d1536890a7e4b026366df48441d503f491f782cfd388797b5782443c5c7"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "b326dd4836ffb42f3903586c", "attempt_id": "22d2af67cab8", "run_group_id": "E4_core:A4|closed|qwen3b:h1", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "d83aa60f20d2879d02898cb9d9d8bd1983ed2d8984cae017c9153cafee3a779a", "git_commit": null, "timestamp_utc": "2026-10-30T06:18:28.684379Z", "task_family": "synthetic", "document_id": "syn_03000_2000", "document_hash": "c9cc27c20b05b1f6e592846e84990ee9b69c42f9720a8e338ec2f6fca777fdbc", "question_id": "q06_negation", "split": "test", "cluster_id": "syn_03000_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "laptop-m5max-48g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 1, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1904, "query_tokens": 592, "message_tokens": 511, "message_utf8_bytes": 1090, "cache_tensor_bytes": null, "serialized_payload_bytes": 1090, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 22814.76537499111, "hops": {"1": 22814.76537499111}, "receiver_prefill_ms": 356.7971249576658, "ttft_ms": 51.169333048164845, "decode_ms": 38.1657499819994}, "total_wall_ms": 23261.79991592653, "ttft_ms": 51.169333048164845, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "no", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 29, "retention_rate": 0.4461538461538462, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5, "status": 1.0, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1090}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 29, "retention_rate": 0.4461538461538462, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5, "status": 1.0, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1090}}}, "reference_key_id": "keys/synthetic/syn_03000_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "562010df7e3ceb9bcd9a843b326413164596adc2f7ca548f1a2eb3ede9dfbb44", "output_hash": "9390298f3fb0c5b160498935d79cb139aef28e1c47358b4bbba61862b9c26e59"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "56a9a8eb9c84ee617d0991a3", "attempt_id": "a08e4f4122d1", "run_group_id": "E4_core:A4|closed|qwen3b:h1", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "d83aa60f20d2879d02898cb9d9d8bd1983ed2d8984cae017c9153cafee3a779a", "git_commit": null, "timestamp_utc": "2026-10-30T06:18:29.148102Z", "task_family": "synthetic", "document_id": "syn_03000_2000", "document_hash": "c9cc27c20b05b1f6e592846e84990ee9b69c42f9720a8e338ec2f6fca777fdbc", "question_id": "q07_negation", "split": "test", "cluster_id": "syn_03000_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "laptop-m5max-48g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 1, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1904, "query_tokens": 592, "message_tokens": 511, "message_utf8_bytes": 1090, "cache_tensor_bytes": null, "serialized_payload_bytes": 1090, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 22814.76537499111, "hops": {"1": 22814.76537499111}, "receiver_prefill_ms": 563.4479169966653, "ttft_ms": 31.627667020075023, "decode_ms": 38.04345801472664}, "total_wall_ms": 23448.58062500134, "ttft_ms": 31.627667020075023, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "no", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 29, "retention_rate": 0.4461538461538462, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5, "status": 1.0, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1090}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 29, "retention_rate": 0.4461538461538462, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5, "status": 1.0, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1090}}}, "reference_key_id": "keys/synthetic/syn_03000_2000.json", "task_scores": {"correct": 0.0, "abstained": 0.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "562010df7e3ceb9bcd9a843b326413164596adc2f7ca548f1a2eb3ede9dfbb44", "output_hash": "9390298f3fb0c5b160498935d79cb139aef28e1c47358b4bbba61862b9c26e59"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "ca8a4d6f36e9c334b80c6617", "attempt_id": "13fec92dfbb4", "run_group_id": "E4_core:A4|closed|qwen3b:h1", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "d83aa60f20d2879d02898cb9d9d8bd1983ed2d8984cae017c9153cafee3a779a", "git_commit": null, "timestamp_utc": "2026-10-30T06:18:29.797824Z", "task_family": "synthetic", "document_id": "syn_03000_2000", "document_hash": "c9cc27c20b05b1f6e592846e84990ee9b69c42f9720a8e338ec2f6fca777fdbc", "question_id": "q08_revision", "split": "test", "cluster_id": "syn_03000_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "laptop-m5max-48g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 1, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1904, "query_tokens": 607, "message_tokens": 511, "message_utf8_bytes": 1090, "cache_tensor_bytes": null, "serialized_payload_bytes": 1090, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 22814.76537499111, "hops": {"1": 22814.76537499111}, "receiver_prefill_ms": 403.94716698210686, "ttft_ms": 40.32362496946007, "decode_ms": 305.66879198886454}, "total_wall_ms": 23565.41354104411, "ttft_ms": 40.32362496946007, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 9}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "2,705,965", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 29, "retention_rate": 0.4461538461538462, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5, "status": 1.0, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1090}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 29, "retention_rate": 0.4461538461538462, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5, "status": 1.0, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1090}}}, "reference_key_id": "keys/synthetic/syn_03000_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "562010df7e3ceb9bcd9a843b326413164596adc2f7ca548f1a2eb3ede9dfbb44", "output_hash": "dbedb6f5d00e413d915513b85d8ed195b18b9acf392ece342a85e15e3752bcc8"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "e35c03de58e2fabdbe763157", "attempt_id": "b02e6f8a1de2", "run_group_id": "E4_core:A4|closed|qwen3b:h1", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "d83aa60f20d2879d02898cb9d9d8bd1983ed2d8984cae017c9153cafee3a779a", "git_commit": null, "timestamp_utc": "2026-10-30T06:18:30.564904Z", "task_family": "synthetic", "document_id": "syn_03000_2000", "document_hash": "c9cc27c20b05b1f6e592846e84990ee9b69c42f9720a8e338ec2f6fca777fdbc", "question_id": "q09_revision", "split": "test", "cluster_id": "syn_03000_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "laptop-m5max-48g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 1, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1904, "query_tokens": 603, "message_tokens": 511, "message_utf8_bytes": 1090, "cache_tensor_bytes": null, "serialized_payload_bytes": 1090, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 22814.76537499111, "hops": {"1": 22814.76537499111}, "receiver_prefill_ms": 403.65450002718717, "ttft_ms": 67.14083289261907, "decode_ms": 411.6140410769731}, "total_wall_ms": 23697.818791959435, "ttft_ms": 67.14083289261907, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 7}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "9127965", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 29, "retention_rate": 0.4461538461538462, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5, "status": 1.0, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1090}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 29, "retention_rate": 0.4461538461538462, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5, "status": 1.0, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1090}}}, "reference_key_id": "keys/synthetic/syn_03000_2000.json", "task_scores": {"correct": 0.0, "abstained": 0.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "562010df7e3ceb9bcd9a843b326413164596adc2f7ca548f1a2eb3ede9dfbb44", "output_hash": "a0c871941224304cec47bb1b3d64b8706843b299f14ab94899d97df14f433da7"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "476e62de33a0775348e3f08d", "attempt_id": "c031f9909df0", "run_group_id": "E4_core:A4|closed|qwen3b:h1", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "d83aa60f20d2879d02898cb9d9d8bd1983ed2d8984cae017c9153cafee3a779a", "git_commit": null, "timestamp_utc": "2026-10-30T06:18:31.466526Z", "task_family": "synthetic", "document_id": "syn_03000_2000", "document_hash": "c9cc27c20b05b1f6e592846e84990ee9b69c42f9720a8e338ec2f6fca777fdbc", "question_id": "q10_temporal", "split": "test", "cluster_id": "syn_03000_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "laptop-m5max-48g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 1, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1904, "query_tokens": 609, "message_tokens": 511, "message_utf8_bytes": 1090, "cache_tensor_bytes": null, "serialized_payload_bytes": 1090, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 22814.76537499111, "hops": {"1": 22814.76537499111}, "receiver_prefill_ms": 402.1220830036327, "ttft_ms": 33.50187500473112, "decode_ms": 157.9035420436412}, "total_wall_ms": 23408.931208890863, "ttft_ms": 33.50187500473112, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 5}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "Felaldnak-7", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 29, "retention_rate": 0.4461538461538462, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5, "status": 1.0, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1090}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 29, "retention_rate": 0.4461538461538462, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5, "status": 1.0, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1090}}}, "reference_key_id": "keys/synthetic/syn_03000_2000.json", "task_scores": {"correct": 0.0, "abstained": 0.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "562010df7e3ceb9bcd9a843b326413164596adc2f7ca548f1a2eb3ede9dfbb44", "output_hash": "2b9a4cf7221f1b6f31508a6b3078ec521f211026a932da5020cddd512ecc5fa2"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "c0473fc71c82c52ff9a019ea", "attempt_id": "f3287a69dc90", "run_group_id": "E4_core:A4|closed|qwen3b:h1", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "d83aa60f20d2879d02898cb9d9d8bd1983ed2d8984cae017c9153cafee3a779a", "git_commit": null, "timestamp_utc": "2026-10-30T06:18:32.077774Z", "task_family": "synthetic", "document_id": "syn_03000_2000", "document_hash": "c9cc27c20b05b1f6e592846e84990ee9b69c42f9720a8e338ec2f6fca777fdbc", "question_id": "q11_aggregation", "split": "test", "cluster_id": "syn_03000_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "laptop-m5max-48g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 1, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1904, "query_tokens": 604, "message_tokens": 511, "message_utf8_bytes": 1090, "cache_tensor_bytes": null, "serialized_payload_bytes": 1090, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 22814.76537499111, "hops": {"1": 22814.76537499111}, "receiver_prefill_ms": 358.8229580782354, "ttft_ms": 46.36241600383073, "decode_ms": 461.7095839930698}, "total_wall_ms": 23682.46858404018, "ttft_ms": 46.36241600383073, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 7}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "9220745", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 29, "retention_rate": 0.4461538461538462, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5, "status": 1.0, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1090}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 29, "retention_rate": 0.4461538461538462, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5, "status": 1.0, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1090}}}, "reference_key_id": "keys/synthetic/syn_03000_2000.json", "task_scores": {"correct": 0.0, "abstained": 0.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "562010df7e3ceb9bcd9a843b326413164596adc2f7ca548f1a2eb3ede9dfbb44", "output_hash": "f5d77aa763f6365bc80d01b9e6e23bc43fb570ac45771d7bd628c75cde727b35"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "b37814debbe1ea7cc396719d", "attempt_id": "ae953d1914a2", "run_group_id": "E4_core:A4|closed|qwen3b:h1", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "d83aa60f20d2879d02898cb9d9d8bd1983ed2d8984cae017c9153cafee3a779a", "git_commit": null, "timestamp_utc": "2026-10-30T06:18:32.961571Z", "task_family": "synthetic", "document_id": "syn_03000_2000", "document_hash": "c9cc27c20b05b1f6e592846e84990ee9b69c42f9720a8e338ec2f6fca777fdbc", "question_id": "q12_unit_conversion", "split": "test", "cluster_id": "syn_03000_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "laptop-m5max-48g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 1, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1904, "query_tokens": 600, "message_tokens": 511, "message_utf8_bytes": 1090, "cache_tensor_bytes": null, "serialized_payload_bytes": 1090, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 22814.76537499111, "hops": {"1": 22814.76537499111}, "receiver_prefill_ms": 392.09012500941753, "ttft_ms": 33.73124997597188, "decode_ms": 30.220082961022854}, "total_wall_ms": 23271.35549997911, "ttft_ms": 33.73124997597188, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 29, "retention_rate": 0.4461538461538462, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5, "status": 1.0, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1090}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 29, "retention_rate": 0.4461538461538462, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5, "status": 1.0, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1090}}}, "reference_key_id": "keys/synthetic/syn_03000_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "562010df7e3ceb9bcd9a843b326413164596adc2f7ca548f1a2eb3ede9dfbb44", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "5c98bdf6143e346e07b9f52c", "attempt_id": "4ae62669621c", "run_group_id": "E4_core:A4|closed|qwen3b:h1", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "d83aa60f20d2879d02898cb9d9d8bd1983ed2d8984cae017c9153cafee3a779a", "git_commit": null, "timestamp_utc": "2026-10-30T06:18:33.434037Z", "task_family": "synthetic", "document_id": "syn_03000_2000", "document_hash": "c9cc27c20b05b1f6e592846e84990ee9b69c42f9720a8e338ec2f6fca777fdbc", "question_id": "q13_unanswerable", "split": "test", "cluster_id": "syn_03000_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "laptop-m5max-48g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 1, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1904, "query_tokens": 604, "message_tokens": 511, "message_utf8_bytes": 1090, "cache_tensor_bytes": null, "serialized_payload_bytes": 1090, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 22814.76537499111, "hops": {"1": 22814.76537499111}, "receiver_prefill_ms": 396.46945893764496, "ttft_ms": 23.039791965857148, "decode_ms": 23.29187502618879}, "total_wall_ms": 23258.194167050533, "ttft_ms": 23.039791965857148, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 29, "retention_rate": 0.4461538461538462, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5, "status": 1.0, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1090}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 29, "retention_rate": 0.4461538461538462, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5, "status": 1.0, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1090}}}, "reference_key_id": "keys/synthetic/syn_03000_2000.json", "task_scores": {"correct": 1.0, "abstained": 1.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "562010df7e3ceb9bcd9a843b326413164596adc2f7ca548f1a2eb3ede9dfbb44", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "b5a3ae685c72ac99d960cf7c", "attempt_id": "a5b68c991711", "run_group_id": "E4_core:A4|closed|qwen3b:h1", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "d83aa60f20d2879d02898cb9d9d8bd1983ed2d8984cae017c9153cafee3a779a", "git_commit": null, "timestamp_utc": "2026-10-30T06:18:33.893576Z", "task_family": "synthetic", "document_id": "syn_03000_2000", "document_hash": "c9cc27c20b05b1f6e592846e84990ee9b69c42f9720a8e338ec2f6fca777fdbc", "question_id": "q14_unanswerable", "split": "test", "cluster_id": "syn_03000_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "laptop-m5max-48g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 1, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1904, "query_tokens": 606, "message_tokens": 511, "message_utf8_bytes": 1090, "cache_tensor_bytes": null, "serialized_payload_bytes": 1090, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 22814.76537499111, "hops": {"1": 22814.76537499111}, "receiver_prefill_ms": 383.866292075254, "ttft_ms": 52.32125008478761, "decode_ms": 46.2514579994604}, "total_wall_ms": 23297.910458990373, "ttft_ms": 52.32125008478761, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 29, "retention_rate": 0.4461538461538462, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5, "status": 1.0, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1090}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 29, "retention_rate": 0.4461538461538462, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5, "status": 1.0, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1090}}}, "reference_key_id": "keys/synthetic/syn_03000_2000.json", "task_scores": {"correct": 1.0, "abstained": 1.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "562010df7e3ceb9bcd9a843b326413164596adc2f7ca548f1a2eb3ede9dfbb44", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "8813403cd8fe29d1afcc14d9", "attempt_id": "4b7ef67108b1", "run_group_id": "E4_core:A4|closed|qwen3b:h2", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "d83aa60f20d2879d02898cb9d9d8bd1983ed2d8984cae017c9153cafee3a779a", "git_commit": null, "timestamp_utc": "2026-10-30T06:18:56.636246Z", "task_family": "synthetic", "document_id": "syn_03000_2000", "document_hash": "c9cc27c20b05b1f6e592846e84990ee9b69c42f9720a8e338ec2f6fca777fdbc", "question_id": "q00_single_fact", "split": "test", "cluster_id": "syn_03000_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "laptop-m5max-48g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 2, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1904, "query_tokens": 593, "message_tokens": 512, "message_utf8_bytes": 1093, "cache_tensor_bytes": null, "serialized_payload_bytes": 1093, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 45056.899791932665, "hops": {"1": 22814.76537499111, "2": 22242.134416941553}, "receiver_prefill_ms": 417.7792079281062, "ttft_ms": 26.94070804864168, "decode_ms": 36.514375009573996}, "total_wall_ms": 45538.796499953605, "ttft_ms": 26.94070804864168, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 29, "retention_rate": 0.4461538461538462, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5, "status": 1.0, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1093}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 29, "retention_rate": 0.4461538461538462, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5, "status": 1.0, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1090}, "2": {"facts_total": 65, "facts_retained": 29, "retention_rate": 0.4461538461538462, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5, "status": 1.0, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1093}}}, "reference_key_id": "keys/synthetic/syn_03000_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "eba3052ac902ffbeeb4c56e0006f98338aa07742b3d8c7095b3defabe4f67f23", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "4a1c8fee570b40306eb5c9af", "attempt_id": "6c9d0a016a50", "run_group_id": "E4_core:A4|closed|qwen3b:h2", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "d83aa60f20d2879d02898cb9d9d8bd1983ed2d8984cae017c9153cafee3a779a", "git_commit": null, "timestamp_utc": "2026-10-30T06:18:57.136571Z", "task_family": "synthetic", "document_id": "syn_03000_2000", "document_hash": "c9cc27c20b05b1f6e592846e84990ee9b69c42f9720a8e338ec2f6fca777fdbc", "question_id": "q01_single_fact", "split": "test", "cluster_id": "syn_03000_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "laptop-m5max-48g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 2, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1904, "query_tokens": 598, "message_tokens": 512, "message_utf8_bytes": 1093, "cache_tensor_bytes": null, "serialized_payload_bytes": 1093, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 45056.899791932665, "hops": {"1": 22814.76537499111, "2": 22242.134416941553}, "receiver_prefill_ms": 380.7745419908315, "ttft_ms": 55.3709160303697, "decode_ms": 535.0932499859482}, "total_wall_ms": 46028.787125018425, "ttft_ms": 55.3709160303697, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 9}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "Lot 19-N, Gaspé", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 29, "retention_rate": 0.4461538461538462, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5, "status": 1.0, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1093}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 29, "retention_rate": 0.4461538461538462, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5, "status": 1.0, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1090}, "2": {"facts_total": 65, "facts_retained": 29, "retention_rate": 0.4461538461538462, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5, "status": 1.0, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1093}}}, "reference_key_id": "keys/synthetic/syn_03000_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 0.8571428571428571}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "eba3052ac902ffbeeb4c56e0006f98338aa07742b3d8c7095b3defabe4f67f23", "output_hash": "2e8cf782bd118a1c6d4e23417785d72fb92e930806b5752cd1ec405a0c9b95e5"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "686ab44b5b14bb9e27a785a6", "attempt_id": "c4a54b731023", "run_group_id": "E4_core:A4|closed|qwen3b:h2", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "d83aa60f20d2879d02898cb9d9d8bd1983ed2d8984cae017c9153cafee3a779a", "git_commit": null, "timestamp_utc": "2026-10-30T06:18:58.124696Z", "task_family": "synthetic", "document_id": "syn_03000_2000", "document_hash": "c9cc27c20b05b1f6e592846e84990ee9b69c42f9720a8e338ec2f6fca777fdbc", "question_id": "q02_exact_copy", "split": "test", "cluster_id": "syn_03000_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "laptop-m5max-48g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 2, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1904, "query_tokens": 597, "message_tokens": 512, "message_utf8_bytes": 1093, "cache_tensor_bytes": null, "serialized_payload_bytes": 1093, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 45056.899791932665, "hops": {"1": 22814.76537499111, "2": 22242.134416941553}, "receiver_prefill_ms": 388.04666604846716, "ttft_ms": 31.45445801783353, "decode_ms": 41.94899997673929}, "total_wall_ms": 45518.95458390936, "ttft_ms": 31.45445801783353, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 29, "retention_rate": 0.4461538461538462, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5, "status": 1.0, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1093}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 29, "retention_rate": 0.4461538461538462, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5, "status": 1.0, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1090}, "2": {"facts_total": 65, "facts_retained": 29, "retention_rate": 0.4461538461538462, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5, "status": 1.0, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1093}}}, "reference_key_id": "keys/synthetic/syn_03000_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "eba3052ac902ffbeeb4c56e0006f98338aa07742b3d8c7095b3defabe4f67f23", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "9e9330f6f4355e5e787ae1ac", "attempt_id": "838b42808d79", "run_group_id": "E4_core:A4|closed|qwen3b:h2", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "d83aa60f20d2879d02898cb9d9d8bd1983ed2d8984cae017c9153cafee3a779a", "git_commit": null, "timestamp_utc": "2026-10-30T06:18:58.603431Z", "task_family": "synthetic", "document_id": "syn_03000_2000", "document_hash": "c9cc27c20b05b1f6e592846e84990ee9b69c42f9720a8e338ec2f6fca777fdbc", "question_id": "q03_exact_copy", "split": "test", "cluster_id": "syn_03000_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "laptop-m5max-48g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 2, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1904, "query_tokens": 596, "message_tokens": 512, "message_utf8_bytes": 1093, "cache_tensor_bytes": null, "serialized_payload_bytes": 1093, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 45056.899791932665, "hops": {"1": 22814.76537499111, "2": 22242.134416941553}, "receiver_prefill_ms": 386.52991596609354, "ttft_ms": 19.984625047072768, "decode_ms": 209.54441698268056}, "total_wall_ms": 45673.89137588907, "ttft_ms": 19.984625047072768, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 7}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "Kwame Haldorsen", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 29, "retention_rate": 0.4461538461538462, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5, "status": 1.0, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1093}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 29, "retention_rate": 0.4461538461538462, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5, "status": 1.0, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1090}, "2": {"facts_total": 65, "facts_retained": 29, "retention_rate": 0.4461538461538462, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5, "status": 1.0, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1093}}}, "reference_key_id": "keys/synthetic/syn_03000_2000.json", "task_scores": {"correct": 0.0, "abstained": 0.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "eba3052ac902ffbeeb4c56e0006f98338aa07742b3d8c7095b3defabe4f67f23", "output_hash": "175253e7cf463934ec4d54e361a781f323cd87290f01306ed368933bcbd2e25d"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "99d5c1542e02e436dd3bb222", "attempt_id": "70e56352891a", "run_group_id": "E4_core:A4|closed|qwen3b:h2", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "d83aa60f20d2879d02898cb9d9d8bd1983ed2d8984cae017c9153cafee3a779a", "git_commit": null, "timestamp_utc": "2026-10-30T06:18:59.237401Z", "task_family": "synthetic", "document_id": "syn_03000_2000", "document_hash": "c9cc27c20b05b1f6e592846e84990ee9b69c42f9720a8e338ec2f6fca777fdbc", "question_id": "q04_multi_hop", "split": "test", "cluster_id": "syn_03000_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "laptop-m5max-48g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 2, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1904, "query_tokens": 602, "message_tokens": 512, "message_utf8_bytes": 1093, "cache_tensor_bytes": null, "serialized_payload_bytes": 1093, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 45056.899791932665, "hops": {"1": 22814.76537499111, "2": 22242.134416941553}, "receiver_prefill_ms": 550.4078330704942, "ttft_ms": 24.867333937436342, "decode_ms": 298.0961250141263}, "total_wall_ms": 45931.304374942556, "ttft_ms": 24.867333937436342, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 9}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "9,220,745", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 29, "retention_rate": 0.4461538461538462, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5, "status": 1.0, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1093}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 29, "retention_rate": 0.4461538461538462, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5, "status": 1.0, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1090}, "2": {"facts_total": 65, "facts_retained": 29, "retention_rate": 0.4461538461538462, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5, "status": 1.0, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1093}}}, "reference_key_id": "keys/synthetic/syn_03000_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "eba3052ac902ffbeeb4c56e0006f98338aa07742b3d8c7095b3defabe4f67f23", "output_hash": "ae55ba88b2cb1b0f1a4e0960ae312e0a1c177c79a6d14b67bff1677cec19b649"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "5e4f9848d286efbf69ea51cf", "attempt_id": "230da8ef5589", "run_group_id": "E4_core:A4|closed|qwen3b:h2", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "d83aa60f20d2879d02898cb9d9d8bd1983ed2d8984cae017c9153cafee3a779a", "git_commit": null, "timestamp_utc": "2026-10-30T06:19:00.129082Z", "task_family": "synthetic", "document_id": "syn_03000_2000", "document_hash": "c9cc27c20b05b1f6e592846e84990ee9b69c42f9720a8e338ec2f6fca777fdbc", "question_id": "q05_multi_hop", "split": "test", "cluster_id": "syn_03000_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "laptop-m5max-48g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 2, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1904, "query_tokens": 604, "message_tokens": 512, "message_utf8_bytes": 1093, "cache_tensor_bytes": null, "serialized_payload_bytes": 1093, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 45056.899791932665, "hops": {"1": 22814.76537499111, "2": 22242.134416941553}, "receiver_prefill_ms": 414.3490409478545, "ttft_ms": 39.70479196868837, "decode_ms": 392.1610830584541}, "total_wall_ms": 45903.89458392747, "ttft_ms": 39.70479196868837, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 10}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "LY-51E-3199", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 29, "retention_rate": 0.4461538461538462, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5, "status": 1.0, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1093}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 29, "retention_rate": 0.4461538461538462, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5, "status": 1.0, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1090}, "2": {"facts_total": 65, "facts_retained": 29, "retention_rate": 0.4461538461538462, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5, "status": 1.0, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1093}}}, "reference_key_id": "keys/synthetic/syn_03000_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "eba3052ac902ffbeeb4c56e0006f98338aa07742b3d8c7095b3defabe4f67f23", "output_hash": "a1245d1536890a7e4b026366df48441d503f491f782cfd388797b5782443c5c7"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "ab062de107a71f7ea92f89cc", "attempt_id": "dc8baa15e01f", "run_group_id": "E4_core:A4|closed|qwen3b:h2", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "d83aa60f20d2879d02898cb9d9d8bd1983ed2d8984cae017c9153cafee3a779a", "git_commit": null, "timestamp_utc": "2026-10-30T06:19:00.994321Z", "task_family": "synthetic", "document_id": "syn_03000_2000", "document_hash": "c9cc27c20b05b1f6e592846e84990ee9b69c42f9720a8e338ec2f6fca777fdbc", "question_id": "q06_negation", "split": "test", "cluster_id": "syn_03000_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "laptop-m5max-48g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 2, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1904, "query_tokens": 593, "message_tokens": 512, "message_utf8_bytes": 1093, "cache_tensor_bytes": null, "serialized_payload_bytes": 1093, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 45056.899791932665, "hops": {"1": 22814.76537499111, "2": 22242.134416941553}, "receiver_prefill_ms": 521.9191249925643, "ttft_ms": 23.564000031910837, "decode_ms": 38.065000087954104}, "total_wall_ms": 45641.197624849156, "ttft_ms": 23.564000031910837, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "no", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 29, "retention_rate": 0.4461538461538462, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5, "status": 1.0, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1093}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 29, "retention_rate": 0.4461538461538462, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5, "status": 1.0, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1090}, "2": {"facts_total": 65, "facts_retained": 29, "retention_rate": 0.4461538461538462, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5, "status": 1.0, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1093}}}, "reference_key_id": "keys/synthetic/syn_03000_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "eba3052ac902ffbeeb4c56e0006f98338aa07742b3d8c7095b3defabe4f67f23", "output_hash": "9390298f3fb0c5b160498935d79cb139aef28e1c47358b4bbba61862b9c26e59"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "fec8e5a40e259edf5a3b72c2", "attempt_id": "0f17b166f3f9", "run_group_id": "E4_core:A4|closed|qwen3b:h2", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "d83aa60f20d2879d02898cb9d9d8bd1983ed2d8984cae017c9153cafee3a779a", "git_commit": null, "timestamp_utc": "2026-10-30T06:19:01.597396Z", "task_family": "synthetic", "document_id": "syn_03000_2000", "document_hash": "c9cc27c20b05b1f6e592846e84990ee9b69c42f9720a8e338ec2f6fca777fdbc", "question_id": "q07_negation", "split": "test", "cluster_id": "syn_03000_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "laptop-m5max-48g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 2, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1904, "query_tokens": 593, "message_tokens": 512, "message_utf8_bytes": 1093, "cache_tensor_bytes": null, "serialized_payload_bytes": 1093, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 45056.899791932665, "hops": {"1": 22814.76537499111, "2": 22242.134416941553}, "receiver_prefill_ms": 410.35254194866866, "ttft_ms": 27.618209016509354, "decode_ms": 37.0837920345366}, "total_wall_ms": 45532.565000932664, "ttft_ms": 27.618209016509354, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "no", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 29, "retention_rate": 0.4461538461538462, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5, "status": 1.0, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1093}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 29, "retention_rate": 0.4461538461538462, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5, "status": 1.0, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1090}, "2": {"facts_total": 65, "facts_retained": 29, "retention_rate": 0.4461538461538462, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5, "status": 1.0, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1093}}}, "reference_key_id": "keys/synthetic/syn_03000_2000.json", "task_scores": {"correct": 0.0, "abstained": 0.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "eba3052ac902ffbeeb4c56e0006f98338aa07742b3d8c7095b3defabe4f67f23", "output_hash": "9390298f3fb0c5b160498935d79cb139aef28e1c47358b4bbba61862b9c26e59"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "fe78009f0defe22852a4576b", "attempt_id": "3d5b1c8fd163", "run_group_id": "E4_core:A4|closed|qwen3b:h2", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "d83aa60f20d2879d02898cb9d9d8bd1983ed2d8984cae017c9153cafee3a779a", "git_commit": null, "timestamp_utc": "2026-10-30T06:19:02.090588Z", "task_family": "synthetic", "document_id": "syn_03000_2000", "document_hash": "c9cc27c20b05b1f6e592846e84990ee9b69c42f9720a8e338ec2f6fca777fdbc", "question_id": "q08_revision", "split": "test", "cluster_id": "syn_03000_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "laptop-m5max-48g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 2, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1904, "query_tokens": 608, "message_tokens": 512, "message_utf8_bytes": 1093, "cache_tensor_bytes": null, "serialized_payload_bytes": 1093, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 45056.899791932665, "hops": {"1": 22814.76537499111, "2": 22242.134416941553}, "receiver_prefill_ms": 359.63662492576987, "ttft_ms": 52.374834078364074, "decode_ms": 520.3682080609724}, "total_wall_ms": 45989.991749986075, "ttft_ms": 52.374834078364074, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 9}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "2,705,965", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 29, "retention_rate": 0.4461538461538462, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5, "status": 1.0, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1093}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 29, "retention_rate": 0.4461538461538462, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5, "status": 1.0, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1090}, "2": {"facts_total": 65, "facts_retained": 29, "retention_rate": 0.4461538461538462, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5, "status": 1.0, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1093}}}, "reference_key_id": "keys/synthetic/syn_03000_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "eba3052ac902ffbeeb4c56e0006f98338aa07742b3d8c7095b3defabe4f67f23", "output_hash": "dbedb6f5d00e413d915513b85d8ed195b18b9acf392ece342a85e15e3752bcc8"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "fde3bc94674a1d327deb5b49", "attempt_id": "af26862f993a", "run_group_id": "E4_core:A4|closed|qwen3b:h2", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "d83aa60f20d2879d02898cb9d9d8bd1983ed2d8984cae017c9153cafee3a779a", "git_commit": null, "timestamp_utc": "2026-10-30T06:19:03.041077Z", "task_family": "synthetic", "document_id": "syn_03000_2000", "document_hash": "c9cc27c20b05b1f6e592846e84990ee9b69c42f9720a8e338ec2f6fca777fdbc", "question_id": "q09_revision", "split": "test", "cluster_id": "syn_03000_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "laptop-m5max-48g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 2, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1904, "query_tokens": 604, "message_tokens": 512, "message_utf8_bytes": 1093, "cache_tensor_bytes": null, "serialized_payload_bytes": 1093, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 45056.899791932665, "hops": {"1": 22814.76537499111, "2": 22242.134416941553}, "receiver_prefill_ms": 397.4099160404876, "ttft_ms": 39.05604197643697, "decode_ms": 218.73037505429238}, "total_wall_ms": 45712.77904196177, "ttft_ms": 39.05604197643697, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 7}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "9127965", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 29, "retention_rate": 0.4461538461538462, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5, "status": 1.0, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1093}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 29, "retention_rate": 0.4461538461538462, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5, "status": 1.0, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1090}, "2": {"facts_total": 65, "facts_retained": 29, "retention_rate": 0.4461538461538462, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5, "status": 1.0, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1093}}}, "reference_key_id": "keys/synthetic/syn_03000_2000.json", "task_scores": {"correct": 0.0, "abstained": 0.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "eba3052ac902ffbeeb4c56e0006f98338aa07742b3d8c7095b3defabe4f67f23", "output_hash": "a0c871941224304cec47bb1b3d64b8706843b299f14ab94899d97df14f433da7"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "e3d7493ce24117dcda8eb173", "attempt_id": "584810114b0a", "run_group_id": "E4_core:A4|closed|qwen3b:h2", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "d83aa60f20d2879d02898cb9d9d8bd1983ed2d8984cae017c9153cafee3a779a", "git_commit": null, "timestamp_utc": "2026-10-30T06:19:03.713597Z", "task_family": "synthetic", "document_id": "syn_03000_2000", "document_hash": "c9cc27c20b05b1f6e592846e84990ee9b69c42f9720a8e338ec2f6fca777fdbc", "question_id": "q10_temporal", "split": "test", "cluster_id": "syn_03000_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "laptop-m5max-48g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 2, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1904, "query_tokens": 610, "message_tokens": 512, "message_utf8_bytes": 1093, "cache_tensor_bytes": null, "serialized_payload_bytes": 1093, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 45056.899791932665, "hops": {"1": 22814.76537499111, "2": 22242.134416941553}, "receiver_prefill_ms": 352.2907909937203, "ttft_ms": 46.61529196891934, "decode_ms": 324.52550006564707}, "total_wall_ms": 45781.17070894223, "ttft_ms": 46.61529196891934, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 5}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "Felaldnak-7", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 29, "retention_rate": 0.4461538461538462, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5, "status": 1.0, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1093}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 29, "retention_rate": 0.4461538461538462, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5, "status": 1.0, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1090}, "2": {"facts_total": 65, "facts_retained": 29, "retention_rate": 0.4461538461538462, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5, "status": 1.0, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1093}}}, "reference_key_id": "keys/synthetic/syn_03000_2000.json", "task_scores": {"correct": 0.0, "abstained": 0.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "eba3052ac902ffbeeb4c56e0006f98338aa07742b3d8c7095b3defabe4f67f23", "output_hash": "2b9a4cf7221f1b6f31508a6b3078ec521f211026a932da5020cddd512ecc5fa2"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "5fbac960bdc849f2385ce32c", "attempt_id": "1c610ce6b388", "run_group_id": "E4_core:A4|closed|qwen3b:h2", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "d83aa60f20d2879d02898cb9d9d8bd1983ed2d8984cae017c9153cafee3a779a", "git_commit": null, "timestamp_utc": "2026-10-30T06:19:04.454891Z", "task_family": "synthetic", "document_id": "syn_03000_2000", "document_hash": "c9cc27c20b05b1f6e592846e84990ee9b69c42f9720a8e338ec2f6fca777fdbc", "question_id": "q11_aggregation", "split": "test", "cluster_id": "syn_03000_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "laptop-m5max-48g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 2, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1904, "query_tokens": 605, "message_tokens": 512, "message_utf8_bytes": 1093, "cache_tensor_bytes": null, "serialized_payload_bytes": 1093, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 45056.899791932665, "hops": {"1": 22814.76537499111, "2": 22242.134416941553}, "receiver_prefill_ms": 471.24666697345674, "ttft_ms": 33.17520907148719, "decode_ms": 221.2172499857843}, "total_wall_ms": 45783.24533288833, "ttft_ms": 33.17520907148719, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 7}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "9220745", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 29, "retention_rate": 0.4461538461538462, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5, "status": 1.0, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1093}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 29, "retention_rate": 0.4461538461538462, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5, "status": 1.0, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1090}, "2": {"facts_total": 65, "facts_retained": 29, "retention_rate": 0.4461538461538462, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5, "status": 1.0, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1093}}}, "reference_key_id": "keys/synthetic/syn_03000_2000.json", "task_scores": {"correct": 0.0, "abstained": 0.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "eba3052ac902ffbeeb4c56e0006f98338aa07742b3d8c7095b3defabe4f67f23", "output_hash": "f5d77aa763f6365bc80d01b9e6e23bc43fb570ac45771d7bd628c75cde727b35"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "269faf8ca294114493ea3b1d", "attempt_id": "8fce7f565a82", "run_group_id": "E4_core:A4|closed|qwen3b:h2", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "d83aa60f20d2879d02898cb9d9d8bd1983ed2d8984cae017c9153cafee3a779a", "git_commit": null, "timestamp_utc": "2026-10-30T06:19:05.197756Z", "task_family": "synthetic", "document_id": "syn_03000_2000", "document_hash": "c9cc27c20b05b1f6e592846e84990ee9b69c42f9720a8e338ec2f6fca777fdbc", "question_id": "q12_unit_conversion", "split": "test", "cluster_id": "syn_03000_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "laptop-m5max-48g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 2, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1904, "query_tokens": 601, "message_tokens": 512, "message_utf8_bytes": 1093, "cache_tensor_bytes": null, "serialized_payload_bytes": 1093, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 45056.899791932665, "hops": {"1": 22814.76537499111, "2": 22242.134416941553}, "receiver_prefill_ms": 374.5994580676779, "ttft_ms": 19.870541989803314, "decode_ms": 57.566832983866334}, "total_wall_ms": 45509.596082847565, "ttft_ms": 19.870541989803314, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 29, "retention_rate": 0.4461538461538462, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5, "status": 1.0, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1093}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 29, "retention_rate": 0.4461538461538462, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5, "status": 1.0, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1090}, "2": {"facts_total": 65, "facts_retained": 29, "retention_rate": 0.4461538461538462, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5, "status": 1.0, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1093}}}, "reference_key_id": "keys/synthetic/syn_03000_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "eba3052ac902ffbeeb4c56e0006f98338aa07742b3d8c7095b3defabe4f67f23", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "11e5af53dd71fbcffe07cd5d", "attempt_id": "78e937433131", "run_group_id": "E4_core:A4|closed|qwen3b:h2", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "d83aa60f20d2879d02898cb9d9d8bd1983ed2d8984cae017c9153cafee3a779a", "git_commit": null, "timestamp_utc": "2026-10-30T06:19:05.667225Z", "task_family": "synthetic", "document_id": "syn_03000_2000", "document_hash": "c9cc27c20b05b1f6e592846e84990ee9b69c42f9720a8e338ec2f6fca777fdbc", "question_id": "q13_unanswerable", "split": "test", "cluster_id": "syn_03000_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "laptop-m5max-48g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 2, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1904, "query_tokens": 605, "message_tokens": 512, "message_utf8_bytes": 1093, "cache_tensor_bytes": null, "serialized_payload_bytes": 1093, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 45056.899791932665, "hops": {"1": 22814.76537499111, "2": 22242.134416941553}, "receiver_prefill_ms": 386.82395801879466, "ttft_ms": 42.738125077448785, "decode_ms": 87.25770900491625}, "total_wall_ms": 45574.52554197516, "ttft_ms": 42.738125077448785, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 29, "retention_rate": 0.4461538461538462, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5, "status": 1.0, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1093}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 29, "retention_rate": 0.4461538461538462, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5, "status": 1.0, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1090}, "2": {"facts_total": 65, "facts_retained": 29, "retention_rate": 0.4461538461538462, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5, "status": 1.0, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1093}}}, "reference_key_id": "keys/synthetic/syn_03000_2000.json", "task_scores": {"correct": 1.0, "abstained": 1.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "eba3052ac902ffbeeb4c56e0006f98338aa07742b3d8c7095b3defabe4f67f23", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "7e6e87708f26668cf148c2d1", "attempt_id": "71f774b94caa", "run_group_id": "E4_core:A4|closed|qwen3b:h2", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "d83aa60f20d2879d02898cb9d9d8bd1983ed2d8984cae017c9153cafee3a779a", "git_commit": null, "timestamp_utc": "2026-10-30T06:19:06.202927Z", "task_family": "synthetic", "document_id": "syn_03000_2000", "document_hash": "c9cc27c20b05b1f6e592846e84990ee9b69c42f9720a8e338ec2f6fca777fdbc", "question_id": "q14_unanswerable", "split": "test", "cluster_id": "syn_03000_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "laptop-m5max-48g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 2, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1904, "query_tokens": 607, "message_tokens": 512, "message_utf8_bytes": 1093, "cache_tensor_bytes": null, "serialized_payload_bytes": 1093, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 45056.899791932665, "hops": {"1": 22814.76537499111, "2": 22242.134416941553}, "receiver_prefill_ms": 467.1932499622926, "ttft_ms": 38.42274995986372, "decode_ms": 26.6235409071669}, "total_wall_ms": 45589.72187596373, "ttft_ms": 38.42274995986372, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 29, "retention_rate": 0.4461538461538462, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5, "status": 1.0, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1093}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 29, "retention_rate": 0.4461538461538462, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5, "status": 1.0, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1090}, "2": {"facts_total": 65, "facts_retained": 29, "retention_rate": 0.4461538461538462, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5, "status": 1.0, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1093}}}, "reference_key_id": "keys/synthetic/syn_03000_2000.json", "task_scores": {"correct": 1.0, "abstained": 1.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "eba3052ac902ffbeeb4c56e0006f98338aa07742b3d8c7095b3defabe4f67f23", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "44f164648ee5f2e7b0725645", "attempt_id": "a6d2c28a4e8a", "run_group_id": "E4_core:A4|closed|qwen3b:h4", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "d83aa60f20d2879d02898cb9d9d8bd1983ed2d8984cae017c9153cafee3a779a", "git_commit": null, "timestamp_utc": "2026-10-30T06:19:50.763376Z", "task_family": "synthetic", "document_id": "syn_03000_2000", "document_hash": "c9cc27c20b05b1f6e592846e84990ee9b69c42f9720a8e338ec2f6fca777fdbc", "question_id": "q00_single_fact", "split": "test", "cluster_id": "syn_03000_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "laptop-m5max-48g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 4, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1904, "query_tokens": 593, "message_tokens": 512, "message_utf8_bytes": 1093, "cache_tensor_bytes": null, "serialized_payload_bytes": 1093, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 89064.99858293682, "hops": {"1": 22814.76537499111, "2": 22242.134416941553, "3": 21933.44883294776, "4": 22074.649958056398}, "receiver_prefill_ms": 361.20483302511275, "ttft_ms": 42.18916699755937, "decode_ms": 46.98849993292242}, "total_wall_ms": 89516.2011248758, "ttft_ms": 42.18916699755937, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 29, "retention_rate": 0.4461538461538462, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5, "status": 1.0, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1093}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 29, "retention_rate": 0.4461538461538462, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5, "status": 1.0, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1090}, "2": {"facts_total": 65, "facts_retained": 29, "retention_rate": 0.4461538461538462, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5, "status": 1.0, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1093}, "3": {"facts_total": 65, "facts_retained": 29, "retention_rate": 0.4461538461538462, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5, "status": 1.0, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1089}, "4": {"facts_total": 65, "facts_retained": 29, "retention_rate": 0.4461538461538462, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5, "status": 1.0, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1093}}}, "reference_key_id": "keys/synthetic/syn_03000_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "eba3052ac902ffbeeb4c56e0006f98338aa07742b3d8c7095b3defabe4f67f23", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "554a1265129d9ecc7a1eef2c", "attempt_id": "1fa123b9952d", "run_group_id": "E4_core:A4|closed|qwen3b:h4", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "d83aa60f20d2879d02898cb9d9d8bd1983ed2d8984cae017c9153cafee3a779a", "git_commit": null, "timestamp_utc": "2026-10-30T06:19:51.232444Z", "task_family": "synthetic", "document_id": "syn_03000_2000", "document_hash": "c9cc27c20b05b1f6e592846e84990ee9b69c42f9720a8e338ec2f6fca777fdbc", "question_id": "q01_single_fact", "split": "test", "cluster_id": "syn_03000_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "laptop-m5max-48g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 4, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1904, "query_tokens": 598, "message_tokens": 512, "message_utf8_bytes": 1093, "cache_tensor_bytes": null, "serialized_payload_bytes": 1093, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 89064.99858293682, "hops": {"1": 22814.76537499111, "2": 22242.134416941553, "3": 21933.44883294776, "4": 22074.649958056398}, "receiver_prefill_ms": 512.0456670410931, "ttft_ms": 33.16612495109439, "decode_ms": 309.13687497377396}, "total_wall_ms": 89920.03854096401, "ttft_ms": 33.16612495109439, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 9}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "Lot 19-N, Gaspé", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 29, "retention_rate": 0.4461538461538462, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5, "status": 1.0, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1093}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 29, "retention_rate": 0.4461538461538462, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5, "status": 1.0, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1090}, "2": {"facts_total": 65, "facts_retained": 29, "retention_rate": 0.4461538461538462, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5, "status": 1.0, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1093}, "3": {"facts_total": 65, "facts_retained": 29, "retention_rate": 0.4461538461538462, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5, "status": 1.0, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1089}, "4": {"facts_total": 65, "facts_retained": 29, "retention_rate": 0.4461538461538462, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5, "status": 1.0, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1093}}}, "reference_key_id": "keys/synthetic/syn_03000_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 0.8571428571428571}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "eba3052ac902ffbeeb4c56e0006f98338aa07742b3d8c7095b3defabe4f67f23", "output_hash": "2e8cf782bd118a1c6d4e23417785d72fb92e930806b5752cd1ec405a0c9b95e5"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "47e5982e94af627a0ccb6347", "attempt_id": "24f9f8f6f9fc", "run_group_id": "E4_core:A4|closed|qwen3b:h4", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "d83aa60f20d2879d02898cb9d9d8bd1983ed2d8984cae017c9153cafee3a779a", "git_commit": null, "timestamp_utc": "2026-10-30T06:19:52.104169Z", "task_family": "synthetic", "document_id": "syn_03000_2000", "document_hash": "c9cc27c20b05b1f6e592846e84990ee9b69c42f9720a8e338ec2f6fca777fdbc", "question_id": "q02_exact_copy", "split": "test", "cluster_id": "syn_03000_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "laptop-m5max-48g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 4, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1904, "query_tokens": 597, "message_tokens": 512, "message_utf8_bytes": 1093, "cache_tensor_bytes": null, "serialized_payload_bytes": 1093, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 89064.99858293682, "hops": {"1": 22814.76537499111, "2": 22242.134416941553, "3": 21933.44883294776, "4": 22074.649958056398}, "receiver_prefill_ms": 382.54683394916356, "ttft_ms": 20.289666019380093, "decode_ms": 20.50258289091289}, "total_wall_ms": 89488.9422499109, "ttft_ms": 20.289666019380093, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 29, "retention_rate": 0.4461538461538462, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5, "status": 1.0, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1093}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 29, "retention_rate": 0.4461538461538462, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5, "status": 1.0, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1090}, "2": {"facts_total": 65, "facts_retained": 29, "retention_rate": 0.4461538461538462, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5, "status": 1.0, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1093}, "3": {"facts_total": 65, "facts_retained": 29, "retention_rate": 0.4461538461538462, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5, "status": 1.0, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1089}, "4": {"facts_total": 65, "facts_retained": 29, "retention_rate": 0.4461538461538462, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5, "status": 1.0, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1093}}}, "reference_key_id": "keys/synthetic/syn_03000_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "eba3052ac902ffbeeb4c56e0006f98338aa07742b3d8c7095b3defabe4f67f23", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "35b259907ed7946658d698d5", "attempt_id": "8389587b6b87", "run_group_id": "E4_core:A4|closed|qwen3b:h4", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "d83aa60f20d2879d02898cb9d9d8bd1983ed2d8984cae017c9153cafee3a779a", "git_commit": null, "timestamp_utc": "2026-10-30T06:19:52.544966Z", "task_family": "synthetic", "document_id": "syn_03000_2000", "document_hash": "c9cc27c20b05b1f6e592846e84990ee9b69c42f9720a8e338ec2f6fca777fdbc", "question_id": "q03_exact_copy", "split": "test", "cluster_id": "syn_03000_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "laptop-m5max-48g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 4, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1904, "query_tokens": 596, "message_tokens": 512, "message_utf8_bytes": 1093, "cache_tensor_bytes": null, "serialized_payload_bytes": 1093, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 89064.99858293682, "hops": {"1": 22814.76537499111, "2": 22242.134416941553, "3": 21933.44883294776, "4": 22074.649958056398}, "receiver_prefill_ms": 378.74716706573963, "ttft_ms": 46.51391599327326, "decode_ms": 433.9874580036849}, "total_wall_ms": 89924.95554091875, "ttft_ms": 46.51391599327326, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 7}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "Kwame Haldorsen", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 29, "retention_rate": 0.4461538461538462, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5, "status": 1.0, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1093}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 29, "retention_rate": 0.4461538461538462, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5, "status": 1.0, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1090}, "2": {"facts_total": 65, "facts_retained": 29, "retention_rate": 0.4461538461538462, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5, "status": 1.0, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1093}, "3": {"facts_total": 65, "facts_retained": 29, "retention_rate": 0.4461538461538462, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5, "status": 1.0, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1089}, "4": {"facts_total": 65, "facts_retained": 29, "retention_rate": 0.4461538461538462, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5, "status": 1.0, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1093}}}, "reference_key_id": "keys/synthetic/syn_03000_2000.json", "task_scores": {"correct": 0.0, "abstained": 0.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "eba3052ac902ffbeeb4c56e0006f98338aa07742b3d8c7095b3defabe4f67f23", "output_hash": "175253e7cf463934ec4d54e361a781f323cd87290f01306ed368933bcbd2e25d"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "a61045cddb28e699e1855868", "attempt_id": "734b0afdeb0e", "run_group_id": "E4_core:A4|closed|qwen3b:h4", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "d83aa60f20d2879d02898cb9d9d8bd1983ed2d8984cae017c9153cafee3a779a", "git_commit": null, "timestamp_utc": "2026-10-30T06:19:53.423708Z", "task_family": "synthetic", "document_id": "syn_03000_2000", "document_hash": "c9cc27c20b05b1f6e592846e84990ee9b69c42f9720a8e338ec2f6fca777fdbc", "question_id": "q04_multi_hop", "split": "test", "cluster_id": "syn_03000_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "laptop-m5max-48g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 4, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1904, "query_tokens": 602, "message_tokens": 512, "message_utf8_bytes": 1093, "cache_tensor_bytes": null, "serialized_payload_bytes": 1093, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 89064.99858293682, "hops": {"1": 22814.76537499111, "2": 22242.134416941553, "3": 21933.44883294776, "4": 22074.649958056398}, "receiver_prefill_ms": 420.9924169117585, "ttft_ms": 20.388833014294505, "decode_ms": 273.7671249778941}, "total_wall_ms": 89780.7689999463, "ttft_ms": 20.388833014294505, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 9}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "9,220,745", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 29, "retention_rate": 0.4461538461538462, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5, "status": 1.0, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1093}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 29, "retention_rate": 0.4461538461538462, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5, "status": 1.0, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1090}, "2": {"facts_total": 65, "facts_retained": 29, "retention_rate": 0.4461538461538462, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5, "status": 1.0, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1093}, "3": {"facts_total": 65, "facts_retained": 29, "retention_rate": 0.4461538461538462, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5, "status": 1.0, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1089}, "4": {"facts_total": 65, "facts_retained": 29, "retention_rate": 0.4461538461538462, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5, "status": 1.0, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1093}}}, "reference_key_id": "keys/synthetic/syn_03000_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "eba3052ac902ffbeeb4c56e0006f98338aa07742b3d8c7095b3defabe4f67f23", "output_hash": "ae55ba88b2cb1b0f1a4e0960ae312e0a1c177c79a6d14b67bff1677cec19b649"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "d25ca6c03ff1ce73fb5e893b", "attempt_id": "aa8bbe1b4704", "run_group_id": "E4_core:A4|closed|qwen3b:h4", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "d83aa60f20d2879d02898cb9d9d8bd1983ed2d8984cae017c9153cafee3a779a", "git_commit": null, "timestamp_utc": "2026-10-30T06:19:54.156273Z", "task_family": "synthetic", "document_id": "syn_03000_2000", "document_hash": "c9cc27c20b05b1f6e592846e84990ee9b69c42f9720a8e338ec2f6fca777fdbc", "question_id": "q05_multi_hop", "split": "test", "cluster_id": "syn_03000_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "laptop-m5max-48g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 4, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1904, "query_tokens": 604, "message_tokens": 512, "message_utf8_bytes": 1093, "cache_tensor_bytes": null, "serialized_payload_bytes": 1093, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 89064.99858293682, "hops": {"1": 22814.76537499111, "2": 22242.134416941553, "3": 21933.44883294776, "4": 22074.649958056398}, "receiver_prefill_ms": 390.47629199922085, "ttft_ms": 54.488625028170645, "decode_ms": 506.95725006517023}, "total_wall_ms": 90017.57541589905, "ttft_ms": 54.488625028170645, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 10}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "LY-51E-3199", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 29, "retention_rate": 0.4461538461538462, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5, "status": 1.0, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1093}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 29, "retention_rate": 0.4461538461538462, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5, "status": 1.0, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1090}, "2": {"facts_total": 65, "facts_retained": 29, "retention_rate": 0.4461538461538462, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5, "status": 1.0, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1093}, "3": {"facts_total": 65, "facts_retained": 29, "retention_rate": 0.4461538461538462, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5, "status": 1.0, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1089}, "4": {"facts_total": 65, "facts_retained": 29, "retention_rate": 0.4461538461538462, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5, "status": 1.0, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1093}}}, "reference_key_id": "keys/synthetic/syn_03000_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "eba3052ac902ffbeeb4c56e0006f98338aa07742b3d8c7095b3defabe4f67f23", "output_hash": "a1245d1536890a7e4b026366df48441d503f491f782cfd388797b5782443c5c7"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "d7878e969948e25ab2e16879", "attempt_id": "7e264ade60e4", "run_group_id": "E4_core:A4|closed|qwen3b:h4", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "d83aa60f20d2879d02898cb9d9d8bd1983ed2d8984cae017c9153cafee3a779a", "git_commit": null, "timestamp_utc": "2026-10-30T06:19:55.127852Z", "task_family": "synthetic", "document_id": "syn_03000_2000", "document_hash": "c9cc27c20b05b1f6e592846e84990ee9b69c42f9720a8e338ec2f6fca777fdbc", "question_id": "q06_negation", "split": "test", "cluster_id": "syn_03000_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "laptop-m5max-48g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 4, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1904, "query_tokens": 593, "message_tokens": 512, "message_utf8_bytes": 1093, "cache_tensor_bytes": null, "serialized_payload_bytes": 1093, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 89064.99858293682, "hops": {"1": 22814.76537499111, "2": 22242.134416941553, "3": 21933.44883294776, "4": 22074.649958056398}, "receiver_prefill_ms": 402.01787499245256, "ttft_ms": 34.773999941535294, "decode_ms": 29.154542018659413}, "total_wall_ms": 89531.48562496062, "ttft_ms": 34.773999941535294, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "no", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 29, "retention_rate": 0.4461538461538462, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5, "status": 1.0, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1093}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 29, "retention_rate": 0.4461538461538462, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5, "status": 1.0, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1090}, "2": {"facts_total": 65, "facts_retained": 29, "retention_rate": 0.4461538461538462, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5, "status": 1.0, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1093}, "3": {"facts_total": 65, "facts_retained": 29, "retention_rate": 0.4461538461538462, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5, "status": 1.0, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1089}, "4": {"facts_total": 65, "facts_retained": 29, "retention_rate": 0.4461538461538462, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5, "status": 1.0, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1093}}}, "reference_key_id": "keys/synthetic/syn_03000_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "eba3052ac902ffbeeb4c56e0006f98338aa07742b3d8c7095b3defabe4f67f23", "output_hash": "9390298f3fb0c5b160498935d79cb139aef28e1c47358b4bbba61862b9c26e59"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "5591b3ad5596ac36e017017e", "attempt_id": "f1f931b86a58", "run_group_id": "E4_core:A4|closed|qwen3b:h4", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "d83aa60f20d2879d02898cb9d9d8bd1983ed2d8984cae017c9153cafee3a779a", "git_commit": null, "timestamp_utc": "2026-10-30T06:19:55.610088Z", "task_family": "synthetic", "document_id": "syn_03000_2000", "document_hash": "c9cc27c20b05b1f6e592846e84990ee9b69c42f9720a8e338ec2f6fca777fdbc", "question_id": "q07_negation", "split": "test", "cluster_id": "syn_03000_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "laptop-m5max-48g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 4, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1904, "query_tokens": 593, "message_tokens": 512, "message_utf8_bytes": 1093, "cache_tensor_bytes": null, "serialized_payload_bytes": 1093, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 89064.99858293682, "hops": {"1": 22814.76537499111, "2": 22242.134416941553, "3": 21933.44883294776, "4": 22074.649958056398}, "receiver_prefill_ms": 332.03433302696794, "ttft_ms": 36.80420794989914, "decode_ms": 54.73487498238683}, "total_wall_ms": 89489.46795798838, "ttft_ms": 36.80420794989914, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "no", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 29, "retention_rate": 0.4461538461538462, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5, "status": 1.0, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1093}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 29, "retention_rate": 0.4461538461538462, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5, "status": 1.0, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1090}, "2": {"facts_total": 65, "facts_retained": 29, "retention_rate": 0.4461538461538462, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5, "status": 1.0, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1093}, "3": {"facts_total": 65, "facts_retained": 29, "retention_rate": 0.4461538461538462, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5, "status": 1.0, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1089}, "4": {"facts_total": 65, "facts_retained": 29, "retention_rate": 0.4461538461538462, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5, "status": 1.0, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1093}}}, "reference_key_id": "keys/synthetic/syn_03000_2000.json", "task_scores": {"correct": 0.0, "abstained": 0.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "eba3052ac902ffbeeb4c56e0006f98338aa07742b3d8c7095b3defabe4f67f23", "output_hash": "9390298f3fb0c5b160498935d79cb139aef28e1c47358b4bbba61862b9c26e59"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "c2747f9009ac51d5f23283ff", "attempt_id": "fee622c68cc1", "run_group_id": "E4_core:A4|closed|qwen3b:h4", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "d83aa60f20d2879d02898cb9d9d8bd1983ed2d8984cae017c9153cafee3a779a", "git_commit": null, "timestamp_utc": "2026-10-30T06:19:56.051807Z", "task_family": "synthetic", "document_id": "syn_03000_2000", "document_hash": "c9cc27c20b05b1f6e592846e84990ee9b69c42f9720a8e338ec2f6fca777fdbc", "question_id": "q08_revision", "split": "test", "cluster_id": "syn_03000_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "laptop-m5max-48g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 4, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1904, "query_tokens": 608, "message_tokens": 512, "message_utf8_bytes": 1093, "cache_tensor_bytes": null, "serialized_payload_bytes": 1093, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 89064.99858293682, "hops": {"1": 22814.76537499111, "2": 22242.134416941553, "3": 21933.44883294776, "4": 22074.649958056398}, "receiver_prefill_ms": 427.5656669633463, "ttft_ms": 138.7366249691695, "decode_ms": 308.35883307736367}, "total_wall_ms": 89940.31033292413, "ttft_ms": 138.7366249691695, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 9}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "2,705,965", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 29, "retention_rate": 0.4461538461538462, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5, "status": 1.0, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1093}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 29, "retention_rate": 0.4461538461538462, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5, "status": 1.0, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1090}, "2": {"facts_total": 65, "facts_retained": 29, "retention_rate": 0.4461538461538462, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5, "status": 1.0, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1093}, "3": {"facts_total": 65, "facts_retained": 29, "retention_rate": 0.4461538461538462, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5, "status": 1.0, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1089}, "4": {"facts_total": 65, "facts_retained": 29, "retention_rate": 0.4461538461538462, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5, "status": 1.0, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1093}}}, "reference_key_id": "keys/synthetic/syn_03000_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "eba3052ac902ffbeeb4c56e0006f98338aa07742b3d8c7095b3defabe4f67f23", "output_hash": "dbedb6f5d00e413d915513b85d8ed195b18b9acf392ece342a85e15e3752bcc8"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "7d8ef9d005a1a95d537d647b", "attempt_id": "dad255aaf0e2", "run_group_id": "E4_core:A4|closed|qwen3b:h4", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "d83aa60f20d2879d02898cb9d9d8bd1983ed2d8984cae017c9153cafee3a779a", "git_commit": null, "timestamp_utc": "2026-10-30T06:19:56.943768Z", "task_family": "synthetic", "document_id": "syn_03000_2000", "document_hash": "c9cc27c20b05b1f6e592846e84990ee9b69c42f9720a8e338ec2f6fca777fdbc", "question_id": "q09_revision", "split": "test", "cluster_id": "syn_03000_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "laptop-m5max-48g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 4, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1904, "query_tokens": 604, "message_tokens": 512, "message_utf8_bytes": 1093, "cache_tensor_bytes": null, "serialized_payload_bytes": 1093, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 89064.99858293682, "hops": {"1": 22814.76537499111, "2": 22242.134416941553, "3": 21933.44883294776, "4": 22074.649958056398}, "receiver_prefill_ms": 403.0348750529811, "ttft_ms": 39.727124967612326, "decode_ms": 232.01500007417053}, "total_wall_ms": 89740.5188329285, "ttft_ms": 39.727124967612326, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 7}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "9127965", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 29, "retention_rate": 0.4461538461538462, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5, "status": 1.0, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1093}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 29, "retention_rate": 0.4461538461538462, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5, "status": 1.0, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1090}, "2": {"facts_total": 65, "facts_retained": 29, "retention_rate": 0.4461538461538462, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5, "status": 1.0, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1093}, "3": {"facts_total": 65, "facts_retained": 29, "retention_rate": 0.4461538461538462, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5, "status": 1.0, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1089}, "4": {"facts_total": 65, "facts_retained": 29, "retention_rate": 0.4461538461538462, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5, "status": 1.0, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1093}}}, "reference_key_id": "keys/synthetic/syn_03000_2000.json", "task_scores": {"correct": 0.0, "abstained": 0.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "eba3052ac902ffbeeb4c56e0006f98338aa07742b3d8c7095b3defabe4f67f23", "output_hash": "a0c871941224304cec47bb1b3d64b8706843b299f14ab94899d97df14f433da7"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "7a53a84ae8cfcdf11b8b2f37", "attempt_id": "ec6daf52465d", "run_group_id": "E4_core:A4|closed|qwen3b:h4", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "d83aa60f20d2879d02898cb9d9d8bd1983ed2d8984cae017c9153cafee3a779a", "git_commit": null, "timestamp_utc": "2026-10-30T06:19:57.635972Z", "task_family": "synthetic", "document_id": "syn_03000_2000", "document_hash": "c9cc27c20b05b1f6e592846e84990ee9b69c42f9720a8e338ec2f6fca777fdbc", "question_id": "q10_temporal", "split": "test", "cluster_id": "syn_03000_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "laptop-m5max-48g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 4, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1904, "query_tokens": 610, "message_tokens": 512, "message_utf8_bytes": 1093, "cache_tensor_bytes": null, "serialized_payload_bytes": 1093, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 89064.99858293682, "hops": {"1": 22814.76537499111, "2": 22242.134416941553, "3": 21933.44883294776, "4": 22074.649958056398}, "receiver_prefill_ms": 467.3550829757005, "ttft_ms": 124.41525002941489, "decode_ms": 174.48304197750986}, "total_wall_ms": 89831.88516600057, "ttft_ms": 124.41525002941489, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 5}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "Felaldnak-7", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 29, "retention_rate": 0.4461538461538462, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5, "status": 1.0, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1093}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 29, "retention_rate": 0.4461538461538462, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5, "status": 1.0, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1090}, "2": {"facts_total": 65, "facts_retained": 29, "retention_rate": 0.4461538461538462, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5, "status": 1.0, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1093}, "3": {"facts_total": 65, "facts_retained": 29, "retention_rate": 0.4461538461538462, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5, "status": 1.0, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1089}, "4": {"facts_total": 65, "facts_retained": 29, "retention_rate": 0.4461538461538462, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5, "status": 1.0, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1093}}}, "reference_key_id": "keys/synthetic/syn_03000_2000.json", "task_scores": {"correct": 0.0, "abstained": 0.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "eba3052ac902ffbeeb4c56e0006f98338aa07742b3d8c7095b3defabe4f67f23", "output_hash": "2b9a4cf7221f1b6f31508a6b3078ec521f211026a932da5020cddd512ecc5fa2"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "693b193ed3fd87db57b11a26", "attempt_id": "45dc5d212de9", "run_group_id": "E4_core:A4|closed|qwen3b:h4", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "d83aa60f20d2879d02898cb9d9d8bd1983ed2d8984cae017c9153cafee3a779a", "git_commit": null, "timestamp_utc": "2026-10-30T06:19:58.419544Z", "task_family": "synthetic", "document_id": "syn_03000_2000", "document_hash": "c9cc27c20b05b1f6e592846e84990ee9b69c42f9720a8e338ec2f6fca777fdbc", "question_id": "q11_aggregation", "split": "test", "cluster_id": "syn_03000_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "laptop-m5max-48g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 4, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1904, "query_tokens": 605, "message_tokens": 512, "message_utf8_bytes": 1093, "cache_tensor_bytes": null, "serialized_payload_bytes": 1093, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 89064.99858293682, "hops": {"1": 22814.76537499111, "2": 22242.134416941553, "3": 21933.44883294776, "4": 22074.649958056398}, "receiver_prefill_ms": 398.0105830123648, "ttft_ms": 24.239875026978552, "decode_ms": 249.00270905345678}, "total_wall_ms": 89736.86924995854, "ttft_ms": 24.239875026978552, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 7}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "9220745", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 29, "retention_rate": 0.4461538461538462, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5, "status": 1.0, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1093}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 29, "retention_rate": 0.4461538461538462, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5, "status": 1.0, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1090}, "2": {"facts_total": 65, "facts_retained": 29, "retention_rate": 0.4461538461538462, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5, "status": 1.0, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1093}, "3": {"facts_total": 65, "facts_retained": 29, "retention_rate": 0.4461538461538462, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5, "status": 1.0, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1089}, "4": {"facts_total": 65, "facts_retained": 29, "retention_rate": 0.4461538461538462, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5, "status": 1.0, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1093}}}, "reference_key_id": "keys/synthetic/syn_03000_2000.json", "task_scores": {"correct": 0.0, "abstained": 0.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "eba3052ac902ffbeeb4c56e0006f98338aa07742b3d8c7095b3defabe4f67f23", "output_hash": "f5d77aa763f6365bc80d01b9e6e23bc43fb570ac45771d7bd628c75cde727b35"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "56003d3ae56967de91a5bacf", "attempt_id": "c72d71e01e95", "run_group_id": "E4_core:A4|closed|qwen3b:h4", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "d83aa60f20d2879d02898cb9d9d8bd1983ed2d8984cae017c9153cafee3a779a", "git_commit": null, "timestamp_utc": "2026-10-30T06:19:59.108209Z", "task_family": "synthetic", "document_id": "syn_03000_2000", "document_hash": "c9cc27c20b05b1f6e592846e84990ee9b69c42f9720a8e338ec2f6fca777fdbc", "question_id": "q12_unit_conversion", "split": "test", "cluster_id": "syn_03000_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "laptop-m5max-48g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 4, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1904, "query_tokens": 601, "message_tokens": 512, "message_utf8_bytes": 1093, "cache_tensor_bytes": null, "serialized_payload_bytes": 1093, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 89064.99858293682, "hops": {"1": 22814.76537499111, "2": 22242.134416941553, "3": 21933.44883294776, "4": 22074.649958056398}, "receiver_prefill_ms": 419.62162498384714, "ttft_ms": 40.42637499514967, "decode_ms": 38.98183396086097}, "total_wall_ms": 89564.99745801557, "ttft_ms": 40.42637499514967, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 29, "retention_rate": 0.4461538461538462, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5, "status": 1.0, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1093}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 29, "retention_rate": 0.4461538461538462, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5, "status": 1.0, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1090}, "2": {"facts_total": 65, "facts_retained": 29, "retention_rate": 0.4461538461538462, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5, "status": 1.0, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1093}, "3": {"facts_total": 65, "facts_retained": 29, "retention_rate": 0.4461538461538462, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5, "status": 1.0, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1089}, "4": {"facts_total": 65, "facts_retained": 29, "retention_rate": 0.4461538461538462, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5, "status": 1.0, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1093}}}, "reference_key_id": "keys/synthetic/syn_03000_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "eba3052ac902ffbeeb4c56e0006f98338aa07742b3d8c7095b3defabe4f67f23", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "9be4e0e0dbe9f3c0909aba0d", "attempt_id": "726d925310a4", "run_group_id": "E4_core:A4|closed|qwen3b:h4", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "d83aa60f20d2879d02898cb9d9d8bd1983ed2d8984cae017c9153cafee3a779a", "git_commit": null, "timestamp_utc": "2026-10-30T06:19:59.625934Z", "task_family": "synthetic", "document_id": "syn_03000_2000", "document_hash": "c9cc27c20b05b1f6e592846e84990ee9b69c42f9720a8e338ec2f6fca777fdbc", "question_id": "q13_unanswerable", "split": "test", "cluster_id": "syn_03000_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "laptop-m5max-48g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 4, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1904, "query_tokens": 605, "message_tokens": 512, "message_utf8_bytes": 1093, "cache_tensor_bytes": null, "serialized_payload_bytes": 1093, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 89064.99858293682, "hops": {"1": 22814.76537499111, "2": 22242.134416941553, "3": 21933.44883294776, "4": 22074.649958056398}, "receiver_prefill_ms": 517.6649589557201, "ttft_ms": 37.52737492322922, "decode_ms": 26.787209091708064}, "total_wall_ms": 89647.66233297996, "ttft_ms": 37.52737492322922, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 29, "retention_rate": 0.4461538461538462, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5, "status": 1.0, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1093}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 29, "retention_rate": 0.4461538461538462, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5, "status": 1.0, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1090}, "2": {"facts_total": 65, "facts_retained": 29, "retention_rate": 0.4461538461538462, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5, "status": 1.0, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1093}, "3": {"facts_total": 65, "facts_retained": 29, "retention_rate": 0.4461538461538462, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5, "status": 1.0, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1089}, "4": {"facts_total": 65, "facts_retained": 29, "retention_rate": 0.4461538461538462, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5, "status": 1.0, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1093}}}, "reference_key_id": "keys/synthetic/syn_03000_2000.json", "task_scores": {"correct": 1.0, "abstained": 1.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "eba3052ac902ffbeeb4c56e0006f98338aa07742b3d8c7095b3defabe4f67f23", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "813815ba579c86e483d35034", "attempt_id": "e2a6292d820c", "run_group_id": "E4_core:A4|closed|qwen3b:h4", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "d83aa60f20d2879d02898cb9d9d8bd1983ed2d8984cae017c9153cafee3a779a", "git_commit": null, "timestamp_utc": "2026-10-30T06:20:00.224871Z", "task_family": "synthetic", "document_id": "syn_03000_2000", "document_hash": "c9cc27c20b05b1f6e592846e84990ee9b69c42f9720a8e338ec2f6fca777fdbc", "question_id": "q14_unanswerable", "split": "test", "cluster_id": "syn_03000_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "laptop-m5max-48g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 4, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1904, "query_tokens": 607, "message_tokens": 512, "message_utf8_bytes": 1093, "cache_tensor_bytes": null, "serialized_payload_bytes": 1093, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 89064.99858293682, "hops": {"1": 22814.76537499111, "2": 22242.134416941553, "3": 21933.44883294776, "4": 22074.649958056398}, "receiver_prefill_ms": 402.01675007119775, "ttft_ms": 39.02612498495728, "decode_ms": 23.520249989815056}, "total_wall_ms": 89530.24104097858, "ttft_ms": 39.02612498495728, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 29, "retention_rate": 0.4461538461538462, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5, "status": 1.0, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1093}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 29, "retention_rate": 0.4461538461538462, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5, "status": 1.0, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1090}, "2": {"facts_total": 65, "facts_retained": 29, "retention_rate": 0.4461538461538462, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5, "status": 1.0, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1093}, "3": {"facts_total": 65, "facts_retained": 29, "retention_rate": 0.4461538461538462, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5, "status": 1.0, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1089}, "4": {"facts_total": 65, "facts_retained": 29, "retention_rate": 0.4461538461538462, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5, "status": 1.0, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1093}}}, "reference_key_id": "keys/synthetic/syn_03000_2000.json", "task_scores": {"correct": 1.0, "abstained": 1.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "eba3052ac902ffbeeb4c56e0006f98338aa07742b3d8c7095b3defabe4f67f23", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "42d02f76fa2a626582c70bca", "attempt_id": "1d29a7f02aed", "run_group_id": "E4_core:A4|closed|qwen3b:h8", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "d83aa60f20d2879d02898cb9d9d8bd1983ed2d8984cae017c9153cafee3a779a", "git_commit": null, "timestamp_utc": "2026-10-30T06:21:28.735800Z", "task_family": "synthetic", "document_id": "syn_03000_2000", "document_hash": "c9cc27c20b05b1f6e592846e84990ee9b69c42f9720a8e338ec2f6fca777fdbc", "question_id": "q00_single_fact", "split": "test", "cluster_id": "syn_03000_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "laptop-m5max-48g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 8, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1904, "query_tokens": 593, "message_tokens": 512, "message_utf8_bytes": 1093, "cache_tensor_bytes": null, "serialized_payload_bytes": 1093, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 177088.284624042, "hops": {"1": 22814.76537499111, "2": 22242.134416941553, "3": 21933.44883294776, "4": 22074.649958056398, "5": 21888.49525002297, "6": 21871.213166974485, "7": 21996.883291052654, "8": 22266.694333055057}, "receiver_prefill_ms": 399.0699580172077, "ttft_ms": 33.116625039838254, "decode_ms": 31.105291098356247}, "total_wall_ms": 177552.15016601142, "ttft_ms": 33.116625039838254, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 29, "retention_rate": 0.4461538461538462, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5, "status": 1.0, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1093}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 29, "retention_rate": 0.4461538461538462, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5, "status": 1.0, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1090}, "2": {"facts_total": 65, "facts_retained": 29, "retention_rate": 0.4461538461538462, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5, "status": 1.0, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1093}, "3": {"facts_total": 65, "facts_retained": 29, "retention_rate": 0.4461538461538462, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5, "status": 1.0, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1089}, "4": {"facts_total": 65, "facts_retained": 29, "retention_rate": 0.4461538461538462, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5, "status": 1.0, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1093}, "5": {"facts_total": 65, "facts_retained": 29, "retention_rate": 0.4461538461538462, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5, "status": 1.0, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1089}, "6": {"facts_total": 65, "facts_retained": 29, "retention_rate": 0.4461538461538462, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5, "status": 1.0, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1093}, "7": {"facts_total": 65, "facts_retained": 29, "retention_rate": 0.4461538461538462, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5, "status": 1.0, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1089}, "8": {"facts_total": 65, "facts_retained": 29, "retention_rate": 0.4461538461538462, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5, "status": 1.0, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1093}}}, "reference_key_id": "keys/synthetic/syn_03000_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "eba3052ac902ffbeeb4c56e0006f98338aa07742b3d8c7095b3defabe4f67f23", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "577638731eeefc72dd063dbb", "attempt_id": "c829bff18082", "run_group_id": "E4_core:A4|closed|qwen3b:h8", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "d83aa60f20d2879d02898cb9d9d8bd1983ed2d8984cae017c9153cafee3a779a", "git_commit": null, "timestamp_utc": "2026-10-30T06:21:29.216193Z", "task_family": "synthetic", "document_id": "syn_03000_2000", "document_hash": "c9cc27c20b05b1f6e592846e84990ee9b69c42f9720a8e338ec2f6fca777fdbc", "question_id": "q01_single_fact", "split": "test", "cluster_id": "syn_03000_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "laptop-m5max-48g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 8, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1904, "query_tokens": 598, "message_tokens": 512, "message_utf8_bytes": 1093, "cache_tensor_bytes": null, "serialized_payload_bytes": 1093, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 177088.284624042, "hops": {"1": 22814.76537499111, "2": 22242.134416941553, "3": 21933.44883294776, "4": 22074.649958056398, "5": 21888.49525002297, "6": 21871.213166974485, "7": 21996.883291052654, "8": 22266.694333055057}, "receiver_prefill_ms": 380.9112919261679, "ttft_ms": 18.17783305887133, "decode_ms": 615.8417500555515}, "total_wall_ms": 178103.8758319337, "ttft_ms": 18.17783305887133, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 9}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "Lot 19-N, Gaspé", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 29, "retention_rate": 0.4461538461538462, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5, "status": 1.0, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1093}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 29, "retention_rate": 0.4461538461538462, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5, "status": 1.0, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1090}, "2": {"facts_total": 65, "facts_retained": 29, "retention_rate": 0.4461538461538462, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5, "status": 1.0, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1093}, "3": {"facts_total": 65, "facts_retained": 29, "retention_rate": 0.4461538461538462, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5, "status": 1.0, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1089}, "4": {"facts_total": 65, "facts_retained": 29, "retention_rate": 0.4461538461538462, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5, "status": 1.0, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1093}, "5": {"facts_total": 65, "facts_retained": 29, "retention_rate": 0.4461538461538462, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5, "status": 1.0, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1089}, "6": {"facts_total": 65, "facts_retained": 29, "retention_rate": 0.4461538461538462, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5, "status": 1.0, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1093}, "7": {"facts_total": 65, "facts_retained": 29, "retention_rate": 0.4461538461538462, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5, "status": 1.0, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1089}, "8": {"facts_total": 65, "facts_retained": 29, "retention_rate": 0.4461538461538462, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5, "status": 1.0, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1093}}}, "reference_key_id": "keys/synthetic/syn_03000_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 0.8571428571428571}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "eba3052ac902ffbeeb4c56e0006f98338aa07742b3d8c7095b3defabe4f67f23", "output_hash": "2e8cf782bd118a1c6d4e23417785d72fb92e930806b5752cd1ec405a0c9b95e5"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "71d01a17ab42e71a0f5961fd", "attempt_id": "c37588159f0d", "run_group_id": "E4_core:A4|closed|qwen3b:h8", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "d83aa60f20d2879d02898cb9d9d8bd1983ed2d8984cae017c9153cafee3a779a", "git_commit": null, "timestamp_utc": "2026-10-30T06:21:30.248179Z", "task_family": "synthetic", "document_id": "syn_03000_2000", "document_hash": "c9cc27c20b05b1f6e592846e84990ee9b69c42f9720a8e338ec2f6fca777fdbc", "question_id": "q02_exact_copy", "split": "test", "cluster_id": "syn_03000_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "laptop-m5max-48g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 8, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1904, "query_tokens": 597, "message_tokens": 512, "message_utf8_bytes": 1093, "cache_tensor_bytes": null, "serialized_payload_bytes": 1093, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 177088.284624042, "hops": {"1": 22814.76537499111, "2": 22242.134416941553, "3": 21933.44883294776, "4": 22074.649958056398, "5": 21888.49525002297, "6": 21871.213166974485, "7": 21996.883291052654, "8": 22266.694333055057}, "receiver_prefill_ms": 412.5935409683734, "ttft_ms": 26.246832916513085, "decode_ms": 37.82904101535678}, "total_wall_ms": 177565.65066508483, "ttft_ms": 26.246832916513085, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 29, "retention_rate": 0.4461538461538462, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5, "status": 1.0, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1093}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 29, "retention_rate": 0.4461538461538462, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5, "status": 1.0, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1090}, "2": {"facts_total": 65, "facts_retained": 29, "retention_rate": 0.4461538461538462, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5, "status": 1.0, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1093}, "3": {"facts_total": 65, "facts_retained": 29, "retention_rate": 0.4461538461538462, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5, "status": 1.0, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1089}, "4": {"facts_total": 65, "facts_retained": 29, "retention_rate": 0.4461538461538462, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5, "status": 1.0, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1093}, "5": {"facts_total": 65, "facts_retained": 29, "retention_rate": 0.4461538461538462, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5, "status": 1.0, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1089}, "6": {"facts_total": 65, "facts_retained": 29, "retention_rate": 0.4461538461538462, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5, "status": 1.0, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1093}, "7": {"facts_total": 65, "facts_retained": 29, "retention_rate": 0.4461538461538462, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5, "status": 1.0, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1089}, "8": {"facts_total": 65, "facts_retained": 29, "retention_rate": 0.4461538461538462, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5, "status": 1.0, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1093}}}, "reference_key_id": "keys/synthetic/syn_03000_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "eba3052ac902ffbeeb4c56e0006f98338aa07742b3d8c7095b3defabe4f67f23", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "2fc5e9f9316477544704e319", "attempt_id": "6bc895eb032c", "run_group_id": "E4_core:A4|closed|qwen3b:h8", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "d83aa60f20d2879d02898cb9d9d8bd1983ed2d8984cae017c9153cafee3a779a", "git_commit": null, "timestamp_utc": "2026-10-30T06:21:30.742250Z", "task_family": "synthetic", "document_id": "syn_03000_2000", "document_hash": "c9cc27c20b05b1f6e592846e84990ee9b69c42f9720a8e338ec2f6fca777fdbc", "question_id": "q03_exact_copy", "split": "test", "cluster_id": "syn_03000_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "laptop-m5max-48g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 8, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1904, "query_tokens": 596, "message_tokens": 512, "message_utf8_bytes": 1093, "cache_tensor_bytes": null, "serialized_payload_bytes": 1093, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 177088.284624042, "hops": {"1": 22814.76537499111, "2": 22242.134416941553, "3": 21933.44883294776, "4": 22074.649958056398, "5": 21888.49525002297, "6": 21871.213166974485, "7": 21996.883291052654, "8": 22266.694333055057}, "receiver_prefill_ms": 399.46324995253235, "ttft_ms": 22.425875067710876, "decode_ms": 330.62341704498976}, "total_wall_ms": 177841.4878330659, "ttft_ms": 22.425875067710876, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 7}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "Kwame Haldorsen", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 29, "retention_rate": 0.4461538461538462, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5, "status": 1.0, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1093}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 29, "retention_rate": 0.4461538461538462, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5, "status": 1.0, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1090}, "2": {"facts_total": 65, "facts_retained": 29, "retention_rate": 0.4461538461538462, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5, "status": 1.0, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1093}, "3": {"facts_total": 65, "facts_retained": 29, "retention_rate": 0.4461538461538462, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5, "status": 1.0, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1089}, "4": {"facts_total": 65, "facts_retained": 29, "retention_rate": 0.4461538461538462, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5, "status": 1.0, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1093}, "5": {"facts_total": 65, "facts_retained": 29, "retention_rate": 0.4461538461538462, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5, "status": 1.0, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1089}, "6": {"facts_total": 65, "facts_retained": 29, "retention_rate": 0.4461538461538462, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5, "status": 1.0, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1093}, "7": {"facts_total": 65, "facts_retained": 29, "retention_rate": 0.4461538461538462, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5, "status": 1.0, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1089}, "8": {"facts_total": 65, "facts_retained": 29, "retention_rate": 0.4461538461538462, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5, "status": 1.0, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1093}}}, "reference_key_id": "keys/synthetic/syn_03000_2000.json", "task_scores": {"correct": 0.0, "abstained": 0.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "eba3052ac902ffbeeb4c56e0006f98338aa07742b3d8c7095b3defabe4f67f23", "output_hash": "175253e7cf463934ec4d54e361a781f323cd87290f01306ed368933bcbd2e25d"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "0694a700cc98a67f51e75793", "attempt_id": "f023874f2b7a", "run_group_id": "E4_core:A4|closed|qwen3b:h8", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "d83aa60f20d2879d02898cb9d9d8bd1983ed2d8984cae017c9153cafee3a779a", "git_commit": null, "timestamp_utc": "2026-10-30T06:21:31.513464Z", "task_family": "synthetic", "document_id": "syn_03000_2000", "document_hash": "c9cc27c20b05b1f6e592846e84990ee9b69c42f9720a8e338ec2f6fca777fdbc", "question_id": "q04_multi_hop", "split": "test", "cluster_id": "syn_03000_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "laptop-m5max-48g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 8, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1904, "query_tokens": 602, "message_tokens": 512, "message_utf8_bytes": 1093, "cache_tensor_bytes": null, "serialized_payload_bytes": 1093, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 177088.284624042, "hops": {"1": 22814.76537499111, "2": 22242.134416941553, "3": 21933.44883294776, "4": 22074.649958056398, "5": 21888.49525002297, "6": 21871.213166974485, "7": 21996.883291052654, "8": 22266.694333055057}, "receiver_prefill_ms": 524.2136250017211, "ttft_ms": 37.737500038929284, "decode_ms": 300.4890000447631}, "total_wall_ms": 177951.68995810673, "ttft_ms": 37.737500038929284, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 9}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "9,220,745", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 29, "retention_rate": 0.4461538461538462, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5, "status": 1.0, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1093}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 29, "retention_rate": 0.4461538461538462, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5, "status": 1.0, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1090}, "2": {"facts_total": 65, "facts_retained": 29, "retention_rate": 0.4461538461538462, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5, "status": 1.0, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1093}, "3": {"facts_total": 65, "facts_retained": 29, "retention_rate": 0.4461538461538462, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5, "status": 1.0, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1089}, "4": {"facts_total": 65, "facts_retained": 29, "retention_rate": 0.4461538461538462, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5, "status": 1.0, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1093}, "5": {"facts_total": 65, "facts_retained": 29, "retention_rate": 0.4461538461538462, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5, "status": 1.0, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1089}, "6": {"facts_total": 65, "facts_retained": 29, "retention_rate": 0.4461538461538462, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5, "status": 1.0, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1093}, "7": {"facts_total": 65, "facts_retained": 29, "retention_rate": 0.4461538461538462, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5, "status": 1.0, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1089}, "8": {"facts_total": 65, "facts_retained": 29, "retention_rate": 0.4461538461538462, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5, "status": 1.0, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1093}}}, "reference_key_id": "keys/synthetic/syn_03000_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "eba3052ac902ffbeeb4c56e0006f98338aa07742b3d8c7095b3defabe4f67f23", "output_hash": "ae55ba88b2cb1b0f1a4e0960ae312e0a1c177c79a6d14b67bff1677cec19b649"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "85ad6d01c1e426dbdc8bd97b", "attempt_id": "82ed51cd7dbb", "run_group_id": "E4_core:A4|closed|qwen3b:h8", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "d83aa60f20d2879d02898cb9d9d8bd1983ed2d8984cae017c9153cafee3a779a", "git_commit": null, "timestamp_utc": "2026-10-30T06:21:32.394054Z", "task_family": "synthetic", "document_id": "syn_03000_2000", "document_hash": "c9cc27c20b05b1f6e592846e84990ee9b69c42f9720a8e338ec2f6fca777fdbc", "question_id": "q05_multi_hop", "split": "test", "cluster_id": "syn_03000_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "laptop-m5max-48g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 8, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1904, "query_tokens": 604, "message_tokens": 512, "message_utf8_bytes": 1093, "cache_tensor_bytes": null, "serialized_payload_bytes": 1093, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 177088.284624042, "hops": {"1": 22814.76537499111, "2": 22242.134416941553, "3": 21933.44883294776, "4": 22074.649958056398, "5": 21888.49525002297, "6": 21871.213166974485, "7": 21996.883291052654, "8": 22266.694333055057}, "receiver_prefill_ms": 390.0492920074612, "ttft_ms": 20.52116603590548, "decode_ms": 667.0180419459939}, "total_wall_ms": 178166.71816608869, "ttft_ms": 20.52116603590548, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 10}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "LY-51E-3199", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 29, "retention_rate": 0.4461538461538462, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5, "status": 1.0, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1093}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 29, "retention_rate": 0.4461538461538462, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5, "status": 1.0, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1090}, "2": {"facts_total": 65, "facts_retained": 29, "retention_rate": 0.4461538461538462, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5, "status": 1.0, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1093}, "3": {"facts_total": 65, "facts_retained": 29, "retention_rate": 0.4461538461538462, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5, "status": 1.0, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1089}, "4": {"facts_total": 65, "facts_retained": 29, "retention_rate": 0.4461538461538462, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5, "status": 1.0, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1093}, "5": {"facts_total": 65, "facts_retained": 29, "retention_rate": 0.4461538461538462, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5, "status": 1.0, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1089}, "6": {"facts_total": 65, "facts_retained": 29, "retention_rate": 0.4461538461538462, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5, "status": 1.0, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1093}, "7": {"facts_total": 65, "facts_retained": 29, "retention_rate": 0.4461538461538462, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5, "status": 1.0, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1089}, "8": {"facts_total": 65, "facts_retained": 29, "retention_rate": 0.4461538461538462, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5, "status": 1.0, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1093}}}, "reference_key_id": "keys/synthetic/syn_03000_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "eba3052ac902ffbeeb4c56e0006f98338aa07742b3d8c7095b3defabe4f67f23", "output_hash": "a1245d1536890a7e4b026366df48441d503f491f782cfd388797b5782443c5c7"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "a551563af449f9efaf33fd96", "attempt_id": "13ac0a8e5ef3", "run_group_id": "E4_core:A4|closed|qwen3b:h8", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "d83aa60f20d2879d02898cb9d9d8bd1983ed2d8984cae017c9153cafee3a779a", "git_commit": null, "timestamp_utc": "2026-10-30T06:21:33.489395Z", "task_family": "synthetic", "document_id": "syn_03000_2000", "document_hash": "c9cc27c20b05b1f6e592846e84990ee9b69c42f9720a8e338ec2f6fca777fdbc", "question_id": "q06_negation", "split": "test", "cluster_id": "syn_03000_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "laptop-m5max-48g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 8, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1904, "query_tokens": 593, "message_tokens": 512, "message_utf8_bytes": 1093, "cache_tensor_bytes": null, "serialized_payload_bytes": 1093, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 177088.284624042, "hops": {"1": 22814.76537499111, "2": 22242.134416941553, "3": 21933.44883294776, "4": 22074.649958056398, "5": 21888.49525002297, "6": 21871.213166974485, "7": 21996.883291052654, "8": 22266.694333055057}, "receiver_prefill_ms": 410.5361660476774, "ttft_ms": 42.67645790241659, "decode_ms": 25.915749953128397}, "total_wall_ms": 177568.1564990664, "ttft_ms": 42.67645790241659, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "no", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 29, "retention_rate": 0.4461538461538462, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5, "status": 1.0, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1093}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 29, "retention_rate": 0.4461538461538462, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5, "status": 1.0, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1090}, "2": {"facts_total": 65, "facts_retained": 29, "retention_rate": 0.4461538461538462, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5, "status": 1.0, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1093}, "3": {"facts_total": 65, "facts_retained": 29, "retention_rate": 0.4461538461538462, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5, "status": 1.0, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1089}, "4": {"facts_total": 65, "facts_retained": 29, "retention_rate": 0.4461538461538462, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5, "status": 1.0, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1093}, "5": {"facts_total": 65, "facts_retained": 29, "retention_rate": 0.4461538461538462, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5, "status": 1.0, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1089}, "6": {"facts_total": 65, "facts_retained": 29, "retention_rate": 0.4461538461538462, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5, "status": 1.0, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1093}, "7": {"facts_total": 65, "facts_retained": 29, "retention_rate": 0.4461538461538462, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5, "status": 1.0, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1089}, "8": {"facts_total": 65, "facts_retained": 29, "retention_rate": 0.4461538461538462, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5, "status": 1.0, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1093}}}, "reference_key_id": "keys/synthetic/syn_03000_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "eba3052ac902ffbeeb4c56e0006f98338aa07742b3d8c7095b3defabe4f67f23", "output_hash": "9390298f3fb0c5b160498935d79cb139aef28e1c47358b4bbba61862b9c26e59"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "deaea8fd7da828d1e3fe6b11", "attempt_id": "528079f50e59", "run_group_id": "E4_core:A4|closed|qwen3b:h8", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "d83aa60f20d2879d02898cb9d9d8bd1983ed2d8984cae017c9153cafee3a779a", "git_commit": null, "timestamp_utc": "2026-10-30T06:21:33.985582Z", "task_family": "synthetic", "document_id": "syn_03000_2000", "document_hash": "c9cc27c20b05b1f6e592846e84990ee9b69c42f9720a8e338ec2f6fca777fdbc", "question_id": "q07_negation", "split": "test", "cluster_id": "syn_03000_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "laptop-m5max-48g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 8, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1904, "query_tokens": 593, "message_tokens": 512, "message_utf8_bytes": 1093, "cache_tensor_bytes": null, "serialized_payload_bytes": 1093, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 177088.284624042, "hops": {"1": 22814.76537499111, "2": 22242.134416941553, "3": 21933.44883294776, "4": 22074.649958056398, "5": 21888.49525002297, "6": 21871.213166974485, "7": 21996.883291052654, "8": 22266.694333055057}, "receiver_prefill_ms": 379.5152499806136, "ttft_ms": 13.012249954044819, "decode_ms": 40.089167072437704}, "total_wall_ms": 177521.79654093925, "ttft_ms": 13.012249954044819, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "no", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 29, "retention_rate": 0.4461538461538462, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5, "status": 1.0, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1093}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 29, "retention_rate": 0.4461538461538462, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5, "status": 1.0, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1090}, "2": {"facts_total": 65, "facts_retained": 29, "retention_rate": 0.4461538461538462, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5, "status": 1.0, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1093}, "3": {"facts_total": 65, "facts_retained": 29, "retention_rate": 0.4461538461538462, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5, "status": 1.0, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1089}, "4": {"facts_total": 65, "facts_retained": 29, "retention_rate": 0.4461538461538462, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5, "status": 1.0, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1093}, "5": {"facts_total": 65, "facts_retained": 29, "retention_rate": 0.4461538461538462, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5, "status": 1.0, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1089}, "6": {"facts_total": 65, "facts_retained": 29, "retention_rate": 0.4461538461538462, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5, "status": 1.0, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1093}, "7": {"facts_total": 65, "facts_retained": 29, "retention_rate": 0.4461538461538462, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5, "status": 1.0, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1089}, "8": {"facts_total": 65, "facts_retained": 29, "retention_rate": 0.4461538461538462, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5, "status": 1.0, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1093}}}, "reference_key_id": "keys/synthetic/syn_03000_2000.json", "task_scores": {"correct": 0.0, "abstained": 0.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "eba3052ac902ffbeeb4c56e0006f98338aa07742b3d8c7095b3defabe4f67f23", "output_hash": "9390298f3fb0c5b160498935d79cb139aef28e1c47358b4bbba61862b9c26e59"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "bc57bb85951801bdd3a160ad", "attempt_id": "25861a17b3a5", "run_group_id": "E4_core:A4|closed|qwen3b:h8", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "d83aa60f20d2879d02898cb9d9d8bd1983ed2d8984cae017c9153cafee3a779a", "git_commit": null, "timestamp_utc": "2026-10-30T06:21:34.436207Z", "task_family": "synthetic", "document_id": "syn_03000_2000", "document_hash": "c9cc27c20b05b1f6e592846e84990ee9b69c42f9720a8e338ec2f6fca777fdbc", "question_id": "q08_revision", "split": "test", "cluster_id": "syn_03000_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "laptop-m5max-48g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 8, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1904, "query_tokens": 608, "message_tokens": 512, "message_utf8_bytes": 1093, "cache_tensor_bytes": null, "serialized_payload_bytes": 1093, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 177088.284624042, "hops": {"1": 22814.76537499111, "2": 22242.134416941553, "3": 21933.44883294776, "4": 22074.649958056398, "5": 21888.49525002297, "6": 21871.213166974485, "7": 21996.883291052654, "8": 22266.694333055057}, "receiver_prefill_ms": 404.36450007837266, "ttft_ms": 53.316625067964196, "decode_ms": 451.90300000831485}, "total_wall_ms": 177998.54133301415, "ttft_ms": 53.316625067964196, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 9}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "2,705,965", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 29, "retention_rate": 0.4461538461538462, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5, "status": 1.0, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1093}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 29, "retention_rate": 0.4461538461538462, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5, "status": 1.0, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1090}, "2": {"facts_total": 65, "facts_retained": 29, "retention_rate": 0.4461538461538462, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5, "status": 1.0, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1093}, "3": {"facts_total": 65, "facts_retained": 29, "retention_rate": 0.4461538461538462, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5, "status": 1.0, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1089}, "4": {"facts_total": 65, "facts_retained": 29, "retention_rate": 0.4461538461538462, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5, "status": 1.0, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1093}, "5": {"facts_total": 65, "facts_retained": 29, "retention_rate": 0.4461538461538462, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5, "status": 1.0, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1089}, "6": {"facts_total": 65, "facts_retained": 29, "retention_rate": 0.4461538461538462, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5, "status": 1.0, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1093}, "7": {"facts_total": 65, "facts_retained": 29, "retention_rate": 0.4461538461538462, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5, "status": 1.0, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1089}, "8": {"facts_total": 65, "facts_retained": 29, "retention_rate": 0.4461538461538462, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5, "status": 1.0, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1093}}}, "reference_key_id": "keys/synthetic/syn_03000_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "eba3052ac902ffbeeb4c56e0006f98338aa07742b3d8c7095b3defabe4f67f23", "output_hash": "dbedb6f5d00e413d915513b85d8ed195b18b9acf392ece342a85e15e3752bcc8"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "3db88bfda2976491d3a914c8", "attempt_id": "9cd92f5992f8", "run_group_id": "E4_core:A4|closed|qwen3b:h8", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "d83aa60f20d2879d02898cb9d9d8bd1983ed2d8984cae017c9153cafee3a779a", "git_commit": null, "timestamp_utc": "2026-10-30T06:21:35.364151Z", "task_family": "synthetic", "document_id": "syn_03000_2000", "document_hash": "c9cc27c20b05b1f6e592846e84990ee9b69c42f9720a8e338ec2f6fca777fdbc", "question_id": "q09_revision", "split": "test", "cluster_id": "syn_03000_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "laptop-m5max-48g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 8, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1904, "query_tokens": 604, "message_tokens": 512, "message_utf8_bytes": 1093, "cache_tensor_bytes": null, "serialized_payload_bytes": 1093, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 177088.284624042, "hops": {"1": 22814.76537499111, "2": 22242.134416941553, "3": 21933.44883294776, "4": 22074.649958056398, "5": 21888.49525002297, "6": 21871.213166974485, "7": 21996.883291052654, "8": 22266.694333055057}, "receiver_prefill_ms": 401.1902919737622, "ttft_ms": 38.42691704630852, "decode_ms": 205.4971670731902}, "total_wall_ms": 177734.09387399442, "ttft_ms": 38.42691704630852, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 7}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "9127965", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 29, "retention_rate": 0.4461538461538462, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5, "status": 1.0, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1093}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 29, "retention_rate": 0.4461538461538462, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5, "status": 1.0, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1090}, "2": {"facts_total": 65, "facts_retained": 29, "retention_rate": 0.4461538461538462, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5, "status": 1.0, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1093}, "3": {"facts_total": 65, "facts_retained": 29, "retention_rate": 0.4461538461538462, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5, "status": 1.0, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1089}, "4": {"facts_total": 65, "facts_retained": 29, "retention_rate": 0.4461538461538462, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5, "status": 1.0, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1093}, "5": {"facts_total": 65, "facts_retained": 29, "retention_rate": 0.4461538461538462, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5, "status": 1.0, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1089}, "6": {"facts_total": 65, "facts_retained": 29, "retention_rate": 0.4461538461538462, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5, "status": 1.0, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1093}, "7": {"facts_total": 65, "facts_retained": 29, "retention_rate": 0.4461538461538462, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5, "status": 1.0, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1089}, "8": {"facts_total": 65, "facts_retained": 29, "retention_rate": 0.4461538461538462, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5, "status": 1.0, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1093}}}, "reference_key_id": "keys/synthetic/syn_03000_2000.json", "task_scores": {"correct": 0.0, "abstained": 0.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "eba3052ac902ffbeeb4c56e0006f98338aa07742b3d8c7095b3defabe4f67f23", "output_hash": "a0c871941224304cec47bb1b3d64b8706843b299f14ab94899d97df14f433da7"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "c6424181723c4ff56e152e6b", "attempt_id": "fc4ede698928", "run_group_id": "E4_core:A4|closed|qwen3b:h8", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "d83aa60f20d2879d02898cb9d9d8bd1983ed2d8984cae017c9153cafee3a779a", "git_commit": null, "timestamp_utc": "2026-10-30T06:21:36.026306Z", "task_family": "synthetic", "document_id": "syn_03000_2000", "document_hash": "c9cc27c20b05b1f6e592846e84990ee9b69c42f9720a8e338ec2f6fca777fdbc", "question_id": "q10_temporal", "split": "test", "cluster_id": "syn_03000_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "laptop-m5max-48g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 8, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1904, "query_tokens": 610, "message_tokens": 512, "message_utf8_bytes": 1093, "cache_tensor_bytes": null, "serialized_payload_bytes": 1093, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 177088.284624042, "hops": {"1": 22814.76537499111, "2": 22242.134416941553, "3": 21933.44883294776, "4": 22074.649958056398, "5": 21888.49525002297, "6": 21871.213166974485, "7": 21996.883291052654, "8": 22266.694333055057}, "receiver_prefill_ms": 387.2169590322301, "ttft_ms": 49.37275010161102, "decode_ms": 357.2142079938203}, "total_wall_ms": 177882.75699911173, "ttft_ms": 49.37275010161102, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 5}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "Felaldnak-7", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 29, "retention_rate": 0.4461538461538462, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5, "status": 1.0, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1093}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 29, "retention_rate": 0.4461538461538462, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5, "status": 1.0, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1090}, "2": {"facts_total": 65, "facts_retained": 29, "retention_rate": 0.4461538461538462, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5, "status": 1.0, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1093}, "3": {"facts_total": 65, "facts_retained": 29, "retention_rate": 0.4461538461538462, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5, "status": 1.0, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1089}, "4": {"facts_total": 65, "facts_retained": 29, "retention_rate": 0.4461538461538462, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5, "status": 1.0, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1093}, "5": {"facts_total": 65, "facts_retained": 29, "retention_rate": 0.4461538461538462, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5, "status": 1.0, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1089}, "6": {"facts_total": 65, "facts_retained": 29, "retention_rate": 0.4461538461538462, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5, "status": 1.0, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1093}, "7": {"facts_total": 65, "facts_retained": 29, "retention_rate": 0.4461538461538462, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5, "status": 1.0, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1089}, "8": {"facts_total": 65, "facts_retained": 29, "retention_rate": 0.4461538461538462, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5, "status": 1.0, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1093}}}, "reference_key_id": "keys/synthetic/syn_03000_2000.json", "task_scores": {"correct": 0.0, "abstained": 0.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "eba3052ac902ffbeeb4c56e0006f98338aa07742b3d8c7095b3defabe4f67f23", "output_hash": "2b9a4cf7221f1b6f31508a6b3078ec521f211026a932da5020cddd512ecc5fa2"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "fc12cfaefb943b2f4054374c", "attempt_id": "9309dcb8449f", "run_group_id": "E4_core:A4|closed|qwen3b:h8", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "d83aa60f20d2879d02898cb9d9d8bd1983ed2d8984cae017c9153cafee3a779a", "git_commit": null, "timestamp_utc": "2026-10-30T06:21:36.837290Z", "task_family": "synthetic", "document_id": "syn_03000_2000", "document_hash": "c9cc27c20b05b1f6e592846e84990ee9b69c42f9720a8e338ec2f6fca777fdbc", "question_id": "q11_aggregation", "split": "test", "cluster_id": "syn_03000_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "laptop-m5max-48g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 8, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1904, "query_tokens": 605, "message_tokens": 512, "message_utf8_bytes": 1093, "cache_tensor_bytes": null, "serialized_payload_bytes": 1093, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 177088.284624042, "hops": {"1": 22814.76537499111, "2": 22242.134416941553, "3": 21933.44883294776, "4": 22074.649958056398, "5": 21888.49525002297, "6": 21871.213166974485, "7": 21996.883291052654, "8": 22266.694333055057}, "receiver_prefill_ms": 409.1772499959916, "ttft_ms": 27.88349997717887, "decode_ms": 228.86704199481755}, "total_wall_ms": 177754.8338320339, "ttft_ms": 27.88349997717887, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 7}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "9220745", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 29, "retention_rate": 0.4461538461538462, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5, "status": 1.0, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1093}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 29, "retention_rate": 0.4461538461538462, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5, "status": 1.0, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1090}, "2": {"facts_total": 65, "facts_retained": 29, "retention_rate": 0.4461538461538462, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5, "status": 1.0, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1093}, "3": {"facts_total": 65, "facts_retained": 29, "retention_rate": 0.4461538461538462, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5, "status": 1.0, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1089}, "4": {"facts_total": 65, "facts_retained": 29, "retention_rate": 0.4461538461538462, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5, "status": 1.0, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1093}, "5": {"facts_total": 65, "facts_retained": 29, "retention_rate": 0.4461538461538462, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5, "status": 1.0, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1089}, "6": {"facts_total": 65, "facts_retained": 29, "retention_rate": 0.4461538461538462, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5, "status": 1.0, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1093}, "7": {"facts_total": 65, "facts_retained": 29, "retention_rate": 0.4461538461538462, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5, "status": 1.0, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1089}, "8": {"facts_total": 65, "facts_retained": 29, "retention_rate": 0.4461538461538462, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5, "status": 1.0, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1093}}}, "reference_key_id": "keys/synthetic/syn_03000_2000.json", "task_scores": {"correct": 0.0, "abstained": 0.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "eba3052ac902ffbeeb4c56e0006f98338aa07742b3d8c7095b3defabe4f67f23", "output_hash": "f5d77aa763f6365bc80d01b9e6e23bc43fb570ac45771d7bd628c75cde727b35"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "e544459f217d49101763b90f", "attempt_id": "b7b51447738e", "run_group_id": "E4_core:A4|closed|qwen3b:h8", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "d83aa60f20d2879d02898cb9d9d8bd1983ed2d8984cae017c9153cafee3a779a", "git_commit": null, "timestamp_utc": "2026-10-30T06:21:37.520544Z", "task_family": "synthetic", "document_id": "syn_03000_2000", "document_hash": "c9cc27c20b05b1f6e592846e84990ee9b69c42f9720a8e338ec2f6fca777fdbc", "question_id": "q12_unit_conversion", "split": "test", "cluster_id": "syn_03000_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "laptop-m5max-48g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 8, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1904, "query_tokens": 601, "message_tokens": 512, "message_utf8_bytes": 1093, "cache_tensor_bytes": null, "serialized_payload_bytes": 1093, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 177088.284624042, "hops": {"1": 22814.76537499111, "2": 22242.134416941553, "3": 21933.44883294776, "4": 22074.649958056398, "5": 21888.49525002297, "6": 21871.213166974485, "7": 21996.883291052654, "8": 22266.694333055057}, "receiver_prefill_ms": 346.79679200053215, "ttft_ms": 39.74845807533711, "decode_ms": 38.50504197180271}, "total_wall_ms": 177514.02683206834, "ttft_ms": 39.74845807533711, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 29, "retention_rate": 0.4461538461538462, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5, "status": 1.0, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1093}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 29, "retention_rate": 0.4461538461538462, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5, "status": 1.0, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1090}, "2": {"facts_total": 65, "facts_retained": 29, "retention_rate": 0.4461538461538462, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5, "status": 1.0, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1093}, "3": {"facts_total": 65, "facts_retained": 29, "retention_rate": 0.4461538461538462, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5, "status": 1.0, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1089}, "4": {"facts_total": 65, "facts_retained": 29, "retention_rate": 0.4461538461538462, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5, "status": 1.0, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1093}, "5": {"facts_total": 65, "facts_retained": 29, "retention_rate": 0.4461538461538462, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5, "status": 1.0, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1089}, "6": {"facts_total": 65, "facts_retained": 29, "retention_rate": 0.4461538461538462, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5, "status": 1.0, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1093}, "7": {"facts_total": 65, "facts_retained": 29, "retention_rate": 0.4461538461538462, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5, "status": 1.0, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1089}, "8": {"facts_total": 65, "facts_retained": 29, "retention_rate": 0.4461538461538462, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5, "status": 1.0, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1093}}}, "reference_key_id": "keys/synthetic/syn_03000_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "eba3052ac902ffbeeb4c56e0006f98338aa07742b3d8c7095b3defabe4f67f23", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "ae3401f64c17ec318de002f8", "attempt_id": "e58890dc1bde", "run_group_id": "E4_core:A4|closed|qwen3b:h8", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "d83aa60f20d2879d02898cb9d9d8bd1983ed2d8984cae017c9153cafee3a779a", "git_commit": null, "timestamp_utc": "2026-10-30T06:21:37.962262Z", "task_family": "synthetic", "document_id": "syn_03000_2000", "document_hash": "c9cc27c20b05b1f6e592846e84990ee9b69c42f9720a8e338ec2f6fca777fdbc", "question_id": "q13_unanswerable", "split": "test", "cluster_id": "syn_03000_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "laptop-m5max-48g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 8, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1904, "query_tokens": 605, "message_tokens": 512, "message_utf8_bytes": 1093, "cache_tensor_bytes": null, "serialized_payload_bytes": 1093, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 177088.284624042, "hops": {"1": 22814.76537499111, "2": 22242.134416941553, "3": 21933.44883294776, "4": 22074.649958056398, "5": 21888.49525002297, "6": 21871.213166974485, "7": 21996.883291052654, "8": 22266.694333055057}, "receiver_prefill_ms": 566.7922500288114, "ttft_ms": 29.47508404031396, "decode_ms": 41.78304108791053}, "total_wall_ms": 177727.02654101886, "ttft_ms": 29.47508404031396, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 29, "retention_rate": 0.4461538461538462, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5, "status": 1.0, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1093}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 29, "retention_rate": 0.4461538461538462, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5, "status": 1.0, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1090}, "2": {"facts_total": 65, "facts_retained": 29, "retention_rate": 0.4461538461538462, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5, "status": 1.0, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1093}, "3": {"facts_total": 65, "facts_retained": 29, "retention_rate": 0.4461538461538462, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5, "status": 1.0, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1089}, "4": {"facts_total": 65, "facts_retained": 29, "retention_rate": 0.4461538461538462, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5, "status": 1.0, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1093}, "5": {"facts_total": 65, "facts_retained": 29, "retention_rate": 0.4461538461538462, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5, "status": 1.0, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1089}, "6": {"facts_total": 65, "facts_retained": 29, "retention_rate": 0.4461538461538462, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5, "status": 1.0, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1093}, "7": {"facts_total": 65, "facts_retained": 29, "retention_rate": 0.4461538461538462, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5, "status": 1.0, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1089}, "8": {"facts_total": 65, "facts_retained": 29, "retention_rate": 0.4461538461538462, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5, "status": 1.0, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1093}}}, "reference_key_id": "keys/synthetic/syn_03000_2000.json", "task_scores": {"correct": 1.0, "abstained": 1.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "eba3052ac902ffbeeb4c56e0006f98338aa07742b3d8c7095b3defabe4f67f23", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "cd09ca989da2c4fe2c66f655", "attempt_id": "554b462bdd97", "run_group_id": "E4_core:A4|closed|qwen3b:h8", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "d83aa60f20d2879d02898cb9d9d8bd1983ed2d8984cae017c9153cafee3a779a", "git_commit": null, "timestamp_utc": "2026-10-30T06:21:38.618206Z", "task_family": "synthetic", "document_id": "syn_03000_2000", "document_hash": "c9cc27c20b05b1f6e592846e84990ee9b69c42f9720a8e338ec2f6fca777fdbc", "question_id": "q14_unanswerable", "split": "test", "cluster_id": "syn_03000_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "laptop-m5max-48g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 8, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1904, "query_tokens": 607, "message_tokens": 512, "message_utf8_bytes": 1093, "cache_tensor_bytes": null, "serialized_payload_bytes": 1093, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 177088.284624042, "hops": {"1": 22814.76537499111, "2": 22242.134416941553, "3": 21933.44883294776, "4": 22074.649958056398, "5": 21888.49525002297, "6": 21871.213166974485, "7": 21996.883291052654, "8": 22266.694333055057}, "receiver_prefill_ms": 406.8365420680493, "ttft_ms": 31.674250029027462, "decode_ms": 44.37975003384054}, "total_wall_ms": 177571.75858307164, "ttft_ms": 31.674250029027462, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 29, "retention_rate": 0.4461538461538462, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5, "status": 1.0, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1093}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 29, "retention_rate": 0.4461538461538462, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5, "status": 1.0, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1090}, "2": {"facts_total": 65, "facts_retained": 29, "retention_rate": 0.4461538461538462, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5, "status": 1.0, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1093}, "3": {"facts_total": 65, "facts_retained": 29, "retention_rate": 0.4461538461538462, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5, "status": 1.0, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1089}, "4": {"facts_total": 65, "facts_retained": 29, "retention_rate": 0.4461538461538462, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5, "status": 1.0, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1093}, "5": {"facts_total": 65, "facts_retained": 29, "retention_rate": 0.4461538461538462, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5, "status": 1.0, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1089}, "6": {"facts_total": 65, "facts_retained": 29, "retention_rate": 0.4461538461538462, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5, "status": 1.0, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1093}, "7": {"facts_total": 65, "facts_retained": 29, "retention_rate": 0.4461538461538462, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5, "status": 1.0, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1089}, "8": {"facts_total": 65, "facts_retained": 29, "retention_rate": 0.4461538461538462, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5, "status": 1.0, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1093}}}, "reference_key_id": "keys/synthetic/syn_03000_2000.json", "task_scores": {"correct": 1.0, "abstained": 1.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "eba3052ac902ffbeeb4c56e0006f98338aa07742b3d8c7095b3defabe4f67f23", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "2c2ec357486881beb319dceb", "attempt_id": "2954bb4cf237", "run_group_id": "E4_core:A5|closed|qwen3b:h1", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "d83aa60f20d2879d02898cb9d9d8bd1983ed2d8984cae017c9153cafee3a779a", "git_commit": null, "timestamp_utc": "2026-10-30T06:22:02.150227Z", "task_family": "synthetic", "document_id": "syn_03000_2000", "document_hash": "c9cc27c20b05b1f6e592846e84990ee9b69c42f9720a8e338ec2f6fca777fdbc", "question_id": "q00_single_fact", "split": "test", "cluster_id": "syn_03000_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "laptop-m5max-48g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 1, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1904, "query_tokens": 464, "message_tokens": 383, "message_utf8_bytes": 911, "cache_tensor_bytes": null, "serialized_payload_bytes": 911, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 23029.02170899324, "hops": {"1": 23029.02170899324}, "receiver_prefill_ms": 327.61612499598414, "ttft_ms": 36.93858301267028, "decode_ms": 28.12916599214077}, "total_wall_ms": 23422.30841703713, "ttft_ms": 36.93858301267028, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 15, "retention_rate": 0.23076923076923078, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.09090909090909091, "amount": 0.16666666666666666, "status": 0.5, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 855}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 15, "retention_rate": 0.23076923076923078, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.09090909090909091, "amount": 0.16666666666666666, "status": 0.5, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 855}}}, "reference_key_id": "keys/synthetic/syn_03000_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "a409b88ff601f1f9bf4d221a0c042117fadd2be19621e0b0dd2800495cfbe0cc", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "a5bd9e5f086519983035b262", "attempt_id": "a595f22fffc0", "run_group_id": "E4_core:A5|closed|qwen3b:h1", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "d83aa60f20d2879d02898cb9d9d8bd1983ed2d8984cae017c9153cafee3a779a", "git_commit": null, "timestamp_utc": "2026-10-30T06:22:02.560599Z", "task_family": "synthetic", "document_id": "syn_03000_2000", "document_hash": "c9cc27c20b05b1f6e592846e84990ee9b69c42f9720a8e338ec2f6fca777fdbc", "question_id": "q01_single_fact", "split": "test", "cluster_id": "syn_03000_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "laptop-m5max-48g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 1, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1904, "query_tokens": 469, "message_tokens": 383, "message_utf8_bytes": 911, "cache_tensor_bytes": null, "serialized_payload_bytes": 911, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 23029.02170899324, "hops": {"1": 23029.02170899324}, "receiver_prefill_ms": 320.53245801944286, "ttft_ms": 37.108666030690074, "decode_ms": 461.53220802079886}, "total_wall_ms": 23848.84124994278, "ttft_ms": 37.108666030690074, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 11}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "Lot 409-H in Senneterre", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 15, "retention_rate": 0.23076923076923078, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.09090909090909091, "amount": 0.16666666666666666, "status": 0.5, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 855}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 15, "retention_rate": 0.23076923076923078, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.09090909090909091, "amount": 0.16666666666666666, "status": 0.5, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 855}}}, "reference_key_id": "keys/synthetic/syn_03000_2000.json", "task_scores": {"correct": 0.0, "abstained": 0.0, "f1": 0.5}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "a409b88ff601f1f9bf4d221a0c042117fadd2be19621e0b0dd2800495cfbe0cc", "output_hash": "928c825f0fee3b69cf0901e3721b2d6c4467e52d2c6707afe17dd750a93d5e31"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "5e511e01dd7d363d059a875e", "attempt_id": "6e4101896f02", "run_group_id": "E4_core:A5|closed|qwen3b:h1", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "d83aa60f20d2879d02898cb9d9d8bd1983ed2d8984cae017c9153cafee3a779a", "git_commit": null, "timestamp_utc": "2026-10-30T06:22:03.396579Z", "task_family": "synthetic", "document_id": "syn_03000_2000", "document_hash": "c9cc27c20b05b1f6e592846e84990ee9b69c42f9720a8e338ec2f6fca777fdbc", "question_id": "q02_exact_copy", "split": "test", "cluster_id": "syn_03000_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "laptop-m5max-48g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 1, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1904, "query_tokens": 468, "message_tokens": 383, "message_utf8_bytes": 911, "cache_tensor_bytes": null, "serialized_payload_bytes": 911, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 23029.02170899324, "hops": {"1": 23029.02170899324}, "receiver_prefill_ms": 446.661917027086, "ttft_ms": 39.75174995139241, "decode_ms": 23.460750002413988}, "total_wall_ms": 23539.48529297486, "ttft_ms": 39.75174995139241, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 15, "retention_rate": 0.23076923076923078, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.09090909090909091, "amount": 0.16666666666666666, "status": 0.5, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 855}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 15, "retention_rate": 0.23076923076923078, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.09090909090909091, "amount": 0.16666666666666666, "status": 0.5, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 855}}}, "reference_key_id": "keys/synthetic/syn_03000_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "a409b88ff601f1f9bf4d221a0c042117fadd2be19621e0b0dd2800495cfbe0cc", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "035b7644d508fa6b9cc18aca", "attempt_id": "45953aff471c", "run_group_id": "E4_core:A5|closed|qwen3b:h1", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "d83aa60f20d2879d02898cb9d9d8bd1983ed2d8984cae017c9153cafee3a779a", "git_commit": null, "timestamp_utc": "2026-10-30T06:22:03.924539Z", "task_family": "synthetic", "document_id": "syn_03000_2000", "document_hash": "c9cc27c20b05b1f6e592846e84990ee9b69c42f9720a8e338ec2f6fca777fdbc", "question_id": "q03_exact_copy", "split": "test", "cluster_id": "syn_03000_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "laptop-m5max-48g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 1, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1904, "query_tokens": 467, "message_tokens": 383, "message_utf8_bytes": 911, "cache_tensor_bytes": null, "serialized_payload_bytes": 911, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 23029.02170899324, "hops": {"1": 23029.02170899324}, "receiver_prefill_ms": 314.80954203289, "ttft_ms": 38.433333043940365, "decode_ms": 26.600958895869553}, "total_wall_ms": 23409.379917080514, "ttft_ms": 38.433333043940365, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 15, "retention_rate": 0.23076923076923078, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.09090909090909091, "amount": 0.16666666666666666, "status": 0.5, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 855}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 15, "retention_rate": 0.23076923076923078, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.09090909090909091, "amount": 0.16666666666666666, "status": 0.5, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 855}}}, "reference_key_id": "keys/synthetic/syn_03000_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "a409b88ff601f1f9bf4d221a0c042117fadd2be19621e0b0dd2800495cfbe0cc", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "3f790b66d16e9fe6d2c424b7", "attempt_id": "d7a004352006", "run_group_id": "E4_core:A5|closed|qwen3b:h1", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "d83aa60f20d2879d02898cb9d9d8bd1983ed2d8984cae017c9153cafee3a779a", "git_commit": null, "timestamp_utc": "2026-10-30T06:22:04.321107Z", "task_family": "synthetic", "document_id": "syn_03000_2000", "document_hash": "c9cc27c20b05b1f6e592846e84990ee9b69c42f9720a8e338ec2f6fca777fdbc", "question_id": "q04_multi_hop", "split": "test", "cluster_id": "syn_03000_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "laptop-m5max-48g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 1, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1904, "query_tokens": 473, "message_tokens": 383, "message_utf8_bytes": 911, "cache_tensor_bytes": null, "serialized_payload_bytes": 911, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 23029.02170899324, "hops": {"1": 23029.02170899324}, "receiver_prefill_ms": 326.6994580626488, "ttft_ms": 19.81654204428196, "decode_ms": 38.43312500976026}, "total_wall_ms": 23414.64841691777, "ttft_ms": 19.81654204428196, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 15, "retention_rate": 0.23076923076923078, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.09090909090909091, "amount": 0.16666666666666666, "status": 0.5, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 855}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 15, "retention_rate": 0.23076923076923078, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.09090909090909091, "amount": 0.16666666666666666, "status": 0.5, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 855}}}, "reference_key_id": "keys/synthetic/syn_03000_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "a409b88ff601f1f9bf4d221a0c042117fadd2be19621e0b0dd2800495cfbe0cc", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "2c76721ec7a255f2fe015a9f", "attempt_id": "45ef16e30902", "run_group_id": "E4_core:A5|closed|qwen3b:h1", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "d83aa60f20d2879d02898cb9d9d8bd1983ed2d8984cae017c9153cafee3a779a", "git_commit": null, "timestamp_utc": "2026-10-30T06:22:04.723294Z", "task_family": "synthetic", "document_id": "syn_03000_2000", "document_hash": "c9cc27c20b05b1f6e592846e84990ee9b69c42f9720a8e338ec2f6fca777fdbc", "question_id": "q05_multi_hop", "split": "test", "cluster_id": "syn_03000_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "laptop-m5max-48g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 1, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1904, "query_tokens": 475, "message_tokens": 383, "message_utf8_bytes": 911, "cache_tensor_bytes": null, "serialized_payload_bytes": 911, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 23029.02170899324, "hops": {"1": 23029.02170899324}, "receiver_prefill_ms": 359.3574579572305, "ttft_ms": 48.78258309327066, "decode_ms": 498.6306669889018}, "total_wall_ms": 23936.4372510463, "ttft_ms": 48.78258309327066, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 10}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "LY-51E-3199", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 15, "retention_rate": 0.23076923076923078, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.09090909090909091, "amount": 0.16666666666666666, "status": 0.5, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 855}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 15, "retention_rate": 0.23076923076923078, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.09090909090909091, "amount": 0.16666666666666666, "status": 0.5, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 855}}}, "reference_key_id": "keys/synthetic/syn_03000_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "a409b88ff601f1f9bf4d221a0c042117fadd2be19621e0b0dd2800495cfbe0cc", "output_hash": "a1245d1536890a7e4b026366df48441d503f491f782cfd388797b5782443c5c7"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "ae6f52eee76d30db32a293f4", "attempt_id": "96d720115cdc", "run_group_id": "E4_core:A5|closed|qwen3b:h1", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "d83aa60f20d2879d02898cb9d9d8bd1983ed2d8984cae017c9153cafee3a779a", "git_commit": null, "timestamp_utc": "2026-10-30T06:22:05.648862Z", "task_family": "synthetic", "document_id": "syn_03000_2000", "document_hash": "c9cc27c20b05b1f6e592846e84990ee9b69c42f9720a8e338ec2f6fca777fdbc", "question_id": "q06_negation", "split": "test", "cluster_id": "syn_03000_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "laptop-m5max-48g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 1, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1904, "query_tokens": 464, "message_tokens": 383, "message_utf8_bytes": 911, "cache_tensor_bytes": null, "serialized_payload_bytes": 911, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 23029.02170899324, "hops": {"1": 23029.02170899324}, "receiver_prefill_ms": 336.35470794979483, "ttft_ms": 38.66758302319795, "decode_ms": 24.32054199744016}, "total_wall_ms": 23428.861041902564, "ttft_ms": 38.66758302319795, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "no", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 15, "retention_rate": 0.23076923076923078, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.09090909090909091, "amount": 0.16666666666666666, "status": 0.5, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 855}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 15, "retention_rate": 0.23076923076923078, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.09090909090909091, "amount": 0.16666666666666666, "status": 0.5, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 855}}}, "reference_key_id": "keys/synthetic/syn_03000_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "a409b88ff601f1f9bf4d221a0c042117fadd2be19621e0b0dd2800495cfbe0cc", "output_hash": "9390298f3fb0c5b160498935d79cb139aef28e1c47358b4bbba61862b9c26e59"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "6a4a469d1fa30b89259fde4e", "attempt_id": "6be303c2bbd2", "run_group_id": "E4_core:A5|closed|qwen3b:h1", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "d83aa60f20d2879d02898cb9d9d8bd1983ed2d8984cae017c9153cafee3a779a", "git_commit": null, "timestamp_utc": "2026-10-30T06:22:06.065781Z", "task_family": "synthetic", "document_id": "syn_03000_2000", "document_hash": "c9cc27c20b05b1f6e592846e84990ee9b69c42f9720a8e338ec2f6fca777fdbc", "question_id": "q07_negation", "split": "test", "cluster_id": "syn_03000_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "laptop-m5max-48g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 1, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1904, "query_tokens": 464, "message_tokens": 383, "message_utf8_bytes": 911, "cache_tensor_bytes": null, "serialized_payload_bytes": 911, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 23029.02170899324, "hops": {"1": 23029.02170899324}, "receiver_prefill_ms": 288.88666594866663, "ttft_ms": 57.73391597904265, "decode_ms": 77.0186249865219}, "total_wall_ms": 23453.23629199993, "ttft_ms": 57.73391597904265, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 15, "retention_rate": 0.23076923076923078, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.09090909090909091, "amount": 0.16666666666666666, "status": 0.5, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 855}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 15, "retention_rate": 0.23076923076923078, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.09090909090909091, "amount": 0.16666666666666666, "status": 0.5, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 855}}}, "reference_key_id": "keys/synthetic/syn_03000_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "a409b88ff601f1f9bf4d221a0c042117fadd2be19621e0b0dd2800495cfbe0cc", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "06ace68638941a5aa717dc99", "attempt_id": "1b6b9e228d76", "run_group_id": "E4_core:A5|closed|qwen3b:h1", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "d83aa60f20d2879d02898cb9d9d8bd1983ed2d8984cae017c9153cafee3a779a", "git_commit": null, "timestamp_utc": "2026-10-30T06:22:06.506735Z", "task_family": "synthetic", "document_id": "syn_03000_2000", "document_hash": "c9cc27c20b05b1f6e592846e84990ee9b69c42f9720a8e338ec2f6fca777fdbc", "question_id": "q08_revision", "split": "test", "cluster_id": "syn_03000_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "laptop-m5max-48g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 1, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1904, "query_tokens": 479, "message_tokens": 383, "message_utf8_bytes": 911, "cache_tensor_bytes": null, "serialized_payload_bytes": 911, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 23029.02170899324, "hops": {"1": 23029.02170899324}, "receiver_prefill_ms": 445.6051660235971, "ttft_ms": 49.39237504731864, "decode_ms": 287.52345801331103}, "total_wall_ms": 23812.0936260093, "ttft_ms": 49.39237504731864, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 9}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "2,705,965", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 15, "retention_rate": 0.23076923076923078, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.09090909090909091, "amount": 0.16666666666666666, "status": 0.5, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 855}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 15, "retention_rate": 0.23076923076923078, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.09090909090909091, "amount": 0.16666666666666666, "status": 0.5, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 855}}}, "reference_key_id": "keys/synthetic/syn_03000_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "a409b88ff601f1f9bf4d221a0c042117fadd2be19621e0b0dd2800495cfbe0cc", "output_hash": "dbedb6f5d00e413d915513b85d8ed195b18b9acf392ece342a85e15e3752bcc8"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "149abc726dfbd11bec4d7533", "attempt_id": "c5e871457c23", "run_group_id": "E4_core:A5|closed|qwen3b:h1", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "d83aa60f20d2879d02898cb9d9d8bd1983ed2d8984cae017c9153cafee3a779a", "git_commit": null, "timestamp_utc": "2026-10-30T06:22:07.307565Z", "task_family": "synthetic", "document_id": "syn_03000_2000", "document_hash": "c9cc27c20b05b1f6e592846e84990ee9b69c42f9720a8e338ec2f6fca777fdbc", "question_id": "q09_revision", "split": "test", "cluster_id": "syn_03000_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "laptop-m5max-48g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 1, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1904, "query_tokens": 475, "message_tokens": 383, "message_utf8_bytes": 911, "cache_tensor_bytes": null, "serialized_payload_bytes": 911, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 23029.02170899324, "hops": {"1": 23029.02170899324}, "receiver_prefill_ms": 341.18620795197785, "ttft_ms": 30.319540994241834, "decode_ms": 32.64083410613239}, "total_wall_ms": 23433.67204198148, "ttft_ms": 30.319540994241834, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 15, "retention_rate": 0.23076923076923078, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.09090909090909091, "amount": 0.16666666666666666, "status": 0.5, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 855}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 15, "retention_rate": 0.23076923076923078, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.09090909090909091, "amount": 0.16666666666666666, "status": 0.5, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 855}}}, "reference_key_id": "keys/synthetic/syn_03000_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "a409b88ff601f1f9bf4d221a0c042117fadd2be19621e0b0dd2800495cfbe0cc", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "e5a64051f0cc31cdb4a83736", "attempt_id": "6acc70a8626d", "run_group_id": "E4_core:A5|closed|qwen3b:h1", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "d83aa60f20d2879d02898cb9d9d8bd1983ed2d8984cae017c9153cafee3a779a", "git_commit": null, "timestamp_utc": "2026-10-30T06:22:07.728630Z", "task_family": "synthetic", "document_id": "syn_03000_2000", "document_hash": "c9cc27c20b05b1f6e592846e84990ee9b69c42f9720a8e338ec2f6fca777fdbc", "question_id": "q10_temporal", "split": "test", "cluster_id": "syn_03000_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "laptop-m5max-48g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 1, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1904, "query_tokens": 481, "message_tokens": 383, "message_utf8_bytes": 911, "cache_tensor_bytes": null, "serialized_payload_bytes": 911, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 23029.02170899324, "hops": {"1": 23029.02170899324}, "receiver_prefill_ms": 309.67791692819446, "ttft_ms": 36.21950000524521, "decode_ms": 52.28616704698652}, "total_wall_ms": 23427.857624948956, "ttft_ms": 36.21950000524521, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 15, "retention_rate": 0.23076923076923078, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.09090909090909091, "amount": 0.16666666666666666, "status": 0.5, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 855}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 15, "retention_rate": 0.23076923076923078, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.09090909090909091, "amount": 0.16666666666666666, "status": 0.5, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 855}}}, "reference_key_id": "keys/synthetic/syn_03000_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "a409b88ff601f1f9bf4d221a0c042117fadd2be19621e0b0dd2800495cfbe0cc", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "e4f6c66b2e92ec1a4504faa7", "attempt_id": "c09d4f268fea", "run_group_id": "E4_core:A5|closed|qwen3b:h1", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "d83aa60f20d2879d02898cb9d9d8bd1983ed2d8984cae017c9153cafee3a779a", "git_commit": null, "timestamp_utc": "2026-10-30T06:22:08.144454Z", "task_family": "synthetic", "document_id": "syn_03000_2000", "document_hash": "c9cc27c20b05b1f6e592846e84990ee9b69c42f9720a8e338ec2f6fca777fdbc", "question_id": "q11_aggregation", "split": "test", "cluster_id": "syn_03000_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "laptop-m5max-48g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 1, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1904, "query_tokens": 476, "message_tokens": 383, "message_utf8_bytes": 911, "cache_tensor_bytes": null, "serialized_payload_bytes": 911, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 23029.02170899324, "hops": {"1": 23029.02170899324}, "receiver_prefill_ms": 413.226707954891, "ttft_ms": 83.41600000858307, "decode_ms": 40.62862507998943}, "total_wall_ms": 23566.81691808626, "ttft_ms": 83.41600000858307, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 15, "retention_rate": 0.23076923076923078, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.09090909090909091, "amount": 0.16666666666666666, "status": 0.5, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 855}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 15, "retention_rate": 0.23076923076923078, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.09090909090909091, "amount": 0.16666666666666666, "status": 0.5, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 855}}}, "reference_key_id": "keys/synthetic/syn_03000_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "a409b88ff601f1f9bf4d221a0c042117fadd2be19621e0b0dd2800495cfbe0cc", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "6f5be344103aeba603bbe1ef", "attempt_id": "4dbd426f52c4", "run_group_id": "E4_core:A5|closed|qwen3b:h1", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "d83aa60f20d2879d02898cb9d9d8bd1983ed2d8984cae017c9153cafee3a779a", "git_commit": null, "timestamp_utc": "2026-10-30T06:22:08.699896Z", "task_family": "synthetic", "document_id": "syn_03000_2000", "document_hash": "c9cc27c20b05b1f6e592846e84990ee9b69c42f9720a8e338ec2f6fca777fdbc", "question_id": "q12_unit_conversion", "split": "test", "cluster_id": "syn_03000_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "laptop-m5max-48g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 1, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1904, "query_tokens": 472, "message_tokens": 383, "message_utf8_bytes": 911, "cache_tensor_bytes": null, "serialized_payload_bytes": 911, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 23029.02170899324, "hops": {"1": 23029.02170899324}, "receiver_prefill_ms": 310.80541701521724, "ttft_ms": 42.133916984312236, "decode_ms": 165.56150000542402}, "total_wall_ms": 23548.036750988103, "ttft_ms": 42.133916984312236, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 5}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "20.46", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 15, "retention_rate": 0.23076923076923078, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.09090909090909091, "amount": 0.16666666666666666, "status": 0.5, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 855}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 15, "retention_rate": 0.23076923076923078, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.09090909090909091, "amount": 0.16666666666666666, "status": 0.5, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 855}}}, "reference_key_id": "keys/synthetic/syn_03000_2000.json", "task_scores": {"correct": 0.0, "abstained": 0.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "a409b88ff601f1f9bf4d221a0c042117fadd2be19621e0b0dd2800495cfbe0cc", "output_hash": "c057e11a5e818ca86a03c336373f3b4e4a2ddd9756e0acd7f5d7d655fd0267fe"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "3fb03cfac4079b43d9fad2a4", "attempt_id": "98df04e23401", "run_group_id": "E4_core:A5|closed|qwen3b:h1", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "d83aa60f20d2879d02898cb9d9d8bd1983ed2d8984cae017c9153cafee3a779a", "git_commit": null, "timestamp_utc": "2026-10-30T06:22:09.234759Z", "task_family": "synthetic", "document_id": "syn_03000_2000", "document_hash": "c9cc27c20b05b1f6e592846e84990ee9b69c42f9720a8e338ec2f6fca777fdbc", "question_id": "q13_unanswerable", "split": "test", "cluster_id": "syn_03000_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "laptop-m5max-48g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 1, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1904, "query_tokens": 476, "message_tokens": 383, "message_utf8_bytes": 911, "cache_tensor_bytes": null, "serialized_payload_bytes": 911, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 23029.02170899324, "hops": {"1": 23029.02170899324}, "receiver_prefill_ms": 304.4374999590218, "ttft_ms": 20.423583104275167, "decode_ms": 20.70308302063495}, "total_wall_ms": 23375.242625945248, "ttft_ms": 20.423583104275167, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 15, "retention_rate": 0.23076923076923078, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.09090909090909091, "amount": 0.16666666666666666, "status": 0.5, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 855}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 15, "retention_rate": 0.23076923076923078, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.09090909090909091, "amount": 0.16666666666666666, "status": 0.5, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 855}}}, "reference_key_id": "keys/synthetic/syn_03000_2000.json", "task_scores": {"correct": 1.0, "abstained": 1.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "a409b88ff601f1f9bf4d221a0c042117fadd2be19621e0b0dd2800495cfbe0cc", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "c63aeb6eb5410018405dcead", "attempt_id": "b53da9c13e68", "run_group_id": "E4_core:A5|closed|qwen3b:h1", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "d83aa60f20d2879d02898cb9d9d8bd1983ed2d8984cae017c9153cafee3a779a", "git_commit": null, "timestamp_utc": "2026-10-30T06:22:09.596298Z", "task_family": "synthetic", "document_id": "syn_03000_2000", "document_hash": "c9cc27c20b05b1f6e592846e84990ee9b69c42f9720a8e338ec2f6fca777fdbc", "question_id": "q14_unanswerable", "split": "test", "cluster_id": "syn_03000_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "laptop-m5max-48g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 1, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1904, "query_tokens": 478, "message_tokens": 383, "message_utf8_bytes": 911, "cache_tensor_bytes": null, "serialized_payload_bytes": 911, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 23029.02170899324, "hops": {"1": 23029.02170899324}, "receiver_prefill_ms": 302.4841248989105, "ttft_ms": 73.9789999788627, "decode_ms": 65.11054211296141}, "total_wall_ms": 23471.33654204663, "ttft_ms": 73.9789999788627, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 15, "retention_rate": 0.23076923076923078, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.09090909090909091, "amount": 0.16666666666666666, "status": 0.5, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 855}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 15, "retention_rate": 0.23076923076923078, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.09090909090909091, "amount": 0.16666666666666666, "status": 0.5, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 855}}}, "reference_key_id": "keys/synthetic/syn_03000_2000.json", "task_scores": {"correct": 1.0, "abstained": 1.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "a409b88ff601f1f9bf4d221a0c042117fadd2be19621e0b0dd2800495cfbe0cc", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "2b98a6c83664d81342265a5f", "attempt_id": "efb9cff605b0", "run_group_id": "E4_core:A5|closed|qwen3b:h2", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "d83aa60f20d2879d02898cb9d9d8bd1983ed2d8984cae017c9153cafee3a779a", "git_commit": null, "timestamp_utc": "2026-10-30T06:22:26.104060Z", "task_family": "synthetic", "document_id": "syn_03000_2000", "document_hash": "c9cc27c20b05b1f6e592846e84990ee9b69c42f9720a8e338ec2f6fca777fdbc", "question_id": "q00_single_fact", "split": "test", "cluster_id": "syn_03000_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "laptop-m5max-48g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 2, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1904, "query_tokens": 441, "message_tokens": 359, "message_utf8_bytes": 843, "cache_tensor_bytes": null, "serialized_payload_bytes": 843, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 39075.39233402349, "hops": {"1": 23029.02170899324, "2": 16046.37062503025}, "receiver_prefill_ms": 312.77725007385015, "ttft_ms": 42.24933299701661, "decode_ms": 59.635500074364245}, "total_wall_ms": 39490.84754299838, "ttft_ms": 42.24933299701661, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 14, "retention_rate": 0.2153846153846154, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.09090909090909091, "amount": 0.16666666666666666, "status": 0.5, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 787}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 15, "retention_rate": 0.23076923076923078, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.09090909090909091, "amount": 0.16666666666666666, "status": 0.5, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 855}, "2": {"facts_total": 65, "facts_retained": 14, "retention_rate": 0.2153846153846154, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.09090909090909091, "amount": 0.16666666666666666, "status": 0.5, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 787}}}, "reference_key_id": "keys/synthetic/syn_03000_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "f97ad5bade54fd65d2c8bafa5ade9463e557d5220c248210b1ca6dea042e54dc", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "83b6d0dee04a98e0815b9fbc", "attempt_id": "6664331ea7b0", "run_group_id": "E4_core:A5|closed|qwen3b:h2", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "d83aa60f20d2879d02898cb9d9d8bd1983ed2d8984cae017c9153cafee3a779a", "git_commit": null, "timestamp_utc": "2026-10-30T06:22:26.536833Z", "task_family": "synthetic", "document_id": "syn_03000_2000", "document_hash": "c9cc27c20b05b1f6e592846e84990ee9b69c42f9720a8e338ec2f6fca777fdbc", "question_id": "q01_single_fact", "split": "test", "cluster_id": "syn_03000_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "laptop-m5max-48g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 2, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1904, "query_tokens": 446, "message_tokens": 359, "message_utf8_bytes": 843, "cache_tensor_bytes": null, "serialized_payload_bytes": 843, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 39075.39233402349, "hops": {"1": 23029.02170899324, "2": 16046.37062503025}, "receiver_prefill_ms": 418.18637505639344, "ttft_ms": 38.13558409456164, "decode_ms": 30.44829098507762}, "total_wall_ms": 39562.644083984196, "ttft_ms": 38.13558409456164, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 14, "retention_rate": 0.2153846153846154, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.09090909090909091, "amount": 0.16666666666666666, "status": 0.5, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 787}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 15, "retention_rate": 0.23076923076923078, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.09090909090909091, "amount": 0.16666666666666666, "status": 0.5, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 855}, "2": {"facts_total": 65, "facts_retained": 14, "retention_rate": 0.2153846153846154, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.09090909090909091, "amount": 0.16666666666666666, "status": 0.5, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 787}}}, "reference_key_id": "keys/synthetic/syn_03000_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "f97ad5bade54fd65d2c8bafa5ade9463e557d5220c248210b1ca6dea042e54dc", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "ec4486436d0acd9570f3a9f2", "attempt_id": "34bf1660b25f", "run_group_id": "E4_core:A5|closed|qwen3b:h2", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "d83aa60f20d2879d02898cb9d9d8bd1983ed2d8984cae017c9153cafee3a779a", "git_commit": null, "timestamp_utc": "2026-10-30T06:22:27.040135Z", "task_family": "synthetic", "document_id": "syn_03000_2000", "document_hash": "c9cc27c20b05b1f6e592846e84990ee9b69c42f9720a8e338ec2f6fca777fdbc", "question_id": "q02_exact_copy", "split": "test", "cluster_id": "syn_03000_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "laptop-m5max-48g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 2, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1904, "query_tokens": 445, "message_tokens": 359, "message_utf8_bytes": 843, "cache_tensor_bytes": null, "serialized_payload_bytes": 843, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 39075.39233402349, "hops": {"1": 23029.02170899324, "2": 16046.37062503025}, "receiver_prefill_ms": 281.3642079709098, "ttft_ms": 42.29783301707357, "decode_ms": 25.020208093337715}, "total_wall_ms": 39424.534250982106, "ttft_ms": 42.29783301707357, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 14, "retention_rate": 0.2153846153846154, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.09090909090909091, "amount": 0.16666666666666666, "status": 0.5, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 787}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 15, "retention_rate": 0.23076923076923078, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.09090909090909091, "amount": 0.16666666666666666, "status": 0.5, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 855}, "2": {"facts_total": 65, "facts_retained": 14, "retention_rate": 0.2153846153846154, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.09090909090909091, "amount": 0.16666666666666666, "status": 0.5, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 787}}}, "reference_key_id": "keys/synthetic/syn_03000_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "f97ad5bade54fd65d2c8bafa5ade9463e557d5220c248210b1ca6dea042e54dc", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "9f82bacbd1338a41da02c938", "attempt_id": "ccb84bddb68c", "run_group_id": "E4_core:A5|closed|qwen3b:h2", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "d83aa60f20d2879d02898cb9d9d8bd1983ed2d8984cae017c9153cafee3a779a", "git_commit": null, "timestamp_utc": "2026-10-30T06:22:27.405430Z", "task_family": "synthetic", "document_id": "syn_03000_2000", "document_hash": "c9cc27c20b05b1f6e592846e84990ee9b69c42f9720a8e338ec2f6fca777fdbc", "question_id": "q03_exact_copy", "split": "test", "cluster_id": "syn_03000_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "laptop-m5max-48g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 2, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1904, "query_tokens": 444, "message_tokens": 359, "message_utf8_bytes": 843, "cache_tensor_bytes": null, "serialized_payload_bytes": 843, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 39075.39233402349, "hops": {"1": 23029.02170899324, "2": 16046.37062503025}, "receiver_prefill_ms": 297.58975002914667, "ttft_ms": 24.137083091773093, "decode_ms": 18.404374946840107}, "total_wall_ms": 39416.00604203995, "ttft_ms": 24.137083091773093, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 14, "retention_rate": 0.2153846153846154, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.09090909090909091, "amount": 0.16666666666666666, "status": 0.5, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 787}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 15, "retention_rate": 0.23076923076923078, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.09090909090909091, "amount": 0.16666666666666666, "status": 0.5, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 855}, "2": {"facts_total": 65, "facts_retained": 14, "retention_rate": 0.2153846153846154, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.09090909090909091, "amount": 0.16666666666666666, "status": 0.5, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 787}}}, "reference_key_id": "keys/synthetic/syn_03000_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "f97ad5bade54fd65d2c8bafa5ade9463e557d5220c248210b1ca6dea042e54dc", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "e45f3d0d7fcad88ee7e6689d", "attempt_id": "a938f381dcbe", "run_group_id": "E4_core:A5|closed|qwen3b:h2", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "d83aa60f20d2879d02898cb9d9d8bd1983ed2d8984cae017c9153cafee3a779a", "git_commit": null, "timestamp_utc": "2026-10-30T06:22:27.762404Z", "task_family": "synthetic", "document_id": "syn_03000_2000", "document_hash": "c9cc27c20b05b1f6e592846e84990ee9b69c42f9720a8e338ec2f6fca777fdbc", "question_id": "q04_multi_hop", "split": "test", "cluster_id": "syn_03000_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "laptop-m5max-48g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 2, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1904, "query_tokens": 450, "message_tokens": 359, "message_utf8_bytes": 843, "cache_tensor_bytes": null, "serialized_payload_bytes": 843, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 39075.39233402349, "hops": {"1": 23029.02170899324, "2": 16046.37062503025}, "receiver_prefill_ms": 332.0914580253884, "ttft_ms": 53.6324999993667, "decode_ms": 57.24404100328684}, "total_wall_ms": 39518.988459021784, "ttft_ms": 53.6324999993667, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 14, "retention_rate": 0.2153846153846154, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.09090909090909091, "amount": 0.16666666666666666, "status": 0.5, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 787}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 15, "retention_rate": 0.23076923076923078, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.09090909090909091, "amount": 0.16666666666666666, "status": 0.5, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 855}, "2": {"facts_total": 65, "facts_retained": 14, "retention_rate": 0.2153846153846154, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.09090909090909091, "amount": 0.16666666666666666, "status": 0.5, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 787}}}, "reference_key_id": "keys/synthetic/syn_03000_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "f97ad5bade54fd65d2c8bafa5ade9463e557d5220c248210b1ca6dea042e54dc", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "65297d7d02f88bd8b67849d6", "attempt_id": "df6415628733", "run_group_id": "E4_core:A5|closed|qwen3b:h2", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "d83aa60f20d2879d02898cb9d9d8bd1983ed2d8984cae017c9153cafee3a779a", "git_commit": null, "timestamp_utc": "2026-10-30T06:22:28.225309Z", "task_family": "synthetic", "document_id": "syn_03000_2000", "document_hash": "c9cc27c20b05b1f6e592846e84990ee9b69c42f9720a8e338ec2f6fca777fdbc", "question_id": "q05_multi_hop", "split": "test", "cluster_id": "syn_03000_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "laptop-m5max-48g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 2, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1904, "query_tokens": 452, "message_tokens": 359, "message_utf8_bytes": 843, "cache_tensor_bytes": null, "serialized_payload_bytes": 843, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 39075.39233402349, "hops": {"1": 23029.02170899324, "2": 16046.37062503025}, "receiver_prefill_ms": 452.93512498028576, "ttft_ms": 24.422041955403984, "decode_ms": 356.033582938835}, "total_wall_ms": 39909.262542030774, "ttft_ms": 24.422041955403984, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 10}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "LY-51E-3199", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 14, "retention_rate": 0.2153846153846154, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.09090909090909091, "amount": 0.16666666666666666, "status": 0.5, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 787}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 15, "retention_rate": 0.23076923076923078, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.09090909090909091, "amount": 0.16666666666666666, "status": 0.5, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 855}, "2": {"facts_total": 65, "facts_retained": 14, "retention_rate": 0.2153846153846154, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.09090909090909091, "amount": 0.16666666666666666, "status": 0.5, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 787}}}, "reference_key_id": "keys/synthetic/syn_03000_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "f97ad5bade54fd65d2c8bafa5ade9463e557d5220c248210b1ca6dea042e54dc", "output_hash": "a1245d1536890a7e4b026366df48441d503f491f782cfd388797b5782443c5c7"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "5407cc5c974d0fcec9aca298", "attempt_id": "723f1a3c8f74", "run_group_id": "E4_core:A5|closed|qwen3b:h2", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "d83aa60f20d2879d02898cb9d9d8bd1983ed2d8984cae017c9153cafee3a779a", "git_commit": null, "timestamp_utc": "2026-10-30T06:22:29.075572Z", "task_family": "synthetic", "document_id": "syn_03000_2000", "document_hash": "c9cc27c20b05b1f6e592846e84990ee9b69c42f9720a8e338ec2f6fca777fdbc", "question_id": "q06_negation", "split": "test", "cluster_id": "syn_03000_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "laptop-m5max-48g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 2, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1904, "query_tokens": 441, "message_tokens": 359, "message_utf8_bytes": 843, "cache_tensor_bytes": null, "serialized_payload_bytes": 843, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 39075.39233402349, "hops": {"1": 23029.02170899324, "2": 16046.37062503025}, "receiver_prefill_ms": 275.936417048797, "ttft_ms": 21.14479197189212, "decode_ms": 68.9888329943642}, "total_wall_ms": 39442.09712604061, "ttft_ms": 21.14479197189212, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "no", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 14, "retention_rate": 0.2153846153846154, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.09090909090909091, "amount": 0.16666666666666666, "status": 0.5, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 787}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 15, "retention_rate": 0.23076923076923078, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.09090909090909091, "amount": 0.16666666666666666, "status": 0.5, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 855}, "2": {"facts_total": 65, "facts_retained": 14, "retention_rate": 0.2153846153846154, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.09090909090909091, "amount": 0.16666666666666666, "status": 0.5, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 787}}}, "reference_key_id": "keys/synthetic/syn_03000_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "f97ad5bade54fd65d2c8bafa5ade9463e557d5220c248210b1ca6dea042e54dc", "output_hash": "9390298f3fb0c5b160498935d79cb139aef28e1c47358b4bbba61862b9c26e59"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "36f7aabf81d81f193667f48f", "attempt_id": "22d1a446cad0", "run_group_id": "E4_core:A5|closed|qwen3b:h2", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "d83aa60f20d2879d02898cb9d9d8bd1983ed2d8984cae017c9153cafee3a779a", "git_commit": null, "timestamp_utc": "2026-10-30T06:22:29.458929Z", "task_family": "synthetic", "document_id": "syn_03000_2000", "document_hash": "c9cc27c20b05b1f6e592846e84990ee9b69c42f9720a8e338ec2f6fca777fdbc", "question_id": "q07_negation", "split": "test", "cluster_id": "syn_03000_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "laptop-m5max-48g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 2, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1904, "query_tokens": 441, "message_tokens": 359, "message_utf8_bytes": 843, "cache_tensor_bytes": null, "serialized_payload_bytes": 843, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 39075.39233402349, "hops": {"1": 23029.02170899324, "2": 16046.37062503025}, "receiver_prefill_ms": 284.87487498205155, "ttft_ms": 37.664499948732555, "decode_ms": 109.4439160078764}, "total_wall_ms": 39508.16725101322, "ttft_ms": 37.664499948732555, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 14, "retention_rate": 0.2153846153846154, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.09090909090909091, "amount": 0.16666666666666666, "status": 0.5, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 787}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 15, "retention_rate": 0.23076923076923078, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.09090909090909091, "amount": 0.16666666666666666, "status": 0.5, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 855}, "2": {"facts_total": 65, "facts_retained": 14, "retention_rate": 0.2153846153846154, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.09090909090909091, "amount": 0.16666666666666666, "status": 0.5, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 787}}}, "reference_key_id": "keys/synthetic/syn_03000_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "f97ad5bade54fd65d2c8bafa5ade9463e557d5220c248210b1ca6dea042e54dc", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "c500afb94cde9f0361458a94", "attempt_id": "6611154ccd4a", "run_group_id": "E4_core:A5|closed|qwen3b:h2", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "d83aa60f20d2879d02898cb9d9d8bd1983ed2d8984cae017c9153cafee3a779a", "git_commit": null, "timestamp_utc": "2026-10-30T06:22:29.908144Z", "task_family": "synthetic", "document_id": "syn_03000_2000", "document_hash": "c9cc27c20b05b1f6e592846e84990ee9b69c42f9720a8e338ec2f6fca777fdbc", "question_id": "q08_revision", "split": "test", "cluster_id": "syn_03000_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "laptop-m5max-48g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 2, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1904, "query_tokens": 456, "message_tokens": 359, "message_utf8_bytes": 843, "cache_tensor_bytes": null, "serialized_payload_bytes": 843, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 39075.39233402349, "hops": {"1": 23029.02170899324, "2": 16046.37062503025}, "receiver_prefill_ms": 392.3410839634016, "ttft_ms": 38.65141700953245, "decode_ms": 293.68279210757464}, "total_wall_ms": 39800.832126056775, "ttft_ms": 38.65141700953245, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 9}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "2,705,965", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 14, "retention_rate": 0.2153846153846154, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.09090909090909091, "amount": 0.16666666666666666, "status": 0.5, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 787}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 15, "retention_rate": 0.23076923076923078, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.09090909090909091, "amount": 0.16666666666666666, "status": 0.5, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 855}, "2": {"facts_total": 65, "facts_retained": 14, "retention_rate": 0.2153846153846154, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.09090909090909091, "amount": 0.16666666666666666, "status": 0.5, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 787}}}, "reference_key_id": "keys/synthetic/syn_03000_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "f97ad5bade54fd65d2c8bafa5ade9463e557d5220c248210b1ca6dea042e54dc", "output_hash": "dbedb6f5d00e413d915513b85d8ed195b18b9acf392ece342a85e15e3752bcc8"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "65e7f5a4b5ece396ebbc3c6d", "attempt_id": "a0c4847984c8", "run_group_id": "E4_core:A5|closed|qwen3b:h2", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "d83aa60f20d2879d02898cb9d9d8bd1983ed2d8984cae017c9153cafee3a779a", "git_commit": null, "timestamp_utc": "2026-10-30T06:22:30.653232Z", "task_family": "synthetic", "document_id": "syn_03000_2000", "document_hash": "c9cc27c20b05b1f6e592846e84990ee9b69c42f9720a8e338ec2f6fca777fdbc", "question_id": "q09_revision", "split": "test", "cluster_id": "syn_03000_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "laptop-m5max-48g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 2, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1904, "query_tokens": 452, "message_tokens": 359, "message_utf8_bytes": 843, "cache_tensor_bytes": null, "serialized_payload_bytes": 843, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 39075.39233402349, "hops": {"1": 23029.02170899324, "2": 16046.37062503025}, "receiver_prefill_ms": 316.522250068374, "ttft_ms": 20.68524993956089, "decode_ms": 40.977458003908396}, "total_wall_ms": 39454.4681670377, "ttft_ms": 20.68524993956089, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 14, "retention_rate": 0.2153846153846154, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.09090909090909091, "amount": 0.16666666666666666, "status": 0.5, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 787}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 15, "retention_rate": 0.23076923076923078, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.09090909090909091, "amount": 0.16666666666666666, "status": 0.5, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 855}, "2": {"facts_total": 65, "facts_retained": 14, "retention_rate": 0.2153846153846154, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.09090909090909091, "amount": 0.16666666666666666, "status": 0.5, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 787}}}, "reference_key_id": "keys/synthetic/syn_03000_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "f97ad5bade54fd65d2c8bafa5ade9463e557d5220c248210b1ca6dea042e54dc", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "0e7db3e5979bc3cbb200a2f1", "attempt_id": "d30326a64712", "run_group_id": "E4_core:A5|closed|qwen3b:h2", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "d83aa60f20d2879d02898cb9d9d8bd1983ed2d8984cae017c9153cafee3a779a", "git_commit": null, "timestamp_utc": "2026-10-30T06:22:31.048876Z", "task_family": "synthetic", "document_id": "syn_03000_2000", "document_hash": "c9cc27c20b05b1f6e592846e84990ee9b69c42f9720a8e338ec2f6fca777fdbc", "question_id": "q10_temporal", "split": "test", "cluster_id": "syn_03000_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "laptop-m5max-48g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 2, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1904, "query_tokens": 458, "message_tokens": 359, "message_utf8_bytes": 843, "cache_tensor_bytes": null, "serialized_payload_bytes": 843, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 39075.39233402349, "hops": {"1": 23029.02170899324, "2": 16046.37062503025}, "receiver_prefill_ms": 302.6189999654889, "ttft_ms": 43.46020799130201, "decode_ms": 61.03962496854365}, "total_wall_ms": 39483.42995904386, "ttft_ms": 43.46020799130201, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 14, "retention_rate": 0.2153846153846154, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.09090909090909091, "amount": 0.16666666666666666, "status": 0.5, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 787}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 15, "retention_rate": 0.23076923076923078, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.09090909090909091, "amount": 0.16666666666666666, "status": 0.5, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 855}, "2": {"facts_total": 65, "facts_retained": 14, "retention_rate": 0.2153846153846154, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.09090909090909091, "amount": 0.16666666666666666, "status": 0.5, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 787}}}, "reference_key_id": "keys/synthetic/syn_03000_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "f97ad5bade54fd65d2c8bafa5ade9463e557d5220c248210b1ca6dea042e54dc", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "73c7f667650939cf1f35e41a", "attempt_id": "b97940b51b61", "run_group_id": "E4_core:A5|closed|qwen3b:h2", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "d83aa60f20d2879d02898cb9d9d8bd1983ed2d8984cae017c9153cafee3a779a", "git_commit": null, "timestamp_utc": "2026-10-30T06:22:31.475889Z", "task_family": "synthetic", "document_id": "syn_03000_2000", "document_hash": "c9cc27c20b05b1f6e592846e84990ee9b69c42f9720a8e338ec2f6fca777fdbc", "question_id": "q11_aggregation", "split": "test", "cluster_id": "syn_03000_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "laptop-m5max-48g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 2, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1904, "query_tokens": 453, "message_tokens": 359, "message_utf8_bytes": 843, "cache_tensor_bytes": null, "serialized_payload_bytes": 843, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 39075.39233402349, "hops": {"1": 23029.02170899324, "2": 16046.37062503025}, "receiver_prefill_ms": 452.216292032972, "ttft_ms": 29.73083406686783, "decode_ms": 41.11300001386553}, "total_wall_ms": 39599.12233403884, "ttft_ms": 29.73083406686783, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 14, "retention_rate": 0.2153846153846154, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.09090909090909091, "amount": 0.16666666666666666, "status": 0.5, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 787}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 15, "retention_rate": 0.23076923076923078, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.09090909090909091, "amount": 0.16666666666666666, "status": 0.5, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 855}, "2": {"facts_total": 65, "facts_retained": 14, "retention_rate": 0.2153846153846154, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.09090909090909091, "amount": 0.16666666666666666, "status": 0.5, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 787}}}, "reference_key_id": "keys/synthetic/syn_03000_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "f97ad5bade54fd65d2c8bafa5ade9463e557d5220c248210b1ca6dea042e54dc", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "a0a7d4e85d467f4ccce09aff", "attempt_id": "970e36930a84", "run_group_id": "E4_core:A5|closed|qwen3b:h2", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "d83aa60f20d2879d02898cb9d9d8bd1983ed2d8984cae017c9153cafee3a779a", "git_commit": null, "timestamp_utc": "2026-10-30T06:22:32.016745Z", "task_family": "synthetic", "document_id": "syn_03000_2000", "document_hash": "c9cc27c20b05b1f6e592846e84990ee9b69c42f9720a8e338ec2f6fca777fdbc", "question_id": "q12_unit_conversion", "split": "test", "cluster_id": "syn_03000_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "laptop-m5max-48g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 2, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1904, "query_tokens": 449, "message_tokens": 359, "message_utf8_bytes": 843, "cache_tensor_bytes": null, "serialized_payload_bytes": 843, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 39075.39233402349, "hops": {"1": 23029.02170899324, "2": 16046.37062503025}, "receiver_prefill_ms": 358.24929107911885, "ttft_ms": 35.09224997833371, "decode_ms": 153.3962079556659}, "total_wall_ms": 39622.64983402565, "ttft_ms": 35.09224997833371, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 5}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "20.46", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 14, "retention_rate": 0.2153846153846154, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.09090909090909091, "amount": 0.16666666666666666, "status": 0.5, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 787}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 15, "retention_rate": 0.23076923076923078, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.09090909090909091, "amount": 0.16666666666666666, "status": 0.5, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 855}, "2": {"facts_total": 65, "facts_retained": 14, "retention_rate": 0.2153846153846154, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.09090909090909091, "amount": 0.16666666666666666, "status": 0.5, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 787}}}, "reference_key_id": "keys/synthetic/syn_03000_2000.json", "task_scores": {"correct": 0.0, "abstained": 0.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "f97ad5bade54fd65d2c8bafa5ade9463e557d5220c248210b1ca6dea042e54dc", "output_hash": "c057e11a5e818ca86a03c336373f3b4e4a2ddd9756e0acd7f5d7d655fd0267fe"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "779a646334ef3ae02c28e249", "attempt_id": "2d82b691a56e", "run_group_id": "E4_core:A5|closed|qwen3b:h2", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "d83aa60f20d2879d02898cb9d9d8bd1983ed2d8984cae017c9153cafee3a779a", "git_commit": null, "timestamp_utc": "2026-10-30T06:22:32.579593Z", "task_family": "synthetic", "document_id": "syn_03000_2000", "document_hash": "c9cc27c20b05b1f6e592846e84990ee9b69c42f9720a8e338ec2f6fca777fdbc", "question_id": "q13_unanswerable", "split": "test", "cluster_id": "syn_03000_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "laptop-m5max-48g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 2, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1904, "query_tokens": 453, "message_tokens": 359, "message_utf8_bytes": 843, "cache_tensor_bytes": null, "serialized_payload_bytes": 843, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 39075.39233402349, "hops": {"1": 23029.02170899324, "2": 16046.37062503025}, "receiver_prefill_ms": 319.36724996194243, "ttft_ms": 52.82683402765542, "decode_ms": 61.57233298290521}, "total_wall_ms": 39509.78408404626, "ttft_ms": 52.82683402765542, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 14, "retention_rate": 0.2153846153846154, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.09090909090909091, "amount": 0.16666666666666666, "status": 0.5, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 787}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 15, "retention_rate": 0.23076923076923078, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.09090909090909091, "amount": 0.16666666666666666, "status": 0.5, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 855}, "2": {"facts_total": 65, "facts_retained": 14, "retention_rate": 0.2153846153846154, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.09090909090909091, "amount": 0.16666666666666666, "status": 0.5, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 787}}}, "reference_key_id": "keys/synthetic/syn_03000_2000.json", "task_scores": {"correct": 1.0, "abstained": 1.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "f97ad5bade54fd65d2c8bafa5ade9463e557d5220c248210b1ca6dea042e54dc", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "7263d888d8f91c70ff4f1ba0", "attempt_id": "7d7d0145723c", "run_group_id": "E4_core:A5|closed|qwen3b:h2", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "d83aa60f20d2879d02898cb9d9d8bd1983ed2d8984cae017c9153cafee3a779a", "git_commit": null, "timestamp_utc": "2026-10-30T06:22:33.031136Z", "task_family": "synthetic", "document_id": "syn_03000_2000", "document_hash": "c9cc27c20b05b1f6e592846e84990ee9b69c42f9720a8e338ec2f6fca777fdbc", "question_id": "q14_unanswerable", "split": "test", "cluster_id": "syn_03000_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "laptop-m5max-48g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 2, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1904, "query_tokens": 455, "message_tokens": 359, "message_utf8_bytes": 843, "cache_tensor_bytes": null, "serialized_payload_bytes": 843, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 39075.39233402349, "hops": {"1": 23029.02170899324, "2": 16046.37062503025}, "receiver_prefill_ms": 464.8899999447167, "ttft_ms": 37.10654203314334, "decode_ms": 24.120041984133422}, "total_wall_ms": 39602.00991795864, "ttft_ms": 37.10654203314334, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 14, "retention_rate": 0.2153846153846154, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.09090909090909091, "amount": 0.16666666666666666, "status": 0.5, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 787}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 15, "retention_rate": 0.23076923076923078, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.09090909090909091, "amount": 0.16666666666666666, "status": 0.5, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 855}, "2": {"facts_total": 65, "facts_retained": 14, "retention_rate": 0.2153846153846154, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.09090909090909091, "amount": 0.16666666666666666, "status": 0.5, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 787}}}, "reference_key_id": "keys/synthetic/syn_03000_2000.json", "task_scores": {"correct": 1.0, "abstained": 1.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "f97ad5bade54fd65d2c8bafa5ade9463e557d5220c248210b1ca6dea042e54dc", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "cfea9b7180236ff7f8703662", "attempt_id": "7a52bc293030", "run_group_id": "E4_core:A5|closed|qwen3b:h4", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "d83aa60f20d2879d02898cb9d9d8bd1983ed2d8984cae017c9153cafee3a779a", "git_commit": null, "timestamp_utc": "2026-10-30T06:23:05.786642Z", "task_family": "synthetic", "document_id": "syn_03000_2000", "document_hash": "c9cc27c20b05b1f6e592846e84990ee9b69c42f9720a8e338ec2f6fca777fdbc", "question_id": "q00_single_fact", "split": "test", "cluster_id": "syn_03000_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "laptop-m5max-48g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 4, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1904, "query_tokens": 441, "message_tokens": 359, "message_utf8_bytes": 843, "cache_tensor_bytes": null, "serialized_payload_bytes": 843, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 71283.19483390078, "hops": {"1": 23029.02170899324, "2": 16046.37062503025, "3": 16191.568707930855, "4": 16016.233791946433}, "receiver_prefill_ms": 442.3644590424374, "ttft_ms": 38.307499955408275, "decode_ms": 45.300416997633874}, "total_wall_ms": 71809.67875092756, "ttft_ms": 38.307499955408275, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 14, "retention_rate": 0.2153846153846154, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.09090909090909091, "amount": 0.16666666666666666, "status": 0.5, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 787}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 15, "retention_rate": 0.23076923076923078, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.09090909090909091, "amount": 0.16666666666666666, "status": 0.5, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 855}, "2": {"facts_total": 65, "facts_retained": 14, "retention_rate": 0.2153846153846154, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.09090909090909091, "amount": 0.16666666666666666, "status": 0.5, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 787}, "3": {"facts_total": 65, "facts_retained": 14, "retention_rate": 0.2153846153846154, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.09090909090909091, "amount": 0.16666666666666666, "status": 0.5, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 787}, "4": {"facts_total": 65, "facts_retained": 14, "retention_rate": 0.2153846153846154, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.09090909090909091, "amount": 0.16666666666666666, "status": 0.5, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 787}}}, "reference_key_id": "keys/synthetic/syn_03000_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "f97ad5bade54fd65d2c8bafa5ade9463e557d5220c248210b1ca6dea042e54dc", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "b3487ae4b9e2052a6e0838ee", "attempt_id": "70b812eb83d6", "run_group_id": "E4_core:A5|closed|qwen3b:h4", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "d83aa60f20d2879d02898cb9d9d8bd1983ed2d8984cae017c9153cafee3a779a", "git_commit": null, "timestamp_utc": "2026-10-30T06:23:06.329901Z", "task_family": "synthetic", "document_id": "syn_03000_2000", "document_hash": "c9cc27c20b05b1f6e592846e84990ee9b69c42f9720a8e338ec2f6fca777fdbc", "question_id": "q01_single_fact", "split": "test", "cluster_id": "syn_03000_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "laptop-m5max-48g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 4, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1904, "query_tokens": 446, "message_tokens": 359, "message_utf8_bytes": 843, "cache_tensor_bytes": null, "serialized_payload_bytes": 843, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 71283.19483390078, "hops": {"1": 23029.02170899324, "2": 16046.37062503025, "3": 16191.568707930855, "4": 16016.233791946433}, "receiver_prefill_ms": 285.6697500683367, "ttft_ms": 36.86495800502598, "decode_ms": 42.154749971814454}, "total_wall_ms": 71648.43404188287, "ttft_ms": 36.86495800502598, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 14, "retention_rate": 0.2153846153846154, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.09090909090909091, "amount": 0.16666666666666666, "status": 0.5, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 787}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 15, "retention_rate": 0.23076923076923078, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.09090909090909091, "amount": 0.16666666666666666, "status": 0.5, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 855}, "2": {"facts_total": 65, "facts_retained": 14, "retention_rate": 0.2153846153846154, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.09090909090909091, "amount": 0.16666666666666666, "status": 0.5, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 787}, "3": {"facts_total": 65, "facts_retained": 14, "retention_rate": 0.2153846153846154, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.09090909090909091, "amount": 0.16666666666666666, "status": 0.5, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 787}, "4": {"facts_total": 65, "facts_retained": 14, "retention_rate": 0.2153846153846154, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.09090909090909091, "amount": 0.16666666666666666, "status": 0.5, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 787}}}, "reference_key_id": "keys/synthetic/syn_03000_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "f97ad5bade54fd65d2c8bafa5ade9463e557d5220c248210b1ca6dea042e54dc", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "a20375da1c01b63d2ecc6cb0", "attempt_id": "a42b7fb8f1fc", "run_group_id": "E4_core:A5|closed|qwen3b:h4", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "d83aa60f20d2879d02898cb9d9d8bd1983ed2d8984cae017c9153cafee3a779a", "git_commit": null, "timestamp_utc": "2026-10-30T06:23:06.711736Z", "task_family": "synthetic", "document_id": "syn_03000_2000", "document_hash": "c9cc27c20b05b1f6e592846e84990ee9b69c42f9720a8e338ec2f6fca777fdbc", "question_id": "q02_exact_copy", "split": "test", "cluster_id": "syn_03000_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "laptop-m5max-48g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 4, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1904, "query_tokens": 445, "message_tokens": 359, "message_utf8_bytes": 843, "cache_tensor_bytes": null, "serialized_payload_bytes": 843, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 71283.19483390078, "hops": {"1": 23029.02170899324, "2": 16046.37062503025, "3": 16191.568707930855, "4": 16016.233791946433}, "receiver_prefill_ms": 290.5840000603348, "ttft_ms": 24.0307500353083, "decode_ms": 21.707999985665083}, "total_wall_ms": 71620.03491690848, "ttft_ms": 24.0307500353083, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 14, "retention_rate": 0.2153846153846154, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.09090909090909091, "amount": 0.16666666666666666, "status": 0.5, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 787}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 15, "retention_rate": 0.23076923076923078, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.09090909090909091, "amount": 0.16666666666666666, "status": 0.5, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 855}, "2": {"facts_total": 65, "facts_retained": 14, "retention_rate": 0.2153846153846154, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.09090909090909091, "amount": 0.16666666666666666, "status": 0.5, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 787}, "3": {"facts_total": 65, "facts_retained": 14, "retention_rate": 0.2153846153846154, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.09090909090909091, "amount": 0.16666666666666666, "status": 0.5, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 787}, "4": {"facts_total": 65, "facts_retained": 14, "retention_rate": 0.2153846153846154, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.09090909090909091, "amount": 0.16666666666666666, "status": 0.5, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 787}}}, "reference_key_id": "keys/synthetic/syn_03000_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "f97ad5bade54fd65d2c8bafa5ade9463e557d5220c248210b1ca6dea042e54dc", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "f7f480f50a343142b951b3d8", "attempt_id": "6e8284165ecf", "run_group_id": "E4_core:A5|closed|qwen3b:h4", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "d83aa60f20d2879d02898cb9d9d8bd1983ed2d8984cae017c9153cafee3a779a", "git_commit": null, "timestamp_utc": "2026-10-30T06:23:07.064815Z", "task_family": "synthetic", "document_id": "syn_03000_2000", "document_hash": "c9cc27c20b05b1f6e592846e84990ee9b69c42f9720a8e338ec2f6fca777fdbc", "question_id": "q03_exact_copy", "split": "test", "cluster_id": "syn_03000_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "laptop-m5max-48g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 4, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1904, "query_tokens": 444, "message_tokens": 359, "message_utf8_bytes": 843, "cache_tensor_bytes": null, "serialized_payload_bytes": 843, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 71283.19483390078, "hops": {"1": 23029.02170899324, "2": 16046.37062503025, "3": 16191.568707930855, "4": 16016.233791946433}, "receiver_prefill_ms": 343.3469169540331, "ttft_ms": 55.19187496975064, "decode_ms": 97.76679193601012}, "total_wall_ms": 71780.17054288648, "ttft_ms": 55.19187496975064, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 14, "retention_rate": 0.2153846153846154, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.09090909090909091, "amount": 0.16666666666666666, "status": 0.5, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 787}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 15, "retention_rate": 0.23076923076923078, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.09090909090909091, "amount": 0.16666666666666666, "status": 0.5, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 855}, "2": {"facts_total": 65, "facts_retained": 14, "retention_rate": 0.2153846153846154, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.09090909090909091, "amount": 0.16666666666666666, "status": 0.5, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 787}, "3": {"facts_total": 65, "facts_retained": 14, "retention_rate": 0.2153846153846154, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.09090909090909091, "amount": 0.16666666666666666, "status": 0.5, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 787}, "4": {"facts_total": 65, "facts_retained": 14, "retention_rate": 0.2153846153846154, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.09090909090909091, "amount": 0.16666666666666666, "status": 0.5, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 787}}}, "reference_key_id": "keys/synthetic/syn_03000_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "f97ad5bade54fd65d2c8bafa5ade9463e557d5220c248210b1ca6dea042e54dc", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "7b99cdabed492c785604a2b9", "attempt_id": "0e6abd9d063c", "run_group_id": "E4_core:A5|closed|qwen3b:h4", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "d83aa60f20d2879d02898cb9d9d8bd1983ed2d8984cae017c9153cafee3a779a", "git_commit": null, "timestamp_utc": "2026-10-30T06:23:07.580753Z", "task_family": "synthetic", "document_id": "syn_03000_2000", "document_hash": "c9cc27c20b05b1f6e592846e84990ee9b69c42f9720a8e338ec2f6fca777fdbc", "question_id": "q04_multi_hop", "split": "test", "cluster_id": "syn_03000_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "laptop-m5max-48g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 4, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1904, "query_tokens": 450, "message_tokens": 359, "message_utf8_bytes": 843, "cache_tensor_bytes": null, "serialized_payload_bytes": 843, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 71283.19483390078, "hops": {"1": 23029.02170899324, "2": 16046.37062503025, "3": 16191.568707930855, "4": 16016.233791946433}, "receiver_prefill_ms": 402.74791698902845, "ttft_ms": 38.60829200129956, "decode_ms": 24.58375005517155}, "total_wall_ms": 71749.63683390524, "ttft_ms": 38.60829200129956, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 14, "retention_rate": 0.2153846153846154, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.09090909090909091, "amount": 0.16666666666666666, "status": 0.5, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 787}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 15, "retention_rate": 0.23076923076923078, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.09090909090909091, "amount": 0.16666666666666666, "status": 0.5, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 855}, "2": {"facts_total": 65, "facts_retained": 14, "retention_rate": 0.2153846153846154, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.09090909090909091, "amount": 0.16666666666666666, "status": 0.5, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 787}, "3": {"facts_total": 65, "facts_retained": 14, "retention_rate": 0.2153846153846154, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.09090909090909091, "amount": 0.16666666666666666, "status": 0.5, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 787}, "4": {"facts_total": 65, "facts_retained": 14, "retention_rate": 0.2153846153846154, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.09090909090909091, "amount": 0.16666666666666666, "status": 0.5, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 787}}}, "reference_key_id": "keys/synthetic/syn_03000_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "f97ad5bade54fd65d2c8bafa5ade9463e557d5220c248210b1ca6dea042e54dc", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "a8eaa73cc55ead63136b081d", "attempt_id": "d56eedda7c0b", "run_group_id": "E4_core:A5|closed|qwen3b:h4", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "d83aa60f20d2879d02898cb9d9d8bd1983ed2d8984cae017c9153cafee3a779a", "git_commit": null, "timestamp_utc": "2026-10-30T06:23:08.066506Z", "task_family": "synthetic", "document_id": "syn_03000_2000", "document_hash": "c9cc27c20b05b1f6e592846e84990ee9b69c42f9720a8e338ec2f6fca777fdbc", "question_id": "q05_multi_hop", "split": "test", "cluster_id": "syn_03000_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "laptop-m5max-48g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 4, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1904, "query_tokens": 452, "message_tokens": 359, "message_utf8_bytes": 843, "cache_tensor_bytes": null, "serialized_payload_bytes": 843, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 71283.19483390078, "hops": {"1": 23029.02170899324, "2": 16046.37062503025, "3": 16191.568707930855, "4": 16016.233791946433}, "receiver_prefill_ms": 320.84858405869454, "ttft_ms": 39.277125033549964, "decode_ms": 312.6839589094743}, "total_wall_ms": 71956.61050081253, "ttft_ms": 39.277125033549964, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 10}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "LY-51E-3199", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 14, "retention_rate": 0.2153846153846154, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.09090909090909091, "amount": 0.16666666666666666, "status": 0.5, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 787}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 15, "retention_rate": 0.23076923076923078, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.09090909090909091, "amount": 0.16666666666666666, "status": 0.5, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 855}, "2": {"facts_total": 65, "facts_retained": 14, "retention_rate": 0.2153846153846154, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.09090909090909091, "amount": 0.16666666666666666, "status": 0.5, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 787}, "3": {"facts_total": 65, "facts_retained": 14, "retention_rate": 0.2153846153846154, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.09090909090909091, "amount": 0.16666666666666666, "status": 0.5, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 787}, "4": {"facts_total": 65, "facts_retained": 14, "retention_rate": 0.2153846153846154, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.09090909090909091, "amount": 0.16666666666666666, "status": 0.5, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 787}}}, "reference_key_id": "keys/synthetic/syn_03000_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "f97ad5bade54fd65d2c8bafa5ade9463e557d5220c248210b1ca6dea042e54dc", "output_hash": "a1245d1536890a7e4b026366df48441d503f491f782cfd388797b5782443c5c7"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "46461602287783508479448b", "attempt_id": "40ad6a6a4e7c", "run_group_id": "E4_core:A5|closed|qwen3b:h4", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "d83aa60f20d2879d02898cb9d9d8bd1983ed2d8984cae017c9153cafee3a779a", "git_commit": null, "timestamp_utc": "2026-10-30T06:23:08.756752Z", "task_family": "synthetic", "document_id": "syn_03000_2000", "document_hash": "c9cc27c20b05b1f6e592846e84990ee9b69c42f9720a8e338ec2f6fca777fdbc", "question_id": "q06_negation", "split": "test", "cluster_id": "syn_03000_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "laptop-m5max-48g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 4, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1904, "query_tokens": 441, "message_tokens": 359, "message_utf8_bytes": 843, "cache_tensor_bytes": null, "serialized_payload_bytes": 843, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 71283.19483390078, "hops": {"1": 23029.02170899324, "2": 16046.37062503025, "3": 16191.568707930855, "4": 16016.233791946433}, "receiver_prefill_ms": 286.1999999731779, "ttft_ms": 44.66737504117191, "decode_ms": 114.49679103679955}, "total_wall_ms": 71729.11379195284, "ttft_ms": 44.66737504117191, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "no", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 14, "retention_rate": 0.2153846153846154, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.09090909090909091, "amount": 0.16666666666666666, "status": 0.5, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 787}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 15, "retention_rate": 0.23076923076923078, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.09090909090909091, "amount": 0.16666666666666666, "status": 0.5, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 855}, "2": {"facts_total": 65, "facts_retained": 14, "retention_rate": 0.2153846153846154, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.09090909090909091, "amount": 0.16666666666666666, "status": 0.5, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 787}, "3": {"facts_total": 65, "facts_retained": 14, "retention_rate": 0.2153846153846154, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.09090909090909091, "amount": 0.16666666666666666, "status": 0.5, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 787}, "4": {"facts_total": 65, "facts_retained": 14, "retention_rate": 0.2153846153846154, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.09090909090909091, "amount": 0.16666666666666666, "status": 0.5, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 787}}}, "reference_key_id": "keys/synthetic/syn_03000_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "f97ad5bade54fd65d2c8bafa5ade9463e557d5220c248210b1ca6dea042e54dc", "output_hash": "9390298f3fb0c5b160498935d79cb139aef28e1c47358b4bbba61862b9c26e59"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "27fd4d9dc56949dc868015ff", "attempt_id": "4bf885b8794e", "run_group_id": "E4_core:A5|closed|qwen3b:h4", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "d83aa60f20d2879d02898cb9d9d8bd1983ed2d8984cae017c9153cafee3a779a", "git_commit": null, "timestamp_utc": "2026-10-30T06:23:09.220727Z", "task_family": "synthetic", "document_id": "syn_03000_2000", "document_hash": "c9cc27c20b05b1f6e592846e84990ee9b69c42f9720a8e338ec2f6fca777fdbc", "question_id": "q07_negation", "split": "test", "cluster_id": "syn_03000_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "laptop-m5max-48g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 4, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1904, "query_tokens": 441, "message_tokens": 359, "message_utf8_bytes": 843, "cache_tensor_bytes": null, "serialized_payload_bytes": 843, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 71283.19483390078, "hops": {"1": 23029.02170899324, "2": 16046.37062503025, "3": 16191.568707930855, "4": 16016.233791946433}, "receiver_prefill_ms": 373.2789580244571, "ttft_ms": 40.499166934750974, "decode_ms": 30.990083003416657}, "total_wall_ms": 71728.4232089296, "ttft_ms": 40.499166934750974, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 14, "retention_rate": 0.2153846153846154, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.09090909090909091, "amount": 0.16666666666666666, "status": 0.5, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 787}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 15, "retention_rate": 0.23076923076923078, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.09090909090909091, "amount": 0.16666666666666666, "status": 0.5, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 855}, "2": {"facts_total": 65, "facts_retained": 14, "retention_rate": 0.2153846153846154, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.09090909090909091, "amount": 0.16666666666666666, "status": 0.5, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 787}, "3": {"facts_total": 65, "facts_retained": 14, "retention_rate": 0.2153846153846154, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.09090909090909091, "amount": 0.16666666666666666, "status": 0.5, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 787}, "4": {"facts_total": 65, "facts_retained": 14, "retention_rate": 0.2153846153846154, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.09090909090909091, "amount": 0.16666666666666666, "status": 0.5, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 787}}}, "reference_key_id": "keys/synthetic/syn_03000_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "f97ad5bade54fd65d2c8bafa5ade9463e557d5220c248210b1ca6dea042e54dc", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "80eaa85fb6daffd2e1803a58", "attempt_id": "dee61f2104cf", "run_group_id": "E4_core:A5|closed|qwen3b:h4", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "d83aa60f20d2879d02898cb9d9d8bd1983ed2d8984cae017c9153cafee3a779a", "git_commit": null, "timestamp_utc": "2026-10-30T06:23:09.684207Z", "task_family": "synthetic", "document_id": "syn_03000_2000", "document_hash": "c9cc27c20b05b1f6e592846e84990ee9b69c42f9720a8e338ec2f6fca777fdbc", "question_id": "q08_revision", "split": "test", "cluster_id": "syn_03000_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "laptop-m5max-48g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 4, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1904, "query_tokens": 456, "message_tokens": 359, "message_utf8_bytes": 843, "cache_tensor_bytes": null, "serialized_payload_bytes": 843, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 71283.19483390078, "hops": {"1": 23029.02170899324, "2": 16046.37062503025, "3": 16191.568707930855, "4": 16016.233791946433}, "receiver_prefill_ms": 322.046541958116, "ttft_ms": 35.339833004400134, "decode_ms": 275.4991250112653}, "total_wall_ms": 71916.73158388585, "ttft_ms": 35.339833004400134, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 9}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "2,705,965", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 14, "retention_rate": 0.2153846153846154, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.09090909090909091, "amount": 0.16666666666666666, "status": 0.5, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 787}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 15, "retention_rate": 0.23076923076923078, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.09090909090909091, "amount": 0.16666666666666666, "status": 0.5, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 855}, "2": {"facts_total": 65, "facts_retained": 14, "retention_rate": 0.2153846153846154, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.09090909090909091, "amount": 0.16666666666666666, "status": 0.5, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 787}, "3": {"facts_total": 65, "facts_retained": 14, "retention_rate": 0.2153846153846154, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.09090909090909091, "amount": 0.16666666666666666, "status": 0.5, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 787}, "4": {"facts_total": 65, "facts_retained": 14, "retention_rate": 0.2153846153846154, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.09090909090909091, "amount": 0.16666666666666666, "status": 0.5, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 787}}}, "reference_key_id": "keys/synthetic/syn_03000_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "f97ad5bade54fd65d2c8bafa5ade9463e557d5220c248210b1ca6dea042e54dc", "output_hash": "dbedb6f5d00e413d915513b85d8ed195b18b9acf392ece342a85e15e3752bcc8"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "9d817ca05a08d0f8424a8511", "attempt_id": "2ceec73e2fe3", "run_group_id": "E4_core:A5|closed|qwen3b:h4", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "d83aa60f20d2879d02898cb9d9d8bd1983ed2d8984cae017c9153cafee3a779a", "git_commit": null, "timestamp_utc": "2026-10-30T06:23:10.333730Z", "task_family": "synthetic", "document_id": "syn_03000_2000", "document_hash": "c9cc27c20b05b1f6e592846e84990ee9b69c42f9720a8e338ec2f6fca777fdbc", "question_id": "q09_revision", "split": "test", "cluster_id": "syn_03000_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "laptop-m5max-48g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 4, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1904, "query_tokens": 452, "message_tokens": 359, "message_utf8_bytes": 843, "cache_tensor_bytes": null, "serialized_payload_bytes": 843, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 71283.19483390078, "hops": {"1": 23029.02170899324, "2": 16046.37062503025, "3": 16191.568707930855, "4": 16016.233791946433}, "receiver_prefill_ms": 343.78025005571544, "ttft_ms": 65.2504579629749, "decode_ms": 93.32295809872448}, "total_wall_ms": 71786.33454197552, "ttft_ms": 65.2504579629749, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 14, "retention_rate": 0.2153846153846154, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.09090909090909091, "amount": 0.16666666666666666, "status": 0.5, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 787}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 15, "retention_rate": 0.23076923076923078, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.09090909090909091, "amount": 0.16666666666666666, "status": 0.5, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 855}, "2": {"facts_total": 65, "facts_retained": 14, "retention_rate": 0.2153846153846154, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.09090909090909091, "amount": 0.16666666666666666, "status": 0.5, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 787}, "3": {"facts_total": 65, "facts_retained": 14, "retention_rate": 0.2153846153846154, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.09090909090909091, "amount": 0.16666666666666666, "status": 0.5, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 787}, "4": {"facts_total": 65, "facts_retained": 14, "retention_rate": 0.2153846153846154, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.09090909090909091, "amount": 0.16666666666666666, "status": 0.5, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 787}}}, "reference_key_id": "keys/synthetic/syn_03000_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "f97ad5bade54fd65d2c8bafa5ade9463e557d5220c248210b1ca6dea042e54dc", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "76fe6c81b22ce024f1bf2338", "attempt_id": "c2472063dca6", "run_group_id": "E4_core:A5|closed|qwen3b:h4", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "d83aa60f20d2879d02898cb9d9d8bd1983ed2d8984cae017c9153cafee3a779a", "git_commit": null, "timestamp_utc": "2026-10-30T06:23:10.855973Z", "task_family": "synthetic", "document_id": "syn_03000_2000", "document_hash": "c9cc27c20b05b1f6e592846e84990ee9b69c42f9720a8e338ec2f6fca777fdbc", "question_id": "q10_temporal", "split": "test", "cluster_id": "syn_03000_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "laptop-m5max-48g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 4, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1904, "query_tokens": 458, "message_tokens": 359, "message_utf8_bytes": 843, "cache_tensor_bytes": null, "serialized_payload_bytes": 843, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 71283.19483390078, "hops": {"1": 23029.02170899324, "2": 16046.37062503025, "3": 16191.568707930855, "4": 16016.233791946433}, "receiver_prefill_ms": 387.7357500605285, "ttft_ms": 37.61716594453901, "decode_ms": 24.228833033703268}, "total_wall_ms": 71733.27433387749, "ttft_ms": 37.61716594453901, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 14, "retention_rate": 0.2153846153846154, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.09090909090909091, "amount": 0.16666666666666666, "status": 0.5, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 787}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 15, "retention_rate": 0.23076923076923078, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.09090909090909091, "amount": 0.16666666666666666, "status": 0.5, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 855}, "2": {"facts_total": 65, "facts_retained": 14, "retention_rate": 0.2153846153846154, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.09090909090909091, "amount": 0.16666666666666666, "status": 0.5, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 787}, "3": {"facts_total": 65, "facts_retained": 14, "retention_rate": 0.2153846153846154, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.09090909090909091, "amount": 0.16666666666666666, "status": 0.5, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 787}, "4": {"facts_total": 65, "facts_retained": 14, "retention_rate": 0.2153846153846154, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.09090909090909091, "amount": 0.16666666666666666, "status": 0.5, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 787}}}, "reference_key_id": "keys/synthetic/syn_03000_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "f97ad5bade54fd65d2c8bafa5ade9463e557d5220c248210b1ca6dea042e54dc", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "8b9c7594a0b803f0a2e5155f", "attempt_id": "ef37c4acdeab", "run_group_id": "E4_core:A5|closed|qwen3b:h4", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "d83aa60f20d2879d02898cb9d9d8bd1983ed2d8984cae017c9153cafee3a779a", "git_commit": null, "timestamp_utc": "2026-10-30T06:23:11.322662Z", "task_family": "synthetic", "document_id": "syn_03000_2000", "document_hash": "c9cc27c20b05b1f6e592846e84990ee9b69c42f9720a8e338ec2f6fca777fdbc", "question_id": "q11_aggregation", "split": "test", "cluster_id": "syn_03000_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "laptop-m5max-48g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 4, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1904, "query_tokens": 453, "message_tokens": 359, "message_utf8_bytes": 843, "cache_tensor_bytes": null, "serialized_payload_bytes": 843, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 71283.19483390078, "hops": {"1": 23029.02170899324, "2": 16046.37062503025, "3": 16191.568707930855, "4": 16016.233791946433}, "receiver_prefill_ms": 340.2045830152929, "ttft_ms": 30.94812505878508, "decode_ms": 40.478750015608966}, "total_wall_ms": 71695.31029183418, "ttft_ms": 30.94812505878508, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 14, "retention_rate": 0.2153846153846154, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.09090909090909091, "amount": 0.16666666666666666, "status": 0.5, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 787}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 15, "retention_rate": 0.23076923076923078, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.09090909090909091, "amount": 0.16666666666666666, "status": 0.5, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 855}, "2": {"facts_total": 65, "facts_retained": 14, "retention_rate": 0.2153846153846154, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.09090909090909091, "amount": 0.16666666666666666, "status": 0.5, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 787}, "3": {"facts_total": 65, "facts_retained": 14, "retention_rate": 0.2153846153846154, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.09090909090909091, "amount": 0.16666666666666666, "status": 0.5, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 787}, "4": {"facts_total": 65, "facts_retained": 14, "retention_rate": 0.2153846153846154, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.09090909090909091, "amount": 0.16666666666666666, "status": 0.5, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 787}}}, "reference_key_id": "keys/synthetic/syn_03000_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "f97ad5bade54fd65d2c8bafa5ade9463e557d5220c248210b1ca6dea042e54dc", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "a1d08d3e132cad6e68721aef", "attempt_id": "57e0dbe59f80", "run_group_id": "E4_core:A5|closed|qwen3b:h4", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "d83aa60f20d2879d02898cb9d9d8bd1983ed2d8984cae017c9153cafee3a779a", "git_commit": null, "timestamp_utc": "2026-10-30T06:23:11.751810Z", "task_family": "synthetic", "document_id": "syn_03000_2000", "document_hash": "c9cc27c20b05b1f6e592846e84990ee9b69c42f9720a8e338ec2f6fca777fdbc", "question_id": "q12_unit_conversion", "split": "test", "cluster_id": "syn_03000_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "laptop-m5max-48g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 4, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1904, "query_tokens": 449, "message_tokens": 359, "message_utf8_bytes": 843, "cache_tensor_bytes": null, "serialized_payload_bytes": 843, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 71283.19483390078, "hops": {"1": 23029.02170899324, "2": 16046.37062503025, "3": 16191.568707930855, "4": 16016.233791946433}, "receiver_prefill_ms": 276.6477910336107, "ttft_ms": 44.85345899593085, "decode_ms": 270.3088748967275}, "total_wall_ms": 71875.5506669404, "ttft_ms": 44.85345899593085, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 5}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "20.46", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 14, "retention_rate": 0.2153846153846154, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.09090909090909091, "amount": 0.16666666666666666, "status": 0.5, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 787}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 15, "retention_rate": 0.23076923076923078, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.09090909090909091, "amount": 0.16666666666666666, "status": 0.5, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 855}, "2": {"facts_total": 65, "facts_retained": 14, "retention_rate": 0.2153846153846154, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.09090909090909091, "amount": 0.16666666666666666, "status": 0.5, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 787}, "3": {"facts_total": 65, "facts_retained": 14, "retention_rate": 0.2153846153846154, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.09090909090909091, "amount": 0.16666666666666666, "status": 0.5, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 787}, "4": {"facts_total": 65, "facts_retained": 14, "retention_rate": 0.2153846153846154, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.09090909090909091, "amount": 0.16666666666666666, "status": 0.5, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 787}}}, "reference_key_id": "keys/synthetic/syn_03000_2000.json", "task_scores": {"correct": 0.0, "abstained": 0.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "f97ad5bade54fd65d2c8bafa5ade9463e557d5220c248210b1ca6dea042e54dc", "output_hash": "c057e11a5e818ca86a03c336373f3b4e4a2ddd9756e0acd7f5d7d655fd0267fe"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "b5867ff0946f6865e994f2f0", "attempt_id": "102a1ea8e6bd", "run_group_id": "E4_core:A5|closed|qwen3b:h4", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "d83aa60f20d2879d02898cb9d9d8bd1983ed2d8984cae017c9153cafee3a779a", "git_commit": null, "timestamp_utc": "2026-10-30T06:23:12.361508Z", "task_family": "synthetic", "document_id": "syn_03000_2000", "document_hash": "c9cc27c20b05b1f6e592846e84990ee9b69c42f9720a8e338ec2f6fca777fdbc", "question_id": "q13_unanswerable", "split": "test", "cluster_id": "syn_03000_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "laptop-m5max-48g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 4, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1904, "query_tokens": 453, "message_tokens": 359, "message_utf8_bytes": 843, "cache_tensor_bytes": null, "serialized_payload_bytes": 843, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 71283.19483390078, "hops": {"1": 23029.02170899324, "2": 16046.37062503025, "3": 16191.568707930855, "4": 16016.233791946433}, "receiver_prefill_ms": 441.6507499990985, "ttft_ms": 36.60404193215072, "decode_ms": 26.871541980654}, "total_wall_ms": 71788.8813338941, "ttft_ms": 36.60404193215072, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 14, "retention_rate": 0.2153846153846154, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.09090909090909091, "amount": 0.16666666666666666, "status": 0.5, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 787}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 15, "retention_rate": 0.23076923076923078, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.09090909090909091, "amount": 0.16666666666666666, "status": 0.5, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 855}, "2": {"facts_total": 65, "facts_retained": 14, "retention_rate": 0.2153846153846154, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.09090909090909091, "amount": 0.16666666666666666, "status": 0.5, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 787}, "3": {"facts_total": 65, "facts_retained": 14, "retention_rate": 0.2153846153846154, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.09090909090909091, "amount": 0.16666666666666666, "status": 0.5, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 787}, "4": {"facts_total": 65, "facts_retained": 14, "retention_rate": 0.2153846153846154, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.09090909090909091, "amount": 0.16666666666666666, "status": 0.5, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 787}}}, "reference_key_id": "keys/synthetic/syn_03000_2000.json", "task_scores": {"correct": 1.0, "abstained": 1.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "f97ad5bade54fd65d2c8bafa5ade9463e557d5220c248210b1ca6dea042e54dc", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "b34d1f10ae04f8aa8597d99f", "attempt_id": "7aa7dc4704a1", "run_group_id": "E4_core:A5|closed|qwen3b:h4", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "d83aa60f20d2879d02898cb9d9d8bd1983ed2d8984cae017c9153cafee3a779a", "git_commit": null, "timestamp_utc": "2026-10-30T06:23:12.884126Z", "task_family": "synthetic", "document_id": "syn_03000_2000", "document_hash": "c9cc27c20b05b1f6e592846e84990ee9b69c42f9720a8e338ec2f6fca777fdbc", "question_id": "q14_unanswerable", "split": "test", "cluster_id": "syn_03000_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "laptop-m5max-48g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 4, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1904, "query_tokens": 455, "message_tokens": 359, "message_utf8_bytes": 843, "cache_tensor_bytes": null, "serialized_payload_bytes": 843, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 71283.19483390078, "hops": {"1": 23029.02170899324, "2": 16046.37062503025, "3": 16191.568707930855, "4": 16016.233791946433}, "receiver_prefill_ms": 317.7424999885261, "ttft_ms": 37.284707999788225, "decode_ms": 44.59391697309911}, "total_wall_ms": 71683.3429998951, "ttft_ms": 37.284707999788225, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 14, "retention_rate": 0.2153846153846154, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.09090909090909091, "amount": 0.16666666666666666, "status": 0.5, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 787}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 15, "retention_rate": 0.23076923076923078, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.09090909090909091, "amount": 0.16666666666666666, "status": 0.5, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 855}, "2": {"facts_total": 65, "facts_retained": 14, "retention_rate": 0.2153846153846154, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.09090909090909091, "amount": 0.16666666666666666, "status": 0.5, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 787}, "3": {"facts_total": 65, "facts_retained": 14, "retention_rate": 0.2153846153846154, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.09090909090909091, "amount": 0.16666666666666666, "status": 0.5, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 787}, "4": {"facts_total": 65, "facts_retained": 14, "retention_rate": 0.2153846153846154, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.09090909090909091, "amount": 0.16666666666666666, "status": 0.5, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 787}}}, "reference_key_id": "keys/synthetic/syn_03000_2000.json", "task_scores": {"correct": 1.0, "abstained": 1.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "f97ad5bade54fd65d2c8bafa5ade9463e557d5220c248210b1ca6dea042e54dc", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "f8e699a0187431b700d44c5d", "attempt_id": "a59936f966e2", "run_group_id": "E4_core:A5|closed|qwen3b:h8", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "d83aa60f20d2879d02898cb9d9d8bd1983ed2d8984cae017c9153cafee3a779a", "git_commit": null, "timestamp_utc": "2026-10-30T06:24:17.631307Z", "task_family": "synthetic", "document_id": "syn_03000_2000", "document_hash": "c9cc27c20b05b1f6e592846e84990ee9b69c42f9720a8e338ec2f6fca777fdbc", "question_id": "q00_single_fact", "split": "test", "cluster_id": "syn_03000_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "laptop-m5max-48g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 8, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1904, "query_tokens": 441, "message_tokens": 359, "message_utf8_bytes": 843, "cache_tensor_bytes": null, "serialized_payload_bytes": 843, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 135609.5230007777, "hops": {"1": 23029.02170899324, "2": 16046.37062503025, "3": 16191.568707930855, "4": 16016.233791946433, "5": 16240.511958952993, "6": 16246.597249992192, "7": 15958.910165936686, "8": 15880.308791995049}, "receiver_prefill_ms": 268.51008308585733, "ttft_ms": 61.59058294724673, "decode_ms": 62.20954202581197}, "total_wall_ms": 136002.43612576742, "ttft_ms": 61.59058294724673, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 14, "retention_rate": 0.2153846153846154, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.09090909090909091, "amount": 0.16666666666666666, "status": 0.5, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 787}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 15, "retention_rate": 0.23076923076923078, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.09090909090909091, "amount": 0.16666666666666666, "status": 0.5, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 855}, "2": {"facts_total": 65, "facts_retained": 14, "retention_rate": 0.2153846153846154, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.09090909090909091, "amount": 0.16666666666666666, "status": 0.5, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 787}, "3": {"facts_total": 65, "facts_retained": 14, "retention_rate": 0.2153846153846154, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.09090909090909091, "amount": 0.16666666666666666, "status": 0.5, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 787}, "4": {"facts_total": 65, "facts_retained": 14, "retention_rate": 0.2153846153846154, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.09090909090909091, "amount": 0.16666666666666666, "status": 0.5, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 787}, "5": {"facts_total": 65, "facts_retained": 14, "retention_rate": 0.2153846153846154, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.09090909090909091, "amount": 0.16666666666666666, "status": 0.5, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 787}, "6": {"facts_total": 65, "facts_retained": 14, "retention_rate": 0.2153846153846154, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.09090909090909091, "amount": 0.16666666666666666, "status": 0.5, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 787}, "7": {"facts_total": 65, "facts_retained": 14, "retention_rate": 0.2153846153846154, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.09090909090909091, "amount": 0.16666666666666666, "status": 0.5, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 787}, "8": {"facts_total": 65, "facts_retained": 14, "retention_rate": 0.2153846153846154, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.09090909090909091, "amount": 0.16666666666666666, "status": 0.5, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 787}}}, "reference_key_id": "keys/synthetic/syn_03000_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "f97ad5bade54fd65d2c8bafa5ade9463e557d5220c248210b1ca6dea042e54dc", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "6a953052582da4732eed803f", "attempt_id": "924cb6cbe897", "run_group_id": "E4_core:A5|closed|qwen3b:h8", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "d83aa60f20d2879d02898cb9d9d8bd1983ed2d8984cae017c9153cafee3a779a", "git_commit": null, "timestamp_utc": "2026-10-30T06:24:18.043300Z", "task_family": "synthetic", "document_id": "syn_03000_2000", "document_hash": "c9cc27c20b05b1f6e592846e84990ee9b69c42f9720a8e338ec2f6fca777fdbc", "question_id": "q01_single_fact", "split": "test", "cluster_id": "syn_03000_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "laptop-m5max-48g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 8, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1904, "query_tokens": 446, "message_tokens": 359, "message_utf8_bytes": 843, "cache_tensor_bytes": null, "serialized_payload_bytes": 843, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 135609.5230007777, "hops": {"1": 23029.02170899324, "2": 16046.37062503025, "3": 16191.568707930855, "4": 16016.233791946433, "5": 16240.511958952993, "6": 16246.597249992192, "7": 15958.910165936686, "8": 15880.308791995049}, "receiver_prefill_ms": 441.47345901001245, "ttft_ms": 24.053834029473364, "decode_ms": 42.428791988641024}, "total_wall_ms": 136118.03258385044, "ttft_ms": 24.053834029473364, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 14, "retention_rate": 0.2153846153846154, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.09090909090909091, "amount": 0.16666666666666666, "status": 0.5, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 787}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 15, "retention_rate": 0.23076923076923078, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.09090909090909091, "amount": 0.16666666666666666, "status": 0.5, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 855}, "2": {"facts_total": 65, "facts_retained": 14, "retention_rate": 0.2153846153846154, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.09090909090909091, "amount": 0.16666666666666666, "status": 0.5, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 787}, "3": {"facts_total": 65, "facts_retained": 14, "retention_rate": 0.2153846153846154, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.09090909090909091, "amount": 0.16666666666666666, "status": 0.5, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 787}, "4": {"facts_total": 65, "facts_retained": 14, "retention_rate": 0.2153846153846154, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.09090909090909091, "amount": 0.16666666666666666, "status": 0.5, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 787}, "5": {"facts_total": 65, "facts_retained": 14, "retention_rate": 0.2153846153846154, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.09090909090909091, "amount": 0.16666666666666666, "status": 0.5, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 787}, "6": {"facts_total": 65, "facts_retained": 14, "retention_rate": 0.2153846153846154, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.09090909090909091, "amount": 0.16666666666666666, "status": 0.5, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 787}, "7": {"facts_total": 65, "facts_retained": 14, "retention_rate": 0.2153846153846154, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.09090909090909091, "amount": 0.16666666666666666, "status": 0.5, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 787}, "8": {"facts_total": 65, "facts_retained": 14, "retention_rate": 0.2153846153846154, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.09090909090909091, "amount": 0.16666666666666666, "status": 0.5, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 787}}}, "reference_key_id": "keys/synthetic/syn_03000_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "f97ad5bade54fd65d2c8bafa5ade9463e557d5220c248210b1ca6dea042e54dc", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "6e3d38d20754713d9c864d8d", "attempt_id": "4780ef8bf85c", "run_group_id": "E4_core:A5|closed|qwen3b:h8", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "d83aa60f20d2879d02898cb9d9d8bd1983ed2d8984cae017c9153cafee3a779a", "git_commit": null, "timestamp_utc": "2026-10-30T06:24:18.568817Z", "task_family": "synthetic", "document_id": "syn_03000_2000", "document_hash": "c9cc27c20b05b1f6e592846e84990ee9b69c42f9720a8e338ec2f6fca777fdbc", "question_id": "q02_exact_copy", "split": "test", "cluster_id": "syn_03000_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "laptop-m5max-48g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 8, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1904, "query_tokens": 445, "message_tokens": 359, "message_utf8_bytes": 843, "cache_tensor_bytes": null, "serialized_payload_bytes": 843, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 135609.5230007777, "hops": {"1": 23029.02170899324, "2": 16046.37062503025, "3": 16191.568707930855, "4": 16016.233791946433, "5": 16240.511958952993, "6": 16246.597249992192, "7": 15958.910165936686, "8": 15880.308791995049}, "receiver_prefill_ms": 289.6134170005098, "ttft_ms": 38.599499966949224, "decode_ms": 43.08129195123911}, "total_wall_ms": 135981.41366778873, "ttft_ms": 38.599499966949224, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 14, "retention_rate": 0.2153846153846154, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.09090909090909091, "amount": 0.16666666666666666, "status": 0.5, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 787}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 15, "retention_rate": 0.23076923076923078, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.09090909090909091, "amount": 0.16666666666666666, "status": 0.5, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 855}, "2": {"facts_total": 65, "facts_retained": 14, "retention_rate": 0.2153846153846154, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.09090909090909091, "amount": 0.16666666666666666, "status": 0.5, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 787}, "3": {"facts_total": 65, "facts_retained": 14, "retention_rate": 0.2153846153846154, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.09090909090909091, "amount": 0.16666666666666666, "status": 0.5, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 787}, "4": {"facts_total": 65, "facts_retained": 14, "retention_rate": 0.2153846153846154, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.09090909090909091, "amount": 0.16666666666666666, "status": 0.5, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 787}, "5": {"facts_total": 65, "facts_retained": 14, "retention_rate": 0.2153846153846154, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.09090909090909091, "amount": 0.16666666666666666, "status": 0.5, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 787}, "6": {"facts_total": 65, "facts_retained": 14, "retention_rate": 0.2153846153846154, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.09090909090909091, "amount": 0.16666666666666666, "status": 0.5, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 787}, "7": {"facts_total": 65, "facts_retained": 14, "retention_rate": 0.2153846153846154, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.09090909090909091, "amount": 0.16666666666666666, "status": 0.5, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 787}, "8": {"facts_total": 65, "facts_retained": 14, "retention_rate": 0.2153846153846154, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.09090909090909091, "amount": 0.16666666666666666, "status": 0.5, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 787}}}, "reference_key_id": "keys/synthetic/syn_03000_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "f97ad5bade54fd65d2c8bafa5ade9463e557d5220c248210b1ca6dea042e54dc", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "b9e17d1aa3b359eacc7a4320", "attempt_id": "4149f3f70393", "run_group_id": "E4_core:A5|closed|qwen3b:h8", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "d83aa60f20d2879d02898cb9d9d8bd1983ed2d8984cae017c9153cafee3a779a", "git_commit": null, "timestamp_utc": "2026-10-30T06:24:18.957158Z", "task_family": "synthetic", "document_id": "syn_03000_2000", "document_hash": "c9cc27c20b05b1f6e592846e84990ee9b69c42f9720a8e338ec2f6fca777fdbc", "question_id": "q03_exact_copy", "split": "test", "cluster_id": "syn_03000_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "laptop-m5max-48g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 8, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1904, "query_tokens": 444, "message_tokens": 359, "message_utf8_bytes": 843, "cache_tensor_bytes": null, "serialized_payload_bytes": 843, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 135609.5230007777, "hops": {"1": 23029.02170899324, "2": 16046.37062503025, "3": 16191.568707930855, "4": 16016.233791946433, "5": 16240.511958952993, "6": 16246.597249992192, "7": 15958.910165936686, "8": 15880.308791995049}, "receiver_prefill_ms": 296.66704195551574, "ttft_ms": 23.522999952547252, "decode_ms": 23.37058400735259}, "total_wall_ms": 135953.5620007664, "ttft_ms": 23.522999952547252, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 14, "retention_rate": 0.2153846153846154, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.09090909090909091, "amount": 0.16666666666666666, "status": 0.5, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 787}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 15, "retention_rate": 0.23076923076923078, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.09090909090909091, "amount": 0.16666666666666666, "status": 0.5, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 855}, "2": {"facts_total": 65, "facts_retained": 14, "retention_rate": 0.2153846153846154, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.09090909090909091, "amount": 0.16666666666666666, "status": 0.5, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 787}, "3": {"facts_total": 65, "facts_retained": 14, "retention_rate": 0.2153846153846154, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.09090909090909091, "amount": 0.16666666666666666, "status": 0.5, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 787}, "4": {"facts_total": 65, "facts_retained": 14, "retention_rate": 0.2153846153846154, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.09090909090909091, "amount": 0.16666666666666666, "status": 0.5, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 787}, "5": {"facts_total": 65, "facts_retained": 14, "retention_rate": 0.2153846153846154, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.09090909090909091, "amount": 0.16666666666666666, "status": 0.5, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 787}, "6": {"facts_total": 65, "facts_retained": 14, "retention_rate": 0.2153846153846154, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.09090909090909091, "amount": 0.16666666666666666, "status": 0.5, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 787}, "7": {"facts_total": 65, "facts_retained": 14, "retention_rate": 0.2153846153846154, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.09090909090909091, "amount": 0.16666666666666666, "status": 0.5, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 787}, "8": {"facts_total": 65, "facts_retained": 14, "retention_rate": 0.2153846153846154, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.09090909090909091, "amount": 0.16666666666666666, "status": 0.5, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 787}}}, "reference_key_id": "keys/synthetic/syn_03000_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "f97ad5bade54fd65d2c8bafa5ade9463e557d5220c248210b1ca6dea042e54dc", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "8fc4d90a41b785c2fe8abe2a", "attempt_id": "36644be7a4d4", "run_group_id": "E4_core:A5|closed|qwen3b:h8", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "d83aa60f20d2879d02898cb9d9d8bd1983ed2d8984cae017c9153cafee3a779a", "git_commit": null, "timestamp_utc": "2026-10-30T06:24:19.316986Z", "task_family": "synthetic", "document_id": "syn_03000_2000", "document_hash": "c9cc27c20b05b1f6e592846e84990ee9b69c42f9720a8e338ec2f6fca777fdbc", "question_id": "q04_multi_hop", "split": "test", "cluster_id": "syn_03000_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "laptop-m5max-48g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 8, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1904, "query_tokens": 450, "message_tokens": 359, "message_utf8_bytes": 843, "cache_tensor_bytes": null, "serialized_payload_bytes": 843, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 135609.5230007777, "hops": {"1": 23029.02170899324, "2": 16046.37062503025, "3": 16191.568707930855, "4": 16016.233791946433, "5": 16240.511958952993, "6": 16246.597249992192, "7": 15958.910165936686, "8": 15880.308791995049}, "receiver_prefill_ms": 301.62549996748567, "ttft_ms": 61.42837496008724, "decode_ms": 74.20608401298523}, "total_wall_ms": 136047.34137572814, "ttft_ms": 61.42837496008724, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 14, "retention_rate": 0.2153846153846154, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.09090909090909091, "amount": 0.16666666666666666, "status": 0.5, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 787}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 15, "retention_rate": 0.23076923076923078, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.09090909090909091, "amount": 0.16666666666666666, "status": 0.5, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 855}, "2": {"facts_total": 65, "facts_retained": 14, "retention_rate": 0.2153846153846154, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.09090909090909091, "amount": 0.16666666666666666, "status": 0.5, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 787}, "3": {"facts_total": 65, "facts_retained": 14, "retention_rate": 0.2153846153846154, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.09090909090909091, "amount": 0.16666666666666666, "status": 0.5, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 787}, "4": {"facts_total": 65, "facts_retained": 14, "retention_rate": 0.2153846153846154, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.09090909090909091, "amount": 0.16666666666666666, "status": 0.5, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 787}, "5": {"facts_total": 65, "facts_retained": 14, "retention_rate": 0.2153846153846154, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.09090909090909091, "amount": 0.16666666666666666, "status": 0.5, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 787}, "6": {"facts_total": 65, "facts_retained": 14, "retention_rate": 0.2153846153846154, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.09090909090909091, "amount": 0.16666666666666666, "status": 0.5, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 787}, "7": {"facts_total": 65, "facts_retained": 14, "retention_rate": 0.2153846153846154, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.09090909090909091, "amount": 0.16666666666666666, "status": 0.5, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 787}, "8": {"facts_total": 65, "facts_retained": 14, "retention_rate": 0.2153846153846154, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.09090909090909091, "amount": 0.16666666666666666, "status": 0.5, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 787}}}, "reference_key_id": "keys/synthetic/syn_03000_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "f97ad5bade54fd65d2c8bafa5ade9463e557d5220c248210b1ca6dea042e54dc", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "0b405a533eb7a813a43c6a56", "attempt_id": "50d6218ed0a4", "run_group_id": "E4_core:A5|closed|qwen3b:h8", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "d83aa60f20d2879d02898cb9d9d8bd1983ed2d8984cae017c9153cafee3a779a", "git_commit": null, "timestamp_utc": "2026-10-30T06:24:19.773463Z", "task_family": "synthetic", "document_id": "syn_03000_2000", "document_hash": "c9cc27c20b05b1f6e592846e84990ee9b69c42f9720a8e338ec2f6fca777fdbc", "question_id": "q05_multi_hop", "split": "test", "cluster_id": "syn_03000_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "laptop-m5max-48g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 8, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1904, "query_tokens": 452, "message_tokens": 359, "message_utf8_bytes": 843, "cache_tensor_bytes": null, "serialized_payload_bytes": 843, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 135609.5230007777, "hops": {"1": 23029.02170899324, "2": 16046.37062503025, "3": 16191.568707930855, "4": 16016.233791946433, "5": 16240.511958952993, "6": 16246.597249992192, "7": 15958.910165936686, "8": 15880.308791995049}, "receiver_prefill_ms": 463.6094159213826, "ttft_ms": 29.293332947418094, "decode_ms": 363.3773330366239}, "total_wall_ms": 136466.31575073116, "ttft_ms": 29.293332947418094, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 10}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "LY-51E-3199", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 14, "retention_rate": 0.2153846153846154, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.09090909090909091, "amount": 0.16666666666666666, "status": 0.5, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 787}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 15, "retention_rate": 0.23076923076923078, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.09090909090909091, "amount": 0.16666666666666666, "status": 0.5, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 855}, "2": {"facts_total": 65, "facts_retained": 14, "retention_rate": 0.2153846153846154, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.09090909090909091, "amount": 0.16666666666666666, "status": 0.5, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 787}, "3": {"facts_total": 65, "facts_retained": 14, "retention_rate": 0.2153846153846154, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.09090909090909091, "amount": 0.16666666666666666, "status": 0.5, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 787}, "4": {"facts_total": 65, "facts_retained": 14, "retention_rate": 0.2153846153846154, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.09090909090909091, "amount": 0.16666666666666666, "status": 0.5, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 787}, "5": {"facts_total": 65, "facts_retained": 14, "retention_rate": 0.2153846153846154, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.09090909090909091, "amount": 0.16666666666666666, "status": 0.5, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 787}, "6": {"facts_total": 65, "facts_retained": 14, "retention_rate": 0.2153846153846154, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.09090909090909091, "amount": 0.16666666666666666, "status": 0.5, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 787}, "7": {"facts_total": 65, "facts_retained": 14, "retention_rate": 0.2153846153846154, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.09090909090909091, "amount": 0.16666666666666666, "status": 0.5, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 787}, "8": {"facts_total": 65, "facts_retained": 14, "retention_rate": 0.2153846153846154, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.09090909090909091, "amount": 0.16666666666666666, "status": 0.5, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 787}}}, "reference_key_id": "keys/synthetic/syn_03000_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "f97ad5bade54fd65d2c8bafa5ade9463e557d5220c248210b1ca6dea042e54dc", "output_hash": "a1245d1536890a7e4b026366df48441d503f491f782cfd388797b5782443c5c7"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "42a97bc39420ba18263dc237", "attempt_id": "e199cd6270b7", "run_group_id": "E4_core:A5|closed|qwen3b:h8", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "d83aa60f20d2879d02898cb9d9d8bd1983ed2d8984cae017c9153cafee3a779a", "git_commit": null, "timestamp_utc": "2026-10-30T06:24:20.648603Z", "task_family": "synthetic", "document_id": "syn_03000_2000", "document_hash": "c9cc27c20b05b1f6e592846e84990ee9b69c42f9720a8e338ec2f6fca777fdbc", "question_id": "q06_negation", "split": "test", "cluster_id": "syn_03000_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "laptop-m5max-48g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 8, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1904, "query_tokens": 441, "message_tokens": 359, "message_utf8_bytes": 843, "cache_tensor_bytes": null, "serialized_payload_bytes": 843, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 135609.5230007777, "hops": {"1": 23029.02170899324, "2": 16046.37062503025, "3": 16191.568707930855, "4": 16016.233791946433, "5": 16240.511958952993, "6": 16246.597249992192, "7": 15958.910165936686, "8": 15880.308791995049}, "receiver_prefill_ms": 271.55295899137855, "ttft_ms": 22.290791966952384, "decode_ms": 71.12641609273851}, "total_wall_ms": 135975.007917732, "ttft_ms": 22.290791966952384, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "no", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 14, "retention_rate": 0.2153846153846154, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.09090909090909091, "amount": 0.16666666666666666, "status": 0.5, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 787}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 15, "retention_rate": 0.23076923076923078, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.09090909090909091, "amount": 0.16666666666666666, "status": 0.5, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 855}, "2": {"facts_total": 65, "facts_retained": 14, "retention_rate": 0.2153846153846154, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.09090909090909091, "amount": 0.16666666666666666, "status": 0.5, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 787}, "3": {"facts_total": 65, "facts_retained": 14, "retention_rate": 0.2153846153846154, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.09090909090909091, "amount": 0.16666666666666666, "status": 0.5, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 787}, "4": {"facts_total": 65, "facts_retained": 14, "retention_rate": 0.2153846153846154, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.09090909090909091, "amount": 0.16666666666666666, "status": 0.5, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 787}, "5": {"facts_total": 65, "facts_retained": 14, "retention_rate": 0.2153846153846154, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.09090909090909091, "amount": 0.16666666666666666, "status": 0.5, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 787}, "6": {"facts_total": 65, "facts_retained": 14, "retention_rate": 0.2153846153846154, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.09090909090909091, "amount": 0.16666666666666666, "status": 0.5, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 787}, "7": {"facts_total": 65, "facts_retained": 14, "retention_rate": 0.2153846153846154, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.09090909090909091, "amount": 0.16666666666666666, "status": 0.5, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 787}, "8": {"facts_total": 65, "facts_retained": 14, "retention_rate": 0.2153846153846154, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.09090909090909091, "amount": 0.16666666666666666, "status": 0.5, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 787}}}, "reference_key_id": "keys/synthetic/syn_03000_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "f97ad5bade54fd65d2c8bafa5ade9463e557d5220c248210b1ca6dea042e54dc", "output_hash": "9390298f3fb0c5b160498935d79cb139aef28e1c47358b4bbba61862b9c26e59"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "3e85137399c4fc3828da776e", "attempt_id": "a70e031b04b2", "run_group_id": "E4_core:A5|closed|qwen3b:h8", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "d83aa60f20d2879d02898cb9d9d8bd1983ed2d8984cae017c9153cafee3a779a", "git_commit": null, "timestamp_utc": "2026-10-30T06:24:21.030625Z", "task_family": "synthetic", "document_id": "syn_03000_2000", "document_hash": "c9cc27c20b05b1f6e592846e84990ee9b69c42f9720a8e338ec2f6fca777fdbc", "question_id": "q07_negation", "split": "test", "cluster_id": "syn_03000_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "laptop-m5max-48g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 8, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1904, "query_tokens": 441, "message_tokens": 359, "message_utf8_bytes": 843, "cache_tensor_bytes": null, "serialized_payload_bytes": 843, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 135609.5230007777, "hops": {"1": 23029.02170899324, "2": 16046.37062503025, "3": 16191.568707930855, "4": 16016.233791946433, "5": 16240.511958952993, "6": 16246.597249992192, "7": 15958.910165936686, "8": 15880.308791995049}, "receiver_prefill_ms": 312.46333406306803, "ttft_ms": 111.30229104310274, "decode_ms": 136.93137501832098}, "total_wall_ms": 136170.76245881617, "ttft_ms": 111.30229104310274, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 14, "retention_rate": 0.2153846153846154, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.09090909090909091, "amount": 0.16666666666666666, "status": 0.5, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 787}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 15, "retention_rate": 0.23076923076923078, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.09090909090909091, "amount": 0.16666666666666666, "status": 0.5, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 855}, "2": {"facts_total": 65, "facts_retained": 14, "retention_rate": 0.2153846153846154, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.09090909090909091, "amount": 0.16666666666666666, "status": 0.5, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 787}, "3": {"facts_total": 65, "facts_retained": 14, "retention_rate": 0.2153846153846154, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.09090909090909091, "amount": 0.16666666666666666, "status": 0.5, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 787}, "4": {"facts_total": 65, "facts_retained": 14, "retention_rate": 0.2153846153846154, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.09090909090909091, "amount": 0.16666666666666666, "status": 0.5, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 787}, "5": {"facts_total": 65, "facts_retained": 14, "retention_rate": 0.2153846153846154, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.09090909090909091, "amount": 0.16666666666666666, "status": 0.5, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 787}, "6": {"facts_total": 65, "facts_retained": 14, "retention_rate": 0.2153846153846154, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.09090909090909091, "amount": 0.16666666666666666, "status": 0.5, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 787}, "7": {"facts_total": 65, "facts_retained": 14, "retention_rate": 0.2153846153846154, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.09090909090909091, "amount": 0.16666666666666666, "status": 0.5, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 787}, "8": {"facts_total": 65, "facts_retained": 14, "retention_rate": 0.2153846153846154, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.09090909090909091, "amount": 0.16666666666666666, "status": 0.5, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 787}}}, "reference_key_id": "keys/synthetic/syn_03000_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "f97ad5bade54fd65d2c8bafa5ade9463e557d5220c248210b1ca6dea042e54dc", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "ae6455de7251143afe1f7331", "attempt_id": "c0e689873e58", "run_group_id": "E4_core:A5|closed|qwen3b:h8", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "d83aa60f20d2879d02898cb9d9d8bd1983ed2d8984cae017c9153cafee3a779a", "git_commit": null, "timestamp_utc": "2026-10-30T06:24:21.609132Z", "task_family": "synthetic", "document_id": "syn_03000_2000", "document_hash": "c9cc27c20b05b1f6e592846e84990ee9b69c42f9720a8e338ec2f6fca777fdbc", "question_id": "q08_revision", "split": "test", "cluster_id": "syn_03000_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "laptop-m5max-48g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 8, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1904, "query_tokens": 456, "message_tokens": 359, "message_utf8_bytes": 843, "cache_tensor_bytes": null, "serialized_payload_bytes": 843, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 135609.5230007777, "hops": {"1": 23029.02170899324, "2": 16046.37062503025, "3": 16191.568707930855, "4": 16016.233791946433, "5": 16240.511958952993, "6": 16246.597249992192, "7": 15958.910165936686, "8": 15880.308791995049}, "receiver_prefill_ms": 313.8032080605626, "ttft_ms": 40.083458996377885, "decode_ms": 325.1834580441937}, "total_wall_ms": 136289.0494178282, "ttft_ms": 40.083458996377885, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 9}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "2,705,965", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 14, "retention_rate": 0.2153846153846154, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.09090909090909091, "amount": 0.16666666666666666, "status": 0.5, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 787}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 15, "retention_rate": 0.23076923076923078, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.09090909090909091, "amount": 0.16666666666666666, "status": 0.5, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 855}, "2": {"facts_total": 65, "facts_retained": 14, "retention_rate": 0.2153846153846154, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.09090909090909091, "amount": 0.16666666666666666, "status": 0.5, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 787}, "3": {"facts_total": 65, "facts_retained": 14, "retention_rate": 0.2153846153846154, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.09090909090909091, "amount": 0.16666666666666666, "status": 0.5, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 787}, "4": {"facts_total": 65, "facts_retained": 14, "retention_rate": 0.2153846153846154, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.09090909090909091, "amount": 0.16666666666666666, "status": 0.5, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 787}, "5": {"facts_total": 65, "facts_retained": 14, "retention_rate": 0.2153846153846154, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.09090909090909091, "amount": 0.16666666666666666, "status": 0.5, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 787}, "6": {"facts_total": 65, "facts_retained": 14, "retention_rate": 0.2153846153846154, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.09090909090909091, "amount": 0.16666666666666666, "status": 0.5, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 787}, "7": {"facts_total": 65, "facts_retained": 14, "retention_rate": 0.2153846153846154, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.09090909090909091, "amount": 0.16666666666666666, "status": 0.5, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 787}, "8": {"facts_total": 65, "facts_retained": 14, "retention_rate": 0.2153846153846154, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.09090909090909091, "amount": 0.16666666666666666, "status": 0.5, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 787}}}, "reference_key_id": "keys/synthetic/syn_03000_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "f97ad5bade54fd65d2c8bafa5ade9463e557d5220c248210b1ca6dea042e54dc", "output_hash": "dbedb6f5d00e413d915513b85d8ed195b18b9acf392ece342a85e15e3752bcc8"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "c5bf79c0b03dd8d01a7735a4", "attempt_id": "2e791ebeede3", "run_group_id": "E4_core:A5|closed|qwen3b:h8", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "d83aa60f20d2879d02898cb9d9d8bd1983ed2d8984cae017c9153cafee3a779a", "git_commit": null, "timestamp_utc": "2026-10-30T06:24:22.305136Z", "task_family": "synthetic", "document_id": "syn_03000_2000", "document_hash": "c9cc27c20b05b1f6e592846e84990ee9b69c42f9720a8e338ec2f6fca777fdbc", "question_id": "q09_revision", "split": "test", "cluster_id": "syn_03000_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "laptop-m5max-48g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 8, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1904, "query_tokens": 452, "message_tokens": 359, "message_utf8_bytes": 843, "cache_tensor_bytes": null, "serialized_payload_bytes": 843, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 135609.5230007777, "hops": {"1": 23029.02170899324, "2": 16046.37062503025, "3": 16191.568707930855, "4": 16016.233791946433, "5": 16240.511958952993, "6": 16246.597249992192, "7": 15958.910165936686, "8": 15880.308791995049}, "receiver_prefill_ms": 291.3120419252664, "ttft_ms": 57.82162491232157, "decode_ms": 77.52091600559652}, "total_wall_ms": 136036.69387567788, "ttft_ms": 57.82162491232157, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 14, "retention_rate": 0.2153846153846154, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.09090909090909091, "amount": 0.16666666666666666, "status": 0.5, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 787}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 15, "retention_rate": 0.23076923076923078, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.09090909090909091, "amount": 0.16666666666666666, "status": 0.5, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 855}, "2": {"facts_total": 65, "facts_retained": 14, "retention_rate": 0.2153846153846154, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.09090909090909091, "amount": 0.16666666666666666, "status": 0.5, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 787}, "3": {"facts_total": 65, "facts_retained": 14, "retention_rate": 0.2153846153846154, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.09090909090909091, "amount": 0.16666666666666666, "status": 0.5, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 787}, "4": {"facts_total": 65, "facts_retained": 14, "retention_rate": 0.2153846153846154, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.09090909090909091, "amount": 0.16666666666666666, "status": 0.5, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 787}, "5": {"facts_total": 65, "facts_retained": 14, "retention_rate": 0.2153846153846154, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.09090909090909091, "amount": 0.16666666666666666, "status": 0.5, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 787}, "6": {"facts_total": 65, "facts_retained": 14, "retention_rate": 0.2153846153846154, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.09090909090909091, "amount": 0.16666666666666666, "status": 0.5, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 787}, "7": {"facts_total": 65, "facts_retained": 14, "retention_rate": 0.2153846153846154, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.09090909090909091, "amount": 0.16666666666666666, "status": 0.5, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 787}, "8": {"facts_total": 65, "facts_retained": 14, "retention_rate": 0.2153846153846154, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.09090909090909091, "amount": 0.16666666666666666, "status": 0.5, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 787}}}, "reference_key_id": "keys/synthetic/syn_03000_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "f97ad5bade54fd65d2c8bafa5ade9463e557d5220c248210b1ca6dea042e54dc", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "b71b38f032123e9202338185", "attempt_id": "9ea3c0d15a5a", "run_group_id": "E4_core:A5|closed|qwen3b:h8", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "d83aa60f20d2879d02898cb9d9d8bd1983ed2d8984cae017c9153cafee3a779a", "git_commit": null, "timestamp_utc": "2026-10-30T06:24:22.748875Z", "task_family": "synthetic", "document_id": "syn_03000_2000", "document_hash": "c9cc27c20b05b1f6e592846e84990ee9b69c42f9720a8e338ec2f6fca777fdbc", "question_id": "q10_temporal", "split": "test", "cluster_id": "syn_03000_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "laptop-m5max-48g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 8, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1904, "query_tokens": 458, "message_tokens": 359, "message_utf8_bytes": 843, "cache_tensor_bytes": null, "serialized_payload_bytes": 843, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 135609.5230007777, "hops": {"1": 23029.02170899324, "2": 16046.37062503025, "3": 16191.568707930855, "4": 16016.233791946433, "5": 16240.511958952993, "6": 16246.597249992192, "7": 15958.910165936686, "8": 15880.308791995049}, "receiver_prefill_ms": 346.115457941778, "ttft_ms": 131.92237506154925, "decode_ms": 41.2073329789564}, "total_wall_ms": 136129.44950081874, "ttft_ms": 131.92237506154925, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 14, "retention_rate": 0.2153846153846154, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.09090909090909091, "amount": 0.16666666666666666, "status": 0.5, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 787}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 15, "retention_rate": 0.23076923076923078, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.09090909090909091, "amount": 0.16666666666666666, "status": 0.5, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 855}, "2": {"facts_total": 65, "facts_retained": 14, "retention_rate": 0.2153846153846154, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.09090909090909091, "amount": 0.16666666666666666, "status": 0.5, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 787}, "3": {"facts_total": 65, "facts_retained": 14, "retention_rate": 0.2153846153846154, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.09090909090909091, "amount": 0.16666666666666666, "status": 0.5, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 787}, "4": {"facts_total": 65, "facts_retained": 14, "retention_rate": 0.2153846153846154, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.09090909090909091, "amount": 0.16666666666666666, "status": 0.5, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 787}, "5": {"facts_total": 65, "facts_retained": 14, "retention_rate": 0.2153846153846154, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.09090909090909091, "amount": 0.16666666666666666, "status": 0.5, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 787}, "6": {"facts_total": 65, "facts_retained": 14, "retention_rate": 0.2153846153846154, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.09090909090909091, "amount": 0.16666666666666666, "status": 0.5, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 787}, "7": {"facts_total": 65, "facts_retained": 14, "retention_rate": 0.2153846153846154, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.09090909090909091, "amount": 0.16666666666666666, "status": 0.5, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 787}, "8": {"facts_total": 65, "facts_retained": 14, "retention_rate": 0.2153846153846154, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.09090909090909091, "amount": 0.16666666666666666, "status": 0.5, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 787}}}, "reference_key_id": "keys/synthetic/syn_03000_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "f97ad5bade54fd65d2c8bafa5ade9463e557d5220c248210b1ca6dea042e54dc", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "2c956258f44f19249a90b0aa", "attempt_id": "66d3a085f650", "run_group_id": "E4_core:A5|closed|qwen3b:h8", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "d83aa60f20d2879d02898cb9d9d8bd1983ed2d8984cae017c9153cafee3a779a", "git_commit": null, "timestamp_utc": "2026-10-30T06:24:23.287525Z", "task_family": "synthetic", "document_id": "syn_03000_2000", "document_hash": "c9cc27c20b05b1f6e592846e84990ee9b69c42f9720a8e338ec2f6fca777fdbc", "question_id": "q11_aggregation", "split": "test", "cluster_id": "syn_03000_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "laptop-m5max-48g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 8, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1904, "query_tokens": 453, "message_tokens": 359, "message_utf8_bytes": 843, "cache_tensor_bytes": null, "serialized_payload_bytes": 843, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 135609.5230007777, "hops": {"1": 23029.02170899324, "2": 16046.37062503025, "3": 16191.568707930855, "4": 16016.233791946433, "5": 16240.511958952993, "6": 16246.597249992192, "7": 15958.910165936686, "8": 15880.308791995049}, "receiver_prefill_ms": 331.6220829728991, "ttft_ms": 36.513666971586645, "decode_ms": 26.54754207469523}, "total_wall_ms": 136004.7234587837, "ttft_ms": 36.513666971586645, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 14, "retention_rate": 0.2153846153846154, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.09090909090909091, "amount": 0.16666666666666666, "status": 0.5, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 787}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 15, "retention_rate": 0.23076923076923078, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.09090909090909091, "amount": 0.16666666666666666, "status": 0.5, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 855}, "2": {"facts_total": 65, "facts_retained": 14, "retention_rate": 0.2153846153846154, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.09090909090909091, "amount": 0.16666666666666666, "status": 0.5, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 787}, "3": {"facts_total": 65, "facts_retained": 14, "retention_rate": 0.2153846153846154, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.09090909090909091, "amount": 0.16666666666666666, "status": 0.5, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 787}, "4": {"facts_total": 65, "facts_retained": 14, "retention_rate": 0.2153846153846154, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.09090909090909091, "amount": 0.16666666666666666, "status": 0.5, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 787}, "5": {"facts_total": 65, "facts_retained": 14, "retention_rate": 0.2153846153846154, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.09090909090909091, "amount": 0.16666666666666666, "status": 0.5, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 787}, "6": {"facts_total": 65, "facts_retained": 14, "retention_rate": 0.2153846153846154, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.09090909090909091, "amount": 0.16666666666666666, "status": 0.5, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 787}, "7": {"facts_total": 65, "facts_retained": 14, "retention_rate": 0.2153846153846154, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.09090909090909091, "amount": 0.16666666666666666, "status": 0.5, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 787}, "8": {"facts_total": 65, "facts_retained": 14, "retention_rate": 0.2153846153846154, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.09090909090909091, "amount": 0.16666666666666666, "status": 0.5, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 787}}}, "reference_key_id": "keys/synthetic/syn_03000_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "f97ad5bade54fd65d2c8bafa5ade9463e557d5220c248210b1ca6dea042e54dc", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "7d322dba55d5b629d492e723", "attempt_id": "f5fd8501fded", "run_group_id": "E4_core:A5|closed|qwen3b:h8", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "d83aa60f20d2879d02898cb9d9d8bd1983ed2d8984cae017c9153cafee3a779a", "git_commit": null, "timestamp_utc": "2026-10-30T06:24:23.699062Z", "task_family": "synthetic", "document_id": "syn_03000_2000", "document_hash": "c9cc27c20b05b1f6e592846e84990ee9b69c42f9720a8e338ec2f6fca777fdbc", "question_id": "q12_unit_conversion", "split": "test", "cluster_id": "syn_03000_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "laptop-m5max-48g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 8, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1904, "query_tokens": 449, "message_tokens": 359, "message_utf8_bytes": 843, "cache_tensor_bytes": null, "serialized_payload_bytes": 843, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 135609.5230007777, "hops": {"1": 23029.02170899324, "2": 16046.37062503025, "3": 16191.568707930855, "4": 16016.233791946433, "5": 16240.511958952993, "6": 16246.597249992192, "7": 15958.910165936686, "8": 15880.308791995049}, "receiver_prefill_ms": 261.24095800332725, "ttft_ms": 40.685915970243514, "decode_ms": 153.67520798463374}, "total_wall_ms": 136065.89362572413, "ttft_ms": 40.685915970243514, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 5}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "20.46", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 14, "retention_rate": 0.2153846153846154, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.09090909090909091, "amount": 0.16666666666666666, "status": 0.5, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 787}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 15, "retention_rate": 0.23076923076923078, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.09090909090909091, "amount": 0.16666666666666666, "status": 0.5, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 855}, "2": {"facts_total": 65, "facts_retained": 14, "retention_rate": 0.2153846153846154, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.09090909090909091, "amount": 0.16666666666666666, "status": 0.5, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 787}, "3": {"facts_total": 65, "facts_retained": 14, "retention_rate": 0.2153846153846154, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.09090909090909091, "amount": 0.16666666666666666, "status": 0.5, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 787}, "4": {"facts_total": 65, "facts_retained": 14, "retention_rate": 0.2153846153846154, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.09090909090909091, "amount": 0.16666666666666666, "status": 0.5, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 787}, "5": {"facts_total": 65, "facts_retained": 14, "retention_rate": 0.2153846153846154, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.09090909090909091, "amount": 0.16666666666666666, "status": 0.5, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 787}, "6": {"facts_total": 65, "facts_retained": 14, "retention_rate": 0.2153846153846154, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.09090909090909091, "amount": 0.16666666666666666, "status": 0.5, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 787}, "7": {"facts_total": 65, "facts_retained": 14, "retention_rate": 0.2153846153846154, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.09090909090909091, "amount": 0.16666666666666666, "status": 0.5, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 787}, "8": {"facts_total": 65, "facts_retained": 14, "retention_rate": 0.2153846153846154, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.09090909090909091, "amount": 0.16666666666666666, "status": 0.5, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 787}}}, "reference_key_id": "keys/synthetic/syn_03000_2000.json", "task_scores": {"correct": 0.0, "abstained": 0.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "f97ad5bade54fd65d2c8bafa5ade9463e557d5220c248210b1ca6dea042e54dc", "output_hash": "c057e11a5e818ca86a03c336373f3b4e4a2ddd9756e0acd7f5d7d655fd0267fe"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "8b501bd8be68fb570a710d6b", "attempt_id": "885c6bc90c96", "run_group_id": "E4_core:A5|closed|qwen3b:h8", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "d83aa60f20d2879d02898cb9d9d8bd1983ed2d8984cae017c9153cafee3a779a", "git_commit": null, "timestamp_utc": "2026-10-30T06:24:24.172637Z", "task_family": "synthetic", "document_id": "syn_03000_2000", "document_hash": "c9cc27c20b05b1f6e592846e84990ee9b69c42f9720a8e338ec2f6fca777fdbc", "question_id": "q13_unanswerable", "split": "test", "cluster_id": "syn_03000_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "laptop-m5max-48g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 8, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1904, "query_tokens": 453, "message_tokens": 359, "message_utf8_bytes": 843, "cache_tensor_bytes": null, "serialized_payload_bytes": 843, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 135609.5230007777, "hops": {"1": 23029.02170899324, "2": 16046.37062503025, "3": 16191.568707930855, "4": 16016.233791946433, "5": 16240.511958952993, "6": 16246.597249992192, "7": 15958.910165936686, "8": 15880.308791995049}, "receiver_prefill_ms": 346.48579196073115, "ttft_ms": 58.29916591756046, "decode_ms": 45.02712492831051}, "total_wall_ms": 136059.88645879552, "ttft_ms": 58.29916591756046, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 14, "retention_rate": 0.2153846153846154, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.09090909090909091, "amount": 0.16666666666666666, "status": 0.5, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 787}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 15, "retention_rate": 0.23076923076923078, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.09090909090909091, "amount": 0.16666666666666666, "status": 0.5, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 855}, "2": {"facts_total": 65, "facts_retained": 14, "retention_rate": 0.2153846153846154, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.09090909090909091, "amount": 0.16666666666666666, "status": 0.5, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 787}, "3": {"facts_total": 65, "facts_retained": 14, "retention_rate": 0.2153846153846154, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.09090909090909091, "amount": 0.16666666666666666, "status": 0.5, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 787}, "4": {"facts_total": 65, "facts_retained": 14, "retention_rate": 0.2153846153846154, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.09090909090909091, "amount": 0.16666666666666666, "status": 0.5, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 787}, "5": {"facts_total": 65, "facts_retained": 14, "retention_rate": 0.2153846153846154, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.09090909090909091, "amount": 0.16666666666666666, "status": 0.5, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 787}, "6": {"facts_total": 65, "facts_retained": 14, "retention_rate": 0.2153846153846154, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.09090909090909091, "amount": 0.16666666666666666, "status": 0.5, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 787}, "7": {"facts_total": 65, "facts_retained": 14, "retention_rate": 0.2153846153846154, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.09090909090909091, "amount": 0.16666666666666666, "status": 0.5, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 787}, "8": {"facts_total": 65, "facts_retained": 14, "retention_rate": 0.2153846153846154, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.09090909090909091, "amount": 0.16666666666666666, "status": 0.5, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 787}}}, "reference_key_id": "keys/synthetic/syn_03000_2000.json", "task_scores": {"correct": 1.0, "abstained": 1.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "f97ad5bade54fd65d2c8bafa5ade9463e557d5220c248210b1ca6dea042e54dc", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "2c90034e75a8f7050baf89f9", "attempt_id": "54a442da1bdd", "run_group_id": "E4_core:A5|closed|qwen3b:h8", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "d83aa60f20d2879d02898cb9d9d8bd1983ed2d8984cae017c9153cafee3a779a", "git_commit": null, "timestamp_utc": "2026-10-30T06:24:24.641817Z", "task_family": "synthetic", "document_id": "syn_03000_2000", "document_hash": "c9cc27c20b05b1f6e592846e84990ee9b69c42f9720a8e338ec2f6fca777fdbc", "question_id": "q14_unanswerable", "split": "test", "cluster_id": "syn_03000_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "laptop-m5max-48g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 8, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1904, "query_tokens": 455, "message_tokens": 359, "message_utf8_bytes": 843, "cache_tensor_bytes": null, "serialized_payload_bytes": 843, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 135609.5230007777, "hops": {"1": 23029.02170899324, "2": 16046.37062503025, "3": 16191.568707930855, "4": 16016.233791946433, "5": 16240.511958952993, "6": 16246.597249992192, "7": 15958.910165936686, "8": 15880.308791995049}, "receiver_prefill_ms": 439.15033305529505, "ttft_ms": 37.53954207058996, "decode_ms": 44.2579590016976}, "total_wall_ms": 136131.02050079033, "ttft_ms": 37.53954207058996, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 14, "retention_rate": 0.2153846153846154, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.09090909090909091, "amount": 0.16666666666666666, "status": 0.5, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 787}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 15, "retention_rate": 0.23076923076923078, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.09090909090909091, "amount": 0.16666666666666666, "status": 0.5, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 855}, "2": {"facts_total": 65, "facts_retained": 14, "retention_rate": 0.2153846153846154, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.09090909090909091, "amount": 0.16666666666666666, "status": 0.5, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 787}, "3": {"facts_total": 65, "facts_retained": 14, "retention_rate": 0.2153846153846154, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.09090909090909091, "amount": 0.16666666666666666, "status": 0.5, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 787}, "4": {"facts_total": 65, "facts_retained": 14, "retention_rate": 0.2153846153846154, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.09090909090909091, "amount": 0.16666666666666666, "status": 0.5, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 787}, "5": {"facts_total": 65, "facts_retained": 14, "retention_rate": 0.2153846153846154, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.09090909090909091, "amount": 0.16666666666666666, "status": 0.5, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 787}, "6": {"facts_total": 65, "facts_retained": 14, "retention_rate": 0.2153846153846154, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.09090909090909091, "amount": 0.16666666666666666, "status": 0.5, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 787}, "7": {"facts_total": 65, "facts_retained": 14, "retention_rate": 0.2153846153846154, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.09090909090909091, "amount": 0.16666666666666666, "status": 0.5, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 787}, "8": {"facts_total": 65, "facts_retained": 14, "retention_rate": 0.2153846153846154, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.09090909090909091, "amount": 0.16666666666666666, "status": 0.5, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 787}}}, "reference_key_id": "keys/synthetic/syn_03000_2000.json", "task_scores": {"correct": 1.0, "abstained": 1.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "f97ad5bade54fd65d2c8bafa5ade9463e557d5220c248210b1ca6dea042e54dc", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "31b99c630be4a79d66bbd537", "attempt_id": "45aed9f06b72", "run_group_id": "E4_core:A4|repairable|qwen3b:h1", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "d83aa60f20d2879d02898cb9d9d8bd1983ed2d8984cae017c9153cafee3a779a", "git_commit": null, "timestamp_utc": "2026-10-30T06:24:47.961286Z", "task_family": "synthetic", "document_id": "syn_03000_2000", "document_hash": "c9cc27c20b05b1f6e592846e84990ee9b69c42f9720a8e338ec2f6fca777fdbc", "question_id": "q00_single_fact", "split": "test", "cluster_id": "syn_03000_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "laptop-m5max-48g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 1, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1904, "query_tokens": 592, "message_tokens": 511, "message_utf8_bytes": 1090, "cache_tensor_bytes": null, "serialized_payload_bytes": 1090, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 22777.715417090803, "hops": {"1": 22777.715417090803}, "receiver_prefill_ms": 452.5007080519572, "ttft_ms": 79.98529099859297, "decode_ms": 113.00137499347329}, "total_wall_ms": 23423.81254211068, "ttft_ms": 79.98529099859297, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 29, "retention_rate": 0.4461538461538462, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5, "status": 1.0, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1090}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 29, "retention_rate": 0.4461538461538462, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5, "status": 1.0, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1090}}}, "reference_key_id": "keys/synthetic/syn_03000_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "562010df7e3ceb9bcd9a843b326413164596adc2f7ca548f1a2eb3ede9dfbb44", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "7f2edcb64ddad749ccb8df94", "attempt_id": "70006bda2e3a", "run_group_id": "E4_core:A4|repairable|qwen3b:h1", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "d83aa60f20d2879d02898cb9d9d8bd1983ed2d8984cae017c9153cafee3a779a", "git_commit": null, "timestamp_utc": "2026-10-30T06:24:48.626052Z", "task_family": "synthetic", "document_id": "syn_03000_2000", "document_hash": "c9cc27c20b05b1f6e592846e84990ee9b69c42f9720a8e338ec2f6fca777fdbc", "question_id": "q01_single_fact", "split": "test", "cluster_id": "syn_03000_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "laptop-m5max-48g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 1, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1904, "query_tokens": 597, "message_tokens": 511, "message_utf8_bytes": 1090, "cache_tensor_bytes": null, "serialized_payload_bytes": 1090, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 22777.715417090803, "hops": {"1": 22777.715417090803}, "receiver_prefill_ms": 399.36129201669246, "ttft_ms": 39.53412501141429, "decode_ms": 307.03808297403157}, "total_wall_ms": 23524.238208075985, "ttft_ms": 39.53412501141429, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 9}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "Lot 19-N, Gaspé", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 29, "retention_rate": 0.4461538461538462, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5, "status": 1.0, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1090}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 29, "retention_rate": 0.4461538461538462, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5, "status": 1.0, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1090}}}, "reference_key_id": "keys/synthetic/syn_03000_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 0.8571428571428571}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "562010df7e3ceb9bcd9a843b326413164596adc2f7ca548f1a2eb3ede9dfbb44", "output_hash": "2e8cf782bd118a1c6d4e23417785d72fb92e930806b5752cd1ec405a0c9b95e5"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "14d984e305884db38206fa42", "attempt_id": "4fbb52c89d6e", "run_group_id": "E4_core:A4|repairable|qwen3b:h1", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "d83aa60f20d2879d02898cb9d9d8bd1983ed2d8984cae017c9153cafee3a779a", "git_commit": null, "timestamp_utc": "2026-10-30T06:24:49.389318Z", "task_family": "synthetic", "document_id": "syn_03000_2000", "document_hash": "c9cc27c20b05b1f6e592846e84990ee9b69c42f9720a8e338ec2f6fca777fdbc", "question_id": "q02_exact_copy", "split": "test", "cluster_id": "syn_03000_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "laptop-m5max-48g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 1, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1904, "query_tokens": 596, "message_tokens": 511, "message_utf8_bytes": 1090, "cache_tensor_bytes": null, "serialized_payload_bytes": 1090, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 22777.715417090803, "hops": {"1": 22777.715417090803}, "receiver_prefill_ms": 372.5291659357026, "ttft_ms": 46.63662507664412, "decode_ms": 49.45862491149455}, "total_wall_ms": 23247.016625129618, "ttft_ms": 46.63662507664412, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 29, "retention_rate": 0.4461538461538462, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5, "status": 1.0, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1090}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 29, "retention_rate": 0.4461538461538462, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5, "status": 1.0, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1090}}}, "reference_key_id": "keys/synthetic/syn_03000_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "562010df7e3ceb9bcd9a843b326413164596adc2f7ca548f1a2eb3ede9dfbb44", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "7de0196c96e7c38387f3a121", "attempt_id": "b2e79a61d1e1", "run_group_id": "E4_core:A4|repairable|qwen3b:h1", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "d83aa60f20d2879d02898cb9d9d8bd1983ed2d8984cae017c9153cafee3a779a", "git_commit": null, "timestamp_utc": "2026-10-30T06:24:49.875488Z", "task_family": "synthetic", "document_id": "syn_03000_2000", "document_hash": "c9cc27c20b05b1f6e592846e84990ee9b69c42f9720a8e338ec2f6fca777fdbc", "question_id": "q03_exact_copy", "split": "test", "cluster_id": "syn_03000_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "laptop-m5max-48g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 1, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1904, "query_tokens": 595, "message_tokens": 511, "message_utf8_bytes": 1090, "cache_tensor_bytes": null, "serialized_payload_bytes": 1090, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 22777.715417090803, "hops": {"1": 22777.715417090803}, "receiver_prefill_ms": 510.7871249783784, "ttft_ms": 39.54858402721584, "decode_ms": 214.79583298787475}, "total_wall_ms": 23543.622000142932, "ttft_ms": 39.54858402721584, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 7}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "Kwame Haldorsen", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 29, "retention_rate": 0.4461538461538462, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5, "status": 1.0, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1090}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 29, "retention_rate": 0.4461538461538462, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5, "status": 1.0, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1090}}}, "reference_key_id": "keys/synthetic/syn_03000_2000.json", "task_scores": {"correct": 0.0, "abstained": 0.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "562010df7e3ceb9bcd9a843b326413164596adc2f7ca548f1a2eb3ede9dfbb44", "output_hash": "175253e7cf463934ec4d54e361a781f323cd87290f01306ed368933bcbd2e25d"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "ffd0153241236178c953bbf0", "attempt_id": "0f4539e35cec", "run_group_id": "E4_core:A4|repairable|qwen3b:h1", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "d83aa60f20d2879d02898cb9d9d8bd1983ed2d8984cae017c9153cafee3a779a", "git_commit": null, "timestamp_utc": "2026-10-30T06:24:50.658422Z", "task_family": "synthetic", "document_id": "syn_03000_2000", "document_hash": "c9cc27c20b05b1f6e592846e84990ee9b69c42f9720a8e338ec2f6fca777fdbc", "question_id": "q04_multi_hop", "split": "test", "cluster_id": "syn_03000_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "laptop-m5max-48g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 1, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1904, "query_tokens": 601, "message_tokens": 511, "message_utf8_bytes": 1090, "cache_tensor_bytes": null, "serialized_payload_bytes": 1090, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 22777.715417090803, "hops": {"1": 22777.715417090803}, "receiver_prefill_ms": 386.73612498678267, "ttft_ms": 27.113625081256032, "decode_ms": 304.89637493155897}, "total_wall_ms": 23497.477709082887, "ttft_ms": 27.113625081256032, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 9}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "9,220,745", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 29, "retention_rate": 0.4461538461538462, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5, "status": 1.0, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1090}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 29, "retention_rate": 0.4461538461538462, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5, "status": 1.0, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1090}}}, "reference_key_id": "keys/synthetic/syn_03000_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "562010df7e3ceb9bcd9a843b326413164596adc2f7ca548f1a2eb3ede9dfbb44", "output_hash": "ae55ba88b2cb1b0f1a4e0960ae312e0a1c177c79a6d14b67bff1677cec19b649"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "85b8d31e6cd200dd4241e4a2", "attempt_id": "ad3f02c18cbb", "run_group_id": "E4_core:A4|repairable|qwen3b:h1", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "d83aa60f20d2879d02898cb9d9d8bd1983ed2d8984cae017c9153cafee3a779a", "git_commit": null, "timestamp_utc": "2026-10-30T06:24:51.396065Z", "task_family": "synthetic", "document_id": "syn_03000_2000", "document_hash": "c9cc27c20b05b1f6e592846e84990ee9b69c42f9720a8e338ec2f6fca777fdbc", "question_id": "q05_multi_hop", "split": "test", "cluster_id": "syn_03000_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "laptop-m5max-48g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 1, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1904, "query_tokens": 603, "message_tokens": 511, "message_utf8_bytes": 1090, "cache_tensor_bytes": null, "serialized_payload_bytes": 1090, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 22777.715417090803, "hops": {"1": 22777.715417090803}, "receiver_prefill_ms": 412.96133399009705, "ttft_ms": 62.02008295804262, "decode_ms": 400.7444999879226}, "total_wall_ms": 23654.150376096368, "ttft_ms": 62.02008295804262, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 10}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "LY-51E-3199", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 29, "retention_rate": 0.4461538461538462, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5, "status": 1.0, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1090}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 29, "retention_rate": 0.4461538461538462, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5, "status": 1.0, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1090}}}, "reference_key_id": "keys/synthetic/syn_03000_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "562010df7e3ceb9bcd9a843b326413164596adc2f7ca548f1a2eb3ede9dfbb44", "output_hash": "a1245d1536890a7e4b026366df48441d503f491f782cfd388797b5782443c5c7"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "9d4c9569cd8b5934688225fc", "attempt_id": "045a54f58894", "run_group_id": "E4_core:A4|repairable|qwen3b:h1", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "d83aa60f20d2879d02898cb9d9d8bd1983ed2d8984cae017c9153cafee3a779a", "git_commit": null, "timestamp_utc": "2026-10-30T06:24:52.288948Z", "task_family": "synthetic", "document_id": "syn_03000_2000", "document_hash": "c9cc27c20b05b1f6e592846e84990ee9b69c42f9720a8e338ec2f6fca777fdbc", "question_id": "q06_negation", "split": "test", "cluster_id": "syn_03000_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "laptop-m5max-48g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 1, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1904, "query_tokens": 592, "message_tokens": 511, "message_utf8_bytes": 1090, "cache_tensor_bytes": null, "serialized_payload_bytes": 1090, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 22777.715417090803, "hops": {"1": 22777.715417090803}, "receiver_prefill_ms": 399.3773750262335, "ttft_ms": 23.716374998912215, "decode_ms": 40.364707936532795}, "total_wall_ms": 23241.87075102236, "ttft_ms": 23.716374998912215, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "no", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 29, "retention_rate": 0.4461538461538462, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5, "status": 1.0, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1090}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 29, "retention_rate": 0.4461538461538462, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5, "status": 1.0, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1090}}}, "reference_key_id": "keys/synthetic/syn_03000_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "562010df7e3ceb9bcd9a843b326413164596adc2f7ca548f1a2eb3ede9dfbb44", "output_hash": "9390298f3fb0c5b160498935d79cb139aef28e1c47358b4bbba61862b9c26e59"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "d82e88dbb2d4f138d6005106", "attempt_id": "136c26f2dc13", "run_group_id": "E4_core:A4|repairable|qwen3b:h1", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "d83aa60f20d2879d02898cb9d9d8bd1983ed2d8984cae017c9153cafee3a779a", "git_commit": null, "timestamp_utc": "2026-10-30T06:24:52.770051Z", "task_family": "synthetic", "document_id": "syn_03000_2000", "document_hash": "c9cc27c20b05b1f6e592846e84990ee9b69c42f9720a8e338ec2f6fca777fdbc", "question_id": "q07_negation", "split": "test", "cluster_id": "syn_03000_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "laptop-m5max-48g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 1, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1904, "query_tokens": 592, "message_tokens": 511, "message_utf8_bytes": 1090, "cache_tensor_bytes": null, "serialized_payload_bytes": 1090, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 22777.715417090803, "hops": {"1": 22777.715417090803}, "receiver_prefill_ms": 352.2431660676375, "ttft_ms": 47.272250056266785, "decode_ms": 64.9741250090301}, "total_wall_ms": 23242.882626131177, "ttft_ms": 47.272250056266785, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "no", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 29, "retention_rate": 0.4461538461538462, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5, "status": 1.0, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1090}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 29, "retention_rate": 0.4461538461538462, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5, "status": 1.0, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1090}}}, "reference_key_id": "keys/synthetic/syn_03000_2000.json", "task_scores": {"correct": 0.0, "abstained": 0.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "562010df7e3ceb9bcd9a843b326413164596adc2f7ca548f1a2eb3ede9dfbb44", "output_hash": "9390298f3fb0c5b160498935d79cb139aef28e1c47358b4bbba61862b9c26e59"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "f4def938873c40ade4701f3b", "attempt_id": "3afafbac4f35", "run_group_id": "E4_core:A4|repairable|qwen3b:h1", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "d83aa60f20d2879d02898cb9d9d8bd1983ed2d8984cae017c9153cafee3a779a", "git_commit": null, "timestamp_utc": "2026-10-30T06:24:53.251298Z", "task_family": "synthetic", "document_id": "syn_03000_2000", "document_hash": "c9cc27c20b05b1f6e592846e84990ee9b69c42f9720a8e338ec2f6fca777fdbc", "question_id": "q08_revision", "split": "test", "cluster_id": "syn_03000_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "laptop-m5max-48g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 1, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1904, "query_tokens": 607, "message_tokens": 511, "message_utf8_bytes": 1090, "cache_tensor_bytes": null, "serialized_payload_bytes": 1090, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 22777.715417090803, "hops": {"1": 22777.715417090803}, "receiver_prefill_ms": 549.0838750265539, "ttft_ms": 36.88374999910593, "decode_ms": 287.6435419311747}, "total_wall_ms": 23652.019000146538, "ttft_ms": 36.88374999910593, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 9}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "2,705,965", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 29, "retention_rate": 0.4461538461538462, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5, "status": 1.0, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1090}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 29, "retention_rate": 0.4461538461538462, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5, "status": 1.0, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1090}}}, "reference_key_id": "keys/synthetic/syn_03000_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "562010df7e3ceb9bcd9a843b326413164596adc2f7ca548f1a2eb3ede9dfbb44", "output_hash": "dbedb6f5d00e413d915513b85d8ed195b18b9acf392ece342a85e15e3752bcc8"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "570fa886d789e1e68d82705e", "attempt_id": "864968e1b6a4", "run_group_id": "E4_core:A4|repairable|qwen3b:h1", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "d83aa60f20d2879d02898cb9d9d8bd1983ed2d8984cae017c9153cafee3a779a", "git_commit": null, "timestamp_utc": "2026-10-30T06:24:54.143228Z", "task_family": "synthetic", "document_id": "syn_03000_2000", "document_hash": "c9cc27c20b05b1f6e592846e84990ee9b69c42f9720a8e338ec2f6fca777fdbc", "question_id": "q09_revision", "split": "test", "cluster_id": "syn_03000_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "laptop-m5max-48g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 1, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1904, "query_tokens": 603, "message_tokens": 511, "message_utf8_bytes": 1090, "cache_tensor_bytes": null, "serialized_payload_bytes": 1090, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 22777.715417090803, "hops": {"1": 22777.715417090803}, "receiver_prefill_ms": 402.2530000656843, "ttft_ms": 24.703333037905395, "decode_ms": 371.5175000252202}, "total_wall_ms": 23577.123125083745, "ttft_ms": 24.703333037905395, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 7}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "9127965", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 29, "retention_rate": 0.4461538461538462, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5, "status": 1.0, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1090}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 29, "retention_rate": 0.4461538461538462, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5, "status": 1.0, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1090}}}, "reference_key_id": "keys/synthetic/syn_03000_2000.json", "task_scores": {"correct": 0.0, "abstained": 0.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "562010df7e3ceb9bcd9a843b326413164596adc2f7ca548f1a2eb3ede9dfbb44", "output_hash": "a0c871941224304cec47bb1b3d64b8706843b299f14ab94899d97df14f433da7"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "1283348ebd0213a4c781fa68", "attempt_id": "38624a5e2521", "run_group_id": "E4_core:A4|repairable|qwen3b:h1", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "d83aa60f20d2879d02898cb9d9d8bd1983ed2d8984cae017c9153cafee3a779a", "git_commit": null, "timestamp_utc": "2026-10-30T06:24:54.959820Z", "task_family": "synthetic", "document_id": "syn_03000_2000", "document_hash": "c9cc27c20b05b1f6e592846e84990ee9b69c42f9720a8e338ec2f6fca777fdbc", "question_id": "q10_temporal", "split": "test", "cluster_id": "syn_03000_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "laptop-m5max-48g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 1, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1904, "query_tokens": 609, "message_tokens": 511, "message_utf8_bytes": 1090, "cache_tensor_bytes": null, "serialized_payload_bytes": 1090, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 22777.715417090803, "hops": {"1": 22777.715417090803}, "receiver_prefill_ms": 530.646999948658, "ttft_ms": 36.20366705581546, "decode_ms": 171.96349997539073}, "total_wall_ms": 23517.28950010147, "ttft_ms": 36.20366705581546, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 5}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "Felaldnak-7", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 29, "retention_rate": 0.4461538461538462, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5, "status": 1.0, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1090}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 29, "retention_rate": 0.4461538461538462, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5, "status": 1.0, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1090}}}, "reference_key_id": "keys/synthetic/syn_03000_2000.json", "task_scores": {"correct": 0.0, "abstained": 0.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "562010df7e3ceb9bcd9a843b326413164596adc2f7ca548f1a2eb3ede9dfbb44", "output_hash": "2b9a4cf7221f1b6f31508a6b3078ec521f211026a932da5020cddd512ecc5fa2"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "6a745abb41130c8d658362a3", "attempt_id": "06977cccd023", "run_group_id": "E4_core:A4|repairable|qwen3b:h1", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "d83aa60f20d2879d02898cb9d9d8bd1983ed2d8984cae017c9153cafee3a779a", "git_commit": null, "timestamp_utc": "2026-10-30T06:24:55.716438Z", "task_family": "synthetic", "document_id": "syn_03000_2000", "document_hash": "c9cc27c20b05b1f6e592846e84990ee9b69c42f9720a8e338ec2f6fca777fdbc", "question_id": "q11_aggregation", "split": "test", "cluster_id": "syn_03000_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "laptop-m5max-48g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 1, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1904, "query_tokens": 604, "message_tokens": 511, "message_utf8_bytes": 1090, "cache_tensor_bytes": null, "serialized_payload_bytes": 1090, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 22777.715417090803, "hops": {"1": 22777.715417090803}, "receiver_prefill_ms": 405.97687498666346, "ttft_ms": 31.003542011603713, "decode_ms": 354.8971670679748}, "total_wall_ms": 23570.254542049952, "ttft_ms": 31.003542011603713, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 7}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "9220745", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 29, "retention_rate": 0.4461538461538462, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5, "status": 1.0, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1090}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 29, "retention_rate": 0.4461538461538462, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5, "status": 1.0, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1090}}}, "reference_key_id": "keys/synthetic/syn_03000_2000.json", "task_scores": {"correct": 0.0, "abstained": 0.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "562010df7e3ceb9bcd9a843b326413164596adc2f7ca548f1a2eb3ede9dfbb44", "output_hash": "f5d77aa763f6365bc80d01b9e6e23bc43fb570ac45771d7bd628c75cde727b35"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "1100277a26ba05e146c1313c", "attempt_id": "795cabb4a0ca", "run_group_id": "E4_core:A4|repairable|qwen3b:h1", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "d83aa60f20d2879d02898cb9d9d8bd1983ed2d8984cae017c9153cafee3a779a", "git_commit": null, "timestamp_utc": "2026-10-30T06:24:56.525178Z", "task_family": "synthetic", "document_id": "syn_03000_2000", "document_hash": "c9cc27c20b05b1f6e592846e84990ee9b69c42f9720a8e338ec2f6fca777fdbc", "question_id": "q12_unit_conversion", "split": "test", "cluster_id": "syn_03000_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "laptop-m5max-48g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 1, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1904, "query_tokens": 600, "message_tokens": 511, "message_utf8_bytes": 1090, "cache_tensor_bytes": null, "serialized_payload_bytes": 1090, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 22777.715417090803, "hops": {"1": 22777.715417090803}, "receiver_prefill_ms": 523.4964579576626, "ttft_ms": 42.7394158905372, "decode_ms": 30.16933298204094}, "total_wall_ms": 23374.717459082603, "ttft_ms": 42.7394158905372, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 29, "retention_rate": 0.4461538461538462, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5, "status": 1.0, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1090}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 29, "retention_rate": 0.4461538461538462, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5, "status": 1.0, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1090}}}, "reference_key_id": "keys/synthetic/syn_03000_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "562010df7e3ceb9bcd9a843b326413164596adc2f7ca548f1a2eb3ede9dfbb44", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "05e63cc9740aab4bbd45d196", "attempt_id": "73561bee63b0", "run_group_id": "E4_core:A4|repairable|qwen3b:h1", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "d83aa60f20d2879d02898cb9d9d8bd1983ed2d8984cae017c9153cafee3a779a", "git_commit": null, "timestamp_utc": "2026-10-30T06:24:57.139985Z", "task_family": "synthetic", "document_id": "syn_03000_2000", "document_hash": "c9cc27c20b05b1f6e592846e84990ee9b69c42f9720a8e338ec2f6fca777fdbc", "question_id": "q13_unanswerable", "split": "test", "cluster_id": "syn_03000_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "laptop-m5max-48g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 1, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1904, "query_tokens": 604, "message_tokens": 511, "message_utf8_bytes": 1090, "cache_tensor_bytes": null, "serialized_payload_bytes": 1090, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 22777.715417090803, "hops": {"1": 22777.715417090803}, "receiver_prefill_ms": 404.84470897354186, "ttft_ms": 39.79329101275653, "decode_ms": 23.352332995273173}, "total_wall_ms": 23246.24933314044, "ttft_ms": 39.79329101275653, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 29, "retention_rate": 0.4461538461538462, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5, "status": 1.0, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1090}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 29, "retention_rate": 0.4461538461538462, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5, "status": 1.0, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1090}}}, "reference_key_id": "keys/synthetic/syn_03000_2000.json", "task_scores": {"correct": 1.0, "abstained": 1.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "562010df7e3ceb9bcd9a843b326413164596adc2f7ca548f1a2eb3ede9dfbb44", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "514a95db58e0cf54a6d53903", "attempt_id": "39f5448dbd36", "run_group_id": "E4_core:A4|repairable|qwen3b:h1", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "d83aa60f20d2879d02898cb9d9d8bd1983ed2d8984cae017c9153cafee3a779a", "git_commit": null, "timestamp_utc": "2026-10-30T06:24:57.624476Z", "task_family": "synthetic", "document_id": "syn_03000_2000", "document_hash": "c9cc27c20b05b1f6e592846e84990ee9b69c42f9720a8e338ec2f6fca777fdbc", "question_id": "q14_unanswerable", "split": "test", "cluster_id": "syn_03000_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "laptop-m5max-48g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 1, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1904, "query_tokens": 606, "message_tokens": 511, "message_utf8_bytes": 1090, "cache_tensor_bytes": null, "serialized_payload_bytes": 1090, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 22777.715417090803, "hops": {"1": 22777.715417090803}, "receiver_prefill_ms": 362.3766669770703, "ttft_ms": 54.438958992250264, "decode_ms": 50.701500033028424}, "total_wall_ms": 23246.018042089418, "ttft_ms": 54.438958992250264, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 29, "retention_rate": 0.4461538461538462, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5, "status": 1.0, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1090}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 29, "retention_rate": 0.4461538461538462, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5, "status": 1.0, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1090}}}, "reference_key_id": "keys/synthetic/syn_03000_2000.json", "task_scores": {"correct": 1.0, "abstained": 1.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "562010df7e3ceb9bcd9a843b326413164596adc2f7ca548f1a2eb3ede9dfbb44", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "44bb9431b63139ed9e660497", "attempt_id": "b3d177243b84", "run_group_id": "E4_core:A4|repairable|qwen3b:h2", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "d83aa60f20d2879d02898cb9d9d8bd1983ed2d8984cae017c9153cafee3a779a", "git_commit": null, "timestamp_utc": "2026-10-30T06:25:21.953198Z", "task_family": "synthetic", "document_id": "syn_03000_2000", "document_hash": "c9cc27c20b05b1f6e592846e84990ee9b69c42f9720a8e338ec2f6fca777fdbc", "question_id": "q00_single_fact", "split": "test", "cluster_id": "syn_03000_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "laptop-m5max-48g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 2, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1904, "query_tokens": 576, "message_tokens": 495, "message_utf8_bytes": 994, "cache_tensor_bytes": null, "serialized_payload_bytes": 994, "side_information_bytes": 0, "fetched_bytes": 1228, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 46620.01320812851, "hops": {"1": 22777.715417090803, "2": 23842.29779103771}, "receiver_prefill_ms": 382.17154203448445, "ttft_ms": 25.18204099033028, "decode_ms": 37.32387500349432}, "total_wall_ms": 47065.26329112239, "ttft_ms": 25.18204099033028, "input_usage": {"lookups": 2}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 29, "retention_rate": 0.4461538461538462, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5, "status": 1.0, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 994}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 29, "retention_rate": 0.4461538461538462, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5, "status": 1.0, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1090}, "2": {"facts_total": 65, "facts_retained": 29, "retention_rate": 0.4461538461538462, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5, "status": 1.0, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 994}}}, "reference_key_id": "keys/synthetic/syn_03000_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "2dd4d43b9eb5a3a78d9b93791c358e57add647eceb6c72336d92751acf94f66d", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "335a12d041d8214fb25d0e95", "attempt_id": "880ba4916168", "run_group_id": "E4_core:A4|repairable|qwen3b:h2", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "d83aa60f20d2879d02898cb9d9d8bd1983ed2d8984cae017c9153cafee3a779a", "git_commit": null, "timestamp_utc": "2026-10-30T06:25:22.415658Z", "task_family": "synthetic", "document_id": "syn_03000_2000", "document_hash": "c9cc27c20b05b1f6e592846e84990ee9b69c42f9720a8e338ec2f6fca777fdbc", "question_id": "q01_single_fact", "split": "test", "cluster_id": "syn_03000_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "laptop-m5max-48g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 2, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1904, "query_tokens": 581, "message_tokens": 495, "message_utf8_bytes": 994, "cache_tensor_bytes": null, "serialized_payload_bytes": 994, "side_information_bytes": 0, "fetched_bytes": 1228, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 46620.01320812851, "hops": {"1": 22777.715417090803, "2": 23842.29779103771}, "receiver_prefill_ms": 349.790082895197, "ttft_ms": 42.86845901515335, "decode_ms": 542.4186660675332}, "total_wall_ms": 47555.74695812538, "ttft_ms": 42.86845901515335, "input_usage": {"lookups": 2}, "output_usage": {"output_tokens": 9}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "Lot 19-N, Gaspé", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 29, "retention_rate": 0.4461538461538462, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5, "status": 1.0, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 994}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 29, "retention_rate": 0.4461538461538462, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5, "status": 1.0, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1090}, "2": {"facts_total": 65, "facts_retained": 29, "retention_rate": 0.4461538461538462, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5, "status": 1.0, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 994}}}, "reference_key_id": "keys/synthetic/syn_03000_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 0.8571428571428571}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "2dd4d43b9eb5a3a78d9b93791c358e57add647eceb6c72336d92751acf94f66d", "output_hash": "2e8cf782bd118a1c6d4e23417785d72fb92e930806b5752cd1ec405a0c9b95e5"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "0968763762fd1dfdb1de1d2e", "attempt_id": "6e86f24357d4", "run_group_id": "E4_core:A4|repairable|qwen3b:h2", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "d83aa60f20d2879d02898cb9d9d8bd1983ed2d8984cae017c9153cafee3a779a", "git_commit": null, "timestamp_utc": "2026-10-30T06:25:23.367798Z", "task_family": "synthetic", "document_id": "syn_03000_2000", "document_hash": "c9cc27c20b05b1f6e592846e84990ee9b69c42f9720a8e338ec2f6fca777fdbc", "question_id": "q02_exact_copy", "split": "test", "cluster_id": "syn_03000_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "laptop-m5max-48g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 2, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1904, "query_tokens": 580, "message_tokens": 495, "message_utf8_bytes": 994, "cache_tensor_bytes": null, "serialized_payload_bytes": 994, "side_information_bytes": 0, "fetched_bytes": 1228, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 46620.01320812851, "hops": {"1": 22777.715417090803, "2": 23842.29779103771}, "receiver_prefill_ms": 381.31475006230175, "ttft_ms": 21.756166010163724, "decode_ms": 40.73287500068545}, "total_wall_ms": 47064.43233310711, "ttft_ms": 21.756166010163724, "input_usage": {"lookups": 2}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 29, "retention_rate": 0.4461538461538462, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5, "status": 1.0, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 994}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 29, "retention_rate": 0.4461538461538462, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5, "status": 1.0, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1090}, "2": {"facts_total": 65, "facts_retained": 29, "retention_rate": 0.4461538461538462, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5, "status": 1.0, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 994}}}, "reference_key_id": "keys/synthetic/syn_03000_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "2dd4d43b9eb5a3a78d9b93791c358e57add647eceb6c72336d92751acf94f66d", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "899543721962476b30b2e21d", "attempt_id": "b5b35d14ed66", "run_group_id": "E4_core:A4|repairable|qwen3b:h2", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "d83aa60f20d2879d02898cb9d9d8bd1983ed2d8984cae017c9153cafee3a779a", "git_commit": null, "timestamp_utc": "2026-10-30T06:25:23.829090Z", "task_family": "synthetic", "document_id": "syn_03000_2000", "document_hash": "c9cc27c20b05b1f6e592846e84990ee9b69c42f9720a8e338ec2f6fca777fdbc", "question_id": "q03_exact_copy", "split": "test", "cluster_id": "syn_03000_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "laptop-m5max-48g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 2, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1904, "query_tokens": 579, "message_tokens": 495, "message_utf8_bytes": 994, "cache_tensor_bytes": null, "serialized_payload_bytes": 994, "side_information_bytes": 0, "fetched_bytes": 1228, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 46620.01320812851, "hops": {"1": 22777.715417090803, "2": 23842.29779103771}, "receiver_prefill_ms": 384.7573749953881, "ttft_ms": 17.89449993520975, "decode_ms": 358.1150830723345}, "total_wall_ms": 47381.66366610676, "ttft_ms": 17.89449993520975, "input_usage": {"lookups": 2}, "output_usage": {"output_tokens": 7}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "Kwame Haldorsen", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 29, "retention_rate": 0.4461538461538462, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5, "status": 1.0, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 994}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 29, "retention_rate": 0.4461538461538462, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5, "status": 1.0, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1090}, "2": {"facts_total": 65, "facts_retained": 29, "retention_rate": 0.4461538461538462, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5, "status": 1.0, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 994}}}, "reference_key_id": "keys/synthetic/syn_03000_2000.json", "task_scores": {"correct": 0.0, "abstained": 0.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "2dd4d43b9eb5a3a78d9b93791c358e57add647eceb6c72336d92751acf94f66d", "output_hash": "175253e7cf463934ec4d54e361a781f323cd87290f01306ed368933bcbd2e25d"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "f058c525be11e443d016d8fe", "attempt_id": "079fa51dc0fc", "run_group_id": "E4_core:A4|repairable|qwen3b:h2", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "d83aa60f20d2879d02898cb9d9d8bd1983ed2d8984cae017c9153cafee3a779a", "git_commit": null, "timestamp_utc": "2026-10-30T06:25:24.607036Z", "task_family": "synthetic", "document_id": "syn_03000_2000", "document_hash": "c9cc27c20b05b1f6e592846e84990ee9b69c42f9720a8e338ec2f6fca777fdbc", "question_id": "q04_multi_hop", "split": "test", "cluster_id": "syn_03000_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "laptop-m5max-48g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 2, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1904, "query_tokens": 585, "message_tokens": 495, "message_utf8_bytes": 994, "cache_tensor_bytes": null, "serialized_payload_bytes": 994, "side_information_bytes": 0, "fetched_bytes": 1228, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 46620.01320812851, "hops": {"1": 22777.715417090803, "2": 23842.29779103771}, "receiver_prefill_ms": 510.54020796436816, "ttft_ms": 37.51612501218915, "decode_ms": 281.1572919599712}, "total_wall_ms": 47449.910750146955, "ttft_ms": 37.51612501218915, "input_usage": {"lookups": 2}, "output_usage": {"output_tokens": 9}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "9,220,745", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 29, "retention_rate": 0.4461538461538462, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5, "status": 1.0, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 994}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 29, "retention_rate": 0.4461538461538462, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5, "status": 1.0, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1090}, "2": {"facts_total": 65, "facts_retained": 29, "retention_rate": 0.4461538461538462, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5, "status": 1.0, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 994}}}, "reference_key_id": "keys/synthetic/syn_03000_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "2dd4d43b9eb5a3a78d9b93791c358e57add647eceb6c72336d92751acf94f66d", "output_hash": "ae55ba88b2cb1b0f1a4e0960ae312e0a1c177c79a6d14b67bff1677cec19b649"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "0c5093cc811db4cfb1ba6ae0", "attempt_id": "c0430ee6e3e8", "run_group_id": "E4_core:A4|repairable|qwen3b:h2", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "d83aa60f20d2879d02898cb9d9d8bd1983ed2d8984cae017c9153cafee3a779a", "git_commit": null, "timestamp_utc": "2026-10-30T06:25:25.453464Z", "task_family": "synthetic", "document_id": "syn_03000_2000", "document_hash": "c9cc27c20b05b1f6e592846e84990ee9b69c42f9720a8e338ec2f6fca777fdbc", "question_id": "q05_multi_hop", "split": "test", "cluster_id": "syn_03000_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "laptop-m5max-48g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 2, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1904, "query_tokens": 587, "message_tokens": 495, "message_utf8_bytes": 994, "cache_tensor_bytes": null, "serialized_payload_bytes": 994, "side_information_bytes": 0, "fetched_bytes": 1228, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 46620.01320812851, "hops": {"1": 22777.715417090803, "2": 23842.29779103771}, "receiver_prefill_ms": 371.0110420361161, "ttft_ms": 21.744875004515052, "decode_ms": 648.430333007127}, "total_wall_ms": 47661.81158309337, "ttft_ms": 21.744875004515052, "input_usage": {"lookups": 2}, "output_usage": {"output_tokens": 10}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "LY-51E-3199", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 29, "retention_rate": 0.4461538461538462, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5, "status": 1.0, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 994}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 29, "retention_rate": 0.4461538461538462, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5, "status": 1.0, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1090}, "2": {"facts_total": 65, "facts_retained": 29, "retention_rate": 0.4461538461538462, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5, "status": 1.0, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 994}}}, "reference_key_id": "keys/synthetic/syn_03000_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "2dd4d43b9eb5a3a78d9b93791c358e57add647eceb6c72336d92751acf94f66d", "output_hash": "a1245d1536890a7e4b026366df48441d503f491f782cfd388797b5782443c5c7"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "851385b85a039ce3229ecc42", "attempt_id": "c0f38b16e56d", "run_group_id": "E4_core:A4|repairable|qwen3b:h2", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "d83aa60f20d2879d02898cb9d9d8bd1983ed2d8984cae017c9153cafee3a779a", "git_commit": null, "timestamp_utc": "2026-10-30T06:25:26.511383Z", "task_family": "synthetic", "document_id": "syn_03000_2000", "document_hash": "c9cc27c20b05b1f6e592846e84990ee9b69c42f9720a8e338ec2f6fca777fdbc", "question_id": "q06_negation", "split": "test", "cluster_id": "syn_03000_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "laptop-m5max-48g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 2, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1904, "query_tokens": 576, "message_tokens": 495, "message_utf8_bytes": 994, "cache_tensor_bytes": null, "serialized_payload_bytes": 994, "side_information_bytes": 0, "fetched_bytes": 1228, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 46620.01320812851, "hops": {"1": 22777.715417090803, "2": 23842.29779103771}, "receiver_prefill_ms": 372.5313749164343, "ttft_ms": 38.25041698291898, "decode_ms": 27.673291973769665}, "total_wall_ms": 47059.076916077174, "ttft_ms": 38.25041698291898, "input_usage": {"lookups": 2}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "no", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 29, "retention_rate": 0.4461538461538462, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5, "status": 1.0, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 994}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 29, "retention_rate": 0.4461538461538462, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5, "status": 1.0, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1090}, "2": {"facts_total": 65, "facts_retained": 29, "retention_rate": 0.4461538461538462, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5, "status": 1.0, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 994}}}, "reference_key_id": "keys/synthetic/syn_03000_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "2dd4d43b9eb5a3a78d9b93791c358e57add647eceb6c72336d92751acf94f66d", "output_hash": "9390298f3fb0c5b160498935d79cb139aef28e1c47358b4bbba61862b9c26e59"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "ca99c3e6058d6899ae437ddd", "attempt_id": "a6792594e8ff", "run_group_id": "E4_core:A4|repairable|qwen3b:h2", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "d83aa60f20d2879d02898cb9d9d8bd1983ed2d8984cae017c9153cafee3a779a", "git_commit": null, "timestamp_utc": "2026-10-30T06:25:26.966839Z", "task_family": "synthetic", "document_id": "syn_03000_2000", "document_hash": "c9cc27c20b05b1f6e592846e84990ee9b69c42f9720a8e338ec2f6fca777fdbc", "question_id": "q07_negation", "split": "test", "cluster_id": "syn_03000_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "laptop-m5max-48g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 2, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1904, "query_tokens": 576, "message_tokens": 495, "message_utf8_bytes": 994, "cache_tensor_bytes": null, "serialized_payload_bytes": 994, "side_information_bytes": 0, "fetched_bytes": 1228, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 46620.01320812851, "hops": {"1": 22777.715417090803, "2": 23842.29779103771}, "receiver_prefill_ms": 378.83645796682686, "ttft_ms": 20.807042019441724, "decode_ms": 21.328584058210254}, "total_wall_ms": 47041.533000185154, "ttft_ms": 20.807042019441724, "input_usage": {"lookups": 2}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "no", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 29, "retention_rate": 0.4461538461538462, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5, "status": 1.0, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 994}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 29, "retention_rate": 0.4461538461538462, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5, "status": 1.0, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1090}, "2": {"facts_total": 65, "facts_retained": 29, "retention_rate": 0.4461538461538462, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5, "status": 1.0, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 994}}}, "reference_key_id": "keys/synthetic/syn_03000_2000.json", "task_scores": {"correct": 0.0, "abstained": 0.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "2dd4d43b9eb5a3a78d9b93791c358e57add647eceb6c72336d92751acf94f66d", "output_hash": "9390298f3fb0c5b160498935d79cb139aef28e1c47358b4bbba61862b9c26e59"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "7671e8d589633ef46e129f81", "attempt_id": "8f20294e2ca2", "run_group_id": "E4_core:A4|repairable|qwen3b:h2", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "d83aa60f20d2879d02898cb9d9d8bd1983ed2d8984cae017c9153cafee3a779a", "git_commit": null, "timestamp_utc": "2026-10-30T06:25:27.404947Z", "task_family": "synthetic", "document_id": "syn_03000_2000", "document_hash": "c9cc27c20b05b1f6e592846e84990ee9b69c42f9720a8e338ec2f6fca777fdbc", "question_id": "q08_revision", "split": "test", "cluster_id": "syn_03000_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "laptop-m5max-48g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 2, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1904, "query_tokens": 591, "message_tokens": 495, "message_utf8_bytes": 994, "cache_tensor_bytes": null, "serialized_payload_bytes": 994, "side_information_bytes": 0, "fetched_bytes": 1228, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 46620.01320812851, "hops": {"1": 22777.715417090803, "2": 23842.29779103771}, "receiver_prefill_ms": 438.32170800305903, "ttft_ms": 39.40404194872826, "decode_ms": 481.18687502574176}, "total_wall_ms": 47579.703542171046, "ttft_ms": 39.40404194872826, "input_usage": {"lookups": 2}, "output_usage": {"output_tokens": 9}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "2,705,965", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 29, "retention_rate": 0.4461538461538462, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5, "status": 1.0, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 994}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 29, "retention_rate": 0.4461538461538462, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5, "status": 1.0, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1090}, "2": {"facts_total": 65, "facts_retained": 29, "retention_rate": 0.4461538461538462, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5, "status": 1.0, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 994}}}, "reference_key_id": "keys/synthetic/syn_03000_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "2dd4d43b9eb5a3a78d9b93791c358e57add647eceb6c72336d92751acf94f66d", "output_hash": "dbedb6f5d00e413d915513b85d8ed195b18b9acf392ece342a85e15e3752bcc8"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "f50327a27bba52af2d49eada", "attempt_id": "7bcd2e8805a7", "run_group_id": "E4_core:A4|repairable|qwen3b:h2", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "d83aa60f20d2879d02898cb9d9d8bd1983ed2d8984cae017c9153cafee3a779a", "git_commit": null, "timestamp_utc": "2026-10-30T06:25:28.383914Z", "task_family": "synthetic", "document_id": "syn_03000_2000", "document_hash": "c9cc27c20b05b1f6e592846e84990ee9b69c42f9720a8e338ec2f6fca777fdbc", "question_id": "q09_revision", "split": "test", "cluster_id": "syn_03000_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "laptop-m5max-48g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 2, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1904, "query_tokens": 587, "message_tokens": 495, "message_utf8_bytes": 994, "cache_tensor_bytes": null, "serialized_payload_bytes": 994, "side_information_bytes": 0, "fetched_bytes": 1228, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 46620.01320812851, "hops": {"1": 22777.715417090803, "2": 23842.29779103771}, "receiver_prefill_ms": 392.92895805556327, "ttft_ms": 33.54216704610735, "decode_ms": 295.93683301936835}, "total_wall_ms": 47343.12075015623, "ttft_ms": 33.54216704610735, "input_usage": {"lookups": 2}, "output_usage": {"output_tokens": 9}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "9,127,965", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 29, "retention_rate": 0.4461538461538462, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5, "status": 1.0, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 994}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 29, "retention_rate": 0.4461538461538462, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5, "status": 1.0, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1090}, "2": {"facts_total": 65, "facts_retained": 29, "retention_rate": 0.4461538461538462, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5, "status": 1.0, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 994}}}, "reference_key_id": "keys/synthetic/syn_03000_2000.json", "task_scores": {"correct": 0.0, "abstained": 0.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "2dd4d43b9eb5a3a78d9b93791c358e57add647eceb6c72336d92751acf94f66d", "output_hash": "f30700451becb8c76f6e83d5016caa83d25106e6dceeffd8fdeca27984218444"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "973fc5e2acb950503c4e4246", "attempt_id": "5b566a396bea", "run_group_id": "E4_core:A4|repairable|qwen3b:h2", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "d83aa60f20d2879d02898cb9d9d8bd1983ed2d8984cae017c9153cafee3a779a", "git_commit": null, "timestamp_utc": "2026-10-30T06:25:29.123853Z", "task_family": "synthetic", "document_id": "syn_03000_2000", "document_hash": "c9cc27c20b05b1f6e592846e84990ee9b69c42f9720a8e338ec2f6fca777fdbc", "question_id": "q10_temporal", "split": "test", "cluster_id": "syn_03000_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "laptop-m5max-48g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 2, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1904, "query_tokens": 593, "message_tokens": 495, "message_utf8_bytes": 994, "cache_tensor_bytes": null, "serialized_payload_bytes": 994, "side_information_bytes": 0, "fetched_bytes": 1228, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 46620.01320812851, "hops": {"1": 22777.715417090803, "2": 23842.29779103771}, "receiver_prefill_ms": 377.41987500339746, "ttft_ms": 89.82662495691329, "decode_ms": 262.42054195608944}, "total_wall_ms": 47350.32799909823, "ttft_ms": 89.82662495691329, "input_usage": {"lookups": 2}, "output_usage": {"output_tokens": 5}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "Felaldnak-7", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 29, "retention_rate": 0.4461538461538462, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5, "status": 1.0, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 994}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 29, "retention_rate": 0.4461538461538462, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5, "status": 1.0, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1090}, "2": {"facts_total": 65, "facts_retained": 29, "retention_rate": 0.4461538461538462, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5, "status": 1.0, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 994}}}, "reference_key_id": "keys/synthetic/syn_03000_2000.json", "task_scores": {"correct": 0.0, "abstained": 0.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "2dd4d43b9eb5a3a78d9b93791c358e57add647eceb6c72336d92751acf94f66d", "output_hash": "2b9a4cf7221f1b6f31508a6b3078ec521f211026a932da5020cddd512ecc5fa2"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "e3fc17b66bfeccb66d514a5f", "attempt_id": "e323a63b1fe2", "run_group_id": "E4_core:A4|repairable|qwen3b:h2", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "d83aa60f20d2879d02898cb9d9d8bd1983ed2d8984cae017c9153cafee3a779a", "git_commit": null, "timestamp_utc": "2026-10-30T06:25:29.871712Z", "task_family": "synthetic", "document_id": "syn_03000_2000", "document_hash": "c9cc27c20b05b1f6e592846e84990ee9b69c42f9720a8e338ec2f6fca777fdbc", "question_id": "q11_aggregation", "split": "test", "cluster_id": "syn_03000_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "laptop-m5max-48g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 2, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1904, "query_tokens": 588, "message_tokens": 495, "message_utf8_bytes": 994, "cache_tensor_bytes": null, "serialized_payload_bytes": 994, "side_information_bytes": 0, "fetched_bytes": 1228, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 46620.01320812851, "hops": {"1": 22777.715417090803, "2": 23842.29779103771}, "receiver_prefill_ms": 392.06687500700355, "ttft_ms": 23.412375012412667, "decode_ms": 237.44991689454764}, "total_wall_ms": 47273.633166216314, "ttft_ms": 23.412375012412667, "input_usage": {"lookups": 2}, "output_usage": {"output_tokens": 7}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "9220745", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 29, "retention_rate": 0.4461538461538462, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5, "status": 1.0, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 994}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 29, "retention_rate": 0.4461538461538462, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5, "status": 1.0, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1090}, "2": {"facts_total": 65, "facts_retained": 29, "retention_rate": 0.4461538461538462, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5, "status": 1.0, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 994}}}, "reference_key_id": "keys/synthetic/syn_03000_2000.json", "task_scores": {"correct": 0.0, "abstained": 0.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "2dd4d43b9eb5a3a78d9b93791c358e57add647eceb6c72336d92751acf94f66d", "output_hash": "f5d77aa763f6365bc80d01b9e6e23bc43fb570ac45771d7bd628c75cde727b35"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "26c57603d973ffef59fc9c5e", "attempt_id": "567ee0057fc0", "run_group_id": "E4_core:A4|repairable|qwen3b:h2", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "d83aa60f20d2879d02898cb9d9d8bd1983ed2d8984cae017c9153cafee3a779a", "git_commit": null, "timestamp_utc": "2026-10-30T06:25:30.541870Z", "task_family": "synthetic", "document_id": "syn_03000_2000", "document_hash": "c9cc27c20b05b1f6e592846e84990ee9b69c42f9720a8e338ec2f6fca777fdbc", "question_id": "q12_unit_conversion", "split": "test", "cluster_id": "syn_03000_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "laptop-m5max-48g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 2, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1904, "query_tokens": 584, "message_tokens": 495, "message_utf8_bytes": 994, "cache_tensor_bytes": null, "serialized_payload_bytes": 994, "side_information_bytes": 0, "fetched_bytes": 1228, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 46620.01320812851, "hops": {"1": 22777.715417090803, "2": 23842.29779103771}, "receiver_prefill_ms": 381.70008396264166, "ttft_ms": 48.753166920505464, "decode_ms": 51.29879096057266}, "total_wall_ms": 47102.52466611564, "ttft_ms": 48.753166920505464, "input_usage": {"lookups": 2}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 29, "retention_rate": 0.4461538461538462, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5, "status": 1.0, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 994}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 29, "retention_rate": 0.4461538461538462, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5, "status": 1.0, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1090}, "2": {"facts_total": 65, "facts_retained": 29, "retention_rate": 0.4461538461538462, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5, "status": 1.0, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 994}}}, "reference_key_id": "keys/synthetic/syn_03000_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "2dd4d43b9eb5a3a78d9b93791c358e57add647eceb6c72336d92751acf94f66d", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "6cd145d9a536a62f36297d3c", "attempt_id": "149c802cc17b", "run_group_id": "E4_core:A4|repairable|qwen3b:h2", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "d83aa60f20d2879d02898cb9d9d8bd1983ed2d8984cae017c9153cafee3a779a", "git_commit": null, "timestamp_utc": "2026-10-30T06:25:31.041745Z", "task_family": "synthetic", "document_id": "syn_03000_2000", "document_hash": "c9cc27c20b05b1f6e592846e84990ee9b69c42f9720a8e338ec2f6fca777fdbc", "question_id": "q13_unanswerable", "split": "test", "cluster_id": "syn_03000_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "laptop-m5max-48g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 2, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1904, "query_tokens": 588, "message_tokens": 495, "message_utf8_bytes": 994, "cache_tensor_bytes": null, "serialized_payload_bytes": 994, "side_information_bytes": 0, "fetched_bytes": 1228, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 46620.01320812851, "hops": {"1": 22777.715417090803, "2": 23842.29779103771}, "receiver_prefill_ms": 554.2078749276698, "ttft_ms": 31.165124964900315, "decode_ms": 40.04083399195224}, "total_wall_ms": 47246.12304114271, "ttft_ms": 31.165124964900315, "input_usage": {"lookups": 2}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 29, "retention_rate": 0.4461538461538462, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5, "status": 1.0, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 994}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 29, "retention_rate": 0.4461538461538462, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5, "status": 1.0, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1090}, "2": {"facts_total": 65, "facts_retained": 29, "retention_rate": 0.4461538461538462, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5, "status": 1.0, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 994}}}, "reference_key_id": "keys/synthetic/syn_03000_2000.json", "task_scores": {"correct": 1.0, "abstained": 1.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "2dd4d43b9eb5a3a78d9b93791c358e57add647eceb6c72336d92751acf94f66d", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "561ac972ea2726991fbc7de5", "attempt_id": "001ea787da5f", "run_group_id": "E4_core:A4|repairable|qwen3b:h2", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "d83aa60f20d2879d02898cb9d9d8bd1983ed2d8984cae017c9153cafee3a779a", "git_commit": null, "timestamp_utc": "2026-10-30T06:25:31.686232Z", "task_family": "synthetic", "document_id": "syn_03000_2000", "document_hash": "c9cc27c20b05b1f6e592846e84990ee9b69c42f9720a8e338ec2f6fca777fdbc", "question_id": "q14_unanswerable", "split": "test", "cluster_id": "syn_03000_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "laptop-m5max-48g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 2, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1904, "query_tokens": 590, "message_tokens": 495, "message_utf8_bytes": 994, "cache_tensor_bytes": null, "serialized_payload_bytes": 994, "side_information_bytes": 0, "fetched_bytes": 1228, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 46620.01320812851, "hops": {"1": 22777.715417090803, "2": 23842.29779103771}, "receiver_prefill_ms": 401.5621249563992, "ttft_ms": 33.67258294019848, "decode_ms": 38.02850004285574}, "total_wall_ms": 47094.037250149995, "ttft_ms": 33.67258294019848, "input_usage": {"lookups": 2}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 29, "retention_rate": 0.4461538461538462, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5, "status": 1.0, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 994}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 29, "retention_rate": 0.4461538461538462, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5, "status": 1.0, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1090}, "2": {"facts_total": 65, "facts_retained": 29, "retention_rate": 0.4461538461538462, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5, "status": 1.0, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 994}}}, "reference_key_id": "keys/synthetic/syn_03000_2000.json", "task_scores": {"correct": 1.0, "abstained": 1.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "2dd4d43b9eb5a3a78d9b93791c358e57add647eceb6c72336d92751acf94f66d", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "b1bf8ad68014dae62ad2bfcc", "attempt_id": "62ad7ef9dfe0", "run_group_id": "E4_core:A4|repairable|qwen3b:h4", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "d83aa60f20d2879d02898cb9d9d8bd1983ed2d8984cae017c9153cafee3a779a", "git_commit": null, "timestamp_utc": "2026-10-30T06:26:15.298326Z", "task_family": "synthetic", "document_id": "syn_03000_2000", "document_hash": "c9cc27c20b05b1f6e592846e84990ee9b69c42f9720a8e338ec2f6fca777fdbc", "question_id": "q00_single_fact", "split": "test", "cluster_id": "syn_03000_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "laptop-m5max-48g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 4, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1904, "query_tokens": 576, "message_tokens": 495, "message_utf8_bytes": 994, "cache_tensor_bytes": null, "serialized_payload_bytes": 994, "side_information_bytes": 0, "fetched_bytes": 3684, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 89737.4447081238, "hops": {"1": 22777.715417090803, "2": 23842.29779103771, "3": 21627.477917005308, "4": 21489.953582989983}, "receiver_prefill_ms": 456.46191702689976, "ttft_ms": 39.4563329173252, "decode_ms": 24.237625068053603}, "total_wall_ms": 90258.29366606195, "ttft_ms": 39.4563329173252, "input_usage": {"lookups": 6}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 29, "retention_rate": 0.4461538461538462, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5, "status": 1.0, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 994}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 29, "retention_rate": 0.4461538461538462, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5, "status": 1.0, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1090}, "2": {"facts_total": 65, "facts_retained": 29, "retention_rate": 0.4461538461538462, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5, "status": 1.0, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 994}, "3": {"facts_total": 65, "facts_retained": 29, "retention_rate": 0.4461538461538462, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5, "status": 1.0, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 994}, "4": {"facts_total": 65, "facts_retained": 29, "retention_rate": 0.4461538461538462, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5, "status": 1.0, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 994}}}, "reference_key_id": "keys/synthetic/syn_03000_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "2dd4d43b9eb5a3a78d9b93791c358e57add647eceb6c72336d92751acf94f66d", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "d6895fc620bad515c8cbaaac", "attempt_id": "b1fe7e5fbd0c", "run_group_id": "E4_core:A4|repairable|qwen3b:h4", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "d83aa60f20d2879d02898cb9d9d8bd1983ed2d8984cae017c9153cafee3a779a", "git_commit": null, "timestamp_utc": "2026-10-30T06:26:15.839182Z", "task_family": "synthetic", "document_id": "syn_03000_2000", "document_hash": "c9cc27c20b05b1f6e592846e84990ee9b69c42f9720a8e338ec2f6fca777fdbc", "question_id": "q01_single_fact", "split": "test", "cluster_id": "syn_03000_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "laptop-m5max-48g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 4, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1904, "query_tokens": 581, "message_tokens": 495, "message_utf8_bytes": 994, "cache_tensor_bytes": null, "serialized_payload_bytes": 994, "side_information_bytes": 0, "fetched_bytes": 3684, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 89737.4447081238, "hops": {"1": 22777.715417090803, "2": 23842.29779103771, "3": 21627.477917005308, "4": 21489.953582989983}, "receiver_prefill_ms": 403.51024991832674, "ttft_ms": 34.88512500189245, "decode_ms": 312.3252500081435}, "total_wall_ms": 90488.84004214779, "ttft_ms": 34.88512500189245, "input_usage": {"lookups": 6}, "output_usage": {"output_tokens": 9}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "Lot 19-N, Gaspé", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 29, "retention_rate": 0.4461538461538462, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5, "status": 1.0, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 994}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 29, "retention_rate": 0.4461538461538462, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5, "status": 1.0, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1090}, "2": {"facts_total": 65, "facts_retained": 29, "retention_rate": 0.4461538461538462, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5, "status": 1.0, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 994}, "3": {"facts_total": 65, "facts_retained": 29, "retention_rate": 0.4461538461538462, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5, "status": 1.0, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 994}, "4": {"facts_total": 65, "facts_retained": 29, "retention_rate": 0.4461538461538462, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5, "status": 1.0, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 994}}}, "reference_key_id": "keys/synthetic/syn_03000_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 0.8571428571428571}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "2dd4d43b9eb5a3a78d9b93791c358e57add647eceb6c72336d92751acf94f66d", "output_hash": "2e8cf782bd118a1c6d4e23417785d72fb92e930806b5752cd1ec405a0c9b95e5"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "5be2e271904451695a958b23", "attempt_id": "79626da2f2f6", "run_group_id": "E4_core:A4|repairable|qwen3b:h4", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "d83aa60f20d2879d02898cb9d9d8bd1983ed2d8984cae017c9153cafee3a779a", "git_commit": null, "timestamp_utc": "2026-10-30T06:26:16.606911Z", "task_family": "synthetic", "document_id": "syn_03000_2000", "document_hash": "c9cc27c20b05b1f6e592846e84990ee9b69c42f9720a8e338ec2f6fca777fdbc", "question_id": "q02_exact_copy", "split": "test", "cluster_id": "syn_03000_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "laptop-m5max-48g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 4, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1904, "query_tokens": 580, "message_tokens": 495, "message_utf8_bytes": 994, "cache_tensor_bytes": null, "serialized_payload_bytes": 994, "side_information_bytes": 0, "fetched_bytes": 3684, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 89737.4447081238, "hops": {"1": 22777.715417090803, "2": 23842.29779103771, "3": 21627.477917005308, "4": 21489.953582989983}, "receiver_prefill_ms": 525.8983749663457, "ttft_ms": 39.225458982400596, "decode_ms": 20.31087502837181}, "total_wall_ms": 90323.62066709902, "ttft_ms": 39.225458982400596, "input_usage": {"lookups": 6}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 29, "retention_rate": 0.4461538461538462, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5, "status": 1.0, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 994}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 29, "retention_rate": 0.4461538461538462, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5, "status": 1.0, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1090}, "2": {"facts_total": 65, "facts_retained": 29, "retention_rate": 0.4461538461538462, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5, "status": 1.0, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 994}, "3": {"facts_total": 65, "facts_retained": 29, "retention_rate": 0.4461538461538462, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5, "status": 1.0, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 994}, "4": {"facts_total": 65, "facts_retained": 29, "retention_rate": 0.4461538461538462, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5, "status": 1.0, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 994}}}, "reference_key_id": "keys/synthetic/syn_03000_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "2dd4d43b9eb5a3a78d9b93791c358e57add647eceb6c72336d92751acf94f66d", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "7615e8cbb96aa9f765dc5bc4", "attempt_id": "fbbc4093835c", "run_group_id": "E4_core:A4|repairable|qwen3b:h4", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "d83aa60f20d2879d02898cb9d9d8bd1983ed2d8984cae017c9153cafee3a779a", "git_commit": null, "timestamp_utc": "2026-10-30T06:26:17.210275Z", "task_family": "synthetic", "document_id": "syn_03000_2000", "document_hash": "c9cc27c20b05b1f6e592846e84990ee9b69c42f9720a8e338ec2f6fca777fdbc", "question_id": "q03_exact_copy", "split": "test", "cluster_id": "syn_03000_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "laptop-m5max-48g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 4, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1904, "query_tokens": 579, "message_tokens": 495, "message_utf8_bytes": 994, "cache_tensor_bytes": null, "serialized_payload_bytes": 994, "side_information_bytes": 0, "fetched_bytes": 3684, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 89737.4447081238, "hops": {"1": 22777.715417090803, "2": 23842.29779103771, "3": 21627.477917005308, "4": 21489.953582989983}, "receiver_prefill_ms": 396.6160420095548, "ttft_ms": 26.329500018619, "decode_ms": 230.718708015047}, "total_wall_ms": 90391.78916718811, "ttft_ms": 26.329500018619, "input_usage": {"lookups": 6}, "output_usage": {"output_tokens": 7}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "Kwame Haldorsen", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 29, "retention_rate": 0.4461538461538462, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5, "status": 1.0, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 994}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 29, "retention_rate": 0.4461538461538462, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5, "status": 1.0, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1090}, "2": {"facts_total": 65, "facts_retained": 29, "retention_rate": 0.4461538461538462, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5, "status": 1.0, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 994}, "3": {"facts_total": 65, "facts_retained": 29, "retention_rate": 0.4461538461538462, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5, "status": 1.0, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 994}, "4": {"facts_total": 65, "facts_retained": 29, "retention_rate": 0.4461538461538462, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5, "status": 1.0, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 994}}}, "reference_key_id": "keys/synthetic/syn_03000_2000.json", "task_scores": {"correct": 0.0, "abstained": 0.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "2dd4d43b9eb5a3a78d9b93791c358e57add647eceb6c72336d92751acf94f66d", "output_hash": "175253e7cf463934ec4d54e361a781f323cd87290f01306ed368933bcbd2e25d"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "57717f5f2f421d3e55eeb6a0", "attempt_id": "c8d1a4d028e4", "run_group_id": "E4_core:A4|repairable|qwen3b:h4", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "d83aa60f20d2879d02898cb9d9d8bd1983ed2d8984cae017c9153cafee3a779a", "git_commit": null, "timestamp_utc": "2026-10-30T06:26:17.881031Z", "task_family": "synthetic", "document_id": "syn_03000_2000", "document_hash": "c9cc27c20b05b1f6e592846e84990ee9b69c42f9720a8e338ec2f6fca777fdbc", "question_id": "q04_multi_hop", "split": "test", "cluster_id": "syn_03000_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "laptop-m5max-48g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 4, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1904, "query_tokens": 585, "message_tokens": 495, "message_utf8_bytes": 994, "cache_tensor_bytes": null, "serialized_payload_bytes": 994, "side_information_bytes": 0, "fetched_bytes": 3684, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 89737.4447081238, "hops": {"1": 22777.715417090803, "2": 23842.29779103771, "3": 21627.477917005308, "4": 21489.953582989983}, "receiver_prefill_ms": 364.5249999826774, "ttft_ms": 44.10604201257229, "decode_ms": 548.3271250268444}, "total_wall_ms": 90695.13029209338, "ttft_ms": 44.10604201257229, "input_usage": {"lookups": 6}, "output_usage": {"output_tokens": 9}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "9,220,745", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 29, "retention_rate": 0.4461538461538462, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5, "status": 1.0, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 994}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 29, "retention_rate": 0.4461538461538462, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5, "status": 1.0, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1090}, "2": {"facts_total": 65, "facts_retained": 29, "retention_rate": 0.4461538461538462, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5, "status": 1.0, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 994}, "3": {"facts_total": 65, "facts_retained": 29, "retention_rate": 0.4461538461538462, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5, "status": 1.0, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 994}, "4": {"facts_total": 65, "facts_retained": 29, "retention_rate": 0.4461538461538462, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5, "status": 1.0, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 994}}}, "reference_key_id": "keys/synthetic/syn_03000_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "2dd4d43b9eb5a3a78d9b93791c358e57add647eceb6c72336d92751acf94f66d", "output_hash": "ae55ba88b2cb1b0f1a4e0960ae312e0a1c177c79a6d14b67bff1677cec19b649"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "9a65e7c85aba5ff10cf32b4c", "attempt_id": "999736d1afb5", "run_group_id": "E4_core:A4|repairable|qwen3b:h4", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "d83aa60f20d2879d02898cb9d9d8bd1983ed2d8984cae017c9153cafee3a779a", "git_commit": null, "timestamp_utc": "2026-10-30T06:26:18.855820Z", "task_family": "synthetic", "document_id": "syn_03000_2000", "document_hash": "c9cc27c20b05b1f6e592846e84990ee9b69c42f9720a8e338ec2f6fca777fdbc", "question_id": "q05_multi_hop", "split": "test", "cluster_id": "syn_03000_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "laptop-m5max-48g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 4, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1904, "query_tokens": 587, "message_tokens": 495, "message_utf8_bytes": 994, "cache_tensor_bytes": null, "serialized_payload_bytes": 994, "side_information_bytes": 0, "fetched_bytes": 3684, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 89737.4447081238, "hops": {"1": 22777.715417090803, "2": 23842.29779103771, "3": 21627.477917005308, "4": 21489.953582989983}, "receiver_prefill_ms": 413.8276659650728, "ttft_ms": 26.210749987512827, "decode_ms": 309.58799994550645}, "total_wall_ms": 90487.70579113625, "ttft_ms": 26.210749987512827, "input_usage": {"lookups": 6}, "output_usage": {"output_tokens": 10}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "LY-51E-3199", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 29, "retention_rate": 0.4461538461538462, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5, "status": 1.0, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 994}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 29, "retention_rate": 0.4461538461538462, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5, "status": 1.0, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1090}, "2": {"facts_total": 65, "facts_retained": 29, "retention_rate": 0.4461538461538462, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5, "status": 1.0, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 994}, "3": {"facts_total": 65, "facts_retained": 29, "retention_rate": 0.4461538461538462, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5, "status": 1.0, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 994}, "4": {"facts_total": 65, "facts_retained": 29, "retention_rate": 0.4461538461538462, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5, "status": 1.0, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 994}}}, "reference_key_id": "keys/synthetic/syn_03000_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "2dd4d43b9eb5a3a78d9b93791c358e57add647eceb6c72336d92751acf94f66d", "output_hash": "a1245d1536890a7e4b026366df48441d503f491f782cfd388797b5782443c5c7"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "b8e08990bcc485fa3bc0d286", "attempt_id": "75d5ab7fc41c", "run_group_id": "E4_core:A4|repairable|qwen3b:h4", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "d83aa60f20d2879d02898cb9d9d8bd1983ed2d8984cae017c9153cafee3a779a", "git_commit": null, "timestamp_utc": "2026-10-30T06:26:19.623210Z", "task_family": "synthetic", "document_id": "syn_03000_2000", "document_hash": "c9cc27c20b05b1f6e592846e84990ee9b69c42f9720a8e338ec2f6fca777fdbc", "question_id": "q06_negation", "split": "test", "cluster_id": "syn_03000_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "laptop-m5max-48g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 4, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1904, "query_tokens": 576, "message_tokens": 495, "message_utf8_bytes": 994, "cache_tensor_bytes": null, "serialized_payload_bytes": 994, "side_information_bytes": 0, "fetched_bytes": 3684, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 89737.4447081238, "hops": {"1": 22777.715417090803, "2": 23842.29779103771, "3": 21627.477917005308, "4": 21489.953582989983}, "receiver_prefill_ms": 353.01595903001726, "ttft_ms": 51.98041605763137, "decode_ms": 57.99270805437118}, "total_wall_ms": 90201.0831660591, "ttft_ms": 51.98041605763137, "input_usage": {"lookups": 6}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "no", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 29, "retention_rate": 0.4461538461538462, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5, "status": 1.0, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 994}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 29, "retention_rate": 0.4461538461538462, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5, "status": 1.0, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1090}, "2": {"facts_total": 65, "facts_retained": 29, "retention_rate": 0.4461538461538462, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5, "status": 1.0, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 994}, "3": {"facts_total": 65, "facts_retained": 29, "retention_rate": 0.4461538461538462, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5, "status": 1.0, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 994}, "4": {"facts_total": 65, "facts_retained": 29, "retention_rate": 0.4461538461538462, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5, "status": 1.0, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 994}}}, "reference_key_id": "keys/synthetic/syn_03000_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "2dd4d43b9eb5a3a78d9b93791c358e57add647eceb6c72336d92751acf94f66d", "output_hash": "9390298f3fb0c5b160498935d79cb139aef28e1c47358b4bbba61862b9c26e59"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "653e88adf6ba4cc5f134f501", "attempt_id": "98bc3b42060e", "run_group_id": "E4_core:A4|repairable|qwen3b:h4", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "d83aa60f20d2879d02898cb9d9d8bd1983ed2d8984cae017c9153cafee3a779a", "git_commit": null, "timestamp_utc": "2026-10-30T06:26:20.104121Z", "task_family": "synthetic", "document_id": "syn_03000_2000", "document_hash": "c9cc27c20b05b1f6e592846e84990ee9b69c42f9720a8e338ec2f6fca777fdbc", "question_id": "q07_negation", "split": "test", "cluster_id": "syn_03000_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "laptop-m5max-48g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 4, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1904, "query_tokens": 576, "message_tokens": 495, "message_utf8_bytes": 994, "cache_tensor_bytes": null, "serialized_payload_bytes": 994, "side_information_bytes": 0, "fetched_bytes": 3684, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 89737.4447081238, "hops": {"1": 22777.715417090803, "2": 23842.29779103771, "3": 21627.477917005308, "4": 21489.953582989983}, "receiver_prefill_ms": 497.1636669943109, "ttft_ms": 27.16470800805837, "decode_ms": 37.3656660085544}, "total_wall_ms": 90299.66883303132, "ttft_ms": 27.16470800805837, "input_usage": {"lookups": 6}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "no", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 29, "retention_rate": 0.4461538461538462, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5, "status": 1.0, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 994}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 29, "retention_rate": 0.4461538461538462, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5, "status": 1.0, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1090}, "2": {"facts_total": 65, "facts_retained": 29, "retention_rate": 0.4461538461538462, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5, "status": 1.0, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 994}, "3": {"facts_total": 65, "facts_retained": 29, "retention_rate": 0.4461538461538462, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5, "status": 1.0, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 994}, "4": {"facts_total": 65, "facts_retained": 29, "retention_rate": 0.4461538461538462, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5, "status": 1.0, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 994}}}, "reference_key_id": "keys/synthetic/syn_03000_2000.json", "task_scores": {"correct": 0.0, "abstained": 0.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "2dd4d43b9eb5a3a78d9b93791c358e57add647eceb6c72336d92751acf94f66d", "output_hash": "9390298f3fb0c5b160498935d79cb139aef28e1c47358b4bbba61862b9c26e59"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "f8bd715cf3fb12e76982af73", "attempt_id": "25572d4862c5", "run_group_id": "E4_core:A4|repairable|qwen3b:h4", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "d83aa60f20d2879d02898cb9d9d8bd1983ed2d8984cae017c9153cafee3a779a", "git_commit": null, "timestamp_utc": "2026-10-30T06:26:20.684581Z", "task_family": "synthetic", "document_id": "syn_03000_2000", "document_hash": "c9cc27c20b05b1f6e592846e84990ee9b69c42f9720a8e338ec2f6fca777fdbc", "question_id": "q08_revision", "split": "test", "cluster_id": "syn_03000_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "laptop-m5max-48g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 4, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1904, "query_tokens": 591, "message_tokens": 495, "message_utf8_bytes": 994, "cache_tensor_bytes": null, "serialized_payload_bytes": 994, "side_information_bytes": 0, "fetched_bytes": 3684, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 89737.4447081238, "hops": {"1": 22777.715417090803, "2": 23842.29779103771, "3": 21627.477917005308, "4": 21489.953582989983}, "receiver_prefill_ms": 407.5375000247732, "ttft_ms": 34.913084004074335, "decode_ms": 283.52416690904647}, "total_wall_ms": 90464.06308317091, "ttft_ms": 34.913084004074335, "input_usage": {"lookups": 6}, "output_usage": {"output_tokens": 9}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "2,705,965", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 29, "retention_rate": 0.4461538461538462, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5, "status": 1.0, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 994}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 29, "retention_rate": 0.4461538461538462, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5, "status": 1.0, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1090}, "2": {"facts_total": 65, "facts_retained": 29, "retention_rate": 0.4461538461538462, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5, "status": 1.0, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 994}, "3": {"facts_total": 65, "facts_retained": 29, "retention_rate": 0.4461538461538462, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5, "status": 1.0, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 994}, "4": {"facts_total": 65, "facts_retained": 29, "retention_rate": 0.4461538461538462, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5, "status": 1.0, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 994}}}, "reference_key_id": "keys/synthetic/syn_03000_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "2dd4d43b9eb5a3a78d9b93791c358e57add647eceb6c72336d92751acf94f66d", "output_hash": "dbedb6f5d00e413d915513b85d8ed195b18b9acf392ece342a85e15e3752bcc8"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "7525a9b9c4f7ee10bf17c4a2", "attempt_id": "a0392b449f9a", "run_group_id": "E4_core:A4|repairable|qwen3b:h4", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "d83aa60f20d2879d02898cb9d9d8bd1983ed2d8984cae017c9153cafee3a779a", "git_commit": null, "timestamp_utc": "2026-10-30T06:26:21.427956Z", "task_family": "synthetic", "document_id": "syn_03000_2000", "document_hash": "c9cc27c20b05b1f6e592846e84990ee9b69c42f9720a8e338ec2f6fca777fdbc", "question_id": "q09_revision", "split": "test", "cluster_id": "syn_03000_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "laptop-m5max-48g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 4, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1904, "query_tokens": 587, "message_tokens": 495, "message_utf8_bytes": 994, "cache_tensor_bytes": null, "serialized_payload_bytes": 994, "side_information_bytes": 0, "fetched_bytes": 3684, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 89737.4447081238, "hops": {"1": 22777.715417090803, "2": 23842.29779103771, "3": 21627.477917005308, "4": 21489.953582989983}, "receiver_prefill_ms": 387.8211249830201, "ttft_ms": 96.37795796152204, "decode_ms": 414.2479168949649}, "total_wall_ms": 90636.62979111541, "ttft_ms": 96.37795796152204, "input_usage": {"lookups": 6}, "output_usage": {"output_tokens": 9}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "9,127,965", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 29, "retention_rate": 0.4461538461538462, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5, "status": 1.0, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 994}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 29, "retention_rate": 0.4461538461538462, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5, "status": 1.0, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1090}, "2": {"facts_total": 65, "facts_retained": 29, "retention_rate": 0.4461538461538462, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5, "status": 1.0, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 994}, "3": {"facts_total": 65, "facts_retained": 29, "retention_rate": 0.4461538461538462, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5, "status": 1.0, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 994}, "4": {"facts_total": 65, "facts_retained": 29, "retention_rate": 0.4461538461538462, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5, "status": 1.0, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 994}}}, "reference_key_id": "keys/synthetic/syn_03000_2000.json", "task_scores": {"correct": 0.0, "abstained": 0.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "2dd4d43b9eb5a3a78d9b93791c358e57add647eceb6c72336d92751acf94f66d", "output_hash": "f30700451becb8c76f6e83d5016caa83d25106e6dceeffd8fdeca27984218444"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "bff87eecbc2e55fc5d8024ab", "attempt_id": "9c50db394202", "run_group_id": "E4_core:A4|repairable|qwen3b:h4", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "d83aa60f20d2879d02898cb9d9d8bd1983ed2d8984cae017c9153cafee3a779a", "git_commit": null, "timestamp_utc": "2026-10-30T06:26:22.344398Z", "task_family": "synthetic", "document_id": "syn_03000_2000", "document_hash": "c9cc27c20b05b1f6e592846e84990ee9b69c42f9720a8e338ec2f6fca777fdbc", "question_id": "q10_temporal", "split": "test", "cluster_id": "syn_03000_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "laptop-m5max-48g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 4, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1904, "query_tokens": 593, "message_tokens": 495, "message_utf8_bytes": 994, "cache_tensor_bytes": null, "serialized_payload_bytes": 994, "side_information_bytes": 0, "fetched_bytes": 3684, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 89737.4447081238, "hops": {"1": 22777.715417090803, "2": 23842.29779103771, "3": 21627.477917005308, "4": 21489.953582989983}, "receiver_prefill_ms": 400.0500839902088, "ttft_ms": 35.15304101165384, "decode_ms": 146.16558398120105}, "total_wall_ms": 90319.39333316404, "ttft_ms": 35.15304101165384, "input_usage": {"lookups": 6}, "output_usage": {"output_tokens": 5}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "Felaldnak-7", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 29, "retention_rate": 0.4461538461538462, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5, "status": 1.0, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 994}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 29, "retention_rate": 0.4461538461538462, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5, "status": 1.0, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1090}, "2": {"facts_total": 65, "facts_retained": 29, "retention_rate": 0.4461538461538462, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5, "status": 1.0, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 994}, "3": {"facts_total": 65, "facts_retained": 29, "retention_rate": 0.4461538461538462, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5, "status": 1.0, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 994}, "4": {"facts_total": 65, "facts_retained": 29, "retention_rate": 0.4461538461538462, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5, "status": 1.0, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 994}}}, "reference_key_id": "keys/synthetic/syn_03000_2000.json", "task_scores": {"correct": 0.0, "abstained": 0.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "2dd4d43b9eb5a3a78d9b93791c358e57add647eceb6c72336d92751acf94f66d", "output_hash": "2b9a4cf7221f1b6f31508a6b3078ec521f211026a932da5020cddd512ecc5fa2"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "daa4a81f0926aa1bccd6b5df", "attempt_id": "d3276d673c89", "run_group_id": "E4_core:A4|repairable|qwen3b:h4", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "d83aa60f20d2879d02898cb9d9d8bd1983ed2d8984cae017c9153cafee3a779a", "git_commit": null, "timestamp_utc": "2026-10-30T06:26:22.943674Z", "task_family": "synthetic", "document_id": "syn_03000_2000", "document_hash": "c9cc27c20b05b1f6e592846e84990ee9b69c42f9720a8e338ec2f6fca777fdbc", "question_id": "q11_aggregation", "split": "test", "cluster_id": "syn_03000_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "laptop-m5max-48g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 4, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1904, "query_tokens": 588, "message_tokens": 495, "message_utf8_bytes": 994, "cache_tensor_bytes": null, "serialized_payload_bytes": 994, "side_information_bytes": 0, "fetched_bytes": 3684, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 89737.4447081238, "hops": {"1": 22777.715417090803, "2": 23842.29779103771, "3": 21627.477917005308, "4": 21489.953582989983}, "receiver_prefill_ms": 419.1465419717133, "ttft_ms": 55.228334036655724, "decode_ms": 407.7339159557596}, "total_wall_ms": 90620.19516713917, "ttft_ms": 55.228334036655724, "input_usage": {"lookups": 6}, "output_usage": {"output_tokens": 7}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "9220745", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 29, "retention_rate": 0.4461538461538462, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5, "status": 1.0, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 994}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 29, "retention_rate": 0.4461538461538462, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5, "status": 1.0, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1090}, "2": {"facts_total": 65, "facts_retained": 29, "retention_rate": 0.4461538461538462, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5, "status": 1.0, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 994}, "3": {"facts_total": 65, "facts_retained": 29, "retention_rate": 0.4461538461538462, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5, "status": 1.0, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 994}, "4": {"facts_total": 65, "facts_retained": 29, "retention_rate": 0.4461538461538462, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5, "status": 1.0, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 994}}}, "reference_key_id": "keys/synthetic/syn_03000_2000.json", "task_scores": {"correct": 0.0, "abstained": 0.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "2dd4d43b9eb5a3a78d9b93791c358e57add647eceb6c72336d92751acf94f66d", "output_hash": "f5d77aa763f6365bc80d01b9e6e23bc43fb570ac45771d7bd628c75cde727b35"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "0e188252d13e8ee31dcd3490", "attempt_id": "e0f8256f8b42", "run_group_id": "E4_core:A4|repairable|qwen3b:h4", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "d83aa60f20d2879d02898cb9d9d8bd1983ed2d8984cae017c9153cafee3a779a", "git_commit": null, "timestamp_utc": "2026-10-30T06:26:23.844736Z", "task_family": "synthetic", "document_id": "syn_03000_2000", "document_hash": "c9cc27c20b05b1f6e592846e84990ee9b69c42f9720a8e338ec2f6fca777fdbc", "question_id": "q12_unit_conversion", "split": "test", "cluster_id": "syn_03000_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "laptop-m5max-48g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 4, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1904, "query_tokens": 584, "message_tokens": 495, "message_utf8_bytes": 994, "cache_tensor_bytes": null, "serialized_payload_bytes": 994, "side_information_bytes": 0, "fetched_bytes": 3684, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 89737.4447081238, "hops": {"1": 22777.715417090803, "2": 23842.29779103771, "3": 21627.477917005308, "4": 21489.953582989983}, "receiver_prefill_ms": 404.21566599979997, "ttft_ms": 19.10391601268202, "decode_ms": 41.94175009615719}, "total_wall_ms": 90203.44958314672, "ttft_ms": 19.10391601268202, "input_usage": {"lookups": 6}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 29, "retention_rate": 0.4461538461538462, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5, "status": 1.0, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 994}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 29, "retention_rate": 0.4461538461538462, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5, "status": 1.0, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1090}, "2": {"facts_total": 65, "facts_retained": 29, "retention_rate": 0.4461538461538462, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5, "status": 1.0, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 994}, "3": {"facts_total": 65, "facts_retained": 29, "retention_rate": 0.4461538461538462, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5, "status": 1.0, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 994}, "4": {"facts_total": 65, "facts_retained": 29, "retention_rate": 0.4461538461538462, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5, "status": 1.0, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 994}}}, "reference_key_id": "keys/synthetic/syn_03000_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "2dd4d43b9eb5a3a78d9b93791c358e57add647eceb6c72336d92751acf94f66d", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "90da30c5f1169f028be18697", "attempt_id": "c557be484c01", "run_group_id": "E4_core:A4|repairable|qwen3b:h4", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "d83aa60f20d2879d02898cb9d9d8bd1983ed2d8984cae017c9153cafee3a779a", "git_commit": null, "timestamp_utc": "2026-10-30T06:26:24.328142Z", "task_family": "synthetic", "document_id": "syn_03000_2000", "document_hash": "c9cc27c20b05b1f6e592846e84990ee9b69c42f9720a8e338ec2f6fca777fdbc", "question_id": "q13_unanswerable", "split": "test", "cluster_id": "syn_03000_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "laptop-m5max-48g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 4, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1904, "query_tokens": 588, "message_tokens": 495, "message_utf8_bytes": 994, "cache_tensor_bytes": null, "serialized_payload_bytes": 994, "side_information_bytes": 0, "fetched_bytes": 3684, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 89737.4447081238, "hops": {"1": 22777.715417090803, "2": 23842.29779103771, "3": 21627.477917005308, "4": 21489.953582989983}, "receiver_prefill_ms": 373.7751670414582, "ttft_ms": 46.5746249537915, "decode_ms": 65.43304107617587}, "total_wall_ms": 90223.8702911418, "ttft_ms": 46.5746249537915, "input_usage": {"lookups": 6}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 29, "retention_rate": 0.4461538461538462, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5, "status": 1.0, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 994}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 29, "retention_rate": 0.4461538461538462, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5, "status": 1.0, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1090}, "2": {"facts_total": 65, "facts_retained": 29, "retention_rate": 0.4461538461538462, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5, "status": 1.0, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 994}, "3": {"facts_total": 65, "facts_retained": 29, "retention_rate": 0.4461538461538462, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5, "status": 1.0, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 994}, "4": {"facts_total": 65, "facts_retained": 29, "retention_rate": 0.4461538461538462, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5, "status": 1.0, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 994}}}, "reference_key_id": "keys/synthetic/syn_03000_2000.json", "task_scores": {"correct": 1.0, "abstained": 1.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "2dd4d43b9eb5a3a78d9b93791c358e57add647eceb6c72336d92751acf94f66d", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "4d7b025222f62b7f6dda6cfa", "attempt_id": "9233e66a3342", "run_group_id": "E4_core:A4|repairable|qwen3b:h4", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "d83aa60f20d2879d02898cb9d9d8bd1983ed2d8984cae017c9153cafee3a779a", "git_commit": null, "timestamp_utc": "2026-10-30T06:26:24.830700Z", "task_family": "synthetic", "document_id": "syn_03000_2000", "document_hash": "c9cc27c20b05b1f6e592846e84990ee9b69c42f9720a8e338ec2f6fca777fdbc", "question_id": "q14_unanswerable", "split": "test", "cluster_id": "syn_03000_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "laptop-m5max-48g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 4, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1904, "query_tokens": 590, "message_tokens": 495, "message_utf8_bytes": 994, "cache_tensor_bytes": null, "serialized_payload_bytes": 994, "side_information_bytes": 0, "fetched_bytes": 3684, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 89737.4447081238, "hops": {"1": 22777.715417090803, "2": 23842.29779103771, "3": 21627.477917005308, "4": 21489.953582989983}, "receiver_prefill_ms": 585.6464999960735, "ttft_ms": 42.444709106348455, "decode_ms": 29.163707979023457}, "total_wall_ms": 90395.35733312368, "ttft_ms": 42.444709106348455, "input_usage": {"lookups": 6}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 29, "retention_rate": 0.4461538461538462, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5, "status": 1.0, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 994}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 29, "retention_rate": 0.4461538461538462, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5, "status": 1.0, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1090}, "2": {"facts_total": 65, "facts_retained": 29, "retention_rate": 0.4461538461538462, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5, "status": 1.0, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 994}, "3": {"facts_total": 65, "facts_retained": 29, "retention_rate": 0.4461538461538462, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5, "status": 1.0, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 994}, "4": {"facts_total": 65, "facts_retained": 29, "retention_rate": 0.4461538461538462, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5, "status": 1.0, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 994}}}, "reference_key_id": "keys/synthetic/syn_03000_2000.json", "task_scores": {"correct": 1.0, "abstained": 1.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "2dd4d43b9eb5a3a78d9b93791c358e57add647eceb6c72336d92751acf94f66d", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "dacc2c7616ce0c7db1b55e60", "attempt_id": "69dbe81ccad9", "run_group_id": "E4_core:A4|repairable|qwen3b:h8", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "d83aa60f20d2879d02898cb9d9d8bd1983ed2d8984cae017c9153cafee3a779a", "git_commit": null, "timestamp_utc": "2026-10-30T06:27:51.698628Z", "task_family": "synthetic", "document_id": "syn_03000_2000", "document_hash": "c9cc27c20b05b1f6e592846e84990ee9b69c42f9720a8e338ec2f6fca777fdbc", "question_id": "q00_single_fact", "split": "test", "cluster_id": "syn_03000_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "laptop-m5max-48g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 8, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1904, "query_tokens": 576, "message_tokens": 495, "message_utf8_bytes": 994, "cache_tensor_bytes": null, "serialized_payload_bytes": 994, "side_information_bytes": 0, "fetched_bytes": 8596, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 175923.93000016455, "hops": {"1": 22777.715417090803, "2": 23842.29779103771, "3": 21627.477917005308, "4": 21489.953582989983, "5": 21196.44504203461, "6": 21617.7212910261, "7": 21700.178666971624, "8": 21672.140292008407}, "receiver_prefill_ms": 372.14487500023097, "ttft_ms": 21.170292049646378, "decode_ms": 38.4662919677794}, "total_wall_ms": 176356.5159161808, "ttft_ms": 21.170292049646378, "input_usage": {"lookups": 14}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 29, "retention_rate": 0.4461538461538462, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5, "status": 1.0, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 994}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 29, "retention_rate": 0.4461538461538462, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5, "status": 1.0, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1090}, "2": {"facts_total": 65, "facts_retained": 29, "retention_rate": 0.4461538461538462, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5, "status": 1.0, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 994}, "3": {"facts_total": 65, "facts_retained": 29, "retention_rate": 0.4461538461538462, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5, "status": 1.0, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 994}, "4": {"facts_total": 65, "facts_retained": 29, "retention_rate": 0.4461538461538462, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5, "status": 1.0, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 994}, "5": {"facts_total": 65, "facts_retained": 29, "retention_rate": 0.4461538461538462, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5, "status": 1.0, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 994}, "6": {"facts_total": 65, "facts_retained": 29, "retention_rate": 0.4461538461538462, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5, "status": 1.0, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 994}, "7": {"facts_total": 65, "facts_retained": 29, "retention_rate": 0.4461538461538462, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5, "status": 1.0, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 994}, "8": {"facts_total": 65, "facts_retained": 29, "retention_rate": 0.4461538461538462, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5, "status": 1.0, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 994}}}, "reference_key_id": "keys/synthetic/syn_03000_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "2dd4d43b9eb5a3a78d9b93791c358e57add647eceb6c72336d92751acf94f66d", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "01da541bde1b794052a0efe8", "attempt_id": "1c9316b18c8f", "run_group_id": "E4_core:A4|repairable|qwen3b:h8", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "d83aa60f20d2879d02898cb9d9d8bd1983ed2d8984cae017c9153cafee3a779a", "git_commit": null, "timestamp_utc": "2026-10-30T06:27:52.148928Z", "task_family": "synthetic", "document_id": "syn_03000_2000", "document_hash": "c9cc27c20b05b1f6e592846e84990ee9b69c42f9720a8e338ec2f6fca777fdbc", "question_id": "q01_single_fact", "split": "test", "cluster_id": "syn_03000_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "laptop-m5max-48g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 8, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1904, "query_tokens": 581, "message_tokens": 495, "message_utf8_bytes": 994, "cache_tensor_bytes": null, "serialized_payload_bytes": 994, "side_information_bytes": 0, "fetched_bytes": 8596, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 175923.93000016455, "hops": {"1": 22777.715417090803, "2": 23842.29779103771, "3": 21627.477917005308, "4": 21489.953582989983, "5": 21196.44504203461, "6": 21617.7212910261, "7": 21700.178666971624, "8": 21672.140292008407}, "receiver_prefill_ms": 378.20825004018843, "ttft_ms": 38.148832973092794, "decode_ms": 312.31566704809666}, "total_wall_ms": 176653.51779211778, "ttft_ms": 38.148832973092794, "input_usage": {"lookups": 14}, "output_usage": {"output_tokens": 9}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "Lot 19-N, Gaspé", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 29, "retention_rate": 0.4461538461538462, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5, "status": 1.0, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 994}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 29, "retention_rate": 0.4461538461538462, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5, "status": 1.0, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1090}, "2": {"facts_total": 65, "facts_retained": 29, "retention_rate": 0.4461538461538462, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5, "status": 1.0, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 994}, "3": {"facts_total": 65, "facts_retained": 29, "retention_rate": 0.4461538461538462, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5, "status": 1.0, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 994}, "4": {"facts_total": 65, "facts_retained": 29, "retention_rate": 0.4461538461538462, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5, "status": 1.0, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 994}, "5": {"facts_total": 65, "facts_retained": 29, "retention_rate": 0.4461538461538462, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5, "status": 1.0, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 994}, "6": {"facts_total": 65, "facts_retained": 29, "retention_rate": 0.4461538461538462, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5, "status": 1.0, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 994}, "7": {"facts_total": 65, "facts_retained": 29, "retention_rate": 0.4461538461538462, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5, "status": 1.0, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 994}, "8": {"facts_total": 65, "facts_retained": 29, "retention_rate": 0.4461538461538462, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5, "status": 1.0, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 994}}}, "reference_key_id": "keys/synthetic/syn_03000_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 0.8571428571428571}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "2dd4d43b9eb5a3a78d9b93791c358e57add647eceb6c72336d92751acf94f66d", "output_hash": "2e8cf782bd118a1c6d4e23417785d72fb92e930806b5752cd1ec405a0c9b95e5"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "c38dc89cb17a318cb91590ce", "attempt_id": "6ab3b74286b8", "run_group_id": "E4_core:A4|repairable|qwen3b:h8", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "d83aa60f20d2879d02898cb9d9d8bd1983ed2d8984cae017c9153cafee3a779a", "git_commit": null, "timestamp_utc": "2026-10-30T06:27:52.895760Z", "task_family": "synthetic", "document_id": "syn_03000_2000", "document_hash": "c9cc27c20b05b1f6e592846e84990ee9b69c42f9720a8e338ec2f6fca777fdbc", "question_id": "q02_exact_copy", "split": "test", "cluster_id": "syn_03000_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "laptop-m5max-48g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 8, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1904, "query_tokens": 580, "message_tokens": 495, "message_utf8_bytes": 994, "cache_tensor_bytes": null, "serialized_payload_bytes": 994, "side_information_bytes": 0, "fetched_bytes": 8596, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 175923.93000016455, "hops": {"1": 22777.715417090803, "2": 23842.29779103771, "3": 21627.477917005308, "4": 21489.953582989983, "5": 21196.44504203461, "6": 21617.7212910261, "7": 21700.178666971624, "8": 21672.140292008407}, "receiver_prefill_ms": 460.7057499233633, "ttft_ms": 68.06716695427895, "decode_ms": 36.534624989144504}, "total_wall_ms": 176489.90504222456, "ttft_ms": 68.06716695427895, "input_usage": {"lookups": 14}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 29, "retention_rate": 0.4461538461538462, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5, "status": 1.0, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 994}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 29, "retention_rate": 0.4461538461538462, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5, "status": 1.0, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1090}, "2": {"facts_total": 65, "facts_retained": 29, "retention_rate": 0.4461538461538462, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5, "status": 1.0, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 994}, "3": {"facts_total": 65, "facts_retained": 29, "retention_rate": 0.4461538461538462, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5, "status": 1.0, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 994}, "4": {"facts_total": 65, "facts_retained": 29, "retention_rate": 0.4461538461538462, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5, "status": 1.0, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 994}, "5": {"facts_total": 65, "facts_retained": 29, "retention_rate": 0.4461538461538462, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5, "status": 1.0, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 994}, "6": {"facts_total": 65, "facts_retained": 29, "retention_rate": 0.4461538461538462, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5, "status": 1.0, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 994}, "7": {"facts_total": 65, "facts_retained": 29, "retention_rate": 0.4461538461538462, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5, "status": 1.0, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 994}, "8": {"facts_total": 65, "facts_retained": 29, "retention_rate": 0.4461538461538462, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5, "status": 1.0, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 994}}}, "reference_key_id": "keys/synthetic/syn_03000_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "2dd4d43b9eb5a3a78d9b93791c358e57add647eceb6c72336d92751acf94f66d", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "5c6ead58c896573839c6cbd2", "attempt_id": "c64f646e1c5b", "run_group_id": "E4_core:A4|repairable|qwen3b:h8", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "d83aa60f20d2879d02898cb9d9d8bd1983ed2d8984cae017c9153cafee3a779a", "git_commit": null, "timestamp_utc": "2026-10-30T06:27:53.479425Z", "task_family": "synthetic", "document_id": "syn_03000_2000", "document_hash": "c9cc27c20b05b1f6e592846e84990ee9b69c42f9720a8e338ec2f6fca777fdbc", "question_id": "q03_exact_copy", "split": "test", "cluster_id": "syn_03000_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "laptop-m5max-48g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 8, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1904, "query_tokens": 579, "message_tokens": 495, "message_utf8_bytes": 994, "cache_tensor_bytes": null, "serialized_payload_bytes": 994, "side_information_bytes": 0, "fetched_bytes": 8596, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 175923.93000016455, "hops": {"1": 22777.715417090803, "2": 23842.29779103771, "3": 21627.477917005308, "4": 21489.953582989983, "5": 21196.44504203461, "6": 21617.7212910261, "7": 21700.178666971624, "8": 21672.140292008407}, "receiver_prefill_ms": 403.7026250734925, "ttft_ms": 39.763416978530586, "decode_ms": 214.01645801961422}, "total_wall_ms": 176582.14525005314, "ttft_ms": 39.763416978530586, "input_usage": {"lookups": 14}, "output_usage": {"output_tokens": 7}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "Kwame Haldorsen", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 29, "retention_rate": 0.4461538461538462, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5, "status": 1.0, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 994}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 29, "retention_rate": 0.4461538461538462, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5, "status": 1.0, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1090}, "2": {"facts_total": 65, "facts_retained": 29, "retention_rate": 0.4461538461538462, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5, "status": 1.0, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 994}, "3": {"facts_total": 65, "facts_retained": 29, "retention_rate": 0.4461538461538462, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5, "status": 1.0, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 994}, "4": {"facts_total": 65, "facts_retained": 29, "retention_rate": 0.4461538461538462, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5, "status": 1.0, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 994}, "5": {"facts_total": 65, "facts_retained": 29, "retention_rate": 0.4461538461538462, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5, "status": 1.0, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 994}, "6": {"facts_total": 65, "facts_retained": 29, "retention_rate": 0.4461538461538462, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5, "status": 1.0, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 994}, "7": {"facts_total": 65, "facts_retained": 29, "retention_rate": 0.4461538461538462, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5, "status": 1.0, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 994}, "8": {"facts_total": 65, "facts_retained": 29, "retention_rate": 0.4461538461538462, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5, "status": 1.0, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 994}}}, "reference_key_id": "keys/synthetic/syn_03000_2000.json", "task_scores": {"correct": 0.0, "abstained": 0.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "2dd4d43b9eb5a3a78d9b93791c358e57add647eceb6c72336d92751acf94f66d", "output_hash": "175253e7cf463934ec4d54e361a781f323cd87290f01306ed368933bcbd2e25d"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "00b021d74fd9b3e98c01e282", "attempt_id": "70470a6b2ac3", "run_group_id": "E4_core:A4|repairable|qwen3b:h8", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "d83aa60f20d2879d02898cb9d9d8bd1983ed2d8984cae017c9153cafee3a779a", "git_commit": null, "timestamp_utc": "2026-10-30T06:27:54.154461Z", "task_family": "synthetic", "document_id": "syn_03000_2000", "document_hash": "c9cc27c20b05b1f6e592846e84990ee9b69c42f9720a8e338ec2f6fca777fdbc", "question_id": "q04_multi_hop", "split": "test", "cluster_id": "syn_03000_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "laptop-m5max-48g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 8, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1904, "query_tokens": 585, "message_tokens": 495, "message_utf8_bytes": 994, "cache_tensor_bytes": null, "serialized_payload_bytes": 994, "side_information_bytes": 0, "fetched_bytes": 8596, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 175923.93000016455, "hops": {"1": 22777.715417090803, "2": 23842.29779103771, "3": 21627.477917005308, "4": 21489.953582989983, "5": 21196.44504203461, "6": 21617.7212910261, "7": 21700.178666971624, "8": 21672.140292008407}, "receiver_prefill_ms": 384.90070798434317, "ttft_ms": 51.5132499858737, "decode_ms": 559.0171249350533}, "total_wall_ms": 176920.04933417775, "ttft_ms": 51.5132499858737, "input_usage": {"lookups": 14}, "output_usage": {"output_tokens": 9}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "9,220,745", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 29, "retention_rate": 0.4461538461538462, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5, "status": 1.0, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 994}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 29, "retention_rate": 0.4461538461538462, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5, "status": 1.0, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1090}, "2": {"facts_total": 65, "facts_retained": 29, "retention_rate": 0.4461538461538462, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5, "status": 1.0, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 994}, "3": {"facts_total": 65, "facts_retained": 29, "retention_rate": 0.4461538461538462, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5, "status": 1.0, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 994}, "4": {"facts_total": 65, "facts_retained": 29, "retention_rate": 0.4461538461538462, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5, "status": 1.0, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 994}, "5": {"facts_total": 65, "facts_retained": 29, "retention_rate": 0.4461538461538462, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5, "status": 1.0, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 994}, "6": {"facts_total": 65, "facts_retained": 29, "retention_rate": 0.4461538461538462, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5, "status": 1.0, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 994}, "7": {"facts_total": 65, "facts_retained": 29, "retention_rate": 0.4461538461538462, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5, "status": 1.0, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 994}, "8": {"facts_total": 65, "facts_retained": 29, "retention_rate": 0.4461538461538462, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5, "status": 1.0, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 994}}}, "reference_key_id": "keys/synthetic/syn_03000_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "2dd4d43b9eb5a3a78d9b93791c358e57add647eceb6c72336d92751acf94f66d", "output_hash": "ae55ba88b2cb1b0f1a4e0960ae312e0a1c177c79a6d14b67bff1677cec19b649"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "a214d9879f512923633ee848", "attempt_id": "86bb8270fa04", "run_group_id": "E4_core:A4|repairable|qwen3b:h8", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "d83aa60f20d2879d02898cb9d9d8bd1983ed2d8984cae017c9153cafee3a779a", "git_commit": null, "timestamp_utc": "2026-10-30T06:27:55.167764Z", "task_family": "synthetic", "document_id": "syn_03000_2000", "document_hash": "c9cc27c20b05b1f6e592846e84990ee9b69c42f9720a8e338ec2f6fca777fdbc", "question_id": "q05_multi_hop", "split": "test", "cluster_id": "syn_03000_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "laptop-m5max-48g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 8, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1904, "query_tokens": 587, "message_tokens": 495, "message_utf8_bytes": 994, "cache_tensor_bytes": null, "serialized_payload_bytes": 994, "side_information_bytes": 0, "fetched_bytes": 8596, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 175923.93000016455, "hops": {"1": 22777.715417090803, "2": 23842.29779103771, "3": 21627.477917005308, "4": 21489.953582989983, "5": 21196.44504203461, "6": 21617.7212910261, "7": 21700.178666971624, "8": 21672.140292008407}, "receiver_prefill_ms": 421.3557919720188, "ttft_ms": 27.547416975721717, "decode_ms": 315.28558395802975}, "total_wall_ms": 176688.95816721488, "ttft_ms": 27.547416975721717, "input_usage": {"lookups": 14}, "output_usage": {"output_tokens": 10}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "LY-51E-3199", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 29, "retention_rate": 0.4461538461538462, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5, "status": 1.0, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 994}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 29, "retention_rate": 0.4461538461538462, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5, "status": 1.0, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1090}, "2": {"facts_total": 65, "facts_retained": 29, "retention_rate": 0.4461538461538462, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5, "status": 1.0, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 994}, "3": {"facts_total": 65, "facts_retained": 29, "retention_rate": 0.4461538461538462, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5, "status": 1.0, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 994}, "4": {"facts_total": 65, "facts_retained": 29, "retention_rate": 0.4461538461538462, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5, "status": 1.0, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 994}, "5": {"facts_total": 65, "facts_retained": 29, "retention_rate": 0.4461538461538462, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5, "status": 1.0, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 994}, "6": {"facts_total": 65, "facts_retained": 29, "retention_rate": 0.4461538461538462, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5, "status": 1.0, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 994}, "7": {"facts_total": 65, "facts_retained": 29, "retention_rate": 0.4461538461538462, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5, "status": 1.0, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 994}, "8": {"facts_total": 65, "facts_retained": 29, "retention_rate": 0.4461538461538462, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5, "status": 1.0, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 994}}}, "reference_key_id": "keys/synthetic/syn_03000_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "2dd4d43b9eb5a3a78d9b93791c358e57add647eceb6c72336d92751acf94f66d", "output_hash": "a1245d1536890a7e4b026366df48441d503f491f782cfd388797b5782443c5c7"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "635dec17042a73bc0c6363c0", "attempt_id": "664f1e9d0a30", "run_group_id": "E4_core:A4|repairable|qwen3b:h8", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "d83aa60f20d2879d02898cb9d9d8bd1983ed2d8984cae017c9153cafee3a779a", "git_commit": null, "timestamp_utc": "2026-10-30T06:27:55.950377Z", "task_family": "synthetic", "document_id": "syn_03000_2000", "document_hash": "c9cc27c20b05b1f6e592846e84990ee9b69c42f9720a8e338ec2f6fca777fdbc", "question_id": "q06_negation", "split": "test", "cluster_id": "syn_03000_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "laptop-m5max-48g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 8, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1904, "query_tokens": 576, "message_tokens": 495, "message_utf8_bytes": 994, "cache_tensor_bytes": null, "serialized_payload_bytes": 994, "side_information_bytes": 0, "fetched_bytes": 8596, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 175923.93000016455, "hops": {"1": 22777.715417090803, "2": 23842.29779103771, "3": 21627.477917005308, "4": 21489.953582989983, "5": 21196.44504203461, "6": 21617.7212910261, "7": 21700.178666971624, "8": 21672.140292008407}, "receiver_prefill_ms": 404.81166704557836, "ttft_ms": 89.26558296661824, "decode_ms": 139.793417067267}, "total_wall_ms": 176558.41716716532, "ttft_ms": 89.26558296661824, "input_usage": {"lookups": 14}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "no", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 29, "retention_rate": 0.4461538461538462, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5, "status": 1.0, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 994}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 29, "retention_rate": 0.4461538461538462, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5, "status": 1.0, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1090}, "2": {"facts_total": 65, "facts_retained": 29, "retention_rate": 0.4461538461538462, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5, "status": 1.0, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 994}, "3": {"facts_total": 65, "facts_retained": 29, "retention_rate": 0.4461538461538462, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5, "status": 1.0, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 994}, "4": {"facts_total": 65, "facts_retained": 29, "retention_rate": 0.4461538461538462, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5, "status": 1.0, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 994}, "5": {"facts_total": 65, "facts_retained": 29, "retention_rate": 0.4461538461538462, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5, "status": 1.0, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 994}, "6": {"facts_total": 65, "facts_retained": 29, "retention_rate": 0.4461538461538462, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5, "status": 1.0, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 994}, "7": {"facts_total": 65, "facts_retained": 29, "retention_rate": 0.4461538461538462, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5, "status": 1.0, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 994}, "8": {"facts_total": 65, "facts_retained": 29, "retention_rate": 0.4461538461538462, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5, "status": 1.0, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 994}}}, "reference_key_id": "keys/synthetic/syn_03000_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "2dd4d43b9eb5a3a78d9b93791c358e57add647eceb6c72336d92751acf94f66d", "output_hash": "9390298f3fb0c5b160498935d79cb139aef28e1c47358b4bbba61862b9c26e59"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "d5e88785854d725b630b9b28", "attempt_id": "5ebda8a1022c", "run_group_id": "E4_core:A4|repairable|qwen3b:h8", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "d83aa60f20d2879d02898cb9d9d8bd1983ed2d8984cae017c9153cafee3a779a", "git_commit": null, "timestamp_utc": "2026-10-30T06:27:56.601576Z", "task_family": "synthetic", "document_id": "syn_03000_2000", "document_hash": "c9cc27c20b05b1f6e592846e84990ee9b69c42f9720a8e338ec2f6fca777fdbc", "question_id": "q07_negation", "split": "test", "cluster_id": "syn_03000_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "laptop-m5max-48g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 8, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1904, "query_tokens": 576, "message_tokens": 495, "message_utf8_bytes": 994, "cache_tensor_bytes": null, "serialized_payload_bytes": 994, "side_information_bytes": 0, "fetched_bytes": 8596, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 175923.93000016455, "hops": {"1": 22777.715417090803, "2": 23842.29779103771, "3": 21627.477917005308, "4": 21489.953582989983, "5": 21196.44504203461, "6": 21617.7212910261, "7": 21700.178666971624, "8": 21672.140292008407}, "receiver_prefill_ms": 375.79708406701684, "ttft_ms": 23.277916945517063, "decode_ms": 41.4033749839291}, "total_wall_ms": 176365.13912514783, "ttft_ms": 23.277916945517063, "input_usage": {"lookups": 14}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "no", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 29, "retention_rate": 0.4461538461538462, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5, "status": 1.0, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 994}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 29, "retention_rate": 0.4461538461538462, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5, "status": 1.0, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1090}, "2": {"facts_total": 65, "facts_retained": 29, "retention_rate": 0.4461538461538462, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5, "status": 1.0, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 994}, "3": {"facts_total": 65, "facts_retained": 29, "retention_rate": 0.4461538461538462, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5, "status": 1.0, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 994}, "4": {"facts_total": 65, "facts_retained": 29, "retention_rate": 0.4461538461538462, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5, "status": 1.0, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 994}, "5": {"facts_total": 65, "facts_retained": 29, "retention_rate": 0.4461538461538462, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5, "status": 1.0, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 994}, "6": {"facts_total": 65, "facts_retained": 29, "retention_rate": 0.4461538461538462, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5, "status": 1.0, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 994}, "7": {"facts_total": 65, "facts_retained": 29, "retention_rate": 0.4461538461538462, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5, "status": 1.0, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 994}, "8": {"facts_total": 65, "facts_retained": 29, "retention_rate": 0.4461538461538462, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5, "status": 1.0, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 994}}}, "reference_key_id": "keys/synthetic/syn_03000_2000.json", "task_scores": {"correct": 0.0, "abstained": 0.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "2dd4d43b9eb5a3a78d9b93791c358e57add647eceb6c72336d92751acf94f66d", "output_hash": "9390298f3fb0c5b160498935d79cb139aef28e1c47358b4bbba61862b9c26e59"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "a387aba3c3d61fbc16bfc36d", "attempt_id": "252af84aaf13", "run_group_id": "E4_core:A4|repairable|qwen3b:h8", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "d83aa60f20d2879d02898cb9d9d8bd1983ed2d8984cae017c9153cafee3a779a", "git_commit": null, "timestamp_utc": "2026-10-30T06:27:57.059379Z", "task_family": "synthetic", "document_id": "syn_03000_2000", "document_hash": "c9cc27c20b05b1f6e592846e84990ee9b69c42f9720a8e338ec2f6fca777fdbc", "question_id": "q08_revision", "split": "test", "cluster_id": "syn_03000_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "laptop-m5max-48g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 8, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1904, "query_tokens": 591, "message_tokens": 495, "message_utf8_bytes": 994, "cache_tensor_bytes": null, "serialized_payload_bytes": 994, "side_information_bytes": 0, "fetched_bytes": 8596, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 175923.93000016455, "hops": {"1": 22777.715417090803, "2": 23842.29779103771, "3": 21627.477917005308, "4": 21489.953582989983, "5": 21196.44504203461, "6": 21617.7212910261, "7": 21700.178666971624, "8": 21672.140292008407}, "receiver_prefill_ms": 399.17587500531226, "ttft_ms": 23.088958929292858, "decode_ms": 392.8399169817567}, "total_wall_ms": 176739.82733418234, "ttft_ms": 23.088958929292858, "input_usage": {"lookups": 14}, "output_usage": {"output_tokens": 9}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "2,705,965", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 29, "retention_rate": 0.4461538461538462, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5, "status": 1.0, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 994}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 29, "retention_rate": 0.4461538461538462, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5, "status": 1.0, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1090}, "2": {"facts_total": 65, "facts_retained": 29, "retention_rate": 0.4461538461538462, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5, "status": 1.0, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 994}, "3": {"facts_total": 65, "facts_retained": 29, "retention_rate": 0.4461538461538462, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5, "status": 1.0, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 994}, "4": {"facts_total": 65, "facts_retained": 29, "retention_rate": 0.4461538461538462, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5, "status": 1.0, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 994}, "5": {"facts_total": 65, "facts_retained": 29, "retention_rate": 0.4461538461538462, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5, "status": 1.0, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 994}, "6": {"facts_total": 65, "facts_retained": 29, "retention_rate": 0.4461538461538462, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5, "status": 1.0, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 994}, "7": {"facts_total": 65, "facts_retained": 29, "retention_rate": 0.4461538461538462, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5, "status": 1.0, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 994}, "8": {"facts_total": 65, "facts_retained": 29, "retention_rate": 0.4461538461538462, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5, "status": 1.0, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 994}}}, "reference_key_id": "keys/synthetic/syn_03000_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "2dd4d43b9eb5a3a78d9b93791c358e57add647eceb6c72336d92751acf94f66d", "output_hash": "dbedb6f5d00e413d915513b85d8ed195b18b9acf392ece342a85e15e3752bcc8"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "3f2759e786a684bced99e510", "attempt_id": "ba449448851b", "run_group_id": "E4_core:A4|repairable|qwen3b:h8", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "d83aa60f20d2879d02898cb9d9d8bd1983ed2d8984cae017c9153cafee3a779a", "git_commit": null, "timestamp_utc": "2026-10-30T06:27:57.891934Z", "task_family": "synthetic", "document_id": "syn_03000_2000", "document_hash": "c9cc27c20b05b1f6e592846e84990ee9b69c42f9720a8e338ec2f6fca777fdbc", "question_id": "q09_revision", "split": "test", "cluster_id": "syn_03000_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "laptop-m5max-48g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 8, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1904, "query_tokens": 587, "message_tokens": 495, "message_utf8_bytes": 994, "cache_tensor_bytes": null, "serialized_payload_bytes": 994, "side_information_bytes": 0, "fetched_bytes": 8596, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 175923.93000016455, "hops": {"1": 22777.715417090803, "2": 23842.29779103771, "3": 21627.477917005308, "4": 21489.953582989983, "5": 21196.44504203461, "6": 21617.7212910261, "7": 21700.178666971624, "8": 21672.140292008407}, "receiver_prefill_ms": 518.0228750687093, "ttft_ms": 38.335332996211946, "decode_ms": 320.1622919877991}, "total_wall_ms": 176801.20829120278, "ttft_ms": 38.335332996211946, "input_usage": {"lookups": 14}, "output_usage": {"output_tokens": 9}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "9,127,965", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 29, "retention_rate": 0.4461538461538462, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5, "status": 1.0, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 994}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 29, "retention_rate": 0.4461538461538462, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5, "status": 1.0, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1090}, "2": {"facts_total": 65, "facts_retained": 29, "retention_rate": 0.4461538461538462, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5, "status": 1.0, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 994}, "3": {"facts_total": 65, "facts_retained": 29, "retention_rate": 0.4461538461538462, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5, "status": 1.0, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 994}, "4": {"facts_total": 65, "facts_retained": 29, "retention_rate": 0.4461538461538462, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5, "status": 1.0, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 994}, "5": {"facts_total": 65, "facts_retained": 29, "retention_rate": 0.4461538461538462, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5, "status": 1.0, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 994}, "6": {"facts_total": 65, "facts_retained": 29, "retention_rate": 0.4461538461538462, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5, "status": 1.0, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 994}, "7": {"facts_total": 65, "facts_retained": 29, "retention_rate": 0.4461538461538462, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5, "status": 1.0, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 994}, "8": {"facts_total": 65, "facts_retained": 29, "retention_rate": 0.4461538461538462, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5, "status": 1.0, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 994}}}, "reference_key_id": "keys/synthetic/syn_03000_2000.json", "task_scores": {"correct": 0.0, "abstained": 0.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "2dd4d43b9eb5a3a78d9b93791c358e57add647eceb6c72336d92751acf94f66d", "output_hash": "f30700451becb8c76f6e83d5016caa83d25106e6dceeffd8fdeca27984218444"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "964b628aa30a9af97df54848", "attempt_id": "f5cbbf402c1f", "run_group_id": "E4_core:A4|repairable|qwen3b:h8", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "d83aa60f20d2879d02898cb9d9d8bd1983ed2d8984cae017c9153cafee3a779a", "git_commit": null, "timestamp_utc": "2026-10-30T06:27:58.785753Z", "task_family": "synthetic", "document_id": "syn_03000_2000", "document_hash": "c9cc27c20b05b1f6e592846e84990ee9b69c42f9720a8e338ec2f6fca777fdbc", "question_id": "q10_temporal", "split": "test", "cluster_id": "syn_03000_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "laptop-m5max-48g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 8, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1904, "query_tokens": 593, "message_tokens": 495, "message_utf8_bytes": 994, "cache_tensor_bytes": null, "serialized_payload_bytes": 994, "side_information_bytes": 0, "fetched_bytes": 8596, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 175923.93000016455, "hops": {"1": 22777.715417090803, "2": 23842.29779103771, "3": 21627.477917005308, "4": 21489.953582989983, "5": 21196.44504203461, "6": 21617.7212910261, "7": 21700.178666971624, "8": 21672.140292008407}, "receiver_prefill_ms": 383.76033306121826, "ttft_ms": 24.501541978679597, "decode_ms": 315.70604199077934}, "total_wall_ms": 176648.5977912089, "ttft_ms": 24.501541978679597, "input_usage": {"lookups": 14}, "output_usage": {"output_tokens": 5}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "Felaldnak-7", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 29, "retention_rate": 0.4461538461538462, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5, "status": 1.0, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 994}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 29, "retention_rate": 0.4461538461538462, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5, "status": 1.0, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1090}, "2": {"facts_total": 65, "facts_retained": 29, "retention_rate": 0.4461538461538462, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5, "status": 1.0, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 994}, "3": {"facts_total": 65, "facts_retained": 29, "retention_rate": 0.4461538461538462, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5, "status": 1.0, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 994}, "4": {"facts_total": 65, "facts_retained": 29, "retention_rate": 0.4461538461538462, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5, "status": 1.0, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 994}, "5": {"facts_total": 65, "facts_retained": 29, "retention_rate": 0.4461538461538462, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5, "status": 1.0, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 994}, "6": {"facts_total": 65, "facts_retained": 29, "retention_rate": 0.4461538461538462, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5, "status": 1.0, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 994}, "7": {"facts_total": 65, "facts_retained": 29, "retention_rate": 0.4461538461538462, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5, "status": 1.0, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 994}, "8": {"facts_total": 65, "facts_retained": 29, "retention_rate": 0.4461538461538462, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5, "status": 1.0, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 994}}}, "reference_key_id": "keys/synthetic/syn_03000_2000.json", "task_scores": {"correct": 0.0, "abstained": 0.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "2dd4d43b9eb5a3a78d9b93791c358e57add647eceb6c72336d92751acf94f66d", "output_hash": "2b9a4cf7221f1b6f31508a6b3078ec521f211026a932da5020cddd512ecc5fa2"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "77277990b02d059e0af9e1f5", "attempt_id": "f0fb3732bf21", "run_group_id": "E4_core:A4|repairable|qwen3b:h8", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "d83aa60f20d2879d02898cb9d9d8bd1983ed2d8984cae017c9153cafee3a779a", "git_commit": null, "timestamp_utc": "2026-10-30T06:27:59.527059Z", "task_family": "synthetic", "document_id": "syn_03000_2000", "document_hash": "c9cc27c20b05b1f6e592846e84990ee9b69c42f9720a8e338ec2f6fca777fdbc", "question_id": "q11_aggregation", "split": "test", "cluster_id": "syn_03000_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "laptop-m5max-48g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 8, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1904, "query_tokens": 588, "message_tokens": 495, "message_utf8_bytes": 994, "cache_tensor_bytes": null, "serialized_payload_bytes": 994, "side_information_bytes": 0, "fetched_bytes": 8596, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 175923.93000016455, "hops": {"1": 22777.715417090803, "2": 23842.29779103771, "3": 21627.477917005308, "4": 21489.953582989983, "5": 21196.44504203461, "6": 21617.7212910261, "7": 21700.178666971624, "8": 21672.140292008407}, "receiver_prefill_ms": 524.302874924615, "ttft_ms": 41.09566693659872, "decode_ms": 214.2408749787137}, "total_wall_ms": 176704.2688751826, "ttft_ms": 41.09566693659872, "input_usage": {"lookups": 14}, "output_usage": {"output_tokens": 7}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "9220745", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 29, "retention_rate": 0.4461538461538462, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5, "status": 1.0, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 994}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 29, "retention_rate": 0.4461538461538462, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5, "status": 1.0, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1090}, "2": {"facts_total": 65, "facts_retained": 29, "retention_rate": 0.4461538461538462, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5, "status": 1.0, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 994}, "3": {"facts_total": 65, "facts_retained": 29, "retention_rate": 0.4461538461538462, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5, "status": 1.0, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 994}, "4": {"facts_total": 65, "facts_retained": 29, "retention_rate": 0.4461538461538462, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5, "status": 1.0, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 994}, "5": {"facts_total": 65, "facts_retained": 29, "retention_rate": 0.4461538461538462, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5, "status": 1.0, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 994}, "6": {"facts_total": 65, "facts_retained": 29, "retention_rate": 0.4461538461538462, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5, "status": 1.0, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 994}, "7": {"facts_total": 65, "facts_retained": 29, "retention_rate": 0.4461538461538462, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5, "status": 1.0, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 994}, "8": {"facts_total": 65, "facts_retained": 29, "retention_rate": 0.4461538461538462, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5, "status": 1.0, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 994}}}, "reference_key_id": "keys/synthetic/syn_03000_2000.json", "task_scores": {"correct": 0.0, "abstained": 0.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "2dd4d43b9eb5a3a78d9b93791c358e57add647eceb6c72336d92751acf94f66d", "output_hash": "f5d77aa763f6365bc80d01b9e6e23bc43fb570ac45771d7bd628c75cde727b35"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "f7fcdc52a2ae2322b95a6601", "attempt_id": "794a1d9ecb25", "run_group_id": "E4_core:A4|repairable|qwen3b:h8", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "d83aa60f20d2879d02898cb9d9d8bd1983ed2d8984cae017c9153cafee3a779a", "git_commit": null, "timestamp_utc": "2026-10-30T06:28:00.324020Z", "task_family": "synthetic", "document_id": "syn_03000_2000", "document_hash": "c9cc27c20b05b1f6e592846e84990ee9b69c42f9720a8e338ec2f6fca777fdbc", "question_id": "q12_unit_conversion", "split": "test", "cluster_id": "syn_03000_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "laptop-m5max-48g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 8, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1904, "query_tokens": 584, "message_tokens": 495, "message_utf8_bytes": 994, "cache_tensor_bytes": null, "serialized_payload_bytes": 994, "side_information_bytes": 0, "fetched_bytes": 8596, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 175923.93000016455, "hops": {"1": 22777.715417090803, "2": 23842.29779103771, "3": 21627.477917005308, "4": 21489.953582989983, "5": 21196.44504203461, "6": 21617.7212910261, "7": 21700.178666971624, "8": 21672.140292008407}, "receiver_prefill_ms": 382.3732500895858, "ttft_ms": 33.31245796289295, "decode_ms": 23.03312497679144}, "total_wall_ms": 176363.2313752314, "ttft_ms": 33.31245796289295, "input_usage": {"lookups": 14}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 29, "retention_rate": 0.4461538461538462, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5, "status": 1.0, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 994}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 29, "retention_rate": 0.4461538461538462, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5, "status": 1.0, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1090}, "2": {"facts_total": 65, "facts_retained": 29, "retention_rate": 0.4461538461538462, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5, "status": 1.0, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 994}, "3": {"facts_total": 65, "facts_retained": 29, "retention_rate": 0.4461538461538462, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5, "status": 1.0, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 994}, "4": {"facts_total": 65, "facts_retained": 29, "retention_rate": 0.4461538461538462, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5, "status": 1.0, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 994}, "5": {"facts_total": 65, "facts_retained": 29, "retention_rate": 0.4461538461538462, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5, "status": 1.0, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 994}, "6": {"facts_total": 65, "facts_retained": 29, "retention_rate": 0.4461538461538462, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5, "status": 1.0, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 994}, "7": {"facts_total": 65, "facts_retained": 29, "retention_rate": 0.4461538461538462, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5, "status": 1.0, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 994}, "8": {"facts_total": 65, "facts_retained": 29, "retention_rate": 0.4461538461538462, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5, "status": 1.0, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 994}}}, "reference_key_id": "keys/synthetic/syn_03000_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "2dd4d43b9eb5a3a78d9b93791c358e57add647eceb6c72336d92751acf94f66d", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "75b05847ca56486197d621d6", "attempt_id": "a38203fa04ba", "run_group_id": "E4_core:A4|repairable|qwen3b:h8", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "d83aa60f20d2879d02898cb9d9d8bd1983ed2d8984cae017c9153cafee3a779a", "git_commit": null, "timestamp_utc": "2026-10-30T06:28:00.780401Z", "task_family": "synthetic", "document_id": "syn_03000_2000", "document_hash": "c9cc27c20b05b1f6e592846e84990ee9b69c42f9720a8e338ec2f6fca777fdbc", "question_id": "q13_unanswerable", "split": "test", "cluster_id": "syn_03000_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "laptop-m5max-48g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 8, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1904, "query_tokens": 588, "message_tokens": 495, "message_utf8_bytes": 994, "cache_tensor_bytes": null, "serialized_payload_bytes": 994, "side_information_bytes": 0, "fetched_bytes": 8596, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 175923.93000016455, "hops": {"1": 22777.715417090803, "2": 23842.29779103771, "3": 21627.477917005308, "4": 21489.953582989983, "5": 21196.44504203461, "6": 21617.7212910261, "7": 21700.178666971624, "8": 21672.140292008407}, "receiver_prefill_ms": 374.1677920334041, "ttft_ms": 54.71191694959998, "decode_ms": 40.47029197681695}, "total_wall_ms": 176394.13358317688, "ttft_ms": 54.71191694959998, "input_usage": {"lookups": 14}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 29, "retention_rate": 0.4461538461538462, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5, "status": 1.0, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 994}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 29, "retention_rate": 0.4461538461538462, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5, "status": 1.0, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1090}, "2": {"facts_total": 65, "facts_retained": 29, "retention_rate": 0.4461538461538462, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5, "status": 1.0, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 994}, "3": {"facts_total": 65, "facts_retained": 29, "retention_rate": 0.4461538461538462, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5, "status": 1.0, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 994}, "4": {"facts_total": 65, "facts_retained": 29, "retention_rate": 0.4461538461538462, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5, "status": 1.0, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 994}, "5": {"facts_total": 65, "facts_retained": 29, "retention_rate": 0.4461538461538462, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5, "status": 1.0, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 994}, "6": {"facts_total": 65, "facts_retained": 29, "retention_rate": 0.4461538461538462, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5, "status": 1.0, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 994}, "7": {"facts_total": 65, "facts_retained": 29, "retention_rate": 0.4461538461538462, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5, "status": 1.0, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 994}, "8": {"facts_total": 65, "facts_retained": 29, "retention_rate": 0.4461538461538462, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5, "status": 1.0, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 994}}}, "reference_key_id": "keys/synthetic/syn_03000_2000.json", "task_scores": {"correct": 1.0, "abstained": 1.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "2dd4d43b9eb5a3a78d9b93791c358e57add647eceb6c72336d92751acf94f66d", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "66ad4b0e4f6b92a185d79045", "attempt_id": "e16036111d13", "run_group_id": "E4_core:A4|repairable|qwen3b:h8", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "d83aa60f20d2879d02898cb9d9d8bd1983ed2d8984cae017c9153cafee3a779a", "git_commit": null, "timestamp_utc": "2026-10-30T06:28:01.267520Z", "task_family": "synthetic", "document_id": "syn_03000_2000", "document_hash": "c9cc27c20b05b1f6e592846e84990ee9b69c42f9720a8e338ec2f6fca777fdbc", "question_id": "q14_unanswerable", "split": "test", "cluster_id": "syn_03000_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "laptop-m5max-48g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 8, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1904, "query_tokens": 590, "message_tokens": 495, "message_utf8_bytes": 994, "cache_tensor_bytes": null, "serialized_payload_bytes": 994, "side_information_bytes": 0, "fetched_bytes": 8596, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 175923.93000016455, "hops": {"1": 22777.715417090803, "2": 23842.29779103771, "3": 21627.477917005308, "4": 21489.953582989983, "5": 21196.44504203461, "6": 21617.7212910261, "7": 21700.178666971624, "8": 21672.140292008407}, "receiver_prefill_ms": 522.8634590748698, "ttft_ms": 28.145499993115664, "decode_ms": 41.9800829840824}, "total_wall_ms": 176517.68437516876, "ttft_ms": 28.145499993115664, "input_usage": {"lookups": 14}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 29, "retention_rate": 0.4461538461538462, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5, "status": 1.0, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 994}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 29, "retention_rate": 0.4461538461538462, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5, "status": 1.0, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1090}, "2": {"facts_total": 65, "facts_retained": 29, "retention_rate": 0.4461538461538462, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5, "status": 1.0, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 994}, "3": {"facts_total": 65, "facts_retained": 29, "retention_rate": 0.4461538461538462, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5, "status": 1.0, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 994}, "4": {"facts_total": 65, "facts_retained": 29, "retention_rate": 0.4461538461538462, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5, "status": 1.0, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 994}, "5": {"facts_total": 65, "facts_retained": 29, "retention_rate": 0.4461538461538462, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5, "status": 1.0, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 994}, "6": {"facts_total": 65, "facts_retained": 29, "retention_rate": 0.4461538461538462, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5, "status": 1.0, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 994}, "7": {"facts_total": 65, "facts_retained": 29, "retention_rate": 0.4461538461538462, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5, "status": 1.0, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 994}, "8": {"facts_total": 65, "facts_retained": 29, "retention_rate": 0.4461538461538462, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5, "status": 1.0, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 994}}}, "reference_key_id": "keys/synthetic/syn_03000_2000.json", "task_scores": {"correct": 1.0, "abstained": 1.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "2dd4d43b9eb5a3a78d9b93791c358e57add647eceb6c72336d92751acf94f66d", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "0d1f3cddc5937d87c1f780bf", "attempt_id": "9b426e2add2c", "run_group_id": "E4_core:A5|repairable|qwen3b:h1", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T06:28:46.199104Z", "task_family": "synthetic", "document_id": "syn_03000_2000", "document_hash": "c9cc27c20b05b1f6e592846e84990ee9b69c42f9720a8e338ec2f6fca777fdbc", "question_id": "q00_single_fact", "split": "test", "cluster_id": "syn_03000_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 1, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1904, "query_tokens": 464, "message_tokens": 383, "message_utf8_bytes": 911, "cache_tensor_bytes": null, "serialized_payload_bytes": 911, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 22956.80054195691, "hops": {"1": 22956.80054195691}, "receiver_prefill_ms": 322.1704170573503, "ttft_ms": 38.84220798499882, "decode_ms": 23.477000067941844}, "total_wall_ms": 23341.871125041507, "ttft_ms": 38.84220798499882, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 15, "retention_rate": 0.23076923076923078, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.09090909090909091, "amount": 0.16666666666666666, "status": 0.5, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 855}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 15, "retention_rate": 0.23076923076923078, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.09090909090909091, "amount": 0.16666666666666666, "status": 0.5, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 855}}}, "reference_key_id": "keys/synthetic/syn_03000_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "a409b88ff601f1f9bf4d221a0c042117fadd2be19621e0b0dd2800495cfbe0cc", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "3efacfec20b92046320d9704", "attempt_id": "d1246842f214", "run_group_id": "E4_core:A5|repairable|qwen3b:h1", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T06:28:46.601868Z", "task_family": "synthetic", "document_id": "syn_03000_2000", "document_hash": "c9cc27c20b05b1f6e592846e84990ee9b69c42f9720a8e338ec2f6fca777fdbc", "question_id": "q01_single_fact", "split": "test", "cluster_id": "syn_03000_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 1, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1904, "query_tokens": 469, "message_tokens": 383, "message_utf8_bytes": 911, "cache_tensor_bytes": null, "serialized_payload_bytes": 911, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 22956.80054195691, "hops": {"1": 22956.80054195691}, "receiver_prefill_ms": 316.71141600236297, "ttft_ms": 20.78937494661659, "decode_ms": 615.7680839532986}, "total_wall_ms": 23910.678916960023, "ttft_ms": 20.78937494661659, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 11}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "Lot 409-H in Senneterre", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 15, "retention_rate": 0.23076923076923078, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.09090909090909091, "amount": 0.16666666666666666, "status": 0.5, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 855}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 15, "retention_rate": 0.23076923076923078, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.09090909090909091, "amount": 0.16666666666666666, "status": 0.5, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 855}}}, "reference_key_id": "keys/synthetic/syn_03000_2000.json", "task_scores": {"correct": 0.0, "abstained": 0.0, "f1": 0.5}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "a409b88ff601f1f9bf4d221a0c042117fadd2be19621e0b0dd2800495cfbe0cc", "output_hash": "928c825f0fee3b69cf0901e3721b2d6c4467e52d2c6707afe17dd750a93d5e31"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "b040e2ca527cd6a2315c203b", "attempt_id": "59bf11082cc2", "run_group_id": "E4_core:A5|repairable|qwen3b:h1", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T06:28:47.572662Z", "task_family": "synthetic", "document_id": "syn_03000_2000", "document_hash": "c9cc27c20b05b1f6e592846e84990ee9b69c42f9720a8e338ec2f6fca777fdbc", "question_id": "q02_exact_copy", "split": "test", "cluster_id": "syn_03000_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 1, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1904, "query_tokens": 468, "message_tokens": 383, "message_utf8_bytes": 911, "cache_tensor_bytes": null, "serialized_payload_bytes": 911, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 22956.80054195691, "hops": {"1": 22956.80054195691}, "receiver_prefill_ms": 319.70816594548523, "ttft_ms": 37.81212505418807, "decode_ms": 34.805333940312266}, "total_wall_ms": 23349.59745791275, "ttft_ms": 37.81212505418807, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 15, "retention_rate": 0.23076923076923078, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.09090909090909091, "amount": 0.16666666666666666, "status": 0.5, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 855}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 15, "retention_rate": 0.23076923076923078, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.09090909090909091, "amount": 0.16666666666666666, "status": 0.5, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 855}}}, "reference_key_id": "keys/synthetic/syn_03000_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "a409b88ff601f1f9bf4d221a0c042117fadd2be19621e0b0dd2800495cfbe0cc", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "5d398b6e88ba28e6888b1c10", "attempt_id": "038cf6b5a062", "run_group_id": "E4_core:A5|repairable|qwen3b:h1", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T06:28:47.981297Z", "task_family": "synthetic", "document_id": "syn_03000_2000", "document_hash": "c9cc27c20b05b1f6e592846e84990ee9b69c42f9720a8e338ec2f6fca777fdbc", "question_id": "q03_exact_copy", "split": "test", "cluster_id": "syn_03000_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 1, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1904, "query_tokens": 467, "message_tokens": 383, "message_utf8_bytes": 911, "cache_tensor_bytes": null, "serialized_payload_bytes": 911, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 22956.80054195691, "hops": {"1": 22956.80054195691}, "receiver_prefill_ms": 327.0198750542477, "ttft_ms": 24.933415930718184, "decode_ms": 31.909666024148464}, "total_wall_ms": 23341.242874972522, "ttft_ms": 24.933415930718184, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 15, "retention_rate": 0.23076923076923078, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.09090909090909091, "amount": 0.16666666666666666, "status": 0.5, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 855}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 15, "retention_rate": 0.23076923076923078, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.09090909090909091, "amount": 0.16666666666666666, "status": 0.5, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 855}}}, "reference_key_id": "keys/synthetic/syn_03000_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "a409b88ff601f1f9bf4d221a0c042117fadd2be19621e0b0dd2800495cfbe0cc", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "61fefe56f727535f4e8c9c8c", "attempt_id": "4697bb2bcbaa", "run_group_id": "E4_core:A5|repairable|qwen3b:h1", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T06:28:48.381859Z", "task_family": "synthetic", "document_id": "syn_03000_2000", "document_hash": "c9cc27c20b05b1f6e592846e84990ee9b69c42f9720a8e338ec2f6fca777fdbc", "question_id": "q04_multi_hop", "split": "test", "cluster_id": "syn_03000_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 1, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1904, "query_tokens": 473, "message_tokens": 383, "message_utf8_bytes": 911, "cache_tensor_bytes": null, "serialized_payload_bytes": 911, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 22956.80054195691, "hops": {"1": 22956.80054195691}, "receiver_prefill_ms": 303.0640829820186, "ttft_ms": 61.40124995727092, "decode_ms": 46.89770797267556}, "total_wall_ms": 23368.970416951925, "ttft_ms": 61.40124995727092, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 15, "retention_rate": 0.23076923076923078, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.09090909090909091, "amount": 0.16666666666666666, "status": 0.5, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 855}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 15, "retention_rate": 0.23076923076923078, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.09090909090909091, "amount": 0.16666666666666666, "status": 0.5, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 855}}}, "reference_key_id": "keys/synthetic/syn_03000_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "a409b88ff601f1f9bf4d221a0c042117fadd2be19621e0b0dd2800495cfbe0cc", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "f4a38f0e480c8e9674d77a15", "attempt_id": "2a17822588c3", "run_group_id": "E4_core:A5|repairable|qwen3b:h1", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T06:28:48.810021Z", "task_family": "synthetic", "document_id": "syn_03000_2000", "document_hash": "c9cc27c20b05b1f6e592846e84990ee9b69c42f9720a8e338ec2f6fca777fdbc", "question_id": "q05_multi_hop", "split": "test", "cluster_id": "syn_03000_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 1, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1904, "query_tokens": 475, "message_tokens": 383, "message_utf8_bytes": 911, "cache_tensor_bytes": null, "serialized_payload_bytes": 911, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 22956.80054195691, "hops": {"1": 22956.80054195691}, "receiver_prefill_ms": 377.7105830376968, "ttft_ms": 98.53108401875943, "decode_ms": 356.0785409063101}, "total_wall_ms": 23789.694666978903, "ttft_ms": 98.53108401875943, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 10}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "LY-51E-3199", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 15, "retention_rate": 0.23076923076923078, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.09090909090909091, "amount": 0.16666666666666666, "status": 0.5, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 855}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 15, "retention_rate": 0.23076923076923078, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.09090909090909091, "amount": 0.16666666666666666, "status": 0.5, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 855}}}, "reference_key_id": "keys/synthetic/syn_03000_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "a409b88ff601f1f9bf4d221a0c042117fadd2be19621e0b0dd2800495cfbe0cc", "output_hash": "a1245d1536890a7e4b026366df48441d503f491f782cfd388797b5782443c5c7"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "3e6fb1644c8fcc28e3b8fa9b", "attempt_id": "94aa62f04202", "run_group_id": "E4_core:A5|repairable|qwen3b:h1", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T06:28:49.658830Z", "task_family": "synthetic", "document_id": "syn_03000_2000", "document_hash": "c9cc27c20b05b1f6e592846e84990ee9b69c42f9720a8e338ec2f6fca777fdbc", "question_id": "q06_negation", "split": "test", "cluster_id": "syn_03000_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 1, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1904, "query_tokens": 464, "message_tokens": 383, "message_utf8_bytes": 911, "cache_tensor_bytes": null, "serialized_payload_bytes": 911, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 22956.80054195691, "hops": {"1": 22956.80054195691}, "receiver_prefill_ms": 303.71900007594377, "ttft_ms": 35.83454200997949, "decode_ms": 39.247042033821344}, "total_wall_ms": 23336.069333949126, "ttft_ms": 35.83454200997949, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "no", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 15, "retention_rate": 0.23076923076923078, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.09090909090909091, "amount": 0.16666666666666666, "status": 0.5, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 855}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 15, "retention_rate": 0.23076923076923078, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.09090909090909091, "amount": 0.16666666666666666, "status": 0.5, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 855}}}, "reference_key_id": "keys/synthetic/syn_03000_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "a409b88ff601f1f9bf4d221a0c042117fadd2be19621e0b0dd2800495cfbe0cc", "output_hash": "9390298f3fb0c5b160498935d79cb139aef28e1c47358b4bbba61862b9c26e59"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "5f53a06ed5a91048c4226a0e", "attempt_id": "58fe233a5893", "run_group_id": "E4_core:A5|repairable|qwen3b:h1", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T06:28:50.053792Z", "task_family": "synthetic", "document_id": "syn_03000_2000", "document_hash": "c9cc27c20b05b1f6e592846e84990ee9b69c42f9720a8e338ec2f6fca777fdbc", "question_id": "q07_negation", "split": "test", "cluster_id": "syn_03000_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 1, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1904, "query_tokens": 464, "message_tokens": 383, "message_utf8_bytes": 911, "cache_tensor_bytes": null, "serialized_payload_bytes": 911, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 22956.80054195691, "hops": {"1": 22956.80054195691}, "receiver_prefill_ms": 280.9419580735266, "ttft_ms": 54.670415935106575, "decode_ms": 53.11999993864447}, "total_wall_ms": 23346.06816689484, "ttft_ms": 54.670415935106575, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 15, "retention_rate": 0.23076923076923078, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.09090909090909091, "amount": 0.16666666666666666, "status": 0.5, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 855}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 15, "retention_rate": 0.23076923076923078, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.09090909090909091, "amount": 0.16666666666666666, "status": 0.5, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 855}}}, "reference_key_id": "keys/synthetic/syn_03000_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "a409b88ff601f1f9bf4d221a0c042117fadd2be19621e0b0dd2800495cfbe0cc", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "dfa5987c064480080bc710af", "attempt_id": "cd24c0733fdf", "run_group_id": "E4_core:A5|repairable|qwen3b:h1", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T06:28:50.459458Z", "task_family": "synthetic", "document_id": "syn_03000_2000", "document_hash": "c9cc27c20b05b1f6e592846e84990ee9b69c42f9720a8e338ec2f6fca777fdbc", "question_id": "q08_revision", "split": "test", "cluster_id": "syn_03000_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 1, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1904, "query_tokens": 479, "message_tokens": 383, "message_utf8_bytes": 911, "cache_tensor_bytes": null, "serialized_payload_bytes": 911, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 22956.80054195691, "hops": {"1": 22956.80054195691}, "receiver_prefill_ms": 408.42820797115564, "ttft_ms": 72.48904195148498, "decode_ms": 328.50437506567687}, "total_wall_ms": 23766.897874884307, "ttft_ms": 72.48904195148498, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 9}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "2,705,965", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 15, "retention_rate": 0.23076923076923078, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.09090909090909091, "amount": 0.16666666666666666, "status": 0.5, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 855}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 15, "retention_rate": 0.23076923076923078, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.09090909090909091, "amount": 0.16666666666666666, "status": 0.5, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 855}}}, "reference_key_id": "keys/synthetic/syn_03000_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "a409b88ff601f1f9bf4d221a0c042117fadd2be19621e0b0dd2800495cfbe0cc", "output_hash": "dbedb6f5d00e413d915513b85d8ed195b18b9acf392ece342a85e15e3752bcc8"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "41a2144913bdb0a55651b7c0", "attempt_id": "2a9231d45f53", "run_group_id": "E4_core:A5|repairable|qwen3b:h1", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T06:28:51.285569Z", "task_family": "synthetic", "document_id": "syn_03000_2000", "document_hash": "c9cc27c20b05b1f6e592846e84990ee9b69c42f9720a8e338ec2f6fca777fdbc", "question_id": "q09_revision", "split": "test", "cluster_id": "syn_03000_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 1, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1904, "query_tokens": 475, "message_tokens": 383, "message_utf8_bytes": 911, "cache_tensor_bytes": null, "serialized_payload_bytes": 911, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 22956.80054195691, "hops": {"1": 22956.80054195691}, "receiver_prefill_ms": 321.7267079744488, "ttft_ms": 27.581791975535452, "decode_ms": 37.87833300884813}, "total_wall_ms": 23344.44041689858, "ttft_ms": 27.581791975535452, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 15, "retention_rate": 0.23076923076923078, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.09090909090909091, "amount": 0.16666666666666666, "status": 0.5, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 855}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 15, "retention_rate": 0.23076923076923078, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.09090909090909091, "amount": 0.16666666666666666, "status": 0.5, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 855}}}, "reference_key_id": "keys/synthetic/syn_03000_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "a409b88ff601f1f9bf4d221a0c042117fadd2be19621e0b0dd2800495cfbe0cc", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "17f7991bae18d15b9eb1c3f5", "attempt_id": "8818e5c9e8ac", "run_group_id": "E4_core:A5|repairable|qwen3b:h1", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T06:28:51.689500Z", "task_family": "synthetic", "document_id": "syn_03000_2000", "document_hash": "c9cc27c20b05b1f6e592846e84990ee9b69c42f9720a8e338ec2f6fca777fdbc", "question_id": "q10_temporal", "split": "test", "cluster_id": "syn_03000_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 1, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1904, "query_tokens": 481, "message_tokens": 383, "message_utf8_bytes": 911, "cache_tensor_bytes": null, "serialized_payload_bytes": 911, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 22956.80054195691, "hops": {"1": 22956.80054195691}, "receiver_prefill_ms": 332.1241249796003, "ttft_ms": 45.18970905337483, "decode_ms": 68.98341607302427}, "total_wall_ms": 23403.6968749715, "ttft_ms": 45.18970905337483, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 15, "retention_rate": 0.23076923076923078, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.09090909090909091, "amount": 0.16666666666666666, "status": 0.5, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 855}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 15, "retention_rate": 0.23076923076923078, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.09090909090909091, "amount": 0.16666666666666666, "status": 0.5, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 855}}}, "reference_key_id": "keys/synthetic/syn_03000_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "a409b88ff601f1f9bf4d221a0c042117fadd2be19621e0b0dd2800495cfbe0cc", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "4a56ad784864770ec9a9984d", "attempt_id": "fa8f437e1967", "run_group_id": "E4_core:A5|repairable|qwen3b:h1", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T06:28:52.152356Z", "task_family": "synthetic", "document_id": "syn_03000_2000", "document_hash": "c9cc27c20b05b1f6e592846e84990ee9b69c42f9720a8e338ec2f6fca777fdbc", "question_id": "q11_aggregation", "split": "test", "cluster_id": "syn_03000_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 1, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1904, "query_tokens": 476, "message_tokens": 383, "message_utf8_bytes": 911, "cache_tensor_bytes": null, "serialized_payload_bytes": 911, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 22956.80054195691, "hops": {"1": 22956.80054195691}, "receiver_prefill_ms": 487.56429203785956, "ttft_ms": 34.771583043038845, "decode_ms": 31.820125062949955}, "total_wall_ms": 23511.552916956134, "ttft_ms": 34.771583043038845, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 15, "retention_rate": 0.23076923076923078, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.09090909090909091, "amount": 0.16666666666666666, "status": 0.5, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 855}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 15, "retention_rate": 0.23076923076923078, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.09090909090909091, "amount": 0.16666666666666666, "status": 0.5, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 855}}}, "reference_key_id": "keys/synthetic/syn_03000_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "a409b88ff601f1f9bf4d221a0c042117fadd2be19621e0b0dd2800495cfbe0cc", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "4a77669ec375456c84410b25", "attempt_id": "37c82d9328c5", "run_group_id": "E4_core:A5|repairable|qwen3b:h1", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T06:28:52.723867Z", "task_family": "synthetic", "document_id": "syn_03000_2000", "document_hash": "c9cc27c20b05b1f6e592846e84990ee9b69c42f9720a8e338ec2f6fca777fdbc", "question_id": "q12_unit_conversion", "split": "test", "cluster_id": "syn_03000_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 1, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1904, "query_tokens": 472, "message_tokens": 383, "message_utf8_bytes": 911, "cache_tensor_bytes": null, "serialized_payload_bytes": 911, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 22956.80054195691, "hops": {"1": 22956.80054195691}, "receiver_prefill_ms": 317.66879197675735, "ttft_ms": 33.811124972999096, "decode_ms": 176.32145795505494}, "total_wall_ms": 23485.221791896038, "ttft_ms": 33.811124972999096, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 5}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "20.46", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 15, "retention_rate": 0.23076923076923078, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.09090909090909091, "amount": 0.16666666666666666, "status": 0.5, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 855}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 15, "retention_rate": 0.23076923076923078, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.09090909090909091, "amount": 0.16666666666666666, "status": 0.5, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 855}}}, "reference_key_id": "keys/synthetic/syn_03000_2000.json", "task_scores": {"correct": 0.0, "abstained": 0.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "a409b88ff601f1f9bf4d221a0c042117fadd2be19621e0b0dd2800495cfbe0cc", "output_hash": "c057e11a5e818ca86a03c336373f3b4e4a2ddd9756e0acd7f5d7d655fd0267fe"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "9b122fb122f0db1c0f1a1f6e", "attempt_id": "b25b5f962c3d", "run_group_id": "E4_core:A5|repairable|qwen3b:h1", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T06:28:53.268196Z", "task_family": "synthetic", "document_id": "syn_03000_2000", "document_hash": "c9cc27c20b05b1f6e592846e84990ee9b69c42f9720a8e338ec2f6fca777fdbc", "question_id": "q13_unanswerable", "split": "test", "cluster_id": "syn_03000_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 1, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1904, "query_tokens": 476, "message_tokens": 383, "message_utf8_bytes": 911, "cache_tensor_bytes": null, "serialized_payload_bytes": 911, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 22956.80054195691, "hops": {"1": 22956.80054195691}, "receiver_prefill_ms": 270.20608400925994, "ttft_ms": 62.81808391213417, "decode_ms": 78.54066707659513}, "total_wall_ms": 23369.04370796401, "ttft_ms": 62.81808391213417, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 15, "retention_rate": 0.23076923076923078, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.09090909090909091, "amount": 0.16666666666666666, "status": 0.5, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 855}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 15, "retention_rate": 0.23076923076923078, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.09090909090909091, "amount": 0.16666666666666666, "status": 0.5, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 855}}}, "reference_key_id": "keys/synthetic/syn_03000_2000.json", "task_scores": {"correct": 1.0, "abstained": 1.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "a409b88ff601f1f9bf4d221a0c042117fadd2be19621e0b0dd2800495cfbe0cc", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "e5b7c8856afc60ba817636b1", "attempt_id": "342bdd4c70f2", "run_group_id": "E4_core:A5|repairable|qwen3b:h1", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T06:28:53.696015Z", "task_family": "synthetic", "document_id": "syn_03000_2000", "document_hash": "c9cc27c20b05b1f6e592846e84990ee9b69c42f9720a8e338ec2f6fca777fdbc", "question_id": "q14_unanswerable", "split": "test", "cluster_id": "syn_03000_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 1, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1904, "query_tokens": 478, "message_tokens": 383, "message_utf8_bytes": 911, "cache_tensor_bytes": null, "serialized_payload_bytes": 911, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 22956.80054195691, "hops": {"1": 22956.80054195691}, "receiver_prefill_ms": 348.921624943614, "ttft_ms": 127.63529201038182, "decode_ms": 42.51237504649907}, "total_wall_ms": 23476.494541973807, "ttft_ms": 127.63529201038182, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 15, "retention_rate": 0.23076923076923078, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.09090909090909091, "amount": 0.16666666666666666, "status": 0.5, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 855}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 15, "retention_rate": 0.23076923076923078, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.09090909090909091, "amount": 0.16666666666666666, "status": 0.5, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 855}}}, "reference_key_id": "keys/synthetic/syn_03000_2000.json", "task_scores": {"correct": 1.0, "abstained": 1.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "a409b88ff601f1f9bf4d221a0c042117fadd2be19621e0b0dd2800495cfbe0cc", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "beb6c3c41bb200aaf965a5a0", "attempt_id": "d6e3c8e0a833", "run_group_id": "E4_core:A5|repairable|qwen3b:h2", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T06:29:10.068111Z", "task_family": "synthetic", "document_id": "syn_03000_2000", "document_hash": "c9cc27c20b05b1f6e592846e84990ee9b69c42f9720a8e338ec2f6fca777fdbc", "question_id": "q00_single_fact", "split": "test", "cluster_id": "syn_03000_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 2, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1904, "query_tokens": 444, "message_tokens": 363, "message_utf8_bytes": 848, "cache_tensor_bytes": null, "serialized_payload_bytes": 848, "side_information_bytes": 0, "fetched_bytes": 1228, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 38790.371541981585, "hops": {"1": 22956.80054195691, "2": 15833.571000024676}, "receiver_prefill_ms": 293.3926669647917, "ttft_ms": 46.14774999208748, "decode_ms": 66.46812497638166}, "total_wall_ms": 39197.05333397724, "ttft_ms": 46.14774999208748, "input_usage": {"lookups": 2}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 14, "retention_rate": 0.2153846153846154, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.09090909090909091, "amount": 0.16666666666666666, "status": 0.5, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 786}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 15, "retention_rate": 0.23076923076923078, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.09090909090909091, "amount": 0.16666666666666666, "status": 0.5, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 855}, "2": {"facts_total": 65, "facts_retained": 14, "retention_rate": 0.2153846153846154, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.09090909090909091, "amount": 0.16666666666666666, "status": 0.5, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 786}}}, "reference_key_id": "keys/synthetic/syn_03000_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "ffaf2ec78fb536984dc613873a800b081c1318d5d46ca5c4de575785729d8cbb", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "64117cc4d853e68aaf4781b5", "attempt_id": "e8bc9791e896", "run_group_id": "E4_core:A5|repairable|qwen3b:h2", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T06:29:10.493862Z", "task_family": "synthetic", "document_id": "syn_03000_2000", "document_hash": "c9cc27c20b05b1f6e592846e84990ee9b69c42f9720a8e338ec2f6fca777fdbc", "question_id": "q01_single_fact", "split": "test", "cluster_id": "syn_03000_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 2, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1904, "query_tokens": 449, "message_tokens": 363, "message_utf8_bytes": 848, "cache_tensor_bytes": null, "serialized_payload_bytes": 848, "side_information_bytes": 0, "fetched_bytes": 1228, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 38790.371541981585, "hops": {"1": 22956.80054195691, "2": 15833.571000024676}, "receiver_prefill_ms": 417.81545896083117, "ttft_ms": 26.855416013859212, "decode_ms": 387.93874997645617}, "total_wall_ms": 39623.629292007536, "ttft_ms": 26.855416013859212, "input_usage": {"lookups": 2}, "output_usage": {"output_tokens": 11}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "Lot 409-H in Senneterre", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 14, "retention_rate": 0.2153846153846154, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.09090909090909091, "amount": 0.16666666666666666, "status": 0.5, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 786}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 15, "retention_rate": 0.23076923076923078, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.09090909090909091, "amount": 0.16666666666666666, "status": 0.5, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 855}, "2": {"facts_total": 65, "facts_retained": 14, "retention_rate": 0.2153846153846154, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.09090909090909091, "amount": 0.16666666666666666, "status": 0.5, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 786}}}, "reference_key_id": "keys/synthetic/syn_03000_2000.json", "task_scores": {"correct": 0.0, "abstained": 0.0, "f1": 0.5}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "ffaf2ec78fb536984dc613873a800b081c1318d5d46ca5c4de575785729d8cbb", "output_hash": "928c825f0fee3b69cf0901e3721b2d6c4467e52d2c6707afe17dd750a93d5e31"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "f987fcbc8a8854586f6c8183", "attempt_id": "74a391e13901", "run_group_id": "E4_core:A5|repairable|qwen3b:h2", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T06:29:11.344884Z", "task_family": "synthetic", "document_id": "syn_03000_2000", "document_hash": "c9cc27c20b05b1f6e592846e84990ee9b69c42f9720a8e338ec2f6fca777fdbc", "question_id": "q02_exact_copy", "split": "test", "cluster_id": "syn_03000_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 2, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1904, "query_tokens": 448, "message_tokens": 363, "message_utf8_bytes": 848, "cache_tensor_bytes": null, "serialized_payload_bytes": 848, "side_information_bytes": 0, "fetched_bytes": 1228, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 38790.371541981585, "hops": {"1": 22956.80054195691, "2": 15833.571000024676}, "receiver_prefill_ms": 287.208209047094, "ttft_ms": 15.604250016622245, "decode_ms": 37.10883401799947}, "total_wall_ms": 39130.86799997836, "ttft_ms": 15.604250016622245, "input_usage": {"lookups": 2}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 14, "retention_rate": 0.2153846153846154, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.09090909090909091, "amount": 0.16666666666666666, "status": 0.5, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 786}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 15, "retention_rate": 0.23076923076923078, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.09090909090909091, "amount": 0.16666666666666666, "status": 0.5, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 855}, "2": {"facts_total": 65, "facts_retained": 14, "retention_rate": 0.2153846153846154, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.09090909090909091, "amount": 0.16666666666666666, "status": 0.5, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 786}}}, "reference_key_id": "keys/synthetic/syn_03000_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "ffaf2ec78fb536984dc613873a800b081c1318d5d46ca5c4de575785729d8cbb", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "3f3ecafaa8bebd8226eaac12", "attempt_id": "7508bbdd8eab", "run_group_id": "E4_core:A5|repairable|qwen3b:h2", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T06:29:11.702239Z", "task_family": "synthetic", "document_id": "syn_03000_2000", "document_hash": "c9cc27c20b05b1f6e592846e84990ee9b69c42f9720a8e338ec2f6fca777fdbc", "question_id": "q03_exact_copy", "split": "test", "cluster_id": "syn_03000_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 2, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1904, "query_tokens": 447, "message_tokens": 363, "message_utf8_bytes": 848, "cache_tensor_bytes": null, "serialized_payload_bytes": 848, "side_information_bytes": 0, "fetched_bytes": 1228, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 38790.371541981585, "hops": {"1": 22956.80054195691, "2": 15833.571000024676}, "receiver_prefill_ms": 322.25074991583824, "ttft_ms": 45.34758301451802, "decode_ms": 105.30570906121284}, "total_wall_ms": 39263.872666982934, "ttft_ms": 45.34758301451802, "input_usage": {"lookups": 2}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 14, "retention_rate": 0.2153846153846154, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.09090909090909091, "amount": 0.16666666666666666, "status": 0.5, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 786}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 15, "retention_rate": 0.23076923076923078, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.09090909090909091, "amount": 0.16666666666666666, "status": 0.5, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 855}, "2": {"facts_total": 65, "facts_retained": 14, "retention_rate": 0.2153846153846154, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.09090909090909091, "amount": 0.16666666666666666, "status": 0.5, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 786}}}, "reference_key_id": "keys/synthetic/syn_03000_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "ffaf2ec78fb536984dc613873a800b081c1318d5d46ca5c4de575785729d8cbb", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "f58726a4918660fe3c0fdcbd", "attempt_id": "9ac1d1c89a5d", "run_group_id": "E4_core:A5|repairable|qwen3b:h2", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T06:29:12.194106Z", "task_family": "synthetic", "document_id": "syn_03000_2000", "document_hash": "c9cc27c20b05b1f6e592846e84990ee9b69c42f9720a8e338ec2f6fca777fdbc", "question_id": "q04_multi_hop", "split": "test", "cluster_id": "syn_03000_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 2, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1904, "query_tokens": 453, "message_tokens": 363, "message_utf8_bytes": 848, "cache_tensor_bytes": null, "serialized_payload_bytes": 848, "side_information_bytes": 0, "fetched_bytes": 1228, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 38790.371541981585, "hops": {"1": 22956.80054195691, "2": 15833.571000024676}, "receiver_prefill_ms": 390.9143749624491, "ttft_ms": 39.33720896020532, "decode_ms": 25.839374982751906}, "total_wall_ms": 39247.0855419524, "ttft_ms": 39.33720896020532, "input_usage": {"lookups": 2}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 14, "retention_rate": 0.2153846153846154, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.09090909090909091, "amount": 0.16666666666666666, "status": 0.5, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 786}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 15, "retention_rate": 0.23076923076923078, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.09090909090909091, "amount": 0.16666666666666666, "status": 0.5, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 855}, "2": {"facts_total": 65, "facts_retained": 14, "retention_rate": 0.2153846153846154, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.09090909090909091, "amount": 0.16666666666666666, "status": 0.5, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 786}}}, "reference_key_id": "keys/synthetic/syn_03000_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "ffaf2ec78fb536984dc613873a800b081c1318d5d46ca5c4de575785729d8cbb", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "91134d30b90b88538c495fe5", "attempt_id": "c17a77b79b13", "run_group_id": "E4_core:A5|repairable|qwen3b:h2", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T06:29:12.669329Z", "task_family": "synthetic", "document_id": "syn_03000_2000", "document_hash": "c9cc27c20b05b1f6e592846e84990ee9b69c42f9720a8e338ec2f6fca777fdbc", "question_id": "q05_multi_hop", "split": "test", "cluster_id": "syn_03000_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 2, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1904, "query_tokens": 455, "message_tokens": 363, "message_utf8_bytes": 848, "cache_tensor_bytes": null, "serialized_payload_bytes": 848, "side_information_bytes": 0, "fetched_bytes": 1228, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 38790.371541981585, "hops": {"1": 22956.80054195691, "2": 15833.571000024676}, "receiver_prefill_ms": 321.2504159891978, "ttft_ms": 40.35962501075119, "decode_ms": 319.060750072822}, "total_wall_ms": 39472.09933400154, "ttft_ms": 40.35962501075119, "input_usage": {"lookups": 2}, "output_usage": {"output_tokens": 10}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "LY-51E-3199", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 14, "retention_rate": 0.2153846153846154, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.09090909090909091, "amount": 0.16666666666666666, "status": 0.5, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 786}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 15, "retention_rate": 0.23076923076923078, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.09090909090909091, "amount": 0.16666666666666666, "status": 0.5, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 855}, "2": {"facts_total": 65, "facts_retained": 14, "retention_rate": 0.2153846153846154, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.09090909090909091, "amount": 0.16666666666666666, "status": 0.5, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 786}}}, "reference_key_id": "keys/synthetic/syn_03000_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "ffaf2ec78fb536984dc613873a800b081c1318d5d46ca5c4de575785729d8cbb", "output_hash": "a1245d1536890a7e4b026366df48441d503f491f782cfd388797b5782443c5c7"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "792d07780fc788f19971dec4", "attempt_id": "54c5e48197cd", "run_group_id": "E4_core:A5|repairable|qwen3b:h2", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T06:29:13.368662Z", "task_family": "synthetic", "document_id": "syn_03000_2000", "document_hash": "c9cc27c20b05b1f6e592846e84990ee9b69c42f9720a8e338ec2f6fca777fdbc", "question_id": "q06_negation", "split": "test", "cluster_id": "syn_03000_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 2, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1904, "query_tokens": 444, "message_tokens": 363, "message_utf8_bytes": 848, "cache_tensor_bytes": null, "serialized_payload_bytes": 848, "side_information_bytes": 0, "fetched_bytes": 1228, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 38790.371541981585, "hops": {"1": 22956.80054195691, "2": 15833.571000024676}, "receiver_prefill_ms": 351.1523340130225, "ttft_ms": 110.9736249782145, "decode_ms": 103.15412492491305}, "total_wall_ms": 39356.2077509705, "ttft_ms": 110.9736249782145, "input_usage": {"lookups": 2}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "no", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 14, "retention_rate": 0.2153846153846154, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.09090909090909091, "amount": 0.16666666666666666, "status": 0.5, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 786}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 15, "retention_rate": 0.23076923076923078, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.09090909090909091, "amount": 0.16666666666666666, "status": 0.5, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 855}, "2": {"facts_total": 65, "facts_retained": 14, "retention_rate": 0.2153846153846154, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.09090909090909091, "amount": 0.16666666666666666, "status": 0.5, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 786}}}, "reference_key_id": "keys/synthetic/syn_03000_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "ffaf2ec78fb536984dc613873a800b081c1318d5d46ca5c4de575785729d8cbb", "output_hash": "9390298f3fb0c5b160498935d79cb139aef28e1c47358b4bbba61862b9c26e59"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "90be4af210a71362e298ec50", "attempt_id": "91f0029ddcb3", "run_group_id": "E4_core:A5|repairable|qwen3b:h2", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T06:29:13.952077Z", "task_family": "synthetic", "document_id": "syn_03000_2000", "document_hash": "c9cc27c20b05b1f6e592846e84990ee9b69c42f9720a8e338ec2f6fca777fdbc", "question_id": "q07_negation", "split": "test", "cluster_id": "syn_03000_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 2, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1904, "query_tokens": 444, "message_tokens": 363, "message_utf8_bytes": 848, "cache_tensor_bytes": null, "serialized_payload_bytes": 848, "side_information_bytes": 0, "fetched_bytes": 1228, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 38790.371541981585, "hops": {"1": 22956.80054195691, "2": 15833.571000024676}, "receiver_prefill_ms": 316.94508309010416, "ttft_ms": 31.32204106077552, "decode_ms": 42.562542017549276}, "total_wall_ms": 39181.67804204859, "ttft_ms": 31.32204106077552, "input_usage": {"lookups": 2}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 14, "retention_rate": 0.2153846153846154, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.09090909090909091, "amount": 0.16666666666666666, "status": 0.5, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 786}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 15, "retention_rate": 0.23076923076923078, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.09090909090909091, "amount": 0.16666666666666666, "status": 0.5, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 855}, "2": {"facts_total": 65, "facts_retained": 14, "retention_rate": 0.2153846153846154, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.09090909090909091, "amount": 0.16666666666666666, "status": 0.5, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 786}}}, "reference_key_id": "keys/synthetic/syn_03000_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "ffaf2ec78fb536984dc613873a800b081c1318d5d46ca5c4de575785729d8cbb", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "34437d9dae1e07c3c85d75b4", "attempt_id": "66a79b9b3b60", "run_group_id": "E4_core:A5|repairable|qwen3b:h2", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T06:29:14.359794Z", "task_family": "synthetic", "document_id": "syn_03000_2000", "document_hash": "c9cc27c20b05b1f6e592846e84990ee9b69c42f9720a8e338ec2f6fca777fdbc", "question_id": "q08_revision", "split": "test", "cluster_id": "syn_03000_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 2, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1904, "query_tokens": 459, "message_tokens": 363, "message_utf8_bytes": 848, "cache_tensor_bytes": null, "serialized_payload_bytes": 848, "side_information_bytes": 0, "fetched_bytes": 1228, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 38790.371541981585, "hops": {"1": 22956.80054195691, "2": 15833.571000024676}, "receiver_prefill_ms": 320.6247090129182, "ttft_ms": 30.38345801178366, "decode_ms": 391.7285419302061}, "total_wall_ms": 39533.69287599344, "ttft_ms": 30.38345801178366, "input_usage": {"lookups": 2}, "output_usage": {"output_tokens": 9}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "2,705,965", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 14, "retention_rate": 0.2153846153846154, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.09090909090909091, "amount": 0.16666666666666666, "status": 0.5, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 786}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 15, "retention_rate": 0.23076923076923078, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.09090909090909091, "amount": 0.16666666666666666, "status": 0.5, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 855}, "2": {"facts_total": 65, "facts_retained": 14, "retention_rate": 0.2153846153846154, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.09090909090909091, "amount": 0.16666666666666666, "status": 0.5, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 786}}}, "reference_key_id": "keys/synthetic/syn_03000_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "ffaf2ec78fb536984dc613873a800b081c1318d5d46ca5c4de575785729d8cbb", "output_hash": "dbedb6f5d00e413d915513b85d8ed195b18b9acf392ece342a85e15e3752bcc8"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "f37462eafe4f2da5eea1cf6f", "attempt_id": "1d5690b8a401", "run_group_id": "E4_core:A5|repairable|qwen3b:h2", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T06:29:15.119824Z", "task_family": "synthetic", "document_id": "syn_03000_2000", "document_hash": "c9cc27c20b05b1f6e592846e84990ee9b69c42f9720a8e338ec2f6fca777fdbc", "question_id": "q09_revision", "split": "test", "cluster_id": "syn_03000_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 2, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1904, "query_tokens": 455, "message_tokens": 363, "message_utf8_bytes": 848, "cache_tensor_bytes": null, "serialized_payload_bytes": 848, "side_information_bytes": 0, "fetched_bytes": 1228, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 38790.371541981585, "hops": {"1": 22956.80054195691, "2": 15833.571000024676}, "receiver_prefill_ms": 470.84708290640265, "ttft_ms": 42.01179195661098, "decode_ms": 27.064249967224896}, "total_wall_ms": 39330.97029197961, "ttft_ms": 42.01179195661098, "input_usage": {"lookups": 2}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 14, "retention_rate": 0.2153846153846154, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.09090909090909091, "amount": 0.16666666666666666, "status": 0.5, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 786}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 15, "retention_rate": 0.23076923076923078, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.09090909090909091, "amount": 0.16666666666666666, "status": 0.5, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 855}, "2": {"facts_total": 65, "facts_retained": 14, "retention_rate": 0.2153846153846154, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.09090909090909091, "amount": 0.16666666666666666, "status": 0.5, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 786}}}, "reference_key_id": "keys/synthetic/syn_03000_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "ffaf2ec78fb536984dc613873a800b081c1318d5d46ca5c4de575785729d8cbb", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "c46940c0679443a609e6aa78", "attempt_id": "03e946e2148e", "run_group_id": "E4_core:A5|repairable|qwen3b:h2", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T06:29:15.677720Z", "task_family": "synthetic", "document_id": "syn_03000_2000", "document_hash": "c9cc27c20b05b1f6e592846e84990ee9b69c42f9720a8e338ec2f6fca777fdbc", "question_id": "q10_temporal", "split": "test", "cluster_id": "syn_03000_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 2, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1904, "query_tokens": 461, "message_tokens": 363, "message_utf8_bytes": 848, "cache_tensor_bytes": null, "serialized_payload_bytes": 848, "side_information_bytes": 0, "fetched_bytes": 1228, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 38790.371541981585, "hops": {"1": 22956.80054195691, "2": 15833.571000024676}, "receiver_prefill_ms": 313.1416670512408, "ttft_ms": 35.644166986458, "decode_ms": 38.866041926667094}, "total_wall_ms": 39178.50883398205, "ttft_ms": 35.644166986458, "input_usage": {"lookups": 2}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 14, "retention_rate": 0.2153846153846154, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.09090909090909091, "amount": 0.16666666666666666, "status": 0.5, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 786}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 15, "retention_rate": 0.23076923076923078, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.09090909090909091, "amount": 0.16666666666666666, "status": 0.5, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 855}, "2": {"facts_total": 65, "facts_retained": 14, "retention_rate": 0.2153846153846154, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.09090909090909091, "amount": 0.16666666666666666, "status": 0.5, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 786}}}, "reference_key_id": "keys/synthetic/syn_03000_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "ffaf2ec78fb536984dc613873a800b081c1318d5d46ca5c4de575785729d8cbb", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "d0d4da0deaf886dda933942f", "attempt_id": "b4e842661b07", "run_group_id": "E4_core:A5|repairable|qwen3b:h2", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T06:29:16.082617Z", "task_family": "synthetic", "document_id": "syn_03000_2000", "document_hash": "c9cc27c20b05b1f6e592846e84990ee9b69c42f9720a8e338ec2f6fca777fdbc", "question_id": "q11_aggregation", "split": "test", "cluster_id": "syn_03000_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 2, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1904, "query_tokens": 456, "message_tokens": 363, "message_utf8_bytes": 848, "cache_tensor_bytes": null, "serialized_payload_bytes": 848, "side_information_bytes": 0, "fetched_bytes": 1228, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 38790.371541981585, "hops": {"1": 22956.80054195691, "2": 15833.571000024676}, "receiver_prefill_ms": 311.681833001785, "ttft_ms": 18.966916017234325, "decode_ms": 38.67545898538083}, "total_wall_ms": 39160.28737497982, "ttft_ms": 18.966916017234325, "input_usage": {"lookups": 2}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 14, "retention_rate": 0.2153846153846154, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.09090909090909091, "amount": 0.16666666666666666, "status": 0.5, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 786}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 15, "retention_rate": 0.23076923076923078, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.09090909090909091, "amount": 0.16666666666666666, "status": 0.5, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 855}, "2": {"facts_total": 65, "facts_retained": 14, "retention_rate": 0.2153846153846154, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.09090909090909091, "amount": 0.16666666666666666, "status": 0.5, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 786}}}, "reference_key_id": "keys/synthetic/syn_03000_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "ffaf2ec78fb536984dc613873a800b081c1318d5d46ca5c4de575785729d8cbb", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "ef3623636e75be523fc14d99", "attempt_id": "1eae6eb10970", "run_group_id": "E4_core:A5|repairable|qwen3b:h2", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T06:29:16.468796Z", "task_family": "synthetic", "document_id": "syn_03000_2000", "document_hash": "c9cc27c20b05b1f6e592846e84990ee9b69c42f9720a8e338ec2f6fca777fdbc", "question_id": "q12_unit_conversion", "split": "test", "cluster_id": "syn_03000_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 2, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1904, "query_tokens": 452, "message_tokens": 363, "message_utf8_bytes": 848, "cache_tensor_bytes": null, "serialized_payload_bytes": 848, "side_information_bytes": 0, "fetched_bytes": 1228, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 38790.371541981585, "hops": {"1": 22956.80054195691, "2": 15833.571000024676}, "receiver_prefill_ms": 277.0671669859439, "ttft_ms": 56.559375021606684, "decode_ms": 417.943959007971}, "total_wall_ms": 39542.718500015326, "ttft_ms": 56.559375021606684, "input_usage": {"lookups": 2}, "output_usage": {"output_tokens": 5}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "20.46", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 14, "retention_rate": 0.2153846153846154, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.09090909090909091, "amount": 0.16666666666666666, "status": 0.5, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 786}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 15, "retention_rate": 0.23076923076923078, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.09090909090909091, "amount": 0.16666666666666666, "status": 0.5, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 855}, "2": {"facts_total": 65, "facts_retained": 14, "retention_rate": 0.2153846153846154, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.09090909090909091, "amount": 0.16666666666666666, "status": 0.5, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 786}}}, "reference_key_id": "keys/synthetic/syn_03000_2000.json", "task_scores": {"correct": 0.0, "abstained": 0.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "ffaf2ec78fb536984dc613873a800b081c1318d5d46ca5c4de575785729d8cbb", "output_hash": "c057e11a5e818ca86a03c336373f3b4e4a2ddd9756e0acd7f5d7d655fd0267fe"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "f7093db659f57ee9b9c617cf", "attempt_id": "85f21d00ef75", "run_group_id": "E4_core:A5|repairable|qwen3b:h2", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T06:29:17.237291Z", "task_family": "synthetic", "document_id": "syn_03000_2000", "document_hash": "c9cc27c20b05b1f6e592846e84990ee9b69c42f9720a8e338ec2f6fca777fdbc", "question_id": "q13_unanswerable", "split": "test", "cluster_id": "syn_03000_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 2, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1904, "query_tokens": 456, "message_tokens": 363, "message_utf8_bytes": 848, "cache_tensor_bytes": null, "serialized_payload_bytes": 848, "side_information_bytes": 0, "fetched_bytes": 1228, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 38790.371541981585, "hops": {"1": 22956.80054195691, "2": 15833.571000024676}, "receiver_prefill_ms": 321.87187497038394, "ttft_ms": 21.437040995806456, "decode_ms": 41.36537492740899}, "total_wall_ms": 39175.68512505386, "ttft_ms": 21.437040995806456, "input_usage": {"lookups": 2}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 14, "retention_rate": 0.2153846153846154, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.09090909090909091, "amount": 0.16666666666666666, "status": 0.5, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 786}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 15, "retention_rate": 0.23076923076923078, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.09090909090909091, "amount": 0.16666666666666666, "status": 0.5, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 855}, "2": {"facts_total": 65, "facts_retained": 14, "retention_rate": 0.2153846153846154, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.09090909090909091, "amount": 0.16666666666666666, "status": 0.5, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 786}}}, "reference_key_id": "keys/synthetic/syn_03000_2000.json", "task_scores": {"correct": 1.0, "abstained": 1.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "ffaf2ec78fb536984dc613873a800b081c1318d5d46ca5c4de575785729d8cbb", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "a2947b1e9464175df5fb1888", "attempt_id": "3a417a832ab5", "run_group_id": "E4_core:A5|repairable|qwen3b:h2", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T06:29:17.638836Z", "task_family": "synthetic", "document_id": "syn_03000_2000", "document_hash": "c9cc27c20b05b1f6e592846e84990ee9b69c42f9720a8e338ec2f6fca777fdbc", "question_id": "q14_unanswerable", "split": "test", "cluster_id": "syn_03000_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 2, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1904, "query_tokens": 458, "message_tokens": 363, "message_utf8_bytes": 848, "cache_tensor_bytes": null, "serialized_payload_bytes": 848, "side_information_bytes": 0, "fetched_bytes": 1228, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 38790.371541981585, "hops": {"1": 22956.80054195691, "2": 15833.571000024676}, "receiver_prefill_ms": 321.9007500447333, "ttft_ms": 18.836500006727874, "decode_ms": 38.26837497763336}, "total_wall_ms": 39169.89512590226, "ttft_ms": 18.836500006727874, "input_usage": {"lookups": 2}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 14, "retention_rate": 0.2153846153846154, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.09090909090909091, "amount": 0.16666666666666666, "status": 0.5, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 786}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 15, "retention_rate": 0.23076923076923078, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.09090909090909091, "amount": 0.16666666666666666, "status": 0.5, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 855}, "2": {"facts_total": 65, "facts_retained": 14, "retention_rate": 0.2153846153846154, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.09090909090909091, "amount": 0.16666666666666666, "status": 0.5, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 786}}}, "reference_key_id": "keys/synthetic/syn_03000_2000.json", "task_scores": {"correct": 1.0, "abstained": 1.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "ffaf2ec78fb536984dc613873a800b081c1318d5d46ca5c4de575785729d8cbb", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "67e42321aca6294231a6c1d2", "attempt_id": "cbc1131f0db6", "run_group_id": "E4_core:A5|repairable|qwen3b:h4", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T06:29:49.595727Z", "task_family": "synthetic", "document_id": "syn_03000_2000", "document_hash": "c9cc27c20b05b1f6e592846e84990ee9b69c42f9720a8e338ec2f6fca777fdbc", "question_id": "q00_single_fact", "split": "test", "cluster_id": "syn_03000_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 4, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1904, "query_tokens": 444, "message_tokens": 363, "message_utf8_bytes": 848, "cache_tensor_bytes": null, "serialized_payload_bytes": 848, "side_information_bytes": 0, "fetched_bytes": 2428, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 70384.45262587629, "hops": {"1": 22956.80054195691, "2": 15833.571000024676, "3": 15950.301249977201, "4": 15643.779833917506}, "receiver_prefill_ms": 318.703250028193, "ttft_ms": 38.3424999890849, "decode_ms": 62.283582985401154}, "total_wall_ms": 70804.44137589075, "ttft_ms": 38.3424999890849, "input_usage": {"lookups": 4}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 14, "retention_rate": 0.2153846153846154, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.09090909090909091, "amount": 0.16666666666666666, "status": 0.5, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 786}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 15, "retention_rate": 0.23076923076923078, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.09090909090909091, "amount": 0.16666666666666666, "status": 0.5, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 855}, "2": {"facts_total": 65, "facts_retained": 14, "retention_rate": 0.2153846153846154, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.09090909090909091, "amount": 0.16666666666666666, "status": 0.5, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 786}, "3": {"facts_total": 65, "facts_retained": 14, "retention_rate": 0.2153846153846154, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.09090909090909091, "amount": 0.16666666666666666, "status": 0.5, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 786}, "4": {"facts_total": 65, "facts_retained": 14, "retention_rate": 0.2153846153846154, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.09090909090909091, "amount": 0.16666666666666666, "status": 0.5, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 786}}}, "reference_key_id": "keys/synthetic/syn_03000_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "ffaf2ec78fb536984dc613873a800b081c1318d5d46ca5c4de575785729d8cbb", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "103237245fa71f78dcac10ac", "attempt_id": "b473782e2534", "run_group_id": "E4_core:A5|repairable|qwen3b:h4", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T06:29:50.034892Z", "task_family": "synthetic", "document_id": "syn_03000_2000", "document_hash": "c9cc27c20b05b1f6e592846e84990ee9b69c42f9720a8e338ec2f6fca777fdbc", "question_id": "q01_single_fact", "split": "test", "cluster_id": "syn_03000_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 4, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1904, "query_tokens": 449, "message_tokens": 363, "message_utf8_bytes": 848, "cache_tensor_bytes": null, "serialized_payload_bytes": 848, "side_information_bytes": 0, "fetched_bytes": 2428, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 70384.45262587629, "hops": {"1": 22956.80054195691, "2": 15833.571000024676, "3": 15950.301249977201, "4": 15643.779833917506}, "receiver_prefill_ms": 405.66316607873887, "ttft_ms": 38.80841692443937, "decode_ms": 337.67437492497265}, "total_wall_ms": 71167.18995990232, "ttft_ms": 38.80841692443937, "input_usage": {"lookups": 4}, "output_usage": {"output_tokens": 11}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "Lot 409-H in Senneterre", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 14, "retention_rate": 0.2153846153846154, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.09090909090909091, "amount": 0.16666666666666666, "status": 0.5, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 786}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 15, "retention_rate": 0.23076923076923078, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.09090909090909091, "amount": 0.16666666666666666, "status": 0.5, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 855}, "2": {"facts_total": 65, "facts_retained": 14, "retention_rate": 0.2153846153846154, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.09090909090909091, "amount": 0.16666666666666666, "status": 0.5, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 786}, "3": {"facts_total": 65, "facts_retained": 14, "retention_rate": 0.2153846153846154, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.09090909090909091, "amount": 0.16666666666666666, "status": 0.5, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 786}, "4": {"facts_total": 65, "facts_retained": 14, "retention_rate": 0.2153846153846154, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.09090909090909091, "amount": 0.16666666666666666, "status": 0.5, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 786}}}, "reference_key_id": "keys/synthetic/syn_03000_2000.json", "task_scores": {"correct": 0.0, "abstained": 0.0, "f1": 0.5}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "ffaf2ec78fb536984dc613873a800b081c1318d5d46ca5c4de575785729d8cbb", "output_hash": "928c825f0fee3b69cf0901e3721b2d6c4467e52d2c6707afe17dd750a93d5e31"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "f8ff3fb9b162866afc1e6a28", "attempt_id": "3aa99be404b3", "run_group_id": "E4_core:A5|repairable|qwen3b:h4", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T06:29:50.833961Z", "task_family": "synthetic", "document_id": "syn_03000_2000", "document_hash": "c9cc27c20b05b1f6e592846e84990ee9b69c42f9720a8e338ec2f6fca777fdbc", "question_id": "q02_exact_copy", "split": "test", "cluster_id": "syn_03000_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 4, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1904, "query_tokens": 448, "message_tokens": 363, "message_utf8_bytes": 848, "cache_tensor_bytes": null, "serialized_payload_bytes": 848, "side_information_bytes": 0, "fetched_bytes": 2428, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 70384.45262587629, "hops": {"1": 22956.80054195691, "2": 15833.571000024676, "3": 15950.301249977201, "4": 15643.779833917506}, "receiver_prefill_ms": 298.3372499002144, "ttft_ms": 23.6924160271883, "decode_ms": 17.57604198064655}, "total_wall_ms": 70724.57225085236, "ttft_ms": 23.6924160271883, "input_usage": {"lookups": 4}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 14, "retention_rate": 0.2153846153846154, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.09090909090909091, "amount": 0.16666666666666666, "status": 0.5, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 786}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 15, "retention_rate": 0.23076923076923078, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.09090909090909091, "amount": 0.16666666666666666, "status": 0.5, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 855}, "2": {"facts_total": 65, "facts_retained": 14, "retention_rate": 0.2153846153846154, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.09090909090909091, "amount": 0.16666666666666666, "status": 0.5, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 786}, "3": {"facts_total": 65, "facts_retained": 14, "retention_rate": 0.2153846153846154, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.09090909090909091, "amount": 0.16666666666666666, "status": 0.5, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 786}, "4": {"facts_total": 65, "facts_retained": 14, "retention_rate": 0.2153846153846154, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.09090909090909091, "amount": 0.16666666666666666, "status": 0.5, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 786}}}, "reference_key_id": "keys/synthetic/syn_03000_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "ffaf2ec78fb536984dc613873a800b081c1318d5d46ca5c4de575785729d8cbb", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "f1d80a185145ffa049f280fc", "attempt_id": "7b5f1493dbcd", "run_group_id": "E4_core:A5|repairable|qwen3b:h4", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T06:29:51.190800Z", "task_family": "synthetic", "document_id": "syn_03000_2000", "document_hash": "c9cc27c20b05b1f6e592846e84990ee9b69c42f9720a8e338ec2f6fca777fdbc", "question_id": "q03_exact_copy", "split": "test", "cluster_id": "syn_03000_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 4, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1904, "query_tokens": 447, "message_tokens": 363, "message_utf8_bytes": 848, "cache_tensor_bytes": null, "serialized_payload_bytes": 848, "side_information_bytes": 0, "fetched_bytes": 2428, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 70384.45262587629, "hops": {"1": 22956.80054195691, "2": 15833.571000024676, "3": 15950.301249977201, "4": 15643.779833917506}, "receiver_prefill_ms": 315.0550409918651, "ttft_ms": 52.85675008781254, "decode_ms": 48.18529193289578}, "total_wall_ms": 70801.08404287603, "ttft_ms": 52.85675008781254, "input_usage": {"lookups": 4}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 14, "retention_rate": 0.2153846153846154, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.09090909090909091, "amount": 0.16666666666666666, "status": 0.5, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 786}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 15, "retention_rate": 0.23076923076923078, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.09090909090909091, "amount": 0.16666666666666666, "status": 0.5, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 855}, "2": {"facts_total": 65, "facts_retained": 14, "retention_rate": 0.2153846153846154, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.09090909090909091, "amount": 0.16666666666666666, "status": 0.5, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 786}, "3": {"facts_total": 65, "facts_retained": 14, "retention_rate": 0.2153846153846154, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.09090909090909091, "amount": 0.16666666666666666, "status": 0.5, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 786}, "4": {"facts_total": 65, "facts_retained": 14, "retention_rate": 0.2153846153846154, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.09090909090909091, "amount": 0.16666666666666666, "status": 0.5, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 786}}}, "reference_key_id": "keys/synthetic/syn_03000_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "ffaf2ec78fb536984dc613873a800b081c1318d5d46ca5c4de575785729d8cbb", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "fc39f87d4492d749ff53fe08", "attempt_id": "b5d7d4229a19", "run_group_id": "E4_core:A5|repairable|qwen3b:h4", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T06:29:51.626229Z", "task_family": "synthetic", "document_id": "syn_03000_2000", "document_hash": "c9cc27c20b05b1f6e592846e84990ee9b69c42f9720a8e338ec2f6fca777fdbc", "question_id": "q04_multi_hop", "split": "test", "cluster_id": "syn_03000_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 4, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1904, "query_tokens": 453, "message_tokens": 363, "message_utf8_bytes": 848, "cache_tensor_bytes": null, "serialized_payload_bytes": 848, "side_information_bytes": 0, "fetched_bytes": 2428, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 70384.45262587629, "hops": {"1": 22956.80054195691, "2": 15833.571000024676, "3": 15950.301249977201, "4": 15643.779833917506}, "receiver_prefill_ms": 455.6271251058206, "ttft_ms": 31.06508299242705, "decode_ms": 42.20779100432992}, "total_wall_ms": 70914.00491783861, "ttft_ms": 31.06508299242705, "input_usage": {"lookups": 4}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 14, "retention_rate": 0.2153846153846154, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.09090909090909091, "amount": 0.16666666666666666, "status": 0.5, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 786}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 15, "retention_rate": 0.23076923076923078, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.09090909090909091, "amount": 0.16666666666666666, "status": 0.5, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 855}, "2": {"facts_total": 65, "facts_retained": 14, "retention_rate": 0.2153846153846154, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.09090909090909091, "amount": 0.16666666666666666, "status": 0.5, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 786}, "3": {"facts_total": 65, "facts_retained": 14, "retention_rate": 0.2153846153846154, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.09090909090909091, "amount": 0.16666666666666666, "status": 0.5, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 786}, "4": {"facts_total": 65, "facts_retained": 14, "retention_rate": 0.2153846153846154, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.09090909090909091, "amount": 0.16666666666666666, "status": 0.5, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 786}}}, "reference_key_id": "keys/synthetic/syn_03000_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "ffaf2ec78fb536984dc613873a800b081c1318d5d46ca5c4de575785729d8cbb", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "f562a6574b7a5dcc100d0044", "attempt_id": "33223d1656f1", "run_group_id": "E4_core:A5|repairable|qwen3b:h4", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T06:29:52.172416Z", "task_family": "synthetic", "document_id": "syn_03000_2000", "document_hash": "c9cc27c20b05b1f6e592846e84990ee9b69c42f9720a8e338ec2f6fca777fdbc", "question_id": "q05_multi_hop", "split": "test", "cluster_id": "syn_03000_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 4, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1904, "query_tokens": 455, "message_tokens": 363, "message_utf8_bytes": 848, "cache_tensor_bytes": null, "serialized_payload_bytes": 848, "side_information_bytes": 0, "fetched_bytes": 2428, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 70384.45262587629, "hops": {"1": 22956.80054195691, "2": 15833.571000024676, "3": 15950.301249977201, "4": 15643.779833917506}, "receiver_prefill_ms": 328.49141699261963, "ttft_ms": 38.54312503244728, "decode_ms": 295.4104579985142}, "total_wall_ms": 71047.65716788825, "ttft_ms": 38.54312503244728, "input_usage": {"lookups": 4}, "output_usage": {"output_tokens": 10}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "LY-51E-3199", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 14, "retention_rate": 0.2153846153846154, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.09090909090909091, "amount": 0.16666666666666666, "status": 0.5, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 786}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 15, "retention_rate": 0.23076923076923078, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.09090909090909091, "amount": 0.16666666666666666, "status": 0.5, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 855}, "2": {"facts_total": 65, "facts_retained": 14, "retention_rate": 0.2153846153846154, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.09090909090909091, "amount": 0.16666666666666666, "status": 0.5, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 786}, "3": {"facts_total": 65, "facts_retained": 14, "retention_rate": 0.2153846153846154, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.09090909090909091, "amount": 0.16666666666666666, "status": 0.5, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 786}, "4": {"facts_total": 65, "facts_retained": 14, "retention_rate": 0.2153846153846154, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.09090909090909091, "amount": 0.16666666666666666, "status": 0.5, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 786}}}, "reference_key_id": "keys/synthetic/syn_03000_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "ffaf2ec78fb536984dc613873a800b081c1318d5d46ca5c4de575785729d8cbb", "output_hash": "a1245d1536890a7e4b026366df48441d503f491f782cfd388797b5782443c5c7"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "d52ff8b18b3737158b0aa461", "attempt_id": "18fee3adf3d0", "run_group_id": "E4_core:A5|repairable|qwen3b:h4", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T06:29:52.851836Z", "task_family": "synthetic", "document_id": "syn_03000_2000", "document_hash": "c9cc27c20b05b1f6e592846e84990ee9b69c42f9720a8e338ec2f6fca777fdbc", "question_id": "q06_negation", "split": "test", "cluster_id": "syn_03000_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 4, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1904, "query_tokens": 444, "message_tokens": 363, "message_utf8_bytes": 848, "cache_tensor_bytes": null, "serialized_payload_bytes": 848, "side_information_bytes": 0, "fetched_bytes": 2428, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 70384.45262587629, "hops": {"1": 22956.80054195691, "2": 15833.571000024676, "3": 15950.301249977201, "4": 15643.779833917506}, "receiver_prefill_ms": 289.4149589119479, "ttft_ms": 58.83004202041775, "decode_ms": 115.76841596979648}, "total_wall_ms": 70849.14000087883, "ttft_ms": 58.83004202041775, "input_usage": {"lookups": 4}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "no", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 14, "retention_rate": 0.2153846153846154, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.09090909090909091, "amount": 0.16666666666666666, "status": 0.5, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 786}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 15, "retention_rate": 0.23076923076923078, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.09090909090909091, "amount": 0.16666666666666666, "status": 0.5, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 855}, "2": {"facts_total": 65, "facts_retained": 14, "retention_rate": 0.2153846153846154, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.09090909090909091, "amount": 0.16666666666666666, "status": 0.5, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 786}, "3": {"facts_total": 65, "facts_retained": 14, "retention_rate": 0.2153846153846154, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.09090909090909091, "amount": 0.16666666666666666, "status": 0.5, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 786}, "4": {"facts_total": 65, "facts_retained": 14, "retention_rate": 0.2153846153846154, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.09090909090909091, "amount": 0.16666666666666666, "status": 0.5, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 786}}}, "reference_key_id": "keys/synthetic/syn_03000_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "ffaf2ec78fb536984dc613873a800b081c1318d5d46ca5c4de575785729d8cbb", "output_hash": "9390298f3fb0c5b160498935d79cb139aef28e1c47358b4bbba61862b9c26e59"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "6294831c6d782ad7d2182cb8", "attempt_id": "d31389c715ab", "run_group_id": "E4_core:A5|repairable|qwen3b:h4", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T06:29:53.333699Z", "task_family": "synthetic", "document_id": "syn_03000_2000", "document_hash": "c9cc27c20b05b1f6e592846e84990ee9b69c42f9720a8e338ec2f6fca777fdbc", "question_id": "q07_negation", "split": "test", "cluster_id": "syn_03000_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 4, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1904, "query_tokens": 444, "message_tokens": 363, "message_utf8_bytes": 848, "cache_tensor_bytes": null, "serialized_payload_bytes": 848, "side_information_bytes": 0, "fetched_bytes": 2428, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 70384.45262587629, "hops": {"1": 22956.80054195691, "2": 15833.571000024676, "3": 15950.301249977201, "4": 15643.779833917506}, "receiver_prefill_ms": 375.20437501370907, "ttft_ms": 40.201166993938386, "decode_ms": 30.001290957443416}, "total_wall_ms": 70830.4167088354, "ttft_ms": 40.201166993938386, "input_usage": {"lookups": 4}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 14, "retention_rate": 0.2153846153846154, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.09090909090909091, "amount": 0.16666666666666666, "status": 0.5, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 786}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 15, "retention_rate": 0.23076923076923078, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.09090909090909091, "amount": 0.16666666666666666, "status": 0.5, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 855}, "2": {"facts_total": 65, "facts_retained": 14, "retention_rate": 0.2153846153846154, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.09090909090909091, "amount": 0.16666666666666666, "status": 0.5, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 786}, "3": {"facts_total": 65, "facts_retained": 14, "retention_rate": 0.2153846153846154, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.09090909090909091, "amount": 0.16666666666666666, "status": 0.5, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 786}, "4": {"facts_total": 65, "facts_retained": 14, "retention_rate": 0.2153846153846154, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.09090909090909091, "amount": 0.16666666666666666, "status": 0.5, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 786}}}, "reference_key_id": "keys/synthetic/syn_03000_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "ffaf2ec78fb536984dc613873a800b081c1318d5d46ca5c4de575785729d8cbb", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "15c70387edf8b5c37c6f0bda", "attempt_id": "065fb82d9763", "run_group_id": "E4_core:A5|repairable|qwen3b:h4", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T06:29:53.798199Z", "task_family": "synthetic", "document_id": "syn_03000_2000", "document_hash": "c9cc27c20b05b1f6e592846e84990ee9b69c42f9720a8e338ec2f6fca777fdbc", "question_id": "q08_revision", "split": "test", "cluster_id": "syn_03000_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 4, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1904, "query_tokens": 459, "message_tokens": 363, "message_utf8_bytes": 848, "cache_tensor_bytes": null, "serialized_payload_bytes": 848, "side_information_bytes": 0, "fetched_bytes": 2428, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 70384.45262587629, "hops": {"1": 22956.80054195691, "2": 15833.571000024676, "3": 15950.301249977201, "4": 15643.779833917506}, "receiver_prefill_ms": 313.99658299051225, "ttft_ms": 39.52395904343575, "decode_ms": 286.4894999656826}, "total_wall_ms": 71025.06229188293, "ttft_ms": 39.52395904343575, "input_usage": {"lookups": 4}, "output_usage": {"output_tokens": 9}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "2,705,965", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 14, "retention_rate": 0.2153846153846154, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.09090909090909091, "amount": 0.16666666666666666, "status": 0.5, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 786}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 15, "retention_rate": 0.23076923076923078, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.09090909090909091, "amount": 0.16666666666666666, "status": 0.5, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 855}, "2": {"facts_total": 65, "facts_retained": 14, "retention_rate": 0.2153846153846154, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.09090909090909091, "amount": 0.16666666666666666, "status": 0.5, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 786}, "3": {"facts_total": 65, "facts_retained": 14, "retention_rate": 0.2153846153846154, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.09090909090909091, "amount": 0.16666666666666666, "status": 0.5, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 786}, "4": {"facts_total": 65, "facts_retained": 14, "retention_rate": 0.2153846153846154, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.09090909090909091, "amount": 0.16666666666666666, "status": 0.5, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 786}}}, "reference_key_id": "keys/synthetic/syn_03000_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "ffaf2ec78fb536984dc613873a800b081c1318d5d46ca5c4de575785729d8cbb", "output_hash": "dbedb6f5d00e413d915513b85d8ed195b18b9acf392ece342a85e15e3752bcc8"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "d674e13bcaf98072f09a2b9e", "attempt_id": "bae7760162b1", "run_group_id": "E4_core:A5|repairable|qwen3b:h4", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T06:29:54.455825Z", "task_family": "synthetic", "document_id": "syn_03000_2000", "document_hash": "c9cc27c20b05b1f6e592846e84990ee9b69c42f9720a8e338ec2f6fca777fdbc", "question_id": "q09_revision", "split": "test", "cluster_id": "syn_03000_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 4, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1904, "query_tokens": 455, "message_tokens": 363, "message_utf8_bytes": 848, "cache_tensor_bytes": null, "serialized_payload_bytes": 848, "side_information_bytes": 0, "fetched_bytes": 2428, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 70384.45262587629, "hops": {"1": 22956.80054195691, "2": 15833.571000024676, "3": 15950.301249977201, "4": 15643.779833917506}, "receiver_prefill_ms": 299.4611670728773, "ttft_ms": 43.25266706291586, "decode_ms": 37.74975007399917}, "total_wall_ms": 70765.42287587654, "ttft_ms": 43.25266706291586, "input_usage": {"lookups": 4}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 14, "retention_rate": 0.2153846153846154, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.09090909090909091, "amount": 0.16666666666666666, "status": 0.5, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 786}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 15, "retention_rate": 0.23076923076923078, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.09090909090909091, "amount": 0.16666666666666666, "status": 0.5, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 855}, "2": {"facts_total": 65, "facts_retained": 14, "retention_rate": 0.2153846153846154, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.09090909090909091, "amount": 0.16666666666666666, "status": 0.5, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 786}, "3": {"facts_total": 65, "facts_retained": 14, "retention_rate": 0.2153846153846154, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.09090909090909091, "amount": 0.16666666666666666, "status": 0.5, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 786}, "4": {"facts_total": 65, "facts_retained": 14, "retention_rate": 0.2153846153846154, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.09090909090909091, "amount": 0.16666666666666666, "status": 0.5, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 786}}}, "reference_key_id": "keys/synthetic/syn_03000_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "ffaf2ec78fb536984dc613873a800b081c1318d5d46ca5c4de575785729d8cbb", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "c89bd310d3982003ae364da5", "attempt_id": "0501a5d02faf", "run_group_id": "E4_core:A5|repairable|qwen3b:h4", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T06:29:54.853870Z", "task_family": "synthetic", "document_id": "syn_03000_2000", "document_hash": "c9cc27c20b05b1f6e592846e84990ee9b69c42f9720a8e338ec2f6fca777fdbc", "question_id": "q10_temporal", "split": "test", "cluster_id": "syn_03000_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 4, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1904, "query_tokens": 461, "message_tokens": 363, "message_utf8_bytes": 848, "cache_tensor_bytes": null, "serialized_payload_bytes": 848, "side_information_bytes": 0, "fetched_bytes": 2428, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 70384.45262587629, "hops": {"1": 22956.80054195691, "2": 15833.571000024676, "3": 15950.301249977201, "4": 15643.779833917506}, "receiver_prefill_ms": 445.30666701029986, "ttft_ms": 40.03937495872378, "decode_ms": 25.663708918727934}, "total_wall_ms": 70896.00220881402, "ttft_ms": 40.03937495872378, "input_usage": {"lookups": 4}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 14, "retention_rate": 0.2153846153846154, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.09090909090909091, "amount": 0.16666666666666666, "status": 0.5, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 786}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 15, "retention_rate": 0.23076923076923078, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.09090909090909091, "amount": 0.16666666666666666, "status": 0.5, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 855}, "2": {"facts_total": 65, "facts_retained": 14, "retention_rate": 0.2153846153846154, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.09090909090909091, "amount": 0.16666666666666666, "status": 0.5, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 786}, "3": {"facts_total": 65, "facts_retained": 14, "retention_rate": 0.2153846153846154, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.09090909090909091, "amount": 0.16666666666666666, "status": 0.5, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 786}, "4": {"facts_total": 65, "facts_retained": 14, "retention_rate": 0.2153846153846154, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.09090909090909091, "amount": 0.16666666666666666, "status": 0.5, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 786}}}, "reference_key_id": "keys/synthetic/syn_03000_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "ffaf2ec78fb536984dc613873a800b081c1318d5d46ca5c4de575785729d8cbb", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "f34d1d11fc747f398a036e26", "attempt_id": "875b08446339", "run_group_id": "E4_core:A5|repairable|qwen3b:h4", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T06:29:55.381943Z", "task_family": "synthetic", "document_id": "syn_03000_2000", "document_hash": "c9cc27c20b05b1f6e592846e84990ee9b69c42f9720a8e338ec2f6fca777fdbc", "question_id": "q11_aggregation", "split": "test", "cluster_id": "syn_03000_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 4, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1904, "query_tokens": 456, "message_tokens": 363, "message_utf8_bytes": 848, "cache_tensor_bytes": null, "serialized_payload_bytes": 848, "side_information_bytes": 0, "fetched_bytes": 2428, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 70384.45262587629, "hops": {"1": 22956.80054195691, "2": 15833.571000024676, "3": 15950.301249977201, "4": 15643.779833917506}, "receiver_prefill_ms": 317.62412493117154, "ttft_ms": 41.412458987906575, "decode_ms": 31.432542018592358}, "total_wall_ms": 70775.46500088647, "ttft_ms": 41.412458987906575, "input_usage": {"lookups": 4}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 14, "retention_rate": 0.2153846153846154, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.09090909090909091, "amount": 0.16666666666666666, "status": 0.5, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 786}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 15, "retention_rate": 0.23076923076923078, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.09090909090909091, "amount": 0.16666666666666666, "status": 0.5, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 855}, "2": {"facts_total": 65, "facts_retained": 14, "retention_rate": 0.2153846153846154, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.09090909090909091, "amount": 0.16666666666666666, "status": 0.5, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 786}, "3": {"facts_total": 65, "facts_retained": 14, "retention_rate": 0.2153846153846154, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.09090909090909091, "amount": 0.16666666666666666, "status": 0.5, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 786}, "4": {"facts_total": 65, "facts_retained": 14, "retention_rate": 0.2153846153846154, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.09090909090909091, "amount": 0.16666666666666666, "status": 0.5, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 786}}}, "reference_key_id": "keys/synthetic/syn_03000_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "ffaf2ec78fb536984dc613873a800b081c1318d5d46ca5c4de575785729d8cbb", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "00efb554ee3ca4e765ceaba1", "attempt_id": "8511e11539e8", "run_group_id": "E4_core:A5|repairable|qwen3b:h4", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T06:29:55.789506Z", "task_family": "synthetic", "document_id": "syn_03000_2000", "document_hash": "c9cc27c20b05b1f6e592846e84990ee9b69c42f9720a8e338ec2f6fca777fdbc", "question_id": "q12_unit_conversion", "split": "test", "cluster_id": "syn_03000_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 4, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1904, "query_tokens": 452, "message_tokens": 363, "message_utf8_bytes": 848, "cache_tensor_bytes": null, "serialized_payload_bytes": 848, "side_information_bytes": 0, "fetched_bytes": 2428, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 70384.45262587629, "hops": {"1": 22956.80054195691, "2": 15833.571000024676, "3": 15950.301249977201, "4": 15643.779833917506}, "receiver_prefill_ms": 306.5441249636933, "ttft_ms": 21.03237493429333, "decode_ms": 232.99024999141693}, "total_wall_ms": 70945.55675098673, "ttft_ms": 21.03237493429333, "input_usage": {"lookups": 4}, "output_usage": {"output_tokens": 5}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "20.46", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 14, "retention_rate": 0.2153846153846154, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.09090909090909091, "amount": 0.16666666666666666, "status": 0.5, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 786}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 15, "retention_rate": 0.23076923076923078, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.09090909090909091, "amount": 0.16666666666666666, "status": 0.5, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 855}, "2": {"facts_total": 65, "facts_retained": 14, "retention_rate": 0.2153846153846154, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.09090909090909091, "amount": 0.16666666666666666, "status": 0.5, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 786}, "3": {"facts_total": 65, "facts_retained": 14, "retention_rate": 0.2153846153846154, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.09090909090909091, "amount": 0.16666666666666666, "status": 0.5, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 786}, "4": {"facts_total": 65, "facts_retained": 14, "retention_rate": 0.2153846153846154, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.09090909090909091, "amount": 0.16666666666666666, "status": 0.5, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 786}}}, "reference_key_id": "keys/synthetic/syn_03000_2000.json", "task_scores": {"correct": 0.0, "abstained": 0.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "ffaf2ec78fb536984dc613873a800b081c1318d5d46ca5c4de575785729d8cbb", "output_hash": "c057e11a5e818ca86a03c336373f3b4e4a2ddd9756e0acd7f5d7d655fd0267fe"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "785a01a454ab6cabde16c044", "attempt_id": "e946b0ddcb11", "run_group_id": "E4_core:A5|repairable|qwen3b:h4", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T06:29:56.366153Z", "task_family": "synthetic", "document_id": "syn_03000_2000", "document_hash": "c9cc27c20b05b1f6e592846e84990ee9b69c42f9720a8e338ec2f6fca777fdbc", "question_id": "q13_unanswerable", "split": "test", "cluster_id": "syn_03000_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 4, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1904, "query_tokens": 456, "message_tokens": 363, "message_utf8_bytes": 848, "cache_tensor_bytes": null, "serialized_payload_bytes": 848, "side_information_bytes": 0, "fetched_bytes": 2428, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 70384.45262587629, "hops": {"1": 22956.80054195691, "2": 15833.571000024676, "3": 15950.301249977201, "4": 15643.779833917506}, "receiver_prefill_ms": 359.87795901019126, "ttft_ms": 103.17350004334003, "decode_ms": 40.11654097121209}, "total_wall_ms": 70888.3115008939, "ttft_ms": 103.17350004334003, "input_usage": {"lookups": 4}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 14, "retention_rate": 0.2153846153846154, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.09090909090909091, "amount": 0.16666666666666666, "status": 0.5, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 786}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 15, "retention_rate": 0.23076923076923078, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.09090909090909091, "amount": 0.16666666666666666, "status": 0.5, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 855}, "2": {"facts_total": 65, "facts_retained": 14, "retention_rate": 0.2153846153846154, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.09090909090909091, "amount": 0.16666666666666666, "status": 0.5, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 786}, "3": {"facts_total": 65, "facts_retained": 14, "retention_rate": 0.2153846153846154, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.09090909090909091, "amount": 0.16666666666666666, "status": 0.5, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 786}, "4": {"facts_total": 65, "facts_retained": 14, "retention_rate": 0.2153846153846154, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.09090909090909091, "amount": 0.16666666666666666, "status": 0.5, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 786}}}, "reference_key_id": "keys/synthetic/syn_03000_2000.json", "task_scores": {"correct": 1.0, "abstained": 1.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "ffaf2ec78fb536984dc613873a800b081c1318d5d46ca5c4de575785729d8cbb", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "31df0f5e7491c92f7fdc8fcd", "attempt_id": "fbbba38af69b", "run_group_id": "E4_core:A5|repairable|qwen3b:h4", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T06:29:56.887342Z", "task_family": "synthetic", "document_id": "syn_03000_2000", "document_hash": "c9cc27c20b05b1f6e592846e84990ee9b69c42f9720a8e338ec2f6fca777fdbc", "question_id": "q14_unanswerable", "split": "test", "cluster_id": "syn_03000_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 4, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1904, "query_tokens": 458, "message_tokens": 363, "message_utf8_bytes": 848, "cache_tensor_bytes": null, "serialized_payload_bytes": 848, "side_information_bytes": 0, "fetched_bytes": 2428, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 70384.45262587629, "hops": {"1": 22956.80054195691, "2": 15833.571000024676, "3": 15950.301249977201, "4": 15643.779833917506}, "receiver_prefill_ms": 324.75675002206117, "ttft_ms": 29.149207985028625, "decode_ms": 42.863417067565024}, "total_wall_ms": 70781.88000083901, "ttft_ms": 29.149207985028625, "input_usage": {"lookups": 4}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 14, "retention_rate": 0.2153846153846154, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.09090909090909091, "amount": 0.16666666666666666, "status": 0.5, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 786}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 15, "retention_rate": 0.23076923076923078, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.09090909090909091, "amount": 0.16666666666666666, "status": 0.5, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 855}, "2": {"facts_total": 65, "facts_retained": 14, "retention_rate": 0.2153846153846154, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.09090909090909091, "amount": 0.16666666666666666, "status": 0.5, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 786}, "3": {"facts_total": 65, "facts_retained": 14, "retention_rate": 0.2153846153846154, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.09090909090909091, "amount": 0.16666666666666666, "status": 0.5, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 786}, "4": {"facts_total": 65, "facts_retained": 14, "retention_rate": 0.2153846153846154, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.09090909090909091, "amount": 0.16666666666666666, "status": 0.5, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 786}}}, "reference_key_id": "keys/synthetic/syn_03000_2000.json", "task_scores": {"correct": 1.0, "abstained": 1.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "ffaf2ec78fb536984dc613873a800b081c1318d5d46ca5c4de575785729d8cbb", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "2f9f2ffe2c7a7785b33e74e7", "attempt_id": "0621d39a725a", "run_group_id": "E4_core:A5|repairable|qwen3b:h8", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T06:31:00.723310Z", "task_family": "synthetic", "document_id": "syn_03000_2000", "document_hash": "c9cc27c20b05b1f6e592846e84990ee9b69c42f9720a8e338ec2f6fca777fdbc", "question_id": "q00_single_fact", "split": "test", "cluster_id": "syn_03000_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 8, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1904, "query_tokens": 444, "message_tokens": 363, "message_utf8_bytes": 848, "cache_tensor_bytes": null, "serialized_payload_bytes": 848, "side_information_bytes": 0, "fetched_bytes": 4828, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 133805.3459180519, "hops": {"1": 22956.80054195691, "2": 15833.571000024676, "3": 15950.301249977201, "4": 15643.779833917506, "5": 16041.412000078708, "6": 15729.852208052762, "7": 16016.10441703815, "8": 15633.524667005986}, "receiver_prefill_ms": 347.3070830805227, "ttft_ms": 84.72266606986523, "decode_ms": 129.58416703622788}, "total_wall_ms": 134367.53816809505, "ttft_ms": 84.72266606986523, "input_usage": {"lookups": 8}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 14, "retention_rate": 0.2153846153846154, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.09090909090909091, "amount": 0.16666666666666666, "status": 0.5, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 786}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 15, "retention_rate": 0.23076923076923078, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.09090909090909091, "amount": 0.16666666666666666, "status": 0.5, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 855}, "2": {"facts_total": 65, "facts_retained": 14, "retention_rate": 0.2153846153846154, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.09090909090909091, "amount": 0.16666666666666666, "status": 0.5, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 786}, "3": {"facts_total": 65, "facts_retained": 14, "retention_rate": 0.2153846153846154, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.09090909090909091, "amount": 0.16666666666666666, "status": 0.5, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 786}, "4": {"facts_total": 65, "facts_retained": 14, "retention_rate": 0.2153846153846154, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.09090909090909091, "amount": 0.16666666666666666, "status": 0.5, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 786}, "5": {"facts_total": 65, "facts_retained": 14, "retention_rate": 0.2153846153846154, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.09090909090909091, "amount": 0.16666666666666666, "status": 0.5, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 786}, "6": {"facts_total": 65, "facts_retained": 14, "retention_rate": 0.2153846153846154, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.09090909090909091, "amount": 0.16666666666666666, "status": 0.5, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 786}, "7": {"facts_total": 65, "facts_retained": 14, "retention_rate": 0.2153846153846154, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.09090909090909091, "amount": 0.16666666666666666, "status": 0.5, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 786}, "8": {"facts_total": 65, "facts_retained": 14, "retention_rate": 0.2153846153846154, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.09090909090909091, "amount": 0.16666666666666666, "status": 0.5, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 786}}}, "reference_key_id": "keys/synthetic/syn_03000_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "ffaf2ec78fb536984dc613873a800b081c1318d5d46ca5c4de575785729d8cbb", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "9c71372036d465f13c33d312", "attempt_id": "b89d8362b8c5", "run_group_id": "E4_core:A5|repairable|qwen3b:h8", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T06:31:01.303301Z", "task_family": "synthetic", "document_id": "syn_03000_2000", "document_hash": "c9cc27c20b05b1f6e592846e84990ee9b69c42f9720a8e338ec2f6fca777fdbc", "question_id": "q01_single_fact", "split": "test", "cluster_id": "syn_03000_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 8, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1904, "query_tokens": 449, "message_tokens": 363, "message_utf8_bytes": 848, "cache_tensor_bytes": null, "serialized_payload_bytes": 848, "side_information_bytes": 0, "fetched_bytes": 4828, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 133805.3459180519, "hops": {"1": 22956.80054195691, "2": 15833.571000024676, "3": 15950.301249977201, "4": 15643.779833917506, "5": 16041.412000078708, "6": 15729.852208052762, "7": 16016.10441703815, "8": 15633.524667005986}, "receiver_prefill_ms": 293.8090410316363, "ttft_ms": 39.438458043150604, "decode_ms": 24.397834087722003}, "total_wall_ms": 134163.5220850585, "ttft_ms": 39.438458043150604, "input_usage": {"lookups": 8}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 14, "retention_rate": 0.2153846153846154, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.09090909090909091, "amount": 0.16666666666666666, "status": 0.5, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 786}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 15, "retention_rate": 0.23076923076923078, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.09090909090909091, "amount": 0.16666666666666666, "status": 0.5, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 855}, "2": {"facts_total": 65, "facts_retained": 14, "retention_rate": 0.2153846153846154, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.09090909090909091, "amount": 0.16666666666666666, "status": 0.5, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 786}, "3": {"facts_total": 65, "facts_retained": 14, "retention_rate": 0.2153846153846154, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.09090909090909091, "amount": 0.16666666666666666, "status": 0.5, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 786}, "4": {"facts_total": 65, "facts_retained": 14, "retention_rate": 0.2153846153846154, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.09090909090909091, "amount": 0.16666666666666666, "status": 0.5, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 786}, "5": {"facts_total": 65, "facts_retained": 14, "retention_rate": 0.2153846153846154, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.09090909090909091, "amount": 0.16666666666666666, "status": 0.5, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 786}, "6": {"facts_total": 65, "facts_retained": 14, "retention_rate": 0.2153846153846154, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.09090909090909091, "amount": 0.16666666666666666, "status": 0.5, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 786}, "7": {"facts_total": 65, "facts_retained": 14, "retention_rate": 0.2153846153846154, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.09090909090909091, "amount": 0.16666666666666666, "status": 0.5, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 786}, "8": {"facts_total": 65, "facts_retained": 14, "retention_rate": 0.2153846153846154, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.09090909090909091, "amount": 0.16666666666666666, "status": 0.5, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 786}}}, "reference_key_id": "keys/synthetic/syn_03000_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "ffaf2ec78fb536984dc613873a800b081c1318d5d46ca5c4de575785729d8cbb", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "b869b44fa91213db217467ea", "attempt_id": "4b4a0fa0529a", "run_group_id": "E4_core:A5|repairable|qwen3b:h8", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T06:31:01.678086Z", "task_family": "synthetic", "document_id": "syn_03000_2000", "document_hash": "c9cc27c20b05b1f6e592846e84990ee9b69c42f9720a8e338ec2f6fca777fdbc", "question_id": "q02_exact_copy", "split": "test", "cluster_id": "syn_03000_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 8, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1904, "query_tokens": 448, "message_tokens": 363, "message_utf8_bytes": 848, "cache_tensor_bytes": null, "serialized_payload_bytes": 848, "side_information_bytes": 0, "fetched_bytes": 4828, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 133805.3459180519, "hops": {"1": 22956.80054195691, "2": 15833.571000024676, "3": 15950.301249977201, "4": 15643.779833917506, "5": 16041.412000078708, "6": 15729.852208052762, "7": 16016.10441703815, "8": 15633.524667005986}, "receiver_prefill_ms": 299.3487910134718, "ttft_ms": 27.62029110454023, "decode_ms": 35.702915978617966}, "total_wall_ms": 134168.59241807833, "ttft_ms": 27.62029110454023, "input_usage": {"lookups": 8}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 14, "retention_rate": 0.2153846153846154, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.09090909090909091, "amount": 0.16666666666666666, "status": 0.5, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 786}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 15, "retention_rate": 0.23076923076923078, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.09090909090909091, "amount": 0.16666666666666666, "status": 0.5, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 855}, "2": {"facts_total": 65, "facts_retained": 14, "retention_rate": 0.2153846153846154, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.09090909090909091, "amount": 0.16666666666666666, "status": 0.5, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 786}, "3": {"facts_total": 65, "facts_retained": 14, "retention_rate": 0.2153846153846154, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.09090909090909091, "amount": 0.16666666666666666, "status": 0.5, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 786}, "4": {"facts_total": 65, "facts_retained": 14, "retention_rate": 0.2153846153846154, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.09090909090909091, "amount": 0.16666666666666666, "status": 0.5, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 786}, "5": {"facts_total": 65, "facts_retained": 14, "retention_rate": 0.2153846153846154, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.09090909090909091, "amount": 0.16666666666666666, "status": 0.5, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 786}, "6": {"facts_total": 65, "facts_retained": 14, "retention_rate": 0.2153846153846154, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.09090909090909091, "amount": 0.16666666666666666, "status": 0.5, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 786}, "7": {"facts_total": 65, "facts_retained": 14, "retention_rate": 0.2153846153846154, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.09090909090909091, "amount": 0.16666666666666666, "status": 0.5, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 786}, "8": {"facts_total": 65, "facts_retained": 14, "retention_rate": 0.2153846153846154, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.09090909090909091, "amount": 0.16666666666666666, "status": 0.5, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 786}}}, "reference_key_id": "keys/synthetic/syn_03000_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "ffaf2ec78fb536984dc613873a800b081c1318d5d46ca5c4de575785729d8cbb", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "2896d5913882817961bfec15", "attempt_id": "371b6606180f", "run_group_id": "E4_core:A5|repairable|qwen3b:h8", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T06:31:02.057903Z", "task_family": "synthetic", "document_id": "syn_03000_2000", "document_hash": "c9cc27c20b05b1f6e592846e84990ee9b69c42f9720a8e338ec2f6fca777fdbc", "question_id": "q03_exact_copy", "split": "test", "cluster_id": "syn_03000_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 8, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1904, "query_tokens": 447, "message_tokens": 363, "message_utf8_bytes": 848, "cache_tensor_bytes": null, "serialized_payload_bytes": 848, "side_information_bytes": 0, "fetched_bytes": 4828, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 133805.3459180519, "hops": {"1": 22956.80054195691, "2": 15833.571000024676, "3": 15950.301249977201, "4": 15643.779833917506, "5": 16041.412000078708, "6": 15729.852208052762, "7": 16016.10441703815, "8": 15633.524667005986}, "receiver_prefill_ms": 263.3667920017615, "ttft_ms": 44.17116695549339, "decode_ms": 62.333667068742216}, "total_wall_ms": 134175.7562099956, "ttft_ms": 44.17116695549339, "input_usage": {"lookups": 8}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 14, "retention_rate": 0.2153846153846154, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.09090909090909091, "amount": 0.16666666666666666, "status": 0.5, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 786}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 15, "retention_rate": 0.23076923076923078, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.09090909090909091, "amount": 0.16666666666666666, "status": 0.5, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 855}, "2": {"facts_total": 65, "facts_retained": 14, "retention_rate": 0.2153846153846154, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.09090909090909091, "amount": 0.16666666666666666, "status": 0.5, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 786}, "3": {"facts_total": 65, "facts_retained": 14, "retention_rate": 0.2153846153846154, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.09090909090909091, "amount": 0.16666666666666666, "status": 0.5, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 786}, "4": {"facts_total": 65, "facts_retained": 14, "retention_rate": 0.2153846153846154, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.09090909090909091, "amount": 0.16666666666666666, "status": 0.5, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 786}, "5": {"facts_total": 65, "facts_retained": 14, "retention_rate": 0.2153846153846154, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.09090909090909091, "amount": 0.16666666666666666, "status": 0.5, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 786}, "6": {"facts_total": 65, "facts_retained": 14, "retention_rate": 0.2153846153846154, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.09090909090909091, "amount": 0.16666666666666666, "status": 0.5, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 786}, "7": {"facts_total": 65, "facts_retained": 14, "retention_rate": 0.2153846153846154, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.09090909090909091, "amount": 0.16666666666666666, "status": 0.5, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 786}, "8": {"facts_total": 65, "facts_retained": 14, "retention_rate": 0.2153846153846154, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.09090909090909091, "amount": 0.16666666666666666, "status": 0.5, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 786}}}, "reference_key_id": "keys/synthetic/syn_03000_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "ffaf2ec78fb536984dc613873a800b081c1318d5d46ca5c4de575785729d8cbb", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "b3cfb7f254e295be5ead1ef9", "attempt_id": "c2260727e681", "run_group_id": "E4_core:A5|repairable|qwen3b:h8", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T06:31:02.444656Z", "task_family": "synthetic", "document_id": "syn_03000_2000", "document_hash": "c9cc27c20b05b1f6e592846e84990ee9b69c42f9720a8e338ec2f6fca777fdbc", "question_id": "q04_multi_hop", "split": "test", "cluster_id": "syn_03000_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 8, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1904, "query_tokens": 453, "message_tokens": 363, "message_utf8_bytes": 848, "cache_tensor_bytes": null, "serialized_payload_bytes": 848, "side_information_bytes": 0, "fetched_bytes": 4828, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 133805.3459180519, "hops": {"1": 22956.80054195691, "2": 15833.571000024676, "3": 15950.301249977201, "4": 15643.779833917506, "5": 16041.412000078708, "6": 15729.852208052762, "7": 16016.10441703815, "8": 15633.524667005986}, "receiver_prefill_ms": 297.23766597453505, "ttft_ms": 91.87945793382823, "decode_ms": 136.56970905140042}, "total_wall_ms": 134331.5922090551, "ttft_ms": 91.87945793382823, "input_usage": {"lookups": 8}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 14, "retention_rate": 0.2153846153846154, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.09090909090909091, "amount": 0.16666666666666666, "status": 0.5, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 786}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 15, "retention_rate": 0.23076923076923078, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.09090909090909091, "amount": 0.16666666666666666, "status": 0.5, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 855}, "2": {"facts_total": 65, "facts_retained": 14, "retention_rate": 0.2153846153846154, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.09090909090909091, "amount": 0.16666666666666666, "status": 0.5, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 786}, "3": {"facts_total": 65, "facts_retained": 14, "retention_rate": 0.2153846153846154, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.09090909090909091, "amount": 0.16666666666666666, "status": 0.5, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 786}, "4": {"facts_total": 65, "facts_retained": 14, "retention_rate": 0.2153846153846154, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.09090909090909091, "amount": 0.16666666666666666, "status": 0.5, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 786}, "5": {"facts_total": 65, "facts_retained": 14, "retention_rate": 0.2153846153846154, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.09090909090909091, "amount": 0.16666666666666666, "status": 0.5, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 786}, "6": {"facts_total": 65, "facts_retained": 14, "retention_rate": 0.2153846153846154, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.09090909090909091, "amount": 0.16666666666666666, "status": 0.5, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 786}, "7": {"facts_total": 65, "facts_retained": 14, "retention_rate": 0.2153846153846154, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.09090909090909091, "amount": 0.16666666666666666, "status": 0.5, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 786}, "8": {"facts_total": 65, "facts_retained": 14, "retention_rate": 0.2153846153846154, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.09090909090909091, "amount": 0.16666666666666666, "status": 0.5, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 786}}}, "reference_key_id": "keys/synthetic/syn_03000_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "ffaf2ec78fb536984dc613873a800b081c1318d5d46ca5c4de575785729d8cbb", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "5abbfd615a48be1941ce12ad", "attempt_id": "cc9df8dcacb7", "run_group_id": "E4_core:A5|repairable|qwen3b:h8", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T06:31:02.987833Z", "task_family": "synthetic", "document_id": "syn_03000_2000", "document_hash": "c9cc27c20b05b1f6e592846e84990ee9b69c42f9720a8e338ec2f6fca777fdbc", "question_id": "q05_multi_hop", "split": "test", "cluster_id": "syn_03000_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 8, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1904, "query_tokens": 455, "message_tokens": 363, "message_utf8_bytes": 848, "cache_tensor_bytes": null, "serialized_payload_bytes": 848, "side_information_bytes": 0, "fetched_bytes": 4828, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 133805.3459180519, "hops": {"1": 22956.80054195691, "2": 15833.571000024676, "3": 15950.301249977201, "4": 15643.779833917506, "5": 16041.412000078708, "6": 15729.852208052762, "7": 16016.10441703815, "8": 15633.524667005986}, "receiver_prefill_ms": 309.4686249969527, "ttft_ms": 27.980624930933118, "decode_ms": 381.99316593818367}, "total_wall_ms": 134525.40341799613, "ttft_ms": 27.980624930933118, "input_usage": {"lookups": 8}, "output_usage": {"output_tokens": 10}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "LY-51E-3199", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 14, "retention_rate": 0.2153846153846154, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.09090909090909091, "amount": 0.16666666666666666, "status": 0.5, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 786}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 15, "retention_rate": 0.23076923076923078, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.09090909090909091, "amount": 0.16666666666666666, "status": 0.5, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 855}, "2": {"facts_total": 65, "facts_retained": 14, "retention_rate": 0.2153846153846154, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.09090909090909091, "amount": 0.16666666666666666, "status": 0.5, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 786}, "3": {"facts_total": 65, "facts_retained": 14, "retention_rate": 0.2153846153846154, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.09090909090909091, "amount": 0.16666666666666666, "status": 0.5, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 786}, "4": {"facts_total": 65, "facts_retained": 14, "retention_rate": 0.2153846153846154, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.09090909090909091, "amount": 0.16666666666666666, "status": 0.5, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 786}, "5": {"facts_total": 65, "facts_retained": 14, "retention_rate": 0.2153846153846154, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.09090909090909091, "amount": 0.16666666666666666, "status": 0.5, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 786}, "6": {"facts_total": 65, "facts_retained": 14, "retention_rate": 0.2153846153846154, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.09090909090909091, "amount": 0.16666666666666666, "status": 0.5, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 786}, "7": {"facts_total": 65, "facts_retained": 14, "retention_rate": 0.2153846153846154, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.09090909090909091, "amount": 0.16666666666666666, "status": 0.5, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 786}, "8": {"facts_total": 65, "facts_retained": 14, "retention_rate": 0.2153846153846154, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.09090909090909091, "amount": 0.16666666666666666, "status": 0.5, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 786}}}, "reference_key_id": "keys/synthetic/syn_03000_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "ffaf2ec78fb536984dc613873a800b081c1318d5d46ca5c4de575785729d8cbb", "output_hash": "a1245d1536890a7e4b026366df48441d503f491f782cfd388797b5782443c5c7"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "d67b2d3edd96f1f6e075cc06", "attempt_id": "810dc3a30e30", "run_group_id": "E4_core:A5|repairable|qwen3b:h8", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T06:31:03.725471Z", "task_family": "synthetic", "document_id": "syn_03000_2000", "document_hash": "c9cc27c20b05b1f6e592846e84990ee9b69c42f9720a8e338ec2f6fca777fdbc", "question_id": "q06_negation", "split": "test", "cluster_id": "syn_03000_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 8, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1904, "query_tokens": 444, "message_tokens": 363, "message_utf8_bytes": 848, "cache_tensor_bytes": null, "serialized_payload_bytes": 848, "side_information_bytes": 0, "fetched_bytes": 4828, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 133805.3459180519, "hops": {"1": 22956.80054195691, "2": 15833.571000024676, "3": 15950.301249977201, "4": 15643.779833917506, "5": 16041.412000078708, "6": 15729.852208052762, "7": 16016.10441703815, "8": 15633.524667005986}, "receiver_prefill_ms": 277.81679190229625, "ttft_ms": 61.25087500549853, "decode_ms": 47.48804192058742}, "total_wall_ms": 134192.49166804366, "ttft_ms": 61.25087500549853, "input_usage": {"lookups": 8}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "no", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 14, "retention_rate": 0.2153846153846154, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.09090909090909091, "amount": 0.16666666666666666, "status": 0.5, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 786}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 15, "retention_rate": 0.23076923076923078, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.09090909090909091, "amount": 0.16666666666666666, "status": 0.5, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 855}, "2": {"facts_total": 65, "facts_retained": 14, "retention_rate": 0.2153846153846154, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.09090909090909091, "amount": 0.16666666666666666, "status": 0.5, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 786}, "3": {"facts_total": 65, "facts_retained": 14, "retention_rate": 0.2153846153846154, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.09090909090909091, "amount": 0.16666666666666666, "status": 0.5, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 786}, "4": {"facts_total": 65, "facts_retained": 14, "retention_rate": 0.2153846153846154, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.09090909090909091, "amount": 0.16666666666666666, "status": 0.5, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 786}, "5": {"facts_total": 65, "facts_retained": 14, "retention_rate": 0.2153846153846154, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.09090909090909091, "amount": 0.16666666666666666, "status": 0.5, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 786}, "6": {"facts_total": 65, "facts_retained": 14, "retention_rate": 0.2153846153846154, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.09090909090909091, "amount": 0.16666666666666666, "status": 0.5, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 786}, "7": {"facts_total": 65, "facts_retained": 14, "retention_rate": 0.2153846153846154, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.09090909090909091, "amount": 0.16666666666666666, "status": 0.5, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 786}, "8": {"facts_total": 65, "facts_retained": 14, "retention_rate": 0.2153846153846154, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.09090909090909091, "amount": 0.16666666666666666, "status": 0.5, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 786}}}, "reference_key_id": "keys/synthetic/syn_03000_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "ffaf2ec78fb536984dc613873a800b081c1318d5d46ca5c4de575785729d8cbb", "output_hash": "9390298f3fb0c5b160498935d79cb139aef28e1c47358b4bbba61862b9c26e59"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "79d287e5213fa29d27372e58", "attempt_id": "80d8e4ec905d", "run_group_id": "E4_core:A5|repairable|qwen3b:h8", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T06:31:04.129711Z", "task_family": "synthetic", "document_id": "syn_03000_2000", "document_hash": "c9cc27c20b05b1f6e592846e84990ee9b69c42f9720a8e338ec2f6fca777fdbc", "question_id": "q07_negation", "split": "test", "cluster_id": "syn_03000_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 8, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1904, "query_tokens": 444, "message_tokens": 363, "message_utf8_bytes": 848, "cache_tensor_bytes": null, "serialized_payload_bytes": 848, "side_information_bytes": 0, "fetched_bytes": 4828, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 133805.3459180519, "hops": {"1": 22956.80054195691, "2": 15833.571000024676, "3": 15950.301249977201, "4": 15643.779833917506, "5": 16041.412000078708, "6": 15729.852208052762, "7": 16016.10441703815, "8": 15633.524667005986}, "receiver_prefill_ms": 326.40445802826434, "ttft_ms": 122.76191695127636, "decode_ms": 38.080999976955354}, "total_wall_ms": 134293.11337706167, "ttft_ms": 122.76191695127636, "input_usage": {"lookups": 8}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 14, "retention_rate": 0.2153846153846154, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.09090909090909091, "amount": 0.16666666666666666, "status": 0.5, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 786}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 15, "retention_rate": 0.23076923076923078, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.09090909090909091, "amount": 0.16666666666666666, "status": 0.5, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 855}, "2": {"facts_total": 65, "facts_retained": 14, "retention_rate": 0.2153846153846154, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.09090909090909091, "amount": 0.16666666666666666, "status": 0.5, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 786}, "3": {"facts_total": 65, "facts_retained": 14, "retention_rate": 0.2153846153846154, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.09090909090909091, "amount": 0.16666666666666666, "status": 0.5, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 786}, "4": {"facts_total": 65, "facts_retained": 14, "retention_rate": 0.2153846153846154, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.09090909090909091, "amount": 0.16666666666666666, "status": 0.5, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 786}, "5": {"facts_total": 65, "facts_retained": 14, "retention_rate": 0.2153846153846154, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.09090909090909091, "amount": 0.16666666666666666, "status": 0.5, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 786}, "6": {"facts_total": 65, "facts_retained": 14, "retention_rate": 0.2153846153846154, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.09090909090909091, "amount": 0.16666666666666666, "status": 0.5, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 786}, "7": {"facts_total": 65, "facts_retained": 14, "retention_rate": 0.2153846153846154, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.09090909090909091, "amount": 0.16666666666666666, "status": 0.5, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 786}, "8": {"facts_total": 65, "facts_retained": 14, "retention_rate": 0.2153846153846154, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.09090909090909091, "amount": 0.16666666666666666, "status": 0.5, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 786}}}, "reference_key_id": "keys/synthetic/syn_03000_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "ffaf2ec78fb536984dc613873a800b081c1318d5d46ca5c4de575785729d8cbb", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "005d91a63e31c14e981a2877", "attempt_id": "7f030cd1ba4c", "run_group_id": "E4_core:A5|repairable|qwen3b:h8", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T06:31:04.634410Z", "task_family": "synthetic", "document_id": "syn_03000_2000", "document_hash": "c9cc27c20b05b1f6e592846e84990ee9b69c42f9720a8e338ec2f6fca777fdbc", "question_id": "q08_revision", "split": "test", "cluster_id": "syn_03000_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 8, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1904, "query_tokens": 459, "message_tokens": 363, "message_utf8_bytes": 848, "cache_tensor_bytes": null, "serialized_payload_bytes": 848, "side_information_bytes": 0, "fetched_bytes": 4828, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 133805.3459180519, "hops": {"1": 22956.80054195691, "2": 15833.571000024676, "3": 15950.301249977201, "4": 15643.779833917506, "5": 16041.412000078708, "6": 15729.852208052762, "7": 16016.10441703815, "8": 15633.524667005986}, "receiver_prefill_ms": 314.230167074129, "ttft_ms": 32.35849994234741, "decode_ms": 342.04316698014736}, "total_wall_ms": 134494.5201270748, "ttft_ms": 32.35849994234741, "input_usage": {"lookups": 8}, "output_usage": {"output_tokens": 9}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "2,705,965", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 14, "retention_rate": 0.2153846153846154, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.09090909090909091, "amount": 0.16666666666666666, "status": 0.5, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 786}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 15, "retention_rate": 0.23076923076923078, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.09090909090909091, "amount": 0.16666666666666666, "status": 0.5, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 855}, "2": {"facts_total": 65, "facts_retained": 14, "retention_rate": 0.2153846153846154, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.09090909090909091, "amount": 0.16666666666666666, "status": 0.5, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 786}, "3": {"facts_total": 65, "facts_retained": 14, "retention_rate": 0.2153846153846154, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.09090909090909091, "amount": 0.16666666666666666, "status": 0.5, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 786}, "4": {"facts_total": 65, "facts_retained": 14, "retention_rate": 0.2153846153846154, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.09090909090909091, "amount": 0.16666666666666666, "status": 0.5, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 786}, "5": {"facts_total": 65, "facts_retained": 14, "retention_rate": 0.2153846153846154, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.09090909090909091, "amount": 0.16666666666666666, "status": 0.5, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 786}, "6": {"facts_total": 65, "facts_retained": 14, "retention_rate": 0.2153846153846154, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.09090909090909091, "amount": 0.16666666666666666, "status": 0.5, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 786}, "7": {"facts_total": 65, "facts_retained": 14, "retention_rate": 0.2153846153846154, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.09090909090909091, "amount": 0.16666666666666666, "status": 0.5, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 786}, "8": {"facts_total": 65, "facts_retained": 14, "retention_rate": 0.2153846153846154, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.09090909090909091, "amount": 0.16666666666666666, "status": 0.5, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 786}}}, "reference_key_id": "keys/synthetic/syn_03000_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "ffaf2ec78fb536984dc613873a800b081c1318d5d46ca5c4de575785729d8cbb", "output_hash": "dbedb6f5d00e413d915513b85d8ed195b18b9acf392ece342a85e15e3752bcc8"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "e50cb825b50cf59fcba83801", "attempt_id": "33e7dcb595f5", "run_group_id": "E4_core:A5|repairable|qwen3b:h8", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T06:31:05.339942Z", "task_family": "synthetic", "document_id": "syn_03000_2000", "document_hash": "c9cc27c20b05b1f6e592846e84990ee9b69c42f9720a8e338ec2f6fca777fdbc", "question_id": "q09_revision", "split": "test", "cluster_id": "syn_03000_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 8, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1904, "query_tokens": 455, "message_tokens": 363, "message_utf8_bytes": 848, "cache_tensor_bytes": null, "serialized_payload_bytes": 848, "side_information_bytes": 0, "fetched_bytes": 4828, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 133805.3459180519, "hops": {"1": 22956.80054195691, "2": 15833.571000024676, "3": 15950.301249977201, "4": 15643.779833917506, "5": 16041.412000078708, "6": 15729.852208052762, "7": 16016.10441703815, "8": 15633.524667005986}, "receiver_prefill_ms": 276.582624996081, "ttft_ms": 48.72554203029722, "decode_ms": 60.80591701902449}, "total_wall_ms": 134192.3301680945, "ttft_ms": 48.72554203029722, "input_usage": {"lookups": 8}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 14, "retention_rate": 0.2153846153846154, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.09090909090909091, "amount": 0.16666666666666666, "status": 0.5, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 786}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 15, "retention_rate": 0.23076923076923078, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.09090909090909091, "amount": 0.16666666666666666, "status": 0.5, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 855}, "2": {"facts_total": 65, "facts_retained": 14, "retention_rate": 0.2153846153846154, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.09090909090909091, "amount": 0.16666666666666666, "status": 0.5, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 786}, "3": {"facts_total": 65, "facts_retained": 14, "retention_rate": 0.2153846153846154, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.09090909090909091, "amount": 0.16666666666666666, "status": 0.5, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 786}, "4": {"facts_total": 65, "facts_retained": 14, "retention_rate": 0.2153846153846154, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.09090909090909091, "amount": 0.16666666666666666, "status": 0.5, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 786}, "5": {"facts_total": 65, "facts_retained": 14, "retention_rate": 0.2153846153846154, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.09090909090909091, "amount": 0.16666666666666666, "status": 0.5, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 786}, "6": {"facts_total": 65, "facts_retained": 14, "retention_rate": 0.2153846153846154, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.09090909090909091, "amount": 0.16666666666666666, "status": 0.5, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 786}, "7": {"facts_total": 65, "facts_retained": 14, "retention_rate": 0.2153846153846154, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.09090909090909091, "amount": 0.16666666666666666, "status": 0.5, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 786}, "8": {"facts_total": 65, "facts_retained": 14, "retention_rate": 0.2153846153846154, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.09090909090909091, "amount": 0.16666666666666666, "status": 0.5, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 786}}}, "reference_key_id": "keys/synthetic/syn_03000_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "ffaf2ec78fb536984dc613873a800b081c1318d5d46ca5c4de575785729d8cbb", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "d22951f53c8976923effd82a", "attempt_id": "4c2250cae6aa", "run_group_id": "E4_core:A5|repairable|qwen3b:h8", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T06:31:05.743875Z", "task_family": "synthetic", "document_id": "syn_03000_2000", "document_hash": "c9cc27c20b05b1f6e592846e84990ee9b69c42f9720a8e338ec2f6fca777fdbc", "question_id": "q10_temporal", "split": "test", "cluster_id": "syn_03000_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 8, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1904, "query_tokens": 461, "message_tokens": 363, "message_utf8_bytes": 848, "cache_tensor_bytes": null, "serialized_payload_bytes": 848, "side_information_bytes": 0, "fetched_bytes": 4828, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 133805.3459180519, "hops": {"1": 22956.80054195691, "2": 15833.571000024676, "3": 15950.301249977201, "4": 15643.779833917506, "5": 16041.412000078708, "6": 15729.852208052762, "7": 16016.10441703815, "8": 15633.524667005986}, "receiver_prefill_ms": 477.51666605472565, "ttft_ms": 31.52404201682657, "decode_ms": 39.01791607495397}, "total_wall_ms": 134353.90912601724, "ttft_ms": 31.52404201682657, "input_usage": {"lookups": 8}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 14, "retention_rate": 0.2153846153846154, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.09090909090909091, "amount": 0.16666666666666666, "status": 0.5, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 786}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 15, "retention_rate": 0.23076923076923078, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.09090909090909091, "amount": 0.16666666666666666, "status": 0.5, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 855}, "2": {"facts_total": 65, "facts_retained": 14, "retention_rate": 0.2153846153846154, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.09090909090909091, "amount": 0.16666666666666666, "status": 0.5, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 786}, "3": {"facts_total": 65, "facts_retained": 14, "retention_rate": 0.2153846153846154, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.09090909090909091, "amount": 0.16666666666666666, "status": 0.5, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 786}, "4": {"facts_total": 65, "facts_retained": 14, "retention_rate": 0.2153846153846154, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.09090909090909091, "amount": 0.16666666666666666, "status": 0.5, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 786}, "5": {"facts_total": 65, "facts_retained": 14, "retention_rate": 0.2153846153846154, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.09090909090909091, "amount": 0.16666666666666666, "status": 0.5, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 786}, "6": {"facts_total": 65, "facts_retained": 14, "retention_rate": 0.2153846153846154, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.09090909090909091, "amount": 0.16666666666666666, "status": 0.5, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 786}, "7": {"facts_total": 65, "facts_retained": 14, "retention_rate": 0.2153846153846154, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.09090909090909091, "amount": 0.16666666666666666, "status": 0.5, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 786}, "8": {"facts_total": 65, "facts_retained": 14, "retention_rate": 0.2153846153846154, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.09090909090909091, "amount": 0.16666666666666666, "status": 0.5, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 786}}}, "reference_key_id": "keys/synthetic/syn_03000_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "ffaf2ec78fb536984dc613873a800b081c1318d5d46ca5c4de575785729d8cbb", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "1a6dacd32c2432c1b374d6a2", "attempt_id": "1c26f4f749b5", "run_group_id": "E4_core:A5|repairable|qwen3b:h8", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T06:31:06.309052Z", "task_family": "synthetic", "document_id": "syn_03000_2000", "document_hash": "c9cc27c20b05b1f6e592846e84990ee9b69c42f9720a8e338ec2f6fca777fdbc", "question_id": "q11_aggregation", "split": "test", "cluster_id": "syn_03000_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 8, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1904, "query_tokens": 456, "message_tokens": 363, "message_utf8_bytes": 848, "cache_tensor_bytes": null, "serialized_payload_bytes": 848, "side_information_bytes": 0, "fetched_bytes": 4828, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 133805.3459180519, "hops": {"1": 22956.80054195691, "2": 15833.571000024676, "3": 15950.301249977201, "4": 15643.779833917506, "5": 16041.412000078708, "6": 15729.852208052762, "7": 16016.10441703815, "8": 15633.524667005986}, "receiver_prefill_ms": 334.73870903253555, "ttft_ms": 39.183042012155056, "decode_ms": 25.161041994579136}, "total_wall_ms": 134204.9457510002, "ttft_ms": 39.183042012155056, "input_usage": {"lookups": 8}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 14, "retention_rate": 0.2153846153846154, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.09090909090909091, "amount": 0.16666666666666666, "status": 0.5, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 786}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 15, "retention_rate": 0.23076923076923078, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.09090909090909091, "amount": 0.16666666666666666, "status": 0.5, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 855}, "2": {"facts_total": 65, "facts_retained": 14, "retention_rate": 0.2153846153846154, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.09090909090909091, "amount": 0.16666666666666666, "status": 0.5, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 786}, "3": {"facts_total": 65, "facts_retained": 14, "retention_rate": 0.2153846153846154, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.09090909090909091, "amount": 0.16666666666666666, "status": 0.5, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 786}, "4": {"facts_total": 65, "facts_retained": 14, "retention_rate": 0.2153846153846154, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.09090909090909091, "amount": 0.16666666666666666, "status": 0.5, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 786}, "5": {"facts_total": 65, "facts_retained": 14, "retention_rate": 0.2153846153846154, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.09090909090909091, "amount": 0.16666666666666666, "status": 0.5, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 786}, "6": {"facts_total": 65, "facts_retained": 14, "retention_rate": 0.2153846153846154, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.09090909090909091, "amount": 0.16666666666666666, "status": 0.5, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 786}, "7": {"facts_total": 65, "facts_retained": 14, "retention_rate": 0.2153846153846154, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.09090909090909091, "amount": 0.16666666666666666, "status": 0.5, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 786}, "8": {"facts_total": 65, "facts_retained": 14, "retention_rate": 0.2153846153846154, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.09090909090909091, "amount": 0.16666666666666666, "status": 0.5, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 786}}}, "reference_key_id": "keys/synthetic/syn_03000_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "ffaf2ec78fb536984dc613873a800b081c1318d5d46ca5c4de575785729d8cbb", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "394934ab0774837762c9c8d5", "attempt_id": "a6163a16fb82", "run_group_id": "E4_core:A5|repairable|qwen3b:h8", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T06:31:06.725919Z", "task_family": "synthetic", "document_id": "syn_03000_2000", "document_hash": "c9cc27c20b05b1f6e592846e84990ee9b69c42f9720a8e338ec2f6fca777fdbc", "question_id": "q12_unit_conversion", "split": "test", "cluster_id": "syn_03000_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 8, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1904, "query_tokens": 452, "message_tokens": 363, "message_utf8_bytes": 848, "cache_tensor_bytes": null, "serialized_payload_bytes": 848, "side_information_bytes": 0, "fetched_bytes": 4828, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 133805.3459180519, "hops": {"1": 22956.80054195691, "2": 15833.571000024676, "3": 15950.301249977201, "4": 15643.779833917506, "5": 16041.412000078708, "6": 15729.852208052762, "7": 16016.10441703815, "8": 15633.524667005986}, "receiver_prefill_ms": 309.97895810287446, "ttft_ms": 21.231874939985573, "decode_ms": 160.82529199775308}, "total_wall_ms": 134297.94637614395, "ttft_ms": 21.231874939985573, "input_usage": {"lookups": 8}, "output_usage": {"output_tokens": 5}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "20.46", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 14, "retention_rate": 0.2153846153846154, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.09090909090909091, "amount": 0.16666666666666666, "status": 0.5, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 786}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 15, "retention_rate": 0.23076923076923078, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.09090909090909091, "amount": 0.16666666666666666, "status": 0.5, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 855}, "2": {"facts_total": 65, "facts_retained": 14, "retention_rate": 0.2153846153846154, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.09090909090909091, "amount": 0.16666666666666666, "status": 0.5, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 786}, "3": {"facts_total": 65, "facts_retained": 14, "retention_rate": 0.2153846153846154, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.09090909090909091, "amount": 0.16666666666666666, "status": 0.5, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 786}, "4": {"facts_total": 65, "facts_retained": 14, "retention_rate": 0.2153846153846154, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.09090909090909091, "amount": 0.16666666666666666, "status": 0.5, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 786}, "5": {"facts_total": 65, "facts_retained": 14, "retention_rate": 0.2153846153846154, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.09090909090909091, "amount": 0.16666666666666666, "status": 0.5, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 786}, "6": {"facts_total": 65, "facts_retained": 14, "retention_rate": 0.2153846153846154, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.09090909090909091, "amount": 0.16666666666666666, "status": 0.5, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 786}, "7": {"facts_total": 65, "facts_retained": 14, "retention_rate": 0.2153846153846154, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.09090909090909091, "amount": 0.16666666666666666, "status": 0.5, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 786}, "8": {"facts_total": 65, "facts_retained": 14, "retention_rate": 0.2153846153846154, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.09090909090909091, "amount": 0.16666666666666666, "status": 0.5, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 786}}}, "reference_key_id": "keys/synthetic/syn_03000_2000.json", "task_scores": {"correct": 0.0, "abstained": 0.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "ffaf2ec78fb536984dc613873a800b081c1318d5d46ca5c4de575785729d8cbb", "output_hash": "c057e11a5e818ca86a03c336373f3b4e4a2ddd9756e0acd7f5d7d655fd0267fe"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "ebcc468a919fe5557531bde5", "attempt_id": "346a4c74a6af", "run_group_id": "E4_core:A5|repairable|qwen3b:h8", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T06:31:07.235103Z", "task_family": "synthetic", "document_id": "syn_03000_2000", "document_hash": "c9cc27c20b05b1f6e592846e84990ee9b69c42f9720a8e338ec2f6fca777fdbc", "question_id": "q13_unanswerable", "split": "test", "cluster_id": "syn_03000_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 8, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1904, "query_tokens": 456, "message_tokens": 363, "message_utf8_bytes": 848, "cache_tensor_bytes": null, "serialized_payload_bytes": 848, "side_information_bytes": 0, "fetched_bytes": 4828, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 133805.3459180519, "hops": {"1": 22956.80054195691, "2": 15833.571000024676, "3": 15950.301249977201, "4": 15643.779833917506, "5": 16041.412000078708, "6": 15729.852208052762, "7": 16016.10441703815, "8": 15633.524667005986}, "receiver_prefill_ms": 360.3075409773737, "ttft_ms": 91.79145796224475, "decode_ms": 143.76279199495912}, "total_wall_ms": 134401.71858505346, "ttft_ms": 91.79145796224475, "input_usage": {"lookups": 8}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 14, "retention_rate": 0.2153846153846154, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.09090909090909091, "amount": 0.16666666666666666, "status": 0.5, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 786}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 15, "retention_rate": 0.23076923076923078, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.09090909090909091, "amount": 0.16666666666666666, "status": 0.5, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 855}, "2": {"facts_total": 65, "facts_retained": 14, "retention_rate": 0.2153846153846154, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.09090909090909091, "amount": 0.16666666666666666, "status": 0.5, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 786}, "3": {"facts_total": 65, "facts_retained": 14, "retention_rate": 0.2153846153846154, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.09090909090909091, "amount": 0.16666666666666666, "status": 0.5, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 786}, "4": {"facts_total": 65, "facts_retained": 14, "retention_rate": 0.2153846153846154, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.09090909090909091, "amount": 0.16666666666666666, "status": 0.5, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 786}, "5": {"facts_total": 65, "facts_retained": 14, "retention_rate": 0.2153846153846154, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.09090909090909091, "amount": 0.16666666666666666, "status": 0.5, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 786}, "6": {"facts_total": 65, "facts_retained": 14, "retention_rate": 0.2153846153846154, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.09090909090909091, "amount": 0.16666666666666666, "status": 0.5, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 786}, "7": {"facts_total": 65, "facts_retained": 14, "retention_rate": 0.2153846153846154, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.09090909090909091, "amount": 0.16666666666666666, "status": 0.5, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 786}, "8": {"facts_total": 65, "facts_retained": 14, "retention_rate": 0.2153846153846154, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.09090909090909091, "amount": 0.16666666666666666, "status": 0.5, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 786}}}, "reference_key_id": "keys/synthetic/syn_03000_2000.json", "task_scores": {"correct": 1.0, "abstained": 1.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "ffaf2ec78fb536984dc613873a800b081c1318d5d46ca5c4de575785729d8cbb", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "f35f687059ca31ef446b29d9", "attempt_id": "f6987491a1f5", "run_group_id": "E4_core:A5|repairable|qwen3b:h8", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T06:31:07.849653Z", "task_family": "synthetic", "document_id": "syn_03000_2000", "document_hash": "c9cc27c20b05b1f6e592846e84990ee9b69c42f9720a8e338ec2f6fca777fdbc", "question_id": "q14_unanswerable", "split": "test", "cluster_id": "syn_03000_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 8, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1904, "query_tokens": 458, "message_tokens": 363, "message_utf8_bytes": 848, "cache_tensor_bytes": null, "serialized_payload_bytes": 848, "side_information_bytes": 0, "fetched_bytes": 4828, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 133805.3459180519, "hops": {"1": 22956.80054195691, "2": 15833.571000024676, "3": 15950.301249977201, "4": 15643.779833917506, "5": 16041.412000078708, "6": 15729.852208052762, "7": 16016.10441703815, "8": 15633.524667005986}, "receiver_prefill_ms": 317.2491660807282, "ttft_ms": 41.32320801727474, "decode_ms": 34.59729207679629}, "total_wall_ms": 134198.98970914073, "ttft_ms": 41.32320801727474, "input_usage": {"lookups": 8}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 14, "retention_rate": 0.2153846153846154, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.09090909090909091, "amount": 0.16666666666666666, "status": 0.5, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 786}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 15, "retention_rate": 0.23076923076923078, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.09090909090909091, "amount": 0.16666666666666666, "status": 0.5, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 855}, "2": {"facts_total": 65, "facts_retained": 14, "retention_rate": 0.2153846153846154, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.09090909090909091, "amount": 0.16666666666666666, "status": 0.5, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 786}, "3": {"facts_total": 65, "facts_retained": 14, "retention_rate": 0.2153846153846154, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.09090909090909091, "amount": 0.16666666666666666, "status": 0.5, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 786}, "4": {"facts_total": 65, "facts_retained": 14, "retention_rate": 0.2153846153846154, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.09090909090909091, "amount": 0.16666666666666666, "status": 0.5, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 786}, "5": {"facts_total": 65, "facts_retained": 14, "retention_rate": 0.2153846153846154, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.09090909090909091, "amount": 0.16666666666666666, "status": 0.5, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 786}, "6": {"facts_total": 65, "facts_retained": 14, "retention_rate": 0.2153846153846154, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.09090909090909091, "amount": 0.16666666666666666, "status": 0.5, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 786}, "7": {"facts_total": 65, "facts_retained": 14, "retention_rate": 0.2153846153846154, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.09090909090909091, "amount": 0.16666666666666666, "status": 0.5, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 786}, "8": {"facts_total": 65, "facts_retained": 14, "retention_rate": 0.2153846153846154, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.09090909090909091, "amount": 0.16666666666666666, "status": 0.5, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 786}}}, "reference_key_id": "keys/synthetic/syn_03000_2000.json", "task_scores": {"correct": 1.0, "abstained": 1.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "ffaf2ec78fb536984dc613873a800b081c1318d5d46ca5c4de575785729d8cbb", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "f0d39a04710791826f5568f8", "attempt_id": "c75882ea6e8a", "run_group_id": "E4_core:A4|closed|qwen3b-llama3b:h1", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T06:31:36.459629Z", "task_family": "synthetic", "document_id": "syn_03000_2000", "document_hash": "c9cc27c20b05b1f6e592846e84990ee9b69c42f9720a8e338ec2f6fca777fdbc", "question_id": "q00_single_fact", "split": "test", "cluster_id": "syn_03000_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 1, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1904, "query_tokens": 592, "message_tokens": 511, "message_utf8_bytes": 1090, "cache_tensor_bytes": null, "serialized_payload_bytes": 1090, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 22739.89845905453, "hops": {"1": 22739.89845905453}, "receiver_prefill_ms": 349.8036250239238, "ttft_ms": 46.911833924241364, "decode_ms": 52.807291969656944}, "total_wall_ms": 23190.144375083037, "ttft_ms": 46.911833924241364, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 29, "retention_rate": 0.4461538461538462, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5, "status": 1.0, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1090}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 29, "retention_rate": 0.4461538461538462, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5, "status": 1.0, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1090}}}, "reference_key_id": "keys/synthetic/syn_03000_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "562010df7e3ceb9bcd9a843b326413164596adc2f7ca548f1a2eb3ede9dfbb44", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "03a0876f73c19e0e8aa03dfb", "attempt_id": "82f7763b1a88", "run_group_id": "E4_core:A4|closed|qwen3b-llama3b:h1", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T06:31:36.928207Z", "task_family": "synthetic", "document_id": "syn_03000_2000", "document_hash": "c9cc27c20b05b1f6e592846e84990ee9b69c42f9720a8e338ec2f6fca777fdbc", "question_id": "q01_single_fact", "split": "test", "cluster_id": "syn_03000_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 1, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1904, "query_tokens": 597, "message_tokens": 511, "message_utf8_bytes": 1090, "cache_tensor_bytes": null, "serialized_payload_bytes": 1090, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 22739.89845905453, "hops": {"1": 22739.89845905453}, "receiver_prefill_ms": 519.2318750778213, "ttft_ms": 44.524290948174894, "decode_ms": 300.3529169363901}, "total_wall_ms": 23604.762083967216, "ttft_ms": 44.524290948174894, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 9}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "Lot 19-N, Gaspé", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 29, "retention_rate": 0.4461538461538462, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5, "status": 1.0, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1090}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 29, "retention_rate": 0.4461538461538462, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5, "status": 1.0, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1090}}}, "reference_key_id": "keys/synthetic/syn_03000_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 0.8571428571428571}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "562010df7e3ceb9bcd9a843b326413164596adc2f7ca548f1a2eb3ede9dfbb44", "output_hash": "2e8cf782bd118a1c6d4e23417785d72fb92e930806b5752cd1ec405a0c9b95e5"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "2a694ce175b3052218e2b9d1", "attempt_id": "bfa486c135d9", "run_group_id": "E4_core:A4|closed|qwen3b-llama3b:h1", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T06:31:37.811507Z", "task_family": "synthetic", "document_id": "syn_03000_2000", "document_hash": "c9cc27c20b05b1f6e592846e84990ee9b69c42f9720a8e338ec2f6fca777fdbc", "question_id": "q02_exact_copy", "split": "test", "cluster_id": "syn_03000_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 1, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1904, "query_tokens": 596, "message_tokens": 511, "message_utf8_bytes": 1090, "cache_tensor_bytes": null, "serialized_payload_bytes": 1090, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 22739.89845905453, "hops": {"1": 22739.89845905453}, "receiver_prefill_ms": 408.2035419996828, "ttft_ms": 22.887709084898233, "decode_ms": 42.31070808600634}, "total_wall_ms": 23213.96387508139, "ttft_ms": 22.887709084898233, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 29, "retention_rate": 0.4461538461538462, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5, "status": 1.0, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1090}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 29, "retention_rate": 0.4461538461538462, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5, "status": 1.0, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1090}}}, "reference_key_id": "keys/synthetic/syn_03000_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "562010df7e3ceb9bcd9a843b326413164596adc2f7ca548f1a2eb3ede9dfbb44", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "4250d24c2619005e062246ef", "attempt_id": "21c983a1b6c9", "run_group_id": "E4_core:A4|closed|qwen3b-llama3b:h1", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T06:31:38.303411Z", "task_family": "synthetic", "document_id": "syn_03000_2000", "document_hash": "c9cc27c20b05b1f6e592846e84990ee9b69c42f9720a8e338ec2f6fca777fdbc", "question_id": "q03_exact_copy", "split": "test", "cluster_id": "syn_03000_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 1, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1904, "query_tokens": 595, "message_tokens": 511, "message_utf8_bytes": 1090, "cache_tensor_bytes": null, "serialized_payload_bytes": 1090, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 22739.89845905453, "hops": {"1": 22739.89845905453}, "receiver_prefill_ms": 359.2195000965148, "ttft_ms": 43.0368329398334, "decode_ms": 463.3370420197025}, "total_wall_ms": 23606.173000996932, "ttft_ms": 43.0368329398334, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 7}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "Kwame Haldorsen", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 29, "retention_rate": 0.4461538461538462, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5, "status": 1.0, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1090}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 29, "retention_rate": 0.4461538461538462, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5, "status": 1.0, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1090}}}, "reference_key_id": "keys/synthetic/syn_03000_2000.json", "task_scores": {"correct": 0.0, "abstained": 0.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "562010df7e3ceb9bcd9a843b326413164596adc2f7ca548f1a2eb3ede9dfbb44", "output_hash": "175253e7cf463934ec4d54e361a781f323cd87290f01306ed368933bcbd2e25d"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "e3b45016b008de101e15db79", "attempt_id": "02db33113a6a", "run_group_id": "E4_core:A4|closed|qwen3b-llama3b:h1", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T06:31:39.187405Z", "task_family": "synthetic", "document_id": "syn_03000_2000", "document_hash": "c9cc27c20b05b1f6e592846e84990ee9b69c42f9720a8e338ec2f6fca777fdbc", "question_id": "q04_multi_hop", "split": "test", "cluster_id": "syn_03000_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 1, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1904, "query_tokens": 601, "message_tokens": 511, "message_utf8_bytes": 1090, "cache_tensor_bytes": null, "serialized_payload_bytes": 1090, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 22739.89845905453, "hops": {"1": 22739.89845905453}, "receiver_prefill_ms": 400.2849590033293, "ttft_ms": 42.89024998433888, "decode_ms": 282.92687505017966}, "total_wall_ms": 23466.631459072232, "ttft_ms": 42.89024998433888, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 9}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "9,220,745", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 29, "retention_rate": 0.4461538461538462, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5, "status": 1.0, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1090}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 29, "retention_rate": 0.4461538461538462, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5, "status": 1.0, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1090}}}, "reference_key_id": "keys/synthetic/syn_03000_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "562010df7e3ceb9bcd9a843b326413164596adc2f7ca548f1a2eb3ede9dfbb44", "output_hash": "ae55ba88b2cb1b0f1a4e0960ae312e0a1c177c79a6d14b67bff1677cec19b649"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "026728888a746828af9f7d1c", "attempt_id": "74747c4f5caf", "run_group_id": "E4_core:A4|closed|qwen3b-llama3b:h1", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T06:31:39.931616Z", "task_family": "synthetic", "document_id": "syn_03000_2000", "document_hash": "c9cc27c20b05b1f6e592846e84990ee9b69c42f9720a8e338ec2f6fca777fdbc", "question_id": "q05_multi_hop", "split": "test", "cluster_id": "syn_03000_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 1, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1904, "query_tokens": 603, "message_tokens": 511, "message_utf8_bytes": 1090, "cache_tensor_bytes": null, "serialized_payload_bytes": 1090, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 22739.89845905453, "hops": {"1": 22739.89845905453}, "receiver_prefill_ms": 426.20866699144244, "ttft_ms": 45.56225007399917, "decode_ms": 527.3226669523865}, "total_wall_ms": 23739.609334035777, "ttft_ms": 45.56225007399917, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 10}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "LY-51E-3199", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 29, "retention_rate": 0.4461538461538462, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5, "status": 1.0, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1090}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 29, "retention_rate": 0.4461538461538462, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5, "status": 1.0, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1090}}}, "reference_key_id": "keys/synthetic/syn_03000_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "562010df7e3ceb9bcd9a843b326413164596adc2f7ca548f1a2eb3ede9dfbb44", "output_hash": "a1245d1536890a7e4b026366df48441d503f491f782cfd388797b5782443c5c7"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "d7019c269803956eed0f9e9d", "attempt_id": "f54226c35689", "run_group_id": "E4_core:A4|closed|qwen3b-llama3b:h1", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T06:31:40.949425Z", "task_family": "synthetic", "document_id": "syn_03000_2000", "document_hash": "c9cc27c20b05b1f6e592846e84990ee9b69c42f9720a8e338ec2f6fca777fdbc", "question_id": "q06_negation", "split": "test", "cluster_id": "syn_03000_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 1, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1904, "query_tokens": 592, "message_tokens": 511, "message_utf8_bytes": 1090, "cache_tensor_bytes": null, "serialized_payload_bytes": 1090, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 22739.89845905453, "hops": {"1": 22739.89845905453}, "receiver_prefill_ms": 418.2868750067428, "ttft_ms": 37.108249962329865, "decode_ms": 29.205959057435393}, "total_wall_ms": 23225.0419590855, "ttft_ms": 37.108249962329865, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "no", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 29, "retention_rate": 0.4461538461538462, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5, "status": 1.0, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1090}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 29, "retention_rate": 0.4461538461538462, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5, "status": 1.0, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1090}}}, "reference_key_id": "keys/synthetic/syn_03000_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "562010df7e3ceb9bcd9a843b326413164596adc2f7ca548f1a2eb3ede9dfbb44", "output_hash": "9390298f3fb0c5b160498935d79cb139aef28e1c47358b4bbba61862b9c26e59"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "da40ba2052dd9e5e647a8a2a", "attempt_id": "19a4a05763e1", "run_group_id": "E4_core:A4|closed|qwen3b-llama3b:h1", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T06:31:41.452186Z", "task_family": "synthetic", "document_id": "syn_03000_2000", "document_hash": "c9cc27c20b05b1f6e592846e84990ee9b69c42f9720a8e338ec2f6fca777fdbc", "question_id": "q07_negation", "split": "test", "cluster_id": "syn_03000_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 1, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1904, "query_tokens": 592, "message_tokens": 511, "message_utf8_bytes": 1090, "cache_tensor_bytes": null, "serialized_payload_bytes": 1090, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 22739.89845905453, "hops": {"1": 22739.89845905453}, "receiver_prefill_ms": 355.05433299113065, "ttft_ms": 54.4337920146063, "decode_ms": 44.23133295495063}, "total_wall_ms": 23194.272001157515, "ttft_ms": 54.4337920146063, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "no", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 29, "retention_rate": 0.4461538461538462, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5, "status": 1.0, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1090}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 29, "retention_rate": 0.4461538461538462, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5, "status": 1.0, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1090}}}, "reference_key_id": "keys/synthetic/syn_03000_2000.json", "task_scores": {"correct": 0.0, "abstained": 0.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "562010df7e3ceb9bcd9a843b326413164596adc2f7ca548f1a2eb3ede9dfbb44", "output_hash": "9390298f3fb0c5b160498935d79cb139aef28e1c47358b4bbba61862b9c26e59"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "4baa733f46586286d6a1d038", "attempt_id": "2ddb97274fcb", "run_group_id": "E4_core:A4|closed|qwen3b-llama3b:h1", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T06:31:41.923514Z", "task_family": "synthetic", "document_id": "syn_03000_2000", "document_hash": "c9cc27c20b05b1f6e592846e84990ee9b69c42f9720a8e338ec2f6fca777fdbc", "question_id": "q08_revision", "split": "test", "cluster_id": "syn_03000_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 1, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1904, "query_tokens": 607, "message_tokens": 511, "message_utf8_bytes": 1090, "cache_tensor_bytes": null, "serialized_payload_bytes": 1090, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 22739.89845905453, "hops": {"1": 22739.89845905453}, "receiver_prefill_ms": 524.1343749221414, "ttft_ms": 37.61595790274441, "decode_ms": 315.74008299503475}, "total_wall_ms": 23618.10450104531, "ttft_ms": 37.61595790274441, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 9}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "2,705,965", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 29, "retention_rate": 0.4461538461538462, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5, "status": 1.0, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1090}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 29, "retention_rate": 0.4461538461538462, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5, "status": 1.0, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1090}}}, "reference_key_id": "keys/synthetic/syn_03000_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "562010df7e3ceb9bcd9a843b326413164596adc2f7ca548f1a2eb3ede9dfbb44", "output_hash": "dbedb6f5d00e413d915513b85d8ed195b18b9acf392ece342a85e15e3752bcc8"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "60e80522b9161db2aeef120b", "attempt_id": "c8f8ee9115d7", "run_group_id": "E4_core:A4|closed|qwen3b-llama3b:h1", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T06:31:42.818752Z", "task_family": "synthetic", "document_id": "syn_03000_2000", "document_hash": "c9cc27c20b05b1f6e592846e84990ee9b69c42f9720a8e338ec2f6fca777fdbc", "question_id": "q09_revision", "split": "test", "cluster_id": "syn_03000_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 1, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1904, "query_tokens": 603, "message_tokens": 511, "message_utf8_bytes": 1090, "cache_tensor_bytes": null, "serialized_payload_bytes": 1090, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 22739.89845905453, "hops": {"1": 22739.89845905453}, "receiver_prefill_ms": 402.1519999951124, "ttft_ms": 23.916958016343415, "decode_ms": 376.3976249611005}, "total_wall_ms": 23543.04795898497, "ttft_ms": 23.916958016343415, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 7}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "9127965", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 29, "retention_rate": 0.4461538461538462, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5, "status": 1.0, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1090}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 29, "retention_rate": 0.4461538461538462, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5, "status": 1.0, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1090}}}, "reference_key_id": "keys/synthetic/syn_03000_2000.json", "task_scores": {"correct": 0.0, "abstained": 0.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "562010df7e3ceb9bcd9a843b326413164596adc2f7ca548f1a2eb3ede9dfbb44", "output_hash": "a0c871941224304cec47bb1b3d64b8706843b299f14ab94899d97df14f433da7"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "ce1490f700b1a08ee0fb72a3", "attempt_id": "2f90fd96eab4", "run_group_id": "E4_core:A4|closed|qwen3b-llama3b:h1", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T06:31:43.638671Z", "task_family": "synthetic", "document_id": "syn_03000_2000", "document_hash": "c9cc27c20b05b1f6e592846e84990ee9b69c42f9720a8e338ec2f6fca777fdbc", "question_id": "q10_temporal", "split": "test", "cluster_id": "syn_03000_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 1, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1904, "query_tokens": 609, "message_tokens": 511, "message_utf8_bytes": 1090, "cache_tensor_bytes": null, "serialized_payload_bytes": 1090, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 22739.89845905453, "hops": {"1": 22739.89845905453}, "receiver_prefill_ms": 533.5919170174748, "ttft_ms": 27.343833004124463, "decode_ms": 176.83787504211068}, "total_wall_ms": 23478.291375096887, "ttft_ms": 27.343833004124463, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 5}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "Felaldnak-7", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 29, "retention_rate": 0.4461538461538462, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5, "status": 1.0, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1090}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 29, "retention_rate": 0.4461538461538462, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5, "status": 1.0, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1090}}}, "reference_key_id": "keys/synthetic/syn_03000_2000.json", "task_scores": {"correct": 0.0, "abstained": 0.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "562010df7e3ceb9bcd9a843b326413164596adc2f7ca548f1a2eb3ede9dfbb44", "output_hash": "2b9a4cf7221f1b6f31508a6b3078ec521f211026a932da5020cddd512ecc5fa2"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "31423b4b39ff7445d38fa5d8", "attempt_id": "a91aa0bb9414", "run_group_id": "E4_core:A4|closed|qwen3b-llama3b:h1", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T06:31:44.394232Z", "task_family": "synthetic", "document_id": "syn_03000_2000", "document_hash": "c9cc27c20b05b1f6e592846e84990ee9b69c42f9720a8e338ec2f6fca777fdbc", "question_id": "q11_aggregation", "split": "test", "cluster_id": "syn_03000_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 1, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1904, "query_tokens": 604, "message_tokens": 511, "message_utf8_bytes": 1090, "cache_tensor_bytes": null, "serialized_payload_bytes": 1090, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 22739.89845905453, "hops": {"1": 22739.89845905453}, "receiver_prefill_ms": 397.177750011906, "ttft_ms": 24.082500021904707, "decode_ms": 324.37612500507385}, "total_wall_ms": 23486.355834058486, "ttft_ms": 24.082500021904707, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 7}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "9220745", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 29, "retention_rate": 0.4461538461538462, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5, "status": 1.0, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1090}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 29, "retention_rate": 0.4461538461538462, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5, "status": 1.0, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1090}}}, "reference_key_id": "keys/synthetic/syn_03000_2000.json", "task_scores": {"correct": 0.0, "abstained": 0.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "562010df7e3ceb9bcd9a843b326413164596adc2f7ca548f1a2eb3ede9dfbb44", "output_hash": "f5d77aa763f6365bc80d01b9e6e23bc43fb570ac45771d7bd628c75cde727b35"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "2993794ba4b03b58ec53da09", "attempt_id": "621f5ff1503c", "run_group_id": "E4_core:A4|closed|qwen3b-llama3b:h1", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T06:31:45.158456Z", "task_family": "synthetic", "document_id": "syn_03000_2000", "document_hash": "c9cc27c20b05b1f6e592846e84990ee9b69c42f9720a8e338ec2f6fca777fdbc", "question_id": "q12_unit_conversion", "split": "test", "cluster_id": "syn_03000_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 1, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1904, "query_tokens": 600, "message_tokens": 511, "message_utf8_bytes": 1090, "cache_tensor_bytes": null, "serialized_payload_bytes": 1090, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 22739.89845905453, "hops": {"1": 22739.89845905453}, "receiver_prefill_ms": 524.4762500515208, "ttft_ms": 41.55570804141462, "decode_ms": 30.207332922145724}, "total_wall_ms": 23336.886291974224, "ttft_ms": 41.55570804141462, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 29, "retention_rate": 0.4461538461538462, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5, "status": 1.0, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1090}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 29, "retention_rate": 0.4461538461538462, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5, "status": 1.0, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1090}}}, "reference_key_id": "keys/synthetic/syn_03000_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "562010df7e3ceb9bcd9a843b326413164596adc2f7ca548f1a2eb3ede9dfbb44", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "ca7edf27e556fb839950f868", "attempt_id": "c926736576e2", "run_group_id": "E4_core:A4|closed|qwen3b-llama3b:h1", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T06:31:45.772613Z", "task_family": "synthetic", "document_id": "syn_03000_2000", "document_hash": "c9cc27c20b05b1f6e592846e84990ee9b69c42f9720a8e338ec2f6fca777fdbc", "question_id": "q13_unanswerable", "split": "test", "cluster_id": "syn_03000_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 1, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1904, "query_tokens": 604, "message_tokens": 511, "message_utf8_bytes": 1090, "cache_tensor_bytes": null, "serialized_payload_bytes": 1090, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 22739.89845905453, "hops": {"1": 22739.89845905453}, "receiver_prefill_ms": 391.88875001855195, "ttft_ms": 40.140458033420146, "decode_ms": 25.673083029687405}, "total_wall_ms": 23198.191167088225, "ttft_ms": 40.140458033420146, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 29, "retention_rate": 0.4461538461538462, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5, "status": 1.0, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1090}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 29, "retention_rate": 0.4461538461538462, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5, "status": 1.0, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1090}}}, "reference_key_id": "keys/synthetic/syn_03000_2000.json", "task_scores": {"correct": 1.0, "abstained": 1.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "562010df7e3ceb9bcd9a843b326413164596adc2f7ca548f1a2eb3ede9dfbb44", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "39ccf3aa041aa7a6a387f16a", "attempt_id": "ac026d9e9e02", "run_group_id": "E4_core:A4|closed|qwen3b-llama3b:h1", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T06:31:46.247324Z", "task_family": "synthetic", "document_id": "syn_03000_2000", "document_hash": "c9cc27c20b05b1f6e592846e84990ee9b69c42f9720a8e338ec2f6fca777fdbc", "question_id": "q14_unanswerable", "split": "test", "cluster_id": "syn_03000_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 1, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1904, "query_tokens": 606, "message_tokens": 511, "message_utf8_bytes": 1090, "cache_tensor_bytes": null, "serialized_payload_bytes": 1090, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 22739.89845905453, "hops": {"1": 22739.89845905453}, "receiver_prefill_ms": 370.8214160287753, "ttft_ms": 50.088874995708466, "decode_ms": 54.219207959249616}, "total_wall_ms": 23215.704584144987, "ttft_ms": 50.088874995708466, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 29, "retention_rate": 0.4461538461538462, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5, "status": 1.0, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1090}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 29, "retention_rate": 0.4461538461538462, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5, "status": 1.0, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1090}}}, "reference_key_id": "keys/synthetic/syn_03000_2000.json", "task_scores": {"correct": 1.0, "abstained": 1.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "562010df7e3ceb9bcd9a843b326413164596adc2f7ca548f1a2eb3ede9dfbb44", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "0a583db0b97835854c7caa94", "attempt_id": "bb52c60d4fe1", "run_group_id": "E4_core:A4|closed|qwen3b-llama3b:h2", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T06:32:03.353015Z", "task_family": "synthetic", "document_id": "syn_03000_2000", "document_hash": "c9cc27c20b05b1f6e592846e84990ee9b69c42f9720a8e338ec2f6fca777fdbc", "question_id": "q00_single_fact", "split": "test", "cluster_id": "syn_03000_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "tokenizer_hashes": {"receiver": "b9d8872cf3a99da99e1fc4b0f71232a451632dc49ca4c6771549b4efa5c1c763"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 2, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1715, "query_tokens": 525, "message_tokens": 422, "message_utf8_bytes": 1287, "cache_tensor_bytes": null, "serialized_payload_bytes": 1287, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 39351.19591699913, "hops": {"1": 22739.89845905453, "2": 16611.2974579446}, "receiver_prefill_ms": 363.9373329933733, "ttft_ms": 85.30466700904071, "decode_ms": 123.32920799963176}, "total_wall_ms": 39924.55058405176, "ttft_ms": 85.30466700904071, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 29, "retention_rate": 0.4461538461538462, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5, "status": 1.0, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1287}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 29, "retention_rate": 0.4461538461538462, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5, "status": 1.0, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1090}, "2": {"facts_total": 65, "facts_retained": 29, "retention_rate": 0.4461538461538462, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5, "status": 1.0, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1287}}}, "reference_key_id": "keys/synthetic/syn_03000_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "7d13c3599ad215a84dcdeaa88517f112a834b20d3470dddfe0f02da2fd7cec73", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "b1e71ce6ffd30617ca499280", "attempt_id": "2924d9b8cd5d", "run_group_id": "E4_core:A4|closed|qwen3b-llama3b:h2", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T06:32:03.943942Z", "task_family": "synthetic", "document_id": "syn_03000_2000", "document_hash": "c9cc27c20b05b1f6e592846e84990ee9b69c42f9720a8e338ec2f6fca777fdbc", "question_id": "q01_single_fact", "split": "test", "cluster_id": "syn_03000_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "tokenizer_hashes": {"receiver": "b9d8872cf3a99da99e1fc4b0f71232a451632dc49ca4c6771549b4efa5c1c763"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 2, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1715, "query_tokens": 530, "message_tokens": 422, "message_utf8_bytes": 1287, "cache_tensor_bytes": null, "serialized_payload_bytes": 1287, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 39351.19591699913, "hops": {"1": 22739.89845905453, "2": 16611.2974579446}, "receiver_prefill_ms": 329.9567500362173, "ttft_ms": 26.667416910640895, "decode_ms": 293.7521660933271}, "total_wall_ms": 40002.208458958194, "ttft_ms": 26.667416910640895, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 9}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "Lot 19-N, Gaspé.", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 29, "retention_rate": 0.4461538461538462, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5, "status": 1.0, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1287}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 29, "retention_rate": 0.4461538461538462, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5, "status": 1.0, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1090}, "2": {"facts_total": 65, "facts_retained": 29, "retention_rate": 0.4461538461538462, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5, "status": 1.0, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1287}}}, "reference_key_id": "keys/synthetic/syn_03000_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 0.8571428571428571}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "7d13c3599ad215a84dcdeaa88517f112a834b20d3470dddfe0f02da2fd7cec73", "output_hash": "ab9cc24b811f5b1342aae731f584e8b0d24b63b61ba59a068b8414deffdc10ae"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "d16f6f623219d6a8d3f5c0be", "attempt_id": "5a8ae7d068ca", "run_group_id": "E4_core:A4|closed|qwen3b-llama3b:h2", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T06:32:04.611499Z", "task_family": "synthetic", "document_id": "syn_03000_2000", "document_hash": "c9cc27c20b05b1f6e592846e84990ee9b69c42f9720a8e338ec2f6fca777fdbc", "question_id": "q02_exact_copy", "split": "test", "cluster_id": "syn_03000_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "tokenizer_hashes": {"receiver": "b9d8872cf3a99da99e1fc4b0f71232a451632dc49ca4c6771549b4efa5c1c763"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 2, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1715, "query_tokens": 529, "message_tokens": 422, "message_utf8_bytes": 1287, "cache_tensor_bytes": null, "serialized_payload_bytes": 1287, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 39351.19591699913, "hops": {"1": 22739.89845905453, "2": 16611.2974579446}, "receiver_prefill_ms": 311.2165000056848, "ttft_ms": 46.184708015061915, "decode_ms": 62.15283402707428}, "total_wall_ms": 39771.63208392449, "ttft_ms": 46.184708015061915, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 29, "retention_rate": 0.4461538461538462, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5, "status": 1.0, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1287}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 29, "retention_rate": 0.4461538461538462, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5, "status": 1.0, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1090}, "2": {"facts_total": 65, "facts_retained": 29, "retention_rate": 0.4461538461538462, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5, "status": 1.0, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1287}}}, "reference_key_id": "keys/synthetic/syn_03000_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "7d13c3599ad215a84dcdeaa88517f112a834b20d3470dddfe0f02da2fd7cec73", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "44e06c3698c1daf351ef7047", "attempt_id": "4ea827678619", "run_group_id": "E4_core:A4|closed|qwen3b-llama3b:h2", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T06:32:05.049181Z", "task_family": "synthetic", "document_id": "syn_03000_2000", "document_hash": "c9cc27c20b05b1f6e592846e84990ee9b69c42f9720a8e338ec2f6fca777fdbc", "question_id": "q03_exact_copy", "split": "test", "cluster_id": "syn_03000_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "tokenizer_hashes": {"receiver": "b9d8872cf3a99da99e1fc4b0f71232a451632dc49ca4c6771549b4efa5c1c763"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 2, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1715, "query_tokens": 527, "message_tokens": 422, "message_utf8_bytes": 1287, "cache_tensor_bytes": null, "serialized_payload_bytes": 1287, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 39351.19591699913, "hops": {"1": 22739.89845905453, "2": 16611.2974579446}, "receiver_prefill_ms": 326.6723749693483, "ttft_ms": 101.29962500650436, "decode_ms": 146.316249971278}, "total_wall_ms": 39926.066417014226, "ttft_ms": 101.29962500650436, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 29, "retention_rate": 0.4461538461538462, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5, "status": 1.0, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1287}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 29, "retention_rate": 0.4461538461538462, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5, "status": 1.0, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1090}, "2": {"facts_total": 65, "facts_retained": 29, "retention_rate": 0.4461538461538462, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5, "status": 1.0, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1287}}}, "reference_key_id": "keys/synthetic/syn_03000_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "7d13c3599ad215a84dcdeaa88517f112a834b20d3470dddfe0f02da2fd7cec73", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "d13c54c1f22c3268a7ae407a", "attempt_id": "df372866237d", "run_group_id": "E4_core:A4|closed|qwen3b-llama3b:h2", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T06:32:05.640976Z", "task_family": "synthetic", "document_id": "syn_03000_2000", "document_hash": "c9cc27c20b05b1f6e592846e84990ee9b69c42f9720a8e338ec2f6fca777fdbc", "question_id": "q04_multi_hop", "split": "test", "cluster_id": "syn_03000_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "tokenizer_hashes": {"receiver": "b9d8872cf3a99da99e1fc4b0f71232a451632dc49ca4c6771549b4efa5c1c763"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 2, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1715, "query_tokens": 534, "message_tokens": 422, "message_utf8_bytes": 1287, "cache_tensor_bytes": null, "serialized_payload_bytes": 1287, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 39351.19591699913, "hops": {"1": 22739.89845905453, "2": 16611.2974579446}, "receiver_prefill_ms": 332.17762503772974, "ttft_ms": 32.07054198719561, "decode_ms": 160.61908297706395}, "total_wall_ms": 39876.71312503517, "ttft_ms": 32.07054198719561, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 5}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "9,220,745", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 29, "retention_rate": 0.4461538461538462, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5, "status": 1.0, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1287}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 29, "retention_rate": 0.4461538461538462, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5, "status": 1.0, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1090}, "2": {"facts_total": 65, "facts_retained": 29, "retention_rate": 0.4461538461538462, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5, "status": 1.0, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1287}}}, "reference_key_id": "keys/synthetic/syn_03000_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "7d13c3599ad215a84dcdeaa88517f112a834b20d3470dddfe0f02da2fd7cec73", "output_hash": "ae55ba88b2cb1b0f1a4e0960ae312e0a1c177c79a6d14b67bff1677cec19b649"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "08edbb6f0b941c0fe895a781", "attempt_id": "1cda69e6d980", "run_group_id": "E4_core:A4|closed|qwen3b-llama3b:h2", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T06:32:06.183179Z", "task_family": "synthetic", "document_id": "syn_03000_2000", "document_hash": "c9cc27c20b05b1f6e592846e84990ee9b69c42f9720a8e338ec2f6fca777fdbc", "question_id": "q05_multi_hop", "split": "test", "cluster_id": "syn_03000_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "tokenizer_hashes": {"receiver": "b9d8872cf3a99da99e1fc4b0f71232a451632dc49ca4c6771549b4efa5c1c763"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 2, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1715, "query_tokens": 534, "message_tokens": 422, "message_utf8_bytes": 1287, "cache_tensor_bytes": null, "serialized_payload_bytes": 1287, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 39351.19591699913, "hops": {"1": 22739.89845905453, "2": 16611.2974579446}, "receiver_prefill_ms": 335.3806249797344, "ttft_ms": 37.33520791865885, "decode_ms": 244.3447910482064}, "total_wall_ms": 39968.88237504754, "ttft_ms": 37.33520791865885, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 7}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "LY-51E-3199", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 29, "retention_rate": 0.4461538461538462, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5, "status": 1.0, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1287}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 29, "retention_rate": 0.4461538461538462, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5, "status": 1.0, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1090}, "2": {"facts_total": 65, "facts_retained": 29, "retention_rate": 0.4461538461538462, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5, "status": 1.0, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1287}}}, "reference_key_id": "keys/synthetic/syn_03000_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "7d13c3599ad215a84dcdeaa88517f112a834b20d3470dddfe0f02da2fd7cec73", "output_hash": "a1245d1536890a7e4b026366df48441d503f491f782cfd388797b5782443c5c7"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "a5a38ca82577dfc26373515c", "attempt_id": "12aae6de1833", "run_group_id": "E4_core:A4|closed|qwen3b-llama3b:h2", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T06:32:06.817884Z", "task_family": "synthetic", "document_id": "syn_03000_2000", "document_hash": "c9cc27c20b05b1f6e592846e84990ee9b69c42f9720a8e338ec2f6fca777fdbc", "question_id": "q06_negation", "split": "test", "cluster_id": "syn_03000_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "tokenizer_hashes": {"receiver": "b9d8872cf3a99da99e1fc4b0f71232a451632dc49ca4c6771549b4efa5c1c763"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 2, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1715, "query_tokens": 525, "message_tokens": 422, "message_utf8_bytes": 1287, "cache_tensor_bytes": null, "serialized_payload_bytes": 1287, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 39351.19591699913, "hops": {"1": 22739.89845905453, "2": 16611.2974579446}, "receiver_prefill_ms": 394.6048329817131, "ttft_ms": 110.57062505278736, "decode_ms": 58.40670899488032}, "total_wall_ms": 39915.379957994446, "ttft_ms": 110.57062505278736, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 2}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "No.", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 29, "retention_rate": 0.4461538461538462, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5, "status": 1.0, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1287}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 29, "retention_rate": 0.4461538461538462, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5, "status": 1.0, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1090}, "2": {"facts_total": 65, "facts_retained": 29, "retention_rate": 0.4461538461538462, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5, "status": 1.0, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1287}}}, "reference_key_id": "keys/synthetic/syn_03000_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "7d13c3599ad215a84dcdeaa88517f112a834b20d3470dddfe0f02da2fd7cec73", "output_hash": "38178a20b470cfd18299fb1593dd4ba706f1a83321dac051d01086efd8b7a96f"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "7846ce2cb6c43fa96e3dba01", "attempt_id": "568342adffdc", "run_group_id": "E4_core:A4|closed|qwen3b-llama3b:h2", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T06:32:07.399512Z", "task_family": "synthetic", "document_id": "syn_03000_2000", "document_hash": "c9cc27c20b05b1f6e592846e84990ee9b69c42f9720a8e338ec2f6fca777fdbc", "question_id": "q07_negation", "split": "test", "cluster_id": "syn_03000_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "tokenizer_hashes": {"receiver": "b9d8872cf3a99da99e1fc4b0f71232a451632dc49ca4c6771549b4efa5c1c763"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 2, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1715, "query_tokens": 525, "message_tokens": 422, "message_utf8_bytes": 1287, "cache_tensor_bytes": null, "serialized_payload_bytes": 1287, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 39351.19591699913, "hops": {"1": 22739.89845905453, "2": 16611.2974579446}, "receiver_prefill_ms": 335.934582981281, "ttft_ms": 38.056832971051335, "decode_ms": 59.960167040117085}, "total_wall_ms": 39785.67220806144, "ttft_ms": 38.056832971051335, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 2}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "No.", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 29, "retention_rate": 0.4461538461538462, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5, "status": 1.0, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1287}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 29, "retention_rate": 0.4461538461538462, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5, "status": 1.0, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1090}, "2": {"facts_total": 65, "facts_retained": 29, "retention_rate": 0.4461538461538462, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5, "status": 1.0, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1287}}}, "reference_key_id": "keys/synthetic/syn_03000_2000.json", "task_scores": {"correct": 0.0, "abstained": 0.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "7d13c3599ad215a84dcdeaa88517f112a834b20d3470dddfe0f02da2fd7cec73", "output_hash": "38178a20b470cfd18299fb1593dd4ba706f1a83321dac051d01086efd8b7a96f"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "260f8c48d7cfe8b35e8f60a3", "attempt_id": "e8c8cace21eb", "run_group_id": "E4_core:A4|closed|qwen3b-llama3b:h2", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T06:32:07.850629Z", "task_family": "synthetic", "document_id": "syn_03000_2000", "document_hash": "c9cc27c20b05b1f6e592846e84990ee9b69c42f9720a8e338ec2f6fca777fdbc", "question_id": "q08_revision", "split": "test", "cluster_id": "syn_03000_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "tokenizer_hashes": {"receiver": "b9d8872cf3a99da99e1fc4b0f71232a451632dc49ca4c6771549b4efa5c1c763"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 2, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1715, "query_tokens": 540, "message_tokens": 422, "message_utf8_bytes": 1287, "cache_tensor_bytes": null, "serialized_payload_bytes": 1287, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 39351.19591699913, "hops": {"1": 22739.89845905453, "2": 16611.2974579446}, "receiver_prefill_ms": 349.10579095594585, "ttft_ms": 24.260042002424598, "decode_ms": 143.76570796594024}, "total_wall_ms": 39868.95662499592, "ttft_ms": 24.260042002424598, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 5}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "2,705,965", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 29, "retention_rate": 0.4461538461538462, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5, "status": 1.0, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1287}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 29, "retention_rate": 0.4461538461538462, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5, "status": 1.0, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1090}, "2": {"facts_total": 65, "facts_retained": 29, "retention_rate": 0.4461538461538462, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5, "status": 1.0, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1287}}}, "reference_key_id": "keys/synthetic/syn_03000_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "7d13c3599ad215a84dcdeaa88517f112a834b20d3470dddfe0f02da2fd7cec73", "output_hash": "dbedb6f5d00e413d915513b85d8ed195b18b9acf392ece342a85e15e3752bcc8"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "8d3bda0a96a97760fbe951a7", "attempt_id": "382d095ad79e", "run_group_id": "E4_core:A4|closed|qwen3b-llama3b:h2", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T06:32:08.385564Z", "task_family": "synthetic", "document_id": "syn_03000_2000", "document_hash": "c9cc27c20b05b1f6e592846e84990ee9b69c42f9720a8e338ec2f6fca777fdbc", "question_id": "q09_revision", "split": "test", "cluster_id": "syn_03000_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "tokenizer_hashes": {"receiver": "b9d8872cf3a99da99e1fc4b0f71232a451632dc49ca4c6771549b4efa5c1c763"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 2, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1715, "query_tokens": 535, "message_tokens": 422, "message_utf8_bytes": 1287, "cache_tensor_bytes": null, "serialized_payload_bytes": 1287, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 39351.19591699913, "hops": {"1": 22739.89845905453, "2": 16611.2974579446}, "receiver_prefill_ms": 345.91754199936986, "ttft_ms": 54.70604193396866, "decode_ms": 331.8763340357691}, "total_wall_ms": 40084.40408401657, "ttft_ms": 54.70604193396866, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 5}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "9,127,965", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 29, "retention_rate": 0.4461538461538462, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5, "status": 1.0, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1287}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 29, "retention_rate": 0.4461538461538462, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5, "status": 1.0, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1090}, "2": {"facts_total": 65, "facts_retained": 29, "retention_rate": 0.4461538461538462, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5, "status": 1.0, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1287}}}, "reference_key_id": "keys/synthetic/syn_03000_2000.json", "task_scores": {"correct": 0.0, "abstained": 0.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "7d13c3599ad215a84dcdeaa88517f112a834b20d3470dddfe0f02da2fd7cec73", "output_hash": "f30700451becb8c76f6e83d5016caa83d25106e6dceeffd8fdeca27984218444"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "a0048d8fb2a9bc5ef53dcf8b", "attempt_id": "10335a71ae13", "run_group_id": "E4_core:A4|closed|qwen3b-llama3b:h2", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T06:32:09.136076Z", "task_family": "synthetic", "document_id": "syn_03000_2000", "document_hash": "c9cc27c20b05b1f6e592846e84990ee9b69c42f9720a8e338ec2f6fca777fdbc", "question_id": "q10_temporal", "split": "test", "cluster_id": "syn_03000_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "tokenizer_hashes": {"receiver": "b9d8872cf3a99da99e1fc4b0f71232a451632dc49ca4c6771549b4efa5c1c763"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 2, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1715, "query_tokens": 542, "message_tokens": 422, "message_utf8_bytes": 1287, "cache_tensor_bytes": null, "serialized_payload_bytes": 1287, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 39351.19591699913, "hops": {"1": 22739.89845905453, "2": 16611.2974579446}, "receiver_prefill_ms": 351.06124996673316, "ttft_ms": 32.24004106596112, "decode_ms": 31.877957982942462}, "total_wall_ms": 39766.96425001137, "ttft_ms": 32.24004106596112, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 29, "retention_rate": 0.4461538461538462, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5, "status": 1.0, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1287}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 29, "retention_rate": 0.4461538461538462, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5, "status": 1.0, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1090}, "2": {"facts_total": 65, "facts_retained": 29, "retention_rate": 0.4461538461538462, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5, "status": 1.0, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1287}}}, "reference_key_id": "keys/synthetic/syn_03000_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "7d13c3599ad215a84dcdeaa88517f112a834b20d3470dddfe0f02da2fd7cec73", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "356d2ee706b9315ae31cb8fc", "attempt_id": "686bb2f581de", "run_group_id": "E4_core:A4|closed|qwen3b-llama3b:h2", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T06:32:09.569157Z", "task_family": "synthetic", "document_id": "syn_03000_2000", "document_hash": "c9cc27c20b05b1f6e592846e84990ee9b69c42f9720a8e338ec2f6fca777fdbc", "question_id": "q11_aggregation", "split": "test", "cluster_id": "syn_03000_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "tokenizer_hashes": {"receiver": "b9d8872cf3a99da99e1fc4b0f71232a451632dc49ca4c6771549b4efa5c1c763"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 2, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1715, "query_tokens": 537, "message_tokens": 422, "message_utf8_bytes": 1287, "cache_tensor_bytes": null, "serialized_payload_bytes": 1287, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 39351.19591699913, "hops": {"1": 22739.89845905453, "2": 16611.2974579446}, "receiver_prefill_ms": 336.30487497430295, "ttft_ms": 25.61920799780637, "decode_ms": 165.08299997076392}, "total_wall_ms": 39878.82958399132, "ttft_ms": 25.61920799780637, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 5}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "9,220,745", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 29, "retention_rate": 0.4461538461538462, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5, "status": 1.0, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1287}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 29, "retention_rate": 0.4461538461538462, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5, "status": 1.0, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1090}, "2": {"facts_total": 65, "facts_retained": 29, "retention_rate": 0.4461538461538462, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5, "status": 1.0, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1287}}}, "reference_key_id": "keys/synthetic/syn_03000_2000.json", "task_scores": {"correct": 0.0, "abstained": 0.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "7d13c3599ad215a84dcdeaa88517f112a834b20d3470dddfe0f02da2fd7cec73", "output_hash": "ae55ba88b2cb1b0f1a4e0960ae312e0a1c177c79a6d14b67bff1677cec19b649"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "d53e14d6987a45d5452a9f53", "attempt_id": "56d4d6adbf55", "run_group_id": "E4_core:A4|closed|qwen3b-llama3b:h2", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T06:32:10.113406Z", "task_family": "synthetic", "document_id": "syn_03000_2000", "document_hash": "c9cc27c20b05b1f6e592846e84990ee9b69c42f9720a8e338ec2f6fca777fdbc", "question_id": "q12_unit_conversion", "split": "test", "cluster_id": "syn_03000_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "tokenizer_hashes": {"receiver": "b9d8872cf3a99da99e1fc4b0f71232a451632dc49ca4c6771549b4efa5c1c763"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 2, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1715, "query_tokens": 533, "message_tokens": 422, "message_utf8_bytes": 1287, "cache_tensor_bytes": null, "serialized_payload_bytes": 1287, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 39351.19591699913, "hops": {"1": 22739.89845905453, "2": 16611.2974579446}, "receiver_prefill_ms": 338.44166703056544, "ttft_ms": 36.39691707212478, "decode_ms": 83.53212499059737}, "total_wall_ms": 39810.300791985355, "ttft_ms": 36.39691707212478, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 29, "retention_rate": 0.4461538461538462, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5, "status": 1.0, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1287}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 29, "retention_rate": 0.4461538461538462, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5, "status": 1.0, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1090}, "2": {"facts_total": 65, "facts_retained": 29, "retention_rate": 0.4461538461538462, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5, "status": 1.0, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1287}}}, "reference_key_id": "keys/synthetic/syn_03000_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "7d13c3599ad215a84dcdeaa88517f112a834b20d3470dddfe0f02da2fd7cec73", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "1dcdf250c93acda3f3902617", "attempt_id": "287b116d5a8b", "run_group_id": "E4_core:A4|closed|qwen3b-llama3b:h2", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T06:32:10.590532Z", "task_family": "synthetic", "document_id": "syn_03000_2000", "document_hash": "c9cc27c20b05b1f6e592846e84990ee9b69c42f9720a8e338ec2f6fca777fdbc", "question_id": "q13_unanswerable", "split": "test", "cluster_id": "syn_03000_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "tokenizer_hashes": {"receiver": "b9d8872cf3a99da99e1fc4b0f71232a451632dc49ca4c6771549b4efa5c1c763"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 2, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1715, "query_tokens": 536, "message_tokens": 422, "message_utf8_bytes": 1287, "cache_tensor_bytes": null, "serialized_payload_bytes": 1287, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 39351.19591699913, "hops": {"1": 22739.89845905453, "2": 16611.2974579446}, "receiver_prefill_ms": 416.4477090816945, "ttft_ms": 37.958166911266744, "decode_ms": 25.16266703605652}, "total_wall_ms": 39831.42158307601, "ttft_ms": 37.958166911266744, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 29, "retention_rate": 0.4461538461538462, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5, "status": 1.0, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1287}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 29, "retention_rate": 0.4461538461538462, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5, "status": 1.0, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1090}, "2": {"facts_total": 65, "facts_retained": 29, "retention_rate": 0.4461538461538462, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5, "status": 1.0, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1287}}}, "reference_key_id": "keys/synthetic/syn_03000_2000.json", "task_scores": {"correct": 1.0, "abstained": 1.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "7d13c3599ad215a84dcdeaa88517f112a834b20d3470dddfe0f02da2fd7cec73", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "34bbf2591ca4fd1c882a8f05", "attempt_id": "d480da589730", "run_group_id": "E4_core:A4|closed|qwen3b-llama3b:h2", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T06:32:11.090320Z", "task_family": "synthetic", "document_id": "syn_03000_2000", "document_hash": "c9cc27c20b05b1f6e592846e84990ee9b69c42f9720a8e338ec2f6fca777fdbc", "question_id": "q14_unanswerable", "split": "test", "cluster_id": "syn_03000_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "tokenizer_hashes": {"receiver": "b9d8872cf3a99da99e1fc4b0f71232a451632dc49ca4c6771549b4efa5c1c763"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 2, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1715, "query_tokens": 539, "message_tokens": 422, "message_utf8_bytes": 1287, "cache_tensor_bytes": null, "serialized_payload_bytes": 1287, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 39351.19591699913, "hops": {"1": 22739.89845905453, "2": 16611.2974579446}, "receiver_prefill_ms": 337.6418750267476, "ttft_ms": 39.02466595172882, "decode_ms": 24.101209011860192}, "total_wall_ms": 39752.4943749886, "ttft_ms": 39.02466595172882, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 29, "retention_rate": 0.4461538461538462, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5, "status": 1.0, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1287}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 29, "retention_rate": 0.4461538461538462, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5, "status": 1.0, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1090}, "2": {"facts_total": 65, "facts_retained": 29, "retention_rate": 0.4461538461538462, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5, "status": 1.0, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1287}}}, "reference_key_id": "keys/synthetic/syn_03000_2000.json", "task_scores": {"correct": 1.0, "abstained": 1.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "7d13c3599ad215a84dcdeaa88517f112a834b20d3470dddfe0f02da2fd7cec73", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "6564339c8625d4cbe7a6484c", "attempt_id": "f524043777ff", "run_group_id": "E4_core:A4|closed|qwen3b-llama3b:h4", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T06:32:50.407209Z", "task_family": "synthetic", "document_id": "syn_03000_2000", "document_hash": "c9cc27c20b05b1f6e592846e84990ee9b69c42f9720a8e338ec2f6fca777fdbc", "question_id": "q00_single_fact", "split": "test", "cluster_id": "syn_03000_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "tokenizer_hashes": {"receiver": "b9d8872cf3a99da99e1fc4b0f71232a451632dc49ca4c6771549b4efa5c1c763"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 4, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1715, "query_tokens": 529, "message_tokens": 426, "message_utf8_bytes": 1247, "cache_tensor_bytes": null, "serialized_payload_bytes": 1247, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 78247.3377090646, "hops": {"1": 22739.89845905453, "2": 16611.2974579446, "3": 21987.99737507943, "4": 16908.144416986033}, "receiver_prefill_ms": 496.1395839927718, "ttft_ms": 24.370333994738758, "decode_ms": 40.202542091719806}, "total_wall_ms": 78808.75612597447, "ttft_ms": 24.370333994738758, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 29, "retention_rate": 0.4461538461538462, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5, "status": 1.0, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1247}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 29, "retention_rate": 0.4461538461538462, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5, "status": 1.0, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1090}, "2": {"facts_total": 65, "facts_retained": 29, "retention_rate": 0.4461538461538462, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5, "status": 1.0, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1287}, "3": {"facts_total": 65, "facts_retained": 29, "retention_rate": 0.4461538461538462, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5, "status": 1.0, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1232}, "4": {"facts_total": 65, "facts_retained": 29, "retention_rate": 0.4461538461538462, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5, "status": 1.0, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1247}}}, "reference_key_id": "keys/synthetic/syn_03000_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "afb8dd9e0494587cde2c5fb6281f534ae0073a68d145eea31da6ae6240dceb66", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "176fbbbf8181875f41ef59ac", "attempt_id": "547dbfb91356", "run_group_id": "E4_core:A4|closed|qwen3b-llama3b:h4", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T06:32:50.986371Z", "task_family": "synthetic", "document_id": "syn_03000_2000", "document_hash": "c9cc27c20b05b1f6e592846e84990ee9b69c42f9720a8e338ec2f6fca777fdbc", "question_id": "q01_single_fact", "split": "test", "cluster_id": "syn_03000_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "tokenizer_hashes": {"receiver": "b9d8872cf3a99da99e1fc4b0f71232a451632dc49ca4c6771549b4efa5c1c763"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 4, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1715, "query_tokens": 534, "message_tokens": 426, "message_utf8_bytes": 1247, "cache_tensor_bytes": null, "serialized_payload_bytes": 1247, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 78247.3377090646, "hops": {"1": 22739.89845905453, "2": 16611.2974579446, "3": 21987.99737507943, "4": 16908.144416986033}, "receiver_prefill_ms": 342.4125420860946, "ttft_ms": 27.245708974078298, "decode_ms": 266.28779096063226}, "total_wall_ms": 78883.87166708708, "ttft_ms": 27.245708974078298, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 9}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "Lot 19-N, Gaspé.", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 29, "retention_rate": 0.4461538461538462, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5, "status": 1.0, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1247}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 29, "retention_rate": 0.4461538461538462, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5, "status": 1.0, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1090}, "2": {"facts_total": 65, "facts_retained": 29, "retention_rate": 0.4461538461538462, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5, "status": 1.0, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1287}, "3": {"facts_total": 65, "facts_retained": 29, "retention_rate": 0.4461538461538462, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5, "status": 1.0, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1232}, "4": {"facts_total": 65, "facts_retained": 29, "retention_rate": 0.4461538461538462, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5, "status": 1.0, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1247}}}, "reference_key_id": "keys/synthetic/syn_03000_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 0.8571428571428571}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "afb8dd9e0494587cde2c5fb6281f534ae0073a68d145eea31da6ae6240dceb66", "output_hash": "ab9cc24b811f5b1342aae731f584e8b0d24b63b61ba59a068b8414deffdc10ae"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "fb7a38a99916dc0e1942f50b", "attempt_id": "3362efc0fd9d", "run_group_id": "E4_core:A4|closed|qwen3b-llama3b:h4", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T06:32:51.640744Z", "task_family": "synthetic", "document_id": "syn_03000_2000", "document_hash": "c9cc27c20b05b1f6e592846e84990ee9b69c42f9720a8e338ec2f6fca777fdbc", "question_id": "q02_exact_copy", "split": "test", "cluster_id": "syn_03000_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "tokenizer_hashes": {"receiver": "b9d8872cf3a99da99e1fc4b0f71232a451632dc49ca4c6771549b4efa5c1c763"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 4, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1715, "query_tokens": 533, "message_tokens": 426, "message_utf8_bytes": 1247, "cache_tensor_bytes": null, "serialized_payload_bytes": 1247, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 78247.3377090646, "hops": {"1": 22739.89845905453, "2": 16611.2974579446, "3": 21987.99737507943, "4": 16908.144416986033}, "receiver_prefill_ms": 292.87433391436934, "ttft_ms": 55.57658290490508, "decode_ms": 59.55066601745784}, "total_wall_ms": 78656.09379298985, "ttft_ms": 55.57658290490508, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 29, "retention_rate": 0.4461538461538462, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5, "status": 1.0, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1247}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 29, "retention_rate": 0.4461538461538462, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5, "status": 1.0, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1090}, "2": {"facts_total": 65, "facts_retained": 29, "retention_rate": 0.4461538461538462, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5, "status": 1.0, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1287}, "3": {"facts_total": 65, "facts_retained": 29, "retention_rate": 0.4461538461538462, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5, "status": 1.0, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1232}, "4": {"facts_total": 65, "facts_retained": 29, "retention_rate": 0.4461538461538462, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5, "status": 1.0, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1247}}}, "reference_key_id": "keys/synthetic/syn_03000_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "afb8dd9e0494587cde2c5fb6281f534ae0073a68d145eea31da6ae6240dceb66", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "5ed4364f5578248d95b6f272", "attempt_id": "538d89f9d7e7", "run_group_id": "E4_core:A4|closed|qwen3b-llama3b:h4", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T06:32:52.066234Z", "task_family": "synthetic", "document_id": "syn_03000_2000", "document_hash": "c9cc27c20b05b1f6e592846e84990ee9b69c42f9720a8e338ec2f6fca777fdbc", "question_id": "q03_exact_copy", "split": "test", "cluster_id": "syn_03000_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "tokenizer_hashes": {"receiver": "b9d8872cf3a99da99e1fc4b0f71232a451632dc49ca4c6771549b4efa5c1c763"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 4, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1715, "query_tokens": 531, "message_tokens": 426, "message_utf8_bytes": 1247, "cache_tensor_bytes": null, "serialized_payload_bytes": 1247, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 78247.3377090646, "hops": {"1": 22739.89845905453, "2": 16611.2974579446, "3": 21987.99737507943, "4": 16908.144416986033}, "receiver_prefill_ms": 454.1254580253735, "ttft_ms": 37.53891703672707, "decode_ms": 23.034083074890077}, "total_wall_ms": 78762.62283406686, "ttft_ms": 37.53891703672707, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 29, "retention_rate": 0.4461538461538462, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5, "status": 1.0, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1247}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 29, "retention_rate": 0.4461538461538462, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5, "status": 1.0, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1090}, "2": {"facts_total": 65, "facts_retained": 29, "retention_rate": 0.4461538461538462, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5, "status": 1.0, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1287}, "3": {"facts_total": 65, "facts_retained": 29, "retention_rate": 0.4461538461538462, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5, "status": 1.0, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1232}, "4": {"facts_total": 65, "facts_retained": 29, "retention_rate": 0.4461538461538462, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5, "status": 1.0, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1247}}}, "reference_key_id": "keys/synthetic/syn_03000_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "afb8dd9e0494587cde2c5fb6281f534ae0073a68d145eea31da6ae6240dceb66", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "283c39c0bdd14a762d82258f", "attempt_id": "72773a53285c", "run_group_id": "E4_core:A4|closed|qwen3b-llama3b:h4", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T06:32:52.598073Z", "task_family": "synthetic", "document_id": "syn_03000_2000", "document_hash": "c9cc27c20b05b1f6e592846e84990ee9b69c42f9720a8e338ec2f6fca777fdbc", "question_id": "q04_multi_hop", "split": "test", "cluster_id": "syn_03000_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "tokenizer_hashes": {"receiver": "b9d8872cf3a99da99e1fc4b0f71232a451632dc49ca4c6771549b4efa5c1c763"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 4, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1715, "query_tokens": 538, "message_tokens": 426, "message_utf8_bytes": 1247, "cache_tensor_bytes": null, "serialized_payload_bytes": 1247, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 78247.3377090646, "hops": {"1": 22739.89845905453, "2": 16611.2974579446, "3": 21987.99737507943, "4": 16908.144416986033}, "receiver_prefill_ms": 334.2221659841016, "ttft_ms": 37.75854199193418, "decode_ms": 152.30824996251613}, "total_wall_ms": 78772.34229200985, "ttft_ms": 37.75854199193418, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 5}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "9,220,745", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 29, "retention_rate": 0.4461538461538462, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5, "status": 1.0, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1247}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 29, "retention_rate": 0.4461538461538462, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5, "status": 1.0, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1090}, "2": {"facts_total": 65, "facts_retained": 29, "retention_rate": 0.4461538461538462, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5, "status": 1.0, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1287}, "3": {"facts_total": 65, "facts_retained": 29, "retention_rate": 0.4461538461538462, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5, "status": 1.0, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1232}, "4": {"facts_total": 65, "facts_retained": 29, "retention_rate": 0.4461538461538462, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5, "status": 1.0, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1247}}}, "reference_key_id": "keys/synthetic/syn_03000_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "afb8dd9e0494587cde2c5fb6281f534ae0073a68d145eea31da6ae6240dceb66", "output_hash": "ae55ba88b2cb1b0f1a4e0960ae312e0a1c177c79a6d14b67bff1677cec19b649"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "a3537227225927b8100f4305", "attempt_id": "974fb2d02354", "run_group_id": "E4_core:A4|closed|qwen3b-llama3b:h4", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T06:32:53.139325Z", "task_family": "synthetic", "document_id": "syn_03000_2000", "document_hash": "c9cc27c20b05b1f6e592846e84990ee9b69c42f9720a8e338ec2f6fca777fdbc", "question_id": "q05_multi_hop", "split": "test", "cluster_id": "syn_03000_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "tokenizer_hashes": {"receiver": "b9d8872cf3a99da99e1fc4b0f71232a451632dc49ca4c6771549b4efa5c1c763"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 4, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1715, "query_tokens": 538, "message_tokens": 426, "message_utf8_bytes": 1247, "cache_tensor_bytes": null, "serialized_payload_bytes": 1247, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 78247.3377090646, "hops": {"1": 22739.89845905453, "2": 16611.2974579446, "3": 21987.99737507943, "4": 16908.144416986033}, "receiver_prefill_ms": 340.4420419828966, "ttft_ms": 20.51554201170802, "decode_ms": 282.8936669975519}, "total_wall_ms": 78892.19179202337, "ttft_ms": 20.51554201170802, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 7}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "LY-51E-3199", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 29, "retention_rate": 0.4461538461538462, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5, "status": 1.0, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1247}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 29, "retention_rate": 0.4461538461538462, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5, "status": 1.0, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1090}, "2": {"facts_total": 65, "facts_retained": 29, "retention_rate": 0.4461538461538462, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5, "status": 1.0, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1287}, "3": {"facts_total": 65, "facts_retained": 29, "retention_rate": 0.4461538461538462, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5, "status": 1.0, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1232}, "4": {"facts_total": 65, "facts_retained": 29, "retention_rate": 0.4461538461538462, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5, "status": 1.0, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1247}}}, "reference_key_id": "keys/synthetic/syn_03000_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "afb8dd9e0494587cde2c5fb6281f534ae0073a68d145eea31da6ae6240dceb66", "output_hash": "a1245d1536890a7e4b026366df48441d503f491f782cfd388797b5782443c5c7"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "e012d8ce2b357cc8c9bea2a2", "attempt_id": "797168f293f5", "run_group_id": "E4_core:A4|closed|qwen3b-llama3b:h4", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T06:32:53.801484Z", "task_family": "synthetic", "document_id": "syn_03000_2000", "document_hash": "c9cc27c20b05b1f6e592846e84990ee9b69c42f9720a8e338ec2f6fca777fdbc", "question_id": "q06_negation", "split": "test", "cluster_id": "syn_03000_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "tokenizer_hashes": {"receiver": "b9d8872cf3a99da99e1fc4b0f71232a451632dc49ca4c6771549b4efa5c1c763"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 4, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1715, "query_tokens": 529, "message_tokens": 426, "message_utf8_bytes": 1247, "cache_tensor_bytes": null, "serialized_payload_bytes": 1247, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 78247.3377090646, "hops": {"1": 22739.89845905453, "2": 16611.2974579446, "3": 21987.99737507943, "4": 16908.144416986033}, "receiver_prefill_ms": 462.8084589494392, "ttft_ms": 35.93004203867167, "decode_ms": 71.04354200419039}, "total_wall_ms": 78817.76929204352, "ttft_ms": 35.93004203867167, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 2}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "No.", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 29, "retention_rate": 0.4461538461538462, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5, "status": 1.0, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1247}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 29, "retention_rate": 0.4461538461538462, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5, "status": 1.0, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1090}, "2": {"facts_total": 65, "facts_retained": 29, "retention_rate": 0.4461538461538462, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5, "status": 1.0, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1287}, "3": {"facts_total": 65, "facts_retained": 29, "retention_rate": 0.4461538461538462, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5, "status": 1.0, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1232}, "4": {"facts_total": 65, "facts_retained": 29, "retention_rate": 0.4461538461538462, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5, "status": 1.0, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1247}}}, "reference_key_id": "keys/synthetic/syn_03000_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "afb8dd9e0494587cde2c5fb6281f534ae0073a68d145eea31da6ae6240dceb66", "output_hash": "38178a20b470cfd18299fb1593dd4ba706f1a83321dac051d01086efd8b7a96f"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "f772fc8ddc6aa5f929c894e8", "attempt_id": "5530ea11d092", "run_group_id": "E4_core:A4|closed|qwen3b-llama3b:h4", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T06:32:54.389509Z", "task_family": "synthetic", "document_id": "syn_03000_2000", "document_hash": "c9cc27c20b05b1f6e592846e84990ee9b69c42f9720a8e338ec2f6fca777fdbc", "question_id": "q07_negation", "split": "test", "cluster_id": "syn_03000_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "tokenizer_hashes": {"receiver": "b9d8872cf3a99da99e1fc4b0f71232a451632dc49ca4c6771549b4efa5c1c763"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 4, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1715, "query_tokens": 529, "message_tokens": 426, "message_utf8_bytes": 1247, "cache_tensor_bytes": null, "serialized_payload_bytes": 1247, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 78247.3377090646, "hops": {"1": 22739.89845905453, "2": 16611.2974579446, "3": 21987.99737507943, "4": 16908.144416986033}, "receiver_prefill_ms": 350.52216693293303, "ttft_ms": 41.47104197181761, "decode_ms": 63.36658401414752}, "total_wall_ms": 78703.27216701116, "ttft_ms": 41.47104197181761, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 2}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "No.", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 29, "retention_rate": 0.4461538461538462, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5, "status": 1.0, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1247}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 29, "retention_rate": 0.4461538461538462, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5, "status": 1.0, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1090}, "2": {"facts_total": 65, "facts_retained": 29, "retention_rate": 0.4461538461538462, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5, "status": 1.0, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1287}, "3": {"facts_total": 65, "facts_retained": 29, "retention_rate": 0.4461538461538462, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5, "status": 1.0, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1232}, "4": {"facts_total": 65, "facts_retained": 29, "retention_rate": 0.4461538461538462, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5, "status": 1.0, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1247}}}, "reference_key_id": "keys/synthetic/syn_03000_2000.json", "task_scores": {"correct": 0.0, "abstained": 0.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "afb8dd9e0494587cde2c5fb6281f534ae0073a68d145eea31da6ae6240dceb66", "output_hash": "38178a20b470cfd18299fb1593dd4ba706f1a83321dac051d01086efd8b7a96f"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "c7288cc810f80048391293d7", "attempt_id": "f0fc698bbcad", "run_group_id": "E4_core:A4|closed|qwen3b-llama3b:h4", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T06:32:54.863118Z", "task_family": "synthetic", "document_id": "syn_03000_2000", "document_hash": "c9cc27c20b05b1f6e592846e84990ee9b69c42f9720a8e338ec2f6fca777fdbc", "question_id": "q08_revision", "split": "test", "cluster_id": "syn_03000_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "tokenizer_hashes": {"receiver": "b9d8872cf3a99da99e1fc4b0f71232a451632dc49ca4c6771549b4efa5c1c763"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 4, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1715, "query_tokens": 544, "message_tokens": 426, "message_utf8_bytes": 1247, "cache_tensor_bytes": null, "serialized_payload_bytes": 1247, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 78247.3377090646, "hops": {"1": 22739.89845905453, "2": 16611.2974579446, "3": 21987.99737507943, "4": 16908.144416986033}, "receiver_prefill_ms": 328.0416250927374, "ttft_ms": 23.66987499408424, "decode_ms": 250.89391705114394}, "total_wall_ms": 78850.68641707767, "ttft_ms": 23.66987499408424, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 5}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "2,705,965", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 29, "retention_rate": 0.4461538461538462, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5, "status": 1.0, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1247}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 29, "retention_rate": 0.4461538461538462, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5, "status": 1.0, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1090}, "2": {"facts_total": 65, "facts_retained": 29, "retention_rate": 0.4461538461538462, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5, "status": 1.0, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1287}, "3": {"facts_total": 65, "facts_retained": 29, "retention_rate": 0.4461538461538462, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5, "status": 1.0, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1232}, "4": {"facts_total": 65, "facts_retained": 29, "retention_rate": 0.4461538461538462, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5, "status": 1.0, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1247}}}, "reference_key_id": "keys/synthetic/syn_03000_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "afb8dd9e0494587cde2c5fb6281f534ae0073a68d145eea31da6ae6240dceb66", "output_hash": "dbedb6f5d00e413d915513b85d8ed195b18b9acf392ece342a85e15e3752bcc8"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "ad4654eb6d2374a6882b3114", "attempt_id": "311a92d32160", "run_group_id": "E4_core:A4|closed|qwen3b-llama3b:h4", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T06:32:55.484072Z", "task_family": "synthetic", "document_id": "syn_03000_2000", "document_hash": "c9cc27c20b05b1f6e592846e84990ee9b69c42f9720a8e338ec2f6fca777fdbc", "question_id": "q09_revision", "split": "test", "cluster_id": "syn_03000_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "tokenizer_hashes": {"receiver": "b9d8872cf3a99da99e1fc4b0f71232a451632dc49ca4c6771549b4efa5c1c763"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 4, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1715, "query_tokens": 539, "message_tokens": 426, "message_utf8_bytes": 1247, "cache_tensor_bytes": null, "serialized_payload_bytes": 1247, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 78247.3377090646, "hops": {"1": 22739.89845905453, "2": 16611.2974579446, "3": 21987.99737507943, "4": 16908.144416986033}, "receiver_prefill_ms": 465.3224169742316, "ttft_ms": 25.617499952204525, "decode_ms": 148.50729098543525}, "total_wall_ms": 78887.47570908163, "ttft_ms": 25.617499952204525, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 5}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "9,127,965", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 29, "retention_rate": 0.4461538461538462, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5, "status": 1.0, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1247}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 29, "retention_rate": 0.4461538461538462, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5, "status": 1.0, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1090}, "2": {"facts_total": 65, "facts_retained": 29, "retention_rate": 0.4461538461538462, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5, "status": 1.0, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1287}, "3": {"facts_total": 65, "facts_retained": 29, "retention_rate": 0.4461538461538462, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5, "status": 1.0, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1232}, "4": {"facts_total": 65, "facts_retained": 29, "retention_rate": 0.4461538461538462, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5, "status": 1.0, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1247}}}, "reference_key_id": "keys/synthetic/syn_03000_2000.json", "task_scores": {"correct": 0.0, "abstained": 0.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "afb8dd9e0494587cde2c5fb6281f534ae0073a68d145eea31da6ae6240dceb66", "output_hash": "f30700451becb8c76f6e83d5016caa83d25106e6dceeffd8fdeca27984218444"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "a05c56165e477ce63969315f", "attempt_id": "53d599212f81", "run_group_id": "E4_core:A4|closed|qwen3b-llama3b:h4", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T06:32:56.141721Z", "task_family": "synthetic", "document_id": "syn_03000_2000", "document_hash": "c9cc27c20b05b1f6e592846e84990ee9b69c42f9720a8e338ec2f6fca777fdbc", "question_id": "q10_temporal", "split": "test", "cluster_id": "syn_03000_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "tokenizer_hashes": {"receiver": "b9d8872cf3a99da99e1fc4b0f71232a451632dc49ca4c6771549b4efa5c1c763"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 4, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1715, "query_tokens": 546, "message_tokens": 426, "message_utf8_bytes": 1247, "cache_tensor_bytes": null, "serialized_payload_bytes": 1247, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 78247.3377090646, "hops": {"1": 22739.89845905453, "2": 16611.2974579446, "3": 21987.99737507943, "4": 16908.144416986033}, "receiver_prefill_ms": 350.44833295978606, "ttft_ms": 44.43787503987551, "decode_ms": 24.27299995906651}, "total_wall_ms": 78667.0044590719, "ttft_ms": 44.43787503987551, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 29, "retention_rate": 0.4461538461538462, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5, "status": 1.0, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1247}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 29, "retention_rate": 0.4461538461538462, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5, "status": 1.0, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1090}, "2": {"facts_total": 65, "facts_retained": 29, "retention_rate": 0.4461538461538462, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5, "status": 1.0, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1287}, "3": {"facts_total": 65, "facts_retained": 29, "retention_rate": 0.4461538461538462, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5, "status": 1.0, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1232}, "4": {"facts_total": 65, "facts_retained": 29, "retention_rate": 0.4461538461538462, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5, "status": 1.0, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1247}}}, "reference_key_id": "keys/synthetic/syn_03000_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "afb8dd9e0494587cde2c5fb6281f534ae0073a68d145eea31da6ae6240dceb66", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "5207659b7d36dcfd388b7339", "attempt_id": "136f7f91dd6e", "run_group_id": "E4_core:A4|closed|qwen3b-llama3b:h4", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T06:32:56.578617Z", "task_family": "synthetic", "document_id": "syn_03000_2000", "document_hash": "c9cc27c20b05b1f6e592846e84990ee9b69c42f9720a8e338ec2f6fca777fdbc", "question_id": "q11_aggregation", "split": "test", "cluster_id": "syn_03000_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "tokenizer_hashes": {"receiver": "b9d8872cf3a99da99e1fc4b0f71232a451632dc49ca4c6771549b4efa5c1c763"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 4, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1715, "query_tokens": 541, "message_tokens": 426, "message_utf8_bytes": 1247, "cache_tensor_bytes": null, "serialized_payload_bytes": 1247, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 78247.3377090646, "hops": {"1": 22739.89845905453, "2": 16611.2974579446, "3": 21987.99737507943, "4": 16908.144416986033}, "receiver_prefill_ms": 315.190166933462, "ttft_ms": 11.54437498189509, "decode_ms": 215.89395799674094}, "total_wall_ms": 78790.65583401825, "ttft_ms": 11.54437498189509, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 5}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "9,220,745", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 29, "retention_rate": 0.4461538461538462, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5, "status": 1.0, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1247}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 29, "retention_rate": 0.4461538461538462, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5, "status": 1.0, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1090}, "2": {"facts_total": 65, "facts_retained": 29, "retention_rate": 0.4461538461538462, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5, "status": 1.0, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1287}, "3": {"facts_total": 65, "facts_retained": 29, "retention_rate": 0.4461538461538462, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5, "status": 1.0, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1232}, "4": {"facts_total": 65, "facts_retained": 29, "retention_rate": 0.4461538461538462, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5, "status": 1.0, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1247}}}, "reference_key_id": "keys/synthetic/syn_03000_2000.json", "task_scores": {"correct": 0.0, "abstained": 0.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "afb8dd9e0494587cde2c5fb6281f534ae0073a68d145eea31da6ae6240dceb66", "output_hash": "ae55ba88b2cb1b0f1a4e0960ae312e0a1c177c79a6d14b67bff1677cec19b649"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "33b078d5268709adc0a35f4a", "attempt_id": "927e7a99be88", "run_group_id": "E4_core:A4|closed|qwen3b-llama3b:h4", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T06:32:57.138322Z", "task_family": "synthetic", "document_id": "syn_03000_2000", "document_hash": "c9cc27c20b05b1f6e592846e84990ee9b69c42f9720a8e338ec2f6fca777fdbc", "question_id": "q12_unit_conversion", "split": "test", "cluster_id": "syn_03000_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "tokenizer_hashes": {"receiver": "b9d8872cf3a99da99e1fc4b0f71232a451632dc49ca4c6771549b4efa5c1c763"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 4, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1715, "query_tokens": 537, "message_tokens": 426, "message_utf8_bytes": 1247, "cache_tensor_bytes": null, "serialized_payload_bytes": 1247, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 78247.3377090646, "hops": {"1": 22739.89845905453, "2": 16611.2974579446, "3": 21987.99737507943, "4": 16908.144416986033}, "receiver_prefill_ms": 408.5147500736639, "ttft_ms": 87.1826249640435, "decode_ms": 29.559207963757217}, "total_wall_ms": 78773.23954214808, "ttft_ms": 87.1826249640435, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 29, "retention_rate": 0.4461538461538462, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5, "status": 1.0, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1247}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 29, "retention_rate": 0.4461538461538462, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5, "status": 1.0, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1090}, "2": {"facts_total": 65, "facts_retained": 29, "retention_rate": 0.4461538461538462, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5, "status": 1.0, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1287}, "3": {"facts_total": 65, "facts_retained": 29, "retention_rate": 0.4461538461538462, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5, "status": 1.0, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1232}, "4": {"facts_total": 65, "facts_retained": 29, "retention_rate": 0.4461538461538462, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5, "status": 1.0, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1247}}}, "reference_key_id": "keys/synthetic/syn_03000_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "afb8dd9e0494587cde2c5fb6281f534ae0073a68d145eea31da6ae6240dceb66", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "46b9d455062103b2864d72d7", "attempt_id": "8eb9bfb15134", "run_group_id": "E4_core:A4|closed|qwen3b-llama3b:h4", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T06:32:57.681091Z", "task_family": "synthetic", "document_id": "syn_03000_2000", "document_hash": "c9cc27c20b05b1f6e592846e84990ee9b69c42f9720a8e338ec2f6fca777fdbc", "question_id": "q13_unanswerable", "split": "test", "cluster_id": "syn_03000_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "tokenizer_hashes": {"receiver": "b9d8872cf3a99da99e1fc4b0f71232a451632dc49ca4c6771549b4efa5c1c763"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 4, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1715, "query_tokens": 540, "message_tokens": 426, "message_utf8_bytes": 1247, "cache_tensor_bytes": null, "serialized_payload_bytes": 1247, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 78247.3377090646, "hops": {"1": 22739.89845905453, "2": 16611.2974579446, "3": 21987.99737507943, "4": 16908.144416986033}, "receiver_prefill_ms": 339.3750420073047, "ttft_ms": 38.22262492030859, "decode_ms": 24.70891596749425}, "total_wall_ms": 78650.46854305547, "ttft_ms": 38.22262492030859, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 29, "retention_rate": 0.4461538461538462, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5, "status": 1.0, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1247}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 29, "retention_rate": 0.4461538461538462, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5, "status": 1.0, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1090}, "2": {"facts_total": 65, "facts_retained": 29, "retention_rate": 0.4461538461538462, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5, "status": 1.0, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1287}, "3": {"facts_total": 65, "facts_retained": 29, "retention_rate": 0.4461538461538462, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5, "status": 1.0, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1232}, "4": {"facts_total": 65, "facts_retained": 29, "retention_rate": 0.4461538461538462, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5, "status": 1.0, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1247}}}, "reference_key_id": "keys/synthetic/syn_03000_2000.json", "task_scores": {"correct": 1.0, "abstained": 1.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "afb8dd9e0494587cde2c5fb6281f534ae0073a68d145eea31da6ae6240dceb66", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "f548b2a0ffdc103784273dc8", "attempt_id": "33d20a67ad57", "run_group_id": "E4_core:A4|closed|qwen3b-llama3b:h4", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T06:32:58.102697Z", "task_family": "synthetic", "document_id": "syn_03000_2000", "document_hash": "c9cc27c20b05b1f6e592846e84990ee9b69c42f9720a8e338ec2f6fca777fdbc", "question_id": "q14_unanswerable", "split": "test", "cluster_id": "syn_03000_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "tokenizer_hashes": {"receiver": "b9d8872cf3a99da99e1fc4b0f71232a451632dc49ca4c6771549b4efa5c1c763"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 4, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1715, "query_tokens": 543, "message_tokens": 426, "message_utf8_bytes": 1247, "cache_tensor_bytes": null, "serialized_payload_bytes": 1247, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 78247.3377090646, "hops": {"1": 22739.89845905453, "2": 16611.2974579446, "3": 21987.99737507943, "4": 16908.144416986033}, "receiver_prefill_ms": 342.00770896859467, "ttft_ms": 36.39374999329448, "decode_ms": 19.91091703530401}, "total_wall_ms": 78646.39420900494, "ttft_ms": 36.39374999329448, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 29, "retention_rate": 0.4461538461538462, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5, "status": 1.0, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1247}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 29, "retention_rate": 0.4461538461538462, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5, "status": 1.0, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1090}, "2": {"facts_total": 65, "facts_retained": 29, "retention_rate": 0.4461538461538462, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5, "status": 1.0, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1287}, "3": {"facts_total": 65, "facts_retained": 29, "retention_rate": 0.4461538461538462, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5, "status": 1.0, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1232}, "4": {"facts_total": 65, "facts_retained": 29, "retention_rate": 0.4461538461538462, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5, "status": 1.0, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1247}}}, "reference_key_id": "keys/synthetic/syn_03000_2000.json", "task_scores": {"correct": 1.0, "abstained": 1.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "afb8dd9e0494587cde2c5fb6281f534ae0073a68d145eea31da6ae6240dceb66", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "fef4946258f06773b698441a", "attempt_id": "af9832e58b78", "run_group_id": "E4_core:A4|closed|qwen3b-llama3b:h8", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T06:34:13.894874Z", "task_family": "synthetic", "document_id": "syn_03000_2000", "document_hash": "c9cc27c20b05b1f6e592846e84990ee9b69c42f9720a8e338ec2f6fca777fdbc", "question_id": "q00_single_fact", "split": "test", "cluster_id": "syn_03000_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "tokenizer_hashes": {"receiver": "b9d8872cf3a99da99e1fc4b0f71232a451632dc49ca4c6771549b4efa5c1c763"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 8, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1715, "query_tokens": 511, "message_tokens": 408, "message_utf8_bytes": 1148, "cache_tensor_bytes": null, "serialized_payload_bytes": 1148, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 153616.3691249676, "hops": {"1": 22739.89845905453, "2": 16611.2974579446, "3": 21987.99737507943, "4": 16908.144416986033, "5": 21572.464791010134, "6": 16617.637541959994, "7": 21230.376082938164, "8": 15948.552999994718}, "receiver_prefill_ms": 345.5822910182178, "ttft_ms": 43.16324996761978, "decode_ms": 56.386042037047446}, "total_wall_ms": 154062.4306659447, "ttft_ms": 43.16324996761978, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 29, "retention_rate": 0.4461538461538462, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5, "status": 1.0, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1148}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 29, "retention_rate": 0.4461538461538462, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5, "status": 1.0, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1090}, "2": {"facts_total": 65, "facts_retained": 29, "retention_rate": 0.4461538461538462, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5, "status": 1.0, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1287}, "3": {"facts_total": 65, "facts_retained": 29, "retention_rate": 0.4461538461538462, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5, "status": 1.0, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1232}, "4": {"facts_total": 65, "facts_retained": 29, "retention_rate": 0.4461538461538462, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5, "status": 1.0, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1247}, "5": {"facts_total": 65, "facts_retained": 29, "retention_rate": 0.4461538461538462, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5, "status": 1.0, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1201}, "6": {"facts_total": 65, "facts_retained": 29, "retention_rate": 0.4461538461538462, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5, "status": 1.0, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1231}, "7": {"facts_total": 65, "facts_retained": 29, "retention_rate": 0.4461538461538462, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5, "status": 1.0, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1132}, "8": {"facts_total": 65, "facts_retained": 29, "retention_rate": 0.4461538461538462, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5, "status": 1.0, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1148}}}, "reference_key_id": "keys/synthetic/syn_03000_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "08f00da002c76636d348608e60cedacdd59f6bcfb42a217242f2b543f322eb65", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "f2392ea4d4db1c7c8421b0c6", "attempt_id": "8b9a1ff4478a", "run_group_id": "E4_core:A4|closed|qwen3b-llama3b:h8", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T06:34:14.360601Z", "task_family": "synthetic", "document_id": "syn_03000_2000", "document_hash": "c9cc27c20b05b1f6e592846e84990ee9b69c42f9720a8e338ec2f6fca777fdbc", "question_id": "q01_single_fact", "split": "test", "cluster_id": "syn_03000_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "tokenizer_hashes": {"receiver": "b9d8872cf3a99da99e1fc4b0f71232a451632dc49ca4c6771549b4efa5c1c763"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 8, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1715, "query_tokens": 516, "message_tokens": 408, "message_utf8_bytes": 1148, "cache_tensor_bytes": null, "serialized_payload_bytes": 1148, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 153616.3691249676, "hops": {"1": 22739.89845905453, "2": 16611.2974579446, "3": 21987.99737507943, "4": 16908.144416986033, "5": 21572.464791010134, "6": 16617.637541959994, "7": 21230.376082938164, "8": 15948.552999994718}, "receiver_prefill_ms": 486.70258396305144, "ttft_ms": 37.43429097812623, "decode_ms": 273.5648340312764}, "total_wall_ms": 154414.79395795614, "ttft_ms": 37.43429097812623, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 9}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "Lot 19-N, Gaspé.", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 29, "retention_rate": 0.4461538461538462, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5, "status": 1.0, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1148}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 29, "retention_rate": 0.4461538461538462, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5, "status": 1.0, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1090}, "2": {"facts_total": 65, "facts_retained": 29, "retention_rate": 0.4461538461538462, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5, "status": 1.0, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1287}, "3": {"facts_total": 65, "facts_retained": 29, "retention_rate": 0.4461538461538462, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5, "status": 1.0, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1232}, "4": {"facts_total": 65, "facts_retained": 29, "retention_rate": 0.4461538461538462, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5, "status": 1.0, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1247}, "5": {"facts_total": 65, "facts_retained": 29, "retention_rate": 0.4461538461538462, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5, "status": 1.0, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1201}, "6": {"facts_total": 65, "facts_retained": 29, "retention_rate": 0.4461538461538462, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5, "status": 1.0, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1231}, "7": {"facts_total": 65, "facts_retained": 29, "retention_rate": 0.4461538461538462, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5, "status": 1.0, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1132}, "8": {"facts_total": 65, "facts_retained": 29, "retention_rate": 0.4461538461538462, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5, "status": 1.0, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1148}}}, "reference_key_id": "keys/synthetic/syn_03000_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 0.8571428571428571}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "08f00da002c76636d348608e60cedacdd59f6bcfb42a217242f2b543f322eb65", "output_hash": "ab9cc24b811f5b1342aae731f584e8b0d24b63b61ba59a068b8414deffdc10ae"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "57adc8320dff2f9c37cf5a93", "attempt_id": "228f3782a9bd", "run_group_id": "E4_core:A4|closed|qwen3b-llama3b:h8", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T06:34:15.177614Z", "task_family": "synthetic", "document_id": "syn_03000_2000", "document_hash": "c9cc27c20b05b1f6e592846e84990ee9b69c42f9720a8e338ec2f6fca777fdbc", "question_id": "q02_exact_copy", "split": "test", "cluster_id": "syn_03000_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "tokenizer_hashes": {"receiver": "b9d8872cf3a99da99e1fc4b0f71232a451632dc49ca4c6771549b4efa5c1c763"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 8, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1715, "query_tokens": 515, "message_tokens": 408, "message_utf8_bytes": 1148, "cache_tensor_bytes": null, "serialized_payload_bytes": 1148, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 153616.3691249676, "hops": {"1": 22739.89845905453, "2": 16611.2974579446, "3": 21987.99737507943, "4": 16908.144416986033, "5": 21572.464791010134, "6": 16617.637541959994, "7": 21230.376082938164, "8": 15948.552999994718}, "receiver_prefill_ms": 334.49425001163036, "ttft_ms": 37.7269999589771, "decode_ms": 25.132875074632466}, "total_wall_ms": 154014.32987500448, "ttft_ms": 37.7269999589771, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 29, "retention_rate": 0.4461538461538462, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5, "status": 1.0, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1148}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 29, "retention_rate": 0.4461538461538462, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5, "status": 1.0, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1090}, "2": {"facts_total": 65, "facts_retained": 29, "retention_rate": 0.4461538461538462, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5, "status": 1.0, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1287}, "3": {"facts_total": 65, "facts_retained": 29, "retention_rate": 0.4461538461538462, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5, "status": 1.0, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1232}, "4": {"facts_total": 65, "facts_retained": 29, "retention_rate": 0.4461538461538462, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5, "status": 1.0, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1247}, "5": {"facts_total": 65, "facts_retained": 29, "retention_rate": 0.4461538461538462, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5, "status": 1.0, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1201}, "6": {"facts_total": 65, "facts_retained": 29, "retention_rate": 0.4461538461538462, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5, "status": 1.0, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1231}, "7": {"facts_total": 65, "facts_retained": 29, "retention_rate": 0.4461538461538462, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5, "status": 1.0, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1132}, "8": {"facts_total": 65, "facts_retained": 29, "retention_rate": 0.4461538461538462, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5, "status": 1.0, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1148}}}, "reference_key_id": "keys/synthetic/syn_03000_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "08f00da002c76636d348608e60cedacdd59f6bcfb42a217242f2b543f322eb65", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "a349b07465e4ae6671be458b", "attempt_id": "0d5512736d75", "run_group_id": "E4_core:A4|closed|qwen3b-llama3b:h8", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T06:34:15.593598Z", "task_family": "synthetic", "document_id": "syn_03000_2000", "document_hash": "c9cc27c20b05b1f6e592846e84990ee9b69c42f9720a8e338ec2f6fca777fdbc", "question_id": "q03_exact_copy", "split": "test", "cluster_id": "syn_03000_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "tokenizer_hashes": {"receiver": "b9d8872cf3a99da99e1fc4b0f71232a451632dc49ca4c6771549b4efa5c1c763"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 8, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1715, "query_tokens": 513, "message_tokens": 408, "message_utf8_bytes": 1148, "cache_tensor_bytes": null, "serialized_payload_bytes": 1148, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 153616.3691249676, "hops": {"1": 22739.89845905453, "2": 16611.2974579446, "3": 21987.99737507943, "4": 16908.144416986033, "5": 21572.464791010134, "6": 16617.637541959994, "7": 21230.376082938164, "8": 15948.552999994718}, "receiver_prefill_ms": 337.0840420247987, "ttft_ms": 39.54699996393174, "decode_ms": 35.06849997211248}, "total_wall_ms": 154029.0004160488, "ttft_ms": 39.54699996393174, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 29, "retention_rate": 0.4461538461538462, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5, "status": 1.0, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1148}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 29, "retention_rate": 0.4461538461538462, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5, "status": 1.0, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1090}, "2": {"facts_total": 65, "facts_retained": 29, "retention_rate": 0.4461538461538462, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5, "status": 1.0, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1287}, "3": {"facts_total": 65, "facts_retained": 29, "retention_rate": 0.4461538461538462, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5, "status": 1.0, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1232}, "4": {"facts_total": 65, "facts_retained": 29, "retention_rate": 0.4461538461538462, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5, "status": 1.0, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1247}, "5": {"facts_total": 65, "facts_retained": 29, "retention_rate": 0.4461538461538462, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5, "status": 1.0, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1201}, "6": {"facts_total": 65, "facts_retained": 29, "retention_rate": 0.4461538461538462, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5, "status": 1.0, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1231}, "7": {"facts_total": 65, "facts_retained": 29, "retention_rate": 0.4461538461538462, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5, "status": 1.0, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1132}, "8": {"facts_total": 65, "facts_retained": 29, "retention_rate": 0.4461538461538462, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5, "status": 1.0, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1148}}}, "reference_key_id": "keys/synthetic/syn_03000_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "08f00da002c76636d348608e60cedacdd59f6bcfb42a217242f2b543f322eb65", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "f509b8d29ea78ce22336e584", "attempt_id": "c20b6995deb6", "run_group_id": "E4_core:A4|closed|qwen3b-llama3b:h8", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T06:34:16.023710Z", "task_family": "synthetic", "document_id": "syn_03000_2000", "document_hash": "c9cc27c20b05b1f6e592846e84990ee9b69c42f9720a8e338ec2f6fca777fdbc", "question_id": "q04_multi_hop", "split": "test", "cluster_id": "syn_03000_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "tokenizer_hashes": {"receiver": "b9d8872cf3a99da99e1fc4b0f71232a451632dc49ca4c6771549b4efa5c1c763"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 8, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1715, "query_tokens": 520, "message_tokens": 408, "message_utf8_bytes": 1148, "cache_tensor_bytes": null, "serialized_payload_bytes": 1148, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 153616.3691249676, "hops": {"1": 22739.89845905453, "2": 16611.2974579446, "3": 21987.99737507943, "4": 16908.144416986033, "5": 21572.464791010134, "6": 16617.637541959994, "7": 21230.376082938164, "8": 15948.552999994718}, "receiver_prefill_ms": 354.2139589553699, "ttft_ms": 41.942583047784865, "decode_ms": 243.74470906332135}, "total_wall_ms": 154256.86320790555, "ttft_ms": 41.942583047784865, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 5}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "9,220,745", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 29, "retention_rate": 0.4461538461538462, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5, "status": 1.0, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1148}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 29, "retention_rate": 0.4461538461538462, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5, "status": 1.0, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1090}, "2": {"facts_total": 65, "facts_retained": 29, "retention_rate": 0.4461538461538462, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5, "status": 1.0, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1287}, "3": {"facts_total": 65, "facts_retained": 29, "retention_rate": 0.4461538461538462, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5, "status": 1.0, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1232}, "4": {"facts_total": 65, "facts_retained": 29, "retention_rate": 0.4461538461538462, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5, "status": 1.0, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1247}, "5": {"facts_total": 65, "facts_retained": 29, "retention_rate": 0.4461538461538462, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5, "status": 1.0, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1201}, "6": {"facts_total": 65, "facts_retained": 29, "retention_rate": 0.4461538461538462, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5, "status": 1.0, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1231}, "7": {"facts_total": 65, "facts_retained": 29, "retention_rate": 0.4461538461538462, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5, "status": 1.0, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1132}, "8": {"facts_total": 65, "facts_retained": 29, "retention_rate": 0.4461538461538462, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5, "status": 1.0, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1148}}}, "reference_key_id": "keys/synthetic/syn_03000_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "08f00da002c76636d348608e60cedacdd59f6bcfb42a217242f2b543f322eb65", "output_hash": "ae55ba88b2cb1b0f1a4e0960ae312e0a1c177c79a6d14b67bff1677cec19b649"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "af6f98fb98cbc55fa06a8632", "attempt_id": "4c4f6f66e2d6", "run_group_id": "E4_core:A4|closed|qwen3b-llama3b:h8", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T06:34:16.681949Z", "task_family": "synthetic", "document_id": "syn_03000_2000", "document_hash": "c9cc27c20b05b1f6e592846e84990ee9b69c42f9720a8e338ec2f6fca777fdbc", "question_id": "q05_multi_hop", "split": "test", "cluster_id": "syn_03000_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "tokenizer_hashes": {"receiver": "b9d8872cf3a99da99e1fc4b0f71232a451632dc49ca4c6771549b4efa5c1c763"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 8, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1715, "query_tokens": 520, "message_tokens": 408, "message_utf8_bytes": 1148, "cache_tensor_bytes": null, "serialized_payload_bytes": 1148, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 153616.3691249676, "hops": {"1": 22739.89845905453, "2": 16611.2974579446, "3": 21987.99737507943, "4": 16908.144416986033, "5": 21572.464791010134, "6": 16617.637541959994, "7": 21230.376082938164, "8": 15948.552999994718}, "receiver_prefill_ms": 333.4366249619052, "ttft_ms": 26.73745807260275, "decode_ms": 231.54416598845273}, "total_wall_ms": 154208.61912495457, "ttft_ms": 26.73745807260275, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 7}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "LY-51E-3199", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 29, "retention_rate": 0.4461538461538462, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5, "status": 1.0, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1148}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 29, "retention_rate": 0.4461538461538462, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5, "status": 1.0, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1090}, "2": {"facts_total": 65, "facts_retained": 29, "retention_rate": 0.4461538461538462, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5, "status": 1.0, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1287}, "3": {"facts_total": 65, "facts_retained": 29, "retention_rate": 0.4461538461538462, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5, "status": 1.0, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1232}, "4": {"facts_total": 65, "facts_retained": 29, "retention_rate": 0.4461538461538462, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5, "status": 1.0, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1247}, "5": {"facts_total": 65, "facts_retained": 29, "retention_rate": 0.4461538461538462, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5, "status": 1.0, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1201}, "6": {"facts_total": 65, "facts_retained": 29, "retention_rate": 0.4461538461538462, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5, "status": 1.0, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1231}, "7": {"facts_total": 65, "facts_retained": 29, "retention_rate": 0.4461538461538462, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5, "status": 1.0, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1132}, "8": {"facts_total": 65, "facts_retained": 29, "retention_rate": 0.4461538461538462, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5, "status": 1.0, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1148}}}, "reference_key_id": "keys/synthetic/syn_03000_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "08f00da002c76636d348608e60cedacdd59f6bcfb42a217242f2b543f322eb65", "output_hash": "a1245d1536890a7e4b026366df48441d503f491f782cfd388797b5782443c5c7"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "9c2e4068d47ccd7e86257e2a", "attempt_id": "9e4bdcbc59d5", "run_group_id": "E4_core:A4|closed|qwen3b-llama3b:h8", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T06:34:17.291703Z", "task_family": "synthetic", "document_id": "syn_03000_2000", "document_hash": "c9cc27c20b05b1f6e592846e84990ee9b69c42f9720a8e338ec2f6fca777fdbc", "question_id": "q06_negation", "split": "test", "cluster_id": "syn_03000_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "tokenizer_hashes": {"receiver": "b9d8872cf3a99da99e1fc4b0f71232a451632dc49ca4c6771549b4efa5c1c763"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 8, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1715, "query_tokens": 511, "message_tokens": 408, "message_utf8_bytes": 1148, "cache_tensor_bytes": null, "serialized_payload_bytes": 1148, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 153616.3691249676, "hops": {"1": 22739.89845905453, "2": 16611.2974579446, "3": 21987.99737507943, "4": 16908.144416986033, "5": 21572.464791010134, "6": 16617.637541959994, "7": 21230.376082938164, "8": 15948.552999994718}, "receiver_prefill_ms": 345.4167080344632, "ttft_ms": 58.82625002413988, "decode_ms": 121.92333396524191}, "total_wall_ms": 154143.42929201666, "ttft_ms": 58.82625002413988, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 2}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "No.", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 29, "retention_rate": 0.4461538461538462, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5, "status": 1.0, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1148}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 29, "retention_rate": 0.4461538461538462, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5, "status": 1.0, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1090}, "2": {"facts_total": 65, "facts_retained": 29, "retention_rate": 0.4461538461538462, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5, "status": 1.0, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1287}, "3": {"facts_total": 65, "facts_retained": 29, "retention_rate": 0.4461538461538462, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5, "status": 1.0, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1232}, "4": {"facts_total": 65, "facts_retained": 29, "retention_rate": 0.4461538461538462, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5, "status": 1.0, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1247}, "5": {"facts_total": 65, "facts_retained": 29, "retention_rate": 0.4461538461538462, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5, "status": 1.0, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1201}, "6": {"facts_total": 65, "facts_retained": 29, "retention_rate": 0.4461538461538462, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5, "status": 1.0, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1231}, "7": {"facts_total": 65, "facts_retained": 29, "retention_rate": 0.4461538461538462, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5, "status": 1.0, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1132}, "8": {"facts_total": 65, "facts_retained": 29, "retention_rate": 0.4461538461538462, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5, "status": 1.0, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1148}}}, "reference_key_id": "keys/synthetic/syn_03000_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "08f00da002c76636d348608e60cedacdd59f6bcfb42a217242f2b543f322eb65", "output_hash": "38178a20b470cfd18299fb1593dd4ba706f1a83321dac051d01086efd8b7a96f"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "5c5ac93755b30413acb519ab", "attempt_id": "f52e38b2a53c", "run_group_id": "E4_core:A4|closed|qwen3b-llama3b:h8", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T06:34:17.836162Z", "task_family": "synthetic", "document_id": "syn_03000_2000", "document_hash": "c9cc27c20b05b1f6e592846e84990ee9b69c42f9720a8e338ec2f6fca777fdbc", "question_id": "q07_negation", "split": "test", "cluster_id": "syn_03000_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "tokenizer_hashes": {"receiver": "b9d8872cf3a99da99e1fc4b0f71232a451632dc49ca4c6771549b4efa5c1c763"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 8, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1715, "query_tokens": 511, "message_tokens": 408, "message_utf8_bytes": 1148, "cache_tensor_bytes": null, "serialized_payload_bytes": 1148, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 153616.3691249676, "hops": {"1": 22739.89845905453, "2": 16611.2974579446, "3": 21987.99737507943, "4": 16908.144416986033, "5": 21572.464791010134, "6": 16617.637541959994, "7": 21230.376082938164, "8": 15948.552999994718}, "receiver_prefill_ms": 488.5217910632491, "ttft_ms": 35.384582937695086, "decode_ms": 60.77262491453439}, "total_wall_ms": 154201.723541948, "ttft_ms": 35.384582937695086, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 2}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "No.", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 29, "retention_rate": 0.4461538461538462, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5, "status": 1.0, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1148}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 29, "retention_rate": 0.4461538461538462, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5, "status": 1.0, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1090}, "2": {"facts_total": 65, "facts_retained": 29, "retention_rate": 0.4461538461538462, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5, "status": 1.0, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1287}, "3": {"facts_total": 65, "facts_retained": 29, "retention_rate": 0.4461538461538462, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5, "status": 1.0, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1232}, "4": {"facts_total": 65, "facts_retained": 29, "retention_rate": 0.4461538461538462, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5, "status": 1.0, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1247}, "5": {"facts_total": 65, "facts_retained": 29, "retention_rate": 0.4461538461538462, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5, "status": 1.0, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1201}, "6": {"facts_total": 65, "facts_retained": 29, "retention_rate": 0.4461538461538462, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5, "status": 1.0, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1231}, "7": {"facts_total": 65, "facts_retained": 29, "retention_rate": 0.4461538461538462, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5, "status": 1.0, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1132}, "8": {"facts_total": 65, "facts_retained": 29, "retention_rate": 0.4461538461538462, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5, "status": 1.0, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1148}}}, "reference_key_id": "keys/synthetic/syn_03000_2000.json", "task_scores": {"correct": 0.0, "abstained": 0.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "08f00da002c76636d348608e60cedacdd59f6bcfb42a217242f2b543f322eb65", "output_hash": "38178a20b470cfd18299fb1593dd4ba706f1a83321dac051d01086efd8b7a96f"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "e01505334d0c3a891d0072f5", "attempt_id": "de46f2ebe6ae", "run_group_id": "E4_core:A4|closed|qwen3b-llama3b:h8", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T06:34:18.438864Z", "task_family": "synthetic", "document_id": "syn_03000_2000", "document_hash": "c9cc27c20b05b1f6e592846e84990ee9b69c42f9720a8e338ec2f6fca777fdbc", "question_id": "q08_revision", "split": "test", "cluster_id": "syn_03000_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "tokenizer_hashes": {"receiver": "b9d8872cf3a99da99e1fc4b0f71232a451632dc49ca4c6771549b4efa5c1c763"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 8, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1715, "query_tokens": 526, "message_tokens": 408, "message_utf8_bytes": 1148, "cache_tensor_bytes": null, "serialized_payload_bytes": 1148, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 153616.3691249676, "hops": {"1": 22739.89845905453, "2": 16611.2974579446, "3": 21987.99737507943, "4": 16908.144416986033, "5": 21572.464791010134, "6": 16617.637541959994, "7": 21230.376082938164, "8": 15948.552999994718}, "receiver_prefill_ms": 347.7600420592353, "ttft_ms": 34.88345898222178, "decode_ms": 153.23758299928159}, "total_wall_ms": 154152.80691697262, "ttft_ms": 34.88345898222178, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 5}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "2,705,965", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 29, "retention_rate": 0.4461538461538462, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5, "status": 1.0, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1148}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 29, "retention_rate": 0.4461538461538462, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5, "status": 1.0, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1090}, "2": {"facts_total": 65, "facts_retained": 29, "retention_rate": 0.4461538461538462, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5, "status": 1.0, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1287}, "3": {"facts_total": 65, "facts_retained": 29, "retention_rate": 0.4461538461538462, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5, "status": 1.0, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1232}, "4": {"facts_total": 65, "facts_retained": 29, "retention_rate": 0.4461538461538462, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5, "status": 1.0, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1247}, "5": {"facts_total": 65, "facts_retained": 29, "retention_rate": 0.4461538461538462, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5, "status": 1.0, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1201}, "6": {"facts_total": 65, "facts_retained": 29, "retention_rate": 0.4461538461538462, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5, "status": 1.0, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1231}, "7": {"facts_total": 65, "facts_retained": 29, "retention_rate": 0.4461538461538462, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5, "status": 1.0, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1132}, "8": {"facts_total": 65, "facts_retained": 29, "retention_rate": 0.4461538461538462, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5, "status": 1.0, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1148}}}, "reference_key_id": "keys/synthetic/syn_03000_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "08f00da002c76636d348608e60cedacdd59f6bcfb42a217242f2b543f322eb65", "output_hash": "dbedb6f5d00e413d915513b85d8ed195b18b9acf392ece342a85e15e3752bcc8"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "2dbdb962205cb27a66c262b3", "attempt_id": "2626999054f6", "run_group_id": "E4_core:A4|closed|qwen3b-llama3b:h8", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T06:34:18.993419Z", "task_family": "synthetic", "document_id": "syn_03000_2000", "document_hash": "c9cc27c20b05b1f6e592846e84990ee9b69c42f9720a8e338ec2f6fca777fdbc", "question_id": "q09_revision", "split": "test", "cluster_id": "syn_03000_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "tokenizer_hashes": {"receiver": "b9d8872cf3a99da99e1fc4b0f71232a451632dc49ca4c6771549b4efa5c1c763"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 8, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1715, "query_tokens": 521, "message_tokens": 408, "message_utf8_bytes": 1148, "cache_tensor_bytes": null, "serialized_payload_bytes": 1148, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 153616.3691249676, "hops": {"1": 22739.89845905453, "2": 16611.2974579446, "3": 21987.99737507943, "4": 16908.144416986033, "5": 21572.464791010134, "6": 16617.637541959994, "7": 21230.376082938164, "8": 15948.552999994718}, "receiver_prefill_ms": 331.34204195812345, "ttft_ms": 42.520874994806945, "decode_ms": 406.41366702038795}, "total_wall_ms": 154397.23450003657, "ttft_ms": 42.520874994806945, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 5}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "9,127,965", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 29, "retention_rate": 0.4461538461538462, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5, "status": 1.0, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1148}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 29, "retention_rate": 0.4461538461538462, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5, "status": 1.0, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1090}, "2": {"facts_total": 65, "facts_retained": 29, "retention_rate": 0.4461538461538462, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5, "status": 1.0, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1287}, "3": {"facts_total": 65, "facts_retained": 29, "retention_rate": 0.4461538461538462, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5, "status": 1.0, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1232}, "4": {"facts_total": 65, "facts_retained": 29, "retention_rate": 0.4461538461538462, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5, "status": 1.0, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1247}, "5": {"facts_total": 65, "facts_retained": 29, "retention_rate": 0.4461538461538462, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5, "status": 1.0, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1201}, "6": {"facts_total": 65, "facts_retained": 29, "retention_rate": 0.4461538461538462, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5, "status": 1.0, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1231}, "7": {"facts_total": 65, "facts_retained": 29, "retention_rate": 0.4461538461538462, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5, "status": 1.0, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1132}, "8": {"facts_total": 65, "facts_retained": 29, "retention_rate": 0.4461538461538462, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5, "status": 1.0, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1148}}}, "reference_key_id": "keys/synthetic/syn_03000_2000.json", "task_scores": {"correct": 0.0, "abstained": 0.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "08f00da002c76636d348608e60cedacdd59f6bcfb42a217242f2b543f322eb65", "output_hash": "f30700451becb8c76f6e83d5016caa83d25106e6dceeffd8fdeca27984218444"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "a5376671c0b75205d95b586e", "attempt_id": "6ffd798aa7c1", "run_group_id": "E4_core:A4|closed|qwen3b-llama3b:h8", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T06:34:19.791548Z", "task_family": "synthetic", "document_id": "syn_03000_2000", "document_hash": "c9cc27c20b05b1f6e592846e84990ee9b69c42f9720a8e338ec2f6fca777fdbc", "question_id": "q10_temporal", "split": "test", "cluster_id": "syn_03000_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "tokenizer_hashes": {"receiver": "b9d8872cf3a99da99e1fc4b0f71232a451632dc49ca4c6771549b4efa5c1c763"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 8, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1715, "query_tokens": 528, "message_tokens": 408, "message_utf8_bytes": 1148, "cache_tensor_bytes": null, "serialized_payload_bytes": 1148, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 153616.3691249676, "hops": {"1": 22739.89845905453, "2": 16611.2974579446, "3": 21987.99737507943, "4": 16908.144416986033, "5": 21572.464791010134, "6": 16617.637541959994, "7": 21230.376082938164, "8": 15948.552999994718}, "receiver_prefill_ms": 331.1803339747712, "ttft_ms": 40.37762503139675, "decode_ms": 24.030333966948092}, "total_wall_ms": 154012.49637501314, "ttft_ms": 40.37762503139675, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 29, "retention_rate": 0.4461538461538462, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5, "status": 1.0, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1148}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 29, "retention_rate": 0.4461538461538462, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5, "status": 1.0, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1090}, "2": {"facts_total": 65, "facts_retained": 29, "retention_rate": 0.4461538461538462, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5, "status": 1.0, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1287}, "3": {"facts_total": 65, "facts_retained": 29, "retention_rate": 0.4461538461538462, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5, "status": 1.0, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1232}, "4": {"facts_total": 65, "facts_retained": 29, "retention_rate": 0.4461538461538462, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5, "status": 1.0, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1247}, "5": {"facts_total": 65, "facts_retained": 29, "retention_rate": 0.4461538461538462, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5, "status": 1.0, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1201}, "6": {"facts_total": 65, "facts_retained": 29, "retention_rate": 0.4461538461538462, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5, "status": 1.0, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1231}, "7": {"facts_total": 65, "facts_retained": 29, "retention_rate": 0.4461538461538462, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5, "status": 1.0, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1132}, "8": {"facts_total": 65, "facts_retained": 29, "retention_rate": 0.4461538461538462, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5, "status": 1.0, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1148}}}, "reference_key_id": "keys/synthetic/syn_03000_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "08f00da002c76636d348608e60cedacdd59f6bcfb42a217242f2b543f322eb65", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "6352dfe2c3b98dbbb5090bbd", "attempt_id": "2b0cbb51572b", "run_group_id": "E4_core:A4|closed|qwen3b-llama3b:h8", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T06:34:20.205651Z", "task_family": "synthetic", "document_id": "syn_03000_2000", "document_hash": "c9cc27c20b05b1f6e592846e84990ee9b69c42f9720a8e338ec2f6fca777fdbc", "question_id": "q11_aggregation", "split": "test", "cluster_id": "syn_03000_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "tokenizer_hashes": {"receiver": "b9d8872cf3a99da99e1fc4b0f71232a451632dc49ca4c6771549b4efa5c1c763"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 8, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1715, "query_tokens": 523, "message_tokens": 408, "message_utf8_bytes": 1148, "cache_tensor_bytes": null, "serialized_payload_bytes": 1148, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 153616.3691249676, "hops": {"1": 22739.89845905453, "2": 16611.2974579446, "3": 21987.99737507943, "4": 16908.144416986033, "5": 21572.464791010134, "6": 16617.637541959994, "7": 21230.376082938164, "8": 15948.552999994718}, "receiver_prefill_ms": 336.6600829176605, "ttft_ms": 38.700625067576766, "decode_ms": 143.18637503311038}, "total_wall_ms": 154135.49574988429, "ttft_ms": 38.700625067576766, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 5}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "9,220,745", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 29, "retention_rate": 0.4461538461538462, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5, "status": 1.0, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1148}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 29, "retention_rate": 0.4461538461538462, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5, "status": 1.0, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1090}, "2": {"facts_total": 65, "facts_retained": 29, "retention_rate": 0.4461538461538462, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5, "status": 1.0, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1287}, "3": {"facts_total": 65, "facts_retained": 29, "retention_rate": 0.4461538461538462, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5, "status": 1.0, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1232}, "4": {"facts_total": 65, "facts_retained": 29, "retention_rate": 0.4461538461538462, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5, "status": 1.0, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1247}, "5": {"facts_total": 65, "facts_retained": 29, "retention_rate": 0.4461538461538462, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5, "status": 1.0, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1201}, "6": {"facts_total": 65, "facts_retained": 29, "retention_rate": 0.4461538461538462, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5, "status": 1.0, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1231}, "7": {"facts_total": 65, "facts_retained": 29, "retention_rate": 0.4461538461538462, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5, "status": 1.0, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1132}, "8": {"facts_total": 65, "facts_retained": 29, "retention_rate": 0.4461538461538462, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5, "status": 1.0, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1148}}}, "reference_key_id": "keys/synthetic/syn_03000_2000.json", "task_scores": {"correct": 0.0, "abstained": 0.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "08f00da002c76636d348608e60cedacdd59f6bcfb42a217242f2b543f322eb65", "output_hash": "ae55ba88b2cb1b0f1a4e0960ae312e0a1c177c79a6d14b67bff1677cec19b649"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "00e006320203355e6447a71d", "attempt_id": "de2a35d7e75e", "run_group_id": "E4_core:A4|closed|qwen3b-llama3b:h8", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T06:34:20.742268Z", "task_family": "synthetic", "document_id": "syn_03000_2000", "document_hash": "c9cc27c20b05b1f6e592846e84990ee9b69c42f9720a8e338ec2f6fca777fdbc", "question_id": "q12_unit_conversion", "split": "test", "cluster_id": "syn_03000_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "tokenizer_hashes": {"receiver": "b9d8872cf3a99da99e1fc4b0f71232a451632dc49ca4c6771549b4efa5c1c763"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 8, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1715, "query_tokens": 519, "message_tokens": 408, "message_utf8_bytes": 1148, "cache_tensor_bytes": null, "serialized_payload_bytes": 1148, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 153616.3691249676, "hops": {"1": 22739.89845905453, "2": 16611.2974579446, "3": 21987.99737507943, "4": 16908.144416986033, "5": 21572.464791010134, "6": 16617.637541959994, "7": 21230.376082938164, "8": 15948.552999994718}, "receiver_prefill_ms": 310.1782499579713, "ttft_ms": 38.79341599531472, "decode_ms": 58.043790981173515}, "total_wall_ms": 154024.15612491313, "ttft_ms": 38.79341599531472, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 29, "retention_rate": 0.4461538461538462, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5, "status": 1.0, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1148}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 29, "retention_rate": 0.4461538461538462, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5, "status": 1.0, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1090}, "2": {"facts_total": 65, "facts_retained": 29, "retention_rate": 0.4461538461538462, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5, "status": 1.0, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1287}, "3": {"facts_total": 65, "facts_retained": 29, "retention_rate": 0.4461538461538462, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5, "status": 1.0, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1232}, "4": {"facts_total": 65, "facts_retained": 29, "retention_rate": 0.4461538461538462, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5, "status": 1.0, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1247}, "5": {"facts_total": 65, "facts_retained": 29, "retention_rate": 0.4461538461538462, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5, "status": 1.0, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1201}, "6": {"facts_total": 65, "facts_retained": 29, "retention_rate": 0.4461538461538462, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5, "status": 1.0, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1231}, "7": {"facts_total": 65, "facts_retained": 29, "retention_rate": 0.4461538461538462, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5, "status": 1.0, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1132}, "8": {"facts_total": 65, "facts_retained": 29, "retention_rate": 0.4461538461538462, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5, "status": 1.0, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1148}}}, "reference_key_id": "keys/synthetic/syn_03000_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "08f00da002c76636d348608e60cedacdd59f6bcfb42a217242f2b543f322eb65", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "90e828a02fd239cd2ae7f934", "attempt_id": "f9036f2b2b04", "run_group_id": "E4_core:A4|closed|qwen3b-llama3b:h8", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T06:34:21.168185Z", "task_family": "synthetic", "document_id": "syn_03000_2000", "document_hash": "c9cc27c20b05b1f6e592846e84990ee9b69c42f9720a8e338ec2f6fca777fdbc", "question_id": "q13_unanswerable", "split": "test", "cluster_id": "syn_03000_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "tokenizer_hashes": {"receiver": "b9d8872cf3a99da99e1fc4b0f71232a451632dc49ca4c6771549b4efa5c1c763"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 8, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1715, "query_tokens": 522, "message_tokens": 408, "message_utf8_bytes": 1148, "cache_tensor_bytes": null, "serialized_payload_bytes": 1148, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 153616.3691249676, "hops": {"1": 22739.89845905453, "2": 16611.2974579446, "3": 21987.99737507943, "4": 16908.144416986033, "5": 21572.464791010134, "6": 16617.637541959994, "7": 21230.376082938164, "8": 15948.552999994718}, "receiver_prefill_ms": 471.9435829902068, "ttft_ms": 37.7612910233438, "decode_ms": 24.53191694803536}, "total_wall_ms": 154151.29645902198, "ttft_ms": 37.7612910233438, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 29, "retention_rate": 0.4461538461538462, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5, "status": 1.0, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1148}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 29, "retention_rate": 0.4461538461538462, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5, "status": 1.0, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1090}, "2": {"facts_total": 65, "facts_retained": 29, "retention_rate": 0.4461538461538462, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5, "status": 1.0, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1287}, "3": {"facts_total": 65, "facts_retained": 29, "retention_rate": 0.4461538461538462, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5, "status": 1.0, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1232}, "4": {"facts_total": 65, "facts_retained": 29, "retention_rate": 0.4461538461538462, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5, "status": 1.0, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1247}, "5": {"facts_total": 65, "facts_retained": 29, "retention_rate": 0.4461538461538462, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5, "status": 1.0, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1201}, "6": {"facts_total": 65, "facts_retained": 29, "retention_rate": 0.4461538461538462, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5, "status": 1.0, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1231}, "7": {"facts_total": 65, "facts_retained": 29, "retention_rate": 0.4461538461538462, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5, "status": 1.0, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1132}, "8": {"facts_total": 65, "facts_retained": 29, "retention_rate": 0.4461538461538462, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5, "status": 1.0, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1148}}}, "reference_key_id": "keys/synthetic/syn_03000_2000.json", "task_scores": {"correct": 1.0, "abstained": 1.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "08f00da002c76636d348608e60cedacdd59f6bcfb42a217242f2b543f322eb65", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "c1176a7585aad5a549e3c001", "attempt_id": "55e34fc3c6e1", "run_group_id": "E4_core:A4|closed|qwen3b-llama3b:h8", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T06:34:21.721336Z", "task_family": "synthetic", "document_id": "syn_03000_2000", "document_hash": "c9cc27c20b05b1f6e592846e84990ee9b69c42f9720a8e338ec2f6fca777fdbc", "question_id": "q14_unanswerable", "split": "test", "cluster_id": "syn_03000_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "tokenizer_hashes": {"receiver": "b9d8872cf3a99da99e1fc4b0f71232a451632dc49ca4c6771549b4efa5c1c763"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 8, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1715, "query_tokens": 525, "message_tokens": 408, "message_utf8_bytes": 1148, "cache_tensor_bytes": null, "serialized_payload_bytes": 1148, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 153616.3691249676, "hops": {"1": 22739.89845905453, "2": 16611.2974579446, "3": 21987.99737507943, "4": 16908.144416986033, "5": 21572.464791010134, "6": 16617.637541959994, "7": 21230.376082938164, "8": 15948.552999994718}, "receiver_prefill_ms": 334.6303750295192, "ttft_ms": 31.477667042054236, "decode_ms": 31.299249967560172}, "total_wall_ms": 154014.36116697732, "ttft_ms": 31.477667042054236, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 29, "retention_rate": 0.4461538461538462, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5, "status": 1.0, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1148}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 29, "retention_rate": 0.4461538461538462, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5, "status": 1.0, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1090}, "2": {"facts_total": 65, "facts_retained": 29, "retention_rate": 0.4461538461538462, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5, "status": 1.0, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1287}, "3": {"facts_total": 65, "facts_retained": 29, "retention_rate": 0.4461538461538462, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5, "status": 1.0, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1232}, "4": {"facts_total": 65, "facts_retained": 29, "retention_rate": 0.4461538461538462, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5, "status": 1.0, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1247}, "5": {"facts_total": 65, "facts_retained": 29, "retention_rate": 0.4461538461538462, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5, "status": 1.0, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1201}, "6": {"facts_total": 65, "facts_retained": 29, "retention_rate": 0.4461538461538462, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5, "status": 1.0, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1231}, "7": {"facts_total": 65, "facts_retained": 29, "retention_rate": 0.4461538461538462, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5, "status": 1.0, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1132}, "8": {"facts_total": 65, "facts_retained": 29, "retention_rate": 0.4461538461538462, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5, "status": 1.0, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1148}}}, "reference_key_id": "keys/synthetic/syn_03000_2000.json", "task_scores": {"correct": 1.0, "abstained": 1.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "08f00da002c76636d348608e60cedacdd59f6bcfb42a217242f2b543f322eb65", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "637e4a1b4d6febccbdf84df5", "attempt_id": "54170c57b8a6", "run_group_id": "E4_core:A5|closed|qwen3b-llama3b:h1", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T06:34:44.634657Z", "task_family": "synthetic", "document_id": "syn_03000_2000", "document_hash": "c9cc27c20b05b1f6e592846e84990ee9b69c42f9720a8e338ec2f6fca777fdbc", "question_id": "q00_single_fact", "split": "test", "cluster_id": "syn_03000_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 1, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1904, "query_tokens": 464, "message_tokens": 383, "message_utf8_bytes": 911, "cache_tensor_bytes": null, "serialized_payload_bytes": 911, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 22495.10891607497, "hops": {"1": 22495.10891607497}, "receiver_prefill_ms": 363.0052920198068, "ttft_ms": 43.056416092440486, "decode_ms": 110.4935830226168}, "total_wall_ms": 23012.416708050296, "ttft_ms": 43.056416092440486, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 15, "retention_rate": 0.23076923076923078, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.09090909090909091, "amount": 0.16666666666666666, "status": 0.5, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 855}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 15, "retention_rate": 0.23076923076923078, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.09090909090909091, "amount": 0.16666666666666666, "status": 0.5, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 855}}}, "reference_key_id": "keys/synthetic/syn_03000_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "a409b88ff601f1f9bf4d221a0c042117fadd2be19621e0b0dd2800495cfbe0cc", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "dcf760c93f8c54b977e7191d", "attempt_id": "055a28f179f8", "run_group_id": "E4_core:A5|closed|qwen3b-llama3b:h1", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T06:34:45.172229Z", "task_family": "synthetic", "document_id": "syn_03000_2000", "document_hash": "c9cc27c20b05b1f6e592846e84990ee9b69c42f9720a8e338ec2f6fca777fdbc", "question_id": "q01_single_fact", "split": "test", "cluster_id": "syn_03000_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 1, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1904, "query_tokens": 469, "message_tokens": 383, "message_utf8_bytes": 911, "cache_tensor_bytes": null, "serialized_payload_bytes": 911, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 22495.10891607497, "hops": {"1": 22495.10891607497}, "receiver_prefill_ms": 397.38079195376486, "ttft_ms": 39.878041949123144, "decode_ms": 366.593417013064}, "total_wall_ms": 23299.53074909281, "ttft_ms": 39.878041949123144, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 11}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "Lot 409-H in Senneterre", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 15, "retention_rate": 0.23076923076923078, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.09090909090909091, "amount": 0.16666666666666666, "status": 0.5, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 855}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 15, "retention_rate": 0.23076923076923078, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.09090909090909091, "amount": 0.16666666666666666, "status": 0.5, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 855}}}, "reference_key_id": "keys/synthetic/syn_03000_2000.json", "task_scores": {"correct": 0.0, "abstained": 0.0, "f1": 0.5}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "a409b88ff601f1f9bf4d221a0c042117fadd2be19621e0b0dd2800495cfbe0cc", "output_hash": "928c825f0fee3b69cf0901e3721b2d6c4467e52d2c6707afe17dd750a93d5e31"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "5e59ec343b3f88d230fdcc77", "attempt_id": "da3a383c4b1f", "run_group_id": "E4_core:A5|closed|qwen3b-llama3b:h1", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T06:34:45.997227Z", "task_family": "synthetic", "document_id": "syn_03000_2000", "document_hash": "c9cc27c20b05b1f6e592846e84990ee9b69c42f9720a8e338ec2f6fca777fdbc", "question_id": "q02_exact_copy", "split": "test", "cluster_id": "syn_03000_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 1, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1904, "query_tokens": 468, "message_tokens": 383, "message_utf8_bytes": 911, "cache_tensor_bytes": null, "serialized_payload_bytes": 911, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 22495.10891607497, "hops": {"1": 22495.10891607497}, "receiver_prefill_ms": 281.6635830095038, "ttft_ms": 57.62920901179314, "decode_ms": 59.31483302265406}, "total_wall_ms": 22894.484291085973, "ttft_ms": 57.62920901179314, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 15, "retention_rate": 0.23076923076923078, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.09090909090909091, "amount": 0.16666666666666666, "status": 0.5, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 855}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 15, "retention_rate": 0.23076923076923078, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.09090909090909091, "amount": 0.16666666666666666, "status": 0.5, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 855}}}, "reference_key_id": "keys/synthetic/syn_03000_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "a409b88ff601f1f9bf4d221a0c042117fadd2be19621e0b0dd2800495cfbe0cc", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "01d64fc7d1a8a7967b8b0cb5", "attempt_id": "e4bf4b995037", "run_group_id": "E4_core:A5|closed|qwen3b-llama3b:h1", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T06:34:46.414425Z", "task_family": "synthetic", "document_id": "syn_03000_2000", "document_hash": "c9cc27c20b05b1f6e592846e84990ee9b69c42f9720a8e338ec2f6fca777fdbc", "question_id": "q03_exact_copy", "split": "test", "cluster_id": "syn_03000_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 1, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1904, "query_tokens": 467, "message_tokens": 383, "message_utf8_bytes": 911, "cache_tensor_bytes": null, "serialized_payload_bytes": 911, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 22495.10891607497, "hops": {"1": 22495.10891607497}, "receiver_prefill_ms": 397.49558293260634, "ttft_ms": 123.08124999981374, "decode_ms": 39.108125027269125}, "total_wall_ms": 23055.50337501336, "ttft_ms": 123.08124999981374, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 15, "retention_rate": 0.23076923076923078, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.09090909090909091, "amount": 0.16666666666666666, "status": 0.5, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 855}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 15, "retention_rate": 0.23076923076923078, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.09090909090909091, "amount": 0.16666666666666666, "status": 0.5, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 855}}}, "reference_key_id": "keys/synthetic/syn_03000_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "a409b88ff601f1f9bf4d221a0c042117fadd2be19621e0b0dd2800495cfbe0cc", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "e138fe576db99fa4e558f31c", "attempt_id": "3a675ef2ba09", "run_group_id": "E4_core:A5|closed|qwen3b-llama3b:h1", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T06:34:46.992238Z", "task_family": "synthetic", "document_id": "syn_03000_2000", "document_hash": "c9cc27c20b05b1f6e592846e84990ee9b69c42f9720a8e338ec2f6fca777fdbc", "question_id": "q04_multi_hop", "split": "test", "cluster_id": "syn_03000_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 1, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1904, "query_tokens": 473, "message_tokens": 383, "message_utf8_bytes": 911, "cache_tensor_bytes": null, "serialized_payload_bytes": 911, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 22495.10891607497, "hops": {"1": 22495.10891607497}, "receiver_prefill_ms": 326.31699997000396, "ttft_ms": 42.267458979040384, "decode_ms": 33.019083086401224}, "total_wall_ms": 22897.283624159172, "ttft_ms": 42.267458979040384, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 15, "retention_rate": 0.23076923076923078, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.09090909090909091, "amount": 0.16666666666666666, "status": 0.5, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 855}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 15, "retention_rate": 0.23076923076923078, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.09090909090909091, "amount": 0.16666666666666666, "status": 0.5, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 855}}}, "reference_key_id": "keys/synthetic/syn_03000_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "a409b88ff601f1f9bf4d221a0c042117fadd2be19621e0b0dd2800495cfbe0cc", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "7bc119b3ac45ea0aa8054f35", "attempt_id": "da7d63eb3c56", "run_group_id": "E4_core:A5|closed|qwen3b-llama3b:h1", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T06:34:47.411323Z", "task_family": "synthetic", "document_id": "syn_03000_2000", "document_hash": "c9cc27c20b05b1f6e592846e84990ee9b69c42f9720a8e338ec2f6fca777fdbc", "question_id": "q05_multi_hop", "split": "test", "cluster_id": "syn_03000_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 1, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1904, "query_tokens": 475, "message_tokens": 383, "message_utf8_bytes": 911, "cache_tensor_bytes": null, "serialized_payload_bytes": 911, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 22495.10891607497, "hops": {"1": 22495.10891607497}, "receiver_prefill_ms": 305.1549999509007, "ttft_ms": 31.827291939407587, "decode_ms": 466.37300006113946}, "total_wall_ms": 23299.098000046797, "ttft_ms": 31.827291939407587, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 10}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "LY-51E-3199", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 15, "retention_rate": 0.23076923076923078, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.09090909090909091, "amount": 0.16666666666666666, "status": 0.5, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 855}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 15, "retention_rate": 0.23076923076923078, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.09090909090909091, "amount": 0.16666666666666666, "status": 0.5, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 855}}}, "reference_key_id": "keys/synthetic/syn_03000_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "a409b88ff601f1f9bf4d221a0c042117fadd2be19621e0b0dd2800495cfbe0cc", "output_hash": "a1245d1536890a7e4b026366df48441d503f491f782cfd388797b5782443c5c7"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "1bf1727e4d3b02030258ce58", "attempt_id": "b1d3ed5d4ee2", "run_group_id": "E4_core:A5|closed|qwen3b-llama3b:h1", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T06:34:48.231881Z", "task_family": "synthetic", "document_id": "syn_03000_2000", "document_hash": "c9cc27c20b05b1f6e592846e84990ee9b69c42f9720a8e338ec2f6fca777fdbc", "question_id": "q06_negation", "split": "test", "cluster_id": "syn_03000_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 1, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1904, "query_tokens": 464, "message_tokens": 383, "message_utf8_bytes": 911, "cache_tensor_bytes": null, "serialized_payload_bytes": 911, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 22495.10891607497, "hops": {"1": 22495.10891607497}, "receiver_prefill_ms": 455.38295805454254, "ttft_ms": 40.11962504591793, "decode_ms": 43.083333992399275}, "total_wall_ms": 23034.307957044803, "ttft_ms": 40.11962504591793, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "no", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 15, "retention_rate": 0.23076923076923078, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.09090909090909091, "amount": 0.16666666666666666, "status": 0.5, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 855}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 15, "retention_rate": 0.23076923076923078, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.09090909090909091, "amount": 0.16666666666666666, "status": 0.5, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 855}}}, "reference_key_id": "keys/synthetic/syn_03000_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "a409b88ff601f1f9bf4d221a0c042117fadd2be19621e0b0dd2800495cfbe0cc", "output_hash": "9390298f3fb0c5b160498935d79cb139aef28e1c47358b4bbba61862b9c26e59"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "61a72ee2878da200acff1515", "attempt_id": "5dcdead8cf6c", "run_group_id": "E4_core:A5|closed|qwen3b-llama3b:h1", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T06:34:48.789133Z", "task_family": "synthetic", "document_id": "syn_03000_2000", "document_hash": "c9cc27c20b05b1f6e592846e84990ee9b69c42f9720a8e338ec2f6fca777fdbc", "question_id": "q07_negation", "split": "test", "cluster_id": "syn_03000_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 1, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1904, "query_tokens": 464, "message_tokens": 383, "message_utf8_bytes": 911, "cache_tensor_bytes": null, "serialized_payload_bytes": 911, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 22495.10891607497, "hops": {"1": 22495.10891607497}, "receiver_prefill_ms": 325.8715409319848, "ttft_ms": 35.42095795273781, "decode_ms": 37.980749970301986}, "total_wall_ms": 22894.840082037263, "ttft_ms": 35.42095795273781, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 15, "retention_rate": 0.23076923076923078, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.09090909090909091, "amount": 0.16666666666666666, "status": 0.5, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 855}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 15, "retention_rate": 0.23076923076923078, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.09090909090909091, "amount": 0.16666666666666666, "status": 0.5, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 855}}}, "reference_key_id": "keys/synthetic/syn_03000_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "a409b88ff601f1f9bf4d221a0c042117fadd2be19621e0b0dd2800495cfbe0cc", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "878f29fa6626ffe7d71c6505", "attempt_id": "9324716fda60", "run_group_id": "E4_core:A5|closed|qwen3b-llama3b:h1", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T06:34:49.206455Z", "task_family": "synthetic", "document_id": "syn_03000_2000", "document_hash": "c9cc27c20b05b1f6e592846e84990ee9b69c42f9720a8e338ec2f6fca777fdbc", "question_id": "q08_revision", "split": "test", "cluster_id": "syn_03000_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 1, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1904, "query_tokens": 479, "message_tokens": 383, "message_utf8_bytes": 911, "cache_tensor_bytes": null, "serialized_payload_bytes": 911, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 22495.10891607497, "hops": {"1": 22495.10891607497}, "receiver_prefill_ms": 293.43491594772786, "ttft_ms": 19.918875070288777, "decode_ms": 608.1010000780225}, "total_wall_ms": 23417.3792070942, "ttft_ms": 19.918875070288777, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 9}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "2,705,965", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 15, "retention_rate": 0.23076923076923078, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.09090909090909091, "amount": 0.16666666666666666, "status": 0.5, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 855}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 15, "retention_rate": 0.23076923076923078, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.09090909090909091, "amount": 0.16666666666666666, "status": 0.5, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 855}}}, "reference_key_id": "keys/synthetic/syn_03000_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "a409b88ff601f1f9bf4d221a0c042117fadd2be19621e0b0dd2800495cfbe0cc", "output_hash": "dbedb6f5d00e413d915513b85d8ed195b18b9acf392ece342a85e15e3752bcc8"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "b447db55173d1a0f599a7fb1", "attempt_id": "eb9d704ea257", "run_group_id": "E4_core:A5|closed|qwen3b-llama3b:h1", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T06:34:50.145821Z", "task_family": "synthetic", "document_id": "syn_03000_2000", "document_hash": "c9cc27c20b05b1f6e592846e84990ee9b69c42f9720a8e338ec2f6fca777fdbc", "question_id": "q09_revision", "split": "test", "cluster_id": "syn_03000_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 1, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1904, "query_tokens": 475, "message_tokens": 383, "message_utf8_bytes": 911, "cache_tensor_bytes": null, "serialized_payload_bytes": 911, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 22495.10891607497, "hops": {"1": 22495.10891607497}, "receiver_prefill_ms": 335.26329207234085, "ttft_ms": 40.942166931927204, "decode_ms": 22.03612495213747}, "total_wall_ms": 22893.91195809003, "ttft_ms": 40.942166931927204, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 15, "retention_rate": 0.23076923076923078, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.09090909090909091, "amount": 0.16666666666666666, "status": 0.5, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 855}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 15, "retention_rate": 0.23076923076923078, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.09090909090909091, "amount": 0.16666666666666666, "status": 0.5, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 855}}}, "reference_key_id": "keys/synthetic/syn_03000_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "a409b88ff601f1f9bf4d221a0c042117fadd2be19621e0b0dd2800495cfbe0cc", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "b9a9f50b7510f70d86ee6011", "attempt_id": "4b42f81766d3", "run_group_id": "E4_core:A5|closed|qwen3b-llama3b:h1", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T06:34:50.561677Z", "task_family": "synthetic", "document_id": "syn_03000_2000", "document_hash": "c9cc27c20b05b1f6e592846e84990ee9b69c42f9720a8e338ec2f6fca777fdbc", "question_id": "q10_temporal", "split": "test", "cluster_id": "syn_03000_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 1, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1904, "query_tokens": 481, "message_tokens": 383, "message_utf8_bytes": 911, "cache_tensor_bytes": null, "serialized_payload_bytes": 911, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 22495.10891607497, "hops": {"1": 22495.10891607497}, "receiver_prefill_ms": 316.1107499618083, "ttft_ms": 37.5951660098508, "decode_ms": 30.141166993416846}, "total_wall_ms": 22879.42454102449, "ttft_ms": 37.5951660098508, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 15, "retention_rate": 0.23076923076923078, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.09090909090909091, "amount": 0.16666666666666666, "status": 0.5, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 855}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 15, "retention_rate": 0.23076923076923078, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.09090909090909091, "amount": 0.16666666666666666, "status": 0.5, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 855}}}, "reference_key_id": "keys/synthetic/syn_03000_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "a409b88ff601f1f9bf4d221a0c042117fadd2be19621e0b0dd2800495cfbe0cc", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "7898a6010fdd479d2e23ed8c", "attempt_id": "d051d90fc76d", "run_group_id": "E4_core:A5|closed|qwen3b-llama3b:h1", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T06:34:50.963052Z", "task_family": "synthetic", "document_id": "syn_03000_2000", "document_hash": "c9cc27c20b05b1f6e592846e84990ee9b69c42f9720a8e338ec2f6fca777fdbc", "question_id": "q11_aggregation", "split": "test", "cluster_id": "syn_03000_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 1, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1904, "query_tokens": 476, "message_tokens": 383, "message_utf8_bytes": 911, "cache_tensor_bytes": null, "serialized_payload_bytes": 911, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 22495.10891607497, "hops": {"1": 22495.10891607497}, "receiver_prefill_ms": 300.62945804093033, "ttft_ms": 45.216709026135504, "decode_ms": 61.427999986335635}, "total_wall_ms": 22902.965332148597, "ttft_ms": 45.216709026135504, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 15, "retention_rate": 0.23076923076923078, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.09090909090909091, "amount": 0.16666666666666666, "status": 0.5, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 855}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 15, "retention_rate": 0.23076923076923078, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.09090909090909091, "amount": 0.16666666666666666, "status": 0.5, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 855}}}, "reference_key_id": "keys/synthetic/syn_03000_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "a409b88ff601f1f9bf4d221a0c042117fadd2be19621e0b0dd2800495cfbe0cc", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "b3010217fc25a59996535eb1", "attempt_id": "505ff3991c3b", "run_group_id": "E4_core:A5|closed|qwen3b-llama3b:h1", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T06:34:51.387779Z", "task_family": "synthetic", "document_id": "syn_03000_2000", "document_hash": "c9cc27c20b05b1f6e592846e84990ee9b69c42f9720a8e338ec2f6fca777fdbc", "question_id": "q12_unit_conversion", "split": "test", "cluster_id": "syn_03000_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 1, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1904, "query_tokens": 472, "message_tokens": 383, "message_utf8_bytes": 911, "cache_tensor_bytes": null, "serialized_payload_bytes": 911, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 22495.10891607497, "hops": {"1": 22495.10891607497}, "receiver_prefill_ms": 441.33312499616295, "ttft_ms": 37.53420803695917, "decode_ms": 165.5785000184551}, "total_wall_ms": 23140.173290972598, "ttft_ms": 37.53420803695917, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 5}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "20.46", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 15, "retention_rate": 0.23076923076923078, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.09090909090909091, "amount": 0.16666666666666666, "status": 0.5, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 855}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 15, "retention_rate": 0.23076923076923078, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.09090909090909091, "amount": 0.16666666666666666, "status": 0.5, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 855}}}, "reference_key_id": "keys/synthetic/syn_03000_2000.json", "task_scores": {"correct": 0.0, "abstained": 0.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "a409b88ff601f1f9bf4d221a0c042117fadd2be19621e0b0dd2800495cfbe0cc", "output_hash": "c057e11a5e818ca86a03c336373f3b4e4a2ddd9756e0acd7f5d7d655fd0267fe"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "223a1be9b3ddb6959307a3c5", "attempt_id": "4789a9e27dff", "run_group_id": "E4_core:A5|closed|qwen3b-llama3b:h1", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T06:34:52.050347Z", "task_family": "synthetic", "document_id": "syn_03000_2000", "document_hash": "c9cc27c20b05b1f6e592846e84990ee9b69c42f9720a8e338ec2f6fca777fdbc", "question_id": "q13_unanswerable", "split": "test", "cluster_id": "syn_03000_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 1, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1904, "query_tokens": 476, "message_tokens": 383, "message_utf8_bytes": 911, "cache_tensor_bytes": null, "serialized_payload_bytes": 911, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 22495.10891607497, "hops": {"1": 22495.10891607497}, "receiver_prefill_ms": 334.63758404832333, "ttft_ms": 39.31716701481491, "decode_ms": 24.966166936792433}, "total_wall_ms": 22894.584458088502, "ttft_ms": 39.31716701481491, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 15, "retention_rate": 0.23076923076923078, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.09090909090909091, "amount": 0.16666666666666666, "status": 0.5, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 855}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 15, "retention_rate": 0.23076923076923078, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.09090909090909091, "amount": 0.16666666666666666, "status": 0.5, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 855}}}, "reference_key_id": "keys/synthetic/syn_03000_2000.json", "task_scores": {"correct": 1.0, "abstained": 1.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "a409b88ff601f1f9bf4d221a0c042117fadd2be19621e0b0dd2800495cfbe0cc", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "281da8cfc4c0b4f337d0c494", "attempt_id": "844b54cf26d2", "run_group_id": "E4_core:A5|closed|qwen3b-llama3b:h1", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T06:34:52.468793Z", "task_family": "synthetic", "document_id": "syn_03000_2000", "document_hash": "c9cc27c20b05b1f6e592846e84990ee9b69c42f9720a8e338ec2f6fca777fdbc", "question_id": "q14_unanswerable", "split": "test", "cluster_id": "syn_03000_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 1, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1904, "query_tokens": 478, "message_tokens": 383, "message_utf8_bytes": 911, "cache_tensor_bytes": null, "serialized_payload_bytes": 911, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 22495.10891607497, "hops": {"1": 22495.10891607497}, "receiver_prefill_ms": 296.59966693725437, "ttft_ms": 11.849166941829026, "decode_ms": 20.35612496547401}, "total_wall_ms": 22824.872166034766, "ttft_ms": 11.849166941829026, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 15, "retention_rate": 0.23076923076923078, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.09090909090909091, "amount": 0.16666666666666666, "status": 0.5, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 855}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 15, "retention_rate": 0.23076923076923078, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.09090909090909091, "amount": 0.16666666666666666, "status": 0.5, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 855}}}, "reference_key_id": "keys/synthetic/syn_03000_2000.json", "task_scores": {"correct": 1.0, "abstained": 1.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "a409b88ff601f1f9bf4d221a0c042117fadd2be19621e0b0dd2800495cfbe0cc", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "cccbbedf4440b921790c8587", "attempt_id": "7b7d9be5def1", "run_group_id": "E4_core:A5|closed|qwen3b-llama3b:h2", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T06:35:07.720578Z", "task_family": "synthetic", "document_id": "syn_03000_2000", "document_hash": "c9cc27c20b05b1f6e592846e84990ee9b69c42f9720a8e338ec2f6fca777fdbc", "question_id": "q00_single_fact", "split": "test", "cluster_id": "syn_03000_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "tokenizer_hashes": {"receiver": "b9d8872cf3a99da99e1fc4b0f71232a451632dc49ca4c6771549b4efa5c1c763"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 2, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1715, "query_tokens": 482, "message_tokens": 379, "message_utf8_bytes": 1042, "cache_tensor_bytes": null, "serialized_payload_bytes": 1042, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 37398.12691602856, "hops": {"1": 22495.10891607497, "2": 14903.01799995359}, "receiver_prefill_ms": 358.4892079234123, "ttft_ms": 22.34912500716746, "decode_ms": 20.07083292119205}, "total_wall_ms": 37799.7619160451, "ttft_ms": 22.34912500716746, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 15, "retention_rate": 0.23076923076923078, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.09090909090909091, "amount": 0.16666666666666666, "status": 0.5, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1042}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 15, "retention_rate": 0.23076923076923078, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.09090909090909091, "amount": 0.16666666666666666, "status": 0.5, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 855}, "2": {"facts_total": 65, "facts_retained": 15, "retention_rate": 0.23076923076923078, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.09090909090909091, "amount": 0.16666666666666666, "status": 0.5, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1042}}}, "reference_key_id": "keys/synthetic/syn_03000_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "e02c931be6afeca96f03fda7275aec7c99638064565cf064ecff55bf536a2d82", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "6e4830c566bd7343c62838fe", "attempt_id": "5756af5368c1", "run_group_id": "E4_core:A5|closed|qwen3b-llama3b:h2", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T06:35:08.140050Z", "task_family": "synthetic", "document_id": "syn_03000_2000", "document_hash": "c9cc27c20b05b1f6e592846e84990ee9b69c42f9720a8e338ec2f6fca777fdbc", "question_id": "q01_single_fact", "split": "test", "cluster_id": "syn_03000_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "tokenizer_hashes": {"receiver": "b9d8872cf3a99da99e1fc4b0f71232a451632dc49ca4c6771549b4efa5c1c763"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 2, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1715, "query_tokens": 487, "message_tokens": 379, "message_utf8_bytes": 1042, "cache_tensor_bytes": null, "serialized_payload_bytes": 1042, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 37398.12691602856, "hops": {"1": 22495.10891607497, "2": 14903.01799995359}, "receiver_prefill_ms": 350.58183304499835, "ttft_ms": 43.54495892766863, "decode_ms": 52.420750027522445}, "total_wall_ms": 37845.265083014965, "ttft_ms": 43.54495892766863, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 15, "retention_rate": 0.23076923076923078, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.09090909090909091, "amount": 0.16666666666666666, "status": 0.5, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1042}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 15, "retention_rate": 0.23076923076923078, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.09090909090909091, "amount": 0.16666666666666666, "status": 0.5, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 855}, "2": {"facts_total": 65, "facts_retained": 15, "retention_rate": 0.23076923076923078, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.09090909090909091, "amount": 0.16666666666666666, "status": 0.5, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1042}}}, "reference_key_id": "keys/synthetic/syn_03000_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "e02c931be6afeca96f03fda7275aec7c99638064565cf064ecff55bf536a2d82", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "92d0a2a9e66cbaae3cabbe90", "attempt_id": "53d7a55e5c2d", "run_group_id": "E4_core:A5|closed|qwen3b-llama3b:h2", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T06:35:08.606002Z", "task_family": "synthetic", "document_id": "syn_03000_2000", "document_hash": "c9cc27c20b05b1f6e592846e84990ee9b69c42f9720a8e338ec2f6fca777fdbc", "question_id": "q02_exact_copy", "split": "test", "cluster_id": "syn_03000_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "tokenizer_hashes": {"receiver": "b9d8872cf3a99da99e1fc4b0f71232a451632dc49ca4c6771549b4efa5c1c763"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 2, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1715, "query_tokens": 486, "message_tokens": 379, "message_utf8_bytes": 1042, "cache_tensor_bytes": null, "serialized_payload_bytes": 1042, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 37398.12691602856, "hops": {"1": 22495.10891607497, "2": 14903.01799995359}, "receiver_prefill_ms": 481.02916695643216, "ttft_ms": 35.07016703952104, "decode_ms": 31.575208995491266}, "total_wall_ms": 37946.255166083574, "ttft_ms": 35.07016703952104, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 15, "retention_rate": 0.23076923076923078, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.09090909090909091, "amount": 0.16666666666666666, "status": 0.5, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1042}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 15, "retention_rate": 0.23076923076923078, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.09090909090909091, "amount": 0.16666666666666666, "status": 0.5, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 855}, "2": {"facts_total": 65, "facts_retained": 15, "retention_rate": 0.23076923076923078, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.09090909090909091, "amount": 0.16666666666666666, "status": 0.5, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1042}}}, "reference_key_id": "keys/synthetic/syn_03000_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "e02c931be6afeca96f03fda7275aec7c99638064565cf064ecff55bf536a2d82", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "cd93114b9d81d4b8fa0ac071", "attempt_id": "e1b844bd60b5", "run_group_id": "E4_core:A5|closed|qwen3b-llama3b:h2", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T06:35:09.171798Z", "task_family": "synthetic", "document_id": "syn_03000_2000", "document_hash": "c9cc27c20b05b1f6e592846e84990ee9b69c42f9720a8e338ec2f6fca777fdbc", "question_id": "q03_exact_copy", "split": "test", "cluster_id": "syn_03000_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "tokenizer_hashes": {"receiver": "b9d8872cf3a99da99e1fc4b0f71232a451632dc49ca4c6771549b4efa5c1c763"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 2, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1715, "query_tokens": 484, "message_tokens": 379, "message_utf8_bytes": 1042, "cache_tensor_bytes": null, "serialized_payload_bytes": 1042, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 37398.12691602856, "hops": {"1": 22495.10891607497, "2": 14903.01799995359}, "receiver_prefill_ms": 366.0552080255002, "ttft_ms": 20.503917010501027, "decode_ms": 33.80520804785192}, "total_wall_ms": 37819.06145706307, "ttft_ms": 20.503917010501027, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 15, "retention_rate": 0.23076923076923078, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.09090909090909091, "amount": 0.16666666666666666, "status": 0.5, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1042}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 15, "retention_rate": 0.23076923076923078, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.09090909090909091, "amount": 0.16666666666666666, "status": 0.5, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 855}, "2": {"facts_total": 65, "facts_retained": 15, "retention_rate": 0.23076923076923078, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.09090909090909091, "amount": 0.16666666666666666, "status": 0.5, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1042}}}, "reference_key_id": "keys/synthetic/syn_03000_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "e02c931be6afeca96f03fda7275aec7c99638064565cf064ecff55bf536a2d82", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "c252d0b3343edf3541fab31f", "attempt_id": "7cd5845af5fa", "run_group_id": "E4_core:A5|closed|qwen3b-llama3b:h2", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T06:35:09.610377Z", "task_family": "synthetic", "document_id": "syn_03000_2000", "document_hash": "c9cc27c20b05b1f6e592846e84990ee9b69c42f9720a8e338ec2f6fca777fdbc", "question_id": "q04_multi_hop", "split": "test", "cluster_id": "syn_03000_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "tokenizer_hashes": {"receiver": "b9d8872cf3a99da99e1fc4b0f71232a451632dc49ca4c6771549b4efa5c1c763"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 2, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1715, "query_tokens": 491, "message_tokens": 379, "message_utf8_bytes": 1042, "cache_tensor_bytes": null, "serialized_payload_bytes": 1042, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 37398.12691602856, "hops": {"1": 22495.10891607497, "2": 14903.01799995359}, "receiver_prefill_ms": 344.89229205064476, "ttft_ms": 37.348957965150476, "decode_ms": 42.75575000792742}, "total_wall_ms": 37823.93245806452, "ttft_ms": 37.348957965150476, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 15, "retention_rate": 0.23076923076923078, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.09090909090909091, "amount": 0.16666666666666666, "status": 0.5, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1042}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 15, "retention_rate": 0.23076923076923078, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.09090909090909091, "amount": 0.16666666666666666, "status": 0.5, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 855}, "2": {"facts_total": 65, "facts_retained": 15, "retention_rate": 0.23076923076923078, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.09090909090909091, "amount": 0.16666666666666666, "status": 0.5, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1042}}}, "reference_key_id": "keys/synthetic/syn_03000_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "e02c931be6afeca96f03fda7275aec7c99638064565cf064ecff55bf536a2d82", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "cadbbcbe26df76955c0f254d", "attempt_id": "901b69fedad9", "run_group_id": "E4_core:A5|closed|qwen3b-llama3b:h2", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T06:35:10.053885Z", "task_family": "synthetic", "document_id": "syn_03000_2000", "document_hash": "c9cc27c20b05b1f6e592846e84990ee9b69c42f9720a8e338ec2f6fca777fdbc", "question_id": "q05_multi_hop", "split": "test", "cluster_id": "syn_03000_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "tokenizer_hashes": {"receiver": "b9d8872cf3a99da99e1fc4b0f71232a451632dc49ca4c6771549b4efa5c1c763"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 2, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1715, "query_tokens": 491, "message_tokens": 379, "message_utf8_bytes": 1042, "cache_tensor_bytes": null, "serialized_payload_bytes": 1042, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 37398.12691602856, "hops": {"1": 22495.10891607497, "2": 14903.01799995359}, "receiver_prefill_ms": 422.1110000507906, "ttft_ms": 86.839959025383, "decode_ms": 223.73637498822063}, "total_wall_ms": 38131.33137405384, "ttft_ms": 86.839959025383, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 7}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "LY-51E-3199", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 15, "retention_rate": 0.23076923076923078, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.09090909090909091, "amount": 0.16666666666666666, "status": 0.5, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1042}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 15, "retention_rate": 0.23076923076923078, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.09090909090909091, "amount": 0.16666666666666666, "status": 0.5, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 855}, "2": {"facts_total": 65, "facts_retained": 15, "retention_rate": 0.23076923076923078, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.09090909090909091, "amount": 0.16666666666666666, "status": 0.5, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1042}}}, "reference_key_id": "keys/synthetic/syn_03000_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "e02c931be6afeca96f03fda7275aec7c99638064565cf064ecff55bf536a2d82", "output_hash": "a1245d1536890a7e4b026366df48441d503f491f782cfd388797b5782443c5c7"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "eae6f30ca58ccaab00363dc4", "attempt_id": "ff8b27cf6fca", "run_group_id": "E4_core:A5|closed|qwen3b-llama3b:h2", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T06:35:10.804567Z", "task_family": "synthetic", "document_id": "syn_03000_2000", "document_hash": "c9cc27c20b05b1f6e592846e84990ee9b69c42f9720a8e338ec2f6fca777fdbc", "question_id": "q06_negation", "split": "test", "cluster_id": "syn_03000_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "tokenizer_hashes": {"receiver": "b9d8872cf3a99da99e1fc4b0f71232a451632dc49ca4c6771549b4efa5c1c763"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 2, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1715, "query_tokens": 482, "message_tokens": 379, "message_utf8_bytes": 1042, "cache_tensor_bytes": null, "serialized_payload_bytes": 1042, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 37398.12691602856, "hops": {"1": 22495.10891607497, "2": 14903.01799995359}, "receiver_prefill_ms": 374.9642090406269, "ttft_ms": 39.63512508198619, "decode_ms": 23.201167001388967}, "total_wall_ms": 37836.472415947355, "ttft_ms": 39.63512508198619, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "No", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 15, "retention_rate": 0.23076923076923078, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.09090909090909091, "amount": 0.16666666666666666, "status": 0.5, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1042}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 15, "retention_rate": 0.23076923076923078, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.09090909090909091, "amount": 0.16666666666666666, "status": 0.5, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 855}, "2": {"facts_total": 65, "facts_retained": 15, "retention_rate": 0.23076923076923078, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.09090909090909091, "amount": 0.16666666666666666, "status": 0.5, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1042}}}, "reference_key_id": "keys/synthetic/syn_03000_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "e02c931be6afeca96f03fda7275aec7c99638064565cf064ecff55bf536a2d82", "output_hash": "1ea442a134b2a184bd5d40104401f2a37fbc09ccf3f4bc9da161c6099be3691d"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "1b5234424ee1c4dc3eb0e780", "attempt_id": "b834ee72aff6", "run_group_id": "E4_core:A5|closed|qwen3b-llama3b:h2", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T06:35:11.260080Z", "task_family": "synthetic", "document_id": "syn_03000_2000", "document_hash": "c9cc27c20b05b1f6e592846e84990ee9b69c42f9720a8e338ec2f6fca777fdbc", "question_id": "q07_negation", "split": "test", "cluster_id": "syn_03000_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "tokenizer_hashes": {"receiver": "b9d8872cf3a99da99e1fc4b0f71232a451632dc49ca4c6771549b4efa5c1c763"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 2, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1715, "query_tokens": 482, "message_tokens": 379, "message_utf8_bytes": 1042, "cache_tensor_bytes": null, "serialized_payload_bytes": 1042, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 37398.12691602856, "hops": {"1": 22495.10891607497, "2": 14903.01799995359}, "receiver_prefill_ms": 349.16783298831433, "ttft_ms": 62.739624991081655, "decode_ms": 55.16454204916954}, "total_wall_ms": 37865.98104098812, "ttft_ms": 62.739624991081655, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "No", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 15, "retention_rate": 0.23076923076923078, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.09090909090909091, "amount": 0.16666666666666666, "status": 0.5, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1042}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 15, "retention_rate": 0.23076923076923078, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.09090909090909091, "amount": 0.16666666666666666, "status": 0.5, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 855}, "2": {"facts_total": 65, "facts_retained": 15, "retention_rate": 0.23076923076923078, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.09090909090909091, "amount": 0.16666666666666666, "status": 0.5, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1042}}}, "reference_key_id": "keys/synthetic/syn_03000_2000.json", "task_scores": {"correct": 0.0, "abstained": 0.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "e02c931be6afeca96f03fda7275aec7c99638064565cf064ecff55bf536a2d82", "output_hash": "1ea442a134b2a184bd5d40104401f2a37fbc09ccf3f4bc9da161c6099be3691d"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "5f936c4165a32958e17c8809", "attempt_id": "a5790f3d1f9f", "run_group_id": "E4_core:A5|closed|qwen3b-llama3b:h2", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T06:35:11.746318Z", "task_family": "synthetic", "document_id": "syn_03000_2000", "document_hash": "c9cc27c20b05b1f6e592846e84990ee9b69c42f9720a8e338ec2f6fca777fdbc", "question_id": "q08_revision", "split": "test", "cluster_id": "syn_03000_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "tokenizer_hashes": {"receiver": "b9d8872cf3a99da99e1fc4b0f71232a451632dc49ca4c6771549b4efa5c1c763"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 2, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1715, "query_tokens": 497, "message_tokens": 379, "message_utf8_bytes": 1042, "cache_tensor_bytes": null, "serialized_payload_bytes": 1042, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 37398.12691602856, "hops": {"1": 22495.10891607497, "2": 14903.01799995359}, "receiver_prefill_ms": 482.70225001033396, "ttft_ms": 38.07154099922627, "decode_ms": 151.72891691327095}, "total_wall_ms": 38071.16966601461, "ttft_ms": 38.07154099922627, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 5}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "2,705,965", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 15, "retention_rate": 0.23076923076923078, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.09090909090909091, "amount": 0.16666666666666666, "status": 0.5, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1042}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 15, "retention_rate": 0.23076923076923078, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.09090909090909091, "amount": 0.16666666666666666, "status": 0.5, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 855}, "2": {"facts_total": 65, "facts_retained": 15, "retention_rate": 0.23076923076923078, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.09090909090909091, "amount": 0.16666666666666666, "status": 0.5, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1042}}}, "reference_key_id": "keys/synthetic/syn_03000_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "e02c931be6afeca96f03fda7275aec7c99638064565cf064ecff55bf536a2d82", "output_hash": "dbedb6f5d00e413d915513b85d8ed195b18b9acf392ece342a85e15e3752bcc8"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "ad72a04a7b2e113244c51435", "attempt_id": "b2a8886591db", "run_group_id": "E4_core:A5|closed|qwen3b-llama3b:h2", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T06:35:12.436554Z", "task_family": "synthetic", "document_id": "syn_03000_2000", "document_hash": "c9cc27c20b05b1f6e592846e84990ee9b69c42f9720a8e338ec2f6fca777fdbc", "question_id": "q09_revision", "split": "test", "cluster_id": "syn_03000_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "tokenizer_hashes": {"receiver": "b9d8872cf3a99da99e1fc4b0f71232a451632dc49ca4c6771549b4efa5c1c763"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 2, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1715, "query_tokens": 492, "message_tokens": 379, "message_utf8_bytes": 1042, "cache_tensor_bytes": null, "serialized_payload_bytes": 1042, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 37398.12691602856, "hops": {"1": 22495.10891607497, "2": 14903.01799995359}, "receiver_prefill_ms": 382.9362919786945, "ttft_ms": 33.61608402337879, "decode_ms": 21.668792003765702}, "total_wall_ms": 37836.80516597815, "ttft_ms": 33.61608402337879, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 15, "retention_rate": 0.23076923076923078, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.09090909090909091, "amount": 0.16666666666666666, "status": 0.5, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1042}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 15, "retention_rate": 0.23076923076923078, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.09090909090909091, "amount": 0.16666666666666666, "status": 0.5, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 855}, "2": {"facts_total": 65, "facts_retained": 15, "retention_rate": 0.23076923076923078, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.09090909090909091, "amount": 0.16666666666666666, "status": 0.5, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1042}}}, "reference_key_id": "keys/synthetic/syn_03000_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "e02c931be6afeca96f03fda7275aec7c99638064565cf064ecff55bf536a2d82", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "a733914840a27ef7977339eb", "attempt_id": "976df3efdff2", "run_group_id": "E4_core:A5|closed|qwen3b-llama3b:h2", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T06:35:12.892932Z", "task_family": "synthetic", "document_id": "syn_03000_2000", "document_hash": "c9cc27c20b05b1f6e592846e84990ee9b69c42f9720a8e338ec2f6fca777fdbc", "question_id": "q10_temporal", "split": "test", "cluster_id": "syn_03000_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "tokenizer_hashes": {"receiver": "b9d8872cf3a99da99e1fc4b0f71232a451632dc49ca4c6771549b4efa5c1c763"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 2, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1715, "query_tokens": 499, "message_tokens": 379, "message_utf8_bytes": 1042, "cache_tensor_bytes": null, "serialized_payload_bytes": 1042, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 37398.12691602856, "hops": {"1": 22495.10891607497, "2": 14903.01799995359}, "receiver_prefill_ms": 331.947207916528, "ttft_ms": 45.014041010290384, "decode_ms": 54.516667034476995}, "total_wall_ms": 37830.42804105207, "ttft_ms": 45.014041010290384, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 15, "retention_rate": 0.23076923076923078, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.09090909090909091, "amount": 0.16666666666666666, "status": 0.5, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1042}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 15, "retention_rate": 0.23076923076923078, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.09090909090909091, "amount": 0.16666666666666666, "status": 0.5, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 855}, "2": {"facts_total": 65, "facts_retained": 15, "retention_rate": 0.23076923076923078, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.09090909090909091, "amount": 0.16666666666666666, "status": 0.5, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1042}}}, "reference_key_id": "keys/synthetic/syn_03000_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "e02c931be6afeca96f03fda7275aec7c99638064565cf064ecff55bf536a2d82", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "4b4aedb3d1b3520a67b68fb0", "attempt_id": "dc5b560b26ef", "run_group_id": "E4_core:A5|closed|qwen3b-llama3b:h2", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T06:35:13.343495Z", "task_family": "synthetic", "document_id": "syn_03000_2000", "document_hash": "c9cc27c20b05b1f6e592846e84990ee9b69c42f9720a8e338ec2f6fca777fdbc", "question_id": "q11_aggregation", "split": "test", "cluster_id": "syn_03000_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "tokenizer_hashes": {"receiver": "b9d8872cf3a99da99e1fc4b0f71232a451632dc49ca4c6771549b4efa5c1c763"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 2, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1715, "query_tokens": 494, "message_tokens": 379, "message_utf8_bytes": 1042, "cache_tensor_bytes": null, "serialized_payload_bytes": 1042, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 37398.12691602856, "hops": {"1": 22495.10891607497, "2": 14903.01799995359}, "receiver_prefill_ms": 480.3249581018463, "ttft_ms": 36.15916601847857, "decode_ms": 44.3366669351235}, "total_wall_ms": 37959.5955820987, "ttft_ms": 36.15916601847857, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 15, "retention_rate": 0.23076923076923078, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.09090909090909091, "amount": 0.16666666666666666, "status": 0.5, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1042}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 15, "retention_rate": 0.23076923076923078, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.09090909090909091, "amount": 0.16666666666666666, "status": 0.5, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 855}, "2": {"facts_total": 65, "facts_retained": 15, "retention_rate": 0.23076923076923078, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.09090909090909091, "amount": 0.16666666666666666, "status": 0.5, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1042}}}, "reference_key_id": "keys/synthetic/syn_03000_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "e02c931be6afeca96f03fda7275aec7c99638064565cf064ecff55bf536a2d82", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "fa9017da949c1908e16b5eb5", "attempt_id": "c7469303dfcd", "run_group_id": "E4_core:A5|closed|qwen3b-llama3b:h2", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T06:35:13.925000Z", "task_family": "synthetic", "document_id": "syn_03000_2000", "document_hash": "c9cc27c20b05b1f6e592846e84990ee9b69c42f9720a8e338ec2f6fca777fdbc", "question_id": "q12_unit_conversion", "split": "test", "cluster_id": "syn_03000_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "tokenizer_hashes": {"receiver": "b9d8872cf3a99da99e1fc4b0f71232a451632dc49ca4c6771549b4efa5c1c763"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 2, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1715, "query_tokens": 490, "message_tokens": 379, "message_utf8_bytes": 1042, "cache_tensor_bytes": null, "serialized_payload_bytes": 1042, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 37398.12691602856, "hops": {"1": 22495.10891607497, "2": 14903.01799995359}, "receiver_prefill_ms": 359.3976669944823, "ttft_ms": 35.32991698011756, "decode_ms": 37.635707994922996}, "total_wall_ms": 37831.122206989676, "ttft_ms": 35.32991698011756, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 15, "retention_rate": 0.23076923076923078, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.09090909090909091, "amount": 0.16666666666666666, "status": 0.5, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1042}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 15, "retention_rate": 0.23076923076923078, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.09090909090909091, "amount": 0.16666666666666666, "status": 0.5, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 855}, "2": {"facts_total": 65, "facts_retained": 15, "retention_rate": 0.23076923076923078, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.09090909090909091, "amount": 0.16666666666666666, "status": 0.5, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1042}}}, "reference_key_id": "keys/synthetic/syn_03000_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "e02c931be6afeca96f03fda7275aec7c99638064565cf064ecff55bf536a2d82", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "808fee1d724402f9b48ecff0", "attempt_id": "69662e9d3973", "run_group_id": "E4_core:A5|closed|qwen3b-llama3b:h2", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T06:35:14.375394Z", "task_family": "synthetic", "document_id": "syn_03000_2000", "document_hash": "c9cc27c20b05b1f6e592846e84990ee9b69c42f9720a8e338ec2f6fca777fdbc", "question_id": "q13_unanswerable", "split": "test", "cluster_id": "syn_03000_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "tokenizer_hashes": {"receiver": "b9d8872cf3a99da99e1fc4b0f71232a451632dc49ca4c6771549b4efa5c1c763"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 2, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1715, "query_tokens": 493, "message_tokens": 379, "message_utf8_bytes": 1042, "cache_tensor_bytes": null, "serialized_payload_bytes": 1042, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 37398.12691602856, "hops": {"1": 22495.10891607497, "2": 14903.01799995359}, "receiver_prefill_ms": 351.2060830835253, "ttft_ms": 19.655374926514924, "decode_ms": 18.634250038303435}, "total_wall_ms": 37788.3477070136, "ttft_ms": 19.655374926514924, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 15, "retention_rate": 0.23076923076923078, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.09090909090909091, "amount": 0.16666666666666666, "status": 0.5, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1042}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 15, "retention_rate": 0.23076923076923078, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.09090909090909091, "amount": 0.16666666666666666, "status": 0.5, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 855}, "2": {"facts_total": 65, "facts_retained": 15, "retention_rate": 0.23076923076923078, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.09090909090909091, "amount": 0.16666666666666666, "status": 0.5, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1042}}}, "reference_key_id": "keys/synthetic/syn_03000_2000.json", "task_scores": {"correct": 1.0, "abstained": 1.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "e02c931be6afeca96f03fda7275aec7c99638064565cf064ecff55bf536a2d82", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "4745ef3f504280ce473ffd4e", "attempt_id": "eb23fccfd4ee", "run_group_id": "E4_core:A5|closed|qwen3b-llama3b:h2", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T06:35:14.783605Z", "task_family": "synthetic", "document_id": "syn_03000_2000", "document_hash": "c9cc27c20b05b1f6e592846e84990ee9b69c42f9720a8e338ec2f6fca777fdbc", "question_id": "q14_unanswerable", "split": "test", "cluster_id": "syn_03000_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "tokenizer_hashes": {"receiver": "b9d8872cf3a99da99e1fc4b0f71232a451632dc49ca4c6771549b4efa5c1c763"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 2, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1715, "query_tokens": 496, "message_tokens": 379, "message_utf8_bytes": 1042, "cache_tensor_bytes": null, "serialized_payload_bytes": 1042, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 37398.12691602856, "hops": {"1": 22495.10891607497, "2": 14903.01799995359}, "receiver_prefill_ms": 368.24474995955825, "ttft_ms": 38.704207981936634, "decode_ms": 107.83133294899017}, "total_wall_ms": 37913.66333293263, "ttft_ms": 38.704207981936634, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 15, "retention_rate": 0.23076923076923078, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.09090909090909091, "amount": 0.16666666666666666, "status": 0.5, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1042}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 15, "retention_rate": 0.23076923076923078, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.09090909090909091, "amount": 0.16666666666666666, "status": 0.5, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 855}, "2": {"facts_total": 65, "facts_retained": 15, "retention_rate": 0.23076923076923078, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.09090909090909091, "amount": 0.16666666666666666, "status": 0.5, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1042}}}, "reference_key_id": "keys/synthetic/syn_03000_2000.json", "task_scores": {"correct": 1.0, "abstained": 1.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "e02c931be6afeca96f03fda7275aec7c99638064565cf064ecff55bf536a2d82", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "1027ac31f7d3db6446ea5a7d", "attempt_id": "c4409b219acd", "run_group_id": "E4_core:A5|closed|qwen3b-llama3b:h4", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T06:35:48.761935Z", "task_family": "synthetic", "document_id": "syn_03000_2000", "document_hash": "c9cc27c20b05b1f6e592846e84990ee9b69c42f9720a8e338ec2f6fca777fdbc", "question_id": "q00_single_fact", "split": "test", "cluster_id": "syn_03000_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "tokenizer_hashes": {"receiver": "b9d8872cf3a99da99e1fc4b0f71232a451632dc49ca4c6771549b4efa5c1c763"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 4, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1715, "query_tokens": 482, "message_tokens": 379, "message_utf8_bytes": 1042, "cache_tensor_bytes": null, "serialized_payload_bytes": 1042, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 70837.43058191612, "hops": {"1": 22495.10891607497, "2": 14903.01799995359, "3": 18322.75574991945, "4": 15116.547915968113}, "receiver_prefill_ms": 376.03683304041624, "ttft_ms": 33.407833077944815, "decode_ms": 29.96641700156033}, "total_wall_ms": 71277.32004097197, "ttft_ms": 33.407833077944815, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 15, "retention_rate": 0.23076923076923078, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.09090909090909091, "amount": 0.16666666666666666, "status": 0.5, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1042}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 15, "retention_rate": 0.23076923076923078, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.09090909090909091, "amount": 0.16666666666666666, "status": 0.5, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 855}, "2": {"facts_total": 65, "facts_retained": 15, "retention_rate": 0.23076923076923078, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.09090909090909091, "amount": 0.16666666666666666, "status": 0.5, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1042}, "3": {"facts_total": 65, "facts_retained": 15, "retention_rate": 0.23076923076923078, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.09090909090909091, "amount": 0.16666666666666666, "status": 0.5, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1042}, "4": {"facts_total": 65, "facts_retained": 15, "retention_rate": 0.23076923076923078, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.09090909090909091, "amount": 0.16666666666666666, "status": 0.5, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1042}}}, "reference_key_id": "keys/synthetic/syn_03000_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "e02c931be6afeca96f03fda7275aec7c99638064565cf064ecff55bf536a2d82", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "58de34134164555640431b68", "attempt_id": "2108397029c0", "run_group_id": "E4_core:A5|closed|qwen3b-llama3b:h4", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T06:35:49.219628Z", "task_family": "synthetic", "document_id": "syn_03000_2000", "document_hash": "c9cc27c20b05b1f6e592846e84990ee9b69c42f9720a8e338ec2f6fca777fdbc", "question_id": "q01_single_fact", "split": "test", "cluster_id": "syn_03000_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "tokenizer_hashes": {"receiver": "b9d8872cf3a99da99e1fc4b0f71232a451632dc49ca4c6771549b4efa5c1c763"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 4, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1715, "query_tokens": 487, "message_tokens": 379, "message_utf8_bytes": 1042, "cache_tensor_bytes": null, "serialized_payload_bytes": 1042, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 70837.43058191612, "hops": {"1": 22495.10891607497, "2": 14903.01799995359, "3": 18322.75574991945, "4": 15116.547915968113}, "receiver_prefill_ms": 342.4685839563608, "ttft_ms": 42.29433310683817, "decode_ms": 73.96162499208003}, "total_wall_ms": 71296.74274893478, "ttft_ms": 42.29433310683817, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 15, "retention_rate": 0.23076923076923078, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.09090909090909091, "amount": 0.16666666666666666, "status": 0.5, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1042}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 15, "retention_rate": 0.23076923076923078, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.09090909090909091, "amount": 0.16666666666666666, "status": 0.5, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 855}, "2": {"facts_total": 65, "facts_retained": 15, "retention_rate": 0.23076923076923078, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.09090909090909091, "amount": 0.16666666666666666, "status": 0.5, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1042}, "3": {"facts_total": 65, "facts_retained": 15, "retention_rate": 0.23076923076923078, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.09090909090909091, "amount": 0.16666666666666666, "status": 0.5, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1042}, "4": {"facts_total": 65, "facts_retained": 15, "retention_rate": 0.23076923076923078, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.09090909090909091, "amount": 0.16666666666666666, "status": 0.5, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1042}}}, "reference_key_id": "keys/synthetic/syn_03000_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "e02c931be6afeca96f03fda7275aec7c99638064565cf064ecff55bf536a2d82", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "f98ee364d6ddd504ebc7f955", "attempt_id": "4f588d6e05e5", "run_group_id": "E4_core:A5|closed|qwen3b-llama3b:h4", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T06:35:49.696962Z", "task_family": "synthetic", "document_id": "syn_03000_2000", "document_hash": "c9cc27c20b05b1f6e592846e84990ee9b69c42f9720a8e338ec2f6fca777fdbc", "question_id": "q02_exact_copy", "split": "test", "cluster_id": "syn_03000_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "tokenizer_hashes": {"receiver": "b9d8872cf3a99da99e1fc4b0f71232a451632dc49ca4c6771549b4efa5c1c763"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 4, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1715, "query_tokens": 486, "message_tokens": 379, "message_utf8_bytes": 1042, "cache_tensor_bytes": null, "serialized_payload_bytes": 1042, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 70837.43058191612, "hops": {"1": 22495.10891607497, "2": 14903.01799995359, "3": 18322.75574991945, "4": 15116.547915968113}, "receiver_prefill_ms": 489.3476669676602, "ttft_ms": 37.2660830616951, "decode_ms": 28.803540975786746}, "total_wall_ms": 71393.35208188277, "ttft_ms": 37.2660830616951, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 15, "retention_rate": 0.23076923076923078, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.09090909090909091, "amount": 0.16666666666666666, "status": 0.5, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1042}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 15, "retention_rate": 0.23076923076923078, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.09090909090909091, "amount": 0.16666666666666666, "status": 0.5, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 855}, "2": {"facts_total": 65, "facts_retained": 15, "retention_rate": 0.23076923076923078, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.09090909090909091, "amount": 0.16666666666666666, "status": 0.5, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1042}, "3": {"facts_total": 65, "facts_retained": 15, "retention_rate": 0.23076923076923078, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.09090909090909091, "amount": 0.16666666666666666, "status": 0.5, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1042}, "4": {"facts_total": 65, "facts_retained": 15, "retention_rate": 0.23076923076923078, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.09090909090909091, "amount": 0.16666666666666666, "status": 0.5, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1042}}}, "reference_key_id": "keys/synthetic/syn_03000_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "e02c931be6afeca96f03fda7275aec7c99638064565cf064ecff55bf536a2d82", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "02189319e1364cddea48a1ab", "attempt_id": "5a0fdc8a1f5e", "run_group_id": "E4_core:A5|closed|qwen3b-llama3b:h4", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T06:35:50.271314Z", "task_family": "synthetic", "document_id": "syn_03000_2000", "document_hash": "c9cc27c20b05b1f6e592846e84990ee9b69c42f9720a8e338ec2f6fca777fdbc", "question_id": "q03_exact_copy", "split": "test", "cluster_id": "syn_03000_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "tokenizer_hashes": {"receiver": "b9d8872cf3a99da99e1fc4b0f71232a451632dc49ca4c6771549b4efa5c1c763"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 4, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1715, "query_tokens": 484, "message_tokens": 379, "message_utf8_bytes": 1042, "cache_tensor_bytes": null, "serialized_payload_bytes": 1042, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 70837.43058191612, "hops": {"1": 22495.10891607497, "2": 14903.01799995359, "3": 18322.75574991945, "4": 15116.547915968113}, "receiver_prefill_ms": 353.19345898460597, "ttft_ms": 27.278415975160897, "decode_ms": 39.04279100243002}, "total_wall_ms": 71257.51699891407, "ttft_ms": 27.278415975160897, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 15, "retention_rate": 0.23076923076923078, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.09090909090909091, "amount": 0.16666666666666666, "status": 0.5, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1042}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 15, "retention_rate": 0.23076923076923078, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.09090909090909091, "amount": 0.16666666666666666, "status": 0.5, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 855}, "2": {"facts_total": 65, "facts_retained": 15, "retention_rate": 0.23076923076923078, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.09090909090909091, "amount": 0.16666666666666666, "status": 0.5, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1042}, "3": {"facts_total": 65, "facts_retained": 15, "retention_rate": 0.23076923076923078, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.09090909090909091, "amount": 0.16666666666666666, "status": 0.5, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1042}, "4": {"facts_total": 65, "facts_retained": 15, "retention_rate": 0.23076923076923078, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.09090909090909091, "amount": 0.16666666666666666, "status": 0.5, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1042}}}, "reference_key_id": "keys/synthetic/syn_03000_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "e02c931be6afeca96f03fda7275aec7c99638064565cf064ecff55bf536a2d82", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "dcda2a1dcb9215d4fe1b70bd", "attempt_id": "ca7472e8651e", "run_group_id": "E4_core:A5|closed|qwen3b-llama3b:h4", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T06:35:50.708568Z", "task_family": "synthetic", "document_id": "syn_03000_2000", "document_hash": "c9cc27c20b05b1f6e592846e84990ee9b69c42f9720a8e338ec2f6fca777fdbc", "question_id": "q04_multi_hop", "split": "test", "cluster_id": "syn_03000_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "tokenizer_hashes": {"receiver": "b9d8872cf3a99da99e1fc4b0f71232a451632dc49ca4c6771549b4efa5c1c763"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 4, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1715, "query_tokens": 491, "message_tokens": 379, "message_utf8_bytes": 1042, "cache_tensor_bytes": null, "serialized_payload_bytes": 1042, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 70837.43058191612, "hops": {"1": 22495.10891607497, "2": 14903.01799995359, "3": 18322.75574991945, "4": 15116.547915968113}, "receiver_prefill_ms": 326.4059160137549, "ttft_ms": 72.91116600390524, "decode_ms": 63.03512502927333}, "total_wall_ms": 71300.31083186623, "ttft_ms": 72.91116600390524, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 15, "retention_rate": 0.23076923076923078, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.09090909090909091, "amount": 0.16666666666666666, "status": 0.5, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1042}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 15, "retention_rate": 0.23076923076923078, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.09090909090909091, "amount": 0.16666666666666666, "status": 0.5, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 855}, "2": {"facts_total": 65, "facts_retained": 15, "retention_rate": 0.23076923076923078, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.09090909090909091, "amount": 0.16666666666666666, "status": 0.5, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1042}, "3": {"facts_total": 65, "facts_retained": 15, "retention_rate": 0.23076923076923078, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.09090909090909091, "amount": 0.16666666666666666, "status": 0.5, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1042}, "4": {"facts_total": 65, "facts_retained": 15, "retention_rate": 0.23076923076923078, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.09090909090909091, "amount": 0.16666666666666666, "status": 0.5, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1042}}}, "reference_key_id": "keys/synthetic/syn_03000_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "e02c931be6afeca96f03fda7275aec7c99638064565cf064ecff55bf536a2d82", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "2c48b380da71590c654fefcb", "attempt_id": "79b6df85dde1", "run_group_id": "E4_core:A5|closed|qwen3b-llama3b:h4", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T06:35:51.188609Z", "task_family": "synthetic", "document_id": "syn_03000_2000", "document_hash": "c9cc27c20b05b1f6e592846e84990ee9b69c42f9720a8e338ec2f6fca777fdbc", "question_id": "q05_multi_hop", "split": "test", "cluster_id": "syn_03000_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "tokenizer_hashes": {"receiver": "b9d8872cf3a99da99e1fc4b0f71232a451632dc49ca4c6771549b4efa5c1c763"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 4, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1715, "query_tokens": 491, "message_tokens": 379, "message_utf8_bytes": 1042, "cache_tensor_bytes": null, "serialized_payload_bytes": 1042, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 70837.43058191612, "hops": {"1": 22495.10891607497, "2": 14903.01799995359, "3": 18322.75574991945, "4": 15116.547915968113}, "receiver_prefill_ms": 499.02870901860297, "ttft_ms": 35.25208297651261, "decode_ms": 217.43541699834168}, "total_wall_ms": 71589.75883189123, "ttft_ms": 35.25208297651261, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 7}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "LY-51E-3199", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 15, "retention_rate": 0.23076923076923078, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.09090909090909091, "amount": 0.16666666666666666, "status": 0.5, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1042}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 15, "retention_rate": 0.23076923076923078, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.09090909090909091, "amount": 0.16666666666666666, "status": 0.5, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 855}, "2": {"facts_total": 65, "facts_retained": 15, "retention_rate": 0.23076923076923078, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.09090909090909091, "amount": 0.16666666666666666, "status": 0.5, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1042}, "3": {"facts_total": 65, "facts_retained": 15, "retention_rate": 0.23076923076923078, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.09090909090909091, "amount": 0.16666666666666666, "status": 0.5, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1042}, "4": {"facts_total": 65, "facts_retained": 15, "retention_rate": 0.23076923076923078, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.09090909090909091, "amount": 0.16666666666666666, "status": 0.5, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1042}}}, "reference_key_id": "keys/synthetic/syn_03000_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "e02c931be6afeca96f03fda7275aec7c99638064565cf064ecff55bf536a2d82", "output_hash": "a1245d1536890a7e4b026366df48441d503f491f782cfd388797b5782443c5c7"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "0f7fcc42ce215acd5293f6a4", "attempt_id": "1ed086af3107", "run_group_id": "E4_core:A5|closed|qwen3b-llama3b:h4", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T06:35:51.959209Z", "task_family": "synthetic", "document_id": "syn_03000_2000", "document_hash": "c9cc27c20b05b1f6e592846e84990ee9b69c42f9720a8e338ec2f6fca777fdbc", "question_id": "q06_negation", "split": "test", "cluster_id": "syn_03000_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "tokenizer_hashes": {"receiver": "b9d8872cf3a99da99e1fc4b0f71232a451632dc49ca4c6771549b4efa5c1c763"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 4, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1715, "query_tokens": 482, "message_tokens": 379, "message_utf8_bytes": 1042, "cache_tensor_bytes": null, "serialized_payload_bytes": 1042, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 70837.43058191612, "hops": {"1": 22495.10891607497, "2": 14903.01799995359, "3": 18322.75574991945, "4": 15116.547915968113}, "receiver_prefill_ms": 371.0012079682201, "ttft_ms": 26.00841608364135, "decode_ms": 35.6585409026593}, "total_wall_ms": 71270.53795696702, "ttft_ms": 26.00841608364135, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "No", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 15, "retention_rate": 0.23076923076923078, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.09090909090909091, "amount": 0.16666666666666666, "status": 0.5, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1042}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 15, "retention_rate": 0.23076923076923078, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.09090909090909091, "amount": 0.16666666666666666, "status": 0.5, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 855}, "2": {"facts_total": 65, "facts_retained": 15, "retention_rate": 0.23076923076923078, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.09090909090909091, "amount": 0.16666666666666666, "status": 0.5, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1042}, "3": {"facts_total": 65, "facts_retained": 15, "retention_rate": 0.23076923076923078, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.09090909090909091, "amount": 0.16666666666666666, "status": 0.5, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1042}, "4": {"facts_total": 65, "facts_retained": 15, "retention_rate": 0.23076923076923078, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.09090909090909091, "amount": 0.16666666666666666, "status": 0.5, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1042}}}, "reference_key_id": "keys/synthetic/syn_03000_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "e02c931be6afeca96f03fda7275aec7c99638064565cf064ecff55bf536a2d82", "output_hash": "1ea442a134b2a184bd5d40104401f2a37fbc09ccf3f4bc9da161c6099be3691d"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "bd4ea38281160a69e38a1d92", "attempt_id": "dd706a64c29f", "run_group_id": "E4_core:A5|closed|qwen3b-llama3b:h4", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T06:35:52.409332Z", "task_family": "synthetic", "document_id": "syn_03000_2000", "document_hash": "c9cc27c20b05b1f6e592846e84990ee9b69c42f9720a8e338ec2f6fca777fdbc", "question_id": "q07_negation", "split": "test", "cluster_id": "syn_03000_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "tokenizer_hashes": {"receiver": "b9d8872cf3a99da99e1fc4b0f71232a451632dc49ca4c6771549b4efa5c1c763"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 4, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1715, "query_tokens": 482, "message_tokens": 379, "message_utf8_bytes": 1042, "cache_tensor_bytes": null, "serialized_payload_bytes": 1042, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 70837.43058191612, "hops": {"1": 22495.10891607497, "2": 14903.01799995359, "3": 18322.75574991945, "4": 15116.547915968113}, "receiver_prefill_ms": 340.43058403767645, "ttft_ms": 54.13687496911734, "decode_ms": 65.83658303134143}, "total_wall_ms": 71298.36216499098, "ttft_ms": 54.13687496911734, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "No", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 15, "retention_rate": 0.23076923076923078, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.09090909090909091, "amount": 0.16666666666666666, "status": 0.5, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1042}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 15, "retention_rate": 0.23076923076923078, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.09090909090909091, "amount": 0.16666666666666666, "status": 0.5, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 855}, "2": {"facts_total": 65, "facts_retained": 15, "retention_rate": 0.23076923076923078, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.09090909090909091, "amount": 0.16666666666666666, "status": 0.5, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1042}, "3": {"facts_total": 65, "facts_retained": 15, "retention_rate": 0.23076923076923078, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.09090909090909091, "amount": 0.16666666666666666, "status": 0.5, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1042}, "4": {"facts_total": 65, "facts_retained": 15, "retention_rate": 0.23076923076923078, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.09090909090909091, "amount": 0.16666666666666666, "status": 0.5, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1042}}}, "reference_key_id": "keys/synthetic/syn_03000_2000.json", "task_scores": {"correct": 0.0, "abstained": 0.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "e02c931be6afeca96f03fda7275aec7c99638064565cf064ecff55bf536a2d82", "output_hash": "1ea442a134b2a184bd5d40104401f2a37fbc09ccf3f4bc9da161c6099be3691d"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "f9f3ce91a24d0c667af2b321", "attempt_id": "3b9202a513f5", "run_group_id": "E4_core:A5|closed|qwen3b-llama3b:h4", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T06:35:52.887364Z", "task_family": "synthetic", "document_id": "syn_03000_2000", "document_hash": "c9cc27c20b05b1f6e592846e84990ee9b69c42f9720a8e338ec2f6fca777fdbc", "question_id": "q08_revision", "split": "test", "cluster_id": "syn_03000_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "tokenizer_hashes": {"receiver": "b9d8872cf3a99da99e1fc4b0f71232a451632dc49ca4c6771549b4efa5c1c763"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 4, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1715, "query_tokens": 497, "message_tokens": 379, "message_utf8_bytes": 1042, "cache_tensor_bytes": null, "serialized_payload_bytes": 1042, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 70837.43058191612, "hops": {"1": 22495.10891607497, "2": 14903.01799995359, "3": 18322.75574991945, "4": 15116.547915968113}, "receiver_prefill_ms": 485.39666598662734, "ttft_ms": 43.09475002810359, "decode_ms": 151.0742079699412}, "total_wall_ms": 71517.4977069255, "ttft_ms": 43.09475002810359, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 5}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "2,705,965", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 15, "retention_rate": 0.23076923076923078, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.09090909090909091, "amount": 0.16666666666666666, "status": 0.5, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1042}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 15, "retention_rate": 0.23076923076923078, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.09090909090909091, "amount": 0.16666666666666666, "status": 0.5, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 855}, "2": {"facts_total": 65, "facts_retained": 15, "retention_rate": 0.23076923076923078, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.09090909090909091, "amount": 0.16666666666666666, "status": 0.5, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1042}, "3": {"facts_total": 65, "facts_retained": 15, "retention_rate": 0.23076923076923078, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.09090909090909091, "amount": 0.16666666666666666, "status": 0.5, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1042}, "4": {"facts_total": 65, "facts_retained": 15, "retention_rate": 0.23076923076923078, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.09090909090909091, "amount": 0.16666666666666666, "status": 0.5, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1042}}}, "reference_key_id": "keys/synthetic/syn_03000_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "e02c931be6afeca96f03fda7275aec7c99638064565cf064ecff55bf536a2d82", "output_hash": "dbedb6f5d00e413d915513b85d8ed195b18b9acf392ece342a85e15e3752bcc8"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "3f472d2354c7884828915d2f", "attempt_id": "7f01cd81d1b9", "run_group_id": "E4_core:A5|closed|qwen3b-llama3b:h4", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T06:35:53.585741Z", "task_family": "synthetic", "document_id": "syn_03000_2000", "document_hash": "c9cc27c20b05b1f6e592846e84990ee9b69c42f9720a8e338ec2f6fca777fdbc", "question_id": "q09_revision", "split": "test", "cluster_id": "syn_03000_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "tokenizer_hashes": {"receiver": "b9d8872cf3a99da99e1fc4b0f71232a451632dc49ca4c6771549b4efa5c1c763"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 4, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1715, "query_tokens": 492, "message_tokens": 379, "message_utf8_bytes": 1042, "cache_tensor_bytes": null, "serialized_payload_bytes": 1042, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 70837.43058191612, "hops": {"1": 22495.10891607497, "2": 14903.01799995359, "3": 18322.75574991945, "4": 15116.547915968113}, "receiver_prefill_ms": 375.5066249286756, "ttft_ms": 26.171917095780373, "decode_ms": 33.667624928057194}, "total_wall_ms": 71273.27795687597, "ttft_ms": 26.171917095780373, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 15, "retention_rate": 0.23076923076923078, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.09090909090909091, "amount": 0.16666666666666666, "status": 0.5, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1042}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 15, "retention_rate": 0.23076923076923078, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.09090909090909091, "amount": 0.16666666666666666, "status": 0.5, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 855}, "2": {"facts_total": 65, "facts_retained": 15, "retention_rate": 0.23076923076923078, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.09090909090909091, "amount": 0.16666666666666666, "status": 0.5, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1042}, "3": {"facts_total": 65, "facts_retained": 15, "retention_rate": 0.23076923076923078, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.09090909090909091, "amount": 0.16666666666666666, "status": 0.5, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1042}, "4": {"facts_total": 65, "facts_retained": 15, "retention_rate": 0.23076923076923078, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.09090909090909091, "amount": 0.16666666666666666, "status": 0.5, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1042}}}, "reference_key_id": "keys/synthetic/syn_03000_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "e02c931be6afeca96f03fda7275aec7c99638064565cf064ecff55bf536a2d82", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "f8b36b217abea2c46fac467f", "attempt_id": "56be3ce5b60b", "run_group_id": "E4_core:A5|closed|qwen3b-llama3b:h4", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T06:35:54.038833Z", "task_family": "synthetic", "document_id": "syn_03000_2000", "document_hash": "c9cc27c20b05b1f6e592846e84990ee9b69c42f9720a8e338ec2f6fca777fdbc", "question_id": "q10_temporal", "split": "test", "cluster_id": "syn_03000_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "tokenizer_hashes": {"receiver": "b9d8872cf3a99da99e1fc4b0f71232a451632dc49ca4c6771549b4efa5c1c763"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 4, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1715, "query_tokens": 499, "message_tokens": 379, "message_utf8_bytes": 1042, "cache_tensor_bytes": null, "serialized_payload_bytes": 1042, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 70837.43058191612, "hops": {"1": 22495.10891607497, "2": 14903.01799995359, "3": 18322.75574991945, "4": 15116.547915968113}, "receiver_prefill_ms": 341.9617920881137, "ttft_ms": 42.26099990773946, "decode_ms": 37.687249947339296}, "total_wall_ms": 71260.11504000053, "ttft_ms": 42.26099990773946, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 15, "retention_rate": 0.23076923076923078, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.09090909090909091, "amount": 0.16666666666666666, "status": 0.5, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1042}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 15, "retention_rate": 0.23076923076923078, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.09090909090909091, "amount": 0.16666666666666666, "status": 0.5, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 855}, "2": {"facts_total": 65, "facts_retained": 15, "retention_rate": 0.23076923076923078, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.09090909090909091, "amount": 0.16666666666666666, "status": 0.5, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1042}, "3": {"facts_total": 65, "facts_retained": 15, "retention_rate": 0.23076923076923078, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.09090909090909091, "amount": 0.16666666666666666, "status": 0.5, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1042}, "4": {"facts_total": 65, "facts_retained": 15, "retention_rate": 0.23076923076923078, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.09090909090909091, "amount": 0.16666666666666666, "status": 0.5, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1042}}}, "reference_key_id": "keys/synthetic/syn_03000_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "e02c931be6afeca96f03fda7275aec7c99638064565cf064ecff55bf536a2d82", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "82ab26c2936a3223b44897bc", "attempt_id": "d1e87ea31819", "run_group_id": "E4_core:A5|closed|qwen3b-llama3b:h4", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T06:35:54.478275Z", "task_family": "synthetic", "document_id": "syn_03000_2000", "document_hash": "c9cc27c20b05b1f6e592846e84990ee9b69c42f9720a8e338ec2f6fca777fdbc", "question_id": "q11_aggregation", "split": "test", "cluster_id": "syn_03000_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "tokenizer_hashes": {"receiver": "b9d8872cf3a99da99e1fc4b0f71232a451632dc49ca4c6771549b4efa5c1c763"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 4, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1715, "query_tokens": 494, "message_tokens": 379, "message_utf8_bytes": 1042, "cache_tensor_bytes": null, "serialized_payload_bytes": 1042, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 70837.43058191612, "hops": {"1": 22495.10891607497, "2": 14903.01799995359, "3": 18322.75574991945, "4": 15116.547915968113}, "receiver_prefill_ms": 499.2738750297576, "ttft_ms": 35.4580000275746, "decode_ms": 42.994625051505864}, "total_wall_ms": 71415.84991593845, "ttft_ms": 35.4580000275746, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 15, "retention_rate": 0.23076923076923078, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.09090909090909091, "amount": 0.16666666666666666, "status": 0.5, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1042}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 15, "retention_rate": 0.23076923076923078, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.09090909090909091, "amount": 0.16666666666666666, "status": 0.5, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 855}, "2": {"facts_total": 65, "facts_retained": 15, "retention_rate": 0.23076923076923078, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.09090909090909091, "amount": 0.16666666666666666, "status": 0.5, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1042}, "3": {"facts_total": 65, "facts_retained": 15, "retention_rate": 0.23076923076923078, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.09090909090909091, "amount": 0.16666666666666666, "status": 0.5, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1042}, "4": {"facts_total": 65, "facts_retained": 15, "retention_rate": 0.23076923076923078, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.09090909090909091, "amount": 0.16666666666666666, "status": 0.5, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1042}}}, "reference_key_id": "keys/synthetic/syn_03000_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "e02c931be6afeca96f03fda7275aec7c99638064565cf064ecff55bf536a2d82", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "12992555225cfc0cfd4b1460", "attempt_id": "66f0fd88fbdc", "run_group_id": "E4_core:A5|closed|qwen3b-llama3b:h4", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T06:35:55.073362Z", "task_family": "synthetic", "document_id": "syn_03000_2000", "document_hash": "c9cc27c20b05b1f6e592846e84990ee9b69c42f9720a8e338ec2f6fca777fdbc", "question_id": "q12_unit_conversion", "split": "test", "cluster_id": "syn_03000_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "tokenizer_hashes": {"receiver": "b9d8872cf3a99da99e1fc4b0f71232a451632dc49ca4c6771549b4efa5c1c763"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 4, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1715, "query_tokens": 490, "message_tokens": 379, "message_utf8_bytes": 1042, "cache_tensor_bytes": null, "serialized_payload_bytes": 1042, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 70837.43058191612, "hops": {"1": 22495.10891607497, "2": 14903.01799995359, "3": 18322.75574991945, "4": 15116.547915968113}, "receiver_prefill_ms": 375.8600000292063, "ttft_ms": 24.02450004592538, "decode_ms": 38.13670901581645}, "total_wall_ms": 71275.99233190995, "ttft_ms": 24.02450004592538, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 15, "retention_rate": 0.23076923076923078, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.09090909090909091, "amount": 0.16666666666666666, "status": 0.5, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1042}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 15, "retention_rate": 0.23076923076923078, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.09090909090909091, "amount": 0.16666666666666666, "status": 0.5, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 855}, "2": {"facts_total": 65, "facts_retained": 15, "retention_rate": 0.23076923076923078, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.09090909090909091, "amount": 0.16666666666666666, "status": 0.5, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1042}, "3": {"facts_total": 65, "facts_retained": 15, "retention_rate": 0.23076923076923078, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.09090909090909091, "amount": 0.16666666666666666, "status": 0.5, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1042}, "4": {"facts_total": 65, "facts_retained": 15, "retention_rate": 0.23076923076923078, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.09090909090909091, "amount": 0.16666666666666666, "status": 0.5, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1042}}}, "reference_key_id": "keys/synthetic/syn_03000_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "e02c931be6afeca96f03fda7275aec7c99638064565cf064ecff55bf536a2d82", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "330b94d472fcd27eec92e045", "attempt_id": "1d5428b5faa8", "run_group_id": "E4_core:A5|closed|qwen3b-llama3b:h4", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T06:35:55.529659Z", "task_family": "synthetic", "document_id": "syn_03000_2000", "document_hash": "c9cc27c20b05b1f6e592846e84990ee9b69c42f9720a8e338ec2f6fca777fdbc", "question_id": "q13_unanswerable", "split": "test", "cluster_id": "syn_03000_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "tokenizer_hashes": {"receiver": "b9d8872cf3a99da99e1fc4b0f71232a451632dc49ca4c6771549b4efa5c1c763"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 4, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1715, "query_tokens": 493, "message_tokens": 379, "message_utf8_bytes": 1042, "cache_tensor_bytes": null, "serialized_payload_bytes": 1042, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 70837.43058191612, "hops": {"1": 22495.10891607497, "2": 14903.01799995359, "3": 18322.75574991945, "4": 15116.547915968113}, "receiver_prefill_ms": 355.08466698229313, "ttft_ms": 39.9424999486655, "decode_ms": 57.94958397746086}, "total_wall_ms": 71290.97983194515, "ttft_ms": 39.9424999486655, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 15, "retention_rate": 0.23076923076923078, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.09090909090909091, "amount": 0.16666666666666666, "status": 0.5, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1042}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 15, "retention_rate": 0.23076923076923078, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.09090909090909091, "amount": 0.16666666666666666, "status": 0.5, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 855}, "2": {"facts_total": 65, "facts_retained": 15, "retention_rate": 0.23076923076923078, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.09090909090909091, "amount": 0.16666666666666666, "status": 0.5, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1042}, "3": {"facts_total": 65, "facts_retained": 15, "retention_rate": 0.23076923076923078, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.09090909090909091, "amount": 0.16666666666666666, "status": 0.5, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1042}, "4": {"facts_total": 65, "facts_retained": 15, "retention_rate": 0.23076923076923078, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.09090909090909091, "amount": 0.16666666666666666, "status": 0.5, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1042}}}, "reference_key_id": "keys/synthetic/syn_03000_2000.json", "task_scores": {"correct": 1.0, "abstained": 1.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "e02c931be6afeca96f03fda7275aec7c99638064565cf064ecff55bf536a2d82", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "f980b99bd27d21f7286fac41", "attempt_id": "8d3aac141ac5", "run_group_id": "E4_core:A5|closed|qwen3b-llama3b:h4", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T06:35:56.000965Z", "task_family": "synthetic", "document_id": "syn_03000_2000", "document_hash": "c9cc27c20b05b1f6e592846e84990ee9b69c42f9720a8e338ec2f6fca777fdbc", "question_id": "q14_unanswerable", "split": "test", "cluster_id": "syn_03000_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "tokenizer_hashes": {"receiver": "b9d8872cf3a99da99e1fc4b0f71232a451632dc49ca4c6771549b4efa5c1c763"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 4, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1715, "query_tokens": 496, "message_tokens": 379, "message_utf8_bytes": 1042, "cache_tensor_bytes": null, "serialized_payload_bytes": 1042, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 70837.43058191612, "hops": {"1": 22495.10891607497, "2": 14903.01799995359, "3": 18322.75574991945, "4": 15116.547915968113}, "receiver_prefill_ms": 385.45733399223536, "ttft_ms": 148.9403749583289, "decode_ms": 37.37595793791115}, "total_wall_ms": 71409.72058195621, "ttft_ms": 148.9403749583289, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 15, "retention_rate": 0.23076923076923078, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.09090909090909091, "amount": 0.16666666666666666, "status": 0.5, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1042}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 15, "retention_rate": 0.23076923076923078, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.09090909090909091, "amount": 0.16666666666666666, "status": 0.5, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 855}, "2": {"facts_total": 65, "facts_retained": 15, "retention_rate": 0.23076923076923078, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.09090909090909091, "amount": 0.16666666666666666, "status": 0.5, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1042}, "3": {"facts_total": 65, "facts_retained": 15, "retention_rate": 0.23076923076923078, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.09090909090909091, "amount": 0.16666666666666666, "status": 0.5, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1042}, "4": {"facts_total": 65, "facts_retained": 15, "retention_rate": 0.23076923076923078, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.09090909090909091, "amount": 0.16666666666666666, "status": 0.5, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1042}}}, "reference_key_id": "keys/synthetic/syn_03000_2000.json", "task_scores": {"correct": 1.0, "abstained": 1.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "e02c931be6afeca96f03fda7275aec7c99638064565cf064ecff55bf536a2d82", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "c2e9dac859036ce01f398057", "attempt_id": "169fbac66c54", "run_group_id": "E4_core:A5|closed|qwen3b-llama3b:h8", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T06:37:03.951843Z", "task_family": "synthetic", "document_id": "syn_03000_2000", "document_hash": "c9cc27c20b05b1f6e592846e84990ee9b69c42f9720a8e338ec2f6fca777fdbc", "question_id": "q00_single_fact", "split": "test", "cluster_id": "syn_03000_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "tokenizer_hashes": {"receiver": "b9d8872cf3a99da99e1fc4b0f71232a451632dc49ca4c6771549b4efa5c1c763"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 8, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1715, "query_tokens": 482, "message_tokens": 379, "message_utf8_bytes": 1042, "cache_tensor_bytes": null, "serialized_payload_bytes": 1042, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 138193.43433191534, "hops": {"1": 22495.10891607497, "2": 14903.01799995359, "3": 18322.75574991945, "4": 15116.547915968113, "5": 18017.787417047657, "6": 15147.36470906064, "7": 18164.166332921013, "8": 16026.685290969908}, "receiver_prefill_ms": 433.3449580008164, "ttft_ms": 56.10199994407594, "decode_ms": 48.48304099868983}, "total_wall_ms": 138731.94178997073, "ttft_ms": 56.10199994407594, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 15, "retention_rate": 0.23076923076923078, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.09090909090909091, "amount": 0.16666666666666666, "status": 0.5, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1042}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 15, "retention_rate": 0.23076923076923078, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.09090909090909091, "amount": 0.16666666666666666, "status": 0.5, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 855}, "2": {"facts_total": 65, "facts_retained": 15, "retention_rate": 0.23076923076923078, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.09090909090909091, "amount": 0.16666666666666666, "status": 0.5, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1042}, "3": {"facts_total": 65, "facts_retained": 15, "retention_rate": 0.23076923076923078, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.09090909090909091, "amount": 0.16666666666666666, "status": 0.5, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1042}, "4": {"facts_total": 65, "facts_retained": 15, "retention_rate": 0.23076923076923078, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.09090909090909091, "amount": 0.16666666666666666, "status": 0.5, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1042}, "5": {"facts_total": 65, "facts_retained": 15, "retention_rate": 0.23076923076923078, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.09090909090909091, "amount": 0.16666666666666666, "status": 0.5, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1042}, "6": {"facts_total": 65, "facts_retained": 15, "retention_rate": 0.23076923076923078, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.09090909090909091, "amount": 0.16666666666666666, "status": 0.5, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1042}, "7": {"facts_total": 65, "facts_retained": 15, "retention_rate": 0.23076923076923078, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.09090909090909091, "amount": 0.16666666666666666, "status": 0.5, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1042}, "8": {"facts_total": 65, "facts_retained": 15, "retention_rate": 0.23076923076923078, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.09090909090909091, "amount": 0.16666666666666666, "status": 0.5, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1042}}}, "reference_key_id": "keys/synthetic/syn_03000_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "e02c931be6afeca96f03fda7275aec7c99638064565cf064ecff55bf536a2d82", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "b2e29c0c5fbb747d02182f4a", "attempt_id": "a4525eff0944", "run_group_id": "E4_core:A5|closed|qwen3b-llama3b:h8", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T06:37:04.508162Z", "task_family": "synthetic", "document_id": "syn_03000_2000", "document_hash": "c9cc27c20b05b1f6e592846e84990ee9b69c42f9720a8e338ec2f6fca777fdbc", "question_id": "q01_single_fact", "split": "test", "cluster_id": "syn_03000_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "tokenizer_hashes": {"receiver": "b9d8872cf3a99da99e1fc4b0f71232a451632dc49ca4c6771549b4efa5c1c763"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 8, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1715, "query_tokens": 487, "message_tokens": 379, "message_utf8_bytes": 1042, "cache_tensor_bytes": null, "serialized_payload_bytes": 1042, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 138193.43433191534, "hops": {"1": 22495.10891607497, "2": 14903.01799995359, "3": 18322.75574991945, "4": 15116.547915968113, "5": 18017.787417047657, "6": 15147.36470906064, "7": 18164.166332921013, "8": 16026.685290969908}, "receiver_prefill_ms": 437.34908290207386, "ttft_ms": 49.665375030599535, "decode_ms": 85.88354208040982}, "total_wall_ms": 138766.92612294573, "ttft_ms": 49.665375030599535, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 15, "retention_rate": 0.23076923076923078, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.09090909090909091, "amount": 0.16666666666666666, "status": 0.5, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1042}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 15, "retention_rate": 0.23076923076923078, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.09090909090909091, "amount": 0.16666666666666666, "status": 0.5, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 855}, "2": {"facts_total": 65, "facts_retained": 15, "retention_rate": 0.23076923076923078, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.09090909090909091, "amount": 0.16666666666666666, "status": 0.5, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1042}, "3": {"facts_total": 65, "facts_retained": 15, "retention_rate": 0.23076923076923078, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.09090909090909091, "amount": 0.16666666666666666, "status": 0.5, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1042}, "4": {"facts_total": 65, "facts_retained": 15, "retention_rate": 0.23076923076923078, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.09090909090909091, "amount": 0.16666666666666666, "status": 0.5, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1042}, "5": {"facts_total": 65, "facts_retained": 15, "retention_rate": 0.23076923076923078, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.09090909090909091, "amount": 0.16666666666666666, "status": 0.5, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1042}, "6": {"facts_total": 65, "facts_retained": 15, "retention_rate": 0.23076923076923078, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.09090909090909091, "amount": 0.16666666666666666, "status": 0.5, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1042}, "7": {"facts_total": 65, "facts_retained": 15, "retention_rate": 0.23076923076923078, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.09090909090909091, "amount": 0.16666666666666666, "status": 0.5, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1042}, "8": {"facts_total": 65, "facts_retained": 15, "retention_rate": 0.23076923076923078, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.09090909090909091, "amount": 0.16666666666666666, "status": 0.5, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1042}}}, "reference_key_id": "keys/synthetic/syn_03000_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "e02c931be6afeca96f03fda7275aec7c99638064565cf064ecff55bf536a2d82", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "42bbcd70295a46a2ac7ecb71", "attempt_id": "d5903ebd8b26", "run_group_id": "E4_core:A5|closed|qwen3b-llama3b:h8", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T06:37:05.099081Z", "task_family": "synthetic", "document_id": "syn_03000_2000", "document_hash": "c9cc27c20b05b1f6e592846e84990ee9b69c42f9720a8e338ec2f6fca777fdbc", "question_id": "q02_exact_copy", "split": "test", "cluster_id": "syn_03000_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "tokenizer_hashes": {"receiver": "b9d8872cf3a99da99e1fc4b0f71232a451632dc49ca4c6771549b4efa5c1c763"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 8, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1715, "query_tokens": 486, "message_tokens": 379, "message_utf8_bytes": 1042, "cache_tensor_bytes": null, "serialized_payload_bytes": 1042, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 138193.43433191534, "hops": {"1": 22495.10891607497, "2": 14903.01799995359, "3": 18322.75574991945, "4": 15116.547915968113, "5": 18017.787417047657, "6": 15147.36470906064, "7": 18164.166332921013, "8": 16026.685290969908}, "receiver_prefill_ms": 454.17204091791064, "ttft_ms": 52.590749925002456, "decode_ms": 48.71783300768584}, "total_wall_ms": 138749.54724893905, "ttft_ms": 52.590749925002456, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 15, "retention_rate": 0.23076923076923078, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.09090909090909091, "amount": 0.16666666666666666, "status": 0.5, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1042}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 15, "retention_rate": 0.23076923076923078, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.09090909090909091, "amount": 0.16666666666666666, "status": 0.5, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 855}, "2": {"facts_total": 65, "facts_retained": 15, "retention_rate": 0.23076923076923078, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.09090909090909091, "amount": 0.16666666666666666, "status": 0.5, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1042}, "3": {"facts_total": 65, "facts_retained": 15, "retention_rate": 0.23076923076923078, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.09090909090909091, "amount": 0.16666666666666666, "status": 0.5, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1042}, "4": {"facts_total": 65, "facts_retained": 15, "retention_rate": 0.23076923076923078, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.09090909090909091, "amount": 0.16666666666666666, "status": 0.5, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1042}, "5": {"facts_total": 65, "facts_retained": 15, "retention_rate": 0.23076923076923078, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.09090909090909091, "amount": 0.16666666666666666, "status": 0.5, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1042}, "6": {"facts_total": 65, "facts_retained": 15, "retention_rate": 0.23076923076923078, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.09090909090909091, "amount": 0.16666666666666666, "status": 0.5, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1042}, "7": {"facts_total": 65, "facts_retained": 15, "retention_rate": 0.23076923076923078, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.09090909090909091, "amount": 0.16666666666666666, "status": 0.5, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1042}, "8": {"facts_total": 65, "facts_retained": 15, "retention_rate": 0.23076923076923078, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.09090909090909091, "amount": 0.16666666666666666, "status": 0.5, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1042}}}, "reference_key_id": "keys/synthetic/syn_03000_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "e02c931be6afeca96f03fda7275aec7c99638064565cf064ecff55bf536a2d82", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "2bd2c5178bb7f935855663cd", "attempt_id": "382485950bf1", "run_group_id": "E4_core:A5|closed|qwen3b-llama3b:h8", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T06:37:05.673100Z", "task_family": "synthetic", "document_id": "syn_03000_2000", "document_hash": "c9cc27c20b05b1f6e592846e84990ee9b69c42f9720a8e338ec2f6fca777fdbc", "question_id": "q03_exact_copy", "split": "test", "cluster_id": "syn_03000_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "tokenizer_hashes": {"receiver": "b9d8872cf3a99da99e1fc4b0f71232a451632dc49ca4c6771549b4efa5c1c763"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 8, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1715, "query_tokens": 484, "message_tokens": 379, "message_utf8_bytes": 1042, "cache_tensor_bytes": null, "serialized_payload_bytes": 1042, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 138193.43433191534, "hops": {"1": 22495.10891607497, "2": 14903.01799995359, "3": 18322.75574991945, "4": 15116.547915968113, "5": 18017.787417047657, "6": 15147.36470906064, "7": 18164.166332921013, "8": 16026.685290969908}, "receiver_prefill_ms": 423.8163329428062, "ttft_ms": 47.44866699911654, "decode_ms": 46.777875046245754}, "total_wall_ms": 138712.02266484033, "ttft_ms": 47.44866699911654, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 15, "retention_rate": 0.23076923076923078, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.09090909090909091, "amount": 0.16666666666666666, "status": 0.5, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1042}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 15, "retention_rate": 0.23076923076923078, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.09090909090909091, "amount": 0.16666666666666666, "status": 0.5, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 855}, "2": {"facts_total": 65, "facts_retained": 15, "retention_rate": 0.23076923076923078, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.09090909090909091, "amount": 0.16666666666666666, "status": 0.5, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1042}, "3": {"facts_total": 65, "facts_retained": 15, "retention_rate": 0.23076923076923078, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.09090909090909091, "amount": 0.16666666666666666, "status": 0.5, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1042}, "4": {"facts_total": 65, "facts_retained": 15, "retention_rate": 0.23076923076923078, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.09090909090909091, "amount": 0.16666666666666666, "status": 0.5, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1042}, "5": {"facts_total": 65, "facts_retained": 15, "retention_rate": 0.23076923076923078, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.09090909090909091, "amount": 0.16666666666666666, "status": 0.5, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1042}, "6": {"facts_total": 65, "facts_retained": 15, "retention_rate": 0.23076923076923078, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.09090909090909091, "amount": 0.16666666666666666, "status": 0.5, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1042}, "7": {"facts_total": 65, "facts_retained": 15, "retention_rate": 0.23076923076923078, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.09090909090909091, "amount": 0.16666666666666666, "status": 0.5, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1042}, "8": {"facts_total": 65, "facts_retained": 15, "retention_rate": 0.23076923076923078, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.09090909090909091, "amount": 0.16666666666666666, "status": 0.5, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1042}}}, "reference_key_id": "keys/synthetic/syn_03000_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "e02c931be6afeca96f03fda7275aec7c99638064565cf064ecff55bf536a2d82", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "14e5ec48fb36500bcc96cbc9", "attempt_id": "eda2191044d4", "run_group_id": "E4_core:A5|closed|qwen3b-llama3b:h8", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T06:37:06.208930Z", "task_family": "synthetic", "document_id": "syn_03000_2000", "document_hash": "c9cc27c20b05b1f6e592846e84990ee9b69c42f9720a8e338ec2f6fca777fdbc", "question_id": "q04_multi_hop", "split": "test", "cluster_id": "syn_03000_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "tokenizer_hashes": {"receiver": "b9d8872cf3a99da99e1fc4b0f71232a451632dc49ca4c6771549b4efa5c1c763"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 8, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1715, "query_tokens": 491, "message_tokens": 379, "message_utf8_bytes": 1042, "cache_tensor_bytes": null, "serialized_payload_bytes": 1042, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 138193.43433191534, "hops": {"1": 22495.10891607497, "2": 14903.01799995359, "3": 18322.75574991945, "4": 15116.547915968113, "5": 18017.787417047657, "6": 15147.36470906064, "7": 18164.166332921013, "8": 16026.685290969908}, "receiver_prefill_ms": 447.75954203214496, "ttft_ms": 84.12645908538252, "decode_ms": 52.01529199257493}, "total_wall_ms": 138777.8691238491, "ttft_ms": 84.12645908538252, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 15, "retention_rate": 0.23076923076923078, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.09090909090909091, "amount": 0.16666666666666666, "status": 0.5, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1042}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 15, "retention_rate": 0.23076923076923078, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.09090909090909091, "amount": 0.16666666666666666, "status": 0.5, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 855}, "2": {"facts_total": 65, "facts_retained": 15, "retention_rate": 0.23076923076923078, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.09090909090909091, "amount": 0.16666666666666666, "status": 0.5, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1042}, "3": {"facts_total": 65, "facts_retained": 15, "retention_rate": 0.23076923076923078, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.09090909090909091, "amount": 0.16666666666666666, "status": 0.5, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1042}, "4": {"facts_total": 65, "facts_retained": 15, "retention_rate": 0.23076923076923078, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.09090909090909091, "amount": 0.16666666666666666, "status": 0.5, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1042}, "5": {"facts_total": 65, "facts_retained": 15, "retention_rate": 0.23076923076923078, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.09090909090909091, "amount": 0.16666666666666666, "status": 0.5, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1042}, "6": {"facts_total": 65, "facts_retained": 15, "retention_rate": 0.23076923076923078, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.09090909090909091, "amount": 0.16666666666666666, "status": 0.5, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1042}, "7": {"facts_total": 65, "facts_retained": 15, "retention_rate": 0.23076923076923078, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.09090909090909091, "amount": 0.16666666666666666, "status": 0.5, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1042}, "8": {"facts_total": 65, "facts_retained": 15, "retention_rate": 0.23076923076923078, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.09090909090909091, "amount": 0.16666666666666666, "status": 0.5, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1042}}}, "reference_key_id": "keys/synthetic/syn_03000_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "e02c931be6afeca96f03fda7275aec7c99638064565cf064ecff55bf536a2d82", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "8f16a643ba1642bcc96d2b46", "attempt_id": "0b83587aff48", "run_group_id": "E4_core:A5|closed|qwen3b-llama3b:h8", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T06:37:06.809718Z", "task_family": "synthetic", "document_id": "syn_03000_2000", "document_hash": "c9cc27c20b05b1f6e592846e84990ee9b69c42f9720a8e338ec2f6fca777fdbc", "question_id": "q05_multi_hop", "split": "test", "cluster_id": "syn_03000_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "tokenizer_hashes": {"receiver": "b9d8872cf3a99da99e1fc4b0f71232a451632dc49ca4c6771549b4efa5c1c763"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 8, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1715, "query_tokens": 491, "message_tokens": 379, "message_utf8_bytes": 1042, "cache_tensor_bytes": null, "serialized_payload_bytes": 1042, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 138193.43433191534, "hops": {"1": 22495.10891607497, "2": 14903.01799995359, "3": 18322.75574991945, "4": 15116.547915968113, "5": 18017.787417047657, "6": 15147.36470906064, "7": 18164.166332921013, "8": 16026.685290969908}, "receiver_prefill_ms": 435.1180420489982, "ttft_ms": 48.139375052414834, "decode_ms": 397.99337496515363}, "total_wall_ms": 139075.20641596057, "ttft_ms": 48.139375052414834, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 7}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "LY-51E-3199", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 15, "retention_rate": 0.23076923076923078, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.09090909090909091, "amount": 0.16666666666666666, "status": 0.5, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1042}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 15, "retention_rate": 0.23076923076923078, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.09090909090909091, "amount": 0.16666666666666666, "status": 0.5, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 855}, "2": {"facts_total": 65, "facts_retained": 15, "retention_rate": 0.23076923076923078, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.09090909090909091, "amount": 0.16666666666666666, "status": 0.5, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1042}, "3": {"facts_total": 65, "facts_retained": 15, "retention_rate": 0.23076923076923078, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.09090909090909091, "amount": 0.16666666666666666, "status": 0.5, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1042}, "4": {"facts_total": 65, "facts_retained": 15, "retention_rate": 0.23076923076923078, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.09090909090909091, "amount": 0.16666666666666666, "status": 0.5, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1042}, "5": {"facts_total": 65, "facts_retained": 15, "retention_rate": 0.23076923076923078, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.09090909090909091, "amount": 0.16666666666666666, "status": 0.5, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1042}, "6": {"facts_total": 65, "facts_retained": 15, "retention_rate": 0.23076923076923078, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.09090909090909091, "amount": 0.16666666666666666, "status": 0.5, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1042}, "7": {"facts_total": 65, "facts_retained": 15, "retention_rate": 0.23076923076923078, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.09090909090909091, "amount": 0.16666666666666666, "status": 0.5, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1042}, "8": {"facts_total": 65, "facts_retained": 15, "retention_rate": 0.23076923076923078, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.09090909090909091, "amount": 0.16666666666666666, "status": 0.5, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1042}}}, "reference_key_id": "keys/synthetic/syn_03000_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "e02c931be6afeca96f03fda7275aec7c99638064565cf064ecff55bf536a2d82", "output_hash": "a1245d1536890a7e4b026366df48441d503f491f782cfd388797b5782443c5c7"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "7c716f3a3a7e63869731a99d", "attempt_id": "d6ac8de31b40", "run_group_id": "E4_core:A5|closed|qwen3b-llama3b:h8", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T06:37:07.708253Z", "task_family": "synthetic", "document_id": "syn_03000_2000", "document_hash": "c9cc27c20b05b1f6e592846e84990ee9b69c42f9720a8e338ec2f6fca777fdbc", "question_id": "q06_negation", "split": "test", "cluster_id": "syn_03000_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "tokenizer_hashes": {"receiver": "b9d8872cf3a99da99e1fc4b0f71232a451632dc49ca4c6771549b4efa5c1c763"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 8, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1715, "query_tokens": 482, "message_tokens": 379, "message_utf8_bytes": 1042, "cache_tensor_bytes": null, "serialized_payload_bytes": 1042, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 138193.43433191534, "hops": {"1": 22495.10891607497, "2": 14903.01799995359, "3": 18322.75574991945, "4": 15116.547915968113, "5": 18017.787417047657, "6": 15147.36470906064, "7": 18164.166332921013, "8": 16026.685290969908}, "receiver_prefill_ms": 311.7721660528332, "ttft_ms": 11.910791974514723, "decode_ms": 11.919917073100805}, "total_wall_ms": 138529.69666488934, "ttft_ms": 11.910791974514723, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "No", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 15, "retention_rate": 0.23076923076923078, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.09090909090909091, "amount": 0.16666666666666666, "status": 0.5, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1042}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 15, "retention_rate": 0.23076923076923078, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.09090909090909091, "amount": 0.16666666666666666, "status": 0.5, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 855}, "2": {"facts_total": 65, "facts_retained": 15, "retention_rate": 0.23076923076923078, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.09090909090909091, "amount": 0.16666666666666666, "status": 0.5, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1042}, "3": {"facts_total": 65, "facts_retained": 15, "retention_rate": 0.23076923076923078, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.09090909090909091, "amount": 0.16666666666666666, "status": 0.5, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1042}, "4": {"facts_total": 65, "facts_retained": 15, "retention_rate": 0.23076923076923078, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.09090909090909091, "amount": 0.16666666666666666, "status": 0.5, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1042}, "5": {"facts_total": 65, "facts_retained": 15, "retention_rate": 0.23076923076923078, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.09090909090909091, "amount": 0.16666666666666666, "status": 0.5, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1042}, "6": {"facts_total": 65, "facts_retained": 15, "retention_rate": 0.23076923076923078, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.09090909090909091, "amount": 0.16666666666666666, "status": 0.5, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1042}, "7": {"facts_total": 65, "facts_retained": 15, "retention_rate": 0.23076923076923078, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.09090909090909091, "amount": 0.16666666666666666, "status": 0.5, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1042}, "8": {"facts_total": 65, "facts_retained": 15, "retention_rate": 0.23076923076923078, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.09090909090909091, "amount": 0.16666666666666666, "status": 0.5, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1042}}}, "reference_key_id": "keys/synthetic/syn_03000_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "e02c931be6afeca96f03fda7275aec7c99638064565cf064ecff55bf536a2d82", "output_hash": "1ea442a134b2a184bd5d40104401f2a37fbc09ccf3f4bc9da161c6099be3691d"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "42f0d49eac9f503e8d6819e7", "attempt_id": "30774da35574", "run_group_id": "E4_core:A5|closed|qwen3b-llama3b:h8", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T06:37:08.061464Z", "task_family": "synthetic", "document_id": "syn_03000_2000", "document_hash": "c9cc27c20b05b1f6e592846e84990ee9b69c42f9720a8e338ec2f6fca777fdbc", "question_id": "q07_negation", "split": "test", "cluster_id": "syn_03000_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "tokenizer_hashes": {"receiver": "b9d8872cf3a99da99e1fc4b0f71232a451632dc49ca4c6771549b4efa5c1c763"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 8, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1715, "query_tokens": 482, "message_tokens": 379, "message_utf8_bytes": 1042, "cache_tensor_bytes": null, "serialized_payload_bytes": 1042, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 138193.43433191534, "hops": {"1": 22495.10891607497, "2": 14903.01799995359, "3": 18322.75574991945, "4": 15116.547915968113, "5": 18017.787417047657, "6": 15147.36470906064, "7": 18164.166332921013, "8": 16026.685290969908}, "receiver_prefill_ms": 207.8140409430489, "ttft_ms": 11.70958299189806, "decode_ms": 11.71966700349003}, "total_wall_ms": 138425.25791481603, "ttft_ms": 11.70958299189806, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "No", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 15, "retention_rate": 0.23076923076923078, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.09090909090909091, "amount": 0.16666666666666666, "status": 0.5, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1042}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 15, "retention_rate": 0.23076923076923078, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.09090909090909091, "amount": 0.16666666666666666, "status": 0.5, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 855}, "2": {"facts_total": 65, "facts_retained": 15, "retention_rate": 0.23076923076923078, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.09090909090909091, "amount": 0.16666666666666666, "status": 0.5, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1042}, "3": {"facts_total": 65, "facts_retained": 15, "retention_rate": 0.23076923076923078, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.09090909090909091, "amount": 0.16666666666666666, "status": 0.5, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1042}, "4": {"facts_total": 65, "facts_retained": 15, "retention_rate": 0.23076923076923078, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.09090909090909091, "amount": 0.16666666666666666, "status": 0.5, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1042}, "5": {"facts_total": 65, "facts_retained": 15, "retention_rate": 0.23076923076923078, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.09090909090909091, "amount": 0.16666666666666666, "status": 0.5, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1042}, "6": {"facts_total": 65, "facts_retained": 15, "retention_rate": 0.23076923076923078, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.09090909090909091, "amount": 0.16666666666666666, "status": 0.5, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1042}, "7": {"facts_total": 65, "facts_retained": 15, "retention_rate": 0.23076923076923078, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.09090909090909091, "amount": 0.16666666666666666, "status": 0.5, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1042}, "8": {"facts_total": 65, "facts_retained": 15, "retention_rate": 0.23076923076923078, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.09090909090909091, "amount": 0.16666666666666666, "status": 0.5, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1042}}}, "reference_key_id": "keys/synthetic/syn_03000_2000.json", "task_scores": {"correct": 0.0, "abstained": 0.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "e02c931be6afeca96f03fda7275aec7c99638064565cf064ecff55bf536a2d82", "output_hash": "1ea442a134b2a184bd5d40104401f2a37fbc09ccf3f4bc9da161c6099be3691d"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "520f9081c5ea749ded6a81a3", "attempt_id": "98b55e340426", "run_group_id": "E4_core:A5|closed|qwen3b-llama3b:h8", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T06:37:08.310652Z", "task_family": "synthetic", "document_id": "syn_03000_2000", "document_hash": "c9cc27c20b05b1f6e592846e84990ee9b69c42f9720a8e338ec2f6fca777fdbc", "question_id": "q08_revision", "split": "test", "cluster_id": "syn_03000_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "tokenizer_hashes": {"receiver": "b9d8872cf3a99da99e1fc4b0f71232a451632dc49ca4c6771549b4efa5c1c763"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 8, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1715, "query_tokens": 497, "message_tokens": 379, "message_utf8_bytes": 1042, "cache_tensor_bytes": null, "serialized_payload_bytes": 1042, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 138193.43433191534, "hops": {"1": 22495.10891607497, "2": 14903.01799995359, "3": 18322.75574991945, "4": 15116.547915968113, "5": 18017.787417047657, "6": 15147.36470906064, "7": 18164.166332921013, "8": 16026.685290969908}, "receiver_prefill_ms": 354.6623749425635, "ttft_ms": 53.416125010699034, "decode_ms": 342.29724993929267}, "total_wall_ms": 138944.41728980746, "ttft_ms": 53.416125010699034, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 5}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "2,705,965", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 15, "retention_rate": 0.23076923076923078, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.09090909090909091, "amount": 0.16666666666666666, "status": 0.5, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1042}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 15, "retention_rate": 0.23076923076923078, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.09090909090909091, "amount": 0.16666666666666666, "status": 0.5, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 855}, "2": {"facts_total": 65, "facts_retained": 15, "retention_rate": 0.23076923076923078, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.09090909090909091, "amount": 0.16666666666666666, "status": 0.5, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1042}, "3": {"facts_total": 65, "facts_retained": 15, "retention_rate": 0.23076923076923078, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.09090909090909091, "amount": 0.16666666666666666, "status": 0.5, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1042}, "4": {"facts_total": 65, "facts_retained": 15, "retention_rate": 0.23076923076923078, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.09090909090909091, "amount": 0.16666666666666666, "status": 0.5, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1042}, "5": {"facts_total": 65, "facts_retained": 15, "retention_rate": 0.23076923076923078, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.09090909090909091, "amount": 0.16666666666666666, "status": 0.5, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1042}, "6": {"facts_total": 65, "facts_retained": 15, "retention_rate": 0.23076923076923078, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.09090909090909091, "amount": 0.16666666666666666, "status": 0.5, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1042}, "7": {"facts_total": 65, "facts_retained": 15, "retention_rate": 0.23076923076923078, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.09090909090909091, "amount": 0.16666666666666666, "status": 0.5, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1042}, "8": {"facts_total": 65, "facts_retained": 15, "retention_rate": 0.23076923076923078, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.09090909090909091, "amount": 0.16666666666666666, "status": 0.5, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1042}}}, "reference_key_id": "keys/synthetic/syn_03000_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "e02c931be6afeca96f03fda7275aec7c99638064565cf064ecff55bf536a2d82", "output_hash": "dbedb6f5d00e413d915513b85d8ed195b18b9acf392ece342a85e15e3752bcc8"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "4334f93427a0a6f65cfc9236", "attempt_id": "43a2b418cf42", "run_group_id": "E4_core:A5|closed|qwen3b-llama3b:h8", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T06:37:09.078793Z", "task_family": "synthetic", "document_id": "syn_03000_2000", "document_hash": "c9cc27c20b05b1f6e592846e84990ee9b69c42f9720a8e338ec2f6fca777fdbc", "question_id": "q09_revision", "split": "test", "cluster_id": "syn_03000_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "tokenizer_hashes": {"receiver": "b9d8872cf3a99da99e1fc4b0f71232a451632dc49ca4c6771549b4efa5c1c763"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 8, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1715, "query_tokens": 492, "message_tokens": 379, "message_utf8_bytes": 1042, "cache_tensor_bytes": null, "serialized_payload_bytes": 1042, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 138193.43433191534, "hops": {"1": 22495.10891607497, "2": 14903.01799995359, "3": 18322.75574991945, "4": 15116.547915968113, "5": 18017.787417047657, "6": 15147.36470906064, "7": 18164.166332921013, "8": 16026.685290969908}, "receiver_prefill_ms": 371.1742080049589, "ttft_ms": 32.90824999567121, "decode_ms": 34.73183396272361}, "total_wall_ms": 138632.73962389212, "ttft_ms": 32.90824999567121, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 15, "retention_rate": 0.23076923076923078, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.09090909090909091, "amount": 0.16666666666666666, "status": 0.5, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1042}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 15, "retention_rate": 0.23076923076923078, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.09090909090909091, "amount": 0.16666666666666666, "status": 0.5, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 855}, "2": {"facts_total": 65, "facts_retained": 15, "retention_rate": 0.23076923076923078, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.09090909090909091, "amount": 0.16666666666666666, "status": 0.5, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1042}, "3": {"facts_total": 65, "facts_retained": 15, "retention_rate": 0.23076923076923078, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.09090909090909091, "amount": 0.16666666666666666, "status": 0.5, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1042}, "4": {"facts_total": 65, "facts_retained": 15, "retention_rate": 0.23076923076923078, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.09090909090909091, "amount": 0.16666666666666666, "status": 0.5, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1042}, "5": {"facts_total": 65, "facts_retained": 15, "retention_rate": 0.23076923076923078, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.09090909090909091, "amount": 0.16666666666666666, "status": 0.5, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1042}, "6": {"facts_total": 65, "facts_retained": 15, "retention_rate": 0.23076923076923078, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.09090909090909091, "amount": 0.16666666666666666, "status": 0.5, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1042}, "7": {"facts_total": 65, "facts_retained": 15, "retention_rate": 0.23076923076923078, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.09090909090909091, "amount": 0.16666666666666666, "status": 0.5, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1042}, "8": {"facts_total": 65, "facts_retained": 15, "retention_rate": 0.23076923076923078, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.09090909090909091, "amount": 0.16666666666666666, "status": 0.5, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1042}}}, "reference_key_id": "keys/synthetic/syn_03000_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "e02c931be6afeca96f03fda7275aec7c99638064565cf064ecff55bf536a2d82", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "383740f16d56b4f6384b1529", "attempt_id": "52ca69c10e6c", "run_group_id": "E4_core:A5|closed|qwen3b-llama3b:h8", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T06:37:09.535613Z", "task_family": "synthetic", "document_id": "syn_03000_2000", "document_hash": "c9cc27c20b05b1f6e592846e84990ee9b69c42f9720a8e338ec2f6fca777fdbc", "question_id": "q10_temporal", "split": "test", "cluster_id": "syn_03000_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "tokenizer_hashes": {"receiver": "b9d8872cf3a99da99e1fc4b0f71232a451632dc49ca4c6771549b4efa5c1c763"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 8, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1715, "query_tokens": 499, "message_tokens": 379, "message_utf8_bytes": 1042, "cache_tensor_bytes": null, "serialized_payload_bytes": 1042, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 138193.43433191534, "hops": {"1": 22495.10891607497, "2": 14903.01799995359, "3": 18322.75574991945, "4": 15116.547915968113, "5": 18017.787417047657, "6": 15147.36470906064, "7": 18164.166332921013, "8": 16026.685290969908}, "receiver_prefill_ms": 363.3734170580283, "ttft_ms": 18.25908396858722, "decode_ms": 21.41633303835988}, "total_wall_ms": 138597.1078738803, "ttft_ms": 18.25908396858722, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 15, "retention_rate": 0.23076923076923078, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.09090909090909091, "amount": 0.16666666666666666, "status": 0.5, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1042}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 15, "retention_rate": 0.23076923076923078, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.09090909090909091, "amount": 0.16666666666666666, "status": 0.5, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 855}, "2": {"facts_total": 65, "facts_retained": 15, "retention_rate": 0.23076923076923078, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.09090909090909091, "amount": 0.16666666666666666, "status": 0.5, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1042}, "3": {"facts_total": 65, "facts_retained": 15, "retention_rate": 0.23076923076923078, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.09090909090909091, "amount": 0.16666666666666666, "status": 0.5, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1042}, "4": {"facts_total": 65, "facts_retained": 15, "retention_rate": 0.23076923076923078, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.09090909090909091, "amount": 0.16666666666666666, "status": 0.5, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1042}, "5": {"facts_total": 65, "facts_retained": 15, "retention_rate": 0.23076923076923078, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.09090909090909091, "amount": 0.16666666666666666, "status": 0.5, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1042}, "6": {"facts_total": 65, "facts_retained": 15, "retention_rate": 0.23076923076923078, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.09090909090909091, "amount": 0.16666666666666666, "status": 0.5, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1042}, "7": {"facts_total": 65, "facts_retained": 15, "retention_rate": 0.23076923076923078, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.09090909090909091, "amount": 0.16666666666666666, "status": 0.5, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1042}, "8": {"facts_total": 65, "facts_retained": 15, "retention_rate": 0.23076923076923078, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.09090909090909091, "amount": 0.16666666666666666, "status": 0.5, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1042}}}, "reference_key_id": "keys/synthetic/syn_03000_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "e02c931be6afeca96f03fda7275aec7c99638064565cf064ecff55bf536a2d82", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "4e5b3eaeaa7db7907508e72a", "attempt_id": "39ca7f8e0e8c", "run_group_id": "E4_core:A5|closed|qwen3b-llama3b:h8", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T06:37:09.956309Z", "task_family": "synthetic", "document_id": "syn_03000_2000", "document_hash": "c9cc27c20b05b1f6e592846e84990ee9b69c42f9720a8e338ec2f6fca777fdbc", "question_id": "q11_aggregation", "split": "test", "cluster_id": "syn_03000_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "tokenizer_hashes": {"receiver": "b9d8872cf3a99da99e1fc4b0f71232a451632dc49ca4c6771549b4efa5c1c763"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 8, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1715, "query_tokens": 494, "message_tokens": 379, "message_utf8_bytes": 1042, "cache_tensor_bytes": null, "serialized_payload_bytes": 1042, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 138193.43433191534, "hops": {"1": 22495.10891607497, "2": 14903.01799995359, "3": 18322.75574991945, "4": 15116.547915968113, "5": 18017.787417047657, "6": 15147.36470906064, "7": 18164.166332921013, "8": 16026.685290969908}, "receiver_prefill_ms": 347.8399580344558, "ttft_ms": 45.09595897980034, "decode_ms": 62.21029092557728}, "total_wall_ms": 138649.3017069297, "ttft_ms": 45.09595897980034, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 15, "retention_rate": 0.23076923076923078, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.09090909090909091, "amount": 0.16666666666666666, "status": 0.5, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1042}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 15, "retention_rate": 0.23076923076923078, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.09090909090909091, "amount": 0.16666666666666666, "status": 0.5, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 855}, "2": {"facts_total": 65, "facts_retained": 15, "retention_rate": 0.23076923076923078, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.09090909090909091, "amount": 0.16666666666666666, "status": 0.5, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1042}, "3": {"facts_total": 65, "facts_retained": 15, "retention_rate": 0.23076923076923078, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.09090909090909091, "amount": 0.16666666666666666, "status": 0.5, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1042}, "4": {"facts_total": 65, "facts_retained": 15, "retention_rate": 0.23076923076923078, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.09090909090909091, "amount": 0.16666666666666666, "status": 0.5, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1042}, "5": {"facts_total": 65, "facts_retained": 15, "retention_rate": 0.23076923076923078, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.09090909090909091, "amount": 0.16666666666666666, "status": 0.5, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1042}, "6": {"facts_total": 65, "facts_retained": 15, "retention_rate": 0.23076923076923078, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.09090909090909091, "amount": 0.16666666666666666, "status": 0.5, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1042}, "7": {"facts_total": 65, "facts_retained": 15, "retention_rate": 0.23076923076923078, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.09090909090909091, "amount": 0.16666666666666666, "status": 0.5, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1042}, "8": {"facts_total": 65, "facts_retained": 15, "retention_rate": 0.23076923076923078, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.09090909090909091, "amount": 0.16666666666666666, "status": 0.5, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1042}}}, "reference_key_id": "keys/synthetic/syn_03000_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "e02c931be6afeca96f03fda7275aec7c99638064565cf064ecff55bf536a2d82", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "2bea9272aa2d8ef926077c44", "attempt_id": "86d23f2a27e8", "run_group_id": "E4_core:A5|closed|qwen3b-llama3b:h8", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T06:37:10.431597Z", "task_family": "synthetic", "document_id": "syn_03000_2000", "document_hash": "c9cc27c20b05b1f6e592846e84990ee9b69c42f9720a8e338ec2f6fca777fdbc", "question_id": "q12_unit_conversion", "split": "test", "cluster_id": "syn_03000_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "tokenizer_hashes": {"receiver": "b9d8872cf3a99da99e1fc4b0f71232a451632dc49ca4c6771549b4efa5c1c763"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 8, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1715, "query_tokens": 490, "message_tokens": 379, "message_utf8_bytes": 1042, "cache_tensor_bytes": null, "serialized_payload_bytes": 1042, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 138193.43433191534, "hops": {"1": 22495.10891607497, "2": 14903.01799995359, "3": 18322.75574991945, "4": 15116.547915968113, "5": 18017.787417047657, "6": 15147.36470906064, "7": 18164.166332921013, "8": 16026.685290969908}, "receiver_prefill_ms": 473.1176249915734, "ttft_ms": 38.33408304490149, "decode_ms": 23.723041987977922}, "total_wall_ms": 138729.10033201333, "ttft_ms": 38.33408304490149, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 15, "retention_rate": 0.23076923076923078, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.09090909090909091, "amount": 0.16666666666666666, "status": 0.5, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1042}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 15, "retention_rate": 0.23076923076923078, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.09090909090909091, "amount": 0.16666666666666666, "status": 0.5, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 855}, "2": {"facts_total": 65, "facts_retained": 15, "retention_rate": 0.23076923076923078, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.09090909090909091, "amount": 0.16666666666666666, "status": 0.5, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1042}, "3": {"facts_total": 65, "facts_retained": 15, "retention_rate": 0.23076923076923078, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.09090909090909091, "amount": 0.16666666666666666, "status": 0.5, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1042}, "4": {"facts_total": 65, "facts_retained": 15, "retention_rate": 0.23076923076923078, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.09090909090909091, "amount": 0.16666666666666666, "status": 0.5, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1042}, "5": {"facts_total": 65, "facts_retained": 15, "retention_rate": 0.23076923076923078, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.09090909090909091, "amount": 0.16666666666666666, "status": 0.5, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1042}, "6": {"facts_total": 65, "facts_retained": 15, "retention_rate": 0.23076923076923078, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.09090909090909091, "amount": 0.16666666666666666, "status": 0.5, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1042}, "7": {"facts_total": 65, "facts_retained": 15, "retention_rate": 0.23076923076923078, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.09090909090909091, "amount": 0.16666666666666666, "status": 0.5, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1042}, "8": {"facts_total": 65, "facts_retained": 15, "retention_rate": 0.23076923076923078, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.09090909090909091, "amount": 0.16666666666666666, "status": 0.5, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1042}}}, "reference_key_id": "keys/synthetic/syn_03000_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "e02c931be6afeca96f03fda7275aec7c99638064565cf064ecff55bf536a2d82", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "f841adfbf4d4f02b8683fa72", "attempt_id": "b96cd5d26878", "run_group_id": "E4_core:A5|closed|qwen3b-llama3b:h8", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T06:37:10.984595Z", "task_family": "synthetic", "document_id": "syn_03000_2000", "document_hash": "c9cc27c20b05b1f6e592846e84990ee9b69c42f9720a8e338ec2f6fca777fdbc", "question_id": "q13_unanswerable", "split": "test", "cluster_id": "syn_03000_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "tokenizer_hashes": {"receiver": "b9d8872cf3a99da99e1fc4b0f71232a451632dc49ca4c6771549b4efa5c1c763"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 8, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1715, "query_tokens": 493, "message_tokens": 379, "message_utf8_bytes": 1042, "cache_tensor_bytes": null, "serialized_payload_bytes": 1042, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 138193.43433191534, "hops": {"1": 22495.10891607497, "2": 14903.01799995359, "3": 18322.75574991945, "4": 15116.547915968113, "5": 18017.787417047657, "6": 15147.36470906064, "7": 18164.166332921013, "8": 16026.685290969908}, "receiver_prefill_ms": 371.91949994303286, "ttft_ms": 32.87970903329551, "decode_ms": 28.85395905468613}, "total_wall_ms": 138627.54345696885, "ttft_ms": 32.87970903329551, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 15, "retention_rate": 0.23076923076923078, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.09090909090909091, "amount": 0.16666666666666666, "status": 0.5, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1042}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 15, "retention_rate": 0.23076923076923078, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.09090909090909091, "amount": 0.16666666666666666, "status": 0.5, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 855}, "2": {"facts_total": 65, "facts_retained": 15, "retention_rate": 0.23076923076923078, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.09090909090909091, "amount": 0.16666666666666666, "status": 0.5, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1042}, "3": {"facts_total": 65, "facts_retained": 15, "retention_rate": 0.23076923076923078, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.09090909090909091, "amount": 0.16666666666666666, "status": 0.5, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1042}, "4": {"facts_total": 65, "facts_retained": 15, "retention_rate": 0.23076923076923078, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.09090909090909091, "amount": 0.16666666666666666, "status": 0.5, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1042}, "5": {"facts_total": 65, "facts_retained": 15, "retention_rate": 0.23076923076923078, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.09090909090909091, "amount": 0.16666666666666666, "status": 0.5, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1042}, "6": {"facts_total": 65, "facts_retained": 15, "retention_rate": 0.23076923076923078, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.09090909090909091, "amount": 0.16666666666666666, "status": 0.5, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1042}, "7": {"facts_total": 65, "facts_retained": 15, "retention_rate": 0.23076923076923078, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.09090909090909091, "amount": 0.16666666666666666, "status": 0.5, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1042}, "8": {"facts_total": 65, "facts_retained": 15, "retention_rate": 0.23076923076923078, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.09090909090909091, "amount": 0.16666666666666666, "status": 0.5, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1042}}}, "reference_key_id": "keys/synthetic/syn_03000_2000.json", "task_scores": {"correct": 1.0, "abstained": 1.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "e02c931be6afeca96f03fda7275aec7c99638064565cf064ecff55bf536a2d82", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "6b0a40e493c0377590bc9be3", "attempt_id": "22871c57cbb7", "run_group_id": "E4_core:A5|closed|qwen3b-llama3b:h8", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T06:37:11.435443Z", "task_family": "synthetic", "document_id": "syn_03000_2000", "document_hash": "c9cc27c20b05b1f6e592846e84990ee9b69c42f9720a8e338ec2f6fca777fdbc", "question_id": "q14_unanswerable", "split": "test", "cluster_id": "syn_03000_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "tokenizer_hashes": {"receiver": "b9d8872cf3a99da99e1fc4b0f71232a451632dc49ca4c6771549b4efa5c1c763"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 8, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1715, "query_tokens": 496, "message_tokens": 379, "message_utf8_bytes": 1042, "cache_tensor_bytes": null, "serialized_payload_bytes": 1042, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 138193.43433191534, "hops": {"1": 22495.10891607497, "2": 14903.01799995359, "3": 18322.75574991945, "4": 15116.547915968113, "5": 18017.787417047657, "6": 15147.36470906064, "7": 18164.166332921013, "8": 16026.685290969908}, "receiver_prefill_ms": 339.59562494419515, "ttft_ms": 37.68908302299678, "decode_ms": 35.70179198868573}, "total_wall_ms": 138607.28012386244, "ttft_ms": 37.68908302299678, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 15, "retention_rate": 0.23076923076923078, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.09090909090909091, "amount": 0.16666666666666666, "status": 0.5, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1042}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 15, "retention_rate": 0.23076923076923078, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.09090909090909091, "amount": 0.16666666666666666, "status": 0.5, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 855}, "2": {"facts_total": 65, "facts_retained": 15, "retention_rate": 0.23076923076923078, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.09090909090909091, "amount": 0.16666666666666666, "status": 0.5, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1042}, "3": {"facts_total": 65, "facts_retained": 15, "retention_rate": 0.23076923076923078, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.09090909090909091, "amount": 0.16666666666666666, "status": 0.5, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1042}, "4": {"facts_total": 65, "facts_retained": 15, "retention_rate": 0.23076923076923078, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.09090909090909091, "amount": 0.16666666666666666, "status": 0.5, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1042}, "5": {"facts_total": 65, "facts_retained": 15, "retention_rate": 0.23076923076923078, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.09090909090909091, "amount": 0.16666666666666666, "status": 0.5, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1042}, "6": {"facts_total": 65, "facts_retained": 15, "retention_rate": 0.23076923076923078, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.09090909090909091, "amount": 0.16666666666666666, "status": 0.5, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1042}, "7": {"facts_total": 65, "facts_retained": 15, "retention_rate": 0.23076923076923078, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.09090909090909091, "amount": 0.16666666666666666, "status": 0.5, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1042}, "8": {"facts_total": 65, "facts_retained": 15, "retention_rate": 0.23076923076923078, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.09090909090909091, "amount": 0.16666666666666666, "status": 0.5, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1042}}}, "reference_key_id": "keys/synthetic/syn_03000_2000.json", "task_scores": {"correct": 1.0, "abstained": 1.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "e02c931be6afeca96f03fda7275aec7c99638064565cf064ecff55bf536a2d82", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "520f81b0ec40ebdec5542c73", "attempt_id": "20af7d3b0ba0", "run_group_id": "E4_core:A4|closed|qwen3b:h1", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T06:37:34.851178Z", "task_family": "synthetic", "document_id": "syn_03003_2000", "document_hash": "dddca1a38e3224870fcfae927c701c400fd885c19123ea840ee06a76a140f407", "question_id": "q00_single_fact", "split": "test", "cluster_id": "syn_03003_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 1, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1866, "query_tokens": 594, "message_tokens": 512, "message_utf8_bytes": 1200, "cache_tensor_bytes": null, "serialized_payload_bytes": 1200, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 22906.635708990507, "hops": {"1": 22906.635708990507}, "receiver_prefill_ms": 406.78112499881536, "ttft_ms": 20.551208057440817, "decode_ms": 619.701666990295}, "total_wall_ms": 23954.510291921906, "ttft_ms": 20.551208057440817, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 10}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "GR-79S-5839", "evidence_artifact": {"ledger": {"facts_total": 58, "facts_retained": 32, "retention_rate": 0.5517241379310345, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.7142857142857143, "amount": 0.6666666666666666, "status": 0.6666666666666666, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1200}, "ledger_by_hop": {"1": {"facts_total": 58, "facts_retained": 32, "retention_rate": 0.5517241379310345, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.7142857142857143, "amount": 0.6666666666666666, "status": 0.6666666666666666, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1200}}}, "reference_key_id": "keys/synthetic/syn_03003_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "6d2a2bad9425191f4568f7eb115d63c953e27080a86953f62a982ad73340e770", "output_hash": "720cc40cd69e85e9728f6bea00b17a95ca07eee643e143a131840763958303cd"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "772bc1ee09163c788f7f462b", "attempt_id": "316690ba4563", "run_group_id": "E4_core:A4|closed|qwen3b:h1", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T06:37:35.918227Z", "task_family": "synthetic", "document_id": "syn_03003_2000", "document_hash": "dddca1a38e3224870fcfae927c701c400fd885c19123ea840ee06a76a140f407", "question_id": "q01_single_fact", "split": "test", "cluster_id": "syn_03003_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 1, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1866, "query_tokens": 598, "message_tokens": 512, "message_utf8_bytes": 1200, "cache_tensor_bytes": null, "serialized_payload_bytes": 1200, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 22906.635708990507, "hops": {"1": 22906.635708990507}, "receiver_prefill_ms": 390.48995811026543, "ttft_ms": 45.547665911726654, "decode_ms": 361.899874988012}, "total_wall_ms": 23705.20516706165, "ttft_ms": 45.547665911726654, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 11}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "Lot 344-B, Témiscaming", "evidence_artifact": {"ledger": {"facts_total": 58, "facts_retained": 32, "retention_rate": 0.5517241379310345, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.7142857142857143, "amount": 0.6666666666666666, "status": 0.6666666666666666, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1200}, "ledger_by_hop": {"1": {"facts_total": 58, "facts_retained": 32, "retention_rate": 0.5517241379310345, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.7142857142857143, "amount": 0.6666666666666666, "status": 0.6666666666666666, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1200}}}, "reference_key_id": "keys/synthetic/syn_03003_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 0.8571428571428571}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "6d2a2bad9425191f4568f7eb115d63c953e27080a86953f62a982ad73340e770", "output_hash": "0197d491a5a04a142dc8e372360735a1d41757e7a5aea7e81f23b0e1ef6ee010"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "220fe241699f2da02a3390fe", "attempt_id": "2f75c4843682", "run_group_id": "E4_core:A4|closed|qwen3b:h1", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T06:37:36.734654Z", "task_family": "synthetic", "document_id": "syn_03003_2000", "document_hash": "dddca1a38e3224870fcfae927c701c400fd885c19123ea840ee06a76a140f407", "question_id": "q02_exact_copy", "split": "test", "cluster_id": "syn_03003_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 1, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1866, "query_tokens": 598, "message_tokens": 512, "message_utf8_bytes": 1200, "cache_tensor_bytes": null, "serialized_payload_bytes": 1200, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 22906.635708990507, "hops": {"1": 22906.635708990507}, "receiver_prefill_ms": 358.6242089513689, "ttft_ms": 46.992916031740606, "decode_ms": 53.518875036388636}, "total_wall_ms": 23366.645583999343, "ttft_ms": 46.992916031740606, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 58, "facts_retained": 32, "retention_rate": 0.5517241379310345, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.7142857142857143, "amount": 0.6666666666666666, "status": 0.6666666666666666, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1200}, "ledger_by_hop": {"1": {"facts_total": 58, "facts_retained": 32, "retention_rate": 0.5517241379310345, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.7142857142857143, "amount": 0.6666666666666666, "status": 0.6666666666666666, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1200}}}, "reference_key_id": "keys/synthetic/syn_03003_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "6d2a2bad9425191f4568f7eb115d63c953e27080a86953f62a982ad73340e770", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "bcc1831770971d5270f6d6e2", "attempt_id": "9bb2a18920dd", "run_group_id": "E4_core:A4|closed|qwen3b:h1", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T06:37:37.212645Z", "task_family": "synthetic", "document_id": "syn_03003_2000", "document_hash": "dddca1a38e3224870fcfae927c701c400fd885c19123ea840ee06a76a140f407", "question_id": "q03_exact_copy", "split": "test", "cluster_id": "syn_03003_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 1, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1866, "query_tokens": 593, "message_tokens": 512, "message_utf8_bytes": 1200, "cache_tensor_bytes": null, "serialized_payload_bytes": 1200, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 22906.635708990507, "hops": {"1": 22906.635708990507}, "receiver_prefill_ms": 518.80991703365, "ttft_ms": 41.17549990769476, "decode_ms": 205.51562507171184}, "total_wall_ms": 23672.78650100343, "ttft_ms": 41.17549990769476, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 6}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "Wilhelmina Achterberg", "evidence_artifact": {"ledger": {"facts_total": 58, "facts_retained": 32, "retention_rate": 0.5517241379310345, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.7142857142857143, "amount": 0.6666666666666666, "status": 0.6666666666666666, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1200}, "ledger_by_hop": {"1": {"facts_total": 58, "facts_retained": 32, "retention_rate": 0.5517241379310345, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.7142857142857143, "amount": 0.6666666666666666, "status": 0.6666666666666666, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1200}}}, "reference_key_id": "keys/synthetic/syn_03003_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "6d2a2bad9425191f4568f7eb115d63c953e27080a86953f62a982ad73340e770", "output_hash": "2a62606853c315bba8769371d195e3ab3d6a8cd7ef959205575c5119740850f8"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "38e426832be9e7a17456ff81", "attempt_id": "be8fbcd30307", "run_group_id": "E4_core:A4|closed|qwen3b:h1", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T06:37:37.996600Z", "task_family": "synthetic", "document_id": "syn_03003_2000", "document_hash": "dddca1a38e3224870fcfae927c701c400fd885c19123ea840ee06a76a140f407", "question_id": "q04_multi_hop", "split": "test", "cluster_id": "syn_03003_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 1, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1866, "query_tokens": 602, "message_tokens": 512, "message_utf8_bytes": 1200, "cache_tensor_bytes": null, "serialized_payload_bytes": 1200, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 22906.635708990507, "hops": {"1": 22906.635708990507}, "receiver_prefill_ms": 395.02287504728884, "ttft_ms": 22.922208067029715, "decode_ms": 38.12299994751811}, "total_wall_ms": 23363.384249969386, "ttft_ms": 22.922208067029715, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 58, "facts_retained": 32, "retention_rate": 0.5517241379310345, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.7142857142857143, "amount": 0.6666666666666666, "status": 0.6666666666666666, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1200}, "ledger_by_hop": {"1": {"facts_total": 58, "facts_retained": 32, "retention_rate": 0.5517241379310345, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.7142857142857143, "amount": 0.6666666666666666, "status": 0.6666666666666666, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1200}}}, "reference_key_id": "keys/synthetic/syn_03003_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "6d2a2bad9425191f4568f7eb115d63c953e27080a86953f62a982ad73340e770", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "757787f709be1a8291a0745e", "attempt_id": "5aad79e0693d", "run_group_id": "E4_core:A4|closed|qwen3b:h1", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T06:37:38.470905Z", "task_family": "synthetic", "document_id": "syn_03003_2000", "document_hash": "dddca1a38e3224870fcfae927c701c400fd885c19123ea840ee06a76a140f407", "question_id": "q05_multi_hop", "split": "test", "cluster_id": "syn_03003_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 1, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1866, "query_tokens": 604, "message_tokens": 512, "message_utf8_bytes": 1200, "cache_tensor_bytes": null, "serialized_payload_bytes": 1200, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 22906.635708990507, "hops": {"1": 22906.635708990507}, "receiver_prefill_ms": 392.7377089858055, "ttft_ms": 38.2033750647679, "decode_ms": 557.5709579279646}, "total_wall_ms": 23895.87504195515, "ttft_ms": 38.2033750647679, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 10}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "NK-95Z-2165", "evidence_artifact": {"ledger": {"facts_total": 58, "facts_retained": 32, "retention_rate": 0.5517241379310345, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.7142857142857143, "amount": 0.6666666666666666, "status": 0.6666666666666666, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1200}, "ledger_by_hop": {"1": {"facts_total": 58, "facts_retained": 32, "retention_rate": 0.5517241379310345, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.7142857142857143, "amount": 0.6666666666666666, "status": 0.6666666666666666, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1200}}}, "reference_key_id": "keys/synthetic/syn_03003_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "6d2a2bad9425191f4568f7eb115d63c953e27080a86953f62a982ad73340e770", "output_hash": "bd8f809b02304c6b9dc7c1e116ecccc2328c976a5c071e89a94d661b48e73aa7"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "d90045c9e2aee154c2624cf8", "attempt_id": "2761cdf9da8e", "run_group_id": "E4_core:A4|closed|qwen3b:h1", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T06:37:39.477876Z", "task_family": "synthetic", "document_id": "syn_03003_2000", "document_hash": "dddca1a38e3224870fcfae927c701c400fd885c19123ea840ee06a76a140f407", "question_id": "q06_negation", "split": "test", "cluster_id": "syn_03003_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 1, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1866, "query_tokens": 596, "message_tokens": 512, "message_utf8_bytes": 1200, "cache_tensor_bytes": null, "serialized_payload_bytes": 1200, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 22906.635708990507, "hops": {"1": 22906.635708990507}, "receiver_prefill_ms": 404.19283392839134, "ttft_ms": 36.35045897681266, "decode_ms": 25.60066699516028}, "total_wall_ms": 23373.367667896673, "ttft_ms": 36.35045897681266, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "yes", "evidence_artifact": {"ledger": {"facts_total": 58, "facts_retained": 32, "retention_rate": 0.5517241379310345, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.7142857142857143, "amount": 0.6666666666666666, "status": 0.6666666666666666, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1200}, "ledger_by_hop": {"1": {"facts_total": 58, "facts_retained": 32, "retention_rate": 0.5517241379310345, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.7142857142857143, "amount": 0.6666666666666666, "status": 0.6666666666666666, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1200}}}, "reference_key_id": "keys/synthetic/syn_03003_2000.json", "task_scores": {"correct": 0.0, "abstained": 0.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "6d2a2bad9425191f4568f7eb115d63c953e27080a86953f62a982ad73340e770", "output_hash": "8a798890fe93817163b10b5f7bd2ca4d25d84c52739a645a889c173eee7d9d3d"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "e02e59a18863bdb4e867ed45", "attempt_id": "af92169979d5", "run_group_id": "E4_core:A4|closed|qwen3b:h1", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T06:37:39.962243Z", "task_family": "synthetic", "document_id": "syn_03003_2000", "document_hash": "dddca1a38e3224870fcfae927c701c400fd885c19123ea840ee06a76a140f407", "question_id": "q07_negation", "split": "test", "cluster_id": "syn_03003_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 1, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1866, "query_tokens": 593, "message_tokens": 512, "message_utf8_bytes": 1200, "cache_tensor_bytes": null, "serialized_payload_bytes": 1200, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 22906.635708990507, "hops": {"1": 22906.635708990507}, "receiver_prefill_ms": 362.7725000260398, "ttft_ms": 49.232083023525774, "decode_ms": 57.65816604252905}, "total_wall_ms": 23376.971833989955, "ttft_ms": 49.232083023525774, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "no", "evidence_artifact": {"ledger": {"facts_total": 58, "facts_retained": 32, "retention_rate": 0.5517241379310345, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.7142857142857143, "amount": 0.6666666666666666, "status": 0.6666666666666666, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1200}, "ledger_by_hop": {"1": {"facts_total": 58, "facts_retained": 32, "retention_rate": 0.5517241379310345, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.7142857142857143, "amount": 0.6666666666666666, "status": 0.6666666666666666, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1200}}}, "reference_key_id": "keys/synthetic/syn_03003_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "6d2a2bad9425191f4568f7eb115d63c953e27080a86953f62a982ad73340e770", "output_hash": "9390298f3fb0c5b160498935d79cb139aef28e1c47358b4bbba61862b9c26e59"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "85ad56e87d20ea16f54135eb", "attempt_id": "d52ea5fd17f6", "run_group_id": "E4_core:A4|closed|qwen3b:h1", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T06:37:40.450089Z", "task_family": "synthetic", "document_id": "syn_03003_2000", "document_hash": "dddca1a38e3224870fcfae927c701c400fd885c19123ea840ee06a76a140f407", "question_id": "q08_revision", "split": "test", "cluster_id": "syn_03003_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 1, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1866, "query_tokens": 611, "message_tokens": 512, "message_utf8_bytes": 1200, "cache_tensor_bytes": null, "serialized_payload_bytes": 1200, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 22906.635708990507, "hops": {"1": 22906.635708990507}, "receiver_prefill_ms": 545.4107919940725, "ttft_ms": 33.37049996480346, "decode_ms": 223.45704096369445}, "total_wall_ms": 23709.619543049484, "ttft_ms": 33.37049996480346, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 7}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "911,925", "evidence_artifact": {"ledger": {"facts_total": 58, "facts_retained": 32, "retention_rate": 0.5517241379310345, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.7142857142857143, "amount": 0.6666666666666666, "status": 0.6666666666666666, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1200}, "ledger_by_hop": {"1": {"facts_total": 58, "facts_retained": 32, "retention_rate": 0.5517241379310345, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.7142857142857143, "amount": 0.6666666666666666, "status": 0.6666666666666666, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1200}}}, "reference_key_id": "keys/synthetic/syn_03003_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "6d2a2bad9425191f4568f7eb115d63c953e27080a86953f62a982ad73340e770", "output_hash": "86be78720fc4b79f93e603e7d9e4861fcd25102612f73209fffc17d787056007"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "1918611a9d8a68457013d9a6", "attempt_id": "31731a85e29b", "run_group_id": "E4_core:A4|closed|qwen3b:h1", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T06:37:41.270491Z", "task_family": "synthetic", "document_id": "syn_03003_2000", "document_hash": "dddca1a38e3224870fcfae927c701c400fd885c19123ea840ee06a76a140f407", "question_id": "q09_revision", "split": "test", "cluster_id": "syn_03003_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 1, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1866, "query_tokens": 604, "message_tokens": 512, "message_utf8_bytes": 1200, "cache_tensor_bytes": null, "serialized_payload_bytes": 1200, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 22906.635708990507, "hops": {"1": 22906.635708990507}, "receiver_prefill_ms": 406.2775420024991, "ttft_ms": 39.247334003448486, "decode_ms": 230.95075006131083}, "total_wall_ms": 23584.138125996105, "ttft_ms": 39.247334003448486, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 7}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "911,925", "evidence_artifact": {"ledger": {"facts_total": 58, "facts_retained": 32, "retention_rate": 0.5517241379310345, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.7142857142857143, "amount": 0.6666666666666666, "status": 0.6666666666666666, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1200}, "ledger_by_hop": {"1": {"facts_total": 58, "facts_retained": 32, "retention_rate": 0.5517241379310345, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.7142857142857143, "amount": 0.6666666666666666, "status": 0.6666666666666666, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1200}}}, "reference_key_id": "keys/synthetic/syn_03003_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "6d2a2bad9425191f4568f7eb115d63c953e27080a86953f62a982ad73340e770", "output_hash": "86be78720fc4b79f93e603e7d9e4861fcd25102612f73209fffc17d787056007"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "23f998250e20090472a20284", "attempt_id": "10e233f8aa69", "run_group_id": "E4_core:A4|closed|qwen3b:h1", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T06:37:41.965478Z", "task_family": "synthetic", "document_id": "syn_03003_2000", "document_hash": "dddca1a38e3224870fcfae927c701c400fd885c19123ea840ee06a76a140f407", "question_id": "q10_temporal", "split": "test", "cluster_id": "syn_03003_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 1, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1866, "query_tokens": 611, "message_tokens": 512, "message_utf8_bytes": 1200, "cache_tensor_bytes": null, "serialized_payload_bytes": 1200, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 22906.635708990507, "hops": {"1": 22906.635708990507}, "receiver_prefill_ms": 367.16824991162866, "ttft_ms": 95.88683396577835, "decode_ms": 304.62875007651746}, "total_wall_ms": 23675.17233407125, "ttft_ms": 95.88683396577835, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 6}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "Gamtiszen-2", "evidence_artifact": {"ledger": {"facts_total": 58, "facts_retained": 32, "retention_rate": 0.5517241379310345, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.7142857142857143, "amount": 0.6666666666666666, "status": 0.6666666666666666, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1200}, "ledger_by_hop": {"1": {"facts_total": 58, "facts_retained": 32, "retention_rate": 0.5517241379310345, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.7142857142857143, "amount": 0.6666666666666666, "status": 0.6666666666666666, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1200}}}, "reference_key_id": "keys/synthetic/syn_03003_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "6d2a2bad9425191f4568f7eb115d63c953e27080a86953f62a982ad73340e770", "output_hash": "32e3af428886434823d82102e1c482765eafdf6077fe8d3fba5752ec5e29826c"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "46881f1b6399940515f08f08", "attempt_id": "05866fdbe648", "run_group_id": "E4_core:A4|closed|qwen3b:h1", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T06:37:42.751498Z", "task_family": "synthetic", "document_id": "syn_03003_2000", "document_hash": "dddca1a38e3224870fcfae927c701c400fd885c19123ea840ee06a76a140f407", "question_id": "q11_aggregation", "split": "test", "cluster_id": "syn_03003_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 1, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1866, "query_tokens": 605, "message_tokens": 512, "message_utf8_bytes": 1200, "cache_tensor_bytes": null, "serialized_payload_bytes": 1200, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 22906.635708990507, "hops": {"1": 22906.635708990507}, "receiver_prefill_ms": 402.286708005704, "ttft_ms": 29.896666994318366, "decode_ms": 230.23345903493464}, "total_wall_ms": 23569.74970898591, "ttft_ms": 29.896666994318366, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 7}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "3397775", "evidence_artifact": {"ledger": {"facts_total": 58, "facts_retained": 32, "retention_rate": 0.5517241379310345, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.7142857142857143, "amount": 0.6666666666666666, "status": 0.6666666666666666, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1200}, "ledger_by_hop": {"1": {"facts_total": 58, "facts_retained": 32, "retention_rate": 0.5517241379310345, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.7142857142857143, "amount": 0.6666666666666666, "status": 0.6666666666666666, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1200}}}, "reference_key_id": "keys/synthetic/syn_03003_2000.json", "task_scores": {"correct": 0.0, "abstained": 0.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "6d2a2bad9425191f4568f7eb115d63c953e27080a86953f62a982ad73340e770", "output_hash": "e971a34f9c497d953f98fb387e189b1587b11a7f6725b2f81ac6d0f3ef8ac845"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "10ec7d7d07f62339ff8d2d4a", "attempt_id": "65caff9ff324", "run_group_id": "E4_core:A4|closed|qwen3b:h1", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T06:37:43.431604Z", "task_family": "synthetic", "document_id": "syn_03003_2000", "document_hash": "dddca1a38e3224870fcfae927c701c400fd885c19123ea840ee06a76a140f407", "question_id": "q12_unit_conversion", "split": "test", "cluster_id": "syn_03003_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 1, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1866, "query_tokens": 602, "message_tokens": 512, "message_utf8_bytes": 1200, "cache_tensor_bytes": null, "serialized_payload_bytes": 1200, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 22906.635708990507, "hops": {"1": 22906.635708990507}, "receiver_prefill_ms": 338.8707499252632, "ttft_ms": 67.08008307032287, "decode_ms": 390.6322499969974}, "total_wall_ms": 23703.90837593004, "ttft_ms": 67.08008307032287, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 5}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "17430", "evidence_artifact": {"ledger": {"facts_total": 58, "facts_retained": 32, "retention_rate": 0.5517241379310345, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.7142857142857143, "amount": 0.6666666666666666, "status": 0.6666666666666666, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1200}, "ledger_by_hop": {"1": {"facts_total": 58, "facts_retained": 32, "retention_rate": 0.5517241379310345, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.7142857142857143, "amount": 0.6666666666666666, "status": 0.6666666666666666, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1200}}}, "reference_key_id": "keys/synthetic/syn_03003_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "6d2a2bad9425191f4568f7eb115d63c953e27080a86953f62a982ad73340e770", "output_hash": "2575eaf3b748efc209b676518766751eb038c88f74578951345818bfe7f7194b"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "c6ab2f7402852b7c26d3787a", "attempt_id": "8d20a7904fb1", "run_group_id": "E4_core:A4|closed|qwen3b:h1", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T06:37:44.245690Z", "task_family": "synthetic", "document_id": "syn_03003_2000", "document_hash": "dddca1a38e3224870fcfae927c701c400fd885c19123ea840ee06a76a140f407", "question_id": "q13_unanswerable", "split": "test", "cluster_id": "syn_03003_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 1, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1866, "query_tokens": 606, "message_tokens": 512, "message_utf8_bytes": 1200, "cache_tensor_bytes": null, "serialized_payload_bytes": 1200, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 22906.635708990507, "hops": {"1": 22906.635708990507}, "receiver_prefill_ms": 407.6799580361694, "ttft_ms": 26.5637090196833, "decode_ms": 37.35754103399813}, "total_wall_ms": 23378.82774998434, "ttft_ms": 26.5637090196833, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 58, "facts_retained": 32, "retention_rate": 0.5517241379310345, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.7142857142857143, "amount": 0.6666666666666666, "status": 0.6666666666666666, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1200}, "ledger_by_hop": {"1": {"facts_total": 58, "facts_retained": 32, "retention_rate": 0.5517241379310345, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.7142857142857143, "amount": 0.6666666666666666, "status": 0.6666666666666666, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1200}}}, "reference_key_id": "keys/synthetic/syn_03003_2000.json", "task_scores": {"correct": 1.0, "abstained": 1.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "6d2a2bad9425191f4568f7eb115d63c953e27080a86953f62a982ad73340e770", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "227e7b504d02d19220fddfb9", "attempt_id": "8b01b65f6c0e", "run_group_id": "E4_core:A4|closed|qwen3b:h1", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T06:37:44.734844Z", "task_family": "synthetic", "document_id": "syn_03003_2000", "document_hash": "dddca1a38e3224870fcfae927c701c400fd885c19123ea840ee06a76a140f407", "question_id": "q14_unanswerable", "split": "test", "cluster_id": "syn_03003_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 1, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1866, "query_tokens": 607, "message_tokens": 512, "message_utf8_bytes": 1200, "cache_tensor_bytes": null, "serialized_payload_bytes": 1200, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 22906.635708990507, "hops": {"1": 22906.635708990507}, "receiver_prefill_ms": 396.2232079356909, "ttft_ms": 30.694041051901877, "decode_ms": 19.391666050069034}, "total_wall_ms": 23353.56408392545, "ttft_ms": 30.694041051901877, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 58, "facts_retained": 32, "retention_rate": 0.5517241379310345, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.7142857142857143, "amount": 0.6666666666666666, "status": 0.6666666666666666, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1200}, "ledger_by_hop": {"1": {"facts_total": 58, "facts_retained": 32, "retention_rate": 0.5517241379310345, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.7142857142857143, "amount": 0.6666666666666666, "status": 0.6666666666666666, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1200}}}, "reference_key_id": "keys/synthetic/syn_03003_2000.json", "task_scores": {"correct": 1.0, "abstained": 1.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "6d2a2bad9425191f4568f7eb115d63c953e27080a86953f62a982ad73340e770", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "998836e2c2e7be03e3c00ce8", "attempt_id": "22fbdfef107b", "run_group_id": "E4_core:A4|closed|qwen3b:h2", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T06:38:07.233424Z", "task_family": "synthetic", "document_id": "syn_03003_2000", "document_hash": "dddca1a38e3224870fcfae927c701c400fd885c19123ea840ee06a76a140f407", "question_id": "q00_single_fact", "split": "test", "cluster_id": "syn_03003_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 2, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1866, "query_tokens": 595, "message_tokens": 512, "message_utf8_bytes": 1193, "cache_tensor_bytes": null, "serialized_payload_bytes": 1193, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 44939.650501008146, "hops": {"1": 22906.635708990507, "2": 22033.01479201764}, "receiver_prefill_ms": 541.3750420557335, "ttft_ms": 22.709375014528632, "decode_ms": 315.1026249397546}, "total_wall_ms": 45819.597335066646, "ttft_ms": 22.709375014528632, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 10}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "GR-79S-5839", "evidence_artifact": {"ledger": {"facts_total": 58, "facts_retained": 32, "retention_rate": 0.5517241379310345, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.7142857142857143, "amount": 0.6666666666666666, "status": 0.6666666666666666, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1193}, "ledger_by_hop": {"1": {"facts_total": 58, "facts_retained": 32, "retention_rate": 0.5517241379310345, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.7142857142857143, "amount": 0.6666666666666666, "status": 0.6666666666666666, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1200}, "2": {"facts_total": 58, "facts_retained": 32, "retention_rate": 0.5517241379310345, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.7142857142857143, "amount": 0.6666666666666666, "status": 0.6666666666666666, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1193}}}, "reference_key_id": "keys/synthetic/syn_03003_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "72b629618498e6857b5e5c18413e492e46455d51f70d74a18ff4e73267a76b13", "output_hash": "720cc40cd69e85e9728f6bea00b17a95ca07eee643e143a131840763958303cd"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "0b551ee3b44d92a5b668f191", "attempt_id": "02545dbe2614", "run_group_id": "E4_core:A4|closed|qwen3b:h2", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T06:38:08.131851Z", "task_family": "synthetic", "document_id": "syn_03003_2000", "document_hash": "dddca1a38e3224870fcfae927c701c400fd885c19123ea840ee06a76a140f407", "question_id": "q01_single_fact", "split": "test", "cluster_id": "syn_03003_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 2, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1866, "query_tokens": 599, "message_tokens": 512, "message_utf8_bytes": 1193, "cache_tensor_bytes": null, "serialized_payload_bytes": 1193, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 44939.650501008146, "hops": {"1": 22906.635708990507, "2": 22033.01479201764}, "receiver_prefill_ms": 405.00833303667605, "ttft_ms": 23.492333944886923, "decode_ms": 645.575457951054}, "total_wall_ms": 46014.382792054676, "ttft_ms": 23.492333944886923, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 11}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "Lot 344-B, Témiscaming", "evidence_artifact": {"ledger": {"facts_total": 58, "facts_retained": 32, "retention_rate": 0.5517241379310345, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.7142857142857143, "amount": 0.6666666666666666, "status": 0.6666666666666666, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1193}, "ledger_by_hop": {"1": {"facts_total": 58, "facts_retained": 32, "retention_rate": 0.5517241379310345, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.7142857142857143, "amount": 0.6666666666666666, "status": 0.6666666666666666, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1200}, "2": {"facts_total": 58, "facts_retained": 32, "retention_rate": 0.5517241379310345, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.7142857142857143, "amount": 0.6666666666666666, "status": 0.6666666666666666, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1193}}}, "reference_key_id": "keys/synthetic/syn_03003_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 0.8571428571428571}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "72b629618498e6857b5e5c18413e492e46455d51f70d74a18ff4e73267a76b13", "output_hash": "0197d491a5a04a142dc8e372360735a1d41757e7a5aea7e81f23b0e1ef6ee010"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "b9acf2e4b9042da8b3c8b4c0", "attempt_id": "0958679cab43", "run_group_id": "E4_core:A4|closed|qwen3b:h2", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T06:38:09.224045Z", "task_family": "synthetic", "document_id": "syn_03003_2000", "document_hash": "dddca1a38e3224870fcfae927c701c400fd885c19123ea840ee06a76a140f407", "question_id": "q02_exact_copy", "split": "test", "cluster_id": "syn_03003_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 2, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1866, "query_tokens": 599, "message_tokens": 512, "message_utf8_bytes": 1193, "cache_tensor_bytes": null, "serialized_payload_bytes": 1193, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 44939.650501008146, "hops": {"1": 22906.635708990507, "2": 22033.01479201764}, "receiver_prefill_ms": 406.45875001791865, "ttft_ms": 32.672417000867426, "decode_ms": 29.58179102279246}, "total_wall_ms": 45408.929001074284, "ttft_ms": 32.672417000867426, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 58, "facts_retained": 32, "retention_rate": 0.5517241379310345, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.7142857142857143, "amount": 0.6666666666666666, "status": 0.6666666666666666, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1193}, "ledger_by_hop": {"1": {"facts_total": 58, "facts_retained": 32, "retention_rate": 0.5517241379310345, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.7142857142857143, "amount": 0.6666666666666666, "status": 0.6666666666666666, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1200}, "2": {"facts_total": 58, "facts_retained": 32, "retention_rate": 0.5517241379310345, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.7142857142857143, "amount": 0.6666666666666666, "status": 0.6666666666666666, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1193}}}, "reference_key_id": "keys/synthetic/syn_03003_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "72b629618498e6857b5e5c18413e492e46455d51f70d74a18ff4e73267a76b13", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "385fd6b3bb3ef659ec30c14d", "attempt_id": "194a2724cd53", "run_group_id": "E4_core:A4|closed|qwen3b:h2", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T06:38:09.710820Z", "task_family": "synthetic", "document_id": "syn_03003_2000", "document_hash": "dddca1a38e3224870fcfae927c701c400fd885c19123ea840ee06a76a140f407", "question_id": "q03_exact_copy", "split": "test", "cluster_id": "syn_03003_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 2, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1866, "query_tokens": 594, "message_tokens": 512, "message_utf8_bytes": 1193, "cache_tensor_bytes": null, "serialized_payload_bytes": 1193, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 44939.650501008146, "hops": {"1": 22906.635708990507, "2": 22033.01479201764}, "receiver_prefill_ms": 406.74112492706627, "ttft_ms": 23.73429189901799, "decode_ms": 287.00966690666974}, "total_wall_ms": 45657.90412609931, "ttft_ms": 23.73429189901799, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 6}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "Wilhelmina Achterberg", "evidence_artifact": {"ledger": {"facts_total": 58, "facts_retained": 32, "retention_rate": 0.5517241379310345, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.7142857142857143, "amount": 0.6666666666666666, "status": 0.6666666666666666, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1193}, "ledger_by_hop": {"1": {"facts_total": 58, "facts_retained": 32, "retention_rate": 0.5517241379310345, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.7142857142857143, "amount": 0.6666666666666666, "status": 0.6666666666666666, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1200}, "2": {"facts_total": 58, "facts_retained": 32, "retention_rate": 0.5517241379310345, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.7142857142857143, "amount": 0.6666666666666666, "status": 0.6666666666666666, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1193}}}, "reference_key_id": "keys/synthetic/syn_03003_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "72b629618498e6857b5e5c18413e492e46455d51f70d74a18ff4e73267a76b13", "output_hash": "2a62606853c315bba8769371d195e3ab3d6a8cd7ef959205575c5119740850f8"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "77cf3b6b6217cabc4b4fa292", "attempt_id": "0412ae2106e0", "run_group_id": "E4_core:A4|closed|qwen3b:h2", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T06:38:10.446101Z", "task_family": "synthetic", "document_id": "syn_03003_2000", "document_hash": "dddca1a38e3224870fcfae927c701c400fd885c19123ea840ee06a76a140f407", "question_id": "q04_multi_hop", "split": "test", "cluster_id": "syn_03003_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 2, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1866, "query_tokens": 603, "message_tokens": 512, "message_utf8_bytes": 1193, "cache_tensor_bytes": null, "serialized_payload_bytes": 1193, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 44939.650501008146, "hops": {"1": 22906.635708990507, "2": 22033.01479201764}, "receiver_prefill_ms": 519.4494579918683, "ttft_ms": 24.791583069600165, "decode_ms": 38.88349991757423}, "total_wall_ms": 45523.54725101031, "ttft_ms": 24.791583069600165, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 58, "facts_retained": 32, "retention_rate": 0.5517241379310345, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.7142857142857143, "amount": 0.6666666666666666, "status": 0.6666666666666666, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1193}, "ledger_by_hop": {"1": {"facts_total": 58, "facts_retained": 32, "retention_rate": 0.5517241379310345, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.7142857142857143, "amount": 0.6666666666666666, "status": 0.6666666666666666, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1200}, "2": {"facts_total": 58, "facts_retained": 32, "retention_rate": 0.5517241379310345, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.7142857142857143, "amount": 0.6666666666666666, "status": 0.6666666666666666, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1193}}}, "reference_key_id": "keys/synthetic/syn_03003_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "72b629618498e6857b5e5c18413e492e46455d51f70d74a18ff4e73267a76b13", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "ff4405db77d5d32782036121", "attempt_id": "13a7a495469c", "run_group_id": "E4_core:A4|closed|qwen3b:h2", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T06:38:11.047327Z", "task_family": "synthetic", "document_id": "syn_03003_2000", "document_hash": "dddca1a38e3224870fcfae927c701c400fd885c19123ea840ee06a76a140f407", "question_id": "q05_multi_hop", "split": "test", "cluster_id": "syn_03003_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 2, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1866, "query_tokens": 605, "message_tokens": 512, "message_utf8_bytes": 1193, "cache_tensor_bytes": null, "serialized_payload_bytes": 1193, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 44939.650501008146, "hops": {"1": 22906.635708990507, "2": 22033.01479201764}, "receiver_prefill_ms": 421.49445903487504, "ttft_ms": 43.08716696687043, "decode_ms": 329.2437921045348}, "total_wall_ms": 45734.13000092842, "ttft_ms": 43.08716696687043, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 10}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "NK-95Z-2165", "evidence_artifact": {"ledger": {"facts_total": 58, "facts_retained": 32, "retention_rate": 0.5517241379310345, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.7142857142857143, "amount": 0.6666666666666666, "status": 0.6666666666666666, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1193}, "ledger_by_hop": {"1": {"facts_total": 58, "facts_retained": 32, "retention_rate": 0.5517241379310345, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.7142857142857143, "amount": 0.6666666666666666, "status": 0.6666666666666666, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1200}, "2": {"facts_total": 58, "facts_retained": 32, "retention_rate": 0.5517241379310345, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.7142857142857143, "amount": 0.6666666666666666, "status": 0.6666666666666666, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1193}}}, "reference_key_id": "keys/synthetic/syn_03003_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "72b629618498e6857b5e5c18413e492e46455d51f70d74a18ff4e73267a76b13", "output_hash": "bd8f809b02304c6b9dc7c1e116ecccc2328c976a5c071e89a94d661b48e73aa7"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "301b0d6c34e8972611af457e", "attempt_id": "e9ca48e11e30", "run_group_id": "E4_core:A4|closed|qwen3b:h2", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T06:38:11.859562Z", "task_family": "synthetic", "document_id": "syn_03003_2000", "document_hash": "dddca1a38e3224870fcfae927c701c400fd885c19123ea840ee06a76a140f407", "question_id": "q06_negation", "split": "test", "cluster_id": "syn_03003_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 2, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1866, "query_tokens": 597, "message_tokens": 512, "message_utf8_bytes": 1193, "cache_tensor_bytes": null, "serialized_payload_bytes": 1193, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 44939.650501008146, "hops": {"1": 22906.635708990507, "2": 22033.01479201764}, "receiver_prefill_ms": 467.5518749281764, "ttft_ms": 144.66441702097654, "decode_ms": 41.336792055517435}, "total_wall_ms": 45593.930584029295, "ttft_ms": 144.66441702097654, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "yes", "evidence_artifact": {"ledger": {"facts_total": 58, "facts_retained": 32, "retention_rate": 0.5517241379310345, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.7142857142857143, "amount": 0.6666666666666666, "status": 0.6666666666666666, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1193}, "ledger_by_hop": {"1": {"facts_total": 58, "facts_retained": 32, "retention_rate": 0.5517241379310345, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.7142857142857143, "amount": 0.6666666666666666, "status": 0.6666666666666666, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1200}, "2": {"facts_total": 58, "facts_retained": 32, "retention_rate": 0.5517241379310345, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.7142857142857143, "amount": 0.6666666666666666, "status": 0.6666666666666666, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1193}}}, "reference_key_id": "keys/synthetic/syn_03003_2000.json", "task_scores": {"correct": 0.0, "abstained": 0.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "72b629618498e6857b5e5c18413e492e46455d51f70d74a18ff4e73267a76b13", "output_hash": "8a798890fe93817163b10b5f7bd2ca4d25d84c52739a645a889c173eee7d9d3d"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "033a075fe4ea7fab1aa2ee0d", "attempt_id": "485b9b62507c", "run_group_id": "E4_core:A4|closed|qwen3b:h2", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T06:38:12.532052Z", "task_family": "synthetic", "document_id": "syn_03003_2000", "document_hash": "dddca1a38e3224870fcfae927c701c400fd885c19123ea840ee06a76a140f407", "question_id": "q07_negation", "split": "test", "cluster_id": "syn_03003_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 2, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1866, "query_tokens": 594, "message_tokens": 512, "message_utf8_bytes": 1193, "cache_tensor_bytes": null, "serialized_payload_bytes": 1193, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 44939.650501008146, "hops": {"1": 22906.635708990507, "2": 22033.01479201764}, "receiver_prefill_ms": 408.9601249434054, "ttft_ms": 21.093624993227422, "decode_ms": 44.17387500870973}, "total_wall_ms": 45414.4747100072, "ttft_ms": 21.093624993227422, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "no", "evidence_artifact": {"ledger": {"facts_total": 58, "facts_retained": 32, "retention_rate": 0.5517241379310345, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.7142857142857143, "amount": 0.6666666666666666, "status": 0.6666666666666666, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1193}, "ledger_by_hop": {"1": {"facts_total": 58, "facts_retained": 32, "retention_rate": 0.5517241379310345, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.7142857142857143, "amount": 0.6666666666666666, "status": 0.6666666666666666, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1200}, "2": {"facts_total": 58, "facts_retained": 32, "retention_rate": 0.5517241379310345, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.7142857142857143, "amount": 0.6666666666666666, "status": 0.6666666666666666, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1193}}}, "reference_key_id": "keys/synthetic/syn_03003_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "72b629618498e6857b5e5c18413e492e46455d51f70d74a18ff4e73267a76b13", "output_hash": "9390298f3fb0c5b160498935d79cb139aef28e1c47358b4bbba61862b9c26e59"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "d2c3f7a380de1006da5f7192", "attempt_id": "ff9f18d73fab", "run_group_id": "E4_core:A4|closed|qwen3b:h2", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T06:38:13.023607Z", "task_family": "synthetic", "document_id": "syn_03003_2000", "document_hash": "dddca1a38e3224870fcfae927c701c400fd885c19123ea840ee06a76a140f407", "question_id": "q08_revision", "split": "test", "cluster_id": "syn_03003_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 2, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1866, "query_tokens": 612, "message_tokens": 512, "message_utf8_bytes": 1193, "cache_tensor_bytes": null, "serialized_payload_bytes": 1193, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 44939.650501008146, "hops": {"1": 22906.635708990507, "2": 22033.01479201764}, "receiver_prefill_ms": 405.0626669777557, "ttft_ms": 21.253666956909, "decode_ms": 316.88937498256564}, "total_wall_ms": 45683.69408498984, "ttft_ms": 21.253666956909, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 7}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "911,925", "evidence_artifact": {"ledger": {"facts_total": 58, "facts_retained": 32, "retention_rate": 0.5517241379310345, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.7142857142857143, "amount": 0.6666666666666666, "status": 0.6666666666666666, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1193}, "ledger_by_hop": {"1": {"facts_total": 58, "facts_retained": 32, "retention_rate": 0.5517241379310345, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.7142857142857143, "amount": 0.6666666666666666, "status": 0.6666666666666666, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1200}, "2": {"facts_total": 58, "facts_retained": 32, "retention_rate": 0.5517241379310345, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.7142857142857143, "amount": 0.6666666666666666, "status": 0.6666666666666666, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1193}}}, "reference_key_id": "keys/synthetic/syn_03003_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "72b629618498e6857b5e5c18413e492e46455d51f70d74a18ff4e73267a76b13", "output_hash": "86be78720fc4b79f93e603e7d9e4861fcd25102612f73209fffc17d787056007"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "f624e1eb6803d73d2d245abd", "attempt_id": "03dafbfe0e4c", "run_group_id": "E4_core:A4|closed|qwen3b:h2", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T06:38:13.784776Z", "task_family": "synthetic", "document_id": "syn_03003_2000", "document_hash": "dddca1a38e3224870fcfae927c701c400fd885c19123ea840ee06a76a140f407", "question_id": "q09_revision", "split": "test", "cluster_id": "syn_03003_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 2, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1866, "query_tokens": 605, "message_tokens": 512, "message_utf8_bytes": 1193, "cache_tensor_bytes": null, "serialized_payload_bytes": 1193, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 44939.650501008146, "hops": {"1": 22906.635708990507, "2": 22033.01479201764}, "receiver_prefill_ms": 516.4762090425938, "ttft_ms": 39.635583059862256, "decode_ms": 213.866458972916}, "total_wall_ms": 45710.30154300388, "ttft_ms": 39.635583059862256, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 7}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "911,925", "evidence_artifact": {"ledger": {"facts_total": 58, "facts_retained": 32, "retention_rate": 0.5517241379310345, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.7142857142857143, "amount": 0.6666666666666666, "status": 0.6666666666666666, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1193}, "ledger_by_hop": {"1": {"facts_total": 58, "facts_retained": 32, "retention_rate": 0.5517241379310345, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.7142857142857143, "amount": 0.6666666666666666, "status": 0.6666666666666666, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1200}, "2": {"facts_total": 58, "facts_retained": 32, "retention_rate": 0.5517241379310345, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.7142857142857143, "amount": 0.6666666666666666, "status": 0.6666666666666666, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1193}}}, "reference_key_id": "keys/synthetic/syn_03003_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "72b629618498e6857b5e5c18413e492e46455d51f70d74a18ff4e73267a76b13", "output_hash": "86be78720fc4b79f93e603e7d9e4861fcd25102612f73209fffc17d787056007"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "f3496d9de3ea08b83d575918", "attempt_id": "a1bfe8f4d5af", "run_group_id": "E4_core:A4|closed|qwen3b:h2", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T06:38:14.573246Z", "task_family": "synthetic", "document_id": "syn_03003_2000", "document_hash": "dddca1a38e3224870fcfae927c701c400fd885c19123ea840ee06a76a140f407", "question_id": "q10_temporal", "split": "test", "cluster_id": "syn_03003_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 2, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1866, "query_tokens": 612, "message_tokens": 512, "message_utf8_bytes": 1193, "cache_tensor_bytes": null, "serialized_payload_bytes": 1193, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 44939.650501008146, "hops": {"1": 22906.635708990507, "2": 22033.01479201764}, "receiver_prefill_ms": 393.0339169455692, "ttft_ms": 23.18462496623397, "decode_ms": 205.557583947666}, "total_wall_ms": 45562.35979299527, "ttft_ms": 23.18462496623397, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 6}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "Gamtiszen-2", "evidence_artifact": {"ledger": {"facts_total": 58, "facts_retained": 32, "retention_rate": 0.5517241379310345, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.7142857142857143, "amount": 0.6666666666666666, "status": 0.6666666666666666, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1193}, "ledger_by_hop": {"1": {"facts_total": 58, "facts_retained": 32, "retention_rate": 0.5517241379310345, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.7142857142857143, "amount": 0.6666666666666666, "status": 0.6666666666666666, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1200}, "2": {"facts_total": 58, "facts_retained": 32, "retention_rate": 0.5517241379310345, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.7142857142857143, "amount": 0.6666666666666666, "status": 0.6666666666666666, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1193}}}, "reference_key_id": "keys/synthetic/syn_03003_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "72b629618498e6857b5e5c18413e492e46455d51f70d74a18ff4e73267a76b13", "output_hash": "32e3af428886434823d82102e1c482765eafdf6077fe8d3fba5752ec5e29826c"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "1663a57e02243473df012d02", "attempt_id": "049885dcfe2d", "run_group_id": "E4_core:A4|closed|qwen3b:h2", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T06:38:15.215141Z", "task_family": "synthetic", "document_id": "syn_03003_2000", "document_hash": "dddca1a38e3224870fcfae927c701c400fd885c19123ea840ee06a76a140f407", "question_id": "q11_aggregation", "split": "test", "cluster_id": "syn_03003_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 2, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1866, "query_tokens": 606, "message_tokens": 512, "message_utf8_bytes": 1193, "cache_tensor_bytes": null, "serialized_payload_bytes": 1193, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 44939.650501008146, "hops": {"1": 22906.635708990507, "2": 22033.01479201764}, "receiver_prefill_ms": 401.0928329080343, "ttft_ms": 73.83966608904302, "decode_ms": 327.12975004687905}, "total_wall_ms": 45742.529667913914, "ttft_ms": 73.83966608904302, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 7}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "3397775", "evidence_artifact": {"ledger": {"facts_total": 58, "facts_retained": 32, "retention_rate": 0.5517241379310345, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.7142857142857143, "amount": 0.6666666666666666, "status": 0.6666666666666666, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1193}, "ledger_by_hop": {"1": {"facts_total": 58, "facts_retained": 32, "retention_rate": 0.5517241379310345, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.7142857142857143, "amount": 0.6666666666666666, "status": 0.6666666666666666, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1200}, "2": {"facts_total": 58, "facts_retained": 32, "retention_rate": 0.5517241379310345, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.7142857142857143, "amount": 0.6666666666666666, "status": 0.6666666666666666, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1193}}}, "reference_key_id": "keys/synthetic/syn_03003_2000.json", "task_scores": {"correct": 0.0, "abstained": 0.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "72b629618498e6857b5e5c18413e492e46455d51f70d74a18ff4e73267a76b13", "output_hash": "e971a34f9c497d953f98fb387e189b1587b11a7f6725b2f81ac6d0f3ef8ac845"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "048a3542850befafa3b30db5", "attempt_id": "664806af2097", "run_group_id": "E4_core:A4|closed|qwen3b:h2", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T06:38:16.035385Z", "task_family": "synthetic", "document_id": "syn_03003_2000", "document_hash": "dddca1a38e3224870fcfae927c701c400fd885c19123ea840ee06a76a140f407", "question_id": "q12_unit_conversion", "split": "test", "cluster_id": "syn_03003_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 2, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1866, "query_tokens": 603, "message_tokens": 512, "message_utf8_bytes": 1193, "cache_tensor_bytes": null, "serialized_payload_bytes": 1193, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 44939.650501008146, "hops": {"1": 22906.635708990507, "2": 22033.01479201764}, "receiver_prefill_ms": 385.9296659938991, "ttft_ms": 40.646292036399245, "decode_ms": 147.3567500943318}, "total_wall_ms": 45514.3121680012, "ttft_ms": 40.646292036399245, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 5}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "17436", "evidence_artifact": {"ledger": {"facts_total": 58, "facts_retained": 32, "retention_rate": 0.5517241379310345, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.7142857142857143, "amount": 0.6666666666666666, "status": 0.6666666666666666, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1193}, "ledger_by_hop": {"1": {"facts_total": 58, "facts_retained": 32, "retention_rate": 0.5517241379310345, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.7142857142857143, "amount": 0.6666666666666666, "status": 0.6666666666666666, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1200}, "2": {"facts_total": 58, "facts_retained": 32, "retention_rate": 0.5517241379310345, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.7142857142857143, "amount": 0.6666666666666666, "status": 0.6666666666666666, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1193}}}, "reference_key_id": "keys/synthetic/syn_03003_2000.json", "task_scores": {"correct": 0.0, "abstained": 0.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "72b629618498e6857b5e5c18413e492e46455d51f70d74a18ff4e73267a76b13", "output_hash": "ab6902e72b98e6503bf8b93f8c2674caf792f64181732f233b84e3e4d2abce06"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "87688ab828b543e85e222203", "attempt_id": "8244f1794753", "run_group_id": "E4_core:A4|closed|qwen3b:h2", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T06:38:16.628954Z", "task_family": "synthetic", "document_id": "syn_03003_2000", "document_hash": "dddca1a38e3224870fcfae927c701c400fd885c19123ea840ee06a76a140f407", "question_id": "q13_unanswerable", "split": "test", "cluster_id": "syn_03003_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 2, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1866, "query_tokens": 607, "message_tokens": 512, "message_utf8_bytes": 1193, "cache_tensor_bytes": null, "serialized_payload_bytes": 1193, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 44939.650501008146, "hops": {"1": 22906.635708990507, "2": 22033.01479201764}, "receiver_prefill_ms": 345.41333303786814, "ttft_ms": 50.32958404626697, "decode_ms": 56.919332942925394}, "total_wall_ms": 45393.00175092649, "ttft_ms": 50.32958404626697, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 58, "facts_retained": 32, "retention_rate": 0.5517241379310345, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.7142857142857143, "amount": 0.6666666666666666, "status": 0.6666666666666666, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1193}, "ledger_by_hop": {"1": {"facts_total": 58, "facts_retained": 32, "retention_rate": 0.5517241379310345, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.7142857142857143, "amount": 0.6666666666666666, "status": 0.6666666666666666, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1200}, "2": {"facts_total": 58, "facts_retained": 32, "retention_rate": 0.5517241379310345, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.7142857142857143, "amount": 0.6666666666666666, "status": 0.6666666666666666, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1193}}}, "reference_key_id": "keys/synthetic/syn_03003_2000.json", "task_scores": {"correct": 1.0, "abstained": 1.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "72b629618498e6857b5e5c18413e492e46455d51f70d74a18ff4e73267a76b13", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "eb30a98a4b9f18b12225332e", "attempt_id": "2a7e7007455d", "run_group_id": "E4_core:A4|closed|qwen3b:h2", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T06:38:17.100162Z", "task_family": "synthetic", "document_id": "syn_03003_2000", "document_hash": "dddca1a38e3224870fcfae927c701c400fd885c19123ea840ee06a76a140f407", "question_id": "q14_unanswerable", "split": "test", "cluster_id": "syn_03003_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 2, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1866, "query_tokens": 608, "message_tokens": 512, "message_utf8_bytes": 1193, "cache_tensor_bytes": null, "serialized_payload_bytes": 1193, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 44939.650501008146, "hops": {"1": 22906.635708990507, "2": 22033.01479201764}, "receiver_prefill_ms": 535.1177080301568, "ttft_ms": 30.205000075511634, "decode_ms": 42.39687498193234}, "total_wall_ms": 45548.12737600878, "ttft_ms": 30.205000075511634, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 58, "facts_retained": 32, "retention_rate": 0.5517241379310345, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.7142857142857143, "amount": 0.6666666666666666, "status": 0.6666666666666666, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1193}, "ledger_by_hop": {"1": {"facts_total": 58, "facts_retained": 32, "retention_rate": 0.5517241379310345, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.7142857142857143, "amount": 0.6666666666666666, "status": 0.6666666666666666, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1200}, "2": {"facts_total": 58, "facts_retained": 32, "retention_rate": 0.5517241379310345, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.7142857142857143, "amount": 0.6666666666666666, "status": 0.6666666666666666, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1193}}}, "reference_key_id": "keys/synthetic/syn_03003_2000.json", "task_scores": {"correct": 1.0, "abstained": 1.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "72b629618498e6857b5e5c18413e492e46455d51f70d74a18ff4e73267a76b13", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "cb37f05e61aa90488d615005", "attempt_id": "cf365fcfeb28", "run_group_id": "E4_core:A4|closed|qwen3b:h4", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T06:39:01.591541Z", "task_family": "synthetic", "document_id": "syn_03003_2000", "document_hash": "dddca1a38e3224870fcfae927c701c400fd885c19123ea840ee06a76a140f407", "question_id": "q00_single_fact", "split": "test", "cluster_id": "syn_03003_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 4, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1866, "query_tokens": 595, "message_tokens": 512, "message_utf8_bytes": 1193, "cache_tensor_bytes": null, "serialized_payload_bytes": 1193, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 88801.63508397527, "hops": {"1": 22906.635708990507, "2": 22033.01479201764, "3": 21749.766708002426, "4": 22112.2178749647}, "receiver_prefill_ms": 385.00758400186896, "ttft_ms": 20.2014590613544, "decode_ms": 619.6241669822484}, "total_wall_ms": 89827.42216694169, "ttft_ms": 20.2014590613544, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 10}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "GR-79S-5839", "evidence_artifact": {"ledger": {"facts_total": 58, "facts_retained": 32, "retention_rate": 0.5517241379310345, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.7142857142857143, "amount": 0.6666666666666666, "status": 0.6666666666666666, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1193}, "ledger_by_hop": {"1": {"facts_total": 58, "facts_retained": 32, "retention_rate": 0.5517241379310345, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.7142857142857143, "amount": 0.6666666666666666, "status": 0.6666666666666666, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1200}, "2": {"facts_total": 58, "facts_retained": 32, "retention_rate": 0.5517241379310345, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.7142857142857143, "amount": 0.6666666666666666, "status": 0.6666666666666666, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1193}, "3": {"facts_total": 58, "facts_retained": 32, "retention_rate": 0.5517241379310345, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.7142857142857143, "amount": 0.6666666666666666, "status": 0.6666666666666666, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1193}, "4": {"facts_total": 58, "facts_retained": 32, "retention_rate": 0.5517241379310345, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.7142857142857143, "amount": 0.6666666666666666, "status": 0.6666666666666666, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1193}}}, "reference_key_id": "keys/synthetic/syn_03003_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "72b629618498e6857b5e5c18413e492e46455d51f70d74a18ff4e73267a76b13", "output_hash": "720cc40cd69e85e9728f6bea00b17a95ca07eee643e143a131840763958303cd"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "6bcb20541894f14a50b0f1b6", "attempt_id": "04bf6e0b4d35", "run_group_id": "E4_core:A4|closed|qwen3b:h4", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T06:39:02.635268Z", "task_family": "synthetic", "document_id": "syn_03003_2000", "document_hash": "dddca1a38e3224870fcfae927c701c400fd885c19123ea840ee06a76a140f407", "question_id": "q01_single_fact", "split": "test", "cluster_id": "syn_03003_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 4, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1866, "query_tokens": 599, "message_tokens": 512, "message_utf8_bytes": 1193, "cache_tensor_bytes": null, "serialized_payload_bytes": 1193, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 88801.63508397527, "hops": {"1": 22906.635708990507, "2": 22033.01479201764, "3": 21749.766708002426, "4": 22112.2178749647}, "receiver_prefill_ms": 411.88049991615117, "ttft_ms": 35.044083022512496, "decode_ms": 348.6071249935776}, "total_wall_ms": 89597.85520902369, "ttft_ms": 35.044083022512496, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 11}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "Lot 344-B, Témiscaming", "evidence_artifact": {"ledger": {"facts_total": 58, "facts_retained": 32, "retention_rate": 0.5517241379310345, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.7142857142857143, "amount": 0.6666666666666666, "status": 0.6666666666666666, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1193}, "ledger_by_hop": {"1": {"facts_total": 58, "facts_retained": 32, "retention_rate": 0.5517241379310345, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.7142857142857143, "amount": 0.6666666666666666, "status": 0.6666666666666666, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1200}, "2": {"facts_total": 58, "facts_retained": 32, "retention_rate": 0.5517241379310345, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.7142857142857143, "amount": 0.6666666666666666, "status": 0.6666666666666666, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1193}, "3": {"facts_total": 58, "facts_retained": 32, "retention_rate": 0.5517241379310345, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.7142857142857143, "amount": 0.6666666666666666, "status": 0.6666666666666666, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1193}, "4": {"facts_total": 58, "facts_retained": 32, "retention_rate": 0.5517241379310345, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.7142857142857143, "amount": 0.6666666666666666, "status": 0.6666666666666666, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1193}}}, "reference_key_id": "keys/synthetic/syn_03003_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 0.8571428571428571}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "72b629618498e6857b5e5c18413e492e46455d51f70d74a18ff4e73267a76b13", "output_hash": "0197d491a5a04a142dc8e372360735a1d41757e7a5aea7e81f23b0e1ef6ee010"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "bfac34076685b39a8d2ab3d2", "attempt_id": "797bc19ff823", "run_group_id": "E4_core:A4|closed|qwen3b:h4", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T06:39:03.449156Z", "task_family": "synthetic", "document_id": "syn_03003_2000", "document_hash": "dddca1a38e3224870fcfae927c701c400fd885c19123ea840ee06a76a140f407", "question_id": "q02_exact_copy", "split": "test", "cluster_id": "syn_03003_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 4, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1866, "query_tokens": 599, "message_tokens": 512, "message_utf8_bytes": 1193, "cache_tensor_bytes": null, "serialized_payload_bytes": 1193, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 88801.63508397527, "hops": {"1": 22906.635708990507, "2": 22033.01479201764, "3": 21749.766708002426, "4": 22112.2178749647}, "receiver_prefill_ms": 358.9104590937495, "ttft_ms": 50.516791990958154, "decode_ms": 39.884749916382134}, "total_wall_ms": 89251.68791797478, "ttft_ms": 50.516791990958154, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 58, "facts_retained": 32, "retention_rate": 0.5517241379310345, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.7142857142857143, "amount": 0.6666666666666666, "status": 0.6666666666666666, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1193}, "ledger_by_hop": {"1": {"facts_total": 58, "facts_retained": 32, "retention_rate": 0.5517241379310345, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.7142857142857143, "amount": 0.6666666666666666, "status": 0.6666666666666666, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1200}, "2": {"facts_total": 58, "facts_retained": 32, "retention_rate": 0.5517241379310345, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.7142857142857143, "amount": 0.6666666666666666, "status": 0.6666666666666666, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1193}, "3": {"facts_total": 58, "facts_retained": 32, "retention_rate": 0.5517241379310345, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.7142857142857143, "amount": 0.6666666666666666, "status": 0.6666666666666666, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1193}, "4": {"facts_total": 58, "facts_retained": 32, "retention_rate": 0.5517241379310345, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.7142857142857143, "amount": 0.6666666666666666, "status": 0.6666666666666666, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1193}}}, "reference_key_id": "keys/synthetic/syn_03003_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "72b629618498e6857b5e5c18413e492e46455d51f70d74a18ff4e73267a76b13", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "9874a0ad6666b1c8ab52debe", "attempt_id": "23ff6d2bb508", "run_group_id": "E4_core:A4|closed|qwen3b:h4", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T06:39:03.918117Z", "task_family": "synthetic", "document_id": "syn_03003_2000", "document_hash": "dddca1a38e3224870fcfae927c701c400fd885c19123ea840ee06a76a140f407", "question_id": "q03_exact_copy", "split": "test", "cluster_id": "syn_03003_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 4, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1866, "query_tokens": 594, "message_tokens": 512, "message_utf8_bytes": 1193, "cache_tensor_bytes": null, "serialized_payload_bytes": 1193, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 88801.63508397527, "hops": {"1": 22906.635708990507, "2": 22033.01479201764, "3": 21749.766708002426, "4": 22112.2178749647}, "receiver_prefill_ms": 520.593540975824, "ttft_ms": 40.21795897278935, "decode_ms": 203.20700004231185}, "total_wall_ms": 89566.40629295725, "ttft_ms": 40.21795897278935, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 6}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "Wilhelmina Achterberg", "evidence_artifact": {"ledger": {"facts_total": 58, "facts_retained": 32, "retention_rate": 0.5517241379310345, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.7142857142857143, "amount": 0.6666666666666666, "status": 0.6666666666666666, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1193}, "ledger_by_hop": {"1": {"facts_total": 58, "facts_retained": 32, "retention_rate": 0.5517241379310345, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.7142857142857143, "amount": 0.6666666666666666, "status": 0.6666666666666666, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1200}, "2": {"facts_total": 58, "facts_retained": 32, "retention_rate": 0.5517241379310345, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.7142857142857143, "amount": 0.6666666666666666, "status": 0.6666666666666666, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1193}, "3": {"facts_total": 58, "facts_retained": 32, "retention_rate": 0.5517241379310345, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.7142857142857143, "amount": 0.6666666666666666, "status": 0.6666666666666666, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1193}, "4": {"facts_total": 58, "facts_retained": 32, "retention_rate": 0.5517241379310345, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.7142857142857143, "amount": 0.6666666666666666, "status": 0.6666666666666666, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1193}}}, "reference_key_id": "keys/synthetic/syn_03003_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "72b629618498e6857b5e5c18413e492e46455d51f70d74a18ff4e73267a76b13", "output_hash": "2a62606853c315bba8769371d195e3ab3d6a8cd7ef959205575c5119740850f8"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "63fa9264a22ae6a966878ff0", "attempt_id": "b8861b7d8773", "run_group_id": "E4_core:A4|closed|qwen3b:h4", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T06:39:04.700400Z", "task_family": "synthetic", "document_id": "syn_03003_2000", "document_hash": "dddca1a38e3224870fcfae927c701c400fd885c19123ea840ee06a76a140f407", "question_id": "q04_multi_hop", "split": "test", "cluster_id": "syn_03003_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 4, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1866, "query_tokens": 603, "message_tokens": 512, "message_utf8_bytes": 1193, "cache_tensor_bytes": null, "serialized_payload_bytes": 1193, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 88801.63508397527, "hops": {"1": 22906.635708990507, "2": 22033.01479201764, "3": 21749.766708002426, "4": 22112.2178749647}, "receiver_prefill_ms": 397.3695409949869, "ttft_ms": 26.3688339618966, "decode_ms": 36.551374942064285}, "total_wall_ms": 89262.4877509661, "ttft_ms": 26.3688339618966, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 58, "facts_retained": 32, "retention_rate": 0.5517241379310345, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.7142857142857143, "amount": 0.6666666666666666, "status": 0.6666666666666666, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1193}, "ledger_by_hop": {"1": {"facts_total": 58, "facts_retained": 32, "retention_rate": 0.5517241379310345, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.7142857142857143, "amount": 0.6666666666666666, "status": 0.6666666666666666, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1200}, "2": {"facts_total": 58, "facts_retained": 32, "retention_rate": 0.5517241379310345, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.7142857142857143, "amount": 0.6666666666666666, "status": 0.6666666666666666, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1193}, "3": {"facts_total": 58, "facts_retained": 32, "retention_rate": 0.5517241379310345, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.7142857142857143, "amount": 0.6666666666666666, "status": 0.6666666666666666, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1193}, "4": {"facts_total": 58, "facts_retained": 32, "retention_rate": 0.5517241379310345, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.7142857142857143, "amount": 0.6666666666666666, "status": 0.6666666666666666, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1193}}}, "reference_key_id": "keys/synthetic/syn_03003_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "72b629618498e6857b5e5c18413e492e46455d51f70d74a18ff4e73267a76b13", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "777c1101df2ed9fde710efea", "attempt_id": "d5f7b6df1dd1", "run_group_id": "E4_core:A4|closed|qwen3b:h4", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T06:39:05.178240Z", "task_family": "synthetic", "document_id": "syn_03003_2000", "document_hash": "dddca1a38e3224870fcfae927c701c400fd885c19123ea840ee06a76a140f407", "question_id": "q05_multi_hop", "split": "test", "cluster_id": "syn_03003_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 4, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1866, "query_tokens": 605, "message_tokens": 512, "message_utf8_bytes": 1193, "cache_tensor_bytes": null, "serialized_payload_bytes": 1193, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 88801.63508397527, "hops": {"1": 22906.635708990507, "2": 22033.01479201764, "3": 21749.766708002426, "4": 22112.2178749647}, "receiver_prefill_ms": 358.4045000607148, "ttft_ms": 47.28504200465977, "decode_ms": 520.8295000484213}, "total_wall_ms": 89728.96554192994, "ttft_ms": 47.28504200465977, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 10}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "NK-95Z-2165", "evidence_artifact": {"ledger": {"facts_total": 58, "facts_retained": 32, "retention_rate": 0.5517241379310345, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.7142857142857143, "amount": 0.6666666666666666, "status": 0.6666666666666666, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1193}, "ledger_by_hop": {"1": {"facts_total": 58, "facts_retained": 32, "retention_rate": 0.5517241379310345, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.7142857142857143, "amount": 0.6666666666666666, "status": 0.6666666666666666, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1200}, "2": {"facts_total": 58, "facts_retained": 32, "retention_rate": 0.5517241379310345, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.7142857142857143, "amount": 0.6666666666666666, "status": 0.6666666666666666, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1193}, "3": {"facts_total": 58, "facts_retained": 32, "retention_rate": 0.5517241379310345, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.7142857142857143, "amount": 0.6666666666666666, "status": 0.6666666666666666, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1193}, "4": {"facts_total": 58, "facts_retained": 32, "retention_rate": 0.5517241379310345, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.7142857142857143, "amount": 0.6666666666666666, "status": 0.6666666666666666, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1193}}}, "reference_key_id": "keys/synthetic/syn_03003_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "72b629618498e6857b5e5c18413e492e46455d51f70d74a18ff4e73267a76b13", "output_hash": "bd8f809b02304c6b9dc7c1e116ecccc2328c976a5c071e89a94d661b48e73aa7"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "bd563a96e7a51dcd0037ea65", "attempt_id": "0a65d0996b43", "run_group_id": "E4_core:A4|closed|qwen3b:h4", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T06:39:06.123286Z", "task_family": "synthetic", "document_id": "syn_03003_2000", "document_hash": "dddca1a38e3224870fcfae927c701c400fd885c19123ea840ee06a76a140f407", "question_id": "q06_negation", "split": "test", "cluster_id": "syn_03003_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 4, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1866, "query_tokens": 597, "message_tokens": 512, "message_utf8_bytes": 1193, "cache_tensor_bytes": null, "serialized_payload_bytes": 1193, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 88801.63508397527, "hops": {"1": 22906.635708990507, "2": 22033.01479201764, "3": 21749.766708002426, "4": 22112.2178749647}, "receiver_prefill_ms": 400.722041958943, "ttft_ms": 34.822790999896824, "decode_ms": 35.19187506753951}, "total_wall_ms": 89272.97870896291, "ttft_ms": 34.822790999896824, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "yes", "evidence_artifact": {"ledger": {"facts_total": 58, "facts_retained": 32, "retention_rate": 0.5517241379310345, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.7142857142857143, "amount": 0.6666666666666666, "status": 0.6666666666666666, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1193}, "ledger_by_hop": {"1": {"facts_total": 58, "facts_retained": 32, "retention_rate": 0.5517241379310345, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.7142857142857143, "amount": 0.6666666666666666, "status": 0.6666666666666666, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1200}, "2": {"facts_total": 58, "facts_retained": 32, "retention_rate": 0.5517241379310345, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.7142857142857143, "amount": 0.6666666666666666, "status": 0.6666666666666666, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1193}, "3": {"facts_total": 58, "facts_retained": 32, "retention_rate": 0.5517241379310345, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.7142857142857143, "amount": 0.6666666666666666, "status": 0.6666666666666666, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1193}, "4": {"facts_total": 58, "facts_retained": 32, "retention_rate": 0.5517241379310345, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.7142857142857143, "amount": 0.6666666666666666, "status": 0.6666666666666666, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1193}}}, "reference_key_id": "keys/synthetic/syn_03003_2000.json", "task_scores": {"correct": 0.0, "abstained": 0.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "72b629618498e6857b5e5c18413e492e46455d51f70d74a18ff4e73267a76b13", "output_hash": "8a798890fe93817163b10b5f7bd2ca4d25d84c52739a645a889c173eee7d9d3d"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "57a3b086daa663c3e3256450", "attempt_id": "be4d4b31137d", "run_group_id": "E4_core:A4|closed|qwen3b:h4", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T06:39:06.612441Z", "task_family": "synthetic", "document_id": "syn_03003_2000", "document_hash": "dddca1a38e3224870fcfae927c701c400fd885c19123ea840ee06a76a140f407", "question_id": "q07_negation", "split": "test", "cluster_id": "syn_03003_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 4, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1866, "query_tokens": 594, "message_tokens": 512, "message_utf8_bytes": 1193, "cache_tensor_bytes": null, "serialized_payload_bytes": 1193, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 88801.63508397527, "hops": {"1": 22906.635708990507, "2": 22033.01479201764, "3": 21749.766708002426, "4": 22112.2178749647}, "receiver_prefill_ms": 342.92341698892415, "ttft_ms": 53.97258303128183, "decode_ms": 47.36949992366135}, "total_wall_ms": 89246.53470888734, "ttft_ms": 53.97258303128183, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "no", "evidence_artifact": {"ledger": {"facts_total": 58, "facts_retained": 32, "retention_rate": 0.5517241379310345, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.7142857142857143, "amount": 0.6666666666666666, "status": 0.6666666666666666, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1193}, "ledger_by_hop": {"1": {"facts_total": 58, "facts_retained": 32, "retention_rate": 0.5517241379310345, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.7142857142857143, "amount": 0.6666666666666666, "status": 0.6666666666666666, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1200}, "2": {"facts_total": 58, "facts_retained": 32, "retention_rate": 0.5517241379310345, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.7142857142857143, "amount": 0.6666666666666666, "status": 0.6666666666666666, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1193}, "3": {"facts_total": 58, "facts_retained": 32, "retention_rate": 0.5517241379310345, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.7142857142857143, "amount": 0.6666666666666666, "status": 0.6666666666666666, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1193}, "4": {"facts_total": 58, "facts_retained": 32, "retention_rate": 0.5517241379310345, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.7142857142857143, "amount": 0.6666666666666666, "status": 0.6666666666666666, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1193}}}, "reference_key_id": "keys/synthetic/syn_03003_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "72b629618498e6857b5e5c18413e492e46455d51f70d74a18ff4e73267a76b13", "output_hash": "9390298f3fb0c5b160498935d79cb139aef28e1c47358b4bbba61862b9c26e59"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "8dffdf969db390fffa72a144", "attempt_id": "42a5181768ff", "run_group_id": "E4_core:A4|closed|qwen3b:h4", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T06:39:07.074452Z", "task_family": "synthetic", "document_id": "syn_03003_2000", "document_hash": "dddca1a38e3224870fcfae927c701c400fd885c19123ea840ee06a76a140f407", "question_id": "q08_revision", "split": "test", "cluster_id": "syn_03003_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 4, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1866, "query_tokens": 612, "message_tokens": 512, "message_utf8_bytes": 1193, "cache_tensor_bytes": null, "serialized_payload_bytes": 1193, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 88801.63508397527, "hops": {"1": 22906.635708990507, "2": 22033.01479201764, "3": 21749.766708002426, "4": 22112.2178749647}, "receiver_prefill_ms": 375.81866700202227, "ttft_ms": 88.8307080604136, "decode_ms": 349.3796670809388}, "total_wall_ms": 89616.35541694704, "ttft_ms": 88.8307080604136, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 7}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "911,925", "evidence_artifact": {"ledger": {"facts_total": 58, "facts_retained": 32, "retention_rate": 0.5517241379310345, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.7142857142857143, "amount": 0.6666666666666666, "status": 0.6666666666666666, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1193}, "ledger_by_hop": {"1": {"facts_total": 58, "facts_retained": 32, "retention_rate": 0.5517241379310345, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.7142857142857143, "amount": 0.6666666666666666, "status": 0.6666666666666666, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1200}, "2": {"facts_total": 58, "facts_retained": 32, "retention_rate": 0.5517241379310345, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.7142857142857143, "amount": 0.6666666666666666, "status": 0.6666666666666666, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1193}, "3": {"facts_total": 58, "facts_retained": 32, "retention_rate": 0.5517241379310345, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.7142857142857143, "amount": 0.6666666666666666, "status": 0.6666666666666666, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1193}, "4": {"facts_total": 58, "facts_retained": 32, "retention_rate": 0.5517241379310345, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.7142857142857143, "amount": 0.6666666666666666, "status": 0.6666666666666666, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1193}}}, "reference_key_id": "keys/synthetic/syn_03003_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "72b629618498e6857b5e5c18413e492e46455d51f70d74a18ff4e73267a76b13", "output_hash": "86be78720fc4b79f93e603e7d9e4861fcd25102612f73209fffc17d787056007"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "b19f1181c64b4b29c162f666", "attempt_id": "9fbdd840adaa", "run_group_id": "E4_core:A4|closed|qwen3b:h4", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T06:39:07.906822Z", "task_family": "synthetic", "document_id": "syn_03003_2000", "document_hash": "dddca1a38e3224870fcfae927c701c400fd885c19123ea840ee06a76a140f407", "question_id": "q09_revision", "split": "test", "cluster_id": "syn_03003_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 4, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1866, "query_tokens": 605, "message_tokens": 512, "message_utf8_bytes": 1193, "cache_tensor_bytes": null, "serialized_payload_bytes": 1193, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 88801.63508397527, "hops": {"1": 22906.635708990507, "2": 22033.01479201764, "3": 21749.766708002426, "4": 22112.2178749647}, "receiver_prefill_ms": 412.98633394762874, "ttft_ms": 24.43812496494502, "decode_ms": 208.20095902308822}, "total_wall_ms": 89447.93154194485, "ttft_ms": 24.43812496494502, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 7}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "911,925", "evidence_artifact": {"ledger": {"facts_total": 58, "facts_retained": 32, "retention_rate": 0.5517241379310345, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.7142857142857143, "amount": 0.6666666666666666, "status": 0.6666666666666666, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1193}, "ledger_by_hop": {"1": {"facts_total": 58, "facts_retained": 32, "retention_rate": 0.5517241379310345, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.7142857142857143, "amount": 0.6666666666666666, "status": 0.6666666666666666, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1200}, "2": {"facts_total": 58, "facts_retained": 32, "retention_rate": 0.5517241379310345, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.7142857142857143, "amount": 0.6666666666666666, "status": 0.6666666666666666, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1193}, "3": {"facts_total": 58, "facts_retained": 32, "retention_rate": 0.5517241379310345, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.7142857142857143, "amount": 0.6666666666666666, "status": 0.6666666666666666, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1193}, "4": {"facts_total": 58, "facts_retained": 32, "retention_rate": 0.5517241379310345, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.7142857142857143, "amount": 0.6666666666666666, "status": 0.6666666666666666, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1193}}}, "reference_key_id": "keys/synthetic/syn_03003_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "72b629618498e6857b5e5c18413e492e46455d51f70d74a18ff4e73267a76b13", "output_hash": "86be78720fc4b79f93e603e7d9e4861fcd25102612f73209fffc17d787056007"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "0e860c06fc283f04a6e9f9e4", "attempt_id": "9092c8da5f00", "run_group_id": "E4_core:A4|closed|qwen3b:h4", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T06:39:08.570393Z", "task_family": "synthetic", "document_id": "syn_03003_2000", "document_hash": "dddca1a38e3224870fcfae927c701c400fd885c19123ea840ee06a76a140f407", "question_id": "q10_temporal", "split": "test", "cluster_id": "syn_03003_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 4, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1866, "query_tokens": 612, "message_tokens": 512, "message_utf8_bytes": 1193, "cache_tensor_bytes": null, "serialized_payload_bytes": 1193, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 88801.63508397527, "hops": {"1": 22906.635708990507, "2": 22033.01479201764, "3": 21749.766708002426, "4": 22112.2178749647}, "receiver_prefill_ms": 431.4220000524074, "ttft_ms": 49.94262498803437, "decode_ms": 345.18029203172773}, "total_wall_ms": 89628.83383396547, "ttft_ms": 49.94262498803437, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 6}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "Gamtiszen-2", "evidence_artifact": {"ledger": {"facts_total": 58, "facts_retained": 32, "retention_rate": 0.5517241379310345, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.7142857142857143, "amount": 0.6666666666666666, "status": 0.6666666666666666, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1193}, "ledger_by_hop": {"1": {"facts_total": 58, "facts_retained": 32, "retention_rate": 0.5517241379310345, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.7142857142857143, "amount": 0.6666666666666666, "status": 0.6666666666666666, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1200}, "2": {"facts_total": 58, "facts_retained": 32, "retention_rate": 0.5517241379310345, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.7142857142857143, "amount": 0.6666666666666666, "status": 0.6666666666666666, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1193}, "3": {"facts_total": 58, "facts_retained": 32, "retention_rate": 0.5517241379310345, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.7142857142857143, "amount": 0.6666666666666666, "status": 0.6666666666666666, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1193}, "4": {"facts_total": 58, "facts_retained": 32, "retention_rate": 0.5517241379310345, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.7142857142857143, "amount": 0.6666666666666666, "status": 0.6666666666666666, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1193}}}, "reference_key_id": "keys/synthetic/syn_03003_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "72b629618498e6857b5e5c18413e492e46455d51f70d74a18ff4e73267a76b13", "output_hash": "32e3af428886434823d82102e1c482765eafdf6077fe8d3fba5752ec5e29826c"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "eb45077cca440d81b3746400", "attempt_id": "3bf7c363d4d8", "run_group_id": "E4_core:A4|closed|qwen3b:h4", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T06:39:09.419751Z", "task_family": "synthetic", "document_id": "syn_03003_2000", "document_hash": "dddca1a38e3224870fcfae927c701c400fd885c19123ea840ee06a76a140f407", "question_id": "q11_aggregation", "split": "test", "cluster_id": "syn_03003_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 4, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1866, "query_tokens": 606, "message_tokens": 512, "message_utf8_bytes": 1193, "cache_tensor_bytes": null, "serialized_payload_bytes": 1193, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 88801.63508397527, "hops": {"1": 22906.635708990507, "2": 22033.01479201764, "3": 21749.766708002426, "4": 22112.2178749647}, "receiver_prefill_ms": 398.6943330382928, "ttft_ms": 33.265208010561764, "decode_ms": 222.02574997209013}, "total_wall_ms": 89456.2735420186, "ttft_ms": 33.265208010561764, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 7}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "3397775", "evidence_artifact": {"ledger": {"facts_total": 58, "facts_retained": 32, "retention_rate": 0.5517241379310345, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.7142857142857143, "amount": 0.6666666666666666, "status": 0.6666666666666666, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1193}, "ledger_by_hop": {"1": {"facts_total": 58, "facts_retained": 32, "retention_rate": 0.5517241379310345, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.7142857142857143, "amount": 0.6666666666666666, "status": 0.6666666666666666, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1200}, "2": {"facts_total": 58, "facts_retained": 32, "retention_rate": 0.5517241379310345, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.7142857142857143, "amount": 0.6666666666666666, "status": 0.6666666666666666, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1193}, "3": {"facts_total": 58, "facts_retained": 32, "retention_rate": 0.5517241379310345, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.7142857142857143, "amount": 0.6666666666666666, "status": 0.6666666666666666, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1193}, "4": {"facts_total": 58, "facts_retained": 32, "retention_rate": 0.5517241379310345, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.7142857142857143, "amount": 0.6666666666666666, "status": 0.6666666666666666, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1193}}}, "reference_key_id": "keys/synthetic/syn_03003_2000.json", "task_scores": {"correct": 0.0, "abstained": 0.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "72b629618498e6857b5e5c18413e492e46455d51f70d74a18ff4e73267a76b13", "output_hash": "e971a34f9c497d953f98fb387e189b1587b11a7f6725b2f81ac6d0f3ef8ac845"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "2e461a5c6f53912f653568ab", "attempt_id": "152f22cf45bb", "run_group_id": "E4_core:A4|closed|qwen3b:h4", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T06:39:10.091083Z", "task_family": "synthetic", "document_id": "syn_03003_2000", "document_hash": "dddca1a38e3224870fcfae927c701c400fd885c19123ea840ee06a76a140f407", "question_id": "q12_unit_conversion", "split": "test", "cluster_id": "syn_03003_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 4, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1866, "query_tokens": 603, "message_tokens": 512, "message_utf8_bytes": 1193, "cache_tensor_bytes": null, "serialized_payload_bytes": 1193, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 88801.63508397527, "hops": {"1": 22906.635708990507, "2": 22033.01479201764, "3": 21749.766708002426, "4": 22112.2178749647}, "receiver_prefill_ms": 350.8188339183107, "ttft_ms": 41.28000000491738, "decode_ms": 399.175375001505}, "total_wall_ms": 89593.6406670371, "ttft_ms": 41.28000000491738, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 5}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "17436", "evidence_artifact": {"ledger": {"facts_total": 58, "facts_retained": 32, "retention_rate": 0.5517241379310345, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.7142857142857143, "amount": 0.6666666666666666, "status": 0.6666666666666666, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1193}, "ledger_by_hop": {"1": {"facts_total": 58, "facts_retained": 32, "retention_rate": 0.5517241379310345, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.7142857142857143, "amount": 0.6666666666666666, "status": 0.6666666666666666, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1200}, "2": {"facts_total": 58, "facts_retained": 32, "retention_rate": 0.5517241379310345, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.7142857142857143, "amount": 0.6666666666666666, "status": 0.6666666666666666, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1193}, "3": {"facts_total": 58, "facts_retained": 32, "retention_rate": 0.5517241379310345, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.7142857142857143, "amount": 0.6666666666666666, "status": 0.6666666666666666, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1193}, "4": {"facts_total": 58, "facts_retained": 32, "retention_rate": 0.5517241379310345, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.7142857142857143, "amount": 0.6666666666666666, "status": 0.6666666666666666, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1193}}}, "reference_key_id": "keys/synthetic/syn_03003_2000.json", "task_scores": {"correct": 0.0, "abstained": 0.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "72b629618498e6857b5e5c18413e492e46455d51f70d74a18ff4e73267a76b13", "output_hash": "ab6902e72b98e6503bf8b93f8c2674caf792f64181732f233b84e3e4d2abce06"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "cf14d2d9d7ef0fc4c4bd8ba7", "attempt_id": "ac0d3eb17b18", "run_group_id": "E4_core:A4|closed|qwen3b:h4", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T06:39:10.901449Z", "task_family": "synthetic", "document_id": "syn_03003_2000", "document_hash": "dddca1a38e3224870fcfae927c701c400fd885c19123ea840ee06a76a140f407", "question_id": "q13_unanswerable", "split": "test", "cluster_id": "syn_03003_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 4, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1866, "query_tokens": 607, "message_tokens": 512, "message_utf8_bytes": 1193, "cache_tensor_bytes": null, "serialized_payload_bytes": 1193, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 88801.63508397527, "hops": {"1": 22906.635708990507, "2": 22033.01479201764, "3": 21749.766708002426, "4": 22112.2178749647}, "receiver_prefill_ms": 408.71262503787875, "ttft_ms": 41.59387503750622, "decode_ms": 21.664875093847513}, "total_wall_ms": 89274.24545888789, "ttft_ms": 41.59387503750622, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 58, "facts_retained": 32, "retention_rate": 0.5517241379310345, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.7142857142857143, "amount": 0.6666666666666666, "status": 0.6666666666666666, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1193}, "ledger_by_hop": {"1": {"facts_total": 58, "facts_retained": 32, "retention_rate": 0.5517241379310345, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.7142857142857143, "amount": 0.6666666666666666, "status": 0.6666666666666666, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1200}, "2": {"facts_total": 58, "facts_retained": 32, "retention_rate": 0.5517241379310345, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.7142857142857143, "amount": 0.6666666666666666, "status": 0.6666666666666666, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1193}, "3": {"facts_total": 58, "facts_retained": 32, "retention_rate": 0.5517241379310345, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.7142857142857143, "amount": 0.6666666666666666, "status": 0.6666666666666666, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1193}, "4": {"facts_total": 58, "facts_retained": 32, "retention_rate": 0.5517241379310345, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.7142857142857143, "amount": 0.6666666666666666, "status": 0.6666666666666666, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1193}}}, "reference_key_id": "keys/synthetic/syn_03003_2000.json", "task_scores": {"correct": 1.0, "abstained": 1.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "72b629618498e6857b5e5c18413e492e46455d51f70d74a18ff4e73267a76b13", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "2dfc478da0211905d7639adf", "attempt_id": "156cff10ae09", "run_group_id": "E4_core:A4|closed|qwen3b:h4", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T06:39:11.391228Z", "task_family": "synthetic", "document_id": "syn_03003_2000", "document_hash": "dddca1a38e3224870fcfae927c701c400fd885c19123ea840ee06a76a140f407", "question_id": "q14_unanswerable", "split": "test", "cluster_id": "syn_03003_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 4, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1866, "query_tokens": 608, "message_tokens": 512, "message_utf8_bytes": 1193, "cache_tensor_bytes": null, "serialized_payload_bytes": 1193, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 88801.63508397527, "hops": {"1": 22906.635708990507, "2": 22033.01479201764, "3": 21749.766708002426, "4": 22112.2178749647}, "receiver_prefill_ms": 397.3264580126852, "ttft_ms": 35.967917065136135, "decode_ms": 33.33241702057421}, "total_wall_ms": 89268.93291703891, "ttft_ms": 35.967917065136135, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 58, "facts_retained": 32, "retention_rate": 0.5517241379310345, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.7142857142857143, "amount": 0.6666666666666666, "status": 0.6666666666666666, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1193}, "ledger_by_hop": {"1": {"facts_total": 58, "facts_retained": 32, "retention_rate": 0.5517241379310345, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.7142857142857143, "amount": 0.6666666666666666, "status": 0.6666666666666666, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1200}, "2": {"facts_total": 58, "facts_retained": 32, "retention_rate": 0.5517241379310345, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.7142857142857143, "amount": 0.6666666666666666, "status": 0.6666666666666666, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1193}, "3": {"facts_total": 58, "facts_retained": 32, "retention_rate": 0.5517241379310345, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.7142857142857143, "amount": 0.6666666666666666, "status": 0.6666666666666666, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1193}, "4": {"facts_total": 58, "facts_retained": 32, "retention_rate": 0.5517241379310345, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.7142857142857143, "amount": 0.6666666666666666, "status": 0.6666666666666666, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1193}}}, "reference_key_id": "keys/synthetic/syn_03003_2000.json", "task_scores": {"correct": 1.0, "abstained": 1.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "72b629618498e6857b5e5c18413e492e46455d51f70d74a18ff4e73267a76b13", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "9a3260da84e5331caf408bce", "attempt_id": "8f0e82e00e7c", "run_group_id": "E4_core:A4|closed|qwen3b:h8", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T06:40:40.067127Z", "task_family": "synthetic", "document_id": "syn_03003_2000", "document_hash": "dddca1a38e3224870fcfae927c701c400fd885c19123ea840ee06a76a140f407", "question_id": "q00_single_fact", "split": "test", "cluster_id": "syn_03003_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 8, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1866, "query_tokens": 595, "message_tokens": 512, "message_utf8_bytes": 1193, "cache_tensor_bytes": null, "serialized_payload_bytes": 1193, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 176987.9467500141, "hops": {"1": 22906.635708990507, "2": 22033.01479201764, "3": 21749.766708002426, "4": 22112.2178749647, "5": 21950.30216604937, "6": 22076.25124999322, "7": 22108.02762489766, "8": 22051.73062509857}, "receiver_prefill_ms": 391.61887497175485, "ttft_ms": 24.9967499403283, "decode_ms": 626.6917500179261}, "total_wall_ms": 178032.0417500334, "ttft_ms": 24.9967499403283, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 10}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "GR-79S-5839", "evidence_artifact": {"ledger": {"facts_total": 58, "facts_retained": 32, "retention_rate": 0.5517241379310345, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.7142857142857143, "amount": 0.6666666666666666, "status": 0.6666666666666666, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1193}, "ledger_by_hop": {"1": {"facts_total": 58, "facts_retained": 32, "retention_rate": 0.5517241379310345, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.7142857142857143, "amount": 0.6666666666666666, "status": 0.6666666666666666, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1200}, "2": {"facts_total": 58, "facts_retained": 32, "retention_rate": 0.5517241379310345, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.7142857142857143, "amount": 0.6666666666666666, "status": 0.6666666666666666, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1193}, "3": {"facts_total": 58, "facts_retained": 32, "retention_rate": 0.5517241379310345, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.7142857142857143, "amount": 0.6666666666666666, "status": 0.6666666666666666, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1193}, "4": {"facts_total": 58, "facts_retained": 32, "retention_rate": 0.5517241379310345, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.7142857142857143, "amount": 0.6666666666666666, "status": 0.6666666666666666, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1193}, "5": {"facts_total": 58, "facts_retained": 32, "retention_rate": 0.5517241379310345, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.7142857142857143, "amount": 0.6666666666666666, "status": 0.6666666666666666, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1193}, "6": {"facts_total": 58, "facts_retained": 32, "retention_rate": 0.5517241379310345, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.7142857142857143, "amount": 0.6666666666666666, "status": 0.6666666666666666, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1193}, "7": {"facts_total": 58, "facts_retained": 32, "retention_rate": 0.5517241379310345, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.7142857142857143, "amount": 0.6666666666666666, "status": 0.6666666666666666, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1193}, "8": {"facts_total": 58, "facts_retained": 32, "retention_rate": 0.5517241379310345, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.7142857142857143, "amount": 0.6666666666666666, "status": 0.6666666666666666, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1193}}}, "reference_key_id": "keys/synthetic/syn_03003_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "72b629618498e6857b5e5c18413e492e46455d51f70d74a18ff4e73267a76b13", "output_hash": "720cc40cd69e85e9728f6bea00b17a95ca07eee643e143a131840763958303cd"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "66769e685dc64257db3ce108", "attempt_id": "c9bd10f6fec6", "run_group_id": "E4_core:A4|closed|qwen3b:h8", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T06:40:41.130120Z", "task_family": "synthetic", "document_id": "syn_03003_2000", "document_hash": "dddca1a38e3224870fcfae927c701c400fd885c19123ea840ee06a76a140f407", "question_id": "q01_single_fact", "split": "test", "cluster_id": "syn_03003_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 8, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1866, "query_tokens": 599, "message_tokens": 512, "message_utf8_bytes": 1193, "cache_tensor_bytes": null, "serialized_payload_bytes": 1193, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 176987.9467500141, "hops": {"1": 22906.635708990507, "2": 22033.01479201764, "3": 21749.766708002426, "4": 22112.2178749647, "5": 21950.30216604937, "6": 22076.25124999322, "7": 22108.02762489766, "8": 22051.73062509857}, "receiver_prefill_ms": 396.6583330184221, "ttft_ms": 35.801124991849065, "decode_ms": 350.968417013064}, "total_wall_ms": 177772.00487512164, "ttft_ms": 35.801124991849065, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 11}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "Lot 344-B, Témiscaming", "evidence_artifact": {"ledger": {"facts_total": 58, "facts_retained": 32, "retention_rate": 0.5517241379310345, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.7142857142857143, "amount": 0.6666666666666666, "status": 0.6666666666666666, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1193}, "ledger_by_hop": {"1": {"facts_total": 58, "facts_retained": 32, "retention_rate": 0.5517241379310345, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.7142857142857143, "amount": 0.6666666666666666, "status": 0.6666666666666666, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1200}, "2": {"facts_total": 58, "facts_retained": 32, "retention_rate": 0.5517241379310345, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.7142857142857143, "amount": 0.6666666666666666, "status": 0.6666666666666666, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1193}, "3": {"facts_total": 58, "facts_retained": 32, "retention_rate": 0.5517241379310345, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.7142857142857143, "amount": 0.6666666666666666, "status": 0.6666666666666666, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1193}, "4": {"facts_total": 58, "facts_retained": 32, "retention_rate": 0.5517241379310345, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.7142857142857143, "amount": 0.6666666666666666, "status": 0.6666666666666666, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1193}, "5": {"facts_total": 58, "facts_retained": 32, "retention_rate": 0.5517241379310345, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.7142857142857143, "amount": 0.6666666666666666, "status": 0.6666666666666666, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1193}, "6": {"facts_total": 58, "facts_retained": 32, "retention_rate": 0.5517241379310345, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.7142857142857143, "amount": 0.6666666666666666, "status": 0.6666666666666666, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1193}, "7": {"facts_total": 58, "facts_retained": 32, "retention_rate": 0.5517241379310345, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.7142857142857143, "amount": 0.6666666666666666, "status": 0.6666666666666666, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1193}, "8": {"facts_total": 58, "facts_retained": 32, "retention_rate": 0.5517241379310345, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.7142857142857143, "amount": 0.6666666666666666, "status": 0.6666666666666666, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1193}}}, "reference_key_id": "keys/synthetic/syn_03003_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 0.8571428571428571}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "72b629618498e6857b5e5c18413e492e46455d51f70d74a18ff4e73267a76b13", "output_hash": "0197d491a5a04a142dc8e372360735a1d41757e7a5aea7e81f23b0e1ef6ee010"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "87112b9e6f4375e137cb7b60", "attempt_id": "62faf42d4ae0", "run_group_id": "E4_core:A4|closed|qwen3b:h8", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T06:40:41.931713Z", "task_family": "synthetic", "document_id": "syn_03003_2000", "document_hash": "dddca1a38e3224870fcfae927c701c400fd885c19123ea840ee06a76a140f407", "question_id": "q02_exact_copy", "split": "test", "cluster_id": "syn_03003_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 8, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1866, "query_tokens": 599, "message_tokens": 512, "message_utf8_bytes": 1193, "cache_tensor_bytes": null, "serialized_payload_bytes": 1193, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 176987.9467500141, "hops": {"1": 22906.635708990507, "2": 22033.01479201764, "3": 21749.766708002426, "4": 22112.2178749647, "5": 21950.30216604937, "6": 22076.25124999322, "7": 22108.02762489766, "8": 22051.73062509857}, "receiver_prefill_ms": 368.7538329977542, "ttft_ms": 49.84420898836106, "decode_ms": 45.41720903944224}, "total_wall_ms": 177452.63566693757, "ttft_ms": 49.84420898836106, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 58, "facts_retained": 32, "retention_rate": 0.5517241379310345, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.7142857142857143, "amount": 0.6666666666666666, "status": 0.6666666666666666, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1193}, "ledger_by_hop": {"1": {"facts_total": 58, "facts_retained": 32, "retention_rate": 0.5517241379310345, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.7142857142857143, "amount": 0.6666666666666666, "status": 0.6666666666666666, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1200}, "2": {"facts_total": 58, "facts_retained": 32, "retention_rate": 0.5517241379310345, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.7142857142857143, "amount": 0.6666666666666666, "status": 0.6666666666666666, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1193}, "3": {"facts_total": 58, "facts_retained": 32, "retention_rate": 0.5517241379310345, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.7142857142857143, "amount": 0.6666666666666666, "status": 0.6666666666666666, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1193}, "4": {"facts_total": 58, "facts_retained": 32, "retention_rate": 0.5517241379310345, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.7142857142857143, "amount": 0.6666666666666666, "status": 0.6666666666666666, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1193}, "5": {"facts_total": 58, "facts_retained": 32, "retention_rate": 0.5517241379310345, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.7142857142857143, "amount": 0.6666666666666666, "status": 0.6666666666666666, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1193}, "6": {"facts_total": 58, "facts_retained": 32, "retention_rate": 0.5517241379310345, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.7142857142857143, "amount": 0.6666666666666666, "status": 0.6666666666666666, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1193}, "7": {"facts_total": 58, "facts_retained": 32, "retention_rate": 0.5517241379310345, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.7142857142857143, "amount": 0.6666666666666666, "status": 0.6666666666666666, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1193}, "8": {"facts_total": 58, "facts_retained": 32, "retention_rate": 0.5517241379310345, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.7142857142857143, "amount": 0.6666666666666666, "status": 0.6666666666666666, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1193}}}, "reference_key_id": "keys/synthetic/syn_03003_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "72b629618498e6857b5e5c18413e492e46455d51f70d74a18ff4e73267a76b13", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "abc26168fdc1b1653847b70b", "attempt_id": "99166135472a", "run_group_id": "E4_core:A4|closed|qwen3b:h8", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T06:40:42.414928Z", "task_family": "synthetic", "document_id": "syn_03003_2000", "document_hash": "dddca1a38e3224870fcfae927c701c400fd885c19123ea840ee06a76a140f407", "question_id": "q03_exact_copy", "split": "test", "cluster_id": "syn_03003_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 8, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1866, "query_tokens": 594, "message_tokens": 512, "message_utf8_bytes": 1193, "cache_tensor_bytes": null, "serialized_payload_bytes": 1193, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 176987.9467500141, "hops": {"1": 22906.635708990507, "2": 22033.01479201764, "3": 21749.766708002426, "4": 22112.2178749647, "5": 21950.30216604937, "6": 22076.25124999322, "7": 22108.02762489766, "8": 22051.73062509857}, "receiver_prefill_ms": 525.5227499874309, "ttft_ms": 23.453874979168177, "decode_ms": 194.85187507234514}, "total_wall_ms": 177732.57945897058, "ttft_ms": 23.453874979168177, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 6}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "Wilhelmina Achterberg", "evidence_artifact": {"ledger": {"facts_total": 58, "facts_retained": 32, "retention_rate": 0.5517241379310345, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.7142857142857143, "amount": 0.6666666666666666, "status": 0.6666666666666666, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1193}, "ledger_by_hop": {"1": {"facts_total": 58, "facts_retained": 32, "retention_rate": 0.5517241379310345, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.7142857142857143, "amount": 0.6666666666666666, "status": 0.6666666666666666, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1200}, "2": {"facts_total": 58, "facts_retained": 32, "retention_rate": 0.5517241379310345, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.7142857142857143, "amount": 0.6666666666666666, "status": 0.6666666666666666, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1193}, "3": {"facts_total": 58, "facts_retained": 32, "retention_rate": 0.5517241379310345, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.7142857142857143, "amount": 0.6666666666666666, "status": 0.6666666666666666, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1193}, "4": {"facts_total": 58, "facts_retained": 32, "retention_rate": 0.5517241379310345, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.7142857142857143, "amount": 0.6666666666666666, "status": 0.6666666666666666, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1193}, "5": {"facts_total": 58, "facts_retained": 32, "retention_rate": 0.5517241379310345, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.7142857142857143, "amount": 0.6666666666666666, "status": 0.6666666666666666, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1193}, "6": {"facts_total": 58, "facts_retained": 32, "retention_rate": 0.5517241379310345, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.7142857142857143, "amount": 0.6666666666666666, "status": 0.6666666666666666, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1193}, "7": {"facts_total": 58, "facts_retained": 32, "retention_rate": 0.5517241379310345, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.7142857142857143, "amount": 0.6666666666666666, "status": 0.6666666666666666, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1193}, "8": {"facts_total": 58, "facts_retained": 32, "retention_rate": 0.5517241379310345, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.7142857142857143, "amount": 0.6666666666666666, "status": 0.6666666666666666, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1193}}}, "reference_key_id": "keys/synthetic/syn_03003_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "72b629618498e6857b5e5c18413e492e46455d51f70d74a18ff4e73267a76b13", "output_hash": "2a62606853c315bba8769371d195e3ab3d6a8cd7ef959205575c5119740850f8"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "b228071279db3b0da26031de", "attempt_id": "65d560ecfba8", "run_group_id": "E4_core:A4|closed|qwen3b:h8", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T06:40:43.178218Z", "task_family": "synthetic", "document_id": "syn_03003_2000", "document_hash": "dddca1a38e3224870fcfae927c701c400fd885c19123ea840ee06a76a140f407", "question_id": "q04_multi_hop", "split": "test", "cluster_id": "syn_03003_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 8, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1866, "query_tokens": 603, "message_tokens": 512, "message_utf8_bytes": 1193, "cache_tensor_bytes": null, "serialized_payload_bytes": 1193, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 176987.9467500141, "hops": {"1": 22906.635708990507, "2": 22033.01479201764, "3": 21749.766708002426, "4": 22112.2178749647, "5": 21950.30216604937, "6": 22076.25124999322, "7": 22108.02762489766, "8": 22051.73062509857}, "receiver_prefill_ms": 397.14287500828505, "ttft_ms": 26.266040978953242, "decode_ms": 36.74691601190716}, "total_wall_ms": 177448.7660001032, "ttft_ms": 26.266040978953242, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 58, "facts_retained": 32, "retention_rate": 0.5517241379310345, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.7142857142857143, "amount": 0.6666666666666666, "status": 0.6666666666666666, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1193}, "ledger_by_hop": {"1": {"facts_total": 58, "facts_retained": 32, "retention_rate": 0.5517241379310345, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.7142857142857143, "amount": 0.6666666666666666, "status": 0.6666666666666666, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1200}, "2": {"facts_total": 58, "facts_retained": 32, "retention_rate": 0.5517241379310345, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.7142857142857143, "amount": 0.6666666666666666, "status": 0.6666666666666666, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1193}, "3": {"facts_total": 58, "facts_retained": 32, "retention_rate": 0.5517241379310345, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.7142857142857143, "amount": 0.6666666666666666, "status": 0.6666666666666666, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1193}, "4": {"facts_total": 58, "facts_retained": 32, "retention_rate": 0.5517241379310345, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.7142857142857143, "amount": 0.6666666666666666, "status": 0.6666666666666666, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1193}, "5": {"facts_total": 58, "facts_retained": 32, "retention_rate": 0.5517241379310345, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.7142857142857143, "amount": 0.6666666666666666, "status": 0.6666666666666666, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1193}, "6": {"facts_total": 58, "facts_retained": 32, "retention_rate": 0.5517241379310345, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.7142857142857143, "amount": 0.6666666666666666, "status": 0.6666666666666666, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1193}, "7": {"facts_total": 58, "facts_retained": 32, "retention_rate": 0.5517241379310345, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.7142857142857143, "amount": 0.6666666666666666, "status": 0.6666666666666666, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1193}, "8": {"facts_total": 58, "facts_retained": 32, "retention_rate": 0.5517241379310345, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.7142857142857143, "amount": 0.6666666666666666, "status": 0.6666666666666666, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1193}}}, "reference_key_id": "keys/synthetic/syn_03003_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "72b629618498e6857b5e5c18413e492e46455d51f70d74a18ff4e73267a76b13", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "a156b23dcd7a408071ce74f9", "attempt_id": "010411c3a430", "run_group_id": "E4_core:A4|closed|qwen3b:h8", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T06:40:43.656127Z", "task_family": "synthetic", "document_id": "syn_03003_2000", "document_hash": "dddca1a38e3224870fcfae927c701c400fd885c19123ea840ee06a76a140f407", "question_id": "q05_multi_hop", "split": "test", "cluster_id": "syn_03003_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 8, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1866, "query_tokens": 605, "message_tokens": 512, "message_utf8_bytes": 1193, "cache_tensor_bytes": null, "serialized_payload_bytes": 1193, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 176987.9467500141, "hops": {"1": 22906.635708990507, "2": 22033.01479201764, "3": 21749.766708002426, "4": 22112.2178749647, "5": 21950.30216604937, "6": 22076.25124999322, "7": 22108.02762489766, "8": 22051.73062509857}, "receiver_prefill_ms": 339.49787507299334, "ttft_ms": 50.547000020742416, "decode_ms": 541.8471250450239}, "total_wall_ms": 177920.55804200936, "ttft_ms": 50.547000020742416, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 10}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "NK-95Z-2165", "evidence_artifact": {"ledger": {"facts_total": 58, "facts_retained": 32, "retention_rate": 0.5517241379310345, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.7142857142857143, "amount": 0.6666666666666666, "status": 0.6666666666666666, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1193}, "ledger_by_hop": {"1": {"facts_total": 58, "facts_retained": 32, "retention_rate": 0.5517241379310345, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.7142857142857143, "amount": 0.6666666666666666, "status": 0.6666666666666666, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1200}, "2": {"facts_total": 58, "facts_retained": 32, "retention_rate": 0.5517241379310345, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.7142857142857143, "amount": 0.6666666666666666, "status": 0.6666666666666666, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1193}, "3": {"facts_total": 58, "facts_retained": 32, "retention_rate": 0.5517241379310345, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.7142857142857143, "amount": 0.6666666666666666, "status": 0.6666666666666666, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1193}, "4": {"facts_total": 58, "facts_retained": 32, "retention_rate": 0.5517241379310345, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.7142857142857143, "amount": 0.6666666666666666, "status": 0.6666666666666666, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1193}, "5": {"facts_total": 58, "facts_retained": 32, "retention_rate": 0.5517241379310345, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.7142857142857143, "amount": 0.6666666666666666, "status": 0.6666666666666666, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1193}, "6": {"facts_total": 58, "facts_retained": 32, "retention_rate": 0.5517241379310345, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.7142857142857143, "amount": 0.6666666666666666, "status": 0.6666666666666666, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1193}, "7": {"facts_total": 58, "facts_retained": 32, "retention_rate": 0.5517241379310345, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.7142857142857143, "amount": 0.6666666666666666, "status": 0.6666666666666666, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1193}, "8": {"facts_total": 58, "facts_retained": 32, "retention_rate": 0.5517241379310345, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.7142857142857143, "amount": 0.6666666666666666, "status": 0.6666666666666666, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1193}}}, "reference_key_id": "keys/synthetic/syn_03003_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "72b629618498e6857b5e5c18413e492e46455d51f70d74a18ff4e73267a76b13", "output_hash": "bd8f809b02304c6b9dc7c1e116ecccc2328c976a5c071e89a94d661b48e73aa7"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "ed29834b1be3ef9b3d662e28", "attempt_id": "64052bffc903", "run_group_id": "E4_core:A4|closed|qwen3b:h8", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T06:40:44.607732Z", "task_family": "synthetic", "document_id": "syn_03003_2000", "document_hash": "dddca1a38e3224870fcfae927c701c400fd885c19123ea840ee06a76a140f407", "question_id": "q06_negation", "split": "test", "cluster_id": "syn_03003_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 8, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1866, "query_tokens": 597, "message_tokens": 512, "message_utf8_bytes": 1193, "cache_tensor_bytes": null, "serialized_payload_bytes": 1193, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 176987.9467500141, "hops": {"1": 22906.635708990507, "2": 22033.01479201764, "3": 21749.766708002426, "4": 22112.2178749647, "5": 21950.30216604937, "6": 22076.25124999322, "7": 22108.02762489766, "8": 22051.73062509857}, "receiver_prefill_ms": 403.1206249492243, "ttft_ms": 32.86737494636327, "decode_ms": 35.761083010584116}, "total_wall_ms": 177460.2995419409, "ttft_ms": 32.86737494636327, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "yes", "evidence_artifact": {"ledger": {"facts_total": 58, "facts_retained": 32, "retention_rate": 0.5517241379310345, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.7142857142857143, "amount": 0.6666666666666666, "status": 0.6666666666666666, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1193}, "ledger_by_hop": {"1": {"facts_total": 58, "facts_retained": 32, "retention_rate": 0.5517241379310345, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.7142857142857143, "amount": 0.6666666666666666, "status": 0.6666666666666666, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1200}, "2": {"facts_total": 58, "facts_retained": 32, "retention_rate": 0.5517241379310345, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.7142857142857143, "amount": 0.6666666666666666, "status": 0.6666666666666666, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1193}, "3": {"facts_total": 58, "facts_retained": 32, "retention_rate": 0.5517241379310345, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.7142857142857143, "amount": 0.6666666666666666, "status": 0.6666666666666666, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1193}, "4": {"facts_total": 58, "facts_retained": 32, "retention_rate": 0.5517241379310345, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.7142857142857143, "amount": 0.6666666666666666, "status": 0.6666666666666666, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1193}, "5": {"facts_total": 58, "facts_retained": 32, "retention_rate": 0.5517241379310345, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.7142857142857143, "amount": 0.6666666666666666, "status": 0.6666666666666666, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1193}, "6": {"facts_total": 58, "facts_retained": 32, "retention_rate": 0.5517241379310345, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.7142857142857143, "amount": 0.6666666666666666, "status": 0.6666666666666666, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1193}, "7": {"facts_total": 58, "facts_retained": 32, "retention_rate": 0.5517241379310345, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.7142857142857143, "amount": 0.6666666666666666, "status": 0.6666666666666666, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1193}, "8": {"facts_total": 58, "facts_retained": 32, "retention_rate": 0.5517241379310345, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.7142857142857143, "amount": 0.6666666666666666, "status": 0.6666666666666666, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1193}}}, "reference_key_id": "keys/synthetic/syn_03003_2000.json", "task_scores": {"correct": 0.0, "abstained": 0.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "72b629618498e6857b5e5c18413e492e46455d51f70d74a18ff4e73267a76b13", "output_hash": "8a798890fe93817163b10b5f7bd2ca4d25d84c52739a645a889c173eee7d9d3d"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "bcfe7b99b1fba0f1b8da672d", "attempt_id": "f6fbdbbd4103", "run_group_id": "E4_core:A4|closed|qwen3b:h8", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T06:40:45.096872Z", "task_family": "synthetic", "document_id": "syn_03003_2000", "document_hash": "dddca1a38e3224870fcfae927c701c400fd885c19123ea840ee06a76a140f407", "question_id": "q07_negation", "split": "test", "cluster_id": "syn_03003_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 8, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1866, "query_tokens": 594, "message_tokens": 512, "message_utf8_bytes": 1193, "cache_tensor_bytes": null, "serialized_payload_bytes": 1193, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 176987.9467500141, "hops": {"1": 22906.635708990507, "2": 22033.01479201764, "3": 21749.766708002426, "4": 22112.2178749647, "5": 21950.30216604937, "6": 22076.25124999322, "7": 22108.02762489766, "8": 22051.73062509857}, "receiver_prefill_ms": 363.50433400366455, "ttft_ms": 38.219000096432865, "decode_ms": 74.1792080225423}, "total_wall_ms": 177464.50625010766, "ttft_ms": 38.219000096432865, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "no", "evidence_artifact": {"ledger": {"facts_total": 58, "facts_retained": 32, "retention_rate": 0.5517241379310345, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.7142857142857143, "amount": 0.6666666666666666, "status": 0.6666666666666666, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1193}, "ledger_by_hop": {"1": {"facts_total": 58, "facts_retained": 32, "retention_rate": 0.5517241379310345, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.7142857142857143, "amount": 0.6666666666666666, "status": 0.6666666666666666, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1200}, "2": {"facts_total": 58, "facts_retained": 32, "retention_rate": 0.5517241379310345, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.7142857142857143, "amount": 0.6666666666666666, "status": 0.6666666666666666, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1193}, "3": {"facts_total": 58, "facts_retained": 32, "retention_rate": 0.5517241379310345, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.7142857142857143, "amount": 0.6666666666666666, "status": 0.6666666666666666, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1193}, "4": {"facts_total": 58, "facts_retained": 32, "retention_rate": 0.5517241379310345, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.7142857142857143, "amount": 0.6666666666666666, "status": 0.6666666666666666, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1193}, "5": {"facts_total": 58, "facts_retained": 32, "retention_rate": 0.5517241379310345, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.7142857142857143, "amount": 0.6666666666666666, "status": 0.6666666666666666, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1193}, "6": {"facts_total": 58, "facts_retained": 32, "retention_rate": 0.5517241379310345, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.7142857142857143, "amount": 0.6666666666666666, "status": 0.6666666666666666, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1193}, "7": {"facts_total": 58, "facts_retained": 32, "retention_rate": 0.5517241379310345, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.7142857142857143, "amount": 0.6666666666666666, "status": 0.6666666666666666, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1193}, "8": {"facts_total": 58, "facts_retained": 32, "retention_rate": 0.5517241379310345, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.7142857142857143, "amount": 0.6666666666666666, "status": 0.6666666666666666, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1193}}}, "reference_key_id": "keys/synthetic/syn_03003_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "72b629618498e6857b5e5c18413e492e46455d51f70d74a18ff4e73267a76b13", "output_hash": "9390298f3fb0c5b160498935d79cb139aef28e1c47358b4bbba61862b9c26e59"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "afb1a0682486ba523d122b9f", "attempt_id": "ca637b72acff", "run_group_id": "E4_core:A4|closed|qwen3b:h8", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T06:40:45.590804Z", "task_family": "synthetic", "document_id": "syn_03003_2000", "document_hash": "dddca1a38e3224870fcfae927c701c400fd885c19123ea840ee06a76a140f407", "question_id": "q08_revision", "split": "test", "cluster_id": "syn_03003_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 8, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1866, "query_tokens": 612, "message_tokens": 512, "message_utf8_bytes": 1193, "cache_tensor_bytes": null, "serialized_payload_bytes": 1193, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 176987.9467500141, "hops": {"1": 22906.635708990507, "2": 22033.01479201764, "3": 21749.766708002426, "4": 22112.2178749647, "5": 21950.30216604937, "6": 22076.25124999322, "7": 22108.02762489766, "8": 22051.73062509857}, "receiver_prefill_ms": 570.0728329829872, "ttft_ms": 23.68737489450723, "decode_ms": 244.2740419646725}, "total_wall_ms": 177826.6667081043, "ttft_ms": 23.68737489450723, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 7}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "911,925", "evidence_artifact": {"ledger": {"facts_total": 58, "facts_retained": 32, "retention_rate": 0.5517241379310345, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.7142857142857143, "amount": 0.6666666666666666, "status": 0.6666666666666666, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1193}, "ledger_by_hop": {"1": {"facts_total": 58, "facts_retained": 32, "retention_rate": 0.5517241379310345, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.7142857142857143, "amount": 0.6666666666666666, "status": 0.6666666666666666, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1200}, "2": {"facts_total": 58, "facts_retained": 32, "retention_rate": 0.5517241379310345, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.7142857142857143, "amount": 0.6666666666666666, "status": 0.6666666666666666, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1193}, "3": {"facts_total": 58, "facts_retained": 32, "retention_rate": 0.5517241379310345, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.7142857142857143, "amount": 0.6666666666666666, "status": 0.6666666666666666, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1193}, "4": {"facts_total": 58, "facts_retained": 32, "retention_rate": 0.5517241379310345, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.7142857142857143, "amount": 0.6666666666666666, "status": 0.6666666666666666, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1193}, "5": {"facts_total": 58, "facts_retained": 32, "retention_rate": 0.5517241379310345, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.7142857142857143, "amount": 0.6666666666666666, "status": 0.6666666666666666, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1193}, "6": {"facts_total": 58, "facts_retained": 32, "retention_rate": 0.5517241379310345, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.7142857142857143, "amount": 0.6666666666666666, "status": 0.6666666666666666, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1193}, "7": {"facts_total": 58, "facts_retained": 32, "retention_rate": 0.5517241379310345, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.7142857142857143, "amount": 0.6666666666666666, "status": 0.6666666666666666, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1193}, "8": {"facts_total": 58, "facts_retained": 32, "retention_rate": 0.5517241379310345, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.7142857142857143, "amount": 0.6666666666666666, "status": 0.6666666666666666, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1193}}}, "reference_key_id": "keys/synthetic/syn_03003_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "72b629618498e6857b5e5c18413e492e46455d51f70d74a18ff4e73267a76b13", "output_hash": "86be78720fc4b79f93e603e7d9e4861fcd25102612f73209fffc17d787056007"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "9f128bae98e4f6f6243d0e3e", "attempt_id": "3ec5d5f73cd6", "run_group_id": "E4_core:A4|closed|qwen3b:h8", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T06:40:46.447997Z", "task_family": "synthetic", "document_id": "syn_03003_2000", "document_hash": "dddca1a38e3224870fcfae927c701c400fd885c19123ea840ee06a76a140f407", "question_id": "q09_revision", "split": "test", "cluster_id": "syn_03003_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 8, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1866, "query_tokens": 605, "message_tokens": 512, "message_utf8_bytes": 1193, "cache_tensor_bytes": null, "serialized_payload_bytes": 1193, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 176987.9467500141, "hops": {"1": 22906.635708990507, "2": 22033.01479201764, "3": 21749.766708002426, "4": 22112.2178749647, "5": 21950.30216604937, "6": 22076.25124999322, "7": 22108.02762489766, "8": 22051.73062509857}, "receiver_prefill_ms": 410.5601250194013, "ttft_ms": 38.13724999781698, "decode_ms": 237.2307920595631}, "total_wall_ms": 177674.62858301587, "ttft_ms": 38.13724999781698, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 7}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "911,925", "evidence_artifact": {"ledger": {"facts_total": 58, "facts_retained": 32, "retention_rate": 0.5517241379310345, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.7142857142857143, "amount": 0.6666666666666666, "status": 0.6666666666666666, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1193}, "ledger_by_hop": {"1": {"facts_total": 58, "facts_retained": 32, "retention_rate": 0.5517241379310345, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.7142857142857143, "amount": 0.6666666666666666, "status": 0.6666666666666666, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1200}, "2": {"facts_total": 58, "facts_retained": 32, "retention_rate": 0.5517241379310345, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.7142857142857143, "amount": 0.6666666666666666, "status": 0.6666666666666666, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1193}, "3": {"facts_total": 58, "facts_retained": 32, "retention_rate": 0.5517241379310345, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.7142857142857143, "amount": 0.6666666666666666, "status": 0.6666666666666666, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1193}, "4": {"facts_total": 58, "facts_retained": 32, "retention_rate": 0.5517241379310345, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.7142857142857143, "amount": 0.6666666666666666, "status": 0.6666666666666666, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1193}, "5": {"facts_total": 58, "facts_retained": 32, "retention_rate": 0.5517241379310345, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.7142857142857143, "amount": 0.6666666666666666, "status": 0.6666666666666666, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1193}, "6": {"facts_total": 58, "facts_retained": 32, "retention_rate": 0.5517241379310345, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.7142857142857143, "amount": 0.6666666666666666, "status": 0.6666666666666666, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1193}, "7": {"facts_total": 58, "facts_retained": 32, "retention_rate": 0.5517241379310345, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.7142857142857143, "amount": 0.6666666666666666, "status": 0.6666666666666666, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1193}, "8": {"facts_total": 58, "facts_retained": 32, "retention_rate": 0.5517241379310345, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.7142857142857143, "amount": 0.6666666666666666, "status": 0.6666666666666666, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1193}}}, "reference_key_id": "keys/synthetic/syn_03003_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "72b629618498e6857b5e5c18413e492e46455d51f70d74a18ff4e73267a76b13", "output_hash": "86be78720fc4b79f93e603e7d9e4861fcd25102612f73209fffc17d787056007"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "897f34d43dfd75a2bf375359", "attempt_id": "b1a4be776146", "run_group_id": "E4_core:A4|closed|qwen3b:h8", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T06:40:47.152373Z", "task_family": "synthetic", "document_id": "syn_03003_2000", "document_hash": "dddca1a38e3224870fcfae927c701c400fd885c19123ea840ee06a76a140f407", "question_id": "q10_temporal", "split": "test", "cluster_id": "syn_03003_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 8, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1866, "query_tokens": 612, "message_tokens": 512, "message_utf8_bytes": 1193, "cache_tensor_bytes": null, "serialized_payload_bytes": 1193, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 176987.9467500141, "hops": {"1": 22906.635708990507, "2": 22033.01479201764, "3": 21749.766708002426, "4": 22112.2178749647, "5": 21950.30216604937, "6": 22076.25124999322, "7": 22108.02762489766, "8": 22051.73062509857}, "receiver_prefill_ms": 494.545416906476, "ttft_ms": 97.11879200767726, "decode_ms": 210.6802500784397}, "total_wall_ms": 177790.99887504708, "ttft_ms": 97.11879200767726, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 6}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "Gamtiszen-2", "evidence_artifact": {"ledger": {"facts_total": 58, "facts_retained": 32, "retention_rate": 0.5517241379310345, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.7142857142857143, "amount": 0.6666666666666666, "status": 0.6666666666666666, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1193}, "ledger_by_hop": {"1": {"facts_total": 58, "facts_retained": 32, "retention_rate": 0.5517241379310345, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.7142857142857143, "amount": 0.6666666666666666, "status": 0.6666666666666666, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1200}, "2": {"facts_total": 58, "facts_retained": 32, "retention_rate": 0.5517241379310345, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.7142857142857143, "amount": 0.6666666666666666, "status": 0.6666666666666666, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1193}, "3": {"facts_total": 58, "facts_retained": 32, "retention_rate": 0.5517241379310345, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.7142857142857143, "amount": 0.6666666666666666, "status": 0.6666666666666666, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1193}, "4": {"facts_total": 58, "facts_retained": 32, "retention_rate": 0.5517241379310345, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.7142857142857143, "amount": 0.6666666666666666, "status": 0.6666666666666666, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1193}, "5": {"facts_total": 58, "facts_retained": 32, "retention_rate": 0.5517241379310345, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.7142857142857143, "amount": 0.6666666666666666, "status": 0.6666666666666666, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1193}, "6": {"facts_total": 58, "facts_retained": 32, "retention_rate": 0.5517241379310345, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.7142857142857143, "amount": 0.6666666666666666, "status": 0.6666666666666666, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1193}, "7": {"facts_total": 58, "facts_retained": 32, "retention_rate": 0.5517241379310345, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.7142857142857143, "amount": 0.6666666666666666, "status": 0.6666666666666666, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1193}, "8": {"facts_total": 58, "facts_retained": 32, "retention_rate": 0.5517241379310345, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.7142857142857143, "amount": 0.6666666666666666, "status": 0.6666666666666666, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1193}}}, "reference_key_id": "keys/synthetic/syn_03003_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "72b629618498e6857b5e5c18413e492e46455d51f70d74a18ff4e73267a76b13", "output_hash": "32e3af428886434823d82102e1c482765eafdf6077fe8d3fba5752ec5e29826c"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "59915243529b34ff2977e30d", "attempt_id": "e32a08f2f5f2", "run_group_id": "E4_core:A4|closed|qwen3b:h8", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T06:40:47.973298Z", "task_family": "synthetic", "document_id": "syn_03003_2000", "document_hash": "dddca1a38e3224870fcfae927c701c400fd885c19123ea840ee06a76a140f407", "question_id": "q11_aggregation", "split": "test", "cluster_id": "syn_03003_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 8, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1866, "query_tokens": 606, "message_tokens": 512, "message_utf8_bytes": 1193, "cache_tensor_bytes": null, "serialized_payload_bytes": 1193, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 176987.9467500141, "hops": {"1": 22906.635708990507, "2": 22033.01479201764, "3": 21749.766708002426, "4": 22112.2178749647, "5": 21950.30216604937, "6": 22076.25124999322, "7": 22108.02762489766, "8": 22051.73062509857}, "receiver_prefill_ms": 415.7853330252692, "ttft_ms": 42.042374960146844, "decode_ms": 193.7463330104947}, "total_wall_ms": 177640.39237506222, "ttft_ms": 42.042374960146844, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 7}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "3397775", "evidence_artifact": {"ledger": {"facts_total": 58, "facts_retained": 32, "retention_rate": 0.5517241379310345, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.7142857142857143, "amount": 0.6666666666666666, "status": 0.6666666666666666, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1193}, "ledger_by_hop": {"1": {"facts_total": 58, "facts_retained": 32, "retention_rate": 0.5517241379310345, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.7142857142857143, "amount": 0.6666666666666666, "status": 0.6666666666666666, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1200}, "2": {"facts_total": 58, "facts_retained": 32, "retention_rate": 0.5517241379310345, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.7142857142857143, "amount": 0.6666666666666666, "status": 0.6666666666666666, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1193}, "3": {"facts_total": 58, "facts_retained": 32, "retention_rate": 0.5517241379310345, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.7142857142857143, "amount": 0.6666666666666666, "status": 0.6666666666666666, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1193}, "4": {"facts_total": 58, "facts_retained": 32, "retention_rate": 0.5517241379310345, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.7142857142857143, "amount": 0.6666666666666666, "status": 0.6666666666666666, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1193}, "5": {"facts_total": 58, "facts_retained": 32, "retention_rate": 0.5517241379310345, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.7142857142857143, "amount": 0.6666666666666666, "status": 0.6666666666666666, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1193}, "6": {"facts_total": 58, "facts_retained": 32, "retention_rate": 0.5517241379310345, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.7142857142857143, "amount": 0.6666666666666666, "status": 0.6666666666666666, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1193}, "7": {"facts_total": 58, "facts_retained": 32, "retention_rate": 0.5517241379310345, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.7142857142857143, "amount": 0.6666666666666666, "status": 0.6666666666666666, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1193}, "8": {"facts_total": 58, "facts_retained": 32, "retention_rate": 0.5517241379310345, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.7142857142857143, "amount": 0.6666666666666666, "status": 0.6666666666666666, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1193}}}, "reference_key_id": "keys/synthetic/syn_03003_2000.json", "task_scores": {"correct": 0.0, "abstained": 0.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "72b629618498e6857b5e5c18413e492e46455d51f70d74a18ff4e73267a76b13", "output_hash": "e971a34f9c497d953f98fb387e189b1587b11a7f6725b2f81ac6d0f3ef8ac845"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "1684f9933d25bb91fd2c2365", "attempt_id": "62563a1313a3", "run_group_id": "E4_core:A4|closed|qwen3b:h8", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T06:40:48.643278Z", "task_family": "synthetic", "document_id": "syn_03003_2000", "document_hash": "dddca1a38e3224870fcfae927c701c400fd885c19123ea840ee06a76a140f407", "question_id": "q12_unit_conversion", "split": "test", "cluster_id": "syn_03003_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 8, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1866, "query_tokens": 603, "message_tokens": 512, "message_utf8_bytes": 1193, "cache_tensor_bytes": null, "serialized_payload_bytes": 1193, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 176987.9467500141, "hops": {"1": 22906.635708990507, "2": 22033.01479201764, "3": 21749.766708002426, "4": 22112.2178749647, "5": 21950.30216604937, "6": 22076.25124999322, "7": 22108.02762489766, "8": 22051.73062509857}, "receiver_prefill_ms": 376.6137500060722, "ttft_ms": 36.38979198876768, "decode_ms": 345.7265830365941}, "total_wall_ms": 177747.39591707475, "ttft_ms": 36.38979198876768, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 5}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "17430", "evidence_artifact": {"ledger": {"facts_total": 58, "facts_retained": 32, "retention_rate": 0.5517241379310345, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.7142857142857143, "amount": 0.6666666666666666, "status": 0.6666666666666666, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1193}, "ledger_by_hop": {"1": {"facts_total": 58, "facts_retained": 32, "retention_rate": 0.5517241379310345, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.7142857142857143, "amount": 0.6666666666666666, "status": 0.6666666666666666, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1200}, "2": {"facts_total": 58, "facts_retained": 32, "retention_rate": 0.5517241379310345, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.7142857142857143, "amount": 0.6666666666666666, "status": 0.6666666666666666, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1193}, "3": {"facts_total": 58, "facts_retained": 32, "retention_rate": 0.5517241379310345, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.7142857142857143, "amount": 0.6666666666666666, "status": 0.6666666666666666, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1193}, "4": {"facts_total": 58, "facts_retained": 32, "retention_rate": 0.5517241379310345, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.7142857142857143, "amount": 0.6666666666666666, "status": 0.6666666666666666, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1193}, "5": {"facts_total": 58, "facts_retained": 32, "retention_rate": 0.5517241379310345, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.7142857142857143, "amount": 0.6666666666666666, "status": 0.6666666666666666, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1193}, "6": {"facts_total": 58, "facts_retained": 32, "retention_rate": 0.5517241379310345, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.7142857142857143, "amount": 0.6666666666666666, "status": 0.6666666666666666, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1193}, "7": {"facts_total": 58, "facts_retained": 32, "retention_rate": 0.5517241379310345, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.7142857142857143, "amount": 0.6666666666666666, "status": 0.6666666666666666, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1193}, "8": {"facts_total": 58, "facts_retained": 32, "retention_rate": 0.5517241379310345, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.7142857142857143, "amount": 0.6666666666666666, "status": 0.6666666666666666, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1193}}}, "reference_key_id": "keys/synthetic/syn_03003_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "72b629618498e6857b5e5c18413e492e46455d51f70d74a18ff4e73267a76b13", "output_hash": "2575eaf3b748efc209b676518766751eb038c88f74578951345818bfe7f7194b"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "46584e3958242b421b4eb449", "attempt_id": "e783a3bd58ed", "run_group_id": "E4_core:A4|closed|qwen3b:h8", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T06:40:49.421343Z", "task_family": "synthetic", "document_id": "syn_03003_2000", "document_hash": "dddca1a38e3224870fcfae927c701c400fd885c19123ea840ee06a76a140f407", "question_id": "q13_unanswerable", "split": "test", "cluster_id": "syn_03003_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 8, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1866, "query_tokens": 607, "message_tokens": 512, "message_utf8_bytes": 1193, "cache_tensor_bytes": null, "serialized_payload_bytes": 1193, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 176987.9467500141, "hops": {"1": 22906.635708990507, "2": 22033.01479201764, "3": 21749.766708002426, "4": 22112.2178749647, "5": 21950.30216604937, "6": 22076.25124999322, "7": 22108.02762489766, "8": 22051.73062509857}, "receiver_prefill_ms": 403.54608302004635, "ttft_ms": 39.90220802370459, "decode_ms": 25.70958307478577}, "total_wall_ms": 177457.87975005805, "ttft_ms": 39.90220802370459, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 58, "facts_retained": 32, "retention_rate": 0.5517241379310345, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.7142857142857143, "amount": 0.6666666666666666, "status": 0.6666666666666666, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1193}, "ledger_by_hop": {"1": {"facts_total": 58, "facts_retained": 32, "retention_rate": 0.5517241379310345, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.7142857142857143, "amount": 0.6666666666666666, "status": 0.6666666666666666, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1200}, "2": {"facts_total": 58, "facts_retained": 32, "retention_rate": 0.5517241379310345, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.7142857142857143, "amount": 0.6666666666666666, "status": 0.6666666666666666, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1193}, "3": {"facts_total": 58, "facts_retained": 32, "retention_rate": 0.5517241379310345, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.7142857142857143, "amount": 0.6666666666666666, "status": 0.6666666666666666, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1193}, "4": {"facts_total": 58, "facts_retained": 32, "retention_rate": 0.5517241379310345, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.7142857142857143, "amount": 0.6666666666666666, "status": 0.6666666666666666, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1193}, "5": {"facts_total": 58, "facts_retained": 32, "retention_rate": 0.5517241379310345, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.7142857142857143, "amount": 0.6666666666666666, "status": 0.6666666666666666, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1193}, "6": {"facts_total": 58, "facts_retained": 32, "retention_rate": 0.5517241379310345, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.7142857142857143, "amount": 0.6666666666666666, "status": 0.6666666666666666, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1193}, "7": {"facts_total": 58, "facts_retained": 32, "retention_rate": 0.5517241379310345, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.7142857142857143, "amount": 0.6666666666666666, "status": 0.6666666666666666, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1193}, "8": {"facts_total": 58, "facts_retained": 32, "retention_rate": 0.5517241379310345, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.7142857142857143, "amount": 0.6666666666666666, "status": 0.6666666666666666, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1193}}}, "reference_key_id": "keys/synthetic/syn_03003_2000.json", "task_scores": {"correct": 1.0, "abstained": 1.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "72b629618498e6857b5e5c18413e492e46455d51f70d74a18ff4e73267a76b13", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "a3199586a24d2f16d23cbb6a", "attempt_id": "b981546f4c7d", "run_group_id": "E4_core:A4|closed|qwen3b:h8", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T06:40:49.908179Z", "task_family": "synthetic", "document_id": "syn_03003_2000", "document_hash": "dddca1a38e3224870fcfae927c701c400fd885c19123ea840ee06a76a140f407", "question_id": "q14_unanswerable", "split": "test", "cluster_id": "syn_03003_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 8, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1866, "query_tokens": 608, "message_tokens": 512, "message_utf8_bytes": 1193, "cache_tensor_bytes": null, "serialized_payload_bytes": 1193, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 176987.9467500141, "hops": {"1": 22906.635708990507, "2": 22033.01479201764, "3": 21749.766708002426, "4": 22112.2178749647, "5": 21950.30216604937, "6": 22076.25124999322, "7": 22108.02762489766, "8": 22051.73062509857}, "receiver_prefill_ms": 405.3830410121009, "ttft_ms": 24.380667018704116, "decode_ms": 17.673959024250507}, "total_wall_ms": 177436.0474590212, "ttft_ms": 24.380667018704116, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 58, "facts_retained": 32, "retention_rate": 0.5517241379310345, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.7142857142857143, "amount": 0.6666666666666666, "status": 0.6666666666666666, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1193}, "ledger_by_hop": {"1": {"facts_total": 58, "facts_retained": 32, "retention_rate": 0.5517241379310345, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.7142857142857143, "amount": 0.6666666666666666, "status": 0.6666666666666666, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1200}, "2": {"facts_total": 58, "facts_retained": 32, "retention_rate": 0.5517241379310345, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.7142857142857143, "amount": 0.6666666666666666, "status": 0.6666666666666666, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1193}, "3": {"facts_total": 58, "facts_retained": 32, "retention_rate": 0.5517241379310345, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.7142857142857143, "amount": 0.6666666666666666, "status": 0.6666666666666666, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1193}, "4": {"facts_total": 58, "facts_retained": 32, "retention_rate": 0.5517241379310345, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.7142857142857143, "amount": 0.6666666666666666, "status": 0.6666666666666666, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1193}, "5": {"facts_total": 58, "facts_retained": 32, "retention_rate": 0.5517241379310345, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.7142857142857143, "amount": 0.6666666666666666, "status": 0.6666666666666666, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1193}, "6": {"facts_total": 58, "facts_retained": 32, "retention_rate": 0.5517241379310345, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.7142857142857143, "amount": 0.6666666666666666, "status": 0.6666666666666666, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1193}, "7": {"facts_total": 58, "facts_retained": 32, "retention_rate": 0.5517241379310345, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.7142857142857143, "amount": 0.6666666666666666, "status": 0.6666666666666666, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1193}, "8": {"facts_total": 58, "facts_retained": 32, "retention_rate": 0.5517241379310345, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.7142857142857143, "amount": 0.6666666666666666, "status": 0.6666666666666666, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1193}}}, "reference_key_id": "keys/synthetic/syn_03003_2000.json", "task_scores": {"correct": 1.0, "abstained": 1.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "72b629618498e6857b5e5c18413e492e46455d51f70d74a18ff4e73267a76b13", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "7c42592645bd7e37b0f51d5f", "attempt_id": "3f172b44fcac", "run_group_id": "E4_core:A5|closed|qwen3b:h1", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T06:41:13.346786Z", "task_family": "synthetic", "document_id": "syn_03003_2000", "document_hash": "dddca1a38e3224870fcfae927c701c400fd885c19123ea840ee06a76a140f407", "question_id": "q00_single_fact", "split": "test", "cluster_id": "syn_03003_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 1, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1866, "query_tokens": 435, "message_tokens": 352, "message_utf8_bytes": 830, "cache_tensor_bytes": null, "serialized_payload_bytes": 830, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 22970.73791595176, "hops": {"1": 22970.73791595176}, "receiver_prefill_ms": 303.74833394307643, "ttft_ms": 22.769083036109805, "decode_ms": 303.6712909815833}, "total_wall_ms": 23601.56083293259, "ttft_ms": 22.769083036109805, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 10}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "GR-79S-5839", "evidence_artifact": {"ledger": {"facts_total": 58, "facts_retained": 15, "retention_rate": 0.25862068965517243, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.2222222222222222, "status": 0.75, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 766}, "ledger_by_hop": {"1": {"facts_total": 58, "facts_retained": 15, "retention_rate": 0.25862068965517243, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.2222222222222222, "status": 0.75, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 766}}}, "reference_key_id": "keys/synthetic/syn_03003_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "a613cf19d60b1dbe7d9f51bbbf98459d9f181f51f14c39d755912e891ae7fd2f", "output_hash": "720cc40cd69e85e9728f6bea00b17a95ca07eee643e143a131840763958303cd"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "9f6afddc67a279474a3515ab", "attempt_id": "cda8ce4cb746", "run_group_id": "E4_core:A5|closed|qwen3b:h1", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T06:41:13.995996Z", "task_family": "synthetic", "document_id": "syn_03003_2000", "document_hash": "dddca1a38e3224870fcfae927c701c400fd885c19123ea840ee06a76a140f407", "question_id": "q01_single_fact", "split": "test", "cluster_id": "syn_03003_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 1, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1866, "query_tokens": 439, "message_tokens": 352, "message_utf8_bytes": 830, "cache_tensor_bytes": null, "serialized_payload_bytes": 830, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 22970.73791595176, "hops": {"1": 22970.73791595176}, "receiver_prefill_ms": 334.48170800693333, "ttft_ms": 57.07379092928022, "decode_ms": 94.43229204043746}, "total_wall_ms": 23457.379957893863, "ttft_ms": 57.07379092928022, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 58, "facts_retained": 15, "retention_rate": 0.25862068965517243, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.2222222222222222, "status": 0.75, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 766}, "ledger_by_hop": {"1": {"facts_total": 58, "facts_retained": 15, "retention_rate": 0.25862068965517243, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.2222222222222222, "status": 0.75, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 766}}}, "reference_key_id": "keys/synthetic/syn_03003_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "a613cf19d60b1dbe7d9f51bbbf98459d9f181f51f14c39d755912e891ae7fd2f", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "27f819d307eb16341b41894a", "attempt_id": "8062c6fdcbbc", "run_group_id": "E4_core:A5|closed|qwen3b:h1", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T06:41:14.500925Z", "task_family": "synthetic", "document_id": "syn_03003_2000", "document_hash": "dddca1a38e3224870fcfae927c701c400fd885c19123ea840ee06a76a140f407", "question_id": "q02_exact_copy", "split": "test", "cluster_id": "syn_03003_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 1, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1866, "query_tokens": 439, "message_tokens": 352, "message_utf8_bytes": 830, "cache_tensor_bytes": null, "serialized_payload_bytes": 830, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 22970.73791595176, "hops": {"1": 22970.73791595176}, "receiver_prefill_ms": 373.5436659771949, "ttft_ms": 42.03104192856699, "decode_ms": 23.680082987993956}, "total_wall_ms": 23410.457540885545, "ttft_ms": 42.03104192856699, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 58, "facts_retained": 15, "retention_rate": 0.25862068965517243, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.2222222222222222, "status": 0.75, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 766}, "ledger_by_hop": {"1": {"facts_total": 58, "facts_retained": 15, "retention_rate": 0.25862068965517243, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.2222222222222222, "status": 0.75, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 766}}}, "reference_key_id": "keys/synthetic/syn_03003_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "a613cf19d60b1dbe7d9f51bbbf98459d9f181f51f14c39d755912e891ae7fd2f", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "d801a135865034b5de954b22", "attempt_id": "983a031bce89", "run_group_id": "E4_core:A5|closed|qwen3b:h1", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T06:41:14.959993Z", "task_family": "synthetic", "document_id": "syn_03003_2000", "document_hash": "dddca1a38e3224870fcfae927c701c400fd885c19123ea840ee06a76a140f407", "question_id": "q03_exact_copy", "split": "test", "cluster_id": "syn_03003_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 1, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1866, "query_tokens": 434, "message_tokens": 352, "message_utf8_bytes": 830, "cache_tensor_bytes": null, "serialized_payload_bytes": 830, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 22970.73791595176, "hops": {"1": 22970.73791595176}, "receiver_prefill_ms": 288.13295799773186, "ttft_ms": 24.545625085011125, "decode_ms": 36.7517078993842}, "total_wall_ms": 23320.727581973188, "ttft_ms": 24.545625085011125, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 58, "facts_retained": 15, "retention_rate": 0.25862068965517243, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.2222222222222222, "status": 0.75, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 766}, "ledger_by_hop": {"1": {"facts_total": 58, "facts_retained": 15, "retention_rate": 0.25862068965517243, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.2222222222222222, "status": 0.75, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 766}}}, "reference_key_id": "keys/synthetic/syn_03003_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "a613cf19d60b1dbe7d9f51bbbf98459d9f181f51f14c39d755912e891ae7fd2f", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "683ef6eedf7ceff990704571", "attempt_id": "0174e714645f", "run_group_id": "E4_core:A5|closed|qwen3b:h1", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T06:41:15.327212Z", "task_family": "synthetic", "document_id": "syn_03003_2000", "document_hash": "dddca1a38e3224870fcfae927c701c400fd885c19123ea840ee06a76a140f407", "question_id": "q04_multi_hop", "split": "test", "cluster_id": "syn_03003_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 1, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1866, "query_tokens": 443, "message_tokens": 352, "message_utf8_bytes": 830, "cache_tensor_bytes": null, "serialized_payload_bytes": 830, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 22970.73791595176, "hops": {"1": 22970.73791595176}, "receiver_prefill_ms": 287.48299996368587, "ttft_ms": 14.60891601163894, "decode_ms": 11.813833960331976}, "total_wall_ms": 23285.1275409339, "ttft_ms": 14.60891601163894, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 58, "facts_retained": 15, "retention_rate": 0.25862068965517243, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.2222222222222222, "status": 0.75, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 766}, "ledger_by_hop": {"1": {"facts_total": 58, "facts_retained": 15, "retention_rate": 0.25862068965517243, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.2222222222222222, "status": 0.75, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 766}}}, "reference_key_id": "keys/synthetic/syn_03003_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "a613cf19d60b1dbe7d9f51bbbf98459d9f181f51f14c39d755912e891ae7fd2f", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "eb3d86b51a478dcefece3135", "attempt_id": "71961e30737d", "run_group_id": "E4_core:A5|closed|qwen3b:h1", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T06:41:15.658808Z", "task_family": "synthetic", "document_id": "syn_03003_2000", "document_hash": "dddca1a38e3224870fcfae927c701c400fd885c19123ea840ee06a76a140f407", "question_id": "q05_multi_hop", "split": "test", "cluster_id": "syn_03003_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 1, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1866, "query_tokens": 445, "message_tokens": 352, "message_utf8_bytes": 830, "cache_tensor_bytes": null, "serialized_payload_bytes": 830, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 22970.73791595176, "hops": {"1": 22970.73791595176}, "receiver_prefill_ms": 296.10433301422745, "ttft_ms": 37.25925006438047, "decode_ms": 62.05404095817357}, "total_wall_ms": 23367.039415985346, "ttft_ms": 37.25925006438047, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 58, "facts_retained": 15, "retention_rate": 0.25862068965517243, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.2222222222222222, "status": 0.75, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 766}, "ledger_by_hop": {"1": {"facts_total": 58, "facts_retained": 15, "retention_rate": 0.25862068965517243, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.2222222222222222, "status": 0.75, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 766}}}, "reference_key_id": "keys/synthetic/syn_03003_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "a613cf19d60b1dbe7d9f51bbbf98459d9f181f51f14c39d755912e891ae7fd2f", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "3ac94741b2c515b8c45c0391", "attempt_id": "5965747b5d96", "run_group_id": "E4_core:A5|closed|qwen3b:h1", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T06:41:16.075819Z", "task_family": "synthetic", "document_id": "syn_03003_2000", "document_hash": "dddca1a38e3224870fcfae927c701c400fd885c19123ea840ee06a76a140f407", "question_id": "q06_negation", "split": "test", "cluster_id": "syn_03003_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 1, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1866, "query_tokens": 437, "message_tokens": 352, "message_utf8_bytes": 830, "cache_tensor_bytes": null, "serialized_payload_bytes": 830, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 22970.73791595176, "hops": {"1": 22970.73791595176}, "receiver_prefill_ms": 429.8334170598537, "ttft_ms": 38.32054208032787, "decode_ms": 31.410375027917325}, "total_wall_ms": 23470.996415941045, "ttft_ms": 38.32054208032787, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "no", "evidence_artifact": {"ledger": {"facts_total": 58, "facts_retained": 15, "retention_rate": 0.25862068965517243, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.2222222222222222, "status": 0.75, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 766}, "ledger_by_hop": {"1": {"facts_total": 58, "facts_retained": 15, "retention_rate": 0.25862068965517243, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.2222222222222222, "status": 0.75, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 766}}}, "reference_key_id": "keys/synthetic/syn_03003_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "a613cf19d60b1dbe7d9f51bbbf98459d9f181f51f14c39d755912e891ae7fd2f", "output_hash": "9390298f3fb0c5b160498935d79cb139aef28e1c47358b4bbba61862b9c26e59"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "628c5fa308db3d2085f3ddf6", "attempt_id": "cfbd095eaffc", "run_group_id": "E4_core:A5|closed|qwen3b:h1", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T06:41:16.593363Z", "task_family": "synthetic", "document_id": "syn_03003_2000", "document_hash": "dddca1a38e3224870fcfae927c701c400fd885c19123ea840ee06a76a140f407", "question_id": "q07_negation", "split": "test", "cluster_id": "syn_03003_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 1, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1866, "query_tokens": 434, "message_tokens": 352, "message_utf8_bytes": 830, "cache_tensor_bytes": null, "serialized_payload_bytes": 830, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 22970.73791595176, "hops": {"1": 22970.73791595176}, "receiver_prefill_ms": 291.2692080717534, "ttft_ms": 41.75012500490993, "decode_ms": 30.331416055560112}, "total_wall_ms": 23334.830207983032, "ttft_ms": 41.75012500490993, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 58, "facts_retained": 15, "retention_rate": 0.25862068965517243, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.2222222222222222, "status": 0.75, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 766}, "ledger_by_hop": {"1": {"facts_total": 58, "facts_retained": 15, "retention_rate": 0.25862068965517243, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.2222222222222222, "status": 0.75, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 766}}}, "reference_key_id": "keys/synthetic/syn_03003_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "a613cf19d60b1dbe7d9f51bbbf98459d9f181f51f14c39d755912e891ae7fd2f", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "0bab48ddd5bdeed3d2fd5055", "attempt_id": "462b08001042", "run_group_id": "E4_core:A5|closed|qwen3b:h1", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T06:41:16.975922Z", "task_family": "synthetic", "document_id": "syn_03003_2000", "document_hash": "dddca1a38e3224870fcfae927c701c400fd885c19123ea840ee06a76a140f407", "question_id": "q08_revision", "split": "test", "cluster_id": "syn_03003_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 1, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1866, "query_tokens": 452, "message_tokens": 352, "message_utf8_bytes": 830, "cache_tensor_bytes": null, "serialized_payload_bytes": 830, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 22970.73791595176, "hops": {"1": 22970.73791595176}, "receiver_prefill_ms": 313.34429199341685, "ttft_ms": 23.755750036798418, "decode_ms": 366.53808306436986}, "total_wall_ms": 23674.968290957622, "ttft_ms": 23.755750036798418, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 7}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "911,925", "evidence_artifact": {"ledger": {"facts_total": 58, "facts_retained": 15, "retention_rate": 0.25862068965517243, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.2222222222222222, "status": 0.75, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 766}, "ledger_by_hop": {"1": {"facts_total": 58, "facts_retained": 15, "retention_rate": 0.25862068965517243, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.2222222222222222, "status": 0.75, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 766}}}, "reference_key_id": "keys/synthetic/syn_03003_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "a613cf19d60b1dbe7d9f51bbbf98459d9f181f51f14c39d755912e891ae7fd2f", "output_hash": "86be78720fc4b79f93e603e7d9e4861fcd25102612f73209fffc17d787056007"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "dbf56d401f8e40582f21eb19", "attempt_id": "b73e20dadf88", "run_group_id": "E4_core:A5|closed|qwen3b:h1", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T06:41:17.696723Z", "task_family": "synthetic", "document_id": "syn_03003_2000", "document_hash": "dddca1a38e3224870fcfae927c701c400fd885c19123ea840ee06a76a140f407", "question_id": "q09_revision", "split": "test", "cluster_id": "syn_03003_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 1, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1866, "query_tokens": 445, "message_tokens": 352, "message_utf8_bytes": 830, "cache_tensor_bytes": null, "serialized_payload_bytes": 830, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 22970.73791595176, "hops": {"1": 22970.73791595176}, "receiver_prefill_ms": 410.92775005381554, "ttft_ms": 38.16108300816268, "decode_ms": 196.65658404119313}, "total_wall_ms": 23617.295666015707, "ttft_ms": 38.16108300816268, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 6}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "911925", "evidence_artifact": {"ledger": {"facts_total": 58, "facts_retained": 15, "retention_rate": 0.25862068965517243, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.2222222222222222, "status": 0.75, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 766}, "ledger_by_hop": {"1": {"facts_total": 58, "facts_retained": 15, "retention_rate": 0.25862068965517243, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.2222222222222222, "status": 0.75, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 766}}}, "reference_key_id": "keys/synthetic/syn_03003_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "a613cf19d60b1dbe7d9f51bbbf98459d9f181f51f14c39d755912e891ae7fd2f", "output_hash": "679a540dc4739f0ace623448b8fd925fb234f554838d7e87da223228ee81b112"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "0b0c6e5a6855699274b7c277", "attempt_id": "ada4bcd4c3cb", "run_group_id": "E4_core:A5|closed|qwen3b:h1", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T06:41:18.361091Z", "task_family": "synthetic", "document_id": "syn_03003_2000", "document_hash": "dddca1a38e3224870fcfae927c701c400fd885c19123ea840ee06a76a140f407", "question_id": "q10_temporal", "split": "test", "cluster_id": "syn_03003_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 1, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1866, "query_tokens": 452, "message_tokens": 352, "message_utf8_bytes": 830, "cache_tensor_bytes": null, "serialized_payload_bytes": 830, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 22970.73791595176, "hops": {"1": 22970.73791595176}, "receiver_prefill_ms": 326.5096670947969, "ttft_ms": 31.632250058464706, "decode_ms": 29.559749993495643}, "total_wall_ms": 23358.935540891252, "ttft_ms": 31.632250058464706, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 58, "facts_retained": 15, "retention_rate": 0.25862068965517243, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.2222222222222222, "status": 0.75, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 766}, "ledger_by_hop": {"1": {"facts_total": 58, "facts_retained": 15, "retention_rate": 0.25862068965517243, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.2222222222222222, "status": 0.75, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 766}}}, "reference_key_id": "keys/synthetic/syn_03003_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "a613cf19d60b1dbe7d9f51bbbf98459d9f181f51f14c39d755912e891ae7fd2f", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "701d9fab0e6584fe171a8d3b", "attempt_id": "3c27f5c2e026", "run_group_id": "E4_core:A5|closed|qwen3b:h1", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T06:41:18.766403Z", "task_family": "synthetic", "document_id": "syn_03003_2000", "document_hash": "dddca1a38e3224870fcfae927c701c400fd885c19123ea840ee06a76a140f407", "question_id": "q11_aggregation", "split": "test", "cluster_id": "syn_03003_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 1, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1866, "query_tokens": 446, "message_tokens": 352, "message_utf8_bytes": 830, "cache_tensor_bytes": null, "serialized_payload_bytes": 830, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 22970.73791595176, "hops": {"1": 22970.73791595176}, "receiver_prefill_ms": 310.5568749597296, "ttft_ms": 48.042999929748476, "decode_ms": 54.72745792940259}, "total_wall_ms": 23384.75412398111, "ttft_ms": 48.042999929748476, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 58, "facts_retained": 15, "retention_rate": 0.25862068965517243, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.2222222222222222, "status": 0.75, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 766}, "ledger_by_hop": {"1": {"facts_total": 58, "facts_retained": 15, "retention_rate": 0.25862068965517243, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.2222222222222222, "status": 0.75, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 766}}}, "reference_key_id": "keys/synthetic/syn_03003_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "a613cf19d60b1dbe7d9f51bbbf98459d9f181f51f14c39d755912e891ae7fd2f", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "861272397043d414c316031a", "attempt_id": "cff59aa53fb6", "run_group_id": "E4_core:A5|closed|qwen3b:h1", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T06:41:19.197741Z", "task_family": "synthetic", "document_id": "syn_03003_2000", "document_hash": "dddca1a38e3224870fcfae927c701c400fd885c19123ea840ee06a76a140f407", "question_id": "q12_unit_conversion", "split": "test", "cluster_id": "syn_03003_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 1, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1866, "query_tokens": 443, "message_tokens": 352, "message_utf8_bytes": 830, "cache_tensor_bytes": null, "serialized_payload_bytes": 830, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 22970.73791595176, "hops": {"1": 22970.73791595176}, "receiver_prefill_ms": 427.549833082594, "ttft_ms": 24.35295796021819, "decode_ms": 170.79858295619488}, "total_wall_ms": 23593.94366585184, "ttft_ms": 24.35295796021819, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 5}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "17430", "evidence_artifact": {"ledger": {"facts_total": 58, "facts_retained": 15, "retention_rate": 0.25862068965517243, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.2222222222222222, "status": 0.75, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 766}, "ledger_by_hop": {"1": {"facts_total": 58, "facts_retained": 15, "retention_rate": 0.25862068965517243, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.2222222222222222, "status": 0.75, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 766}}}, "reference_key_id": "keys/synthetic/syn_03003_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "a613cf19d60b1dbe7d9f51bbbf98459d9f181f51f14c39d755912e891ae7fd2f", "output_hash": "2575eaf3b748efc209b676518766751eb038c88f74578951345818bfe7f7194b"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "38d74dda6ff788760ce0dd3b", "attempt_id": "fca819f04e01", "run_group_id": "E4_core:A5|closed|qwen3b:h1", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T06:41:19.839062Z", "task_family": "synthetic", "document_id": "syn_03003_2000", "document_hash": "dddca1a38e3224870fcfae927c701c400fd885c19123ea840ee06a76a140f407", "question_id": "q13_unanswerable", "split": "test", "cluster_id": "syn_03003_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 1, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1866, "query_tokens": 447, "message_tokens": 352, "message_utf8_bytes": 830, "cache_tensor_bytes": null, "serialized_payload_bytes": 830, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 22970.73791595176, "hops": {"1": 22970.73791595176}, "receiver_prefill_ms": 301.3527080183849, "ttft_ms": 26.857250020839274, "decode_ms": 38.206166005693376}, "total_wall_ms": 23337.625249987468, "ttft_ms": 26.857250020839274, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 58, "facts_retained": 15, "retention_rate": 0.25862068965517243, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.2222222222222222, "status": 0.75, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 766}, "ledger_by_hop": {"1": {"facts_total": 58, "facts_retained": 15, "retention_rate": 0.25862068965517243, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.2222222222222222, "status": 0.75, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 766}}}, "reference_key_id": "keys/synthetic/syn_03003_2000.json", "task_scores": {"correct": 1.0, "abstained": 1.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "a613cf19d60b1dbe7d9f51bbbf98459d9f181f51f14c39d755912e891ae7fd2f", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "c66adedfe3b8be4273b7e6da", "attempt_id": "db8b4f11b21a", "run_group_id": "E4_core:A5|closed|qwen3b:h1", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T06:41:20.222491Z", "task_family": "synthetic", "document_id": "syn_03003_2000", "document_hash": "dddca1a38e3224870fcfae927c701c400fd885c19123ea840ee06a76a140f407", "question_id": "q14_unanswerable", "split": "test", "cluster_id": "syn_03003_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 1, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1866, "query_tokens": 448, "message_tokens": 352, "message_utf8_bytes": 830, "cache_tensor_bytes": null, "serialized_payload_bytes": 830, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 22970.73791595176, "hops": {"1": 22970.73791595176}, "receiver_prefill_ms": 273.2112919911742, "ttft_ms": 17.448875005356967, "decode_ms": 49.00095798075199}, "total_wall_ms": 23310.942207928747, "ttft_ms": 17.448875005356967, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 58, "facts_retained": 15, "retention_rate": 0.25862068965517243, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.2222222222222222, "status": 0.75, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 766}, "ledger_by_hop": {"1": {"facts_total": 58, "facts_retained": 15, "retention_rate": 0.25862068965517243, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.2222222222222222, "status": 0.75, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 766}}}, "reference_key_id": "keys/synthetic/syn_03003_2000.json", "task_scores": {"correct": 1.0, "abstained": 1.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "a613cf19d60b1dbe7d9f51bbbf98459d9f181f51f14c39d755912e891ae7fd2f", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "3fa080bf603e07ab93cfd1d4", "attempt_id": "0df66927a4dc", "run_group_id": "E4_core:A5|closed|qwen3b:h2", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T06:41:36.377822Z", "task_family": "synthetic", "document_id": "syn_03003_2000", "document_hash": "dddca1a38e3224870fcfae927c701c400fd885c19123ea840ee06a76a140f407", "question_id": "q00_single_fact", "split": "test", "cluster_id": "syn_03003_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 2, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1866, "query_tokens": 435, "message_tokens": 352, "message_utf8_bytes": 830, "cache_tensor_bytes": null, "serialized_payload_bytes": 830, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 38767.85674900748, "hops": {"1": 22970.73791595176, "2": 15797.11883305572}, "receiver_prefill_ms": 292.0515420846641, "ttft_ms": 39.32262503076345, "decode_ms": 322.8866670979187}, "total_wall_ms": 39423.08324901387, "ttft_ms": 39.32262503076345, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 10}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "GR-79S-5839", "evidence_artifact": {"ledger": {"facts_total": 58, "facts_retained": 15, "retention_rate": 0.25862068965517243, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.2222222222222222, "status": 0.75, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 766}, "ledger_by_hop": {"1": {"facts_total": 58, "facts_retained": 15, "retention_rate": 0.25862068965517243, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.2222222222222222, "status": 0.75, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 766}, "2": {"facts_total": 58, "facts_retained": 15, "retention_rate": 0.25862068965517243, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.2222222222222222, "status": 0.75, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 766}}}, "reference_key_id": "keys/synthetic/syn_03003_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "a613cf19d60b1dbe7d9f51bbbf98459d9f181f51f14c39d755912e891ae7fd2f", "output_hash": "720cc40cd69e85e9728f6bea00b17a95ca07eee643e143a131840763958303cd"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "a354d7689f83d73ba6a529d6", "attempt_id": "7d97758add85", "run_group_id": "E4_core:A5|closed|qwen3b:h2", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T06:41:37.050751Z", "task_family": "synthetic", "document_id": "syn_03003_2000", "document_hash": "dddca1a38e3224870fcfae927c701c400fd885c19123ea840ee06a76a140f407", "question_id": "q01_single_fact", "split": "test", "cluster_id": "syn_03003_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 2, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1866, "query_tokens": 439, "message_tokens": 352, "message_utf8_bytes": 830, "cache_tensor_bytes": null, "serialized_payload_bytes": 830, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 38767.85674900748, "hops": {"1": 22970.73791595176, "2": 15797.11883305572}, "receiver_prefill_ms": 292.94249997474253, "ttft_ms": 51.744624972343445, "decode_ms": 91.27125004306436}, "total_wall_ms": 39204.45104094688, "ttft_ms": 51.744624972343445, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 58, "facts_retained": 15, "retention_rate": 0.25862068965517243, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.2222222222222222, "status": 0.75, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 766}, "ledger_by_hop": {"1": {"facts_total": 58, "facts_retained": 15, "retention_rate": 0.25862068965517243, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.2222222222222222, "status": 0.75, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 766}, "2": {"facts_total": 58, "facts_retained": 15, "retention_rate": 0.25862068965517243, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.2222222222222222, "status": 0.75, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 766}}}, "reference_key_id": "keys/synthetic/syn_03003_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "a613cf19d60b1dbe7d9f51bbbf98459d9f181f51f14c39d755912e891ae7fd2f", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "0c7c1c50b6904e021ed47eb6", "attempt_id": "9cafd080598d", "run_group_id": "E4_core:A5|closed|qwen3b:h2", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T06:41:37.504617Z", "task_family": "synthetic", "document_id": "syn_03003_2000", "document_hash": "dddca1a38e3224870fcfae927c701c400fd885c19123ea840ee06a76a140f407", "question_id": "q02_exact_copy", "split": "test", "cluster_id": "syn_03003_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 2, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1866, "query_tokens": 439, "message_tokens": 352, "message_utf8_bytes": 830, "cache_tensor_bytes": null, "serialized_payload_bytes": 830, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 38767.85674900748, "hops": {"1": 22970.73791595176, "2": 15797.11883305572}, "receiver_prefill_ms": 375.313083990477, "ttft_ms": 39.61625008378178, "decode_ms": 24.70549999270588}, "total_wall_ms": 39208.11991603114, "ttft_ms": 39.61625008378178, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 58, "facts_retained": 15, "retention_rate": 0.25862068965517243, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.2222222222222222, "status": 0.75, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 766}, "ledger_by_hop": {"1": {"facts_total": 58, "facts_retained": 15, "retention_rate": 0.25862068965517243, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.2222222222222222, "status": 0.75, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 766}, "2": {"facts_total": 58, "facts_retained": 15, "retention_rate": 0.25862068965517243, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.2222222222222222, "status": 0.75, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 766}}}, "reference_key_id": "keys/synthetic/syn_03003_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "a613cf19d60b1dbe7d9f51bbbf98459d9f181f51f14c39d755912e891ae7fd2f", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "5f3d353ce8f9923000b3cd23", "attempt_id": "ed323bc4abca", "run_group_id": "E4_core:A5|closed|qwen3b:h2", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T06:41:37.964339Z", "task_family": "synthetic", "document_id": "syn_03003_2000", "document_hash": "dddca1a38e3224870fcfae927c701c400fd885c19123ea840ee06a76a140f407", "question_id": "q03_exact_copy", "split": "test", "cluster_id": "syn_03003_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 2, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1866, "query_tokens": 434, "message_tokens": 352, "message_utf8_bytes": 830, "cache_tensor_bytes": null, "serialized_payload_bytes": 830, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 38767.85674900748, "hops": {"1": 22970.73791595176, "2": 15797.11883305572}, "receiver_prefill_ms": 299.3129580281675, "ttft_ms": 40.65995803102851, "decode_ms": 24.86683300230652}, "total_wall_ms": 39133.24533298146, "ttft_ms": 40.65995803102851, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 58, "facts_retained": 15, "retention_rate": 0.25862068965517243, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.2222222222222222, "status": 0.75, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 766}, "ledger_by_hop": {"1": {"facts_total": 58, "facts_retained": 15, "retention_rate": 0.25862068965517243, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.2222222222222222, "status": 0.75, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 766}, "2": {"facts_total": 58, "facts_retained": 15, "retention_rate": 0.25862068965517243, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.2222222222222222, "status": 0.75, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 766}}}, "reference_key_id": "keys/synthetic/syn_03003_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "a613cf19d60b1dbe7d9f51bbbf98459d9f181f51f14c39d755912e891ae7fd2f", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "1e139dedc1f012370801463f", "attempt_id": "8d0075d93736", "run_group_id": "E4_core:A5|closed|qwen3b:h2", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T06:41:38.347217Z", "task_family": "synthetic", "document_id": "syn_03003_2000", "document_hash": "dddca1a38e3224870fcfae927c701c400fd885c19123ea840ee06a76a140f407", "question_id": "q04_multi_hop", "split": "test", "cluster_id": "syn_03003_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 2, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1866, "query_tokens": 443, "message_tokens": 352, "message_utf8_bytes": 830, "cache_tensor_bytes": null, "serialized_payload_bytes": 830, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 38767.85674900748, "hops": {"1": 22970.73791595176, "2": 15797.11883305572}, "receiver_prefill_ms": 279.65341706294566, "ttft_ms": 51.60937493201345, "decode_ms": 51.2202080572024}, "total_wall_ms": 39151.02870797273, "ttft_ms": 51.60937493201345, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 58, "facts_retained": 15, "retention_rate": 0.25862068965517243, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.2222222222222222, "status": 0.75, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 766}, "ledger_by_hop": {"1": {"facts_total": 58, "facts_retained": 15, "retention_rate": 0.25862068965517243, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.2222222222222222, "status": 0.75, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 766}, "2": {"facts_total": 58, "facts_retained": 15, "retention_rate": 0.25862068965517243, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.2222222222222222, "status": 0.75, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 766}}}, "reference_key_id": "keys/synthetic/syn_03003_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "a613cf19d60b1dbe7d9f51bbbf98459d9f181f51f14c39d755912e891ae7fd2f", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "cbd7f112a04e6aa69d589a18", "attempt_id": "a21d3ce5caec", "run_group_id": "E4_core:A5|closed|qwen3b:h2", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T06:41:38.748019Z", "task_family": "synthetic", "document_id": "syn_03003_2000", "document_hash": "dddca1a38e3224870fcfae927c701c400fd885c19123ea840ee06a76a140f407", "question_id": "q05_multi_hop", "split": "test", "cluster_id": "syn_03003_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 2, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1866, "query_tokens": 445, "message_tokens": 352, "message_utf8_bytes": 830, "cache_tensor_bytes": null, "serialized_payload_bytes": 830, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 38767.85674900748, "hops": {"1": 22970.73791595176, "2": 15797.11883305572}, "receiver_prefill_ms": 293.8662909436971, "ttft_ms": 91.9795410009101, "decode_ms": 60.43595902156085}, "total_wall_ms": 39214.654040988535, "ttft_ms": 91.9795410009101, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 58, "facts_retained": 15, "retention_rate": 0.25862068965517243, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.2222222222222222, "status": 0.75, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 766}, "ledger_by_hop": {"1": {"facts_total": 58, "facts_retained": 15, "retention_rate": 0.25862068965517243, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.2222222222222222, "status": 0.75, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 766}, "2": {"facts_total": 58, "facts_retained": 15, "retention_rate": 0.25862068965517243, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.2222222222222222, "status": 0.75, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 766}}}, "reference_key_id": "keys/synthetic/syn_03003_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "a613cf19d60b1dbe7d9f51bbbf98459d9f181f51f14c39d755912e891ae7fd2f", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "f0de80f1abd8037ce43e1680", "attempt_id": "79e25e6b201c", "run_group_id": "E4_core:A5|closed|qwen3b:h2", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T06:41:39.211617Z", "task_family": "synthetic", "document_id": "syn_03003_2000", "document_hash": "dddca1a38e3224870fcfae927c701c400fd885c19123ea840ee06a76a140f407", "question_id": "q06_negation", "split": "test", "cluster_id": "syn_03003_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 2, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1866, "query_tokens": 437, "message_tokens": 352, "message_utf8_bytes": 830, "cache_tensor_bytes": null, "serialized_payload_bytes": 830, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 38767.85674900748, "hops": {"1": 22970.73791595176, "2": 15797.11883305572}, "receiver_prefill_ms": 371.590334107168, "ttft_ms": 23.480208008550107, "decode_ms": 38.66695798933506}, "total_wall_ms": 39202.07804092206, "ttft_ms": 23.480208008550107, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "no", "evidence_artifact": {"ledger": {"facts_total": 58, "facts_retained": 15, "retention_rate": 0.25862068965517243, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.2222222222222222, "status": 0.75, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 766}, "ledger_by_hop": {"1": {"facts_total": 58, "facts_retained": 15, "retention_rate": 0.25862068965517243, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.2222222222222222, "status": 0.75, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 766}, "2": {"facts_total": 58, "facts_retained": 15, "retention_rate": 0.25862068965517243, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.2222222222222222, "status": 0.75, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 766}}}, "reference_key_id": "keys/synthetic/syn_03003_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "a613cf19d60b1dbe7d9f51bbbf98459d9f181f51f14c39d755912e891ae7fd2f", "output_hash": "9390298f3fb0c5b160498935d79cb139aef28e1c47358b4bbba61862b9c26e59"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "ba4b243631c6ece8920247e7", "attempt_id": "074a3a435f47", "run_group_id": "E4_core:A5|closed|qwen3b:h2", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T06:41:39.662754Z", "task_family": "synthetic", "document_id": "syn_03003_2000", "document_hash": "dddca1a38e3224870fcfae927c701c400fd885c19123ea840ee06a76a140f407", "question_id": "q07_negation", "split": "test", "cluster_id": "syn_03003_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 2, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1866, "query_tokens": 434, "message_tokens": 352, "message_utf8_bytes": 830, "cache_tensor_bytes": null, "serialized_payload_bytes": 830, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 38767.85674900748, "hops": {"1": 22970.73791595176, "2": 15797.11883305572}, "receiver_prefill_ms": 309.47525007650256, "ttft_ms": 44.03916699811816, "decode_ms": 24.823207990266383}, "total_wall_ms": 39146.84716600459, "ttft_ms": 44.03916699811816, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 58, "facts_retained": 15, "retention_rate": 0.25862068965517243, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.2222222222222222, "status": 0.75, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 766}, "ledger_by_hop": {"1": {"facts_total": 58, "facts_retained": 15, "retention_rate": 0.25862068965517243, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.2222222222222222, "status": 0.75, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 766}, "2": {"facts_total": 58, "facts_retained": 15, "retention_rate": 0.25862068965517243, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.2222222222222222, "status": 0.75, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 766}}}, "reference_key_id": "keys/synthetic/syn_03003_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "a613cf19d60b1dbe7d9f51bbbf98459d9f181f51f14c39d755912e891ae7fd2f", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "31669efada00ad4a24466f68", "attempt_id": "1115cea385db", "run_group_id": "E4_core:A5|closed|qwen3b:h2", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T06:41:40.059037Z", "task_family": "synthetic", "document_id": "syn_03003_2000", "document_hash": "dddca1a38e3224870fcfae927c701c400fd885c19123ea840ee06a76a140f407", "question_id": "q08_revision", "split": "test", "cluster_id": "syn_03003_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 2, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1866, "query_tokens": 452, "message_tokens": 352, "message_utf8_bytes": 830, "cache_tensor_bytes": null, "serialized_payload_bytes": 830, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 38767.85674900748, "hops": {"1": 22970.73791595176, "2": 15797.11883305572}, "receiver_prefill_ms": 282.73870900738984, "ttft_ms": 56.368833989836276, "decode_ms": 473.22870790958405}, "total_wall_ms": 39580.73553990107, "ttft_ms": 56.368833989836276, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 6}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "911925", "evidence_artifact": {"ledger": {"facts_total": 58, "facts_retained": 15, "retention_rate": 0.25862068965517243, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.2222222222222222, "status": 0.75, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 766}, "ledger_by_hop": {"1": {"facts_total": 58, "facts_retained": 15, "retention_rate": 0.25862068965517243, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.2222222222222222, "status": 0.75, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 766}, "2": {"facts_total": 58, "facts_retained": 15, "retention_rate": 0.25862068965517243, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.2222222222222222, "status": 0.75, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 766}}}, "reference_key_id": "keys/synthetic/syn_03003_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "a613cf19d60b1dbe7d9f51bbbf98459d9f181f51f14c39d755912e891ae7fd2f", "output_hash": "679a540dc4739f0ace623448b8fd925fb234f554838d7e87da223228ee81b112"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "5b7f5a0cc2f14eeba4700cf8", "attempt_id": "65ace63393c4", "run_group_id": "E4_core:A5|closed|qwen3b:h2", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T06:41:40.889423Z", "task_family": "synthetic", "document_id": "syn_03003_2000", "document_hash": "dddca1a38e3224870fcfae927c701c400fd885c19123ea840ee06a76a140f407", "question_id": "q09_revision", "split": "test", "cluster_id": "syn_03003_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 2, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1866, "query_tokens": 445, "message_tokens": 352, "message_utf8_bytes": 830, "cache_tensor_bytes": null, "serialized_payload_bytes": 830, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 38767.85674900748, "hops": {"1": 22970.73791595176, "2": 15797.11883305572}, "receiver_prefill_ms": 297.00970789417624, "ttft_ms": 39.38704205211252, "decode_ms": 207.21579191740602}, "total_wall_ms": 39312.06516607199, "ttft_ms": 39.38704205211252, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 6}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "911925", "evidence_artifact": {"ledger": {"facts_total": 58, "facts_retained": 15, "retention_rate": 0.25862068965517243, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.2222222222222222, "status": 0.75, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 766}, "ledger_by_hop": {"1": {"facts_total": 58, "facts_retained": 15, "retention_rate": 0.25862068965517243, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.2222222222222222, "status": 0.75, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 766}, "2": {"facts_total": 58, "facts_retained": 15, "retention_rate": 0.25862068965517243, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.2222222222222222, "status": 0.75, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 766}}}, "reference_key_id": "keys/synthetic/syn_03003_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "a613cf19d60b1dbe7d9f51bbbf98459d9f181f51f14c39d755912e891ae7fd2f", "output_hash": "679a540dc4739f0ace623448b8fd925fb234f554838d7e87da223228ee81b112"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "e9df468e4c1947f104fb3930", "attempt_id": "732bed9b60ab", "run_group_id": "E4_core:A5|closed|qwen3b:h2", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T06:41:41.450957Z", "task_family": "synthetic", "document_id": "syn_03003_2000", "document_hash": "dddca1a38e3224870fcfae927c701c400fd885c19123ea840ee06a76a140f407", "question_id": "q10_temporal", "split": "test", "cluster_id": "syn_03003_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 2, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1866, "query_tokens": 452, "message_tokens": 352, "message_utf8_bytes": 830, "cache_tensor_bytes": null, "serialized_payload_bytes": 830, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 38767.85674900748, "hops": {"1": 22970.73791595176, "2": 15797.11883305572}, "receiver_prefill_ms": 317.9137089755386, "ttft_ms": 22.79691700823605, "decode_ms": 17.724749981425703}, "total_wall_ms": 39126.76553998608, "ttft_ms": 22.79691700823605, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 58, "facts_retained": 15, "retention_rate": 0.25862068965517243, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.2222222222222222, "status": 0.75, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 766}, "ledger_by_hop": {"1": {"facts_total": 58, "facts_retained": 15, "retention_rate": 0.25862068965517243, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.2222222222222222, "status": 0.75, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 766}, "2": {"facts_total": 58, "facts_retained": 15, "retention_rate": 0.25862068965517243, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.2222222222222222, "status": 0.75, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 766}}}, "reference_key_id": "keys/synthetic/syn_03003_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "a613cf19d60b1dbe7d9f51bbbf98459d9f181f51f14c39d755912e891ae7fd2f", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "807de852d7e3293a8a9d7a89", "attempt_id": "7819a262a93b", "run_group_id": "E4_core:A5|closed|qwen3b:h2", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T06:41:41.827294Z", "task_family": "synthetic", "document_id": "syn_03003_2000", "document_hash": "dddca1a38e3224870fcfae927c701c400fd885c19123ea840ee06a76a140f407", "question_id": "q11_aggregation", "split": "test", "cluster_id": "syn_03003_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 2, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1866, "query_tokens": 446, "message_tokens": 352, "message_utf8_bytes": 830, "cache_tensor_bytes": null, "serialized_payload_bytes": 830, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 38767.85674900748, "hops": {"1": 22970.73791595176, "2": 15797.11883305572}, "receiver_prefill_ms": 338.0155840422958, "ttft_ms": 62.34829092863947, "decode_ms": 96.42937500029802}, "total_wall_ms": 39265.34141600132, "ttft_ms": 62.34829092863947, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 58, "facts_retained": 15, "retention_rate": 0.25862068965517243, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.2222222222222222, "status": 0.75, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 766}, "ledger_by_hop": {"1": {"facts_total": 58, "facts_retained": 15, "retention_rate": 0.25862068965517243, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.2222222222222222, "status": 0.75, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 766}, "2": {"facts_total": 58, "facts_retained": 15, "retention_rate": 0.25862068965517243, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.2222222222222222, "status": 0.75, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 766}}}, "reference_key_id": "keys/synthetic/syn_03003_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "a613cf19d60b1dbe7d9f51bbbf98459d9f181f51f14c39d755912e891ae7fd2f", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "d240a195acc97353cd39893a", "attempt_id": "18808a1394c2", "run_group_id": "E4_core:A5|closed|qwen3b:h2", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T06:41:42.345303Z", "task_family": "synthetic", "document_id": "syn_03003_2000", "document_hash": "dddca1a38e3224870fcfae927c701c400fd885c19123ea840ee06a76a140f407", "question_id": "q12_unit_conversion", "split": "test", "cluster_id": "syn_03003_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 2, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1866, "query_tokens": 443, "message_tokens": 352, "message_utf8_bytes": 830, "cache_tensor_bytes": null, "serialized_payload_bytes": 830, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 38767.85674900748, "hops": {"1": 22970.73791595176, "2": 15797.11883305572}, "receiver_prefill_ms": 399.1651670075953, "ttft_ms": 31.177042052149773, "decode_ms": 171.67787498328835}, "total_wall_ms": 39370.442957035266, "ttft_ms": 31.177042052149773, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 5}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "17430", "evidence_artifact": {"ledger": {"facts_total": 58, "facts_retained": 15, "retention_rate": 0.25862068965517243, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.2222222222222222, "status": 0.75, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 766}, "ledger_by_hop": {"1": {"facts_total": 58, "facts_retained": 15, "retention_rate": 0.25862068965517243, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.2222222222222222, "status": 0.75, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 766}, "2": {"facts_total": 58, "facts_retained": 15, "retention_rate": 0.25862068965517243, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.2222222222222222, "status": 0.75, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 766}}}, "reference_key_id": "keys/synthetic/syn_03003_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "a613cf19d60b1dbe7d9f51bbbf98459d9f181f51f14c39d755912e891ae7fd2f", "output_hash": "2575eaf3b748efc209b676518766751eb038c88f74578951345818bfe7f7194b"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "d7cee787dca92f2cc8e3a2e5", "attempt_id": "430cde3fded1", "run_group_id": "E4_core:A5|closed|qwen3b:h2", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T06:41:42.967716Z", "task_family": "synthetic", "document_id": "syn_03003_2000", "document_hash": "dddca1a38e3224870fcfae927c701c400fd885c19123ea840ee06a76a140f407", "question_id": "q13_unanswerable", "split": "test", "cluster_id": "syn_03003_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 2, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1866, "query_tokens": 447, "message_tokens": 352, "message_utf8_bytes": 830, "cache_tensor_bytes": null, "serialized_payload_bytes": 830, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 38767.85674900748, "hops": {"1": 22970.73791595176, "2": 15797.11883305572}, "receiver_prefill_ms": 309.24691702239215, "ttft_ms": 40.20258295349777, "decode_ms": 26.08354203402996}, "total_wall_ms": 39144.0135409357, "ttft_ms": 40.20258295349777, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 58, "facts_retained": 15, "retention_rate": 0.25862068965517243, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.2222222222222222, "status": 0.75, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 766}, "ledger_by_hop": {"1": {"facts_total": 58, "facts_retained": 15, "retention_rate": 0.25862068965517243, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.2222222222222222, "status": 0.75, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 766}, "2": {"facts_total": 58, "facts_retained": 15, "retention_rate": 0.25862068965517243, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.2222222222222222, "status": 0.75, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 766}}}, "reference_key_id": "keys/synthetic/syn_03003_2000.json", "task_scores": {"correct": 1.0, "abstained": 1.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "a613cf19d60b1dbe7d9f51bbbf98459d9f181f51f14c39d755912e891ae7fd2f", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "a481af1c766ba6c94767cc4e", "attempt_id": "b8b69da4021a", "run_group_id": "E4_core:A5|closed|qwen3b:h2", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T06:41:43.361234Z", "task_family": "synthetic", "document_id": "syn_03003_2000", "document_hash": "dddca1a38e3224870fcfae927c701c400fd885c19123ea840ee06a76a140f407", "question_id": "q14_unanswerable", "split": "test", "cluster_id": "syn_03003_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 2, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1866, "query_tokens": 448, "message_tokens": 352, "message_utf8_bytes": 830, "cache_tensor_bytes": null, "serialized_payload_bytes": 830, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 38767.85674900748, "hops": {"1": 22970.73791595176, "2": 15797.11883305572}, "receiver_prefill_ms": 269.70770908519626, "ttft_ms": 53.50933305453509, "decode_ms": 53.17166598979384}, "total_wall_ms": 39144.78262397461, "ttft_ms": 53.50933305453509, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 58, "facts_retained": 15, "retention_rate": 0.25862068965517243, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.2222222222222222, "status": 0.75, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 766}, "ledger_by_hop": {"1": {"facts_total": 58, "facts_retained": 15, "retention_rate": 0.25862068965517243, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.2222222222222222, "status": 0.75, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 766}, "2": {"facts_total": 58, "facts_retained": 15, "retention_rate": 0.25862068965517243, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.2222222222222222, "status": 0.75, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 766}}}, "reference_key_id": "keys/synthetic/syn_03003_2000.json", "task_scores": {"correct": 1.0, "abstained": 1.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "a613cf19d60b1dbe7d9f51bbbf98459d9f181f51f14c39d755912e891ae7fd2f", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "6c4dcf3e9b3f801452ac0eb3", "attempt_id": "655f00b0b8cc", "run_group_id": "E4_core:A5|closed|qwen3b:h4", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T06:42:15.280329Z", "task_family": "synthetic", "document_id": "syn_03003_2000", "document_hash": "dddca1a38e3224870fcfae927c701c400fd885c19123ea840ee06a76a140f407", "question_id": "q00_single_fact", "split": "test", "cluster_id": "syn_03003_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 4, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1866, "query_tokens": 435, "message_tokens": 352, "message_utf8_bytes": 830, "cache_tensor_bytes": null, "serialized_payload_bytes": 830, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 70289.44795799907, "hops": {"1": 22970.73791595176, "2": 15797.11883305572, "3": 15682.611500029452, "4": 15838.979708962142}, "receiver_prefill_ms": 373.01470793318003, "ttft_ms": 33.08516601100564, "decode_ms": 325.825916021131}, "total_wall_ms": 71021.99054101948, "ttft_ms": 33.08516601100564, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 10}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "GR-79S-5839", "evidence_artifact": {"ledger": {"facts_total": 58, "facts_retained": 15, "retention_rate": 0.25862068965517243, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.2222222222222222, "status": 0.75, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 766}, "ledger_by_hop": {"1": {"facts_total": 58, "facts_retained": 15, "retention_rate": 0.25862068965517243, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.2222222222222222, "status": 0.75, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 766}, "2": {"facts_total": 58, "facts_retained": 15, "retention_rate": 0.25862068965517243, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.2222222222222222, "status": 0.75, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 766}, "3": {"facts_total": 58, "facts_retained": 15, "retention_rate": 0.25862068965517243, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.2222222222222222, "status": 0.75, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 766}, "4": {"facts_total": 58, "facts_retained": 15, "retention_rate": 0.25862068965517243, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.2222222222222222, "status": 0.75, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 766}}}, "reference_key_id": "keys/synthetic/syn_03003_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "a613cf19d60b1dbe7d9f51bbbf98459d9f181f51f14c39d755912e891ae7fd2f", "output_hash": "720cc40cd69e85e9728f6bea00b17a95ca07eee643e143a131840763958303cd"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "6c333c3df6a81cbcd842a476", "attempt_id": "3754b924323e", "run_group_id": "E4_core:A5|closed|qwen3b:h4", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T06:42:16.032792Z", "task_family": "synthetic", "document_id": "syn_03003_2000", "document_hash": "dddca1a38e3224870fcfae927c701c400fd885c19123ea840ee06a76a140f407", "question_id": "q01_single_fact", "split": "test", "cluster_id": "syn_03003_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 4, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1866, "query_tokens": 439, "message_tokens": 352, "message_utf8_bytes": 830, "cache_tensor_bytes": null, "serialized_payload_bytes": 830, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 70289.44795799907, "hops": {"1": 22970.73791595176, "2": 15797.11883305572, "3": 15682.611500029452, "4": 15838.979708962142}, "receiver_prefill_ms": 286.4320839289576, "ttft_ms": 19.928416935727, "decode_ms": 41.551167028956115}, "total_wall_ms": 70637.8994579427, "ttft_ms": 19.928416935727, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 58, "facts_retained": 15, "retention_rate": 0.25862068965517243, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.2222222222222222, "status": 0.75, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 766}, "ledger_by_hop": {"1": {"facts_total": 58, "facts_retained": 15, "retention_rate": 0.25862068965517243, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.2222222222222222, "status": 0.75, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 766}, "2": {"facts_total": 58, "facts_retained": 15, "retention_rate": 0.25862068965517243, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.2222222222222222, "status": 0.75, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 766}, "3": {"facts_total": 58, "facts_retained": 15, "retention_rate": 0.25862068965517243, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.2222222222222222, "status": 0.75, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 766}, "4": {"facts_total": 58, "facts_retained": 15, "retention_rate": 0.25862068965517243, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.2222222222222222, "status": 0.75, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 766}}}, "reference_key_id": "keys/synthetic/syn_03003_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "a613cf19d60b1dbe7d9f51bbbf98459d9f181f51f14c39d755912e891ae7fd2f", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "42a14328ecd0001d92f04fd3", "attempt_id": "7c916195b7f4", "run_group_id": "E4_core:A5|closed|qwen3b:h4", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T06:42:16.399433Z", "task_family": "synthetic", "document_id": "syn_03003_2000", "document_hash": "dddca1a38e3224870fcfae927c701c400fd885c19123ea840ee06a76a140f407", "question_id": "q02_exact_copy", "split": "test", "cluster_id": "syn_03003_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 4, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1866, "query_tokens": 439, "message_tokens": 352, "message_utf8_bytes": 830, "cache_tensor_bytes": null, "serialized_payload_bytes": 830, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 70289.44795799907, "hops": {"1": 22970.73791595176, "2": 15797.11883305572, "3": 15682.611500029452, "4": 15838.979708962142}, "receiver_prefill_ms": 311.9228750001639, "ttft_ms": 42.16208297293633, "decode_ms": 89.81770800892264}, "total_wall_ms": 70733.87287394144, "ttft_ms": 42.16208297293633, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 58, "facts_retained": 15, "retention_rate": 0.25862068965517243, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.2222222222222222, "status": 0.75, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 766}, "ledger_by_hop": {"1": {"facts_total": 58, "facts_retained": 15, "retention_rate": 0.25862068965517243, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.2222222222222222, "status": 0.75, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 766}, "2": {"facts_total": 58, "facts_retained": 15, "retention_rate": 0.25862068965517243, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.2222222222222222, "status": 0.75, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 766}, "3": {"facts_total": 58, "facts_retained": 15, "retention_rate": 0.25862068965517243, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.2222222222222222, "status": 0.75, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 766}, "4": {"facts_total": 58, "facts_retained": 15, "retention_rate": 0.25862068965517243, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.2222222222222222, "status": 0.75, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 766}}}, "reference_key_id": "keys/synthetic/syn_03003_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "a613cf19d60b1dbe7d9f51bbbf98459d9f181f51f14c39d755912e891ae7fd2f", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "7f76032e85b6564f03477663", "attempt_id": "532addb120c5", "run_group_id": "E4_core:A5|closed|qwen3b:h4", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T06:42:16.861467Z", "task_family": "synthetic", "document_id": "syn_03003_2000", "document_hash": "dddca1a38e3224870fcfae927c701c400fd885c19123ea840ee06a76a140f407", "question_id": "q03_exact_copy", "split": "test", "cluster_id": "syn_03003_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 4, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1866, "query_tokens": 434, "message_tokens": 352, "message_utf8_bytes": 830, "cache_tensor_bytes": null, "serialized_payload_bytes": 830, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 70289.44795799907, "hops": {"1": 22970.73791595176, "2": 15797.11883305572, "3": 15682.611500029452, "4": 15838.979708962142}, "receiver_prefill_ms": 379.47325001005083, "ttft_ms": 35.21258302498609, "decode_ms": 43.78012497909367}, "total_wall_ms": 70748.5569579294, "ttft_ms": 35.21258302498609, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 58, "facts_retained": 15, "retention_rate": 0.25862068965517243, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.2222222222222222, "status": 0.75, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 766}, "ledger_by_hop": {"1": {"facts_total": 58, "facts_retained": 15, "retention_rate": 0.25862068965517243, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.2222222222222222, "status": 0.75, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 766}, "2": {"facts_total": 58, "facts_retained": 15, "retention_rate": 0.25862068965517243, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.2222222222222222, "status": 0.75, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 766}, "3": {"facts_total": 58, "facts_retained": 15, "retention_rate": 0.25862068965517243, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.2222222222222222, "status": 0.75, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 766}, "4": {"facts_total": 58, "facts_retained": 15, "retention_rate": 0.25862068965517243, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.2222222222222222, "status": 0.75, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 766}}}, "reference_key_id": "keys/synthetic/syn_03003_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "a613cf19d60b1dbe7d9f51bbbf98459d9f181f51f14c39d755912e891ae7fd2f", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "e9cee9a4d8ec8b5b9fe07f7c", "attempt_id": "616308f0fd35", "run_group_id": "E4_core:A5|closed|qwen3b:h4", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T06:42:17.340179Z", "task_family": "synthetic", "document_id": "syn_03003_2000", "document_hash": "dddca1a38e3224870fcfae927c701c400fd885c19123ea840ee06a76a140f407", "question_id": "q04_multi_hop", "split": "test", "cluster_id": "syn_03003_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 4, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1866, "query_tokens": 443, "message_tokens": 352, "message_utf8_bytes": 830, "cache_tensor_bytes": null, "serialized_payload_bytes": 830, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 70289.44795799907, "hops": {"1": 22970.73791595176, "2": 15797.11883305572, "3": 15682.611500029452, "4": 15838.979708962142}, "receiver_prefill_ms": 285.37541709374636, "ttft_ms": 28.708791942335665, "decode_ms": 48.05924999527633}, "total_wall_ms": 70652.12870808318, "ttft_ms": 28.708791942335665, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 58, "facts_retained": 15, "retention_rate": 0.25862068965517243, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.2222222222222222, "status": 0.75, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 766}, "ledger_by_hop": {"1": {"facts_total": 58, "facts_retained": 15, "retention_rate": 0.25862068965517243, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.2222222222222222, "status": 0.75, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 766}, "2": {"facts_total": 58, "facts_retained": 15, "retention_rate": 0.25862068965517243, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.2222222222222222, "status": 0.75, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 766}, "3": {"facts_total": 58, "facts_retained": 15, "retention_rate": 0.25862068965517243, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.2222222222222222, "status": 0.75, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 766}, "4": {"facts_total": 58, "facts_retained": 15, "retention_rate": 0.25862068965517243, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.2222222222222222, "status": 0.75, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 766}}}, "reference_key_id": "keys/synthetic/syn_03003_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "a613cf19d60b1dbe7d9f51bbbf98459d9f181f51f14c39d755912e891ae7fd2f", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "bfdaf678d2922bc3a6de49e6", "attempt_id": "60959b07584f", "run_group_id": "E4_core:A5|closed|qwen3b:h4", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T06:42:17.719484Z", "task_family": "synthetic", "document_id": "syn_03003_2000", "document_hash": "dddca1a38e3224870fcfae927c701c400fd885c19123ea840ee06a76a140f407", "question_id": "q05_multi_hop", "split": "test", "cluster_id": "syn_03003_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 4, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1866, "query_tokens": 445, "message_tokens": 352, "message_utf8_bytes": 830, "cache_tensor_bytes": null, "serialized_payload_bytes": 830, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 70289.44795799907, "hops": {"1": 22970.73791595176, "2": 15797.11883305572, "3": 15682.611500029452, "4": 15838.979708962142}, "receiver_prefill_ms": 273.03925005253404, "ttft_ms": 19.644125015474856, "decode_ms": 39.46416696999222}, "total_wall_ms": 70622.424916015, "ttft_ms": 19.644125015474856, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 58, "facts_retained": 15, "retention_rate": 0.25862068965517243, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.2222222222222222, "status": 0.75, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 766}, "ledger_by_hop": {"1": {"facts_total": 58, "facts_retained": 15, "retention_rate": 0.25862068965517243, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.2222222222222222, "status": 0.75, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 766}, "2": {"facts_total": 58, "facts_retained": 15, "retention_rate": 0.25862068965517243, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.2222222222222222, "status": 0.75, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 766}, "3": {"facts_total": 58, "facts_retained": 15, "retention_rate": 0.25862068965517243, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.2222222222222222, "status": 0.75, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 766}, "4": {"facts_total": 58, "facts_retained": 15, "retention_rate": 0.25862068965517243, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.2222222222222222, "status": 0.75, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 766}}}, "reference_key_id": "keys/synthetic/syn_03003_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "a613cf19d60b1dbe7d9f51bbbf98459d9f181f51f14c39d755912e891ae7fd2f", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "04531eef38cb094263edc8a6", "attempt_id": "d26728083837", "run_group_id": "E4_core:A5|closed|qwen3b:h4", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T06:42:18.070397Z", "task_family": "synthetic", "document_id": "syn_03003_2000", "document_hash": "dddca1a38e3224870fcfae927c701c400fd885c19123ea840ee06a76a140f407", "question_id": "q06_negation", "split": "test", "cluster_id": "syn_03003_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 4, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1866, "query_tokens": 437, "message_tokens": 352, "message_utf8_bytes": 830, "cache_tensor_bytes": null, "serialized_payload_bytes": 830, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 70289.44795799907, "hops": {"1": 22970.73791595176, "2": 15797.11883305572, "3": 15682.611500029452, "4": 15838.979708962142}, "receiver_prefill_ms": 277.9434169642627, "ttft_ms": 57.755500078201294, "decode_ms": 55.3307089721784}, "total_wall_ms": 70681.01399997249, "ttft_ms": 57.755500078201294, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "no", "evidence_artifact": {"ledger": {"facts_total": 58, "facts_retained": 15, "retention_rate": 0.25862068965517243, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.2222222222222222, "status": 0.75, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 766}, "ledger_by_hop": {"1": {"facts_total": 58, "facts_retained": 15, "retention_rate": 0.25862068965517243, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.2222222222222222, "status": 0.75, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 766}, "2": {"facts_total": 58, "facts_retained": 15, "retention_rate": 0.25862068965517243, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.2222222222222222, "status": 0.75, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 766}, "3": {"facts_total": 58, "facts_retained": 15, "retention_rate": 0.25862068965517243, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.2222222222222222, "status": 0.75, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 766}, "4": {"facts_total": 58, "facts_retained": 15, "retention_rate": 0.25862068965517243, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.2222222222222222, "status": 0.75, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 766}}}, "reference_key_id": "keys/synthetic/syn_03003_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "a613cf19d60b1dbe7d9f51bbbf98459d9f181f51f14c39d755912e891ae7fd2f", "output_hash": "9390298f3fb0c5b160498935d79cb139aef28e1c47358b4bbba61862b9c26e59"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "26dfdc3f56a7bc5fa709e2a6", "attempt_id": "44f205a3ed0e", "run_group_id": "E4_core:A5|closed|qwen3b:h4", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T06:42:18.479284Z", "task_family": "synthetic", "document_id": "syn_03003_2000", "document_hash": "dddca1a38e3224870fcfae927c701c400fd885c19123ea840ee06a76a140f407", "question_id": "q07_negation", "split": "test", "cluster_id": "syn_03003_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 4, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1866, "query_tokens": 434, "message_tokens": 352, "message_utf8_bytes": 830, "cache_tensor_bytes": null, "serialized_payload_bytes": 830, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 70289.44795799907, "hops": {"1": 22970.73791595176, "2": 15797.11883305572, "3": 15682.611500029452, "4": 15838.979708962142}, "receiver_prefill_ms": 426.9248330965638, "ttft_ms": 31.48491692263633, "decode_ms": 42.35825000796467}, "total_wall_ms": 70790.89216596913, "ttft_ms": 31.48491692263633, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 58, "facts_retained": 15, "retention_rate": 0.25862068965517243, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.2222222222222222, "status": 0.75, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 766}, "ledger_by_hop": {"1": {"facts_total": 58, "facts_retained": 15, "retention_rate": 0.25862068965517243, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.2222222222222222, "status": 0.75, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 766}, "2": {"facts_total": 58, "facts_retained": 15, "retention_rate": 0.25862068965517243, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.2222222222222222, "status": 0.75, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 766}, "3": {"facts_total": 58, "facts_retained": 15, "retention_rate": 0.25862068965517243, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.2222222222222222, "status": 0.75, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 766}, "4": {"facts_total": 58, "facts_retained": 15, "retention_rate": 0.25862068965517243, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.2222222222222222, "status": 0.75, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 766}}}, "reference_key_id": "keys/synthetic/syn_03003_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "a613cf19d60b1dbe7d9f51bbbf98459d9f181f51f14c39d755912e891ae7fd2f", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "d233a97b7e418cfafc12409c", "attempt_id": "ac8ced44d39b", "run_group_id": "E4_core:A5|closed|qwen3b:h4", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T06:42:18.997836Z", "task_family": "synthetic", "document_id": "syn_03003_2000", "document_hash": "dddca1a38e3224870fcfae927c701c400fd885c19123ea840ee06a76a140f407", "question_id": "q08_revision", "split": "test", "cluster_id": "syn_03003_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 4, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1866, "query_tokens": 452, "message_tokens": 352, "message_utf8_bytes": 830, "cache_tensor_bytes": null, "serialized_payload_bytes": 830, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 70289.44795799907, "hops": {"1": 22970.73791595176, "2": 15797.11883305572, "3": 15682.611500029452, "4": 15838.979708962142}, "receiver_prefill_ms": 315.0080000050366, "ttft_ms": 35.124124959111214, "decode_ms": 210.4980000294745}, "total_wall_ms": 70850.63733300194, "ttft_ms": 35.124124959111214, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 6}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "911925", "evidence_artifact": {"ledger": {"facts_total": 58, "facts_retained": 15, "retention_rate": 0.25862068965517243, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.2222222222222222, "status": 0.75, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 766}, "ledger_by_hop": {"1": {"facts_total": 58, "facts_retained": 15, "retention_rate": 0.25862068965517243, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.2222222222222222, "status": 0.75, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 766}, "2": {"facts_total": 58, "facts_retained": 15, "retention_rate": 0.25862068965517243, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.2222222222222222, "status": 0.75, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 766}, "3": {"facts_total": 58, "facts_retained": 15, "retention_rate": 0.25862068965517243, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.2222222222222222, "status": 0.75, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 766}, "4": {"facts_total": 58, "facts_retained": 15, "retention_rate": 0.25862068965517243, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.2222222222222222, "status": 0.75, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 766}}}, "reference_key_id": "keys/synthetic/syn_03003_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "a613cf19d60b1dbe7d9f51bbbf98459d9f181f51f14c39d755912e891ae7fd2f", "output_hash": "679a540dc4739f0ace623448b8fd925fb234f554838d7e87da223228ee81b112"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "d714d75d203cdfdb575bf3af", "attempt_id": "b09fafdcf65b", "run_group_id": "E4_core:A5|closed|qwen3b:h4", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T06:42:19.576232Z", "task_family": "synthetic", "document_id": "syn_03003_2000", "document_hash": "dddca1a38e3224870fcfae927c701c400fd885c19123ea840ee06a76a140f407", "question_id": "q09_revision", "split": "test", "cluster_id": "syn_03003_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 4, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1866, "query_tokens": 445, "message_tokens": 352, "message_utf8_bytes": 830, "cache_tensor_bytes": null, "serialized_payload_bytes": 830, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 70289.44795799907, "hops": {"1": 22970.73791595176, "2": 15797.11883305572, "3": 15682.611500029452, "4": 15838.979708962142}, "receiver_prefill_ms": 306.08233297243714, "ttft_ms": 43.346916092559695, "decode_ms": 413.9759580139071}, "total_wall_ms": 71053.35970793385, "ttft_ms": 43.346916092559695, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 6}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "911925", "evidence_artifact": {"ledger": {"facts_total": 58, "facts_retained": 15, "retention_rate": 0.25862068965517243, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.2222222222222222, "status": 0.75, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 766}, "ledger_by_hop": {"1": {"facts_total": 58, "facts_retained": 15, "retention_rate": 0.25862068965517243, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.2222222222222222, "status": 0.75, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 766}, "2": {"facts_total": 58, "facts_retained": 15, "retention_rate": 0.25862068965517243, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.2222222222222222, "status": 0.75, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 766}, "3": {"facts_total": 58, "facts_retained": 15, "retention_rate": 0.25862068965517243, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.2222222222222222, "status": 0.75, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 766}, "4": {"facts_total": 58, "facts_retained": 15, "retention_rate": 0.25862068965517243, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.2222222222222222, "status": 0.75, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 766}}}, "reference_key_id": "keys/synthetic/syn_03003_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "a613cf19d60b1dbe7d9f51bbbf98459d9f181f51f14c39d755912e891ae7fd2f", "output_hash": "679a540dc4739f0ace623448b8fd925fb234f554838d7e87da223228ee81b112"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "de6cd0484cd8f6f8756ec34c", "attempt_id": "8235f1617393", "run_group_id": "E4_core:A5|closed|qwen3b:h4", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T06:42:20.357571Z", "task_family": "synthetic", "document_id": "syn_03003_2000", "document_hash": "dddca1a38e3224870fcfae927c701c400fd885c19123ea840ee06a76a140f407", "question_id": "q10_temporal", "split": "test", "cluster_id": "syn_03003_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 4, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1866, "query_tokens": 452, "message_tokens": 352, "message_utf8_bytes": 830, "cache_tensor_bytes": null, "serialized_payload_bytes": 830, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 70289.44795799907, "hops": {"1": 22970.73791595176, "2": 15797.11883305572, "3": 15682.611500029452, "4": 15838.979708962142}, "receiver_prefill_ms": 316.5455419803038, "ttft_ms": 34.87762506119907, "decode_ms": 21.703209029510617}, "total_wall_ms": 70663.10900007375, "ttft_ms": 34.87762506119907, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 58, "facts_retained": 15, "retention_rate": 0.25862068965517243, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.2222222222222222, "status": 0.75, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 766}, "ledger_by_hop": {"1": {"facts_total": 58, "facts_retained": 15, "retention_rate": 0.25862068965517243, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.2222222222222222, "status": 0.75, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 766}, "2": {"facts_total": 58, "facts_retained": 15, "retention_rate": 0.25862068965517243, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.2222222222222222, "status": 0.75, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 766}, "3": {"facts_total": 58, "facts_retained": 15, "retention_rate": 0.25862068965517243, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.2222222222222222, "status": 0.75, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 766}, "4": {"facts_total": 58, "facts_retained": 15, "retention_rate": 0.25862068965517243, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.2222222222222222, "status": 0.75, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 766}}}, "reference_key_id": "keys/synthetic/syn_03003_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "a613cf19d60b1dbe7d9f51bbbf98459d9f181f51f14c39d755912e891ae7fd2f", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "4f56aa8a1e13622bdcc0662c", "attempt_id": "9a7ca60ea370", "run_group_id": "E4_core:A5|closed|qwen3b:h4", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T06:42:20.748023Z", "task_family": "synthetic", "document_id": "syn_03003_2000", "document_hash": "dddca1a38e3224870fcfae927c701c400fd885c19123ea840ee06a76a140f407", "question_id": "q11_aggregation", "split": "test", "cluster_id": "syn_03003_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 4, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1866, "query_tokens": 446, "message_tokens": 352, "message_utf8_bytes": 830, "cache_tensor_bytes": null, "serialized_payload_bytes": 830, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 70289.44795799907, "hops": {"1": 22970.73791595176, "2": 15797.11883305572, "3": 15682.611500029452, "4": 15838.979708962142}, "receiver_prefill_ms": 295.1619579689577, "ttft_ms": 31.84237494133413, "decode_ms": 29.244833975099027}, "total_wall_ms": 70646.11662505195, "ttft_ms": 31.84237494133413, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 58, "facts_retained": 15, "retention_rate": 0.25862068965517243, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.2222222222222222, "status": 0.75, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 766}, "ledger_by_hop": {"1": {"facts_total": 58, "facts_retained": 15, "retention_rate": 0.25862068965517243, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.2222222222222222, "status": 0.75, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 766}, "2": {"facts_total": 58, "facts_retained": 15, "retention_rate": 0.25862068965517243, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.2222222222222222, "status": 0.75, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 766}, "3": {"facts_total": 58, "facts_retained": 15, "retention_rate": 0.25862068965517243, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.2222222222222222, "status": 0.75, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 766}, "4": {"facts_total": 58, "facts_retained": 15, "retention_rate": 0.25862068965517243, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.2222222222222222, "status": 0.75, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 766}}}, "reference_key_id": "keys/synthetic/syn_03003_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "a613cf19d60b1dbe7d9f51bbbf98459d9f181f51f14c39d755912e891ae7fd2f", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "6c1e9644b9fd3759b3363286", "attempt_id": "3e81086d4410", "run_group_id": "E4_core:A5|closed|qwen3b:h4", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T06:42:21.120936Z", "task_family": "synthetic", "document_id": "syn_03003_2000", "document_hash": "dddca1a38e3224870fcfae927c701c400fd885c19123ea840ee06a76a140f407", "question_id": "q12_unit_conversion", "split": "test", "cluster_id": "syn_03003_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 4, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1866, "query_tokens": 443, "message_tokens": 352, "message_utf8_bytes": 830, "cache_tensor_bytes": null, "serialized_payload_bytes": 830, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 70289.44795799907, "hops": {"1": 22970.73791595176, "2": 15797.11883305572, "3": 15682.611500029452, "4": 15838.979708962142}, "receiver_prefill_ms": 264.7398749832064, "ttft_ms": 39.28983397781849, "decode_ms": 308.5510419914499}, "total_wall_ms": 70902.65724889468, "ttft_ms": 39.28983397781849, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 5}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "17430", "evidence_artifact": {"ledger": {"facts_total": 58, "facts_retained": 15, "retention_rate": 0.25862068965517243, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.2222222222222222, "status": 0.75, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 766}, "ledger_by_hop": {"1": {"facts_total": 58, "facts_retained": 15, "retention_rate": 0.25862068965517243, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.2222222222222222, "status": 0.75, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 766}, "2": {"facts_total": 58, "facts_retained": 15, "retention_rate": 0.25862068965517243, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.2222222222222222, "status": 0.75, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 766}, "3": {"facts_total": 58, "facts_retained": 15, "retention_rate": 0.25862068965517243, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.2222222222222222, "status": 0.75, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 766}, "4": {"facts_total": 58, "facts_retained": 15, "retention_rate": 0.25862068965517243, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.2222222222222222, "status": 0.75, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 766}}}, "reference_key_id": "keys/synthetic/syn_03003_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "a613cf19d60b1dbe7d9f51bbbf98459d9f181f51f14c39d755912e891ae7fd2f", "output_hash": "2575eaf3b748efc209b676518766751eb038c88f74578951345818bfe7f7194b"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "dd744bafb302f2cd6c2ba42a", "attempt_id": "97ab5a48a5a8", "run_group_id": "E4_core:A5|closed|qwen3b:h4", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T06:42:21.751652Z", "task_family": "synthetic", "document_id": "syn_03003_2000", "document_hash": "dddca1a38e3224870fcfae927c701c400fd885c19123ea840ee06a76a140f407", "question_id": "q13_unanswerable", "split": "test", "cluster_id": "syn_03003_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 4, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1866, "query_tokens": 447, "message_tokens": 352, "message_utf8_bytes": 830, "cache_tensor_bytes": null, "serialized_payload_bytes": 830, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 70289.44795799907, "hops": {"1": 22970.73791595176, "2": 15797.11883305572, "3": 15682.611500029452, "4": 15838.979708962142}, "receiver_prefill_ms": 385.03929204307497, "ttft_ms": 35.600541974417865, "decode_ms": 21.949874935671687}, "total_wall_ms": 70732.55266598426, "ttft_ms": 35.600541974417865, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 58, "facts_retained": 15, "retention_rate": 0.25862068965517243, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.2222222222222222, "status": 0.75, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 766}, "ledger_by_hop": {"1": {"facts_total": 58, "facts_retained": 15, "retention_rate": 0.25862068965517243, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.2222222222222222, "status": 0.75, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 766}, "2": {"facts_total": 58, "facts_retained": 15, "retention_rate": 0.25862068965517243, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.2222222222222222, "status": 0.75, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 766}, "3": {"facts_total": 58, "facts_retained": 15, "retention_rate": 0.25862068965517243, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.2222222222222222, "status": 0.75, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 766}, "4": {"facts_total": 58, "facts_retained": 15, "retention_rate": 0.25862068965517243, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.2222222222222222, "status": 0.75, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 766}}}, "reference_key_id": "keys/synthetic/syn_03003_2000.json", "task_scores": {"correct": 1.0, "abstained": 1.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "a613cf19d60b1dbe7d9f51bbbf98459d9f181f51f14c39d755912e891ae7fd2f", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "3b3a7f22e5de6d0285956168", "attempt_id": "405db0dfb56b", "run_group_id": "E4_core:A5|closed|qwen3b:h4", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T06:42:22.214503Z", "task_family": "synthetic", "document_id": "syn_03003_2000", "document_hash": "dddca1a38e3224870fcfae927c701c400fd885c19123ea840ee06a76a140f407", "question_id": "q14_unanswerable", "split": "test", "cluster_id": "syn_03003_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 4, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1866, "query_tokens": 448, "message_tokens": 352, "message_utf8_bytes": 830, "cache_tensor_bytes": null, "serialized_payload_bytes": 830, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 70289.44795799907, "hops": {"1": 22970.73791595176, "2": 15797.11883305572, "3": 15682.611500029452, "4": 15838.979708962142}, "receiver_prefill_ms": 295.15458398964256, "ttft_ms": 40.437291958369315, "decode_ms": 25.096500059589744}, "total_wall_ms": 70650.66929196473, "ttft_ms": 40.437291958369315, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 58, "facts_retained": 15, "retention_rate": 0.25862068965517243, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.2222222222222222, "status": 0.75, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 766}, "ledger_by_hop": {"1": {"facts_total": 58, "facts_retained": 15, "retention_rate": 0.25862068965517243, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.2222222222222222, "status": 0.75, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 766}, "2": {"facts_total": 58, "facts_retained": 15, "retention_rate": 0.25862068965517243, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.2222222222222222, "status": 0.75, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 766}, "3": {"facts_total": 58, "facts_retained": 15, "retention_rate": 0.25862068965517243, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.2222222222222222, "status": 0.75, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 766}, "4": {"facts_total": 58, "facts_retained": 15, "retention_rate": 0.25862068965517243, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.2222222222222222, "status": 0.75, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 766}}}, "reference_key_id": "keys/synthetic/syn_03003_2000.json", "task_scores": {"correct": 1.0, "abstained": 1.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "a613cf19d60b1dbe7d9f51bbbf98459d9f181f51f14c39d755912e891ae7fd2f", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "e5947dba30433ab519cd3dbf", "attempt_id": "1b24b10f4328", "run_group_id": "E4_core:A5|closed|qwen3b:h8", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T06:43:25.649404Z", "task_family": "synthetic", "document_id": "syn_03003_2000", "document_hash": "dddca1a38e3224870fcfae927c701c400fd885c19123ea840ee06a76a140f407", "question_id": "q00_single_fact", "split": "test", "cluster_id": "syn_03003_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 8, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1866, "query_tokens": 435, "message_tokens": 352, "message_utf8_bytes": 830, "cache_tensor_bytes": null, "serialized_payload_bytes": 830, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 133341.68820711784, "hops": {"1": 22970.73791595176, "2": 15797.11883305572, "3": 15682.611500029452, "4": 15838.979708962142, "5": 15930.869291070849, "6": 15628.01954196766, "7": 15836.956083076075, "8": 15656.395333004184}, "receiver_prefill_ms": 277.61120791547, "ttft_ms": 83.64495798014104, "decode_ms": 458.7848329683766}, "total_wall_ms": 134162.2896241024, "ttft_ms": 83.64495798014104, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 10}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "GR-79S-5839", "evidence_artifact": {"ledger": {"facts_total": 58, "facts_retained": 15, "retention_rate": 0.25862068965517243, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.2222222222222222, "status": 0.75, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 766}, "ledger_by_hop": {"1": {"facts_total": 58, "facts_retained": 15, "retention_rate": 0.25862068965517243, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.2222222222222222, "status": 0.75, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 766}, "2": {"facts_total": 58, "facts_retained": 15, "retention_rate": 0.25862068965517243, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.2222222222222222, "status": 0.75, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 766}, "3": {"facts_total": 58, "facts_retained": 15, "retention_rate": 0.25862068965517243, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.2222222222222222, "status": 0.75, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 766}, "4": {"facts_total": 58, "facts_retained": 15, "retention_rate": 0.25862068965517243, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.2222222222222222, "status": 0.75, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 766}, "5": {"facts_total": 58, "facts_retained": 15, "retention_rate": 0.25862068965517243, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.2222222222222222, "status": 0.75, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 766}, "6": {"facts_total": 58, "facts_retained": 15, "retention_rate": 0.25862068965517243, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.2222222222222222, "status": 0.75, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 766}, "7": {"facts_total": 58, "facts_retained": 15, "retention_rate": 0.25862068965517243, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.2222222222222222, "status": 0.75, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 766}, "8": {"facts_total": 58, "facts_retained": 15, "retention_rate": 0.25862068965517243, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.2222222222222222, "status": 0.75, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 766}}}, "reference_key_id": "keys/synthetic/syn_03003_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "a613cf19d60b1dbe7d9f51bbbf98459d9f181f51f14c39d755912e891ae7fd2f", "output_hash": "720cc40cd69e85e9728f6bea00b17a95ca07eee643e143a131840763958303cd"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "63a05712bf58ffdb49789e39", "attempt_id": "db51f69e56e1", "run_group_id": "E4_core:A5|closed|qwen3b:h8", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T06:43:26.487537Z", "task_family": "synthetic", "document_id": "syn_03003_2000", "document_hash": "dddca1a38e3224870fcfae927c701c400fd885c19123ea840ee06a76a140f407", "question_id": "q01_single_fact", "split": "test", "cluster_id": "syn_03003_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 8, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1866, "query_tokens": 439, "message_tokens": 352, "message_utf8_bytes": 830, "cache_tensor_bytes": null, "serialized_payload_bytes": 830, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 133341.68820711784, "hops": {"1": 22970.73791595176, "2": 15797.11883305572, "3": 15682.611500029452, "4": 15838.979708962142, "5": 15930.869291070849, "6": 15628.01954196766, "7": 15836.956083076075, "8": 15656.395333004184}, "receiver_prefill_ms": 314.375750022009, "ttft_ms": 34.037375007756054, "decode_ms": 28.972083004191518}, "total_wall_ms": 133719.49941513594, "ttft_ms": 34.037375007756054, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 58, "facts_retained": 15, "retention_rate": 0.25862068965517243, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.2222222222222222, "status": 0.75, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 766}, "ledger_by_hop": {"1": {"facts_total": 58, "facts_retained": 15, "retention_rate": 0.25862068965517243, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.2222222222222222, "status": 0.75, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 766}, "2": {"facts_total": 58, "facts_retained": 15, "retention_rate": 0.25862068965517243, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.2222222222222222, "status": 0.75, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 766}, "3": {"facts_total": 58, "facts_retained": 15, "retention_rate": 0.25862068965517243, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.2222222222222222, "status": 0.75, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 766}, "4": {"facts_total": 58, "facts_retained": 15, "retention_rate": 0.25862068965517243, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.2222222222222222, "status": 0.75, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 766}, "5": {"facts_total": 58, "facts_retained": 15, "retention_rate": 0.25862068965517243, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.2222222222222222, "status": 0.75, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 766}, "6": {"facts_total": 58, "facts_retained": 15, "retention_rate": 0.25862068965517243, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.2222222222222222, "status": 0.75, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 766}, "7": {"facts_total": 58, "facts_retained": 15, "retention_rate": 0.25862068965517243, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.2222222222222222, "status": 0.75, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 766}, "8": {"facts_total": 58, "facts_retained": 15, "retention_rate": 0.25862068965517243, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.2222222222222222, "status": 0.75, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 766}}}, "reference_key_id": "keys/synthetic/syn_03003_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "a613cf19d60b1dbe7d9f51bbbf98459d9f181f51f14c39d755912e891ae7fd2f", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "cae050d1b6ef871aff4dbf70", "attempt_id": "703600239b06", "run_group_id": "E4_core:A5|closed|qwen3b:h8", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T06:43:26.883323Z", "task_family": "synthetic", "document_id": "syn_03003_2000", "document_hash": "dddca1a38e3224870fcfae927c701c400fd885c19123ea840ee06a76a140f407", "question_id": "q02_exact_copy", "split": "test", "cluster_id": "syn_03003_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 8, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1866, "query_tokens": 439, "message_tokens": 352, "message_utf8_bytes": 830, "cache_tensor_bytes": null, "serialized_payload_bytes": 830, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 133341.68820711784, "hops": {"1": 22970.73791595176, "2": 15797.11883305572, "3": 15682.611500029452, "4": 15838.979708962142, "5": 15930.869291070849, "6": 15628.01954196766, "7": 15836.956083076075, "8": 15656.395333004184}, "receiver_prefill_ms": 300.1935830106959, "ttft_ms": 61.47295795381069, "decode_ms": 64.16883296333253}, "total_wall_ms": 133768.12699914444, "ttft_ms": 61.47295795381069, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 58, "facts_retained": 15, "retention_rate": 0.25862068965517243, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.2222222222222222, "status": 0.75, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 766}, "ledger_by_hop": {"1": {"facts_total": 58, "facts_retained": 15, "retention_rate": 0.25862068965517243, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.2222222222222222, "status": 0.75, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 766}, "2": {"facts_total": 58, "facts_retained": 15, "retention_rate": 0.25862068965517243, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.2222222222222222, "status": 0.75, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 766}, "3": {"facts_total": 58, "facts_retained": 15, "retention_rate": 0.25862068965517243, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.2222222222222222, "status": 0.75, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 766}, "4": {"facts_total": 58, "facts_retained": 15, "retention_rate": 0.25862068965517243, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.2222222222222222, "status": 0.75, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 766}, "5": {"facts_total": 58, "facts_retained": 15, "retention_rate": 0.25862068965517243, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.2222222222222222, "status": 0.75, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 766}, "6": {"facts_total": 58, "facts_retained": 15, "retention_rate": 0.25862068965517243, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.2222222222222222, "status": 0.75, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 766}, "7": {"facts_total": 58, "facts_retained": 15, "retention_rate": 0.25862068965517243, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.2222222222222222, "status": 0.75, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 766}, "8": {"facts_total": 58, "facts_retained": 15, "retention_rate": 0.25862068965517243, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.2222222222222222, "status": 0.75, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 766}}}, "reference_key_id": "keys/synthetic/syn_03003_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "a613cf19d60b1dbe7d9f51bbbf98459d9f181f51f14c39d755912e891ae7fd2f", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "2f52325094cb1fee0eec2ed2", "attempt_id": "4774534e0d5f", "run_group_id": "E4_core:A5|closed|qwen3b:h8", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T06:43:27.326544Z", "task_family": "synthetic", "document_id": "syn_03003_2000", "document_hash": "dddca1a38e3224870fcfae927c701c400fd885c19123ea840ee06a76a140f407", "question_id": "q03_exact_copy", "split": "test", "cluster_id": "syn_03003_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 8, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1866, "query_tokens": 434, "message_tokens": 352, "message_utf8_bytes": 830, "cache_tensor_bytes": null, "serialized_payload_bytes": 830, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 133341.68820711784, "hops": {"1": 22970.73791595176, "2": 15797.11883305572, "3": 15682.611500029452, "4": 15838.979708962142, "5": 15930.869291070849, "6": 15628.01954196766, "7": 15836.956083076075, "8": 15656.395333004184}, "receiver_prefill_ms": 426.34266591630876, "ttft_ms": 33.10241608414799, "decode_ms": 33.027625060640275}, "total_wall_ms": 133834.858957096, "ttft_ms": 33.10241608414799, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 58, "facts_retained": 15, "retention_rate": 0.25862068965517243, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.2222222222222222, "status": 0.75, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 766}, "ledger_by_hop": {"1": {"facts_total": 58, "facts_retained": 15, "retention_rate": 0.25862068965517243, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.2222222222222222, "status": 0.75, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 766}, "2": {"facts_total": 58, "facts_retained": 15, "retention_rate": 0.25862068965517243, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.2222222222222222, "status": 0.75, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 766}, "3": {"facts_total": 58, "facts_retained": 15, "retention_rate": 0.25862068965517243, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.2222222222222222, "status": 0.75, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 766}, "4": {"facts_total": 58, "facts_retained": 15, "retention_rate": 0.25862068965517243, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.2222222222222222, "status": 0.75, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 766}, "5": {"facts_total": 58, "facts_retained": 15, "retention_rate": 0.25862068965517243, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.2222222222222222, "status": 0.75, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 766}, "6": {"facts_total": 58, "facts_retained": 15, "retention_rate": 0.25862068965517243, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.2222222222222222, "status": 0.75, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 766}, "7": {"facts_total": 58, "facts_retained": 15, "retention_rate": 0.25862068965517243, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.2222222222222222, "status": 0.75, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 766}, "8": {"facts_total": 58, "facts_retained": 15, "retention_rate": 0.25862068965517243, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.2222222222222222, "status": 0.75, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 766}}}, "reference_key_id": "keys/synthetic/syn_03003_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "a613cf19d60b1dbe7d9f51bbbf98459d9f181f51f14c39d755912e891ae7fd2f", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "09c4da4df2fc6d8e8ef6029c", "attempt_id": "d45c44e5fe88", "run_group_id": "E4_core:A5|closed|qwen3b:h8", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T06:43:27.837035Z", "task_family": "synthetic", "document_id": "syn_03003_2000", "document_hash": "dddca1a38e3224870fcfae927c701c400fd885c19123ea840ee06a76a140f407", "question_id": "q04_multi_hop", "split": "test", "cluster_id": "syn_03003_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 8, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1866, "query_tokens": 443, "message_tokens": 352, "message_utf8_bytes": 830, "cache_tensor_bytes": null, "serialized_payload_bytes": 830, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 133341.68820711784, "hops": {"1": 22970.73791595176, "2": 15797.11883305572, "3": 15682.611500029452, "4": 15838.979708962142, "5": 15930.869291070849, "6": 15628.01954196766, "7": 15836.956083076075, "8": 15656.395333004184}, "receiver_prefill_ms": 284.1750829247758, "ttft_ms": 24.927541962824762, "decode_ms": 44.21225003898144}, "total_wall_ms": 133695.53291611373, "ttft_ms": 24.927541962824762, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 58, "facts_retained": 15, "retention_rate": 0.25862068965517243, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.2222222222222222, "status": 0.75, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 766}, "ledger_by_hop": {"1": {"facts_total": 58, "facts_retained": 15, "retention_rate": 0.25862068965517243, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.2222222222222222, "status": 0.75, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 766}, "2": {"facts_total": 58, "facts_retained": 15, "retention_rate": 0.25862068965517243, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.2222222222222222, "status": 0.75, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 766}, "3": {"facts_total": 58, "facts_retained": 15, "retention_rate": 0.25862068965517243, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.2222222222222222, "status": 0.75, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 766}, "4": {"facts_total": 58, "facts_retained": 15, "retention_rate": 0.25862068965517243, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.2222222222222222, "status": 0.75, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 766}, "5": {"facts_total": 58, "facts_retained": 15, "retention_rate": 0.25862068965517243, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.2222222222222222, "status": 0.75, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 766}, "6": {"facts_total": 58, "facts_retained": 15, "retention_rate": 0.25862068965517243, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.2222222222222222, "status": 0.75, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 766}, "7": {"facts_total": 58, "facts_retained": 15, "retention_rate": 0.25862068965517243, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.2222222222222222, "status": 0.75, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 766}, "8": {"facts_total": 58, "facts_retained": 15, "retention_rate": 0.25862068965517243, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.2222222222222222, "status": 0.75, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 766}}}, "reference_key_id": "keys/synthetic/syn_03003_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "a613cf19d60b1dbe7d9f51bbbf98459d9f181f51f14c39d755912e891ae7fd2f", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "abe10e4f54ad508288fa3a2a", "attempt_id": "92076de595ea", "run_group_id": "E4_core:A5|closed|qwen3b:h8", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T06:43:28.208201Z", "task_family": "synthetic", "document_id": "syn_03003_2000", "document_hash": "dddca1a38e3224870fcfae927c701c400fd885c19123ea840ee06a76a140f407", "question_id": "q05_multi_hop", "split": "test", "cluster_id": "syn_03003_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 8, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1866, "query_tokens": 445, "message_tokens": 352, "message_utf8_bytes": 830, "cache_tensor_bytes": null, "serialized_payload_bytes": 830, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 133341.68820711784, "hops": {"1": 22970.73791595176, "2": 15797.11883305572, "3": 15682.611500029452, "4": 15838.979708962142, "5": 15930.869291070849, "6": 15628.01954196766, "7": 15836.956083076075, "8": 15656.395333004184}, "receiver_prefill_ms": 300.3198329824954, "ttft_ms": 21.3363750372082, "decode_ms": 40.35599995404482}, "total_wall_ms": 133704.21295717824, "ttft_ms": 21.3363750372082, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 58, "facts_retained": 15, "retention_rate": 0.25862068965517243, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.2222222222222222, "status": 0.75, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 766}, "ledger_by_hop": {"1": {"facts_total": 58, "facts_retained": 15, "retention_rate": 0.25862068965517243, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.2222222222222222, "status": 0.75, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 766}, "2": {"facts_total": 58, "facts_retained": 15, "retention_rate": 0.25862068965517243, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.2222222222222222, "status": 0.75, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 766}, "3": {"facts_total": 58, "facts_retained": 15, "retention_rate": 0.25862068965517243, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.2222222222222222, "status": 0.75, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 766}, "4": {"facts_total": 58, "facts_retained": 15, "retention_rate": 0.25862068965517243, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.2222222222222222, "status": 0.75, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 766}, "5": {"facts_total": 58, "facts_retained": 15, "retention_rate": 0.25862068965517243, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.2222222222222222, "status": 0.75, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 766}, "6": {"facts_total": 58, "facts_retained": 15, "retention_rate": 0.25862068965517243, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.2222222222222222, "status": 0.75, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 766}, "7": {"facts_total": 58, "facts_retained": 15, "retention_rate": 0.25862068965517243, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.2222222222222222, "status": 0.75, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 766}, "8": {"facts_total": 58, "facts_retained": 15, "retention_rate": 0.25862068965517243, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.2222222222222222, "status": 0.75, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 766}}}, "reference_key_id": "keys/synthetic/syn_03003_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "a613cf19d60b1dbe7d9f51bbbf98459d9f181f51f14c39d755912e891ae7fd2f", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "8fb1a88067ba951725a4c906", "attempt_id": "2e77c4c52715", "run_group_id": "E4_core:A5|closed|qwen3b:h8", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T06:43:28.587734Z", "task_family": "synthetic", "document_id": "syn_03003_2000", "document_hash": "dddca1a38e3224870fcfae927c701c400fd885c19123ea840ee06a76a140f407", "question_id": "q06_negation", "split": "test", "cluster_id": "syn_03003_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 8, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1866, "query_tokens": 437, "message_tokens": 352, "message_utf8_bytes": 830, "cache_tensor_bytes": null, "serialized_payload_bytes": 830, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 133341.68820711784, "hops": {"1": 22970.73791595176, "2": 15797.11883305572, "3": 15682.611500029452, "4": 15838.979708962142, "5": 15930.869291070849, "6": 15628.01954196766, "7": 15836.956083076075, "8": 15656.395333004184}, "receiver_prefill_ms": 256.7252499284223, "ttft_ms": 43.106333003379405, "decode_ms": 59.88304200582206}, "total_wall_ms": 133701.95683208294, "ttft_ms": 43.106333003379405, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "no", "evidence_artifact": {"ledger": {"facts_total": 58, "facts_retained": 15, "retention_rate": 0.25862068965517243, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.2222222222222222, "status": 0.75, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 766}, "ledger_by_hop": {"1": {"facts_total": 58, "facts_retained": 15, "retention_rate": 0.25862068965517243, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.2222222222222222, "status": 0.75, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 766}, "2": {"facts_total": 58, "facts_retained": 15, "retention_rate": 0.25862068965517243, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.2222222222222222, "status": 0.75, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 766}, "3": {"facts_total": 58, "facts_retained": 15, "retention_rate": 0.25862068965517243, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.2222222222222222, "status": 0.75, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 766}, "4": {"facts_total": 58, "facts_retained": 15, "retention_rate": 0.25862068965517243, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.2222222222222222, "status": 0.75, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 766}, "5": {"facts_total": 58, "facts_retained": 15, "retention_rate": 0.25862068965517243, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.2222222222222222, "status": 0.75, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 766}, "6": {"facts_total": 58, "facts_retained": 15, "retention_rate": 0.25862068965517243, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.2222222222222222, "status": 0.75, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 766}, "7": {"facts_total": 58, "facts_retained": 15, "retention_rate": 0.25862068965517243, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.2222222222222222, "status": 0.75, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 766}, "8": {"facts_total": 58, "facts_retained": 15, "retention_rate": 0.25862068965517243, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.2222222222222222, "status": 0.75, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 766}}}, "reference_key_id": "keys/synthetic/syn_03003_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "a613cf19d60b1dbe7d9f51bbbf98459d9f181f51f14c39d755912e891ae7fd2f", "output_hash": "9390298f3fb0c5b160498935d79cb139aef28e1c47358b4bbba61862b9c26e59"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "a20247d89710ae62402ce005", "attempt_id": "89afc9ab331c", "run_group_id": "E4_core:A5|closed|qwen3b:h8", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T06:43:28.965187Z", "task_family": "synthetic", "document_id": "syn_03003_2000", "document_hash": "dddca1a38e3224870fcfae927c701c400fd885c19123ea840ee06a76a140f407", "question_id": "q07_negation", "split": "test", "cluster_id": "syn_03003_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 8, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1866, "query_tokens": 434, "message_tokens": 352, "message_utf8_bytes": 830, "cache_tensor_bytes": null, "serialized_payload_bytes": 830, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 133341.68820711784, "hops": {"1": 22970.73791595176, "2": 15797.11883305572, "3": 15682.611500029452, "4": 15838.979708962142, "5": 15930.869291070849, "6": 15628.01954196766, "7": 15836.956083076075, "8": 15656.395333004184}, "receiver_prefill_ms": 451.51012495625764, "ttft_ms": 38.24541706126183, "decode_ms": 32.55412494763732}, "total_wall_ms": 133864.54483203124, "ttft_ms": 38.24541706126183, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 58, "facts_retained": 15, "retention_rate": 0.25862068965517243, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.2222222222222222, "status": 0.75, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 766}, "ledger_by_hop": {"1": {"facts_total": 58, "facts_retained": 15, "retention_rate": 0.25862068965517243, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.2222222222222222, "status": 0.75, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 766}, "2": {"facts_total": 58, "facts_retained": 15, "retention_rate": 0.25862068965517243, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.2222222222222222, "status": 0.75, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 766}, "3": {"facts_total": 58, "facts_retained": 15, "retention_rate": 0.25862068965517243, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.2222222222222222, "status": 0.75, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 766}, "4": {"facts_total": 58, "facts_retained": 15, "retention_rate": 0.25862068965517243, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.2222222222222222, "status": 0.75, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 766}, "5": {"facts_total": 58, "facts_retained": 15, "retention_rate": 0.25862068965517243, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.2222222222222222, "status": 0.75, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 766}, "6": {"facts_total": 58, "facts_retained": 15, "retention_rate": 0.25862068965517243, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.2222222222222222, "status": 0.75, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 766}, "7": {"facts_total": 58, "facts_retained": 15, "retention_rate": 0.25862068965517243, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.2222222222222222, "status": 0.75, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 766}, "8": {"facts_total": 58, "facts_retained": 15, "retention_rate": 0.25862068965517243, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.2222222222222222, "status": 0.75, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 766}}}, "reference_key_id": "keys/synthetic/syn_03003_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "a613cf19d60b1dbe7d9f51bbbf98459d9f181f51f14c39d755912e891ae7fd2f", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "08a346e8bdc51a33640a59d4", "attempt_id": "1099d2687559", "run_group_id": "E4_core:A5|closed|qwen3b:h8", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T06:43:29.506072Z", "task_family": "synthetic", "document_id": "syn_03003_2000", "document_hash": "dddca1a38e3224870fcfae927c701c400fd885c19123ea840ee06a76a140f407", "question_id": "q08_revision", "split": "test", "cluster_id": "syn_03003_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 8, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1866, "query_tokens": 452, "message_tokens": 352, "message_utf8_bytes": 830, "cache_tensor_bytes": null, "serialized_payload_bytes": 830, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 133341.68820711784, "hops": {"1": 22970.73791595176, "2": 15797.11883305572, "3": 15682.611500029452, "4": 15838.979708962142, "5": 15930.869291070849, "6": 15628.01954196766, "7": 15836.956083076075, "8": 15656.395333004184}, "receiver_prefill_ms": 309.98541601002216, "ttft_ms": 39.936708053573966, "decode_ms": 222.75283304043114}, "total_wall_ms": 133914.9266660679, "ttft_ms": 39.936708053573966, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 6}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "911925", "evidence_artifact": {"ledger": {"facts_total": 58, "facts_retained": 15, "retention_rate": 0.25862068965517243, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.2222222222222222, "status": 0.75, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 766}, "ledger_by_hop": {"1": {"facts_total": 58, "facts_retained": 15, "retention_rate": 0.25862068965517243, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.2222222222222222, "status": 0.75, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 766}, "2": {"facts_total": 58, "facts_retained": 15, "retention_rate": 0.25862068965517243, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.2222222222222222, "status": 0.75, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 766}, "3": {"facts_total": 58, "facts_retained": 15, "retention_rate": 0.25862068965517243, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.2222222222222222, "status": 0.75, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 766}, "4": {"facts_total": 58, "facts_retained": 15, "retention_rate": 0.25862068965517243, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.2222222222222222, "status": 0.75, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 766}, "5": {"facts_total": 58, "facts_retained": 15, "retention_rate": 0.25862068965517243, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.2222222222222222, "status": 0.75, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 766}, "6": {"facts_total": 58, "facts_retained": 15, "retention_rate": 0.25862068965517243, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.2222222222222222, "status": 0.75, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 766}, "7": {"facts_total": 58, "facts_retained": 15, "retention_rate": 0.25862068965517243, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.2222222222222222, "status": 0.75, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 766}, "8": {"facts_total": 58, "facts_retained": 15, "retention_rate": 0.25862068965517243, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.2222222222222222, "status": 0.75, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 766}}}, "reference_key_id": "keys/synthetic/syn_03003_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "a613cf19d60b1dbe7d9f51bbbf98459d9f181f51f14c39d755912e891ae7fd2f", "output_hash": "679a540dc4739f0ace623448b8fd925fb234f554838d7e87da223228ee81b112"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "c9574d8766ddbd41575e4d79", "attempt_id": "01753d51ad57", "run_group_id": "E4_core:A5|closed|qwen3b:h8", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T06:43:30.097754Z", "task_family": "synthetic", "document_id": "syn_03003_2000", "document_hash": "dddca1a38e3224870fcfae927c701c400fd885c19123ea840ee06a76a140f407", "question_id": "q09_revision", "split": "test", "cluster_id": "syn_03003_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 8, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1866, "query_tokens": 445, "message_tokens": 352, "message_utf8_bytes": 830, "cache_tensor_bytes": null, "serialized_payload_bytes": 830, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 133341.68820711784, "hops": {"1": 22970.73791595176, "2": 15797.11883305572, "3": 15682.611500029452, "4": 15838.979708962142, "5": 15930.869291070849, "6": 15628.01954196766, "7": 15836.956083076075, "8": 15656.395333004184}, "receiver_prefill_ms": 270.7115829689428, "ttft_ms": 45.580417034216225, "decode_ms": 465.79545794520527}, "total_wall_ms": 134124.36699902173, "ttft_ms": 45.580417034216225, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 6}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "911925", "evidence_artifact": {"ledger": {"facts_total": 58, "facts_retained": 15, "retention_rate": 0.25862068965517243, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.2222222222222222, "status": 0.75, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 766}, "ledger_by_hop": {"1": {"facts_total": 58, "facts_retained": 15, "retention_rate": 0.25862068965517243, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.2222222222222222, "status": 0.75, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 766}, "2": {"facts_total": 58, "facts_retained": 15, "retention_rate": 0.25862068965517243, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.2222222222222222, "status": 0.75, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 766}, "3": {"facts_total": 58, "facts_retained": 15, "retention_rate": 0.25862068965517243, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.2222222222222222, "status": 0.75, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 766}, "4": {"facts_total": 58, "facts_retained": 15, "retention_rate": 0.25862068965517243, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.2222222222222222, "status": 0.75, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 766}, "5": {"facts_total": 58, "facts_retained": 15, "retention_rate": 0.25862068965517243, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.2222222222222222, "status": 0.75, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 766}, "6": {"facts_total": 58, "facts_retained": 15, "retention_rate": 0.25862068965517243, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.2222222222222222, "status": 0.75, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 766}, "7": {"facts_total": 58, "facts_retained": 15, "retention_rate": 0.25862068965517243, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.2222222222222222, "status": 0.75, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 766}, "8": {"facts_total": 58, "facts_retained": 15, "retention_rate": 0.25862068965517243, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.2222222222222222, "status": 0.75, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 766}}}, "reference_key_id": "keys/synthetic/syn_03003_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "a613cf19d60b1dbe7d9f51bbbf98459d9f181f51f14c39d755912e891ae7fd2f", "output_hash": "679a540dc4739f0ace623448b8fd925fb234f554838d7e87da223228ee81b112"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "15b09571080df228f532305b", "attempt_id": "085337381b7e", "run_group_id": "E4_core:A5|closed|qwen3b:h8", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T06:43:30.898094Z", "task_family": "synthetic", "document_id": "syn_03003_2000", "document_hash": "dddca1a38e3224870fcfae927c701c400fd885c19123ea840ee06a76a140f407", "question_id": "q10_temporal", "split": "test", "cluster_id": "syn_03003_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 8, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1866, "query_tokens": 452, "message_tokens": 352, "message_utf8_bytes": 830, "cache_tensor_bytes": null, "serialized_payload_bytes": 830, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 133341.68820711784, "hops": {"1": 22970.73791595176, "2": 15797.11883305572, "3": 15682.611500029452, "4": 15838.979708962142, "5": 15930.869291070849, "6": 15628.01954196766, "7": 15836.956083076075, "8": 15656.395333004184}, "receiver_prefill_ms": 316.45491695962846, "ttft_ms": 36.80825000628829, "decode_ms": 25.504499906674027}, "total_wall_ms": 133720.8981651347, "ttft_ms": 36.80825000628829, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 58, "facts_retained": 15, "retention_rate": 0.25862068965517243, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.2222222222222222, "status": 0.75, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 766}, "ledger_by_hop": {"1": {"facts_total": 58, "facts_retained": 15, "retention_rate": 0.25862068965517243, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.2222222222222222, "status": 0.75, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 766}, "2": {"facts_total": 58, "facts_retained": 15, "retention_rate": 0.25862068965517243, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.2222222222222222, "status": 0.75, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 766}, "3": {"facts_total": 58, "facts_retained": 15, "retention_rate": 0.25862068965517243, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.2222222222222222, "status": 0.75, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 766}, "4": {"facts_total": 58, "facts_retained": 15, "retention_rate": 0.25862068965517243, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.2222222222222222, "status": 0.75, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 766}, "5": {"facts_total": 58, "facts_retained": 15, "retention_rate": 0.25862068965517243, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.2222222222222222, "status": 0.75, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 766}, "6": {"facts_total": 58, "facts_retained": 15, "retention_rate": 0.25862068965517243, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.2222222222222222, "status": 0.75, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 766}, "7": {"facts_total": 58, "facts_retained": 15, "retention_rate": 0.25862068965517243, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.2222222222222222, "status": 0.75, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 766}, "8": {"facts_total": 58, "facts_retained": 15, "retention_rate": 0.25862068965517243, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.2222222222222222, "status": 0.75, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 766}}}, "reference_key_id": "keys/synthetic/syn_03003_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "a613cf19d60b1dbe7d9f51bbbf98459d9f181f51f14c39d755912e891ae7fd2f", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "8a9f134052cf2066b93133c1", "attempt_id": "dcb51604bf63", "run_group_id": "E4_core:A5|closed|qwen3b:h8", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T06:43:31.293823Z", "task_family": "synthetic", "document_id": "syn_03003_2000", "document_hash": "dddca1a38e3224870fcfae927c701c400fd885c19123ea840ee06a76a140f407", "question_id": "q11_aggregation", "split": "test", "cluster_id": "syn_03003_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 8, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1866, "query_tokens": 446, "message_tokens": 352, "message_utf8_bytes": 830, "cache_tensor_bytes": null, "serialized_payload_bytes": 830, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 133341.68820711784, "hops": {"1": 22970.73791595176, "2": 15797.11883305572, "3": 15682.611500029452, "4": 15838.979708962142, "5": 15930.869291070849, "6": 15628.01954196766, "7": 15836.956083076075, "8": 15656.395333004184}, "receiver_prefill_ms": 295.1079170452431, "ttft_ms": 33.44495804049075, "decode_ms": 35.43845901731402}, "total_wall_ms": 133706.18008205201, "ttft_ms": 33.44495804049075, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 58, "facts_retained": 15, "retention_rate": 0.25862068965517243, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.2222222222222222, "status": 0.75, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 766}, "ledger_by_hop": {"1": {"facts_total": 58, "facts_retained": 15, "retention_rate": 0.25862068965517243, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.2222222222222222, "status": 0.75, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 766}, "2": {"facts_total": 58, "facts_retained": 15, "retention_rate": 0.25862068965517243, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.2222222222222222, "status": 0.75, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 766}, "3": {"facts_total": 58, "facts_retained": 15, "retention_rate": 0.25862068965517243, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.2222222222222222, "status": 0.75, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 766}, "4": {"facts_total": 58, "facts_retained": 15, "retention_rate": 0.25862068965517243, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.2222222222222222, "status": 0.75, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 766}, "5": {"facts_total": 58, "facts_retained": 15, "retention_rate": 0.25862068965517243, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.2222222222222222, "status": 0.75, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 766}, "6": {"facts_total": 58, "facts_retained": 15, "retention_rate": 0.25862068965517243, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.2222222222222222, "status": 0.75, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 766}, "7": {"facts_total": 58, "facts_retained": 15, "retention_rate": 0.25862068965517243, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.2222222222222222, "status": 0.75, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 766}, "8": {"facts_total": 58, "facts_retained": 15, "retention_rate": 0.25862068965517243, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.2222222222222222, "status": 0.75, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 766}}}, "reference_key_id": "keys/synthetic/syn_03003_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "a613cf19d60b1dbe7d9f51bbbf98459d9f181f51f14c39d755912e891ae7fd2f", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "995a2b49dd5d50c93d72ba0e", "attempt_id": "452ead529ffe", "run_group_id": "E4_core:A5|closed|qwen3b:h8", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T06:43:31.675971Z", "task_family": "synthetic", "document_id": "syn_03003_2000", "document_hash": "dddca1a38e3224870fcfae927c701c400fd885c19123ea840ee06a76a140f407", "question_id": "q12_unit_conversion", "split": "test", "cluster_id": "syn_03003_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 8, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1866, "query_tokens": 443, "message_tokens": 352, "message_utf8_bytes": 830, "cache_tensor_bytes": null, "serialized_payload_bytes": 830, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 133341.68820711784, "hops": {"1": 22970.73791595176, "2": 15797.11883305572, "3": 15682.611500029452, "4": 15838.979708962142, "5": 15930.869291070849, "6": 15628.01954196766, "7": 15836.956083076075, "8": 15656.395333004184}, "receiver_prefill_ms": 259.6864169463515, "ttft_ms": 58.22087510023266, "decode_ms": 259.3665000749752}, "total_wall_ms": 133919.50441512745, "ttft_ms": 58.22087510023266, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 5}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "17430", "evidence_artifact": {"ledger": {"facts_total": 58, "facts_retained": 15, "retention_rate": 0.25862068965517243, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.2222222222222222, "status": 0.75, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 766}, "ledger_by_hop": {"1": {"facts_total": 58, "facts_retained": 15, "retention_rate": 0.25862068965517243, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.2222222222222222, "status": 0.75, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 766}, "2": {"facts_total": 58, "facts_retained": 15, "retention_rate": 0.25862068965517243, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.2222222222222222, "status": 0.75, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 766}, "3": {"facts_total": 58, "facts_retained": 15, "retention_rate": 0.25862068965517243, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.2222222222222222, "status": 0.75, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 766}, "4": {"facts_total": 58, "facts_retained": 15, "retention_rate": 0.25862068965517243, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.2222222222222222, "status": 0.75, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 766}, "5": {"facts_total": 58, "facts_retained": 15, "retention_rate": 0.25862068965517243, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.2222222222222222, "status": 0.75, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 766}, "6": {"facts_total": 58, "facts_retained": 15, "retention_rate": 0.25862068965517243, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.2222222222222222, "status": 0.75, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 766}, "7": {"facts_total": 58, "facts_retained": 15, "retention_rate": 0.25862068965517243, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.2222222222222222, "status": 0.75, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 766}, "8": {"facts_total": 58, "facts_retained": 15, "retention_rate": 0.25862068965517243, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.2222222222222222, "status": 0.75, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 766}}}, "reference_key_id": "keys/synthetic/syn_03003_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "a613cf19d60b1dbe7d9f51bbbf98459d9f181f51f14c39d755912e891ae7fd2f", "output_hash": "2575eaf3b748efc209b676518766751eb038c88f74578951345818bfe7f7194b"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "ead6a3fa1aab5a04431fd23c", "attempt_id": "b52ca23bea21", "run_group_id": "E4_core:A5|closed|qwen3b:h8", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T06:43:32.271287Z", "task_family": "synthetic", "document_id": "syn_03003_2000", "document_hash": "dddca1a38e3224870fcfae927c701c400fd885c19123ea840ee06a76a140f407", "question_id": "q13_unanswerable", "split": "test", "cluster_id": "syn_03003_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 8, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1866, "query_tokens": 447, "message_tokens": 352, "message_utf8_bytes": 830, "cache_tensor_bytes": null, "serialized_payload_bytes": 830, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 133341.68820711784, "hops": {"1": 22970.73791595176, "2": 15797.11883305572, "3": 15682.611500029452, "4": 15838.979708962142, "5": 15930.869291070849, "6": 15628.01954196766, "7": 15836.956083076075, "8": 15656.395333004184}, "receiver_prefill_ms": 409.0606670361012, "ttft_ms": 40.21554091013968, "decode_ms": 27.111083967611194}, "total_wall_ms": 133818.52966512088, "ttft_ms": 40.21554091013968, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 58, "facts_retained": 15, "retention_rate": 0.25862068965517243, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.2222222222222222, "status": 0.75, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 766}, "ledger_by_hop": {"1": {"facts_total": 58, "facts_retained": 15, "retention_rate": 0.25862068965517243, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.2222222222222222, "status": 0.75, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 766}, "2": {"facts_total": 58, "facts_retained": 15, "retention_rate": 0.25862068965517243, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.2222222222222222, "status": 0.75, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 766}, "3": {"facts_total": 58, "facts_retained": 15, "retention_rate": 0.25862068965517243, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.2222222222222222, "status": 0.75, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 766}, "4": {"facts_total": 58, "facts_retained": 15, "retention_rate": 0.25862068965517243, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.2222222222222222, "status": 0.75, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 766}, "5": {"facts_total": 58, "facts_retained": 15, "retention_rate": 0.25862068965517243, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.2222222222222222, "status": 0.75, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 766}, "6": {"facts_total": 58, "facts_retained": 15, "retention_rate": 0.25862068965517243, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.2222222222222222, "status": 0.75, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 766}, "7": {"facts_total": 58, "facts_retained": 15, "retention_rate": 0.25862068965517243, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.2222222222222222, "status": 0.75, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 766}, "8": {"facts_total": 58, "facts_retained": 15, "retention_rate": 0.25862068965517243, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.2222222222222222, "status": 0.75, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 766}}}, "reference_key_id": "keys/synthetic/syn_03003_2000.json", "task_scores": {"correct": 1.0, "abstained": 1.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "a613cf19d60b1dbe7d9f51bbbf98459d9f181f51f14c39d755912e891ae7fd2f", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "22c0feb365e2e63938601412", "attempt_id": "077bfbb13c85", "run_group_id": "E4_core:A5|closed|qwen3b:h8", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T06:43:32.765834Z", "task_family": "synthetic", "document_id": "syn_03003_2000", "document_hash": "dddca1a38e3224870fcfae927c701c400fd885c19123ea840ee06a76a140f407", "question_id": "q14_unanswerable", "split": "test", "cluster_id": "syn_03003_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 8, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1866, "query_tokens": 448, "message_tokens": 352, "message_utf8_bytes": 830, "cache_tensor_bytes": null, "serialized_payload_bytes": 830, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 133341.68820711784, "hops": {"1": 22970.73791595176, "2": 15797.11883305572, "3": 15682.611500029452, "4": 15838.979708962142, "5": 15930.869291070849, "6": 15628.01954196766, "7": 15836.956083076075, "8": 15656.395333004184}, "receiver_prefill_ms": 294.93237496353686, "ttft_ms": 40.09729204699397, "decode_ms": 25.09983303025365}, "total_wall_ms": 133702.27483217604, "ttft_ms": 40.09729204699397, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 58, "facts_retained": 15, "retention_rate": 0.25862068965517243, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.2222222222222222, "status": 0.75, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 766}, "ledger_by_hop": {"1": {"facts_total": 58, "facts_retained": 15, "retention_rate": 0.25862068965517243, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.2222222222222222, "status": 0.75, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 766}, "2": {"facts_total": 58, "facts_retained": 15, "retention_rate": 0.25862068965517243, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.2222222222222222, "status": 0.75, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 766}, "3": {"facts_total": 58, "facts_retained": 15, "retention_rate": 0.25862068965517243, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.2222222222222222, "status": 0.75, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 766}, "4": {"facts_total": 58, "facts_retained": 15, "retention_rate": 0.25862068965517243, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.2222222222222222, "status": 0.75, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 766}, "5": {"facts_total": 58, "facts_retained": 15, "retention_rate": 0.25862068965517243, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.2222222222222222, "status": 0.75, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 766}, "6": {"facts_total": 58, "facts_retained": 15, "retention_rate": 0.25862068965517243, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.2222222222222222, "status": 0.75, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 766}, "7": {"facts_total": 58, "facts_retained": 15, "retention_rate": 0.25862068965517243, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.2222222222222222, "status": 0.75, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 766}, "8": {"facts_total": 58, "facts_retained": 15, "retention_rate": 0.25862068965517243, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.2222222222222222, "status": 0.75, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 766}}}, "reference_key_id": "keys/synthetic/syn_03003_2000.json", "task_scores": {"correct": 1.0, "abstained": 1.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "a613cf19d60b1dbe7d9f51bbbf98459d9f181f51f14c39d755912e891ae7fd2f", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "c1ed1bfe9ddb973a29f2bad9", "attempt_id": "6d85468a13c4", "run_group_id": "E4_core:A4|repairable|qwen3b:h1", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T06:43:56.213596Z", "task_family": "synthetic", "document_id": "syn_03003_2000", "document_hash": "dddca1a38e3224870fcfae927c701c400fd885c19123ea840ee06a76a140f407", "question_id": "q00_single_fact", "split": "test", "cluster_id": "syn_03003_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 1, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1866, "query_tokens": 594, "message_tokens": 512, "message_utf8_bytes": 1200, "cache_tensor_bytes": null, "serialized_payload_bytes": 1200, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 23068.98970901966, "hops": {"1": 23068.98970901966}, "receiver_prefill_ms": 405.07683297619224, "ttft_ms": 23.47020898014307, "decode_ms": 326.59574993886054}, "total_wall_ms": 23824.778917012736, "ttft_ms": 23.47020898014307, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 10}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "GR-79S-5839", "evidence_artifact": {"ledger": {"facts_total": 58, "facts_retained": 32, "retention_rate": 0.5517241379310345, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.7142857142857143, "amount": 0.6666666666666666, "status": 0.6666666666666666, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1200}, "ledger_by_hop": {"1": {"facts_total": 58, "facts_retained": 32, "retention_rate": 0.5517241379310345, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.7142857142857143, "amount": 0.6666666666666666, "status": 0.6666666666666666, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1200}}}, "reference_key_id": "keys/synthetic/syn_03003_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "6d2a2bad9425191f4568f7eb115d63c953e27080a86953f62a982ad73340e770", "output_hash": "720cc40cd69e85e9728f6bea00b17a95ca07eee643e143a131840763958303cd"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "db1345062b3f17cc299aedbd", "attempt_id": "5ccfef6f3f29", "run_group_id": "E4_core:A4|repairable|qwen3b:h1", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T06:43:56.986622Z", "task_family": "synthetic", "document_id": "syn_03003_2000", "document_hash": "dddca1a38e3224870fcfae927c701c400fd885c19123ea840ee06a76a140f407", "question_id": "q01_single_fact", "split": "test", "cluster_id": "syn_03003_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 1, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1866, "query_tokens": 598, "message_tokens": 512, "message_utf8_bytes": 1200, "cache_tensor_bytes": null, "serialized_payload_bytes": 1200, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 23068.98970901966, "hops": {"1": 23068.98970901966}, "receiver_prefill_ms": 414.36433396302164, "ttft_ms": 35.411208984442055, "decode_ms": 588.2367079611868}, "total_wall_ms": 24107.809251057915, "ttft_ms": 35.411208984442055, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 11}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "Lot 344-B, Témiscaming", "evidence_artifact": {"ledger": {"facts_total": 58, "facts_retained": 32, "retention_rate": 0.5517241379310345, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.7142857142857143, "amount": 0.6666666666666666, "status": 0.6666666666666666, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1200}, "ledger_by_hop": {"1": {"facts_total": 58, "facts_retained": 32, "retention_rate": 0.5517241379310345, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.7142857142857143, "amount": 0.6666666666666666, "status": 0.6666666666666666, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1200}}}, "reference_key_id": "keys/synthetic/syn_03003_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 0.8571428571428571}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "6d2a2bad9425191f4568f7eb115d63c953e27080a86953f62a982ad73340e770", "output_hash": "0197d491a5a04a142dc8e372360735a1d41757e7a5aea7e81f23b0e1ef6ee010"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "df9240677c9c81854bc2df43", "attempt_id": "2877100d87a2", "run_group_id": "E4_core:A4|repairable|qwen3b:h1", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T06:43:58.045324Z", "task_family": "synthetic", "document_id": "syn_03003_2000", "document_hash": "dddca1a38e3224870fcfae927c701c400fd885c19123ea840ee06a76a140f407", "question_id": "q02_exact_copy", "split": "test", "cluster_id": "syn_03003_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 1, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1866, "query_tokens": 598, "message_tokens": 512, "message_utf8_bytes": 1200, "cache_tensor_bytes": null, "serialized_payload_bytes": 1200, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 23068.98970901966, "hops": {"1": 23068.98970901966}, "receiver_prefill_ms": 412.6571250380948, "ttft_ms": 35.909042111597955, "decode_ms": 38.6912920512259}, "total_wall_ms": 23556.83150002733, "ttft_ms": 35.909042111597955, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 58, "facts_retained": 32, "retention_rate": 0.5517241379310345, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.7142857142857143, "amount": 0.6666666666666666, "status": 0.6666666666666666, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1200}, "ledger_by_hop": {"1": {"facts_total": 58, "facts_retained": 32, "retention_rate": 0.5517241379310345, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.7142857142857143, "amount": 0.6666666666666666, "status": 0.6666666666666666, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1200}}}, "reference_key_id": "keys/synthetic/syn_03003_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "6d2a2bad9425191f4568f7eb115d63c953e27080a86953f62a982ad73340e770", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "40235a047682faf847bbc5a2", "attempt_id": "3e7e51e0c39c", "run_group_id": "E4_core:A4|repairable|qwen3b:h1", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T06:43:58.550976Z", "task_family": "synthetic", "document_id": "syn_03003_2000", "document_hash": "dddca1a38e3224870fcfae927c701c400fd885c19123ea840ee06a76a140f407", "question_id": "q03_exact_copy", "split": "test", "cluster_id": "syn_03003_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 1, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1866, "query_tokens": 593, "message_tokens": 512, "message_utf8_bytes": 1200, "cache_tensor_bytes": null, "serialized_payload_bytes": 1200, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 23068.98970901966, "hops": {"1": 23068.98970901966}, "receiver_prefill_ms": 373.8131249556318, "ttft_ms": 61.725249979645014, "decode_ms": 405.81333392765373}, "total_wall_ms": 23911.020708968863, "ttft_ms": 61.725249979645014, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 6}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "Wilhelmina Achterberg", "evidence_artifact": {"ledger": {"facts_total": 58, "facts_retained": 32, "retention_rate": 0.5517241379310345, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.7142857142857143, "amount": 0.6666666666666666, "status": 0.6666666666666666, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1200}, "ledger_by_hop": {"1": {"facts_total": 58, "facts_retained": 32, "retention_rate": 0.5517241379310345, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.7142857142857143, "amount": 0.6666666666666666, "status": 0.6666666666666666, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1200}}}, "reference_key_id": "keys/synthetic/syn_03003_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "6d2a2bad9425191f4568f7eb115d63c953e27080a86953f62a982ad73340e770", "output_hash": "2a62606853c315bba8769371d195e3ab3d6a8cd7ef959205575c5119740850f8"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "6f08619645f606762eee41b5", "attempt_id": "39a17dccb5f4", "run_group_id": "E4_core:A4|repairable|qwen3b:h1", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T06:43:59.410877Z", "task_family": "synthetic", "document_id": "syn_03003_2000", "document_hash": "dddca1a38e3224870fcfae927c701c400fd885c19123ea840ee06a76a140f407", "question_id": "q04_multi_hop", "split": "test", "cluster_id": "syn_03003_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 1, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1866, "query_tokens": 602, "message_tokens": 512, "message_utf8_bytes": 1200, "cache_tensor_bytes": null, "serialized_payload_bytes": 1200, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 23068.98970901966, "hops": {"1": 23068.98970901966}, "receiver_prefill_ms": 406.0356660047546, "ttft_ms": 27.539375005289912, "decode_ms": 36.25587502028793}, "total_wall_ms": 23539.372334023938, "ttft_ms": 27.539375005289912, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 58, "facts_retained": 32, "retention_rate": 0.5517241379310345, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.7142857142857143, "amount": 0.6666666666666666, "status": 0.6666666666666666, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1200}, "ledger_by_hop": {"1": {"facts_total": 58, "facts_retained": 32, "retention_rate": 0.5517241379310345, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.7142857142857143, "amount": 0.6666666666666666, "status": 0.6666666666666666, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1200}}}, "reference_key_id": "keys/synthetic/syn_03003_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "6d2a2bad9425191f4568f7eb115d63c953e27080a86953f62a982ad73340e770", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "a0be8fd8989be4ed68fce5b3", "attempt_id": "86fc373ae3e0", "run_group_id": "E4_core:A4|repairable|qwen3b:h1", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T06:43:59.898063Z", "task_family": "synthetic", "document_id": "syn_03003_2000", "document_hash": "dddca1a38e3224870fcfae927c701c400fd885c19123ea840ee06a76a140f407", "question_id": "q05_multi_hop", "split": "test", "cluster_id": "syn_03003_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 1, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1866, "query_tokens": 604, "message_tokens": 512, "message_utf8_bytes": 1200, "cache_tensor_bytes": null, "serialized_payload_bytes": 1200, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 23068.98970901966, "hops": {"1": 23068.98970901966}, "receiver_prefill_ms": 386.72416703775525, "ttft_ms": 13.899625046178699, "decode_ms": 629.8095419770107}, "total_wall_ms": 24100.194042082876, "ttft_ms": 13.899625046178699, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 10}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "NK-95Z-2165", "evidence_artifact": {"ledger": {"facts_total": 58, "facts_retained": 32, "retention_rate": 0.5517241379310345, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.7142857142857143, "amount": 0.6666666666666666, "status": 0.6666666666666666, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1200}, "ledger_by_hop": {"1": {"facts_total": 58, "facts_retained": 32, "retention_rate": 0.5517241379310345, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.7142857142857143, "amount": 0.6666666666666666, "status": 0.6666666666666666, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1200}}}, "reference_key_id": "keys/synthetic/syn_03003_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "6d2a2bad9425191f4568f7eb115d63c953e27080a86953f62a982ad73340e770", "output_hash": "bd8f809b02304c6b9dc7c1e116ecccc2328c976a5c071e89a94d661b48e73aa7"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "ea297491284153822c31fcdd", "attempt_id": "ec6135b831e8", "run_group_id": "E4_core:A4|repairable|qwen3b:h1", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T06:44:00.946598Z", "task_family": "synthetic", "document_id": "syn_03003_2000", "document_hash": "dddca1a38e3224870fcfae927c701c400fd885c19123ea840ee06a76a140f407", "question_id": "q06_negation", "split": "test", "cluster_id": "syn_03003_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 1, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1866, "query_tokens": 596, "message_tokens": 512, "message_utf8_bytes": 1200, "cache_tensor_bytes": null, "serialized_payload_bytes": 1200, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 23068.98970901966, "hops": {"1": 23068.98970901966}, "receiver_prefill_ms": 402.4522079853341, "ttft_ms": 26.37858293019235, "decode_ms": 38.641792023554444}, "total_wall_ms": 23537.051251041703, "ttft_ms": 26.37858293019235, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "yes", "evidence_artifact": {"ledger": {"facts_total": 58, "facts_retained": 32, "retention_rate": 0.5517241379310345, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.7142857142857143, "amount": 0.6666666666666666, "status": 0.6666666666666666, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1200}, "ledger_by_hop": {"1": {"facts_total": 58, "facts_retained": 32, "retention_rate": 0.5517241379310345, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.7142857142857143, "amount": 0.6666666666666666, "status": 0.6666666666666666, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1200}}}, "reference_key_id": "keys/synthetic/syn_03003_2000.json", "task_scores": {"correct": 0.0, "abstained": 0.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "6d2a2bad9425191f4568f7eb115d63c953e27080a86953f62a982ad73340e770", "output_hash": "8a798890fe93817163b10b5f7bd2ca4d25d84c52739a645a889c173eee7d9d3d"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "ee6b51fc63f68521dfb5bd36", "attempt_id": "a30f485e01f8", "run_group_id": "E4_core:A4|repairable|qwen3b:h1", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T06:44:01.431804Z", "task_family": "synthetic", "document_id": "syn_03003_2000", "document_hash": "dddca1a38e3224870fcfae927c701c400fd885c19123ea840ee06a76a140f407", "question_id": "q07_negation", "split": "test", "cluster_id": "syn_03003_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 1, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1866, "query_tokens": 593, "message_tokens": 512, "message_utf8_bytes": 1200, "cache_tensor_bytes": null, "serialized_payload_bytes": 1200, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 23068.98970901966, "hops": {"1": 23068.98970901966}, "receiver_prefill_ms": 416.8475840706378, "ttft_ms": 20.560791017487645, "decode_ms": 19.97158396989107}, "total_wall_ms": 23527.0890010288, "ttft_ms": 20.560791017487645, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "no", "evidence_artifact": {"ledger": {"facts_total": 58, "facts_retained": 32, "retention_rate": 0.5517241379310345, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.7142857142857143, "amount": 0.6666666666666666, "status": 0.6666666666666666, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1200}, "ledger_by_hop": {"1": {"facts_total": 58, "facts_retained": 32, "retention_rate": 0.5517241379310345, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.7142857142857143, "amount": 0.6666666666666666, "status": 0.6666666666666666, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1200}}}, "reference_key_id": "keys/synthetic/syn_03003_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "6d2a2bad9425191f4568f7eb115d63c953e27080a86953f62a982ad73340e770", "output_hash": "9390298f3fb0c5b160498935d79cb139aef28e1c47358b4bbba61862b9c26e59"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "153fac8321b9242d5fc8d9d9", "attempt_id": "e71420769a7b", "run_group_id": "E4_core:A4|repairable|qwen3b:h1", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T06:44:01.907557Z", "task_family": "synthetic", "document_id": "syn_03003_2000", "document_hash": "dddca1a38e3224870fcfae927c701c400fd885c19123ea840ee06a76a140f407", "question_id": "q08_revision", "split": "test", "cluster_id": "syn_03003_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 1, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1866, "query_tokens": 611, "message_tokens": 512, "message_utf8_bytes": 1200, "cache_tensor_bytes": null, "serialized_payload_bytes": 1200, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 23068.98970901966, "hops": {"1": 23068.98970901966}, "receiver_prefill_ms": 408.82554196286947, "ttft_ms": 39.890917018055916, "decode_ms": 425.7231659721583}, "total_wall_ms": 23944.09025099594, "ttft_ms": 39.890917018055916, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 7}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "911,925", "evidence_artifact": {"ledger": {"facts_total": 58, "facts_retained": 32, "retention_rate": 0.5517241379310345, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.7142857142857143, "amount": 0.6666666666666666, "status": 0.6666666666666666, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1200}, "ledger_by_hop": {"1": {"facts_total": 58, "facts_retained": 32, "retention_rate": 0.5517241379310345, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.7142857142857143, "amount": 0.6666666666666666, "status": 0.6666666666666666, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1200}}}, "reference_key_id": "keys/synthetic/syn_03003_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "6d2a2bad9425191f4568f7eb115d63c953e27080a86953f62a982ad73340e770", "output_hash": "86be78720fc4b79f93e603e7d9e4861fcd25102612f73209fffc17d787056007"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "3fc0ce1dbc7f000e4dfa759d", "attempt_id": "877451fbad55", "run_group_id": "E4_core:A4|repairable|qwen3b:h1", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T06:44:02.803461Z", "task_family": "synthetic", "document_id": "syn_03003_2000", "document_hash": "dddca1a38e3224870fcfae927c701c400fd885c19123ea840ee06a76a140f407", "question_id": "q09_revision", "split": "test", "cluster_id": "syn_03003_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 1, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1866, "query_tokens": 604, "message_tokens": 512, "message_utf8_bytes": 1200, "cache_tensor_bytes": null, "serialized_payload_bytes": 1200, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 23068.98970901966, "hops": {"1": 23068.98970901966}, "receiver_prefill_ms": 397.94600009918213, "ttft_ms": 21.296791965141892, "decode_ms": 239.08083292189986}, "total_wall_ms": 23727.92550106533, "ttft_ms": 21.296791965141892, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 7}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "911,925", "evidence_artifact": {"ledger": {"facts_total": 58, "facts_retained": 32, "retention_rate": 0.5517241379310345, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.7142857142857143, "amount": 0.6666666666666666, "status": 0.6666666666666666, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1200}, "ledger_by_hop": {"1": {"facts_total": 58, "facts_retained": 32, "retention_rate": 0.5517241379310345, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.7142857142857143, "amount": 0.6666666666666666, "status": 0.6666666666666666, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1200}}}, "reference_key_id": "keys/synthetic/syn_03003_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "6d2a2bad9425191f4568f7eb115d63c953e27080a86953f62a982ad73340e770", "output_hash": "86be78720fc4b79f93e603e7d9e4861fcd25102612f73209fffc17d787056007"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "471de1d3ade48893b9eaba85", "attempt_id": "c6f35883395e", "run_group_id": "E4_core:A4|repairable|qwen3b:h1", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T06:44:03.480347Z", "task_family": "synthetic", "document_id": "syn_03003_2000", "document_hash": "dddca1a38e3224870fcfae927c701c400fd885c19123ea840ee06a76a140f407", "question_id": "q10_temporal", "split": "test", "cluster_id": "syn_03003_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 1, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1866, "query_tokens": 611, "message_tokens": 512, "message_utf8_bytes": 1200, "cache_tensor_bytes": null, "serialized_payload_bytes": 1200, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 23068.98970901966, "hops": {"1": 23068.98970901966}, "receiver_prefill_ms": 356.1161250108853, "ttft_ms": 52.12195799686015, "decode_ms": 421.5335000772029}, "total_wall_ms": 23899.51295906212, "ttft_ms": 52.12195799686015, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 6}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "Gamtiszen-2", "evidence_artifact": {"ledger": {"facts_total": 58, "facts_retained": 32, "retention_rate": 0.5517241379310345, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.7142857142857143, "amount": 0.6666666666666666, "status": 0.6666666666666666, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1200}, "ledger_by_hop": {"1": {"facts_total": 58, "facts_retained": 32, "retention_rate": 0.5517241379310345, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.7142857142857143, "amount": 0.6666666666666666, "status": 0.6666666666666666, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1200}}}, "reference_key_id": "keys/synthetic/syn_03003_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "6d2a2bad9425191f4568f7eb115d63c953e27080a86953f62a982ad73340e770", "output_hash": "32e3af428886434823d82102e1c482765eafdf6077fe8d3fba5752ec5e29826c"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "1fdd0639e5ebe42323846452", "attempt_id": "dc5a59f17dbd", "run_group_id": "E4_core:A4|repairable|qwen3b:h1", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T06:44:04.329411Z", "task_family": "synthetic", "document_id": "syn_03003_2000", "document_hash": "dddca1a38e3224870fcfae927c701c400fd885c19123ea840ee06a76a140f407", "question_id": "q11_aggregation", "split": "test", "cluster_id": "syn_03003_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 1, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1866, "query_tokens": 605, "message_tokens": 512, "message_utf8_bytes": 1200, "cache_tensor_bytes": null, "serialized_payload_bytes": 1200, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 23068.98970901966, "hops": {"1": 23068.98970901966}, "receiver_prefill_ms": 414.44408299867064, "ttft_ms": 31.73166699707508, "decode_ms": 235.6842498993501}, "total_wall_ms": 23751.592876040377, "ttft_ms": 31.73166699707508, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 7}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "3397775", "evidence_artifact": {"ledger": {"facts_total": 58, "facts_retained": 32, "retention_rate": 0.5517241379310345, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.7142857142857143, "amount": 0.6666666666666666, "status": 0.6666666666666666, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1200}, "ledger_by_hop": {"1": {"facts_total": 58, "facts_retained": 32, "retention_rate": 0.5517241379310345, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.7142857142857143, "amount": 0.6666666666666666, "status": 0.6666666666666666, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1200}}}, "reference_key_id": "keys/synthetic/syn_03003_2000.json", "task_scores": {"correct": 0.0, "abstained": 0.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "6d2a2bad9425191f4568f7eb115d63c953e27080a86953f62a982ad73340e770", "output_hash": "e971a34f9c497d953f98fb387e189b1587b11a7f6725b2f81ac6d0f3ef8ac845"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "73ca4485d24f2fb881d547e6", "attempt_id": "5e98828b76c2", "run_group_id": "E4_core:A4|repairable|qwen3b:h1", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T06:44:05.030252Z", "task_family": "synthetic", "document_id": "syn_03003_2000", "document_hash": "dddca1a38e3224870fcfae927c701c400fd885c19123ea840ee06a76a140f407", "question_id": "q12_unit_conversion", "split": "test", "cluster_id": "syn_03003_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 1, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1866, "query_tokens": 602, "message_tokens": 512, "message_utf8_bytes": 1200, "cache_tensor_bytes": null, "serialized_payload_bytes": 1200, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 23068.98970901966, "hops": {"1": 23068.98970901966}, "receiver_prefill_ms": 345.23637499660254, "ttft_ms": 44.31583301629871, "decode_ms": 397.9607500368729}, "total_wall_ms": 23857.262209057808, "ttft_ms": 44.31583301629871, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 5}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "17430", "evidence_artifact": {"ledger": {"facts_total": 58, "facts_retained": 32, "retention_rate": 0.5517241379310345, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.7142857142857143, "amount": 0.6666666666666666, "status": 0.6666666666666666, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1200}, "ledger_by_hop": {"1": {"facts_total": 58, "facts_retained": 32, "retention_rate": 0.5517241379310345, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.7142857142857143, "amount": 0.6666666666666666, "status": 0.6666666666666666, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1200}}}, "reference_key_id": "keys/synthetic/syn_03003_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "6d2a2bad9425191f4568f7eb115d63c953e27080a86953f62a982ad73340e770", "output_hash": "2575eaf3b748efc209b676518766751eb038c88f74578951345818bfe7f7194b"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "5221b65eefe296c2d3f4380a", "attempt_id": "00845ab39dc5", "run_group_id": "E4_core:A4|repairable|qwen3b:h1", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T06:44:05.835244Z", "task_family": "synthetic", "document_id": "syn_03003_2000", "document_hash": "dddca1a38e3224870fcfae927c701c400fd885c19123ea840ee06a76a140f407", "question_id": "q13_unanswerable", "split": "test", "cluster_id": "syn_03003_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 1, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1866, "query_tokens": 606, "message_tokens": 512, "message_utf8_bytes": 1200, "cache_tensor_bytes": null, "serialized_payload_bytes": 1200, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 23068.98970901966, "hops": {"1": 23068.98970901966}, "receiver_prefill_ms": 396.57304191496223, "ttft_ms": 27.52637502271682, "decode_ms": 41.004833998158574}, "total_wall_ms": 23534.68912607059, "ttft_ms": 27.52637502271682, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 58, "facts_retained": 32, "retention_rate": 0.5517241379310345, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.7142857142857143, "amount": 0.6666666666666666, "status": 0.6666666666666666, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1200}, "ledger_by_hop": {"1": {"facts_total": 58, "facts_retained": 32, "retention_rate": 0.5517241379310345, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.7142857142857143, "amount": 0.6666666666666666, "status": 0.6666666666666666, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1200}}}, "reference_key_id": "keys/synthetic/syn_03003_2000.json", "task_scores": {"correct": 1.0, "abstained": 1.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "6d2a2bad9425191f4568f7eb115d63c953e27080a86953f62a982ad73340e770", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "f344c6e37b5b239d57e49993", "attempt_id": "7877a59ac3f3", "run_group_id": "E4_core:A4|repairable|qwen3b:h1", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T06:44:06.318374Z", "task_family": "synthetic", "document_id": "syn_03003_2000", "document_hash": "dddca1a38e3224870fcfae927c701c400fd885c19123ea840ee06a76a140f407", "question_id": "q14_unanswerable", "split": "test", "cluster_id": "syn_03003_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 1, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1866, "query_tokens": 607, "message_tokens": 512, "message_utf8_bytes": 1200, "cache_tensor_bytes": null, "serialized_payload_bytes": 1200, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 23068.98970901966, "hops": {"1": 23068.98970901966}, "receiver_prefill_ms": 404.2461250210181, "ttft_ms": 34.617583034560084, "decode_ms": 34.55870796460658}, "total_wall_ms": 23543.159876018763, "ttft_ms": 34.617583034560084, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 58, "facts_retained": 32, "retention_rate": 0.5517241379310345, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.7142857142857143, "amount": 0.6666666666666666, "status": 0.6666666666666666, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1200}, "ledger_by_hop": {"1": {"facts_total": 58, "facts_retained": 32, "retention_rate": 0.5517241379310345, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.7142857142857143, "amount": 0.6666666666666666, "status": 0.6666666666666666, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1200}}}, "reference_key_id": "keys/synthetic/syn_03003_2000.json", "task_scores": {"correct": 1.0, "abstained": 1.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "6d2a2bad9425191f4568f7eb115d63c953e27080a86953f62a982ad73340e770", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "1ae3d0aeb26d433a94dbaea0", "attempt_id": "ae6f7038ff61", "run_group_id": "E4_core:A4|repairable|qwen3b:h2", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T06:44:28.624121Z", "task_family": "synthetic", "document_id": "syn_03003_2000", "document_hash": "dddca1a38e3224870fcfae927c701c400fd885c19123ea840ee06a76a140f407", "question_id": "q00_single_fact", "split": "test", "cluster_id": "syn_03003_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 2, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1866, "query_tokens": 586, "message_tokens": 504, "message_utf8_bytes": 1121, "cache_tensor_bytes": null, "serialized_payload_bytes": 1121, "side_information_bytes": 0, "fetched_bytes": 763, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 44882.173666963354, "hops": {"1": 23068.98970901966, "2": 21813.183957943693}, "receiver_prefill_ms": 361.8221250362694, "ttft_ms": 49.500082968734205, "decode_ms": 589.2352920491248}, "total_wall_ms": 45883.46200005617, "ttft_ms": 49.500082968734205, "input_usage": {"lookups": 4}, "output_usage": {"output_tokens": 10}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "GR-79S-5839", "evidence_artifact": {"ledger": {"facts_total": 58, "facts_retained": 32, "retention_rate": 0.5517241379310345, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.7142857142857143, "amount": 0.6666666666666666, "status": 0.6666666666666666, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1121}, "ledger_by_hop": {"1": {"facts_total": 58, "facts_retained": 32, "retention_rate": 0.5517241379310345, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.7142857142857143, "amount": 0.6666666666666666, "status": 0.6666666666666666, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1200}, "2": {"facts_total": 58, "facts_retained": 32, "retention_rate": 0.5517241379310345, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.7142857142857143, "amount": 0.6666666666666666, "status": 0.6666666666666666, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1121}}}, "reference_key_id": "keys/synthetic/syn_03003_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "74294db8131d36bd09f61954da68ad856f3e18d9de11e15ac7d3770d376e3fa1", "output_hash": "720cc40cd69e85e9728f6bea00b17a95ca07eee643e143a131840763958303cd"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "a6ab1d584e419651ce44d87d", "attempt_id": "66c0b40d18a0", "run_group_id": "E4_core:A4|repairable|qwen3b:h2", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T06:44:29.643342Z", "task_family": "synthetic", "document_id": "syn_03003_2000", "document_hash": "dddca1a38e3224870fcfae927c701c400fd885c19123ea840ee06a76a140f407", "question_id": "q01_single_fact", "split": "test", "cluster_id": "syn_03003_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 2, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1866, "query_tokens": 590, "message_tokens": 504, "message_utf8_bytes": 1121, "cache_tensor_bytes": null, "serialized_payload_bytes": 1121, "side_information_bytes": 0, "fetched_bytes": 763, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 44882.173666963354, "hops": {"1": 23068.98970901966, "2": 21813.183957943693}, "receiver_prefill_ms": 410.3404999477789, "ttft_ms": 22.818166995421052, "decode_ms": 392.49966701027006}, "total_wall_ms": 45708.69670796674, "ttft_ms": 22.818166995421052, "input_usage": {"lookups": 4}, "output_usage": {"output_tokens": 11}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "Lot 344-B, Témiscaming", "evidence_artifact": {"ledger": {"facts_total": 58, "facts_retained": 32, "retention_rate": 0.5517241379310345, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.7142857142857143, "amount": 0.6666666666666666, "status": 0.6666666666666666, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1121}, "ledger_by_hop": {"1": {"facts_total": 58, "facts_retained": 32, "retention_rate": 0.5517241379310345, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.7142857142857143, "amount": 0.6666666666666666, "status": 0.6666666666666666, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1200}, "2": {"facts_total": 58, "facts_retained": 32, "retention_rate": 0.5517241379310345, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.7142857142857143, "amount": 0.6666666666666666, "status": 0.6666666666666666, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1121}}}, "reference_key_id": "keys/synthetic/syn_03003_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 0.8571428571428571}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "74294db8131d36bd09f61954da68ad856f3e18d9de11e15ac7d3770d376e3fa1", "output_hash": "0197d491a5a04a142dc8e372360735a1d41757e7a5aea7e81f23b0e1ef6ee010"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "b43d3e1cf466f353430c7087", "attempt_id": "aaa468a46016", "run_group_id": "E4_core:A4|repairable|qwen3b:h2", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T06:44:30.488237Z", "task_family": "synthetic", "document_id": "syn_03003_2000", "document_hash": "dddca1a38e3224870fcfae927c701c400fd885c19123ea840ee06a76a140f407", "question_id": "q02_exact_copy", "split": "test", "cluster_id": "syn_03003_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 2, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1866, "query_tokens": 590, "message_tokens": 504, "message_utf8_bytes": 1121, "cache_tensor_bytes": null, "serialized_payload_bytes": 1121, "side_information_bytes": 0, "fetched_bytes": 763, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 44882.173666963354, "hops": {"1": 23068.98970901966, "2": 21813.183957943693}, "receiver_prefill_ms": 427.16645798645914, "ttft_ms": 105.04583409056067, "decode_ms": 39.32279208675027}, "total_wall_ms": 45454.47808399331, "ttft_ms": 105.04583409056067, "input_usage": {"lookups": 4}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 58, "facts_retained": 32, "retention_rate": 0.5517241379310345, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.7142857142857143, "amount": 0.6666666666666666, "status": 0.6666666666666666, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1121}, "ledger_by_hop": {"1": {"facts_total": 58, "facts_retained": 32, "retention_rate": 0.5517241379310345, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.7142857142857143, "amount": 0.6666666666666666, "status": 0.6666666666666666, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1200}, "2": {"facts_total": 58, "facts_retained": 32, "retention_rate": 0.5517241379310345, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.7142857142857143, "amount": 0.6666666666666666, "status": 0.6666666666666666, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1121}}}, "reference_key_id": "keys/synthetic/syn_03003_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "74294db8131d36bd09f61954da68ad856f3e18d9de11e15ac7d3770d376e3fa1", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "15993536f48485069fc72487", "attempt_id": "52dd2c2104ba", "run_group_id": "E4_core:A4|repairable|qwen3b:h2", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T06:44:31.078180Z", "task_family": "synthetic", "document_id": "syn_03003_2000", "document_hash": "dddca1a38e3224870fcfae927c701c400fd885c19123ea840ee06a76a140f407", "question_id": "q03_exact_copy", "split": "test", "cluster_id": "syn_03003_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 2, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1866, "query_tokens": 585, "message_tokens": 504, "message_utf8_bytes": 1121, "cache_tensor_bytes": null, "serialized_payload_bytes": 1121, "side_information_bytes": 0, "fetched_bytes": 763, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 44882.173666963354, "hops": {"1": 23068.98970901966, "2": 21813.183957943693}, "receiver_prefill_ms": 410.3540410287678, "ttft_ms": 34.29195797070861, "decode_ms": 196.97837508283556}, "total_wall_ms": 45524.411041871645, "ttft_ms": 34.29195797070861, "input_usage": {"lookups": 4}, "output_usage": {"output_tokens": 6}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "Wilhelmina Achterberg", "evidence_artifact": {"ledger": {"facts_total": 58, "facts_retained": 32, "retention_rate": 0.5517241379310345, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.7142857142857143, "amount": 0.6666666666666666, "status": 0.6666666666666666, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1121}, "ledger_by_hop": {"1": {"facts_total": 58, "facts_retained": 32, "retention_rate": 0.5517241379310345, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.7142857142857143, "amount": 0.6666666666666666, "status": 0.6666666666666666, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1200}, "2": {"facts_total": 58, "facts_retained": 32, "retention_rate": 0.5517241379310345, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.7142857142857143, "amount": 0.6666666666666666, "status": 0.6666666666666666, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1121}}}, "reference_key_id": "keys/synthetic/syn_03003_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "74294db8131d36bd09f61954da68ad856f3e18d9de11e15ac7d3770d376e3fa1", "output_hash": "2a62606853c315bba8769371d195e3ab3d6a8cd7ef959205575c5119740850f8"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "adec6359cf4d11e1b836e3aa", "attempt_id": "63c346773087", "run_group_id": "E4_core:A4|repairable|qwen3b:h2", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T06:44:31.737393Z", "task_family": "synthetic", "document_id": "syn_03003_2000", "document_hash": "dddca1a38e3224870fcfae927c701c400fd885c19123ea840ee06a76a140f407", "question_id": "q04_multi_hop", "split": "test", "cluster_id": "syn_03003_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 2, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1866, "query_tokens": 594, "message_tokens": 504, "message_utf8_bytes": 1121, "cache_tensor_bytes": null, "serialized_payload_bytes": 1121, "side_information_bytes": 0, "fetched_bytes": 763, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 44882.173666963354, "hops": {"1": 23068.98970901966, "2": 21813.183957943693}, "receiver_prefill_ms": 361.64429201744497, "ttft_ms": 42.3556660534814, "decode_ms": 57.52745806239545}, "total_wall_ms": 45344.718042062595, "ttft_ms": 42.3556660534814, "input_usage": {"lookups": 4}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 58, "facts_retained": 32, "retention_rate": 0.5517241379310345, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.7142857142857143, "amount": 0.6666666666666666, "status": 0.6666666666666666, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1121}, "ledger_by_hop": {"1": {"facts_total": 58, "facts_retained": 32, "retention_rate": 0.5517241379310345, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.7142857142857143, "amount": 0.6666666666666666, "status": 0.6666666666666666, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1200}, "2": {"facts_total": 58, "facts_retained": 32, "retention_rate": 0.5517241379310345, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.7142857142857143, "amount": 0.6666666666666666, "status": 0.6666666666666666, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1121}}}, "reference_key_id": "keys/synthetic/syn_03003_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "74294db8131d36bd09f61954da68ad856f3e18d9de11e15ac7d3770d376e3fa1", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "5f66e675db626de38f51e8aa", "attempt_id": "8e8beef3e11a", "run_group_id": "E4_core:A4|repairable|qwen3b:h2", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T06:44:32.217618Z", "task_family": "synthetic", "document_id": "syn_03003_2000", "document_hash": "dddca1a38e3224870fcfae927c701c400fd885c19123ea840ee06a76a140f407", "question_id": "q05_multi_hop", "split": "test", "cluster_id": "syn_03003_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 2, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1866, "query_tokens": 596, "message_tokens": 504, "message_utf8_bytes": 1121, "cache_tensor_bytes": null, "serialized_payload_bytes": 1121, "side_information_bytes": 0, "fetched_bytes": 763, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 44882.173666963354, "hops": {"1": 23068.98970901966, "2": 21813.183957943693}, "receiver_prefill_ms": 416.76224989350885, "ttft_ms": 115.89562497101724, "decode_ms": 339.6741249598563}, "total_wall_ms": 45755.1372499438, "ttft_ms": 115.89562497101724, "input_usage": {"lookups": 4}, "output_usage": {"output_tokens": 10}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "NK-95Z-2165", "evidence_artifact": {"ledger": {"facts_total": 58, "facts_retained": 32, "retention_rate": 0.5517241379310345, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.7142857142857143, "amount": 0.6666666666666666, "status": 0.6666666666666666, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1121}, "ledger_by_hop": {"1": {"facts_total": 58, "facts_retained": 32, "retention_rate": 0.5517241379310345, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.7142857142857143, "amount": 0.6666666666666666, "status": 0.6666666666666666, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1200}, "2": {"facts_total": 58, "facts_retained": 32, "retention_rate": 0.5517241379310345, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.7142857142857143, "amount": 0.6666666666666666, "status": 0.6666666666666666, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1121}}}, "reference_key_id": "keys/synthetic/syn_03003_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "74294db8131d36bd09f61954da68ad856f3e18d9de11e15ac7d3770d376e3fa1", "output_hash": "bd8f809b02304c6b9dc7c1e116ecccc2328c976a5c071e89a94d661b48e73aa7"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "68452108ae9b86b3cf718231", "attempt_id": "63e6ec324432", "run_group_id": "E4_core:A4|repairable|qwen3b:h2", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T06:44:33.107859Z", "task_family": "synthetic", "document_id": "syn_03003_2000", "document_hash": "dddca1a38e3224870fcfae927c701c400fd885c19123ea840ee06a76a140f407", "question_id": "q06_negation", "split": "test", "cluster_id": "syn_03003_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 2, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1866, "query_tokens": 588, "message_tokens": 504, "message_utf8_bytes": 1121, "cache_tensor_bytes": null, "serialized_payload_bytes": 1121, "side_information_bytes": 0, "fetched_bytes": 763, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 44882.173666963354, "hops": {"1": 23068.98970901966, "2": 21813.183957943693}, "receiver_prefill_ms": 407.6013750163838, "ttft_ms": 24.918166105635464, "decode_ms": 37.48795797582716}, "total_wall_ms": 45352.74245799519, "ttft_ms": 24.918166105635464, "input_usage": {"lookups": 4}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "yes", "evidence_artifact": {"ledger": {"facts_total": 58, "facts_retained": 32, "retention_rate": 0.5517241379310345, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.7142857142857143, "amount": 0.6666666666666666, "status": 0.6666666666666666, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1121}, "ledger_by_hop": {"1": {"facts_total": 58, "facts_retained": 32, "retention_rate": 0.5517241379310345, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.7142857142857143, "amount": 0.6666666666666666, "status": 0.6666666666666666, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1200}, "2": {"facts_total": 58, "facts_retained": 32, "retention_rate": 0.5517241379310345, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.7142857142857143, "amount": 0.6666666666666666, "status": 0.6666666666666666, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1121}}}, "reference_key_id": "keys/synthetic/syn_03003_2000.json", "task_scores": {"correct": 0.0, "abstained": 0.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "74294db8131d36bd09f61954da68ad856f3e18d9de11e15ac7d3770d376e3fa1", "output_hash": "8a798890fe93817163b10b5f7bd2ca4d25d84c52739a645a889c173eee7d9d3d"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "f38211289ddcc6ed74b51580", "attempt_id": "df5881927bfc", "run_group_id": "E4_core:A4|repairable|qwen3b:h2", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T06:44:33.596305Z", "task_family": "synthetic", "document_id": "syn_03003_2000", "document_hash": "dddca1a38e3224870fcfae927c701c400fd885c19123ea840ee06a76a140f407", "question_id": "q07_negation", "split": "test", "cluster_id": "syn_03003_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 2, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1866, "query_tokens": 585, "message_tokens": 504, "message_utf8_bytes": 1121, "cache_tensor_bytes": null, "serialized_payload_bytes": 1121, "side_information_bytes": 0, "fetched_bytes": 763, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 44882.173666963354, "hops": {"1": 23068.98970901966, "2": 21813.183957943693}, "receiver_prefill_ms": 343.5030000982806, "ttft_ms": 53.88487491291016, "decode_ms": 41.5860420325771}, "total_wall_ms": 45321.94016687572, "ttft_ms": 53.88487491291016, "input_usage": {"lookups": 4}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "no", "evidence_artifact": {"ledger": {"facts_total": 58, "facts_retained": 32, "retention_rate": 0.5517241379310345, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.7142857142857143, "amount": 0.6666666666666666, "status": 0.6666666666666666, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1121}, "ledger_by_hop": {"1": {"facts_total": 58, "facts_retained": 32, "retention_rate": 0.5517241379310345, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.7142857142857143, "amount": 0.6666666666666666, "status": 0.6666666666666666, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1200}, "2": {"facts_total": 58, "facts_retained": 32, "retention_rate": 0.5517241379310345, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.7142857142857143, "amount": 0.6666666666666666, "status": 0.6666666666666666, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1121}}}, "reference_key_id": "keys/synthetic/syn_03003_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "74294db8131d36bd09f61954da68ad856f3e18d9de11e15ac7d3770d376e3fa1", "output_hash": "9390298f3fb0c5b160498935d79cb139aef28e1c47358b4bbba61862b9c26e59"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "01c870ed8bb70ea3129e1fa7", "attempt_id": "343182c2ec6f", "run_group_id": "E4_core:A4|repairable|qwen3b:h2", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T06:44:34.054010Z", "task_family": "synthetic", "document_id": "syn_03003_2000", "document_hash": "dddca1a38e3224870fcfae927c701c400fd885c19123ea840ee06a76a140f407", "question_id": "q08_revision", "split": "test", "cluster_id": "syn_03003_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 2, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1866, "query_tokens": 603, "message_tokens": 504, "message_utf8_bytes": 1121, "cache_tensor_bytes": null, "serialized_payload_bytes": 1121, "side_information_bytes": 0, "fetched_bytes": 763, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 44882.173666963354, "hops": {"1": 23068.98970901966, "2": 21813.183957943693}, "receiver_prefill_ms": 528.7046659504995, "ttft_ms": 32.11716702207923, "decode_ms": 235.03866605460644}, "total_wall_ms": 45678.74970892444, "ttft_ms": 32.11716702207923, "input_usage": {"lookups": 4}, "output_usage": {"output_tokens": 7}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "911,925", "evidence_artifact": {"ledger": {"facts_total": 58, "facts_retained": 32, "retention_rate": 0.5517241379310345, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.7142857142857143, "amount": 0.6666666666666666, "status": 0.6666666666666666, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1121}, "ledger_by_hop": {"1": {"facts_total": 58, "facts_retained": 32, "retention_rate": 0.5517241379310345, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.7142857142857143, "amount": 0.6666666666666666, "status": 0.6666666666666666, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1200}, "2": {"facts_total": 58, "facts_retained": 32, "retention_rate": 0.5517241379310345, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.7142857142857143, "amount": 0.6666666666666666, "status": 0.6666666666666666, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1121}}}, "reference_key_id": "keys/synthetic/syn_03003_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "74294db8131d36bd09f61954da68ad856f3e18d9de11e15ac7d3770d376e3fa1", "output_hash": "86be78720fc4b79f93e603e7d9e4861fcd25102612f73209fffc17d787056007"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "167a9e6f7deb94f390c58987", "attempt_id": "b0057d345240", "run_group_id": "E4_core:A4|repairable|qwen3b:h2", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T06:44:34.867836Z", "task_family": "synthetic", "document_id": "syn_03003_2000", "document_hash": "dddca1a38e3224870fcfae927c701c400fd885c19123ea840ee06a76a140f407", "question_id": "q09_revision", "split": "test", "cluster_id": "syn_03003_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 2, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1866, "query_tokens": 596, "message_tokens": 504, "message_utf8_bytes": 1121, "cache_tensor_bytes": null, "serialized_payload_bytes": 1121, "side_information_bytes": 0, "fetched_bytes": 763, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 44882.173666963354, "hops": {"1": 23068.98970901966, "2": 21813.183957943693}, "receiver_prefill_ms": 397.9332079179585, "ttft_ms": 35.022250027395785, "decode_ms": 201.31545898038894}, "total_wall_ms": 45517.07883388735, "ttft_ms": 35.022250027395785, "input_usage": {"lookups": 4}, "output_usage": {"output_tokens": 6}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "911925", "evidence_artifact": {"ledger": {"facts_total": 58, "facts_retained": 32, "retention_rate": 0.5517241379310345, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.7142857142857143, "amount": 0.6666666666666666, "status": 0.6666666666666666, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1121}, "ledger_by_hop": {"1": {"facts_total": 58, "facts_retained": 32, "retention_rate": 0.5517241379310345, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.7142857142857143, "amount": 0.6666666666666666, "status": 0.6666666666666666, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1200}, "2": {"facts_total": 58, "facts_retained": 32, "retention_rate": 0.5517241379310345, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.7142857142857143, "amount": 0.6666666666666666, "status": 0.6666666666666666, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1121}}}, "reference_key_id": "keys/synthetic/syn_03003_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "74294db8131d36bd09f61954da68ad856f3e18d9de11e15ac7d3770d376e3fa1", "output_hash": "679a540dc4739f0ace623448b8fd925fb234f554838d7e87da223228ee81b112"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "79783a4835f55a2f685adb4f", "attempt_id": "532b03f2e067", "run_group_id": "E4_core:A4|repairable|qwen3b:h2", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T06:44:35.519742Z", "task_family": "synthetic", "document_id": "syn_03003_2000", "document_hash": "dddca1a38e3224870fcfae927c701c400fd885c19123ea840ee06a76a140f407", "question_id": "q10_temporal", "split": "test", "cluster_id": "syn_03003_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 2, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1866, "query_tokens": 603, "message_tokens": 504, "message_utf8_bytes": 1121, "cache_tensor_bytes": null, "serialized_payload_bytes": 1121, "side_information_bytes": 0, "fetched_bytes": 763, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 44882.173666963354, "hops": {"1": 23068.98970901966, "2": 21813.183957943693}, "receiver_prefill_ms": 567.9363330127671, "ttft_ms": 43.14908303786069, "decode_ms": 219.79158406611532}, "total_wall_ms": 45713.7286249781, "ttft_ms": 43.14908303786069, "input_usage": {"lookups": 4}, "output_usage": {"output_tokens": 6}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "Gamtiszen-2", "evidence_artifact": {"ledger": {"facts_total": 58, "facts_retained": 32, "retention_rate": 0.5517241379310345, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.7142857142857143, "amount": 0.6666666666666666, "status": 0.6666666666666666, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1121}, "ledger_by_hop": {"1": {"facts_total": 58, "facts_retained": 32, "retention_rate": 0.5517241379310345, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.7142857142857143, "amount": 0.6666666666666666, "status": 0.6666666666666666, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1200}, "2": {"facts_total": 58, "facts_retained": 32, "retention_rate": 0.5517241379310345, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.7142857142857143, "amount": 0.6666666666666666, "status": 0.6666666666666666, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1121}}}, "reference_key_id": "keys/synthetic/syn_03003_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "74294db8131d36bd09f61954da68ad856f3e18d9de11e15ac7d3770d376e3fa1", "output_hash": "32e3af428886434823d82102e1c482765eafdf6077fe8d3fba5752ec5e29826c"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "52ec2bd630b33a38786cf09a", "attempt_id": "826dbd24179c", "run_group_id": "E4_core:A4|repairable|qwen3b:h2", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T06:44:36.369599Z", "task_family": "synthetic", "document_id": "syn_03003_2000", "document_hash": "dddca1a38e3224870fcfae927c701c400fd885c19123ea840ee06a76a140f407", "question_id": "q11_aggregation", "split": "test", "cluster_id": "syn_03003_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 2, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1866, "query_tokens": 597, "message_tokens": 504, "message_utf8_bytes": 1121, "cache_tensor_bytes": null, "serialized_payload_bytes": 1121, "side_information_bytes": 0, "fetched_bytes": 763, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 44882.173666963354, "hops": {"1": 23068.98970901966, "2": 21813.183957943693}, "receiver_prefill_ms": 383.39104200713336, "ttft_ms": 38.54225005488843, "decode_ms": 424.00095891207457}, "total_wall_ms": 45728.79550000653, "ttft_ms": 38.54225005488843, "input_usage": {"lookups": 4}, "output_usage": {"output_tokens": 10}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "33,449,445", "evidence_artifact": {"ledger": {"facts_total": 58, "facts_retained": 32, "retention_rate": 0.5517241379310345, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.7142857142857143, "amount": 0.6666666666666666, "status": 0.6666666666666666, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1121}, "ledger_by_hop": {"1": {"facts_total": 58, "facts_retained": 32, "retention_rate": 0.5517241379310345, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.7142857142857143, "amount": 0.6666666666666666, "status": 0.6666666666666666, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1200}, "2": {"facts_total": 58, "facts_retained": 32, "retention_rate": 0.5517241379310345, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.7142857142857143, "amount": 0.6666666666666666, "status": 0.6666666666666666, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1121}}}, "reference_key_id": "keys/synthetic/syn_03003_2000.json", "task_scores": {"correct": 0.0, "abstained": 0.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "74294db8131d36bd09f61954da68ad856f3e18d9de11e15ac7d3770d376e3fa1", "output_hash": "08d172c2c66fe2edb2ffe283f62b9d1e37bd5ca2c67a79d275c24fd6b43dabcc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "1d9cb89aca98bbd249c27e51", "attempt_id": "e5bfe0f37436", "run_group_id": "E4_core:A4|repairable|qwen3b:h2", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T06:44:37.234369Z", "task_family": "synthetic", "document_id": "syn_03003_2000", "document_hash": "dddca1a38e3224870fcfae927c701c400fd885c19123ea840ee06a76a140f407", "question_id": "q12_unit_conversion", "split": "test", "cluster_id": "syn_03003_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 2, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1866, "query_tokens": 594, "message_tokens": 504, "message_utf8_bytes": 1121, "cache_tensor_bytes": null, "serialized_payload_bytes": 1121, "side_information_bytes": 0, "fetched_bytes": 763, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 44882.173666963354, "hops": {"1": 23068.98970901966, "2": 21813.183957943693}, "receiver_prefill_ms": 530.3034579847008, "ttft_ms": 34.07766600139439, "decode_ms": 179.4127500616014}, "total_wall_ms": 45626.7143759178, "ttft_ms": 34.07766600139439, "input_usage": {"lookups": 4}, "output_usage": {"output_tokens": 5}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "17430", "evidence_artifact": {"ledger": {"facts_total": 58, "facts_retained": 32, "retention_rate": 0.5517241379310345, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.7142857142857143, "amount": 0.6666666666666666, "status": 0.6666666666666666, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1121}, "ledger_by_hop": {"1": {"facts_total": 58, "facts_retained": 32, "retention_rate": 0.5517241379310345, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.7142857142857143, "amount": 0.6666666666666666, "status": 0.6666666666666666, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1200}, "2": {"facts_total": 58, "facts_retained": 32, "retention_rate": 0.5517241379310345, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.7142857142857143, "amount": 0.6666666666666666, "status": 0.6666666666666666, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1121}}}, "reference_key_id": "keys/synthetic/syn_03003_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "74294db8131d36bd09f61954da68ad856f3e18d9de11e15ac7d3770d376e3fa1", "output_hash": "2575eaf3b748efc209b676518766751eb038c88f74578951345818bfe7f7194b"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "0389889abc4576a59dafad37", "attempt_id": "4e756baeb84e", "run_group_id": "E4_core:A4|repairable|qwen3b:h2", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T06:44:37.997623Z", "task_family": "synthetic", "document_id": "syn_03003_2000", "document_hash": "dddca1a38e3224870fcfae927c701c400fd885c19123ea840ee06a76a140f407", "question_id": "q13_unanswerable", "split": "test", "cluster_id": "syn_03003_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 2, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1866, "query_tokens": 598, "message_tokens": 504, "message_utf8_bytes": 1121, "cache_tensor_bytes": null, "serialized_payload_bytes": 1121, "side_information_bytes": 0, "fetched_bytes": 763, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 44882.173666963354, "hops": {"1": 23068.98970901966, "2": 21813.183957943693}, "receiver_prefill_ms": 407.6272089732811, "ttft_ms": 31.7239579744637, "decode_ms": 30.411790939979255}, "total_wall_ms": 45352.6026669424, "ttft_ms": 31.7239579744637, "input_usage": {"lookups": 4}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 58, "facts_retained": 32, "retention_rate": 0.5517241379310345, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.7142857142857143, "amount": 0.6666666666666666, "status": 0.6666666666666666, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1121}, "ledger_by_hop": {"1": {"facts_total": 58, "facts_retained": 32, "retention_rate": 0.5517241379310345, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.7142857142857143, "amount": 0.6666666666666666, "status": 0.6666666666666666, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1200}, "2": {"facts_total": 58, "facts_retained": 32, "retention_rate": 0.5517241379310345, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.7142857142857143, "amount": 0.6666666666666666, "status": 0.6666666666666666, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1121}}}, "reference_key_id": "keys/synthetic/syn_03003_2000.json", "task_scores": {"correct": 1.0, "abstained": 1.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "74294db8131d36bd09f61954da68ad856f3e18d9de11e15ac7d3770d376e3fa1", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "95e15c73c47a06dc5bcf7a0f", "attempt_id": "5a8574c6f64a", "run_group_id": "E4_core:A4|repairable|qwen3b:h2", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T06:44:38.487139Z", "task_family": "synthetic", "document_id": "syn_03003_2000", "document_hash": "dddca1a38e3224870fcfae927c701c400fd885c19123ea840ee06a76a140f407", "question_id": "q14_unanswerable", "split": "test", "cluster_id": "syn_03003_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 2, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1866, "query_tokens": 599, "message_tokens": 504, "message_utf8_bytes": 1121, "cache_tensor_bytes": null, "serialized_payload_bytes": 1121, "side_information_bytes": 0, "fetched_bytes": 763, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 44882.173666963354, "hops": {"1": 23068.98970901966, "2": 21813.183957943693}, "receiver_prefill_ms": 350.12545902282, "ttft_ms": 36.48825001437217, "decode_ms": 63.622584100812674}, "total_wall_ms": 45333.22995796334, "ttft_ms": 36.48825001437217, "input_usage": {"lookups": 4}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 58, "facts_retained": 32, "retention_rate": 0.5517241379310345, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.7142857142857143, "amount": 0.6666666666666666, "status": 0.6666666666666666, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1121}, "ledger_by_hop": {"1": {"facts_total": 58, "facts_retained": 32, "retention_rate": 0.5517241379310345, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.7142857142857143, "amount": 0.6666666666666666, "status": 0.6666666666666666, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1200}, "2": {"facts_total": 58, "facts_retained": 32, "retention_rate": 0.5517241379310345, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.7142857142857143, "amount": 0.6666666666666666, "status": 0.6666666666666666, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1121}}}, "reference_key_id": "keys/synthetic/syn_03003_2000.json", "task_scores": {"correct": 1.0, "abstained": 1.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "74294db8131d36bd09f61954da68ad856f3e18d9de11e15ac7d3770d376e3fa1", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "dabd32d43f8dfe7d11abbbf4", "attempt_id": "6a58ed9f4d6c", "run_group_id": "E4_core:A4|repairable|qwen3b:h4", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T06:45:22.023368Z", "task_family": "synthetic", "document_id": "syn_03003_2000", "document_hash": "dddca1a38e3224870fcfae927c701c400fd885c19123ea840ee06a76a140f407", "question_id": "q00_single_fact", "split": "test", "cluster_id": "syn_03003_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 4, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1866, "query_tokens": 580, "message_tokens": 498, "message_utf8_bytes": 1106, "cache_tensor_bytes": null, "serialized_payload_bytes": 1106, "side_information_bytes": 0, "fetched_bytes": 2289, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 87954.53012594953, "hops": {"1": 23068.98970901966, "2": 21813.183957943693, "3": 21540.921541978605, "4": 21531.434917007573}, "receiver_prefill_ms": 401.6515000257641, "ttft_ms": 25.9505829308182, "decode_ms": 429.4277090812102}, "total_wall_ms": 88812.27320886683, "ttft_ms": 25.9505829308182, "input_usage": {"lookups": 12}, "output_usage": {"output_tokens": 10}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "GR-79S-5839", "evidence_artifact": {"ledger": {"facts_total": 58, "facts_retained": 32, "retention_rate": 0.5517241379310345, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.7142857142857143, "amount": 0.6666666666666666, "status": 0.6666666666666666, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1106}, "ledger_by_hop": {"1": {"facts_total": 58, "facts_retained": 32, "retention_rate": 0.5517241379310345, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.7142857142857143, "amount": 0.6666666666666666, "status": 0.6666666666666666, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1200}, "2": {"facts_total": 58, "facts_retained": 32, "retention_rate": 0.5517241379310345, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.7142857142857143, "amount": 0.6666666666666666, "status": 0.6666666666666666, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1121}, "3": {"facts_total": 58, "facts_retained": 32, "retention_rate": 0.5517241379310345, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.7142857142857143, "amount": 0.6666666666666666, "status": 0.6666666666666666, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1106}, "4": {"facts_total": 58, "facts_retained": 32, "retention_rate": 0.5517241379310345, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.7142857142857143, "amount": 0.6666666666666666, "status": 0.6666666666666666, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1106}}}, "reference_key_id": "keys/synthetic/syn_03003_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "fd539b5069a942d3034d38027c24271a19ab0d754e9dee3800caa6490c9681d4", "output_hash": "720cc40cd69e85e9728f6bea00b17a95ca07eee643e143a131840763958303cd"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "7c32dd619d39037dafd5af6a", "attempt_id": "7e9d98ff1200", "run_group_id": "E4_core:A4|repairable|qwen3b:h4", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T06:45:22.897577Z", "task_family": "synthetic", "document_id": "syn_03003_2000", "document_hash": "dddca1a38e3224870fcfae927c701c400fd885c19123ea840ee06a76a140f407", "question_id": "q01_single_fact", "split": "test", "cluster_id": "syn_03003_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 4, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1866, "query_tokens": 584, "message_tokens": 498, "message_utf8_bytes": 1106, "cache_tensor_bytes": null, "serialized_payload_bytes": 1106, "side_information_bytes": 0, "fetched_bytes": 2289, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 87954.53012594953, "hops": {"1": 23068.98970901966, "2": 21813.183957943693, "3": 21540.921541978605, "4": 21531.434917007573}, "receiver_prefill_ms": 508.4931250894442, "ttft_ms": 32.5467090588063, "decode_ms": 382.5243329629302}, "total_wall_ms": 88878.92216793261, "ttft_ms": 32.5467090588063, "input_usage": {"lookups": 12}, "output_usage": {"output_tokens": 11}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "Lot 344-B, Témiscaming", "evidence_artifact": {"ledger": {"facts_total": 58, "facts_retained": 32, "retention_rate": 0.5517241379310345, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.7142857142857143, "amount": 0.6666666666666666, "status": 0.6666666666666666, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1106}, "ledger_by_hop": {"1": {"facts_total": 58, "facts_retained": 32, "retention_rate": 0.5517241379310345, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.7142857142857143, "amount": 0.6666666666666666, "status": 0.6666666666666666, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1200}, "2": {"facts_total": 58, "facts_retained": 32, "retention_rate": 0.5517241379310345, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.7142857142857143, "amount": 0.6666666666666666, "status": 0.6666666666666666, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1121}, "3": {"facts_total": 58, "facts_retained": 32, "retention_rate": 0.5517241379310345, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.7142857142857143, "amount": 0.6666666666666666, "status": 0.6666666666666666, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1106}, "4": {"facts_total": 58, "facts_retained": 32, "retention_rate": 0.5517241379310345, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.7142857142857143, "amount": 0.6666666666666666, "status": 0.6666666666666666, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1106}}}, "reference_key_id": "keys/synthetic/syn_03003_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 0.8571428571428571}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "fd539b5069a942d3034d38027c24271a19ab0d754e9dee3800caa6490c9681d4", "output_hash": "0197d491a5a04a142dc8e372360735a1d41757e7a5aea7e81f23b0e1ef6ee010"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "b63743082511fd395e58bc54", "attempt_id": "7f2d128ace82", "run_group_id": "E4_core:A4|repairable|qwen3b:h4", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T06:45:23.839082Z", "task_family": "synthetic", "document_id": "syn_03003_2000", "document_hash": "dddca1a38e3224870fcfae927c701c400fd885c19123ea840ee06a76a140f407", "question_id": "q02_exact_copy", "split": "test", "cluster_id": "syn_03003_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 4, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1866, "query_tokens": 584, "message_tokens": 498, "message_utf8_bytes": 1106, "cache_tensor_bytes": null, "serialized_payload_bytes": 1106, "side_information_bytes": 0, "fetched_bytes": 2289, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 87954.53012594953, "hops": {"1": 23068.98970901966, "2": 21813.183957943693, "3": 21540.921541978605, "4": 21531.434917007573}, "receiver_prefill_ms": 359.70779100898653, "ttft_ms": 36.2427500076592, "decode_ms": 61.63370795547962}, "total_wall_ms": 88413.00929302815, "ttft_ms": 36.2427500076592, "input_usage": {"lookups": 12}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 58, "facts_retained": 32, "retention_rate": 0.5517241379310345, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.7142857142857143, "amount": 0.6666666666666666, "status": 0.6666666666666666, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1106}, "ledger_by_hop": {"1": {"facts_total": 58, "facts_retained": 32, "retention_rate": 0.5517241379310345, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.7142857142857143, "amount": 0.6666666666666666, "status": 0.6666666666666666, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1200}, "2": {"facts_total": 58, "facts_retained": 32, "retention_rate": 0.5517241379310345, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.7142857142857143, "amount": 0.6666666666666666, "status": 0.6666666666666666, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1121}, "3": {"facts_total": 58, "facts_retained": 32, "retention_rate": 0.5517241379310345, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.7142857142857143, "amount": 0.6666666666666666, "status": 0.6666666666666666, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1106}, "4": {"facts_total": 58, "facts_retained": 32, "retention_rate": 0.5517241379310345, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.7142857142857143, "amount": 0.6666666666666666, "status": 0.6666666666666666, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1106}}}, "reference_key_id": "keys/synthetic/syn_03003_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "fd539b5069a942d3034d38027c24271a19ab0d754e9dee3800caa6490c9681d4", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "b84861d349e03446807c0cbf", "attempt_id": "a63ab5d82bcf", "run_group_id": "E4_core:A4|repairable|qwen3b:h4", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T06:45:24.315393Z", "task_family": "synthetic", "document_id": "syn_03003_2000", "document_hash": "dddca1a38e3224870fcfae927c701c400fd885c19123ea840ee06a76a140f407", "question_id": "q03_exact_copy", "split": "test", "cluster_id": "syn_03003_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 4, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1866, "query_tokens": 579, "message_tokens": 498, "message_utf8_bytes": 1106, "cache_tensor_bytes": null, "serialized_payload_bytes": 1106, "side_information_bytes": 0, "fetched_bytes": 2289, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 87954.53012594953, "hops": {"1": 23068.98970901966, "2": 21813.183957943693, "3": 21540.921541978605, "4": 21531.434917007573}, "receiver_prefill_ms": 565.7207500189543, "ttft_ms": 48.51049999706447, "decode_ms": 235.5416250647977}, "total_wall_ms": 88805.06491695996, "ttft_ms": 48.51049999706447, "input_usage": {"lookups": 12}, "output_usage": {"output_tokens": 6}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "Wilhelmina Achterberg", "evidence_artifact": {"ledger": {"facts_total": 58, "facts_retained": 32, "retention_rate": 0.5517241379310345, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.7142857142857143, "amount": 0.6666666666666666, "status": 0.6666666666666666, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1106}, "ledger_by_hop": {"1": {"facts_total": 58, "facts_retained": 32, "retention_rate": 0.5517241379310345, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.7142857142857143, "amount": 0.6666666666666666, "status": 0.6666666666666666, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1200}, "2": {"facts_total": 58, "facts_retained": 32, "retention_rate": 0.5517241379310345, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.7142857142857143, "amount": 0.6666666666666666, "status": 0.6666666666666666, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1121}, "3": {"facts_total": 58, "facts_retained": 32, "retention_rate": 0.5517241379310345, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.7142857142857143, "amount": 0.6666666666666666, "status": 0.6666666666666666, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1106}, "4": {"facts_total": 58, "facts_retained": 32, "retention_rate": 0.5517241379310345, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.7142857142857143, "amount": 0.6666666666666666, "status": 0.6666666666666666, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1106}}}, "reference_key_id": "keys/synthetic/syn_03003_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "fd539b5069a942d3034d38027c24271a19ab0d754e9dee3800caa6490c9681d4", "output_hash": "2a62606853c315bba8769371d195e3ab3d6a8cd7ef959205575c5119740850f8"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "9780e9926c1baf58d3508ea4", "attempt_id": "a276ca063102", "run_group_id": "E4_core:A4|repairable|qwen3b:h4", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T06:45:25.181752Z", "task_family": "synthetic", "document_id": "syn_03003_2000", "document_hash": "dddca1a38e3224870fcfae927c701c400fd885c19123ea840ee06a76a140f407", "question_id": "q04_multi_hop", "split": "test", "cluster_id": "syn_03003_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 4, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1866, "query_tokens": 588, "message_tokens": 498, "message_utf8_bytes": 1106, "cache_tensor_bytes": null, "serialized_payload_bytes": 1106, "side_information_bytes": 0, "fetched_bytes": 2289, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 87954.53012594953, "hops": {"1": 23068.98970901966, "2": 21813.183957943693, "3": 21540.921541978605, "4": 21531.434917007573}, "receiver_prefill_ms": 390.279124956578, "ttft_ms": 40.17670906614512, "decode_ms": 39.09329103771597}, "total_wall_ms": 88424.69429294579, "ttft_ms": 40.17670906614512, "input_usage": {"lookups": 12}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 58, "facts_retained": 32, "retention_rate": 0.5517241379310345, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.7142857142857143, "amount": 0.6666666666666666, "status": 0.6666666666666666, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1106}, "ledger_by_hop": {"1": {"facts_total": 58, "facts_retained": 32, "retention_rate": 0.5517241379310345, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.7142857142857143, "amount": 0.6666666666666666, "status": 0.6666666666666666, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1200}, "2": {"facts_total": 58, "facts_retained": 32, "retention_rate": 0.5517241379310345, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.7142857142857143, "amount": 0.6666666666666666, "status": 0.6666666666666666, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1121}, "3": {"facts_total": 58, "facts_retained": 32, "retention_rate": 0.5517241379310345, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.7142857142857143, "amount": 0.6666666666666666, "status": 0.6666666666666666, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1106}, "4": {"facts_total": 58, "facts_retained": 32, "retention_rate": 0.5517241379310345, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.7142857142857143, "amount": 0.6666666666666666, "status": 0.6666666666666666, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1106}}}, "reference_key_id": "keys/synthetic/syn_03003_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "fd539b5069a942d3034d38027c24271a19ab0d754e9dee3800caa6490c9681d4", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "b25d0a5bee15d9d13d859d96", "attempt_id": "c2d7917fdf5e", "run_group_id": "E4_core:A4|repairable|qwen3b:h4", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T06:45:25.669434Z", "task_family": "synthetic", "document_id": "syn_03003_2000", "document_hash": "dddca1a38e3224870fcfae927c701c400fd885c19123ea840ee06a76a140f407", "question_id": "q05_multi_hop", "split": "test", "cluster_id": "syn_03003_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 4, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1866, "query_tokens": 590, "message_tokens": 498, "message_utf8_bytes": 1106, "cache_tensor_bytes": null, "serialized_payload_bytes": 1106, "side_information_bytes": 0, "fetched_bytes": 2289, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 87954.53012594953, "hops": {"1": 23068.98970901966, "2": 21813.183957943693, "3": 21540.921541978605, "4": 21531.434917007573}, "receiver_prefill_ms": 388.39124992955476, "ttft_ms": 51.24620802234858, "decode_ms": 541.9013750506565}, "total_wall_ms": 88936.9265009882, "ttft_ms": 51.24620802234858, "input_usage": {"lookups": 12}, "output_usage": {"output_tokens": 10}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "NK-95Z-2165", "evidence_artifact": {"ledger": {"facts_total": 58, "facts_retained": 32, "retention_rate": 0.5517241379310345, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.7142857142857143, "amount": 0.6666666666666666, "status": 0.6666666666666666, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1106}, "ledger_by_hop": {"1": {"facts_total": 58, "facts_retained": 32, "retention_rate": 0.5517241379310345, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.7142857142857143, "amount": 0.6666666666666666, "status": 0.6666666666666666, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1200}, "2": {"facts_total": 58, "facts_retained": 32, "retention_rate": 0.5517241379310345, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.7142857142857143, "amount": 0.6666666666666666, "status": 0.6666666666666666, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1121}, "3": {"facts_total": 58, "facts_retained": 32, "retention_rate": 0.5517241379310345, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.7142857142857143, "amount": 0.6666666666666666, "status": 0.6666666666666666, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1106}, "4": {"facts_total": 58, "facts_retained": 32, "retention_rate": 0.5517241379310345, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.7142857142857143, "amount": 0.6666666666666666, "status": 0.6666666666666666, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1106}}}, "reference_key_id": "keys/synthetic/syn_03003_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "fd539b5069a942d3034d38027c24271a19ab0d754e9dee3800caa6490c9681d4", "output_hash": "bd8f809b02304c6b9dc7c1e116ecccc2328c976a5c071e89a94d661b48e73aa7"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "c177b3022ca58496f6499e73", "attempt_id": "220ef82598b2", "run_group_id": "E4_core:A4|repairable|qwen3b:h4", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T06:45:26.667745Z", "task_family": "synthetic", "document_id": "syn_03003_2000", "document_hash": "dddca1a38e3224870fcfae927c701c400fd885c19123ea840ee06a76a140f407", "question_id": "q06_negation", "split": "test", "cluster_id": "syn_03003_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 4, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1866, "query_tokens": 582, "message_tokens": 498, "message_utf8_bytes": 1106, "cache_tensor_bytes": null, "serialized_payload_bytes": 1106, "side_information_bytes": 0, "fetched_bytes": 2289, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 87954.53012594953, "hops": {"1": 23068.98970901966, "2": 21813.183957943693, "3": 21540.921541978605, "4": 21531.434917007573}, "receiver_prefill_ms": 410.5999580351636, "ttft_ms": 20.264332997612655, "decode_ms": 37.79579093679786}, "total_wall_ms": 88423.82825096138, "ttft_ms": 20.264332997612655, "input_usage": {"lookups": 12}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "yes", "evidence_artifact": {"ledger": {"facts_total": 58, "facts_retained": 32, "retention_rate": 0.5517241379310345, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.7142857142857143, "amount": 0.6666666666666666, "status": 0.6666666666666666, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1106}, "ledger_by_hop": {"1": {"facts_total": 58, "facts_retained": 32, "retention_rate": 0.5517241379310345, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.7142857142857143, "amount": 0.6666666666666666, "status": 0.6666666666666666, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1200}, "2": {"facts_total": 58, "facts_retained": 32, "retention_rate": 0.5517241379310345, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.7142857142857143, "amount": 0.6666666666666666, "status": 0.6666666666666666, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1121}, "3": {"facts_total": 58, "facts_retained": 32, "retention_rate": 0.5517241379310345, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.7142857142857143, "amount": 0.6666666666666666, "status": 0.6666666666666666, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1106}, "4": {"facts_total": 58, "facts_retained": 32, "retention_rate": 0.5517241379310345, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.7142857142857143, "amount": 0.6666666666666666, "status": 0.6666666666666666, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1106}}}, "reference_key_id": "keys/synthetic/syn_03003_2000.json", "task_scores": {"correct": 0.0, "abstained": 0.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "fd539b5069a942d3034d38027c24271a19ab0d754e9dee3800caa6490c9681d4", "output_hash": "8a798890fe93817163b10b5f7bd2ca4d25d84c52739a645a889c173eee7d9d3d"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "cdb623032ece1b009bda974f", "attempt_id": "8518af459d58", "run_group_id": "E4_core:A4|repairable|qwen3b:h4", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T06:45:27.154103Z", "task_family": "synthetic", "document_id": "syn_03003_2000", "document_hash": "dddca1a38e3224870fcfae927c701c400fd885c19123ea840ee06a76a140f407", "question_id": "q07_negation", "split": "test", "cluster_id": "syn_03003_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 4, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1866, "query_tokens": 579, "message_tokens": 498, "message_utf8_bytes": 1106, "cache_tensor_bytes": null, "serialized_payload_bytes": 1106, "side_information_bytes": 0, "fetched_bytes": 2289, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 87954.53012594953, "hops": {"1": 23068.98970901966, "2": 21813.183957943693, "3": 21540.921541978605, "4": 21531.434917007573}, "receiver_prefill_ms": 384.7130419453606, "ttft_ms": 52.850125008262694, "decode_ms": 50.922916969284415}, "total_wall_ms": 88443.66937596351, "ttft_ms": 52.850125008262694, "input_usage": {"lookups": 12}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "no", "evidence_artifact": {"ledger": {"facts_total": 58, "facts_retained": 32, "retention_rate": 0.5517241379310345, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.7142857142857143, "amount": 0.6666666666666666, "status": 0.6666666666666666, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1106}, "ledger_by_hop": {"1": {"facts_total": 58, "facts_retained": 32, "retention_rate": 0.5517241379310345, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.7142857142857143, "amount": 0.6666666666666666, "status": 0.6666666666666666, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1200}, "2": {"facts_total": 58, "facts_retained": 32, "retention_rate": 0.5517241379310345, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.7142857142857143, "amount": 0.6666666666666666, "status": 0.6666666666666666, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1121}, "3": {"facts_total": 58, "facts_retained": 32, "retention_rate": 0.5517241379310345, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.7142857142857143, "amount": 0.6666666666666666, "status": 0.6666666666666666, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1106}, "4": {"facts_total": 58, "facts_retained": 32, "retention_rate": 0.5517241379310345, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.7142857142857143, "amount": 0.6666666666666666, "status": 0.6666666666666666, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1106}}}, "reference_key_id": "keys/synthetic/syn_03003_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "fd539b5069a942d3034d38027c24271a19ab0d754e9dee3800caa6490c9681d4", "output_hash": "9390298f3fb0c5b160498935d79cb139aef28e1c47358b4bbba61862b9c26e59"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "5a84cbeaf8c3c54b3f85fbf7", "attempt_id": "353c7fb51ce9", "run_group_id": "E4_core:A4|repairable|qwen3b:h4", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T06:45:27.659458Z", "task_family": "synthetic", "document_id": "syn_03003_2000", "document_hash": "dddca1a38e3224870fcfae927c701c400fd885c19123ea840ee06a76a140f407", "question_id": "q08_revision", "split": "test", "cluster_id": "syn_03003_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 4, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1866, "query_tokens": 597, "message_tokens": 498, "message_utf8_bytes": 1106, "cache_tensor_bytes": null, "serialized_payload_bytes": 1106, "side_information_bytes": 0, "fetched_bytes": 2289, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 87954.53012594953, "hops": {"1": 23068.98970901966, "2": 21813.183957943693, "3": 21540.921541978605, "4": 21531.434917007573}, "receiver_prefill_ms": 551.8552920548245, "ttft_ms": 37.10300009697676, "decode_ms": 252.68958299420774}, "total_wall_ms": 88796.96937603876, "ttft_ms": 37.10300009697676, "input_usage": {"lookups": 12}, "output_usage": {"output_tokens": 7}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "911,925", "evidence_artifact": {"ledger": {"facts_total": 58, "facts_retained": 32, "retention_rate": 0.5517241379310345, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.7142857142857143, "amount": 0.6666666666666666, "status": 0.6666666666666666, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1106}, "ledger_by_hop": {"1": {"facts_total": 58, "facts_retained": 32, "retention_rate": 0.5517241379310345, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.7142857142857143, "amount": 0.6666666666666666, "status": 0.6666666666666666, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1200}, "2": {"facts_total": 58, "facts_retained": 32, "retention_rate": 0.5517241379310345, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.7142857142857143, "amount": 0.6666666666666666, "status": 0.6666666666666666, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1121}, "3": {"facts_total": 58, "facts_retained": 32, "retention_rate": 0.5517241379310345, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.7142857142857143, "amount": 0.6666666666666666, "status": 0.6666666666666666, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1106}, "4": {"facts_total": 58, "facts_retained": 32, "retention_rate": 0.5517241379310345, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.7142857142857143, "amount": 0.6666666666666666, "status": 0.6666666666666666, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1106}}}, "reference_key_id": "keys/synthetic/syn_03003_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "fd539b5069a942d3034d38027c24271a19ab0d754e9dee3800caa6490c9681d4", "output_hash": "86be78720fc4b79f93e603e7d9e4861fcd25102612f73209fffc17d787056007"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "7a0ed8709d1b130846da995c", "attempt_id": "93474ddd96b1", "run_group_id": "E4_core:A4|repairable|qwen3b:h4", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T06:45:28.519347Z", "task_family": "synthetic", "document_id": "syn_03003_2000", "document_hash": "dddca1a38e3224870fcfae927c701c400fd885c19123ea840ee06a76a140f407", "question_id": "q09_revision", "split": "test", "cluster_id": "syn_03003_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 4, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1866, "query_tokens": 590, "message_tokens": 498, "message_utf8_bytes": 1106, "cache_tensor_bytes": null, "serialized_payload_bytes": 1106, "side_information_bytes": 0, "fetched_bytes": 2289, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 87954.53012594953, "hops": {"1": 23068.98970901966, "2": 21813.183957943693, "3": 21540.921541978605, "4": 21531.434917007573}, "receiver_prefill_ms": 398.23008398525417, "ttft_ms": 27.180875069461763, "decode_ms": 141.10341703053564}, "total_wall_ms": 88521.84008387849, "ttft_ms": 27.180875069461763, "input_usage": {"lookups": 12}, "output_usage": {"output_tokens": 6}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "911925", "evidence_artifact": {"ledger": {"facts_total": 58, "facts_retained": 32, "retention_rate": 0.5517241379310345, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.7142857142857143, "amount": 0.6666666666666666, "status": 0.6666666666666666, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1106}, "ledger_by_hop": {"1": {"facts_total": 58, "facts_retained": 32, "retention_rate": 0.5517241379310345, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.7142857142857143, "amount": 0.6666666666666666, "status": 0.6666666666666666, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1200}, "2": {"facts_total": 58, "facts_retained": 32, "retention_rate": 0.5517241379310345, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.7142857142857143, "amount": 0.6666666666666666, "status": 0.6666666666666666, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1121}, "3": {"facts_total": 58, "facts_retained": 32, "retention_rate": 0.5517241379310345, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.7142857142857143, "amount": 0.6666666666666666, "status": 0.6666666666666666, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1106}, "4": {"facts_total": 58, "facts_retained": 32, "retention_rate": 0.5517241379310345, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.7142857142857143, "amount": 0.6666666666666666, "status": 0.6666666666666666, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1106}}}, "reference_key_id": "keys/synthetic/syn_03003_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "fd539b5069a942d3034d38027c24271a19ab0d754e9dee3800caa6490c9681d4", "output_hash": "679a540dc4739f0ace623448b8fd925fb234f554838d7e87da223228ee81b112"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "46d407422640e6f9f28018ac", "attempt_id": "b12e9258f0fc", "run_group_id": "E4_core:A4|repairable|qwen3b:h4", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T06:45:29.103346Z", "task_family": "synthetic", "document_id": "syn_03003_2000", "document_hash": "dddca1a38e3224870fcfae927c701c400fd885c19123ea840ee06a76a140f407", "question_id": "q10_temporal", "split": "test", "cluster_id": "syn_03003_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 4, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1866, "query_tokens": 597, "message_tokens": 498, "message_utf8_bytes": 1106, "cache_tensor_bytes": null, "serialized_payload_bytes": 1106, "side_information_bytes": 0, "fetched_bytes": 2289, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 87954.53012594953, "hops": {"1": 23068.98970901966, "2": 21813.183957943693, "3": 21540.921541978605, "4": 21531.434917007573}, "receiver_prefill_ms": 402.2439579712227, "ttft_ms": 49.25695795100182, "decode_ms": 368.53408289607614}, "total_wall_ms": 88775.40450100787, "ttft_ms": 49.25695795100182, "input_usage": {"lookups": 12}, "output_usage": {"output_tokens": 6}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "Gamtiszen-2", "evidence_artifact": {"ledger": {"facts_total": 58, "facts_retained": 32, "retention_rate": 0.5517241379310345, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.7142857142857143, "amount": 0.6666666666666666, "status": 0.6666666666666666, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1106}, "ledger_by_hop": {"1": {"facts_total": 58, "facts_retained": 32, "retention_rate": 0.5517241379310345, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.7142857142857143, "amount": 0.6666666666666666, "status": 0.6666666666666666, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1200}, "2": {"facts_total": 58, "facts_retained": 32, "retention_rate": 0.5517241379310345, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.7142857142857143, "amount": 0.6666666666666666, "status": 0.6666666666666666, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1121}, "3": {"facts_total": 58, "facts_retained": 32, "retention_rate": 0.5517241379310345, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.7142857142857143, "amount": 0.6666666666666666, "status": 0.6666666666666666, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1106}, "4": {"facts_total": 58, "facts_retained": 32, "retention_rate": 0.5517241379310345, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.7142857142857143, "amount": 0.6666666666666666, "status": 0.6666666666666666, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1106}}}, "reference_key_id": "keys/synthetic/syn_03003_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "fd539b5069a942d3034d38027c24271a19ab0d754e9dee3800caa6490c9681d4", "output_hash": "32e3af428886434823d82102e1c482765eafdf6077fe8d3fba5752ec5e29826c"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "7cc23131ca53ed5ed8d5f575", "attempt_id": "c22c25b979f3", "run_group_id": "E4_core:A4|repairable|qwen3b:h4", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T06:45:29.941702Z", "task_family": "synthetic", "document_id": "syn_03003_2000", "document_hash": "dddca1a38e3224870fcfae927c701c400fd885c19123ea840ee06a76a140f407", "question_id": "q11_aggregation", "split": "test", "cluster_id": "syn_03003_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 4, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1866, "query_tokens": 591, "message_tokens": 498, "message_utf8_bytes": 1106, "cache_tensor_bytes": null, "serialized_payload_bytes": 1106, "side_information_bytes": 0, "fetched_bytes": 2289, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 87954.53012594953, "hops": {"1": 23068.98970901966, "2": 21813.183957943693, "3": 21540.921541978605, "4": 21531.434917007573}, "receiver_prefill_ms": 410.4874589247629, "ttft_ms": 27.691125054843724, "decode_ms": 348.78691693302244}, "total_wall_ms": 88742.42641800083, "ttft_ms": 27.691125054843724, "input_usage": {"lookups": 12}, "output_usage": {"output_tokens": 10}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "33,506,345", "evidence_artifact": {"ledger": {"facts_total": 58, "facts_retained": 32, "retention_rate": 0.5517241379310345, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.7142857142857143, "amount": 0.6666666666666666, "status": 0.6666666666666666, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1106}, "ledger_by_hop": {"1": {"facts_total": 58, "facts_retained": 32, "retention_rate": 0.5517241379310345, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.7142857142857143, "amount": 0.6666666666666666, "status": 0.6666666666666666, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1200}, "2": {"facts_total": 58, "facts_retained": 32, "retention_rate": 0.5517241379310345, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.7142857142857143, "amount": 0.6666666666666666, "status": 0.6666666666666666, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1121}, "3": {"facts_total": 58, "facts_retained": 32, "retention_rate": 0.5517241379310345, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.7142857142857143, "amount": 0.6666666666666666, "status": 0.6666666666666666, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1106}, "4": {"facts_total": 58, "facts_retained": 32, "retention_rate": 0.5517241379310345, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.7142857142857143, "amount": 0.6666666666666666, "status": 0.6666666666666666, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1106}}}, "reference_key_id": "keys/synthetic/syn_03003_2000.json", "task_scores": {"correct": 0.0, "abstained": 0.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "fd539b5069a942d3034d38027c24271a19ab0d754e9dee3800caa6490c9681d4", "output_hash": "de275001d02d943760a37daf3a991b5fdb2fa479267a2e7f5723f623b23797db"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "930076d9c2b84e1ad4c05c1a", "attempt_id": "7d6ebb748d3f", "run_group_id": "E4_core:A4|repairable|qwen3b:h4", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T06:45:30.747083Z", "task_family": "synthetic", "document_id": "syn_03003_2000", "document_hash": "dddca1a38e3224870fcfae927c701c400fd885c19123ea840ee06a76a140f407", "question_id": "q12_unit_conversion", "split": "test", "cluster_id": "syn_03003_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 4, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1866, "query_tokens": 588, "message_tokens": 498, "message_utf8_bytes": 1106, "cache_tensor_bytes": null, "serialized_payload_bytes": 1106, "side_information_bytes": 0, "fetched_bytes": 2289, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 87954.53012594953, "hops": {"1": 23068.98970901966, "2": 21813.183957943693, "3": 21540.921541978605, "4": 21531.434917007573}, "receiver_prefill_ms": 374.4612500304356, "ttft_ms": 36.078040953725576, "decode_ms": 332.3022499680519}, "total_wall_ms": 88698.04487598594, "ttft_ms": 36.078040953725576, "input_usage": {"lookups": 12}, "output_usage": {"output_tokens": 5}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "17430", "evidence_artifact": {"ledger": {"facts_total": 58, "facts_retained": 32, "retention_rate": 0.5517241379310345, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.7142857142857143, "amount": 0.6666666666666666, "status": 0.6666666666666666, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1106}, "ledger_by_hop": {"1": {"facts_total": 58, "facts_retained": 32, "retention_rate": 0.5517241379310345, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.7142857142857143, "amount": 0.6666666666666666, "status": 0.6666666666666666, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1200}, "2": {"facts_total": 58, "facts_retained": 32, "retention_rate": 0.5517241379310345, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.7142857142857143, "amount": 0.6666666666666666, "status": 0.6666666666666666, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1121}, "3": {"facts_total": 58, "facts_retained": 32, "retention_rate": 0.5517241379310345, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.7142857142857143, "amount": 0.6666666666666666, "status": 0.6666666666666666, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1106}, "4": {"facts_total": 58, "facts_retained": 32, "retention_rate": 0.5517241379310345, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.7142857142857143, "amount": 0.6666666666666666, "status": 0.6666666666666666, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1106}}}, "reference_key_id": "keys/synthetic/syn_03003_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "fd539b5069a942d3034d38027c24271a19ab0d754e9dee3800caa6490c9681d4", "output_hash": "2575eaf3b748efc209b676518766751eb038c88f74578951345818bfe7f7194b"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "8478b001c9d0d84aad41d62c", "attempt_id": "78167e378df2", "run_group_id": "E4_core:A4|repairable|qwen3b:h4", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T06:45:31.507045Z", "task_family": "synthetic", "document_id": "syn_03003_2000", "document_hash": "dddca1a38e3224870fcfae927c701c400fd885c19123ea840ee06a76a140f407", "question_id": "q13_unanswerable", "split": "test", "cluster_id": "syn_03003_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 4, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1866, "query_tokens": 592, "message_tokens": 498, "message_utf8_bytes": 1106, "cache_tensor_bytes": null, "serialized_payload_bytes": 1106, "side_information_bytes": 0, "fetched_bytes": 2289, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 87954.53012594953, "hops": {"1": 23068.98970901966, "2": 21813.183957943693, "3": 21540.921541978605, "4": 21531.434917007573}, "receiver_prefill_ms": 405.9615839505568, "ttft_ms": 28.971249936148524, "decode_ms": 35.18154192715883}, "total_wall_ms": 88425.2725429833, "ttft_ms": 28.971249936148524, "input_usage": {"lookups": 12}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 58, "facts_retained": 32, "retention_rate": 0.5517241379310345, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.7142857142857143, "amount": 0.6666666666666666, "status": 0.6666666666666666, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1106}, "ledger_by_hop": {"1": {"facts_total": 58, "facts_retained": 32, "retention_rate": 0.5517241379310345, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.7142857142857143, "amount": 0.6666666666666666, "status": 0.6666666666666666, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1200}, "2": {"facts_total": 58, "facts_retained": 32, "retention_rate": 0.5517241379310345, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.7142857142857143, "amount": 0.6666666666666666, "status": 0.6666666666666666, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1121}, "3": {"facts_total": 58, "facts_retained": 32, "retention_rate": 0.5517241379310345, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.7142857142857143, "amount": 0.6666666666666666, "status": 0.6666666666666666, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1106}, "4": {"facts_total": 58, "facts_retained": 32, "retention_rate": 0.5517241379310345, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.7142857142857143, "amount": 0.6666666666666666, "status": 0.6666666666666666, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1106}}}, "reference_key_id": "keys/synthetic/syn_03003_2000.json", "task_scores": {"correct": 1.0, "abstained": 1.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "fd539b5069a942d3034d38027c24271a19ab0d754e9dee3800caa6490c9681d4", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "5ee8c144bd209af4890e70ec", "attempt_id": "039206ca47a8", "run_group_id": "E4_core:A4|repairable|qwen3b:h4", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T06:45:31.994365Z", "task_family": "synthetic", "document_id": "syn_03003_2000", "document_hash": "dddca1a38e3224870fcfae927c701c400fd885c19123ea840ee06a76a140f407", "question_id": "q14_unanswerable", "split": "test", "cluster_id": "syn_03003_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 4, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1866, "query_tokens": 593, "message_tokens": 498, "message_utf8_bytes": 1106, "cache_tensor_bytes": null, "serialized_payload_bytes": 1106, "side_information_bytes": 0, "fetched_bytes": 2289, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 87954.53012594953, "hops": {"1": 23068.98970901966, "2": 21813.183957943693, "3": 21540.921541978605, "4": 21531.434917007573}, "receiver_prefill_ms": 381.3343750080094, "ttft_ms": 21.213000058196485, "decode_ms": 39.58354191854596}, "total_wall_ms": 88397.45350100566, "ttft_ms": 21.213000058196485, "input_usage": {"lookups": 12}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 58, "facts_retained": 32, "retention_rate": 0.5517241379310345, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.7142857142857143, "amount": 0.6666666666666666, "status": 0.6666666666666666, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1106}, "ledger_by_hop": {"1": {"facts_total": 58, "facts_retained": 32, "retention_rate": 0.5517241379310345, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.7142857142857143, "amount": 0.6666666666666666, "status": 0.6666666666666666, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1200}, "2": {"facts_total": 58, "facts_retained": 32, "retention_rate": 0.5517241379310345, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.7142857142857143, "amount": 0.6666666666666666, "status": 0.6666666666666666, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1121}, "3": {"facts_total": 58, "facts_retained": 32, "retention_rate": 0.5517241379310345, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.7142857142857143, "amount": 0.6666666666666666, "status": 0.6666666666666666, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1106}, "4": {"facts_total": 58, "facts_retained": 32, "retention_rate": 0.5517241379310345, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.7142857142857143, "amount": 0.6666666666666666, "status": 0.6666666666666666, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1106}}}, "reference_key_id": "keys/synthetic/syn_03003_2000.json", "task_scores": {"correct": 1.0, "abstained": 1.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "fd539b5069a942d3034d38027c24271a19ab0d754e9dee3800caa6490c9681d4", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "2aefb36b0d2c60064c40814c", "attempt_id": "e93401d41139", "run_group_id": "E4_core:A4|repairable|qwen3b:h8", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T06:46:59.103978Z", "task_family": "synthetic", "document_id": "syn_03003_2000", "document_hash": "dddca1a38e3224870fcfae927c701c400fd885c19123ea840ee06a76a140f407", "question_id": "q00_single_fact", "split": "test", "cluster_id": "syn_03003_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 8, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1866, "query_tokens": 580, "message_tokens": 498, "message_utf8_bytes": 1106, "cache_tensor_bytes": null, "serialized_payload_bytes": 1106, "side_information_bytes": 0, "fetched_bytes": 5341, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 174613.7597510824, "hops": {"1": 23068.98970901966, "2": 21813.183957943693, "3": 21540.921541978605, "4": 21531.434917007573, "5": 21513.298875070177, "6": 21784.19441706501, "7": 21694.188333000056, "8": 21667.547999997623}, "receiver_prefill_ms": 405.5869169533253, "ttft_ms": 20.387833938002586, "decode_ms": 345.2832920011133}, "total_wall_ms": 175385.78237604816, "ttft_ms": 20.387833938002586, "input_usage": {"lookups": 28}, "output_usage": {"output_tokens": 10}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "GR-79S-5839", "evidence_artifact": {"ledger": {"facts_total": 58, "facts_retained": 32, "retention_rate": 0.5517241379310345, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.7142857142857143, "amount": 0.6666666666666666, "status": 0.6666666666666666, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1106}, "ledger_by_hop": {"1": {"facts_total": 58, "facts_retained": 32, "retention_rate": 0.5517241379310345, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.7142857142857143, "amount": 0.6666666666666666, "status": 0.6666666666666666, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1200}, "2": {"facts_total": 58, "facts_retained": 32, "retention_rate": 0.5517241379310345, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.7142857142857143, "amount": 0.6666666666666666, "status": 0.6666666666666666, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1121}, "3": {"facts_total": 58, "facts_retained": 32, "retention_rate": 0.5517241379310345, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.7142857142857143, "amount": 0.6666666666666666, "status": 0.6666666666666666, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1106}, "4": {"facts_total": 58, "facts_retained": 32, "retention_rate": 0.5517241379310345, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.7142857142857143, "amount": 0.6666666666666666, "status": 0.6666666666666666, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1106}, "5": {"facts_total": 58, "facts_retained": 32, "retention_rate": 0.5517241379310345, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.7142857142857143, "amount": 0.6666666666666666, "status": 0.6666666666666666, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1106}, "6": {"facts_total": 58, "facts_retained": 32, "retention_rate": 0.5517241379310345, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.7142857142857143, "amount": 0.6666666666666666, "status": 0.6666666666666666, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1106}, "7": {"facts_total": 58, "facts_retained": 32, "retention_rate": 0.5517241379310345, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.7142857142857143, "amount": 0.6666666666666666, "status": 0.6666666666666666, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1106}, "8": {"facts_total": 58, "facts_retained": 32, "retention_rate": 0.5517241379310345, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.7142857142857143, "amount": 0.6666666666666666, "status": 0.6666666666666666, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1106}}}, "reference_key_id": "keys/synthetic/syn_03003_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "fd539b5069a942d3034d38027c24271a19ab0d754e9dee3800caa6490c9681d4", "output_hash": "720cc40cd69e85e9728f6bea00b17a95ca07eee643e143a131840763958303cd"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "f3702fa4ea24537f39208ff8", "attempt_id": "13d0c4dc2e1d", "run_group_id": "E4_core:A4|repairable|qwen3b:h8", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T06:46:59.894075Z", "task_family": "synthetic", "document_id": "syn_03003_2000", "document_hash": "dddca1a38e3224870fcfae927c701c400fd885c19123ea840ee06a76a140f407", "question_id": "q01_single_fact", "split": "test", "cluster_id": "syn_03003_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 8, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1866, "query_tokens": 584, "message_tokens": 498, "message_utf8_bytes": 1106, "cache_tensor_bytes": null, "serialized_payload_bytes": 1106, "side_information_bytes": 0, "fetched_bytes": 5341, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 174613.7597510824, "hops": {"1": 23068.98970901966, "2": 21813.183957943693, "3": 21540.921541978605, "4": 21531.434917007573, "5": 21513.298875070177, "6": 21784.19441706501, "7": 21694.188333000056, "8": 21667.547999997623}, "receiver_prefill_ms": 392.4339159857482, "ttft_ms": 36.746166995726526, "decode_ms": 594.2188339540735}, "total_wall_ms": 175638.0527090514, "ttft_ms": 36.746166995726526, "input_usage": {"lookups": 28}, "output_usage": {"output_tokens": 11}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "Lot 344-B, Témiscaming", "evidence_artifact": {"ledger": {"facts_total": 58, "facts_retained": 32, "retention_rate": 0.5517241379310345, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.7142857142857143, "amount": 0.6666666666666666, "status": 0.6666666666666666, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1106}, "ledger_by_hop": {"1": {"facts_total": 58, "facts_retained": 32, "retention_rate": 0.5517241379310345, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.7142857142857143, "amount": 0.6666666666666666, "status": 0.6666666666666666, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1200}, "2": {"facts_total": 58, "facts_retained": 32, "retention_rate": 0.5517241379310345, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.7142857142857143, "amount": 0.6666666666666666, "status": 0.6666666666666666, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1121}, "3": {"facts_total": 58, "facts_retained": 32, "retention_rate": 0.5517241379310345, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.7142857142857143, "amount": 0.6666666666666666, "status": 0.6666666666666666, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1106}, "4": {"facts_total": 58, "facts_retained": 32, "retention_rate": 0.5517241379310345, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.7142857142857143, "amount": 0.6666666666666666, "status": 0.6666666666666666, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1106}, "5": {"facts_total": 58, "facts_retained": 32, "retention_rate": 0.5517241379310345, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.7142857142857143, "amount": 0.6666666666666666, "status": 0.6666666666666666, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1106}, "6": {"facts_total": 58, "facts_retained": 32, "retention_rate": 0.5517241379310345, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.7142857142857143, "amount": 0.6666666666666666, "status": 0.6666666666666666, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1106}, "7": {"facts_total": 58, "facts_retained": 32, "retention_rate": 0.5517241379310345, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.7142857142857143, "amount": 0.6666666666666666, "status": 0.6666666666666666, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1106}, "8": {"facts_total": 58, "facts_retained": 32, "retention_rate": 0.5517241379310345, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.7142857142857143, "amount": 0.6666666666666666, "status": 0.6666666666666666, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1106}}}, "reference_key_id": "keys/synthetic/syn_03003_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 0.8571428571428571}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "fd539b5069a942d3034d38027c24271a19ab0d754e9dee3800caa6490c9681d4", "output_hash": "0197d491a5a04a142dc8e372360735a1d41757e7a5aea7e81f23b0e1ef6ee010"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "f4db63fac8c160d86a0a8e9d", "attempt_id": "d0d57c7b4a62", "run_group_id": "E4_core:A4|repairable|qwen3b:h8", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T06:47:00.936083Z", "task_family": "synthetic", "document_id": "syn_03003_2000", "document_hash": "dddca1a38e3224870fcfae927c701c400fd885c19123ea840ee06a76a140f407", "question_id": "q02_exact_copy", "split": "test", "cluster_id": "syn_03003_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 8, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1866, "query_tokens": 584, "message_tokens": 498, "message_utf8_bytes": 1106, "cache_tensor_bytes": null, "serialized_payload_bytes": 1106, "side_information_bytes": 0, "fetched_bytes": 5341, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 174613.7597510824, "hops": {"1": 23068.98970901966, "2": 21813.183957943693, "3": 21540.921541978605, "4": 21531.434917007573, "5": 21513.298875070177, "6": 21784.19441706501, "7": 21694.188333000056, "8": 21667.547999997623}, "receiver_prefill_ms": 414.4523750292137, "ttft_ms": 18.77741690259427, "decode_ms": 36.251540994271636}, "total_wall_ms": 175083.81541713607, "ttft_ms": 18.77741690259427, "input_usage": {"lookups": 28}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 58, "facts_retained": 32, "retention_rate": 0.5517241379310345, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.7142857142857143, "amount": 0.6666666666666666, "status": 0.6666666666666666, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1106}, "ledger_by_hop": {"1": {"facts_total": 58, "facts_retained": 32, "retention_rate": 0.5517241379310345, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.7142857142857143, "amount": 0.6666666666666666, "status": 0.6666666666666666, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1200}, "2": {"facts_total": 58, "facts_retained": 32, "retention_rate": 0.5517241379310345, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.7142857142857143, "amount": 0.6666666666666666, "status": 0.6666666666666666, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1121}, "3": {"facts_total": 58, "facts_retained": 32, "retention_rate": 0.5517241379310345, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.7142857142857143, "amount": 0.6666666666666666, "status": 0.6666666666666666, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1106}, "4": {"facts_total": 58, "facts_retained": 32, "retention_rate": 0.5517241379310345, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.7142857142857143, "amount": 0.6666666666666666, "status": 0.6666666666666666, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1106}, "5": {"facts_total": 58, "facts_retained": 32, "retention_rate": 0.5517241379310345, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.7142857142857143, "amount": 0.6666666666666666, "status": 0.6666666666666666, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1106}, "6": {"facts_total": 58, "facts_retained": 32, "retention_rate": 0.5517241379310345, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.7142857142857143, "amount": 0.6666666666666666, "status": 0.6666666666666666, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1106}, "7": {"facts_total": 58, "facts_retained": 32, "retention_rate": 0.5517241379310345, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.7142857142857143, "amount": 0.6666666666666666, "status": 0.6666666666666666, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1106}, "8": {"facts_total": 58, "facts_retained": 32, "retention_rate": 0.5517241379310345, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.7142857142857143, "amount": 0.6666666666666666, "status": 0.6666666666666666, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1106}}}, "reference_key_id": "keys/synthetic/syn_03003_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "fd539b5069a942d3034d38027c24271a19ab0d754e9dee3800caa6490c9681d4", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "c457e8251eec6e4cd5f2ba95", "attempt_id": "6ceadb25332b", "run_group_id": "E4_core:A4|repairable|qwen3b:h8", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T06:47:01.424738Z", "task_family": "synthetic", "document_id": "syn_03003_2000", "document_hash": "dddca1a38e3224870fcfae927c701c400fd885c19123ea840ee06a76a140f407", "question_id": "q03_exact_copy", "split": "test", "cluster_id": "syn_03003_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 8, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1866, "query_tokens": 579, "message_tokens": 498, "message_utf8_bytes": 1106, "cache_tensor_bytes": null, "serialized_payload_bytes": 1106, "side_information_bytes": 0, "fetched_bytes": 5341, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 174613.7597510824, "hops": {"1": 23068.98970901966, "2": 21813.183957943693, "3": 21540.921541978605, "4": 21531.434917007573, "5": 21513.298875070177, "6": 21784.19441706501, "7": 21694.188333000056, "8": 21667.547999997623}, "receiver_prefill_ms": 354.03258295264095, "ttft_ms": 64.02016698848456, "decode_ms": 432.7584580751136}, "total_wall_ms": 175465.21500113886, "ttft_ms": 64.02016698848456, "input_usage": {"lookups": 28}, "output_usage": {"output_tokens": 6}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "Wilhelmina Achterberg", "evidence_artifact": {"ledger": {"facts_total": 58, "facts_retained": 32, "retention_rate": 0.5517241379310345, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.7142857142857143, "amount": 0.6666666666666666, "status": 0.6666666666666666, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1106}, "ledger_by_hop": {"1": {"facts_total": 58, "facts_retained": 32, "retention_rate": 0.5517241379310345, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.7142857142857143, "amount": 0.6666666666666666, "status": 0.6666666666666666, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1200}, "2": {"facts_total": 58, "facts_retained": 32, "retention_rate": 0.5517241379310345, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.7142857142857143, "amount": 0.6666666666666666, "status": 0.6666666666666666, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1121}, "3": {"facts_total": 58, "facts_retained": 32, "retention_rate": 0.5517241379310345, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.7142857142857143, "amount": 0.6666666666666666, "status": 0.6666666666666666, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1106}, "4": {"facts_total": 58, "facts_retained": 32, "retention_rate": 0.5517241379310345, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.7142857142857143, "amount": 0.6666666666666666, "status": 0.6666666666666666, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1106}, "5": {"facts_total": 58, "facts_retained": 32, "retention_rate": 0.5517241379310345, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.7142857142857143, "amount": 0.6666666666666666, "status": 0.6666666666666666, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1106}, "6": {"facts_total": 58, "facts_retained": 32, "retention_rate": 0.5517241379310345, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.7142857142857143, "amount": 0.6666666666666666, "status": 0.6666666666666666, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1106}, "7": {"facts_total": 58, "facts_retained": 32, "retention_rate": 0.5517241379310345, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.7142857142857143, "amount": 0.6666666666666666, "status": 0.6666666666666666, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1106}, "8": {"facts_total": 58, "facts_retained": 32, "retention_rate": 0.5517241379310345, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.7142857142857143, "amount": 0.6666666666666666, "status": 0.6666666666666666, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1106}}}, "reference_key_id": "keys/synthetic/syn_03003_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "fd539b5069a942d3034d38027c24271a19ab0d754e9dee3800caa6490c9681d4", "output_hash": "2a62606853c315bba8769371d195e3ab3d6a8cd7ef959205575c5119740850f8"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "6dd1c1b3d32c04f0cce20f23", "attempt_id": "f05f50583cf7", "run_group_id": "E4_core:A4|repairable|qwen3b:h8", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T06:47:02.294121Z", "task_family": "synthetic", "document_id": "syn_03003_2000", "document_hash": "dddca1a38e3224870fcfae927c701c400fd885c19123ea840ee06a76a140f407", "question_id": "q04_multi_hop", "split": "test", "cluster_id": "syn_03003_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 8, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1866, "query_tokens": 588, "message_tokens": 498, "message_utf8_bytes": 1106, "cache_tensor_bytes": null, "serialized_payload_bytes": 1106, "side_information_bytes": 0, "fetched_bytes": 5341, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 174613.7597510824, "hops": {"1": 23068.98970901966, "2": 21813.183957943693, "3": 21540.921541978605, "4": 21531.434917007573, "5": 21513.298875070177, "6": 21784.19441706501, "7": 21694.188333000056, "8": 21667.547999997623}, "receiver_prefill_ms": 402.59837498888373, "ttft_ms": 22.061542025767267, "decode_ms": 37.416458944790065}, "total_wall_ms": 175076.52883511037, "ttft_ms": 22.061542025767267, "input_usage": {"lookups": 28}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 58, "facts_retained": 32, "retention_rate": 0.5517241379310345, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.7142857142857143, "amount": 0.6666666666666666, "status": 0.6666666666666666, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1106}, "ledger_by_hop": {"1": {"facts_total": 58, "facts_retained": 32, "retention_rate": 0.5517241379310345, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.7142857142857143, "amount": 0.6666666666666666, "status": 0.6666666666666666, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1200}, "2": {"facts_total": 58, "facts_retained": 32, "retention_rate": 0.5517241379310345, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.7142857142857143, "amount": 0.6666666666666666, "status": 0.6666666666666666, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1121}, "3": {"facts_total": 58, "facts_retained": 32, "retention_rate": 0.5517241379310345, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.7142857142857143, "amount": 0.6666666666666666, "status": 0.6666666666666666, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1106}, "4": {"facts_total": 58, "facts_retained": 32, "retention_rate": 0.5517241379310345, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.7142857142857143, "amount": 0.6666666666666666, "status": 0.6666666666666666, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1106}, "5": {"facts_total": 58, "facts_retained": 32, "retention_rate": 0.5517241379310345, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.7142857142857143, "amount": 0.6666666666666666, "status": 0.6666666666666666, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1106}, "6": {"facts_total": 58, "facts_retained": 32, "retention_rate": 0.5517241379310345, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.7142857142857143, "amount": 0.6666666666666666, "status": 0.6666666666666666, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1106}, "7": {"facts_total": 58, "facts_retained": 32, "retention_rate": 0.5517241379310345, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.7142857142857143, "amount": 0.6666666666666666, "status": 0.6666666666666666, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1106}, "8": {"facts_total": 58, "facts_retained": 32, "retention_rate": 0.5517241379310345, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.7142857142857143, "amount": 0.6666666666666666, "status": 0.6666666666666666, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1106}}}, "reference_key_id": "keys/synthetic/syn_03003_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "fd539b5069a942d3034d38027c24271a19ab0d754e9dee3800caa6490c9681d4", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "3912d20db52caf0c119bc3ec", "attempt_id": "dc28987f033c", "run_group_id": "E4_core:A4|repairable|qwen3b:h8", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T06:47:02.774402Z", "task_family": "synthetic", "document_id": "syn_03003_2000", "document_hash": "dddca1a38e3224870fcfae927c701c400fd885c19123ea840ee06a76a140f407", "question_id": "q05_multi_hop", "split": "test", "cluster_id": "syn_03003_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 8, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1866, "query_tokens": 590, "message_tokens": 498, "message_utf8_bytes": 1106, "cache_tensor_bytes": null, "serialized_payload_bytes": 1106, "side_information_bytes": 0, "fetched_bytes": 5341, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 174613.7597510824, "hops": {"1": 23068.98970901966, "2": 21813.183957943693, "3": 21540.921541978605, "4": 21531.434917007573, "5": 21513.298875070177, "6": 21784.19441706501, "7": 21694.188333000056, "8": 21667.547999997623}, "receiver_prefill_ms": 408.608749974519, "ttft_ms": 30.291375005617738, "decode_ms": 621.2514590006322}, "total_wall_ms": 175674.5632099919, "ttft_ms": 30.291375005617738, "input_usage": {"lookups": 28}, "output_usage": {"output_tokens": 10}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "NK-95Z-2165", "evidence_artifact": {"ledger": {"facts_total": 58, "facts_retained": 32, "retention_rate": 0.5517241379310345, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.7142857142857143, "amount": 0.6666666666666666, "status": 0.6666666666666666, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1106}, "ledger_by_hop": {"1": {"facts_total": 58, "facts_retained": 32, "retention_rate": 0.5517241379310345, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.7142857142857143, "amount": 0.6666666666666666, "status": 0.6666666666666666, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1200}, "2": {"facts_total": 58, "facts_retained": 32, "retention_rate": 0.5517241379310345, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.7142857142857143, "amount": 0.6666666666666666, "status": 0.6666666666666666, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1121}, "3": {"facts_total": 58, "facts_retained": 32, "retention_rate": 0.5517241379310345, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.7142857142857143, "amount": 0.6666666666666666, "status": 0.6666666666666666, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1106}, "4": {"facts_total": 58, "facts_retained": 32, "retention_rate": 0.5517241379310345, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.7142857142857143, "amount": 0.6666666666666666, "status": 0.6666666666666666, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1106}, "5": {"facts_total": 58, "facts_retained": 32, "retention_rate": 0.5517241379310345, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.7142857142857143, "amount": 0.6666666666666666, "status": 0.6666666666666666, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1106}, "6": {"facts_total": 58, "facts_retained": 32, "retention_rate": 0.5517241379310345, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.7142857142857143, "amount": 0.6666666666666666, "status": 0.6666666666666666, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1106}, "7": {"facts_total": 58, "facts_retained": 32, "retention_rate": 0.5517241379310345, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.7142857142857143, "amount": 0.6666666666666666, "status": 0.6666666666666666, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1106}, "8": {"facts_total": 58, "facts_retained": 32, "retention_rate": 0.5517241379310345, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.7142857142857143, "amount": 0.6666666666666666, "status": 0.6666666666666666, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1106}}}, "reference_key_id": "keys/synthetic/syn_03003_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "fd539b5069a942d3034d38027c24271a19ab0d754e9dee3800caa6490c9681d4", "output_hash": "bd8f809b02304c6b9dc7c1e116ecccc2328c976a5c071e89a94d661b48e73aa7"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "84ab7cf29e2fcad148cc3018", "attempt_id": "23b2a2154e70", "run_group_id": "E4_core:A4|repairable|qwen3b:h8", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T06:47:03.853357Z", "task_family": "synthetic", "document_id": "syn_03003_2000", "document_hash": "dddca1a38e3224870fcfae927c701c400fd885c19123ea840ee06a76a140f407", "question_id": "q06_negation", "split": "test", "cluster_id": "syn_03003_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 8, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1866, "query_tokens": 582, "message_tokens": 498, "message_utf8_bytes": 1106, "cache_tensor_bytes": null, "serialized_payload_bytes": 1106, "side_information_bytes": 0, "fetched_bytes": 5341, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 174613.7597510824, "hops": {"1": 23068.98970901966, "2": 21813.183957943693, "3": 21540.921541978605, "4": 21531.434917007573, "5": 21513.298875070177, "6": 21784.19441706501, "7": 21694.188333000056, "8": 21667.547999997623}, "receiver_prefill_ms": 385.17908297944814, "ttft_ms": 40.64091597683728, "decode_ms": 27.40775002166629}, "total_wall_ms": 175067.6207090728, "ttft_ms": 40.64091597683728, "input_usage": {"lookups": 28}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "yes", "evidence_artifact": {"ledger": {"facts_total": 58, "facts_retained": 32, "retention_rate": 0.5517241379310345, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.7142857142857143, "amount": 0.6666666666666666, "status": 0.6666666666666666, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1106}, "ledger_by_hop": {"1": {"facts_total": 58, "facts_retained": 32, "retention_rate": 0.5517241379310345, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.7142857142857143, "amount": 0.6666666666666666, "status": 0.6666666666666666, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1200}, "2": {"facts_total": 58, "facts_retained": 32, "retention_rate": 0.5517241379310345, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.7142857142857143, "amount": 0.6666666666666666, "status": 0.6666666666666666, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1121}, "3": {"facts_total": 58, "facts_retained": 32, "retention_rate": 0.5517241379310345, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.7142857142857143, "amount": 0.6666666666666666, "status": 0.6666666666666666, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1106}, "4": {"facts_total": 58, "facts_retained": 32, "retention_rate": 0.5517241379310345, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.7142857142857143, "amount": 0.6666666666666666, "status": 0.6666666666666666, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1106}, "5": {"facts_total": 58, "facts_retained": 32, "retention_rate": 0.5517241379310345, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.7142857142857143, "amount": 0.6666666666666666, "status": 0.6666666666666666, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1106}, "6": {"facts_total": 58, "facts_retained": 32, "retention_rate": 0.5517241379310345, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.7142857142857143, "amount": 0.6666666666666666, "status": 0.6666666666666666, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1106}, "7": {"facts_total": 58, "facts_retained": 32, "retention_rate": 0.5517241379310345, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.7142857142857143, "amount": 0.6666666666666666, "status": 0.6666666666666666, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1106}, "8": {"facts_total": 58, "facts_retained": 32, "retention_rate": 0.5517241379310345, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.7142857142857143, "amount": 0.6666666666666666, "status": 0.6666666666666666, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1106}}}, "reference_key_id": "keys/synthetic/syn_03003_2000.json", "task_scores": {"correct": 0.0, "abstained": 0.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "fd539b5069a942d3034d38027c24271a19ab0d754e9dee3800caa6490c9681d4", "output_hash": "8a798890fe93817163b10b5f7bd2ca4d25d84c52739a645a889c173eee7d9d3d"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "dae30e9909fd5fac803954c8", "attempt_id": "c68f5ede54ea", "run_group_id": "E4_core:A4|repairable|qwen3b:h8", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T06:47:04.324623Z", "task_family": "synthetic", "document_id": "syn_03003_2000", "document_hash": "dddca1a38e3224870fcfae927c701c400fd885c19123ea840ee06a76a140f407", "question_id": "q07_negation", "split": "test", "cluster_id": "syn_03003_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 8, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1866, "query_tokens": 579, "message_tokens": 498, "message_utf8_bytes": 1106, "cache_tensor_bytes": null, "serialized_payload_bytes": 1106, "side_information_bytes": 0, "fetched_bytes": 5341, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 174613.7597510824, "hops": {"1": 23068.98970901966, "2": 21813.183957943693, "3": 21540.921541978605, "4": 21531.434917007573, "5": 21513.298875070177, "6": 21784.19441706501, "7": 21694.188333000056, "8": 21667.547999997623}, "receiver_prefill_ms": 372.5188340758905, "ttft_ms": 36.27116605639458, "decode_ms": 19.143500016070902}, "total_wall_ms": 175042.4251680961, "ttft_ms": 36.27116605639458, "input_usage": {"lookups": 28}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "no", "evidence_artifact": {"ledger": {"facts_total": 58, "facts_retained": 32, "retention_rate": 0.5517241379310345, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.7142857142857143, "amount": 0.6666666666666666, "status": 0.6666666666666666, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1106}, "ledger_by_hop": {"1": {"facts_total": 58, "facts_retained": 32, "retention_rate": 0.5517241379310345, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.7142857142857143, "amount": 0.6666666666666666, "status": 0.6666666666666666, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1200}, "2": {"facts_total": 58, "facts_retained": 32, "retention_rate": 0.5517241379310345, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.7142857142857143, "amount": 0.6666666666666666, "status": 0.6666666666666666, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1121}, "3": {"facts_total": 58, "facts_retained": 32, "retention_rate": 0.5517241379310345, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.7142857142857143, "amount": 0.6666666666666666, "status": 0.6666666666666666, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1106}, "4": {"facts_total": 58, "facts_retained": 32, "retention_rate": 0.5517241379310345, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.7142857142857143, "amount": 0.6666666666666666, "status": 0.6666666666666666, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1106}, "5": {"facts_total": 58, "facts_retained": 32, "retention_rate": 0.5517241379310345, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.7142857142857143, "amount": 0.6666666666666666, "status": 0.6666666666666666, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1106}, "6": {"facts_total": 58, "facts_retained": 32, "retention_rate": 0.5517241379310345, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.7142857142857143, "amount": 0.6666666666666666, "status": 0.6666666666666666, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1106}, "7": {"facts_total": 58, "facts_retained": 32, "retention_rate": 0.5517241379310345, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.7142857142857143, "amount": 0.6666666666666666, "status": 0.6666666666666666, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1106}, "8": {"facts_total": 58, "facts_retained": 32, "retention_rate": 0.5517241379310345, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.7142857142857143, "amount": 0.6666666666666666, "status": 0.6666666666666666, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1106}}}, "reference_key_id": "keys/synthetic/syn_03003_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "fd539b5069a942d3034d38027c24271a19ab0d754e9dee3800caa6490c9681d4", "output_hash": "9390298f3fb0c5b160498935d79cb139aef28e1c47358b4bbba61862b9c26e59"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "a6dc7234091cbade3f35cc3a", "attempt_id": "6ec5e0ede4b3", "run_group_id": "E4_core:A4|repairable|qwen3b:h8", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T06:47:04.770427Z", "task_family": "synthetic", "document_id": "syn_03003_2000", "document_hash": "dddca1a38e3224870fcfae927c701c400fd885c19123ea840ee06a76a140f407", "question_id": "q08_revision", "split": "test", "cluster_id": "syn_03003_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 8, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1866, "query_tokens": 597, "message_tokens": 498, "message_utf8_bytes": 1106, "cache_tensor_bytes": null, "serialized_payload_bytes": 1106, "side_information_bytes": 0, "fetched_bytes": 5341, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 174613.7597510824, "hops": {"1": 23068.98970901966, "2": 21813.183957943693, "3": 21540.921541978605, "4": 21531.434917007573, "5": 21513.298875070177, "6": 21784.19441706501, "7": 21694.188333000056, "8": 21667.547999997623}, "receiver_prefill_ms": 341.85029193758965, "ttft_ms": 35.74129194021225, "decode_ms": 493.1545000290498}, "total_wall_ms": 175485.19862617832, "ttft_ms": 35.74129194021225, "input_usage": {"lookups": 28}, "output_usage": {"output_tokens": 7}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "911,925", "evidence_artifact": {"ledger": {"facts_total": 58, "facts_retained": 32, "retention_rate": 0.5517241379310345, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.7142857142857143, "amount": 0.6666666666666666, "status": 0.6666666666666666, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1106}, "ledger_by_hop": {"1": {"facts_total": 58, "facts_retained": 32, "retention_rate": 0.5517241379310345, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.7142857142857143, "amount": 0.6666666666666666, "status": 0.6666666666666666, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1200}, "2": {"facts_total": 58, "facts_retained": 32, "retention_rate": 0.5517241379310345, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.7142857142857143, "amount": 0.6666666666666666, "status": 0.6666666666666666, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1121}, "3": {"facts_total": 58, "facts_retained": 32, "retention_rate": 0.5517241379310345, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.7142857142857143, "amount": 0.6666666666666666, "status": 0.6666666666666666, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1106}, "4": {"facts_total": 58, "facts_retained": 32, "retention_rate": 0.5517241379310345, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.7142857142857143, "amount": 0.6666666666666666, "status": 0.6666666666666666, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1106}, "5": {"facts_total": 58, "facts_retained": 32, "retention_rate": 0.5517241379310345, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.7142857142857143, "amount": 0.6666666666666666, "status": 0.6666666666666666, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1106}, "6": {"facts_total": 58, "facts_retained": 32, "retention_rate": 0.5517241379310345, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.7142857142857143, "amount": 0.6666666666666666, "status": 0.6666666666666666, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1106}, "7": {"facts_total": 58, "facts_retained": 32, "retention_rate": 0.5517241379310345, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.7142857142857143, "amount": 0.6666666666666666, "status": 0.6666666666666666, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1106}, "8": {"facts_total": 58, "facts_retained": 32, "retention_rate": 0.5517241379310345, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.7142857142857143, "amount": 0.6666666666666666, "status": 0.6666666666666666, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1106}}}, "reference_key_id": "keys/synthetic/syn_03003_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "fd539b5069a942d3034d38027c24271a19ab0d754e9dee3800caa6490c9681d4", "output_hash": "86be78720fc4b79f93e603e7d9e4861fcd25102612f73209fffc17d787056007"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "ace8e5fb38fc20d6b64be54d", "attempt_id": "9d0f950fa01a", "run_group_id": "E4_core:A4|repairable|qwen3b:h8", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T06:47:05.659271Z", "task_family": "synthetic", "document_id": "syn_03003_2000", "document_hash": "dddca1a38e3224870fcfae927c701c400fd885c19123ea840ee06a76a140f407", "question_id": "q09_revision", "split": "test", "cluster_id": "syn_03003_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 8, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1866, "query_tokens": 590, "message_tokens": 498, "message_utf8_bytes": 1106, "cache_tensor_bytes": null, "serialized_payload_bytes": 1106, "side_information_bytes": 0, "fetched_bytes": 5341, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 174613.7597510824, "hops": {"1": 23068.98970901966, "2": 21813.183957943693, "3": 21540.921541978605, "4": 21531.434917007573, "5": 21513.298875070177, "6": 21784.19441706501, "7": 21694.188333000056, "8": 21667.547999997623}, "receiver_prefill_ms": 379.813416977413, "ttft_ms": 38.12433406710625, "decode_ms": 189.4686659798026}, "total_wall_ms": 175222.00166713446, "ttft_ms": 38.12433406710625, "input_usage": {"lookups": 28}, "output_usage": {"output_tokens": 6}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "911925", "evidence_artifact": {"ledger": {"facts_total": 58, "facts_retained": 32, "retention_rate": 0.5517241379310345, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.7142857142857143, "amount": 0.6666666666666666, "status": 0.6666666666666666, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1106}, "ledger_by_hop": {"1": {"facts_total": 58, "facts_retained": 32, "retention_rate": 0.5517241379310345, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.7142857142857143, "amount": 0.6666666666666666, "status": 0.6666666666666666, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1200}, "2": {"facts_total": 58, "facts_retained": 32, "retention_rate": 0.5517241379310345, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.7142857142857143, "amount": 0.6666666666666666, "status": 0.6666666666666666, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1121}, "3": {"facts_total": 58, "facts_retained": 32, "retention_rate": 0.5517241379310345, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.7142857142857143, "amount": 0.6666666666666666, "status": 0.6666666666666666, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1106}, "4": {"facts_total": 58, "facts_retained": 32, "retention_rate": 0.5517241379310345, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.7142857142857143, "amount": 0.6666666666666666, "status": 0.6666666666666666, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1106}, "5": {"facts_total": 58, "facts_retained": 32, "retention_rate": 0.5517241379310345, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.7142857142857143, "amount": 0.6666666666666666, "status": 0.6666666666666666, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1106}, "6": {"facts_total": 58, "facts_retained": 32, "retention_rate": 0.5517241379310345, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.7142857142857143, "amount": 0.6666666666666666, "status": 0.6666666666666666, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1106}, "7": {"facts_total": 58, "facts_retained": 32, "retention_rate": 0.5517241379310345, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.7142857142857143, "amount": 0.6666666666666666, "status": 0.6666666666666666, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1106}, "8": {"facts_total": 58, "facts_retained": 32, "retention_rate": 0.5517241379310345, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.7142857142857143, "amount": 0.6666666666666666, "status": 0.6666666666666666, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1106}}}, "reference_key_id": "keys/synthetic/syn_03003_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "fd539b5069a942d3034d38027c24271a19ab0d754e9dee3800caa6490c9681d4", "output_hash": "679a540dc4739f0ace623448b8fd925fb234f554838d7e87da223228ee81b112"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "9cfbbcb96cecdf68216be890", "attempt_id": "c9c94a47b4f1", "run_group_id": "E4_core:A4|repairable|qwen3b:h8", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T06:47:06.285962Z", "task_family": "synthetic", "document_id": "syn_03003_2000", "document_hash": "dddca1a38e3224870fcfae927c701c400fd885c19123ea840ee06a76a140f407", "question_id": "q10_temporal", "split": "test", "cluster_id": "syn_03003_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 8, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1866, "query_tokens": 597, "message_tokens": 498, "message_utf8_bytes": 1106, "cache_tensor_bytes": null, "serialized_payload_bytes": 1106, "side_information_bytes": 0, "fetched_bytes": 5341, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 174613.7597510824, "hops": {"1": 23068.98970901966, "2": 21813.183957943693, "3": 21540.921541978605, "4": 21531.434917007573, "5": 21513.298875070177, "6": 21784.19441706501, "7": 21694.188333000056, "8": 21667.547999997623}, "receiver_prefill_ms": 357.0720830466598, "ttft_ms": 49.52124995179474, "decode_ms": 467.58116700220853}, "total_wall_ms": 175488.70533506852, "ttft_ms": 49.52124995179474, "input_usage": {"lookups": 28}, "output_usage": {"output_tokens": 6}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "Gamtiszen-2", "evidence_artifact": {"ledger": {"facts_total": 58, "facts_retained": 32, "retention_rate": 0.5517241379310345, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.7142857142857143, "amount": 0.6666666666666666, "status": 0.6666666666666666, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1106}, "ledger_by_hop": {"1": {"facts_total": 58, "facts_retained": 32, "retention_rate": 0.5517241379310345, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.7142857142857143, "amount": 0.6666666666666666, "status": 0.6666666666666666, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1200}, "2": {"facts_total": 58, "facts_retained": 32, "retention_rate": 0.5517241379310345, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.7142857142857143, "amount": 0.6666666666666666, "status": 0.6666666666666666, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1121}, "3": {"facts_total": 58, "facts_retained": 32, "retention_rate": 0.5517241379310345, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.7142857142857143, "amount": 0.6666666666666666, "status": 0.6666666666666666, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1106}, "4": {"facts_total": 58, "facts_retained": 32, "retention_rate": 0.5517241379310345, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.7142857142857143, "amount": 0.6666666666666666, "status": 0.6666666666666666, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1106}, "5": {"facts_total": 58, "facts_retained": 32, "retention_rate": 0.5517241379310345, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.7142857142857143, "amount": 0.6666666666666666, "status": 0.6666666666666666, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1106}, "6": {"facts_total": 58, "facts_retained": 32, "retention_rate": 0.5517241379310345, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.7142857142857143, "amount": 0.6666666666666666, "status": 0.6666666666666666, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1106}, "7": {"facts_total": 58, "facts_retained": 32, "retention_rate": 0.5517241379310345, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.7142857142857143, "amount": 0.6666666666666666, "status": 0.6666666666666666, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1106}, "8": {"facts_total": 58, "facts_retained": 32, "retention_rate": 0.5517241379310345, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.7142857142857143, "amount": 0.6666666666666666, "status": 0.6666666666666666, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1106}}}, "reference_key_id": "keys/synthetic/syn_03003_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "fd539b5069a942d3034d38027c24271a19ab0d754e9dee3800caa6490c9681d4", "output_hash": "32e3af428886434823d82102e1c482765eafdf6077fe8d3fba5752ec5e29826c"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "a5124725de158e9d25e5e70b", "attempt_id": "a6e10c97cfa2", "run_group_id": "E4_core:A4|repairable|qwen3b:h8", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T06:47:07.179424Z", "task_family": "synthetic", "document_id": "syn_03003_2000", "document_hash": "dddca1a38e3224870fcfae927c701c400fd885c19123ea840ee06a76a140f407", "question_id": "q11_aggregation", "split": "test", "cluster_id": "syn_03003_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 8, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1866, "query_tokens": 591, "message_tokens": 498, "message_utf8_bytes": 1106, "cache_tensor_bytes": null, "serialized_payload_bytes": 1106, "side_information_bytes": 0, "fetched_bytes": 5341, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 174613.7597510824, "hops": {"1": 23068.98970901966, "2": 21813.183957943693, "3": 21540.921541978605, "4": 21531.434917007573, "5": 21513.298875070177, "6": 21784.19441706501, "7": 21694.188333000056, "8": 21667.547999997623}, "receiver_prefill_ms": 402.2698749322444, "ttft_ms": 38.19529199972749, "decode_ms": 338.1054999772459}, "total_wall_ms": 175393.2310420787, "ttft_ms": 38.19529199972749, "input_usage": {"lookups": 28}, "output_usage": {"output_tokens": 10}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "33,506,345", "evidence_artifact": {"ledger": {"facts_total": 58, "facts_retained": 32, "retention_rate": 0.5517241379310345, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.7142857142857143, "amount": 0.6666666666666666, "status": 0.6666666666666666, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1106}, "ledger_by_hop": {"1": {"facts_total": 58, "facts_retained": 32, "retention_rate": 0.5517241379310345, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.7142857142857143, "amount": 0.6666666666666666, "status": 0.6666666666666666, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1200}, "2": {"facts_total": 58, "facts_retained": 32, "retention_rate": 0.5517241379310345, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.7142857142857143, "amount": 0.6666666666666666, "status": 0.6666666666666666, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1121}, "3": {"facts_total": 58, "facts_retained": 32, "retention_rate": 0.5517241379310345, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.7142857142857143, "amount": 0.6666666666666666, "status": 0.6666666666666666, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1106}, "4": {"facts_total": 58, "facts_retained": 32, "retention_rate": 0.5517241379310345, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.7142857142857143, "amount": 0.6666666666666666, "status": 0.6666666666666666, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1106}, "5": {"facts_total": 58, "facts_retained": 32, "retention_rate": 0.5517241379310345, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.7142857142857143, "amount": 0.6666666666666666, "status": 0.6666666666666666, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1106}, "6": {"facts_total": 58, "facts_retained": 32, "retention_rate": 0.5517241379310345, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.7142857142857143, "amount": 0.6666666666666666, "status": 0.6666666666666666, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1106}, "7": {"facts_total": 58, "facts_retained": 32, "retention_rate": 0.5517241379310345, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.7142857142857143, "amount": 0.6666666666666666, "status": 0.6666666666666666, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1106}, "8": {"facts_total": 58, "facts_retained": 32, "retention_rate": 0.5517241379310345, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.7142857142857143, "amount": 0.6666666666666666, "status": 0.6666666666666666, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1106}}}, "reference_key_id": "keys/synthetic/syn_03003_2000.json", "task_scores": {"correct": 0.0, "abstained": 0.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "fd539b5069a942d3034d38027c24271a19ab0d754e9dee3800caa6490c9681d4", "output_hash": "de275001d02d943760a37daf3a991b5fdb2fa479267a2e7f5723f623b23797db"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "e94971cb960be6104065cf90", "attempt_id": "742579042058", "run_group_id": "E4_core:A4|repairable|qwen3b:h8", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T06:47:07.976618Z", "task_family": "synthetic", "document_id": "syn_03003_2000", "document_hash": "dddca1a38e3224870fcfae927c701c400fd885c19123ea840ee06a76a140f407", "question_id": "q12_unit_conversion", "split": "test", "cluster_id": "syn_03003_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 8, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1866, "query_tokens": 588, "message_tokens": 498, "message_utf8_bytes": 1106, "cache_tensor_bytes": null, "serialized_payload_bytes": 1106, "side_information_bytes": 0, "fetched_bytes": 5341, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 174613.7597510824, "hops": {"1": 23068.98970901966, "2": 21813.183957943693, "3": 21540.921541978605, "4": 21531.434917007573, "5": 21513.298875070177, "6": 21784.19441706501, "7": 21694.188333000056, "8": 21667.547999997623}, "receiver_prefill_ms": 336.28250006586313, "ttft_ms": 49.261041916906834, "decode_ms": 397.4472909467295}, "total_wall_ms": 175397.4635010818, "ttft_ms": 49.261041916906834, "input_usage": {"lookups": 28}, "output_usage": {"output_tokens": 5}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "17430", "evidence_artifact": {"ledger": {"facts_total": 58, "facts_retained": 32, "retention_rate": 0.5517241379310345, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.7142857142857143, "amount": 0.6666666666666666, "status": 0.6666666666666666, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1106}, "ledger_by_hop": {"1": {"facts_total": 58, "facts_retained": 32, "retention_rate": 0.5517241379310345, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.7142857142857143, "amount": 0.6666666666666666, "status": 0.6666666666666666, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1200}, "2": {"facts_total": 58, "facts_retained": 32, "retention_rate": 0.5517241379310345, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.7142857142857143, "amount": 0.6666666666666666, "status": 0.6666666666666666, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1121}, "3": {"facts_total": 58, "facts_retained": 32, "retention_rate": 0.5517241379310345, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.7142857142857143, "amount": 0.6666666666666666, "status": 0.6666666666666666, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1106}, "4": {"facts_total": 58, "facts_retained": 32, "retention_rate": 0.5517241379310345, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.7142857142857143, "amount": 0.6666666666666666, "status": 0.6666666666666666, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1106}, "5": {"facts_total": 58, "facts_retained": 32, "retention_rate": 0.5517241379310345, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.7142857142857143, "amount": 0.6666666666666666, "status": 0.6666666666666666, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1106}, "6": {"facts_total": 58, "facts_retained": 32, "retention_rate": 0.5517241379310345, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.7142857142857143, "amount": 0.6666666666666666, "status": 0.6666666666666666, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1106}, "7": {"facts_total": 58, "facts_retained": 32, "retention_rate": 0.5517241379310345, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.7142857142857143, "amount": 0.6666666666666666, "status": 0.6666666666666666, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1106}, "8": {"facts_total": 58, "facts_retained": 32, "retention_rate": 0.5517241379310345, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.7142857142857143, "amount": 0.6666666666666666, "status": 0.6666666666666666, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1106}}}, "reference_key_id": "keys/synthetic/syn_03003_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "fd539b5069a942d3034d38027c24271a19ab0d754e9dee3800caa6490c9681d4", "output_hash": "2575eaf3b748efc209b676518766751eb038c88f74578951345818bfe7f7194b"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "31bc2fb1fd7809bae971acd4", "attempt_id": "9a6c920ba8f8", "run_group_id": "E4_core:A4|repairable|qwen3b:h8", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T06:47:08.777778Z", "task_family": "synthetic", "document_id": "syn_03003_2000", "document_hash": "dddca1a38e3224870fcfae927c701c400fd885c19123ea840ee06a76a140f407", "question_id": "q13_unanswerable", "split": "test", "cluster_id": "syn_03003_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 8, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1866, "query_tokens": 592, "message_tokens": 498, "message_utf8_bytes": 1106, "cache_tensor_bytes": null, "serialized_payload_bytes": 1106, "side_information_bytes": 0, "fetched_bytes": 5341, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 174613.7597510824, "hops": {"1": 23068.98970901966, "2": 21813.183957943693, "3": 21540.921541978605, "4": 21531.434917007573, "5": 21513.298875070177, "6": 21784.19441706501, "7": 21694.188333000056, "8": 21667.547999997623}, "receiver_prefill_ms": 389.10479203332216, "ttft_ms": 23.418833035975695, "decode_ms": 39.8714579641819}, "total_wall_ms": 175066.75220909528, "ttft_ms": 23.418833035975695, "input_usage": {"lookups": 28}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 58, "facts_retained": 32, "retention_rate": 0.5517241379310345, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.7142857142857143, "amount": 0.6666666666666666, "status": 0.6666666666666666, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1106}, "ledger_by_hop": {"1": {"facts_total": 58, "facts_retained": 32, "retention_rate": 0.5517241379310345, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.7142857142857143, "amount": 0.6666666666666666, "status": 0.6666666666666666, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1200}, "2": {"facts_total": 58, "facts_retained": 32, "retention_rate": 0.5517241379310345, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.7142857142857143, "amount": 0.6666666666666666, "status": 0.6666666666666666, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1121}, "3": {"facts_total": 58, "facts_retained": 32, "retention_rate": 0.5517241379310345, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.7142857142857143, "amount": 0.6666666666666666, "status": 0.6666666666666666, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1106}, "4": {"facts_total": 58, "facts_retained": 32, "retention_rate": 0.5517241379310345, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.7142857142857143, "amount": 0.6666666666666666, "status": 0.6666666666666666, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1106}, "5": {"facts_total": 58, "facts_retained": 32, "retention_rate": 0.5517241379310345, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.7142857142857143, "amount": 0.6666666666666666, "status": 0.6666666666666666, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1106}, "6": {"facts_total": 58, "facts_retained": 32, "retention_rate": 0.5517241379310345, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.7142857142857143, "amount": 0.6666666666666666, "status": 0.6666666666666666, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1106}, "7": {"facts_total": 58, "facts_retained": 32, "retention_rate": 0.5517241379310345, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.7142857142857143, "amount": 0.6666666666666666, "status": 0.6666666666666666, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1106}, "8": {"facts_total": 58, "facts_retained": 32, "retention_rate": 0.5517241379310345, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.7142857142857143, "amount": 0.6666666666666666, "status": 0.6666666666666666, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1106}}}, "reference_key_id": "keys/synthetic/syn_03003_2000.json", "task_scores": {"correct": 1.0, "abstained": 1.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "fd539b5069a942d3034d38027c24271a19ab0d754e9dee3800caa6490c9681d4", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "f8cb1b76ff87eb79ca430a1a", "attempt_id": "051a0cb61d9b", "run_group_id": "E4_core:A4|repairable|qwen3b:h8", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T06:47:09.248074Z", "task_family": "synthetic", "document_id": "syn_03003_2000", "document_hash": "dddca1a38e3224870fcfae927c701c400fd885c19123ea840ee06a76a140f407", "question_id": "q14_unanswerable", "split": "test", "cluster_id": "syn_03003_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 8, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1866, "query_tokens": 593, "message_tokens": 498, "message_utf8_bytes": 1106, "cache_tensor_bytes": null, "serialized_payload_bytes": 1106, "side_information_bytes": 0, "fetched_bytes": 5341, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 174613.7597510824, "hops": {"1": 23068.98970901966, "2": 21813.183957943693, "3": 21540.921541978605, "4": 21531.434917007573, "5": 21513.298875070177, "6": 21784.19441706501, "7": 21694.188333000056, "8": 21667.547999997623}, "receiver_prefill_ms": 386.00470894016325, "ttft_ms": 36.39099991414696, "decode_ms": 22.767291986383498}, "total_wall_ms": 175059.50479302555, "ttft_ms": 36.39099991414696, "input_usage": {"lookups": 28}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 58, "facts_retained": 32, "retention_rate": 0.5517241379310345, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.7142857142857143, "amount": 0.6666666666666666, "status": 0.6666666666666666, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1106}, "ledger_by_hop": {"1": {"facts_total": 58, "facts_retained": 32, "retention_rate": 0.5517241379310345, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.7142857142857143, "amount": 0.6666666666666666, "status": 0.6666666666666666, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1200}, "2": {"facts_total": 58, "facts_retained": 32, "retention_rate": 0.5517241379310345, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.7142857142857143, "amount": 0.6666666666666666, "status": 0.6666666666666666, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1121}, "3": {"facts_total": 58, "facts_retained": 32, "retention_rate": 0.5517241379310345, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.7142857142857143, "amount": 0.6666666666666666, "status": 0.6666666666666666, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1106}, "4": {"facts_total": 58, "facts_retained": 32, "retention_rate": 0.5517241379310345, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.7142857142857143, "amount": 0.6666666666666666, "status": 0.6666666666666666, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1106}, "5": {"facts_total": 58, "facts_retained": 32, "retention_rate": 0.5517241379310345, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.7142857142857143, "amount": 0.6666666666666666, "status": 0.6666666666666666, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1106}, "6": {"facts_total": 58, "facts_retained": 32, "retention_rate": 0.5517241379310345, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.7142857142857143, "amount": 0.6666666666666666, "status": 0.6666666666666666, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1106}, "7": {"facts_total": 58, "facts_retained": 32, "retention_rate": 0.5517241379310345, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.7142857142857143, "amount": 0.6666666666666666, "status": 0.6666666666666666, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1106}, "8": {"facts_total": 58, "facts_retained": 32, "retention_rate": 0.5517241379310345, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.7142857142857143, "amount": 0.6666666666666666, "status": 0.6666666666666666, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1106}}}, "reference_key_id": "keys/synthetic/syn_03003_2000.json", "task_scores": {"correct": 1.0, "abstained": 1.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "fd539b5069a942d3034d38027c24271a19ab0d754e9dee3800caa6490c9681d4", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "4be6487714f4a9176cab4595", "attempt_id": "9ed196e82a1b", "run_group_id": "E4_core:A5|repairable|qwen3b:h1", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T06:47:32.346987Z", "task_family": "synthetic", "document_id": "syn_03003_2000", "document_hash": "dddca1a38e3224870fcfae927c701c400fd885c19123ea840ee06a76a140f407", "question_id": "q00_single_fact", "split": "test", "cluster_id": "syn_03003_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 1, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1866, "query_tokens": 435, "message_tokens": 352, "message_utf8_bytes": 830, "cache_tensor_bytes": null, "serialized_payload_bytes": 830, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 22633.649082970805, "hops": {"1": 22633.649082970805}, "receiver_prefill_ms": 424.58445800002664, "ttft_ms": 39.756958023644984, "decode_ms": 357.823041966185}, "total_wall_ms": 23456.35616593063, "ttft_ms": 39.756958023644984, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 10}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "GR-79S-5839", "evidence_artifact": {"ledger": {"facts_total": 58, "facts_retained": 15, "retention_rate": 0.25862068965517243, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.2222222222222222, "status": 0.75, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 766}, "ledger_by_hop": {"1": {"facts_total": 58, "facts_retained": 15, "retention_rate": 0.25862068965517243, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.2222222222222222, "status": 0.75, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 766}}}, "reference_key_id": "keys/synthetic/syn_03003_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "a613cf19d60b1dbe7d9f51bbbf98459d9f181f51f14c39d755912e891ae7fd2f", "output_hash": "720cc40cd69e85e9728f6bea00b17a95ca07eee643e143a131840763958303cd"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "eb7da206c15c951a6119dbbc", "attempt_id": "cebc9cec739c", "run_group_id": "E4_core:A5|repairable|qwen3b:h1", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T06:47:33.187973Z", "task_family": "synthetic", "document_id": "syn_03003_2000", "document_hash": "dddca1a38e3224870fcfae927c701c400fd885c19123ea840ee06a76a140f407", "question_id": "q01_single_fact", "split": "test", "cluster_id": "syn_03003_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 1, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1866, "query_tokens": 439, "message_tokens": 352, "message_utf8_bytes": 830, "cache_tensor_bytes": null, "serialized_payload_bytes": 830, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 22633.649082970805, "hops": {"1": 22633.649082970805}, "receiver_prefill_ms": 278.24179094750434, "ttft_ms": 20.49791591707617, "decode_ms": 19.644082989543676}, "total_wall_ms": 22952.569373999722, "ttft_ms": 20.49791591707617, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 58, "facts_retained": 15, "retention_rate": 0.25862068965517243, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.2222222222222222, "status": 0.75, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 766}, "ledger_by_hop": {"1": {"facts_total": 58, "facts_retained": 15, "retention_rate": 0.25862068965517243, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.2222222222222222, "status": 0.75, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 766}}}, "reference_key_id": "keys/synthetic/syn_03003_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "a613cf19d60b1dbe7d9f51bbbf98459d9f181f51f14c39d755912e891ae7fd2f", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "c03593e32e6b51e6d5e4d259", "attempt_id": "880550d2b041", "run_group_id": "E4_core:A5|repairable|qwen3b:h1", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T06:47:33.524978Z", "task_family": "synthetic", "document_id": "syn_03003_2000", "document_hash": "dddca1a38e3224870fcfae927c701c400fd885c19123ea840ee06a76a140f407", "question_id": "q02_exact_copy", "split": "test", "cluster_id": "syn_03003_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 1, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1866, "query_tokens": 439, "message_tokens": 352, "message_utf8_bytes": 830, "cache_tensor_bytes": null, "serialized_payload_bytes": 830, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 22633.649082970805, "hops": {"1": 22633.649082970805}, "receiver_prefill_ms": 293.4932910138741, "ttft_ms": 43.9436249434948, "decode_ms": 74.21941705979407}, "total_wall_ms": 23045.85141595453, "ttft_ms": 43.9436249434948, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 58, "facts_retained": 15, "retention_rate": 0.25862068965517243, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.2222222222222222, "status": 0.75, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 766}, "ledger_by_hop": {"1": {"facts_total": 58, "facts_retained": 15, "retention_rate": 0.25862068965517243, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.2222222222222222, "status": 0.75, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 766}}}, "reference_key_id": "keys/synthetic/syn_03003_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "a613cf19d60b1dbe7d9f51bbbf98459d9f181f51f14c39d755912e891ae7fd2f", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "3f808b1fd278ff14cbcff420", "attempt_id": "c8783b09703d", "run_group_id": "E4_core:A5|repairable|qwen3b:h1", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T06:47:33.956493Z", "task_family": "synthetic", "document_id": "syn_03003_2000", "document_hash": "dddca1a38e3224870fcfae927c701c400fd885c19123ea840ee06a76a140f407", "question_id": "q03_exact_copy", "split": "test", "cluster_id": "syn_03003_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 1, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1866, "query_tokens": 434, "message_tokens": 352, "message_utf8_bytes": 830, "cache_tensor_bytes": null, "serialized_payload_bytes": 830, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 22633.649082970805, "hops": {"1": 22633.649082970805}, "receiver_prefill_ms": 415.3790419222787, "ttft_ms": 37.823042017407715, "decode_ms": 31.094291945919394}, "total_wall_ms": 23118.5708749108, "ttft_ms": 37.823042017407715, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 58, "facts_retained": 15, "retention_rate": 0.25862068965517243, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.2222222222222222, "status": 0.75, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 766}, "ledger_by_hop": {"1": {"facts_total": 58, "facts_retained": 15, "retention_rate": 0.25862068965517243, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.2222222222222222, "status": 0.75, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 766}}}, "reference_key_id": "keys/synthetic/syn_03003_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "a613cf19d60b1dbe7d9f51bbbf98459d9f181f51f14c39d755912e891ae7fd2f", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "60037fc768416241e1f99889", "attempt_id": "22cca1aab28a", "run_group_id": "E4_core:A5|repairable|qwen3b:h1", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T06:47:34.460328Z", "task_family": "synthetic", "document_id": "syn_03003_2000", "document_hash": "dddca1a38e3224870fcfae927c701c400fd885c19123ea840ee06a76a140f407", "question_id": "q04_multi_hop", "split": "test", "cluster_id": "syn_03003_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 1, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1866, "query_tokens": 443, "message_tokens": 352, "message_utf8_bytes": 830, "cache_tensor_bytes": null, "serialized_payload_bytes": 830, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 22633.649082970805, "hops": {"1": 22633.649082970805}, "receiver_prefill_ms": 285.48466600477695, "ttft_ms": 26.879459037445486, "decode_ms": 41.56845808029175}, "total_wall_ms": 22988.102125003934, "ttft_ms": 26.879459037445486, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 58, "facts_retained": 15, "retention_rate": 0.25862068965517243, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.2222222222222222, "status": 0.75, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 766}, "ledger_by_hop": {"1": {"facts_total": 58, "facts_retained": 15, "retention_rate": 0.25862068965517243, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.2222222222222222, "status": 0.75, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 766}}}, "reference_key_id": "keys/synthetic/syn_03003_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "a613cf19d60b1dbe7d9f51bbbf98459d9f181f51f14c39d755912e891ae7fd2f", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "5c27b7135450077560da6e36", "attempt_id": "13091560e3e2", "run_group_id": "E4_core:A5|repairable|qwen3b:h1", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T06:47:34.831306Z", "task_family": "synthetic", "document_id": "syn_03003_2000", "document_hash": "dddca1a38e3224870fcfae927c701c400fd885c19123ea840ee06a76a140f407", "question_id": "q05_multi_hop", "split": "test", "cluster_id": "syn_03003_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 1, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1866, "query_tokens": 445, "message_tokens": 352, "message_utf8_bytes": 830, "cache_tensor_bytes": null, "serialized_payload_bytes": 830, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 22633.649082970805, "hops": {"1": 22633.649082970805}, "receiver_prefill_ms": 294.73774996586144, "ttft_ms": 24.634917033836246, "decode_ms": 21.957707940600812}, "total_wall_ms": 22975.467665935867, "ttft_ms": 24.634917033836246, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 58, "facts_retained": 15, "retention_rate": 0.25862068965517243, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.2222222222222222, "status": 0.75, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 766}, "ledger_by_hop": {"1": {"facts_total": 58, "facts_retained": 15, "retention_rate": 0.25862068965517243, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.2222222222222222, "status": 0.75, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 766}}}, "reference_key_id": "keys/synthetic/syn_03003_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "a613cf19d60b1dbe7d9f51bbbf98459d9f181f51f14c39d755912e891ae7fd2f", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "fd60ade46e45ed51653d0c5d", "attempt_id": "c53536532d6e", "run_group_id": "E4_core:A5|repairable|qwen3b:h1", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T06:47:35.189841Z", "task_family": "synthetic", "document_id": "syn_03003_2000", "document_hash": "dddca1a38e3224870fcfae927c701c400fd885c19123ea840ee06a76a140f407", "question_id": "q06_negation", "split": "test", "cluster_id": "syn_03003_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 1, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1866, "query_tokens": 437, "message_tokens": 352, "message_utf8_bytes": 830, "cache_tensor_bytes": null, "serialized_payload_bytes": 830, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 22633.649082970805, "hops": {"1": 22633.649082970805}, "receiver_prefill_ms": 265.5492079211399, "ttft_ms": 56.80612497963011, "decode_ms": 77.42562494240701}, "total_wall_ms": 23034.0070000384, "ttft_ms": 56.80612497963011, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "no", "evidence_artifact": {"ledger": {"facts_total": 58, "facts_retained": 15, "retention_rate": 0.25862068965517243, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.2222222222222222, "status": 0.75, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 766}, "ledger_by_hop": {"1": {"facts_total": 58, "facts_retained": 15, "retention_rate": 0.25862068965517243, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.2222222222222222, "status": 0.75, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 766}}}, "reference_key_id": "keys/synthetic/syn_03003_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "a613cf19d60b1dbe7d9f51bbbf98459d9f181f51f14c39d755912e891ae7fd2f", "output_hash": "9390298f3fb0c5b160498935d79cb139aef28e1c47358b4bbba61862b9c26e59"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "e459144c5dcbfdc370de6772", "attempt_id": "f537418c497d", "run_group_id": "E4_core:A5|repairable|qwen3b:h1", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T06:47:35.607144Z", "task_family": "synthetic", "document_id": "syn_03003_2000", "document_hash": "dddca1a38e3224870fcfae927c701c400fd885c19123ea840ee06a76a140f407", "question_id": "q07_negation", "split": "test", "cluster_id": "syn_03003_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 1, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1866, "query_tokens": 434, "message_tokens": 352, "message_utf8_bytes": 830, "cache_tensor_bytes": null, "serialized_payload_bytes": 830, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 22633.649082970805, "hops": {"1": 22633.649082970805}, "receiver_prefill_ms": 438.9530000044033, "ttft_ms": 42.42220800369978, "decode_ms": 25.563833070918918}, "total_wall_ms": 23141.144957975484, "ttft_ms": 42.42220800369978, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 58, "facts_retained": 15, "retention_rate": 0.25862068965517243, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.2222222222222222, "status": 0.75, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 766}, "ledger_by_hop": {"1": {"facts_total": 58, "facts_retained": 15, "retention_rate": 0.25862068965517243, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.2222222222222222, "status": 0.75, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 766}}}, "reference_key_id": "keys/synthetic/syn_03003_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "a613cf19d60b1dbe7d9f51bbbf98459d9f181f51f14c39d755912e891ae7fd2f", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "84af0f3dcc9ccc237e42b6cf", "attempt_id": "d2ee0cb5ab6e", "run_group_id": "E4_core:A5|repairable|qwen3b:h1", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T06:47:36.133216Z", "task_family": "synthetic", "document_id": "syn_03003_2000", "document_hash": "dddca1a38e3224870fcfae927c701c400fd885c19123ea840ee06a76a140f407", "question_id": "q08_revision", "split": "test", "cluster_id": "syn_03003_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 1, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1866, "query_tokens": 452, "message_tokens": 352, "message_utf8_bytes": 830, "cache_tensor_bytes": null, "serialized_payload_bytes": 830, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 22633.649082970805, "hops": {"1": 22633.649082970805}, "receiver_prefill_ms": 352.0471249939874, "ttft_ms": 23.616624996066093, "decode_ms": 208.3935410482809}, "total_wall_ms": 23218.37941603735, "ttft_ms": 23.616624996066093, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 7}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "911,925", "evidence_artifact": {"ledger": {"facts_total": 58, "facts_retained": 15, "retention_rate": 0.25862068965517243, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.2222222222222222, "status": 0.75, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 766}, "ledger_by_hop": {"1": {"facts_total": 58, "facts_retained": 15, "retention_rate": 0.25862068965517243, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.2222222222222222, "status": 0.75, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 766}}}, "reference_key_id": "keys/synthetic/syn_03003_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "a613cf19d60b1dbe7d9f51bbbf98459d9f181f51f14c39d755912e891ae7fd2f", "output_hash": "86be78720fc4b79f93e603e7d9e4861fcd25102612f73209fffc17d787056007"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "3ab157f08ad79e952c994263", "attempt_id": "5684e28b071c", "run_group_id": "E4_core:A5|repairable|qwen3b:h1", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T06:47:36.735014Z", "task_family": "synthetic", "document_id": "syn_03003_2000", "document_hash": "dddca1a38e3224870fcfae927c701c400fd885c19123ea840ee06a76a140f407", "question_id": "q09_revision", "split": "test", "cluster_id": "syn_03003_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 1, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1866, "query_tokens": 445, "message_tokens": 352, "message_utf8_bytes": 830, "cache_tensor_bytes": null, "serialized_payload_bytes": 830, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 22633.649082970805, "hops": {"1": 22633.649082970805}, "receiver_prefill_ms": 250.73412503115833, "ttft_ms": 45.355541980825365, "decode_ms": 437.3211250640452}, "total_wall_ms": 23367.759248940274, "ttft_ms": 45.355541980825365, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 6}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "911925", "evidence_artifact": {"ledger": {"facts_total": 58, "facts_retained": 15, "retention_rate": 0.25862068965517243, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.2222222222222222, "status": 0.75, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 766}, "ledger_by_hop": {"1": {"facts_total": 58, "facts_retained": 15, "retention_rate": 0.25862068965517243, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.2222222222222222, "status": 0.75, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 766}}}, "reference_key_id": "keys/synthetic/syn_03003_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "a613cf19d60b1dbe7d9f51bbbf98459d9f181f51f14c39d755912e891ae7fd2f", "output_hash": "679a540dc4739f0ace623448b8fd925fb234f554838d7e87da223228ee81b112"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "9e0743ccd47859e098eb6ab9", "attempt_id": "fb8f8909fee6", "run_group_id": "E4_core:A5|repairable|qwen3b:h1", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T06:47:37.486386Z", "task_family": "synthetic", "document_id": "syn_03003_2000", "document_hash": "dddca1a38e3224870fcfae927c701c400fd885c19123ea840ee06a76a140f407", "question_id": "q10_temporal", "split": "test", "cluster_id": "syn_03003_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 1, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1866, "query_tokens": 452, "message_tokens": 352, "message_utf8_bytes": 830, "cache_tensor_bytes": null, "serialized_payload_bytes": 830, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 22633.649082970805, "hops": {"1": 22633.649082970805}, "receiver_prefill_ms": 317.19479197636247, "ttft_ms": 37.420208915136755, "decode_ms": 43.729832977987826}, "total_wall_ms": 23032.543958048336, "ttft_ms": 37.420208915136755, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 58, "facts_retained": 15, "retention_rate": 0.25862068965517243, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.2222222222222222, "status": 0.75, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 766}, "ledger_by_hop": {"1": {"facts_total": 58, "facts_retained": 15, "retention_rate": 0.25862068965517243, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.2222222222222222, "status": 0.75, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 766}}}, "reference_key_id": "keys/synthetic/syn_03003_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "a613cf19d60b1dbe7d9f51bbbf98459d9f181f51f14c39d755912e891ae7fd2f", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "0189bff5620eb04d2fe547c4", "attempt_id": "a8f3eb041e47", "run_group_id": "E4_core:A5|repairable|qwen3b:h1", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T06:47:37.903564Z", "task_family": "synthetic", "document_id": "syn_03003_2000", "document_hash": "dddca1a38e3224870fcfae927c701c400fd885c19123ea840ee06a76a140f407", "question_id": "q11_aggregation", "split": "test", "cluster_id": "syn_03003_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 1, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1866, "query_tokens": 446, "message_tokens": 352, "message_utf8_bytes": 830, "cache_tensor_bytes": null, "serialized_payload_bytes": 830, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 22633.649082970805, "hops": {"1": 22633.649082970805}, "receiver_prefill_ms": 314.3257499905303, "ttft_ms": 27.27495797444135, "decode_ms": 34.970249980688095}, "total_wall_ms": 23010.667792055756, "ttft_ms": 27.27495797444135, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 58, "facts_retained": 15, "retention_rate": 0.25862068965517243, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.2222222222222222, "status": 0.75, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 766}, "ledger_by_hop": {"1": {"facts_total": 58, "facts_retained": 15, "retention_rate": 0.25862068965517243, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.2222222222222222, "status": 0.75, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 766}}}, "reference_key_id": "keys/synthetic/syn_03003_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "a613cf19d60b1dbe7d9f51bbbf98459d9f181f51f14c39d755912e891ae7fd2f", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "9f540ef87a4ca9b7f5e3d7a7", "attempt_id": "353ee3172fab", "run_group_id": "E4_core:A5|repairable|qwen3b:h1", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T06:47:38.297970Z", "task_family": "synthetic", "document_id": "syn_03003_2000", "document_hash": "dddca1a38e3224870fcfae927c701c400fd885c19123ea840ee06a76a140f407", "question_id": "q12_unit_conversion", "split": "test", "cluster_id": "syn_03003_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 1, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1866, "query_tokens": 443, "message_tokens": 352, "message_utf8_bytes": 830, "cache_tensor_bytes": null, "serialized_payload_bytes": 830, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 22633.649082970805, "hops": {"1": 22633.649082970805}, "receiver_prefill_ms": 269.2364579997957, "ttft_ms": 56.62100005429238, "decode_ms": 413.8004590058699}, "total_wall_ms": 23373.9547500154, "ttft_ms": 56.62100005429238, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 5}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "17430", "evidence_artifact": {"ledger": {"facts_total": 58, "facts_retained": 15, "retention_rate": 0.25862068965517243, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.2222222222222222, "status": 0.75, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 766}, "ledger_by_hop": {"1": {"facts_total": 58, "facts_retained": 15, "retention_rate": 0.25862068965517243, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.2222222222222222, "status": 0.75, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 766}}}, "reference_key_id": "keys/synthetic/syn_03003_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "a613cf19d60b1dbe7d9f51bbbf98459d9f181f51f14c39d755912e891ae7fd2f", "output_hash": "2575eaf3b748efc209b676518766751eb038c88f74578951345818bfe7f7194b"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "07673c28e1b87df7ad630ebf", "attempt_id": "46c4a7dc7fa9", "run_group_id": "E4_core:A5|repairable|qwen3b:h1", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T06:47:39.058523Z", "task_family": "synthetic", "document_id": "syn_03003_2000", "document_hash": "dddca1a38e3224870fcfae927c701c400fd885c19123ea840ee06a76a140f407", "question_id": "q13_unanswerable", "split": "test", "cluster_id": "syn_03003_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 1, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1866, "query_tokens": 447, "message_tokens": 352, "message_utf8_bytes": 830, "cache_tensor_bytes": null, "serialized_payload_bytes": 830, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 22633.649082970805, "hops": {"1": 22633.649082970805}, "receiver_prefill_ms": 294.04962493572384, "ttft_ms": 39.58170802798122, "decode_ms": 26.20379102881998}, "total_wall_ms": 22994.082123972476, "ttft_ms": 39.58170802798122, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 58, "facts_retained": 15, "retention_rate": 0.25862068965517243, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.2222222222222222, "status": 0.75, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 766}, "ledger_by_hop": {"1": {"facts_total": 58, "facts_retained": 15, "retention_rate": 0.25862068965517243, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.2222222222222222, "status": 0.75, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 766}}}, "reference_key_id": "keys/synthetic/syn_03003_2000.json", "task_scores": {"correct": 1.0, "abstained": 1.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "a613cf19d60b1dbe7d9f51bbbf98459d9f181f51f14c39d755912e891ae7fd2f", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "5e396bee81bd8b6aa3c24aee", "attempt_id": "be0e5030f4a9", "run_group_id": "E4_core:A5|repairable|qwen3b:h1", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T06:47:39.436661Z", "task_family": "synthetic", "document_id": "syn_03003_2000", "document_hash": "dddca1a38e3224870fcfae927c701c400fd885c19123ea840ee06a76a140f407", "question_id": "q14_unanswerable", "split": "test", "cluster_id": "syn_03003_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 1, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1866, "query_tokens": 448, "message_tokens": 352, "message_utf8_bytes": 830, "cache_tensor_bytes": null, "serialized_payload_bytes": 830, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 22633.649082970805, "hops": {"1": 22633.649082970805}, "receiver_prefill_ms": 295.72958301287144, "ttft_ms": 34.6248330315575, "decode_ms": 35.45958292670548}, "total_wall_ms": 22999.951999052428, "ttft_ms": 34.6248330315575, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 58, "facts_retained": 15, "retention_rate": 0.25862068965517243, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.2222222222222222, "status": 0.75, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 766}, "ledger_by_hop": {"1": {"facts_total": 58, "facts_retained": 15, "retention_rate": 0.25862068965517243, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.2222222222222222, "status": 0.75, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 766}}}, "reference_key_id": "keys/synthetic/syn_03003_2000.json", "task_scores": {"correct": 1.0, "abstained": 1.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "a613cf19d60b1dbe7d9f51bbbf98459d9f181f51f14c39d755912e891ae7fd2f", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "923a0771e5a6f00f2ef88c54", "attempt_id": "0d9ebc4f8b43", "run_group_id": "E4_core:A5|repairable|qwen3b:h2", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T06:47:55.772933Z", "task_family": "synthetic", "document_id": "syn_03003_2000", "document_hash": "dddca1a38e3224870fcfae927c701c400fd885c19123ea840ee06a76a140f407", "question_id": "q00_single_fact", "split": "test", "cluster_id": "syn_03003_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 2, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1866, "query_tokens": 444, "message_tokens": 362, "message_utf8_bytes": 859, "cache_tensor_bytes": null, "serialized_payload_bytes": 859, "side_information_bytes": 0, "fetched_bytes": 508, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 38584.62487498764, "hops": {"1": 22633.649082970805, "2": 15950.975792016834}, "receiver_prefill_ms": 301.61475006025285, "ttft_ms": 40.88354192208499, "decode_ms": 293.60737500246614}, "total_wall_ms": 39221.250667003915, "ttft_ms": 40.88354192208499, "input_usage": {"lookups": 1}, "output_usage": {"output_tokens": 10}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "GR-79S-5839", "evidence_artifact": {"ledger": {"facts_total": 58, "facts_retained": 15, "retention_rate": 0.25862068965517243, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.2222222222222222, "status": 0.75, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 795}, "ledger_by_hop": {"1": {"facts_total": 58, "facts_retained": 15, "retention_rate": 0.25862068965517243, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.2222222222222222, "status": 0.75, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 766}, "2": {"facts_total": 58, "facts_retained": 15, "retention_rate": 0.25862068965517243, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.2222222222222222, "status": 0.75, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 795}}}, "reference_key_id": "keys/synthetic/syn_03003_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "0d4ead41380940db4359555678482b867e828b27a6b33c483fba125924e02f66", "output_hash": "720cc40cd69e85e9728f6bea00b17a95ca07eee643e143a131840763958303cd"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "12c3b093ba0d28764c60f8c7", "attempt_id": "884da4327b51", "run_group_id": "E4_core:A5|repairable|qwen3b:h2", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T06:47:56.426907Z", "task_family": "synthetic", "document_id": "syn_03003_2000", "document_hash": "dddca1a38e3224870fcfae927c701c400fd885c19123ea840ee06a76a140f407", "question_id": "q01_single_fact", "split": "test", "cluster_id": "syn_03003_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 2, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1866, "query_tokens": 448, "message_tokens": 362, "message_utf8_bytes": 859, "cache_tensor_bytes": null, "serialized_payload_bytes": 859, "side_information_bytes": 0, "fetched_bytes": 508, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 38584.62487498764, "hops": {"1": 22633.649082970805, "2": 15950.975792016834}, "receiver_prefill_ms": 346.44745802506804, "ttft_ms": 62.63208307791501, "decode_ms": 99.61141704116017}, "total_wall_ms": 39093.878041021526, "ttft_ms": 62.63208307791501, "input_usage": {"lookups": 1}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 58, "facts_retained": 15, "retention_rate": 0.25862068965517243, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.2222222222222222, "status": 0.75, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 795}, "ledger_by_hop": {"1": {"facts_total": 58, "facts_retained": 15, "retention_rate": 0.25862068965517243, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.2222222222222222, "status": 0.75, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 766}, "2": {"facts_total": 58, "facts_retained": 15, "retention_rate": 0.25862068965517243, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.2222222222222222, "status": 0.75, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 795}}}, "reference_key_id": "keys/synthetic/syn_03003_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "0d4ead41380940db4359555678482b867e828b27a6b33c483fba125924e02f66", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "8c72e709244df3622569adc9", "attempt_id": "e7de82418477", "run_group_id": "E4_core:A5|repairable|qwen3b:h2", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T06:47:56.956295Z", "task_family": "synthetic", "document_id": "syn_03003_2000", "document_hash": "dddca1a38e3224870fcfae927c701c400fd885c19123ea840ee06a76a140f407", "question_id": "q02_exact_copy", "split": "test", "cluster_id": "syn_03003_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 2, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1866, "query_tokens": 448, "message_tokens": 362, "message_utf8_bytes": 859, "cache_tensor_bytes": null, "serialized_payload_bytes": 859, "side_information_bytes": 0, "fetched_bytes": 508, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 38584.62487498764, "hops": {"1": 22633.649082970805, "2": 15950.975792016834}, "receiver_prefill_ms": 371.29599996842444, "ttft_ms": 41.674750042147934, "decode_ms": 23.103000014089048}, "total_wall_ms": 39021.15745795891, "ttft_ms": 41.674750042147934, "input_usage": {"lookups": 1}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 58, "facts_retained": 15, "retention_rate": 0.25862068965517243, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.2222222222222222, "status": 0.75, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 795}, "ledger_by_hop": {"1": {"facts_total": 58, "facts_retained": 15, "retention_rate": 0.25862068965517243, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.2222222222222222, "status": 0.75, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 766}, "2": {"facts_total": 58, "facts_retained": 15, "retention_rate": 0.25862068965517243, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.2222222222222222, "status": 0.75, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 795}}}, "reference_key_id": "keys/synthetic/syn_03003_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "0d4ead41380940db4359555678482b867e828b27a6b33c483fba125924e02f66", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "85f2ae1f1fc3ef0e5ab4f397", "attempt_id": "c6a30690e977", "run_group_id": "E4_core:A5|repairable|qwen3b:h2", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T06:47:57.412176Z", "task_family": "synthetic", "document_id": "syn_03003_2000", "document_hash": "dddca1a38e3224870fcfae927c701c400fd885c19123ea840ee06a76a140f407", "question_id": "q03_exact_copy", "split": "test", "cluster_id": "syn_03003_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 2, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1866, "query_tokens": 443, "message_tokens": 362, "message_utf8_bytes": 859, "cache_tensor_bytes": null, "serialized_payload_bytes": 859, "side_information_bytes": 0, "fetched_bytes": 508, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 38584.62487498764, "hops": {"1": 22633.649082970805, "2": 15950.975792016834}, "receiver_prefill_ms": 298.5652090283111, "ttft_ms": 35.52791697438806, "decode_ms": 28.995709028095007}, "total_wall_ms": 38948.27054196503, "ttft_ms": 35.52791697438806, "input_usage": {"lookups": 1}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 58, "facts_retained": 15, "retention_rate": 0.25862068965517243, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.2222222222222222, "status": 0.75, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 795}, "ledger_by_hop": {"1": {"facts_total": 58, "facts_retained": 15, "retention_rate": 0.25862068965517243, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.2222222222222222, "status": 0.75, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 766}, "2": {"facts_total": 58, "facts_retained": 15, "retention_rate": 0.25862068965517243, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.2222222222222222, "status": 0.75, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 795}}}, "reference_key_id": "keys/synthetic/syn_03003_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "0d4ead41380940db4359555678482b867e828b27a6b33c483fba125924e02f66", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "c67854bafd69a740b6ca3895", "attempt_id": "2e3be3445c1d", "run_group_id": "E4_core:A5|repairable|qwen3b:h2", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T06:47:57.792883Z", "task_family": "synthetic", "document_id": "syn_03003_2000", "document_hash": "dddca1a38e3224870fcfae927c701c400fd885c19123ea840ee06a76a140f407", "question_id": "q04_multi_hop", "split": "test", "cluster_id": "syn_03003_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 2, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1866, "query_tokens": 452, "message_tokens": 362, "message_utf8_bytes": 859, "cache_tensor_bytes": null, "serialized_payload_bytes": 859, "side_information_bytes": 0, "fetched_bytes": 508, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 38584.62487498764, "hops": {"1": 22633.649082970805, "2": 15950.975792016834}, "receiver_prefill_ms": 295.4499159241095, "ttft_ms": 58.70841699652374, "decode_ms": 49.66816701926291}, "total_wall_ms": 38989.00058399886, "ttft_ms": 58.70841699652374, "input_usage": {"lookups": 1}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 58, "facts_retained": 15, "retention_rate": 0.25862068965517243, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.2222222222222222, "status": 0.75, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 795}, "ledger_by_hop": {"1": {"facts_total": 58, "facts_retained": 15, "retention_rate": 0.25862068965517243, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.2222222222222222, "status": 0.75, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 766}, "2": {"facts_total": 58, "facts_retained": 15, "retention_rate": 0.25862068965517243, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.2222222222222222, "status": 0.75, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 795}}}, "reference_key_id": "keys/synthetic/syn_03003_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "0d4ead41380940db4359555678482b867e828b27a6b33c483fba125924e02f66", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "00e265ff08ee56cd8ebb0fa3", "attempt_id": "c3dcd1274abf", "run_group_id": "E4_core:A5|repairable|qwen3b:h2", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T06:47:58.214356Z", "task_family": "synthetic", "document_id": "syn_03003_2000", "document_hash": "dddca1a38e3224870fcfae927c701c400fd885c19123ea840ee06a76a140f407", "question_id": "q05_multi_hop", "split": "test", "cluster_id": "syn_03003_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 2, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1866, "query_tokens": 454, "message_tokens": 362, "message_utf8_bytes": 859, "cache_tensor_bytes": null, "serialized_payload_bytes": 859, "side_information_bytes": 0, "fetched_bytes": 508, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 38584.62487498764, "hops": {"1": 22633.649082970805, "2": 15950.975792016834}, "receiver_prefill_ms": 320.5608340213075, "ttft_ms": 94.17770791333169, "decode_ms": 139.55274992622435}, "total_wall_ms": 39139.5137909567, "ttft_ms": 94.17770791333169, "input_usage": {"lookups": 1}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 58, "facts_retained": 15, "retention_rate": 0.25862068965517243, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.2222222222222222, "status": 0.75, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 795}, "ledger_by_hop": {"1": {"facts_total": 58, "facts_retained": 15, "retention_rate": 0.25862068965517243, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.2222222222222222, "status": 0.75, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 766}, "2": {"facts_total": 58, "facts_retained": 15, "retention_rate": 0.25862068965517243, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.2222222222222222, "status": 0.75, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 795}}}, "reference_key_id": "keys/synthetic/syn_03003_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "0d4ead41380940db4359555678482b867e828b27a6b33c483fba125924e02f66", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "5a9b9be3dcd7e9776ce71ea0", "attempt_id": "f5e46f8a2311", "run_group_id": "E4_core:A5|repairable|qwen3b:h2", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T06:47:58.787231Z", "task_family": "synthetic", "document_id": "syn_03003_2000", "document_hash": "dddca1a38e3224870fcfae927c701c400fd885c19123ea840ee06a76a140f407", "question_id": "q06_negation", "split": "test", "cluster_id": "syn_03003_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 2, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1866, "query_tokens": 446, "message_tokens": 362, "message_utf8_bytes": 859, "cache_tensor_bytes": null, "serialized_payload_bytes": 859, "side_information_bytes": 0, "fetched_bytes": 508, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 38584.62487498764, "hops": {"1": 22633.649082970805, "2": 15950.975792016834}, "receiver_prefill_ms": 295.29595898929983, "ttft_ms": 39.86450005322695, "decode_ms": 31.175832962617278}, "total_wall_ms": 38951.65895903483, "ttft_ms": 39.86450005322695, "input_usage": {"lookups": 1}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "no", "evidence_artifact": {"ledger": {"facts_total": 58, "facts_retained": 15, "retention_rate": 0.25862068965517243, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.2222222222222222, "status": 0.75, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 795}, "ledger_by_hop": {"1": {"facts_total": 58, "facts_retained": 15, "retention_rate": 0.25862068965517243, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.2222222222222222, "status": 0.75, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 766}, "2": {"facts_total": 58, "facts_retained": 15, "retention_rate": 0.25862068965517243, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.2222222222222222, "status": 0.75, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 795}}}, "reference_key_id": "keys/synthetic/syn_03003_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "0d4ead41380940db4359555678482b867e828b27a6b33c483fba125924e02f66", "output_hash": "9390298f3fb0c5b160498935d79cb139aef28e1c47358b4bbba61862b9c26e59"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "d5431b81c413aa20f6bc4c6e", "attempt_id": "78bc238600a1", "run_group_id": "E4_core:A5|repairable|qwen3b:h2", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T06:47:59.171679Z", "task_family": "synthetic", "document_id": "syn_03003_2000", "document_hash": "dddca1a38e3224870fcfae927c701c400fd885c19123ea840ee06a76a140f407", "question_id": "q07_negation", "split": "test", "cluster_id": "syn_03003_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 2, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1866, "query_tokens": 443, "message_tokens": 362, "message_utf8_bytes": 859, "cache_tensor_bytes": null, "serialized_payload_bytes": 859, "side_information_bytes": 0, "fetched_bytes": 508, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 38584.62487498764, "hops": {"1": 22633.649082970805, "2": 15950.975792016834}, "receiver_prefill_ms": 291.1348749184981, "ttft_ms": 33.29754096921533, "decode_ms": 36.14695800933987}, "total_wall_ms": 38945.86645800155, "ttft_ms": 33.29754096921533, "input_usage": {"lookups": 1}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 58, "facts_retained": 15, "retention_rate": 0.25862068965517243, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.2222222222222222, "status": 0.75, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 795}, "ledger_by_hop": {"1": {"facts_total": 58, "facts_retained": 15, "retention_rate": 0.25862068965517243, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.2222222222222222, "status": 0.75, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 766}, "2": {"facts_total": 58, "facts_retained": 15, "retention_rate": 0.25862068965517243, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.2222222222222222, "status": 0.75, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 795}}}, "reference_key_id": "keys/synthetic/syn_03003_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "0d4ead41380940db4359555678482b867e828b27a6b33c483fba125924e02f66", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "df237a5a4a480b2692c02614", "attempt_id": "d3a79eda1fa6", "run_group_id": "E4_core:A5|repairable|qwen3b:h2", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T06:47:59.550931Z", "task_family": "synthetic", "document_id": "syn_03003_2000", "document_hash": "dddca1a38e3224870fcfae927c701c400fd885c19123ea840ee06a76a140f407", "question_id": "q08_revision", "split": "test", "cluster_id": "syn_03003_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 2, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1866, "query_tokens": 461, "message_tokens": 362, "message_utf8_bytes": 859, "cache_tensor_bytes": null, "serialized_payload_bytes": 859, "side_information_bytes": 0, "fetched_bytes": 508, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 38584.62487498764, "hops": {"1": 22633.649082970805, "2": 15950.975792016834}, "receiver_prefill_ms": 298.54758409783244, "ttft_ms": 59.52475010417402, "decode_ms": 458.69733300060034}, "total_wall_ms": 39401.96662500966, "ttft_ms": 59.52475010417402, "input_usage": {"lookups": 1}, "output_usage": {"output_tokens": 6}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "911925", "evidence_artifact": {"ledger": {"facts_total": 58, "facts_retained": 15, "retention_rate": 0.25862068965517243, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.2222222222222222, "status": 0.75, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 795}, "ledger_by_hop": {"1": {"facts_total": 58, "facts_retained": 15, "retention_rate": 0.25862068965517243, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.2222222222222222, "status": 0.75, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 766}, "2": {"facts_total": 58, "facts_retained": 15, "retention_rate": 0.25862068965517243, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.2222222222222222, "status": 0.75, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 795}}}, "reference_key_id": "keys/synthetic/syn_03003_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "0d4ead41380940db4359555678482b867e828b27a6b33c483fba125924e02f66", "output_hash": "679a540dc4739f0ace623448b8fd925fb234f554838d7e87da223228ee81b112"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "327acb604e5ccccf9b9319a9", "attempt_id": "021585b1cfde", "run_group_id": "E4_core:A5|repairable|qwen3b:h2", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T06:48:00.385238Z", "task_family": "synthetic", "document_id": "syn_03003_2000", "document_hash": "dddca1a38e3224870fcfae927c701c400fd885c19123ea840ee06a76a140f407", "question_id": "q09_revision", "split": "test", "cluster_id": "syn_03003_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 2, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1866, "query_tokens": 454, "message_tokens": 362, "message_utf8_bytes": 859, "cache_tensor_bytes": null, "serialized_payload_bytes": 859, "side_information_bytes": 0, "fetched_bytes": 508, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 38584.62487498764, "hops": {"1": 22633.649082970805, "2": 15950.975792016834}, "receiver_prefill_ms": 314.35437488835305, "ttft_ms": 38.92958303913474, "decode_ms": 205.54766710847616}, "total_wall_ms": 39143.986999988556, "ttft_ms": 38.92958303913474, "input_usage": {"lookups": 1}, "output_usage": {"output_tokens": 6}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "911925", "evidence_artifact": {"ledger": {"facts_total": 58, "facts_retained": 15, "retention_rate": 0.25862068965517243, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.2222222222222222, "status": 0.75, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 795}, "ledger_by_hop": {"1": {"facts_total": 58, "facts_retained": 15, "retention_rate": 0.25862068965517243, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.2222222222222222, "status": 0.75, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 766}, "2": {"facts_total": 58, "facts_retained": 15, "retention_rate": 0.25862068965517243, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.2222222222222222, "status": 0.75, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 795}}}, "reference_key_id": "keys/synthetic/syn_03003_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "0d4ead41380940db4359555678482b867e828b27a6b33c483fba125924e02f66", "output_hash": "679a540dc4739f0ace623448b8fd925fb234f554838d7e87da223228ee81b112"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "2f661fdf3e29f00dd3263608", "attempt_id": "caf21a4504a3", "run_group_id": "E4_core:A5|repairable|qwen3b:h2", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T06:48:00.961974Z", "task_family": "synthetic", "document_id": "syn_03003_2000", "document_hash": "dddca1a38e3224870fcfae927c701c400fd885c19123ea840ee06a76a140f407", "question_id": "q10_temporal", "split": "test", "cluster_id": "syn_03003_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 2, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1866, "query_tokens": 461, "message_tokens": 362, "message_utf8_bytes": 859, "cache_tensor_bytes": null, "serialized_payload_bytes": 859, "side_information_bytes": 0, "fetched_bytes": 508, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 38584.62487498764, "hops": {"1": 22633.649082970805, "2": 15950.975792016834}, "receiver_prefill_ms": 322.8555410169065, "ttft_ms": 22.60212495457381, "decode_ms": 15.374999959021807}, "total_wall_ms": 38945.938291959465, "ttft_ms": 22.60212495457381, "input_usage": {"lookups": 1}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 58, "facts_retained": 15, "retention_rate": 0.25862068965517243, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.2222222222222222, "status": 0.75, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 795}, "ledger_by_hop": {"1": {"facts_total": 58, "facts_retained": 15, "retention_rate": 0.25862068965517243, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.2222222222222222, "status": 0.75, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 766}, "2": {"facts_total": 58, "facts_retained": 15, "retention_rate": 0.25862068965517243, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.2222222222222222, "status": 0.75, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 795}}}, "reference_key_id": "keys/synthetic/syn_03003_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "0d4ead41380940db4359555678482b867e828b27a6b33c483fba125924e02f66", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "77db9c5ef0aeb77a8dce3c48", "attempt_id": "324905bbdd17", "run_group_id": "E4_core:A5|repairable|qwen3b:h2", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T06:48:01.340503Z", "task_family": "synthetic", "document_id": "syn_03003_2000", "document_hash": "dddca1a38e3224870fcfae927c701c400fd885c19123ea840ee06a76a140f407", "question_id": "q11_aggregation", "split": "test", "cluster_id": "syn_03003_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 2, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1866, "query_tokens": 455, "message_tokens": 362, "message_utf8_bytes": 859, "cache_tensor_bytes": null, "serialized_payload_bytes": 859, "side_information_bytes": 0, "fetched_bytes": 508, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 38584.62487498764, "hops": {"1": 22633.649082970805, "2": 15950.975792016834}, "receiver_prefill_ms": 288.31895801704377, "ttft_ms": 65.66454190760851, "decode_ms": 56.80720810778439}, "total_wall_ms": 38996.01191596594, "ttft_ms": 65.66454190760851, "input_usage": {"lookups": 1}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 58, "facts_retained": 15, "retention_rate": 0.25862068965517243, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.2222222222222222, "status": 0.75, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 795}, "ledger_by_hop": {"1": {"facts_total": 58, "facts_retained": 15, "retention_rate": 0.25862068965517243, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.2222222222222222, "status": 0.75, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 766}, "2": {"facts_total": 58, "facts_retained": 15, "retention_rate": 0.25862068965517243, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.2222222222222222, "status": 0.75, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 795}}}, "reference_key_id": "keys/synthetic/syn_03003_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "0d4ead41380940db4359555678482b867e828b27a6b33c483fba125924e02f66", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "4a6fccf81f01725a9f52d689", "attempt_id": "03940ff84841", "run_group_id": "E4_core:A5|repairable|qwen3b:h2", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T06:48:01.767496Z", "task_family": "synthetic", "document_id": "syn_03003_2000", "document_hash": "dddca1a38e3224870fcfae927c701c400fd885c19123ea840ee06a76a140f407", "question_id": "q12_unit_conversion", "split": "test", "cluster_id": "syn_03003_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 2, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1866, "query_tokens": 452, "message_tokens": 362, "message_utf8_bytes": 859, "cache_tensor_bytes": null, "serialized_payload_bytes": 859, "side_information_bytes": 0, "fetched_bytes": 508, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 38584.62487498764, "hops": {"1": 22633.649082970805, "2": 15950.975792016834}, "receiver_prefill_ms": 465.6832079635933, "ttft_ms": 43.097916059195995, "decode_ms": 153.69995799846947}, "total_wall_ms": 39247.63754196465, "ttft_ms": 43.097916059195995, "input_usage": {"lookups": 1}, "output_usage": {"output_tokens": 5}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "17430", "evidence_artifact": {"ledger": {"facts_total": 58, "facts_retained": 15, "retention_rate": 0.25862068965517243, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.2222222222222222, "status": 0.75, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 795}, "ledger_by_hop": {"1": {"facts_total": 58, "facts_retained": 15, "retention_rate": 0.25862068965517243, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.2222222222222222, "status": 0.75, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 766}, "2": {"facts_total": 58, "facts_retained": 15, "retention_rate": 0.25862068965517243, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.2222222222222222, "status": 0.75, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 795}}}, "reference_key_id": "keys/synthetic/syn_03003_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "0d4ead41380940db4359555678482b867e828b27a6b33c483fba125924e02f66", "output_hash": "2575eaf3b748efc209b676518766751eb038c88f74578951345818bfe7f7194b"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "e78f51fbed4346f7d4824880", "attempt_id": "facf3863cc00", "run_group_id": "E4_core:A5|repairable|qwen3b:h2", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T06:48:02.449260Z", "task_family": "synthetic", "document_id": "syn_03003_2000", "document_hash": "dddca1a38e3224870fcfae927c701c400fd885c19123ea840ee06a76a140f407", "question_id": "q13_unanswerable", "split": "test", "cluster_id": "syn_03003_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 2, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1866, "query_tokens": 456, "message_tokens": 362, "message_utf8_bytes": 859, "cache_tensor_bytes": null, "serialized_payload_bytes": 859, "side_information_bytes": 0, "fetched_bytes": 508, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 38584.62487498764, "hops": {"1": 22633.649082970805, "2": 15950.975792016834}, "receiver_prefill_ms": 323.5949999652803, "ttft_ms": 37.43637492880225, "decode_ms": 31.546875019557774}, "total_wall_ms": 38977.72058297414, "ttft_ms": 37.43637492880225, "input_usage": {"lookups": 1}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 58, "facts_retained": 15, "retention_rate": 0.25862068965517243, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.2222222222222222, "status": 0.75, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 795}, "ledger_by_hop": {"1": {"facts_total": 58, "facts_retained": 15, "retention_rate": 0.25862068965517243, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.2222222222222222, "status": 0.75, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 766}, "2": {"facts_total": 58, "facts_retained": 15, "retention_rate": 0.25862068965517243, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.2222222222222222, "status": 0.75, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 795}}}, "reference_key_id": "keys/synthetic/syn_03003_2000.json", "task_scores": {"correct": 1.0, "abstained": 1.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "0d4ead41380940db4359555678482b867e828b27a6b33c483fba125924e02f66", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "8c8e0abf5b340b3b4b0cf2e4", "attempt_id": "7421dc06b051", "run_group_id": "E4_core:A5|repairable|qwen3b:h2", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T06:48:02.859677Z", "task_family": "synthetic", "document_id": "syn_03003_2000", "document_hash": "dddca1a38e3224870fcfae927c701c400fd885c19123ea840ee06a76a140f407", "question_id": "q14_unanswerable", "split": "test", "cluster_id": "syn_03003_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 2, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1866, "query_tokens": 457, "message_tokens": 362, "message_utf8_bytes": 859, "cache_tensor_bytes": null, "serialized_payload_bytes": 859, "side_information_bytes": 0, "fetched_bytes": 508, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 38584.62487498764, "hops": {"1": 22633.649082970805, "2": 15950.975792016834}, "receiver_prefill_ms": 292.1632499201223, "ttft_ms": 55.096249911002815, "decode_ms": 61.631166958250105}, "total_wall_ms": 38994.06745797023, "ttft_ms": 55.096249911002815, "input_usage": {"lookups": 1}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 58, "facts_retained": 15, "retention_rate": 0.25862068965517243, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.2222222222222222, "status": 0.75, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 795}, "ledger_by_hop": {"1": {"facts_total": 58, "facts_retained": 15, "retention_rate": 0.25862068965517243, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.2222222222222222, "status": 0.75, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 766}, "2": {"facts_total": 58, "facts_retained": 15, "retention_rate": 0.25862068965517243, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.2222222222222222, "status": 0.75, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 795}}}, "reference_key_id": "keys/synthetic/syn_03003_2000.json", "task_scores": {"correct": 1.0, "abstained": 1.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "0d4ead41380940db4359555678482b867e828b27a6b33c483fba125924e02f66", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "698bcfcb682312e4f896a9d1", "attempt_id": "cdb0ea08d7e8", "run_group_id": "E4_core:A5|repairable|qwen3b:h4", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T06:48:35.313864Z", "task_family": "synthetic", "document_id": "syn_03003_2000", "document_hash": "dddca1a38e3224870fcfae927c701c400fd885c19123ea840ee06a76a140f407", "question_id": "q00_single_fact", "split": "test", "cluster_id": "syn_03003_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 4, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1866, "query_tokens": 444, "message_tokens": 362, "message_utf8_bytes": 859, "cache_tensor_bytes": null, "serialized_payload_bytes": 859, "side_information_bytes": 0, "fetched_bytes": 1524, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 70610.8777920017, "hops": {"1": 22633.649082970805, "2": 15950.975792016834, "3": 15936.411041999236, "4": 16089.841875014827}, "receiver_prefill_ms": 311.0643330728635, "ttft_ms": 33.76812499482185, "decode_ms": 305.1580829778686}, "total_wall_ms": 71261.76725002006, "ttft_ms": 33.76812499482185, "input_usage": {"lookups": 3}, "output_usage": {"output_tokens": 10}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "GR-79S-5839", "evidence_artifact": {"ledger": {"facts_total": 58, "facts_retained": 15, "retention_rate": 0.25862068965517243, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.2222222222222222, "status": 0.75, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 795}, "ledger_by_hop": {"1": {"facts_total": 58, "facts_retained": 15, "retention_rate": 0.25862068965517243, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.2222222222222222, "status": 0.75, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 766}, "2": {"facts_total": 58, "facts_retained": 15, "retention_rate": 0.25862068965517243, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.2222222222222222, "status": 0.75, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 795}, "3": {"facts_total": 58, "facts_retained": 15, "retention_rate": 0.25862068965517243, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.2222222222222222, "status": 0.75, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 795}, "4": {"facts_total": 58, "facts_retained": 15, "retention_rate": 0.25862068965517243, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.2222222222222222, "status": 0.75, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 795}}}, "reference_key_id": "keys/synthetic/syn_03003_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "0d4ead41380940db4359555678482b867e828b27a6b33c483fba125924e02f66", "output_hash": "720cc40cd69e85e9728f6bea00b17a95ca07eee643e143a131840763958303cd"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "6d1cbd2ae44d19a9140370f2", "attempt_id": "b5282107eaad", "run_group_id": "E4_core:A5|repairable|qwen3b:h4", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T06:48:35.983849Z", "task_family": "synthetic", "document_id": "syn_03003_2000", "document_hash": "dddca1a38e3224870fcfae927c701c400fd885c19123ea840ee06a76a140f407", "question_id": "q01_single_fact", "split": "test", "cluster_id": "syn_03003_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 4, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1866, "query_tokens": 448, "message_tokens": 362, "message_utf8_bytes": 859, "cache_tensor_bytes": null, "serialized_payload_bytes": 859, "side_information_bytes": 0, "fetched_bytes": 1524, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 70610.8777920017, "hops": {"1": 22633.649082970805, "2": 15950.975792016834, "3": 15936.411041999236, "4": 16089.841875014827}, "receiver_prefill_ms": 327.45920901652426, "ttft_ms": 53.426750004291534, "decode_ms": 114.5436669467017}, "total_wall_ms": 71106.8385419203, "ttft_ms": 53.426750004291534, "input_usage": {"lookups": 3}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 58, "facts_retained": 15, "retention_rate": 0.25862068965517243, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.2222222222222222, "status": 0.75, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 795}, "ledger_by_hop": {"1": {"facts_total": 58, "facts_retained": 15, "retention_rate": 0.25862068965517243, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.2222222222222222, "status": 0.75, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 766}, "2": {"facts_total": 58, "facts_retained": 15, "retention_rate": 0.25862068965517243, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.2222222222222222, "status": 0.75, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 795}, "3": {"facts_total": 58, "facts_retained": 15, "retention_rate": 0.25862068965517243, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.2222222222222222, "status": 0.75, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 795}, "4": {"facts_total": 58, "facts_retained": 15, "retention_rate": 0.25862068965517243, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.2222222222222222, "status": 0.75, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 795}}}, "reference_key_id": "keys/synthetic/syn_03003_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "0d4ead41380940db4359555678482b867e828b27a6b33c483fba125924e02f66", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "c87bae3ffb3069c6a219adba", "attempt_id": "d4277fdb288f", "run_group_id": "E4_core:A5|repairable|qwen3b:h4", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T06:48:36.497194Z", "task_family": "synthetic", "document_id": "syn_03003_2000", "document_hash": "dddca1a38e3224870fcfae927c701c400fd885c19123ea840ee06a76a140f407", "question_id": "q02_exact_copy", "split": "test", "cluster_id": "syn_03003_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 4, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1866, "query_tokens": 448, "message_tokens": 362, "message_utf8_bytes": 859, "cache_tensor_bytes": null, "serialized_payload_bytes": 859, "side_information_bytes": 0, "fetched_bytes": 1524, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 70610.8777920017, "hops": {"1": 22633.649082970805, "2": 15950.975792016834, "3": 15936.411041999236, "4": 16089.841875014827}, "receiver_prefill_ms": 346.4374579489231, "ttft_ms": 38.26462500728667, "decode_ms": 39.010958978906274}, "total_wall_ms": 71035.04949994385, "ttft_ms": 38.26462500728667, "input_usage": {"lookups": 3}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 58, "facts_retained": 15, "retention_rate": 0.25862068965517243, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.2222222222222222, "status": 0.75, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 795}, "ledger_by_hop": {"1": {"facts_total": 58, "facts_retained": 15, "retention_rate": 0.25862068965517243, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.2222222222222222, "status": 0.75, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 766}, "2": {"facts_total": 58, "facts_retained": 15, "retention_rate": 0.25862068965517243, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.2222222222222222, "status": 0.75, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 795}, "3": {"facts_total": 58, "facts_retained": 15, "retention_rate": 0.25862068965517243, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.2222222222222222, "status": 0.75, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 795}, "4": {"facts_total": 58, "facts_retained": 15, "retention_rate": 0.25862068965517243, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.2222222222222222, "status": 0.75, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 795}}}, "reference_key_id": "keys/synthetic/syn_03003_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "0d4ead41380940db4359555678482b867e828b27a6b33c483fba125924e02f66", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "1d4c73bb74ca4234d8350647", "attempt_id": "aa8370b40484", "run_group_id": "E4_core:A5|repairable|qwen3b:h4", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T06:48:36.938392Z", "task_family": "synthetic", "document_id": "syn_03003_2000", "document_hash": "dddca1a38e3224870fcfae927c701c400fd885c19123ea840ee06a76a140f407", "question_id": "q03_exact_copy", "split": "test", "cluster_id": "syn_03003_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 4, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1866, "query_tokens": 443, "message_tokens": 362, "message_utf8_bytes": 859, "cache_tensor_bytes": null, "serialized_payload_bytes": 859, "side_information_bytes": 0, "fetched_bytes": 1524, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 70610.8777920017, "hops": {"1": 22633.649082970805, "2": 15950.975792016834, "3": 15936.411041999236, "4": 16089.841875014827}, "receiver_prefill_ms": 281.87049995176494, "ttft_ms": 35.97299999091774, "decode_ms": 30.139792012050748}, "total_wall_ms": 70959.30374995805, "ttft_ms": 35.97299999091774, "input_usage": {"lookups": 3}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 58, "facts_retained": 15, "retention_rate": 0.25862068965517243, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.2222222222222222, "status": 0.75, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 795}, "ledger_by_hop": {"1": {"facts_total": 58, "facts_retained": 15, "retention_rate": 0.25862068965517243, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.2222222222222222, "status": 0.75, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 766}, "2": {"facts_total": 58, "facts_retained": 15, "retention_rate": 0.25862068965517243, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.2222222222222222, "status": 0.75, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 795}, "3": {"facts_total": 58, "facts_retained": 15, "retention_rate": 0.25862068965517243, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.2222222222222222, "status": 0.75, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 795}, "4": {"facts_total": 58, "facts_retained": 15, "retention_rate": 0.25862068965517243, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.2222222222222222, "status": 0.75, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 795}}}, "reference_key_id": "keys/synthetic/syn_03003_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "0d4ead41380940db4359555678482b867e828b27a6b33c483fba125924e02f66", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "537047c719a315c2f57fbb51", "attempt_id": "d834cd232e4a", "run_group_id": "E4_core:A5|repairable|qwen3b:h4", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T06:48:37.302859Z", "task_family": "synthetic", "document_id": "syn_03003_2000", "document_hash": "dddca1a38e3224870fcfae927c701c400fd885c19123ea840ee06a76a140f407", "question_id": "q04_multi_hop", "split": "test", "cluster_id": "syn_03003_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 4, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1866, "query_tokens": 452, "message_tokens": 362, "message_utf8_bytes": 859, "cache_tensor_bytes": null, "serialized_payload_bytes": 859, "side_information_bytes": 0, "fetched_bytes": 1524, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 70610.8777920017, "hops": {"1": 22633.649082970805, "2": 15950.975792016834, "3": 15936.411041999236, "4": 16089.841875014827}, "receiver_prefill_ms": 300.90004194062203, "ttft_ms": 39.26474996842444, "decode_ms": 52.6696250308305}, "total_wall_ms": 71004.390166956, "ttft_ms": 39.26474996842444, "input_usage": {"lookups": 3}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 58, "facts_retained": 15, "retention_rate": 0.25862068965517243, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.2222222222222222, "status": 0.75, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 795}, "ledger_by_hop": {"1": {"facts_total": 58, "facts_retained": 15, "retention_rate": 0.25862068965517243, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.2222222222222222, "status": 0.75, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 766}, "2": {"facts_total": 58, "facts_retained": 15, "retention_rate": 0.25862068965517243, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.2222222222222222, "status": 0.75, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 795}, "3": {"facts_total": 58, "facts_retained": 15, "retention_rate": 0.25862068965517243, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.2222222222222222, "status": 0.75, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 795}, "4": {"facts_total": 58, "facts_retained": 15, "retention_rate": 0.25862068965517243, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.2222222222222222, "status": 0.75, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 795}}}, "reference_key_id": "keys/synthetic/syn_03003_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "0d4ead41380940db4359555678482b867e828b27a6b33c483fba125924e02f66", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "efe1966e6b3a45e7ce814022", "attempt_id": "c4c50fac6552", "run_group_id": "E4_core:A5|repairable|qwen3b:h4", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T06:48:37.713492Z", "task_family": "synthetic", "document_id": "syn_03003_2000", "document_hash": "dddca1a38e3224870fcfae927c701c400fd885c19123ea840ee06a76a140f407", "question_id": "q05_multi_hop", "split": "test", "cluster_id": "syn_03003_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 4, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1866, "query_tokens": 454, "message_tokens": 362, "message_utf8_bytes": 859, "cache_tensor_bytes": null, "serialized_payload_bytes": 859, "side_information_bytes": 0, "fetched_bytes": 1524, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 70610.8777920017, "hops": {"1": 22633.649082970805, "2": 15950.975792016834, "3": 15936.411041999236, "4": 16089.841875014827}, "receiver_prefill_ms": 418.7838339712471, "ttft_ms": 137.3402919853106, "decode_ms": 45.28179101180285}, "total_wall_ms": 71212.82458398491, "ttft_ms": 137.3402919853106, "input_usage": {"lookups": 3}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 58, "facts_retained": 15, "retention_rate": 0.25862068965517243, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.2222222222222222, "status": 0.75, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 795}, "ledger_by_hop": {"1": {"facts_total": 58, "facts_retained": 15, "retention_rate": 0.25862068965517243, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.2222222222222222, "status": 0.75, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 766}, "2": {"facts_total": 58, "facts_retained": 15, "retention_rate": 0.25862068965517243, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.2222222222222222, "status": 0.75, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 795}, "3": {"facts_total": 58, "facts_retained": 15, "retention_rate": 0.25862068965517243, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.2222222222222222, "status": 0.75, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 795}, "4": {"facts_total": 58, "facts_retained": 15, "retention_rate": 0.25862068965517243, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.2222222222222222, "status": 0.75, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 795}}}, "reference_key_id": "keys/synthetic/syn_03003_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "0d4ead41380940db4359555678482b867e828b27a6b33c483fba125924e02f66", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "5ab1fab849ad028c13da04ba", "attempt_id": "3a29730b903b", "run_group_id": "E4_core:A5|repairable|qwen3b:h4", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T06:48:38.335704Z", "task_family": "synthetic", "document_id": "syn_03003_2000", "document_hash": "dddca1a38e3224870fcfae927c701c400fd885c19123ea840ee06a76a140f407", "question_id": "q06_negation", "split": "test", "cluster_id": "syn_03003_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 4, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1866, "query_tokens": 446, "message_tokens": 362, "message_utf8_bytes": 859, "cache_tensor_bytes": null, "serialized_payload_bytes": 859, "side_information_bytes": 0, "fetched_bytes": 1524, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 70610.8777920017, "hops": {"1": 22633.649082970805, "2": 15950.975792016834, "3": 15936.411041999236, "4": 16089.841875014827}, "receiver_prefill_ms": 290.47720797825605, "ttft_ms": 29.57808307837695, "decode_ms": 45.23812502156943}, "total_wall_ms": 70976.62700107321, "ttft_ms": 29.57808307837695, "input_usage": {"lookups": 3}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "no", "evidence_artifact": {"ledger": {"facts_total": 58, "facts_retained": 15, "retention_rate": 0.25862068965517243, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.2222222222222222, "status": 0.75, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 795}, "ledger_by_hop": {"1": {"facts_total": 58, "facts_retained": 15, "retention_rate": 0.25862068965517243, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.2222222222222222, "status": 0.75, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 766}, "2": {"facts_total": 58, "facts_retained": 15, "retention_rate": 0.25862068965517243, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.2222222222222222, "status": 0.75, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 795}, "3": {"facts_total": 58, "facts_retained": 15, "retention_rate": 0.25862068965517243, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.2222222222222222, "status": 0.75, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 795}, "4": {"facts_total": 58, "facts_retained": 15, "retention_rate": 0.25862068965517243, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.2222222222222222, "status": 0.75, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 795}}}, "reference_key_id": "keys/synthetic/syn_03003_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "0d4ead41380940db4359555678482b867e828b27a6b33c483fba125924e02f66", "output_hash": "9390298f3fb0c5b160498935d79cb139aef28e1c47358b4bbba61862b9c26e59"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "26140bb339843bfc5146ff4d", "attempt_id": "18425191d8e8", "run_group_id": "E4_core:A5|repairable|qwen3b:h4", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T06:48:38.718604Z", "task_family": "synthetic", "document_id": "syn_03003_2000", "document_hash": "dddca1a38e3224870fcfae927c701c400fd885c19123ea840ee06a76a140f407", "question_id": "q07_negation", "split": "test", "cluster_id": "syn_03003_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 4, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1866, "query_tokens": 443, "message_tokens": 362, "message_utf8_bytes": 859, "cache_tensor_bytes": null, "serialized_payload_bytes": 859, "side_information_bytes": 0, "fetched_bytes": 1524, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 70610.8777920017, "hops": {"1": 22633.649082970805, "2": 15950.975792016834, "3": 15936.411041999236, "4": 16089.841875014827}, "receiver_prefill_ms": 290.56262492667884, "ttft_ms": 28.42275006696582, "decode_ms": 40.63629207666963}, "total_wall_ms": 70971.00029198918, "ttft_ms": 28.42275006696582, "input_usage": {"lookups": 3}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 58, "facts_retained": 15, "retention_rate": 0.25862068965517243, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.2222222222222222, "status": 0.75, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 795}, "ledger_by_hop": {"1": {"facts_total": 58, "facts_retained": 15, "retention_rate": 0.25862068965517243, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.2222222222222222, "status": 0.75, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 766}, "2": {"facts_total": 58, "facts_retained": 15, "retention_rate": 0.25862068965517243, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.2222222222222222, "status": 0.75, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 795}, "3": {"facts_total": 58, "facts_retained": 15, "retention_rate": 0.25862068965517243, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.2222222222222222, "status": 0.75, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 795}, "4": {"facts_total": 58, "facts_retained": 15, "retention_rate": 0.25862068965517243, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.2222222222222222, "status": 0.75, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 795}}}, "reference_key_id": "keys/synthetic/syn_03003_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "0d4ead41380940db4359555678482b867e828b27a6b33c483fba125924e02f66", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "4a6600e00ab9e02c423661e4", "attempt_id": "0e0958d7ec5a", "run_group_id": "E4_core:A5|repairable|qwen3b:h4", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T06:48:39.096064Z", "task_family": "synthetic", "document_id": "syn_03003_2000", "document_hash": "dddca1a38e3224870fcfae927c701c400fd885c19123ea840ee06a76a140f407", "question_id": "q08_revision", "split": "test", "cluster_id": "syn_03003_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 4, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1866, "query_tokens": 461, "message_tokens": 362, "message_utf8_bytes": 859, "cache_tensor_bytes": null, "serialized_payload_bytes": 859, "side_information_bytes": 0, "fetched_bytes": 1524, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 70610.8777920017, "hops": {"1": 22633.649082970805, "2": 15950.975792016834, "3": 15936.411041999236, "4": 16089.841875014827}, "receiver_prefill_ms": 324.63675003964454, "ttft_ms": 55.18045905046165, "decode_ms": 429.08745806198567}, "total_wall_ms": 71420.61412497424, "ttft_ms": 55.18045905046165, "input_usage": {"lookups": 3}, "output_usage": {"output_tokens": 6}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "911925", "evidence_artifact": {"ledger": {"facts_total": 58, "facts_retained": 15, "retention_rate": 0.25862068965517243, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.2222222222222222, "status": 0.75, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 795}, "ledger_by_hop": {"1": {"facts_total": 58, "facts_retained": 15, "retention_rate": 0.25862068965517243, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.2222222222222222, "status": 0.75, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 766}, "2": {"facts_total": 58, "facts_retained": 15, "retention_rate": 0.25862068965517243, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.2222222222222222, "status": 0.75, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 795}, "3": {"facts_total": 58, "facts_retained": 15, "retention_rate": 0.25862068965517243, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.2222222222222222, "status": 0.75, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 795}, "4": {"facts_total": 58, "facts_retained": 15, "retention_rate": 0.25862068965517243, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.2222222222222222, "status": 0.75, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 795}}}, "reference_key_id": "keys/synthetic/syn_03003_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "0d4ead41380940db4359555678482b867e828b27a6b33c483fba125924e02f66", "output_hash": "679a540dc4739f0ace623448b8fd925fb234f554838d7e87da223228ee81b112"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "bdfe090e2c0a460e383396f5", "attempt_id": "e7489d867d6d", "run_group_id": "E4_core:A5|repairable|qwen3b:h4", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T06:48:39.923411Z", "task_family": "synthetic", "document_id": "syn_03003_2000", "document_hash": "dddca1a38e3224870fcfae927c701c400fd885c19123ea840ee06a76a140f407", "question_id": "q09_revision", "split": "test", "cluster_id": "syn_03003_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 4, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1866, "query_tokens": 454, "message_tokens": 362, "message_utf8_bytes": 859, "cache_tensor_bytes": null, "serialized_payload_bytes": 859, "side_information_bytes": 0, "fetched_bytes": 1524, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 70610.8777920017, "hops": {"1": 22633.649082970805, "2": 15950.975792016834, "3": 15936.411041999236, "4": 16089.841875014827}, "receiver_prefill_ms": 327.23362499382347, "ttft_ms": 38.902874919585884, "decode_ms": 198.8871250068769}, "total_wall_ms": 71176.51000095066, "ttft_ms": 38.902874919585884, "input_usage": {"lookups": 3}, "output_usage": {"output_tokens": 6}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "911925", "evidence_artifact": {"ledger": {"facts_total": 58, "facts_retained": 15, "retention_rate": 0.25862068965517243, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.2222222222222222, "status": 0.75, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 795}, "ledger_by_hop": {"1": {"facts_total": 58, "facts_retained": 15, "retention_rate": 0.25862068965517243, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.2222222222222222, "status": 0.75, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 766}, "2": {"facts_total": 58, "facts_retained": 15, "retention_rate": 0.25862068965517243, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.2222222222222222, "status": 0.75, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 795}, "3": {"facts_total": 58, "facts_retained": 15, "retention_rate": 0.25862068965517243, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.2222222222222222, "status": 0.75, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 795}, "4": {"facts_total": 58, "facts_retained": 15, "retention_rate": 0.25862068965517243, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.2222222222222222, "status": 0.75, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 795}}}, "reference_key_id": "keys/synthetic/syn_03003_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "0d4ead41380940db4359555678482b867e828b27a6b33c483fba125924e02f66", "output_hash": "679a540dc4739f0ace623448b8fd925fb234f554838d7e87da223228ee81b112"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "3644a8c3aaf4634fcc61da96", "attempt_id": "24331f3380a4", "run_group_id": "E4_core:A5|repairable|qwen3b:h4", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T06:48:40.505701Z", "task_family": "synthetic", "document_id": "syn_03003_2000", "document_hash": "dddca1a38e3224870fcfae927c701c400fd885c19123ea840ee06a76a140f407", "question_id": "q10_temporal", "split": "test", "cluster_id": "syn_03003_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 4, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1866, "query_tokens": 461, "message_tokens": 362, "message_utf8_bytes": 859, "cache_tensor_bytes": null, "serialized_payload_bytes": 859, "side_information_bytes": 0, "fetched_bytes": 1524, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 70610.8777920017, "hops": {"1": 22633.649082970805, "2": 15950.975792016834, "3": 15936.411041999236, "4": 16089.841875014827}, "receiver_prefill_ms": 299.89008395932615, "ttft_ms": 13.353749993257225, "decode_ms": 21.134749986231327}, "total_wall_ms": 70946.03108393494, "ttft_ms": 13.353749993257225, "input_usage": {"lookups": 3}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 58, "facts_retained": 15, "retention_rate": 0.25862068965517243, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.2222222222222222, "status": 0.75, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 795}, "ledger_by_hop": {"1": {"facts_total": 58, "facts_retained": 15, "retention_rate": 0.25862068965517243, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.2222222222222222, "status": 0.75, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 766}, "2": {"facts_total": 58, "facts_retained": 15, "retention_rate": 0.25862068965517243, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.2222222222222222, "status": 0.75, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 795}, "3": {"facts_total": 58, "facts_retained": 15, "retention_rate": 0.25862068965517243, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.2222222222222222, "status": 0.75, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 795}, "4": {"facts_total": 58, "facts_retained": 15, "retention_rate": 0.25862068965517243, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.2222222222222222, "status": 0.75, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 795}}}, "reference_key_id": "keys/synthetic/syn_03003_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "0d4ead41380940db4359555678482b867e828b27a6b33c483fba125924e02f66", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "d406561ec9c9f4bb318e152f", "attempt_id": "e19a94a59325", "run_group_id": "E4_core:A5|repairable|qwen3b:h4", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T06:48:40.858386Z", "task_family": "synthetic", "document_id": "syn_03003_2000", "document_hash": "dddca1a38e3224870fcfae927c701c400fd885c19123ea840ee06a76a140f407", "question_id": "q11_aggregation", "split": "test", "cluster_id": "syn_03003_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 4, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1866, "query_tokens": 455, "message_tokens": 362, "message_utf8_bytes": 859, "cache_tensor_bytes": null, "serialized_payload_bytes": 859, "side_information_bytes": 0, "fetched_bytes": 1524, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 70610.8777920017, "hops": {"1": 22633.649082970805, "2": 15950.975792016834, "3": 15936.411041999236, "4": 16089.841875014827}, "receiver_prefill_ms": 351.87162493821234, "ttft_ms": 45.89287505950779, "decode_ms": 108.97270892746747}, "total_wall_ms": 71118.22449997999, "ttft_ms": 45.89287505950779, "input_usage": {"lookups": 3}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 58, "facts_retained": 15, "retention_rate": 0.25862068965517243, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.2222222222222222, "status": 0.75, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 795}, "ledger_by_hop": {"1": {"facts_total": 58, "facts_retained": 15, "retention_rate": 0.25862068965517243, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.2222222222222222, "status": 0.75, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 766}, "2": {"facts_total": 58, "facts_retained": 15, "retention_rate": 0.25862068965517243, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.2222222222222222, "status": 0.75, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 795}, "3": {"facts_total": 58, "facts_retained": 15, "retention_rate": 0.25862068965517243, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.2222222222222222, "status": 0.75, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 795}, "4": {"facts_total": 58, "facts_retained": 15, "retention_rate": 0.25862068965517243, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.2222222222222222, "status": 0.75, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 795}}}, "reference_key_id": "keys/synthetic/syn_03003_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "0d4ead41380940db4359555678482b867e828b27a6b33c483fba125924e02f66", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "95c9692505755248479f4c04", "attempt_id": "74c3c7ffd295", "run_group_id": "E4_core:A5|repairable|qwen3b:h4", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T06:48:41.383050Z", "task_family": "synthetic", "document_id": "syn_03003_2000", "document_hash": "dddca1a38e3224870fcfae927c701c400fd885c19123ea840ee06a76a140f407", "question_id": "q12_unit_conversion", "split": "test", "cluster_id": "syn_03003_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 4, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1866, "query_tokens": 452, "message_tokens": 362, "message_utf8_bytes": 859, "cache_tensor_bytes": null, "serialized_payload_bytes": 859, "side_information_bytes": 0, "fetched_bytes": 1524, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 70610.8777920017, "hops": {"1": 22633.649082970805, "2": 15950.975792016834, "3": 15936.411041999236, "4": 16089.841875014827}, "receiver_prefill_ms": 395.00037499237806, "ttft_ms": 37.28312498424202, "decode_ms": 149.8420409625396}, "total_wall_ms": 71193.6633340083, "ttft_ms": 37.28312498424202, "input_usage": {"lookups": 3}, "output_usage": {"output_tokens": 5}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "17430", "evidence_artifact": {"ledger": {"facts_total": 58, "facts_retained": 15, "retention_rate": 0.25862068965517243, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.2222222222222222, "status": 0.75, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 795}, "ledger_by_hop": {"1": {"facts_total": 58, "facts_retained": 15, "retention_rate": 0.25862068965517243, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.2222222222222222, "status": 0.75, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 766}, "2": {"facts_total": 58, "facts_retained": 15, "retention_rate": 0.25862068965517243, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.2222222222222222, "status": 0.75, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 795}, "3": {"facts_total": 58, "facts_retained": 15, "retention_rate": 0.25862068965517243, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.2222222222222222, "status": 0.75, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 795}, "4": {"facts_total": 58, "facts_retained": 15, "retention_rate": 0.25862068965517243, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.2222222222222222, "status": 0.75, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 795}}}, "reference_key_id": "keys/synthetic/syn_03003_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "0d4ead41380940db4359555678482b867e828b27a6b33c483fba125924e02f66", "output_hash": "2575eaf3b748efc209b676518766751eb038c88f74578951345818bfe7f7194b"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "72ffcfe6d65eb76d458630c2", "attempt_id": "db5d77f8cfcb", "run_group_id": "E4_core:A5|repairable|qwen3b:h4", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T06:48:41.985887Z", "task_family": "synthetic", "document_id": "syn_03003_2000", "document_hash": "dddca1a38e3224870fcfae927c701c400fd885c19123ea840ee06a76a140f407", "question_id": "q13_unanswerable", "split": "test", "cluster_id": "syn_03003_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 4, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1866, "query_tokens": 456, "message_tokens": 362, "message_utf8_bytes": 859, "cache_tensor_bytes": null, "serialized_payload_bytes": 859, "side_information_bytes": 0, "fetched_bytes": 1524, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 70610.8777920017, "hops": {"1": 22633.649082970805, "2": 15950.975792016834, "3": 15936.411041999236, "4": 16089.841875014827}, "receiver_prefill_ms": 323.43012501951307, "ttft_ms": 40.007042000070214, "decode_ms": 29.689499991945922}, "total_wall_ms": 71004.45208291057, "ttft_ms": 40.007042000070214, "input_usage": {"lookups": 3}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 58, "facts_retained": 15, "retention_rate": 0.25862068965517243, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.2222222222222222, "status": 0.75, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 795}, "ledger_by_hop": {"1": {"facts_total": 58, "facts_retained": 15, "retention_rate": 0.25862068965517243, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.2222222222222222, "status": 0.75, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 766}, "2": {"facts_total": 58, "facts_retained": 15, "retention_rate": 0.25862068965517243, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.2222222222222222, "status": 0.75, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 795}, "3": {"facts_total": 58, "facts_retained": 15, "retention_rate": 0.25862068965517243, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.2222222222222222, "status": 0.75, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 795}, "4": {"facts_total": 58, "facts_retained": 15, "retention_rate": 0.25862068965517243, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.2222222222222222, "status": 0.75, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 795}}}, "reference_key_id": "keys/synthetic/syn_03003_2000.json", "task_scores": {"correct": 1.0, "abstained": 1.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "0d4ead41380940db4359555678482b867e828b27a6b33c483fba125924e02f66", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "19ab7235a6c7d83396589b0a", "attempt_id": "ca37a281e502", "run_group_id": "E4_core:A5|repairable|qwen3b:h4", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T06:48:42.397239Z", "task_family": "synthetic", "document_id": "syn_03003_2000", "document_hash": "dddca1a38e3224870fcfae927c701c400fd885c19123ea840ee06a76a140f407", "question_id": "q14_unanswerable", "split": "test", "cluster_id": "syn_03003_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 4, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1866, "query_tokens": 457, "message_tokens": 362, "message_utf8_bytes": 859, "cache_tensor_bytes": null, "serialized_payload_bytes": 859, "side_information_bytes": 0, "fetched_bytes": 1524, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 70610.8777920017, "hops": {"1": 22633.649082970805, "2": 15950.975792016834, "3": 15936.411041999236, "4": 16089.841875014827}, "receiver_prefill_ms": 298.9968330366537, "ttft_ms": 60.34920806996524, "decode_ms": 78.51162494625896}, "total_wall_ms": 71049.2929170141, "ttft_ms": 60.34920806996524, "input_usage": {"lookups": 3}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 58, "facts_retained": 15, "retention_rate": 0.25862068965517243, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.2222222222222222, "status": 0.75, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 795}, "ledger_by_hop": {"1": {"facts_total": 58, "facts_retained": 15, "retention_rate": 0.25862068965517243, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.2222222222222222, "status": 0.75, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 766}, "2": {"facts_total": 58, "facts_retained": 15, "retention_rate": 0.25862068965517243, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.2222222222222222, "status": 0.75, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 795}, "3": {"facts_total": 58, "facts_retained": 15, "retention_rate": 0.25862068965517243, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.2222222222222222, "status": 0.75, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 795}, "4": {"facts_total": 58, "facts_retained": 15, "retention_rate": 0.25862068965517243, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.2222222222222222, "status": 0.75, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 795}}}, "reference_key_id": "keys/synthetic/syn_03003_2000.json", "task_scores": {"correct": 1.0, "abstained": 1.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "0d4ead41380940db4359555678482b867e828b27a6b33c483fba125924e02f66", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "45a847150ce8802c41b8e2e4", "attempt_id": "50e67d32719a", "run_group_id": "E4_core:A5|repairable|qwen3b:h8", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T06:49:48.490824Z", "task_family": "synthetic", "document_id": "syn_03003_2000", "document_hash": "dddca1a38e3224870fcfae927c701c400fd885c19123ea840ee06a76a140f407", "question_id": "q00_single_fact", "split": "test", "cluster_id": "syn_03003_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 8, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1866, "query_tokens": 444, "message_tokens": 362, "message_utf8_bytes": 859, "cache_tensor_bytes": null, "serialized_payload_bytes": 859, "side_information_bytes": 0, "fetched_bytes": 3556, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 136245.71970908437, "hops": {"1": 22633.649082970805, "2": 15950.975792016834, "3": 15936.411041999236, "4": 16089.841875014827, "5": 17749.737125006504, "6": 15981.693333014846, "7": 15952.465916983783, "8": 15950.945542077534}, "receiver_prefill_ms": 349.3275410728529, "ttft_ms": 36.56495898030698, "decode_ms": 332.1501249447465}, "total_wall_ms": 136964.2777921399, "ttft_ms": 36.56495898030698, "input_usage": {"lookups": 7}, "output_usage": {"output_tokens": 10}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "GR-79S-5839", "evidence_artifact": {"ledger": {"facts_total": 58, "facts_retained": 15, "retention_rate": 0.25862068965517243, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.2222222222222222, "status": 0.75, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 795}, "ledger_by_hop": {"1": {"facts_total": 58, "facts_retained": 15, "retention_rate": 0.25862068965517243, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.2222222222222222, "status": 0.75, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 766}, "2": {"facts_total": 58, "facts_retained": 15, "retention_rate": 0.25862068965517243, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.2222222222222222, "status": 0.75, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 795}, "3": {"facts_total": 58, "facts_retained": 15, "retention_rate": 0.25862068965517243, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.2222222222222222, "status": 0.75, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 795}, "4": {"facts_total": 58, "facts_retained": 15, "retention_rate": 0.25862068965517243, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.2222222222222222, "status": 0.75, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 795}, "5": {"facts_total": 58, "facts_retained": 15, "retention_rate": 0.25862068965517243, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.2222222222222222, "status": 0.75, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 795}, "6": {"facts_total": 58, "facts_retained": 15, "retention_rate": 0.25862068965517243, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.2222222222222222, "status": 0.75, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 795}, "7": {"facts_total": 58, "facts_retained": 15, "retention_rate": 0.25862068965517243, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.2222222222222222, "status": 0.75, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 795}, "8": {"facts_total": 58, "facts_retained": 15, "retention_rate": 0.25862068965517243, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.2222222222222222, "status": 0.75, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 795}}}, "reference_key_id": "keys/synthetic/syn_03003_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "0d4ead41380940db4359555678482b867e828b27a6b33c483fba125924e02f66", "output_hash": "720cc40cd69e85e9728f6bea00b17a95ca07eee643e143a131840763958303cd"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "96400401391825e48712121a", "attempt_id": "8162d9f06bb4", "run_group_id": "E4_core:A5|repairable|qwen3b:h8", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T06:49:49.226684Z", "task_family": "synthetic", "document_id": "syn_03003_2000", "document_hash": "dddca1a38e3224870fcfae927c701c400fd885c19123ea840ee06a76a140f407", "question_id": "q01_single_fact", "split": "test", "cluster_id": "syn_03003_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 8, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1866, "query_tokens": 448, "message_tokens": 362, "message_utf8_bytes": 859, "cache_tensor_bytes": null, "serialized_payload_bytes": 859, "side_information_bytes": 0, "fetched_bytes": 3556, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 136245.71970908437, "hops": {"1": 22633.649082970805, "2": 15950.975792016834, "3": 15936.411041999236, "4": 16089.841875014827, "5": 17749.737125006504, "6": 15981.693333014846, "7": 15952.465916983783, "8": 15950.945542077534}, "receiver_prefill_ms": 278.9101249072701, "ttft_ms": 42.20937506761402, "decode_ms": 71.24399999156594}, "total_wall_ms": 136638.6195430532, "ttft_ms": 42.20937506761402, "input_usage": {"lookups": 7}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 58, "facts_retained": 15, "retention_rate": 0.25862068965517243, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.2222222222222222, "status": 0.75, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 795}, "ledger_by_hop": {"1": {"facts_total": 58, "facts_retained": 15, "retention_rate": 0.25862068965517243, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.2222222222222222, "status": 0.75, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 766}, "2": {"facts_total": 58, "facts_retained": 15, "retention_rate": 0.25862068965517243, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.2222222222222222, "status": 0.75, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 795}, "3": {"facts_total": 58, "facts_retained": 15, "retention_rate": 0.25862068965517243, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.2222222222222222, "status": 0.75, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 795}, "4": {"facts_total": 58, "facts_retained": 15, "retention_rate": 0.25862068965517243, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.2222222222222222, "status": 0.75, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 795}, "5": {"facts_total": 58, "facts_retained": 15, "retention_rate": 0.25862068965517243, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.2222222222222222, "status": 0.75, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 795}, "6": {"facts_total": 58, "facts_retained": 15, "retention_rate": 0.25862068965517243, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.2222222222222222, "status": 0.75, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 795}, "7": {"facts_total": 58, "facts_retained": 15, "retention_rate": 0.25862068965517243, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.2222222222222222, "status": 0.75, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 795}, "8": {"facts_total": 58, "facts_retained": 15, "retention_rate": 0.25862068965517243, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.2222222222222222, "status": 0.75, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 795}}}, "reference_key_id": "keys/synthetic/syn_03003_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "0d4ead41380940db4359555678482b867e828b27a6b33c483fba125924e02f66", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "aa074ad416b5914a33ce00b4", "attempt_id": "aa17afa550ee", "run_group_id": "E4_core:A5|repairable|qwen3b:h8", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T06:49:49.636924Z", "task_family": "synthetic", "document_id": "syn_03003_2000", "document_hash": "dddca1a38e3224870fcfae927c701c400fd885c19123ea840ee06a76a140f407", "question_id": "q02_exact_copy", "split": "test", "cluster_id": "syn_03003_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 8, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1866, "query_tokens": 448, "message_tokens": 362, "message_utf8_bytes": 859, "cache_tensor_bytes": null, "serialized_payload_bytes": 859, "side_information_bytes": 0, "fetched_bytes": 3556, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 136245.71970908437, "hops": {"1": 22633.649082970805, "2": 15950.975792016834, "3": 15936.411041999236, "4": 16089.841875014827, "5": 17749.737125006504, "6": 15981.693333014846, "7": 15952.465916983783, "8": 15950.945542077534}, "receiver_prefill_ms": 338.80600007250905, "ttft_ms": 81.93329209461808, "decode_ms": 106.92187491804361}, "total_wall_ms": 136774.11112608388, "ttft_ms": 81.93329209461808, "input_usage": {"lookups": 7}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 58, "facts_retained": 15, "retention_rate": 0.25862068965517243, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.2222222222222222, "status": 0.75, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 795}, "ledger_by_hop": {"1": {"facts_total": 58, "facts_retained": 15, "retention_rate": 0.25862068965517243, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.2222222222222222, "status": 0.75, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 766}, "2": {"facts_total": 58, "facts_retained": 15, "retention_rate": 0.25862068965517243, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.2222222222222222, "status": 0.75, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 795}, "3": {"facts_total": 58, "facts_retained": 15, "retention_rate": 0.25862068965517243, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.2222222222222222, "status": 0.75, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 795}, "4": {"facts_total": 58, "facts_retained": 15, "retention_rate": 0.25862068965517243, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.2222222222222222, "status": 0.75, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 795}, "5": {"facts_total": 58, "facts_retained": 15, "retention_rate": 0.25862068965517243, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.2222222222222222, "status": 0.75, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 795}, "6": {"facts_total": 58, "facts_retained": 15, "retention_rate": 0.25862068965517243, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.2222222222222222, "status": 0.75, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 795}, "7": {"facts_total": 58, "facts_retained": 15, "retention_rate": 0.25862068965517243, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.2222222222222222, "status": 0.75, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 795}, "8": {"facts_total": 58, "facts_retained": 15, "retention_rate": 0.25862068965517243, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.2222222222222222, "status": 0.75, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 795}}}, "reference_key_id": "keys/synthetic/syn_03003_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "0d4ead41380940db4359555678482b867e828b27a6b33c483fba125924e02f66", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "0e9b41430e85d689c5d1645f", "attempt_id": "7f64b3c04b3f", "run_group_id": "E4_core:A5|repairable|qwen3b:h8", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T06:49:50.183843Z", "task_family": "synthetic", "document_id": "syn_03003_2000", "document_hash": "dddca1a38e3224870fcfae927c701c400fd885c19123ea840ee06a76a140f407", "question_id": "q03_exact_copy", "split": "test", "cluster_id": "syn_03003_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 8, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1866, "query_tokens": 443, "message_tokens": 362, "message_utf8_bytes": 859, "cache_tensor_bytes": null, "serialized_payload_bytes": 859, "side_information_bytes": 0, "fetched_bytes": 3556, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 136245.71970908437, "hops": {"1": 22633.649082970805, "2": 15950.975792016834, "3": 15936.411041999236, "4": 16089.841875014827, "5": 17749.737125006504, "6": 15981.693333014846, "7": 15952.465916983783, "8": 15950.945542077534}, "receiver_prefill_ms": 282.806291943416, "ttft_ms": 25.450749904848635, "decode_ms": 44.777542003430426}, "total_wall_ms": 136599.2813340854, "ttft_ms": 25.450749904848635, "input_usage": {"lookups": 7}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 58, "facts_retained": 15, "retention_rate": 0.25862068965517243, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.2222222222222222, "status": 0.75, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 795}, "ledger_by_hop": {"1": {"facts_total": 58, "facts_retained": 15, "retention_rate": 0.25862068965517243, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.2222222222222222, "status": 0.75, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 766}, "2": {"facts_total": 58, "facts_retained": 15, "retention_rate": 0.25862068965517243, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.2222222222222222, "status": 0.75, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 795}, "3": {"facts_total": 58, "facts_retained": 15, "retention_rate": 0.25862068965517243, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.2222222222222222, "status": 0.75, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 795}, "4": {"facts_total": 58, "facts_retained": 15, "retention_rate": 0.25862068965517243, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.2222222222222222, "status": 0.75, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 795}, "5": {"facts_total": 58, "facts_retained": 15, "retention_rate": 0.25862068965517243, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.2222222222222222, "status": 0.75, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 795}, "6": {"facts_total": 58, "facts_retained": 15, "retention_rate": 0.25862068965517243, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.2222222222222222, "status": 0.75, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 795}, "7": {"facts_total": 58, "facts_retained": 15, "retention_rate": 0.25862068965517243, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.2222222222222222, "status": 0.75, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 795}, "8": {"facts_total": 58, "facts_retained": 15, "retention_rate": 0.25862068965517243, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.2222222222222222, "status": 0.75, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 795}}}, "reference_key_id": "keys/synthetic/syn_03003_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "0d4ead41380940db4359555678482b867e828b27a6b33c483fba125924e02f66", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "0ef9a8a21e947ad838d048d7", "attempt_id": "8b301c47332b", "run_group_id": "E4_core:A5|repairable|qwen3b:h8", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T06:49:50.554189Z", "task_family": "synthetic", "document_id": "syn_03003_2000", "document_hash": "dddca1a38e3224870fcfae927c701c400fd885c19123ea840ee06a76a140f407", "question_id": "q04_multi_hop", "split": "test", "cluster_id": "syn_03003_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 8, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1866, "query_tokens": 452, "message_tokens": 362, "message_utf8_bytes": 859, "cache_tensor_bytes": null, "serialized_payload_bytes": 859, "side_information_bytes": 0, "fetched_bytes": 3556, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 136245.71970908437, "hops": {"1": 22633.649082970805, "2": 15950.975792016834, "3": 15936.411041999236, "4": 16089.841875014827, "5": 17749.737125006504, "6": 15981.693333014846, "7": 15952.465916983783, "8": 15950.945542077534}, "receiver_prefill_ms": 320.49212499987334, "ttft_ms": 33.25554192997515, "decode_ms": 32.51041704788804}, "total_wall_ms": 136632.43795908056, "ttft_ms": 33.25554192997515, "input_usage": {"lookups": 7}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 58, "facts_retained": 15, "retention_rate": 0.25862068965517243, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.2222222222222222, "status": 0.75, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 795}, "ledger_by_hop": {"1": {"facts_total": 58, "facts_retained": 15, "retention_rate": 0.25862068965517243, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.2222222222222222, "status": 0.75, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 766}, "2": {"facts_total": 58, "facts_retained": 15, "retention_rate": 0.25862068965517243, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.2222222222222222, "status": 0.75, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 795}, "3": {"facts_total": 58, "facts_retained": 15, "retention_rate": 0.25862068965517243, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.2222222222222222, "status": 0.75, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 795}, "4": {"facts_total": 58, "facts_retained": 15, "retention_rate": 0.25862068965517243, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.2222222222222222, "status": 0.75, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 795}, "5": {"facts_total": 58, "facts_retained": 15, "retention_rate": 0.25862068965517243, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.2222222222222222, "status": 0.75, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 795}, "6": {"facts_total": 58, "facts_retained": 15, "retention_rate": 0.25862068965517243, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.2222222222222222, "status": 0.75, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 795}, "7": {"facts_total": 58, "facts_retained": 15, "retention_rate": 0.25862068965517243, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.2222222222222222, "status": 0.75, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 795}, "8": {"facts_total": 58, "facts_retained": 15, "retention_rate": 0.25862068965517243, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.2222222222222222, "status": 0.75, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 795}}}, "reference_key_id": "keys/synthetic/syn_03003_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "0d4ead41380940db4359555678482b867e828b27a6b33c483fba125924e02f66", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "e488bd664bba05f4fc471da4", "attempt_id": "8c85c040fef0", "run_group_id": "E4_core:A5|repairable|qwen3b:h8", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T06:49:50.958109Z", "task_family": "synthetic", "document_id": "syn_03003_2000", "document_hash": "dddca1a38e3224870fcfae927c701c400fd885c19123ea840ee06a76a140f407", "question_id": "q05_multi_hop", "split": "test", "cluster_id": "syn_03003_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 8, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1866, "query_tokens": 454, "message_tokens": 362, "message_utf8_bytes": 859, "cache_tensor_bytes": null, "serialized_payload_bytes": 859, "side_information_bytes": 0, "fetched_bytes": 3556, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 136245.71970908437, "hops": {"1": 22633.649082970805, "2": 15950.975792016834, "3": 15936.411041999236, "4": 16089.841875014827, "5": 17749.737125006504, "6": 15981.693333014846, "7": 15952.465916983783, "8": 15950.945542077534}, "receiver_prefill_ms": 282.6463329838589, "ttft_ms": 45.288374996744096, "decode_ms": 68.37624998297542}, "total_wall_ms": 136642.5860421732, "ttft_ms": 45.288374996744096, "input_usage": {"lookups": 7}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 58, "facts_retained": 15, "retention_rate": 0.25862068965517243, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.2222222222222222, "status": 0.75, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 795}, "ledger_by_hop": {"1": {"facts_total": 58, "facts_retained": 15, "retention_rate": 0.25862068965517243, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.2222222222222222, "status": 0.75, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 766}, "2": {"facts_total": 58, "facts_retained": 15, "retention_rate": 0.25862068965517243, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.2222222222222222, "status": 0.75, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 795}, "3": {"facts_total": 58, "facts_retained": 15, "retention_rate": 0.25862068965517243, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.2222222222222222, "status": 0.75, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 795}, "4": {"facts_total": 58, "facts_retained": 15, "retention_rate": 0.25862068965517243, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.2222222222222222, "status": 0.75, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 795}, "5": {"facts_total": 58, "facts_retained": 15, "retention_rate": 0.25862068965517243, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.2222222222222222, "status": 0.75, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 795}, "6": {"facts_total": 58, "facts_retained": 15, "retention_rate": 0.25862068965517243, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.2222222222222222, "status": 0.75, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 795}, "7": {"facts_total": 58, "facts_retained": 15, "retention_rate": 0.25862068965517243, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.2222222222222222, "status": 0.75, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 795}, "8": {"facts_total": 58, "facts_retained": 15, "retention_rate": 0.25862068965517243, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.2222222222222222, "status": 0.75, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 795}}}, "reference_key_id": "keys/synthetic/syn_03003_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "0d4ead41380940db4359555678482b867e828b27a6b33c483fba125924e02f66", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "cf20f60568372e3d77edca82", "attempt_id": "8ca6c087c9a8", "run_group_id": "E4_core:A5|repairable|qwen3b:h8", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T06:49:51.371847Z", "task_family": "synthetic", "document_id": "syn_03003_2000", "document_hash": "dddca1a38e3224870fcfae927c701c400fd885c19123ea840ee06a76a140f407", "question_id": "q06_negation", "split": "test", "cluster_id": "syn_03003_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 8, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1866, "query_tokens": 446, "message_tokens": 362, "message_utf8_bytes": 859, "cache_tensor_bytes": null, "serialized_payload_bytes": 859, "side_information_bytes": 0, "fetched_bytes": 3556, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 136245.71970908437, "hops": {"1": 22633.649082970805, "2": 15950.975792016834, "3": 15936.411041999236, "4": 16089.841875014827, "5": 17749.737125006504, "6": 15981.693333014846, "7": 15952.465916983783, "8": 15950.945542077534}, "receiver_prefill_ms": 308.6596670327708, "ttft_ms": 77.9382080072537, "decode_ms": 126.42458395566791}, "total_wall_ms": 136759.4889170723, "ttft_ms": 77.9382080072537, "input_usage": {"lookups": 7}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "no", "evidence_artifact": {"ledger": {"facts_total": 58, "facts_retained": 15, "retention_rate": 0.25862068965517243, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.2222222222222222, "status": 0.75, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 795}, "ledger_by_hop": {"1": {"facts_total": 58, "facts_retained": 15, "retention_rate": 0.25862068965517243, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.2222222222222222, "status": 0.75, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 766}, "2": {"facts_total": 58, "facts_retained": 15, "retention_rate": 0.25862068965517243, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.2222222222222222, "status": 0.75, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 795}, "3": {"facts_total": 58, "facts_retained": 15, "retention_rate": 0.25862068965517243, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.2222222222222222, "status": 0.75, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 795}, "4": {"facts_total": 58, "facts_retained": 15, "retention_rate": 0.25862068965517243, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.2222222222222222, "status": 0.75, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 795}, "5": {"facts_total": 58, "facts_retained": 15, "retention_rate": 0.25862068965517243, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.2222222222222222, "status": 0.75, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 795}, "6": {"facts_total": 58, "facts_retained": 15, "retention_rate": 0.25862068965517243, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.2222222222222222, "status": 0.75, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 795}, "7": {"facts_total": 58, "facts_retained": 15, "retention_rate": 0.25862068965517243, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.2222222222222222, "status": 0.75, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 795}, "8": {"facts_total": 58, "facts_retained": 15, "retention_rate": 0.25862068965517243, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.2222222222222222, "status": 0.75, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 795}}}, "reference_key_id": "keys/synthetic/syn_03003_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "0d4ead41380940db4359555678482b867e828b27a6b33c483fba125924e02f66", "output_hash": "9390298f3fb0c5b160498935d79cb139aef28e1c47358b4bbba61862b9c26e59"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "3d11fbc98406ea9f9b661dc8", "attempt_id": "65919ffaa7cb", "run_group_id": "E4_core:A5|repairable|qwen3b:h8", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T06:49:51.904515Z", "task_family": "synthetic", "document_id": "syn_03003_2000", "document_hash": "dddca1a38e3224870fcfae927c701c400fd885c19123ea840ee06a76a140f407", "question_id": "q07_negation", "split": "test", "cluster_id": "syn_03003_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 8, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1866, "query_tokens": 443, "message_tokens": 362, "message_utf8_bytes": 859, "cache_tensor_bytes": null, "serialized_payload_bytes": 859, "side_information_bytes": 0, "fetched_bytes": 3556, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 136245.71970908437, "hops": {"1": 22633.649082970805, "2": 15950.975792016834, "3": 15936.411041999236, "4": 16089.841875014827, "5": 17749.737125006504, "6": 15981.693333014846, "7": 15952.465916983783, "8": 15950.945542077534}, "receiver_prefill_ms": 277.1917500067502, "ttft_ms": 41.69320804066956, "decode_ms": 30.81833408214152}, "total_wall_ms": 136595.9666670533, "ttft_ms": 41.69320804066956, "input_usage": {"lookups": 7}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 58, "facts_retained": 15, "retention_rate": 0.25862068965517243, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.2222222222222222, "status": 0.75, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 795}, "ledger_by_hop": {"1": {"facts_total": 58, "facts_retained": 15, "retention_rate": 0.25862068965517243, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.2222222222222222, "status": 0.75, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 766}, "2": {"facts_total": 58, "facts_retained": 15, "retention_rate": 0.25862068965517243, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.2222222222222222, "status": 0.75, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 795}, "3": {"facts_total": 58, "facts_retained": 15, "retention_rate": 0.25862068965517243, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.2222222222222222, "status": 0.75, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 795}, "4": {"facts_total": 58, "facts_retained": 15, "retention_rate": 0.25862068965517243, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.2222222222222222, "status": 0.75, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 795}, "5": {"facts_total": 58, "facts_retained": 15, "retention_rate": 0.25862068965517243, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.2222222222222222, "status": 0.75, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 795}, "6": {"facts_total": 58, "facts_retained": 15, "retention_rate": 0.25862068965517243, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.2222222222222222, "status": 0.75, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 795}, "7": {"facts_total": 58, "facts_retained": 15, "retention_rate": 0.25862068965517243, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.2222222222222222, "status": 0.75, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 795}, "8": {"facts_total": 58, "facts_retained": 15, "retention_rate": 0.25862068965517243, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.2222222222222222, "status": 0.75, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 795}}}, "reference_key_id": "keys/synthetic/syn_03003_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "0d4ead41380940db4359555678482b867e828b27a6b33c483fba125924e02f66", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "cc2092e403b97fec3a880296", "attempt_id": "1f8886fb222d", "run_group_id": "E4_core:A5|repairable|qwen3b:h8", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T06:49:52.273020Z", "task_family": "synthetic", "document_id": "syn_03003_2000", "document_hash": "dddca1a38e3224870fcfae927c701c400fd885c19123ea840ee06a76a140f407", "question_id": "q08_revision", "split": "test", "cluster_id": "syn_03003_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 8, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1866, "query_tokens": 461, "message_tokens": 362, "message_utf8_bytes": 859, "cache_tensor_bytes": null, "serialized_payload_bytes": 859, "side_information_bytes": 0, "fetched_bytes": 3556, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 136245.71970908437, "hops": {"1": 22633.649082970805, "2": 15950.975792016834, "3": 15936.411041999236, "4": 16089.841875014827, "5": 17749.737125006504, "6": 15981.693333014846, "7": 15952.465916983783, "8": 15950.945542077534}, "receiver_prefill_ms": 309.87737502437085, "ttft_ms": 35.7707500224933, "decode_ms": 192.7852500230074}, "total_wall_ms": 136784.70933402423, "ttft_ms": 35.7707500224933, "input_usage": {"lookups": 7}, "output_usage": {"output_tokens": 6}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "911925", "evidence_artifact": {"ledger": {"facts_total": 58, "facts_retained": 15, "retention_rate": 0.25862068965517243, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.2222222222222222, "status": 0.75, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 795}, "ledger_by_hop": {"1": {"facts_total": 58, "facts_retained": 15, "retention_rate": 0.25862068965517243, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.2222222222222222, "status": 0.75, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 766}, "2": {"facts_total": 58, "facts_retained": 15, "retention_rate": 0.25862068965517243, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.2222222222222222, "status": 0.75, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 795}, "3": {"facts_total": 58, "facts_retained": 15, "retention_rate": 0.25862068965517243, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.2222222222222222, "status": 0.75, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 795}, "4": {"facts_total": 58, "facts_retained": 15, "retention_rate": 0.25862068965517243, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.2222222222222222, "status": 0.75, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 795}, "5": {"facts_total": 58, "facts_retained": 15, "retention_rate": 0.25862068965517243, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.2222222222222222, "status": 0.75, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 795}, "6": {"facts_total": 58, "facts_retained": 15, "retention_rate": 0.25862068965517243, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.2222222222222222, "status": 0.75, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 795}, "7": {"facts_total": 58, "facts_retained": 15, "retention_rate": 0.25862068965517243, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.2222222222222222, "status": 0.75, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 795}, "8": {"facts_total": 58, "facts_retained": 15, "retention_rate": 0.25862068965517243, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.2222222222222222, "status": 0.75, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 795}}}, "reference_key_id": "keys/synthetic/syn_03003_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "0d4ead41380940db4359555678482b867e828b27a6b33c483fba125924e02f66", "output_hash": "679a540dc4739f0ace623448b8fd925fb234f554838d7e87da223228ee81b112"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "5b7f2e18b38709914bd1837c", "attempt_id": "fa7740b7b14b", "run_group_id": "E4_core:A5|repairable|qwen3b:h8", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T06:49:52.829122Z", "task_family": "synthetic", "document_id": "syn_03003_2000", "document_hash": "dddca1a38e3224870fcfae927c701c400fd885c19123ea840ee06a76a140f407", "question_id": "q09_revision", "split": "test", "cluster_id": "syn_03003_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 8, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1866, "query_tokens": 454, "message_tokens": 362, "message_utf8_bytes": 859, "cache_tensor_bytes": null, "serialized_payload_bytes": 859, "side_information_bytes": 0, "fetched_bytes": 3556, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 136245.71970908437, "hops": {"1": 22633.649082970805, "2": 15950.975792016834, "3": 15936.411041999236, "4": 16089.841875014827, "5": 17749.737125006504, "6": 15981.693333014846, "7": 15952.465916983783, "8": 15950.945542077534}, "receiver_prefill_ms": 292.5959579879418, "ttft_ms": 56.050874991342425, "decode_ms": 434.28233300801367}, "total_wall_ms": 137029.28483416326, "ttft_ms": 56.050874991342425, "input_usage": {"lookups": 7}, "output_usage": {"output_tokens": 6}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "911925", "evidence_artifact": {"ledger": {"facts_total": 58, "facts_retained": 15, "retention_rate": 0.25862068965517243, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.2222222222222222, "status": 0.75, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 795}, "ledger_by_hop": {"1": {"facts_total": 58, "facts_retained": 15, "retention_rate": 0.25862068965517243, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.2222222222222222, "status": 0.75, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 766}, "2": {"facts_total": 58, "facts_retained": 15, "retention_rate": 0.25862068965517243, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.2222222222222222, "status": 0.75, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 795}, "3": {"facts_total": 58, "facts_retained": 15, "retention_rate": 0.25862068965517243, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.2222222222222222, "status": 0.75, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 795}, "4": {"facts_total": 58, "facts_retained": 15, "retention_rate": 0.25862068965517243, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.2222222222222222, "status": 0.75, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 795}, "5": {"facts_total": 58, "facts_retained": 15, "retention_rate": 0.25862068965517243, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.2222222222222222, "status": 0.75, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 795}, "6": {"facts_total": 58, "facts_retained": 15, "retention_rate": 0.25862068965517243, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.2222222222222222, "status": 0.75, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 795}, "7": {"facts_total": 58, "facts_retained": 15, "retention_rate": 0.25862068965517243, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.2222222222222222, "status": 0.75, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 795}, "8": {"facts_total": 58, "facts_retained": 15, "retention_rate": 0.25862068965517243, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.2222222222222222, "status": 0.75, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 795}}}, "reference_key_id": "keys/synthetic/syn_03003_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "0d4ead41380940db4359555678482b867e828b27a6b33c483fba125924e02f66", "output_hash": "679a540dc4739f0ace623448b8fd925fb234f554838d7e87da223228ee81b112"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "1deb186483209fc52123e1c3", "attempt_id": "11fff719199a", "run_group_id": "E4_core:A5|repairable|qwen3b:h8", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T06:49:53.631794Z", "task_family": "synthetic", "document_id": "syn_03003_2000", "document_hash": "dddca1a38e3224870fcfae927c701c400fd885c19123ea840ee06a76a140f407", "question_id": "q10_temporal", "split": "test", "cluster_id": "syn_03003_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 8, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1866, "query_tokens": 461, "message_tokens": 362, "message_utf8_bytes": 859, "cache_tensor_bytes": null, "serialized_payload_bytes": 859, "side_information_bytes": 0, "fetched_bytes": 3556, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 136245.71970908437, "hops": {"1": 22633.649082970805, "2": 15950.975792016834, "3": 15936.411041999236, "4": 16089.841875014827, "5": 17749.737125006504, "6": 15981.693333014846, "7": 15952.465916983783, "8": 15950.945542077534}, "receiver_prefill_ms": 314.25362499430776, "ttft_ms": 39.420791901648045, "decode_ms": 27.866374934092164}, "total_wall_ms": 136627.7995420387, "ttft_ms": 39.420791901648045, "input_usage": {"lookups": 7}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 58, "facts_retained": 15, "retention_rate": 0.25862068965517243, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.2222222222222222, "status": 0.75, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 795}, "ledger_by_hop": {"1": {"facts_total": 58, "facts_retained": 15, "retention_rate": 0.25862068965517243, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.2222222222222222, "status": 0.75, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 766}, "2": {"facts_total": 58, "facts_retained": 15, "retention_rate": 0.25862068965517243, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.2222222222222222, "status": 0.75, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 795}, "3": {"facts_total": 58, "facts_retained": 15, "retention_rate": 0.25862068965517243, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.2222222222222222, "status": 0.75, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 795}, "4": {"facts_total": 58, "facts_retained": 15, "retention_rate": 0.25862068965517243, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.2222222222222222, "status": 0.75, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 795}, "5": {"facts_total": 58, "facts_retained": 15, "retention_rate": 0.25862068965517243, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.2222222222222222, "status": 0.75, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 795}, "6": {"facts_total": 58, "facts_retained": 15, "retention_rate": 0.25862068965517243, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.2222222222222222, "status": 0.75, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 795}, "7": {"facts_total": 58, "facts_retained": 15, "retention_rate": 0.25862068965517243, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.2222222222222222, "status": 0.75, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 795}, "8": {"facts_total": 58, "facts_retained": 15, "retention_rate": 0.25862068965517243, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.2222222222222222, "status": 0.75, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 795}}}, "reference_key_id": "keys/synthetic/syn_03003_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "0d4ead41380940db4359555678482b867e828b27a6b33c483fba125924e02f66", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "4653a70556bb021254098148", "attempt_id": "93e2d72cd4be", "run_group_id": "E4_core:A5|repairable|qwen3b:h8", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T06:49:54.030747Z", "task_family": "synthetic", "document_id": "syn_03003_2000", "document_hash": "dddca1a38e3224870fcfae927c701c400fd885c19123ea840ee06a76a140f407", "question_id": "q11_aggregation", "split": "test", "cluster_id": "syn_03003_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 8, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1866, "query_tokens": 455, "message_tokens": 362, "message_utf8_bytes": 859, "cache_tensor_bytes": null, "serialized_payload_bytes": 859, "side_information_bytes": 0, "fetched_bytes": 3556, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 136245.71970908437, "hops": {"1": 22633.649082970805, "2": 15950.975792016834, "3": 15936.411041999236, "4": 16089.841875014827, "5": 17749.737125006504, "6": 15981.693333014846, "7": 15952.465916983783, "8": 15950.945542077534}, "receiver_prefill_ms": 324.0102920681238, "ttft_ms": 42.57108399178833, "decode_ms": 23.438042029738426}, "total_wall_ms": 136636.31366810296, "ttft_ms": 42.57108399178833, "input_usage": {"lookups": 7}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 58, "facts_retained": 15, "retention_rate": 0.25862068965517243, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.2222222222222222, "status": 0.75, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 795}, "ledger_by_hop": {"1": {"facts_total": 58, "facts_retained": 15, "retention_rate": 0.25862068965517243, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.2222222222222222, "status": 0.75, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 766}, "2": {"facts_total": 58, "facts_retained": 15, "retention_rate": 0.25862068965517243, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.2222222222222222, "status": 0.75, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 795}, "3": {"facts_total": 58, "facts_retained": 15, "retention_rate": 0.25862068965517243, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.2222222222222222, "status": 0.75, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 795}, "4": {"facts_total": 58, "facts_retained": 15, "retention_rate": 0.25862068965517243, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.2222222222222222, "status": 0.75, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 795}, "5": {"facts_total": 58, "facts_retained": 15, "retention_rate": 0.25862068965517243, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.2222222222222222, "status": 0.75, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 795}, "6": {"facts_total": 58, "facts_retained": 15, "retention_rate": 0.25862068965517243, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.2222222222222222, "status": 0.75, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 795}, "7": {"facts_total": 58, "facts_retained": 15, "retention_rate": 0.25862068965517243, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.2222222222222222, "status": 0.75, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 795}, "8": {"facts_total": 58, "facts_retained": 15, "retention_rate": 0.25862068965517243, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.2222222222222222, "status": 0.75, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 795}}}, "reference_key_id": "keys/synthetic/syn_03003_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "0d4ead41380940db4359555678482b867e828b27a6b33c483fba125924e02f66", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "e53bb432f1649e5a512138d1", "attempt_id": "eeb489d24963", "run_group_id": "E4_core:A5|repairable|qwen3b:h8", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T06:49:54.439309Z", "task_family": "synthetic", "document_id": "syn_03003_2000", "document_hash": "dddca1a38e3224870fcfae927c701c400fd885c19123ea840ee06a76a140f407", "question_id": "q12_unit_conversion", "split": "test", "cluster_id": "syn_03003_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 8, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1866, "query_tokens": 452, "message_tokens": 362, "message_utf8_bytes": 859, "cache_tensor_bytes": null, "serialized_payload_bytes": 859, "side_information_bytes": 0, "fetched_bytes": 3556, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 136245.71970908437, "hops": {"1": 22633.649082970805, "2": 15950.975792016834, "3": 15936.411041999236, "4": 16089.841875014827, "5": 17749.737125006504, "6": 15981.693333014846, "7": 15952.465916983783, "8": 15950.945542077534}, "receiver_prefill_ms": 280.8266670908779, "ttft_ms": 54.58262504544109, "decode_ms": 400.39170894306153}, "total_wall_ms": 136982.11966711096, "ttft_ms": 54.58262504544109, "input_usage": {"lookups": 7}, "output_usage": {"output_tokens": 5}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "17430", "evidence_artifact": {"ledger": {"facts_total": 58, "facts_retained": 15, "retention_rate": 0.25862068965517243, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.2222222222222222, "status": 0.75, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 795}, "ledger_by_hop": {"1": {"facts_total": 58, "facts_retained": 15, "retention_rate": 0.25862068965517243, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.2222222222222222, "status": 0.75, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 766}, "2": {"facts_total": 58, "facts_retained": 15, "retention_rate": 0.25862068965517243, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.2222222222222222, "status": 0.75, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 795}, "3": {"facts_total": 58, "facts_retained": 15, "retention_rate": 0.25862068965517243, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.2222222222222222, "status": 0.75, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 795}, "4": {"facts_total": 58, "facts_retained": 15, "retention_rate": 0.25862068965517243, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.2222222222222222, "status": 0.75, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 795}, "5": {"facts_total": 58, "facts_retained": 15, "retention_rate": 0.25862068965517243, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.2222222222222222, "status": 0.75, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 795}, "6": {"facts_total": 58, "facts_retained": 15, "retention_rate": 0.25862068965517243, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.2222222222222222, "status": 0.75, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 795}, "7": {"facts_total": 58, "facts_retained": 15, "retention_rate": 0.25862068965517243, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.2222222222222222, "status": 0.75, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 795}, "8": {"facts_total": 58, "facts_retained": 15, "retention_rate": 0.25862068965517243, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.2222222222222222, "status": 0.75, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 795}}}, "reference_key_id": "keys/synthetic/syn_03003_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "0d4ead41380940db4359555678482b867e828b27a6b33c483fba125924e02f66", "output_hash": "2575eaf3b748efc209b676518766751eb038c88f74578951345818bfe7f7194b"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "c1fdc95be4ea7b58aba62b29", "attempt_id": "6bb5c9ea1ecd", "run_group_id": "E4_core:A5|repairable|qwen3b:h8", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T06:49:55.193781Z", "task_family": "synthetic", "document_id": "syn_03003_2000", "document_hash": "dddca1a38e3224870fcfae927c701c400fd885c19123ea840ee06a76a140f407", "question_id": "q13_unanswerable", "split": "test", "cluster_id": "syn_03003_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 8, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1866, "query_tokens": 456, "message_tokens": 362, "message_utf8_bytes": 859, "cache_tensor_bytes": null, "serialized_payload_bytes": 859, "side_information_bytes": 0, "fetched_bytes": 3556, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 136245.71970908437, "hops": {"1": 22633.649082970805, "2": 15950.975792016834, "3": 15936.411041999236, "4": 16089.841875014827, "5": 17749.737125006504, "6": 15981.693333014846, "7": 15952.465916983783, "8": 15950.945542077534}, "receiver_prefill_ms": 329.06533300410956, "ttft_ms": 39.87912507727742, "decode_ms": 25.426791980862617}, "total_wall_ms": 136640.66100108903, "ttft_ms": 39.87912507727742, "input_usage": {"lookups": 7}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 58, "facts_retained": 15, "retention_rate": 0.25862068965517243, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.2222222222222222, "status": 0.75, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 795}, "ledger_by_hop": {"1": {"facts_total": 58, "facts_retained": 15, "retention_rate": 0.25862068965517243, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.2222222222222222, "status": 0.75, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 766}, "2": {"facts_total": 58, "facts_retained": 15, "retention_rate": 0.25862068965517243, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.2222222222222222, "status": 0.75, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 795}, "3": {"facts_total": 58, "facts_retained": 15, "retention_rate": 0.25862068965517243, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.2222222222222222, "status": 0.75, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 795}, "4": {"facts_total": 58, "facts_retained": 15, "retention_rate": 0.25862068965517243, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.2222222222222222, "status": 0.75, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 795}, "5": {"facts_total": 58, "facts_retained": 15, "retention_rate": 0.25862068965517243, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.2222222222222222, "status": 0.75, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 795}, "6": {"facts_total": 58, "facts_retained": 15, "retention_rate": 0.25862068965517243, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.2222222222222222, "status": 0.75, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 795}, "7": {"facts_total": 58, "facts_retained": 15, "retention_rate": 0.25862068965517243, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.2222222222222222, "status": 0.75, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 795}, "8": {"facts_total": 58, "facts_retained": 15, "retention_rate": 0.25862068965517243, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.2222222222222222, "status": 0.75, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 795}}}, "reference_key_id": "keys/synthetic/syn_03003_2000.json", "task_scores": {"correct": 1.0, "abstained": 1.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "0d4ead41380940db4359555678482b867e828b27a6b33c483fba125924e02f66", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "1fc18bd8d0b02140e2aa76fd", "attempt_id": "9b7b7512ba70", "run_group_id": "E4_core:A5|repairable|qwen3b:h8", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T06:49:55.606320Z", "task_family": "synthetic", "document_id": "syn_03003_2000", "document_hash": "dddca1a38e3224870fcfae927c701c400fd885c19123ea840ee06a76a140f407", "question_id": "q14_unanswerable", "split": "test", "cluster_id": "syn_03003_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 8, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1866, "query_tokens": 457, "message_tokens": 362, "message_utf8_bytes": 859, "cache_tensor_bytes": null, "serialized_payload_bytes": 859, "side_information_bytes": 0, "fetched_bytes": 3556, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 136245.71970908437, "hops": {"1": 22633.649082970805, "2": 15950.975792016834, "3": 15936.411041999236, "4": 16089.841875014827, "5": 17749.737125006504, "6": 15981.693333014846, "7": 15952.465916983783, "8": 15950.945542077534}, "receiver_prefill_ms": 324.5586659759283, "ttft_ms": 38.07329200208187, "decode_ms": 21.4134999550879}, "total_wall_ms": 136630.22166711744, "ttft_ms": 38.07329200208187, "input_usage": {"lookups": 7}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 58, "facts_retained": 15, "retention_rate": 0.25862068965517243, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.2222222222222222, "status": 0.75, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 795}, "ledger_by_hop": {"1": {"facts_total": 58, "facts_retained": 15, "retention_rate": 0.25862068965517243, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.2222222222222222, "status": 0.75, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 766}, "2": {"facts_total": 58, "facts_retained": 15, "retention_rate": 0.25862068965517243, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.2222222222222222, "status": 0.75, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 795}, "3": {"facts_total": 58, "facts_retained": 15, "retention_rate": 0.25862068965517243, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.2222222222222222, "status": 0.75, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 795}, "4": {"facts_total": 58, "facts_retained": 15, "retention_rate": 0.25862068965517243, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.2222222222222222, "status": 0.75, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 795}, "5": {"facts_total": 58, "facts_retained": 15, "retention_rate": 0.25862068965517243, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.2222222222222222, "status": 0.75, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 795}, "6": {"facts_total": 58, "facts_retained": 15, "retention_rate": 0.25862068965517243, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.2222222222222222, "status": 0.75, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 795}, "7": {"facts_total": 58, "facts_retained": 15, "retention_rate": 0.25862068965517243, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.2222222222222222, "status": 0.75, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 795}, "8": {"facts_total": 58, "facts_retained": 15, "retention_rate": 0.25862068965517243, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.2222222222222222, "status": 0.75, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 795}}}, "reference_key_id": "keys/synthetic/syn_03003_2000.json", "task_scores": {"correct": 1.0, "abstained": 1.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "0d4ead41380940db4359555678482b867e828b27a6b33c483fba125924e02f66", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "1b25e4942f3a4ba248b9b97c", "attempt_id": "a09ffff52906", "run_group_id": "E4_core:A4|closed|qwen3b-llama3b:h1", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T06:50:18.579461Z", "task_family": "synthetic", "document_id": "syn_03003_2000", "document_hash": "dddca1a38e3224870fcfae927c701c400fd885c19123ea840ee06a76a140f407", "question_id": "q00_single_fact", "split": "test", "cluster_id": "syn_03003_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 1, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1866, "query_tokens": 594, "message_tokens": 512, "message_utf8_bytes": 1200, "cache_tensor_bytes": null, "serialized_payload_bytes": 1200, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 22569.42216597963, "hops": {"1": 22569.42216597963}, "receiver_prefill_ms": 514.07079200726, "ttft_ms": 38.53045799769461, "decode_ms": 327.0804160274565}, "total_wall_ms": 23449.859373969957, "ttft_ms": 38.53045799769461, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 10}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "GR-79S-5839", "evidence_artifact": {"ledger": {"facts_total": 58, "facts_retained": 32, "retention_rate": 0.5517241379310345, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.7142857142857143, "amount": 0.6666666666666666, "status": 0.6666666666666666, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1200}, "ledger_by_hop": {"1": {"facts_total": 58, "facts_retained": 32, "retention_rate": 0.5517241379310345, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.7142857142857143, "amount": 0.6666666666666666, "status": 0.6666666666666666, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1200}}}, "reference_key_id": "keys/synthetic/syn_03003_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "6d2a2bad9425191f4568f7eb115d63c953e27080a86953f62a982ad73340e770", "output_hash": "720cc40cd69e85e9728f6bea00b17a95ca07eee643e143a131840763958303cd"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "3bd86e225aa7bf89a542014f", "attempt_id": "3951995b1817", "run_group_id": "E4_core:A4|closed|qwen3b-llama3b:h1", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T06:50:19.479343Z", "task_family": "synthetic", "document_id": "syn_03003_2000", "document_hash": "dddca1a38e3224870fcfae927c701c400fd885c19123ea840ee06a76a140f407", "question_id": "q01_single_fact", "split": "test", "cluster_id": "syn_03003_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 1, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1866, "query_tokens": 598, "message_tokens": 512, "message_utf8_bytes": 1200, "cache_tensor_bytes": null, "serialized_payload_bytes": 1200, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 22569.42216597963, "hops": {"1": 22569.42216597963}, "receiver_prefill_ms": 374.20612503774464, "ttft_ms": 17.576207988895476, "decode_ms": 672.3187500610948}, "total_wall_ms": 23634.20120801311, "ttft_ms": 17.576207988895476, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 11}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "Lot 344-B, Témiscaming", "evidence_artifact": {"ledger": {"facts_total": 58, "facts_retained": 32, "retention_rate": 0.5517241379310345, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.7142857142857143, "amount": 0.6666666666666666, "status": 0.6666666666666666, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1200}, "ledger_by_hop": {"1": {"facts_total": 58, "facts_retained": 32, "retention_rate": 0.5517241379310345, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.7142857142857143, "amount": 0.6666666666666666, "status": 0.6666666666666666, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1200}}}, "reference_key_id": "keys/synthetic/syn_03003_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 0.8571428571428571}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "6d2a2bad9425191f4568f7eb115d63c953e27080a86953f62a982ad73340e770", "output_hash": "0197d491a5a04a142dc8e372360735a1d41757e7a5aea7e81f23b0e1ef6ee010"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "a764a96c6644807989f61818", "attempt_id": "e6967aa785a4", "run_group_id": "E4_core:A4|closed|qwen3b-llama3b:h1", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T06:50:20.561609Z", "task_family": "synthetic", "document_id": "syn_03003_2000", "document_hash": "dddca1a38e3224870fcfae927c701c400fd885c19123ea840ee06a76a140f407", "question_id": "q02_exact_copy", "split": "test", "cluster_id": "syn_03003_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 1, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1866, "query_tokens": 598, "message_tokens": 512, "message_utf8_bytes": 1200, "cache_tensor_bytes": null, "serialized_payload_bytes": 1200, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 22569.42216597963, "hops": {"1": 22569.42216597963}, "receiver_prefill_ms": 417.3753340728581, "ttft_ms": 28.22720806580037, "decode_ms": 36.01245803292841}, "total_wall_ms": 23051.615081960335, "ttft_ms": 28.22720806580037, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 58, "facts_retained": 32, "retention_rate": 0.5517241379310345, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.7142857142857143, "amount": 0.6666666666666666, "status": 0.6666666666666666, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1200}, "ledger_by_hop": {"1": {"facts_total": 58, "facts_retained": 32, "retention_rate": 0.5517241379310345, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.7142857142857143, "amount": 0.6666666666666666, "status": 0.6666666666666666, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1200}}}, "reference_key_id": "keys/synthetic/syn_03003_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "6d2a2bad9425191f4568f7eb115d63c953e27080a86953f62a982ad73340e770", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "88b6a27f6b3c7a1a039f7f91", "attempt_id": "b620bf395102", "run_group_id": "E4_core:A4|closed|qwen3b-llama3b:h1", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T06:50:21.061884Z", "task_family": "synthetic", "document_id": "syn_03003_2000", "document_hash": "dddca1a38e3224870fcfae927c701c400fd885c19123ea840ee06a76a140f407", "question_id": "q03_exact_copy", "split": "test", "cluster_id": "syn_03003_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 1, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1866, "query_tokens": 593, "message_tokens": 512, "message_utf8_bytes": 1200, "cache_tensor_bytes": null, "serialized_payload_bytes": 1200, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 22569.42216597963, "hops": {"1": 22569.42216597963}, "receiver_prefill_ms": 379.33820800390095, "ttft_ms": 18.763125059194863, "decode_ms": 337.36008300911635}, "total_wall_ms": 23305.572082987055, "ttft_ms": 18.763125059194863, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 6}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "Wilhelmina Achterberg", "evidence_artifact": {"ledger": {"facts_total": 58, "facts_retained": 32, "retention_rate": 0.5517241379310345, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.7142857142857143, "amount": 0.6666666666666666, "status": 0.6666666666666666, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1200}, "ledger_by_hop": {"1": {"facts_total": 58, "facts_retained": 32, "retention_rate": 0.5517241379310345, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.7142857142857143, "amount": 0.6666666666666666, "status": 0.6666666666666666, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1200}}}, "reference_key_id": "keys/synthetic/syn_03003_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "6d2a2bad9425191f4568f7eb115d63c953e27080a86953f62a982ad73340e770", "output_hash": "2a62606853c315bba8769371d195e3ab3d6a8cd7ef959205575c5119740850f8"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "8590768d611536a9872f2732", "attempt_id": "a422dd615584", "run_group_id": "E4_core:A4|closed|qwen3b-llama3b:h1", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T06:50:21.815226Z", "task_family": "synthetic", "document_id": "syn_03003_2000", "document_hash": "dddca1a38e3224870fcfae927c701c400fd885c19123ea840ee06a76a140f407", "question_id": "q04_multi_hop", "split": "test", "cluster_id": "syn_03003_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 1, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1866, "query_tokens": 602, "message_tokens": 512, "message_utf8_bytes": 1200, "cache_tensor_bytes": null, "serialized_payload_bytes": 1200, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 22569.42216597963, "hops": {"1": 22569.42216597963}, "receiver_prefill_ms": 538.0569170229137, "ttft_ms": 37.55262494087219, "decode_ms": 44.14100002031773}, "total_wall_ms": 23189.79020707775, "ttft_ms": 37.55262494087219, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 58, "facts_retained": 32, "retention_rate": 0.5517241379310345, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.7142857142857143, "amount": 0.6666666666666666, "status": 0.6666666666666666, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1200}, "ledger_by_hop": {"1": {"facts_total": 58, "facts_retained": 32, "retention_rate": 0.5517241379310345, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.7142857142857143, "amount": 0.6666666666666666, "status": 0.6666666666666666, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1200}}}, "reference_key_id": "keys/synthetic/syn_03003_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "6d2a2bad9425191f4568f7eb115d63c953e27080a86953f62a982ad73340e770", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "9dd7f002d1e1ebe5dbd071ab", "attempt_id": "4d21a5406ac8", "run_group_id": "E4_core:A4|closed|qwen3b-llama3b:h1", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T06:50:22.454492Z", "task_family": "synthetic", "document_id": "syn_03003_2000", "document_hash": "dddca1a38e3224870fcfae927c701c400fd885c19123ea840ee06a76a140f407", "question_id": "q05_multi_hop", "split": "test", "cluster_id": "syn_03003_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 1, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1866, "query_tokens": 604, "message_tokens": 512, "message_utf8_bytes": 1200, "cache_tensor_bytes": null, "serialized_payload_bytes": 1200, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 22569.42216597963, "hops": {"1": 22569.42216597963}, "receiver_prefill_ms": 421.0453330306336, "ttft_ms": 33.234833041206, "decode_ms": 421.90033302176744}, "total_wall_ms": 23446.56649895478, "ttft_ms": 33.234833041206, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 10}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "NK-95Z-2165", "evidence_artifact": {"ledger": {"facts_total": 58, "facts_retained": 32, "retention_rate": 0.5517241379310345, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.7142857142857143, "amount": 0.6666666666666666, "status": 0.6666666666666666, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1200}, "ledger_by_hop": {"1": {"facts_total": 58, "facts_retained": 32, "retention_rate": 0.5517241379310345, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.7142857142857143, "amount": 0.6666666666666666, "status": 0.6666666666666666, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1200}}}, "reference_key_id": "keys/synthetic/syn_03003_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "6d2a2bad9425191f4568f7eb115d63c953e27080a86953f62a982ad73340e770", "output_hash": "bd8f809b02304c6b9dc7c1e116ecccc2328c976a5c071e89a94d661b48e73aa7"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "e0fde4cfaa230a1a419f1fed", "attempt_id": "467e12f86569", "run_group_id": "E4_core:A4|closed|qwen3b-llama3b:h1", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T06:50:23.350131Z", "task_family": "synthetic", "document_id": "syn_03003_2000", "document_hash": "dddca1a38e3224870fcfae927c701c400fd885c19123ea840ee06a76a140f407", "question_id": "q06_negation", "split": "test", "cluster_id": "syn_03003_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 1, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1866, "query_tokens": 596, "message_tokens": 512, "message_utf8_bytes": 1200, "cache_tensor_bytes": null, "serialized_payload_bytes": 1200, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 22569.42216597963, "hops": {"1": 22569.42216597963}, "receiver_prefill_ms": 528.4955829847604, "ttft_ms": 38.29995810519904, "decode_ms": 42.59066598024219}, "total_wall_ms": 23179.596206871793, "ttft_ms": 38.29995810519904, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "yes", "evidence_artifact": {"ledger": {"facts_total": 58, "facts_retained": 32, "retention_rate": 0.5517241379310345, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.7142857142857143, "amount": 0.6666666666666666, "status": 0.6666666666666666, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1200}, "ledger_by_hop": {"1": {"facts_total": 58, "facts_retained": 32, "retention_rate": 0.5517241379310345, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.7142857142857143, "amount": 0.6666666666666666, "status": 0.6666666666666666, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1200}}}, "reference_key_id": "keys/synthetic/syn_03003_2000.json", "task_scores": {"correct": 0.0, "abstained": 0.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "6d2a2bad9425191f4568f7eb115d63c953e27080a86953f62a982ad73340e770", "output_hash": "8a798890fe93817163b10b5f7bd2ca4d25d84c52739a645a889c173eee7d9d3d"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "b91297a41da0ee3c977c11eb", "attempt_id": "924bd33eea56", "run_group_id": "E4_core:A4|closed|qwen3b-llama3b:h1", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T06:50:23.977635Z", "task_family": "synthetic", "document_id": "syn_03003_2000", "document_hash": "dddca1a38e3224870fcfae927c701c400fd885c19123ea840ee06a76a140f407", "question_id": "q07_negation", "split": "test", "cluster_id": "syn_03003_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 1, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1866, "query_tokens": 593, "message_tokens": 512, "message_utf8_bytes": 1200, "cache_tensor_bytes": null, "serialized_payload_bytes": 1200, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 22569.42216597963, "hops": {"1": 22569.42216597963}, "receiver_prefill_ms": 400.08108399342746, "ttft_ms": 34.092457965016365, "decode_ms": 39.57695793360472}, "total_wall_ms": 23043.860665988177, "ttft_ms": 34.092457965016365, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "no", "evidence_artifact": {"ledger": {"facts_total": 58, "facts_retained": 32, "retention_rate": 0.5517241379310345, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.7142857142857143, "amount": 0.6666666666666666, "status": 0.6666666666666666, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1200}, "ledger_by_hop": {"1": {"facts_total": 58, "facts_retained": 32, "retention_rate": 0.5517241379310345, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.7142857142857143, "amount": 0.6666666666666666, "status": 0.6666666666666666, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1200}}}, "reference_key_id": "keys/synthetic/syn_03003_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "6d2a2bad9425191f4568f7eb115d63c953e27080a86953f62a982ad73340e770", "output_hash": "9390298f3fb0c5b160498935d79cb139aef28e1c47358b4bbba61862b9c26e59"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "9798fc131e18ace813fa3ab7", "attempt_id": "44ff2850121b", "run_group_id": "E4_core:A4|closed|qwen3b-llama3b:h1", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T06:50:24.470160Z", "task_family": "synthetic", "document_id": "syn_03003_2000", "document_hash": "dddca1a38e3224870fcfae927c701c400fd885c19123ea840ee06a76a140f407", "question_id": "q08_revision", "split": "test", "cluster_id": "syn_03003_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 1, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1866, "query_tokens": 611, "message_tokens": 512, "message_utf8_bytes": 1200, "cache_tensor_bytes": null, "serialized_payload_bytes": 1200, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 22569.42216597963, "hops": {"1": 22569.42216597963}, "receiver_prefill_ms": 390.3571660630405, "ttft_ms": 56.697041960433125, "decode_ms": 477.7939170598984}, "total_wall_ms": 23495.09574903641, "ttft_ms": 56.697041960433125, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 7}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "911,925", "evidence_artifact": {"ledger": {"facts_total": 58, "facts_retained": 32, "retention_rate": 0.5517241379310345, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.7142857142857143, "amount": 0.6666666666666666, "status": 0.6666666666666666, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1200}, "ledger_by_hop": {"1": {"facts_total": 58, "facts_retained": 32, "retention_rate": 0.5517241379310345, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.7142857142857143, "amount": 0.6666666666666666, "status": 0.6666666666666666, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1200}}}, "reference_key_id": "keys/synthetic/syn_03003_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "6d2a2bad9425191f4568f7eb115d63c953e27080a86953f62a982ad73340e770", "output_hash": "86be78720fc4b79f93e603e7d9e4861fcd25102612f73209fffc17d787056007"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "525b0e037655371bdc6863af", "attempt_id": "2c016e4cabfe", "run_group_id": "E4_core:A4|closed|qwen3b-llama3b:h1", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T06:50:25.413933Z", "task_family": "synthetic", "document_id": "syn_03003_2000", "document_hash": "dddca1a38e3224870fcfae927c701c400fd885c19123ea840ee06a76a140f407", "question_id": "q09_revision", "split": "test", "cluster_id": "syn_03003_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 1, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1866, "query_tokens": 604, "message_tokens": 512, "message_utf8_bytes": 1200, "cache_tensor_bytes": null, "serialized_payload_bytes": 1200, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 22569.42216597963, "hops": {"1": 22569.42216597963}, "receiver_prefill_ms": 395.8340829703957, "ttft_ms": 41.72095796093345, "decode_ms": 213.35108298808336}, "total_wall_ms": 23220.917623955756, "ttft_ms": 41.72095796093345, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 7}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "911,925", "evidence_artifact": {"ledger": {"facts_total": 58, "facts_retained": 32, "retention_rate": 0.5517241379310345, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.7142857142857143, "amount": 0.6666666666666666, "status": 0.6666666666666666, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1200}, "ledger_by_hop": {"1": {"facts_total": 58, "facts_retained": 32, "retention_rate": 0.5517241379310345, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.7142857142857143, "amount": 0.6666666666666666, "status": 0.6666666666666666, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1200}}}, "reference_key_id": "keys/synthetic/syn_03003_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "6d2a2bad9425191f4568f7eb115d63c953e27080a86953f62a982ad73340e770", "output_hash": "86be78720fc4b79f93e603e7d9e4861fcd25102612f73209fffc17d787056007"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "52b780f900a2bf07bb56f12c", "attempt_id": "1e1ccf209c9e", "run_group_id": "E4_core:A4|closed|qwen3b-llama3b:h1", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T06:50:26.082709Z", "task_family": "synthetic", "document_id": "syn_03003_2000", "document_hash": "dddca1a38e3224870fcfae927c701c400fd885c19123ea840ee06a76a140f407", "question_id": "q10_temporal", "split": "test", "cluster_id": "syn_03003_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 1, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1866, "query_tokens": 611, "message_tokens": 512, "message_utf8_bytes": 1200, "cache_tensor_bytes": null, "serialized_payload_bytes": 1200, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 22569.42216597963, "hops": {"1": 22569.42216597963}, "receiver_prefill_ms": 419.04908302240074, "ttft_ms": 57.55358305759728, "decode_ms": 436.1280830344185}, "total_wall_ms": 23482.86404099781, "ttft_ms": 57.55358305759728, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 6}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "Gamtiszen-2", "evidence_artifact": {"ledger": {"facts_total": 58, "facts_retained": 32, "retention_rate": 0.5517241379310345, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.7142857142857143, "amount": 0.6666666666666666, "status": 0.6666666666666666, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1200}, "ledger_by_hop": {"1": {"facts_total": 58, "facts_retained": 32, "retention_rate": 0.5517241379310345, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.7142857142857143, "amount": 0.6666666666666666, "status": 0.6666666666666666, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1200}}}, "reference_key_id": "keys/synthetic/syn_03003_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "6d2a2bad9425191f4568f7eb115d63c953e27080a86953f62a982ad73340e770", "output_hash": "32e3af428886434823d82102e1c482765eafdf6077fe8d3fba5752ec5e29826c"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "df052f466ab36f5f3ab3247b", "attempt_id": "260045587343", "run_group_id": "E4_core:A4|closed|qwen3b-llama3b:h1", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T06:50:27.013713Z", "task_family": "synthetic", "document_id": "syn_03003_2000", "document_hash": "dddca1a38e3224870fcfae927c701c400fd885c19123ea840ee06a76a140f407", "question_id": "q11_aggregation", "split": "test", "cluster_id": "syn_03003_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 1, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1866, "query_tokens": 605, "message_tokens": 512, "message_utf8_bytes": 1200, "cache_tensor_bytes": null, "serialized_payload_bytes": 1200, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 22569.42216597963, "hops": {"1": 22569.42216597963}, "receiver_prefill_ms": 391.89070800784975, "ttft_ms": 40.22204200737178, "decode_ms": 213.14545907080173}, "total_wall_ms": 23215.375790954567, "ttft_ms": 40.22204200737178, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 7}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "3397775", "evidence_artifact": {"ledger": {"facts_total": 58, "facts_retained": 32, "retention_rate": 0.5517241379310345, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.7142857142857143, "amount": 0.6666666666666666, "status": 0.6666666666666666, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1200}, "ledger_by_hop": {"1": {"facts_total": 58, "facts_retained": 32, "retention_rate": 0.5517241379310345, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.7142857142857143, "amount": 0.6666666666666666, "status": 0.6666666666666666, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1200}}}, "reference_key_id": "keys/synthetic/syn_03003_2000.json", "task_scores": {"correct": 0.0, "abstained": 0.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "6d2a2bad9425191f4568f7eb115d63c953e27080a86953f62a982ad73340e770", "output_hash": "e971a34f9c497d953f98fb387e189b1587b11a7f6725b2f81ac6d0f3ef8ac845"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "78f527d6aba5fe3d11aa5286", "attempt_id": "e9a7a6773262", "run_group_id": "E4_core:A4|closed|qwen3b-llama3b:h1", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T06:50:27.676731Z", "task_family": "synthetic", "document_id": "syn_03003_2000", "document_hash": "dddca1a38e3224870fcfae927c701c400fd885c19123ea840ee06a76a140f407", "question_id": "q12_unit_conversion", "split": "test", "cluster_id": "syn_03003_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 1, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1866, "query_tokens": 602, "message_tokens": 512, "message_utf8_bytes": 1200, "cache_tensor_bytes": null, "serialized_payload_bytes": 1200, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 22569.42216597963, "hops": {"1": 22569.42216597963}, "receiver_prefill_ms": 346.38229198753834, "ttft_ms": 42.7784159546718, "decode_ms": 343.4699580539018}, "total_wall_ms": 23302.966458024457, "ttft_ms": 42.7784159546718, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 5}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "17430", "evidence_artifact": {"ledger": {"facts_total": 58, "facts_retained": 32, "retention_rate": 0.5517241379310345, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.7142857142857143, "amount": 0.6666666666666666, "status": 0.6666666666666666, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1200}, "ledger_by_hop": {"1": {"facts_total": 58, "facts_retained": 32, "retention_rate": 0.5517241379310345, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.7142857142857143, "amount": 0.6666666666666666, "status": 0.6666666666666666, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1200}}}, "reference_key_id": "keys/synthetic/syn_03003_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "6d2a2bad9425191f4568f7eb115d63c953e27080a86953f62a982ad73340e770", "output_hash": "2575eaf3b748efc209b676518766751eb038c88f74578951345818bfe7f7194b"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "ec6a33f77c8498fc027b167a", "attempt_id": "1eb0cae70879", "run_group_id": "E4_core:A4|closed|qwen3b-llama3b:h1", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T06:50:28.427694Z", "task_family": "synthetic", "document_id": "syn_03003_2000", "document_hash": "dddca1a38e3224870fcfae927c701c400fd885c19123ea840ee06a76a140f407", "question_id": "q13_unanswerable", "split": "test", "cluster_id": "syn_03003_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 1, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1866, "query_tokens": 606, "message_tokens": 512, "message_utf8_bytes": 1200, "cache_tensor_bytes": null, "serialized_payload_bytes": 1200, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 22569.42216597963, "hops": {"1": 22569.42216597963}, "receiver_prefill_ms": 467.2487919451669, "ttft_ms": 32.53958304412663, "decode_ms": 33.23554201051593}, "total_wall_ms": 23103.078081971034, "ttft_ms": 32.53958304412663, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 58, "facts_retained": 32, "retention_rate": 0.5517241379310345, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.7142857142857143, "amount": 0.6666666666666666, "status": 0.6666666666666666, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1200}, "ledger_by_hop": {"1": {"facts_total": 58, "facts_retained": 32, "retention_rate": 0.5517241379310345, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.7142857142857143, "amount": 0.6666666666666666, "status": 0.6666666666666666, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1200}}}, "reference_key_id": "keys/synthetic/syn_03003_2000.json", "task_scores": {"correct": 1.0, "abstained": 1.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "6d2a2bad9425191f4568f7eb115d63c953e27080a86953f62a982ad73340e770", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "dcf8d0cecf21c23528ae77dc", "attempt_id": "38d4d28491ff", "run_group_id": "E4_core:A4|closed|qwen3b-llama3b:h1", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T06:50:28.978504Z", "task_family": "synthetic", "document_id": "syn_03003_2000", "document_hash": "dddca1a38e3224870fcfae927c701c400fd885c19123ea840ee06a76a140f407", "question_id": "q14_unanswerable", "split": "test", "cluster_id": "syn_03003_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 1, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1866, "query_tokens": 607, "message_tokens": 512, "message_utf8_bytes": 1200, "cache_tensor_bytes": null, "serialized_payload_bytes": 1200, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 22569.42216597963, "hops": {"1": 22569.42216597963}, "receiver_prefill_ms": 402.74262498132885, "ttft_ms": 19.00220802053809, "decode_ms": 40.570583078078926}, "total_wall_ms": 23032.451873994432, "ttft_ms": 19.00220802053809, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 58, "facts_retained": 32, "retention_rate": 0.5517241379310345, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.7142857142857143, "amount": 0.6666666666666666, "status": 0.6666666666666666, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1200}, "ledger_by_hop": {"1": {"facts_total": 58, "facts_retained": 32, "retention_rate": 0.5517241379310345, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.7142857142857143, "amount": 0.6666666666666666, "status": 0.6666666666666666, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1200}}}, "reference_key_id": "keys/synthetic/syn_03003_2000.json", "task_scores": {"correct": 1.0, "abstained": 1.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "6d2a2bad9425191f4568f7eb115d63c953e27080a86953f62a982ad73340e770", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "2e9bd9f78a33994b4fd29f07", "attempt_id": "da8cbff7b8b1", "run_group_id": "E4_core:A4|closed|qwen3b-llama3b:h2", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T06:50:45.081516Z", "task_family": "synthetic", "document_id": "syn_03003_2000", "document_hash": "dddca1a38e3224870fcfae927c701c400fd885c19123ea840ee06a76a140f407", "question_id": "q00_single_fact", "split": "test", "cluster_id": "syn_03003_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "tokenizer_hashes": {"receiver": "b9d8872cf3a99da99e1fc4b0f71232a451632dc49ca4c6771549b4efa5c1c763"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 2, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1722, "query_tokens": 500, "message_tokens": 395, "message_utf8_bytes": 1045, "cache_tensor_bytes": null, "serialized_payload_bytes": 1045, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 38191.03049894329, "hops": {"1": 22569.42216597963, "2": 15621.60833296366}, "receiver_prefill_ms": 470.67608300130814, "ttft_ms": 23.814416956156492, "decode_ms": 220.06175003480166}, "total_wall_ms": 38906.266498961486, "ttft_ms": 23.814416956156492, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 7}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "GR-79S-5839", "evidence_artifact": {"ledger": {"facts_total": 58, "facts_retained": 32, "retention_rate": 0.5517241379310345, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.7142857142857143, "amount": 0.6666666666666666, "status": 0.6666666666666666, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1045}, "ledger_by_hop": {"1": {"facts_total": 58, "facts_retained": 32, "retention_rate": 0.5517241379310345, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.7142857142857143, "amount": 0.6666666666666666, "status": 0.6666666666666666, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1200}, "2": {"facts_total": 58, "facts_retained": 32, "retention_rate": 0.5517241379310345, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.7142857142857143, "amount": 0.6666666666666666, "status": 0.6666666666666666, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1045}}}, "reference_key_id": "keys/synthetic/syn_03003_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "810eb876ea8770014b7b0dfadbbca83a889061ee931dee3a3e0c50a9ac91cdc2", "output_hash": "720cc40cd69e85e9728f6bea00b17a95ca07eee643e143a131840763958303cd"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "c89e60bbcbff18548df0201c", "attempt_id": "6cf7ed670378", "run_group_id": "E4_core:A4|closed|qwen3b-llama3b:h2", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T06:50:45.814094Z", "task_family": "synthetic", "document_id": "syn_03003_2000", "document_hash": "dddca1a38e3224870fcfae927c701c400fd885c19123ea840ee06a76a140f407", "question_id": "q01_single_fact", "split": "test", "cluster_id": "syn_03003_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "tokenizer_hashes": {"receiver": "b9d8872cf3a99da99e1fc4b0f71232a451632dc49ca4c6771549b4efa5c1c763"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 2, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1722, "query_tokens": 504, "message_tokens": 395, "message_utf8_bytes": 1045, "cache_tensor_bytes": null, "serialized_payload_bytes": 1045, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 38191.03049894329, "hops": {"1": 22569.42216597963, "2": 15621.60833296366}, "receiver_prefill_ms": 383.109875023365, "ttft_ms": 18.99833302013576, "decode_ms": 273.89345900155604}, "total_wall_ms": 38867.873831884935, "ttft_ms": 18.99833302013576, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 9}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "Lot 344-B, Témiscaming", "evidence_artifact": {"ledger": {"facts_total": 58, "facts_retained": 32, "retention_rate": 0.5517241379310345, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.7142857142857143, "amount": 0.6666666666666666, "status": 0.6666666666666666, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1045}, "ledger_by_hop": {"1": {"facts_total": 58, "facts_retained": 32, "retention_rate": 0.5517241379310345, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.7142857142857143, "amount": 0.6666666666666666, "status": 0.6666666666666666, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1200}, "2": {"facts_total": 58, "facts_retained": 32, "retention_rate": 0.5517241379310345, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.7142857142857143, "amount": 0.6666666666666666, "status": 0.6666666666666666, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1045}}}, "reference_key_id": "keys/synthetic/syn_03003_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 0.8571428571428571}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "810eb876ea8770014b7b0dfadbbca83a889061ee931dee3a3e0c50a9ac91cdc2", "output_hash": "0197d491a5a04a142dc8e372360735a1d41757e7a5aea7e81f23b0e1ef6ee010"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "b027ef280da08438c2cbff5f", "attempt_id": "3f43eaa27594", "run_group_id": "E4_core:A4|closed|qwen3b-llama3b:h2", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T06:50:46.509489Z", "task_family": "synthetic", "document_id": "syn_03003_2000", "document_hash": "dddca1a38e3224870fcfae927c701c400fd885c19123ea840ee06a76a140f407", "question_id": "q02_exact_copy", "split": "test", "cluster_id": "syn_03003_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "tokenizer_hashes": {"receiver": "b9d8872cf3a99da99e1fc4b0f71232a451632dc49ca4c6771549b4efa5c1c763"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 2, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1722, "query_tokens": 504, "message_tokens": 395, "message_utf8_bytes": 1045, "cache_tensor_bytes": null, "serialized_payload_bytes": 1045, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 38191.03049894329, "hops": {"1": 22569.42216597963, "2": 15621.60833296366}, "receiver_prefill_ms": 424.8610419454053, "ttft_ms": 126.50616699829698, "decode_ms": 35.795166972093284}, "total_wall_ms": 38778.7196659483, "ttft_ms": 126.50616699829698, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 58, "facts_retained": 32, "retention_rate": 0.5517241379310345, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.7142857142857143, "amount": 0.6666666666666666, "status": 0.6666666666666666, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1045}, "ledger_by_hop": {"1": {"facts_total": 58, "facts_retained": 32, "retention_rate": 0.5517241379310345, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.7142857142857143, "amount": 0.6666666666666666, "status": 0.6666666666666666, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1200}, "2": {"facts_total": 58, "facts_retained": 32, "retention_rate": 0.5517241379310345, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.7142857142857143, "amount": 0.6666666666666666, "status": 0.6666666666666666, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1045}}}, "reference_key_id": "keys/synthetic/syn_03003_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "810eb876ea8770014b7b0dfadbbca83a889061ee931dee3a3e0c50a9ac91cdc2", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "adcfa4851f91a0989d04c875", "attempt_id": "63c4a46f2656", "run_group_id": "E4_core:A4|closed|qwen3b-llama3b:h2", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T06:50:47.114161Z", "task_family": "synthetic", "document_id": "syn_03003_2000", "document_hash": "dddca1a38e3224870fcfae927c701c400fd885c19123ea840ee06a76a140f407", "question_id": "q03_exact_copy", "split": "test", "cluster_id": "syn_03003_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "tokenizer_hashes": {"receiver": "b9d8872cf3a99da99e1fc4b0f71232a451632dc49ca4c6771549b4efa5c1c763"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 2, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1722, "query_tokens": 499, "message_tokens": 395, "message_utf8_bytes": 1045, "cache_tensor_bytes": null, "serialized_payload_bytes": 1045, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 38191.03049894329, "hops": {"1": 22569.42216597963, "2": 15621.60833296366}, "receiver_prefill_ms": 352.0789999747649, "ttft_ms": 38.20370906032622, "decode_ms": 195.47287502791733}, "total_wall_ms": 38777.314581908286, "ttft_ms": 38.20370906032622, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 6}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "Wilhelmina Achterberg", "evidence_artifact": {"ledger": {"facts_total": 58, "facts_retained": 32, "retention_rate": 0.5517241379310345, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.7142857142857143, "amount": 0.6666666666666666, "status": 0.6666666666666666, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1045}, "ledger_by_hop": {"1": {"facts_total": 58, "facts_retained": 32, "retention_rate": 0.5517241379310345, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.7142857142857143, "amount": 0.6666666666666666, "status": 0.6666666666666666, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1200}, "2": {"facts_total": 58, "facts_retained": 32, "retention_rate": 0.5517241379310345, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.7142857142857143, "amount": 0.6666666666666666, "status": 0.6666666666666666, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1045}}}, "reference_key_id": "keys/synthetic/syn_03003_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "810eb876ea8770014b7b0dfadbbca83a889061ee931dee3a3e0c50a9ac91cdc2", "output_hash": "2a62606853c315bba8769371d195e3ab3d6a8cd7ef959205575c5119740850f8"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "9159e95e3924f360a1887ba8", "attempt_id": "8ce5b9f93174", "run_group_id": "E4_core:A4|closed|qwen3b-llama3b:h2", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T06:50:47.718207Z", "task_family": "synthetic", "document_id": "syn_03003_2000", "document_hash": "dddca1a38e3224870fcfae927c701c400fd885c19123ea840ee06a76a140f407", "question_id": "q04_multi_hop", "split": "test", "cluster_id": "syn_03003_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "tokenizer_hashes": {"receiver": "b9d8872cf3a99da99e1fc4b0f71232a451632dc49ca4c6771549b4efa5c1c763"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 2, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1722, "query_tokens": 508, "message_tokens": 395, "message_utf8_bytes": 1045, "cache_tensor_bytes": null, "serialized_payload_bytes": 1045, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 38191.03049894329, "hops": {"1": 22569.42216597963, "2": 15621.60833296366}, "receiver_prefill_ms": 355.86016706656665, "ttft_ms": 51.91124998964369, "decode_ms": 56.98166694492102}, "total_wall_ms": 38656.49474901147, "ttft_ms": 51.91124998964369, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 58, "facts_retained": 32, "retention_rate": 0.5517241379310345, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.7142857142857143, "amount": 0.6666666666666666, "status": 0.6666666666666666, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1045}, "ledger_by_hop": {"1": {"facts_total": 58, "facts_retained": 32, "retention_rate": 0.5517241379310345, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.7142857142857143, "amount": 0.6666666666666666, "status": 0.6666666666666666, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1200}, "2": {"facts_total": 58, "facts_retained": 32, "retention_rate": 0.5517241379310345, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.7142857142857143, "amount": 0.6666666666666666, "status": 0.6666666666666666, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1045}}}, "reference_key_id": "keys/synthetic/syn_03003_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "810eb876ea8770014b7b0dfadbbca83a889061ee931dee3a3e0c50a9ac91cdc2", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "f8a962999afdf058a79f8fdd", "attempt_id": "456032f25bd3", "run_group_id": "E4_core:A4|closed|qwen3b-llama3b:h2", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T06:50:48.201051Z", "task_family": "synthetic", "document_id": "syn_03003_2000", "document_hash": "dddca1a38e3224870fcfae927c701c400fd885c19123ea840ee06a76a140f407", "question_id": "q05_multi_hop", "split": "test", "cluster_id": "syn_03003_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "tokenizer_hashes": {"receiver": "b9d8872cf3a99da99e1fc4b0f71232a451632dc49ca4c6771549b4efa5c1c763"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 2, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1722, "query_tokens": 508, "message_tokens": 395, "message_utf8_bytes": 1045, "cache_tensor_bytes": null, "serialized_payload_bytes": 1045, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 38191.03049894329, "hops": {"1": 22569.42216597963, "2": 15621.60833296366}, "receiver_prefill_ms": 498.75237490050495, "ttft_ms": 38.415250019170344, "decode_ms": 204.40833398606628}, "total_wall_ms": 38933.13258187845, "ttft_ms": 38.415250019170344, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 7}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "NK-95Z-2165", "evidence_artifact": {"ledger": {"facts_total": 58, "facts_retained": 32, "retention_rate": 0.5517241379310345, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.7142857142857143, "amount": 0.6666666666666666, "status": 0.6666666666666666, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1045}, "ledger_by_hop": {"1": {"facts_total": 58, "facts_retained": 32, "retention_rate": 0.5517241379310345, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.7142857142857143, "amount": 0.6666666666666666, "status": 0.6666666666666666, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1200}, "2": {"facts_total": 58, "facts_retained": 32, "retention_rate": 0.5517241379310345, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.7142857142857143, "amount": 0.6666666666666666, "status": 0.6666666666666666, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1045}}}, "reference_key_id": "keys/synthetic/syn_03003_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "810eb876ea8770014b7b0dfadbbca83a889061ee931dee3a3e0c50a9ac91cdc2", "output_hash": "bd8f809b02304c6b9dc7c1e116ecccc2328c976a5c071e89a94d661b48e73aa7"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "332d945a731c0b7204867b68", "attempt_id": "4e9371ff7f7d", "run_group_id": "E4_core:A4|closed|qwen3b-llama3b:h2", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T06:50:48.960381Z", "task_family": "synthetic", "document_id": "syn_03003_2000", "document_hash": "dddca1a38e3224870fcfae927c701c400fd885c19123ea840ee06a76a140f407", "question_id": "q06_negation", "split": "test", "cluster_id": "syn_03003_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "tokenizer_hashes": {"receiver": "b9d8872cf3a99da99e1fc4b0f71232a451632dc49ca4c6771549b4efa5c1c763"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 2, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1722, "query_tokens": 501, "message_tokens": 395, "message_utf8_bytes": 1045, "cache_tensor_bytes": null, "serialized_payload_bytes": 1045, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 38191.03049894329, "hops": {"1": 22569.42216597963, "2": 15621.60833296366}, "receiver_prefill_ms": 378.4667090512812, "ttft_ms": 38.517458946444094, "decode_ms": 51.65337503422052}, "total_wall_ms": 38660.19595693797, "ttft_ms": 38.517458946444094, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 2}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "No.", "evidence_artifact": {"ledger": {"facts_total": 58, "facts_retained": 32, "retention_rate": 0.5517241379310345, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.7142857142857143, "amount": 0.6666666666666666, "status": 0.6666666666666666, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1045}, "ledger_by_hop": {"1": {"facts_total": 58, "facts_retained": 32, "retention_rate": 0.5517241379310345, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.7142857142857143, "amount": 0.6666666666666666, "status": 0.6666666666666666, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1200}, "2": {"facts_total": 58, "facts_retained": 32, "retention_rate": 0.5517241379310345, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.7142857142857143, "amount": 0.6666666666666666, "status": 0.6666666666666666, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1045}}}, "reference_key_id": "keys/synthetic/syn_03003_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "810eb876ea8770014b7b0dfadbbca83a889061ee931dee3a3e0c50a9ac91cdc2", "output_hash": "38178a20b470cfd18299fb1593dd4ba706f1a83321dac051d01086efd8b7a96f"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "6d5008f5fbb3747cbcb2a463", "attempt_id": "60135e747b19", "run_group_id": "E4_core:A4|closed|qwen3b-llama3b:h2", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T06:50:49.446425Z", "task_family": "synthetic", "document_id": "syn_03003_2000", "document_hash": "dddca1a38e3224870fcfae927c701c400fd885c19123ea840ee06a76a140f407", "question_id": "q07_negation", "split": "test", "cluster_id": "syn_03003_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "tokenizer_hashes": {"receiver": "b9d8872cf3a99da99e1fc4b0f71232a451632dc49ca4c6771549b4efa5c1c763"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 2, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1722, "query_tokens": 499, "message_tokens": 395, "message_utf8_bytes": 1045, "cache_tensor_bytes": null, "serialized_payload_bytes": 1045, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 38191.03049894329, "hops": {"1": 22569.42216597963, "2": 15621.60833296366}, "receiver_prefill_ms": 353.3479580655694, "ttft_ms": 50.63012498430908, "decode_ms": 39.84208300244063}, "total_wall_ms": 38635.591623955406, "ttft_ms": 50.63012498430908, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "Yes", "evidence_artifact": {"ledger": {"facts_total": 58, "facts_retained": 32, "retention_rate": 0.5517241379310345, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.7142857142857143, "amount": 0.6666666666666666, "status": 0.6666666666666666, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1045}, "ledger_by_hop": {"1": {"facts_total": 58, "facts_retained": 32, "retention_rate": 0.5517241379310345, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.7142857142857143, "amount": 0.6666666666666666, "status": 0.6666666666666666, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1200}, "2": {"facts_total": 58, "facts_retained": 32, "retention_rate": 0.5517241379310345, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.7142857142857143, "amount": 0.6666666666666666, "status": 0.6666666666666666, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1045}}}, "reference_key_id": "keys/synthetic/syn_03003_2000.json", "task_scores": {"correct": 0.0, "abstained": 0.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "810eb876ea8770014b7b0dfadbbca83a889061ee931dee3a3e0c50a9ac91cdc2", "output_hash": "85a39ab345d672ff8ca9b9c6876f3adcacf45ee7c1e2dbd2408fd338bd55e07e"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "52b3c9a462216fa05ebdd4eb", "attempt_id": "8f9fdff34eec", "run_group_id": "E4_core:A4|closed|qwen3b-llama3b:h2", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T06:50:49.908173Z", "task_family": "synthetic", "document_id": "syn_03003_2000", "document_hash": "dddca1a38e3224870fcfae927c701c400fd885c19123ea840ee06a76a140f407", "question_id": "q08_revision", "split": "test", "cluster_id": "syn_03003_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "tokenizer_hashes": {"receiver": "b9d8872cf3a99da99e1fc4b0f71232a451632dc49ca4c6771549b4efa5c1c763"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 2, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1722, "query_tokens": 517, "message_tokens": 395, "message_utf8_bytes": 1045, "cache_tensor_bytes": null, "serialized_payload_bytes": 1045, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 38191.03049894329, "hops": {"1": 22569.42216597963, "2": 15621.60833296366}, "receiver_prefill_ms": 489.26041601225734, "ttft_ms": 19.10541602410376, "decode_ms": 101.36620793491602}, "total_wall_ms": 38801.53553991113, "ttft_ms": 19.10541602410376, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 3}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "911,925", "evidence_artifact": {"ledger": {"facts_total": 58, "facts_retained": 32, "retention_rate": 0.5517241379310345, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.7142857142857143, "amount": 0.6666666666666666, "status": 0.6666666666666666, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1045}, "ledger_by_hop": {"1": {"facts_total": 58, "facts_retained": 32, "retention_rate": 0.5517241379310345, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.7142857142857143, "amount": 0.6666666666666666, "status": 0.6666666666666666, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1200}, "2": {"facts_total": 58, "facts_retained": 32, "retention_rate": 0.5517241379310345, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.7142857142857143, "amount": 0.6666666666666666, "status": 0.6666666666666666, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1045}}}, "reference_key_id": "keys/synthetic/syn_03003_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "810eb876ea8770014b7b0dfadbbca83a889061ee931dee3a3e0c50a9ac91cdc2", "output_hash": "86be78720fc4b79f93e603e7d9e4861fcd25102612f73209fffc17d787056007"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "595df2d42628274adb0f3e9b", "attempt_id": "7a94068bf62f", "run_group_id": "E4_core:A4|closed|qwen3b-llama3b:h2", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T06:50:50.536267Z", "task_family": "synthetic", "document_id": "syn_03003_2000", "document_hash": "dddca1a38e3224870fcfae927c701c400fd885c19123ea840ee06a76a140f407", "question_id": "q09_revision", "split": "test", "cluster_id": "syn_03003_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "tokenizer_hashes": {"receiver": "b9d8872cf3a99da99e1fc4b0f71232a451632dc49ca4c6771549b4efa5c1c763"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 2, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1722, "query_tokens": 510, "message_tokens": 395, "message_utf8_bytes": 1045, "cache_tensor_bytes": null, "serialized_payload_bytes": 1045, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 38191.03049894329, "hops": {"1": 22569.42216597963, "2": 15621.60833296366}, "receiver_prefill_ms": 383.9062500046566, "ttft_ms": 26.609334046952426, "decode_ms": 100.83616699557751}, "total_wall_ms": 38702.91262387764, "ttft_ms": 26.609334046952426, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 3}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "911,925", "evidence_artifact": {"ledger": {"facts_total": 58, "facts_retained": 32, "retention_rate": 0.5517241379310345, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.7142857142857143, "amount": 0.6666666666666666, "status": 0.6666666666666666, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1045}, "ledger_by_hop": {"1": {"facts_total": 58, "facts_retained": 32, "retention_rate": 0.5517241379310345, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.7142857142857143, "amount": 0.6666666666666666, "status": 0.6666666666666666, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1200}, "2": {"facts_total": 58, "facts_retained": 32, "retention_rate": 0.5517241379310345, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.7142857142857143, "amount": 0.6666666666666666, "status": 0.6666666666666666, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1045}}}, "reference_key_id": "keys/synthetic/syn_03003_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "810eb876ea8770014b7b0dfadbbca83a889061ee931dee3a3e0c50a9ac91cdc2", "output_hash": "86be78720fc4b79f93e603e7d9e4861fcd25102612f73209fffc17d787056007"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "f0eb46703f1bc63b82324c9d", "attempt_id": "c850b6f3674c", "run_group_id": "E4_core:A4|closed|qwen3b-llama3b:h2", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T06:50:51.064569Z", "task_family": "synthetic", "document_id": "syn_03003_2000", "document_hash": "dddca1a38e3224870fcfae927c701c400fd885c19123ea840ee06a76a140f407", "question_id": "q10_temporal", "split": "test", "cluster_id": "syn_03003_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "tokenizer_hashes": {"receiver": "b9d8872cf3a99da99e1fc4b0f71232a451632dc49ca4c6771549b4efa5c1c763"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 2, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1722, "query_tokens": 517, "message_tokens": 395, "message_utf8_bytes": 1045, "cache_tensor_bytes": null, "serialized_payload_bytes": 1045, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 38191.03049894329, "hops": {"1": 22569.42216597963, "2": 15621.60833296366}, "receiver_prefill_ms": 306.02170806378126, "ttft_ms": 35.23995797149837, "decode_ms": 289.0170420287177}, "total_wall_ms": 38822.190123959444, "ttft_ms": 35.23995797149837, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 6}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "Gamtiszen-2", "evidence_artifact": {"ledger": {"facts_total": 58, "facts_retained": 32, "retention_rate": 0.5517241379310345, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.7142857142857143, "amount": 0.6666666666666666, "status": 0.6666666666666666, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1045}, "ledger_by_hop": {"1": {"facts_total": 58, "facts_retained": 32, "retention_rate": 0.5517241379310345, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.7142857142857143, "amount": 0.6666666666666666, "status": 0.6666666666666666, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1200}, "2": {"facts_total": 58, "facts_retained": 32, "retention_rate": 0.5517241379310345, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.7142857142857143, "amount": 0.6666666666666666, "status": 0.6666666666666666, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1045}}}, "reference_key_id": "keys/synthetic/syn_03003_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "810eb876ea8770014b7b0dfadbbca83a889061ee931dee3a3e0c50a9ac91cdc2", "output_hash": "32e3af428886434823d82102e1c482765eafdf6077fe8d3fba5752ec5e29826c"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "2575d53ef2c6fcbd7c00ce0b", "attempt_id": "0d502a4ed3e8", "run_group_id": "E4_core:A4|closed|qwen3b-llama3b:h2", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T06:50:51.713133Z", "task_family": "synthetic", "document_id": "syn_03003_2000", "document_hash": "dddca1a38e3224870fcfae927c701c400fd885c19123ea840ee06a76a140f407", "question_id": "q11_aggregation", "split": "test", "cluster_id": "syn_03003_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "tokenizer_hashes": {"receiver": "b9d8872cf3a99da99e1fc4b0f71232a451632dc49ca4c6771549b4efa5c1c763"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 2, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1722, "query_tokens": 511, "message_tokens": 395, "message_utf8_bytes": 1045, "cache_tensor_bytes": null, "serialized_payload_bytes": 1045, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 38191.03049894329, "hops": {"1": 22569.42216597963, "2": 15621.60833296366}, "receiver_prefill_ms": 477.2714169230312, "ttft_ms": 34.574959077872336, "decode_ms": 148.9520421018824}, "total_wall_ms": 38852.339707897045, "ttft_ms": 34.574959077872336, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 5}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "11,165,385", "evidence_artifact": {"ledger": {"facts_total": 58, "facts_retained": 32, "retention_rate": 0.5517241379310345, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.7142857142857143, "amount": 0.6666666666666666, "status": 0.6666666666666666, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1045}, "ledger_by_hop": {"1": {"facts_total": 58, "facts_retained": 32, "retention_rate": 0.5517241379310345, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.7142857142857143, "amount": 0.6666666666666666, "status": 0.6666666666666666, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1200}, "2": {"facts_total": 58, "facts_retained": 32, "retention_rate": 0.5517241379310345, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.7142857142857143, "amount": 0.6666666666666666, "status": 0.6666666666666666, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1045}}}, "reference_key_id": "keys/synthetic/syn_03003_2000.json", "task_scores": {"correct": 0.0, "abstained": 0.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "810eb876ea8770014b7b0dfadbbca83a889061ee931dee3a3e0c50a9ac91cdc2", "output_hash": "f1e5ec0b4bf018da9a3426a6a44b214a3c6d41b2cfd75dda83aeab2e6d4e624e"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "315ac1557e6b08ed4fb5c68b", "attempt_id": "38c28d71af94", "run_group_id": "E4_core:A4|closed|qwen3b-llama3b:h2", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T06:50:52.391330Z", "task_family": "synthetic", "document_id": "syn_03003_2000", "document_hash": "dddca1a38e3224870fcfae927c701c400fd885c19123ea840ee06a76a140f407", "question_id": "q12_unit_conversion", "split": "test", "cluster_id": "syn_03003_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "tokenizer_hashes": {"receiver": "b9d8872cf3a99da99e1fc4b0f71232a451632dc49ca4c6771549b4efa5c1c763"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 2, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1722, "query_tokens": 508, "message_tokens": 395, "message_utf8_bytes": 1045, "cache_tensor_bytes": null, "serialized_payload_bytes": 1045, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 38191.03049894329, "hops": {"1": 22569.42216597963, "2": 15621.60833296366}, "receiver_prefill_ms": 360.93299998901784, "ttft_ms": 22.025250014849007, "decode_ms": 83.29491596668959}, "total_wall_ms": 38657.882915926166, "ttft_ms": 22.025250014849007, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 3}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "17.43", "evidence_artifact": {"ledger": {"facts_total": 58, "facts_retained": 32, "retention_rate": 0.5517241379310345, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.7142857142857143, "amount": 0.6666666666666666, "status": 0.6666666666666666, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1045}, "ledger_by_hop": {"1": {"facts_total": 58, "facts_retained": 32, "retention_rate": 0.5517241379310345, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.7142857142857143, "amount": 0.6666666666666666, "status": 0.6666666666666666, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1200}, "2": {"facts_total": 58, "facts_retained": 32, "retention_rate": 0.5517241379310345, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.7142857142857143, "amount": 0.6666666666666666, "status": 0.6666666666666666, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1045}}}, "reference_key_id": "keys/synthetic/syn_03003_2000.json", "task_scores": {"correct": 0.0, "abstained": 0.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "810eb876ea8770014b7b0dfadbbca83a889061ee931dee3a3e0c50a9ac91cdc2", "output_hash": "1cd677e6c296f31fca744a7c4d96b5514fa6b036913240b8d689f1e277900d17"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "a5cd7fbd91835723b032d215", "attempt_id": "5e625c0f3895", "run_group_id": "E4_core:A4|closed|qwen3b-llama3b:h2", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T06:50:52.875529Z", "task_family": "synthetic", "document_id": "syn_03003_2000", "document_hash": "dddca1a38e3224870fcfae927c701c400fd885c19123ea840ee06a76a140f407", "question_id": "q13_unanswerable", "split": "test", "cluster_id": "syn_03003_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "tokenizer_hashes": {"receiver": "b9d8872cf3a99da99e1fc4b0f71232a451632dc49ca4c6771549b4efa5c1c763"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 2, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1722, "query_tokens": 512, "message_tokens": 395, "message_utf8_bytes": 1045, "cache_tensor_bytes": null, "serialized_payload_bytes": 1045, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 38191.03049894329, "hops": {"1": 22569.42216597963, "2": 15621.60833296366}, "receiver_prefill_ms": 397.012708010152, "ttft_ms": 48.33191691432148, "decode_ms": 393.2277919957414}, "total_wall_ms": 39030.292041017674, "ttft_ms": 48.33191691432148, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 7}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "GR-79S-5839", "evidence_artifact": {"ledger": {"facts_total": 58, "facts_retained": 32, "retention_rate": 0.5517241379310345, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.7142857142857143, "amount": 0.6666666666666666, "status": 0.6666666666666666, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1045}, "ledger_by_hop": {"1": {"facts_total": 58, "facts_retained": 32, "retention_rate": 0.5517241379310345, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.7142857142857143, "amount": 0.6666666666666666, "status": 0.6666666666666666, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1200}, "2": {"facts_total": 58, "facts_retained": 32, "retention_rate": 0.5517241379310345, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.7142857142857143, "amount": 0.6666666666666666, "status": 0.6666666666666666, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1045}}}, "reference_key_id": "keys/synthetic/syn_03003_2000.json", "task_scores": {"correct": 0.0, "abstained": 0.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "810eb876ea8770014b7b0dfadbbca83a889061ee931dee3a3e0c50a9ac91cdc2", "output_hash": "720cc40cd69e85e9728f6bea00b17a95ca07eee643e143a131840763958303cd"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "4e018ce83e32b8123e3df464", "attempt_id": "d06c72255cc0", "run_group_id": "E4_core:A4|closed|qwen3b-llama3b:h2", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T06:50:53.732525Z", "task_family": "synthetic", "document_id": "syn_03003_2000", "document_hash": "dddca1a38e3224870fcfae927c701c400fd885c19123ea840ee06a76a140f407", "question_id": "q14_unanswerable", "split": "test", "cluster_id": "syn_03003_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "tokenizer_hashes": {"receiver": "b9d8872cf3a99da99e1fc4b0f71232a451632dc49ca4c6771549b4efa5c1c763"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 2, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1722, "query_tokens": 513, "message_tokens": 395, "message_utf8_bytes": 1045, "cache_tensor_bytes": null, "serialized_payload_bytes": 1045, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 38191.03049894329, "hops": {"1": 22569.42216597963, "2": 15621.60833296366}, "receiver_prefill_ms": 345.6635830225423, "ttft_ms": 37.8960840171203, "decode_ms": 24.636917049065232}, "total_wall_ms": 38599.752664915286, "ttft_ms": 37.8960840171203, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 58, "facts_retained": 32, "retention_rate": 0.5517241379310345, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.7142857142857143, "amount": 0.6666666666666666, "status": 0.6666666666666666, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1045}, "ledger_by_hop": {"1": {"facts_total": 58, "facts_retained": 32, "retention_rate": 0.5517241379310345, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.7142857142857143, "amount": 0.6666666666666666, "status": 0.6666666666666666, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1200}, "2": {"facts_total": 58, "facts_retained": 32, "retention_rate": 0.5517241379310345, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.7142857142857143, "amount": 0.6666666666666666, "status": 0.6666666666666666, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1045}}}, "reference_key_id": "keys/synthetic/syn_03003_2000.json", "task_scores": {"correct": 1.0, "abstained": 1.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "810eb876ea8770014b7b0dfadbbca83a889061ee931dee3a3e0c50a9ac91cdc2", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "e0f4a045330cf3b601816e15", "attempt_id": "86ae1a8374fa", "run_group_id": "E4_core:A4|closed|qwen3b-llama3b:h4", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T06:51:30.253346Z", "task_family": "synthetic", "document_id": "syn_03003_2000", "document_hash": "dddca1a38e3224870fcfae927c701c400fd885c19123ea840ee06a76a140f407", "question_id": "q00_single_fact", "split": "test", "cluster_id": "syn_03003_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "tokenizer_hashes": {"receiver": "b9d8872cf3a99da99e1fc4b0f71232a451632dc49ca4c6771549b4efa5c1c763"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 4, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1722, "query_tokens": 500, "message_tokens": 395, "message_utf8_bytes": 1044, "cache_tensor_bytes": null, "serialized_payload_bytes": 1044, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 74283.0914988881, "hops": {"1": 22569.42216597963, "2": 15621.60833296366, "3": 20515.35983302165, "4": 15576.701166923158}, "receiver_prefill_ms": 462.7136669587344, "ttft_ms": 71.21041708160192, "decode_ms": 222.8857500012964}, "total_wall_ms": 75040.45266483445, "ttft_ms": 71.21041708160192, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 7}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "GR-79S-5839", "evidence_artifact": {"ledger": {"facts_total": 58, "facts_retained": 32, "retention_rate": 0.5517241379310345, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.7142857142857143, "amount": 0.6666666666666666, "status": 0.6666666666666666, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1044}, "ledger_by_hop": {"1": {"facts_total": 58, "facts_retained": 32, "retention_rate": 0.5517241379310345, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.7142857142857143, "amount": 0.6666666666666666, "status": 0.6666666666666666, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1200}, "2": {"facts_total": 58, "facts_retained": 32, "retention_rate": 0.5517241379310345, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.7142857142857143, "amount": 0.6666666666666666, "status": 0.6666666666666666, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1045}, "3": {"facts_total": 58, "facts_retained": 32, "retention_rate": 0.5517241379310345, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.7142857142857143, "amount": 0.6666666666666666, "status": 0.6666666666666666, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1044}, "4": {"facts_total": 58, "facts_retained": 32, "retention_rate": 0.5517241379310345, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.7142857142857143, "amount": 0.6666666666666666, "status": 0.6666666666666666, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1044}}}, "reference_key_id": "keys/synthetic/syn_03003_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "508f87a419f4b708b6d957677c3ec8f39a369a431efc5130fdf9aa8501d4c113", "output_hash": "720cc40cd69e85e9728f6bea00b17a95ca07eee643e143a131840763958303cd"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "cec39a35bcbf7453b72374e5", "attempt_id": "ace2108ff195", "run_group_id": "E4_core:A4|closed|qwen3b-llama3b:h4", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T06:51:31.028852Z", "task_family": "synthetic", "document_id": "syn_03003_2000", "document_hash": "dddca1a38e3224870fcfae927c701c400fd885c19123ea840ee06a76a140f407", "question_id": "q01_single_fact", "split": "test", "cluster_id": "syn_03003_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "tokenizer_hashes": {"receiver": "b9d8872cf3a99da99e1fc4b0f71232a451632dc49ca4c6771549b4efa5c1c763"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 4, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1722, "query_tokens": 504, "message_tokens": 395, "message_utf8_bytes": 1044, "cache_tensor_bytes": null, "serialized_payload_bytes": 1044, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 74283.0914988881, "hops": {"1": 22569.42216597963, "2": 15621.60833296366, "3": 20515.35983302165, "4": 15576.701166923158}, "receiver_prefill_ms": 380.6222080020234, "ttft_ms": 32.295624958351254, "decode_ms": 239.76266698446125}, "total_wall_ms": 74936.43670692109, "ttft_ms": 32.295624958351254, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 9}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "Lot 344-B, Témiscaming", "evidence_artifact": {"ledger": {"facts_total": 58, "facts_retained": 32, "retention_rate": 0.5517241379310345, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.7142857142857143, "amount": 0.6666666666666666, "status": 0.6666666666666666, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1044}, "ledger_by_hop": {"1": {"facts_total": 58, "facts_retained": 32, "retention_rate": 0.5517241379310345, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.7142857142857143, "amount": 0.6666666666666666, "status": 0.6666666666666666, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1200}, "2": {"facts_total": 58, "facts_retained": 32, "retention_rate": 0.5517241379310345, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.7142857142857143, "amount": 0.6666666666666666, "status": 0.6666666666666666, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1045}, "3": {"facts_total": 58, "facts_retained": 32, "retention_rate": 0.5517241379310345, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.7142857142857143, "amount": 0.6666666666666666, "status": 0.6666666666666666, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1044}, "4": {"facts_total": 58, "facts_retained": 32, "retention_rate": 0.5517241379310345, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.7142857142857143, "amount": 0.6666666666666666, "status": 0.6666666666666666, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1044}}}, "reference_key_id": "keys/synthetic/syn_03003_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 0.8571428571428571}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "508f87a419f4b708b6d957677c3ec8f39a369a431efc5130fdf9aa8501d4c113", "output_hash": "0197d491a5a04a142dc8e372360735a1d41757e7a5aea7e81f23b0e1ef6ee010"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "630a89b2480ee8813d1cf67a", "attempt_id": "05e89f7af422", "run_group_id": "E4_core:A4|closed|qwen3b-llama3b:h4", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T06:51:31.699635Z", "task_family": "synthetic", "document_id": "syn_03003_2000", "document_hash": "dddca1a38e3224870fcfae927c701c400fd885c19123ea840ee06a76a140f407", "question_id": "q02_exact_copy", "split": "test", "cluster_id": "syn_03003_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "tokenizer_hashes": {"receiver": "b9d8872cf3a99da99e1fc4b0f71232a451632dc49ca4c6771549b4efa5c1c763"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 4, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1722, "query_tokens": 504, "message_tokens": 395, "message_utf8_bytes": 1044, "cache_tensor_bytes": null, "serialized_payload_bytes": 1044, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 74283.0914988881, "hops": {"1": 22569.42216597963, "2": 15621.60833296366, "3": 20515.35983302165, "4": 15576.701166923158}, "receiver_prefill_ms": 360.74045789428055, "ttft_ms": 43.44316595233977, "decode_ms": 102.78241604100913}, "total_wall_ms": 74790.77608184889, "ttft_ms": 43.44316595233977, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 58, "facts_retained": 32, "retention_rate": 0.5517241379310345, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.7142857142857143, "amount": 0.6666666666666666, "status": 0.6666666666666666, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1044}, "ledger_by_hop": {"1": {"facts_total": 58, "facts_retained": 32, "retention_rate": 0.5517241379310345, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.7142857142857143, "amount": 0.6666666666666666, "status": 0.6666666666666666, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1200}, "2": {"facts_total": 58, "facts_retained": 32, "retention_rate": 0.5517241379310345, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.7142857142857143, "amount": 0.6666666666666666, "status": 0.6666666666666666, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1045}, "3": {"facts_total": 58, "facts_retained": 32, "retention_rate": 0.5517241379310345, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.7142857142857143, "amount": 0.6666666666666666, "status": 0.6666666666666666, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1044}, "4": {"facts_total": 58, "facts_retained": 32, "retention_rate": 0.5517241379310345, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.7142857142857143, "amount": 0.6666666666666666, "status": 0.6666666666666666, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1044}}}, "reference_key_id": "keys/synthetic/syn_03003_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "508f87a419f4b708b6d957677c3ec8f39a369a431efc5130fdf9aa8501d4c113", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "d69ffbe0bf84c111f644ab85", "attempt_id": "42664e7cdcf2", "run_group_id": "E4_core:A4|closed|qwen3b-llama3b:h4", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T06:51:32.225581Z", "task_family": "synthetic", "document_id": "syn_03003_2000", "document_hash": "dddca1a38e3224870fcfae927c701c400fd885c19123ea840ee06a76a140f407", "question_id": "q03_exact_copy", "split": "test", "cluster_id": "syn_03003_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "tokenizer_hashes": {"receiver": "b9d8872cf3a99da99e1fc4b0f71232a451632dc49ca4c6771549b4efa5c1c763"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 4, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1722, "query_tokens": 499, "message_tokens": 395, "message_utf8_bytes": 1044, "cache_tensor_bytes": null, "serialized_payload_bytes": 1044, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 74283.0914988881, "hops": {"1": 22569.42216597963, "2": 15621.60833296366, "3": 20515.35983302165, "4": 15576.701166923158}, "receiver_prefill_ms": 430.12812500819564, "ttft_ms": 25.299874949268997, "decode_ms": 172.06391599029303}, "total_wall_ms": 74911.07828984968, "ttft_ms": 25.299874949268997, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 6}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "Wilhelmina Achterberg", "evidence_artifact": {"ledger": {"facts_total": 58, "facts_retained": 32, "retention_rate": 0.5517241379310345, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.7142857142857143, "amount": 0.6666666666666666, "status": 0.6666666666666666, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1044}, "ledger_by_hop": {"1": {"facts_total": 58, "facts_retained": 32, "retention_rate": 0.5517241379310345, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.7142857142857143, "amount": 0.6666666666666666, "status": 0.6666666666666666, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1200}, "2": {"facts_total": 58, "facts_retained": 32, "retention_rate": 0.5517241379310345, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.7142857142857143, "amount": 0.6666666666666666, "status": 0.6666666666666666, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1045}, "3": {"facts_total": 58, "facts_retained": 32, "retention_rate": 0.5517241379310345, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.7142857142857143, "amount": 0.6666666666666666, "status": 0.6666666666666666, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1044}, "4": {"facts_total": 58, "facts_retained": 32, "retention_rate": 0.5517241379310345, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.7142857142857143, "amount": 0.6666666666666666, "status": 0.6666666666666666, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1044}}}, "reference_key_id": "keys/synthetic/syn_03003_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "508f87a419f4b708b6d957677c3ec8f39a369a431efc5130fdf9aa8501d4c113", "output_hash": "2a62606853c315bba8769371d195e3ab3d6a8cd7ef959205575c5119740850f8"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "d1c1fb3df79b648d15d7acde", "attempt_id": "fe9fe3dcbd78", "run_group_id": "E4_core:A4|closed|qwen3b-llama3b:h4", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T06:51:32.870019Z", "task_family": "synthetic", "document_id": "syn_03003_2000", "document_hash": "dddca1a38e3224870fcfae927c701c400fd885c19123ea840ee06a76a140f407", "question_id": "q04_multi_hop", "split": "test", "cluster_id": "syn_03003_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "tokenizer_hashes": {"receiver": "b9d8872cf3a99da99e1fc4b0f71232a451632dc49ca4c6771549b4efa5c1c763"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 4, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1722, "query_tokens": 508, "message_tokens": 395, "message_utf8_bytes": 1044, "cache_tensor_bytes": null, "serialized_payload_bytes": 1044, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 74283.0914988881, "hops": {"1": 22569.42216597963, "2": 15621.60833296366, "3": 20515.35983302165, "4": 15576.701166923158}, "receiver_prefill_ms": 379.0726250736043, "ttft_ms": 21.944041014648974, "decode_ms": 17.260249936953187}, "total_wall_ms": 74701.85920689255, "ttft_ms": 21.944041014648974, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 58, "facts_retained": 32, "retention_rate": 0.5517241379310345, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.7142857142857143, "amount": 0.6666666666666666, "status": 0.6666666666666666, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1044}, "ledger_by_hop": {"1": {"facts_total": 58, "facts_retained": 32, "retention_rate": 0.5517241379310345, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.7142857142857143, "amount": 0.6666666666666666, "status": 0.6666666666666666, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1200}, "2": {"facts_total": 58, "facts_retained": 32, "retention_rate": 0.5517241379310345, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.7142857142857143, "amount": 0.6666666666666666, "status": 0.6666666666666666, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1045}, "3": {"facts_total": 58, "facts_retained": 32, "retention_rate": 0.5517241379310345, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.7142857142857143, "amount": 0.6666666666666666, "status": 0.6666666666666666, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1044}, "4": {"facts_total": 58, "facts_retained": 32, "retention_rate": 0.5517241379310345, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.7142857142857143, "amount": 0.6666666666666666, "status": 0.6666666666666666, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1044}}}, "reference_key_id": "keys/synthetic/syn_03003_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "508f87a419f4b708b6d957677c3ec8f39a369a431efc5130fdf9aa8501d4c113", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "9f799e2ab32d7fefb7e83375", "attempt_id": "9dc5f9bb4393", "run_group_id": "E4_core:A4|closed|qwen3b-llama3b:h4", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T06:51:33.305720Z", "task_family": "synthetic", "document_id": "syn_03003_2000", "document_hash": "dddca1a38e3224870fcfae927c701c400fd885c19123ea840ee06a76a140f407", "question_id": "q05_multi_hop", "split": "test", "cluster_id": "syn_03003_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "tokenizer_hashes": {"receiver": "b9d8872cf3a99da99e1fc4b0f71232a451632dc49ca4c6771549b4efa5c1c763"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 4, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1722, "query_tokens": 508, "message_tokens": 395, "message_utf8_bytes": 1044, "cache_tensor_bytes": null, "serialized_payload_bytes": 1044, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 74283.0914988881, "hops": {"1": 22569.42216597963, "2": 15621.60833296366, "3": 20515.35983302165, "4": 15576.701166923158}, "receiver_prefill_ms": 341.22437494806945, "ttft_ms": 35.98475002218038, "decode_ms": 414.2717090435326}, "total_wall_ms": 75075.14920691028, "ttft_ms": 35.98475002218038, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 7}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "NK-95Z-2165", "evidence_artifact": {"ledger": {"facts_total": 58, "facts_retained": 32, "retention_rate": 0.5517241379310345, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.7142857142857143, "amount": 0.6666666666666666, "status": 0.6666666666666666, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1044}, "ledger_by_hop": {"1": {"facts_total": 58, "facts_retained": 32, "retention_rate": 0.5517241379310345, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.7142857142857143, "amount": 0.6666666666666666, "status": 0.6666666666666666, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1200}, "2": {"facts_total": 58, "facts_retained": 32, "retention_rate": 0.5517241379310345, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.7142857142857143, "amount": 0.6666666666666666, "status": 0.6666666666666666, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1045}, "3": {"facts_total": 58, "facts_retained": 32, "retention_rate": 0.5517241379310345, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.7142857142857143, "amount": 0.6666666666666666, "status": 0.6666666666666666, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1044}, "4": {"facts_total": 58, "facts_retained": 32, "retention_rate": 0.5517241379310345, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.7142857142857143, "amount": 0.6666666666666666, "status": 0.6666666666666666, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1044}}}, "reference_key_id": "keys/synthetic/syn_03003_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "508f87a419f4b708b6d957677c3ec8f39a369a431efc5130fdf9aa8501d4c113", "output_hash": "bd8f809b02304c6b9dc7c1e116ecccc2328c976a5c071e89a94d661b48e73aa7"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "bc40c6efd07b6cb2bb452f22", "attempt_id": "9a164de955e0", "run_group_id": "E4_core:A4|closed|qwen3b-llama3b:h4", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T06:51:34.113510Z", "task_family": "synthetic", "document_id": "syn_03003_2000", "document_hash": "dddca1a38e3224870fcfae927c701c400fd885c19123ea840ee06a76a140f407", "question_id": "q06_negation", "split": "test", "cluster_id": "syn_03003_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "tokenizer_hashes": {"receiver": "b9d8872cf3a99da99e1fc4b0f71232a451632dc49ca4c6771549b4efa5c1c763"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 4, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1722, "query_tokens": 501, "message_tokens": 395, "message_utf8_bytes": 1044, "cache_tensor_bytes": null, "serialized_payload_bytes": 1044, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 74283.0914988881, "hops": {"1": 22569.42216597963, "2": 15621.60833296366, "3": 20515.35983302165, "4": 15576.701166923158}, "receiver_prefill_ms": 365.5174580635503, "ttft_ms": 31.651082914322615, "decode_ms": 66.01204199250787}, "total_wall_ms": 74746.85508292168, "ttft_ms": 31.651082914322615, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 2}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "No.", "evidence_artifact": {"ledger": {"facts_total": 58, "facts_retained": 32, "retention_rate": 0.5517241379310345, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.7142857142857143, "amount": 0.6666666666666666, "status": 0.6666666666666666, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1044}, "ledger_by_hop": {"1": {"facts_total": 58, "facts_retained": 32, "retention_rate": 0.5517241379310345, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.7142857142857143, "amount": 0.6666666666666666, "status": 0.6666666666666666, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1200}, "2": {"facts_total": 58, "facts_retained": 32, "retention_rate": 0.5517241379310345, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.7142857142857143, "amount": 0.6666666666666666, "status": 0.6666666666666666, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1045}, "3": {"facts_total": 58, "facts_retained": 32, "retention_rate": 0.5517241379310345, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.7142857142857143, "amount": 0.6666666666666666, "status": 0.6666666666666666, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1044}, "4": {"facts_total": 58, "facts_retained": 32, "retention_rate": 0.5517241379310345, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.7142857142857143, "amount": 0.6666666666666666, "status": 0.6666666666666666, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1044}}}, "reference_key_id": "keys/synthetic/syn_03003_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "508f87a419f4b708b6d957677c3ec8f39a369a431efc5130fdf9aa8501d4c113", "output_hash": "38178a20b470cfd18299fb1593dd4ba706f1a83321dac051d01086efd8b7a96f"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "238f03ae8873cb5400ae92ca", "attempt_id": "994c70499d7e", "run_group_id": "E4_core:A4|closed|qwen3b-llama3b:h4", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T06:51:34.595560Z", "task_family": "synthetic", "document_id": "syn_03003_2000", "document_hash": "dddca1a38e3224870fcfae927c701c400fd885c19123ea840ee06a76a140f407", "question_id": "q07_negation", "split": "test", "cluster_id": "syn_03003_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "tokenizer_hashes": {"receiver": "b9d8872cf3a99da99e1fc4b0f71232a451632dc49ca4c6771549b4efa5c1c763"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 4, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1722, "query_tokens": 499, "message_tokens": 395, "message_utf8_bytes": 1044, "cache_tensor_bytes": null, "serialized_payload_bytes": 1044, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 74283.0914988881, "hops": {"1": 22569.42216597963, "2": 15621.60833296366, "3": 20515.35983302165, "4": 15576.701166923158}, "receiver_prefill_ms": 351.8099579960108, "ttft_ms": 36.18575003929436, "decode_ms": 42.36362501978874}, "total_wall_ms": 74714.2227909062, "ttft_ms": 36.18575003929436, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "No", "evidence_artifact": {"ledger": {"facts_total": 58, "facts_retained": 32, "retention_rate": 0.5517241379310345, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.7142857142857143, "amount": 0.6666666666666666, "status": 0.6666666666666666, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1044}, "ledger_by_hop": {"1": {"facts_total": 58, "facts_retained": 32, "retention_rate": 0.5517241379310345, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.7142857142857143, "amount": 0.6666666666666666, "status": 0.6666666666666666, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1200}, "2": {"facts_total": 58, "facts_retained": 32, "retention_rate": 0.5517241379310345, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.7142857142857143, "amount": 0.6666666666666666, "status": 0.6666666666666666, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1045}, "3": {"facts_total": 58, "facts_retained": 32, "retention_rate": 0.5517241379310345, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.7142857142857143, "amount": 0.6666666666666666, "status": 0.6666666666666666, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1044}, "4": {"facts_total": 58, "facts_retained": 32, "retention_rate": 0.5517241379310345, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.7142857142857143, "amount": 0.6666666666666666, "status": 0.6666666666666666, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1044}}}, "reference_key_id": "keys/synthetic/syn_03003_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "508f87a419f4b708b6d957677c3ec8f39a369a431efc5130fdf9aa8501d4c113", "output_hash": "1ea442a134b2a184bd5d40104401f2a37fbc09ccf3f4bc9da161c6099be3691d"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "568bac165cb527e70a4d191b", "attempt_id": "e7093cb8628d", "run_group_id": "E4_core:A4|closed|qwen3b-llama3b:h4", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T06:51:35.044216Z", "task_family": "synthetic", "document_id": "syn_03003_2000", "document_hash": "dddca1a38e3224870fcfae927c701c400fd885c19123ea840ee06a76a140f407", "question_id": "q08_revision", "split": "test", "cluster_id": "syn_03003_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "tokenizer_hashes": {"receiver": "b9d8872cf3a99da99e1fc4b0f71232a451632dc49ca4c6771549b4efa5c1c763"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 4, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1722, "query_tokens": 517, "message_tokens": 395, "message_utf8_bytes": 1044, "cache_tensor_bytes": null, "serialized_payload_bytes": 1044, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 74283.0914988881, "hops": {"1": 22569.42216597963, "2": 15621.60833296366, "3": 20515.35983302165, "4": 15576.701166923158}, "receiver_prefill_ms": 349.88479199819267, "ttft_ms": 53.72324993368238, "decode_ms": 245.78537500929087}, "total_wall_ms": 74933.06645692792, "ttft_ms": 53.72324993368238, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 3}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "911,925", "evidence_artifact": {"ledger": {"facts_total": 58, "facts_retained": 32, "retention_rate": 0.5517241379310345, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.7142857142857143, "amount": 0.6666666666666666, "status": 0.6666666666666666, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1044}, "ledger_by_hop": {"1": {"facts_total": 58, "facts_retained": 32, "retention_rate": 0.5517241379310345, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.7142857142857143, "amount": 0.6666666666666666, "status": 0.6666666666666666, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1200}, "2": {"facts_total": 58, "facts_retained": 32, "retention_rate": 0.5517241379310345, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.7142857142857143, "amount": 0.6666666666666666, "status": 0.6666666666666666, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1045}, "3": {"facts_total": 58, "facts_retained": 32, "retention_rate": 0.5517241379310345, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.7142857142857143, "amount": 0.6666666666666666, "status": 0.6666666666666666, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1044}, "4": {"facts_total": 58, "facts_retained": 32, "retention_rate": 0.5517241379310345, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.7142857142857143, "amount": 0.6666666666666666, "status": 0.6666666666666666, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1044}}}, "reference_key_id": "keys/synthetic/syn_03003_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "508f87a419f4b708b6d957677c3ec8f39a369a431efc5130fdf9aa8501d4c113", "output_hash": "86be78720fc4b79f93e603e7d9e4861fcd25102612f73209fffc17d787056007"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "004126b7bbe39f0e5037c1e6", "attempt_id": "a271974f54e7", "run_group_id": "E4_core:A4|closed|qwen3b-llama3b:h4", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T06:51:35.711673Z", "task_family": "synthetic", "document_id": "syn_03003_2000", "document_hash": "dddca1a38e3224870fcfae927c701c400fd885c19123ea840ee06a76a140f407", "question_id": "q09_revision", "split": "test", "cluster_id": "syn_03003_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "tokenizer_hashes": {"receiver": "b9d8872cf3a99da99e1fc4b0f71232a451632dc49ca4c6771549b4efa5c1c763"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 4, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1722, "query_tokens": 510, "message_tokens": 395, "message_utf8_bytes": 1044, "cache_tensor_bytes": null, "serialized_payload_bytes": 1044, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 74283.0914988881, "hops": {"1": 22569.42216597963, "2": 15621.60833296366, "3": 20515.35983302165, "4": 15576.701166923158}, "receiver_prefill_ms": 368.57954203151166, "ttft_ms": 26.563541032373905, "decode_ms": 107.85124998074025}, "total_wall_ms": 74786.67337389197, "ttft_ms": 26.563541032373905, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 3}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "911,925", "evidence_artifact": {"ledger": {"facts_total": 58, "facts_retained": 32, "retention_rate": 0.5517241379310345, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.7142857142857143, "amount": 0.6666666666666666, "status": 0.6666666666666666, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1044}, "ledger_by_hop": {"1": {"facts_total": 58, "facts_retained": 32, "retention_rate": 0.5517241379310345, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.7142857142857143, "amount": 0.6666666666666666, "status": 0.6666666666666666, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1200}, "2": {"facts_total": 58, "facts_retained": 32, "retention_rate": 0.5517241379310345, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.7142857142857143, "amount": 0.6666666666666666, "status": 0.6666666666666666, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1045}, "3": {"facts_total": 58, "facts_retained": 32, "retention_rate": 0.5517241379310345, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.7142857142857143, "amount": 0.6666666666666666, "status": 0.6666666666666666, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1044}, "4": {"facts_total": 58, "facts_retained": 32, "retention_rate": 0.5517241379310345, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.7142857142857143, "amount": 0.6666666666666666, "status": 0.6666666666666666, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1044}}}, "reference_key_id": "keys/synthetic/syn_03003_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "508f87a419f4b708b6d957677c3ec8f39a369a431efc5130fdf9aa8501d4c113", "output_hash": "86be78720fc4b79f93e603e7d9e4861fcd25102612f73209fffc17d787056007"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "273dbdf12394c84c5e9a0e31", "attempt_id": "eb11a2f65ecb", "run_group_id": "E4_core:A4|closed|qwen3b-llama3b:h4", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T06:51:36.233067Z", "task_family": "synthetic", "document_id": "syn_03003_2000", "document_hash": "dddca1a38e3224870fcfae927c701c400fd885c19123ea840ee06a76a140f407", "question_id": "q10_temporal", "split": "test", "cluster_id": "syn_03003_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "tokenizer_hashes": {"receiver": "b9d8872cf3a99da99e1fc4b0f71232a451632dc49ca4c6771549b4efa5c1c763"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 4, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1722, "query_tokens": 517, "message_tokens": 395, "message_utf8_bytes": 1044, "cache_tensor_bytes": null, "serialized_payload_bytes": 1044, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 74283.0914988881, "hops": {"1": 22569.42216597963, "2": 15621.60833296366, "3": 20515.35983302165, "4": 15576.701166923158}, "receiver_prefill_ms": 339.06499994918704, "ttft_ms": 23.103208048269153, "decode_ms": 289.24920805729926}, "total_wall_ms": 74935.22154085804, "ttft_ms": 23.103208048269153, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 6}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "Gamtiszen-2", "evidence_artifact": {"ledger": {"facts_total": 58, "facts_retained": 32, "retention_rate": 0.5517241379310345, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.7142857142857143, "amount": 0.6666666666666666, "status": 0.6666666666666666, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1044}, "ledger_by_hop": {"1": {"facts_total": 58, "facts_retained": 32, "retention_rate": 0.5517241379310345, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.7142857142857143, "amount": 0.6666666666666666, "status": 0.6666666666666666, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1200}, "2": {"facts_total": 58, "facts_retained": 32, "retention_rate": 0.5517241379310345, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.7142857142857143, "amount": 0.6666666666666666, "status": 0.6666666666666666, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1045}, "3": {"facts_total": 58, "facts_retained": 32, "retention_rate": 0.5517241379310345, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.7142857142857143, "amount": 0.6666666666666666, "status": 0.6666666666666666, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1044}, "4": {"facts_total": 58, "facts_retained": 32, "retention_rate": 0.5517241379310345, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.7142857142857143, "amount": 0.6666666666666666, "status": 0.6666666666666666, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1044}}}, "reference_key_id": "keys/synthetic/syn_03003_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "508f87a419f4b708b6d957677c3ec8f39a369a431efc5130fdf9aa8501d4c113", "output_hash": "32e3af428886434823d82102e1c482765eafdf6077fe8d3fba5752ec5e29826c"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "66d200937b51bebf8c35f907", "attempt_id": "9e7df95dd930", "run_group_id": "E4_core:A4|closed|qwen3b-llama3b:h4", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T06:51:36.902254Z", "task_family": "synthetic", "document_id": "syn_03003_2000", "document_hash": "dddca1a38e3224870fcfae927c701c400fd885c19123ea840ee06a76a140f407", "question_id": "q11_aggregation", "split": "test", "cluster_id": "syn_03003_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "tokenizer_hashes": {"receiver": "b9d8872cf3a99da99e1fc4b0f71232a451632dc49ca4c6771549b4efa5c1c763"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 4, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1722, "query_tokens": 511, "message_tokens": 395, "message_utf8_bytes": 1044, "cache_tensor_bytes": null, "serialized_payload_bytes": 1044, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 74283.0914988881, "hops": {"1": 22569.42216597963, "2": 15621.60833296366, "3": 20515.35983302165, "4": 15576.701166923158}, "receiver_prefill_ms": 489.35899999924004, "ttft_ms": 24.299167096614838, "decode_ms": 163.64787507336587}, "total_wall_ms": 74961.02708193939, "ttft_ms": 24.299167096614838, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 5}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "11,165,406", "evidence_artifact": {"ledger": {"facts_total": 58, "facts_retained": 32, "retention_rate": 0.5517241379310345, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.7142857142857143, "amount": 0.6666666666666666, "status": 0.6666666666666666, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1044}, "ledger_by_hop": {"1": {"facts_total": 58, "facts_retained": 32, "retention_rate": 0.5517241379310345, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.7142857142857143, "amount": 0.6666666666666666, "status": 0.6666666666666666, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1200}, "2": {"facts_total": 58, "facts_retained": 32, "retention_rate": 0.5517241379310345, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.7142857142857143, "amount": 0.6666666666666666, "status": 0.6666666666666666, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1045}, "3": {"facts_total": 58, "facts_retained": 32, "retention_rate": 0.5517241379310345, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.7142857142857143, "amount": 0.6666666666666666, "status": 0.6666666666666666, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1044}, "4": {"facts_total": 58, "facts_retained": 32, "retention_rate": 0.5517241379310345, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.7142857142857143, "amount": 0.6666666666666666, "status": 0.6666666666666666, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1044}}}, "reference_key_id": "keys/synthetic/syn_03003_2000.json", "task_scores": {"correct": 0.0, "abstained": 0.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "508f87a419f4b708b6d957677c3ec8f39a369a431efc5130fdf9aa8501d4c113", "output_hash": "557c867fd18f226d3d49763e0ac0fd86a4d623de71032ef41a7e1439fa3c20e7"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "19b2e347bf766eea817351cf", "attempt_id": "05e86da50c40", "run_group_id": "E4_core:A4|closed|qwen3b-llama3b:h4", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T06:51:37.597372Z", "task_family": "synthetic", "document_id": "syn_03003_2000", "document_hash": "dddca1a38e3224870fcfae927c701c400fd885c19123ea840ee06a76a140f407", "question_id": "q12_unit_conversion", "split": "test", "cluster_id": "syn_03003_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "tokenizer_hashes": {"receiver": "b9d8872cf3a99da99e1fc4b0f71232a451632dc49ca4c6771549b4efa5c1c763"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 4, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1722, "query_tokens": 508, "message_tokens": 395, "message_utf8_bytes": 1044, "cache_tensor_bytes": null, "serialized_payload_bytes": 1044, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 74283.0914988881, "hops": {"1": 22569.42216597963, "2": 15621.60833296366, "3": 20515.35983302165, "4": 15576.701166923158}, "receiver_prefill_ms": 391.55716705136, "ttft_ms": 17.645334010012448, "decode_ms": 87.160374969244}, "total_wall_ms": 74780.19070683513, "ttft_ms": 17.645334010012448, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 3}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "17.43", "evidence_artifact": {"ledger": {"facts_total": 58, "facts_retained": 32, "retention_rate": 0.5517241379310345, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.7142857142857143, "amount": 0.6666666666666666, "status": 0.6666666666666666, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1044}, "ledger_by_hop": {"1": {"facts_total": 58, "facts_retained": 32, "retention_rate": 0.5517241379310345, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.7142857142857143, "amount": 0.6666666666666666, "status": 0.6666666666666666, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1200}, "2": {"facts_total": 58, "facts_retained": 32, "retention_rate": 0.5517241379310345, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.7142857142857143, "amount": 0.6666666666666666, "status": 0.6666666666666666, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1045}, "3": {"facts_total": 58, "facts_retained": 32, "retention_rate": 0.5517241379310345, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.7142857142857143, "amount": 0.6666666666666666, "status": 0.6666666666666666, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1044}, "4": {"facts_total": 58, "facts_retained": 32, "retention_rate": 0.5517241379310345, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.7142857142857143, "amount": 0.6666666666666666, "status": 0.6666666666666666, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1044}}}, "reference_key_id": "keys/synthetic/syn_03003_2000.json", "task_scores": {"correct": 0.0, "abstained": 0.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "508f87a419f4b708b6d957677c3ec8f39a369a431efc5130fdf9aa8501d4c113", "output_hash": "1cd677e6c296f31fca744a7c4d96b5514fa6b036913240b8d689f1e277900d17"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "76de4acabbb465b031af8ffd", "attempt_id": "b94e49ddb6a0", "run_group_id": "E4_core:A4|closed|qwen3b-llama3b:h4", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T06:51:38.112264Z", "task_family": "synthetic", "document_id": "syn_03003_2000", "document_hash": "dddca1a38e3224870fcfae927c701c400fd885c19123ea840ee06a76a140f407", "question_id": "q13_unanswerable", "split": "test", "cluster_id": "syn_03003_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "tokenizer_hashes": {"receiver": "b9d8872cf3a99da99e1fc4b0f71232a451632dc49ca4c6771549b4efa5c1c763"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 4, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1722, "query_tokens": 512, "message_tokens": 395, "message_utf8_bytes": 1044, "cache_tensor_bytes": null, "serialized_payload_bytes": 1044, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 74283.0914988881, "hops": {"1": 22569.42216597963, "2": 15621.60833296366, "3": 20515.35983302165, "4": 15576.701166923158}, "receiver_prefill_ms": 349.9918340239674, "ttft_ms": 53.727917023934424, "decode_ms": 423.1307079317048}, "total_wall_ms": 75110.6285819551, "ttft_ms": 53.727917023934424, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 7}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "GR-79S-5839", "evidence_artifact": {"ledger": {"facts_total": 58, "facts_retained": 32, "retention_rate": 0.5517241379310345, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.7142857142857143, "amount": 0.6666666666666666, "status": 0.6666666666666666, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1044}, "ledger_by_hop": {"1": {"facts_total": 58, "facts_retained": 32, "retention_rate": 0.5517241379310345, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.7142857142857143, "amount": 0.6666666666666666, "status": 0.6666666666666666, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1200}, "2": {"facts_total": 58, "facts_retained": 32, "retention_rate": 0.5517241379310345, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.7142857142857143, "amount": 0.6666666666666666, "status": 0.6666666666666666, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1045}, "3": {"facts_total": 58, "facts_retained": 32, "retention_rate": 0.5517241379310345, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.7142857142857143, "amount": 0.6666666666666666, "status": 0.6666666666666666, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1044}, "4": {"facts_total": 58, "facts_retained": 32, "retention_rate": 0.5517241379310345, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.7142857142857143, "amount": 0.6666666666666666, "status": 0.6666666666666666, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1044}}}, "reference_key_id": "keys/synthetic/syn_03003_2000.json", "task_scores": {"correct": 0.0, "abstained": 0.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "508f87a419f4b708b6d957677c3ec8f39a369a431efc5130fdf9aa8501d4c113", "output_hash": "720cc40cd69e85e9728f6bea00b17a95ca07eee643e143a131840763958303cd"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "d8b46fc23ccdbd9664793f4e", "attempt_id": "4f796e87e8eb", "run_group_id": "E4_core:A4|closed|qwen3b-llama3b:h4", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T06:51:38.957863Z", "task_family": "synthetic", "document_id": "syn_03003_2000", "document_hash": "dddca1a38e3224870fcfae927c701c400fd885c19123ea840ee06a76a140f407", "question_id": "q14_unanswerable", "split": "test", "cluster_id": "syn_03003_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "tokenizer_hashes": {"receiver": "b9d8872cf3a99da99e1fc4b0f71232a451632dc49ca4c6771549b4efa5c1c763"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 4, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1722, "query_tokens": 513, "message_tokens": 395, "message_utf8_bytes": 1044, "cache_tensor_bytes": null, "serialized_payload_bytes": 1044, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 74283.0914988881, "hops": {"1": 22569.42216597963, "2": 15621.60833296366, "3": 20515.35983302165, "4": 15576.701166923158}, "receiver_prefill_ms": 349.54891609959304, "ttft_ms": 34.435750101692975, "decode_ms": 21.836208063177764}, "total_wall_ms": 74689.53966593836, "ttft_ms": 34.435750101692975, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 58, "facts_retained": 32, "retention_rate": 0.5517241379310345, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.7142857142857143, "amount": 0.6666666666666666, "status": 0.6666666666666666, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1044}, "ledger_by_hop": {"1": {"facts_total": 58, "facts_retained": 32, "retention_rate": 0.5517241379310345, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.7142857142857143, "amount": 0.6666666666666666, "status": 0.6666666666666666, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1200}, "2": {"facts_total": 58, "facts_retained": 32, "retention_rate": 0.5517241379310345, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.7142857142857143, "amount": 0.6666666666666666, "status": 0.6666666666666666, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1045}, "3": {"facts_total": 58, "facts_retained": 32, "retention_rate": 0.5517241379310345, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.7142857142857143, "amount": 0.6666666666666666, "status": 0.6666666666666666, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1044}, "4": {"facts_total": 58, "facts_retained": 32, "retention_rate": 0.5517241379310345, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.7142857142857143, "amount": 0.6666666666666666, "status": 0.6666666666666666, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1044}}}, "reference_key_id": "keys/synthetic/syn_03003_2000.json", "task_scores": {"correct": 1.0, "abstained": 1.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "508f87a419f4b708b6d957677c3ec8f39a369a431efc5130fdf9aa8501d4c113", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "a2b63139128771ebf4188f1d", "attempt_id": "ce96f8e60f3b", "run_group_id": "E4_core:A4|closed|qwen3b-llama3b:h8", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T06:52:49.422133Z", "task_family": "synthetic", "document_id": "syn_03003_2000", "document_hash": "dddca1a38e3224870fcfae927c701c400fd885c19123ea840ee06a76a140f407", "question_id": "q00_single_fact", "split": "test", "cluster_id": "syn_03003_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "tokenizer_hashes": {"receiver": "b9d8872cf3a99da99e1fc4b0f71232a451632dc49ca4c6771549b4efa5c1c763"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 8, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1722, "query_tokens": 490, "message_tokens": 385, "message_utf8_bytes": 1049, "cache_tensor_bytes": null, "serialized_payload_bytes": 1049, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 144318.40237381402, "hops": {"1": 22569.42216597963, "2": 15621.60833296366, "3": 20515.35983302165, "4": 15576.701166923158, "5": 20001.019624993205, "6": 14886.58912491519, "7": 20005.93258300796, "8": 15141.76954200957}, "receiver_prefill_ms": 374.8163749696687, "ttft_ms": 21.575833088718355, "decode_ms": 311.5568329812959}, "total_wall_ms": 145026.956915739, "ttft_ms": 21.575833088718355, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 7}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "GR-79S-5839", "evidence_artifact": {"ledger": {"facts_total": 58, "facts_retained": 32, "retention_rate": 0.5517241379310345, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.7142857142857143, "amount": 0.6666666666666666, "status": 0.6666666666666666, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1049}, "ledger_by_hop": {"1": {"facts_total": 58, "facts_retained": 32, "retention_rate": 0.5517241379310345, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.7142857142857143, "amount": 0.6666666666666666, "status": 0.6666666666666666, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1200}, "2": {"facts_total": 58, "facts_retained": 32, "retention_rate": 0.5517241379310345, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.7142857142857143, "amount": 0.6666666666666666, "status": 0.6666666666666666, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1045}, "3": {"facts_total": 58, "facts_retained": 32, "retention_rate": 0.5517241379310345, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.7142857142857143, "amount": 0.6666666666666666, "status": 0.6666666666666666, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1044}, "4": {"facts_total": 58, "facts_retained": 32, "retention_rate": 0.5517241379310345, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.7142857142857143, "amount": 0.6666666666666666, "status": 0.6666666666666666, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1044}, "5": {"facts_total": 58, "facts_retained": 32, "retention_rate": 0.5517241379310345, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.7142857142857143, "amount": 0.6666666666666666, "status": 0.6666666666666666, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1020}, "6": {"facts_total": 58, "facts_retained": 32, "retention_rate": 0.5517241379310345, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.7142857142857143, "amount": 0.6666666666666666, "status": 0.6666666666666666, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1049}, "7": {"facts_total": 58, "facts_retained": 32, "retention_rate": 0.5517241379310345, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.7142857142857143, "amount": 0.6666666666666666, "status": 0.6666666666666666, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1049}, "8": {"facts_total": 58, "facts_retained": 32, "retention_rate": 0.5517241379310345, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.7142857142857143, "amount": 0.6666666666666666, "status": 0.6666666666666666, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1049}}}, "reference_key_id": "keys/synthetic/syn_03003_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "bcdef32d72e140b171048e34a3d65f86a0141ce1e4df6b6fc2128c34ee471bef", "output_hash": "720cc40cd69e85e9728f6bea00b17a95ca07eee643e143a131840763958303cd"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "810da7e4153ea25c0d49cc6f", "attempt_id": "0206d3311170", "run_group_id": "E4_core:A4|closed|qwen3b-llama3b:h8", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T06:52:50.149113Z", "task_family": "synthetic", "document_id": "syn_03003_2000", "document_hash": "dddca1a38e3224870fcfae927c701c400fd885c19123ea840ee06a76a140f407", "question_id": "q01_single_fact", "split": "test", "cluster_id": "syn_03003_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "tokenizer_hashes": {"receiver": "b9d8872cf3a99da99e1fc4b0f71232a451632dc49ca4c6771549b4efa5c1c763"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 8, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1722, "query_tokens": 494, "message_tokens": 385, "message_utf8_bytes": 1049, "cache_tensor_bytes": null, "serialized_payload_bytes": 1049, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 144318.40237381402, "hops": {"1": 22569.42216597963, "2": 15621.60833296366, "3": 20515.35983302165, "4": 15576.701166923158, "5": 20001.019624993205, "6": 14886.58912491519, "7": 20005.93258300796, "8": 15141.76954200957}, "receiver_prefill_ms": 479.52699998859316, "ttft_ms": 25.87533392943442, "decode_ms": 294.351291959174}, "total_wall_ms": 145118.7199158594, "ttft_ms": 25.87533392943442, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 9}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "Lot 344-B, Témiscaming", "evidence_artifact": {"ledger": {"facts_total": 58, "facts_retained": 32, "retention_rate": 0.5517241379310345, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.7142857142857143, "amount": 0.6666666666666666, "status": 0.6666666666666666, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1049}, "ledger_by_hop": {"1": {"facts_total": 58, "facts_retained": 32, "retention_rate": 0.5517241379310345, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.7142857142857143, "amount": 0.6666666666666666, "status": 0.6666666666666666, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1200}, "2": {"facts_total": 58, "facts_retained": 32, "retention_rate": 0.5517241379310345, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.7142857142857143, "amount": 0.6666666666666666, "status": 0.6666666666666666, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1045}, "3": {"facts_total": 58, "facts_retained": 32, "retention_rate": 0.5517241379310345, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.7142857142857143, "amount": 0.6666666666666666, "status": 0.6666666666666666, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1044}, "4": {"facts_total": 58, "facts_retained": 32, "retention_rate": 0.5517241379310345, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.7142857142857143, "amount": 0.6666666666666666, "status": 0.6666666666666666, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1044}, "5": {"facts_total": 58, "facts_retained": 32, "retention_rate": 0.5517241379310345, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.7142857142857143, "amount": 0.6666666666666666, "status": 0.6666666666666666, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1020}, "6": {"facts_total": 58, "facts_retained": 32, "retention_rate": 0.5517241379310345, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.7142857142857143, "amount": 0.6666666666666666, "status": 0.6666666666666666, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1049}, "7": {"facts_total": 58, "facts_retained": 32, "retention_rate": 0.5517241379310345, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.7142857142857143, "amount": 0.6666666666666666, "status": 0.6666666666666666, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1049}, "8": {"facts_total": 58, "facts_retained": 32, "retention_rate": 0.5517241379310345, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.7142857142857143, "amount": 0.6666666666666666, "status": 0.6666666666666666, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1049}}}, "reference_key_id": "keys/synthetic/syn_03003_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 0.8571428571428571}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "bcdef32d72e140b171048e34a3d65f86a0141ce1e4df6b6fc2128c34ee471bef", "output_hash": "0197d491a5a04a142dc8e372360735a1d41757e7a5aea7e81f23b0e1ef6ee010"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "fea1c76ccdf96a0728671bbd", "attempt_id": "bc8b0439d3b4", "run_group_id": "E4_core:A4|closed|qwen3b-llama3b:h8", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T06:52:50.967160Z", "task_family": "synthetic", "document_id": "syn_03003_2000", "document_hash": "dddca1a38e3224870fcfae927c701c400fd885c19123ea840ee06a76a140f407", "question_id": "q02_exact_copy", "split": "test", "cluster_id": "syn_03003_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "tokenizer_hashes": {"receiver": "b9d8872cf3a99da99e1fc4b0f71232a451632dc49ca4c6771549b4efa5c1c763"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 8, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1722, "query_tokens": 494, "message_tokens": 385, "message_utf8_bytes": 1049, "cache_tensor_bytes": null, "serialized_payload_bytes": 1049, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 144318.40237381402, "hops": {"1": 22569.42216597963, "2": 15621.60833296366, "3": 20515.35983302165, "4": 15576.701166923158, "5": 20001.019624993205, "6": 14886.58912491519, "7": 20005.93258300796, "8": 15141.76954200957}, "receiver_prefill_ms": 373.6615829402581, "ttft_ms": 20.860707969404757, "decode_ms": 18.87545792851597}, "total_wall_ms": 144732.30033181608, "ttft_ms": 20.860707969404757, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 58, "facts_retained": 32, "retention_rate": 0.5517241379310345, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.7142857142857143, "amount": 0.6666666666666666, "status": 0.6666666666666666, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1049}, "ledger_by_hop": {"1": {"facts_total": 58, "facts_retained": 32, "retention_rate": 0.5517241379310345, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.7142857142857143, "amount": 0.6666666666666666, "status": 0.6666666666666666, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1200}, "2": {"facts_total": 58, "facts_retained": 32, "retention_rate": 0.5517241379310345, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.7142857142857143, "amount": 0.6666666666666666, "status": 0.6666666666666666, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1045}, "3": {"facts_total": 58, "facts_retained": 32, "retention_rate": 0.5517241379310345, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.7142857142857143, "amount": 0.6666666666666666, "status": 0.6666666666666666, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1044}, "4": {"facts_total": 58, "facts_retained": 32, "retention_rate": 0.5517241379310345, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.7142857142857143, "amount": 0.6666666666666666, "status": 0.6666666666666666, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1044}, "5": {"facts_total": 58, "facts_retained": 32, "retention_rate": 0.5517241379310345, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.7142857142857143, "amount": 0.6666666666666666, "status": 0.6666666666666666, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1020}, "6": {"facts_total": 58, "facts_retained": 32, "retention_rate": 0.5517241379310345, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.7142857142857143, "amount": 0.6666666666666666, "status": 0.6666666666666666, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1049}, "7": {"facts_total": 58, "facts_retained": 32, "retention_rate": 0.5517241379310345, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.7142857142857143, "amount": 0.6666666666666666, "status": 0.6666666666666666, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1049}, "8": {"facts_total": 58, "facts_retained": 32, "retention_rate": 0.5517241379310345, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.7142857142857143, "amount": 0.6666666666666666, "status": 0.6666666666666666, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1049}}}, "reference_key_id": "keys/synthetic/syn_03003_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "bcdef32d72e140b171048e34a3d65f86a0141ce1e4df6b6fc2128c34ee471bef", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "7331edf26f12dea5f799e89e", "attempt_id": "c72a7cee6cde", "run_group_id": "E4_core:A4|closed|qwen3b-llama3b:h8", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T06:52:51.398823Z", "task_family": "synthetic", "document_id": "syn_03003_2000", "document_hash": "dddca1a38e3224870fcfae927c701c400fd885c19123ea840ee06a76a140f407", "question_id": "q03_exact_copy", "split": "test", "cluster_id": "syn_03003_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "tokenizer_hashes": {"receiver": "b9d8872cf3a99da99e1fc4b0f71232a451632dc49ca4c6771549b4efa5c1c763"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 8, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1722, "query_tokens": 489, "message_tokens": 385, "message_utf8_bytes": 1049, "cache_tensor_bytes": null, "serialized_payload_bytes": 1049, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 144318.40237381402, "hops": {"1": 22569.42216597963, "2": 15621.60833296366, "3": 20515.35983302165, "4": 15576.701166923158, "5": 20001.019624993205, "6": 14886.58912491519, "7": 20005.93258300796, "8": 15141.76954200957}, "receiver_prefill_ms": 360.77554197981954, "ttft_ms": 52.014625049196184, "decode_ms": 374.56154206302017}, "total_wall_ms": 145106.32349876687, "ttft_ms": 52.014625049196184, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 6}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "Wilhelmina Achterberg", "evidence_artifact": {"ledger": {"facts_total": 58, "facts_retained": 32, "retention_rate": 0.5517241379310345, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.7142857142857143, "amount": 0.6666666666666666, "status": 0.6666666666666666, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1049}, "ledger_by_hop": {"1": {"facts_total": 58, "facts_retained": 32, "retention_rate": 0.5517241379310345, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.7142857142857143, "amount": 0.6666666666666666, "status": 0.6666666666666666, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1200}, "2": {"facts_total": 58, "facts_retained": 32, "retention_rate": 0.5517241379310345, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.7142857142857143, "amount": 0.6666666666666666, "status": 0.6666666666666666, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1045}, "3": {"facts_total": 58, "facts_retained": 32, "retention_rate": 0.5517241379310345, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.7142857142857143, "amount": 0.6666666666666666, "status": 0.6666666666666666, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1044}, "4": {"facts_total": 58, "facts_retained": 32, "retention_rate": 0.5517241379310345, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.7142857142857143, "amount": 0.6666666666666666, "status": 0.6666666666666666, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1044}, "5": {"facts_total": 58, "facts_retained": 32, "retention_rate": 0.5517241379310345, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.7142857142857143, "amount": 0.6666666666666666, "status": 0.6666666666666666, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1020}, "6": {"facts_total": 58, "facts_retained": 32, "retention_rate": 0.5517241379310345, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.7142857142857143, "amount": 0.6666666666666666, "status": 0.6666666666666666, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1049}, "7": {"facts_total": 58, "facts_retained": 32, "retention_rate": 0.5517241379310345, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.7142857142857143, "amount": 0.6666666666666666, "status": 0.6666666666666666, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1049}, "8": {"facts_total": 58, "facts_retained": 32, "retention_rate": 0.5517241379310345, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.7142857142857143, "amount": 0.6666666666666666, "status": 0.6666666666666666, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1049}}}, "reference_key_id": "keys/synthetic/syn_03003_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "bcdef32d72e140b171048e34a3d65f86a0141ce1e4df6b6fc2128c34ee471bef", "output_hash": "2a62606853c315bba8769371d195e3ab3d6a8cd7ef959205575c5119740850f8"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "9429a2e08a44179c83f35a0b", "attempt_id": "f58458490e43", "run_group_id": "E4_core:A4|closed|qwen3b-llama3b:h8", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T06:52:52.206345Z", "task_family": "synthetic", "document_id": "syn_03003_2000", "document_hash": "dddca1a38e3224870fcfae927c701c400fd885c19123ea840ee06a76a140f407", "question_id": "q04_multi_hop", "split": "test", "cluster_id": "syn_03003_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "tokenizer_hashes": {"receiver": "b9d8872cf3a99da99e1fc4b0f71232a451632dc49ca4c6771549b4efa5c1c763"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 8, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1722, "query_tokens": 498, "message_tokens": 385, "message_utf8_bytes": 1049, "cache_tensor_bytes": null, "serialized_payload_bytes": 1049, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 144318.40237381402, "hops": {"1": 22569.42216597963, "2": 15621.60833296366, "3": 20515.35983302165, "4": 15576.701166923158, "5": 20001.019624993205, "6": 14886.58912491519, "7": 20005.93258300796, "8": 15141.76954200957}, "receiver_prefill_ms": 357.78041707817465, "ttft_ms": 19.548874930478632, "decode_ms": 37.00450004544109}, "total_wall_ms": 144733.25012379792, "ttft_ms": 19.548874930478632, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 58, "facts_retained": 32, "retention_rate": 0.5517241379310345, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.7142857142857143, "amount": 0.6666666666666666, "status": 0.6666666666666666, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1049}, "ledger_by_hop": {"1": {"facts_total": 58, "facts_retained": 32, "retention_rate": 0.5517241379310345, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.7142857142857143, "amount": 0.6666666666666666, "status": 0.6666666666666666, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1200}, "2": {"facts_total": 58, "facts_retained": 32, "retention_rate": 0.5517241379310345, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.7142857142857143, "amount": 0.6666666666666666, "status": 0.6666666666666666, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1045}, "3": {"facts_total": 58, "facts_retained": 32, "retention_rate": 0.5517241379310345, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.7142857142857143, "amount": 0.6666666666666666, "status": 0.6666666666666666, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1044}, "4": {"facts_total": 58, "facts_retained": 32, "retention_rate": 0.5517241379310345, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.7142857142857143, "amount": 0.6666666666666666, "status": 0.6666666666666666, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1044}, "5": {"facts_total": 58, "facts_retained": 32, "retention_rate": 0.5517241379310345, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.7142857142857143, "amount": 0.6666666666666666, "status": 0.6666666666666666, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1020}, "6": {"facts_total": 58, "facts_retained": 32, "retention_rate": 0.5517241379310345, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.7142857142857143, "amount": 0.6666666666666666, "status": 0.6666666666666666, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1049}, "7": {"facts_total": 58, "facts_retained": 32, "retention_rate": 0.5517241379310345, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.7142857142857143, "amount": 0.6666666666666666, "status": 0.6666666666666666, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1049}, "8": {"facts_total": 58, "facts_retained": 32, "retention_rate": 0.5517241379310345, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.7142857142857143, "amount": 0.6666666666666666, "status": 0.6666666666666666, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1049}}}, "reference_key_id": "keys/synthetic/syn_03003_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "bcdef32d72e140b171048e34a3d65f86a0141ce1e4df6b6fc2128c34ee471bef", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "060c0136bfb1d7a8a809ad6b", "attempt_id": "b4a175114843", "run_group_id": "E4_core:A4|closed|qwen3b-llama3b:h8", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T06:52:52.637985Z", "task_family": "synthetic", "document_id": "syn_03003_2000", "document_hash": "dddca1a38e3224870fcfae927c701c400fd885c19123ea840ee06a76a140f407", "question_id": "q05_multi_hop", "split": "test", "cluster_id": "syn_03003_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "tokenizer_hashes": {"receiver": "b9d8872cf3a99da99e1fc4b0f71232a451632dc49ca4c6771549b4efa5c1c763"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 8, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1722, "query_tokens": 498, "message_tokens": 385, "message_utf8_bytes": 1049, "cache_tensor_bytes": null, "serialized_payload_bytes": 1049, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 144318.40237381402, "hops": {"1": 22569.42216597963, "2": 15621.60833296366, "3": 20515.35983302165, "4": 15576.701166923158, "5": 20001.019624993205, "6": 14886.58912491519, "7": 20005.93258300796, "8": 15141.76954200957}, "receiver_prefill_ms": 352.7799170697108, "ttft_ms": 18.84654094465077, "decode_ms": 363.57541603501886}, "total_wall_ms": 145054.4846658595, "ttft_ms": 18.84654094465077, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 7}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "NK-95Z-2165", "evidence_artifact": {"ledger": {"facts_total": 58, "facts_retained": 32, "retention_rate": 0.5517241379310345, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.7142857142857143, "amount": 0.6666666666666666, "status": 0.6666666666666666, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1049}, "ledger_by_hop": {"1": {"facts_total": 58, "facts_retained": 32, "retention_rate": 0.5517241379310345, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.7142857142857143, "amount": 0.6666666666666666, "status": 0.6666666666666666, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1200}, "2": {"facts_total": 58, "facts_retained": 32, "retention_rate": 0.5517241379310345, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.7142857142857143, "amount": 0.6666666666666666, "status": 0.6666666666666666, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1045}, "3": {"facts_total": 58, "facts_retained": 32, "retention_rate": 0.5517241379310345, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.7142857142857143, "amount": 0.6666666666666666, "status": 0.6666666666666666, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1044}, "4": {"facts_total": 58, "facts_retained": 32, "retention_rate": 0.5517241379310345, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.7142857142857143, "amount": 0.6666666666666666, "status": 0.6666666666666666, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1044}, "5": {"facts_total": 58, "facts_retained": 32, "retention_rate": 0.5517241379310345, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.7142857142857143, "amount": 0.6666666666666666, "status": 0.6666666666666666, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1020}, "6": {"facts_total": 58, "facts_retained": 32, "retention_rate": 0.5517241379310345, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.7142857142857143, "amount": 0.6666666666666666, "status": 0.6666666666666666, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1049}, "7": {"facts_total": 58, "facts_retained": 32, "retention_rate": 0.5517241379310345, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.7142857142857143, "amount": 0.6666666666666666, "status": 0.6666666666666666, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1049}, "8": {"facts_total": 58, "facts_retained": 32, "retention_rate": 0.5517241379310345, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.7142857142857143, "amount": 0.6666666666666666, "status": 0.6666666666666666, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1049}}}, "reference_key_id": "keys/synthetic/syn_03003_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "bcdef32d72e140b171048e34a3d65f86a0141ce1e4df6b6fc2128c34ee471bef", "output_hash": "bd8f809b02304c6b9dc7c1e116ecccc2328c976a5c071e89a94d661b48e73aa7"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "99d790a78c1ad8be99095197", "attempt_id": "0ef1b4b7eeb0", "run_group_id": "E4_core:A4|closed|qwen3b-llama3b:h8", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T06:52:53.391997Z", "task_family": "synthetic", "document_id": "syn_03003_2000", "document_hash": "dddca1a38e3224870fcfae927c701c400fd885c19123ea840ee06a76a140f407", "question_id": "q06_negation", "split": "test", "cluster_id": "syn_03003_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "tokenizer_hashes": {"receiver": "b9d8872cf3a99da99e1fc4b0f71232a451632dc49ca4c6771549b4efa5c1c763"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 8, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1722, "query_tokens": 491, "message_tokens": 385, "message_utf8_bytes": 1049, "cache_tensor_bytes": null, "serialized_payload_bytes": 1049, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 144318.40237381402, "hops": {"1": 22569.42216597963, "2": 15621.60833296366, "3": 20515.35983302165, "4": 15576.701166923158, "5": 20001.019624993205, "6": 14886.58912491519, "7": 20005.93258300796, "8": 15141.76954200957}, "receiver_prefill_ms": 480.49637489020824, "ttft_ms": 39.940041955560446, "decode_ms": 59.7030830103904}, "total_wall_ms": 144899.1292907158, "ttft_ms": 39.940041955560446, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 2}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "No.", "evidence_artifact": {"ledger": {"facts_total": 58, "facts_retained": 32, "retention_rate": 0.5517241379310345, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.7142857142857143, "amount": 0.6666666666666666, "status": 0.6666666666666666, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1049}, "ledger_by_hop": {"1": {"facts_total": 58, "facts_retained": 32, "retention_rate": 0.5517241379310345, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.7142857142857143, "amount": 0.6666666666666666, "status": 0.6666666666666666, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1200}, "2": {"facts_total": 58, "facts_retained": 32, "retention_rate": 0.5517241379310345, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.7142857142857143, "amount": 0.6666666666666666, "status": 0.6666666666666666, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1045}, "3": {"facts_total": 58, "facts_retained": 32, "retention_rate": 0.5517241379310345, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.7142857142857143, "amount": 0.6666666666666666, "status": 0.6666666666666666, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1044}, "4": {"facts_total": 58, "facts_retained": 32, "retention_rate": 0.5517241379310345, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.7142857142857143, "amount": 0.6666666666666666, "status": 0.6666666666666666, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1044}, "5": {"facts_total": 58, "facts_retained": 32, "retention_rate": 0.5517241379310345, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.7142857142857143, "amount": 0.6666666666666666, "status": 0.6666666666666666, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1020}, "6": {"facts_total": 58, "facts_retained": 32, "retention_rate": 0.5517241379310345, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.7142857142857143, "amount": 0.6666666666666666, "status": 0.6666666666666666, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1049}, "7": {"facts_total": 58, "facts_retained": 32, "retention_rate": 0.5517241379310345, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.7142857142857143, "amount": 0.6666666666666666, "status": 0.6666666666666666, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1049}, "8": {"facts_total": 58, "facts_retained": 32, "retention_rate": 0.5517241379310345, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.7142857142857143, "amount": 0.6666666666666666, "status": 0.6666666666666666, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1049}}}, "reference_key_id": "keys/synthetic/syn_03003_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "bcdef32d72e140b171048e34a3d65f86a0141ce1e4df6b6fc2128c34ee471bef", "output_hash": "38178a20b470cfd18299fb1593dd4ba706f1a83321dac051d01086efd8b7a96f"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "e5e0c2db524f37af2207ee0f", "attempt_id": "311667f254d0", "run_group_id": "E4_core:A4|closed|qwen3b-llama3b:h8", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T06:52:53.990174Z", "task_family": "synthetic", "document_id": "syn_03003_2000", "document_hash": "dddca1a38e3224870fcfae927c701c400fd885c19123ea840ee06a76a140f407", "question_id": "q07_negation", "split": "test", "cluster_id": "syn_03003_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "tokenizer_hashes": {"receiver": "b9d8872cf3a99da99e1fc4b0f71232a451632dc49ca4c6771549b4efa5c1c763"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 8, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1722, "query_tokens": 489, "message_tokens": 385, "message_utf8_bytes": 1049, "cache_tensor_bytes": null, "serialized_payload_bytes": 1049, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 144318.40237381402, "hops": {"1": 22569.42216597963, "2": 15621.60833296366, "3": 20515.35983302165, "4": 15576.701166923158, "5": 20001.019624993205, "6": 14886.58912491519, "7": 20005.93258300796, "8": 15141.76954200957}, "receiver_prefill_ms": 382.0356250507757, "ttft_ms": 18.059125053696334, "decode_ms": 55.460624978877604}, "total_wall_ms": 144774.4695817819, "ttft_ms": 18.059125053696334, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 2}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "No.", "evidence_artifact": {"ledger": {"facts_total": 58, "facts_retained": 32, "retention_rate": 0.5517241379310345, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.7142857142857143, "amount": 0.6666666666666666, "status": 0.6666666666666666, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1049}, "ledger_by_hop": {"1": {"facts_total": 58, "facts_retained": 32, "retention_rate": 0.5517241379310345, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.7142857142857143, "amount": 0.6666666666666666, "status": 0.6666666666666666, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1200}, "2": {"facts_total": 58, "facts_retained": 32, "retention_rate": 0.5517241379310345, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.7142857142857143, "amount": 0.6666666666666666, "status": 0.6666666666666666, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1045}, "3": {"facts_total": 58, "facts_retained": 32, "retention_rate": 0.5517241379310345, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.7142857142857143, "amount": 0.6666666666666666, "status": 0.6666666666666666, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1044}, "4": {"facts_total": 58, "facts_retained": 32, "retention_rate": 0.5517241379310345, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.7142857142857143, "amount": 0.6666666666666666, "status": 0.6666666666666666, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1044}, "5": {"facts_total": 58, "facts_retained": 32, "retention_rate": 0.5517241379310345, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.7142857142857143, "amount": 0.6666666666666666, "status": 0.6666666666666666, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1020}, "6": {"facts_total": 58, "facts_retained": 32, "retention_rate": 0.5517241379310345, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.7142857142857143, "amount": 0.6666666666666666, "status": 0.6666666666666666, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1049}, "7": {"facts_total": 58, "facts_retained": 32, "retention_rate": 0.5517241379310345, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.7142857142857143, "amount": 0.6666666666666666, "status": 0.6666666666666666, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1049}, "8": {"facts_total": 58, "facts_retained": 32, "retention_rate": 0.5517241379310345, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.7142857142857143, "amount": 0.6666666666666666, "status": 0.6666666666666666, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1049}}}, "reference_key_id": "keys/synthetic/syn_03003_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "bcdef32d72e140b171048e34a3d65f86a0141ce1e4df6b6fc2128c34ee471bef", "output_hash": "38178a20b470cfd18299fb1593dd4ba706f1a83321dac051d01086efd8b7a96f"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "5e3174c24f68e67c174908b7", "attempt_id": "40fd36000bb9", "run_group_id": "E4_core:A4|closed|qwen3b-llama3b:h8", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T06:52:54.463843Z", "task_family": "synthetic", "document_id": "syn_03003_2000", "document_hash": "dddca1a38e3224870fcfae927c701c400fd885c19123ea840ee06a76a140f407", "question_id": "q08_revision", "split": "test", "cluster_id": "syn_03003_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "tokenizer_hashes": {"receiver": "b9d8872cf3a99da99e1fc4b0f71232a451632dc49ca4c6771549b4efa5c1c763"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 8, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1722, "query_tokens": 507, "message_tokens": 385, "message_utf8_bytes": 1049, "cache_tensor_bytes": null, "serialized_payload_bytes": 1049, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 144318.40237381402, "hops": {"1": 22569.42216597963, "2": 15621.60833296366, "3": 20515.35983302165, "4": 15576.701166923158, "5": 20001.019624993205, "6": 14886.58912491519, "7": 20005.93258300796, "8": 15141.76954200957}, "receiver_prefill_ms": 341.51833306532353, "ttft_ms": 37.60129201691598, "decode_ms": 158.0015419749543}, "total_wall_ms": 144856.28633270971, "ttft_ms": 37.60129201691598, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 3}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "911,925", "evidence_artifact": {"ledger": {"facts_total": 58, "facts_retained": 32, "retention_rate": 0.5517241379310345, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.7142857142857143, "amount": 0.6666666666666666, "status": 0.6666666666666666, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1049}, "ledger_by_hop": {"1": {"facts_total": 58, "facts_retained": 32, "retention_rate": 0.5517241379310345, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.7142857142857143, "amount": 0.6666666666666666, "status": 0.6666666666666666, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1200}, "2": {"facts_total": 58, "facts_retained": 32, "retention_rate": 0.5517241379310345, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.7142857142857143, "amount": 0.6666666666666666, "status": 0.6666666666666666, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1045}, "3": {"facts_total": 58, "facts_retained": 32, "retention_rate": 0.5517241379310345, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.7142857142857143, "amount": 0.6666666666666666, "status": 0.6666666666666666, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1044}, "4": {"facts_total": 58, "facts_retained": 32, "retention_rate": 0.5517241379310345, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.7142857142857143, "amount": 0.6666666666666666, "status": 0.6666666666666666, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1044}, "5": {"facts_total": 58, "facts_retained": 32, "retention_rate": 0.5517241379310345, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.7142857142857143, "amount": 0.6666666666666666, "status": 0.6666666666666666, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1020}, "6": {"facts_total": 58, "facts_retained": 32, "retention_rate": 0.5517241379310345, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.7142857142857143, "amount": 0.6666666666666666, "status": 0.6666666666666666, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1049}, "7": {"facts_total": 58, "facts_retained": 32, "retention_rate": 0.5517241379310345, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.7142857142857143, "amount": 0.6666666666666666, "status": 0.6666666666666666, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1049}, "8": {"facts_total": 58, "facts_retained": 32, "retention_rate": 0.5517241379310345, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.7142857142857143, "amount": 0.6666666666666666, "status": 0.6666666666666666, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1049}}}, "reference_key_id": "keys/synthetic/syn_03003_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "bcdef32d72e140b171048e34a3d65f86a0141ce1e4df6b6fc2128c34ee471bef", "output_hash": "86be78720fc4b79f93e603e7d9e4861fcd25102612f73209fffc17d787056007"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "43772670a4d99c9a6c83491b", "attempt_id": "18bb31144b7e", "run_group_id": "E4_core:A4|closed|qwen3b-llama3b:h8", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T06:52:55.019502Z", "task_family": "synthetic", "document_id": "syn_03003_2000", "document_hash": "dddca1a38e3224870fcfae927c701c400fd885c19123ea840ee06a76a140f407", "question_id": "q09_revision", "split": "test", "cluster_id": "syn_03003_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "tokenizer_hashes": {"receiver": "b9d8872cf3a99da99e1fc4b0f71232a451632dc49ca4c6771549b4efa5c1c763"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 8, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1722, "query_tokens": 500, "message_tokens": 385, "message_utf8_bytes": 1049, "cache_tensor_bytes": null, "serialized_payload_bytes": 1049, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 144318.40237381402, "hops": {"1": 22569.42216597963, "2": 15621.60833296366, "3": 20515.35983302165, "4": 15576.701166923158, "5": 20001.019624993205, "6": 14886.58912491519, "7": 20005.93258300796, "8": 15141.76954200957}, "receiver_prefill_ms": 491.2016249727458, "ttft_ms": 35.57250008452684, "decode_ms": 108.48250007256866}, "total_wall_ms": 144954.19687381946, "ttft_ms": 35.57250008452684, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 3}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "911,925", "evidence_artifact": {"ledger": {"facts_total": 58, "facts_retained": 32, "retention_rate": 0.5517241379310345, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.7142857142857143, "amount": 0.6666666666666666, "status": 0.6666666666666666, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1049}, "ledger_by_hop": {"1": {"facts_total": 58, "facts_retained": 32, "retention_rate": 0.5517241379310345, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.7142857142857143, "amount": 0.6666666666666666, "status": 0.6666666666666666, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1200}, "2": {"facts_total": 58, "facts_retained": 32, "retention_rate": 0.5517241379310345, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.7142857142857143, "amount": 0.6666666666666666, "status": 0.6666666666666666, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1045}, "3": {"facts_total": 58, "facts_retained": 32, "retention_rate": 0.5517241379310345, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.7142857142857143, "amount": 0.6666666666666666, "status": 0.6666666666666666, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1044}, "4": {"facts_total": 58, "facts_retained": 32, "retention_rate": 0.5517241379310345, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.7142857142857143, "amount": 0.6666666666666666, "status": 0.6666666666666666, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1044}, "5": {"facts_total": 58, "facts_retained": 32, "retention_rate": 0.5517241379310345, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.7142857142857143, "amount": 0.6666666666666666, "status": 0.6666666666666666, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1020}, "6": {"facts_total": 58, "facts_retained": 32, "retention_rate": 0.5517241379310345, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.7142857142857143, "amount": 0.6666666666666666, "status": 0.6666666666666666, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1049}, "7": {"facts_total": 58, "facts_retained": 32, "retention_rate": 0.5517241379310345, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.7142857142857143, "amount": 0.6666666666666666, "status": 0.6666666666666666, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1049}, "8": {"facts_total": 58, "facts_retained": 32, "retention_rate": 0.5517241379310345, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.7142857142857143, "amount": 0.6666666666666666, "status": 0.6666666666666666, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1049}}}, "reference_key_id": "keys/synthetic/syn_03003_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "bcdef32d72e140b171048e34a3d65f86a0141ce1e4df6b6fc2128c34ee471bef", "output_hash": "86be78720fc4b79f93e603e7d9e4861fcd25102612f73209fffc17d787056007"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "29fceae2233249df06fc9afa", "attempt_id": "5bfdefdd3260", "run_group_id": "E4_core:A4|closed|qwen3b-llama3b:h8", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T06:52:55.672521Z", "task_family": "synthetic", "document_id": "syn_03003_2000", "document_hash": "dddca1a38e3224870fcfae927c701c400fd885c19123ea840ee06a76a140f407", "question_id": "q10_temporal", "split": "test", "cluster_id": "syn_03003_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "tokenizer_hashes": {"receiver": "b9d8872cf3a99da99e1fc4b0f71232a451632dc49ca4c6771549b4efa5c1c763"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 8, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1722, "query_tokens": 507, "message_tokens": 385, "message_utf8_bytes": 1049, "cache_tensor_bytes": null, "serialized_payload_bytes": 1049, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 144318.40237381402, "hops": {"1": 22569.42216597963, "2": 15621.60833296366, "3": 20515.35983302165, "4": 15576.701166923158, "5": 20001.019624993205, "6": 14886.58912491519, "7": 20005.93258300796, "8": 15141.76954200957}, "receiver_prefill_ms": 382.55849992856383, "ttft_ms": 22.79600000474602, "decode_ms": 178.94954199437052}, "total_wall_ms": 144903.55633187573, "ttft_ms": 22.79600000474602, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 6}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "Gamtiszen-2", "evidence_artifact": {"ledger": {"facts_total": 58, "facts_retained": 32, "retention_rate": 0.5517241379310345, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.7142857142857143, "amount": 0.6666666666666666, "status": 0.6666666666666666, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1049}, "ledger_by_hop": {"1": {"facts_total": 58, "facts_retained": 32, "retention_rate": 0.5517241379310345, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.7142857142857143, "amount": 0.6666666666666666, "status": 0.6666666666666666, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1200}, "2": {"facts_total": 58, "facts_retained": 32, "retention_rate": 0.5517241379310345, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.7142857142857143, "amount": 0.6666666666666666, "status": 0.6666666666666666, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1045}, "3": {"facts_total": 58, "facts_retained": 32, "retention_rate": 0.5517241379310345, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.7142857142857143, "amount": 0.6666666666666666, "status": 0.6666666666666666, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1044}, "4": {"facts_total": 58, "facts_retained": 32, "retention_rate": 0.5517241379310345, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.7142857142857143, "amount": 0.6666666666666666, "status": 0.6666666666666666, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1044}, "5": {"facts_total": 58, "facts_retained": 32, "retention_rate": 0.5517241379310345, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.7142857142857143, "amount": 0.6666666666666666, "status": 0.6666666666666666, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1020}, "6": {"facts_total": 58, "facts_retained": 32, "retention_rate": 0.5517241379310345, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.7142857142857143, "amount": 0.6666666666666666, "status": 0.6666666666666666, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1049}, "7": {"facts_total": 58, "facts_retained": 32, "retention_rate": 0.5517241379310345, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.7142857142857143, "amount": 0.6666666666666666, "status": 0.6666666666666666, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1049}, "8": {"facts_total": 58, "facts_retained": 32, "retention_rate": 0.5517241379310345, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.7142857142857143, "amount": 0.6666666666666666, "status": 0.6666666666666666, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1049}}}, "reference_key_id": "keys/synthetic/syn_03003_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "bcdef32d72e140b171048e34a3d65f86a0141ce1e4df6b6fc2128c34ee471bef", "output_hash": "32e3af428886434823d82102e1c482765eafdf6077fe8d3fba5752ec5e29826c"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "7f3dd269226da7559c647b0d", "attempt_id": "059b180c73f5", "run_group_id": "E4_core:A4|closed|qwen3b-llama3b:h8", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T06:52:56.275588Z", "task_family": "synthetic", "document_id": "syn_03003_2000", "document_hash": "dddca1a38e3224870fcfae927c701c400fd885c19123ea840ee06a76a140f407", "question_id": "q11_aggregation", "split": "test", "cluster_id": "syn_03003_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "tokenizer_hashes": {"receiver": "b9d8872cf3a99da99e1fc4b0f71232a451632dc49ca4c6771549b4efa5c1c763"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 8, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1722, "query_tokens": 501, "message_tokens": 385, "message_utf8_bytes": 1049, "cache_tensor_bytes": null, "serialized_payload_bytes": 1049, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 144318.40237381402, "hops": {"1": 22569.42216597963, "2": 15621.60833296366, "3": 20515.35983302165, "4": 15576.701166923158, "5": 20001.019624993205, "6": 14886.58912491519, "7": 20005.93258300796, "8": 15141.76954200957}, "receiver_prefill_ms": 359.12470798939466, "ttft_ms": 91.32412495091558, "decode_ms": 259.1361249797046}, "total_wall_ms": 145028.58912386, "ttft_ms": 91.32412495091558, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 5}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "11,165,406", "evidence_artifact": {"ledger": {"facts_total": 58, "facts_retained": 32, "retention_rate": 0.5517241379310345, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.7142857142857143, "amount": 0.6666666666666666, "status": 0.6666666666666666, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1049}, "ledger_by_hop": {"1": {"facts_total": 58, "facts_retained": 32, "retention_rate": 0.5517241379310345, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.7142857142857143, "amount": 0.6666666666666666, "status": 0.6666666666666666, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1200}, "2": {"facts_total": 58, "facts_retained": 32, "retention_rate": 0.5517241379310345, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.7142857142857143, "amount": 0.6666666666666666, "status": 0.6666666666666666, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1045}, "3": {"facts_total": 58, "facts_retained": 32, "retention_rate": 0.5517241379310345, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.7142857142857143, "amount": 0.6666666666666666, "status": 0.6666666666666666, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1044}, "4": {"facts_total": 58, "facts_retained": 32, "retention_rate": 0.5517241379310345, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.7142857142857143, "amount": 0.6666666666666666, "status": 0.6666666666666666, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1044}, "5": {"facts_total": 58, "facts_retained": 32, "retention_rate": 0.5517241379310345, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.7142857142857143, "amount": 0.6666666666666666, "status": 0.6666666666666666, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1020}, "6": {"facts_total": 58, "facts_retained": 32, "retention_rate": 0.5517241379310345, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.7142857142857143, "amount": 0.6666666666666666, "status": 0.6666666666666666, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1049}, "7": {"facts_total": 58, "facts_retained": 32, "retention_rate": 0.5517241379310345, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.7142857142857143, "amount": 0.6666666666666666, "status": 0.6666666666666666, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1049}, "8": {"facts_total": 58, "facts_retained": 32, "retention_rate": 0.5517241379310345, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.7142857142857143, "amount": 0.6666666666666666, "status": 0.6666666666666666, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1049}}}, "reference_key_id": "keys/synthetic/syn_03003_2000.json", "task_scores": {"correct": 0.0, "abstained": 0.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "bcdef32d72e140b171048e34a3d65f86a0141ce1e4df6b6fc2128c34ee471bef", "output_hash": "557c867fd18f226d3d49763e0ac0fd86a4d623de71032ef41a7e1439fa3c20e7"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "9c40bad8258963d2024b7c5a", "attempt_id": "b5bce0fefa1d", "run_group_id": "E4_core:A4|closed|qwen3b-llama3b:h8", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T06:52:57.003130Z", "task_family": "synthetic", "document_id": "syn_03003_2000", "document_hash": "dddca1a38e3224870fcfae927c701c400fd885c19123ea840ee06a76a140f407", "question_id": "q12_unit_conversion", "split": "test", "cluster_id": "syn_03003_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "tokenizer_hashes": {"receiver": "b9d8872cf3a99da99e1fc4b0f71232a451632dc49ca4c6771549b4efa5c1c763"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 8, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1722, "query_tokens": 498, "message_tokens": 385, "message_utf8_bytes": 1049, "cache_tensor_bytes": null, "serialized_payload_bytes": 1049, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 144318.40237381402, "hops": {"1": 22569.42216597963, "2": 15621.60833296366, "3": 20515.35983302165, "4": 15576.701166923158, "5": 20001.019624993205, "6": 14886.58912491519, "7": 20005.93258300796, "8": 15141.76954200957}, "receiver_prefill_ms": 374.3130420334637, "ttft_ms": 36.93662502337247, "decode_ms": 88.7399580096826}, "total_wall_ms": 144818.8739567995, "ttft_ms": 36.93662502337247, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 3}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "17.43", "evidence_artifact": {"ledger": {"facts_total": 58, "facts_retained": 32, "retention_rate": 0.5517241379310345, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.7142857142857143, "amount": 0.6666666666666666, "status": 0.6666666666666666, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1049}, "ledger_by_hop": {"1": {"facts_total": 58, "facts_retained": 32, "retention_rate": 0.5517241379310345, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.7142857142857143, "amount": 0.6666666666666666, "status": 0.6666666666666666, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1200}, "2": {"facts_total": 58, "facts_retained": 32, "retention_rate": 0.5517241379310345, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.7142857142857143, "amount": 0.6666666666666666, "status": 0.6666666666666666, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1045}, "3": {"facts_total": 58, "facts_retained": 32, "retention_rate": 0.5517241379310345, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.7142857142857143, "amount": 0.6666666666666666, "status": 0.6666666666666666, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1044}, "4": {"facts_total": 58, "facts_retained": 32, "retention_rate": 0.5517241379310345, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.7142857142857143, "amount": 0.6666666666666666, "status": 0.6666666666666666, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1044}, "5": {"facts_total": 58, "facts_retained": 32, "retention_rate": 0.5517241379310345, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.7142857142857143, "amount": 0.6666666666666666, "status": 0.6666666666666666, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1020}, "6": {"facts_total": 58, "facts_retained": 32, "retention_rate": 0.5517241379310345, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.7142857142857143, "amount": 0.6666666666666666, "status": 0.6666666666666666, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1049}, "7": {"facts_total": 58, "facts_retained": 32, "retention_rate": 0.5517241379310345, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.7142857142857143, "amount": 0.6666666666666666, "status": 0.6666666666666666, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1049}, "8": {"facts_total": 58, "facts_retained": 32, "retention_rate": 0.5517241379310345, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.7142857142857143, "amount": 0.6666666666666666, "status": 0.6666666666666666, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1049}}}, "reference_key_id": "keys/synthetic/syn_03003_2000.json", "task_scores": {"correct": 0.0, "abstained": 0.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "bcdef32d72e140b171048e34a3d65f86a0141ce1e4df6b6fc2128c34ee471bef", "output_hash": "1cd677e6c296f31fca744a7c4d96b5514fa6b036913240b8d689f1e277900d17"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "811f44e716c5bd4ec24f000f", "attempt_id": "2a429da49779", "run_group_id": "E4_core:A4|closed|qwen3b-llama3b:h8", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T06:52:57.521033Z", "task_family": "synthetic", "document_id": "syn_03003_2000", "document_hash": "dddca1a38e3224870fcfae927c701c400fd885c19123ea840ee06a76a140f407", "question_id": "q13_unanswerable", "split": "test", "cluster_id": "syn_03003_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "tokenizer_hashes": {"receiver": "b9d8872cf3a99da99e1fc4b0f71232a451632dc49ca4c6771549b4efa5c1c763"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 8, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1722, "query_tokens": 502, "message_tokens": 385, "message_utf8_bytes": 1049, "cache_tensor_bytes": null, "serialized_payload_bytes": 1049, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 144318.40237381402, "hops": {"1": 22569.42216597963, "2": 15621.60833296366, "3": 20515.35983302165, "4": 15576.701166923158, "5": 20001.019624993205, "6": 14886.58912491519, "7": 20005.93258300796, "8": 15141.76954200957}, "receiver_prefill_ms": 344.13120907265693, "ttft_ms": 34.150582971051335, "decode_ms": 55.28012500144541}, "total_wall_ms": 144752.92324880138, "ttft_ms": 34.150582971051335, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 58, "facts_retained": 32, "retention_rate": 0.5517241379310345, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.7142857142857143, "amount": 0.6666666666666666, "status": 0.6666666666666666, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1049}, "ledger_by_hop": {"1": {"facts_total": 58, "facts_retained": 32, "retention_rate": 0.5517241379310345, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.7142857142857143, "amount": 0.6666666666666666, "status": 0.6666666666666666, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1200}, "2": {"facts_total": 58, "facts_retained": 32, "retention_rate": 0.5517241379310345, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.7142857142857143, "amount": 0.6666666666666666, "status": 0.6666666666666666, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1045}, "3": {"facts_total": 58, "facts_retained": 32, "retention_rate": 0.5517241379310345, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.7142857142857143, "amount": 0.6666666666666666, "status": 0.6666666666666666, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1044}, "4": {"facts_total": 58, "facts_retained": 32, "retention_rate": 0.5517241379310345, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.7142857142857143, "amount": 0.6666666666666666, "status": 0.6666666666666666, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1044}, "5": {"facts_total": 58, "facts_retained": 32, "retention_rate": 0.5517241379310345, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.7142857142857143, "amount": 0.6666666666666666, "status": 0.6666666666666666, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1020}, "6": {"facts_total": 58, "facts_retained": 32, "retention_rate": 0.5517241379310345, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.7142857142857143, "amount": 0.6666666666666666, "status": 0.6666666666666666, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1049}, "7": {"facts_total": 58, "facts_retained": 32, "retention_rate": 0.5517241379310345, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.7142857142857143, "amount": 0.6666666666666666, "status": 0.6666666666666666, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1049}, "8": {"facts_total": 58, "facts_retained": 32, "retention_rate": 0.5517241379310345, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.7142857142857143, "amount": 0.6666666666666666, "status": 0.6666666666666666, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1049}}}, "reference_key_id": "keys/synthetic/syn_03003_2000.json", "task_scores": {"correct": 1.0, "abstained": 1.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "bcdef32d72e140b171048e34a3d65f86a0141ce1e4df6b6fc2128c34ee471bef", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "fb551d9eb4b4136e295d2764", "attempt_id": "b4003d347609", "run_group_id": "E4_core:A4|closed|qwen3b-llama3b:h8", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T06:52:57.973507Z", "task_family": "synthetic", "document_id": "syn_03003_2000", "document_hash": "dddca1a38e3224870fcfae927c701c400fd885c19123ea840ee06a76a140f407", "question_id": "q14_unanswerable", "split": "test", "cluster_id": "syn_03003_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "tokenizer_hashes": {"receiver": "b9d8872cf3a99da99e1fc4b0f71232a451632dc49ca4c6771549b4efa5c1c763"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 8, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1722, "query_tokens": 503, "message_tokens": 385, "message_utf8_bytes": 1049, "cache_tensor_bytes": null, "serialized_payload_bytes": 1049, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 144318.40237381402, "hops": {"1": 22569.42216597963, "2": 15621.60833296366, "3": 20515.35983302165, "4": 15576.701166923158, "5": 20001.019624993205, "6": 14886.58912491519, "7": 20005.93258300796, "8": 15141.76954200957}, "receiver_prefill_ms": 482.1709169773385, "ttft_ms": 59.919459046795964, "decode_ms": 29.74404103588313}, "total_wall_ms": 144890.83004079293, "ttft_ms": 59.919459046795964, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 58, "facts_retained": 32, "retention_rate": 0.5517241379310345, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.7142857142857143, "amount": 0.6666666666666666, "status": 0.6666666666666666, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1049}, "ledger_by_hop": {"1": {"facts_total": 58, "facts_retained": 32, "retention_rate": 0.5517241379310345, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.7142857142857143, "amount": 0.6666666666666666, "status": 0.6666666666666666, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1200}, "2": {"facts_total": 58, "facts_retained": 32, "retention_rate": 0.5517241379310345, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.7142857142857143, "amount": 0.6666666666666666, "status": 0.6666666666666666, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1045}, "3": {"facts_total": 58, "facts_retained": 32, "retention_rate": 0.5517241379310345, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.7142857142857143, "amount": 0.6666666666666666, "status": 0.6666666666666666, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1044}, "4": {"facts_total": 58, "facts_retained": 32, "retention_rate": 0.5517241379310345, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.7142857142857143, "amount": 0.6666666666666666, "status": 0.6666666666666666, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1044}, "5": {"facts_total": 58, "facts_retained": 32, "retention_rate": 0.5517241379310345, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.7142857142857143, "amount": 0.6666666666666666, "status": 0.6666666666666666, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1020}, "6": {"facts_total": 58, "facts_retained": 32, "retention_rate": 0.5517241379310345, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.7142857142857143, "amount": 0.6666666666666666, "status": 0.6666666666666666, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1049}, "7": {"facts_total": 58, "facts_retained": 32, "retention_rate": 0.5517241379310345, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.7142857142857143, "amount": 0.6666666666666666, "status": 0.6666666666666666, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1049}, "8": {"facts_total": 58, "facts_retained": 32, "retention_rate": 0.5517241379310345, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.7142857142857143, "amount": 0.6666666666666666, "status": 0.6666666666666666, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1049}}}, "reference_key_id": "keys/synthetic/syn_03003_2000.json", "task_scores": {"correct": 1.0, "abstained": 1.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "bcdef32d72e140b171048e34a3d65f86a0141ce1e4df6b6fc2128c34ee471bef", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "4e6c366e5a14895fb2904a89", "attempt_id": "b64f9845ca1a", "run_group_id": "E4_core:A5|closed|qwen3b-llama3b:h1", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T06:53:21.177549Z", "task_family": "synthetic", "document_id": "syn_03003_2000", "document_hash": "dddca1a38e3224870fcfae927c701c400fd885c19123ea840ee06a76a140f407", "question_id": "q00_single_fact", "split": "test", "cluster_id": "syn_03003_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 1, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1866, "query_tokens": 435, "message_tokens": 352, "message_utf8_bytes": 830, "cache_tensor_bytes": null, "serialized_payload_bytes": 830, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 22612.123124999925, "hops": {"1": 22612.123124999925}, "receiver_prefill_ms": 294.887249940075, "ttft_ms": 23.88800005428493, "decode_ms": 633.0620419466868}, "total_wall_ms": 23564.83629194554, "ttft_ms": 23.88800005428493, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 10}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "GR-79S-5839", "evidence_artifact": {"ledger": {"facts_total": 58, "facts_retained": 15, "retention_rate": 0.25862068965517243, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.2222222222222222, "status": 0.75, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 766}, "ledger_by_hop": {"1": {"facts_total": 58, "facts_retained": 15, "retention_rate": 0.25862068965517243, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.2222222222222222, "status": 0.75, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 766}}}, "reference_key_id": "keys/synthetic/syn_03003_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "a613cf19d60b1dbe7d9f51bbbf98459d9f181f51f14c39d755912e891ae7fd2f", "output_hash": "720cc40cd69e85e9728f6bea00b17a95ca07eee643e143a131840763958303cd"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "7747e8bccad60eaf10491fbb", "attempt_id": "99aa7bfdd134", "run_group_id": "E4_core:A5|closed|qwen3b-llama3b:h1", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T06:53:22.148805Z", "task_family": "synthetic", "document_id": "syn_03003_2000", "document_hash": "dddca1a38e3224870fcfae927c701c400fd885c19123ea840ee06a76a140f407", "question_id": "q01_single_fact", "split": "test", "cluster_id": "syn_03003_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 1, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1866, "query_tokens": 439, "message_tokens": 352, "message_utf8_bytes": 830, "cache_tensor_bytes": null, "serialized_payload_bytes": 830, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 22612.123124999925, "hops": {"1": 22612.123124999925}, "receiver_prefill_ms": 294.0259169554338, "ttft_ms": 24.719083914533257, "decode_ms": 38.751082960516214}, "total_wall_ms": 22970.141959027387, "ttft_ms": 24.719083914533257, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 58, "facts_retained": 15, "retention_rate": 0.25862068965517243, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.2222222222222222, "status": 0.75, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 766}, "ledger_by_hop": {"1": {"facts_total": 58, "facts_retained": 15, "retention_rate": 0.25862068965517243, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.2222222222222222, "status": 0.75, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 766}}}, "reference_key_id": "keys/synthetic/syn_03003_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "a613cf19d60b1dbe7d9f51bbbf98459d9f181f51f14c39d755912e891ae7fd2f", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "52d5b7fc6b2e5714a72af49e", "attempt_id": "8b297cffc5d9", "run_group_id": "E4_core:A5|closed|qwen3b-llama3b:h1", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T06:53:22.524596Z", "task_family": "synthetic", "document_id": "syn_03003_2000", "document_hash": "dddca1a38e3224870fcfae927c701c400fd885c19123ea840ee06a76a140f407", "question_id": "q02_exact_copy", "split": "test", "cluster_id": "syn_03003_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 1, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1866, "query_tokens": 439, "message_tokens": 352, "message_utf8_bytes": 830, "cache_tensor_bytes": null, "serialized_payload_bytes": 830, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 22612.123124999925, "hops": {"1": 22612.123124999925}, "receiver_prefill_ms": 291.66724998503923, "ttft_ms": 30.02658300101757, "decode_ms": 33.262584009207785}, "total_wall_ms": 22967.67262497451, "ttft_ms": 30.02658300101757, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 58, "facts_retained": 15, "retention_rate": 0.25862068965517243, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.2222222222222222, "status": 0.75, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 766}, "ledger_by_hop": {"1": {"facts_total": 58, "facts_retained": 15, "retention_rate": 0.25862068965517243, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.2222222222222222, "status": 0.75, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 766}}}, "reference_key_id": "keys/synthetic/syn_03003_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "a613cf19d60b1dbe7d9f51bbbf98459d9f181f51f14c39d755912e891ae7fd2f", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "2e2d4f5cfa5e77246ead14b2", "attempt_id": "075bd91da7d3", "run_group_id": "E4_core:A5|closed|qwen3b-llama3b:h1", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T06:53:22.898088Z", "task_family": "synthetic", "document_id": "syn_03003_2000", "document_hash": "dddca1a38e3224870fcfae927c701c400fd885c19123ea840ee06a76a140f407", "question_id": "q03_exact_copy", "split": "test", "cluster_id": "syn_03003_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 1, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1866, "query_tokens": 434, "message_tokens": 352, "message_utf8_bytes": 830, "cache_tensor_bytes": null, "serialized_payload_bytes": 830, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 22612.123124999925, "hops": {"1": 22612.123124999925}, "receiver_prefill_ms": 256.7287500714883, "ttft_ms": 56.296999915502965, "decode_ms": 46.25087499152869}, "total_wall_ms": 22972.12741698604, "ttft_ms": 56.296999915502965, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 58, "facts_retained": 15, "retention_rate": 0.25862068965517243, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.2222222222222222, "status": 0.75, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 766}, "ledger_by_hop": {"1": {"facts_total": 58, "facts_retained": 15, "retention_rate": 0.25862068965517243, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.2222222222222222, "status": 0.75, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 766}}}, "reference_key_id": "keys/synthetic/syn_03003_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "a613cf19d60b1dbe7d9f51bbbf98459d9f181f51f14c39d755912e891ae7fd2f", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "dd826ec44a2d5da4a16e2fab", "attempt_id": "354d671137eb", "run_group_id": "E4_core:A5|closed|qwen3b-llama3b:h1", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T06:53:23.276583Z", "task_family": "synthetic", "document_id": "syn_03003_2000", "document_hash": "dddca1a38e3224870fcfae927c701c400fd885c19123ea840ee06a76a140f407", "question_id": "q04_multi_hop", "split": "test", "cluster_id": "syn_03003_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 1, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1866, "query_tokens": 443, "message_tokens": 352, "message_utf8_bytes": 830, "cache_tensor_bytes": null, "serialized_payload_bytes": 830, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 22612.123124999925, "hops": {"1": 22612.123124999925}, "receiver_prefill_ms": 362.8057499881834, "ttft_ms": 114.06833294313401, "decode_ms": 39.754458935931325}, "total_wall_ms": 23129.34170803055, "ttft_ms": 114.06833294313401, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 58, "facts_retained": 15, "retention_rate": 0.25862068965517243, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.2222222222222222, "status": 0.75, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 766}, "ledger_by_hop": {"1": {"facts_total": 58, "facts_retained": 15, "retention_rate": 0.25862068965517243, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.2222222222222222, "status": 0.75, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 766}}}, "reference_key_id": "keys/synthetic/syn_03003_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "a613cf19d60b1dbe7d9f51bbbf98459d9f181f51f14c39d755912e891ae7fd2f", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "037d074c20cdb29504428177", "attempt_id": "d46bd723ee26", "run_group_id": "E4_core:A5|closed|qwen3b-llama3b:h1", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T06:53:23.811714Z", "task_family": "synthetic", "document_id": "syn_03003_2000", "document_hash": "dddca1a38e3224870fcfae927c701c400fd885c19123ea840ee06a76a140f407", "question_id": "q05_multi_hop", "split": "test", "cluster_id": "syn_03003_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 1, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1866, "query_tokens": 445, "message_tokens": 352, "message_utf8_bytes": 830, "cache_tensor_bytes": null, "serialized_payload_bytes": 830, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 22612.123124999925, "hops": {"1": 22612.123124999925}, "receiver_prefill_ms": 296.3165419641882, "ttft_ms": 24.477709084749222, "decode_ms": 37.49083401635289}, "total_wall_ms": 22970.93333292287, "ttft_ms": 24.477709084749222, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 58, "facts_retained": 15, "retention_rate": 0.25862068965517243, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.2222222222222222, "status": 0.75, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 766}, "ledger_by_hop": {"1": {"facts_total": 58, "facts_retained": 15, "retention_rate": 0.25862068965517243, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.2222222222222222, "status": 0.75, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 766}}}, "reference_key_id": "keys/synthetic/syn_03003_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "a613cf19d60b1dbe7d9f51bbbf98459d9f181f51f14c39d755912e891ae7fd2f", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "49d16f51898ae17a93d8e77b", "attempt_id": "94cfbdf35e9d", "run_group_id": "E4_core:A5|closed|qwen3b-llama3b:h1", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T06:53:24.188037Z", "task_family": "synthetic", "document_id": "syn_03003_2000", "document_hash": "dddca1a38e3224870fcfae927c701c400fd885c19123ea840ee06a76a140f407", "question_id": "q06_negation", "split": "test", "cluster_id": "syn_03003_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 1, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1866, "query_tokens": 437, "message_tokens": 352, "message_utf8_bytes": 830, "cache_tensor_bytes": null, "serialized_payload_bytes": 830, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 22612.123124999925, "hops": {"1": 22612.123124999925}, "receiver_prefill_ms": 306.42204103060067, "ttft_ms": 21.43429103307426, "decode_ms": 37.39333397243172}, "total_wall_ms": 22977.989249979146, "ttft_ms": 21.43429103307426, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "no", "evidence_artifact": {"ledger": {"facts_total": 58, "facts_retained": 15, "retention_rate": 0.25862068965517243, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.2222222222222222, "status": 0.75, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 766}, "ledger_by_hop": {"1": {"facts_total": 58, "facts_retained": 15, "retention_rate": 0.25862068965517243, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.2222222222222222, "status": 0.75, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 766}}}, "reference_key_id": "keys/synthetic/syn_03003_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "a613cf19d60b1dbe7d9f51bbbf98459d9f181f51f14c39d755912e891ae7fd2f", "output_hash": "9390298f3fb0c5b160498935d79cb139aef28e1c47358b4bbba61862b9c26e59"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "af313d4bc82b6e886f204206", "attempt_id": "578f8ae82829", "run_group_id": "E4_core:A5|closed|qwen3b-llama3b:h1", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T06:53:24.571431Z", "task_family": "synthetic", "document_id": "syn_03003_2000", "document_hash": "dddca1a38e3224870fcfae927c701c400fd885c19123ea840ee06a76a140f407", "question_id": "q07_negation", "split": "test", "cluster_id": "syn_03003_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 1, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1866, "query_tokens": 434, "message_tokens": 352, "message_utf8_bytes": 830, "cache_tensor_bytes": null, "serialized_payload_bytes": 830, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 22612.123124999925, "hops": {"1": 22612.123124999925}, "receiver_prefill_ms": 276.987916091457, "ttft_ms": 43.16058289259672, "decode_ms": 46.10137501731515}, "total_wall_ms": 22978.910500067286, "ttft_ms": 43.16058289259672, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 58, "facts_retained": 15, "retention_rate": 0.25862068965517243, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.2222222222222222, "status": 0.75, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 766}, "ledger_by_hop": {"1": {"facts_total": 58, "facts_retained": 15, "retention_rate": 0.25862068965517243, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.2222222222222222, "status": 0.75, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 766}}}, "reference_key_id": "keys/synthetic/syn_03003_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "a613cf19d60b1dbe7d9f51bbbf98459d9f181f51f14c39d755912e891ae7fd2f", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "0ff819582daa9e61a376a3aa", "attempt_id": "18538a07ea9e", "run_group_id": "E4_core:A5|closed|qwen3b-llama3b:h1", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T06:53:24.956301Z", "task_family": "synthetic", "document_id": "syn_03003_2000", "document_hash": "dddca1a38e3224870fcfae927c701c400fd885c19123ea840ee06a76a140f407", "question_id": "q08_revision", "split": "test", "cluster_id": "syn_03003_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 1, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1866, "query_tokens": 452, "message_tokens": 352, "message_utf8_bytes": 830, "cache_tensor_bytes": null, "serialized_payload_bytes": 830, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 22612.123124999925, "hops": {"1": 22612.123124999925}, "receiver_prefill_ms": 494.8640000075102, "ttft_ms": 24.687541066668928, "decode_ms": 229.61204196326435}, "total_wall_ms": 23361.857707961462, "ttft_ms": 24.687541066668928, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 7}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "911,925", "evidence_artifact": {"ledger": {"facts_total": 58, "facts_retained": 15, "retention_rate": 0.25862068965517243, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.2222222222222222, "status": 0.75, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 766}, "ledger_by_hop": {"1": {"facts_total": 58, "facts_retained": 15, "retention_rate": 0.25862068965517243, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.2222222222222222, "status": 0.75, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 766}}}, "reference_key_id": "keys/synthetic/syn_03003_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "a613cf19d60b1dbe7d9f51bbbf98459d9f181f51f14c39d755912e891ae7fd2f", "output_hash": "86be78720fc4b79f93e603e7d9e4861fcd25102612f73209fffc17d787056007"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "00dacf9743031892f1646212", "attempt_id": "c86df84eca91", "run_group_id": "E4_core:A5|closed|qwen3b-llama3b:h1", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T06:53:25.724779Z", "task_family": "synthetic", "document_id": "syn_03003_2000", "document_hash": "dddca1a38e3224870fcfae927c701c400fd885c19123ea840ee06a76a140f407", "question_id": "q09_revision", "split": "test", "cluster_id": "syn_03003_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 1, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1866, "query_tokens": 445, "message_tokens": 352, "message_utf8_bytes": 830, "cache_tensor_bytes": null, "serialized_payload_bytes": 830, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 22612.123124999925, "hops": {"1": 22612.123124999925}, "receiver_prefill_ms": 314.15654195006937, "ttft_ms": 41.692125028930604, "decode_ms": 170.7662909757346}, "total_wall_ms": 23139.365207985975, "ttft_ms": 41.692125028930604, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 6}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "911925", "evidence_artifact": {"ledger": {"facts_total": 58, "facts_retained": 15, "retention_rate": 0.25862068965517243, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.2222222222222222, "status": 0.75, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 766}, "ledger_by_hop": {"1": {"facts_total": 58, "facts_retained": 15, "retention_rate": 0.25862068965517243, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.2222222222222222, "status": 0.75, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 766}}}, "reference_key_id": "keys/synthetic/syn_03003_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "a613cf19d60b1dbe7d9f51bbbf98459d9f181f51f14c39d755912e891ae7fd2f", "output_hash": "679a540dc4739f0ace623448b8fd925fb234f554838d7e87da223228ee81b112"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "1f6a57de89d435da1bc9e877", "attempt_id": "9723b2ecb455", "run_group_id": "E4_core:A5|closed|qwen3b-llama3b:h1", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T06:53:26.270387Z", "task_family": "synthetic", "document_id": "syn_03003_2000", "document_hash": "dddca1a38e3224870fcfae927c701c400fd885c19123ea840ee06a76a140f407", "question_id": "q10_temporal", "split": "test", "cluster_id": "syn_03003_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 1, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1866, "query_tokens": 452, "message_tokens": 352, "message_utf8_bytes": 830, "cache_tensor_bytes": null, "serialized_payload_bytes": 830, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 22612.123124999925, "hops": {"1": 22612.123124999925}, "receiver_prefill_ms": 320.33283403143287, "ttft_ms": 42.068916955031455, "decode_ms": 60.61766599304974}, "total_wall_ms": 23035.904707969166, "ttft_ms": 42.068916955031455, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 58, "facts_retained": 15, "retention_rate": 0.25862068965517243, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.2222222222222222, "status": 0.75, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 766}, "ledger_by_hop": {"1": {"facts_total": 58, "facts_retained": 15, "retention_rate": 0.25862068965517243, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.2222222222222222, "status": 0.75, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 766}}}, "reference_key_id": "keys/synthetic/syn_03003_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "a613cf19d60b1dbe7d9f51bbbf98459d9f181f51f14c39d755912e891ae7fd2f", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "e66bf0381713727d14e67b94", "attempt_id": "dd650681e188", "run_group_id": "E4_core:A5|closed|qwen3b-llama3b:h1", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T06:53:26.713461Z", "task_family": "synthetic", "document_id": "syn_03003_2000", "document_hash": "dddca1a38e3224870fcfae927c701c400fd885c19123ea840ee06a76a140f407", "question_id": "q11_aggregation", "split": "test", "cluster_id": "syn_03003_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 1, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1866, "query_tokens": 446, "message_tokens": 352, "message_utf8_bytes": 830, "cache_tensor_bytes": null, "serialized_payload_bytes": 830, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 22612.123124999925, "hops": {"1": 22612.123124999925}, "receiver_prefill_ms": 429.8944999463856, "ttft_ms": 39.31770892813802, "decode_ms": 32.06958307418972}, "total_wall_ms": 23114.039540989324, "ttft_ms": 39.31770892813802, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 58, "facts_retained": 15, "retention_rate": 0.25862068965517243, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.2222222222222222, "status": 0.75, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 766}, "ledger_by_hop": {"1": {"facts_total": 58, "facts_retained": 15, "retention_rate": 0.25862068965517243, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.2222222222222222, "status": 0.75, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 766}}}, "reference_key_id": "keys/synthetic/syn_03003_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "a613cf19d60b1dbe7d9f51bbbf98459d9f181f51f14c39d755912e891ae7fd2f", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "09f095073036498567b4b894", "attempt_id": "4a97410a2c2a", "run_group_id": "E4_core:A5|closed|qwen3b-llama3b:h1", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T06:53:27.232896Z", "task_family": "synthetic", "document_id": "syn_03003_2000", "document_hash": "dddca1a38e3224870fcfae927c701c400fd885c19123ea840ee06a76a140f407", "question_id": "q12_unit_conversion", "split": "test", "cluster_id": "syn_03003_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 1, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1866, "query_tokens": 443, "message_tokens": 352, "message_utf8_bytes": 830, "cache_tensor_bytes": null, "serialized_payload_bytes": 830, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 22612.123124999925, "hops": {"1": 22612.123124999925}, "receiver_prefill_ms": 281.230500084348, "ttft_ms": 38.01120794378221, "decode_ms": 185.10550004430115}, "total_wall_ms": 23117.12675006129, "ttft_ms": 38.01120794378221, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 5}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "17430", "evidence_artifact": {"ledger": {"facts_total": 58, "facts_retained": 15, "retention_rate": 0.25862068965517243, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.2222222222222222, "status": 0.75, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 766}, "ledger_by_hop": {"1": {"facts_total": 58, "facts_retained": 15, "retention_rate": 0.25862068965517243, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.2222222222222222, "status": 0.75, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 766}}}, "reference_key_id": "keys/synthetic/syn_03003_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "a613cf19d60b1dbe7d9f51bbbf98459d9f181f51f14c39d755912e891ae7fd2f", "output_hash": "2575eaf3b748efc209b676518766751eb038c88f74578951345818bfe7f7194b"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "14bb3c77dfbf1e4c3a060417", "attempt_id": "23c6c353a85c", "run_group_id": "E4_core:A5|closed|qwen3b-llama3b:h1", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T06:53:27.755191Z", "task_family": "synthetic", "document_id": "syn_03003_2000", "document_hash": "dddca1a38e3224870fcfae927c701c400fd885c19123ea840ee06a76a140f407", "question_id": "q13_unanswerable", "split": "test", "cluster_id": "syn_03003_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 1, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1866, "query_tokens": 447, "message_tokens": 352, "message_utf8_bytes": 830, "cache_tensor_bytes": null, "serialized_payload_bytes": 830, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 22612.123124999925, "hops": {"1": 22612.123124999925}, "receiver_prefill_ms": 239.69108297023922, "ttft_ms": 67.83999991603196, "decode_ms": 49.14325010031462}, "total_wall_ms": 22969.360291957855, "ttft_ms": 67.83999991603196, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 58, "facts_retained": 15, "retention_rate": 0.25862068965517243, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.2222222222222222, "status": 0.75, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 766}, "ledger_by_hop": {"1": {"facts_total": 58, "facts_retained": 15, "retention_rate": 0.25862068965517243, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.2222222222222222, "status": 0.75, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 766}}}, "reference_key_id": "keys/synthetic/syn_03003_2000.json", "task_scores": {"correct": 1.0, "abstained": 1.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "a613cf19d60b1dbe7d9f51bbbf98459d9f181f51f14c39d755912e891ae7fd2f", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "b27beee16da326b46a573a62", "attempt_id": "12296869afea", "run_group_id": "E4_core:A5|closed|qwen3b-llama3b:h1", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T06:53:28.130415Z", "task_family": "synthetic", "document_id": "syn_03003_2000", "document_hash": "dddca1a38e3224870fcfae927c701c400fd885c19123ea840ee06a76a140f407", "question_id": "q14_unanswerable", "split": "test", "cluster_id": "syn_03003_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 1, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1866, "query_tokens": 448, "message_tokens": 352, "message_utf8_bytes": 830, "cache_tensor_bytes": null, "serialized_payload_bytes": 830, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 22612.123124999925, "hops": {"1": 22612.123124999925}, "receiver_prefill_ms": 318.60316602978855, "ttft_ms": 87.94962498359382, "decode_ms": 102.18020796310157}, "total_wall_ms": 23121.343082981184, "ttft_ms": 87.94962498359382, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 58, "facts_retained": 15, "retention_rate": 0.25862068965517243, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.2222222222222222, "status": 0.75, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 766}, "ledger_by_hop": {"1": {"facts_total": 58, "facts_retained": 15, "retention_rate": 0.25862068965517243, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.2222222222222222, "status": 0.75, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 766}}}, "reference_key_id": "keys/synthetic/syn_03003_2000.json", "task_scores": {"correct": 1.0, "abstained": 1.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "a613cf19d60b1dbe7d9f51bbbf98459d9f181f51f14c39d755912e891ae7fd2f", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "5c2f2490137d1eecf5eb354c", "attempt_id": "561203b113f1", "run_group_id": "E4_core:A5|closed|qwen3b-llama3b:h2", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T06:53:41.581667Z", "task_family": "synthetic", "document_id": "syn_03003_2000", "document_hash": "dddca1a38e3224870fcfae927c701c400fd885c19123ea840ee06a76a140f407", "question_id": "q00_single_fact", "split": "test", "cluster_id": "syn_03003_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "tokenizer_hashes": {"receiver": "b9d8872cf3a99da99e1fc4b0f71232a451632dc49ca4c6771549b4efa5c1c763"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 2, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1722, "query_tokens": 430, "message_tokens": 326, "message_utf8_bytes": 911, "cache_tensor_bytes": null, "serialized_payload_bytes": 911, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 35535.04054190125, "hops": {"1": 22612.123124999925, "2": 12922.917416901328}, "receiver_prefill_ms": 349.6607909910381, "ttft_ms": 17.629874986596406, "decode_ms": 293.30854199361056}, "total_wall_ms": 36196.41262595542, "ttft_ms": 17.629874986596406, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 7}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "GR-79S-5839", "evidence_artifact": {"ledger": {"facts_total": 58, "facts_retained": 15, "retention_rate": 0.25862068965517243, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.2222222222222222, "status": 0.75, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 911}, "ledger_by_hop": {"1": {"facts_total": 58, "facts_retained": 15, "retention_rate": 0.25862068965517243, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.2222222222222222, "status": 0.75, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 766}, "2": {"facts_total": 58, "facts_retained": 15, "retention_rate": 0.25862068965517243, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.2222222222222222, "status": 0.75, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 911}}}, "reference_key_id": "keys/synthetic/syn_03003_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "3d426d8999e6ba8503401b57d622bad0690314233e836579ad9658a82f039d3a", "output_hash": "720cc40cd69e85e9728f6bea00b17a95ca07eee643e143a131840763958303cd"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "89a13197299e9e5c6c334e66", "attempt_id": "0e123fca507e", "run_group_id": "E4_core:A5|closed|qwen3b-llama3b:h2", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T06:53:42.260769Z", "task_family": "synthetic", "document_id": "syn_03003_2000", "document_hash": "dddca1a38e3224870fcfae927c701c400fd885c19123ea840ee06a76a140f407", "question_id": "q01_single_fact", "split": "test", "cluster_id": "syn_03003_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "tokenizer_hashes": {"receiver": "b9d8872cf3a99da99e1fc4b0f71232a451632dc49ca4c6771549b4efa5c1c763"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 2, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1722, "query_tokens": 434, "message_tokens": 326, "message_utf8_bytes": 911, "cache_tensor_bytes": null, "serialized_payload_bytes": 911, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 35535.04054190125, "hops": {"1": 22612.123124999925, "2": 12922.917416901328}, "receiver_prefill_ms": 434.27000008523464, "ttft_ms": 35.57262499816716, "decode_ms": 28.09645899105817}, "total_wall_ms": 36033.44874992035, "ttft_ms": 35.57262499816716, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 58, "facts_retained": 15, "retention_rate": 0.25862068965517243, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.2222222222222222, "status": 0.75, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 911}, "ledger_by_hop": {"1": {"facts_total": 58, "facts_retained": 15, "retention_rate": 0.25862068965517243, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.2222222222222222, "status": 0.75, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 766}, "2": {"facts_total": 58, "facts_retained": 15, "retention_rate": 0.25862068965517243, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.2222222222222222, "status": 0.75, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 911}}}, "reference_key_id": "keys/synthetic/syn_03003_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "3d426d8999e6ba8503401b57d622bad0690314233e836579ad9658a82f039d3a", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "c405af1f374c3229cd72077b", "attempt_id": "abc4affd44c5", "run_group_id": "E4_core:A5|closed|qwen3b-llama3b:h2", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T06:53:42.776491Z", "task_family": "synthetic", "document_id": "syn_03003_2000", "document_hash": "dddca1a38e3224870fcfae927c701c400fd885c19123ea840ee06a76a140f407", "question_id": "q02_exact_copy", "split": "test", "cluster_id": "syn_03003_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "tokenizer_hashes": {"receiver": "b9d8872cf3a99da99e1fc4b0f71232a451632dc49ca4c6771549b4efa5c1c763"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 2, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1722, "query_tokens": 434, "message_tokens": 326, "message_utf8_bytes": 911, "cache_tensor_bytes": null, "serialized_payload_bytes": 911, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 35535.04054190125, "hops": {"1": 22612.123124999925, "2": 12922.917416901328}, "receiver_prefill_ms": 350.149542093277, "ttft_ms": 18.2618749095127, "decode_ms": 35.72033403906971}, "total_wall_ms": 35939.676583977416, "ttft_ms": 18.2618749095127, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 58, "facts_retained": 15, "retention_rate": 0.25862068965517243, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.2222222222222222, "status": 0.75, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 911}, "ledger_by_hop": {"1": {"facts_total": 58, "facts_retained": 15, "retention_rate": 0.25862068965517243, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.2222222222222222, "status": 0.75, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 766}, "2": {"facts_total": 58, "facts_retained": 15, "retention_rate": 0.25862068965517243, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.2222222222222222, "status": 0.75, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 911}}}, "reference_key_id": "keys/synthetic/syn_03003_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "3d426d8999e6ba8503401b57d622bad0690314233e836579ad9658a82f039d3a", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "5badd34db562457e680307a2", "attempt_id": "a2d3786cfbc5", "run_group_id": "E4_core:A5|closed|qwen3b-llama3b:h2", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T06:53:43.197891Z", "task_family": "synthetic", "document_id": "syn_03003_2000", "document_hash": "dddca1a38e3224870fcfae927c701c400fd885c19123ea840ee06a76a140f407", "question_id": "q03_exact_copy", "split": "test", "cluster_id": "syn_03003_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "tokenizer_hashes": {"receiver": "b9d8872cf3a99da99e1fc4b0f71232a451632dc49ca4c6771549b4efa5c1c763"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 2, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1722, "query_tokens": 429, "message_tokens": 326, "message_utf8_bytes": 911, "cache_tensor_bytes": null, "serialized_payload_bytes": 911, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 35535.04054190125, "hops": {"1": 22612.123124999925, "2": 12922.917416901328}, "receiver_prefill_ms": 330.92570793814957, "ttft_ms": 19.28612496703863, "decode_ms": 16.15566690452397}, "total_wall_ms": 35901.92279184703, "ttft_ms": 19.28612496703863, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 58, "facts_retained": 15, "retention_rate": 0.25862068965517243, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.2222222222222222, "status": 0.75, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 911}, "ledger_by_hop": {"1": {"facts_total": 58, "facts_retained": 15, "retention_rate": 0.25862068965517243, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.2222222222222222, "status": 0.75, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 766}, "2": {"facts_total": 58, "facts_retained": 15, "retention_rate": 0.25862068965517243, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.2222222222222222, "status": 0.75, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 911}}}, "reference_key_id": "keys/synthetic/syn_03003_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "3d426d8999e6ba8503401b57d622bad0690314233e836579ad9658a82f039d3a", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "938cc0e82e24633bcccc0bb5", "attempt_id": "bf2e91dd1b2d", "run_group_id": "E4_core:A5|closed|qwen3b-llama3b:h2", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T06:53:43.582374Z", "task_family": "synthetic", "document_id": "syn_03003_2000", "document_hash": "dddca1a38e3224870fcfae927c701c400fd885c19123ea840ee06a76a140f407", "question_id": "q04_multi_hop", "split": "test", "cluster_id": "syn_03003_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "tokenizer_hashes": {"receiver": "b9d8872cf3a99da99e1fc4b0f71232a451632dc49ca4c6771549b4efa5c1c763"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 2, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1722, "query_tokens": 438, "message_tokens": 326, "message_utf8_bytes": 911, "cache_tensor_bytes": null, "serialized_payload_bytes": 911, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 35535.04054190125, "hops": {"1": 22612.123124999925, "2": 12922.917416901328}, "receiver_prefill_ms": 348.2371250865981, "ttft_ms": 43.035791954025626, "decode_ms": 83.4277089452371}, "total_wall_ms": 36010.25874982588, "ttft_ms": 43.035791954025626, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 58, "facts_retained": 15, "retention_rate": 0.25862068965517243, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.2222222222222222, "status": 0.75, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 911}, "ledger_by_hop": {"1": {"facts_total": 58, "facts_retained": 15, "retention_rate": 0.25862068965517243, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.2222222222222222, "status": 0.75, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 766}, "2": {"facts_total": 58, "facts_retained": 15, "retention_rate": 0.25862068965517243, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.2222222222222222, "status": 0.75, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 911}}}, "reference_key_id": "keys/synthetic/syn_03003_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "3d426d8999e6ba8503401b57d622bad0690314233e836579ad9658a82f039d3a", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "844a355eb9500557b41c6439", "attempt_id": "caf42b142035", "run_group_id": "E4_core:A5|closed|qwen3b-llama3b:h2", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T06:53:44.076963Z", "task_family": "synthetic", "document_id": "syn_03003_2000", "document_hash": "dddca1a38e3224870fcfae927c701c400fd885c19123ea840ee06a76a140f407", "question_id": "q05_multi_hop", "split": "test", "cluster_id": "syn_03003_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "tokenizer_hashes": {"receiver": "b9d8872cf3a99da99e1fc4b0f71232a451632dc49ca4c6771549b4efa5c1c763"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 2, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1722, "query_tokens": 438, "message_tokens": 326, "message_utf8_bytes": 911, "cache_tensor_bytes": null, "serialized_payload_bytes": 911, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 35535.04054190125, "hops": {"1": 22612.123124999925, "2": 12922.917416901328}, "receiver_prefill_ms": 406.5535420086235, "ttft_ms": 40.54862505290657, "decode_ms": 24.734542006626725}, "total_wall_ms": 36007.37520889379, "ttft_ms": 40.54862505290657, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 58, "facts_retained": 15, "retention_rate": 0.25862068965517243, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.2222222222222222, "status": 0.75, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 911}, "ledger_by_hop": {"1": {"facts_total": 58, "facts_retained": 15, "retention_rate": 0.25862068965517243, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.2222222222222222, "status": 0.75, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 766}, "2": {"facts_total": 58, "facts_retained": 15, "retention_rate": 0.25862068965517243, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.2222222222222222, "status": 0.75, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 911}}}, "reference_key_id": "keys/synthetic/syn_03003_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "3d426d8999e6ba8503401b57d622bad0690314233e836579ad9658a82f039d3a", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "8cc6e12dd6f6c44dc453db31", "attempt_id": "4fe38177586a", "run_group_id": "E4_core:A5|closed|qwen3b-llama3b:h2", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T06:53:44.565840Z", "task_family": "synthetic", "document_id": "syn_03003_2000", "document_hash": "dddca1a38e3224870fcfae927c701c400fd885c19123ea840ee06a76a140f407", "question_id": "q06_negation", "split": "test", "cluster_id": "syn_03003_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "tokenizer_hashes": {"receiver": "b9d8872cf3a99da99e1fc4b0f71232a451632dc49ca4c6771549b4efa5c1c763"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 2, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1722, "query_tokens": 431, "message_tokens": 326, "message_utf8_bytes": 911, "cache_tensor_bytes": null, "serialized_payload_bytes": 911, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 35535.04054190125, "hops": {"1": 22612.123124999925, "2": 12922.917416901328}, "receiver_prefill_ms": 352.40304202307016, "ttft_ms": 26.27124998252839, "decode_ms": 50.16245797742158}, "total_wall_ms": 35964.346249820665, "ttft_ms": 26.27124998252839, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 2}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "No.", "evidence_artifact": {"ledger": {"facts_total": 58, "facts_retained": 15, "retention_rate": 0.25862068965517243, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.2222222222222222, "status": 0.75, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 911}, "ledger_by_hop": {"1": {"facts_total": 58, "facts_retained": 15, "retention_rate": 0.25862068965517243, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.2222222222222222, "status": 0.75, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 766}, "2": {"facts_total": 58, "facts_retained": 15, "retention_rate": 0.25862068965517243, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.2222222222222222, "status": 0.75, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 911}}}, "reference_key_id": "keys/synthetic/syn_03003_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "3d426d8999e6ba8503401b57d622bad0690314233e836579ad9658a82f039d3a", "output_hash": "38178a20b470cfd18299fb1593dd4ba706f1a83321dac051d01086efd8b7a96f"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "a27bf94083e0e0cabe510eb6", "attempt_id": "09e5cdfb42aa", "run_group_id": "E4_core:A5|closed|qwen3b-llama3b:h2", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T06:53:45.011954Z", "task_family": "synthetic", "document_id": "syn_03003_2000", "document_hash": "dddca1a38e3224870fcfae927c701c400fd885c19123ea840ee06a76a140f407", "question_id": "q07_negation", "split": "test", "cluster_id": "syn_03003_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "tokenizer_hashes": {"receiver": "b9d8872cf3a99da99e1fc4b0f71232a451632dc49ca4c6771549b4efa5c1c763"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 2, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1722, "query_tokens": 429, "message_tokens": 326, "message_utf8_bytes": 911, "cache_tensor_bytes": null, "serialized_payload_bytes": 911, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 35535.04054190125, "hops": {"1": 22612.123124999925, "2": 12922.917416901328}, "receiver_prefill_ms": 306.18345795664936, "ttft_ms": 46.17637500632554, "decode_ms": 40.48833402339369}, "total_wall_ms": 35928.40516695287, "ttft_ms": 46.17637500632554, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "No", "evidence_artifact": {"ledger": {"facts_total": 58, "facts_retained": 15, "retention_rate": 0.25862068965517243, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.2222222222222222, "status": 0.75, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 911}, "ledger_by_hop": {"1": {"facts_total": 58, "facts_retained": 15, "retention_rate": 0.25862068965517243, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.2222222222222222, "status": 0.75, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 766}, "2": {"facts_total": 58, "facts_retained": 15, "retention_rate": 0.25862068965517243, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.2222222222222222, "status": 0.75, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 911}}}, "reference_key_id": "keys/synthetic/syn_03003_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "3d426d8999e6ba8503401b57d622bad0690314233e836579ad9658a82f039d3a", "output_hash": "1ea442a134b2a184bd5d40104401f2a37fbc09ccf3f4bc9da161c6099be3691d"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "f01839a80a71a1ad25e959de", "attempt_id": "f71a3875f471", "run_group_id": "E4_core:A5|closed|qwen3b-llama3b:h2", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T06:53:45.422338Z", "task_family": "synthetic", "document_id": "syn_03003_2000", "document_hash": "dddca1a38e3224870fcfae927c701c400fd885c19123ea840ee06a76a140f407", "question_id": "q08_revision", "split": "test", "cluster_id": "syn_03003_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "tokenizer_hashes": {"receiver": "b9d8872cf3a99da99e1fc4b0f71232a451632dc49ca4c6771549b4efa5c1c763"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 2, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1722, "query_tokens": 447, "message_tokens": 326, "message_utf8_bytes": 911, "cache_tensor_bytes": null, "serialized_payload_bytes": 911, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 35535.04054190125, "hops": {"1": 22612.123124999925, "2": 12922.917416901328}, "receiver_prefill_ms": 410.7916250359267, "ttft_ms": 82.83824997488409, "decode_ms": 95.97050002776086}, "total_wall_ms": 36125.26395788882, "ttft_ms": 82.83824997488409, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 3}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "911,925", "evidence_artifact": {"ledger": {"facts_total": 58, "facts_retained": 15, "retention_rate": 0.25862068965517243, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.2222222222222222, "status": 0.75, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 911}, "ledger_by_hop": {"1": {"facts_total": 58, "facts_retained": 15, "retention_rate": 0.25862068965517243, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.2222222222222222, "status": 0.75, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 766}, "2": {"facts_total": 58, "facts_retained": 15, "retention_rate": 0.25862068965517243, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.2222222222222222, "status": 0.75, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 911}}}, "reference_key_id": "keys/synthetic/syn_03003_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "3d426d8999e6ba8503401b57d622bad0690314233e836579ad9658a82f039d3a", "output_hash": "86be78720fc4b79f93e603e7d9e4861fcd25102612f73209fffc17d787056007"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "4a1369232892cc316dfa41dc", "attempt_id": "8c2d0b3ea68a", "run_group_id": "E4_core:A5|closed|qwen3b-llama3b:h2", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T06:53:46.030352Z", "task_family": "synthetic", "document_id": "syn_03003_2000", "document_hash": "dddca1a38e3224870fcfae927c701c400fd885c19123ea840ee06a76a140f407", "question_id": "q09_revision", "split": "test", "cluster_id": "syn_03003_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "tokenizer_hashes": {"receiver": "b9d8872cf3a99da99e1fc4b0f71232a451632dc49ca4c6771549b4efa5c1c763"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 2, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1722, "query_tokens": 440, "message_tokens": 326, "message_utf8_bytes": 911, "cache_tensor_bytes": null, "serialized_payload_bytes": 911, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 35535.04054190125, "hops": {"1": 22612.123124999925, "2": 12922.917416901328}, "receiver_prefill_ms": 352.25762496702373, "ttft_ms": 42.964083957485855, "decode_ms": 88.40487501583993}, "total_wall_ms": 36019.16695886757, "ttft_ms": 42.964083957485855, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 3}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "911,925", "evidence_artifact": {"ledger": {"facts_total": 58, "facts_retained": 15, "retention_rate": 0.25862068965517243, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.2222222222222222, "status": 0.75, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 911}, "ledger_by_hop": {"1": {"facts_total": 58, "facts_retained": 15, "retention_rate": 0.25862068965517243, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.2222222222222222, "status": 0.75, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 766}, "2": {"facts_total": 58, "facts_retained": 15, "retention_rate": 0.25862068965517243, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.2222222222222222, "status": 0.75, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 911}}}, "reference_key_id": "keys/synthetic/syn_03003_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "3d426d8999e6ba8503401b57d622bad0690314233e836579ad9658a82f039d3a", "output_hash": "86be78720fc4b79f93e603e7d9e4861fcd25102612f73209fffc17d787056007"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "5e5d390b6321d7dc71744f37", "attempt_id": "6cb7edd30af6", "run_group_id": "E4_core:A5|closed|qwen3b-llama3b:h2", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T06:53:46.532144Z", "task_family": "synthetic", "document_id": "syn_03003_2000", "document_hash": "dddca1a38e3224870fcfae927c701c400fd885c19123ea840ee06a76a140f407", "question_id": "q10_temporal", "split": "test", "cluster_id": "syn_03003_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "tokenizer_hashes": {"receiver": "b9d8872cf3a99da99e1fc4b0f71232a451632dc49ca4c6771549b4efa5c1c763"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 2, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1722, "query_tokens": 447, "message_tokens": 326, "message_utf8_bytes": 911, "cache_tensor_bytes": null, "serialized_payload_bytes": 911, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 35535.04054190125, "hops": {"1": 22612.123124999925, "2": 12922.917416901328}, "receiver_prefill_ms": 301.11141700763255, "ttft_ms": 35.55699996650219, "decode_ms": 61.66704196948558}, "total_wall_ms": 35934.222625917755, "ttft_ms": 35.55699996650219, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 58, "facts_retained": 15, "retention_rate": 0.25862068965517243, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.2222222222222222, "status": 0.75, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 911}, "ledger_by_hop": {"1": {"facts_total": 58, "facts_retained": 15, "retention_rate": 0.25862068965517243, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.2222222222222222, "status": 0.75, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 766}, "2": {"facts_total": 58, "facts_retained": 15, "retention_rate": 0.25862068965517243, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.2222222222222222, "status": 0.75, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 911}}}, "reference_key_id": "keys/synthetic/syn_03003_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "3d426d8999e6ba8503401b57d622bad0690314233e836579ad9658a82f039d3a", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "dd3d1d5f9b0c819a1741ed63", "attempt_id": "649b1c77364d", "run_group_id": "E4_core:A5|closed|qwen3b-llama3b:h2", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T06:53:46.949404Z", "task_family": "synthetic", "document_id": "syn_03003_2000", "document_hash": "dddca1a38e3224870fcfae927c701c400fd885c19123ea840ee06a76a140f407", "question_id": "q11_aggregation", "split": "test", "cluster_id": "syn_03003_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "tokenizer_hashes": {"receiver": "b9d8872cf3a99da99e1fc4b0f71232a451632dc49ca4c6771549b4efa5c1c763"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 2, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1722, "query_tokens": 441, "message_tokens": 326, "message_utf8_bytes": 911, "cache_tensor_bytes": null, "serialized_payload_bytes": 911, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 35535.04054190125, "hops": {"1": 22612.123124999925, "2": 12922.917416901328}, "receiver_prefill_ms": 315.79345802310854, "ttft_ms": 74.2230840260163, "decode_ms": 39.30049994960427}, "total_wall_ms": 35965.02558293287, "ttft_ms": 74.2230840260163, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 58, "facts_retained": 15, "retention_rate": 0.25862068965517243, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.2222222222222222, "status": 0.75, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 911}, "ledger_by_hop": {"1": {"facts_total": 58, "facts_retained": 15, "retention_rate": 0.25862068965517243, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.2222222222222222, "status": 0.75, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 766}, "2": {"facts_total": 58, "facts_retained": 15, "retention_rate": 0.25862068965517243, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.2222222222222222, "status": 0.75, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 911}}}, "reference_key_id": "keys/synthetic/syn_03003_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "3d426d8999e6ba8503401b57d622bad0690314233e836579ad9658a82f039d3a", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "11f7bf3abba289aaf47748b8", "attempt_id": "298db99b7034", "run_group_id": "E4_core:A5|closed|qwen3b-llama3b:h2", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T06:53:47.396690Z", "task_family": "synthetic", "document_id": "syn_03003_2000", "document_hash": "dddca1a38e3224870fcfae927c701c400fd885c19123ea840ee06a76a140f407", "question_id": "q12_unit_conversion", "split": "test", "cluster_id": "syn_03003_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "tokenizer_hashes": {"receiver": "b9d8872cf3a99da99e1fc4b0f71232a451632dc49ca4c6771549b4efa5c1c763"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 2, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1722, "query_tokens": 438, "message_tokens": 326, "message_utf8_bytes": 911, "cache_tensor_bytes": null, "serialized_payload_bytes": 911, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 35535.04054190125, "hops": {"1": 22612.123124999925, "2": 12922.917416901328}, "receiver_prefill_ms": 408.97208300884813, "ttft_ms": 25.603374931961298, "decode_ms": 103.92904095351696}, "total_wall_ms": 36073.98429187015, "ttft_ms": 25.603374931961298, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 3}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "17.43", "evidence_artifact": {"ledger": {"facts_total": 58, "facts_retained": 15, "retention_rate": 0.25862068965517243, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.2222222222222222, "status": 0.75, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 911}, "ledger_by_hop": {"1": {"facts_total": 58, "facts_retained": 15, "retention_rate": 0.25862068965517243, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.2222222222222222, "status": 0.75, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 766}, "2": {"facts_total": 58, "facts_retained": 15, "retention_rate": 0.25862068965517243, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.2222222222222222, "status": 0.75, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 911}}}, "reference_key_id": "keys/synthetic/syn_03003_2000.json", "task_scores": {"correct": 0.0, "abstained": 0.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "3d426d8999e6ba8503401b57d622bad0690314233e836579ad9658a82f039d3a", "output_hash": "1cd677e6c296f31fca744a7c4d96b5514fa6b036913240b8d689f1e277900d17"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "0c8536c32f1edc6b0805df5c", "attempt_id": "603fbf7c9049", "run_group_id": "E4_core:A5|closed|qwen3b-llama3b:h2", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T06:53:47.952246Z", "task_family": "synthetic", "document_id": "syn_03003_2000", "document_hash": "dddca1a38e3224870fcfae927c701c400fd885c19123ea840ee06a76a140f407", "question_id": "q13_unanswerable", "split": "test", "cluster_id": "syn_03003_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "tokenizer_hashes": {"receiver": "b9d8872cf3a99da99e1fc4b0f71232a451632dc49ca4c6771549b4efa5c1c763"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 2, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1722, "query_tokens": 442, "message_tokens": 326, "message_utf8_bytes": 911, "cache_tensor_bytes": null, "serialized_payload_bytes": 911, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 35535.04054190125, "hops": {"1": 22612.123124999925, "2": 12922.917416901328}, "receiver_prefill_ms": 350.26054200716317, "ttft_ms": 17.89437502156943, "decode_ms": 34.906583023257554}, "total_wall_ms": 35938.67241695989, "ttft_ms": 17.89437502156943, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 58, "facts_retained": 15, "retention_rate": 0.25862068965517243, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.2222222222222222, "status": 0.75, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 911}, "ledger_by_hop": {"1": {"facts_total": 58, "facts_retained": 15, "retention_rate": 0.25862068965517243, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.2222222222222222, "status": 0.75, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 766}, "2": {"facts_total": 58, "facts_retained": 15, "retention_rate": 0.25862068965517243, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.2222222222222222, "status": 0.75, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 911}}}, "reference_key_id": "keys/synthetic/syn_03003_2000.json", "task_scores": {"correct": 1.0, "abstained": 1.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "3d426d8999e6ba8503401b57d622bad0690314233e836579ad9658a82f039d3a", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "9779281ed707019e17f0d79b", "attempt_id": "162f9e460a2c", "run_group_id": "E4_core:A5|closed|qwen3b-llama3b:h2", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T06:53:48.372910Z", "task_family": "synthetic", "document_id": "syn_03003_2000", "document_hash": "dddca1a38e3224870fcfae927c701c400fd885c19123ea840ee06a76a140f407", "question_id": "q14_unanswerable", "split": "test", "cluster_id": "syn_03003_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "tokenizer_hashes": {"receiver": "b9d8872cf3a99da99e1fc4b0f71232a451632dc49ca4c6771549b4efa5c1c763"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 2, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1722, "query_tokens": 443, "message_tokens": 326, "message_utf8_bytes": 911, "cache_tensor_bytes": null, "serialized_payload_bytes": 911, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 35535.04054190125, "hops": {"1": 22612.123124999925, "2": 12922.917416901328}, "receiver_prefill_ms": 339.2857079161331, "ttft_ms": 44.00491702836007, "decode_ms": 70.96637494396418}, "total_wall_ms": 35989.79674989823, "ttft_ms": 44.00491702836007, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 58, "facts_retained": 15, "retention_rate": 0.25862068965517243, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.2222222222222222, "status": 0.75, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 911}, "ledger_by_hop": {"1": {"facts_total": 58, "facts_retained": 15, "retention_rate": 0.25862068965517243, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.2222222222222222, "status": 0.75, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 766}, "2": {"facts_total": 58, "facts_retained": 15, "retention_rate": 0.25862068965517243, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.2222222222222222, "status": 0.75, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 911}}}, "reference_key_id": "keys/synthetic/syn_03003_2000.json", "task_scores": {"correct": 1.0, "abstained": 1.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "3d426d8999e6ba8503401b57d622bad0690314233e836579ad9658a82f039d3a", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "223a4310a1aa0ee93e365e34", "attempt_id": "c465772edbd6", "run_group_id": "E4_core:A5|closed|qwen3b-llama3b:h4", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T06:54:17.253217Z", "task_family": "synthetic", "document_id": "syn_03003_2000", "document_hash": "dddca1a38e3224870fcfae927c701c400fd885c19123ea840ee06a76a140f407", "question_id": "q00_single_fact", "split": "test", "cluster_id": "syn_03003_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "tokenizer_hashes": {"receiver": "b9d8872cf3a99da99e1fc4b0f71232a451632dc49ca4c6771549b4efa5c1c763"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 4, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1722, "query_tokens": 430, "message_tokens": 326, "message_utf8_bytes": 911, "cache_tensor_bytes": null, "serialized_payload_bytes": 911, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 63941.011166898534, "hops": {"1": 22612.123124999925, "2": 12922.917416901328, "3": 15539.13074999582, "4": 12866.83987500146}, "receiver_prefill_ms": 296.51491600088775, "ttft_ms": 57.834708015434444, "decode_ms": 470.15170799568295}, "total_wall_ms": 64766.07366686221, "ttft_ms": 57.834708015434444, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 7}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "GR-79S-5839", "evidence_artifact": {"ledger": {"facts_total": 58, "facts_retained": 15, "retention_rate": 0.25862068965517243, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.2222222222222222, "status": 0.75, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 911}, "ledger_by_hop": {"1": {"facts_total": 58, "facts_retained": 15, "retention_rate": 0.25862068965517243, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.2222222222222222, "status": 0.75, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 766}, "2": {"facts_total": 58, "facts_retained": 15, "retention_rate": 0.25862068965517243, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.2222222222222222, "status": 0.75, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 911}, "3": {"facts_total": 58, "facts_retained": 15, "retention_rate": 0.25862068965517243, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.2222222222222222, "status": 0.75, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 911}, "4": {"facts_total": 58, "facts_retained": 15, "retention_rate": 0.25862068965517243, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.2222222222222222, "status": 0.75, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 911}}}, "reference_key_id": "keys/synthetic/syn_03003_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "3d426d8999e6ba8503401b57d622bad0690314233e836579ad9658a82f039d3a", "output_hash": "720cc40cd69e85e9728f6bea00b17a95ca07eee643e143a131840763958303cd"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "7a995ec9ceb3b0fac23a5d8d", "attempt_id": "27b0403680db", "run_group_id": "E4_core:A5|closed|qwen3b-llama3b:h4", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T06:54:18.096645Z", "task_family": "synthetic", "document_id": "syn_03003_2000", "document_hash": "dddca1a38e3224870fcfae927c701c400fd885c19123ea840ee06a76a140f407", "question_id": "q01_single_fact", "split": "test", "cluster_id": "syn_03003_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "tokenizer_hashes": {"receiver": "b9d8872cf3a99da99e1fc4b0f71232a451632dc49ca4c6771549b4efa5c1c763"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 4, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1722, "query_tokens": 434, "message_tokens": 326, "message_utf8_bytes": 911, "cache_tensor_bytes": null, "serialized_payload_bytes": 911, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 63941.011166898534, "hops": {"1": 22612.123124999925, "2": 12922.917416901328, "3": 15539.13074999582, "4": 12866.83987500146}, "receiver_prefill_ms": 349.50066602323204, "ttft_ms": 42.686874978244305, "decode_ms": 27.37237501423806}, "total_wall_ms": 64361.0959998332, "ttft_ms": 42.686874978244305, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 58, "facts_retained": 15, "retention_rate": 0.25862068965517243, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.2222222222222222, "status": 0.75, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 911}, "ledger_by_hop": {"1": {"facts_total": 58, "facts_retained": 15, "retention_rate": 0.25862068965517243, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.2222222222222222, "status": 0.75, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 766}, "2": {"facts_total": 58, "facts_retained": 15, "retention_rate": 0.25862068965517243, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.2222222222222222, "status": 0.75, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 911}, "3": {"facts_total": 58, "facts_retained": 15, "retention_rate": 0.25862068965517243, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.2222222222222222, "status": 0.75, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 911}, "4": {"facts_total": 58, "facts_retained": 15, "retention_rate": 0.25862068965517243, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.2222222222222222, "status": 0.75, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 911}}}, "reference_key_id": "keys/synthetic/syn_03003_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "3d426d8999e6ba8503401b57d622bad0690314233e836579ad9658a82f039d3a", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "c8d9aa8f7ad6a1ce18fdff8d", "attempt_id": "b16c6246812c", "run_group_id": "E4_core:A5|closed|qwen3b-llama3b:h4", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T06:54:18.534303Z", "task_family": "synthetic", "document_id": "syn_03003_2000", "document_hash": "dddca1a38e3224870fcfae927c701c400fd885c19123ea840ee06a76a140f407", "question_id": "q02_exact_copy", "split": "test", "cluster_id": "syn_03003_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "tokenizer_hashes": {"receiver": "b9d8872cf3a99da99e1fc4b0f71232a451632dc49ca4c6771549b4efa5c1c763"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 4, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1722, "query_tokens": 434, "message_tokens": 326, "message_utf8_bytes": 911, "cache_tensor_bytes": null, "serialized_payload_bytes": 911, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 63941.011166898534, "hops": {"1": 22612.123124999925, "2": 12922.917416901328, "3": 15539.13074999582, "4": 12866.83987500146}, "receiver_prefill_ms": 348.4913329593837, "ttft_ms": 17.56462501361966, "decode_ms": 19.03295808006078}, "total_wall_ms": 64326.54783281032, "ttft_ms": 17.56462501361966, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 58, "facts_retained": 15, "retention_rate": 0.25862068965517243, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.2222222222222222, "status": 0.75, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 911}, "ledger_by_hop": {"1": {"facts_total": 58, "facts_retained": 15, "retention_rate": 0.25862068965517243, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.2222222222222222, "status": 0.75, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 766}, "2": {"facts_total": 58, "facts_retained": 15, "retention_rate": 0.25862068965517243, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.2222222222222222, "status": 0.75, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 911}, "3": {"facts_total": 58, "facts_retained": 15, "retention_rate": 0.25862068965517243, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.2222222222222222, "status": 0.75, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 911}, "4": {"facts_total": 58, "facts_retained": 15, "retention_rate": 0.25862068965517243, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.2222222222222222, "status": 0.75, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 911}}}, "reference_key_id": "keys/synthetic/syn_03003_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "3d426d8999e6ba8503401b57d622bad0690314233e836579ad9658a82f039d3a", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "68a53fba4ddcf9f7c6a3c0e7", "attempt_id": "3a37a468364f", "run_group_id": "E4_core:A5|closed|qwen3b-llama3b:h4", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T06:54:18.936599Z", "task_family": "synthetic", "document_id": "syn_03003_2000", "document_hash": "dddca1a38e3224870fcfae927c701c400fd885c19123ea840ee06a76a140f407", "question_id": "q03_exact_copy", "split": "test", "cluster_id": "syn_03003_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "tokenizer_hashes": {"receiver": "b9d8872cf3a99da99e1fc4b0f71232a451632dc49ca4c6771549b4efa5c1c763"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 4, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1722, "query_tokens": 429, "message_tokens": 326, "message_utf8_bytes": 911, "cache_tensor_bytes": null, "serialized_payload_bytes": 911, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 63941.011166898534, "hops": {"1": 22612.123124999925, "2": 12922.917416901328, "3": 15539.13074999582, "4": 12866.83987500146}, "receiver_prefill_ms": 358.08845807332546, "ttft_ms": 42.636957950890064, "decode_ms": 112.55133396480232}, "total_wall_ms": 64454.87870881334, "ttft_ms": 42.636957950890064, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 58, "facts_retained": 15, "retention_rate": 0.25862068965517243, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.2222222222222222, "status": 0.75, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 911}, "ledger_by_hop": {"1": {"facts_total": 58, "facts_retained": 15, "retention_rate": 0.25862068965517243, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.2222222222222222, "status": 0.75, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 766}, "2": {"facts_total": 58, "facts_retained": 15, "retention_rate": 0.25862068965517243, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.2222222222222222, "status": 0.75, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 911}, "3": {"facts_total": 58, "facts_retained": 15, "retention_rate": 0.25862068965517243, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.2222222222222222, "status": 0.75, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 911}, "4": {"facts_total": 58, "facts_retained": 15, "retention_rate": 0.25862068965517243, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.2222222222222222, "status": 0.75, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 911}}}, "reference_key_id": "keys/synthetic/syn_03003_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "3d426d8999e6ba8503401b57d622bad0690314233e836579ad9658a82f039d3a", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "5ff54ddf83dac36316b57733", "attempt_id": "2e1584fac01f", "run_group_id": "E4_core:A5|closed|qwen3b-llama3b:h4", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T06:54:19.469729Z", "task_family": "synthetic", "document_id": "syn_03003_2000", "document_hash": "dddca1a38e3224870fcfae927c701c400fd885c19123ea840ee06a76a140f407", "question_id": "q04_multi_hop", "split": "test", "cluster_id": "syn_03003_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "tokenizer_hashes": {"receiver": "b9d8872cf3a99da99e1fc4b0f71232a451632dc49ca4c6771549b4efa5c1c763"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 4, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1722, "query_tokens": 438, "message_tokens": 326, "message_utf8_bytes": 911, "cache_tensor_bytes": null, "serialized_payload_bytes": 911, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 63941.011166898534, "hops": {"1": 22612.123124999925, "2": 12922.917416901328, "3": 15539.13074999582, "4": 12866.83987500146}, "receiver_prefill_ms": 422.4038749234751, "ttft_ms": 26.402708957903087, "decode_ms": 36.52924997732043}, "total_wall_ms": 64426.768166944385, "ttft_ms": 26.402708957903087, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 58, "facts_retained": 15, "retention_rate": 0.25862068965517243, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.2222222222222222, "status": 0.75, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 911}, "ledger_by_hop": {"1": {"facts_total": 58, "facts_retained": 15, "retention_rate": 0.25862068965517243, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.2222222222222222, "status": 0.75, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 766}, "2": {"facts_total": 58, "facts_retained": 15, "retention_rate": 0.25862068965517243, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.2222222222222222, "status": 0.75, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 911}, "3": {"facts_total": 58, "facts_retained": 15, "retention_rate": 0.25862068965517243, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.2222222222222222, "status": 0.75, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 911}, "4": {"facts_total": 58, "facts_retained": 15, "retention_rate": 0.25862068965517243, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.2222222222222222, "status": 0.75, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 911}}}, "reference_key_id": "keys/synthetic/syn_03003_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "3d426d8999e6ba8503401b57d622bad0690314233e836579ad9658a82f039d3a", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "6571d1021aa3adbeeac60408", "attempt_id": "06d3be546162", "run_group_id": "E4_core:A5|closed|qwen3b-llama3b:h4", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T06:54:19.974356Z", "task_family": "synthetic", "document_id": "syn_03003_2000", "document_hash": "dddca1a38e3224870fcfae927c701c400fd885c19123ea840ee06a76a140f407", "question_id": "q05_multi_hop", "split": "test", "cluster_id": "syn_03003_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "tokenizer_hashes": {"receiver": "b9d8872cf3a99da99e1fc4b0f71232a451632dc49ca4c6771549b4efa5c1c763"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 4, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1722, "query_tokens": 438, "message_tokens": 326, "message_utf8_bytes": 911, "cache_tensor_bytes": null, "serialized_payload_bytes": 911, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 63941.011166898534, "hops": {"1": 22612.123124999925, "2": 12922.917416901328, "3": 15539.13074999582, "4": 12866.83987500146}, "receiver_prefill_ms": 353.47854101564735, "ttft_ms": 21.36449993122369, "decode_ms": 33.694792073220015}, "total_wall_ms": 64350.20562587306, "ttft_ms": 21.36449993122369, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 58, "facts_retained": 15, "retention_rate": 0.25862068965517243, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.2222222222222222, "status": 0.75, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 911}, "ledger_by_hop": {"1": {"facts_total": 58, "facts_retained": 15, "retention_rate": 0.25862068965517243, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.2222222222222222, "status": 0.75, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 766}, "2": {"facts_total": 58, "facts_retained": 15, "retention_rate": 0.25862068965517243, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.2222222222222222, "status": 0.75, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 911}, "3": {"facts_total": 58, "facts_retained": 15, "retention_rate": 0.25862068965517243, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.2222222222222222, "status": 0.75, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 911}, "4": {"facts_total": 58, "facts_retained": 15, "retention_rate": 0.25862068965517243, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.2222222222222222, "status": 0.75, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 911}}}, "reference_key_id": "keys/synthetic/syn_03003_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "3d426d8999e6ba8503401b57d622bad0690314233e836579ad9658a82f039d3a", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "e5b23ef8b440b33b08b003ce", "attempt_id": "b44cc7f33075", "run_group_id": "E4_core:A5|closed|qwen3b-llama3b:h4", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T06:54:20.400633Z", "task_family": "synthetic", "document_id": "syn_03003_2000", "document_hash": "dddca1a38e3224870fcfae927c701c400fd885c19123ea840ee06a76a140f407", "question_id": "q06_negation", "split": "test", "cluster_id": "syn_03003_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "tokenizer_hashes": {"receiver": "b9d8872cf3a99da99e1fc4b0f71232a451632dc49ca4c6771549b4efa5c1c763"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 4, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1722, "query_tokens": 431, "message_tokens": 326, "message_utf8_bytes": 911, "cache_tensor_bytes": null, "serialized_payload_bytes": 911, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 63941.011166898534, "hops": {"1": 22612.123124999925, "2": 12922.917416901328, "3": 15539.13074999582, "4": 12866.83987500146}, "receiver_prefill_ms": 327.0741669693962, "ttft_ms": 57.985124993138015, "decode_ms": 116.04654206894338}, "total_wall_ms": 64442.96195881907, "ttft_ms": 57.985124993138015, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 2}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "No.", "evidence_artifact": {"ledger": {"facts_total": 58, "facts_retained": 15, "retention_rate": 0.25862068965517243, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.2222222222222222, "status": 0.75, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 911}, "ledger_by_hop": {"1": {"facts_total": 58, "facts_retained": 15, "retention_rate": 0.25862068965517243, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.2222222222222222, "status": 0.75, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 766}, "2": {"facts_total": 58, "facts_retained": 15, "retention_rate": 0.25862068965517243, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.2222222222222222, "status": 0.75, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 911}, "3": {"facts_total": 58, "facts_retained": 15, "retention_rate": 0.25862068965517243, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.2222222222222222, "status": 0.75, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 911}, "4": {"facts_total": 58, "facts_retained": 15, "retention_rate": 0.25862068965517243, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.2222222222222222, "status": 0.75, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 911}}}, "reference_key_id": "keys/synthetic/syn_03003_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "3d426d8999e6ba8503401b57d622bad0690314233e836579ad9658a82f039d3a", "output_hash": "38178a20b470cfd18299fb1593dd4ba706f1a83321dac051d01086efd8b7a96f"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "fdddd68febb8c977411850e8", "attempt_id": "6480e9ca8a4c", "run_group_id": "E4_core:A5|closed|qwen3b-llama3b:h4", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T06:54:20.921060Z", "task_family": "synthetic", "document_id": "syn_03003_2000", "document_hash": "dddca1a38e3224870fcfae927c701c400fd885c19123ea840ee06a76a140f407", "question_id": "q07_negation", "split": "test", "cluster_id": "syn_03003_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "tokenizer_hashes": {"receiver": "b9d8872cf3a99da99e1fc4b0f71232a451632dc49ca4c6771549b4efa5c1c763"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 4, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1722, "query_tokens": 429, "message_tokens": 326, "message_utf8_bytes": 911, "cache_tensor_bytes": null, "serialized_payload_bytes": 911, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 63941.011166898534, "hops": {"1": 22612.123124999925, "2": 12922.917416901328, "3": 15539.13074999582, "4": 12866.83987500146}, "receiver_prefill_ms": 456.0537500074133, "ttft_ms": 36.670624976977706, "decode_ms": 18.01904197782278}, "total_wall_ms": 64452.422250877135, "ttft_ms": 36.670624976977706, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "No", "evidence_artifact": {"ledger": {"facts_total": 58, "facts_retained": 15, "retention_rate": 0.25862068965517243, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.2222222222222222, "status": 0.75, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 911}, "ledger_by_hop": {"1": {"facts_total": 58, "facts_retained": 15, "retention_rate": 0.25862068965517243, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.2222222222222222, "status": 0.75, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 766}, "2": {"facts_total": 58, "facts_retained": 15, "retention_rate": 0.25862068965517243, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.2222222222222222, "status": 0.75, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 911}, "3": {"facts_total": 58, "facts_retained": 15, "retention_rate": 0.25862068965517243, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.2222222222222222, "status": 0.75, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 911}, "4": {"facts_total": 58, "facts_retained": 15, "retention_rate": 0.25862068965517243, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.2222222222222222, "status": 0.75, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 911}}}, "reference_key_id": "keys/synthetic/syn_03003_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "3d426d8999e6ba8503401b57d622bad0690314233e836579ad9658a82f039d3a", "output_hash": "1ea442a134b2a184bd5d40104401f2a37fbc09ccf3f4bc9da161c6099be3691d"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "a468a9dce6947b3ecdd69d38", "attempt_id": "1052bdd5523b", "run_group_id": "E4_core:A5|closed|qwen3b-llama3b:h4", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T06:54:21.449526Z", "task_family": "synthetic", "document_id": "syn_03003_2000", "document_hash": "dddca1a38e3224870fcfae927c701c400fd885c19123ea840ee06a76a140f407", "question_id": "q08_revision", "split": "test", "cluster_id": "syn_03003_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "tokenizer_hashes": {"receiver": "b9d8872cf3a99da99e1fc4b0f71232a451632dc49ca4c6771549b4efa5c1c763"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 4, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1722, "query_tokens": 447, "message_tokens": 326, "message_utf8_bytes": 911, "cache_tensor_bytes": null, "serialized_payload_bytes": 911, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 63941.011166898534, "hops": {"1": 22612.123124999925, "2": 12922.917416901328, "3": 15539.13074999582, "4": 12866.83987500146}, "receiver_prefill_ms": 341.8273749994114, "ttft_ms": 21.44733304157853, "decode_ms": 99.41616700962186}, "total_wall_ms": 64404.25725094974, "ttft_ms": 21.44733304157853, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 3}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "911,925", "evidence_artifact": {"ledger": {"facts_total": 58, "facts_retained": 15, "retention_rate": 0.25862068965517243, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.2222222222222222, "status": 0.75, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 911}, "ledger_by_hop": {"1": {"facts_total": 58, "facts_retained": 15, "retention_rate": 0.25862068965517243, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.2222222222222222, "status": 0.75, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 766}, "2": {"facts_total": 58, "facts_retained": 15, "retention_rate": 0.25862068965517243, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.2222222222222222, "status": 0.75, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 911}, "3": {"facts_total": 58, "facts_retained": 15, "retention_rate": 0.25862068965517243, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.2222222222222222, "status": 0.75, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 911}, "4": {"facts_total": 58, "facts_retained": 15, "retention_rate": 0.25862068965517243, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.2222222222222222, "status": 0.75, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 911}}}, "reference_key_id": "keys/synthetic/syn_03003_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "3d426d8999e6ba8503401b57d622bad0690314233e836579ad9658a82f039d3a", "output_hash": "86be78720fc4b79f93e603e7d9e4861fcd25102612f73209fffc17d787056007"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "7a52a545066ab15e28bf2a52", "attempt_id": "810d90fa6932", "run_group_id": "E4_core:A5|closed|qwen3b-llama3b:h4", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T06:54:21.930169Z", "task_family": "synthetic", "document_id": "syn_03003_2000", "document_hash": "dddca1a38e3224870fcfae927c701c400fd885c19123ea840ee06a76a140f407", "question_id": "q09_revision", "split": "test", "cluster_id": "syn_03003_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "tokenizer_hashes": {"receiver": "b9d8872cf3a99da99e1fc4b0f71232a451632dc49ca4c6771549b4efa5c1c763"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 4, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1722, "query_tokens": 440, "message_tokens": 326, "message_utf8_bytes": 911, "cache_tensor_bytes": null, "serialized_payload_bytes": 911, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 63941.011166898534, "hops": {"1": 22612.123124999925, "2": 12922.917416901328, "3": 15539.13074999582, "4": 12866.83987500146}, "receiver_prefill_ms": 302.5024168891832, "ttft_ms": 54.836791940033436, "decode_ms": 185.0304170511663}, "total_wall_ms": 64483.93624986056, "ttft_ms": 54.836791940033436, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 3}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "911,925", "evidence_artifact": {"ledger": {"facts_total": 58, "facts_retained": 15, "retention_rate": 0.25862068965517243, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.2222222222222222, "status": 0.75, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 911}, "ledger_by_hop": {"1": {"facts_total": 58, "facts_retained": 15, "retention_rate": 0.25862068965517243, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.2222222222222222, "status": 0.75, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 766}, "2": {"facts_total": 58, "facts_retained": 15, "retention_rate": 0.25862068965517243, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.2222222222222222, "status": 0.75, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 911}, "3": {"facts_total": 58, "facts_retained": 15, "retention_rate": 0.25862068965517243, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.2222222222222222, "status": 0.75, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 911}, "4": {"facts_total": 58, "facts_retained": 15, "retention_rate": 0.25862068965517243, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.2222222222222222, "status": 0.75, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 911}}}, "reference_key_id": "keys/synthetic/syn_03003_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "3d426d8999e6ba8503401b57d622bad0690314233e836579ad9658a82f039d3a", "output_hash": "86be78720fc4b79f93e603e7d9e4861fcd25102612f73209fffc17d787056007"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "6cb13950843b9ace41fd3ade", "attempt_id": "bffb2f5680a1", "run_group_id": "E4_core:A5|closed|qwen3b-llama3b:h4", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T06:54:22.491285Z", "task_family": "synthetic", "document_id": "syn_03003_2000", "document_hash": "dddca1a38e3224870fcfae927c701c400fd885c19123ea840ee06a76a140f407", "question_id": "q10_temporal", "split": "test", "cluster_id": "syn_03003_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "tokenizer_hashes": {"receiver": "b9d8872cf3a99da99e1fc4b0f71232a451632dc49ca4c6771549b4efa5c1c763"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 4, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1722, "query_tokens": 447, "message_tokens": 326, "message_utf8_bytes": 911, "cache_tensor_bytes": null, "serialized_payload_bytes": 911, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 63941.011166898534, "hops": {"1": 22612.123124999925, "2": 12922.917416901328, "3": 15539.13074999582, "4": 12866.83987500146}, "receiver_prefill_ms": 453.7690830184147, "ttft_ms": 36.57304204534739, "decode_ms": 24.254375020973384}, "total_wall_ms": 64456.2065419741, "ttft_ms": 36.57304204534739, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 58, "facts_retained": 15, "retention_rate": 0.25862068965517243, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.2222222222222222, "status": 0.75, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 911}, "ledger_by_hop": {"1": {"facts_total": 58, "facts_retained": 15, "retention_rate": 0.25862068965517243, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.2222222222222222, "status": 0.75, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 766}, "2": {"facts_total": 58, "facts_retained": 15, "retention_rate": 0.25862068965517243, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.2222222222222222, "status": 0.75, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 911}, "3": {"facts_total": 58, "facts_retained": 15, "retention_rate": 0.25862068965517243, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.2222222222222222, "status": 0.75, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 911}, "4": {"facts_total": 58, "facts_retained": 15, "retention_rate": 0.25862068965517243, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.2222222222222222, "status": 0.75, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 911}}}, "reference_key_id": "keys/synthetic/syn_03003_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "3d426d8999e6ba8503401b57d622bad0690314233e836579ad9658a82f039d3a", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "6117499771ce02bd9353ea9c", "attempt_id": "c4764c7424bb", "run_group_id": "E4_core:A5|closed|qwen3b-llama3b:h4", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T06:54:23.024746Z", "task_family": "synthetic", "document_id": "syn_03003_2000", "document_hash": "dddca1a38e3224870fcfae927c701c400fd885c19123ea840ee06a76a140f407", "question_id": "q11_aggregation", "split": "test", "cluster_id": "syn_03003_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "tokenizer_hashes": {"receiver": "b9d8872cf3a99da99e1fc4b0f71232a451632dc49ca4c6771549b4efa5c1c763"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 4, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1722, "query_tokens": 441, "message_tokens": 326, "message_utf8_bytes": 911, "cache_tensor_bytes": null, "serialized_payload_bytes": 911, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 63941.011166898534, "hops": {"1": 22612.123124999925, "2": 12922.917416901328, "3": 15539.13074999582, "4": 12866.83987500146}, "receiver_prefill_ms": 351.55279107857496, "ttft_ms": 27.292959042824805, "decode_ms": 36.03687498252839}, "total_wall_ms": 64356.32525093388, "ttft_ms": 27.292959042824805, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 58, "facts_retained": 15, "retention_rate": 0.25862068965517243, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.2222222222222222, "status": 0.75, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 911}, "ledger_by_hop": {"1": {"facts_total": 58, "facts_retained": 15, "retention_rate": 0.25862068965517243, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.2222222222222222, "status": 0.75, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 766}, "2": {"facts_total": 58, "facts_retained": 15, "retention_rate": 0.25862068965517243, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.2222222222222222, "status": 0.75, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 911}, "3": {"facts_total": 58, "facts_retained": 15, "retention_rate": 0.25862068965517243, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.2222222222222222, "status": 0.75, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 911}, "4": {"facts_total": 58, "facts_retained": 15, "retention_rate": 0.25862068965517243, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.2222222222222222, "status": 0.75, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 911}}}, "reference_key_id": "keys/synthetic/syn_03003_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "3d426d8999e6ba8503401b57d622bad0690314233e836579ad9658a82f039d3a", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "2018b827cb868215eafe3dd1", "attempt_id": "0d4be97d8bee", "run_group_id": "E4_core:A5|closed|qwen3b-llama3b:h4", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T06:54:23.457108Z", "task_family": "synthetic", "document_id": "syn_03003_2000", "document_hash": "dddca1a38e3224870fcfae927c701c400fd885c19123ea840ee06a76a140f407", "question_id": "q12_unit_conversion", "split": "test", "cluster_id": "syn_03003_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "tokenizer_hashes": {"receiver": "b9d8872cf3a99da99e1fc4b0f71232a451632dc49ca4c6771549b4efa5c1c763"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 4, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1722, "query_tokens": 438, "message_tokens": 326, "message_utf8_bytes": 911, "cache_tensor_bytes": null, "serialized_payload_bytes": 911, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 63941.011166898534, "hops": {"1": 22612.123124999925, "2": 12922.917416901328, "3": 15539.13074999582, "4": 12866.83987500146}, "receiver_prefill_ms": 301.174083026126, "ttft_ms": 58.33666608668864, "decode_ms": 153.66199996788055}, "total_wall_ms": 64454.708624864, "ttft_ms": 58.33666608668864, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 3}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "17.43", "evidence_artifact": {"ledger": {"facts_total": 58, "facts_retained": 15, "retention_rate": 0.25862068965517243, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.2222222222222222, "status": 0.75, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 911}, "ledger_by_hop": {"1": {"facts_total": 58, "facts_retained": 15, "retention_rate": 0.25862068965517243, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.2222222222222222, "status": 0.75, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 766}, "2": {"facts_total": 58, "facts_retained": 15, "retention_rate": 0.25862068965517243, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.2222222222222222, "status": 0.75, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 911}, "3": {"facts_total": 58, "facts_retained": 15, "retention_rate": 0.25862068965517243, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.2222222222222222, "status": 0.75, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 911}, "4": {"facts_total": 58, "facts_retained": 15, "retention_rate": 0.25862068965517243, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.2222222222222222, "status": 0.75, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 911}}}, "reference_key_id": "keys/synthetic/syn_03003_2000.json", "task_scores": {"correct": 0.0, "abstained": 0.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "3d426d8999e6ba8503401b57d622bad0690314233e836579ad9658a82f039d3a", "output_hash": "1cd677e6c296f31fca744a7c4d96b5514fa6b036913240b8d689f1e277900d17"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "8e40a251ea12b46a51fe20d5", "attempt_id": "366e66c9ae50", "run_group_id": "E4_core:A5|closed|qwen3b-llama3b:h4", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T06:54:23.987888Z", "task_family": "synthetic", "document_id": "syn_03003_2000", "document_hash": "dddca1a38e3224870fcfae927c701c400fd885c19123ea840ee06a76a140f407", "question_id": "q13_unanswerable", "split": "test", "cluster_id": "syn_03003_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "tokenizer_hashes": {"receiver": "b9d8872cf3a99da99e1fc4b0f71232a451632dc49ca4c6771549b4efa5c1c763"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 4, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1722, "query_tokens": 442, "message_tokens": 326, "message_utf8_bytes": 911, "cache_tensor_bytes": null, "serialized_payload_bytes": 911, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 63941.011166898534, "hops": {"1": 22612.123124999925, "2": 12922.917416901328, "3": 15539.13074999582, "4": 12866.83987500146}, "receiver_prefill_ms": 475.9938329225406, "ttft_ms": 44.26345799583942, "decode_ms": 18.330292077735066}, "total_wall_ms": 64480.09758384433, "ttft_ms": 44.26345799583942, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 58, "facts_retained": 15, "retention_rate": 0.25862068965517243, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.2222222222222222, "status": 0.75, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 911}, "ledger_by_hop": {"1": {"facts_total": 58, "facts_retained": 15, "retention_rate": 0.25862068965517243, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.2222222222222222, "status": 0.75, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 766}, "2": {"facts_total": 58, "facts_retained": 15, "retention_rate": 0.25862068965517243, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.2222222222222222, "status": 0.75, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 911}, "3": {"facts_total": 58, "facts_retained": 15, "retention_rate": 0.25862068965517243, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.2222222222222222, "status": 0.75, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 911}, "4": {"facts_total": 58, "facts_retained": 15, "retention_rate": 0.25862068965517243, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.2222222222222222, "status": 0.75, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 911}}}, "reference_key_id": "keys/synthetic/syn_03003_2000.json", "task_scores": {"correct": 1.0, "abstained": 1.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "3d426d8999e6ba8503401b57d622bad0690314233e836579ad9658a82f039d3a", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "5d2e7e62cee5b5415e80b60f", "attempt_id": "696761969067", "run_group_id": "E4_core:A5|closed|qwen3b-llama3b:h4", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T06:54:24.544322Z", "task_family": "synthetic", "document_id": "syn_03003_2000", "document_hash": "dddca1a38e3224870fcfae927c701c400fd885c19123ea840ee06a76a140f407", "question_id": "q14_unanswerable", "split": "test", "cluster_id": "syn_03003_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "tokenizer_hashes": {"receiver": "b9d8872cf3a99da99e1fc4b0f71232a451632dc49ca4c6771549b4efa5c1c763"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 4, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1722, "query_tokens": 443, "message_tokens": 326, "message_utf8_bytes": 911, "cache_tensor_bytes": null, "serialized_payload_bytes": 911, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 63941.011166898534, "hops": {"1": 22612.123124999925, "2": 12922.917416901328, "3": 15539.13074999582, "4": 12866.83987500146}, "receiver_prefill_ms": 340.95120802521706, "ttft_ms": 26.262958999723196, "decode_ms": 38.07395894546062}, "total_wall_ms": 64346.764457877725, "ttft_ms": 26.262958999723196, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 58, "facts_retained": 15, "retention_rate": 0.25862068965517243, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.2222222222222222, "status": 0.75, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 911}, "ledger_by_hop": {"1": {"facts_total": 58, "facts_retained": 15, "retention_rate": 0.25862068965517243, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.2222222222222222, "status": 0.75, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 766}, "2": {"facts_total": 58, "facts_retained": 15, "retention_rate": 0.25862068965517243, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.2222222222222222, "status": 0.75, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 911}, "3": {"facts_total": 58, "facts_retained": 15, "retention_rate": 0.25862068965517243, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.2222222222222222, "status": 0.75, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 911}, "4": {"facts_total": 58, "facts_retained": 15, "retention_rate": 0.25862068965517243, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.2222222222222222, "status": 0.75, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 911}}}, "reference_key_id": "keys/synthetic/syn_03003_2000.json", "task_scores": {"correct": 1.0, "abstained": 1.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "3d426d8999e6ba8503401b57d622bad0690314233e836579ad9658a82f039d3a", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "17df3454c8c70c9159d5a537", "attempt_id": "d737b0b0600d", "run_group_id": "E4_core:A5|closed|qwen3b-llama3b:h8", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T06:55:22.571778Z", "task_family": "synthetic", "document_id": "syn_03003_2000", "document_hash": "dddca1a38e3224870fcfae927c701c400fd885c19123ea840ee06a76a140f407", "question_id": "q00_single_fact", "split": "test", "cluster_id": "syn_03003_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "tokenizer_hashes": {"receiver": "b9d8872cf3a99da99e1fc4b0f71232a451632dc49ca4c6771549b4efa5c1c763"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 8, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1722, "query_tokens": 430, "message_tokens": 326, "message_utf8_bytes": 911, "cache_tensor_bytes": null, "serialized_payload_bytes": 911, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 121541.23483377043, "hops": {"1": 22612.123124999925, "2": 12922.917416901328, "3": 15539.13074999582, "4": 12866.83987500146, "5": 15885.767916915938, "6": 12968.57120806817, "7": 15879.040624946356, "8": 12866.843916941434}, "receiver_prefill_ms": 354.08966708928347, "ttft_ms": 20.52500005811453, "decode_ms": 260.1821249118075}, "total_wall_ms": 122176.5873337863, "ttft_ms": 20.52500005811453, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 7}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "GR-79S-5839", "evidence_artifact": {"ledger": {"facts_total": 58, "facts_retained": 15, "retention_rate": 0.25862068965517243, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.2222222222222222, "status": 0.75, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 911}, "ledger_by_hop": {"1": {"facts_total": 58, "facts_retained": 15, "retention_rate": 0.25862068965517243, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.2222222222222222, "status": 0.75, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 766}, "2": {"facts_total": 58, "facts_retained": 15, "retention_rate": 0.25862068965517243, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.2222222222222222, "status": 0.75, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 911}, "3": {"facts_total": 58, "facts_retained": 15, "retention_rate": 0.25862068965517243, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.2222222222222222, "status": 0.75, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 911}, "4": {"facts_total": 58, "facts_retained": 15, "retention_rate": 0.25862068965517243, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.2222222222222222, "status": 0.75, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 911}, "5": {"facts_total": 58, "facts_retained": 15, "retention_rate": 0.25862068965517243, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.2222222222222222, "status": 0.75, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 911}, "6": {"facts_total": 58, "facts_retained": 15, "retention_rate": 0.25862068965517243, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.2222222222222222, "status": 0.75, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 911}, "7": {"facts_total": 58, "facts_retained": 15, "retention_rate": 0.25862068965517243, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.2222222222222222, "status": 0.75, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 911}, "8": {"facts_total": 58, "facts_retained": 15, "retention_rate": 0.25862068965517243, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.2222222222222222, "status": 0.75, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 911}}}, "reference_key_id": "keys/synthetic/syn_03003_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "3d426d8999e6ba8503401b57d622bad0690314233e836579ad9658a82f039d3a", "output_hash": "720cc40cd69e85e9728f6bea00b17a95ca07eee643e143a131840763958303cd"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "3c5826755147084ff8295d2b", "attempt_id": "92f77e4abc57", "run_group_id": "E4_core:A5|closed|qwen3b-llama3b:h8", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T06:55:23.225029Z", "task_family": "synthetic", "document_id": "syn_03003_2000", "document_hash": "dddca1a38e3224870fcfae927c701c400fd885c19123ea840ee06a76a140f407", "question_id": "q01_single_fact", "split": "test", "cluster_id": "syn_03003_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "tokenizer_hashes": {"receiver": "b9d8872cf3a99da99e1fc4b0f71232a451632dc49ca4c6771549b4efa5c1c763"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 8, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1722, "query_tokens": 434, "message_tokens": 326, "message_utf8_bytes": 911, "cache_tensor_bytes": null, "serialized_payload_bytes": 911, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 121541.23483377043, "hops": {"1": 22612.123124999925, "2": 12922.917416901328, "3": 15539.13074999582, "4": 12866.83987500146, "5": 15885.767916915938, "6": 12968.57120806817, "7": 15879.040624946356, "8": 12866.843916941434}, "receiver_prefill_ms": 449.0134170046076, "ttft_ms": 21.922457963228226, "decode_ms": 37.708249990828335}, "total_wall_ms": 122050.35812582355, "ttft_ms": 21.922457963228226, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 58, "facts_retained": 15, "retention_rate": 0.25862068965517243, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.2222222222222222, "status": 0.75, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 911}, "ledger_by_hop": {"1": {"facts_total": 58, "facts_retained": 15, "retention_rate": 0.25862068965517243, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.2222222222222222, "status": 0.75, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 766}, "2": {"facts_total": 58, "facts_retained": 15, "retention_rate": 0.25862068965517243, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.2222222222222222, "status": 0.75, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 911}, "3": {"facts_total": 58, "facts_retained": 15, "retention_rate": 0.25862068965517243, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.2222222222222222, "status": 0.75, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 911}, "4": {"facts_total": 58, "facts_retained": 15, "retention_rate": 0.25862068965517243, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.2222222222222222, "status": 0.75, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 911}, "5": {"facts_total": 58, "facts_retained": 15, "retention_rate": 0.25862068965517243, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.2222222222222222, "status": 0.75, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 911}, "6": {"facts_total": 58, "facts_retained": 15, "retention_rate": 0.25862068965517243, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.2222222222222222, "status": 0.75, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 911}, "7": {"facts_total": 58, "facts_retained": 15, "retention_rate": 0.25862068965517243, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.2222222222222222, "status": 0.75, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 911}, "8": {"facts_total": 58, "facts_retained": 15, "retention_rate": 0.25862068965517243, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.2222222222222222, "status": 0.75, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 911}}}, "reference_key_id": "keys/synthetic/syn_03003_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "3d426d8999e6ba8503401b57d622bad0690314233e836579ad9658a82f039d3a", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "64390401d4ddbb8c51756a97", "attempt_id": "5ed9a1b5723b", "run_group_id": "E4_core:A5|closed|qwen3b-llama3b:h8", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T06:55:23.751582Z", "task_family": "synthetic", "document_id": "syn_03003_2000", "document_hash": "dddca1a38e3224870fcfae927c701c400fd885c19123ea840ee06a76a140f407", "question_id": "q02_exact_copy", "split": "test", "cluster_id": "syn_03003_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "tokenizer_hashes": {"receiver": "b9d8872cf3a99da99e1fc4b0f71232a451632dc49ca4c6771549b4efa5c1c763"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 8, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1722, "query_tokens": 434, "message_tokens": 326, "message_utf8_bytes": 911, "cache_tensor_bytes": null, "serialized_payload_bytes": 911, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 121541.23483377043, "hops": {"1": 22612.123124999925, "2": 12922.917416901328, "3": 15539.13074999582, "4": 12866.83987500146, "5": 15885.767916915938, "6": 12968.57120806817, "7": 15879.040624946356, "8": 12866.843916941434}, "receiver_prefill_ms": 360.7482500374317, "ttft_ms": 21.572500001639128, "decode_ms": 31.333207967691123}, "total_wall_ms": 121955.4445007816, "ttft_ms": 21.572500001639128, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 58, "facts_retained": 15, "retention_rate": 0.25862068965517243, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.2222222222222222, "status": 0.75, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 911}, "ledger_by_hop": {"1": {"facts_total": 58, "facts_retained": 15, "retention_rate": 0.25862068965517243, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.2222222222222222, "status": 0.75, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 766}, "2": {"facts_total": 58, "facts_retained": 15, "retention_rate": 0.25862068965517243, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.2222222222222222, "status": 0.75, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 911}, "3": {"facts_total": 58, "facts_retained": 15, "retention_rate": 0.25862068965517243, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.2222222222222222, "status": 0.75, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 911}, "4": {"facts_total": 58, "facts_retained": 15, "retention_rate": 0.25862068965517243, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.2222222222222222, "status": 0.75, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 911}, "5": {"facts_total": 58, "facts_retained": 15, "retention_rate": 0.25862068965517243, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.2222222222222222, "status": 0.75, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 911}, "6": {"facts_total": 58, "facts_retained": 15, "retention_rate": 0.25862068965517243, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.2222222222222222, "status": 0.75, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 911}, "7": {"facts_total": 58, "facts_retained": 15, "retention_rate": 0.25862068965517243, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.2222222222222222, "status": 0.75, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 911}, "8": {"facts_total": 58, "facts_retained": 15, "retention_rate": 0.25862068965517243, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.2222222222222222, "status": 0.75, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 911}}}, "reference_key_id": "keys/synthetic/syn_03003_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "3d426d8999e6ba8503401b57d622bad0690314233e836579ad9658a82f039d3a", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "88876c5d4540aa6cf07ce268", "attempt_id": "afefa2b7e587", "run_group_id": "E4_core:A5|closed|qwen3b-llama3b:h8", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T06:55:24.183532Z", "task_family": "synthetic", "document_id": "syn_03003_2000", "document_hash": "dddca1a38e3224870fcfae927c701c400fd885c19123ea840ee06a76a140f407", "question_id": "q03_exact_copy", "split": "test", "cluster_id": "syn_03003_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "tokenizer_hashes": {"receiver": "b9d8872cf3a99da99e1fc4b0f71232a451632dc49ca4c6771549b4efa5c1c763"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 8, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1722, "query_tokens": 429, "message_tokens": 326, "message_utf8_bytes": 911, "cache_tensor_bytes": null, "serialized_payload_bytes": 911, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 121541.23483377043, "hops": {"1": 22612.123124999925, "2": 12922.917416901328, "3": 15539.13074999582, "4": 12866.83987500146, "5": 15885.767916915938, "6": 12968.57120806817, "7": 15879.040624946356, "8": 12866.843916941434}, "receiver_prefill_ms": 327.14862504508346, "ttft_ms": 17.74804200977087, "decode_ms": 19.080500002019107}, "total_wall_ms": 121905.71562573314, "ttft_ms": 17.74804200977087, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 58, "facts_retained": 15, "retention_rate": 0.25862068965517243, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.2222222222222222, "status": 0.75, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 911}, "ledger_by_hop": {"1": {"facts_total": 58, "facts_retained": 15, "retention_rate": 0.25862068965517243, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.2222222222222222, "status": 0.75, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 766}, "2": {"facts_total": 58, "facts_retained": 15, "retention_rate": 0.25862068965517243, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.2222222222222222, "status": 0.75, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 911}, "3": {"facts_total": 58, "facts_retained": 15, "retention_rate": 0.25862068965517243, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.2222222222222222, "status": 0.75, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 911}, "4": {"facts_total": 58, "facts_retained": 15, "retention_rate": 0.25862068965517243, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.2222222222222222, "status": 0.75, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 911}, "5": {"facts_total": 58, "facts_retained": 15, "retention_rate": 0.25862068965517243, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.2222222222222222, "status": 0.75, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 911}, "6": {"facts_total": 58, "facts_retained": 15, "retention_rate": 0.25862068965517243, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.2222222222222222, "status": 0.75, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 911}, "7": {"facts_total": 58, "facts_retained": 15, "retention_rate": 0.25862068965517243, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.2222222222222222, "status": 0.75, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 911}, "8": {"facts_total": 58, "facts_retained": 15, "retention_rate": 0.25862068965517243, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.2222222222222222, "status": 0.75, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 911}}}, "reference_key_id": "keys/synthetic/syn_03003_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "3d426d8999e6ba8503401b57d622bad0690314233e836579ad9658a82f039d3a", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "597bf3aff1efefe7e45aef0e", "attempt_id": "317ac684f9ac", "run_group_id": "E4_core:A5|closed|qwen3b-llama3b:h8", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T06:55:24.565592Z", "task_family": "synthetic", "document_id": "syn_03003_2000", "document_hash": "dddca1a38e3224870fcfae927c701c400fd885c19123ea840ee06a76a140f407", "question_id": "q04_multi_hop", "split": "test", "cluster_id": "syn_03003_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "tokenizer_hashes": {"receiver": "b9d8872cf3a99da99e1fc4b0f71232a451632dc49ca4c6771549b4efa5c1c763"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 8, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1722, "query_tokens": 438, "message_tokens": 326, "message_utf8_bytes": 911, "cache_tensor_bytes": null, "serialized_payload_bytes": 911, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 121541.23483377043, "hops": {"1": 22612.123124999925, "2": 12922.917416901328, "3": 15539.13074999582, "4": 12866.83987500146, "5": 15885.767916915938, "6": 12968.57120806817, "7": 15879.040624946356, "8": 12866.843916941434}, "receiver_prefill_ms": 360.29049998614937, "ttft_ms": 34.644666011445224, "decode_ms": 34.71929102670401}, "total_wall_ms": 121971.63904178888, "ttft_ms": 34.644666011445224, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 58, "facts_retained": 15, "retention_rate": 0.25862068965517243, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.2222222222222222, "status": 0.75, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 911}, "ledger_by_hop": {"1": {"facts_total": 58, "facts_retained": 15, "retention_rate": 0.25862068965517243, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.2222222222222222, "status": 0.75, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 766}, "2": {"facts_total": 58, "facts_retained": 15, "retention_rate": 0.25862068965517243, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.2222222222222222, "status": 0.75, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 911}, "3": {"facts_total": 58, "facts_retained": 15, "retention_rate": 0.25862068965517243, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.2222222222222222, "status": 0.75, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 911}, "4": {"facts_total": 58, "facts_retained": 15, "retention_rate": 0.25862068965517243, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.2222222222222222, "status": 0.75, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 911}, "5": {"facts_total": 58, "facts_retained": 15, "retention_rate": 0.25862068965517243, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.2222222222222222, "status": 0.75, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 911}, "6": {"facts_total": 58, "facts_retained": 15, "retention_rate": 0.25862068965517243, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.2222222222222222, "status": 0.75, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 911}, "7": {"facts_total": 58, "facts_retained": 15, "retention_rate": 0.25862068965517243, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.2222222222222222, "status": 0.75, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 911}, "8": {"facts_total": 58, "facts_retained": 15, "retention_rate": 0.25862068965517243, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.2222222222222222, "status": 0.75, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 911}}}, "reference_key_id": "keys/synthetic/syn_03003_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "3d426d8999e6ba8503401b57d622bad0690314233e836579ad9658a82f039d3a", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "52672b23457ce3200c6880fd", "attempt_id": "184aaf3a8b20", "run_group_id": "E4_core:A5|closed|qwen3b-llama3b:h8", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T06:55:25.015496Z", "task_family": "synthetic", "document_id": "syn_03003_2000", "document_hash": "dddca1a38e3224870fcfae927c701c400fd885c19123ea840ee06a76a140f407", "question_id": "q05_multi_hop", "split": "test", "cluster_id": "syn_03003_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "tokenizer_hashes": {"receiver": "b9d8872cf3a99da99e1fc4b0f71232a451632dc49ca4c6771549b4efa5c1c763"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 8, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1722, "query_tokens": 438, "message_tokens": 326, "message_utf8_bytes": 911, "cache_tensor_bytes": null, "serialized_payload_bytes": 911, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 121541.23483377043, "hops": {"1": 22612.123124999925, "2": 12922.917416901328, "3": 15539.13074999582, "4": 12866.83987500146, "5": 15885.767916915938, "6": 12968.57120806817, "7": 15879.040624946356, "8": 12866.843916941434}, "receiver_prefill_ms": 459.59424995817244, "ttft_ms": 31.303874915465713, "decode_ms": 31.149458955042064}, "total_wall_ms": 122063.68258385919, "ttft_ms": 31.303874915465713, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 58, "facts_retained": 15, "retention_rate": 0.25862068965517243, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.2222222222222222, "status": 0.75, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 911}, "ledger_by_hop": {"1": {"facts_total": 58, "facts_retained": 15, "retention_rate": 0.25862068965517243, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.2222222222222222, "status": 0.75, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 766}, "2": {"facts_total": 58, "facts_retained": 15, "retention_rate": 0.25862068965517243, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.2222222222222222, "status": 0.75, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 911}, "3": {"facts_total": 58, "facts_retained": 15, "retention_rate": 0.25862068965517243, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.2222222222222222, "status": 0.75, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 911}, "4": {"facts_total": 58, "facts_retained": 15, "retention_rate": 0.25862068965517243, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.2222222222222222, "status": 0.75, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 911}, "5": {"facts_total": 58, "facts_retained": 15, "retention_rate": 0.25862068965517243, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.2222222222222222, "status": 0.75, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 911}, "6": {"facts_total": 58, "facts_retained": 15, "retention_rate": 0.25862068965517243, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.2222222222222222, "status": 0.75, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 911}, "7": {"facts_total": 58, "facts_retained": 15, "retention_rate": 0.25862068965517243, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.2222222222222222, "status": 0.75, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 911}, "8": {"facts_total": 58, "facts_retained": 15, "retention_rate": 0.25862068965517243, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.2222222222222222, "status": 0.75, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 911}}}, "reference_key_id": "keys/synthetic/syn_03003_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "3d426d8999e6ba8503401b57d622bad0690314233e836579ad9658a82f039d3a", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "0af72ed4bd4b77700b9832e6", "attempt_id": "f194bf880892", "run_group_id": "E4_core:A5|closed|qwen3b-llama3b:h8", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T06:55:25.554699Z", "task_family": "synthetic", "document_id": "syn_03003_2000", "document_hash": "dddca1a38e3224870fcfae927c701c400fd885c19123ea840ee06a76a140f407", "question_id": "q06_negation", "split": "test", "cluster_id": "syn_03003_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "tokenizer_hashes": {"receiver": "b9d8872cf3a99da99e1fc4b0f71232a451632dc49ca4c6771549b4efa5c1c763"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 8, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1722, "query_tokens": 431, "message_tokens": 326, "message_utf8_bytes": 911, "cache_tensor_bytes": null, "serialized_payload_bytes": 911, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 121541.23483377043, "hops": {"1": 22612.123124999925, "2": 12922.917416901328, "3": 15539.13074999582, "4": 12866.83987500146, "5": 15885.767916915938, "6": 12968.57120806817, "7": 15879.040624946356, "8": 12866.843916941434}, "receiver_prefill_ms": 353.5785840358585, "ttft_ms": 18.793791066855192, "decode_ms": 54.74433291237801}, "total_wall_ms": 121968.95479166415, "ttft_ms": 18.793791066855192, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 2}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "No.", "evidence_artifact": {"ledger": {"facts_total": 58, "facts_retained": 15, "retention_rate": 0.25862068965517243, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.2222222222222222, "status": 0.75, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 911}, "ledger_by_hop": {"1": {"facts_total": 58, "facts_retained": 15, "retention_rate": 0.25862068965517243, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.2222222222222222, "status": 0.75, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 766}, "2": {"facts_total": 58, "facts_retained": 15, "retention_rate": 0.25862068965517243, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.2222222222222222, "status": 0.75, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 911}, "3": {"facts_total": 58, "facts_retained": 15, "retention_rate": 0.25862068965517243, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.2222222222222222, "status": 0.75, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 911}, "4": {"facts_total": 58, "facts_retained": 15, "retention_rate": 0.25862068965517243, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.2222222222222222, "status": 0.75, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 911}, "5": {"facts_total": 58, "facts_retained": 15, "retention_rate": 0.25862068965517243, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.2222222222222222, "status": 0.75, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 911}, "6": {"facts_total": 58, "facts_retained": 15, "retention_rate": 0.25862068965517243, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.2222222222222222, "status": 0.75, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 911}, "7": {"facts_total": 58, "facts_retained": 15, "retention_rate": 0.25862068965517243, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.2222222222222222, "status": 0.75, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 911}, "8": {"facts_total": 58, "facts_retained": 15, "retention_rate": 0.25862068965517243, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.2222222222222222, "status": 0.75, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 911}}}, "reference_key_id": "keys/synthetic/syn_03003_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "3d426d8999e6ba8503401b57d622bad0690314233e836579ad9658a82f039d3a", "output_hash": "38178a20b470cfd18299fb1593dd4ba706f1a83321dac051d01086efd8b7a96f"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "b507d719a0e8d26334a0019f", "attempt_id": "f88d55bb5a85", "run_group_id": "E4_core:A5|closed|qwen3b-llama3b:h8", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T06:55:25.999547Z", "task_family": "synthetic", "document_id": "syn_03003_2000", "document_hash": "dddca1a38e3224870fcfae927c701c400fd885c19123ea840ee06a76a140f407", "question_id": "q07_negation", "split": "test", "cluster_id": "syn_03003_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "tokenizer_hashes": {"receiver": "b9d8872cf3a99da99e1fc4b0f71232a451632dc49ca4c6771549b4efa5c1c763"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 8, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1722, "query_tokens": 429, "message_tokens": 326, "message_utf8_bytes": 911, "cache_tensor_bytes": null, "serialized_payload_bytes": 911, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 121541.23483377043, "hops": {"1": 22612.123124999925, "2": 12922.917416901328, "3": 15539.13074999582, "4": 12866.83987500146, "5": 15885.767916915938, "6": 12968.57120806817, "7": 15879.040624946356, "8": 12866.843916941434}, "receiver_prefill_ms": 313.43187496531755, "ttft_ms": 42.29816701263189, "decode_ms": 34.617792000062764}, "total_wall_ms": 121932.24162480328, "ttft_ms": 42.29816701263189, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "No", "evidence_artifact": {"ledger": {"facts_total": 58, "facts_retained": 15, "retention_rate": 0.25862068965517243, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.2222222222222222, "status": 0.75, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 911}, "ledger_by_hop": {"1": {"facts_total": 58, "facts_retained": 15, "retention_rate": 0.25862068965517243, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.2222222222222222, "status": 0.75, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 766}, "2": {"facts_total": 58, "facts_retained": 15, "retention_rate": 0.25862068965517243, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.2222222222222222, "status": 0.75, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 911}, "3": {"facts_total": 58, "facts_retained": 15, "retention_rate": 0.25862068965517243, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.2222222222222222, "status": 0.75, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 911}, "4": {"facts_total": 58, "facts_retained": 15, "retention_rate": 0.25862068965517243, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.2222222222222222, "status": 0.75, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 911}, "5": {"facts_total": 58, "facts_retained": 15, "retention_rate": 0.25862068965517243, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.2222222222222222, "status": 0.75, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 911}, "6": {"facts_total": 58, "facts_retained": 15, "retention_rate": 0.25862068965517243, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.2222222222222222, "status": 0.75, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 911}, "7": {"facts_total": 58, "facts_retained": 15, "retention_rate": 0.25862068965517243, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.2222222222222222, "status": 0.75, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 911}, "8": {"facts_total": 58, "facts_retained": 15, "retention_rate": 0.25862068965517243, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.2222222222222222, "status": 0.75, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 911}}}, "reference_key_id": "keys/synthetic/syn_03003_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "3d426d8999e6ba8503401b57d622bad0690314233e836579ad9658a82f039d3a", "output_hash": "1ea442a134b2a184bd5d40104401f2a37fbc09ccf3f4bc9da161c6099be3691d"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "e98541ffee60ddd99ca9c312", "attempt_id": "8381be52635f", "run_group_id": "E4_core:A5|closed|qwen3b-llama3b:h8", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T06:55:26.408281Z", "task_family": "synthetic", "document_id": "syn_03003_2000", "document_hash": "dddca1a38e3224870fcfae927c701c400fd885c19123ea840ee06a76a140f407", "question_id": "q08_revision", "split": "test", "cluster_id": "syn_03003_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "tokenizer_hashes": {"receiver": "b9d8872cf3a99da99e1fc4b0f71232a451632dc49ca4c6771549b4efa5c1c763"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 8, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1722, "query_tokens": 447, "message_tokens": 326, "message_utf8_bytes": 911, "cache_tensor_bytes": null, "serialized_payload_bytes": 911, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 121541.23483377043, "hops": {"1": 22612.123124999925, "2": 12922.917416901328, "3": 15539.13074999582, "4": 12866.83987500146, "5": 15885.767916915938, "6": 12968.57120806817, "7": 15879.040624946356, "8": 12866.843916941434}, "receiver_prefill_ms": 499.11479209549725, "ttft_ms": 28.28833309467882, "decode_ms": 108.20116708055139}, "total_wall_ms": 122177.39687475841, "ttft_ms": 28.28833309467882, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 3}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "911,925", "evidence_artifact": {"ledger": {"facts_total": 58, "facts_retained": 15, "retention_rate": 0.25862068965517243, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.2222222222222222, "status": 0.75, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 911}, "ledger_by_hop": {"1": {"facts_total": 58, "facts_retained": 15, "retention_rate": 0.25862068965517243, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.2222222222222222, "status": 0.75, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 766}, "2": {"facts_total": 58, "facts_retained": 15, "retention_rate": 0.25862068965517243, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.2222222222222222, "status": 0.75, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 911}, "3": {"facts_total": 58, "facts_retained": 15, "retention_rate": 0.25862068965517243, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.2222222222222222, "status": 0.75, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 911}, "4": {"facts_total": 58, "facts_retained": 15, "retention_rate": 0.25862068965517243, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.2222222222222222, "status": 0.75, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 911}, "5": {"facts_total": 58, "facts_retained": 15, "retention_rate": 0.25862068965517243, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.2222222222222222, "status": 0.75, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 911}, "6": {"facts_total": 58, "facts_retained": 15, "retention_rate": 0.25862068965517243, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.2222222222222222, "status": 0.75, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 911}, "7": {"facts_total": 58, "facts_retained": 15, "retention_rate": 0.25862068965517243, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.2222222222222222, "status": 0.75, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 911}, "8": {"facts_total": 58, "facts_retained": 15, "retention_rate": 0.25862068965517243, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.2222222222222222, "status": 0.75, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 911}}}, "reference_key_id": "keys/synthetic/syn_03003_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "3d426d8999e6ba8503401b57d622bad0690314233e836579ad9658a82f039d3a", "output_hash": "86be78720fc4b79f93e603e7d9e4861fcd25102612f73209fffc17d787056007"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "92121d6959e9e4cd86c2bbf8", "attempt_id": "50a500a0d254", "run_group_id": "E4_core:A5|closed|qwen3b-llama3b:h8", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T06:55:27.061829Z", "task_family": "synthetic", "document_id": "syn_03003_2000", "document_hash": "dddca1a38e3224870fcfae927c701c400fd885c19123ea840ee06a76a140f407", "question_id": "q09_revision", "split": "test", "cluster_id": "syn_03003_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "tokenizer_hashes": {"receiver": "b9d8872cf3a99da99e1fc4b0f71232a451632dc49ca4c6771549b4efa5c1c763"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 8, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1722, "query_tokens": 440, "message_tokens": 326, "message_utf8_bytes": 911, "cache_tensor_bytes": null, "serialized_payload_bytes": 911, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 121541.23483377043, "hops": {"1": 22612.123124999925, "2": 12922.917416901328, "3": 15539.13074999582, "4": 12866.83987500146, "5": 15885.767916915938, "6": 12968.57120806817, "7": 15879.040624946356, "8": 12866.843916941434}, "receiver_prefill_ms": 351.9434999907389, "ttft_ms": 22.162417066283524, "decode_ms": 101.87037498690188}, "total_wall_ms": 122017.7130837692, "ttft_ms": 22.162417066283524, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 3}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "911,925", "evidence_artifact": {"ledger": {"facts_total": 58, "facts_retained": 15, "retention_rate": 0.25862068965517243, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.2222222222222222, "status": 0.75, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 911}, "ledger_by_hop": {"1": {"facts_total": 58, "facts_retained": 15, "retention_rate": 0.25862068965517243, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.2222222222222222, "status": 0.75, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 766}, "2": {"facts_total": 58, "facts_retained": 15, "retention_rate": 0.25862068965517243, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.2222222222222222, "status": 0.75, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 911}, "3": {"facts_total": 58, "facts_retained": 15, "retention_rate": 0.25862068965517243, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.2222222222222222, "status": 0.75, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 911}, "4": {"facts_total": 58, "facts_retained": 15, "retention_rate": 0.25862068965517243, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.2222222222222222, "status": 0.75, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 911}, "5": {"facts_total": 58, "facts_retained": 15, "retention_rate": 0.25862068965517243, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.2222222222222222, "status": 0.75, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 911}, "6": {"facts_total": 58, "facts_retained": 15, "retention_rate": 0.25862068965517243, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.2222222222222222, "status": 0.75, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 911}, "7": {"facts_total": 58, "facts_retained": 15, "retention_rate": 0.25862068965517243, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.2222222222222222, "status": 0.75, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 911}, "8": {"facts_total": 58, "facts_retained": 15, "retention_rate": 0.25862068965517243, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.2222222222222222, "status": 0.75, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 911}}}, "reference_key_id": "keys/synthetic/syn_03003_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "3d426d8999e6ba8503401b57d622bad0690314233e836579ad9658a82f039d3a", "output_hash": "86be78720fc4b79f93e603e7d9e4861fcd25102612f73209fffc17d787056007"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "07c93e1252410aba7e5d8e7b", "attempt_id": "fc462dde05a1", "run_group_id": "E4_core:A5|closed|qwen3b-llama3b:h8", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T06:55:27.555638Z", "task_family": "synthetic", "document_id": "syn_03003_2000", "document_hash": "dddca1a38e3224870fcfae927c701c400fd885c19123ea840ee06a76a140f407", "question_id": "q10_temporal", "split": "test", "cluster_id": "syn_03003_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "tokenizer_hashes": {"receiver": "b9d8872cf3a99da99e1fc4b0f71232a451632dc49ca4c6771549b4efa5c1c763"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 8, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1722, "query_tokens": 447, "message_tokens": 326, "message_utf8_bytes": 911, "cache_tensor_bytes": null, "serialized_payload_bytes": 911, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 121541.23483377043, "hops": {"1": 22612.123124999925, "2": 12922.917416901328, "3": 15539.13074999582, "4": 12866.83987500146, "5": 15885.767916915938, "6": 12968.57120806817, "7": 15879.040624946356, "8": 12866.843916941434}, "receiver_prefill_ms": 307.36133304890245, "ttft_ms": 42.84195799846202, "decode_ms": 65.40741701610386}, "total_wall_ms": 121957.37016678322, "ttft_ms": 42.84195799846202, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 58, "facts_retained": 15, "retention_rate": 0.25862068965517243, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.2222222222222222, "status": 0.75, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 911}, "ledger_by_hop": {"1": {"facts_total": 58, "facts_retained": 15, "retention_rate": 0.25862068965517243, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.2222222222222222, "status": 0.75, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 766}, "2": {"facts_total": 58, "facts_retained": 15, "retention_rate": 0.25862068965517243, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.2222222222222222, "status": 0.75, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 911}, "3": {"facts_total": 58, "facts_retained": 15, "retention_rate": 0.25862068965517243, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.2222222222222222, "status": 0.75, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 911}, "4": {"facts_total": 58, "facts_retained": 15, "retention_rate": 0.25862068965517243, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.2222222222222222, "status": 0.75, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 911}, "5": {"facts_total": 58, "facts_retained": 15, "retention_rate": 0.25862068965517243, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.2222222222222222, "status": 0.75, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 911}, "6": {"facts_total": 58, "facts_retained": 15, "retention_rate": 0.25862068965517243, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.2222222222222222, "status": 0.75, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 911}, "7": {"facts_total": 58, "facts_retained": 15, "retention_rate": 0.25862068965517243, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.2222222222222222, "status": 0.75, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 911}, "8": {"facts_total": 58, "facts_retained": 15, "retention_rate": 0.25862068965517243, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.2222222222222222, "status": 0.75, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 911}}}, "reference_key_id": "keys/synthetic/syn_03003_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "3d426d8999e6ba8503401b57d622bad0690314233e836579ad9658a82f039d3a", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "0d492f2d5a7d2c197b693a26", "attempt_id": "e5da8f6f6aa0", "run_group_id": "E4_core:A5|closed|qwen3b-llama3b:h8", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T06:55:27.989722Z", "task_family": "synthetic", "document_id": "syn_03003_2000", "document_hash": "dddca1a38e3224870fcfae927c701c400fd885c19123ea840ee06a76a140f407", "question_id": "q11_aggregation", "split": "test", "cluster_id": "syn_03003_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "tokenizer_hashes": {"receiver": "b9d8872cf3a99da99e1fc4b0f71232a451632dc49ca4c6771549b4efa5c1c763"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 8, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1722, "query_tokens": 441, "message_tokens": 326, "message_utf8_bytes": 911, "cache_tensor_bytes": null, "serialized_payload_bytes": 911, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 121541.23483377043, "hops": {"1": 22612.123124999925, "2": 12922.917416901328, "3": 15539.13074999582, "4": 12866.83987500146, "5": 15885.767916915938, "6": 12968.57120806817, "7": 15879.040624946356, "8": 12866.843916941434}, "receiver_prefill_ms": 482.27137501817197, "ttft_ms": 31.87395795248449, "decode_ms": 32.26204100064933}, "total_wall_ms": 122088.3179177763, "ttft_ms": 31.87395795248449, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 58, "facts_retained": 15, "retention_rate": 0.25862068965517243, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.2222222222222222, "status": 0.75, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 911}, "ledger_by_hop": {"1": {"facts_total": 58, "facts_retained": 15, "retention_rate": 0.25862068965517243, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.2222222222222222, "status": 0.75, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 766}, "2": {"facts_total": 58, "facts_retained": 15, "retention_rate": 0.25862068965517243, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.2222222222222222, "status": 0.75, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 911}, "3": {"facts_total": 58, "facts_retained": 15, "retention_rate": 0.25862068965517243, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.2222222222222222, "status": 0.75, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 911}, "4": {"facts_total": 58, "facts_retained": 15, "retention_rate": 0.25862068965517243, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.2222222222222222, "status": 0.75, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 911}, "5": {"facts_total": 58, "facts_retained": 15, "retention_rate": 0.25862068965517243, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.2222222222222222, "status": 0.75, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 911}, "6": {"facts_total": 58, "facts_retained": 15, "retention_rate": 0.25862068965517243, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.2222222222222222, "status": 0.75, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 911}, "7": {"facts_total": 58, "facts_retained": 15, "retention_rate": 0.25862068965517243, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.2222222222222222, "status": 0.75, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 911}, "8": {"facts_total": 58, "facts_retained": 15, "retention_rate": 0.25862068965517243, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.2222222222222222, "status": 0.75, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 911}}}, "reference_key_id": "keys/synthetic/syn_03003_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "3d426d8999e6ba8503401b57d622bad0690314233e836579ad9658a82f039d3a", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "2bd267fb1eba96e615a700d8", "attempt_id": "b3bf26c865ad", "run_group_id": "E4_core:A5|closed|qwen3b-llama3b:h8", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T06:55:28.553772Z", "task_family": "synthetic", "document_id": "syn_03003_2000", "document_hash": "dddca1a38e3224870fcfae927c701c400fd885c19123ea840ee06a76a140f407", "question_id": "q12_unit_conversion", "split": "test", "cluster_id": "syn_03003_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "tokenizer_hashes": {"receiver": "b9d8872cf3a99da99e1fc4b0f71232a451632dc49ca4c6771549b4efa5c1c763"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 8, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1722, "query_tokens": 438, "message_tokens": 326, "message_utf8_bytes": 911, "cache_tensor_bytes": null, "serialized_payload_bytes": 911, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 121541.23483377043, "hops": {"1": 22612.123124999925, "2": 12922.917416901328, "3": 15539.13074999582, "4": 12866.83987500146, "5": 15885.767916915938, "6": 12968.57120806817, "7": 15879.040624946356, "8": 12866.843916941434}, "receiver_prefill_ms": 332.10595906712115, "ttft_ms": 40.48479092307389, "decode_ms": 87.63316704425961}, "total_wall_ms": 122001.97362573817, "ttft_ms": 40.48479092307389, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 3}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "17.43", "evidence_artifact": {"ledger": {"facts_total": 58, "facts_retained": 15, "retention_rate": 0.25862068965517243, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.2222222222222222, "status": 0.75, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 911}, "ledger_by_hop": {"1": {"facts_total": 58, "facts_retained": 15, "retention_rate": 0.25862068965517243, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.2222222222222222, "status": 0.75, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 766}, "2": {"facts_total": 58, "facts_retained": 15, "retention_rate": 0.25862068965517243, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.2222222222222222, "status": 0.75, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 911}, "3": {"facts_total": 58, "facts_retained": 15, "retention_rate": 0.25862068965517243, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.2222222222222222, "status": 0.75, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 911}, "4": {"facts_total": 58, "facts_retained": 15, "retention_rate": 0.25862068965517243, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.2222222222222222, "status": 0.75, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 911}, "5": {"facts_total": 58, "facts_retained": 15, "retention_rate": 0.25862068965517243, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.2222222222222222, "status": 0.75, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 911}, "6": {"facts_total": 58, "facts_retained": 15, "retention_rate": 0.25862068965517243, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.2222222222222222, "status": 0.75, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 911}, "7": {"facts_total": 58, "facts_retained": 15, "retention_rate": 0.25862068965517243, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.2222222222222222, "status": 0.75, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 911}, "8": {"facts_total": 58, "facts_retained": 15, "retention_rate": 0.25862068965517243, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.2222222222222222, "status": 0.75, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 911}}}, "reference_key_id": "keys/synthetic/syn_03003_2000.json", "task_scores": {"correct": 0.0, "abstained": 0.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "3d426d8999e6ba8503401b57d622bad0690314233e836579ad9658a82f039d3a", "output_hash": "1cd677e6c296f31fca744a7c4d96b5514fa6b036913240b8d689f1e277900d17"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "0027ebe627bada9418361cbd", "attempt_id": "7f824ea62508", "run_group_id": "E4_core:A5|closed|qwen3b-llama3b:h8", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T06:55:29.032569Z", "task_family": "synthetic", "document_id": "syn_03003_2000", "document_hash": "dddca1a38e3224870fcfae927c701c400fd885c19123ea840ee06a76a140f407", "question_id": "q13_unanswerable", "split": "test", "cluster_id": "syn_03003_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "tokenizer_hashes": {"receiver": "b9d8872cf3a99da99e1fc4b0f71232a451632dc49ca4c6771549b4efa5c1c763"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 8, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1722, "query_tokens": 442, "message_tokens": 326, "message_utf8_bytes": 911, "cache_tensor_bytes": null, "serialized_payload_bytes": 911, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 121541.23483377043, "hops": {"1": 22612.123124999925, "2": 12922.917416901328, "3": 15539.13074999582, "4": 12866.83987500146, "5": 15885.767916915938, "6": 12968.57120806817, "7": 15879.040624946356, "8": 12866.843916941434}, "receiver_prefill_ms": 328.67825007997453, "ttft_ms": 21.478667040355504, "decode_ms": 61.179750016890466}, "total_wall_ms": 121953.25316779781, "ttft_ms": 21.478667040355504, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 58, "facts_retained": 15, "retention_rate": 0.25862068965517243, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.2222222222222222, "status": 0.75, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 911}, "ledger_by_hop": {"1": {"facts_total": 58, "facts_retained": 15, "retention_rate": 0.25862068965517243, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.2222222222222222, "status": 0.75, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 766}, "2": {"facts_total": 58, "facts_retained": 15, "retention_rate": 0.25862068965517243, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.2222222222222222, "status": 0.75, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 911}, "3": {"facts_total": 58, "facts_retained": 15, "retention_rate": 0.25862068965517243, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.2222222222222222, "status": 0.75, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 911}, "4": {"facts_total": 58, "facts_retained": 15, "retention_rate": 0.25862068965517243, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.2222222222222222, "status": 0.75, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 911}, "5": {"facts_total": 58, "facts_retained": 15, "retention_rate": 0.25862068965517243, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.2222222222222222, "status": 0.75, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 911}, "6": {"facts_total": 58, "facts_retained": 15, "retention_rate": 0.25862068965517243, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.2222222222222222, "status": 0.75, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 911}, "7": {"facts_total": 58, "facts_retained": 15, "retention_rate": 0.25862068965517243, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.2222222222222222, "status": 0.75, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 911}, "8": {"facts_total": 58, "facts_retained": 15, "retention_rate": 0.25862068965517243, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.2222222222222222, "status": 0.75, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 911}}}, "reference_key_id": "keys/synthetic/syn_03003_2000.json", "task_scores": {"correct": 1.0, "abstained": 1.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "3d426d8999e6ba8503401b57d622bad0690314233e836579ad9658a82f039d3a", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "a4f17726b7ebb84abd56b3e7", "attempt_id": "0eb6a8eb350d", "run_group_id": "E4_core:A5|closed|qwen3b-llama3b:h8", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T06:55:29.462249Z", "task_family": "synthetic", "document_id": "syn_03003_2000", "document_hash": "dddca1a38e3224870fcfae927c701c400fd885c19123ea840ee06a76a140f407", "question_id": "q14_unanswerable", "split": "test", "cluster_id": "syn_03003_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "tokenizer_hashes": {"receiver": "b9d8872cf3a99da99e1fc4b0f71232a451632dc49ca4c6771549b4efa5c1c763"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 8, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1722, "query_tokens": 443, "message_tokens": 326, "message_utf8_bytes": 911, "cache_tensor_bytes": null, "serialized_payload_bytes": 911, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 121541.23483377043, "hops": {"1": 22612.123124999925, "2": 12922.917416901328, "3": 15539.13074999582, "4": 12866.83987500146, "5": 15885.767916915938, "6": 12968.57120806817, "7": 15879.040624946356, "8": 12866.843916941434}, "receiver_prefill_ms": 319.33241698425263, "ttft_ms": 113.92791604157537, "decode_ms": 137.5934169627726}, "total_wall_ms": 122112.60466685053, "ttft_ms": 113.92791604157537, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 58, "facts_retained": 15, "retention_rate": 0.25862068965517243, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.2222222222222222, "status": 0.75, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 911}, "ledger_by_hop": {"1": {"facts_total": 58, "facts_retained": 15, "retention_rate": 0.25862068965517243, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.2222222222222222, "status": 0.75, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 766}, "2": {"facts_total": 58, "facts_retained": 15, "retention_rate": 0.25862068965517243, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.2222222222222222, "status": 0.75, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 911}, "3": {"facts_total": 58, "facts_retained": 15, "retention_rate": 0.25862068965517243, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.2222222222222222, "status": 0.75, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 911}, "4": {"facts_total": 58, "facts_retained": 15, "retention_rate": 0.25862068965517243, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.2222222222222222, "status": 0.75, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 911}, "5": {"facts_total": 58, "facts_retained": 15, "retention_rate": 0.25862068965517243, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.2222222222222222, "status": 0.75, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 911}, "6": {"facts_total": 58, "facts_retained": 15, "retention_rate": 0.25862068965517243, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.2222222222222222, "status": 0.75, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 911}, "7": {"facts_total": 58, "facts_retained": 15, "retention_rate": 0.25862068965517243, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.2222222222222222, "status": 0.75, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 911}, "8": {"facts_total": 58, "facts_retained": 15, "retention_rate": 0.25862068965517243, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.2222222222222222, "status": 0.75, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 911}}}, "reference_key_id": "keys/synthetic/syn_03003_2000.json", "task_scores": {"correct": 1.0, "abstained": 1.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "3d426d8999e6ba8503401b57d622bad0690314233e836579ad9658a82f039d3a", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "ed7ad96961fd9dd3b4a0ab4c", "attempt_id": "b53b63b873e2", "run_group_id": "E4_core:A4|closed|qwen3b:h1", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T06:55:52.738444Z", "task_family": "synthetic", "document_id": "syn_03006_2000", "document_hash": "89be5e77e0ca114963e9ffdf8dc38fe5610f13a78ac29e7a1956cd9124ce5217", "question_id": "q00_single_fact", "split": "test", "cluster_id": "syn_03006_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 1, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1890, "query_tokens": 595, "message_tokens": 512, "message_utf8_bytes": 1087, "cache_tensor_bytes": null, "serialized_payload_bytes": 1087, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 22603.971540927887, "hops": {"1": 22603.971540927887}, "receiver_prefill_ms": 367.6694589667022, "ttft_ms": 42.53504110965878, "decode_ms": 609.6354579785839}, "total_wall_ms": 23624.763123923913, "ttft_ms": 42.53504110965878, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 10}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "XM-76Q-2958", "evidence_artifact": {"ledger": {"facts_total": 64, "facts_retained": 30, "retention_rate": 0.46875, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.6666666666666666, "status": 0.9166666666666666, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1087}, "ledger_by_hop": {"1": {"facts_total": 64, "facts_retained": 30, "retention_rate": 0.46875, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.6666666666666666, "status": 0.9166666666666666, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1087}}}, "reference_key_id": "keys/synthetic/syn_03006_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "47f843b1989b354c7cbe406d456e62c59b6229490f55dde77bcdb5f70147ff28", "output_hash": "cf42691eec772b9ec74d67643944de2e19e53724d7b772cef1e0f2d5ef3b4926"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "8cc9569a8dfd25bb1598e905", "attempt_id": "295e2f7b1b45", "run_group_id": "E4_core:A4|closed|qwen3b:h1", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T06:55:53.777469Z", "task_family": "synthetic", "document_id": "syn_03006_2000", "document_hash": "89be5e77e0ca114963e9ffdf8dc38fe5610f13a78ac29e7a1956cd9124ce5217", "question_id": "q01_single_fact", "split": "test", "cluster_id": "syn_03006_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 1, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1890, "query_tokens": 600, "message_tokens": 512, "message_utf8_bytes": 1087, "cache_tensor_bytes": null, "serialized_payload_bytes": 1087, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 22603.971540927887, "hops": {"1": 22603.971540927887}, "receiver_prefill_ms": 408.1520000472665, "ttft_ms": 21.074417047202587, "decode_ms": 42.71574993617833}, "total_wall_ms": 23076.541916001588, "ttft_ms": 21.074417047202587, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 64, "facts_retained": 30, "retention_rate": 0.46875, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.6666666666666666, "status": 0.9166666666666666, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1087}, "ledger_by_hop": {"1": {"facts_total": 64, "facts_retained": 30, "retention_rate": 0.46875, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.6666666666666666, "status": 0.9166666666666666, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1087}}}, "reference_key_id": "keys/synthetic/syn_03006_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "47f843b1989b354c7cbe406d456e62c59b6229490f55dde77bcdb5f70147ff28", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "c0f16944b0b3fab81d2317b1", "attempt_id": "e2c0c7f84a6f", "run_group_id": "E4_core:A4|closed|qwen3b:h1", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T06:55:54.267540Z", "task_family": "synthetic", "document_id": "syn_03006_2000", "document_hash": "89be5e77e0ca114963e9ffdf8dc38fe5610f13a78ac29e7a1956cd9124ce5217", "question_id": "q02_exact_copy", "split": "test", "cluster_id": "syn_03006_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 1, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1890, "query_tokens": 599, "message_tokens": 512, "message_utf8_bytes": 1087, "cache_tensor_bytes": null, "serialized_payload_bytes": 1087, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 22603.971540927887, "hops": {"1": 22603.971540927887}, "receiver_prefill_ms": 383.44970799516886, "ttft_ms": 20.905124954879284, "decode_ms": 61.329458956606686}, "total_wall_ms": 23070.286415866576, "ttft_ms": 20.905124954879284, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 64, "facts_retained": 30, "retention_rate": 0.46875, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.6666666666666666, "status": 0.9166666666666666, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1087}, "ledger_by_hop": {"1": {"facts_total": 64, "facts_retained": 30, "retention_rate": 0.46875, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.6666666666666666, "status": 0.9166666666666666, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1087}}}, "reference_key_id": "keys/synthetic/syn_03006_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "47f843b1989b354c7cbe406d456e62c59b6229490f55dde77bcdb5f70147ff28", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "4e817e7695a83e90b04079db", "attempt_id": "b66e8b06e432", "run_group_id": "E4_core:A4|closed|qwen3b:h1", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T06:55:54.750898Z", "task_family": "synthetic", "document_id": "syn_03006_2000", "document_hash": "89be5e77e0ca114963e9ffdf8dc38fe5610f13a78ac29e7a1956cd9124ce5217", "question_id": "q03_exact_copy", "split": "test", "cluster_id": "syn_03006_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 1, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1890, "query_tokens": 594, "message_tokens": 512, "message_utf8_bytes": 1087, "cache_tensor_bytes": null, "serialized_payload_bytes": 1087, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 22603.971540927887, "hops": {"1": 22603.971540927887}, "receiver_prefill_ms": 395.006624981761, "ttft_ms": 93.08320796117187, "decode_ms": 382.6257090549916}, "total_wall_ms": 23475.37204087712, "ttft_ms": 93.08320796117187, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 8}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "Haruki Ouellet-Marchand", "evidence_artifact": {"ledger": {"facts_total": 64, "facts_retained": 30, "retention_rate": 0.46875, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.6666666666666666, "status": 0.9166666666666666, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1087}, "ledger_by_hop": {"1": {"facts_total": 64, "facts_retained": 30, "retention_rate": 0.46875, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.6666666666666666, "status": 0.9166666666666666, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1087}}}, "reference_key_id": "keys/synthetic/syn_03006_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "47f843b1989b354c7cbe406d456e62c59b6229490f55dde77bcdb5f70147ff28", "output_hash": "a3b6e0a44d495387d8b8fc50ba6be0f2ed43121c674d3b3e962f8c4690e15b5c"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "fd8b0e09393f43b4a47985b2", "attempt_id": "84d9535cdc42", "run_group_id": "E4_core:A4|closed|qwen3b:h1", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T06:55:55.639646Z", "task_family": "synthetic", "document_id": "syn_03006_2000", "document_hash": "89be5e77e0ca114963e9ffdf8dc38fe5610f13a78ac29e7a1956cd9124ce5217", "question_id": "q04_multi_hop", "split": "test", "cluster_id": "syn_03006_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 1, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1890, "query_tokens": 603, "message_tokens": 512, "message_utf8_bytes": 1087, "cache_tensor_bytes": null, "serialized_payload_bytes": 1087, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 22603.971540927887, "hops": {"1": 22603.971540927887}, "receiver_prefill_ms": 402.7612500358373, "ttft_ms": 27.684833970852196, "decode_ms": 35.1231669774279}, "total_wall_ms": 23070.129749947228, "ttft_ms": 27.684833970852196, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 64, "facts_retained": 30, "retention_rate": 0.46875, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.6666666666666666, "status": 0.9166666666666666, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1087}, "ledger_by_hop": {"1": {"facts_total": 64, "facts_retained": 30, "retention_rate": 0.46875, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.6666666666666666, "status": 0.9166666666666666, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1087}}}, "reference_key_id": "keys/synthetic/syn_03006_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "47f843b1989b354c7cbe406d456e62c59b6229490f55dde77bcdb5f70147ff28", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "06eddbdb4e2cf65666f629d2", "attempt_id": "9c602a6ea63b", "run_group_id": "E4_core:A4|closed|qwen3b:h1", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T06:55:56.122638Z", "task_family": "synthetic", "document_id": "syn_03006_2000", "document_hash": "89be5e77e0ca114963e9ffdf8dc38fe5610f13a78ac29e7a1956cd9124ce5217", "question_id": "q05_multi_hop", "split": "test", "cluster_id": "syn_03006_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 1, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1890, "query_tokens": 606, "message_tokens": 512, "message_utf8_bytes": 1087, "cache_tensor_bytes": null, "serialized_payload_bytes": 1087, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 22603.971540927887, "hops": {"1": 22603.971540927887}, "receiver_prefill_ms": 357.2169999824837, "ttft_ms": 53.266583010554314, "decode_ms": 569.8349580634385}, "total_wall_ms": 23585.017374833114, "ttft_ms": 53.266583010554314, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 10}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "EP-52G-1330", "evidence_artifact": {"ledger": {"facts_total": 64, "facts_retained": 30, "retention_rate": 0.46875, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.6666666666666666, "status": 0.9166666666666666, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1087}, "ledger_by_hop": {"1": {"facts_total": 64, "facts_retained": 30, "retention_rate": 0.46875, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.6666666666666666, "status": 0.9166666666666666, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1087}}}, "reference_key_id": "keys/synthetic/syn_03006_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "47f843b1989b354c7cbe406d456e62c59b6229490f55dde77bcdb5f70147ff28", "output_hash": "104db207937fd79e23ce16f9b3f10b1a9ed34f6dd0441a99be6ea1b2d0a09d99"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "826bde18083a7918770d059e", "attempt_id": "8f72ebdedf01", "run_group_id": "E4_core:A4|closed|qwen3b:h1", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T06:55:57.120585Z", "task_family": "synthetic", "document_id": "syn_03006_2000", "document_hash": "89be5e77e0ca114963e9ffdf8dc38fe5610f13a78ac29e7a1956cd9124ce5217", "question_id": "q06_negation", "split": "test", "cluster_id": "syn_03006_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 1, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1890, "query_tokens": 594, "message_tokens": 512, "message_utf8_bytes": 1087, "cache_tensor_bytes": null, "serialized_payload_bytes": 1087, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 22603.971540927887, "hops": {"1": 22603.971540927887}, "receiver_prefill_ms": 395.8598750177771, "ttft_ms": 42.91941598057747, "decode_ms": 24.590082932263613}, "total_wall_ms": 23067.98174895812, "ttft_ms": 42.91941598057747, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "no", "evidence_artifact": {"ledger": {"facts_total": 64, "facts_retained": 30, "retention_rate": 0.46875, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.6666666666666666, "status": 0.9166666666666666, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1087}, "ledger_by_hop": {"1": {"facts_total": 64, "facts_retained": 30, "retention_rate": 0.46875, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.6666666666666666, "status": 0.9166666666666666, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1087}}}, "reference_key_id": "keys/synthetic/syn_03006_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "47f843b1989b354c7cbe406d456e62c59b6229490f55dde77bcdb5f70147ff28", "output_hash": "9390298f3fb0c5b160498935d79cb139aef28e1c47358b4bbba61862b9c26e59"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "b8ac9f78bfddf782ea44a485", "attempt_id": "843a968a135f", "run_group_id": "E4_core:A4|closed|qwen3b:h1", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T06:55:57.601331Z", "task_family": "synthetic", "document_id": "syn_03006_2000", "document_hash": "89be5e77e0ca114963e9ffdf8dc38fe5610f13a78ac29e7a1956cd9124ce5217", "question_id": "q07_negation", "split": "test", "cluster_id": "syn_03006_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 1, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1890, "query_tokens": 594, "message_tokens": 512, "message_utf8_bytes": 1087, "cache_tensor_bytes": null, "serialized_payload_bytes": 1087, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 22603.971540927887, "hops": {"1": 22603.971540927887}, "receiver_prefill_ms": 379.8569160280749, "ttft_ms": 21.001209039241076, "decode_ms": 60.80633297096938}, "total_wall_ms": 23066.435165936127, "ttft_ms": 21.001209039241076, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "no", "evidence_artifact": {"ledger": {"facts_total": 64, "facts_retained": 30, "retention_rate": 0.46875, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.6666666666666666, "status": 0.9166666666666666, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1087}, "ledger_by_hop": {"1": {"facts_total": 64, "facts_retained": 30, "retention_rate": 0.46875, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.6666666666666666, "status": 0.9166666666666666, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1087}}}, "reference_key_id": "keys/synthetic/syn_03006_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "47f843b1989b354c7cbe406d456e62c59b6229490f55dde77bcdb5f70147ff28", "output_hash": "9390298f3fb0c5b160498935d79cb139aef28e1c47358b4bbba61862b9c26e59"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "e9da25246bc4b804b91f7ed2", "attempt_id": "ac39709fc9b7", "run_group_id": "E4_core:A4|closed|qwen3b:h1", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T06:55:58.080987Z", "task_family": "synthetic", "document_id": "syn_03006_2000", "document_hash": "89be5e77e0ca114963e9ffdf8dc38fe5610f13a78ac29e7a1956cd9124ce5217", "question_id": "q08_revision", "split": "test", "cluster_id": "syn_03006_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 1, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1890, "query_tokens": 611, "message_tokens": 512, "message_utf8_bytes": 1087, "cache_tensor_bytes": null, "serialized_payload_bytes": 1087, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 22603.971540927887, "hops": {"1": 22603.971540927887}, "receiver_prefill_ms": 399.3899580091238, "ttft_ms": 91.8541670544073, "decode_ms": 349.97099998872727}, "total_wall_ms": 23445.931331953034, "ttft_ms": 91.8541670544073, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 7}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "9112735", "evidence_artifact": {"ledger": {"facts_total": 64, "facts_retained": 30, "retention_rate": 0.46875, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.6666666666666666, "status": 0.9166666666666666, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1087}, "ledger_by_hop": {"1": {"facts_total": 64, "facts_retained": 30, "retention_rate": 0.46875, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.6666666666666666, "status": 0.9166666666666666, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1087}}}, "reference_key_id": "keys/synthetic/syn_03006_2000.json", "task_scores": {"correct": 0.0, "abstained": 0.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "47f843b1989b354c7cbe406d456e62c59b6229490f55dde77bcdb5f70147ff28", "output_hash": "15a3f804636f0e62d4a3d9be7677f001a0ad6b1fb8ae7e8cdc63c44ec7bac31f"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "dfa1ae53f4da34b3412385dc", "attempt_id": "51da567275f7", "run_group_id": "E4_core:A4|closed|qwen3b:h1", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T06:55:58.940410Z", "task_family": "synthetic", "document_id": "syn_03006_2000", "document_hash": "89be5e77e0ca114963e9ffdf8dc38fe5610f13a78ac29e7a1956cd9124ce5217", "question_id": "q09_revision", "split": "test", "cluster_id": "syn_03006_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 1, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1890, "query_tokens": 604, "message_tokens": 512, "message_utf8_bytes": 1087, "cache_tensor_bytes": null, "serialized_payload_bytes": 1087, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 22603.971540927887, "hops": {"1": 22603.971540927887}, "receiver_prefill_ms": 411.48316697217524, "ttft_ms": 23.080167011357844, "decode_ms": 210.49112500622869}, "total_wall_ms": 23249.63574996218, "ttft_ms": 23.080167011357844, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 7}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "9112735", "evidence_artifact": {"ledger": {"facts_total": 64, "facts_retained": 30, "retention_rate": 0.46875, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.6666666666666666, "status": 0.9166666666666666, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1087}, "ledger_by_hop": {"1": {"facts_total": 64, "facts_retained": 30, "retention_rate": 0.46875, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.6666666666666666, "status": 0.9166666666666666, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1087}}}, "reference_key_id": "keys/synthetic/syn_03006_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "47f843b1989b354c7cbe406d456e62c59b6229490f55dde77bcdb5f70147ff28", "output_hash": "15a3f804636f0e62d4a3d9be7677f001a0ad6b1fb8ae7e8cdc63c44ec7bac31f"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "22f1cf53543d3b5f6852a255", "attempt_id": "1b28184b44fa", "run_group_id": "E4_core:A4|closed|qwen3b:h1", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T06:55:59.603735Z", "task_family": "synthetic", "document_id": "syn_03006_2000", "document_hash": "89be5e77e0ca114963e9ffdf8dc38fe5610f13a78ac29e7a1956cd9124ce5217", "question_id": "q10_temporal", "split": "test", "cluster_id": "syn_03006_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 1, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1890, "query_tokens": 613, "message_tokens": 512, "message_utf8_bytes": 1087, "cache_tensor_bytes": null, "serialized_payload_bytes": 1087, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 22603.971540927887, "hops": {"1": 22603.971540927887}, "receiver_prefill_ms": 389.45558306295425, "ttft_ms": 62.649291940033436, "decode_ms": 101.47562494967133}, "total_wall_ms": 23158.396290964447, "ttft_ms": 62.649291940033436, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 64, "facts_retained": 30, "retention_rate": 0.46875, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.6666666666666666, "status": 0.9166666666666666, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1087}, "ledger_by_hop": {"1": {"facts_total": 64, "facts_retained": 30, "retention_rate": 0.46875, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.6666666666666666, "status": 0.9166666666666666, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1087}}}, "reference_key_id": "keys/synthetic/syn_03006_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "47f843b1989b354c7cbe406d456e62c59b6229490f55dde77bcdb5f70147ff28", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "3234e127c579c3643dba69ae", "attempt_id": "9796c5b4f175", "run_group_id": "E4_core:A4|closed|qwen3b:h1", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T06:56:00.177590Z", "task_family": "synthetic", "document_id": "syn_03006_2000", "document_hash": "89be5e77e0ca114963e9ffdf8dc38fe5610f13a78ac29e7a1956cd9124ce5217", "question_id": "q11_aggregation", "split": "test", "cluster_id": "syn_03006_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 1, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1890, "query_tokens": 606, "message_tokens": 512, "message_utf8_bytes": 1087, "cache_tensor_bytes": null, "serialized_payload_bytes": 1087, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 22603.971540927887, "hops": {"1": 22603.971540927887}, "receiver_prefill_ms": 488.2397080073133, "ttft_ms": 33.443208085373044, "decode_ms": 224.89270905498415}, "total_wall_ms": 23351.21062491089, "ttft_ms": 33.443208085373044, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 7}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "8908553", "evidence_artifact": {"ledger": {"facts_total": 64, "facts_retained": 30, "retention_rate": 0.46875, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.6666666666666666, "status": 0.9166666666666666, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1087}, "ledger_by_hop": {"1": {"facts_total": 64, "facts_retained": 30, "retention_rate": 0.46875, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.6666666666666666, "status": 0.9166666666666666, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1087}}}, "reference_key_id": "keys/synthetic/syn_03006_2000.json", "task_scores": {"correct": 0.0, "abstained": 0.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "47f843b1989b354c7cbe406d456e62c59b6229490f55dde77bcdb5f70147ff28", "output_hash": "5ad9e777dc3be36808c74badfc8aae217b41a486754d07adf1abda6f6bce9aca"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "3a24526554a5bfbb2c91e37b", "attempt_id": "ef4a4b1853b5", "run_group_id": "E4_core:A4|closed|qwen3b:h1", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T06:56:00.944431Z", "task_family": "synthetic", "document_id": "syn_03006_2000", "document_hash": "89be5e77e0ca114963e9ffdf8dc38fe5610f13a78ac29e7a1956cd9124ce5217", "question_id": "q12_unit_conversion", "split": "test", "cluster_id": "syn_03006_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 1, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1890, "query_tokens": 603, "message_tokens": 512, "message_utf8_bytes": 1087, "cache_tensor_bytes": null, "serialized_payload_bytes": 1087, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 22603.971540927887, "hops": {"1": 22603.971540927887}, "receiver_prefill_ms": 378.4396250266582, "ttft_ms": 14.01204103603959, "decode_ms": 208.8222079910338}, "total_wall_ms": 23206.001540995203, "ttft_ms": 14.01204103603959, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 5}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "39730", "evidence_artifact": {"ledger": {"facts_total": 64, "facts_retained": 30, "retention_rate": 0.46875, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.6666666666666666, "status": 0.9166666666666666, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1087}, "ledger_by_hop": {"1": {"facts_total": 64, "facts_retained": 30, "retention_rate": 0.46875, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.6666666666666666, "status": 0.9166666666666666, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1087}}}, "reference_key_id": "keys/synthetic/syn_03006_2000.json", "task_scores": {"correct": 0.0, "abstained": 0.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "47f843b1989b354c7cbe406d456e62c59b6229490f55dde77bcdb5f70147ff28", "output_hash": "91306ecd2f87f05ab1c9b3377d592e9ec7f50f091fa1f94b9799c87e2d2a0108"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "1f9c3aaecaa102ceea59307a", "attempt_id": "69887100d4ae", "run_group_id": "E4_core:A4|closed|qwen3b:h1", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T06:56:01.563522Z", "task_family": "synthetic", "document_id": "syn_03006_2000", "document_hash": "89be5e77e0ca114963e9ffdf8dc38fe5610f13a78ac29e7a1956cd9124ce5217", "question_id": "q13_unanswerable", "split": "test", "cluster_id": "syn_03006_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 1, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1890, "query_tokens": 607, "message_tokens": 512, "message_utf8_bytes": 1087, "cache_tensor_bytes": null, "serialized_payload_bytes": 1087, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 22603.971540927887, "hops": {"1": 22603.971540927887}, "receiver_prefill_ms": 552.0399999804795, "ttft_ms": 39.047917001880705, "decode_ms": 25.83437506109476}, "total_wall_ms": 23221.598415984772, "ttft_ms": 39.047917001880705, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 64, "facts_retained": 30, "retention_rate": 0.46875, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.6666666666666666, "status": 0.9166666666666666, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1087}, "ledger_by_hop": {"1": {"facts_total": 64, "facts_retained": 30, "retention_rate": 0.46875, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.6666666666666666, "status": 0.9166666666666666, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1087}}}, "reference_key_id": "keys/synthetic/syn_03006_2000.json", "task_scores": {"correct": 1.0, "abstained": 1.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "47f843b1989b354c7cbe406d456e62c59b6229490f55dde77bcdb5f70147ff28", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "cdaa1c3ffd3a00ce62c5714b", "attempt_id": "f88a3eeb661a", "run_group_id": "E4_core:A4|closed|qwen3b:h1", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T06:56:02.199011Z", "task_family": "synthetic", "document_id": "syn_03006_2000", "document_hash": "89be5e77e0ca114963e9ffdf8dc38fe5610f13a78ac29e7a1956cd9124ce5217", "question_id": "q14_unanswerable", "split": "test", "cluster_id": "syn_03006_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 1, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1890, "query_tokens": 608, "message_tokens": 512, "message_utf8_bytes": 1087, "cache_tensor_bytes": null, "serialized_payload_bytes": 1087, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 22603.971540927887, "hops": {"1": 22603.971540927887}, "receiver_prefill_ms": 403.9077910128981, "ttft_ms": 40.383083978667855, "decode_ms": 30.542250024154782}, "total_wall_ms": 23079.551290953532, "ttft_ms": 40.383083978667855, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 64, "facts_retained": 30, "retention_rate": 0.46875, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.6666666666666666, "status": 0.9166666666666666, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1087}, "ledger_by_hop": {"1": {"facts_total": 64, "facts_retained": 30, "retention_rate": 0.46875, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.6666666666666666, "status": 0.9166666666666666, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1087}}}, "reference_key_id": "keys/synthetic/syn_03006_2000.json", "task_scores": {"correct": 1.0, "abstained": 1.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "47f843b1989b354c7cbe406d456e62c59b6229490f55dde77bcdb5f70147ff28", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "1d9927157acfd6c32ba12358", "attempt_id": "f31ddd4a8bc9", "run_group_id": "E4_core:A4|closed|qwen3b:h2", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T06:56:20.716672Z", "task_family": "synthetic", "document_id": "syn_03006_2000", "document_hash": "89be5e77e0ca114963e9ffdf8dc38fe5610f13a78ac29e7a1956cd9124ce5217", "question_id": "q00_single_fact", "split": "test", "cluster_id": "syn_03006_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 2, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1890, "query_tokens": 496, "message_tokens": 414, "message_utf8_bytes": 912, "cache_tensor_bytes": null, "serialized_payload_bytes": 912, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 40627.17524985783, "hops": {"1": 22603.971540927887, "2": 18023.20370892994}, "receiver_prefill_ms": 359.10645802505314, "ttft_ms": 112.71191597916186, "decode_ms": 452.94716698117554}, "total_wall_ms": 41552.548290928826, "ttft_ms": 112.71191597916186, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 10}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "XM-76Q-2958", "evidence_artifact": {"ledger": {"facts_total": 64, "facts_retained": 30, "retention_rate": 0.46875, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.6666666666666666, "status": 0.9166666666666666, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 912}, "ledger_by_hop": {"1": {"facts_total": 64, "facts_retained": 30, "retention_rate": 0.46875, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.6666666666666666, "status": 0.9166666666666666, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1087}, "2": {"facts_total": 64, "facts_retained": 30, "retention_rate": 0.46875, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.6666666666666666, "status": 0.9166666666666666, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 912}}}, "reference_key_id": "keys/synthetic/syn_03006_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "060968d99413307af0515a687c7ce77ff5260c81e4a1b8709ace3f1df2e644b7", "output_hash": "cf42691eec772b9ec74d67643944de2e19e53724d7b772cef1e0f2d5ef3b4926"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "e03a231210811db5ebc9b2ac", "attempt_id": "f08e2b4aff03", "run_group_id": "E4_core:A4|closed|qwen3b:h2", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T06:56:21.660048Z", "task_family": "synthetic", "document_id": "syn_03006_2000", "document_hash": "89be5e77e0ca114963e9ffdf8dc38fe5610f13a78ac29e7a1956cd9124ce5217", "question_id": "q01_single_fact", "split": "test", "cluster_id": "syn_03006_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 2, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1890, "query_tokens": 501, "message_tokens": 414, "message_utf8_bytes": 912, "cache_tensor_bytes": null, "serialized_payload_bytes": 912, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 40627.17524985783, "hops": {"1": 22603.971540927887, "2": 18023.20370892994}, "receiver_prefill_ms": 326.640082988888, "ttft_ms": 32.854542019777, "decode_ms": 34.94141704868525}, "total_wall_ms": 41022.41062489338, "ttft_ms": 32.854542019777, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 64, "facts_retained": 30, "retention_rate": 0.46875, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.6666666666666666, "status": 0.9166666666666666, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 912}, "ledger_by_hop": {"1": {"facts_total": 64, "facts_retained": 30, "retention_rate": 0.46875, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.6666666666666666, "status": 0.9166666666666666, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1087}, "2": {"facts_total": 64, "facts_retained": 30, "retention_rate": 0.46875, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.6666666666666666, "status": 0.9166666666666666, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 912}}}, "reference_key_id": "keys/synthetic/syn_03006_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "060968d99413307af0515a687c7ce77ff5260c81e4a1b8709ace3f1df2e644b7", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "7288470bde78233d74e6fabc", "attempt_id": "6bc931a77434", "run_group_id": "E4_core:A4|closed|qwen3b:h2", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T06:56:22.071760Z", "task_family": "synthetic", "document_id": "syn_03006_2000", "document_hash": "89be5e77e0ca114963e9ffdf8dc38fe5610f13a78ac29e7a1956cd9124ce5217", "question_id": "q02_exact_copy", "split": "test", "cluster_id": "syn_03006_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 2, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1890, "query_tokens": 500, "message_tokens": 414, "message_utf8_bytes": 912, "cache_tensor_bytes": null, "serialized_payload_bytes": 912, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 40627.17524985783, "hops": {"1": 22603.971540927887, "2": 18023.20370892994}, "receiver_prefill_ms": 283.2828750833869, "ttft_ms": 46.81133397389203, "decode_ms": 45.386708923615515}, "total_wall_ms": 41003.37170786224, "ttft_ms": 46.81133397389203, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 64, "facts_retained": 30, "retention_rate": 0.46875, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.6666666666666666, "status": 0.9166666666666666, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 912}, "ledger_by_hop": {"1": {"facts_total": 64, "facts_retained": 30, "retention_rate": 0.46875, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.6666666666666666, "status": 0.9166666666666666, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1087}, "2": {"facts_total": 64, "facts_retained": 30, "retention_rate": 0.46875, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.6666666666666666, "status": 0.9166666666666666, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 912}}}, "reference_key_id": "keys/synthetic/syn_03006_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "060968d99413307af0515a687c7ce77ff5260c81e4a1b8709ace3f1df2e644b7", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "ed6e672d58a8de7ee253a27a", "attempt_id": "b0b4dcf92523", "run_group_id": "E4_core:A4|closed|qwen3b:h2", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T06:56:22.465698Z", "task_family": "synthetic", "document_id": "syn_03006_2000", "document_hash": "89be5e77e0ca114963e9ffdf8dc38fe5610f13a78ac29e7a1956cd9124ce5217", "question_id": "q03_exact_copy", "split": "test", "cluster_id": "syn_03006_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 2, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1890, "query_tokens": 495, "message_tokens": 414, "message_utf8_bytes": 912, "cache_tensor_bytes": null, "serialized_payload_bytes": 912, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 40627.17524985783, "hops": {"1": 22603.971540927887, "2": 18023.20370892994}, "receiver_prefill_ms": 473.1902920175344, "ttft_ms": 45.19274993799627, "decode_ms": 307.34374991152436}, "total_wall_ms": 41453.4984579077, "ttft_ms": 45.19274993799627, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 8}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "Haruki Ouellet-Marchand", "evidence_artifact": {"ledger": {"facts_total": 64, "facts_retained": 30, "retention_rate": 0.46875, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.6666666666666666, "status": 0.9166666666666666, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 912}, "ledger_by_hop": {"1": {"facts_total": 64, "facts_retained": 30, "retention_rate": 0.46875, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.6666666666666666, "status": 0.9166666666666666, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1087}, "2": {"facts_total": 64, "facts_retained": 30, "retention_rate": 0.46875, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.6666666666666666, "status": 0.9166666666666666, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 912}}}, "reference_key_id": "keys/synthetic/syn_03006_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "060968d99413307af0515a687c7ce77ff5260c81e4a1b8709ace3f1df2e644b7", "output_hash": "a3b6e0a44d495387d8b8fc50ba6be0f2ed43121c674d3b3e962f8c4690e15b5c"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "9662e2b56ed00458a617cbeb", "attempt_id": "3ad9d156b952", "run_group_id": "E4_core:A4|closed|qwen3b:h2", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T06:56:23.311267Z", "task_family": "synthetic", "document_id": "syn_03006_2000", "document_hash": "89be5e77e0ca114963e9ffdf8dc38fe5610f13a78ac29e7a1956cd9124ce5217", "question_id": "q04_multi_hop", "split": "test", "cluster_id": "syn_03006_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 2, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1890, "query_tokens": 504, "message_tokens": 414, "message_utf8_bytes": 912, "cache_tensor_bytes": null, "serialized_payload_bytes": 912, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 40627.17524985783, "hops": {"1": 22603.971540927887, "2": 18023.20370892994}, "receiver_prefill_ms": 317.08104198332876, "ttft_ms": 25.088250054977834, "decode_ms": 40.521415998227894}, "total_wall_ms": 41010.44520782307, "ttft_ms": 25.088250054977834, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 64, "facts_retained": 30, "retention_rate": 0.46875, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.6666666666666666, "status": 0.9166666666666666, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 912}, "ledger_by_hop": {"1": {"facts_total": 64, "facts_retained": 30, "retention_rate": 0.46875, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.6666666666666666, "status": 0.9166666666666666, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1087}, "2": {"facts_total": 64, "facts_retained": 30, "retention_rate": 0.46875, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.6666666666666666, "status": 0.9166666666666666, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 912}}}, "reference_key_id": "keys/synthetic/syn_03006_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "060968d99413307af0515a687c7ce77ff5260c81e4a1b8709ace3f1df2e644b7", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "36a1f6b8d0e65a029b0a0c40", "attempt_id": "c7cc213212c5", "run_group_id": "E4_core:A4|closed|qwen3b:h2", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T06:56:23.711906Z", "task_family": "synthetic", "document_id": "syn_03006_2000", "document_hash": "89be5e77e0ca114963e9ffdf8dc38fe5610f13a78ac29e7a1956cd9124ce5217", "question_id": "q05_multi_hop", "split": "test", "cluster_id": "syn_03006_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 2, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1890, "query_tokens": 507, "message_tokens": 414, "message_utf8_bytes": 912, "cache_tensor_bytes": null, "serialized_payload_bytes": 912, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 40627.17524985783, "hops": {"1": 22603.971540927887, "2": 18023.20370892994}, "receiver_prefill_ms": 285.8177910093218, "ttft_ms": 67.88991601206362, "decode_ms": 585.3333750274032}, "total_wall_ms": 41566.98833277915, "ttft_ms": 67.88991601206362, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 10}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "EP-52G-1330", "evidence_artifact": {"ledger": {"facts_total": 64, "facts_retained": 30, "retention_rate": 0.46875, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.6666666666666666, "status": 0.9166666666666666, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 912}, "ledger_by_hop": {"1": {"facts_total": 64, "facts_retained": 30, "retention_rate": 0.46875, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.6666666666666666, "status": 0.9166666666666666, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1087}, "2": {"facts_total": 64, "facts_retained": 30, "retention_rate": 0.46875, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.6666666666666666, "status": 0.9166666666666666, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 912}}}, "reference_key_id": "keys/synthetic/syn_03006_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "060968d99413307af0515a687c7ce77ff5260c81e4a1b8709ace3f1df2e644b7", "output_hash": "104db207937fd79e23ce16f9b3f10b1a9ed34f6dd0441a99be6ea1b2d0a09d99"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "49e2fa5b6914b6b81e0ea10e", "attempt_id": "c074bc921466", "run_group_id": "E4_core:A4|closed|qwen3b:h2", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T06:56:24.668821Z", "task_family": "synthetic", "document_id": "syn_03006_2000", "document_hash": "89be5e77e0ca114963e9ffdf8dc38fe5610f13a78ac29e7a1956cd9124ce5217", "question_id": "q06_negation", "split": "test", "cluster_id": "syn_03006_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 2, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1890, "query_tokens": 495, "message_tokens": 414, "message_utf8_bytes": 912, "cache_tensor_bytes": null, "serialized_payload_bytes": 912, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 40627.17524985783, "hops": {"1": 22603.971540927887, "2": 18023.20370892994}, "receiver_prefill_ms": 321.9782920787111, "ttft_ms": 43.61858300399035, "decode_ms": 27.93924999423325}, "total_wall_ms": 41021.39891684055, "ttft_ms": 43.61858300399035, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "no", "evidence_artifact": {"ledger": {"facts_total": 64, "facts_retained": 30, "retention_rate": 0.46875, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.6666666666666666, "status": 0.9166666666666666, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 912}, "ledger_by_hop": {"1": {"facts_total": 64, "facts_retained": 30, "retention_rate": 0.46875, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.6666666666666666, "status": 0.9166666666666666, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1087}, "2": {"facts_total": 64, "facts_retained": 30, "retention_rate": 0.46875, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.6666666666666666, "status": 0.9166666666666666, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 912}}}, "reference_key_id": "keys/synthetic/syn_03006_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "060968d99413307af0515a687c7ce77ff5260c81e4a1b8709ace3f1df2e644b7", "output_hash": "9390298f3fb0c5b160498935d79cb139aef28e1c47358b4bbba61862b9c26e59"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "511c89ae1877fe8926589e76", "attempt_id": "5dcf57b5226b", "run_group_id": "E4_core:A4|closed|qwen3b:h2", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T06:56:25.080065Z", "task_family": "synthetic", "document_id": "syn_03006_2000", "document_hash": "89be5e77e0ca114963e9ffdf8dc38fe5610f13a78ac29e7a1956cd9124ce5217", "question_id": "q07_negation", "split": "test", "cluster_id": "syn_03006_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 2, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1890, "query_tokens": 495, "message_tokens": 414, "message_utf8_bytes": 912, "cache_tensor_bytes": null, "serialized_payload_bytes": 912, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 40627.17524985783, "hops": {"1": 22603.971540927887, "2": 18023.20370892994}, "receiver_prefill_ms": 318.25891591142863, "ttft_ms": 20.184500026516616, "decode_ms": 40.155666996724904}, "total_wall_ms": 41006.443124846555, "ttft_ms": 20.184500026516616, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "no", "evidence_artifact": {"ledger": {"facts_total": 64, "facts_retained": 30, "retention_rate": 0.46875, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.6666666666666666, "status": 0.9166666666666666, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 912}, "ledger_by_hop": {"1": {"facts_total": 64, "facts_retained": 30, "retention_rate": 0.46875, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.6666666666666666, "status": 0.9166666666666666, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1087}, "2": {"facts_total": 64, "facts_retained": 30, "retention_rate": 0.46875, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.6666666666666666, "status": 0.9166666666666666, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 912}}}, "reference_key_id": "keys/synthetic/syn_03006_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "060968d99413307af0515a687c7ce77ff5260c81e4a1b8709ace3f1df2e644b7", "output_hash": "9390298f3fb0c5b160498935d79cb139aef28e1c47358b4bbba61862b9c26e59"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "837561e90e8be2b96d9beeda", "attempt_id": "f8db073ddf33", "run_group_id": "E4_core:A4|closed|qwen3b:h2", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T06:56:25.476301Z", "task_family": "synthetic", "document_id": "syn_03006_2000", "document_hash": "89be5e77e0ca114963e9ffdf8dc38fe5610f13a78ac29e7a1956cd9124ce5217", "question_id": "q08_revision", "split": "test", "cluster_id": "syn_03006_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 2, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1890, "query_tokens": 512, "message_tokens": 414, "message_utf8_bytes": 912, "cache_tensor_bytes": null, "serialized_payload_bytes": 912, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 40627.17524985783, "hops": {"1": 22603.971540927887, "2": 18023.20370892994}, "receiver_prefill_ms": 336.74754202365875, "ttft_ms": 42.80820896383375, "decode_ms": 424.5498749660328}, "total_wall_ms": 41432.09879193455, "ttft_ms": 42.80820896383375, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 7}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "9112735", "evidence_artifact": {"ledger": {"facts_total": 64, "facts_retained": 30, "retention_rate": 0.46875, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.6666666666666666, "status": 0.9166666666666666, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 912}, "ledger_by_hop": {"1": {"facts_total": 64, "facts_retained": 30, "retention_rate": 0.46875, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.6666666666666666, "status": 0.9166666666666666, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1087}, "2": {"facts_total": 64, "facts_retained": 30, "retention_rate": 0.46875, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.6666666666666666, "status": 0.9166666666666666, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 912}}}, "reference_key_id": "keys/synthetic/syn_03006_2000.json", "task_scores": {"correct": 0.0, "abstained": 0.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "060968d99413307af0515a687c7ce77ff5260c81e4a1b8709ace3f1df2e644b7", "output_hash": "15a3f804636f0e62d4a3d9be7677f001a0ad6b1fb8ae7e8cdc63c44ec7bac31f"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "baa0413b960710cc448a0a90", "attempt_id": "b664b37c5775", "run_group_id": "E4_core:A4|closed|qwen3b:h2", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T06:56:26.298948Z", "task_family": "synthetic", "document_id": "syn_03006_2000", "document_hash": "89be5e77e0ca114963e9ffdf8dc38fe5610f13a78ac29e7a1956cd9124ce5217", "question_id": "q09_revision", "split": "test", "cluster_id": "syn_03006_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 2, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1890, "query_tokens": 505, "message_tokens": 414, "message_utf8_bytes": 912, "cache_tensor_bytes": null, "serialized_payload_bytes": 912, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 40627.17524985783, "hops": {"1": 22603.971540927887, "2": 18023.20370892994}, "receiver_prefill_ms": 328.60858307685703, "ttft_ms": 37.7388340421021, "decode_ms": 279.6828750288114}, "total_wall_ms": 41273.77129183151, "ttft_ms": 37.7388340421021, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 9}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "9,112,735", "evidence_artifact": {"ledger": {"facts_total": 64, "facts_retained": 30, "retention_rate": 0.46875, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.6666666666666666, "status": 0.9166666666666666, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 912}, "ledger_by_hop": {"1": {"facts_total": 64, "facts_retained": 30, "retention_rate": 0.46875, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.6666666666666666, "status": 0.9166666666666666, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1087}, "2": {"facts_total": 64, "facts_retained": 30, "retention_rate": 0.46875, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.6666666666666666, "status": 0.9166666666666666, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 912}}}, "reference_key_id": "keys/synthetic/syn_03006_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "060968d99413307af0515a687c7ce77ff5260c81e4a1b8709ace3f1df2e644b7", "output_hash": "788fcda908c2029597420995b09d0cc919bfdbf7a28b68dcb0e331db7df7b342"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "7934e140fea87ce115c311df", "attempt_id": "27e8db7ad107", "run_group_id": "E4_core:A4|closed|qwen3b:h2", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T06:56:26.963108Z", "task_family": "synthetic", "document_id": "syn_03006_2000", "document_hash": "89be5e77e0ca114963e9ffdf8dc38fe5610f13a78ac29e7a1956cd9124ce5217", "question_id": "q10_temporal", "split": "test", "cluster_id": "syn_03006_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 2, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1890, "query_tokens": 514, "message_tokens": 414, "message_utf8_bytes": 912, "cache_tensor_bytes": null, "serialized_payload_bytes": 912, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 40627.17524985783, "hops": {"1": 22603.971540927887, "2": 18023.20370892994}, "receiver_prefill_ms": 393.01754091866314, "ttft_ms": 62.31279205530882, "decode_ms": 97.44370798580348}, "total_wall_ms": 41180.63216586597, "ttft_ms": 62.31279205530882, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 64, "facts_retained": 30, "retention_rate": 0.46875, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.6666666666666666, "status": 0.9166666666666666, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 912}, "ledger_by_hop": {"1": {"facts_total": 64, "facts_retained": 30, "retention_rate": 0.46875, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.6666666666666666, "status": 0.9166666666666666, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1087}, "2": {"facts_total": 64, "facts_retained": 30, "retention_rate": 0.46875, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.6666666666666666, "status": 0.9166666666666666, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 912}}}, "reference_key_id": "keys/synthetic/syn_03006_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "060968d99413307af0515a687c7ce77ff5260c81e4a1b8709ace3f1df2e644b7", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "143e8b6cdb8809155b511daf", "attempt_id": "eadc7bee5003", "run_group_id": "E4_core:A4|closed|qwen3b:h2", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T06:56:27.534641Z", "task_family": "synthetic", "document_id": "syn_03006_2000", "document_hash": "89be5e77e0ca114963e9ffdf8dc38fe5610f13a78ac29e7a1956cd9124ce5217", "question_id": "q11_aggregation", "split": "test", "cluster_id": "syn_03006_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 2, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1890, "query_tokens": 507, "message_tokens": 414, "message_utf8_bytes": 912, "cache_tensor_bytes": null, "serialized_payload_bytes": 912, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 40627.17524985783, "hops": {"1": 22603.971540927887, "2": 18023.20370892994}, "receiver_prefill_ms": 394.4724169559777, "ttft_ms": 37.47025004122406, "decode_ms": 254.06404200475663}, "total_wall_ms": 41313.93962493166, "ttft_ms": 37.47025004122406, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 7}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "8908553", "evidence_artifact": {"ledger": {"facts_total": 64, "facts_retained": 30, "retention_rate": 0.46875, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.6666666666666666, "status": 0.9166666666666666, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 912}, "ledger_by_hop": {"1": {"facts_total": 64, "facts_retained": 30, "retention_rate": 0.46875, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.6666666666666666, "status": 0.9166666666666666, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1087}, "2": {"facts_total": 64, "facts_retained": 30, "retention_rate": 0.46875, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.6666666666666666, "status": 0.9166666666666666, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 912}}}, "reference_key_id": "keys/synthetic/syn_03006_2000.json", "task_scores": {"correct": 0.0, "abstained": 0.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "060968d99413307af0515a687c7ce77ff5260c81e4a1b8709ace3f1df2e644b7", "output_hash": "5ad9e777dc3be36808c74badfc8aae217b41a486754d07adf1abda6f6bce9aca"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "de7d4b788221636db81bcdce", "attempt_id": "12ae823ddeff", "run_group_id": "E4_core:A4|closed|qwen3b:h2", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T06:56:28.240824Z", "task_family": "synthetic", "document_id": "syn_03006_2000", "document_hash": "89be5e77e0ca114963e9ffdf8dc38fe5610f13a78ac29e7a1956cd9124ce5217", "question_id": "q12_unit_conversion", "split": "test", "cluster_id": "syn_03006_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 2, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1890, "query_tokens": 504, "message_tokens": 414, "message_utf8_bytes": 912, "cache_tensor_bytes": null, "serialized_payload_bytes": 912, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 40627.17524985783, "hops": {"1": 22603.971540927887, "2": 18023.20370892994}, "receiver_prefill_ms": 335.6405410449952, "ttft_ms": 19.6130839176476, "decode_ms": 216.5382499806583}, "total_wall_ms": 41199.581124819815, "ttft_ms": 19.6130839176476, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 5}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "39700", "evidence_artifact": {"ledger": {"facts_total": 64, "facts_retained": 30, "retention_rate": 0.46875, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.6666666666666666, "status": 0.9166666666666666, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 912}, "ledger_by_hop": {"1": {"facts_total": 64, "facts_retained": 30, "retention_rate": 0.46875, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.6666666666666666, "status": 0.9166666666666666, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1087}, "2": {"facts_total": 64, "facts_retained": 30, "retention_rate": 0.46875, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.6666666666666666, "status": 0.9166666666666666, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 912}}}, "reference_key_id": "keys/synthetic/syn_03006_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "060968d99413307af0515a687c7ce77ff5260c81e4a1b8709ace3f1df2e644b7", "output_hash": "d8f3f4a352330800cb9f3d838e47d965e9c710d166f53c6b3724fdeae73c9a5e"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "8a33e2c6c253f63c2881d293", "attempt_id": "f0aff27e781b", "run_group_id": "E4_core:A4|closed|qwen3b:h2", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T06:56:28.830388Z", "task_family": "synthetic", "document_id": "syn_03006_2000", "document_hash": "89be5e77e0ca114963e9ffdf8dc38fe5610f13a78ac29e7a1956cd9124ce5217", "question_id": "q13_unanswerable", "split": "test", "cluster_id": "syn_03006_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 2, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1890, "query_tokens": 508, "message_tokens": 414, "message_utf8_bytes": 912, "cache_tensor_bytes": null, "serialized_payload_bytes": 912, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 40627.17524985783, "hops": {"1": 22603.971540927887, "2": 18023.20370892994}, "receiver_prefill_ms": 430.41641695890576, "ttft_ms": 63.94970801193267, "decode_ms": 23.891624994575977}, "total_wall_ms": 41146.00633387454, "ttft_ms": 63.94970801193267, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 64, "facts_retained": 30, "retention_rate": 0.46875, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.6666666666666666, "status": 0.9166666666666666, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 912}, "ledger_by_hop": {"1": {"facts_total": 64, "facts_retained": 30, "retention_rate": 0.46875, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.6666666666666666, "status": 0.9166666666666666, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1087}, "2": {"facts_total": 64, "facts_retained": 30, "retention_rate": 0.46875, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.6666666666666666, "status": 0.9166666666666666, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 912}}}, "reference_key_id": "keys/synthetic/syn_03006_2000.json", "task_scores": {"correct": 1.0, "abstained": 1.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "060968d99413307af0515a687c7ce77ff5260c81e4a1b8709ace3f1df2e644b7", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "0f27395d838015fc5595d2f9", "attempt_id": "fb1d017c1d9c", "run_group_id": "E4_core:A4|closed|qwen3b:h2", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T06:56:29.366506Z", "task_family": "synthetic", "document_id": "syn_03006_2000", "document_hash": "89be5e77e0ca114963e9ffdf8dc38fe5610f13a78ac29e7a1956cd9124ce5217", "question_id": "q14_unanswerable", "split": "test", "cluster_id": "syn_03006_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 2, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1890, "query_tokens": 509, "message_tokens": 414, "message_utf8_bytes": 912, "cache_tensor_bytes": null, "serialized_payload_bytes": 912, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 40627.17524985783, "hops": {"1": 22603.971540927887, "2": 18023.20370892994}, "receiver_prefill_ms": 350.3497919300571, "ttft_ms": 22.25354197435081, "decode_ms": 37.40662499330938}, "total_wall_ms": 41037.74737485219, "ttft_ms": 22.25354197435081, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 64, "facts_retained": 30, "retention_rate": 0.46875, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.6666666666666666, "status": 0.9166666666666666, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 912}, "ledger_by_hop": {"1": {"facts_total": 64, "facts_retained": 30, "retention_rate": 0.46875, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.6666666666666666, "status": 0.9166666666666666, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1087}, "2": {"facts_total": 64, "facts_retained": 30, "retention_rate": 0.46875, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.6666666666666666, "status": 0.9166666666666666, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 912}}}, "reference_key_id": "keys/synthetic/syn_03006_2000.json", "task_scores": {"correct": 1.0, "abstained": 1.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "060968d99413307af0515a687c7ce77ff5260c81e4a1b8709ace3f1df2e644b7", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "9243fd112aeedde2ad1d189f", "attempt_id": "19891466d185", "run_group_id": "E4_core:A4|closed|qwen3b:h4", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T06:57:05.290994Z", "task_family": "synthetic", "document_id": "syn_03006_2000", "document_hash": "89be5e77e0ca114963e9ffdf8dc38fe5610f13a78ac29e7a1956cd9124ce5217", "question_id": "q00_single_fact", "split": "test", "cluster_id": "syn_03006_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 4, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1890, "query_tokens": 496, "message_tokens": 414, "message_utf8_bytes": 912, "cache_tensor_bytes": null, "serialized_payload_bytes": 912, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 76121.53695872985, "hops": {"1": 22603.971540927887, "2": 18023.20370892994, "3": 17767.595666926354, "4": 17726.766041945666}, "receiver_prefill_ms": 326.8334170570597, "ttft_ms": 20.054874941706657, "decode_ms": 625.5557499825954}, "total_wall_ms": 77094.92025070358, "ttft_ms": 20.054874941706657, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 10}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "XM-76Q-2958", "evidence_artifact": {"ledger": {"facts_total": 64, "facts_retained": 30, "retention_rate": 0.46875, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.6666666666666666, "status": 0.9166666666666666, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 912}, "ledger_by_hop": {"1": {"facts_total": 64, "facts_retained": 30, "retention_rate": 0.46875, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.6666666666666666, "status": 0.9166666666666666, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1087}, "2": {"facts_total": 64, "facts_retained": 30, "retention_rate": 0.46875, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.6666666666666666, "status": 0.9166666666666666, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 912}, "3": {"facts_total": 64, "facts_retained": 30, "retention_rate": 0.46875, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.6666666666666666, "status": 0.9166666666666666, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 912}, "4": {"facts_total": 64, "facts_retained": 30, "retention_rate": 0.46875, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.6666666666666666, "status": 0.9166666666666666, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 912}}}, "reference_key_id": "keys/synthetic/syn_03006_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "060968d99413307af0515a687c7ce77ff5260c81e4a1b8709ace3f1df2e644b7", "output_hash": "cf42691eec772b9ec74d67643944de2e19e53724d7b772cef1e0f2d5ef3b4926"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "1bb86b9854f02d0345f4ddfe", "attempt_id": "b6e9b70da6d6", "run_group_id": "E4_core:A4|closed|qwen3b:h4", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T06:57:06.282302Z", "task_family": "synthetic", "document_id": "syn_03006_2000", "document_hash": "89be5e77e0ca114963e9ffdf8dc38fe5610f13a78ac29e7a1956cd9124ce5217", "question_id": "q01_single_fact", "split": "test", "cluster_id": "syn_03006_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 4, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1890, "query_tokens": 501, "message_tokens": 414, "message_utf8_bytes": 912, "cache_tensor_bytes": null, "serialized_payload_bytes": 912, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 76121.53695872985, "hops": {"1": 22603.971540927887, "2": 18023.20370892994, "3": 17767.595666926354, "4": 17726.766041945666}, "receiver_prefill_ms": 331.83262508828193, "ttft_ms": 40.884042042307556, "decode_ms": 22.79529196675867}, "total_wall_ms": 76517.65883364715, "ttft_ms": 40.884042042307556, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 64, "facts_retained": 30, "retention_rate": 0.46875, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.6666666666666666, "status": 0.9166666666666666, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 912}, "ledger_by_hop": {"1": {"facts_total": 64, "facts_retained": 30, "retention_rate": 0.46875, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.6666666666666666, "status": 0.9166666666666666, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1087}, "2": {"facts_total": 64, "facts_retained": 30, "retention_rate": 0.46875, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.6666666666666666, "status": 0.9166666666666666, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 912}, "3": {"facts_total": 64, "facts_retained": 30, "retention_rate": 0.46875, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.6666666666666666, "status": 0.9166666666666666, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 912}, "4": {"facts_total": 64, "facts_retained": 30, "retention_rate": 0.46875, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.6666666666666666, "status": 0.9166666666666666, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 912}}}, "reference_key_id": "keys/synthetic/syn_03006_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "060968d99413307af0515a687c7ce77ff5260c81e4a1b8709ace3f1df2e644b7", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "8b37c2724778c2e9816623e2", "attempt_id": "458187eb442e", "run_group_id": "E4_core:A4|closed|qwen3b:h4", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T06:57:06.696612Z", "task_family": "synthetic", "document_id": "syn_03006_2000", "document_hash": "89be5e77e0ca114963e9ffdf8dc38fe5610f13a78ac29e7a1956cd9124ce5217", "question_id": "q02_exact_copy", "split": "test", "cluster_id": "syn_03006_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 4, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1890, "query_tokens": 500, "message_tokens": 414, "message_utf8_bytes": 912, "cache_tensor_bytes": null, "serialized_payload_bytes": 912, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 76121.53695872985, "hops": {"1": 22603.971540927887, "2": 18023.20370892994, "3": 17767.595666926354, "4": 17726.766041945666}, "receiver_prefill_ms": 316.1887499736622, "ttft_ms": 38.3426669286564, "decode_ms": 24.30150005966425}, "total_wall_ms": 76500.83120877389, "ttft_ms": 38.3426669286564, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 64, "facts_retained": 30, "retention_rate": 0.46875, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.6666666666666666, "status": 0.9166666666666666, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 912}, "ledger_by_hop": {"1": {"facts_total": 64, "facts_retained": 30, "retention_rate": 0.46875, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.6666666666666666, "status": 0.9166666666666666, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1087}, "2": {"facts_total": 64, "facts_retained": 30, "retention_rate": 0.46875, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.6666666666666666, "status": 0.9166666666666666, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 912}, "3": {"facts_total": 64, "facts_retained": 30, "retention_rate": 0.46875, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.6666666666666666, "status": 0.9166666666666666, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 912}, "4": {"facts_total": 64, "facts_retained": 30, "retention_rate": 0.46875, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.6666666666666666, "status": 0.9166666666666666, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 912}}}, "reference_key_id": "keys/synthetic/syn_03006_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "060968d99413307af0515a687c7ce77ff5260c81e4a1b8709ace3f1df2e644b7", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "c956db38f5ae5fd74e09c61e", "attempt_id": "38d8159ab588", "run_group_id": "E4_core:A4|closed|qwen3b:h4", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T06:57:07.092578Z", "task_family": "synthetic", "document_id": "syn_03006_2000", "document_hash": "89be5e77e0ca114963e9ffdf8dc38fe5610f13a78ac29e7a1956cd9124ce5217", "question_id": "q03_exact_copy", "split": "test", "cluster_id": "syn_03006_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 4, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1890, "query_tokens": 495, "message_tokens": 414, "message_utf8_bytes": 912, "cache_tensor_bytes": null, "serialized_payload_bytes": 912, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 76121.53695872985, "hops": {"1": 22603.971540927887, "2": 18023.20370892994, "3": 17767.595666926354, "4": 17726.766041945666}, "receiver_prefill_ms": 270.4464590642601, "ttft_ms": 48.346249968744814, "decode_ms": 511.55745796859264}, "total_wall_ms": 76952.45341770351, "ttft_ms": 48.346249968744814, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 8}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "Haruki Ouellet-Marchand", "evidence_artifact": {"ledger": {"facts_total": 64, "facts_retained": 30, "retention_rate": 0.46875, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.6666666666666666, "status": 0.9166666666666666, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 912}, "ledger_by_hop": {"1": {"facts_total": 64, "facts_retained": 30, "retention_rate": 0.46875, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.6666666666666666, "status": 0.9166666666666666, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1087}, "2": {"facts_total": 64, "facts_retained": 30, "retention_rate": 0.46875, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.6666666666666666, "status": 0.9166666666666666, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 912}, "3": {"facts_total": 64, "facts_retained": 30, "retention_rate": 0.46875, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.6666666666666666, "status": 0.9166666666666666, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 912}, "4": {"facts_total": 64, "facts_retained": 30, "retention_rate": 0.46875, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.6666666666666666, "status": 0.9166666666666666, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 912}}}, "reference_key_id": "keys/synthetic/syn_03006_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "060968d99413307af0515a687c7ce77ff5260c81e4a1b8709ace3f1df2e644b7", "output_hash": "a3b6e0a44d495387d8b8fc50ba6be0f2ed43121c674d3b3e962f8c4690e15b5c"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "26a9acb1b774d63085e43ae5", "attempt_id": "84b283b64c69", "run_group_id": "E4_core:A4|closed|qwen3b:h4", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T06:57:07.940364Z", "task_family": "synthetic", "document_id": "syn_03006_2000", "document_hash": "89be5e77e0ca114963e9ffdf8dc38fe5610f13a78ac29e7a1956cd9124ce5217", "question_id": "q04_multi_hop", "split": "test", "cluster_id": "syn_03006_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 4, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1890, "query_tokens": 504, "message_tokens": 414, "message_utf8_bytes": 912, "cache_tensor_bytes": null, "serialized_payload_bytes": 912, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 76121.53695872985, "hops": {"1": 22603.971540927887, "2": 18023.20370892994, "3": 17767.595666926354, "4": 17726.766041945666}, "receiver_prefill_ms": 317.89791700430214, "ttft_ms": 40.565499919466674, "decode_ms": 21.879791980609298}, "total_wall_ms": 76502.43412575219, "ttft_ms": 40.565499919466674, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 64, "facts_retained": 30, "retention_rate": 0.46875, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.6666666666666666, "status": 0.9166666666666666, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 912}, "ledger_by_hop": {"1": {"facts_total": 64, "facts_retained": 30, "retention_rate": 0.46875, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.6666666666666666, "status": 0.9166666666666666, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1087}, "2": {"facts_total": 64, "facts_retained": 30, "retention_rate": 0.46875, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.6666666666666666, "status": 0.9166666666666666, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 912}, "3": {"facts_total": 64, "facts_retained": 30, "retention_rate": 0.46875, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.6666666666666666, "status": 0.9166666666666666, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 912}, "4": {"facts_total": 64, "facts_retained": 30, "retention_rate": 0.46875, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.6666666666666666, "status": 0.9166666666666666, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 912}}}, "reference_key_id": "keys/synthetic/syn_03006_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "060968d99413307af0515a687c7ce77ff5260c81e4a1b8709ace3f1df2e644b7", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "bc6e88367df183ec679f3f2a", "attempt_id": "42a34aff11af", "run_group_id": "E4_core:A4|closed|qwen3b:h4", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T06:57:08.338362Z", "task_family": "synthetic", "document_id": "syn_03006_2000", "document_hash": "89be5e77e0ca114963e9ffdf8dc38fe5610f13a78ac29e7a1956cd9124ce5217", "question_id": "q05_multi_hop", "split": "test", "cluster_id": "syn_03006_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 4, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1890, "query_tokens": 507, "message_tokens": 414, "message_utf8_bytes": 912, "cache_tensor_bytes": null, "serialized_payload_bytes": 912, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 76121.53695872985, "hops": {"1": 22603.971540927887, "2": 18023.20370892994, "3": 17767.595666926354, "4": 17726.766041945666}, "receiver_prefill_ms": 322.9911669623107, "ttft_ms": 40.21237511187792, "decode_ms": 408.2818749593571}, "total_wall_ms": 76893.73699983116, "ttft_ms": 40.21237511187792, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 10}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "EP-52G-1330", "evidence_artifact": {"ledger": {"facts_total": 64, "facts_retained": 30, "retention_rate": 0.46875, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.6666666666666666, "status": 0.9166666666666666, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 912}, "ledger_by_hop": {"1": {"facts_total": 64, "facts_retained": 30, "retention_rate": 0.46875, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.6666666666666666, "status": 0.9166666666666666, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1087}, "2": {"facts_total": 64, "facts_retained": 30, "retention_rate": 0.46875, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.6666666666666666, "status": 0.9166666666666666, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 912}, "3": {"facts_total": 64, "facts_retained": 30, "retention_rate": 0.46875, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.6666666666666666, "status": 0.9166666666666666, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 912}, "4": {"facts_total": 64, "facts_retained": 30, "retention_rate": 0.46875, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.6666666666666666, "status": 0.9166666666666666, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 912}}}, "reference_key_id": "keys/synthetic/syn_03006_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "060968d99413307af0515a687c7ce77ff5260c81e4a1b8709ace3f1df2e644b7", "output_hash": "104db207937fd79e23ce16f9b3f10b1a9ed34f6dd0441a99be6ea1b2d0a09d99"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "b8aa082064bfe35dcf18f072", "attempt_id": "838eae218839", "run_group_id": "E4_core:A4|closed|qwen3b:h4", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T06:57:09.127997Z", "task_family": "synthetic", "document_id": "syn_03006_2000", "document_hash": "89be5e77e0ca114963e9ffdf8dc38fe5610f13a78ac29e7a1956cd9124ce5217", "question_id": "q06_negation", "split": "test", "cluster_id": "syn_03006_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 4, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1890, "query_tokens": 495, "message_tokens": 414, "message_utf8_bytes": 912, "cache_tensor_bytes": null, "serialized_payload_bytes": 912, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 76121.53695872985, "hops": {"1": 22603.971540927887, "2": 18023.20370892994, "3": 17767.595666926354, "4": 17726.766041945666}, "receiver_prefill_ms": 466.39783307909966, "ttft_ms": 21.163959056138992, "decode_ms": 41.01754107978195}, "total_wall_ms": 76650.74991679285, "ttft_ms": 21.163959056138992, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "no", "evidence_artifact": {"ledger": {"facts_total": 64, "facts_retained": 30, "retention_rate": 0.46875, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.6666666666666666, "status": 0.9166666666666666, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 912}, "ledger_by_hop": {"1": {"facts_total": 64, "facts_retained": 30, "retention_rate": 0.46875, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.6666666666666666, "status": 0.9166666666666666, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1087}, "2": {"facts_total": 64, "facts_retained": 30, "retention_rate": 0.46875, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.6666666666666666, "status": 0.9166666666666666, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 912}, "3": {"facts_total": 64, "facts_retained": 30, "retention_rate": 0.46875, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.6666666666666666, "status": 0.9166666666666666, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 912}, "4": {"facts_total": 64, "facts_retained": 30, "retention_rate": 0.46875, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.6666666666666666, "status": 0.9166666666666666, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 912}}}, "reference_key_id": "keys/synthetic/syn_03006_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "060968d99413307af0515a687c7ce77ff5260c81e4a1b8709ace3f1df2e644b7", "output_hash": "9390298f3fb0c5b160498935d79cb139aef28e1c47358b4bbba61862b9c26e59"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "542327d491be1a4cf97546f5", "attempt_id": "2a82124c13d3", "run_group_id": "E4_core:A4|closed|qwen3b:h4", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T06:57:09.674811Z", "task_family": "synthetic", "document_id": "syn_03006_2000", "document_hash": "89be5e77e0ca114963e9ffdf8dc38fe5610f13a78ac29e7a1956cd9124ce5217", "question_id": "q07_negation", "split": "test", "cluster_id": "syn_03006_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 4, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1890, "query_tokens": 495, "message_tokens": 414, "message_utf8_bytes": 912, "cache_tensor_bytes": null, "serialized_payload_bytes": 912, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 76121.53695872985, "hops": {"1": 22603.971540927887, "2": 18023.20370892994, "3": 17767.595666926354, "4": 17726.766041945666}, "receiver_prefill_ms": 334.8360840464011, "ttft_ms": 38.33341703284532, "decode_ms": 22.193583892658353}, "total_wall_ms": 76517.5764587475, "ttft_ms": 38.33341703284532, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "no", "evidence_artifact": {"ledger": {"facts_total": 64, "facts_retained": 30, "retention_rate": 0.46875, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.6666666666666666, "status": 0.9166666666666666, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 912}, "ledger_by_hop": {"1": {"facts_total": 64, "facts_retained": 30, "retention_rate": 0.46875, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.6666666666666666, "status": 0.9166666666666666, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1087}, "2": {"facts_total": 64, "facts_retained": 30, "retention_rate": 0.46875, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.6666666666666666, "status": 0.9166666666666666, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 912}, "3": {"facts_total": 64, "facts_retained": 30, "retention_rate": 0.46875, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.6666666666666666, "status": 0.9166666666666666, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 912}, "4": {"facts_total": 64, "facts_retained": 30, "retention_rate": 0.46875, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.6666666666666666, "status": 0.9166666666666666, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 912}}}, "reference_key_id": "keys/synthetic/syn_03006_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "060968d99413307af0515a687c7ce77ff5260c81e4a1b8709ace3f1df2e644b7", "output_hash": "9390298f3fb0c5b160498935d79cb139aef28e1c47358b4bbba61862b9c26e59"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "be69f34d59c02a606796a0ef", "attempt_id": "4732296d900e", "run_group_id": "E4_core:A4|closed|qwen3b:h4", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T06:57:10.087535Z", "task_family": "synthetic", "document_id": "syn_03006_2000", "document_hash": "89be5e77e0ca114963e9ffdf8dc38fe5610f13a78ac29e7a1956cd9124ce5217", "question_id": "q08_revision", "split": "test", "cluster_id": "syn_03006_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 4, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1890, "query_tokens": 512, "message_tokens": 414, "message_utf8_bytes": 912, "cache_tensor_bytes": null, "serialized_payload_bytes": 912, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 76121.53695872985, "hops": {"1": 22603.971540927887, "2": 18023.20370892994, "3": 17767.595666926354, "4": 17726.766041945666}, "receiver_prefill_ms": 333.0377920065075, "ttft_ms": 19.677250063978136, "decode_ms": 348.34566607605666}, "total_wall_ms": 76823.2509996742, "ttft_ms": 19.677250063978136, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 7}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "9112735", "evidence_artifact": {"ledger": {"facts_total": 64, "facts_retained": 30, "retention_rate": 0.46875, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.6666666666666666, "status": 0.9166666666666666, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 912}, "ledger_by_hop": {"1": {"facts_total": 64, "facts_retained": 30, "retention_rate": 0.46875, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.6666666666666666, "status": 0.9166666666666666, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1087}, "2": {"facts_total": 64, "facts_retained": 30, "retention_rate": 0.46875, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.6666666666666666, "status": 0.9166666666666666, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 912}, "3": {"facts_total": 64, "facts_retained": 30, "retention_rate": 0.46875, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.6666666666666666, "status": 0.9166666666666666, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 912}, "4": {"facts_total": 64, "facts_retained": 30, "retention_rate": 0.46875, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.6666666666666666, "status": 0.9166666666666666, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 912}}}, "reference_key_id": "keys/synthetic/syn_03006_2000.json", "task_scores": {"correct": 0.0, "abstained": 0.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "060968d99413307af0515a687c7ce77ff5260c81e4a1b8709ace3f1df2e644b7", "output_hash": "15a3f804636f0e62d4a3d9be7677f001a0ad6b1fb8ae7e8cdc63c44ec7bac31f"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "ef3ac1b95f23a29984f3e2ca", "attempt_id": "4dba31239375", "run_group_id": "E4_core:A4|closed|qwen3b:h4", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T06:57:10.807542Z", "task_family": "synthetic", "document_id": "syn_03006_2000", "document_hash": "89be5e77e0ca114963e9ffdf8dc38fe5610f13a78ac29e7a1956cd9124ce5217", "question_id": "q09_revision", "split": "test", "cluster_id": "syn_03006_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 4, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1890, "query_tokens": 505, "message_tokens": 414, "message_utf8_bytes": 912, "cache_tensor_bytes": null, "serialized_payload_bytes": 912, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 76121.53695872985, "hops": {"1": 22603.971540927887, "2": 18023.20370892994, "3": 17767.595666926354, "4": 17726.766041945666}, "receiver_prefill_ms": 466.8175419792533, "ttft_ms": 20.716375089250505, "decode_ms": 299.47320895735174}, "total_wall_ms": 76909.29570875596, "ttft_ms": 20.716375089250505, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 9}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "9,112,735", "evidence_artifact": {"ledger": {"facts_total": 64, "facts_retained": 30, "retention_rate": 0.46875, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.6666666666666666, "status": 0.9166666666666666, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 912}, "ledger_by_hop": {"1": {"facts_total": 64, "facts_retained": 30, "retention_rate": 0.46875, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.6666666666666666, "status": 0.9166666666666666, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1087}, "2": {"facts_total": 64, "facts_retained": 30, "retention_rate": 0.46875, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.6666666666666666, "status": 0.9166666666666666, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 912}, "3": {"facts_total": 64, "facts_retained": 30, "retention_rate": 0.46875, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.6666666666666666, "status": 0.9166666666666666, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 912}, "4": {"facts_total": 64, "facts_retained": 30, "retention_rate": 0.46875, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.6666666666666666, "status": 0.9166666666666666, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 912}}}, "reference_key_id": "keys/synthetic/syn_03006_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "060968d99413307af0515a687c7ce77ff5260c81e4a1b8709ace3f1df2e644b7", "output_hash": "788fcda908c2029597420995b09d0cc919bfdbf7a28b68dcb0e331db7df7b342"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "207e5c0ba996b4b9c6d7ae06", "attempt_id": "c2ea60d8812e", "run_group_id": "E4_core:A4|closed|qwen3b:h4", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T06:57:11.612742Z", "task_family": "synthetic", "document_id": "syn_03006_2000", "document_hash": "89be5e77e0ca114963e9ffdf8dc38fe5610f13a78ac29e7a1956cd9124ce5217", "question_id": "q10_temporal", "split": "test", "cluster_id": "syn_03006_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 4, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1890, "query_tokens": 514, "message_tokens": 414, "message_utf8_bytes": 912, "cache_tensor_bytes": null, "serialized_payload_bytes": 912, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 76121.53695872985, "hops": {"1": 22603.971540927887, "2": 18023.20370892994, "3": 17767.595666926354, "4": 17726.766041945666}, "receiver_prefill_ms": 358.83345804177225, "ttft_ms": 21.532833110541105, "decode_ms": 57.46825004462153}, "total_wall_ms": 76560.23579172324, "ttft_ms": 21.532833110541105, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 64, "facts_retained": 30, "retention_rate": 0.46875, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.6666666666666666, "status": 0.9166666666666666, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 912}, "ledger_by_hop": {"1": {"facts_total": 64, "facts_retained": 30, "retention_rate": 0.46875, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.6666666666666666, "status": 0.9166666666666666, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1087}, "2": {"facts_total": 64, "facts_retained": 30, "retention_rate": 0.46875, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.6666666666666666, "status": 0.9166666666666666, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 912}, "3": {"facts_total": 64, "facts_retained": 30, "retention_rate": 0.46875, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.6666666666666666, "status": 0.9166666666666666, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 912}, "4": {"facts_total": 64, "facts_retained": 30, "retention_rate": 0.46875, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.6666666666666666, "status": 0.9166666666666666, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 912}}}, "reference_key_id": "keys/synthetic/syn_03006_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "060968d99413307af0515a687c7ce77ff5260c81e4a1b8709ace3f1df2e644b7", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "dc20bd8b2ece17905dcce15a", "attempt_id": "8b4902c05aa4", "run_group_id": "E4_core:A4|closed|qwen3b:h4", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T06:57:12.069388Z", "task_family": "synthetic", "document_id": "syn_03006_2000", "document_hash": "89be5e77e0ca114963e9ffdf8dc38fe5610f13a78ac29e7a1956cd9124ce5217", "question_id": "q11_aggregation", "split": "test", "cluster_id": "syn_03006_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 4, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1890, "query_tokens": 507, "message_tokens": 414, "message_utf8_bytes": 912, "cache_tensor_bytes": null, "serialized_payload_bytes": 912, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 76121.53695872985, "hops": {"1": 22603.971540927887, "2": 18023.20370892994, "3": 17767.595666926354, "4": 17726.766041945666}, "receiver_prefill_ms": 324.1038750857115, "ttft_ms": 45.43679195921868, "decode_ms": 401.51574998162687}, "total_wall_ms": 76893.1934167631, "ttft_ms": 45.43679195921868, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 7}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "8908553", "evidence_artifact": {"ledger": {"facts_total": 64, "facts_retained": 30, "retention_rate": 0.46875, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.6666666666666666, "status": 0.9166666666666666, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 912}, "ledger_by_hop": {"1": {"facts_total": 64, "facts_retained": 30, "retention_rate": 0.46875, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.6666666666666666, "status": 0.9166666666666666, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1087}, "2": {"facts_total": 64, "facts_retained": 30, "retention_rate": 0.46875, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.6666666666666666, "status": 0.9166666666666666, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 912}, "3": {"facts_total": 64, "facts_retained": 30, "retention_rate": 0.46875, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.6666666666666666, "status": 0.9166666666666666, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 912}, "4": {"facts_total": 64, "facts_retained": 30, "retention_rate": 0.46875, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.6666666666666666, "status": 0.9166666666666666, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 912}}}, "reference_key_id": "keys/synthetic/syn_03006_2000.json", "task_scores": {"correct": 0.0, "abstained": 0.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "060968d99413307af0515a687c7ce77ff5260c81e4a1b8709ace3f1df2e644b7", "output_hash": "5ad9e777dc3be36808c74badfc8aae217b41a486754d07adf1abda6f6bce9aca"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "b15b0eb19a02ee372bc8dda7", "attempt_id": "ecd4c942c411", "run_group_id": "E4_core:A4|closed|qwen3b:h4", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T06:57:12.858477Z", "task_family": "synthetic", "document_id": "syn_03006_2000", "document_hash": "89be5e77e0ca114963e9ffdf8dc38fe5610f13a78ac29e7a1956cd9124ce5217", "question_id": "q12_unit_conversion", "split": "test", "cluster_id": "syn_03006_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 4, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1890, "query_tokens": 504, "message_tokens": 414, "message_utf8_bytes": 912, "cache_tensor_bytes": null, "serialized_payload_bytes": 912, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 76121.53695872985, "hops": {"1": 22603.971540927887, "2": 18023.20370892994, "3": 17767.595666926354, "4": 17726.766041945666}, "receiver_prefill_ms": 325.50216710660607, "ttft_ms": 39.46933301631361, "decode_ms": 174.4388750521466}, "total_wall_ms": 76661.50345874485, "ttft_ms": 39.46933301631361, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 5}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "39700", "evidence_artifact": {"ledger": {"facts_total": 64, "facts_retained": 30, "retention_rate": 0.46875, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.6666666666666666, "status": 0.9166666666666666, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 912}, "ledger_by_hop": {"1": {"facts_total": 64, "facts_retained": 30, "retention_rate": 0.46875, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.6666666666666666, "status": 0.9166666666666666, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1087}, "2": {"facts_total": 64, "facts_retained": 30, "retention_rate": 0.46875, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.6666666666666666, "status": 0.9166666666666666, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 912}, "3": {"facts_total": 64, "facts_retained": 30, "retention_rate": 0.46875, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.6666666666666666, "status": 0.9166666666666666, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 912}, "4": {"facts_total": 64, "facts_retained": 30, "retention_rate": 0.46875, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.6666666666666666, "status": 0.9166666666666666, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 912}}}, "reference_key_id": "keys/synthetic/syn_03006_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "060968d99413307af0515a687c7ce77ff5260c81e4a1b8709ace3f1df2e644b7", "output_hash": "d8f3f4a352330800cb9f3d838e47d965e9c710d166f53c6b3724fdeae73c9a5e"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "1183f8e658ec41e75773aec4", "attempt_id": "03a0b3f9b908", "run_group_id": "E4_core:A4|closed|qwen3b:h4", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T06:57:13.415722Z", "task_family": "synthetic", "document_id": "syn_03006_2000", "document_hash": "89be5e77e0ca114963e9ffdf8dc38fe5610f13a78ac29e7a1956cd9124ce5217", "question_id": "q13_unanswerable", "split": "test", "cluster_id": "syn_03006_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 4, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1890, "query_tokens": 508, "message_tokens": 414, "message_utf8_bytes": 912, "cache_tensor_bytes": null, "serialized_payload_bytes": 912, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 76121.53695872985, "hops": {"1": 22603.971540927887, "2": 18023.20370892994, "3": 17767.595666926354, "4": 17726.766041945666}, "receiver_prefill_ms": 295.99445895291865, "ttft_ms": 56.739790947176516, "decode_ms": 57.57108307443559}, "total_wall_ms": 76532.43950067554, "ttft_ms": 56.739790947176516, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 64, "facts_retained": 30, "retention_rate": 0.46875, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.6666666666666666, "status": 0.9166666666666666, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 912}, "ledger_by_hop": {"1": {"facts_total": 64, "facts_retained": 30, "retention_rate": 0.46875, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.6666666666666666, "status": 0.9166666666666666, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1087}, "2": {"facts_total": 64, "facts_retained": 30, "retention_rate": 0.46875, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.6666666666666666, "status": 0.9166666666666666, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 912}, "3": {"facts_total": 64, "facts_retained": 30, "retention_rate": 0.46875, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.6666666666666666, "status": 0.9166666666666666, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 912}, "4": {"facts_total": 64, "facts_retained": 30, "retention_rate": 0.46875, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.6666666666666666, "status": 0.9166666666666666, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 912}}}, "reference_key_id": "keys/synthetic/syn_03006_2000.json", "task_scores": {"correct": 1.0, "abstained": 1.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "060968d99413307af0515a687c7ce77ff5260c81e4a1b8709ace3f1df2e644b7", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "fa14b60d92f35236b2b33de6", "attempt_id": "1f28b7a8eeff", "run_group_id": "E4_core:A4|closed|qwen3b:h4", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T06:57:13.843908Z", "task_family": "synthetic", "document_id": "syn_03006_2000", "document_hash": "89be5e77e0ca114963e9ffdf8dc38fe5610f13a78ac29e7a1956cd9124ce5217", "question_id": "q14_unanswerable", "split": "test", "cluster_id": "syn_03006_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 4, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1890, "query_tokens": 509, "message_tokens": 414, "message_utf8_bytes": 912, "cache_tensor_bytes": null, "serialized_payload_bytes": 912, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 76121.53695872985, "hops": {"1": 22603.971540927887, "2": 18023.20370892994, "3": 17767.595666926354, "4": 17726.766041945666}, "receiver_prefill_ms": 492.88991594221443, "ttft_ms": 32.576250028796494, "decode_ms": 31.358374981209636}, "total_wall_ms": 76678.89629269484, "ttft_ms": 32.576250028796494, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 64, "facts_retained": 30, "retention_rate": 0.46875, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.6666666666666666, "status": 0.9166666666666666, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 912}, "ledger_by_hop": {"1": {"facts_total": 64, "facts_retained": 30, "retention_rate": 0.46875, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.6666666666666666, "status": 0.9166666666666666, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1087}, "2": {"facts_total": 64, "facts_retained": 30, "retention_rate": 0.46875, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.6666666666666666, "status": 0.9166666666666666, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 912}, "3": {"facts_total": 64, "facts_retained": 30, "retention_rate": 0.46875, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.6666666666666666, "status": 0.9166666666666666, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 912}, "4": {"facts_total": 64, "facts_retained": 30, "retention_rate": 0.46875, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.6666666666666666, "status": 0.9166666666666666, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 912}}}, "reference_key_id": "keys/synthetic/syn_03006_2000.json", "task_scores": {"correct": 1.0, "abstained": 1.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "060968d99413307af0515a687c7ce77ff5260c81e4a1b8709ace3f1df2e644b7", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "274db3dcd3578f71bb378ac8", "attempt_id": "beac4400b0ae", "run_group_id": "E4_core:A4|closed|qwen3b:h8", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T06:58:25.687333Z", "task_family": "synthetic", "document_id": "syn_03006_2000", "document_hash": "89be5e77e0ca114963e9ffdf8dc38fe5610f13a78ac29e7a1956cd9124ce5217", "question_id": "q00_single_fact", "split": "test", "cluster_id": "syn_03006_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 8, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1890, "query_tokens": 496, "message_tokens": 414, "message_utf8_bytes": 912, "cache_tensor_bytes": null, "serialized_payload_bytes": 912, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 147384.49412677437, "hops": {"1": 22603.971540927887, "2": 18023.20370892994, "3": 17767.595666926354, "4": 17726.766041945666, "5": 17904.166499967687, "6": 17769.70225002151, "7": 17855.234208982438, "8": 17733.854209072888}, "receiver_prefill_ms": 325.15962491743267, "ttft_ms": 52.97195794992149, "decode_ms": 579.888042062521}, "total_wall_ms": 148343.21754379198, "ttft_ms": 52.97195794992149, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 10}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "XM-76Q-2958", "evidence_artifact": {"ledger": {"facts_total": 64, "facts_retained": 30, "retention_rate": 0.46875, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.6666666666666666, "status": 0.9166666666666666, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 912}, "ledger_by_hop": {"1": {"facts_total": 64, "facts_retained": 30, "retention_rate": 0.46875, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.6666666666666666, "status": 0.9166666666666666, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1087}, "2": {"facts_total": 64, "facts_retained": 30, "retention_rate": 0.46875, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.6666666666666666, "status": 0.9166666666666666, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 912}, "3": {"facts_total": 64, "facts_retained": 30, "retention_rate": 0.46875, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.6666666666666666, "status": 0.9166666666666666, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 912}, "4": {"facts_total": 64, "facts_retained": 30, "retention_rate": 0.46875, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.6666666666666666, "status": 0.9166666666666666, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 912}, "5": {"facts_total": 64, "facts_retained": 30, "retention_rate": 0.46875, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.6666666666666666, "status": 0.9166666666666666, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 912}, "6": {"facts_total": 64, "facts_retained": 30, "retention_rate": 0.46875, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.6666666666666666, "status": 0.9166666666666666, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 912}, "7": {"facts_total": 64, "facts_retained": 30, "retention_rate": 0.46875, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.6666666666666666, "status": 0.9166666666666666, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 912}, "8": {"facts_total": 64, "facts_retained": 30, "retention_rate": 0.46875, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.6666666666666666, "status": 0.9166666666666666, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 912}}}, "reference_key_id": "keys/synthetic/syn_03006_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "060968d99413307af0515a687c7ce77ff5260c81e4a1b8709ace3f1df2e644b7", "output_hash": "cf42691eec772b9ec74d67643944de2e19e53724d7b772cef1e0f2d5ef3b4926"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "bc6c344739e8c83ad46ee120", "attempt_id": "d78553e66827", "run_group_id": "E4_core:A4|closed|qwen3b:h8", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T06:58:26.663669Z", "task_family": "synthetic", "document_id": "syn_03006_2000", "document_hash": "89be5e77e0ca114963e9ffdf8dc38fe5610f13a78ac29e7a1956cd9124ce5217", "question_id": "q01_single_fact", "split": "test", "cluster_id": "syn_03006_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 8, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1890, "query_tokens": 501, "message_tokens": 414, "message_utf8_bytes": 912, "cache_tensor_bytes": null, "serialized_payload_bytes": 912, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 147384.49412677437, "hops": {"1": 22603.971540927887, "2": 18023.20370892994, "3": 17767.595666926354, "4": 17726.766041945666, "5": 17904.166499967687, "6": 17769.70225002151, "7": 17855.234208982438, "8": 17733.854209072888}, "receiver_prefill_ms": 310.79341599252075, "ttft_ms": 37.20220807008445, "decode_ms": 25.231125066056848}, "total_wall_ms": 147758.34354385734, "ttft_ms": 37.20220807008445, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 64, "facts_retained": 30, "retention_rate": 0.46875, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.6666666666666666, "status": 0.9166666666666666, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 912}, "ledger_by_hop": {"1": {"facts_total": 64, "facts_retained": 30, "retention_rate": 0.46875, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.6666666666666666, "status": 0.9166666666666666, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1087}, "2": {"facts_total": 64, "facts_retained": 30, "retention_rate": 0.46875, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.6666666666666666, "status": 0.9166666666666666, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 912}, "3": {"facts_total": 64, "facts_retained": 30, "retention_rate": 0.46875, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.6666666666666666, "status": 0.9166666666666666, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 912}, "4": {"facts_total": 64, "facts_retained": 30, "retention_rate": 0.46875, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.6666666666666666, "status": 0.9166666666666666, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 912}, "5": {"facts_total": 64, "facts_retained": 30, "retention_rate": 0.46875, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.6666666666666666, "status": 0.9166666666666666, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 912}, "6": {"facts_total": 64, "facts_retained": 30, "retention_rate": 0.46875, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.6666666666666666, "status": 0.9166666666666666, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 912}, "7": {"facts_total": 64, "facts_retained": 30, "retention_rate": 0.46875, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.6666666666666666, "status": 0.9166666666666666, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 912}, "8": {"facts_total": 64, "facts_retained": 30, "retention_rate": 0.46875, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.6666666666666666, "status": 0.9166666666666666, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 912}}}, "reference_key_id": "keys/synthetic/syn_03006_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "060968d99413307af0515a687c7ce77ff5260c81e4a1b8709ace3f1df2e644b7", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "b25ec7a0bf991779f2a6c442", "attempt_id": "97e6d8a8566f", "run_group_id": "E4_core:A4|closed|qwen3b:h8", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T06:58:27.054548Z", "task_family": "synthetic", "document_id": "syn_03006_2000", "document_hash": "89be5e77e0ca114963e9ffdf8dc38fe5610f13a78ac29e7a1956cd9124ce5217", "question_id": "q02_exact_copy", "split": "test", "cluster_id": "syn_03006_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 8, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1890, "query_tokens": 500, "message_tokens": 414, "message_utf8_bytes": 912, "cache_tensor_bytes": null, "serialized_payload_bytes": 912, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 147384.49412677437, "hops": {"1": 22603.971540927887, "2": 18023.20370892994, "3": 17767.595666926354, "4": 17726.766041945666, "5": 17904.166499967687, "6": 17769.70225002151, "7": 17855.234208982438, "8": 17733.854209072888}, "receiver_prefill_ms": 324.2020830512047, "ttft_ms": 37.7116670133546, "decode_ms": 21.381917060352862}, "total_wall_ms": 147768.31250172108, "ttft_ms": 37.7116670133546, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 64, "facts_retained": 30, "retention_rate": 0.46875, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.6666666666666666, "status": 0.9166666666666666, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 912}, "ledger_by_hop": {"1": {"facts_total": 64, "facts_retained": 30, "retention_rate": 0.46875, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.6666666666666666, "status": 0.9166666666666666, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1087}, "2": {"facts_total": 64, "facts_retained": 30, "retention_rate": 0.46875, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.6666666666666666, "status": 0.9166666666666666, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 912}, "3": {"facts_total": 64, "facts_retained": 30, "retention_rate": 0.46875, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.6666666666666666, "status": 0.9166666666666666, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 912}, "4": {"facts_total": 64, "facts_retained": 30, "retention_rate": 0.46875, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.6666666666666666, "status": 0.9166666666666666, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 912}, "5": {"facts_total": 64, "facts_retained": 30, "retention_rate": 0.46875, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.6666666666666666, "status": 0.9166666666666666, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 912}, "6": {"facts_total": 64, "facts_retained": 30, "retention_rate": 0.46875, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.6666666666666666, "status": 0.9166666666666666, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 912}, "7": {"facts_total": 64, "facts_retained": 30, "retention_rate": 0.46875, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.6666666666666666, "status": 0.9166666666666666, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 912}, "8": {"facts_total": 64, "facts_retained": 30, "retention_rate": 0.46875, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.6666666666666666, "status": 0.9166666666666666, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 912}}}, "reference_key_id": "keys/synthetic/syn_03006_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "060968d99413307af0515a687c7ce77ff5260c81e4a1b8709ace3f1df2e644b7", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "7c8307a037f3bbc50b9353e7", "attempt_id": "8293ea1ac315", "run_group_id": "E4_core:A4|closed|qwen3b:h8", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T06:58:27.454890Z", "task_family": "synthetic", "document_id": "syn_03006_2000", "document_hash": "89be5e77e0ca114963e9ffdf8dc38fe5610f13a78ac29e7a1956cd9124ce5217", "question_id": "q03_exact_copy", "split": "test", "cluster_id": "syn_03006_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 8, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1890, "query_tokens": 495, "message_tokens": 414, "message_utf8_bytes": 912, "cache_tensor_bytes": null, "serialized_payload_bytes": 912, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 147384.49412677437, "hops": {"1": 22603.971540927887, "2": 18023.20370892994, "3": 17767.595666926354, "4": 17726.766041945666, "5": 17904.166499967687, "6": 17769.70225002151, "7": 17855.234208982438, "8": 17733.854209072888}, "receiver_prefill_ms": 292.80149994883686, "ttft_ms": 40.866625029593706, "decode_ms": 497.156040975824}, "total_wall_ms": 148215.89708479587, "ttft_ms": 40.866625029593706, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 8}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "Haruki Ouellet-Marchand", "evidence_artifact": {"ledger": {"facts_total": 64, "facts_retained": 30, "retention_rate": 0.46875, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.6666666666666666, "status": 0.9166666666666666, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 912}, "ledger_by_hop": {"1": {"facts_total": 64, "facts_retained": 30, "retention_rate": 0.46875, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.6666666666666666, "status": 0.9166666666666666, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1087}, "2": {"facts_total": 64, "facts_retained": 30, "retention_rate": 0.46875, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.6666666666666666, "status": 0.9166666666666666, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 912}, "3": {"facts_total": 64, "facts_retained": 30, "retention_rate": 0.46875, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.6666666666666666, "status": 0.9166666666666666, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 912}, "4": {"facts_total": 64, "facts_retained": 30, "retention_rate": 0.46875, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.6666666666666666, "status": 0.9166666666666666, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 912}, "5": {"facts_total": 64, "facts_retained": 30, "retention_rate": 0.46875, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.6666666666666666, "status": 0.9166666666666666, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 912}, "6": {"facts_total": 64, "facts_retained": 30, "retention_rate": 0.46875, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.6666666666666666, "status": 0.9166666666666666, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 912}, "7": {"facts_total": 64, "facts_retained": 30, "retention_rate": 0.46875, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.6666666666666666, "status": 0.9166666666666666, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 912}, "8": {"facts_total": 64, "facts_retained": 30, "retention_rate": 0.46875, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.6666666666666666, "status": 0.9166666666666666, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 912}}}, "reference_key_id": "keys/synthetic/syn_03006_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "060968d99413307af0515a687c7ce77ff5260c81e4a1b8709ace3f1df2e644b7", "output_hash": "a3b6e0a44d495387d8b8fc50ba6be0f2ed43121c674d3b3e962f8c4690e15b5c"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "a45d320cb6251e0537fbdd6b", "attempt_id": "b98deabd7c12", "run_group_id": "E4_core:A4|closed|qwen3b:h8", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T06:58:28.305111Z", "task_family": "synthetic", "document_id": "syn_03006_2000", "document_hash": "89be5e77e0ca114963e9ffdf8dc38fe5610f13a78ac29e7a1956cd9124ce5217", "question_id": "q04_multi_hop", "split": "test", "cluster_id": "syn_03006_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 8, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1890, "query_tokens": 504, "message_tokens": 414, "message_utf8_bytes": 912, "cache_tensor_bytes": null, "serialized_payload_bytes": 912, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 147384.49412677437, "hops": {"1": 22603.971540927887, "2": 18023.20370892994, "3": 17767.595666926354, "4": 17726.766041945666, "5": 17904.166499967687, "6": 17769.70225002151, "7": 17855.234208982438, "8": 17733.854209072888}, "receiver_prefill_ms": 342.7441669628024, "ttft_ms": 22.085083997808397, "decode_ms": 44.91841699928045}, "total_wall_ms": 147794.77429378312, "ttft_ms": 22.085083997808397, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 64, "facts_retained": 30, "retention_rate": 0.46875, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.6666666666666666, "status": 0.9166666666666666, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 912}, "ledger_by_hop": {"1": {"facts_total": 64, "facts_retained": 30, "retention_rate": 0.46875, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.6666666666666666, "status": 0.9166666666666666, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1087}, "2": {"facts_total": 64, "facts_retained": 30, "retention_rate": 0.46875, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.6666666666666666, "status": 0.9166666666666666, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 912}, "3": {"facts_total": 64, "facts_retained": 30, "retention_rate": 0.46875, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.6666666666666666, "status": 0.9166666666666666, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 912}, "4": {"facts_total": 64, "facts_retained": 30, "retention_rate": 0.46875, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.6666666666666666, "status": 0.9166666666666666, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 912}, "5": {"facts_total": 64, "facts_retained": 30, "retention_rate": 0.46875, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.6666666666666666, "status": 0.9166666666666666, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 912}, "6": {"facts_total": 64, "facts_retained": 30, "retention_rate": 0.46875, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.6666666666666666, "status": 0.9166666666666666, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 912}, "7": {"facts_total": 64, "facts_retained": 30, "retention_rate": 0.46875, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.6666666666666666, "status": 0.9166666666666666, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 912}, "8": {"facts_total": 64, "facts_retained": 30, "retention_rate": 0.46875, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.6666666666666666, "status": 0.9166666666666666, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 912}}}, "reference_key_id": "keys/synthetic/syn_03006_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "060968d99413307af0515a687c7ce77ff5260c81e4a1b8709ace3f1df2e644b7", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "423d55301313ae23fa0e78a0", "attempt_id": "e0f1731a33b6", "run_group_id": "E4_core:A4|closed|qwen3b:h8", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T06:58:28.731876Z", "task_family": "synthetic", "document_id": "syn_03006_2000", "document_hash": "89be5e77e0ca114963e9ffdf8dc38fe5610f13a78ac29e7a1956cd9124ce5217", "question_id": "q05_multi_hop", "split": "test", "cluster_id": "syn_03006_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 8, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1890, "query_tokens": 507, "message_tokens": 414, "message_utf8_bytes": 912, "cache_tensor_bytes": null, "serialized_payload_bytes": 912, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 147384.49412677437, "hops": {"1": 22603.971540927887, "2": 18023.20370892994, "3": 17767.595666926354, "4": 17726.766041945666, "5": 17904.166499967687, "6": 17769.70225002151, "7": 17855.234208982438, "8": 17733.854209072888}, "receiver_prefill_ms": 323.9122920203954, "ttft_ms": 20.28887497726828, "decode_ms": 605.5855420418084}, "total_wall_ms": 148334.82654381078, "ttft_ms": 20.28887497726828, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 10}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "EP-52G-1330", "evidence_artifact": {"ledger": {"facts_total": 64, "facts_retained": 30, "retention_rate": 0.46875, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.6666666666666666, "status": 0.9166666666666666, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 912}, "ledger_by_hop": {"1": {"facts_total": 64, "facts_retained": 30, "retention_rate": 0.46875, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.6666666666666666, "status": 0.9166666666666666, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1087}, "2": {"facts_total": 64, "facts_retained": 30, "retention_rate": 0.46875, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.6666666666666666, "status": 0.9166666666666666, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 912}, "3": {"facts_total": 64, "facts_retained": 30, "retention_rate": 0.46875, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.6666666666666666, "status": 0.9166666666666666, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 912}, "4": {"facts_total": 64, "facts_retained": 30, "retention_rate": 0.46875, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.6666666666666666, "status": 0.9166666666666666, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 912}, "5": {"facts_total": 64, "facts_retained": 30, "retention_rate": 0.46875, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.6666666666666666, "status": 0.9166666666666666, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 912}, "6": {"facts_total": 64, "facts_retained": 30, "retention_rate": 0.46875, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.6666666666666666, "status": 0.9166666666666666, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 912}, "7": {"facts_total": 64, "facts_retained": 30, "retention_rate": 0.46875, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.6666666666666666, "status": 0.9166666666666666, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 912}, "8": {"facts_total": 64, "facts_retained": 30, "retention_rate": 0.46875, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.6666666666666666, "status": 0.9166666666666666, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 912}}}, "reference_key_id": "keys/synthetic/syn_03006_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "060968d99413307af0515a687c7ce77ff5260c81e4a1b8709ace3f1df2e644b7", "output_hash": "104db207937fd79e23ce16f9b3f10b1a9ed34f6dd0441a99be6ea1b2d0a09d99"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "b6b2151b67936bf6aec67715", "attempt_id": "6b2f12e90618", "run_group_id": "E4_core:A4|closed|qwen3b:h8", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T06:58:29.699125Z", "task_family": "synthetic", "document_id": "syn_03006_2000", "document_hash": "89be5e77e0ca114963e9ffdf8dc38fe5610f13a78ac29e7a1956cd9124ce5217", "question_id": "q06_negation", "split": "test", "cluster_id": "syn_03006_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 8, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1890, "query_tokens": 495, "message_tokens": 414, "message_utf8_bytes": 912, "cache_tensor_bytes": null, "serialized_payload_bytes": 912, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 147384.49412677437, "hops": {"1": 22603.971540927887, "2": 18023.20370892994, "3": 17767.595666926354, "4": 17726.766041945666, "5": 17904.166499967687, "6": 17769.70225002151, "7": 17855.234208982438, "8": 17733.854209072888}, "receiver_prefill_ms": 318.543832981959, "ttft_ms": 40.678916964679956, "decode_ms": 25.288083008490503}, "total_wall_ms": 147769.63141874876, "ttft_ms": 40.678916964679956, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "no", "evidence_artifact": {"ledger": {"facts_total": 64, "facts_retained": 30, "retention_rate": 0.46875, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.6666666666666666, "status": 0.9166666666666666, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 912}, "ledger_by_hop": {"1": {"facts_total": 64, "facts_retained": 30, "retention_rate": 0.46875, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.6666666666666666, "status": 0.9166666666666666, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1087}, "2": {"facts_total": 64, "facts_retained": 30, "retention_rate": 0.46875, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.6666666666666666, "status": 0.9166666666666666, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 912}, "3": {"facts_total": 64, "facts_retained": 30, "retention_rate": 0.46875, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.6666666666666666, "status": 0.9166666666666666, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 912}, "4": {"facts_total": 64, "facts_retained": 30, "retention_rate": 0.46875, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.6666666666666666, "status": 0.9166666666666666, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 912}, "5": {"facts_total": 64, "facts_retained": 30, "retention_rate": 0.46875, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.6666666666666666, "status": 0.9166666666666666, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 912}, "6": {"facts_total": 64, "facts_retained": 30, "retention_rate": 0.46875, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.6666666666666666, "status": 0.9166666666666666, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 912}, "7": {"facts_total": 64, "facts_retained": 30, "retention_rate": 0.46875, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.6666666666666666, "status": 0.9166666666666666, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 912}, "8": {"facts_total": 64, "facts_retained": 30, "retention_rate": 0.46875, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.6666666666666666, "status": 0.9166666666666666, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 912}}}, "reference_key_id": "keys/synthetic/syn_03006_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "060968d99413307af0515a687c7ce77ff5260c81e4a1b8709ace3f1df2e644b7", "output_hash": "9390298f3fb0c5b160498935d79cb139aef28e1c47358b4bbba61862b9c26e59"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "b23bb09f1ebf2e48d7ebe8a7", "attempt_id": "53eeeb2f1a1a", "run_group_id": "E4_core:A4|closed|qwen3b:h8", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T06:58:30.101388Z", "task_family": "synthetic", "document_id": "syn_03006_2000", "document_hash": "89be5e77e0ca114963e9ffdf8dc38fe5610f13a78ac29e7a1956cd9124ce5217", "question_id": "q07_negation", "split": "test", "cluster_id": "syn_03006_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 8, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1890, "query_tokens": 495, "message_tokens": 414, "message_utf8_bytes": 912, "cache_tensor_bytes": null, "serialized_payload_bytes": 912, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 147384.49412677437, "hops": {"1": 22603.971540927887, "2": 18023.20370892994, "3": 17767.595666926354, "4": 17726.766041945666, "5": 17904.166499967687, "6": 17769.70225002151, "7": 17855.234208982438, "8": 17733.854209072888}, "receiver_prefill_ms": 320.8628749707714, "ttft_ms": 38.27300004195422, "decode_ms": 21.282417001202703}, "total_wall_ms": 147765.39083581883, "ttft_ms": 38.27300004195422, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "no", "evidence_artifact": {"ledger": {"facts_total": 64, "facts_retained": 30, "retention_rate": 0.46875, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.6666666666666666, "status": 0.9166666666666666, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 912}, "ledger_by_hop": {"1": {"facts_total": 64, "facts_retained": 30, "retention_rate": 0.46875, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.6666666666666666, "status": 0.9166666666666666, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1087}, "2": {"facts_total": 64, "facts_retained": 30, "retention_rate": 0.46875, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.6666666666666666, "status": 0.9166666666666666, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 912}, "3": {"facts_total": 64, "facts_retained": 30, "retention_rate": 0.46875, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.6666666666666666, "status": 0.9166666666666666, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 912}, "4": {"facts_total": 64, "facts_retained": 30, "retention_rate": 0.46875, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.6666666666666666, "status": 0.9166666666666666, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 912}, "5": {"facts_total": 64, "facts_retained": 30, "retention_rate": 0.46875, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.6666666666666666, "status": 0.9166666666666666, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 912}, "6": {"facts_total": 64, "facts_retained": 30, "retention_rate": 0.46875, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.6666666666666666, "status": 0.9166666666666666, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 912}, "7": {"facts_total": 64, "facts_retained": 30, "retention_rate": 0.46875, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.6666666666666666, "status": 0.9166666666666666, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 912}, "8": {"facts_total": 64, "facts_retained": 30, "retention_rate": 0.46875, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.6666666666666666, "status": 0.9166666666666666, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 912}}}, "reference_key_id": "keys/synthetic/syn_03006_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "060968d99413307af0515a687c7ce77ff5260c81e4a1b8709ace3f1df2e644b7", "output_hash": "9390298f3fb0c5b160498935d79cb139aef28e1c47358b4bbba61862b9c26e59"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "8ff22b18ea7e4cbdf67c6138", "attempt_id": "6c010aa1efa9", "run_group_id": "E4_core:A4|closed|qwen3b:h8", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T06:58:30.498890Z", "task_family": "synthetic", "document_id": "syn_03006_2000", "document_hash": "89be5e77e0ca114963e9ffdf8dc38fe5610f13a78ac29e7a1956cd9124ce5217", "question_id": "q08_revision", "split": "test", "cluster_id": "syn_03006_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 8, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1890, "query_tokens": 512, "message_tokens": 414, "message_utf8_bytes": 912, "cache_tensor_bytes": null, "serialized_payload_bytes": 912, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 147384.49412677437, "hops": {"1": 22603.971540927887, "2": 18023.20370892994, "3": 17767.595666926354, "4": 17726.766041945666, "5": 17904.166499967687, "6": 17769.70225002151, "7": 17855.234208982438, "8": 17733.854209072888}, "receiver_prefill_ms": 297.8832090739161, "ttft_ms": 42.39929106552154, "decode_ms": 477.57662495132536}, "total_wall_ms": 148202.9593347106, "ttft_ms": 42.39929106552154, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 7}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "9112735", "evidence_artifact": {"ledger": {"facts_total": 64, "facts_retained": 30, "retention_rate": 0.46875, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.6666666666666666, "status": 0.9166666666666666, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 912}, "ledger_by_hop": {"1": {"facts_total": 64, "facts_retained": 30, "retention_rate": 0.46875, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.6666666666666666, "status": 0.9166666666666666, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1087}, "2": {"facts_total": 64, "facts_retained": 30, "retention_rate": 0.46875, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.6666666666666666, "status": 0.9166666666666666, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 912}, "3": {"facts_total": 64, "facts_retained": 30, "retention_rate": 0.46875, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.6666666666666666, "status": 0.9166666666666666, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 912}, "4": {"facts_total": 64, "facts_retained": 30, "retention_rate": 0.46875, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.6666666666666666, "status": 0.9166666666666666, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 912}, "5": {"facts_total": 64, "facts_retained": 30, "retention_rate": 0.46875, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.6666666666666666, "status": 0.9166666666666666, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 912}, "6": {"facts_total": 64, "facts_retained": 30, "retention_rate": 0.46875, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.6666666666666666, "status": 0.9166666666666666, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 912}, "7": {"facts_total": 64, "facts_retained": 30, "retention_rate": 0.46875, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.6666666666666666, "status": 0.9166666666666666, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 912}, "8": {"facts_total": 64, "facts_retained": 30, "retention_rate": 0.46875, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.6666666666666666, "status": 0.9166666666666666, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 912}}}, "reference_key_id": "keys/synthetic/syn_03006_2000.json", "task_scores": {"correct": 0.0, "abstained": 0.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "060968d99413307af0515a687c7ce77ff5260c81e4a1b8709ace3f1df2e644b7", "output_hash": "15a3f804636f0e62d4a3d9be7677f001a0ad6b1fb8ae7e8cdc63c44ec7bac31f"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "e47ce7c0827b55a95a803a6d", "attempt_id": "0588a258d761", "run_group_id": "E4_core:A4|closed|qwen3b:h8", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T06:58:31.334490Z", "task_family": "synthetic", "document_id": "syn_03006_2000", "document_hash": "89be5e77e0ca114963e9ffdf8dc38fe5610f13a78ac29e7a1956cd9124ce5217", "question_id": "q09_revision", "split": "test", "cluster_id": "syn_03006_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 8, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1890, "query_tokens": 505, "message_tokens": 414, "message_utf8_bytes": 912, "cache_tensor_bytes": null, "serialized_payload_bytes": 912, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 147384.49412677437, "hops": {"1": 22603.971540927887, "2": 18023.20370892994, "3": 17767.595666926354, "4": 17726.766041945666, "5": 17904.166499967687, "6": 17769.70225002151, "7": 17855.234208982438, "8": 17733.854209072888}, "receiver_prefill_ms": 327.0507500274107, "ttft_ms": 38.875582977198064, "decode_ms": 296.43283400218934}, "total_wall_ms": 148047.39987675566, "ttft_ms": 38.875582977198064, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 9}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "9,112,735", "evidence_artifact": {"ledger": {"facts_total": 64, "facts_retained": 30, "retention_rate": 0.46875, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.6666666666666666, "status": 0.9166666666666666, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 912}, "ledger_by_hop": {"1": {"facts_total": 64, "facts_retained": 30, "retention_rate": 0.46875, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.6666666666666666, "status": 0.9166666666666666, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1087}, "2": {"facts_total": 64, "facts_retained": 30, "retention_rate": 0.46875, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.6666666666666666, "status": 0.9166666666666666, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 912}, "3": {"facts_total": 64, "facts_retained": 30, "retention_rate": 0.46875, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.6666666666666666, "status": 0.9166666666666666, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 912}, "4": {"facts_total": 64, "facts_retained": 30, "retention_rate": 0.46875, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.6666666666666666, "status": 0.9166666666666666, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 912}, "5": {"facts_total": 64, "facts_retained": 30, "retention_rate": 0.46875, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.6666666666666666, "status": 0.9166666666666666, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 912}, "6": {"facts_total": 64, "facts_retained": 30, "retention_rate": 0.46875, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.6666666666666666, "status": 0.9166666666666666, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 912}, "7": {"facts_total": 64, "facts_retained": 30, "retention_rate": 0.46875, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.6666666666666666, "status": 0.9166666666666666, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 912}, "8": {"facts_total": 64, "facts_retained": 30, "retention_rate": 0.46875, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.6666666666666666, "status": 0.9166666666666666, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 912}}}, "reference_key_id": "keys/synthetic/syn_03006_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "060968d99413307af0515a687c7ce77ff5260c81e4a1b8709ace3f1df2e644b7", "output_hash": "788fcda908c2029597420995b09d0cc919bfdbf7a28b68dcb0e331db7df7b342"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "f038d98412d3f0ca3e4efb81", "attempt_id": "a611cf9d05e6", "run_group_id": "E4_core:A4|closed|qwen3b:h8", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T06:58:32.014456Z", "task_family": "synthetic", "document_id": "syn_03006_2000", "document_hash": "89be5e77e0ca114963e9ffdf8dc38fe5610f13a78ac29e7a1956cd9124ce5217", "question_id": "q10_temporal", "split": "test", "cluster_id": "syn_03006_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 8, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1890, "query_tokens": 514, "message_tokens": 414, "message_utf8_bytes": 912, "cache_tensor_bytes": null, "serialized_payload_bytes": 912, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 147384.49412677437, "hops": {"1": 22603.971540927887, "2": 18023.20370892994, "3": 17767.595666926354, "4": 17726.766041945666, "5": 17904.166499967687, "6": 17769.70225002151, "7": 17855.234208982438, "8": 17733.854209072888}, "receiver_prefill_ms": 373.8036249997094, "ttft_ms": 64.32495801709592, "decode_ms": 52.63945902697742}, "total_wall_ms": 147875.85808581207, "ttft_ms": 64.32495801709592, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 64, "facts_retained": 30, "retention_rate": 0.46875, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.6666666666666666, "status": 0.9166666666666666, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 912}, "ledger_by_hop": {"1": {"facts_total": 64, "facts_retained": 30, "retention_rate": 0.46875, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.6666666666666666, "status": 0.9166666666666666, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1087}, "2": {"facts_total": 64, "facts_retained": 30, "retention_rate": 0.46875, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.6666666666666666, "status": 0.9166666666666666, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 912}, "3": {"facts_total": 64, "facts_retained": 30, "retention_rate": 0.46875, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.6666666666666666, "status": 0.9166666666666666, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 912}, "4": {"facts_total": 64, "facts_retained": 30, "retention_rate": 0.46875, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.6666666666666666, "status": 0.9166666666666666, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 912}, "5": {"facts_total": 64, "facts_retained": 30, "retention_rate": 0.46875, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.6666666666666666, "status": 0.9166666666666666, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 912}, "6": {"facts_total": 64, "facts_retained": 30, "retention_rate": 0.46875, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.6666666666666666, "status": 0.9166666666666666, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 912}, "7": {"facts_total": 64, "facts_retained": 30, "retention_rate": 0.46875, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.6666666666666666, "status": 0.9166666666666666, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 912}, "8": {"facts_total": 64, "facts_retained": 30, "retention_rate": 0.46875, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.6666666666666666, "status": 0.9166666666666666, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 912}}}, "reference_key_id": "keys/synthetic/syn_03006_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "060968d99413307af0515a687c7ce77ff5260c81e4a1b8709ace3f1df2e644b7", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "7eeac140001fe5f2e6cad530", "attempt_id": "72e8b51aa883", "run_group_id": "E4_core:A4|closed|qwen3b:h8", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T06:58:32.523096Z", "task_family": "synthetic", "document_id": "syn_03006_2000", "document_hash": "89be5e77e0ca114963e9ffdf8dc38fe5610f13a78ac29e7a1956cd9124ce5217", "question_id": "q11_aggregation", "split": "test", "cluster_id": "syn_03006_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 8, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1890, "query_tokens": 507, "message_tokens": 414, "message_utf8_bytes": 912, "cache_tensor_bytes": null, "serialized_payload_bytes": 912, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 147384.49412677437, "hops": {"1": 22603.971540927887, "2": 18023.20370892994, "3": 17767.595666926354, "4": 17726.766041945666, "5": 17904.166499967687, "6": 17769.70225002151, "7": 17855.234208982438, "8": 17733.854209072888}, "receiver_prefill_ms": 477.85141703207046, "ttft_ms": 42.20254090614617, "decode_ms": 255.63879206310958}, "total_wall_ms": 148161.00633481983, "ttft_ms": 42.20254090614617, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 7}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "8908553", "evidence_artifact": {"ledger": {"facts_total": 64, "facts_retained": 30, "retention_rate": 0.46875, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.6666666666666666, "status": 0.9166666666666666, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 912}, "ledger_by_hop": {"1": {"facts_total": 64, "facts_retained": 30, "retention_rate": 0.46875, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.6666666666666666, "status": 0.9166666666666666, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1087}, "2": {"facts_total": 64, "facts_retained": 30, "retention_rate": 0.46875, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.6666666666666666, "status": 0.9166666666666666, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 912}, "3": {"facts_total": 64, "facts_retained": 30, "retention_rate": 0.46875, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.6666666666666666, "status": 0.9166666666666666, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 912}, "4": {"facts_total": 64, "facts_retained": 30, "retention_rate": 0.46875, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.6666666666666666, "status": 0.9166666666666666, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 912}, "5": {"facts_total": 64, "facts_retained": 30, "retention_rate": 0.46875, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.6666666666666666, "status": 0.9166666666666666, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 912}, "6": {"facts_total": 64, "facts_retained": 30, "retention_rate": 0.46875, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.6666666666666666, "status": 0.9166666666666666, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 912}, "7": {"facts_total": 64, "facts_retained": 30, "retention_rate": 0.46875, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.6666666666666666, "status": 0.9166666666666666, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 912}, "8": {"facts_total": 64, "facts_retained": 30, "retention_rate": 0.46875, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.6666666666666666, "status": 0.9166666666666666, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 912}}}, "reference_key_id": "keys/synthetic/syn_03006_2000.json", "task_scores": {"correct": 0.0, "abstained": 0.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "060968d99413307af0515a687c7ce77ff5260c81e4a1b8709ace3f1df2e644b7", "output_hash": "5ad9e777dc3be36808c74badfc8aae217b41a486754d07adf1abda6f6bce9aca"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "0aeab2a2449a2af46d3528ed", "attempt_id": "21787784d575", "run_group_id": "E4_core:A4|closed|qwen3b:h8", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T06:58:33.317523Z", "task_family": "synthetic", "document_id": "syn_03006_2000", "document_hash": "89be5e77e0ca114963e9ffdf8dc38fe5610f13a78ac29e7a1956cd9124ce5217", "question_id": "q12_unit_conversion", "split": "test", "cluster_id": "syn_03006_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 8, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1890, "query_tokens": 504, "message_tokens": 414, "message_utf8_bytes": 912, "cache_tensor_bytes": null, "serialized_payload_bytes": 912, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 147384.49412677437, "hops": {"1": 22603.971540927887, "2": 18023.20370892994, "3": 17767.595666926354, "4": 17726.766041945666, "5": 17904.166499967687, "6": 17769.70225002151, "7": 17855.234208982438, "8": 17733.854209072888}, "receiver_prefill_ms": 325.03412500955164, "ttft_ms": 24.955707951448858, "decode_ms": 163.44749997369945}, "total_wall_ms": 147898.52662675548, "ttft_ms": 24.955707951448858, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 5}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "39700", "evidence_artifact": {"ledger": {"facts_total": 64, "facts_retained": 30, "retention_rate": 0.46875, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.6666666666666666, "status": 0.9166666666666666, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 912}, "ledger_by_hop": {"1": {"facts_total": 64, "facts_retained": 30, "retention_rate": 0.46875, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.6666666666666666, "status": 0.9166666666666666, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1087}, "2": {"facts_total": 64, "facts_retained": 30, "retention_rate": 0.46875, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.6666666666666666, "status": 0.9166666666666666, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 912}, "3": {"facts_total": 64, "facts_retained": 30, "retention_rate": 0.46875, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.6666666666666666, "status": 0.9166666666666666, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 912}, "4": {"facts_total": 64, "facts_retained": 30, "retention_rate": 0.46875, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.6666666666666666, "status": 0.9166666666666666, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 912}, "5": {"facts_total": 64, "facts_retained": 30, "retention_rate": 0.46875, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.6666666666666666, "status": 0.9166666666666666, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 912}, "6": {"facts_total": 64, "facts_retained": 30, "retention_rate": 0.46875, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.6666666666666666, "status": 0.9166666666666666, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 912}, "7": {"facts_total": 64, "facts_retained": 30, "retention_rate": 0.46875, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.6666666666666666, "status": 0.9166666666666666, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 912}, "8": {"facts_total": 64, "facts_retained": 30, "retention_rate": 0.46875, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.6666666666666666, "status": 0.9166666666666666, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 912}}}, "reference_key_id": "keys/synthetic/syn_03006_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "060968d99413307af0515a687c7ce77ff5260c81e4a1b8709ace3f1df2e644b7", "output_hash": "d8f3f4a352330800cb9f3d838e47d965e9c710d166f53c6b3724fdeae73c9a5e"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "ecda8457d72c97d127be9f46", "attempt_id": "44a353289e9d", "run_group_id": "E4_core:A4|closed|qwen3b:h8", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T06:58:33.848425Z", "task_family": "synthetic", "document_id": "syn_03006_2000", "document_hash": "89be5e77e0ca114963e9ffdf8dc38fe5610f13a78ac29e7a1956cd9124ce5217", "question_id": "q13_unanswerable", "split": "test", "cluster_id": "syn_03006_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 8, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1890, "query_tokens": 508, "message_tokens": 414, "message_utf8_bytes": 912, "cache_tensor_bytes": null, "serialized_payload_bytes": 912, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 147384.49412677437, "hops": {"1": 22603.971540927887, "2": 18023.20370892994, "3": 17767.595666926354, "4": 17726.766041945666, "5": 17904.166499967687, "6": 17769.70225002151, "7": 17855.234208982438, "8": 17733.854209072888}, "receiver_prefill_ms": 317.55366595461965, "ttft_ms": 59.5945839304477, "decode_ms": 57.56300000939518}, "total_wall_ms": 147819.83829278033, "ttft_ms": 59.5945839304477, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 64, "facts_retained": 30, "retention_rate": 0.46875, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.6666666666666666, "status": 0.9166666666666666, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 912}, "ledger_by_hop": {"1": {"facts_total": 64, "facts_retained": 30, "retention_rate": 0.46875, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.6666666666666666, "status": 0.9166666666666666, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1087}, "2": {"facts_total": 64, "facts_retained": 30, "retention_rate": 0.46875, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.6666666666666666, "status": 0.9166666666666666, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 912}, "3": {"facts_total": 64, "facts_retained": 30, "retention_rate": 0.46875, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.6666666666666666, "status": 0.9166666666666666, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 912}, "4": {"facts_total": 64, "facts_retained": 30, "retention_rate": 0.46875, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.6666666666666666, "status": 0.9166666666666666, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 912}, "5": {"facts_total": 64, "facts_retained": 30, "retention_rate": 0.46875, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.6666666666666666, "status": 0.9166666666666666, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 912}, "6": {"facts_total": 64, "facts_retained": 30, "retention_rate": 0.46875, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.6666666666666666, "status": 0.9166666666666666, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 912}, "7": {"facts_total": 64, "facts_retained": 30, "retention_rate": 0.46875, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.6666666666666666, "status": 0.9166666666666666, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 912}, "8": {"facts_total": 64, "facts_retained": 30, "retention_rate": 0.46875, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.6666666666666666, "status": 0.9166666666666666, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 912}}}, "reference_key_id": "keys/synthetic/syn_03006_2000.json", "task_scores": {"correct": 1.0, "abstained": 1.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "060968d99413307af0515a687c7ce77ff5260c81e4a1b8709ace3f1df2e644b7", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "57fd4d06e0c0fa241181546d", "attempt_id": "b589438c03a6", "run_group_id": "E4_core:A4|closed|qwen3b:h8", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T06:58:34.302081Z", "task_family": "synthetic", "document_id": "syn_03006_2000", "document_hash": "89be5e77e0ca114963e9ffdf8dc38fe5610f13a78ac29e7a1956cd9124ce5217", "question_id": "q14_unanswerable", "split": "test", "cluster_id": "syn_03006_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 8, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1890, "query_tokens": 509, "message_tokens": 414, "message_utf8_bytes": 912, "cache_tensor_bytes": null, "serialized_payload_bytes": 912, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 147384.49412677437, "hops": {"1": 22603.971540927887, "2": 18023.20370892994, "3": 17767.595666926354, "4": 17726.766041945666, "5": 17904.166499967687, "6": 17769.70225002151, "7": 17855.234208982438, "8": 17733.854209072888}, "receiver_prefill_ms": 456.4787920098752, "ttft_ms": 33.54645904619247, "decode_ms": 39.62075000163168}, "total_wall_ms": 147914.7441687528, "ttft_ms": 33.54645904619247, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 64, "facts_retained": 30, "retention_rate": 0.46875, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.6666666666666666, "status": 0.9166666666666666, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 912}, "ledger_by_hop": {"1": {"facts_total": 64, "facts_retained": 30, "retention_rate": 0.46875, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.6666666666666666, "status": 0.9166666666666666, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1087}, "2": {"facts_total": 64, "facts_retained": 30, "retention_rate": 0.46875, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.6666666666666666, "status": 0.9166666666666666, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 912}, "3": {"facts_total": 64, "facts_retained": 30, "retention_rate": 0.46875, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.6666666666666666, "status": 0.9166666666666666, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 912}, "4": {"facts_total": 64, "facts_retained": 30, "retention_rate": 0.46875, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.6666666666666666, "status": 0.9166666666666666, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 912}, "5": {"facts_total": 64, "facts_retained": 30, "retention_rate": 0.46875, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.6666666666666666, "status": 0.9166666666666666, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 912}, "6": {"facts_total": 64, "facts_retained": 30, "retention_rate": 0.46875, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.6666666666666666, "status": 0.9166666666666666, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 912}, "7": {"facts_total": 64, "facts_retained": 30, "retention_rate": 0.46875, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.6666666666666666, "status": 0.9166666666666666, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 912}, "8": {"facts_total": 64, "facts_retained": 30, "retention_rate": 0.46875, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.6666666666666666, "status": 0.9166666666666666, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 912}}}, "reference_key_id": "keys/synthetic/syn_03006_2000.json", "task_scores": {"correct": 1.0, "abstained": 1.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "060968d99413307af0515a687c7ce77ff5260c81e4a1b8709ace3f1df2e644b7", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "49dc78540d1a15720b1bba66", "attempt_id": "a64b954dcd2c", "run_group_id": "E4_core:A5|closed|qwen3b:h1", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T06:58:57.513066Z", "task_family": "synthetic", "document_id": "syn_03006_2000", "document_hash": "89be5e77e0ca114963e9ffdf8dc38fe5610f13a78ac29e7a1956cd9124ce5217", "question_id": "q00_single_fact", "split": "test", "cluster_id": "syn_03006_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 1, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1890, "query_tokens": 387, "message_tokens": 304, "message_utf8_bytes": 717, "cache_tensor_bytes": null, "serialized_payload_bytes": 717, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 22661.616917001083, "hops": {"1": 22661.616917001083}, "receiver_prefill_ms": 291.5338330203667, "ttft_ms": 45.12608295772225, "decode_ms": 53.92704205587506}, "total_wall_ms": 23052.95212508645, "ttft_ms": 45.12608295772225, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 64, "facts_retained": 10, "retention_rate": 0.15625, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.2, "amount": 0.16666666666666666, "status": 0.3333333333333333, "place": 0.0, "identifier": 0.0, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 669}, "ledger_by_hop": {"1": {"facts_total": 64, "facts_retained": 10, "retention_rate": 0.15625, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.2, "amount": 0.16666666666666666, "status": 0.3333333333333333, "place": 0.0, "identifier": 0.0, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 669}}}, "reference_key_id": "keys/synthetic/syn_03006_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "a7a61e059050ba2e9e391181c042fc9f413143ee1a9ef5f897c5cddf9f37fe4f", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "4461c394d62921a712fd0cce", "attempt_id": "70bda87daaca", "run_group_id": "E4_core:A5|closed|qwen3b:h1", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T06:58:57.923565Z", "task_family": "synthetic", "document_id": "syn_03006_2000", "document_hash": "89be5e77e0ca114963e9ffdf8dc38fe5610f13a78ac29e7a1956cd9124ce5217", "question_id": "q01_single_fact", "split": "test", "cluster_id": "syn_03006_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 1, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1890, "query_tokens": 392, "message_tokens": 304, "message_utf8_bytes": 717, "cache_tensor_bytes": null, "serialized_payload_bytes": 717, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 22661.616917001083, "hops": {"1": 22661.616917001083}, "receiver_prefill_ms": 431.6069999476895, "ttft_ms": 36.09979199245572, "decode_ms": 41.27341601997614}, "total_wall_ms": 23171.17300105747, "ttft_ms": 36.09979199245572, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 64, "facts_retained": 10, "retention_rate": 0.15625, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.2, "amount": 0.16666666666666666, "status": 0.3333333333333333, "place": 0.0, "identifier": 0.0, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 669}, "ledger_by_hop": {"1": {"facts_total": 64, "facts_retained": 10, "retention_rate": 0.15625, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.2, "amount": 0.16666666666666666, "status": 0.3333333333333333, "place": 0.0, "identifier": 0.0, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 669}}}, "reference_key_id": "keys/synthetic/syn_03006_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "a7a61e059050ba2e9e391181c042fc9f413143ee1a9ef5f897c5cddf9f37fe4f", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "7841c2270484d2f55a3fe156", "attempt_id": "b01b6f33a80d", "run_group_id": "E4_core:A5|closed|qwen3b:h1", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T06:58:58.450231Z", "task_family": "synthetic", "document_id": "syn_03006_2000", "document_hash": "89be5e77e0ca114963e9ffdf8dc38fe5610f13a78ac29e7a1956cd9124ce5217", "question_id": "q02_exact_copy", "split": "test", "cluster_id": "syn_03006_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 1, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1890, "query_tokens": 391, "message_tokens": 304, "message_utf8_bytes": 717, "cache_tensor_bytes": null, "serialized_payload_bytes": 717, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 22661.616917001083, "hops": {"1": 22661.616917001083}, "receiver_prefill_ms": 287.1987499529496, "ttft_ms": 33.12758298125118, "decode_ms": 41.50724993087351}, "total_wall_ms": 23023.94099999219, "ttft_ms": 33.12758298125118, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 64, "facts_retained": 10, "retention_rate": 0.15625, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.2, "amount": 0.16666666666666666, "status": 0.3333333333333333, "place": 0.0, "identifier": 0.0, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 669}, "ledger_by_hop": {"1": {"facts_total": 64, "facts_retained": 10, "retention_rate": 0.15625, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.2, "amount": 0.16666666666666666, "status": 0.3333333333333333, "place": 0.0, "identifier": 0.0, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 669}}}, "reference_key_id": "keys/synthetic/syn_03006_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "a7a61e059050ba2e9e391181c042fc9f413143ee1a9ef5f897c5cddf9f37fe4f", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "eae65f1e98ce21966cb84772", "attempt_id": "0ec93a87eadf", "run_group_id": "E4_core:A5|closed|qwen3b:h1", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T06:58:58.830013Z", "task_family": "synthetic", "document_id": "syn_03006_2000", "document_hash": "89be5e77e0ca114963e9ffdf8dc38fe5610f13a78ac29e7a1956cd9124ce5217", "question_id": "q03_exact_copy", "split": "test", "cluster_id": "syn_03006_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 1, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1890, "query_tokens": 386, "message_tokens": 304, "message_utf8_bytes": 717, "cache_tensor_bytes": null, "serialized_payload_bytes": 717, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 22661.616917001083, "hops": {"1": 22661.616917001083}, "receiver_prefill_ms": 282.31204207986593, "ttft_ms": 36.40033397823572, "decode_ms": 19.097208976745605}, "total_wall_ms": 23000.085333012976, "ttft_ms": 36.40033397823572, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 64, "facts_retained": 10, "retention_rate": 0.15625, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.2, "amount": 0.16666666666666666, "status": 0.3333333333333333, "place": 0.0, "identifier": 0.0, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 669}, "ledger_by_hop": {"1": {"facts_total": 64, "facts_retained": 10, "retention_rate": 0.15625, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.2, "amount": 0.16666666666666666, "status": 0.3333333333333333, "place": 0.0, "identifier": 0.0, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 669}}}, "reference_key_id": "keys/synthetic/syn_03006_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "a7a61e059050ba2e9e391181c042fc9f413143ee1a9ef5f897c5cddf9f37fe4f", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "a9bf0618b84803cb779173e8", "attempt_id": "a869ce997842", "run_group_id": "E4_core:A5|closed|qwen3b:h1", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T06:58:59.186149Z", "task_family": "synthetic", "document_id": "syn_03006_2000", "document_hash": "89be5e77e0ca114963e9ffdf8dc38fe5610f13a78ac29e7a1956cd9124ce5217", "question_id": "q04_multi_hop", "split": "test", "cluster_id": "syn_03006_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 1, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1890, "query_tokens": 395, "message_tokens": 304, "message_utf8_bytes": 717, "cache_tensor_bytes": null, "serialized_payload_bytes": 717, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 22661.616917001083, "hops": {"1": 22661.616917001083}, "receiver_prefill_ms": 341.713041998446, "ttft_ms": 56.365625001490116, "decode_ms": 58.00141696818173}, "total_wall_ms": 23118.24041698128, "ttft_ms": 56.365625001490116, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 64, "facts_retained": 10, "retention_rate": 0.15625, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.2, "amount": 0.16666666666666666, "status": 0.3333333333333333, "place": 0.0, "identifier": 0.0, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 669}, "ledger_by_hop": {"1": {"facts_total": 64, "facts_retained": 10, "retention_rate": 0.15625, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.2, "amount": 0.16666666666666666, "status": 0.3333333333333333, "place": 0.0, "identifier": 0.0, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 669}}}, "reference_key_id": "keys/synthetic/syn_03006_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "a7a61e059050ba2e9e391181c042fc9f413143ee1a9ef5f897c5cddf9f37fe4f", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "4e0f6caa2d4062d7f628c3ea", "attempt_id": "31fef896f57b", "run_group_id": "E4_core:A5|closed|qwen3b:h1", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T06:58:59.661635Z", "task_family": "synthetic", "document_id": "syn_03006_2000", "document_hash": "89be5e77e0ca114963e9ffdf8dc38fe5610f13a78ac29e7a1956cd9124ce5217", "question_id": "q05_multi_hop", "split": "test", "cluster_id": "syn_03006_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 1, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1890, "query_tokens": 398, "message_tokens": 304, "message_utf8_bytes": 717, "cache_tensor_bytes": null, "serialized_payload_bytes": 717, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 22661.616917001083, "hops": {"1": 22661.616917001083}, "receiver_prefill_ms": 418.01491694059223, "ttft_ms": 31.33300004992634, "decode_ms": 46.047499985434115}, "total_wall_ms": 23157.50041708816, "ttft_ms": 31.33300004992634, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 64, "facts_retained": 10, "retention_rate": 0.15625, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.2, "amount": 0.16666666666666666, "status": 0.3333333333333333, "place": 0.0, "identifier": 0.0, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 669}, "ledger_by_hop": {"1": {"facts_total": 64, "facts_retained": 10, "retention_rate": 0.15625, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.2, "amount": 0.16666666666666666, "status": 0.3333333333333333, "place": 0.0, "identifier": 0.0, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 669}}}, "reference_key_id": "keys/synthetic/syn_03006_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "a7a61e059050ba2e9e391181c042fc9f413143ee1a9ef5f897c5cddf9f37fe4f", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "20661b346815e14f09929831", "attempt_id": "f9b07cff4239", "run_group_id": "E4_core:A5|closed|qwen3b:h1", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T06:59:00.173913Z", "task_family": "synthetic", "document_id": "syn_03006_2000", "document_hash": "89be5e77e0ca114963e9ffdf8dc38fe5610f13a78ac29e7a1956cd9124ce5217", "question_id": "q06_negation", "split": "test", "cluster_id": "syn_03006_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 1, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1890, "query_tokens": 386, "message_tokens": 304, "message_utf8_bytes": 717, "cache_tensor_bytes": null, "serialized_payload_bytes": 717, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 22661.616917001083, "hops": {"1": 22661.616917001083}, "receiver_prefill_ms": 281.9023330230266, "ttft_ms": 33.44874992035329, "decode_ms": 38.274041027761996}, "total_wall_ms": 23015.755375963636, "ttft_ms": 33.44874992035329, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 64, "facts_retained": 10, "retention_rate": 0.15625, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.2, "amount": 0.16666666666666666, "status": 0.3333333333333333, "place": 0.0, "identifier": 0.0, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 669}, "ledger_by_hop": {"1": {"facts_total": 64, "facts_retained": 10, "retention_rate": 0.15625, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.2, "amount": 0.16666666666666666, "status": 0.3333333333333333, "place": 0.0, "identifier": 0.0, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 669}}}, "reference_key_id": "keys/synthetic/syn_03006_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "a7a61e059050ba2e9e391181c042fc9f413143ee1a9ef5f897c5cddf9f37fe4f", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "d7ae9cc27bc5d2984813b8c7", "attempt_id": "4325d1618770", "run_group_id": "E4_core:A5|closed|qwen3b:h1", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T06:59:00.544853Z", "task_family": "synthetic", "document_id": "syn_03006_2000", "document_hash": "89be5e77e0ca114963e9ffdf8dc38fe5610f13a78ac29e7a1956cd9124ce5217", "question_id": "q07_negation", "split": "test", "cluster_id": "syn_03006_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 1, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1890, "query_tokens": 386, "message_tokens": 304, "message_utf8_bytes": 717, "cache_tensor_bytes": null, "serialized_payload_bytes": 717, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 22661.616917001083, "hops": {"1": 22661.616917001083}, "receiver_prefill_ms": 278.7012909539044, "ttft_ms": 19.9192090658471, "decode_ms": 13.701332965865731}, "total_wall_ms": 22974.364749970846, "ttft_ms": 19.9192090658471, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "no", "evidence_artifact": {"ledger": {"facts_total": 64, "facts_retained": 10, "retention_rate": 0.15625, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.2, "amount": 0.16666666666666666, "status": 0.3333333333333333, "place": 0.0, "identifier": 0.0, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 669}, "ledger_by_hop": {"1": {"facts_total": 64, "facts_retained": 10, "retention_rate": 0.15625, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.2, "amount": 0.16666666666666666, "status": 0.3333333333333333, "place": 0.0, "identifier": 0.0, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 669}}}, "reference_key_id": "keys/synthetic/syn_03006_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "a7a61e059050ba2e9e391181c042fc9f413143ee1a9ef5f897c5cddf9f37fe4f", "output_hash": "9390298f3fb0c5b160498935d79cb139aef28e1c47358b4bbba61862b9c26e59"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "7bd8ba3deac6058a59941329", "attempt_id": "22d8d17f27a7", "run_group_id": "E4_core:A5|closed|qwen3b:h1", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T06:59:00.874831Z", "task_family": "synthetic", "document_id": "syn_03006_2000", "document_hash": "89be5e77e0ca114963e9ffdf8dc38fe5610f13a78ac29e7a1956cd9124ce5217", "question_id": "q08_revision", "split": "test", "cluster_id": "syn_03006_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 1, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1890, "query_tokens": 403, "message_tokens": 304, "message_utf8_bytes": 717, "cache_tensor_bytes": null, "serialized_payload_bytes": 717, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 22661.616917001083, "hops": {"1": 22661.616917001083}, "receiver_prefill_ms": 326.3220409862697, "ttft_ms": 42.30762494262308, "decode_ms": 455.0647920696065}, "total_wall_ms": 23485.899583087303, "ttft_ms": 42.30762494262308, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 8}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "10107545", "evidence_artifact": {"ledger": {"facts_total": 64, "facts_retained": 10, "retention_rate": 0.15625, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.2, "amount": 0.16666666666666666, "status": 0.3333333333333333, "place": 0.0, "identifier": 0.0, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 669}, "ledger_by_hop": {"1": {"facts_total": 64, "facts_retained": 10, "retention_rate": 0.15625, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.2, "amount": 0.16666666666666666, "status": 0.3333333333333333, "place": 0.0, "identifier": 0.0, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 669}}}, "reference_key_id": "keys/synthetic/syn_03006_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "a7a61e059050ba2e9e391181c042fc9f413143ee1a9ef5f897c5cddf9f37fe4f", "output_hash": "6faba9de4e29949bbdd8934f69fc2110f07076db6801b211aeca9351292537fe"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "004b8350af86ecf27ae9d01e", "attempt_id": "2312a3d3a561", "run_group_id": "E4_core:A5|closed|qwen3b:h1", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T06:59:01.717199Z", "task_family": "synthetic", "document_id": "syn_03006_2000", "document_hash": "89be5e77e0ca114963e9ffdf8dc38fe5610f13a78ac29e7a1956cd9124ce5217", "question_id": "q09_revision", "split": "test", "cluster_id": "syn_03006_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 1, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1890, "query_tokens": 396, "message_tokens": 304, "message_utf8_bytes": 717, "cache_tensor_bytes": null, "serialized_payload_bytes": 717, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 22661.616917001083, "hops": {"1": 22661.616917001083}, "receiver_prefill_ms": 298.7924579065293, "ttft_ms": 22.345458040945232, "decode_ms": 232.87874995730817}, "total_wall_ms": 23216.08566690702, "ttft_ms": 22.345458040945232, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 7}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "9112735", "evidence_artifact": {"ledger": {"facts_total": 64, "facts_retained": 10, "retention_rate": 0.15625, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.2, "amount": 0.16666666666666666, "status": 0.3333333333333333, "place": 0.0, "identifier": 0.0, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 669}, "ledger_by_hop": {"1": {"facts_total": 64, "facts_retained": 10, "retention_rate": 0.15625, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.2, "amount": 0.16666666666666666, "status": 0.3333333333333333, "place": 0.0, "identifier": 0.0, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 669}}}, "reference_key_id": "keys/synthetic/syn_03006_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "a7a61e059050ba2e9e391181c042fc9f413143ee1a9ef5f897c5cddf9f37fe4f", "output_hash": "15a3f804636f0e62d4a3d9be7677f001a0ad6b1fb8ae7e8cdc63c44ec7bac31f"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "5bd7e998fa7ea50f23fc371f", "attempt_id": "8ab28f9b5245", "run_group_id": "E4_core:A5|closed|qwen3b:h1", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T06:59:02.288826Z", "task_family": "synthetic", "document_id": "syn_03006_2000", "document_hash": "89be5e77e0ca114963e9ffdf8dc38fe5610f13a78ac29e7a1956cd9124ce5217", "question_id": "q10_temporal", "split": "test", "cluster_id": "syn_03006_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 1, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1890, "query_tokens": 405, "message_tokens": 304, "message_utf8_bytes": 717, "cache_tensor_bytes": null, "serialized_payload_bytes": 717, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 22661.616917001083, "hops": {"1": 22661.616917001083}, "receiver_prefill_ms": 259.45291691459715, "ttft_ms": 53.74020803719759, "decode_ms": 49.17629202827811}, "total_wall_ms": 23024.690250051208, "ttft_ms": 53.74020803719759, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 64, "facts_retained": 10, "retention_rate": 0.15625, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.2, "amount": 0.16666666666666666, "status": 0.3333333333333333, "place": 0.0, "identifier": 0.0, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 669}, "ledger_by_hop": {"1": {"facts_total": 64, "facts_retained": 10, "retention_rate": 0.15625, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.2, "amount": 0.16666666666666666, "status": 0.3333333333333333, "place": 0.0, "identifier": 0.0, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 669}}}, "reference_key_id": "keys/synthetic/syn_03006_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "a7a61e059050ba2e9e391181c042fc9f413143ee1a9ef5f897c5cddf9f37fe4f", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "dcfdef0ca33726fbb1f276fc", "attempt_id": "d995d435566c", "run_group_id": "E4_core:A5|closed|qwen3b:h1", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T06:59:02.669143Z", "task_family": "synthetic", "document_id": "syn_03006_2000", "document_hash": "89be5e77e0ca114963e9ffdf8dc38fe5610f13a78ac29e7a1956cd9124ce5217", "question_id": "q11_aggregation", "split": "test", "cluster_id": "syn_03006_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 1, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1890, "query_tokens": 398, "message_tokens": 304, "message_utf8_bytes": 717, "cache_tensor_bytes": null, "serialized_payload_bytes": 717, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 22661.616917001083, "hops": {"1": 22661.616917001083}, "receiver_prefill_ms": 453.061957960017, "ttft_ms": 39.099999936297536, "decode_ms": 21.663542021997273}, "total_wall_ms": 23176.175708067603, "ttft_ms": 39.099999936297536, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 64, "facts_retained": 10, "retention_rate": 0.15625, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.2, "amount": 0.16666666666666666, "status": 0.3333333333333333, "place": 0.0, "identifier": 0.0, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 669}, "ledger_by_hop": {"1": {"facts_total": 64, "facts_retained": 10, "retention_rate": 0.15625, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.2, "amount": 0.16666666666666666, "status": 0.3333333333333333, "place": 0.0, "identifier": 0.0, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 669}}}, "reference_key_id": "keys/synthetic/syn_03006_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "a7a61e059050ba2e9e391181c042fc9f413143ee1a9ef5f897c5cddf9f37fe4f", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "0710fbfce26c6127fa17a31c", "attempt_id": "e78f74659e6b", "run_group_id": "E4_core:A5|closed|qwen3b:h1", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T06:59:03.202113Z", "task_family": "synthetic", "document_id": "syn_03006_2000", "document_hash": "89be5e77e0ca114963e9ffdf8dc38fe5610f13a78ac29e7a1956cd9124ce5217", "question_id": "q12_unit_conversion", "split": "test", "cluster_id": "syn_03006_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 1, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1890, "query_tokens": 395, "message_tokens": 304, "message_utf8_bytes": 717, "cache_tensor_bytes": null, "serialized_payload_bytes": 717, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 22661.616917001083, "hops": {"1": 22661.616917001083}, "receiver_prefill_ms": 293.1237079901621, "ttft_ms": 38.776917033828795, "decode_ms": 265.2467089938}, "total_wall_ms": 23259.297209093347, "ttft_ms": 38.776917033828795, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 8}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "11946000", "evidence_artifact": {"ledger": {"facts_total": 64, "facts_retained": 10, "retention_rate": 0.15625, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.2, "amount": 0.16666666666666666, "status": 0.3333333333333333, "place": 0.0, "identifier": 0.0, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 669}, "ledger_by_hop": {"1": {"facts_total": 64, "facts_retained": 10, "retention_rate": 0.15625, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.2, "amount": 0.16666666666666666, "status": 0.3333333333333333, "place": 0.0, "identifier": 0.0, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 669}}}, "reference_key_id": "keys/synthetic/syn_03006_2000.json", "task_scores": {"correct": 0.0, "abstained": 0.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "a7a61e059050ba2e9e391181c042fc9f413143ee1a9ef5f897c5cddf9f37fe4f", "output_hash": "edf27f197d1395ea4e7b86fae1feecd3591246e1299bba6a08655f5724a670ea"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "197fefd8b255071102df65d1", "attempt_id": "ff8c296d388e", "run_group_id": "E4_core:A5|closed|qwen3b:h1", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T06:59:03.818088Z", "task_family": "synthetic", "document_id": "syn_03006_2000", "document_hash": "89be5e77e0ca114963e9ffdf8dc38fe5610f13a78ac29e7a1956cd9124ce5217", "question_id": "q13_unanswerable", "split": "test", "cluster_id": "syn_03006_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 1, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1890, "query_tokens": 399, "message_tokens": 304, "message_utf8_bytes": 717, "cache_tensor_bytes": null, "serialized_payload_bytes": 717, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 22661.616917001083, "hops": {"1": 22661.616917001083}, "receiver_prefill_ms": 270.1662079198286, "ttft_ms": 53.531625075265765, "decode_ms": 43.292707996442914}, "total_wall_ms": 23029.211958986707, "ttft_ms": 53.531625075265765, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 64, "facts_retained": 10, "retention_rate": 0.15625, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.2, "amount": 0.16666666666666666, "status": 0.3333333333333333, "place": 0.0, "identifier": 0.0, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 669}, "ledger_by_hop": {"1": {"facts_total": 64, "facts_retained": 10, "retention_rate": 0.15625, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.2, "amount": 0.16666666666666666, "status": 0.3333333333333333, "place": 0.0, "identifier": 0.0, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 669}}}, "reference_key_id": "keys/synthetic/syn_03006_2000.json", "task_scores": {"correct": 1.0, "abstained": 1.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "a7a61e059050ba2e9e391181c042fc9f413143ee1a9ef5f897c5cddf9f37fe4f", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "4469e765c9df1968496452fa", "attempt_id": "b8f989ef80d9", "run_group_id": "E4_core:A5|closed|qwen3b:h1", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T06:59:04.202782Z", "task_family": "synthetic", "document_id": "syn_03006_2000", "document_hash": "89be5e77e0ca114963e9ffdf8dc38fe5610f13a78ac29e7a1956cd9124ce5217", "question_id": "q14_unanswerable", "split": "test", "cluster_id": "syn_03006_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 1, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1890, "query_tokens": 400, "message_tokens": 304, "message_utf8_bytes": 717, "cache_tensor_bytes": null, "serialized_payload_bytes": 717, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 22661.616917001083, "hops": {"1": 22661.616917001083}, "receiver_prefill_ms": 362.55333304870874, "ttft_ms": 139.63366590905935, "decode_ms": 39.50333304237574}, "total_wall_ms": 23203.987833927386, "ttft_ms": 139.63366590905935, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 64, "facts_retained": 10, "retention_rate": 0.15625, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.2, "amount": 0.16666666666666666, "status": 0.3333333333333333, "place": 0.0, "identifier": 0.0, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 669}, "ledger_by_hop": {"1": {"facts_total": 64, "facts_retained": 10, "retention_rate": 0.15625, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.2, "amount": 0.16666666666666666, "status": 0.3333333333333333, "place": 0.0, "identifier": 0.0, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 669}}}, "reference_key_id": "keys/synthetic/syn_03006_2000.json", "task_scores": {"correct": 1.0, "abstained": 1.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "a7a61e059050ba2e9e391181c042fc9f413143ee1a9ef5f897c5cddf9f37fe4f", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "44ccc078a815779030752b3a", "attempt_id": "05ceeb36e9c4", "run_group_id": "E4_core:A5|closed|qwen3b:h2", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T06:59:16.991495Z", "task_family": "synthetic", "document_id": "syn_03006_2000", "document_hash": "89be5e77e0ca114963e9ffdf8dc38fe5610f13a78ac29e7a1956cd9124ce5217", "question_id": "q00_single_fact", "split": "test", "cluster_id": "syn_03006_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 2, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1890, "query_tokens": 352, "message_tokens": 269, "message_utf8_bytes": 620, "cache_tensor_bytes": null, "serialized_payload_bytes": 620, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 34889.89491702523, "hops": {"1": 22661.616917001083, "2": 12228.278000024147}, "receiver_prefill_ms": 271.1275831097737, "ttft_ms": 43.26987511012703, "decode_ms": 22.167416056618094}, "total_wall_ms": 35226.884375093505, "ttft_ms": 43.26987511012703, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 64, "facts_retained": 9, "retention_rate": 0.140625, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.16666666666666666, "status": 0.3333333333333333, "place": 0.0, "identifier": 0.0, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 572}, "ledger_by_hop": {"1": {"facts_total": 64, "facts_retained": 10, "retention_rate": 0.15625, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.2, "amount": 0.16666666666666666, "status": 0.3333333333333333, "place": 0.0, "identifier": 0.0, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 669}, "2": {"facts_total": 64, "facts_retained": 9, "retention_rate": 0.140625, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.16666666666666666, "status": 0.3333333333333333, "place": 0.0, "identifier": 0.0, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 572}}}, "reference_key_id": "keys/synthetic/syn_03006_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "175d3ea2044e4182288513cfbf6fbcb45bd8241379081d5258569d283ccfe60c", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "1c49e3fdeada614d34cc40cc", "attempt_id": "2d4eb49456ed", "run_group_id": "E4_core:A5|closed|qwen3b:h2", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T06:59:17.345708Z", "task_family": "synthetic", "document_id": "syn_03006_2000", "document_hash": "89be5e77e0ca114963e9ffdf8dc38fe5610f13a78ac29e7a1956cd9124ce5217", "question_id": "q01_single_fact", "split": "test", "cluster_id": "syn_03006_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 2, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1890, "query_tokens": 357, "message_tokens": 269, "message_utf8_bytes": 620, "cache_tensor_bytes": null, "serialized_payload_bytes": 620, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 34889.89491702523, "hops": {"1": 22661.616917001083, "2": 12228.278000024147}, "receiver_prefill_ms": 276.4203749829903, "ttft_ms": 27.300666086375713, "decode_ms": 37.28645795490593}, "total_wall_ms": 35231.329833040945, "ttft_ms": 27.300666086375713, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 64, "facts_retained": 9, "retention_rate": 0.140625, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.16666666666666666, "status": 0.3333333333333333, "place": 0.0, "identifier": 0.0, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 572}, "ledger_by_hop": {"1": {"facts_total": 64, "facts_retained": 10, "retention_rate": 0.15625, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.2, "amount": 0.16666666666666666, "status": 0.3333333333333333, "place": 0.0, "identifier": 0.0, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 669}, "2": {"facts_total": 64, "facts_retained": 9, "retention_rate": 0.140625, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.16666666666666666, "status": 0.3333333333333333, "place": 0.0, "identifier": 0.0, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 572}}}, "reference_key_id": "keys/synthetic/syn_03006_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "175d3ea2044e4182288513cfbf6fbcb45bd8241379081d5258569d283ccfe60c", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "f4c768bc766e64acb81fd23c", "attempt_id": "ffb8cb1396a7", "run_group_id": "E4_core:A5|closed|qwen3b:h2", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T06:59:17.704305Z", "task_family": "synthetic", "document_id": "syn_03006_2000", "document_hash": "89be5e77e0ca114963e9ffdf8dc38fe5610f13a78ac29e7a1956cd9124ce5217", "question_id": "q02_exact_copy", "split": "test", "cluster_id": "syn_03006_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 2, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1890, "query_tokens": 356, "message_tokens": 269, "message_utf8_bytes": 620, "cache_tensor_bytes": null, "serialized_payload_bytes": 620, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 34889.89491702523, "hops": {"1": 22661.616917001083, "2": 12228.278000024147}, "receiver_prefill_ms": 241.93000001832843, "ttft_ms": 38.708040956407785, "decode_ms": 69.6544999955222}, "total_wall_ms": 35240.64445798285, "ttft_ms": 38.708040956407785, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 64, "facts_retained": 9, "retention_rate": 0.140625, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.16666666666666666, "status": 0.3333333333333333, "place": 0.0, "identifier": 0.0, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 572}, "ledger_by_hop": {"1": {"facts_total": 64, "facts_retained": 10, "retention_rate": 0.15625, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.2, "amount": 0.16666666666666666, "status": 0.3333333333333333, "place": 0.0, "identifier": 0.0, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 669}, "2": {"facts_total": 64, "facts_retained": 9, "retention_rate": 0.140625, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.16666666666666666, "status": 0.3333333333333333, "place": 0.0, "identifier": 0.0, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 572}}}, "reference_key_id": "keys/synthetic/syn_03006_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "175d3ea2044e4182288513cfbf6fbcb45bd8241379081d5258569d283ccfe60c", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "0dfc840f30f7d5950916e9a3", "attempt_id": "8dda93b48a77", "run_group_id": "E4_core:A5|closed|qwen3b:h2", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T06:59:18.071923Z", "task_family": "synthetic", "document_id": "syn_03006_2000", "document_hash": "89be5e77e0ca114963e9ffdf8dc38fe5610f13a78ac29e7a1956cd9124ce5217", "question_id": "q03_exact_copy", "split": "test", "cluster_id": "syn_03006_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 2, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1890, "query_tokens": 351, "message_tokens": 269, "message_utf8_bytes": 620, "cache_tensor_bytes": null, "serialized_payload_bytes": 620, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 34889.89491702523, "hops": {"1": 22661.616917001083, "2": 12228.278000024147}, "receiver_prefill_ms": 317.53204204142094, "ttft_ms": 70.91925002168864, "decode_ms": 101.09787504188716}, "total_wall_ms": 35379.90399997216, "ttft_ms": 70.91925002168864, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 64, "facts_retained": 9, "retention_rate": 0.140625, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.16666666666666666, "status": 0.3333333333333333, "place": 0.0, "identifier": 0.0, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 572}, "ledger_by_hop": {"1": {"facts_total": 64, "facts_retained": 10, "retention_rate": 0.15625, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.2, "amount": 0.16666666666666666, "status": 0.3333333333333333, "place": 0.0, "identifier": 0.0, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 669}, "2": {"facts_total": 64, "facts_retained": 9, "retention_rate": 0.140625, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.16666666666666666, "status": 0.3333333333333333, "place": 0.0, "identifier": 0.0, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 572}}}, "reference_key_id": "keys/synthetic/syn_03006_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "175d3ea2044e4182288513cfbf6fbcb45bd8241379081d5258569d283ccfe60c", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "9974080d458292e13ebb9211", "attempt_id": "86aa4239e769", "run_group_id": "E4_core:A5|closed|qwen3b:h2", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T06:59:18.579619Z", "task_family": "synthetic", "document_id": "syn_03006_2000", "document_hash": "89be5e77e0ca114963e9ffdf8dc38fe5610f13a78ac29e7a1956cd9124ce5217", "question_id": "q04_multi_hop", "split": "test", "cluster_id": "syn_03006_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 2, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1890, "query_tokens": 360, "message_tokens": 269, "message_utf8_bytes": 620, "cache_tensor_bytes": null, "serialized_payload_bytes": 620, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 34889.89491702523, "hops": {"1": 22661.616917001083, "2": 12228.278000024147}, "receiver_prefill_ms": 264.46891704108566, "ttft_ms": 34.9962500622496, "decode_ms": 40.38408398628235}, "total_wall_ms": 35230.170125025325, "ttft_ms": 34.9962500622496, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 64, "facts_retained": 9, "retention_rate": 0.140625, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.16666666666666666, "status": 0.3333333333333333, "place": 0.0, "identifier": 0.0, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 572}, "ledger_by_hop": {"1": {"facts_total": 64, "facts_retained": 10, "retention_rate": 0.15625, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.2, "amount": 0.16666666666666666, "status": 0.3333333333333333, "place": 0.0, "identifier": 0.0, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 669}, "2": {"facts_total": 64, "facts_retained": 9, "retention_rate": 0.140625, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.16666666666666666, "status": 0.3333333333333333, "place": 0.0, "identifier": 0.0, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 572}}}, "reference_key_id": "keys/synthetic/syn_03006_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "175d3ea2044e4182288513cfbf6fbcb45bd8241379081d5258569d283ccfe60c", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "1dd9cb410b5285a002e365f1", "attempt_id": "26add4b869ca", "run_group_id": "E4_core:A5|closed|qwen3b:h2", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T06:59:18.936924Z", "task_family": "synthetic", "document_id": "syn_03006_2000", "document_hash": "89be5e77e0ca114963e9ffdf8dc38fe5610f13a78ac29e7a1956cd9124ce5217", "question_id": "q05_multi_hop", "split": "test", "cluster_id": "syn_03006_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 2, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1890, "query_tokens": 363, "message_tokens": 269, "message_utf8_bytes": 620, "cache_tensor_bytes": null, "serialized_payload_bytes": 620, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 34889.89491702523, "hops": {"1": 22661.616917001083, "2": 12228.278000024147}, "receiver_prefill_ms": 270.050875027664, "ttft_ms": 31.883916002698243, "decode_ms": 42.341416934505105}, "total_wall_ms": 35234.57995895296, "ttft_ms": 31.883916002698243, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 64, "facts_retained": 9, "retention_rate": 0.140625, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.16666666666666666, "status": 0.3333333333333333, "place": 0.0, "identifier": 0.0, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 572}, "ledger_by_hop": {"1": {"facts_total": 64, "facts_retained": 10, "retention_rate": 0.15625, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.2, "amount": 0.16666666666666666, "status": 0.3333333333333333, "place": 0.0, "identifier": 0.0, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 669}, "2": {"facts_total": 64, "facts_retained": 9, "retention_rate": 0.140625, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.16666666666666666, "status": 0.3333333333333333, "place": 0.0, "identifier": 0.0, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 572}}}, "reference_key_id": "keys/synthetic/syn_03006_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "175d3ea2044e4182288513cfbf6fbcb45bd8241379081d5258569d283ccfe60c", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "146ab328e26d82b51b277c56", "attempt_id": "0f0b388a5785", "run_group_id": "E4_core:A5|closed|qwen3b:h2", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T06:59:19.298383Z", "task_family": "synthetic", "document_id": "syn_03006_2000", "document_hash": "89be5e77e0ca114963e9ffdf8dc38fe5610f13a78ac29e7a1956cd9124ce5217", "question_id": "q06_negation", "split": "test", "cluster_id": "syn_03006_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 2, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1890, "query_tokens": 351, "message_tokens": 269, "message_utf8_bytes": 620, "cache_tensor_bytes": null, "serialized_payload_bytes": 620, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 34889.89491702523, "hops": {"1": 22661.616917001083, "2": 12228.278000024147}, "receiver_prefill_ms": 251.1514580110088, "ttft_ms": 56.430332944728434, "decode_ms": 44.36212498694658}, "total_wall_ms": 35242.33216699213, "ttft_ms": 56.430332944728434, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 64, "facts_retained": 9, "retention_rate": 0.140625, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.16666666666666666, "status": 0.3333333333333333, "place": 0.0, "identifier": 0.0, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 572}, "ledger_by_hop": {"1": {"facts_total": 64, "facts_retained": 10, "retention_rate": 0.15625, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.2, "amount": 0.16666666666666666, "status": 0.3333333333333333, "place": 0.0, "identifier": 0.0, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 669}, "2": {"facts_total": 64, "facts_retained": 9, "retention_rate": 0.140625, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.16666666666666666, "status": 0.3333333333333333, "place": 0.0, "identifier": 0.0, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 572}}}, "reference_key_id": "keys/synthetic/syn_03006_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "175d3ea2044e4182288513cfbf6fbcb45bd8241379081d5258569d283ccfe60c", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "298b0cfcdbd610e9dafe8ebd", "attempt_id": "6bbadba304e4", "run_group_id": "E4_core:A5|closed|qwen3b:h2", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T06:59:19.667713Z", "task_family": "synthetic", "document_id": "syn_03006_2000", "document_hash": "89be5e77e0ca114963e9ffdf8dc38fe5610f13a78ac29e7a1956cd9124ce5217", "question_id": "q07_negation", "split": "test", "cluster_id": "syn_03006_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 2, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1890, "query_tokens": 351, "message_tokens": 269, "message_utf8_bytes": 620, "cache_tensor_bytes": null, "serialized_payload_bytes": 620, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 34889.89491702523, "hops": {"1": 22661.616917001083, "2": 12228.278000024147}, "receiver_prefill_ms": 265.0046249618754, "ttft_ms": 130.0464590312913, "decode_ms": 102.13283402845263}, "total_wall_ms": 35387.534376000986, "ttft_ms": 130.0464590312913, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 64, "facts_retained": 9, "retention_rate": 0.140625, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.16666666666666666, "status": 0.3333333333333333, "place": 0.0, "identifier": 0.0, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 572}, "ledger_by_hop": {"1": {"facts_total": 64, "facts_retained": 10, "retention_rate": 0.15625, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.2, "amount": 0.16666666666666666, "status": 0.3333333333333333, "place": 0.0, "identifier": 0.0, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 669}, "2": {"facts_total": 64, "facts_retained": 9, "retention_rate": 0.140625, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.16666666666666666, "status": 0.3333333333333333, "place": 0.0, "identifier": 0.0, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 572}}}, "reference_key_id": "keys/synthetic/syn_03006_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "175d3ea2044e4182288513cfbf6fbcb45bd8241379081d5258569d283ccfe60c", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "b4abde16378a5c933048e283", "attempt_id": "1c66895ee0ff", "run_group_id": "E4_core:A5|closed|qwen3b:h2", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T06:59:20.182499Z", "task_family": "synthetic", "document_id": "syn_03006_2000", "document_hash": "89be5e77e0ca114963e9ffdf8dc38fe5610f13a78ac29e7a1956cd9124ce5217", "question_id": "q08_revision", "split": "test", "cluster_id": "syn_03006_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 2, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1890, "query_tokens": 368, "message_tokens": 269, "message_utf8_bytes": 620, "cache_tensor_bytes": null, "serialized_payload_bytes": 620, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 34889.89491702523, "hops": {"1": 22661.616917001083, "2": 12228.278000024147}, "receiver_prefill_ms": 265.27299999725074, "ttft_ms": 44.697584002278745, "decode_ms": 294.7077910648659}, "total_wall_ms": 35495.01345795579, "ttft_ms": 44.697584002278745, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 8}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "10107545", "evidence_artifact": {"ledger": {"facts_total": 64, "facts_retained": 9, "retention_rate": 0.140625, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.16666666666666666, "status": 0.3333333333333333, "place": 0.0, "identifier": 0.0, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 572}, "ledger_by_hop": {"1": {"facts_total": 64, "facts_retained": 10, "retention_rate": 0.15625, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.2, "amount": 0.16666666666666666, "status": 0.3333333333333333, "place": 0.0, "identifier": 0.0, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 669}, "2": {"facts_total": 64, "facts_retained": 9, "retention_rate": 0.140625, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.16666666666666666, "status": 0.3333333333333333, "place": 0.0, "identifier": 0.0, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 572}}}, "reference_key_id": "keys/synthetic/syn_03006_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "175d3ea2044e4182288513cfbf6fbcb45bd8241379081d5258569d283ccfe60c", "output_hash": "6faba9de4e29949bbdd8934f69fc2110f07076db6801b211aeca9351292537fe"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "7b13e7947e827bc7fe9f94de", "attempt_id": "c9b27646dd7a", "run_group_id": "E4_core:A5|closed|qwen3b:h2", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T06:59:20.803969Z", "task_family": "synthetic", "document_id": "syn_03006_2000", "document_hash": "89be5e77e0ca114963e9ffdf8dc38fe5610f13a78ac29e7a1956cd9124ce5217", "question_id": "q09_revision", "split": "test", "cluster_id": "syn_03006_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 2, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1890, "query_tokens": 361, "message_tokens": 269, "message_utf8_bytes": 620, "cache_tensor_bytes": null, "serialized_payload_bytes": 620, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 34889.89491702523, "hops": {"1": 22661.616917001083, "2": 12228.278000024147}, "receiver_prefill_ms": 235.30504200607538, "ttft_ms": 46.896791900508106, "decode_ms": 532.9618330579251}, "total_wall_ms": 35705.51445905585, "ttft_ms": 46.896791900508106, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 7}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "9112735", "evidence_artifact": {"ledger": {"facts_total": 64, "facts_retained": 9, "retention_rate": 0.140625, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.16666666666666666, "status": 0.3333333333333333, "place": 0.0, "identifier": 0.0, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 572}, "ledger_by_hop": {"1": {"facts_total": 64, "facts_retained": 10, "retention_rate": 0.15625, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.2, "amount": 0.16666666666666666, "status": 0.3333333333333333, "place": 0.0, "identifier": 0.0, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 669}, "2": {"facts_total": 64, "facts_retained": 9, "retention_rate": 0.140625, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.16666666666666666, "status": 0.3333333333333333, "place": 0.0, "identifier": 0.0, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 572}}}, "reference_key_id": "keys/synthetic/syn_03006_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "175d3ea2044e4182288513cfbf6fbcb45bd8241379081d5258569d283ccfe60c", "output_hash": "15a3f804636f0e62d4a3d9be7677f001a0ad6b1fb8ae7e8cdc63c44ec7bac31f"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "0a18504e362032770d02c10b", "attempt_id": "d7a2772571d2", "run_group_id": "E4_core:A5|closed|qwen3b:h2", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T06:59:21.637508Z", "task_family": "synthetic", "document_id": "syn_03006_2000", "document_hash": "89be5e77e0ca114963e9ffdf8dc38fe5610f13a78ac29e7a1956cd9124ce5217", "question_id": "q10_temporal", "split": "test", "cluster_id": "syn_03006_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 2, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1890, "query_tokens": 370, "message_tokens": 269, "message_utf8_bytes": 620, "cache_tensor_bytes": null, "serialized_payload_bytes": 620, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 34889.89491702523, "hops": {"1": 22661.616917001083, "2": 12228.278000024147}, "receiver_prefill_ms": 277.38166600465775, "ttft_ms": 39.70683296211064, "decode_ms": 31.863207928836346}, "total_wall_ms": 35239.353708107956, "ttft_ms": 39.70683296211064, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 64, "facts_retained": 9, "retention_rate": 0.140625, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.16666666666666666, "status": 0.3333333333333333, "place": 0.0, "identifier": 0.0, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 572}, "ledger_by_hop": {"1": {"facts_total": 64, "facts_retained": 10, "retention_rate": 0.15625, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.2, "amount": 0.16666666666666666, "status": 0.3333333333333333, "place": 0.0, "identifier": 0.0, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 669}, "2": {"facts_total": 64, "facts_retained": 9, "retention_rate": 0.140625, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.16666666666666666, "status": 0.3333333333333333, "place": 0.0, "identifier": 0.0, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 572}}}, "reference_key_id": "keys/synthetic/syn_03006_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "175d3ea2044e4182288513cfbf6fbcb45bd8241379081d5258569d283ccfe60c", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "e859a1ad534da0240c785463", "attempt_id": "419833fa9c37", "run_group_id": "E4_core:A5|closed|qwen3b:h2", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T06:59:22.004178Z", "task_family": "synthetic", "document_id": "syn_03006_2000", "document_hash": "89be5e77e0ca114963e9ffdf8dc38fe5610f13a78ac29e7a1956cd9124ce5217", "question_id": "q11_aggregation", "split": "test", "cluster_id": "syn_03006_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 2, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1890, "query_tokens": 363, "message_tokens": 269, "message_utf8_bytes": 620, "cache_tensor_bytes": null, "serialized_payload_bytes": 620, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 34889.89491702523, "hops": {"1": 22661.616917001083, "2": 12228.278000024147}, "receiver_prefill_ms": 260.50912495702505, "ttft_ms": 30.33016703557223, "decode_ms": 45.72499997448176}, "total_wall_ms": 35226.894958061166, "ttft_ms": 30.33016703557223, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 64, "facts_retained": 9, "retention_rate": 0.140625, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.16666666666666666, "status": 0.3333333333333333, "place": 0.0, "identifier": 0.0, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 572}, "ledger_by_hop": {"1": {"facts_total": 64, "facts_retained": 10, "retention_rate": 0.15625, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.2, "amount": 0.16666666666666666, "status": 0.3333333333333333, "place": 0.0, "identifier": 0.0, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 669}, "2": {"facts_total": 64, "facts_retained": 9, "retention_rate": 0.140625, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.16666666666666666, "status": 0.3333333333333333, "place": 0.0, "identifier": 0.0, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 572}}}, "reference_key_id": "keys/synthetic/syn_03006_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "175d3ea2044e4182288513cfbf6fbcb45bd8241379081d5258569d283ccfe60c", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "be8a1715d20ccb1af123d514", "attempt_id": "a633fa405945", "run_group_id": "E4_core:A5|closed|qwen3b:h2", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T06:59:22.358108Z", "task_family": "synthetic", "document_id": "syn_03006_2000", "document_hash": "89be5e77e0ca114963e9ffdf8dc38fe5610f13a78ac29e7a1956cd9124ce5217", "question_id": "q12_unit_conversion", "split": "test", "cluster_id": "syn_03006_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 2, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1890, "query_tokens": 360, "message_tokens": 269, "message_utf8_bytes": 620, "cache_tensor_bytes": null, "serialized_payload_bytes": 620, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 34889.89491702523, "hops": {"1": 22661.616917001083, "2": 12228.278000024147}, "receiver_prefill_ms": 241.8085839599371, "ttft_ms": 54.18358405586332, "decode_ms": 554.5205420348793}, "total_wall_ms": 35740.90337497182, "ttft_ms": 54.18358405586332, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 8}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "11946000", "evidence_artifact": {"ledger": {"facts_total": 64, "facts_retained": 9, "retention_rate": 0.140625, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.16666666666666666, "status": 0.3333333333333333, "place": 0.0, "identifier": 0.0, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 572}, "ledger_by_hop": {"1": {"facts_total": 64, "facts_retained": 10, "retention_rate": 0.15625, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.2, "amount": 0.16666666666666666, "status": 0.3333333333333333, "place": 0.0, "identifier": 0.0, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 669}, "2": {"facts_total": 64, "facts_retained": 9, "retention_rate": 0.140625, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.16666666666666666, "status": 0.3333333333333333, "place": 0.0, "identifier": 0.0, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 572}}}, "reference_key_id": "keys/synthetic/syn_03006_2000.json", "task_scores": {"correct": 0.0, "abstained": 0.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "175d3ea2044e4182288513cfbf6fbcb45bd8241379081d5258569d283ccfe60c", "output_hash": "edf27f197d1395ea4e7b86fae1feecd3591246e1299bba6a08655f5724a670ea"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "66e5e40f1555fa0beb4f97b9", "attempt_id": "528f77916f32", "run_group_id": "E4_core:A5|closed|qwen3b:h2", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T06:59:23.226078Z", "task_family": "synthetic", "document_id": "syn_03006_2000", "document_hash": "89be5e77e0ca114963e9ffdf8dc38fe5610f13a78ac29e7a1956cd9124ce5217", "question_id": "q13_unanswerable", "split": "test", "cluster_id": "syn_03006_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 2, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1890, "query_tokens": 364, "message_tokens": 269, "message_utf8_bytes": 620, "cache_tensor_bytes": null, "serialized_payload_bytes": 620, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 34889.89491702523, "hops": {"1": 22661.616917001083, "2": 12228.278000024147}, "receiver_prefill_ms": 272.6043750299141, "ttft_ms": 42.97141602728516, "decode_ms": 24.462874978780746}, "total_wall_ms": 35230.61341699213, "ttft_ms": 42.97141602728516, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 64, "facts_retained": 9, "retention_rate": 0.140625, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.16666666666666666, "status": 0.3333333333333333, "place": 0.0, "identifier": 0.0, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 572}, "ledger_by_hop": {"1": {"facts_total": 64, "facts_retained": 10, "retention_rate": 0.15625, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.2, "amount": 0.16666666666666666, "status": 0.3333333333333333, "place": 0.0, "identifier": 0.0, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 669}, "2": {"facts_total": 64, "facts_retained": 9, "retention_rate": 0.140625, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.16666666666666666, "status": 0.3333333333333333, "place": 0.0, "identifier": 0.0, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 572}}}, "reference_key_id": "keys/synthetic/syn_03006_2000.json", "task_scores": {"correct": 1.0, "abstained": 1.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "175d3ea2044e4182288513cfbf6fbcb45bd8241379081d5258569d283ccfe60c", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "bca8ffe0692da0c681c88063", "attempt_id": "9815b721ab28", "run_group_id": "E4_core:A5|closed|qwen3b:h2", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T06:59:23.583874Z", "task_family": "synthetic", "document_id": "syn_03006_2000", "document_hash": "89be5e77e0ca114963e9ffdf8dc38fe5610f13a78ac29e7a1956cd9124ce5217", "question_id": "q14_unanswerable", "split": "test", "cluster_id": "syn_03006_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 2, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1890, "query_tokens": 365, "message_tokens": 269, "message_utf8_bytes": 620, "cache_tensor_bytes": null, "serialized_payload_bytes": 620, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 34889.89491702523, "hops": {"1": 22661.616917001083, "2": 12228.278000024147}, "receiver_prefill_ms": 276.11466601956636, "ttft_ms": 25.977459037676454, "decode_ms": 38.370874943211675}, "total_wall_ms": 35230.84708303213, "ttft_ms": 25.977459037676454, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 64, "facts_retained": 9, "retention_rate": 0.140625, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.16666666666666666, "status": 0.3333333333333333, "place": 0.0, "identifier": 0.0, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 572}, "ledger_by_hop": {"1": {"facts_total": 64, "facts_retained": 10, "retention_rate": 0.15625, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.2, "amount": 0.16666666666666666, "status": 0.3333333333333333, "place": 0.0, "identifier": 0.0, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 669}, "2": {"facts_total": 64, "facts_retained": 9, "retention_rate": 0.140625, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.16666666666666666, "status": 0.3333333333333333, "place": 0.0, "identifier": 0.0, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 572}}}, "reference_key_id": "keys/synthetic/syn_03006_2000.json", "task_scores": {"correct": 1.0, "abstained": 1.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "175d3ea2044e4182288513cfbf6fbcb45bd8241379081d5258569d283ccfe60c", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "69fbebcaf8adffb7fc0656d9", "attempt_id": "cca96e26e3c4", "run_group_id": "E4_core:A5|closed|qwen3b:h4", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T06:59:48.173722Z", "task_family": "synthetic", "document_id": "syn_03006_2000", "document_hash": "89be5e77e0ca114963e9ffdf8dc38fe5610f13a78ac29e7a1956cd9124ce5217", "question_id": "q00_single_fact", "split": "test", "cluster_id": "syn_03006_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 4, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1890, "query_tokens": 352, "message_tokens": 269, "message_utf8_bytes": 620, "cache_tensor_bytes": null, "serialized_payload_bytes": 620, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 59117.748667136766, "hops": {"1": 22661.616917001083, "2": 12228.278000024147, "3": 12082.4769170722, "4": 12145.376833039336}, "receiver_prefill_ms": 289.9216669611633, "ttft_ms": 33.217832911759615, "decode_ms": 29.69995909370482}, "total_wall_ms": 59470.978917204775, "ttft_ms": 33.217832911759615, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 64, "facts_retained": 9, "retention_rate": 0.140625, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.16666666666666666, "status": 0.3333333333333333, "place": 0.0, "identifier": 0.0, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 572}, "ledger_by_hop": {"1": {"facts_total": 64, "facts_retained": 10, "retention_rate": 0.15625, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.2, "amount": 0.16666666666666666, "status": 0.3333333333333333, "place": 0.0, "identifier": 0.0, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 669}, "2": {"facts_total": 64, "facts_retained": 9, "retention_rate": 0.140625, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.16666666666666666, "status": 0.3333333333333333, "place": 0.0, "identifier": 0.0, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 572}, "3": {"facts_total": 64, "facts_retained": 9, "retention_rate": 0.140625, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.16666666666666666, "status": 0.3333333333333333, "place": 0.0, "identifier": 0.0, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 572}, "4": {"facts_total": 64, "facts_retained": 9, "retention_rate": 0.140625, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.16666666666666666, "status": 0.3333333333333333, "place": 0.0, "identifier": 0.0, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 572}}}, "reference_key_id": "keys/synthetic/syn_03006_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "175d3ea2044e4182288513cfbf6fbcb45bd8241379081d5258569d283ccfe60c", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "014189216ea26ca804013f9f", "attempt_id": "5c4ac563e06b", "run_group_id": "E4_core:A5|closed|qwen3b:h4", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T06:59:48.544473Z", "task_family": "synthetic", "document_id": "syn_03006_2000", "document_hash": "89be5e77e0ca114963e9ffdf8dc38fe5610f13a78ac29e7a1956cd9124ce5217", "question_id": "q01_single_fact", "split": "test", "cluster_id": "syn_03006_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 4, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1890, "query_tokens": 357, "message_tokens": 269, "message_utf8_bytes": 620, "cache_tensor_bytes": null, "serialized_payload_bytes": 620, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 59117.748667136766, "hops": {"1": 22661.616917001083, "2": 12228.278000024147, "3": 12082.4769170722, "4": 12145.376833039336}, "receiver_prefill_ms": 271.26970794051886, "ttft_ms": 41.555625037290156, "decode_ms": 55.86441699415445}, "total_wall_ms": 59486.95866717026, "ttft_ms": 41.555625037290156, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 64, "facts_retained": 9, "retention_rate": 0.140625, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.16666666666666666, "status": 0.3333333333333333, "place": 0.0, "identifier": 0.0, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 572}, "ledger_by_hop": {"1": {"facts_total": 64, "facts_retained": 10, "retention_rate": 0.15625, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.2, "amount": 0.16666666666666666, "status": 0.3333333333333333, "place": 0.0, "identifier": 0.0, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 669}, "2": {"facts_total": 64, "facts_retained": 9, "retention_rate": 0.140625, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.16666666666666666, "status": 0.3333333333333333, "place": 0.0, "identifier": 0.0, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 572}, "3": {"facts_total": 64, "facts_retained": 9, "retention_rate": 0.140625, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.16666666666666666, "status": 0.3333333333333333, "place": 0.0, "identifier": 0.0, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 572}, "4": {"facts_total": 64, "facts_retained": 9, "retention_rate": 0.140625, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.16666666666666666, "status": 0.3333333333333333, "place": 0.0, "identifier": 0.0, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 572}}}, "reference_key_id": "keys/synthetic/syn_03006_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "175d3ea2044e4182288513cfbf6fbcb45bd8241379081d5258569d283ccfe60c", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "5b99908c09cae259979e34a1", "attempt_id": "5503b90c62d8", "run_group_id": "E4_core:A5|closed|qwen3b:h4", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T06:59:48.930904Z", "task_family": "synthetic", "document_id": "syn_03006_2000", "document_hash": "89be5e77e0ca114963e9ffdf8dc38fe5610f13a78ac29e7a1956cd9124ce5217", "question_id": "q02_exact_copy", "split": "test", "cluster_id": "syn_03006_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 4, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1890, "query_tokens": 356, "message_tokens": 269, "message_utf8_bytes": 620, "cache_tensor_bytes": null, "serialized_payload_bytes": 620, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 59117.748667136766, "hops": {"1": 22661.616917001083, "2": 12228.278000024147, "3": 12082.4769170722, "4": 12145.376833039336}, "receiver_prefill_ms": 308.46550001297146, "ttft_ms": 98.75991696026176, "decode_ms": 40.20883294288069}, "total_wall_ms": 59565.84233313333, "ttft_ms": 98.75991696026176, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 64, "facts_retained": 9, "retention_rate": 0.140625, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.16666666666666666, "status": 0.3333333333333333, "place": 0.0, "identifier": 0.0, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 572}, "ledger_by_hop": {"1": {"facts_total": 64, "facts_retained": 10, "retention_rate": 0.15625, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.2, "amount": 0.16666666666666666, "status": 0.3333333333333333, "place": 0.0, "identifier": 0.0, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 669}, "2": {"facts_total": 64, "facts_retained": 9, "retention_rate": 0.140625, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.16666666666666666, "status": 0.3333333333333333, "place": 0.0, "identifier": 0.0, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 572}, "3": {"facts_total": 64, "facts_retained": 9, "retention_rate": 0.140625, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.16666666666666666, "status": 0.3333333333333333, "place": 0.0, "identifier": 0.0, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 572}, "4": {"facts_total": 64, "facts_retained": 9, "retention_rate": 0.140625, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.16666666666666666, "status": 0.3333333333333333, "place": 0.0, "identifier": 0.0, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 572}}}, "reference_key_id": "keys/synthetic/syn_03006_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "175d3ea2044e4182288513cfbf6fbcb45bd8241379081d5258569d283ccfe60c", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "c5c686d3565916ae51fad7be", "attempt_id": "9211384fdddc", "run_group_id": "E4_core:A5|closed|qwen3b:h4", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T06:59:49.397314Z", "task_family": "synthetic", "document_id": "syn_03006_2000", "document_hash": "89be5e77e0ca114963e9ffdf8dc38fe5610f13a78ac29e7a1956cd9124ce5217", "question_id": "q03_exact_copy", "split": "test", "cluster_id": "syn_03006_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 4, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1890, "query_tokens": 351, "message_tokens": 269, "message_utf8_bytes": 620, "cache_tensor_bytes": null, "serialized_payload_bytes": 620, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 59117.748667136766, "hops": {"1": 22661.616917001083, "2": 12228.278000024147, "3": 12082.4769170722, "4": 12145.376833039336}, "receiver_prefill_ms": 287.5047499546781, "ttft_ms": 43.604834005236626, "decode_ms": 26.409249985590577}, "total_wall_ms": 59475.65862617921, "ttft_ms": 43.604834005236626, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 64, "facts_retained": 9, "retention_rate": 0.140625, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.16666666666666666, "status": 0.3333333333333333, "place": 0.0, "identifier": 0.0, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 572}, "ledger_by_hop": {"1": {"facts_total": 64, "facts_retained": 10, "retention_rate": 0.15625, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.2, "amount": 0.16666666666666666, "status": 0.3333333333333333, "place": 0.0, "identifier": 0.0, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 669}, "2": {"facts_total": 64, "facts_retained": 9, "retention_rate": 0.140625, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.16666666666666666, "status": 0.3333333333333333, "place": 0.0, "identifier": 0.0, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 572}, "3": {"facts_total": 64, "facts_retained": 9, "retention_rate": 0.140625, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.16666666666666666, "status": 0.3333333333333333, "place": 0.0, "identifier": 0.0, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 572}, "4": {"facts_total": 64, "facts_retained": 9, "retention_rate": 0.140625, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.16666666666666666, "status": 0.3333333333333333, "place": 0.0, "identifier": 0.0, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 572}}}, "reference_key_id": "keys/synthetic/syn_03006_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "175d3ea2044e4182288513cfbf6fbcb45bd8241379081d5258569d283ccfe60c", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "640228cdd82a437cfa9f422d", "attempt_id": "e82cef308267", "run_group_id": "E4_core:A5|closed|qwen3b:h4", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T06:59:49.772717Z", "task_family": "synthetic", "document_id": "syn_03006_2000", "document_hash": "89be5e77e0ca114963e9ffdf8dc38fe5610f13a78ac29e7a1956cd9124ce5217", "question_id": "q04_multi_hop", "split": "test", "cluster_id": "syn_03006_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 4, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1890, "query_tokens": 360, "message_tokens": 269, "message_utf8_bytes": 620, "cache_tensor_bytes": null, "serialized_payload_bytes": 620, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 59117.748667136766, "hops": {"1": 22661.616917001083, "2": 12228.278000024147, "3": 12082.4769170722, "4": 12145.376833039336}, "receiver_prefill_ms": 274.3856250308454, "ttft_ms": 29.013041988946497, "decode_ms": 35.724167013540864}, "total_wall_ms": 59457.32108410448, "ttft_ms": 29.013041988946497, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 64, "facts_retained": 9, "retention_rate": 0.140625, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.16666666666666666, "status": 0.3333333333333333, "place": 0.0, "identifier": 0.0, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 572}, "ledger_by_hop": {"1": {"facts_total": 64, "facts_retained": 10, "retention_rate": 0.15625, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.2, "amount": 0.16666666666666666, "status": 0.3333333333333333, "place": 0.0, "identifier": 0.0, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 669}, "2": {"facts_total": 64, "facts_retained": 9, "retention_rate": 0.140625, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.16666666666666666, "status": 0.3333333333333333, "place": 0.0, "identifier": 0.0, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 572}, "3": {"facts_total": 64, "facts_retained": 9, "retention_rate": 0.140625, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.16666666666666666, "status": 0.3333333333333333, "place": 0.0, "identifier": 0.0, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 572}, "4": {"facts_total": 64, "facts_retained": 9, "retention_rate": 0.140625, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.16666666666666666, "status": 0.3333333333333333, "place": 0.0, "identifier": 0.0, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 572}}}, "reference_key_id": "keys/synthetic/syn_03006_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "175d3ea2044e4182288513cfbf6fbcb45bd8241379081d5258569d283ccfe60c", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "b76921454ce0a426834a8b1f", "attempt_id": "48569021523a", "run_group_id": "E4_core:A5|closed|qwen3b:h4", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T06:59:50.129084Z", "task_family": "synthetic", "document_id": "syn_03006_2000", "document_hash": "89be5e77e0ca114963e9ffdf8dc38fe5610f13a78ac29e7a1956cd9124ce5217", "question_id": "q05_multi_hop", "split": "test", "cluster_id": "syn_03006_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 4, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1890, "query_tokens": 363, "message_tokens": 269, "message_utf8_bytes": 620, "cache_tensor_bytes": null, "serialized_payload_bytes": 620, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 59117.748667136766, "hops": {"1": 22661.616917001083, "2": 12228.278000024147, "3": 12082.4769170722, "4": 12145.376833039336}, "receiver_prefill_ms": 245.84008299279958, "ttft_ms": 41.18612501770258, "decode_ms": 47.43720800615847}, "total_wall_ms": 59452.67191703897, "ttft_ms": 41.18612501770258, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 64, "facts_retained": 9, "retention_rate": 0.140625, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.16666666666666666, "status": 0.3333333333333333, "place": 0.0, "identifier": 0.0, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 572}, "ledger_by_hop": {"1": {"facts_total": 64, "facts_retained": 10, "retention_rate": 0.15625, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.2, "amount": 0.16666666666666666, "status": 0.3333333333333333, "place": 0.0, "identifier": 0.0, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 669}, "2": {"facts_total": 64, "facts_retained": 9, "retention_rate": 0.140625, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.16666666666666666, "status": 0.3333333333333333, "place": 0.0, "identifier": 0.0, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 572}, "3": {"facts_total": 64, "facts_retained": 9, "retention_rate": 0.140625, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.16666666666666666, "status": 0.3333333333333333, "place": 0.0, "identifier": 0.0, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 572}, "4": {"facts_total": 64, "facts_retained": 9, "retention_rate": 0.140625, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.16666666666666666, "status": 0.3333333333333333, "place": 0.0, "identifier": 0.0, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 572}}}, "reference_key_id": "keys/synthetic/syn_03006_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "175d3ea2044e4182288513cfbf6fbcb45bd8241379081d5258569d283ccfe60c", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "5b297596bd3e3c9e3ab28f3d", "attempt_id": "8510a8047e1c", "run_group_id": "E4_core:A5|closed|qwen3b:h4", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T06:59:50.480944Z", "task_family": "synthetic", "document_id": "syn_03006_2000", "document_hash": "89be5e77e0ca114963e9ffdf8dc38fe5610f13a78ac29e7a1956cd9124ce5217", "question_id": "q06_negation", "split": "test", "cluster_id": "syn_03006_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 4, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1890, "query_tokens": 351, "message_tokens": 269, "message_utf8_bytes": 620, "cache_tensor_bytes": null, "serialized_payload_bytes": 620, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 59117.748667136766, "hops": {"1": 22661.616917001083, "2": 12228.278000024147, "3": 12082.4769170722, "4": 12145.376833039336}, "receiver_prefill_ms": 310.68474997300655, "ttft_ms": 136.16533298045397, "decode_ms": 45.340291922912}, "total_wall_ms": 59610.41541711893, "ttft_ms": 136.16533298045397, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 64, "facts_retained": 9, "retention_rate": 0.140625, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.16666666666666666, "status": 0.3333333333333333, "place": 0.0, "identifier": 0.0, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 572}, "ledger_by_hop": {"1": {"facts_total": 64, "facts_retained": 10, "retention_rate": 0.15625, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.2, "amount": 0.16666666666666666, "status": 0.3333333333333333, "place": 0.0, "identifier": 0.0, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 669}, "2": {"facts_total": 64, "facts_retained": 9, "retention_rate": 0.140625, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.16666666666666666, "status": 0.3333333333333333, "place": 0.0, "identifier": 0.0, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 572}, "3": {"facts_total": 64, "facts_retained": 9, "retention_rate": 0.140625, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.16666666666666666, "status": 0.3333333333333333, "place": 0.0, "identifier": 0.0, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 572}, "4": {"facts_total": 64, "facts_retained": 9, "retention_rate": 0.140625, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.16666666666666666, "status": 0.3333333333333333, "place": 0.0, "identifier": 0.0, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 572}}}, "reference_key_id": "keys/synthetic/syn_03006_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "175d3ea2044e4182288513cfbf6fbcb45bd8241379081d5258569d283ccfe60c", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "6793ac78911ae900c6ab42f2", "attempt_id": "afd873f2a9c3", "run_group_id": "E4_core:A5|closed|qwen3b:h4", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T06:59:50.991726Z", "task_family": "synthetic", "document_id": "syn_03006_2000", "document_hash": "89be5e77e0ca114963e9ffdf8dc38fe5610f13a78ac29e7a1956cd9124ce5217", "question_id": "q07_negation", "split": "test", "cluster_id": "syn_03006_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 4, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1890, "query_tokens": 351, "message_tokens": 269, "message_utf8_bytes": 620, "cache_tensor_bytes": null, "serialized_payload_bytes": 620, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 59117.748667136766, "hops": {"1": 22661.616917001083, "2": 12228.278000024147, "3": 12082.4769170722, "4": 12145.376833039336}, "receiver_prefill_ms": 284.4223330030218, "ttft_ms": 42.55074996035546, "decode_ms": 24.638375034555793}, "total_wall_ms": 59469.829209148884, "ttft_ms": 42.55074996035546, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 64, "facts_retained": 9, "retention_rate": 0.140625, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.16666666666666666, "status": 0.3333333333333333, "place": 0.0, "identifier": 0.0, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 572}, "ledger_by_hop": {"1": {"facts_total": 64, "facts_retained": 10, "retention_rate": 0.15625, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.2, "amount": 0.16666666666666666, "status": 0.3333333333333333, "place": 0.0, "identifier": 0.0, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 669}, "2": {"facts_total": 64, "facts_retained": 9, "retention_rate": 0.140625, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.16666666666666666, "status": 0.3333333333333333, "place": 0.0, "identifier": 0.0, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 572}, "3": {"facts_total": 64, "facts_retained": 9, "retention_rate": 0.140625, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.16666666666666666, "status": 0.3333333333333333, "place": 0.0, "identifier": 0.0, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 572}, "4": {"facts_total": 64, "facts_retained": 9, "retention_rate": 0.140625, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.16666666666666666, "status": 0.3333333333333333, "place": 0.0, "identifier": 0.0, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 572}}}, "reference_key_id": "keys/synthetic/syn_03006_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "175d3ea2044e4182288513cfbf6fbcb45bd8241379081d5258569d283ccfe60c", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "efecb4f645619c0cf45a0d5c", "attempt_id": "6821ee4b6196", "run_group_id": "E4_core:A5|closed|qwen3b:h4", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T06:59:51.361244Z", "task_family": "synthetic", "document_id": "syn_03006_2000", "document_hash": "89be5e77e0ca114963e9ffdf8dc38fe5610f13a78ac29e7a1956cd9124ce5217", "question_id": "q08_revision", "split": "test", "cluster_id": "syn_03006_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 4, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1890, "query_tokens": 368, "message_tokens": 269, "message_utf8_bytes": 620, "cache_tensor_bytes": null, "serialized_payload_bytes": 620, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 59117.748667136766, "hops": {"1": 22661.616917001083, "2": 12228.278000024147, "3": 12082.4769170722, "4": 12145.376833039336}, "receiver_prefill_ms": 278.4819590160623, "ttft_ms": 26.20349999051541, "decode_ms": 335.0726669887081}, "total_wall_ms": 59757.978542125784, "ttft_ms": 26.20349999051541, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 8}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "10107545", "evidence_artifact": {"ledger": {"facts_total": 64, "facts_retained": 9, "retention_rate": 0.140625, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.16666666666666666, "status": 0.3333333333333333, "place": 0.0, "identifier": 0.0, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 572}, "ledger_by_hop": {"1": {"facts_total": 64, "facts_retained": 10, "retention_rate": 0.15625, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.2, "amount": 0.16666666666666666, "status": 0.3333333333333333, "place": 0.0, "identifier": 0.0, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 669}, "2": {"facts_total": 64, "facts_retained": 9, "retention_rate": 0.140625, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.16666666666666666, "status": 0.3333333333333333, "place": 0.0, "identifier": 0.0, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 572}, "3": {"facts_total": 64, "facts_retained": 9, "retention_rate": 0.140625, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.16666666666666666, "status": 0.3333333333333333, "place": 0.0, "identifier": 0.0, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 572}, "4": {"facts_total": 64, "facts_retained": 9, "retention_rate": 0.140625, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.16666666666666666, "status": 0.3333333333333333, "place": 0.0, "identifier": 0.0, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 572}}}, "reference_key_id": "keys/synthetic/syn_03006_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "175d3ea2044e4182288513cfbf6fbcb45bd8241379081d5258569d283ccfe60c", "output_hash": "6faba9de4e29949bbdd8934f69fc2110f07076db6801b211aeca9351292537fe"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "c1a8e29367b599112b4138ac", "attempt_id": "428b17746693", "run_group_id": "E4_core:A5|closed|qwen3b:h4", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T06:59:52.019063Z", "task_family": "synthetic", "document_id": "syn_03006_2000", "document_hash": "89be5e77e0ca114963e9ffdf8dc38fe5610f13a78ac29e7a1956cd9124ce5217", "question_id": "q09_revision", "split": "test", "cluster_id": "syn_03006_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 4, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1890, "query_tokens": 361, "message_tokens": 269, "message_utf8_bytes": 620, "cache_tensor_bytes": null, "serialized_payload_bytes": 620, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 59117.748667136766, "hops": {"1": 22661.616917001083, "2": 12228.278000024147, "3": 12082.4769170722, "4": 12145.376833039336}, "receiver_prefill_ms": 432.62816697824746, "ttft_ms": 40.2778749121353, "decode_ms": 236.39645904768258}, "total_wall_ms": 59827.578834141605, "ttft_ms": 40.2778749121353, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 7}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "9112735", "evidence_artifact": {"ledger": {"facts_total": 64, "facts_retained": 9, "retention_rate": 0.140625, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.16666666666666666, "status": 0.3333333333333333, "place": 0.0, "identifier": 0.0, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 572}, "ledger_by_hop": {"1": {"facts_total": 64, "facts_retained": 10, "retention_rate": 0.15625, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.2, "amount": 0.16666666666666666, "status": 0.3333333333333333, "place": 0.0, "identifier": 0.0, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 669}, "2": {"facts_total": 64, "facts_retained": 9, "retention_rate": 0.140625, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.16666666666666666, "status": 0.3333333333333333, "place": 0.0, "identifier": 0.0, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 572}, "3": {"facts_total": 64, "facts_retained": 9, "retention_rate": 0.140625, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.16666666666666666, "status": 0.3333333333333333, "place": 0.0, "identifier": 0.0, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 572}, "4": {"facts_total": 64, "facts_retained": 9, "retention_rate": 0.140625, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.16666666666666666, "status": 0.3333333333333333, "place": 0.0, "identifier": 0.0, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 572}}}, "reference_key_id": "keys/synthetic/syn_03006_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "175d3ea2044e4182288513cfbf6fbcb45bd8241379081d5258569d283ccfe60c", "output_hash": "15a3f804636f0e62d4a3d9be7677f001a0ad6b1fb8ae7e8cdc63c44ec7bac31f"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "90340d48a2db09e3d89cdbd6", "attempt_id": "c4f0543921da", "run_group_id": "E4_core:A5|closed|qwen3b:h4", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T06:59:52.746608Z", "task_family": "synthetic", "document_id": "syn_03006_2000", "document_hash": "89be5e77e0ca114963e9ffdf8dc38fe5610f13a78ac29e7a1956cd9124ce5217", "question_id": "q10_temporal", "split": "test", "cluster_id": "syn_03006_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 4, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1890, "query_tokens": 370, "message_tokens": 269, "message_utf8_bytes": 620, "cache_tensor_bytes": null, "serialized_payload_bytes": 620, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 59117.748667136766, "hops": {"1": 22661.616917001083, "2": 12228.278000024147, "3": 12082.4769170722, "4": 12145.376833039336}, "receiver_prefill_ms": 273.8064581062645, "ttft_ms": 31.095291953533888, "decode_ms": 26.571166003122926}, "total_wall_ms": 59449.641751125455, "ttft_ms": 31.095291953533888, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 64, "facts_retained": 9, "retention_rate": 0.140625, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.16666666666666666, "status": 0.3333333333333333, "place": 0.0, "identifier": 0.0, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 572}, "ledger_by_hop": {"1": {"facts_total": 64, "facts_retained": 10, "retention_rate": 0.15625, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.2, "amount": 0.16666666666666666, "status": 0.3333333333333333, "place": 0.0, "identifier": 0.0, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 669}, "2": {"facts_total": 64, "facts_retained": 9, "retention_rate": 0.140625, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.16666666666666666, "status": 0.3333333333333333, "place": 0.0, "identifier": 0.0, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 572}, "3": {"facts_total": 64, "facts_retained": 9, "retention_rate": 0.140625, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.16666666666666666, "status": 0.3333333333333333, "place": 0.0, "identifier": 0.0, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 572}, "4": {"facts_total": 64, "facts_retained": 9, "retention_rate": 0.140625, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.16666666666666666, "status": 0.3333333333333333, "place": 0.0, "identifier": 0.0, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 572}}}, "reference_key_id": "keys/synthetic/syn_03006_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "175d3ea2044e4182288513cfbf6fbcb45bd8241379081d5258569d283ccfe60c", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "6843d34195d308ef4fd6e099", "attempt_id": "b2284e5c3b5f", "run_group_id": "E4_core:A5|closed|qwen3b:h4", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T06:59:53.096353Z", "task_family": "synthetic", "document_id": "syn_03006_2000", "document_hash": "89be5e77e0ca114963e9ffdf8dc38fe5610f13a78ac29e7a1956cd9124ce5217", "question_id": "q11_aggregation", "split": "test", "cluster_id": "syn_03006_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 4, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1890, "query_tokens": 363, "message_tokens": 269, "message_utf8_bytes": 620, "cache_tensor_bytes": null, "serialized_payload_bytes": 620, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 59117.748667136766, "hops": {"1": 22661.616917001083, "2": 12228.278000024147, "3": 12082.4769170722, "4": 12145.376833039336}, "receiver_prefill_ms": 243.1956670479849, "ttft_ms": 31.470082933083177, "decode_ms": 48.45041607040912}, "total_wall_ms": 59441.436959197745, "ttft_ms": 31.470082933083177, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 64, "facts_retained": 9, "retention_rate": 0.140625, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.16666666666666666, "status": 0.3333333333333333, "place": 0.0, "identifier": 0.0, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 572}, "ledger_by_hop": {"1": {"facts_total": 64, "facts_retained": 10, "retention_rate": 0.15625, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.2, "amount": 0.16666666666666666, "status": 0.3333333333333333, "place": 0.0, "identifier": 0.0, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 669}, "2": {"facts_total": 64, "facts_retained": 9, "retention_rate": 0.140625, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.16666666666666666, "status": 0.3333333333333333, "place": 0.0, "identifier": 0.0, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 572}, "3": {"facts_total": 64, "facts_retained": 9, "retention_rate": 0.140625, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.16666666666666666, "status": 0.3333333333333333, "place": 0.0, "identifier": 0.0, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 572}, "4": {"facts_total": 64, "facts_retained": 9, "retention_rate": 0.140625, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.16666666666666666, "status": 0.3333333333333333, "place": 0.0, "identifier": 0.0, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 572}}}, "reference_key_id": "keys/synthetic/syn_03006_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "175d3ea2044e4182288513cfbf6fbcb45bd8241379081d5258569d283ccfe60c", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "06e3ba2039718b6ab2ef958a", "attempt_id": "7cba1892169e", "run_group_id": "E4_core:A5|closed|qwen3b:h4", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T06:59:53.437218Z", "task_family": "synthetic", "document_id": "syn_03006_2000", "document_hash": "89be5e77e0ca114963e9ffdf8dc38fe5610f13a78ac29e7a1956cd9124ce5217", "question_id": "q12_unit_conversion", "split": "test", "cluster_id": "syn_03006_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 4, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1890, "query_tokens": 360, "message_tokens": 269, "message_utf8_bytes": 620, "cache_tensor_bytes": null, "serialized_payload_bytes": 620, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 59117.748667136766, "hops": {"1": 22661.616917001083, "2": 12228.278000024147, "3": 12082.4769170722, "4": 12145.376833039336}, "receiver_prefill_ms": 265.3025418985635, "ttft_ms": 35.393749945797026, "decode_ms": 468.66370807401836}, "total_wall_ms": 59887.596292071976, "ttft_ms": 35.393749945797026, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 8}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "11946000", "evidence_artifact": {"ledger": {"facts_total": 64, "facts_retained": 9, "retention_rate": 0.140625, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.16666666666666666, "status": 0.3333333333333333, "place": 0.0, "identifier": 0.0, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 572}, "ledger_by_hop": {"1": {"facts_total": 64, "facts_retained": 10, "retention_rate": 0.15625, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.2, "amount": 0.16666666666666666, "status": 0.3333333333333333, "place": 0.0, "identifier": 0.0, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 669}, "2": {"facts_total": 64, "facts_retained": 9, "retention_rate": 0.140625, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.16666666666666666, "status": 0.3333333333333333, "place": 0.0, "identifier": 0.0, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 572}, "3": {"facts_total": 64, "facts_retained": 9, "retention_rate": 0.140625, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.16666666666666666, "status": 0.3333333333333333, "place": 0.0, "identifier": 0.0, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 572}, "4": {"facts_total": 64, "facts_retained": 9, "retention_rate": 0.140625, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.16666666666666666, "status": 0.3333333333333333, "place": 0.0, "identifier": 0.0, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 572}}}, "reference_key_id": "keys/synthetic/syn_03006_2000.json", "task_scores": {"correct": 0.0, "abstained": 0.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "175d3ea2044e4182288513cfbf6fbcb45bd8241379081d5258569d283ccfe60c", "output_hash": "edf27f197d1395ea4e7b86fae1feecd3591246e1299bba6a08655f5724a670ea"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "4b17ca880034e10269f0999b", "attempt_id": "86d926c265be", "run_group_id": "E4_core:A5|closed|qwen3b:h4", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T06:59:54.224023Z", "task_family": "synthetic", "document_id": "syn_03006_2000", "document_hash": "89be5e77e0ca114963e9ffdf8dc38fe5610f13a78ac29e7a1956cd9124ce5217", "question_id": "q13_unanswerable", "split": "test", "cluster_id": "syn_03006_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 4, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1890, "query_tokens": 364, "message_tokens": 269, "message_utf8_bytes": 620, "cache_tensor_bytes": null, "serialized_payload_bytes": 620, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 59117.748667136766, "hops": {"1": 22661.616917001083, "2": 12228.278000024147, "3": 12082.4769170722, "4": 12145.376833039336}, "receiver_prefill_ms": 274.81420803815126, "ttft_ms": 30.637916992418468, "decode_ms": 40.68458406254649}, "total_wall_ms": 59464.39175005071, "ttft_ms": 30.637916992418468, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 64, "facts_retained": 9, "retention_rate": 0.140625, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.16666666666666666, "status": 0.3333333333333333, "place": 0.0, "identifier": 0.0, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 572}, "ledger_by_hop": {"1": {"facts_total": 64, "facts_retained": 10, "retention_rate": 0.15625, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.2, "amount": 0.16666666666666666, "status": 0.3333333333333333, "place": 0.0, "identifier": 0.0, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 669}, "2": {"facts_total": 64, "facts_retained": 9, "retention_rate": 0.140625, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.16666666666666666, "status": 0.3333333333333333, "place": 0.0, "identifier": 0.0, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 572}, "3": {"facts_total": 64, "facts_retained": 9, "retention_rate": 0.140625, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.16666666666666666, "status": 0.3333333333333333, "place": 0.0, "identifier": 0.0, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 572}, "4": {"facts_total": 64, "facts_retained": 9, "retention_rate": 0.140625, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.16666666666666666, "status": 0.3333333333333333, "place": 0.0, "identifier": 0.0, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 572}}}, "reference_key_id": "keys/synthetic/syn_03006_2000.json", "task_scores": {"correct": 1.0, "abstained": 1.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "175d3ea2044e4182288513cfbf6fbcb45bd8241379081d5258569d283ccfe60c", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "9294eeef316d7bb1aa6d1b74", "attempt_id": "490897f27d78", "run_group_id": "E4_core:A5|closed|qwen3b:h4", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T06:59:54.588571Z", "task_family": "synthetic", "document_id": "syn_03006_2000", "document_hash": "89be5e77e0ca114963e9ffdf8dc38fe5610f13a78ac29e7a1956cd9124ce5217", "question_id": "q14_unanswerable", "split": "test", "cluster_id": "syn_03006_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 4, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1890, "query_tokens": 365, "message_tokens": 269, "message_utf8_bytes": 620, "cache_tensor_bytes": null, "serialized_payload_bytes": 620, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 59117.748667136766, "hops": {"1": 22661.616917001083, "2": 12228.278000024147, "3": 12082.4769170722, "4": 12145.376833039336}, "receiver_prefill_ms": 270.24129207711667, "ttft_ms": 19.055416923947632, "decode_ms": 40.54716695100069}, "total_wall_ms": 59448.23604216799, "ttft_ms": 19.055416923947632, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 64, "facts_retained": 9, "retention_rate": 0.140625, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.16666666666666666, "status": 0.3333333333333333, "place": 0.0, "identifier": 0.0, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 572}, "ledger_by_hop": {"1": {"facts_total": 64, "facts_retained": 10, "retention_rate": 0.15625, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.2, "amount": 0.16666666666666666, "status": 0.3333333333333333, "place": 0.0, "identifier": 0.0, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 669}, "2": {"facts_total": 64, "facts_retained": 9, "retention_rate": 0.140625, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.16666666666666666, "status": 0.3333333333333333, "place": 0.0, "identifier": 0.0, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 572}, "3": {"facts_total": 64, "facts_retained": 9, "retention_rate": 0.140625, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.16666666666666666, "status": 0.3333333333333333, "place": 0.0, "identifier": 0.0, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 572}, "4": {"facts_total": 64, "facts_retained": 9, "retention_rate": 0.140625, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.16666666666666666, "status": 0.3333333333333333, "place": 0.0, "identifier": 0.0, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 572}}}, "reference_key_id": "keys/synthetic/syn_03006_2000.json", "task_scores": {"correct": 1.0, "abstained": 1.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "175d3ea2044e4182288513cfbf6fbcb45bd8241379081d5258569d283ccfe60c", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "1218ef30cfc7a7675b438cb9", "attempt_id": "c81491e0c997", "run_group_id": "E4_core:A5|closed|qwen3b:h8", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:00:42.700859Z", "task_family": "synthetic", "document_id": "syn_03006_2000", "document_hash": "89be5e77e0ca114963e9ffdf8dc38fe5610f13a78ac29e7a1956cd9124ce5217", "question_id": "q00_single_fact", "split": "test", "cluster_id": "syn_03006_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 8, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1890, "query_tokens": 352, "message_tokens": 269, "message_utf8_bytes": 620, "cache_tensor_bytes": null, "serialized_payload_bytes": 620, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 106877.52525007818, "hops": {"1": 22661.616917001083, "2": 12228.278000024147, "3": 12082.4769170722, "4": 12145.376833039336, "5": 11973.832083051093, "6": 12040.75849999208, "7": 11826.985499938019, "8": 11918.200499960221}, "receiver_prefill_ms": 288.2002079859376, "ttft_ms": 43.29141695052385, "decode_ms": 55.96141691785306}, "total_wall_ms": 107265.43416711502, "ttft_ms": 43.29141695052385, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 64, "facts_retained": 9, "retention_rate": 0.140625, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.16666666666666666, "status": 0.3333333333333333, "place": 0.0, "identifier": 0.0, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 572}, "ledger_by_hop": {"1": {"facts_total": 64, "facts_retained": 10, "retention_rate": 0.15625, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.2, "amount": 0.16666666666666666, "status": 0.3333333333333333, "place": 0.0, "identifier": 0.0, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 669}, "2": {"facts_total": 64, "facts_retained": 9, "retention_rate": 0.140625, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.16666666666666666, "status": 0.3333333333333333, "place": 0.0, "identifier": 0.0, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 572}, "3": {"facts_total": 64, "facts_retained": 9, "retention_rate": 0.140625, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.16666666666666666, "status": 0.3333333333333333, "place": 0.0, "identifier": 0.0, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 572}, "4": {"facts_total": 64, "facts_retained": 9, "retention_rate": 0.140625, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.16666666666666666, "status": 0.3333333333333333, "place": 0.0, "identifier": 0.0, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 572}, "5": {"facts_total": 64, "facts_retained": 9, "retention_rate": 0.140625, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.16666666666666666, "status": 0.3333333333333333, "place": 0.0, "identifier": 0.0, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 572}, "6": {"facts_total": 64, "facts_retained": 9, "retention_rate": 0.140625, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.16666666666666666, "status": 0.3333333333333333, "place": 0.0, "identifier": 0.0, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 572}, "7": {"facts_total": 64, "facts_retained": 9, "retention_rate": 0.140625, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.16666666666666666, "status": 0.3333333333333333, "place": 0.0, "identifier": 0.0, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 572}, "8": {"facts_total": 64, "facts_retained": 9, "retention_rate": 0.140625, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.16666666666666666, "status": 0.3333333333333333, "place": 0.0, "identifier": 0.0, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 572}}}, "reference_key_id": "keys/synthetic/syn_03006_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "175d3ea2044e4182288513cfbf6fbcb45bd8241379081d5258569d283ccfe60c", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "5327dca08d83582be5f17b51", "attempt_id": "67aefb1c11bc", "run_group_id": "E4_core:A5|closed|qwen3b:h8", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:00:43.108032Z", "task_family": "synthetic", "document_id": "syn_03006_2000", "document_hash": "89be5e77e0ca114963e9ffdf8dc38fe5610f13a78ac29e7a1956cd9124ce5217", "question_id": "q01_single_fact", "split": "test", "cluster_id": "syn_03006_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 8, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1890, "query_tokens": 357, "message_tokens": 269, "message_utf8_bytes": 620, "cache_tensor_bytes": null, "serialized_payload_bytes": 620, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 106877.52525007818, "hops": {"1": 22661.616917001083, "2": 12228.278000024147, "3": 12082.4769170722, "4": 12145.376833039336, "5": 11973.832083051093, "6": 12040.75849999208, "7": 11826.985499938019, "8": 11918.200499960221}, "receiver_prefill_ms": 408.0460829427466, "ttft_ms": 39.14254193659872, "decode_ms": 28.17845798563212}, "total_wall_ms": 107353.44199999236, "ttft_ms": 39.14254193659872, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 64, "facts_retained": 9, "retention_rate": 0.140625, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.16666666666666666, "status": 0.3333333333333333, "place": 0.0, "identifier": 0.0, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 572}, "ledger_by_hop": {"1": {"facts_total": 64, "facts_retained": 10, "retention_rate": 0.15625, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.2, "amount": 0.16666666666666666, "status": 0.3333333333333333, "place": 0.0, "identifier": 0.0, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 669}, "2": {"facts_total": 64, "facts_retained": 9, "retention_rate": 0.140625, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.16666666666666666, "status": 0.3333333333333333, "place": 0.0, "identifier": 0.0, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 572}, "3": {"facts_total": 64, "facts_retained": 9, "retention_rate": 0.140625, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.16666666666666666, "status": 0.3333333333333333, "place": 0.0, "identifier": 0.0, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 572}, "4": {"facts_total": 64, "facts_retained": 9, "retention_rate": 0.140625, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.16666666666666666, "status": 0.3333333333333333, "place": 0.0, "identifier": 0.0, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 572}, "5": {"facts_total": 64, "facts_retained": 9, "retention_rate": 0.140625, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.16666666666666666, "status": 0.3333333333333333, "place": 0.0, "identifier": 0.0, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 572}, "6": {"facts_total": 64, "facts_retained": 9, "retention_rate": 0.140625, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.16666666666666666, "status": 0.3333333333333333, "place": 0.0, "identifier": 0.0, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 572}, "7": {"facts_total": 64, "facts_retained": 9, "retention_rate": 0.140625, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.16666666666666666, "status": 0.3333333333333333, "place": 0.0, "identifier": 0.0, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 572}, "8": {"facts_total": 64, "facts_retained": 9, "retention_rate": 0.140625, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.16666666666666666, "status": 0.3333333333333333, "place": 0.0, "identifier": 0.0, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 572}}}, "reference_key_id": "keys/synthetic/syn_03006_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "175d3ea2044e4182288513cfbf6fbcb45bd8241379081d5258569d283ccfe60c", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "340735fb3db3e93948f58e84", "attempt_id": "0478bb43036c", "run_group_id": "E4_core:A5|closed|qwen3b:h8", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:00:43.601389Z", "task_family": "synthetic", "document_id": "syn_03006_2000", "document_hash": "89be5e77e0ca114963e9ffdf8dc38fe5610f13a78ac29e7a1956cd9124ce5217", "question_id": "q02_exact_copy", "split": "test", "cluster_id": "syn_03006_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 8, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1890, "query_tokens": 356, "message_tokens": 269, "message_utf8_bytes": 620, "cache_tensor_bytes": null, "serialized_payload_bytes": 620, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 106877.52525007818, "hops": {"1": 22661.616917001083, "2": 12228.278000024147, "3": 12082.4769170722, "4": 12145.376833039336, "5": 11973.832083051093, "6": 12040.75849999208, "7": 11826.985499938019, "8": 11918.200499960221}, "receiver_prefill_ms": 263.2488328963518, "ttft_ms": 32.172415987588465, "decode_ms": 42.16841596644372}, "total_wall_ms": 107215.66387498751, "ttft_ms": 32.172415987588465, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 64, "facts_retained": 9, "retention_rate": 0.140625, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.16666666666666666, "status": 0.3333333333333333, "place": 0.0, "identifier": 0.0, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 572}, "ledger_by_hop": {"1": {"facts_total": 64, "facts_retained": 10, "retention_rate": 0.15625, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.2, "amount": 0.16666666666666666, "status": 0.3333333333333333, "place": 0.0, "identifier": 0.0, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 669}, "2": {"facts_total": 64, "facts_retained": 9, "retention_rate": 0.140625, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.16666666666666666, "status": 0.3333333333333333, "place": 0.0, "identifier": 0.0, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 572}, "3": {"facts_total": 64, "facts_retained": 9, "retention_rate": 0.140625, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.16666666666666666, "status": 0.3333333333333333, "place": 0.0, "identifier": 0.0, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 572}, "4": {"facts_total": 64, "facts_retained": 9, "retention_rate": 0.140625, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.16666666666666666, "status": 0.3333333333333333, "place": 0.0, "identifier": 0.0, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 572}, "5": {"facts_total": 64, "facts_retained": 9, "retention_rate": 0.140625, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.16666666666666666, "status": 0.3333333333333333, "place": 0.0, "identifier": 0.0, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 572}, "6": {"facts_total": 64, "facts_retained": 9, "retention_rate": 0.140625, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.16666666666666666, "status": 0.3333333333333333, "place": 0.0, "identifier": 0.0, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 572}, "7": {"facts_total": 64, "facts_retained": 9, "retention_rate": 0.140625, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.16666666666666666, "status": 0.3333333333333333, "place": 0.0, "identifier": 0.0, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 572}, "8": {"facts_total": 64, "facts_retained": 9, "retention_rate": 0.140625, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.16666666666666666, "status": 0.3333333333333333, "place": 0.0, "identifier": 0.0, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 572}}}, "reference_key_id": "keys/synthetic/syn_03006_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "175d3ea2044e4182288513cfbf6fbcb45bd8241379081d5258569d283ccfe60c", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "e23bf92e57b339bb2f6ff6d0", "attempt_id": "175f1e2879c3", "run_group_id": "E4_core:A5|closed|qwen3b:h8", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:00:43.957066Z", "task_family": "synthetic", "document_id": "syn_03006_2000", "document_hash": "89be5e77e0ca114963e9ffdf8dc38fe5610f13a78ac29e7a1956cd9124ce5217", "question_id": "q03_exact_copy", "split": "test", "cluster_id": "syn_03006_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 8, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1890, "query_tokens": 351, "message_tokens": 269, "message_utf8_bytes": 620, "cache_tensor_bytes": null, "serialized_payload_bytes": 620, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 106877.52525007818, "hops": {"1": 22661.616917001083, "2": 12228.278000024147, "3": 12082.4769170722, "4": 12145.376833039336, "5": 11973.832083051093, "6": 12040.75849999208, "7": 11826.985499938019, "8": 11918.200499960221}, "receiver_prefill_ms": 264.28662496618927, "ttft_ms": 29.55187507905066, "decode_ms": 21.28683403134346}, "total_wall_ms": 107193.20404203609, "ttft_ms": 29.55187507905066, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 64, "facts_retained": 9, "retention_rate": 0.140625, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.16666666666666666, "status": 0.3333333333333333, "place": 0.0, "identifier": 0.0, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 572}, "ledger_by_hop": {"1": {"facts_total": 64, "facts_retained": 10, "retention_rate": 0.15625, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.2, "amount": 0.16666666666666666, "status": 0.3333333333333333, "place": 0.0, "identifier": 0.0, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 669}, "2": {"facts_total": 64, "facts_retained": 9, "retention_rate": 0.140625, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.16666666666666666, "status": 0.3333333333333333, "place": 0.0, "identifier": 0.0, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 572}, "3": {"facts_total": 64, "facts_retained": 9, "retention_rate": 0.140625, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.16666666666666666, "status": 0.3333333333333333, "place": 0.0, "identifier": 0.0, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 572}, "4": {"facts_total": 64, "facts_retained": 9, "retention_rate": 0.140625, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.16666666666666666, "status": 0.3333333333333333, "place": 0.0, "identifier": 0.0, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 572}, "5": {"facts_total": 64, "facts_retained": 9, "retention_rate": 0.140625, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.16666666666666666, "status": 0.3333333333333333, "place": 0.0, "identifier": 0.0, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 572}, "6": {"facts_total": 64, "facts_retained": 9, "retention_rate": 0.140625, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.16666666666666666, "status": 0.3333333333333333, "place": 0.0, "identifier": 0.0, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 572}, "7": {"facts_total": 64, "facts_retained": 9, "retention_rate": 0.140625, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.16666666666666666, "status": 0.3333333333333333, "place": 0.0, "identifier": 0.0, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 572}, "8": {"facts_total": 64, "facts_retained": 9, "retention_rate": 0.140625, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.16666666666666666, "status": 0.3333333333333333, "place": 0.0, "identifier": 0.0, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 572}}}, "reference_key_id": "keys/synthetic/syn_03006_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "175d3ea2044e4182288513cfbf6fbcb45bd8241379081d5258569d283ccfe60c", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "0065ff8fd0ea274286e36ed2", "attempt_id": "dd482c5126c6", "run_group_id": "E4_core:A5|closed|qwen3b:h8", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:00:44.290867Z", "task_family": "synthetic", "document_id": "syn_03006_2000", "document_hash": "89be5e77e0ca114963e9ffdf8dc38fe5610f13a78ac29e7a1956cd9124ce5217", "question_id": "q04_multi_hop", "split": "test", "cluster_id": "syn_03006_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 8, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1890, "query_tokens": 360, "message_tokens": 269, "message_utf8_bytes": 620, "cache_tensor_bytes": null, "serialized_payload_bytes": 620, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 106877.52525007818, "hops": {"1": 22661.616917001083, "2": 12228.278000024147, "3": 12082.4769170722, "4": 12145.376833039336, "5": 11973.832083051093, "6": 12040.75849999208, "7": 11826.985499938019, "8": 11918.200499960221}, "receiver_prefill_ms": 292.0402920572087, "ttft_ms": 68.44945799093693, "decode_ms": 56.864332989789546}, "total_wall_ms": 107295.4994160682, "ttft_ms": 68.44945799093693, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 64, "facts_retained": 9, "retention_rate": 0.140625, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.16666666666666666, "status": 0.3333333333333333, "place": 0.0, "identifier": 0.0, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 572}, "ledger_by_hop": {"1": {"facts_total": 64, "facts_retained": 10, "retention_rate": 0.15625, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.2, "amount": 0.16666666666666666, "status": 0.3333333333333333, "place": 0.0, "identifier": 0.0, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 669}, "2": {"facts_total": 64, "facts_retained": 9, "retention_rate": 0.140625, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.16666666666666666, "status": 0.3333333333333333, "place": 0.0, "identifier": 0.0, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 572}, "3": {"facts_total": 64, "facts_retained": 9, "retention_rate": 0.140625, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.16666666666666666, "status": 0.3333333333333333, "place": 0.0, "identifier": 0.0, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 572}, "4": {"facts_total": 64, "facts_retained": 9, "retention_rate": 0.140625, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.16666666666666666, "status": 0.3333333333333333, "place": 0.0, "identifier": 0.0, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 572}, "5": {"facts_total": 64, "facts_retained": 9, "retention_rate": 0.140625, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.16666666666666666, "status": 0.3333333333333333, "place": 0.0, "identifier": 0.0, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 572}, "6": {"facts_total": 64, "facts_retained": 9, "retention_rate": 0.140625, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.16666666666666666, "status": 0.3333333333333333, "place": 0.0, "identifier": 0.0, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 572}, "7": {"facts_total": 64, "facts_retained": 9, "retention_rate": 0.140625, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.16666666666666666, "status": 0.3333333333333333, "place": 0.0, "identifier": 0.0, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 572}, "8": {"facts_total": 64, "facts_retained": 9, "retention_rate": 0.140625, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.16666666666666666, "status": 0.3333333333333333, "place": 0.0, "identifier": 0.0, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 572}}}, "reference_key_id": "keys/synthetic/syn_03006_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "175d3ea2044e4182288513cfbf6fbcb45bd8241379081d5258569d283ccfe60c", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "963b573bbc0e38fa3c62c8a5", "attempt_id": "e9e78cfd90cc", "run_group_id": "E4_core:A5|closed|qwen3b:h8", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:00:44.727750Z", "task_family": "synthetic", "document_id": "syn_03006_2000", "document_hash": "89be5e77e0ca114963e9ffdf8dc38fe5610f13a78ac29e7a1956cd9124ce5217", "question_id": "q05_multi_hop", "split": "test", "cluster_id": "syn_03006_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 8, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1890, "query_tokens": 363, "message_tokens": 269, "message_utf8_bytes": 620, "cache_tensor_bytes": null, "serialized_payload_bytes": 620, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 106877.52525007818, "hops": {"1": 22661.616917001083, "2": 12228.278000024147, "3": 12082.4769170722, "4": 12145.376833039336, "5": 11973.832083051093, "6": 12040.75849999208, "7": 11826.985499938019, "8": 11918.200499960221}, "receiver_prefill_ms": 410.8293750323355, "ttft_ms": 30.757792061194777, "decode_ms": 42.55808307789266}, "total_wall_ms": 107362.1518331347, "ttft_ms": 30.757792061194777, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 64, "facts_retained": 9, "retention_rate": 0.140625, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.16666666666666666, "status": 0.3333333333333333, "place": 0.0, "identifier": 0.0, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 572}, "ledger_by_hop": {"1": {"facts_total": 64, "facts_retained": 10, "retention_rate": 0.15625, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.2, "amount": 0.16666666666666666, "status": 0.3333333333333333, "place": 0.0, "identifier": 0.0, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 669}, "2": {"facts_total": 64, "facts_retained": 9, "retention_rate": 0.140625, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.16666666666666666, "status": 0.3333333333333333, "place": 0.0, "identifier": 0.0, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 572}, "3": {"facts_total": 64, "facts_retained": 9, "retention_rate": 0.140625, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.16666666666666666, "status": 0.3333333333333333, "place": 0.0, "identifier": 0.0, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 572}, "4": {"facts_total": 64, "facts_retained": 9, "retention_rate": 0.140625, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.16666666666666666, "status": 0.3333333333333333, "place": 0.0, "identifier": 0.0, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 572}, "5": {"facts_total": 64, "facts_retained": 9, "retention_rate": 0.140625, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.16666666666666666, "status": 0.3333333333333333, "place": 0.0, "identifier": 0.0, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 572}, "6": {"facts_total": 64, "facts_retained": 9, "retention_rate": 0.140625, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.16666666666666666, "status": 0.3333333333333333, "place": 0.0, "identifier": 0.0, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 572}, "7": {"facts_total": 64, "facts_retained": 9, "retention_rate": 0.140625, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.16666666666666666, "status": 0.3333333333333333, "place": 0.0, "identifier": 0.0, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 572}, "8": {"facts_total": 64, "facts_retained": 9, "retention_rate": 0.140625, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.16666666666666666, "status": 0.3333333333333333, "place": 0.0, "identifier": 0.0, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 572}}}, "reference_key_id": "keys/synthetic/syn_03006_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "175d3ea2044e4182288513cfbf6fbcb45bd8241379081d5258569d283ccfe60c", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "a4f17e9a72c37c9eb97f9908", "attempt_id": "eeec5f9f50d7", "run_group_id": "E4_core:A5|closed|qwen3b:h8", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:00:45.229223Z", "task_family": "synthetic", "document_id": "syn_03006_2000", "document_hash": "89be5e77e0ca114963e9ffdf8dc38fe5610f13a78ac29e7a1956cd9124ce5217", "question_id": "q06_negation", "split": "test", "cluster_id": "syn_03006_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 8, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1890, "query_tokens": 351, "message_tokens": 269, "message_utf8_bytes": 620, "cache_tensor_bytes": null, "serialized_payload_bytes": 620, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 106877.52525007818, "hops": {"1": 22661.616917001083, "2": 12228.278000024147, "3": 12082.4769170722, "4": 12145.376833039336, "5": 11973.832083051093, "6": 12040.75849999208, "7": 11826.985499938019, "8": 11918.200499960221}, "receiver_prefill_ms": 262.99345795996487, "ttft_ms": 40.24820798076689, "decode_ms": 44.645125046372414}, "total_wall_ms": 107225.83216708153, "ttft_ms": 40.24820798076689, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 64, "facts_retained": 9, "retention_rate": 0.140625, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.16666666666666666, "status": 0.3333333333333333, "place": 0.0, "identifier": 0.0, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 572}, "ledger_by_hop": {"1": {"facts_total": 64, "facts_retained": 10, "retention_rate": 0.15625, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.2, "amount": 0.16666666666666666, "status": 0.3333333333333333, "place": 0.0, "identifier": 0.0, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 669}, "2": {"facts_total": 64, "facts_retained": 9, "retention_rate": 0.140625, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.16666666666666666, "status": 0.3333333333333333, "place": 0.0, "identifier": 0.0, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 572}, "3": {"facts_total": 64, "facts_retained": 9, "retention_rate": 0.140625, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.16666666666666666, "status": 0.3333333333333333, "place": 0.0, "identifier": 0.0, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 572}, "4": {"facts_total": 64, "facts_retained": 9, "retention_rate": 0.140625, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.16666666666666666, "status": 0.3333333333333333, "place": 0.0, "identifier": 0.0, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 572}, "5": {"facts_total": 64, "facts_retained": 9, "retention_rate": 0.140625, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.16666666666666666, "status": 0.3333333333333333, "place": 0.0, "identifier": 0.0, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 572}, "6": {"facts_total": 64, "facts_retained": 9, "retention_rate": 0.140625, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.16666666666666666, "status": 0.3333333333333333, "place": 0.0, "identifier": 0.0, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 572}, "7": {"facts_total": 64, "facts_retained": 9, "retention_rate": 0.140625, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.16666666666666666, "status": 0.3333333333333333, "place": 0.0, "identifier": 0.0, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 572}, "8": {"facts_total": 64, "facts_retained": 9, "retention_rate": 0.140625, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.16666666666666666, "status": 0.3333333333333333, "place": 0.0, "identifier": 0.0, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 572}}}, "reference_key_id": "keys/synthetic/syn_03006_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "175d3ea2044e4182288513cfbf6fbcb45bd8241379081d5258569d283ccfe60c", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "c195667318a07c19bb0a658a", "attempt_id": "16d8642ce1e3", "run_group_id": "E4_core:A5|closed|qwen3b:h8", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:00:45.595182Z", "task_family": "synthetic", "document_id": "syn_03006_2000", "document_hash": "89be5e77e0ca114963e9ffdf8dc38fe5610f13a78ac29e7a1956cd9124ce5217", "question_id": "q07_negation", "split": "test", "cluster_id": "syn_03006_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 8, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1890, "query_tokens": 351, "message_tokens": 269, "message_utf8_bytes": 620, "cache_tensor_bytes": null, "serialized_payload_bytes": 620, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 106877.52525007818, "hops": {"1": 22661.616917001083, "2": 12228.278000024147, "3": 12082.4769170722, "4": 12145.376833039336, "5": 11973.832083051093, "6": 12040.75849999208, "7": 11826.985499938019, "8": 11918.200499960221}, "receiver_prefill_ms": 255.71558298543096, "ttft_ms": 22.169915959239006, "decode_ms": 19.38770804554224}, "total_wall_ms": 107175.39870808832, "ttft_ms": 22.169915959239006, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 64, "facts_retained": 9, "retention_rate": 0.140625, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.16666666666666666, "status": 0.3333333333333333, "place": 0.0, "identifier": 0.0, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 572}, "ledger_by_hop": {"1": {"facts_total": 64, "facts_retained": 10, "retention_rate": 0.15625, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.2, "amount": 0.16666666666666666, "status": 0.3333333333333333, "place": 0.0, "identifier": 0.0, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 669}, "2": {"facts_total": 64, "facts_retained": 9, "retention_rate": 0.140625, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.16666666666666666, "status": 0.3333333333333333, "place": 0.0, "identifier": 0.0, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 572}, "3": {"facts_total": 64, "facts_retained": 9, "retention_rate": 0.140625, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.16666666666666666, "status": 0.3333333333333333, "place": 0.0, "identifier": 0.0, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 572}, "4": {"facts_total": 64, "facts_retained": 9, "retention_rate": 0.140625, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.16666666666666666, "status": 0.3333333333333333, "place": 0.0, "identifier": 0.0, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 572}, "5": {"facts_total": 64, "facts_retained": 9, "retention_rate": 0.140625, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.16666666666666666, "status": 0.3333333333333333, "place": 0.0, "identifier": 0.0, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 572}, "6": {"facts_total": 64, "facts_retained": 9, "retention_rate": 0.140625, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.16666666666666666, "status": 0.3333333333333333, "place": 0.0, "identifier": 0.0, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 572}, "7": {"facts_total": 64, "facts_retained": 9, "retention_rate": 0.140625, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.16666666666666666, "status": 0.3333333333333333, "place": 0.0, "identifier": 0.0, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 572}, "8": {"facts_total": 64, "facts_retained": 9, "retention_rate": 0.140625, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.16666666666666666, "status": 0.3333333333333333, "place": 0.0, "identifier": 0.0, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 572}}}, "reference_key_id": "keys/synthetic/syn_03006_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "175d3ea2044e4182288513cfbf6fbcb45bd8241379081d5258569d283ccfe60c", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "da8ff265e06d1ea5f3e8fa45", "attempt_id": "8bea4a8fbc74", "run_group_id": "E4_core:A5|closed|qwen3b:h8", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:00:45.910867Z", "task_family": "synthetic", "document_id": "syn_03006_2000", "document_hash": "89be5e77e0ca114963e9ffdf8dc38fe5610f13a78ac29e7a1956cd9124ce5217", "question_id": "q08_revision", "split": "test", "cluster_id": "syn_03006_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 8, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1890, "query_tokens": 368, "message_tokens": 269, "message_utf8_bytes": 620, "cache_tensor_bytes": null, "serialized_payload_bytes": 620, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 106877.52525007818, "hops": {"1": 22661.616917001083, "2": 12228.278000024147, "3": 12082.4769170722, "4": 12145.376833039336, "5": 11973.832083051093, "6": 12040.75849999208, "7": 11826.985499938019, "8": 11918.200499960221}, "receiver_prefill_ms": 252.67675006762147, "ttft_ms": 54.50045794714242, "decode_ms": 475.04308295901865}, "total_wall_ms": 107660.20325000864, "ttft_ms": 54.50045794714242, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 8}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "10107545", "evidence_artifact": {"ledger": {"facts_total": 64, "facts_retained": 9, "retention_rate": 0.140625, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.16666666666666666, "status": 0.3333333333333333, "place": 0.0, "identifier": 0.0, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 572}, "ledger_by_hop": {"1": {"facts_total": 64, "facts_retained": 10, "retention_rate": 0.15625, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.2, "amount": 0.16666666666666666, "status": 0.3333333333333333, "place": 0.0, "identifier": 0.0, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 669}, "2": {"facts_total": 64, "facts_retained": 9, "retention_rate": 0.140625, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.16666666666666666, "status": 0.3333333333333333, "place": 0.0, "identifier": 0.0, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 572}, "3": {"facts_total": 64, "facts_retained": 9, "retention_rate": 0.140625, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.16666666666666666, "status": 0.3333333333333333, "place": 0.0, "identifier": 0.0, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 572}, "4": {"facts_total": 64, "facts_retained": 9, "retention_rate": 0.140625, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.16666666666666666, "status": 0.3333333333333333, "place": 0.0, "identifier": 0.0, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 572}, "5": {"facts_total": 64, "facts_retained": 9, "retention_rate": 0.140625, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.16666666666666666, "status": 0.3333333333333333, "place": 0.0, "identifier": 0.0, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 572}, "6": {"facts_total": 64, "facts_retained": 9, "retention_rate": 0.140625, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.16666666666666666, "status": 0.3333333333333333, "place": 0.0, "identifier": 0.0, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 572}, "7": {"facts_total": 64, "facts_retained": 9, "retention_rate": 0.140625, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.16666666666666666, "status": 0.3333333333333333, "place": 0.0, "identifier": 0.0, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 572}, "8": {"facts_total": 64, "facts_retained": 9, "retention_rate": 0.140625, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.16666666666666666, "status": 0.3333333333333333, "place": 0.0, "identifier": 0.0, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 572}}}, "reference_key_id": "keys/synthetic/syn_03006_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "175d3ea2044e4182288513cfbf6fbcb45bd8241379081d5258569d283ccfe60c", "output_hash": "6faba9de4e29949bbdd8934f69fc2110f07076db6801b211aeca9351292537fe"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "a9a6b4ccffe2620bb7848aea", "attempt_id": "34573f92fc92", "run_group_id": "E4_core:A5|closed|qwen3b:h8", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:00:46.712044Z", "task_family": "synthetic", "document_id": "syn_03006_2000", "document_hash": "89be5e77e0ca114963e9ffdf8dc38fe5610f13a78ac29e7a1956cd9124ce5217", "question_id": "q09_revision", "split": "test", "cluster_id": "syn_03006_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 8, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1890, "query_tokens": 361, "message_tokens": 269, "message_utf8_bytes": 620, "cache_tensor_bytes": null, "serialized_payload_bytes": 620, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 106877.52525007818, "hops": {"1": 22661.616917001083, "2": 12228.278000024147, "3": 12082.4769170722, "4": 12145.376833039336, "5": 11973.832083051093, "6": 12040.75849999208, "7": 11826.985499938019, "8": 11918.200499960221}, "receiver_prefill_ms": 272.87054201588035, "ttft_ms": 30.789499985985458, "decode_ms": 325.74879203457385}, "total_wall_ms": 107507.52245809417, "ttft_ms": 30.789499985985458, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 9}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "9,112,735", "evidence_artifact": {"ledger": {"facts_total": 64, "facts_retained": 9, "retention_rate": 0.140625, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.16666666666666666, "status": 0.3333333333333333, "place": 0.0, "identifier": 0.0, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 572}, "ledger_by_hop": {"1": {"facts_total": 64, "facts_retained": 10, "retention_rate": 0.15625, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.2, "amount": 0.16666666666666666, "status": 0.3333333333333333, "place": 0.0, "identifier": 0.0, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 669}, "2": {"facts_total": 64, "facts_retained": 9, "retention_rate": 0.140625, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.16666666666666666, "status": 0.3333333333333333, "place": 0.0, "identifier": 0.0, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 572}, "3": {"facts_total": 64, "facts_retained": 9, "retention_rate": 0.140625, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.16666666666666666, "status": 0.3333333333333333, "place": 0.0, "identifier": 0.0, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 572}, "4": {"facts_total": 64, "facts_retained": 9, "retention_rate": 0.140625, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.16666666666666666, "status": 0.3333333333333333, "place": 0.0, "identifier": 0.0, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 572}, "5": {"facts_total": 64, "facts_retained": 9, "retention_rate": 0.140625, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.16666666666666666, "status": 0.3333333333333333, "place": 0.0, "identifier": 0.0, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 572}, "6": {"facts_total": 64, "facts_retained": 9, "retention_rate": 0.140625, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.16666666666666666, "status": 0.3333333333333333, "place": 0.0, "identifier": 0.0, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 572}, "7": {"facts_total": 64, "facts_retained": 9, "retention_rate": 0.140625, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.16666666666666666, "status": 0.3333333333333333, "place": 0.0, "identifier": 0.0, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 572}, "8": {"facts_total": 64, "facts_retained": 9, "retention_rate": 0.140625, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.16666666666666666, "status": 0.3333333333333333, "place": 0.0, "identifier": 0.0, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 572}}}, "reference_key_id": "keys/synthetic/syn_03006_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "175d3ea2044e4182288513cfbf6fbcb45bd8241379081d5258569d283ccfe60c", "output_hash": "788fcda908c2029597420995b09d0cc919bfdbf7a28b68dcb0e331db7df7b342"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "4353c8bb9e26e093e3d8e880", "attempt_id": "e89dc218544f", "run_group_id": "E4_core:A5|closed|qwen3b:h8", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:00:47.360728Z", "task_family": "synthetic", "document_id": "syn_03006_2000", "document_hash": "89be5e77e0ca114963e9ffdf8dc38fe5610f13a78ac29e7a1956cd9124ce5217", "question_id": "q10_temporal", "split": "test", "cluster_id": "syn_03006_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 8, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1890, "query_tokens": 370, "message_tokens": 269, "message_utf8_bytes": 620, "cache_tensor_bytes": null, "serialized_payload_bytes": 620, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 106877.52525007818, "hops": {"1": 22661.616917001083, "2": 12228.278000024147, "3": 12082.4769170722, "4": 12145.376833039336, "5": 11973.832083051093, "6": 12040.75849999208, "7": 11826.985499938019, "8": 11918.200499960221}, "receiver_prefill_ms": 381.8212499609217, "ttft_ms": 48.50204195827246, "decode_ms": 64.48583293240517}, "total_wall_ms": 107372.76583304629, "ttft_ms": 48.50204195827246, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 64, "facts_retained": 9, "retention_rate": 0.140625, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.16666666666666666, "status": 0.3333333333333333, "place": 0.0, "identifier": 0.0, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 572}, "ledger_by_hop": {"1": {"facts_total": 64, "facts_retained": 10, "retention_rate": 0.15625, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.2, "amount": 0.16666666666666666, "status": 0.3333333333333333, "place": 0.0, "identifier": 0.0, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 669}, "2": {"facts_total": 64, "facts_retained": 9, "retention_rate": 0.140625, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.16666666666666666, "status": 0.3333333333333333, "place": 0.0, "identifier": 0.0, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 572}, "3": {"facts_total": 64, "facts_retained": 9, "retention_rate": 0.140625, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.16666666666666666, "status": 0.3333333333333333, "place": 0.0, "identifier": 0.0, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 572}, "4": {"facts_total": 64, "facts_retained": 9, "retention_rate": 0.140625, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.16666666666666666, "status": 0.3333333333333333, "place": 0.0, "identifier": 0.0, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 572}, "5": {"facts_total": 64, "facts_retained": 9, "retention_rate": 0.140625, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.16666666666666666, "status": 0.3333333333333333, "place": 0.0, "identifier": 0.0, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 572}, "6": {"facts_total": 64, "facts_retained": 9, "retention_rate": 0.140625, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.16666666666666666, "status": 0.3333333333333333, "place": 0.0, "identifier": 0.0, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 572}, "7": {"facts_total": 64, "facts_retained": 9, "retention_rate": 0.140625, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.16666666666666666, "status": 0.3333333333333333, "place": 0.0, "identifier": 0.0, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 572}, "8": {"facts_total": 64, "facts_retained": 9, "retention_rate": 0.140625, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.16666666666666666, "status": 0.3333333333333333, "place": 0.0, "identifier": 0.0, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 572}}}, "reference_key_id": "keys/synthetic/syn_03006_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "175d3ea2044e4182288513cfbf6fbcb45bd8241379081d5258569d283ccfe60c", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "4f3ac21e8f64eacdb0bed52f", "attempt_id": "894c4b941437", "run_group_id": "E4_core:A5|closed|qwen3b:h8", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:00:47.873822Z", "task_family": "synthetic", "document_id": "syn_03006_2000", "document_hash": "89be5e77e0ca114963e9ffdf8dc38fe5610f13a78ac29e7a1956cd9124ce5217", "question_id": "q11_aggregation", "split": "test", "cluster_id": "syn_03006_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 8, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1890, "query_tokens": 363, "message_tokens": 269, "message_utf8_bytes": 620, "cache_tensor_bytes": null, "serialized_payload_bytes": 620, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 106877.52525007818, "hops": {"1": 22661.616917001083, "2": 12228.278000024147, "3": 12082.4769170722, "4": 12145.376833039336, "5": 11973.832083051093, "6": 12040.75849999208, "7": 11826.985499938019, "8": 11918.200499960221}, "receiver_prefill_ms": 380.92616701032966, "ttft_ms": 73.65258294157684, "decode_ms": 72.59487500414252}, "total_wall_ms": 107405.15404206235, "ttft_ms": 73.65258294157684, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 64, "facts_retained": 9, "retention_rate": 0.140625, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.16666666666666666, "status": 0.3333333333333333, "place": 0.0, "identifier": 0.0, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 572}, "ledger_by_hop": {"1": {"facts_total": 64, "facts_retained": 10, "retention_rate": 0.15625, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.2, "amount": 0.16666666666666666, "status": 0.3333333333333333, "place": 0.0, "identifier": 0.0, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 669}, "2": {"facts_total": 64, "facts_retained": 9, "retention_rate": 0.140625, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.16666666666666666, "status": 0.3333333333333333, "place": 0.0, "identifier": 0.0, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 572}, "3": {"facts_total": 64, "facts_retained": 9, "retention_rate": 0.140625, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.16666666666666666, "status": 0.3333333333333333, "place": 0.0, "identifier": 0.0, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 572}, "4": {"facts_total": 64, "facts_retained": 9, "retention_rate": 0.140625, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.16666666666666666, "status": 0.3333333333333333, "place": 0.0, "identifier": 0.0, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 572}, "5": {"facts_total": 64, "facts_retained": 9, "retention_rate": 0.140625, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.16666666666666666, "status": 0.3333333333333333, "place": 0.0, "identifier": 0.0, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 572}, "6": {"facts_total": 64, "facts_retained": 9, "retention_rate": 0.140625, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.16666666666666666, "status": 0.3333333333333333, "place": 0.0, "identifier": 0.0, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 572}, "7": {"facts_total": 64, "facts_retained": 9, "retention_rate": 0.140625, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.16666666666666666, "status": 0.3333333333333333, "place": 0.0, "identifier": 0.0, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 572}, "8": {"facts_total": 64, "facts_retained": 9, "retention_rate": 0.140625, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.16666666666666666, "status": 0.3333333333333333, "place": 0.0, "identifier": 0.0, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 572}}}, "reference_key_id": "keys/synthetic/syn_03006_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "175d3ea2044e4182288513cfbf6fbcb45bd8241379081d5258569d283ccfe60c", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "1757a8e92265a6d61877d085", "attempt_id": "e9165c636672", "run_group_id": "E4_core:A5|closed|qwen3b:h8", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:00:48.418396Z", "task_family": "synthetic", "document_id": "syn_03006_2000", "document_hash": "89be5e77e0ca114963e9ffdf8dc38fe5610f13a78ac29e7a1956cd9124ce5217", "question_id": "q12_unit_conversion", "split": "test", "cluster_id": "syn_03006_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 8, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1890, "query_tokens": 360, "message_tokens": 269, "message_utf8_bytes": 620, "cache_tensor_bytes": null, "serialized_payload_bytes": 620, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 106877.52525007818, "hops": {"1": 22661.616917001083, "2": 12228.278000024147, "3": 12082.4769170722, "4": 12145.376833039336, "5": 11973.832083051093, "6": 12040.75849999208, "7": 11826.985499938019, "8": 11918.200499960221}, "receiver_prefill_ms": 361.42662493512034, "ttft_ms": 92.39300002809614, "decode_ms": 464.1377499792725}, "total_wall_ms": 107795.96141714137, "ttft_ms": 92.39300002809614, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 8}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "11946000", "evidence_artifact": {"ledger": {"facts_total": 64, "facts_retained": 9, "retention_rate": 0.140625, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.16666666666666666, "status": 0.3333333333333333, "place": 0.0, "identifier": 0.0, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 572}, "ledger_by_hop": {"1": {"facts_total": 64, "facts_retained": 10, "retention_rate": 0.15625, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.2, "amount": 0.16666666666666666, "status": 0.3333333333333333, "place": 0.0, "identifier": 0.0, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 669}, "2": {"facts_total": 64, "facts_retained": 9, "retention_rate": 0.140625, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.16666666666666666, "status": 0.3333333333333333, "place": 0.0, "identifier": 0.0, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 572}, "3": {"facts_total": 64, "facts_retained": 9, "retention_rate": 0.140625, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.16666666666666666, "status": 0.3333333333333333, "place": 0.0, "identifier": 0.0, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 572}, "4": {"facts_total": 64, "facts_retained": 9, "retention_rate": 0.140625, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.16666666666666666, "status": 0.3333333333333333, "place": 0.0, "identifier": 0.0, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 572}, "5": {"facts_total": 64, "facts_retained": 9, "retention_rate": 0.140625, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.16666666666666666, "status": 0.3333333333333333, "place": 0.0, "identifier": 0.0, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 572}, "6": {"facts_total": 64, "facts_retained": 9, "retention_rate": 0.140625, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.16666666666666666, "status": 0.3333333333333333, "place": 0.0, "identifier": 0.0, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 572}, "7": {"facts_total": 64, "facts_retained": 9, "retention_rate": 0.140625, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.16666666666666666, "status": 0.3333333333333333, "place": 0.0, "identifier": 0.0, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 572}, "8": {"facts_total": 64, "facts_retained": 9, "retention_rate": 0.140625, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.16666666666666666, "status": 0.3333333333333333, "place": 0.0, "identifier": 0.0, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 572}}}, "reference_key_id": "keys/synthetic/syn_03006_2000.json", "task_scores": {"correct": 0.0, "abstained": 0.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "175d3ea2044e4182288513cfbf6fbcb45bd8241379081d5258569d283ccfe60c", "output_hash": "edf27f197d1395ea4e7b86fae1feecd3591246e1299bba6a08655f5724a670ea"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "f1a5d71f9da31fe5566e3f8d", "attempt_id": "f14d22c94910", "run_group_id": "E4_core:A5|closed|qwen3b:h8", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:00:49.354089Z", "task_family": "synthetic", "document_id": "syn_03006_2000", "document_hash": "89be5e77e0ca114963e9ffdf8dc38fe5610f13a78ac29e7a1956cd9124ce5217", "question_id": "q13_unanswerable", "split": "test", "cluster_id": "syn_03006_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 8, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1890, "query_tokens": 364, "message_tokens": 269, "message_utf8_bytes": 620, "cache_tensor_bytes": null, "serialized_payload_bytes": 620, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 106877.52525007818, "hops": {"1": 22661.616917001083, "2": 12228.278000024147, "3": 12082.4769170722, "4": 12145.376833039336, "5": 11973.832083051093, "6": 12040.75849999208, "7": 11826.985499938019, "8": 11918.200499960221}, "receiver_prefill_ms": 359.9790839944035, "ttft_ms": 62.07975000143051, "decode_ms": 52.66833398491144}, "total_wall_ms": 107352.72529197391, "ttft_ms": 62.07975000143051, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 64, "facts_retained": 9, "retention_rate": 0.140625, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.16666666666666666, "status": 0.3333333333333333, "place": 0.0, "identifier": 0.0, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 572}, "ledger_by_hop": {"1": {"facts_total": 64, "facts_retained": 10, "retention_rate": 0.15625, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.2, "amount": 0.16666666666666666, "status": 0.3333333333333333, "place": 0.0, "identifier": 0.0, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 669}, "2": {"facts_total": 64, "facts_retained": 9, "retention_rate": 0.140625, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.16666666666666666, "status": 0.3333333333333333, "place": 0.0, "identifier": 0.0, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 572}, "3": {"facts_total": 64, "facts_retained": 9, "retention_rate": 0.140625, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.16666666666666666, "status": 0.3333333333333333, "place": 0.0, "identifier": 0.0, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 572}, "4": {"facts_total": 64, "facts_retained": 9, "retention_rate": 0.140625, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.16666666666666666, "status": 0.3333333333333333, "place": 0.0, "identifier": 0.0, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 572}, "5": {"facts_total": 64, "facts_retained": 9, "retention_rate": 0.140625, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.16666666666666666, "status": 0.3333333333333333, "place": 0.0, "identifier": 0.0, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 572}, "6": {"facts_total": 64, "facts_retained": 9, "retention_rate": 0.140625, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.16666666666666666, "status": 0.3333333333333333, "place": 0.0, "identifier": 0.0, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 572}, "7": {"facts_total": 64, "facts_retained": 9, "retention_rate": 0.140625, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.16666666666666666, "status": 0.3333333333333333, "place": 0.0, "identifier": 0.0, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 572}, "8": {"facts_total": 64, "facts_retained": 9, "retention_rate": 0.140625, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.16666666666666666, "status": 0.3333333333333333, "place": 0.0, "identifier": 0.0, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 572}}}, "reference_key_id": "keys/synthetic/syn_03006_2000.json", "task_scores": {"correct": 1.0, "abstained": 1.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "175d3ea2044e4182288513cfbf6fbcb45bd8241379081d5258569d283ccfe60c", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "4f3cef2606f846e5f578ab95", "attempt_id": "ed0b81bc7bd9", "run_group_id": "E4_core:A5|closed|qwen3b:h8", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:00:49.846435Z", "task_family": "synthetic", "document_id": "syn_03006_2000", "document_hash": "89be5e77e0ca114963e9ffdf8dc38fe5610f13a78ac29e7a1956cd9124ce5217", "question_id": "q14_unanswerable", "split": "test", "cluster_id": "syn_03006_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 8, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1890, "query_tokens": 365, "message_tokens": 269, "message_utf8_bytes": 620, "cache_tensor_bytes": null, "serialized_payload_bytes": 620, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 106877.52525007818, "hops": {"1": 22661.616917001083, "2": 12228.278000024147, "3": 12082.4769170722, "4": 12145.376833039336, "5": 11973.832083051093, "6": 12040.75849999208, "7": 11826.985499938019, "8": 11918.200499960221}, "receiver_prefill_ms": 367.4690410261974, "ttft_ms": 70.29766694176942, "decode_ms": 90.16770799644291}, "total_wall_ms": 107405.9155411087, "ttft_ms": 70.29766694176942, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 64, "facts_retained": 9, "retention_rate": 0.140625, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.16666666666666666, "status": 0.3333333333333333, "place": 0.0, "identifier": 0.0, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 572}, "ledger_by_hop": {"1": {"facts_total": 64, "facts_retained": 10, "retention_rate": 0.15625, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.2, "amount": 0.16666666666666666, "status": 0.3333333333333333, "place": 0.0, "identifier": 0.0, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 669}, "2": {"facts_total": 64, "facts_retained": 9, "retention_rate": 0.140625, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.16666666666666666, "status": 0.3333333333333333, "place": 0.0, "identifier": 0.0, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 572}, "3": {"facts_total": 64, "facts_retained": 9, "retention_rate": 0.140625, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.16666666666666666, "status": 0.3333333333333333, "place": 0.0, "identifier": 0.0, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 572}, "4": {"facts_total": 64, "facts_retained": 9, "retention_rate": 0.140625, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.16666666666666666, "status": 0.3333333333333333, "place": 0.0, "identifier": 0.0, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 572}, "5": {"facts_total": 64, "facts_retained": 9, "retention_rate": 0.140625, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.16666666666666666, "status": 0.3333333333333333, "place": 0.0, "identifier": 0.0, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 572}, "6": {"facts_total": 64, "facts_retained": 9, "retention_rate": 0.140625, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.16666666666666666, "status": 0.3333333333333333, "place": 0.0, "identifier": 0.0, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 572}, "7": {"facts_total": 64, "facts_retained": 9, "retention_rate": 0.140625, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.16666666666666666, "status": 0.3333333333333333, "place": 0.0, "identifier": 0.0, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 572}, "8": {"facts_total": 64, "facts_retained": 9, "retention_rate": 0.140625, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.16666666666666666, "status": 0.3333333333333333, "place": 0.0, "identifier": 0.0, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 572}}}, "reference_key_id": "keys/synthetic/syn_03006_2000.json", "task_scores": {"correct": 1.0, "abstained": 1.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "175d3ea2044e4182288513cfbf6fbcb45bd8241379081d5258569d283ccfe60c", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "b1d30e0f93949a7163762f83", "attempt_id": "9f3de92cf2e0", "run_group_id": "E4_core:A4|repairable|qwen3b:h1", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:01:13.527909Z", "task_family": "synthetic", "document_id": "syn_03006_2000", "document_hash": "89be5e77e0ca114963e9ffdf8dc38fe5610f13a78ac29e7a1956cd9124ce5217", "question_id": "q00_single_fact", "split": "test", "cluster_id": "syn_03006_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 1, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1890, "query_tokens": 595, "message_tokens": 512, "message_utf8_bytes": 1087, "cache_tensor_bytes": null, "serialized_payload_bytes": 1087, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 23133.45850002952, "hops": {"1": 23133.45850002952}, "receiver_prefill_ms": 417.18375007621944, "ttft_ms": 40.16229195985943, "decode_ms": 317.51920795068145}, "total_wall_ms": 23908.99966598954, "ttft_ms": 40.16229195985943, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 10}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "XM-76Q-2958", "evidence_artifact": {"ledger": {"facts_total": 64, "facts_retained": 30, "retention_rate": 0.46875, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.6666666666666666, "status": 0.9166666666666666, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1087}, "ledger_by_hop": {"1": {"facts_total": 64, "facts_retained": 30, "retention_rate": 0.46875, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.6666666666666666, "status": 0.9166666666666666, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1087}}}, "reference_key_id": "keys/synthetic/syn_03006_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "47f843b1989b354c7cbe406d456e62c59b6229490f55dde77bcdb5f70147ff28", "output_hash": "cf42691eec772b9ec74d67643944de2e19e53724d7b772cef1e0f2d5ef3b4926"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "6673ab1d7e51b51f6462dbac", "attempt_id": "030deec3da47", "run_group_id": "E4_core:A4|repairable|qwen3b:h1", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:01:14.321255Z", "task_family": "synthetic", "document_id": "syn_03006_2000", "document_hash": "89be5e77e0ca114963e9ffdf8dc38fe5610f13a78ac29e7a1956cd9124ce5217", "question_id": "q01_single_fact", "split": "test", "cluster_id": "syn_03006_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 1, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1890, "query_tokens": 600, "message_tokens": 512, "message_utf8_bytes": 1087, "cache_tensor_bytes": null, "serialized_payload_bytes": 1087, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 23133.45850002952, "hops": {"1": 23133.45850002952}, "receiver_prefill_ms": 344.738999963738, "ttft_ms": 50.53666699677706, "decode_ms": 56.96666706353426}, "total_wall_ms": 23586.37170807924, "ttft_ms": 50.53666699677706, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 64, "facts_retained": 30, "retention_rate": 0.46875, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.6666666666666666, "status": 0.9166666666666666, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1087}, "ledger_by_hop": {"1": {"facts_total": 64, "facts_retained": 30, "retention_rate": 0.46875, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.6666666666666666, "status": 0.9166666666666666, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1087}}}, "reference_key_id": "keys/synthetic/syn_03006_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "47f843b1989b354c7cbe406d456e62c59b6229490f55dde77bcdb5f70147ff28", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "dcd13a39ed164e3b3e3f4fee", "attempt_id": "e6cc10cac4af", "run_group_id": "E4_core:A4|repairable|qwen3b:h1", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:01:14.792117Z", "task_family": "synthetic", "document_id": "syn_03006_2000", "document_hash": "89be5e77e0ca114963e9ffdf8dc38fe5610f13a78ac29e7a1956cd9124ce5217", "question_id": "q02_exact_copy", "split": "test", "cluster_id": "syn_03006_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 1, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1890, "query_tokens": 599, "message_tokens": 512, "message_utf8_bytes": 1087, "cache_tensor_bytes": null, "serialized_payload_bytes": 1087, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 23133.45850002952, "hops": {"1": 23133.45850002952}, "receiver_prefill_ms": 544.6670840028673, "ttft_ms": 34.293083008378744, "decode_ms": 40.51029100082815}, "total_wall_ms": 23753.628417034633, "ttft_ms": 34.293083008378744, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 64, "facts_retained": 30, "retention_rate": 0.46875, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.6666666666666666, "status": 0.9166666666666666, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1087}, "ledger_by_hop": {"1": {"facts_total": 64, "facts_retained": 30, "retention_rate": 0.46875, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.6666666666666666, "status": 0.9166666666666666, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1087}}}, "reference_key_id": "keys/synthetic/syn_03006_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "47f843b1989b354c7cbe406d456e62c59b6229490f55dde77bcdb5f70147ff28", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "062c1decf02d1eed37e95c73", "attempt_id": "b14b2bdf6570", "run_group_id": "E4_core:A4|repairable|qwen3b:h1", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:01:15.431155Z", "task_family": "synthetic", "document_id": "syn_03006_2000", "document_hash": "89be5e77e0ca114963e9ffdf8dc38fe5610f13a78ac29e7a1956cd9124ce5217", "question_id": "q03_exact_copy", "split": "test", "cluster_id": "syn_03006_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 1, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1890, "query_tokens": 594, "message_tokens": 512, "message_utf8_bytes": 1087, "cache_tensor_bytes": null, "serialized_payload_bytes": 1087, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 23133.45850002952, "hops": {"1": 23133.45850002952}, "receiver_prefill_ms": 419.2870419938117, "ttft_ms": 27.976499986834824, "decode_ms": 214.25104199443012}, "total_wall_ms": 23795.56387499906, "ttft_ms": 27.976499986834824, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 8}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "Haruki Ouellet-Marchand", "evidence_artifact": {"ledger": {"facts_total": 64, "facts_retained": 30, "retention_rate": 0.46875, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.6666666666666666, "status": 0.9166666666666666, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1087}, "ledger_by_hop": {"1": {"facts_total": 64, "facts_retained": 30, "retention_rate": 0.46875, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.6666666666666666, "status": 0.9166666666666666, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1087}}}, "reference_key_id": "keys/synthetic/syn_03006_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "47f843b1989b354c7cbe406d456e62c59b6229490f55dde77bcdb5f70147ff28", "output_hash": "a3b6e0a44d495387d8b8fc50ba6be0f2ed43121c674d3b3e962f8c4690e15b5c"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "25f616c2efbf8684447acdf7", "attempt_id": "7a8e934d740a", "run_group_id": "E4_core:A4|repairable|qwen3b:h1", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:01:16.110920Z", "task_family": "synthetic", "document_id": "syn_03006_2000", "document_hash": "89be5e77e0ca114963e9ffdf8dc38fe5610f13a78ac29e7a1956cd9124ce5217", "question_id": "q04_multi_hop", "split": "test", "cluster_id": "syn_03006_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 1, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1890, "query_tokens": 603, "message_tokens": 512, "message_utf8_bytes": 1087, "cache_tensor_bytes": null, "serialized_payload_bytes": 1087, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 23133.45850002952, "hops": {"1": 23133.45850002952}, "receiver_prefill_ms": 395.130333956331, "ttft_ms": 49.6735000051558, "decode_ms": 57.115416042506695}, "total_wall_ms": 23636.064833030105, "ttft_ms": 49.6735000051558, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 64, "facts_retained": 30, "retention_rate": 0.46875, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.6666666666666666, "status": 0.9166666666666666, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1087}, "ledger_by_hop": {"1": {"facts_total": 64, "facts_retained": 30, "retention_rate": 0.46875, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.6666666666666666, "status": 0.9166666666666666, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1087}}}, "reference_key_id": "keys/synthetic/syn_03006_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "47f843b1989b354c7cbe406d456e62c59b6229490f55dde77bcdb5f70147ff28", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "3ffd3d0a121063bd5eb36c90", "attempt_id": "11f3330cbe05", "run_group_id": "E4_core:A4|repairable|qwen3b:h1", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:01:16.629502Z", "task_family": "synthetic", "document_id": "syn_03006_2000", "document_hash": "89be5e77e0ca114963e9ffdf8dc38fe5610f13a78ac29e7a1956cd9124ce5217", "question_id": "q05_multi_hop", "split": "test", "cluster_id": "syn_03006_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 1, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1890, "query_tokens": 606, "message_tokens": 512, "message_utf8_bytes": 1087, "cache_tensor_bytes": null, "serialized_payload_bytes": 1087, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 23133.45850002952, "hops": {"1": 23133.45850002952}, "receiver_prefill_ms": 508.6167911067605, "ttft_ms": 39.894582936540246, "decode_ms": 350.79383302945644}, "total_wall_ms": 24033.538916963153, "ttft_ms": 39.894582936540246, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 10}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "EP-52G-1330", "evidence_artifact": {"ledger": {"facts_total": 64, "facts_retained": 30, "retention_rate": 0.46875, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.6666666666666666, "status": 0.9166666666666666, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1087}, "ledger_by_hop": {"1": {"facts_total": 64, "facts_retained": 30, "retention_rate": 0.46875, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.6666666666666666, "status": 0.9166666666666666, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1087}}}, "reference_key_id": "keys/synthetic/syn_03006_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "47f843b1989b354c7cbe406d456e62c59b6229490f55dde77bcdb5f70147ff28", "output_hash": "104db207937fd79e23ce16f9b3f10b1a9ed34f6dd0441a99be6ea1b2d0a09d99"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "18b39f3ca5bea31b6155b96a", "attempt_id": "b010313f51ab", "run_group_id": "E4_core:A4|repairable|qwen3b:h1", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:01:17.546875Z", "task_family": "synthetic", "document_id": "syn_03006_2000", "document_hash": "89be5e77e0ca114963e9ffdf8dc38fe5610f13a78ac29e7a1956cd9124ce5217", "question_id": "q06_negation", "split": "test", "cluster_id": "syn_03006_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 1, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1890, "query_tokens": 594, "message_tokens": 512, "message_utf8_bytes": 1087, "cache_tensor_bytes": null, "serialized_payload_bytes": 1087, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 23133.45850002952, "hops": {"1": 23133.45850002952}, "receiver_prefill_ms": 349.07991590444, "ttft_ms": 44.40633300691843, "decode_ms": 51.85858404729515}, "total_wall_ms": 23579.603541991673, "ttft_ms": 44.40633300691843, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "no", "evidence_artifact": {"ledger": {"facts_total": 64, "facts_retained": 30, "retention_rate": 0.46875, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.6666666666666666, "status": 0.9166666666666666, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1087}, "ledger_by_hop": {"1": {"facts_total": 64, "facts_retained": 30, "retention_rate": 0.46875, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.6666666666666666, "status": 0.9166666666666666, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1087}}}, "reference_key_id": "keys/synthetic/syn_03006_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "47f843b1989b354c7cbe406d456e62c59b6229490f55dde77bcdb5f70147ff28", "output_hash": "9390298f3fb0c5b160498935d79cb139aef28e1c47358b4bbba61862b9c26e59"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "4587d01f6db88c945c3b6b8a", "attempt_id": "3a7d23283924", "run_group_id": "E4_core:A4|repairable|qwen3b:h1", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:01:18.011498Z", "task_family": "synthetic", "document_id": "syn_03006_2000", "document_hash": "89be5e77e0ca114963e9ffdf8dc38fe5610f13a78ac29e7a1956cd9124ce5217", "question_id": "q07_negation", "split": "test", "cluster_id": "syn_03006_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 1, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1890, "query_tokens": 594, "message_tokens": 512, "message_utf8_bytes": 1087, "cache_tensor_bytes": null, "serialized_payload_bytes": 1087, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 23133.45850002952, "hops": {"1": 23133.45850002952}, "receiver_prefill_ms": 453.04674992803484, "ttft_ms": 105.92979204375297, "decode_ms": 43.35912503302097}, "total_wall_ms": 23736.410125042312, "ttft_ms": 105.92979204375297, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "no", "evidence_artifact": {"ledger": {"facts_total": 64, "facts_retained": 30, "retention_rate": 0.46875, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.6666666666666666, "status": 0.9166666666666666, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1087}, "ledger_by_hop": {"1": {"facts_total": 64, "facts_retained": 30, "retention_rate": 0.46875, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.6666666666666666, "status": 0.9166666666666666, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1087}}}, "reference_key_id": "keys/synthetic/syn_03006_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "47f843b1989b354c7cbe406d456e62c59b6229490f55dde77bcdb5f70147ff28", "output_hash": "9390298f3fb0c5b160498935d79cb139aef28e1c47358b4bbba61862b9c26e59"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "667962da44e6701ab8838a9e", "attempt_id": "446d296f65de", "run_group_id": "E4_core:A4|repairable|qwen3b:h1", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:01:18.631638Z", "task_family": "synthetic", "document_id": "syn_03006_2000", "document_hash": "89be5e77e0ca114963e9ffdf8dc38fe5610f13a78ac29e7a1956cd9124ce5217", "question_id": "q08_revision", "split": "test", "cluster_id": "syn_03006_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 1, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1890, "query_tokens": 611, "message_tokens": 512, "message_utf8_bytes": 1087, "cache_tensor_bytes": null, "serialized_payload_bytes": 1087, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 23133.45850002952, "hops": {"1": 23133.45850002952}, "receiver_prefill_ms": 410.0491249701008, "ttft_ms": 21.183541044592857, "decode_ms": 231.33729107212275}, "total_wall_ms": 23796.669500065036, "ttft_ms": 21.183541044592857, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 7}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "9112735", "evidence_artifact": {"ledger": {"facts_total": 64, "facts_retained": 30, "retention_rate": 0.46875, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.6666666666666666, "status": 0.9166666666666666, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1087}, "ledger_by_hop": {"1": {"facts_total": 64, "facts_retained": 30, "retention_rate": 0.46875, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.6666666666666666, "status": 0.9166666666666666, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1087}}}, "reference_key_id": "keys/synthetic/syn_03006_2000.json", "task_scores": {"correct": 0.0, "abstained": 0.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "47f843b1989b354c7cbe406d456e62c59b6229490f55dde77bcdb5f70147ff28", "output_hash": "15a3f804636f0e62d4a3d9be7677f001a0ad6b1fb8ae7e8cdc63c44ec7bac31f"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "1b005258324923abc38104d3", "attempt_id": "db687ef8e77f", "run_group_id": "E4_core:A4|repairable|qwen3b:h1", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:01:19.311515Z", "task_family": "synthetic", "document_id": "syn_03006_2000", "document_hash": "89be5e77e0ca114963e9ffdf8dc38fe5610f13a78ac29e7a1956cd9124ce5217", "question_id": "q09_revision", "split": "test", "cluster_id": "syn_03006_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 1, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1890, "query_tokens": 604, "message_tokens": 512, "message_utf8_bytes": 1087, "cache_tensor_bytes": null, "serialized_payload_bytes": 1087, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 23133.45850002952, "hops": {"1": 23133.45850002952}, "receiver_prefill_ms": 354.9917080672458, "ttft_ms": 48.74066694173962, "decode_ms": 488.7044170172885}, "total_wall_ms": 24026.649207924493, "ttft_ms": 48.74066694173962, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 7}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "9112735", "evidence_artifact": {"ledger": {"facts_total": 64, "facts_retained": 30, "retention_rate": 0.46875, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.6666666666666666, "status": 0.9166666666666666, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1087}, "ledger_by_hop": {"1": {"facts_total": 64, "facts_retained": 30, "retention_rate": 0.46875, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.6666666666666666, "status": 0.9166666666666666, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1087}}}, "reference_key_id": "keys/synthetic/syn_03006_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "47f843b1989b354c7cbe406d456e62c59b6229490f55dde77bcdb5f70147ff28", "output_hash": "15a3f804636f0e62d4a3d9be7677f001a0ad6b1fb8ae7e8cdc63c44ec7bac31f"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "efff9f12315ffe0fa19d0951", "attempt_id": "4ba33288f6df", "run_group_id": "E4_core:A4|repairable|qwen3b:h1", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:01:20.223529Z", "task_family": "synthetic", "document_id": "syn_03006_2000", "document_hash": "89be5e77e0ca114963e9ffdf8dc38fe5610f13a78ac29e7a1956cd9124ce5217", "question_id": "q10_temporal", "split": "test", "cluster_id": "syn_03006_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 1, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1890, "query_tokens": 613, "message_tokens": 512, "message_utf8_bytes": 1087, "cache_tensor_bytes": null, "serialized_payload_bytes": 1087, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 23133.45850002952, "hops": {"1": 23133.45850002952}, "receiver_prefill_ms": 403.9764169137925, "ttft_ms": 23.038791958242655, "decode_ms": 38.53025007992983}, "total_wall_ms": 23599.64825003408, "ttft_ms": 23.038791958242655, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 64, "facts_retained": 30, "retention_rate": 0.46875, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.6666666666666666, "status": 0.9166666666666666, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1087}, "ledger_by_hop": {"1": {"facts_total": 64, "facts_retained": 30, "retention_rate": 0.46875, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.6666666666666666, "status": 0.9166666666666666, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1087}}}, "reference_key_id": "keys/synthetic/syn_03006_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "47f843b1989b354c7cbe406d456e62c59b6229490f55dde77bcdb5f70147ff28", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "e2f9b874ef7f611da993c03b", "attempt_id": "c657a7cd565e", "run_group_id": "E4_core:A4|repairable|qwen3b:h1", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:01:20.707075Z", "task_family": "synthetic", "document_id": "syn_03006_2000", "document_hash": "89be5e77e0ca114963e9ffdf8dc38fe5610f13a78ac29e7a1956cd9124ce5217", "question_id": "q11_aggregation", "split": "test", "cluster_id": "syn_03006_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 1, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1890, "query_tokens": 606, "message_tokens": 512, "message_utf8_bytes": 1087, "cache_tensor_bytes": null, "serialized_payload_bytes": 1087, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 23133.45850002952, "hops": {"1": 23133.45850002952}, "receiver_prefill_ms": 412.15241700410843, "ttft_ms": 29.35737499501556, "decode_ms": 317.75183300487697}, "total_wall_ms": 23893.397792009637, "ttft_ms": 29.35737499501556, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 7}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "8908553", "evidence_artifact": {"ledger": {"facts_total": 64, "facts_retained": 30, "retention_rate": 0.46875, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.6666666666666666, "status": 0.9166666666666666, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1087}, "ledger_by_hop": {"1": {"facts_total": 64, "facts_retained": 30, "retention_rate": 0.46875, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.6666666666666666, "status": 0.9166666666666666, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1087}}}, "reference_key_id": "keys/synthetic/syn_03006_2000.json", "task_scores": {"correct": 0.0, "abstained": 0.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "47f843b1989b354c7cbe406d456e62c59b6229490f55dde77bcdb5f70147ff28", "output_hash": "5ad9e777dc3be36808c74badfc8aae217b41a486754d07adf1abda6f6bce9aca"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "38f90f2d9593185da023a413", "attempt_id": "871ebcc2348b", "run_group_id": "E4_core:A4|repairable|qwen3b:h1", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:01:21.485976Z", "task_family": "synthetic", "document_id": "syn_03006_2000", "document_hash": "89be5e77e0ca114963e9ffdf8dc38fe5610f13a78ac29e7a1956cd9124ce5217", "question_id": "q12_unit_conversion", "split": "test", "cluster_id": "syn_03006_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 1, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1890, "query_tokens": 603, "message_tokens": 512, "message_utf8_bytes": 1087, "cache_tensor_bytes": null, "serialized_payload_bytes": 1087, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 23133.45850002952, "hops": {"1": 23133.45850002952}, "receiver_prefill_ms": 551.8338750116527, "ttft_ms": 35.3613329352811, "decode_ms": 173.36699995212257}, "total_wall_ms": 23894.823000067845, "ttft_ms": 35.3613329352811, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 5}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "39730", "evidence_artifact": {"ledger": {"facts_total": 64, "facts_retained": 30, "retention_rate": 0.46875, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.6666666666666666, "status": 0.9166666666666666, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1087}, "ledger_by_hop": {"1": {"facts_total": 64, "facts_retained": 30, "retention_rate": 0.46875, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.6666666666666666, "status": 0.9166666666666666, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1087}}}, "reference_key_id": "keys/synthetic/syn_03006_2000.json", "task_scores": {"correct": 0.0, "abstained": 0.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "47f843b1989b354c7cbe406d456e62c59b6229490f55dde77bcdb5f70147ff28", "output_hash": "91306ecd2f87f05ab1c9b3377d592e9ec7f50f091fa1f94b9799c87e2d2a0108"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "25a8ba93eebdf91d6bbd38f1", "attempt_id": "6e6f82ea9052", "run_group_id": "E4_core:A4|repairable|qwen3b:h1", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:01:22.265328Z", "task_family": "synthetic", "document_id": "syn_03006_2000", "document_hash": "89be5e77e0ca114963e9ffdf8dc38fe5610f13a78ac29e7a1956cd9124ce5217", "question_id": "q13_unanswerable", "split": "test", "cluster_id": "syn_03006_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 1, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1890, "query_tokens": 607, "message_tokens": 512, "message_utf8_bytes": 1087, "cache_tensor_bytes": null, "serialized_payload_bytes": 1087, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 23133.45850002952, "hops": {"1": 23133.45850002952}, "receiver_prefill_ms": 393.2112919865176, "ttft_ms": 20.688583026640117, "decode_ms": 42.170792003162205}, "total_wall_ms": 23590.342915966175, "ttft_ms": 20.688583026640117, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 64, "facts_retained": 30, "retention_rate": 0.46875, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.6666666666666666, "status": 0.9166666666666666, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1087}, "ledger_by_hop": {"1": {"facts_total": 64, "facts_retained": 30, "retention_rate": 0.46875, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.6666666666666666, "status": 0.9166666666666666, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1087}}}, "reference_key_id": "keys/synthetic/syn_03006_2000.json", "task_scores": {"correct": 1.0, "abstained": 1.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "47f843b1989b354c7cbe406d456e62c59b6229490f55dde77bcdb5f70147ff28", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "ffe55130f955af5319553ee5", "attempt_id": "37186cc259e1", "run_group_id": "E4_core:A4|repairable|qwen3b:h1", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:01:22.739636Z", "task_family": "synthetic", "document_id": "syn_03006_2000", "document_hash": "89be5e77e0ca114963e9ffdf8dc38fe5610f13a78ac29e7a1956cd9124ce5217", "question_id": "q14_unanswerable", "split": "test", "cluster_id": "syn_03006_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 1, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1890, "query_tokens": 608, "message_tokens": 512, "message_utf8_bytes": 1087, "cache_tensor_bytes": null, "serialized_payload_bytes": 1087, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 23133.45850002952, "hops": {"1": 23133.45850002952}, "receiver_prefill_ms": 366.7261250084266, "ttft_ms": 48.54020907077938, "decode_ms": 64.3161250045523}, "total_wall_ms": 23613.78854198847, "ttft_ms": 48.54020907077938, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 64, "facts_retained": 30, "retention_rate": 0.46875, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.6666666666666666, "status": 0.9166666666666666, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1087}, "ledger_by_hop": {"1": {"facts_total": 64, "facts_retained": 30, "retention_rate": 0.46875, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.6666666666666666, "status": 0.9166666666666666, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1087}}}, "reference_key_id": "keys/synthetic/syn_03006_2000.json", "task_scores": {"correct": 1.0, "abstained": 1.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "47f843b1989b354c7cbe406d456e62c59b6229490f55dde77bcdb5f70147ff28", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "b52f85c29c053e5d49cb09dc", "attempt_id": "d27c1aef0630", "run_group_id": "E4_core:A4|repairable|qwen3b:h2", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:01:45.293630Z", "task_family": "synthetic", "document_id": "syn_03006_2000", "document_hash": "89be5e77e0ca114963e9ffdf8dc38fe5610f13a78ac29e7a1956cd9124ce5217", "question_id": "q00_single_fact", "split": "test", "cluster_id": "syn_03006_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 2, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1890, "query_tokens": 594, "message_tokens": 511, "message_utf8_bytes": 1075, "cache_tensor_bytes": null, "serialized_payload_bytes": 1075, "side_information_bytes": 0, "fetched_bytes": 837, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 45188.29654203728, "hops": {"1": 23133.45850002952, "2": 22054.83804200776}, "receiver_prefill_ms": 449.77195793762803, "ttft_ms": 38.740749936550856, "decode_ms": 323.6604999983683}, "total_wall_ms": 46001.329167047516, "ttft_ms": 38.740749936550856, "input_usage": {"lookups": 4}, "output_usage": {"output_tokens": 10}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "XM-76Q-2958", "evidence_artifact": {"ledger": {"facts_total": 64, "facts_retained": 30, "retention_rate": 0.46875, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.6666666666666666, "status": 0.9166666666666666, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1075}, "ledger_by_hop": {"1": {"facts_total": 64, "facts_retained": 30, "retention_rate": 0.46875, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.6666666666666666, "status": 0.9166666666666666, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1087}, "2": {"facts_total": 64, "facts_retained": 30, "retention_rate": 0.46875, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.6666666666666666, "status": 0.9166666666666666, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1075}}}, "reference_key_id": "keys/synthetic/syn_03006_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "42789030d2b9a011f83c6582210309bf81b6016770030d2236e8df8b76ecbe77", "output_hash": "cf42691eec772b9ec74d67643944de2e19e53724d7b772cef1e0f2d5ef3b4926"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "99f9fa6569e136e149925b7f", "attempt_id": "a8b40a922291", "run_group_id": "E4_core:A4|repairable|qwen3b:h2", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:01:46.125632Z", "task_family": "synthetic", "document_id": "syn_03006_2000", "document_hash": "89be5e77e0ca114963e9ffdf8dc38fe5610f13a78ac29e7a1956cd9124ce5217", "question_id": "q01_single_fact", "split": "test", "cluster_id": "syn_03006_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 2, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1890, "query_tokens": 599, "message_tokens": 511, "message_utf8_bytes": 1075, "cache_tensor_bytes": null, "serialized_payload_bytes": 1075, "side_information_bytes": 0, "fetched_bytes": 837, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 45188.29654203728, "hops": {"1": 23133.45850002952, "2": 22054.83804200776}, "receiver_prefill_ms": 360.3984999936074, "ttft_ms": 49.495792016386986, "decode_ms": 43.283000006340444}, "total_wall_ms": 45642.3369590193, "ttft_ms": 49.495792016386986, "input_usage": {"lookups": 4}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 64, "facts_retained": 30, "retention_rate": 0.46875, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.6666666666666666, "status": 0.9166666666666666, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1075}, "ledger_by_hop": {"1": {"facts_total": 64, "facts_retained": 30, "retention_rate": 0.46875, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.6666666666666666, "status": 0.9166666666666666, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1087}, "2": {"facts_total": 64, "facts_retained": 30, "retention_rate": 0.46875, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.6666666666666666, "status": 0.9166666666666666, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1075}}}, "reference_key_id": "keys/synthetic/syn_03006_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "42789030d2b9a011f83c6582210309bf81b6016770030d2236e8df8b76ecbe77", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "5fe9b0c1b6c55f8429dc0027", "attempt_id": "863c2bf84239", "run_group_id": "E4_core:A4|repairable|qwen3b:h2", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:01:46.596832Z", "task_family": "synthetic", "document_id": "syn_03006_2000", "document_hash": "89be5e77e0ca114963e9ffdf8dc38fe5610f13a78ac29e7a1956cd9124ce5217", "question_id": "q02_exact_copy", "split": "test", "cluster_id": "syn_03006_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 2, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1890, "query_tokens": 598, "message_tokens": 511, "message_utf8_bytes": 1075, "cache_tensor_bytes": null, "serialized_payload_bytes": 1075, "side_information_bytes": 0, "fetched_bytes": 837, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 45188.29654203728, "hops": {"1": 23133.45850002952, "2": 22054.83804200776}, "receiver_prefill_ms": 431.5095420461148, "ttft_ms": 100.26474995538592, "decode_ms": 41.763957939110696}, "total_wall_ms": 45762.4925840646, "ttft_ms": 100.26474995538592, "input_usage": {"lookups": 4}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 64, "facts_retained": 30, "retention_rate": 0.46875, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.6666666666666666, "status": 0.9166666666666666, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1075}, "ledger_by_hop": {"1": {"facts_total": 64, "facts_retained": 30, "retention_rate": 0.46875, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.6666666666666666, "status": 0.9166666666666666, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1087}, "2": {"facts_total": 64, "facts_retained": 30, "retention_rate": 0.46875, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.6666666666666666, "status": 0.9166666666666666, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1075}}}, "reference_key_id": "keys/synthetic/syn_03006_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "42789030d2b9a011f83c6582210309bf81b6016770030d2236e8df8b76ecbe77", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "13b5ebae74e7434297653972", "attempt_id": "6ad5e51d061d", "run_group_id": "E4_core:A4|repairable|qwen3b:h2", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:01:47.188433Z", "task_family": "synthetic", "document_id": "syn_03006_2000", "document_hash": "89be5e77e0ca114963e9ffdf8dc38fe5610f13a78ac29e7a1956cd9124ce5217", "question_id": "q03_exact_copy", "split": "test", "cluster_id": "syn_03006_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 2, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1890, "query_tokens": 593, "message_tokens": 511, "message_utf8_bytes": 1075, "cache_tensor_bytes": null, "serialized_payload_bytes": 1075, "side_information_bytes": 0, "fetched_bytes": 837, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 45188.29654203728, "hops": {"1": 23133.45850002952, "2": 22054.83804200776}, "receiver_prefill_ms": 402.5659170001745, "ttft_ms": 23.23645807337016, "decode_ms": 279.5754160033539}, "total_wall_ms": 45894.45241703652, "ttft_ms": 23.23645807337016, "input_usage": {"lookups": 4}, "output_usage": {"output_tokens": 8}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "Haruki Ouellet-Marchand", "evidence_artifact": {"ledger": {"facts_total": 64, "facts_retained": 30, "retention_rate": 0.46875, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.6666666666666666, "status": 0.9166666666666666, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1075}, "ledger_by_hop": {"1": {"facts_total": 64, "facts_retained": 30, "retention_rate": 0.46875, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.6666666666666666, "status": 0.9166666666666666, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1087}, "2": {"facts_total": 64, "facts_retained": 30, "retention_rate": 0.46875, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.6666666666666666, "status": 0.9166666666666666, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1075}}}, "reference_key_id": "keys/synthetic/syn_03006_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "42789030d2b9a011f83c6582210309bf81b6016770030d2236e8df8b76ecbe77", "output_hash": "a3b6e0a44d495387d8b8fc50ba6be0f2ed43121c674d3b3e962f8c4690e15b5c"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "62c273f03d0fc7603da60ea1", "attempt_id": "bf93aca95ad5", "run_group_id": "E4_core:A4|repairable|qwen3b:h2", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:01:47.912000Z", "task_family": "synthetic", "document_id": "syn_03006_2000", "document_hash": "89be5e77e0ca114963e9ffdf8dc38fe5610f13a78ac29e7a1956cd9124ce5217", "question_id": "q04_multi_hop", "split": "test", "cluster_id": "syn_03006_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 2, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1890, "query_tokens": 602, "message_tokens": 511, "message_utf8_bytes": 1075, "cache_tensor_bytes": null, "serialized_payload_bytes": 1075, "side_information_bytes": 0, "fetched_bytes": 837, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 45188.29654203728, "hops": {"1": 23133.45850002952, "2": 22054.83804200776}, "receiver_prefill_ms": 364.78745902422816, "ttft_ms": 42.94341697823256, "decode_ms": 56.561124976724386}, "total_wall_ms": 45653.37129205, "ttft_ms": 42.94341697823256, "input_usage": {"lookups": 4}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 64, "facts_retained": 30, "retention_rate": 0.46875, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.6666666666666666, "status": 0.9166666666666666, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1075}, "ledger_by_hop": {"1": {"facts_total": 64, "facts_retained": 30, "retention_rate": 0.46875, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.6666666666666666, "status": 0.9166666666666666, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1087}, "2": {"facts_total": 64, "facts_retained": 30, "retention_rate": 0.46875, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.6666666666666666, "status": 0.9166666666666666, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1075}}}, "reference_key_id": "keys/synthetic/syn_03006_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "42789030d2b9a011f83c6582210309bf81b6016770030d2236e8df8b76ecbe77", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "4555c6192f8bcc0a9abac89e", "attempt_id": "bc2ffd78283c", "run_group_id": "E4_core:A4|repairable|qwen3b:h2", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:01:48.395970Z", "task_family": "synthetic", "document_id": "syn_03006_2000", "document_hash": "89be5e77e0ca114963e9ffdf8dc38fe5610f13a78ac29e7a1956cd9124ce5217", "question_id": "q05_multi_hop", "split": "test", "cluster_id": "syn_03006_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 2, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1890, "query_tokens": 605, "message_tokens": 511, "message_utf8_bytes": 1075, "cache_tensor_bytes": null, "serialized_payload_bytes": 1075, "side_information_bytes": 0, "fetched_bytes": 837, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 45188.29654203728, "hops": {"1": 23133.45850002952, "2": 22054.83804200776}, "receiver_prefill_ms": 550.4677499411628, "ttft_ms": 28.56187499128282, "decode_ms": 344.4353339727968}, "total_wall_ms": 46112.53808403853, "ttft_ms": 28.56187499128282, "input_usage": {"lookups": 4}, "output_usage": {"output_tokens": 10}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "EP-52G-1330", "evidence_artifact": {"ledger": {"facts_total": 64, "facts_retained": 30, "retention_rate": 0.46875, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.6666666666666666, "status": 0.9166666666666666, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1075}, "ledger_by_hop": {"1": {"facts_total": 64, "facts_retained": 30, "retention_rate": 0.46875, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.6666666666666666, "status": 0.9166666666666666, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1087}, "2": {"facts_total": 64, "facts_retained": 30, "retention_rate": 0.46875, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.6666666666666666, "status": 0.9166666666666666, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1075}}}, "reference_key_id": "keys/synthetic/syn_03006_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "42789030d2b9a011f83c6582210309bf81b6016770030d2236e8df8b76ecbe77", "output_hash": "104db207937fd79e23ce16f9b3f10b1a9ed34f6dd0441a99be6ea1b2d0a09d99"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "984285dc9407266e225c89f5", "attempt_id": "5ce9c4547730", "run_group_id": "E4_core:A4|repairable|qwen3b:h2", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:01:49.337285Z", "task_family": "synthetic", "document_id": "syn_03006_2000", "document_hash": "89be5e77e0ca114963e9ffdf8dc38fe5610f13a78ac29e7a1956cd9124ce5217", "question_id": "q06_negation", "split": "test", "cluster_id": "syn_03006_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 2, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1890, "query_tokens": 593, "message_tokens": 511, "message_utf8_bytes": 1075, "cache_tensor_bytes": null, "serialized_payload_bytes": 1075, "side_information_bytes": 0, "fetched_bytes": 837, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 45188.29654203728, "hops": {"1": 23133.45850002952, "2": 22054.83804200776}, "receiver_prefill_ms": 389.82012507040054, "ttft_ms": 17.422124976292253, "decode_ms": 45.693375053815544}, "total_wall_ms": 45642.17450004071, "ttft_ms": 17.422124976292253, "input_usage": {"lookups": 4}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "no", "evidence_artifact": {"ledger": {"facts_total": 64, "facts_retained": 30, "retention_rate": 0.46875, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.6666666666666666, "status": 0.9166666666666666, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1075}, "ledger_by_hop": {"1": {"facts_total": 64, "facts_retained": 30, "retention_rate": 0.46875, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.6666666666666666, "status": 0.9166666666666666, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1087}, "2": {"facts_total": 64, "facts_retained": 30, "retention_rate": 0.46875, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.6666666666666666, "status": 0.9166666666666666, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1075}}}, "reference_key_id": "keys/synthetic/syn_03006_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "42789030d2b9a011f83c6582210309bf81b6016770030d2236e8df8b76ecbe77", "output_hash": "9390298f3fb0c5b160498935d79cb139aef28e1c47358b4bbba61862b9c26e59"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "4608127cd8ba812fdb44b917", "attempt_id": "2b188accae34", "run_group_id": "E4_core:A4|repairable|qwen3b:h2", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:01:49.809419Z", "task_family": "synthetic", "document_id": "syn_03006_2000", "document_hash": "89be5e77e0ca114963e9ffdf8dc38fe5610f13a78ac29e7a1956cd9124ce5217", "question_id": "q07_negation", "split": "test", "cluster_id": "syn_03006_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 2, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1890, "query_tokens": 593, "message_tokens": 511, "message_utf8_bytes": 1075, "cache_tensor_bytes": null, "serialized_payload_bytes": 1075, "side_information_bytes": 0, "fetched_bytes": 837, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 45188.29654203728, "hops": {"1": 23133.45850002952, "2": 22054.83804200776}, "receiver_prefill_ms": 368.4867910342291, "ttft_ms": 35.65591701772064, "decode_ms": 115.14962499495596}, "total_wall_ms": 45708.3132090047, "ttft_ms": 35.65591701772064, "input_usage": {"lookups": 4}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "no", "evidence_artifact": {"ledger": {"facts_total": 64, "facts_retained": 30, "retention_rate": 0.46875, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.6666666666666666, "status": 0.9166666666666666, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1075}, "ledger_by_hop": {"1": {"facts_total": 64, "facts_retained": 30, "retention_rate": 0.46875, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.6666666666666666, "status": 0.9166666666666666, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1087}, "2": {"facts_total": 64, "facts_retained": 30, "retention_rate": 0.46875, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.6666666666666666, "status": 0.9166666666666666, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1075}}}, "reference_key_id": "keys/synthetic/syn_03006_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "42789030d2b9a011f83c6582210309bf81b6016770030d2236e8df8b76ecbe77", "output_hash": "9390298f3fb0c5b160498935d79cb139aef28e1c47358b4bbba61862b9c26e59"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "199243a3e990e6a9d4503fc2", "attempt_id": "72b35ebddaa8", "run_group_id": "E4_core:A4|repairable|qwen3b:h2", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:01:50.346670Z", "task_family": "synthetic", "document_id": "syn_03006_2000", "document_hash": "89be5e77e0ca114963e9ffdf8dc38fe5610f13a78ac29e7a1956cd9124ce5217", "question_id": "q08_revision", "split": "test", "cluster_id": "syn_03006_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 2, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1890, "query_tokens": 610, "message_tokens": 511, "message_utf8_bytes": 1075, "cache_tensor_bytes": null, "serialized_payload_bytes": 1075, "side_information_bytes": 0, "fetched_bytes": 837, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 45188.29654203728, "hops": {"1": 23133.45850002952, "2": 22054.83804200776}, "receiver_prefill_ms": 477.5704579660669, "ttft_ms": 33.35229097865522, "decode_ms": 316.6987909935415}, "total_wall_ms": 46016.66720805224, "ttft_ms": 33.35229097865522, "input_usage": {"lookups": 4}, "output_usage": {"output_tokens": 9}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "9,112,735", "evidence_artifact": {"ledger": {"facts_total": 64, "facts_retained": 30, "retention_rate": 0.46875, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.6666666666666666, "status": 0.9166666666666666, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1075}, "ledger_by_hop": {"1": {"facts_total": 64, "facts_retained": 30, "retention_rate": 0.46875, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.6666666666666666, "status": 0.9166666666666666, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1087}, "2": {"facts_total": 64, "facts_retained": 30, "retention_rate": 0.46875, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.6666666666666666, "status": 0.9166666666666666, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1075}}}, "reference_key_id": "keys/synthetic/syn_03006_2000.json", "task_scores": {"correct": 0.0, "abstained": 0.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "42789030d2b9a011f83c6582210309bf81b6016770030d2236e8df8b76ecbe77", "output_hash": "788fcda908c2029597420995b09d0cc919bfdbf7a28b68dcb0e331db7df7b342"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "3ff561ab3f36025dd9c5057e", "attempt_id": "beffc2c9e818", "run_group_id": "E4_core:A4|repairable|qwen3b:h2", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:01:51.194874Z", "task_family": "synthetic", "document_id": "syn_03006_2000", "document_hash": "89be5e77e0ca114963e9ffdf8dc38fe5610f13a78ac29e7a1956cd9124ce5217", "question_id": "q09_revision", "split": "test", "cluster_id": "syn_03006_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 2, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1890, "query_tokens": 603, "message_tokens": 511, "message_utf8_bytes": 1075, "cache_tensor_bytes": null, "serialized_payload_bytes": 1075, "side_information_bytes": 0, "fetched_bytes": 837, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 45188.29654203728, "hops": {"1": 23133.45850002952, "2": 22054.83804200776}, "receiver_prefill_ms": 363.9607500517741, "ttft_ms": 52.99333296716213, "decode_ms": 494.38329110853374}, "total_wall_ms": 46100.35254212562, "ttft_ms": 52.99333296716213, "input_usage": {"lookups": 4}, "output_usage": {"output_tokens": 7}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "9112735", "evidence_artifact": {"ledger": {"facts_total": 64, "facts_retained": 30, "retention_rate": 0.46875, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.6666666666666666, "status": 0.9166666666666666, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1075}, "ledger_by_hop": {"1": {"facts_total": 64, "facts_retained": 30, "retention_rate": 0.46875, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.6666666666666666, "status": 0.9166666666666666, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1087}, "2": {"facts_total": 64, "facts_retained": 30, "retention_rate": 0.46875, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.6666666666666666, "status": 0.9166666666666666, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1075}}}, "reference_key_id": "keys/synthetic/syn_03006_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "42789030d2b9a011f83c6582210309bf81b6016770030d2236e8df8b76ecbe77", "output_hash": "15a3f804636f0e62d4a3d9be7677f001a0ad6b1fb8ae7e8cdc63c44ec7bac31f"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "aa397281e2f4fd75c66b406e", "attempt_id": "4ee9378f7f81", "run_group_id": "E4_core:A4|repairable|qwen3b:h2", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:01:52.124692Z", "task_family": "synthetic", "document_id": "syn_03006_2000", "document_hash": "89be5e77e0ca114963e9ffdf8dc38fe5610f13a78ac29e7a1956cd9124ce5217", "question_id": "q10_temporal", "split": "test", "cluster_id": "syn_03006_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 2, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1890, "query_tokens": 612, "message_tokens": 511, "message_utf8_bytes": 1075, "cache_tensor_bytes": null, "serialized_payload_bytes": 1075, "side_information_bytes": 0, "fetched_bytes": 837, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 45188.29654203728, "hops": {"1": 23133.45850002952, "2": 22054.83804200776}, "receiver_prefill_ms": 406.88266698271036, "ttft_ms": 30.352666974067688, "decode_ms": 36.78012499585748}, "total_wall_ms": 45662.99708397128, "ttft_ms": 30.352666974067688, "input_usage": {"lookups": 4}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 64, "facts_retained": 30, "retention_rate": 0.46875, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.6666666666666666, "status": 0.9166666666666666, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1075}, "ledger_by_hop": {"1": {"facts_total": 64, "facts_retained": 30, "retention_rate": 0.46875, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.6666666666666666, "status": 0.9166666666666666, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1087}, "2": {"facts_total": 64, "facts_retained": 30, "retention_rate": 0.46875, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.6666666666666666, "status": 0.9166666666666666, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1075}}}, "reference_key_id": "keys/synthetic/syn_03006_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "42789030d2b9a011f83c6582210309bf81b6016770030d2236e8df8b76ecbe77", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "03f9f1e4adf4834e24b5fa30", "attempt_id": "55a3c848ea60", "run_group_id": "E4_core:A4|repairable|qwen3b:h2", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:01:52.616763Z", "task_family": "synthetic", "document_id": "syn_03006_2000", "document_hash": "89be5e77e0ca114963e9ffdf8dc38fe5610f13a78ac29e7a1956cd9124ce5217", "question_id": "q11_aggregation", "split": "test", "cluster_id": "syn_03006_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 2, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1890, "query_tokens": 605, "message_tokens": 511, "message_utf8_bytes": 1075, "cache_tensor_bytes": null, "serialized_payload_bytes": 1075, "side_information_bytes": 0, "fetched_bytes": 837, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 45188.29654203728, "hops": {"1": 23133.45850002952, "2": 22054.83804200776}, "receiver_prefill_ms": 410.47204192727804, "ttft_ms": 25.584000046364963, "decode_ms": 322.41962500847876}, "total_wall_ms": 45947.47708400246, "ttft_ms": 25.584000046364963, "input_usage": {"lookups": 4}, "output_usage": {"output_tokens": 7}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "8908553", "evidence_artifact": {"ledger": {"facts_total": 64, "facts_retained": 30, "retention_rate": 0.46875, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.6666666666666666, "status": 0.9166666666666666, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1075}, "ledger_by_hop": {"1": {"facts_total": 64, "facts_retained": 30, "retention_rate": 0.46875, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.6666666666666666, "status": 0.9166666666666666, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1087}, "2": {"facts_total": 64, "facts_retained": 30, "retention_rate": 0.46875, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.6666666666666666, "status": 0.9166666666666666, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1075}}}, "reference_key_id": "keys/synthetic/syn_03006_2000.json", "task_scores": {"correct": 0.0, "abstained": 0.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "42789030d2b9a011f83c6582210309bf81b6016770030d2236e8df8b76ecbe77", "output_hash": "5ad9e777dc3be36808c74badfc8aae217b41a486754d07adf1abda6f6bce9aca"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "b85f55b995aecb4899d28aff", "attempt_id": "f85bd42aafd9", "run_group_id": "E4_core:A4|repairable|qwen3b:h2", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:01:53.393247Z", "task_family": "synthetic", "document_id": "syn_03006_2000", "document_hash": "89be5e77e0ca114963e9ffdf8dc38fe5610f13a78ac29e7a1956cd9124ce5217", "question_id": "q12_unit_conversion", "split": "test", "cluster_id": "syn_03006_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 2, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1890, "query_tokens": 602, "message_tokens": 511, "message_utf8_bytes": 1075, "cache_tensor_bytes": null, "serialized_payload_bytes": 1075, "side_information_bytes": 0, "fetched_bytes": 837, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 45188.29654203728, "hops": {"1": 23133.45850002952, "2": 22054.83804200776}, "receiver_prefill_ms": 531.24366607517, "ttft_ms": 24.068832979537547, "decode_ms": 212.30354206636548}, "total_wall_ms": 45956.622251076624, "ttft_ms": 24.068832979537547, "input_usage": {"lookups": 4}, "output_usage": {"output_tokens": 6}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "139400", "evidence_artifact": {"ledger": {"facts_total": 64, "facts_retained": 30, "retention_rate": 0.46875, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.6666666666666666, "status": 0.9166666666666666, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1075}, "ledger_by_hop": {"1": {"facts_total": 64, "facts_retained": 30, "retention_rate": 0.46875, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.6666666666666666, "status": 0.9166666666666666, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1087}, "2": {"facts_total": 64, "facts_retained": 30, "retention_rate": 0.46875, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.6666666666666666, "status": 0.9166666666666666, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1075}}}, "reference_key_id": "keys/synthetic/syn_03006_2000.json", "task_scores": {"correct": 0.0, "abstained": 0.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "42789030d2b9a011f83c6582210309bf81b6016770030d2236e8df8b76ecbe77", "output_hash": "44e2108c521dad3f1a7e752d78e4bb639548505bd781de7df6eb35a3bd8fb735"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "7163ac81092795432613a36d", "attempt_id": "0089880c6706", "run_group_id": "E4_core:A4|repairable|qwen3b:h2", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:01:54.180552Z", "task_family": "synthetic", "document_id": "syn_03006_2000", "document_hash": "89be5e77e0ca114963e9ffdf8dc38fe5610f13a78ac29e7a1956cd9124ce5217", "question_id": "q13_unanswerable", "split": "test", "cluster_id": "syn_03006_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 2, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1890, "query_tokens": 606, "message_tokens": 511, "message_utf8_bytes": 1075, "cache_tensor_bytes": null, "serialized_payload_bytes": 1075, "side_information_bytes": 0, "fetched_bytes": 837, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 45188.29654203728, "hops": {"1": 23133.45850002952, "2": 22054.83804200776}, "receiver_prefill_ms": 390.4730830108747, "ttft_ms": 42.450750013813376, "decode_ms": 22.23758294712752}, "total_wall_ms": 45644.107500091195, "ttft_ms": 42.450750013813376, "input_usage": {"lookups": 4}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 64, "facts_retained": 30, "retention_rate": 0.46875, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.6666666666666666, "status": 0.9166666666666666, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1075}, "ledger_by_hop": {"1": {"facts_total": 64, "facts_retained": 30, "retention_rate": 0.46875, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.6666666666666666, "status": 0.9166666666666666, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1087}, "2": {"facts_total": 64, "facts_retained": 30, "retention_rate": 0.46875, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.6666666666666666, "status": 0.9166666666666666, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1075}}}, "reference_key_id": "keys/synthetic/syn_03006_2000.json", "task_scores": {"correct": 1.0, "abstained": 1.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "42789030d2b9a011f83c6582210309bf81b6016770030d2236e8df8b76ecbe77", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "122f96e3b8f9a88de2473351", "attempt_id": "fed2a4fb179b", "run_group_id": "E4_core:A4|repairable|qwen3b:h2", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:01:54.654765Z", "task_family": "synthetic", "document_id": "syn_03006_2000", "document_hash": "89be5e77e0ca114963e9ffdf8dc38fe5610f13a78ac29e7a1956cd9124ce5217", "question_id": "q14_unanswerable", "split": "test", "cluster_id": "syn_03006_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 2, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1890, "query_tokens": 607, "message_tokens": 511, "message_utf8_bytes": 1075, "cache_tensor_bytes": null, "serialized_payload_bytes": 1075, "side_information_bytes": 0, "fetched_bytes": 837, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 45188.29654203728, "hops": {"1": 23133.45850002952, "2": 22054.83804200776}, "receiver_prefill_ms": 367.8692919202149, "ttft_ms": 49.65137504041195, "decode_ms": 54.254209040664136}, "total_wall_ms": 45660.93229211401, "ttft_ms": 49.65137504041195, "input_usage": {"lookups": 4}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 64, "facts_retained": 30, "retention_rate": 0.46875, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.6666666666666666, "status": 0.9166666666666666, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1075}, "ledger_by_hop": {"1": {"facts_total": 64, "facts_retained": 30, "retention_rate": 0.46875, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.6666666666666666, "status": 0.9166666666666666, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1087}, "2": {"facts_total": 64, "facts_retained": 30, "retention_rate": 0.46875, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.6666666666666666, "status": 0.9166666666666666, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1075}}}, "reference_key_id": "keys/synthetic/syn_03006_2000.json", "task_scores": {"correct": 1.0, "abstained": 1.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "42789030d2b9a011f83c6582210309bf81b6016770030d2236e8df8b76ecbe77", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "3f409b7ea06a149ef1f7d75b", "attempt_id": "f9c74f0c6eb9", "run_group_id": "E4_core:A4|repairable|qwen3b:h4", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:02:39.479798Z", "task_family": "synthetic", "document_id": "syn_03006_2000", "document_hash": "89be5e77e0ca114963e9ffdf8dc38fe5610f13a78ac29e7a1956cd9124ce5217", "question_id": "q00_single_fact", "split": "test", "cluster_id": "syn_03006_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 4, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1890, "query_tokens": 595, "message_tokens": 512, "message_utf8_bytes": 1364, "cache_tensor_bytes": null, "serialized_payload_bytes": 1364, "side_information_bytes": 0, "fetched_bytes": 2511, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 89519.31475102901, "hops": {"1": 23133.45850002952, "2": 22054.83804200776, "3": 22289.791416958906, "4": 22041.22679203283}, "receiver_prefill_ms": 403.83245807606727, "ttft_ms": 31.651000026613474, "decode_ms": 429.90587500389665}, "total_wall_ms": 90385.48554200679, "ttft_ms": 31.651000026613474, "input_usage": {"lookups": 12}, "output_usage": {"output_tokens": 10}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "XM-76Q-2958", "evidence_artifact": {"ledger": {"facts_total": 64, "facts_retained": 34, "retention_rate": 0.53125, "retention_by_kind": {"person": 0.5, "date": 0.3, "amount": 0.6666666666666666, "status": 0.9166666666666666, "place": 0.16666666666666666, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1364}, "ledger_by_hop": {"1": {"facts_total": 64, "facts_retained": 30, "retention_rate": 0.46875, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.6666666666666666, "status": 0.9166666666666666, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1087}, "2": {"facts_total": 64, "facts_retained": 30, "retention_rate": 0.46875, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.6666666666666666, "status": 0.9166666666666666, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1075}, "3": {"facts_total": 64, "facts_retained": 34, "retention_rate": 0.53125, "retention_by_kind": {"person": 0.5, "date": 0.3, "amount": 0.6666666666666666, "status": 0.9166666666666666, "place": 0.16666666666666666, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1364}, "4": {"facts_total": 64, "facts_retained": 34, "retention_rate": 0.53125, "retention_by_kind": {"person": 0.5, "date": 0.3, "amount": 0.6666666666666666, "status": 0.9166666666666666, "place": 0.16666666666666666, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1364}}}, "reference_key_id": "keys/synthetic/syn_03006_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "05248504387de244bb26ede8192b5c4b34d448e31f7043a95c04d8a18c00902a", "output_hash": "cf42691eec772b9ec74d67643944de2e19e53724d7b772cef1e0f2d5ef3b4926"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "56fc985753d89e9494575071", "attempt_id": "fa73070d449f", "run_group_id": "E4_core:A4|repairable|qwen3b:h4", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:02:40.363914Z", "task_family": "synthetic", "document_id": "syn_03006_2000", "document_hash": "89be5e77e0ca114963e9ffdf8dc38fe5610f13a78ac29e7a1956cd9124ce5217", "question_id": "q01_single_fact", "split": "test", "cluster_id": "syn_03006_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 4, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1890, "query_tokens": 600, "message_tokens": 512, "message_utf8_bytes": 1364, "cache_tensor_bytes": null, "serialized_payload_bytes": 1364, "side_information_bytes": 0, "fetched_bytes": 2511, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 89519.31475102901, "hops": {"1": 23133.45850002952, "2": 22054.83804200776, "3": 22289.791416958906, "4": 22041.22679203283}, "receiver_prefill_ms": 555.209832964465, "ttft_ms": 34.7244159784168, "decode_ms": 32.005541026592255}, "total_wall_ms": 90141.90504304133, "ttft_ms": 34.7244159784168, "input_usage": {"lookups": 12}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 64, "facts_retained": 34, "retention_rate": 0.53125, "retention_by_kind": {"person": 0.5, "date": 0.3, "amount": 0.6666666666666666, "status": 0.9166666666666666, "place": 0.16666666666666666, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1364}, "ledger_by_hop": {"1": {"facts_total": 64, "facts_retained": 30, "retention_rate": 0.46875, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.6666666666666666, "status": 0.9166666666666666, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1087}, "2": {"facts_total": 64, "facts_retained": 30, "retention_rate": 0.46875, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.6666666666666666, "status": 0.9166666666666666, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1075}, "3": {"facts_total": 64, "facts_retained": 34, "retention_rate": 0.53125, "retention_by_kind": {"person": 0.5, "date": 0.3, "amount": 0.6666666666666666, "status": 0.9166666666666666, "place": 0.16666666666666666, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1364}, "4": {"facts_total": 64, "facts_retained": 34, "retention_rate": 0.53125, "retention_by_kind": {"person": 0.5, "date": 0.3, "amount": 0.6666666666666666, "status": 0.9166666666666666, "place": 0.16666666666666666, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1364}}}, "reference_key_id": "keys/synthetic/syn_03006_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "05248504387de244bb26ede8192b5c4b34d448e31f7043a95c04d8a18c00902a", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "793d4608cd48819a37f7eb89", "attempt_id": "b7d152ba1df8", "run_group_id": "E4_core:A4|repairable|qwen3b:h4", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:02:41.004317Z", "task_family": "synthetic", "document_id": "syn_03006_2000", "document_hash": "89be5e77e0ca114963e9ffdf8dc38fe5610f13a78ac29e7a1956cd9124ce5217", "question_id": "q02_exact_copy", "split": "test", "cluster_id": "syn_03006_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 4, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1890, "query_tokens": 599, "message_tokens": 512, "message_utf8_bytes": 1364, "cache_tensor_bytes": null, "serialized_payload_bytes": 1364, "side_information_bytes": 0, "fetched_bytes": 2511, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 89519.31475102901, "hops": {"1": 23133.45850002952, "2": 22054.83804200776, "3": 22289.791416958906, "4": 22041.22679203283}, "receiver_prefill_ms": 392.0795830199495, "ttft_ms": 24.72862508147955, "decode_ms": 37.439291016198695}, "total_wall_ms": 89974.4116680231, "ttft_ms": 24.72862508147955, "input_usage": {"lookups": 12}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 64, "facts_retained": 34, "retention_rate": 0.53125, "retention_by_kind": {"person": 0.5, "date": 0.3, "amount": 0.6666666666666666, "status": 0.9166666666666666, "place": 0.16666666666666666, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1364}, "ledger_by_hop": {"1": {"facts_total": 64, "facts_retained": 30, "retention_rate": 0.46875, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.6666666666666666, "status": 0.9166666666666666, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1087}, "2": {"facts_total": 64, "facts_retained": 30, "retention_rate": 0.46875, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.6666666666666666, "status": 0.9166666666666666, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1075}, "3": {"facts_total": 64, "facts_retained": 34, "retention_rate": 0.53125, "retention_by_kind": {"person": 0.5, "date": 0.3, "amount": 0.6666666666666666, "status": 0.9166666666666666, "place": 0.16666666666666666, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1364}, "4": {"facts_total": 64, "facts_retained": 34, "retention_rate": 0.53125, "retention_by_kind": {"person": 0.5, "date": 0.3, "amount": 0.6666666666666666, "status": 0.9166666666666666, "place": 0.16666666666666666, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1364}}}, "reference_key_id": "keys/synthetic/syn_03006_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "05248504387de244bb26ede8192b5c4b34d448e31f7043a95c04d8a18c00902a", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "546fdfd04fe7d36b4bf8acba", "attempt_id": "a5630667e1da", "run_group_id": "E4_core:A4|repairable|qwen3b:h4", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:02:41.476728Z", "task_family": "synthetic", "document_id": "syn_03006_2000", "document_hash": "89be5e77e0ca114963e9ffdf8dc38fe5610f13a78ac29e7a1956cd9124ce5217", "question_id": "q03_exact_copy", "split": "test", "cluster_id": "syn_03006_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 4, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1890, "query_tokens": 594, "message_tokens": 512, "message_utf8_bytes": 1364, "cache_tensor_bytes": null, "serialized_payload_bytes": 1364, "side_information_bytes": 0, "fetched_bytes": 2511, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 89519.31475102901, "hops": {"1": 23133.45850002952, "2": 22054.83804200776, "3": 22289.791416958906, "4": 22041.22679203283}, "receiver_prefill_ms": 357.8727910062298, "ttft_ms": 35.906082950532436, "decode_ms": 534.4462089706212}, "total_wall_ms": 90448.26733402442, "ttft_ms": 35.906082950532436, "input_usage": {"lookups": 12}, "output_usage": {"output_tokens": 8}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "Haruki Ouellet-Marchand", "evidence_artifact": {"ledger": {"facts_total": 64, "facts_retained": 34, "retention_rate": 0.53125, "retention_by_kind": {"person": 0.5, "date": 0.3, "amount": 0.6666666666666666, "status": 0.9166666666666666, "place": 0.16666666666666666, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1364}, "ledger_by_hop": {"1": {"facts_total": 64, "facts_retained": 30, "retention_rate": 0.46875, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.6666666666666666, "status": 0.9166666666666666, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1087}, "2": {"facts_total": 64, "facts_retained": 30, "retention_rate": 0.46875, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.6666666666666666, "status": 0.9166666666666666, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1075}, "3": {"facts_total": 64, "facts_retained": 34, "retention_rate": 0.53125, "retention_by_kind": {"person": 0.5, "date": 0.3, "amount": 0.6666666666666666, "status": 0.9166666666666666, "place": 0.16666666666666666, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1364}, "4": {"facts_total": 64, "facts_retained": 34, "retention_rate": 0.53125, "retention_by_kind": {"person": 0.5, "date": 0.3, "amount": 0.6666666666666666, "status": 0.9166666666666666, "place": 0.16666666666666666, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1364}}}, "reference_key_id": "keys/synthetic/syn_03006_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "05248504387de244bb26ede8192b5c4b34d448e31f7043a95c04d8a18c00902a", "output_hash": "a3b6e0a44d495387d8b8fc50ba6be0f2ed43121c674d3b3e962f8c4690e15b5c"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "94bb1761a358c34644764a81", "attempt_id": "69e734b6aca0", "run_group_id": "E4_core:A4|repairable|qwen3b:h4", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:02:42.423908Z", "task_family": "synthetic", "document_id": "syn_03006_2000", "document_hash": "89be5e77e0ca114963e9ffdf8dc38fe5610f13a78ac29e7a1956cd9124ce5217", "question_id": "q04_multi_hop", "split": "test", "cluster_id": "syn_03006_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 4, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1890, "query_tokens": 603, "message_tokens": 512, "message_utf8_bytes": 1364, "cache_tensor_bytes": null, "serialized_payload_bytes": 1364, "side_information_bytes": 0, "fetched_bytes": 2511, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 89519.31475102901, "hops": {"1": 23133.45850002952, "2": 22054.83804200776, "3": 22289.791416958906, "4": 22041.22679203283}, "receiver_prefill_ms": 396.49095793720335, "ttft_ms": 33.64400006830692, "decode_ms": 32.96404203865677}, "total_wall_ms": 89983.21183410008, "ttft_ms": 33.64400006830692, "input_usage": {"lookups": 12}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 64, "facts_retained": 34, "retention_rate": 0.53125, "retention_by_kind": {"person": 0.5, "date": 0.3, "amount": 0.6666666666666666, "status": 0.9166666666666666, "place": 0.16666666666666666, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1364}, "ledger_by_hop": {"1": {"facts_total": 64, "facts_retained": 30, "retention_rate": 0.46875, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.6666666666666666, "status": 0.9166666666666666, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1087}, "2": {"facts_total": 64, "facts_retained": 30, "retention_rate": 0.46875, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.6666666666666666, "status": 0.9166666666666666, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1075}, "3": {"facts_total": 64, "facts_retained": 34, "retention_rate": 0.53125, "retention_by_kind": {"person": 0.5, "date": 0.3, "amount": 0.6666666666666666, "status": 0.9166666666666666, "place": 0.16666666666666666, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1364}, "4": {"facts_total": 64, "facts_retained": 34, "retention_rate": 0.53125, "retention_by_kind": {"person": 0.5, "date": 0.3, "amount": 0.6666666666666666, "status": 0.9166666666666666, "place": 0.16666666666666666, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1364}}}, "reference_key_id": "keys/synthetic/syn_03006_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "05248504387de244bb26ede8192b5c4b34d448e31f7043a95c04d8a18c00902a", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "e999fefb78c9449f228a509b", "attempt_id": "62bd2ed976b1", "run_group_id": "E4_core:A4|repairable|qwen3b:h4", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:02:42.905543Z", "task_family": "synthetic", "document_id": "syn_03006_2000", "document_hash": "89be5e77e0ca114963e9ffdf8dc38fe5610f13a78ac29e7a1956cd9124ce5217", "question_id": "q05_multi_hop", "split": "test", "cluster_id": "syn_03006_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 4, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1890, "query_tokens": 606, "message_tokens": 512, "message_utf8_bytes": 1364, "cache_tensor_bytes": null, "serialized_payload_bytes": 1364, "side_information_bytes": 0, "fetched_bytes": 2511, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 89519.31475102901, "hops": {"1": 23133.45850002952, "2": 22054.83804200776, "3": 22289.791416958906, "4": 22041.22679203283}, "receiver_prefill_ms": 405.9674160089344, "ttft_ms": 20.24600002914667, "decode_ms": 636.407500016503}, "total_wall_ms": 90582.77825103141, "ttft_ms": 20.24600002914667, "input_usage": {"lookups": 12}, "output_usage": {"output_tokens": 10}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "EP-52G-1330", "evidence_artifact": {"ledger": {"facts_total": 64, "facts_retained": 34, "retention_rate": 0.53125, "retention_by_kind": {"person": 0.5, "date": 0.3, "amount": 0.6666666666666666, "status": 0.9166666666666666, "place": 0.16666666666666666, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1364}, "ledger_by_hop": {"1": {"facts_total": 64, "facts_retained": 30, "retention_rate": 0.46875, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.6666666666666666, "status": 0.9166666666666666, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1087}, "2": {"facts_total": 64, "facts_retained": 30, "retention_rate": 0.46875, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.6666666666666666, "status": 0.9166666666666666, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1075}, "3": {"facts_total": 64, "facts_retained": 34, "retention_rate": 0.53125, "retention_by_kind": {"person": 0.5, "date": 0.3, "amount": 0.6666666666666666, "status": 0.9166666666666666, "place": 0.16666666666666666, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1364}, "4": {"facts_total": 64, "facts_retained": 34, "retention_rate": 0.53125, "retention_by_kind": {"person": 0.5, "date": 0.3, "amount": 0.6666666666666666, "status": 0.9166666666666666, "place": 0.16666666666666666, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1364}}}, "reference_key_id": "keys/synthetic/syn_03006_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "05248504387de244bb26ede8192b5c4b34d448e31f7043a95c04d8a18c00902a", "output_hash": "104db207937fd79e23ce16f9b3f10b1a9ed34f6dd0441a99be6ea1b2d0a09d99"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "496f11f152e6474d10417121", "attempt_id": "534053e26441", "run_group_id": "E4_core:A4|repairable|qwen3b:h4", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:02:43.987260Z", "task_family": "synthetic", "document_id": "syn_03006_2000", "document_hash": "89be5e77e0ca114963e9ffdf8dc38fe5610f13a78ac29e7a1956cd9124ce5217", "question_id": "q06_negation", "split": "test", "cluster_id": "syn_03006_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 4, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1890, "query_tokens": 594, "message_tokens": 512, "message_utf8_bytes": 1364, "cache_tensor_bytes": null, "serialized_payload_bytes": 1364, "side_information_bytes": 0, "fetched_bytes": 2511, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 89519.31475102901, "hops": {"1": 23133.45850002952, "2": 22054.83804200776, "3": 22289.791416958906, "4": 22041.22679203283}, "receiver_prefill_ms": 391.614833031781, "ttft_ms": 31.96304093580693, "decode_ms": 39.5315409405157}, "total_wall_ms": 89983.2210430177, "ttft_ms": 31.96304093580693, "input_usage": {"lookups": 12}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 64, "facts_retained": 34, "retention_rate": 0.53125, "retention_by_kind": {"person": 0.5, "date": 0.3, "amount": 0.6666666666666666, "status": 0.9166666666666666, "place": 0.16666666666666666, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1364}, "ledger_by_hop": {"1": {"facts_total": 64, "facts_retained": 30, "retention_rate": 0.46875, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.6666666666666666, "status": 0.9166666666666666, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1087}, "2": {"facts_total": 64, "facts_retained": 30, "retention_rate": 0.46875, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.6666666666666666, "status": 0.9166666666666666, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1075}, "3": {"facts_total": 64, "facts_retained": 34, "retention_rate": 0.53125, "retention_by_kind": {"person": 0.5, "date": 0.3, "amount": 0.6666666666666666, "status": 0.9166666666666666, "place": 0.16666666666666666, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1364}, "4": {"facts_total": 64, "facts_retained": 34, "retention_rate": 0.53125, "retention_by_kind": {"person": 0.5, "date": 0.3, "amount": 0.6666666666666666, "status": 0.9166666666666666, "place": 0.16666666666666666, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1364}}}, "reference_key_id": "keys/synthetic/syn_03006_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "05248504387de244bb26ede8192b5c4b34d448e31f7043a95c04d8a18c00902a", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "03b13f8d1a342304ae713556", "attempt_id": "ac87e99d296c", "run_group_id": "E4_core:A4|repairable|qwen3b:h4", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:02:44.468507Z", "task_family": "synthetic", "document_id": "syn_03006_2000", "document_hash": "89be5e77e0ca114963e9ffdf8dc38fe5610f13a78ac29e7a1956cd9124ce5217", "question_id": "q07_negation", "split": "test", "cluster_id": "syn_03006_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 4, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1890, "query_tokens": 594, "message_tokens": 512, "message_utf8_bytes": 1364, "cache_tensor_bytes": null, "serialized_payload_bytes": 1364, "side_information_bytes": 0, "fetched_bytes": 2511, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 89519.31475102901, "hops": {"1": 23133.45850002952, "2": 22054.83804200776, "3": 22289.791416958906, "4": 22041.22679203283}, "receiver_prefill_ms": 411.6350410040468, "ttft_ms": 27.360084000974894, "decode_ms": 20.28529194649309}, "total_wall_ms": 89979.39908504486, "ttft_ms": 27.360084000974894, "input_usage": {"lookups": 12}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "no", "evidence_artifact": {"ledger": {"facts_total": 64, "facts_retained": 34, "retention_rate": 0.53125, "retention_by_kind": {"person": 0.5, "date": 0.3, "amount": 0.6666666666666666, "status": 0.9166666666666666, "place": 0.16666666666666666, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1364}, "ledger_by_hop": {"1": {"facts_total": 64, "facts_retained": 30, "retention_rate": 0.46875, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.6666666666666666, "status": 0.9166666666666666, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1087}, "2": {"facts_total": 64, "facts_retained": 30, "retention_rate": 0.46875, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.6666666666666666, "status": 0.9166666666666666, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1075}, "3": {"facts_total": 64, "facts_retained": 34, "retention_rate": 0.53125, "retention_by_kind": {"person": 0.5, "date": 0.3, "amount": 0.6666666666666666, "status": 0.9166666666666666, "place": 0.16666666666666666, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1364}, "4": {"facts_total": 64, "facts_retained": 34, "retention_rate": 0.53125, "retention_by_kind": {"person": 0.5, "date": 0.3, "amount": 0.6666666666666666, "status": 0.9166666666666666, "place": 0.16666666666666666, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1364}}}, "reference_key_id": "keys/synthetic/syn_03006_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "05248504387de244bb26ede8192b5c4b34d448e31f7043a95c04d8a18c00902a", "output_hash": "9390298f3fb0c5b160498935d79cb139aef28e1c47358b4bbba61862b9c26e59"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "a3fc64c772b33687b66f698e", "attempt_id": "9e1494f36266", "run_group_id": "E4_core:A4|repairable|qwen3b:h4", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:02:44.946410Z", "task_family": "synthetic", "document_id": "syn_03006_2000", "document_hash": "89be5e77e0ca114963e9ffdf8dc38fe5610f13a78ac29e7a1956cd9124ce5217", "question_id": "q08_revision", "split": "test", "cluster_id": "syn_03006_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 4, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1890, "query_tokens": 611, "message_tokens": 512, "message_utf8_bytes": 1364, "cache_tensor_bytes": null, "serialized_payload_bytes": 1364, "side_information_bytes": 0, "fetched_bytes": 2511, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 89519.31475102901, "hops": {"1": 23133.45850002952, "2": 22054.83804200776, "3": 22289.791416958906, "4": 22041.22679203283}, "receiver_prefill_ms": 384.2795839300379, "ttft_ms": 37.90375008247793, "decode_ms": 470.7899580243975}, "total_wall_ms": 90413.06337609421, "ttft_ms": 37.90375008247793, "input_usage": {"lookups": 12}, "output_usage": {"output_tokens": 9}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "9,112,735", "evidence_artifact": {"ledger": {"facts_total": 64, "facts_retained": 34, "retention_rate": 0.53125, "retention_by_kind": {"person": 0.5, "date": 0.3, "amount": 0.6666666666666666, "status": 0.9166666666666666, "place": 0.16666666666666666, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1364}, "ledger_by_hop": {"1": {"facts_total": 64, "facts_retained": 30, "retention_rate": 0.46875, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.6666666666666666, "status": 0.9166666666666666, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1087}, "2": {"facts_total": 64, "facts_retained": 30, "retention_rate": 0.46875, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.6666666666666666, "status": 0.9166666666666666, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1075}, "3": {"facts_total": 64, "facts_retained": 34, "retention_rate": 0.53125, "retention_by_kind": {"person": 0.5, "date": 0.3, "amount": 0.6666666666666666, "status": 0.9166666666666666, "place": 0.16666666666666666, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1364}, "4": {"facts_total": 64, "facts_retained": 34, "retention_rate": 0.53125, "retention_by_kind": {"person": 0.5, "date": 0.3, "amount": 0.6666666666666666, "status": 0.9166666666666666, "place": 0.16666666666666666, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1364}}}, "reference_key_id": "keys/synthetic/syn_03006_2000.json", "task_scores": {"correct": 0.0, "abstained": 0.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "05248504387de244bb26ede8192b5c4b34d448e31f7043a95c04d8a18c00902a", "output_hash": "788fcda908c2029597420995b09d0cc919bfdbf7a28b68dcb0e331db7df7b342"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "f7b50fa5dbbbc002713f8b89", "attempt_id": "cf20845a5e81", "run_group_id": "E4_core:A4|repairable|qwen3b:h4", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:02:45.859778Z", "task_family": "synthetic", "document_id": "syn_03006_2000", "document_hash": "89be5e77e0ca114963e9ffdf8dc38fe5610f13a78ac29e7a1956cd9124ce5217", "question_id": "q09_revision", "split": "test", "cluster_id": "syn_03006_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 4, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1890, "query_tokens": 604, "message_tokens": 512, "message_utf8_bytes": 1364, "cache_tensor_bytes": null, "serialized_payload_bytes": 1364, "side_information_bytes": 0, "fetched_bytes": 2511, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 89519.31475102901, "hops": {"1": 23133.45850002952, "2": 22054.83804200776, "3": 22289.791416958906, "4": 22041.22679203283}, "receiver_prefill_ms": 400.78479202929884, "ttft_ms": 42.19029098749161, "decode_ms": 219.19058298226446}, "total_wall_ms": 90182.12616804522, "ttft_ms": 42.19029098749161, "input_usage": {"lookups": 12}, "output_usage": {"output_tokens": 7}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "9112735", "evidence_artifact": {"ledger": {"facts_total": 64, "facts_retained": 34, "retention_rate": 0.53125, "retention_by_kind": {"person": 0.5, "date": 0.3, "amount": 0.6666666666666666, "status": 0.9166666666666666, "place": 0.16666666666666666, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1364}, "ledger_by_hop": {"1": {"facts_total": 64, "facts_retained": 30, "retention_rate": 0.46875, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.6666666666666666, "status": 0.9166666666666666, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1087}, "2": {"facts_total": 64, "facts_retained": 30, "retention_rate": 0.46875, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.6666666666666666, "status": 0.9166666666666666, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1075}, "3": {"facts_total": 64, "facts_retained": 34, "retention_rate": 0.53125, "retention_by_kind": {"person": 0.5, "date": 0.3, "amount": 0.6666666666666666, "status": 0.9166666666666666, "place": 0.16666666666666666, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1364}, "4": {"facts_total": 64, "facts_retained": 34, "retention_rate": 0.53125, "retention_by_kind": {"person": 0.5, "date": 0.3, "amount": 0.6666666666666666, "status": 0.9166666666666666, "place": 0.16666666666666666, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1364}}}, "reference_key_id": "keys/synthetic/syn_03006_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "05248504387de244bb26ede8192b5c4b34d448e31f7043a95c04d8a18c00902a", "output_hash": "15a3f804636f0e62d4a3d9be7677f001a0ad6b1fb8ae7e8cdc63c44ec7bac31f"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "1139c7d262a00fbdb77069ea", "attempt_id": "e1e3c1d7a496", "run_group_id": "E4_core:A4|repairable|qwen3b:h4", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:02:46.540613Z", "task_family": "synthetic", "document_id": "syn_03006_2000", "document_hash": "89be5e77e0ca114963e9ffdf8dc38fe5610f13a78ac29e7a1956cd9124ce5217", "question_id": "q10_temporal", "split": "test", "cluster_id": "syn_03006_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 4, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1890, "query_tokens": 613, "message_tokens": 512, "message_utf8_bytes": 1364, "cache_tensor_bytes": null, "serialized_payload_bytes": 1364, "side_information_bytes": 0, "fetched_bytes": 2511, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 89519.31475102901, "hops": {"1": 23133.45850002952, "2": 22054.83804200776, "3": 22289.791416958906, "4": 22041.22679203283}, "receiver_prefill_ms": 426.16791697219014, "ttft_ms": 57.388334069401026, "decode_ms": 100.02108302433044}, "total_wall_ms": 90103.6463760538, "ttft_ms": 57.388334069401026, "input_usage": {"lookups": 12}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 64, "facts_retained": 34, "retention_rate": 0.53125, "retention_by_kind": {"person": 0.5, "date": 0.3, "amount": 0.6666666666666666, "status": 0.9166666666666666, "place": 0.16666666666666666, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1364}, "ledger_by_hop": {"1": {"facts_total": 64, "facts_retained": 30, "retention_rate": 0.46875, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.6666666666666666, "status": 0.9166666666666666, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1087}, "2": {"facts_total": 64, "facts_retained": 30, "retention_rate": 0.46875, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.6666666666666666, "status": 0.9166666666666666, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1075}, "3": {"facts_total": 64, "facts_retained": 34, "retention_rate": 0.53125, "retention_by_kind": {"person": 0.5, "date": 0.3, "amount": 0.6666666666666666, "status": 0.9166666666666666, "place": 0.16666666666666666, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1364}, "4": {"facts_total": 64, "facts_retained": 34, "retention_rate": 0.53125, "retention_by_kind": {"person": 0.5, "date": 0.3, "amount": 0.6666666666666666, "status": 0.9166666666666666, "place": 0.16666666666666666, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1364}}}, "reference_key_id": "keys/synthetic/syn_03006_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "05248504387de244bb26ede8192b5c4b34d448e31f7043a95c04d8a18c00902a", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "97274cb0395758224a384132", "attempt_id": "36b2fee130c4", "run_group_id": "E4_core:A4|repairable|qwen3b:h4", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:02:47.143856Z", "task_family": "synthetic", "document_id": "syn_03006_2000", "document_hash": "89be5e77e0ca114963e9ffdf8dc38fe5610f13a78ac29e7a1956cd9124ce5217", "question_id": "q11_aggregation", "split": "test", "cluster_id": "syn_03006_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 4, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1890, "query_tokens": 606, "message_tokens": 512, "message_utf8_bytes": 1364, "cache_tensor_bytes": null, "serialized_payload_bytes": 1364, "side_information_bytes": 0, "fetched_bytes": 2511, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 89519.31475102901, "hops": {"1": 23133.45850002952, "2": 22054.83804200776, "3": 22289.791416958906, "4": 22041.22679203283}, "receiver_prefill_ms": 470.80337500665337, "ttft_ms": 36.463917000219226, "decode_ms": 232.00695891864598}, "total_wall_ms": 90259.35754308011, "ttft_ms": 36.463917000219226, "input_usage": {"lookups": 12}, "output_usage": {"output_tokens": 7}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "8908553", "evidence_artifact": {"ledger": {"facts_total": 64, "facts_retained": 34, "retention_rate": 0.53125, "retention_by_kind": {"person": 0.5, "date": 0.3, "amount": 0.6666666666666666, "status": 0.9166666666666666, "place": 0.16666666666666666, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1364}, "ledger_by_hop": {"1": {"facts_total": 64, "facts_retained": 30, "retention_rate": 0.46875, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.6666666666666666, "status": 0.9166666666666666, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1087}, "2": {"facts_total": 64, "facts_retained": 30, "retention_rate": 0.46875, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.6666666666666666, "status": 0.9166666666666666, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1075}, "3": {"facts_total": 64, "facts_retained": 34, "retention_rate": 0.53125, "retention_by_kind": {"person": 0.5, "date": 0.3, "amount": 0.6666666666666666, "status": 0.9166666666666666, "place": 0.16666666666666666, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1364}, "4": {"facts_total": 64, "facts_retained": 34, "retention_rate": 0.53125, "retention_by_kind": {"person": 0.5, "date": 0.3, "amount": 0.6666666666666666, "status": 0.9166666666666666, "place": 0.16666666666666666, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1364}}}, "reference_key_id": "keys/synthetic/syn_03006_2000.json", "task_scores": {"correct": 0.0, "abstained": 0.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "05248504387de244bb26ede8192b5c4b34d448e31f7043a95c04d8a18c00902a", "output_hash": "5ad9e777dc3be36808c74badfc8aae217b41a486754d07adf1abda6f6bce9aca"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "4f52fd57614ecdebf4c0aab9", "attempt_id": "86e80bcea656", "run_group_id": "E4_core:A4|repairable|qwen3b:h4", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:02:47.903632Z", "task_family": "synthetic", "document_id": "syn_03006_2000", "document_hash": "89be5e77e0ca114963e9ffdf8dc38fe5610f13a78ac29e7a1956cd9124ce5217", "question_id": "q12_unit_conversion", "split": "test", "cluster_id": "syn_03006_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 4, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1890, "query_tokens": 603, "message_tokens": 512, "message_utf8_bytes": 1364, "cache_tensor_bytes": null, "serialized_payload_bytes": 1364, "side_information_bytes": 0, "fetched_bytes": 2511, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 89519.31475102901, "hops": {"1": 23133.45850002952, "2": 22054.83804200776, "3": 22289.791416958906, "4": 22041.22679203283}, "receiver_prefill_ms": 373.11666703317314, "ttft_ms": 34.50054198037833, "decode_ms": 274.899541051127}, "total_wall_ms": 90202.51150103286, "ttft_ms": 34.50054198037833, "input_usage": {"lookups": 12}, "output_usage": {"output_tokens": 5}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "39700", "evidence_artifact": {"ledger": {"facts_total": 64, "facts_retained": 34, "retention_rate": 0.53125, "retention_by_kind": {"person": 0.5, "date": 0.3, "amount": 0.6666666666666666, "status": 0.9166666666666666, "place": 0.16666666666666666, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1364}, "ledger_by_hop": {"1": {"facts_total": 64, "facts_retained": 30, "retention_rate": 0.46875, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.6666666666666666, "status": 0.9166666666666666, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1087}, "2": {"facts_total": 64, "facts_retained": 30, "retention_rate": 0.46875, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.6666666666666666, "status": 0.9166666666666666, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1075}, "3": {"facts_total": 64, "facts_retained": 34, "retention_rate": 0.53125, "retention_by_kind": {"person": 0.5, "date": 0.3, "amount": 0.6666666666666666, "status": 0.9166666666666666, "place": 0.16666666666666666, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1364}, "4": {"facts_total": 64, "facts_retained": 34, "retention_rate": 0.53125, "retention_by_kind": {"person": 0.5, "date": 0.3, "amount": 0.6666666666666666, "status": 0.9166666666666666, "place": 0.16666666666666666, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1364}}}, "reference_key_id": "keys/synthetic/syn_03006_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "05248504387de244bb26ede8192b5c4b34d448e31f7043a95c04d8a18c00902a", "output_hash": "d8f3f4a352330800cb9f3d838e47d965e9c710d166f53c6b3724fdeae73c9a5e"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "3a30895150f3c49323456ede", "attempt_id": "66bc8b2ad5f4", "run_group_id": "E4_core:A4|repairable|qwen3b:h4", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:02:48.604277Z", "task_family": "synthetic", "document_id": "syn_03006_2000", "document_hash": "89be5e77e0ca114963e9ffdf8dc38fe5610f13a78ac29e7a1956cd9124ce5217", "question_id": "q13_unanswerable", "split": "test", "cluster_id": "syn_03006_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 4, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1890, "query_tokens": 607, "message_tokens": 512, "message_utf8_bytes": 1364, "cache_tensor_bytes": null, "serialized_payload_bytes": 1364, "side_information_bytes": 0, "fetched_bytes": 2511, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 89519.31475102901, "hops": {"1": 23133.45850002952, "2": 22054.83804200776, "3": 22289.791416958906, "4": 22041.22679203283}, "receiver_prefill_ms": 533.5235419915989, "ttft_ms": 25.73216601740569, "decode_ms": 39.58816605154425}, "total_wall_ms": 90118.92437597271, "ttft_ms": 25.73216601740569, "input_usage": {"lookups": 12}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 64, "facts_retained": 34, "retention_rate": 0.53125, "retention_by_kind": {"person": 0.5, "date": 0.3, "amount": 0.6666666666666666, "status": 0.9166666666666666, "place": 0.16666666666666666, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1364}, "ledger_by_hop": {"1": {"facts_total": 64, "facts_retained": 30, "retention_rate": 0.46875, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.6666666666666666, "status": 0.9166666666666666, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1087}, "2": {"facts_total": 64, "facts_retained": 30, "retention_rate": 0.46875, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.6666666666666666, "status": 0.9166666666666666, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1075}, "3": {"facts_total": 64, "facts_retained": 34, "retention_rate": 0.53125, "retention_by_kind": {"person": 0.5, "date": 0.3, "amount": 0.6666666666666666, "status": 0.9166666666666666, "place": 0.16666666666666666, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1364}, "4": {"facts_total": 64, "facts_retained": 34, "retention_rate": 0.53125, "retention_by_kind": {"person": 0.5, "date": 0.3, "amount": 0.6666666666666666, "status": 0.9166666666666666, "place": 0.16666666666666666, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1364}}}, "reference_key_id": "keys/synthetic/syn_03006_2000.json", "task_scores": {"correct": 1.0, "abstained": 1.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "05248504387de244bb26ede8192b5c4b34d448e31f7043a95c04d8a18c00902a", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "4a7cdd215445b78956576d4d", "attempt_id": "df24c5cfc5d3", "run_group_id": "E4_core:A4|repairable|qwen3b:h4", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:02:49.222298Z", "task_family": "synthetic", "document_id": "syn_03006_2000", "document_hash": "89be5e77e0ca114963e9ffdf8dc38fe5610f13a78ac29e7a1956cd9124ce5217", "question_id": "q14_unanswerable", "split": "test", "cluster_id": "syn_03006_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 4, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1890, "query_tokens": 608, "message_tokens": 512, "message_utf8_bytes": 1364, "cache_tensor_bytes": null, "serialized_payload_bytes": 1364, "side_information_bytes": 0, "fetched_bytes": 2511, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 89519.31475102901, "hops": {"1": 23133.45850002952, "2": 22054.83804200776, "3": 22289.791416958906, "4": 22041.22679203283}, "receiver_prefill_ms": 405.0638750195503, "ttft_ms": 27.308749966323376, "decode_ms": 37.125958944670856}, "total_wall_ms": 89989.41033403389, "ttft_ms": 27.308749966323376, "input_usage": {"lookups": 12}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 64, "facts_retained": 34, "retention_rate": 0.53125, "retention_by_kind": {"person": 0.5, "date": 0.3, "amount": 0.6666666666666666, "status": 0.9166666666666666, "place": 0.16666666666666666, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1364}, "ledger_by_hop": {"1": {"facts_total": 64, "facts_retained": 30, "retention_rate": 0.46875, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.6666666666666666, "status": 0.9166666666666666, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1087}, "2": {"facts_total": 64, "facts_retained": 30, "retention_rate": 0.46875, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.6666666666666666, "status": 0.9166666666666666, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1075}, "3": {"facts_total": 64, "facts_retained": 34, "retention_rate": 0.53125, "retention_by_kind": {"person": 0.5, "date": 0.3, "amount": 0.6666666666666666, "status": 0.9166666666666666, "place": 0.16666666666666666, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1364}, "4": {"facts_total": 64, "facts_retained": 34, "retention_rate": 0.53125, "retention_by_kind": {"person": 0.5, "date": 0.3, "amount": 0.6666666666666666, "status": 0.9166666666666666, "place": 0.16666666666666666, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1364}}}, "reference_key_id": "keys/synthetic/syn_03006_2000.json", "task_scores": {"correct": 1.0, "abstained": 1.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "05248504387de244bb26ede8192b5c4b34d448e31f7043a95c04d8a18c00902a", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "58606eff33a16d5f01060d59", "attempt_id": "73673114432b", "run_group_id": "E4_core:A4|repairable|qwen3b:h8", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:04:18.165806Z", "task_family": "synthetic", "document_id": "syn_03006_2000", "document_hash": "89be5e77e0ca114963e9ffdf8dc38fe5610f13a78ac29e7a1956cd9124ce5217", "question_id": "q00_single_fact", "split": "test", "cluster_id": "syn_03006_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 8, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1890, "query_tokens": 595, "message_tokens": 512, "message_utf8_bytes": 1364, "cache_tensor_bytes": null, "serialized_payload_bytes": 1364, "side_information_bytes": 0, "fetched_bytes": 5859, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 177969.8518749792, "hops": {"1": 23133.45850002952, "2": 22054.83804200776, "3": 22289.791416958906, "4": 22041.22679203283, "5": 22040.885791997425, "6": 21990.91641593259, "7": 22282.63995796442, "8": 22136.094958055764}, "receiver_prefill_ms": 401.3275830075145, "ttft_ms": 26.40845801215619, "decode_ms": 371.4980840450153}, "total_wall_ms": 178770.11658390984, "ttft_ms": 26.40845801215619, "input_usage": {"lookups": 28}, "output_usage": {"output_tokens": 10}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "XM-76Q-2958", "evidence_artifact": {"ledger": {"facts_total": 64, "facts_retained": 34, "retention_rate": 0.53125, "retention_by_kind": {"person": 0.5, "date": 0.3, "amount": 0.6666666666666666, "status": 0.9166666666666666, "place": 0.16666666666666666, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1364}, "ledger_by_hop": {"1": {"facts_total": 64, "facts_retained": 30, "retention_rate": 0.46875, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.6666666666666666, "status": 0.9166666666666666, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1087}, "2": {"facts_total": 64, "facts_retained": 30, "retention_rate": 0.46875, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.6666666666666666, "status": 0.9166666666666666, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1075}, "3": {"facts_total": 64, "facts_retained": 34, "retention_rate": 0.53125, "retention_by_kind": {"person": 0.5, "date": 0.3, "amount": 0.6666666666666666, "status": 0.9166666666666666, "place": 0.16666666666666666, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1364}, "4": {"facts_total": 64, "facts_retained": 34, "retention_rate": 0.53125, "retention_by_kind": {"person": 0.5, "date": 0.3, "amount": 0.6666666666666666, "status": 0.9166666666666666, "place": 0.16666666666666666, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1364}, "5": {"facts_total": 64, "facts_retained": 34, "retention_rate": 0.53125, "retention_by_kind": {"person": 0.5, "date": 0.3, "amount": 0.6666666666666666, "status": 0.9166666666666666, "place": 0.16666666666666666, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1364}, "6": {"facts_total": 64, "facts_retained": 34, "retention_rate": 0.53125, "retention_by_kind": {"person": 0.5, "date": 0.3, "amount": 0.6666666666666666, "status": 0.9166666666666666, "place": 0.16666666666666666, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1364}, "7": {"facts_total": 64, "facts_retained": 34, "retention_rate": 0.53125, "retention_by_kind": {"person": 0.5, "date": 0.3, "amount": 0.6666666666666666, "status": 0.9166666666666666, "place": 0.16666666666666666, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1364}, "8": {"facts_total": 64, "facts_retained": 34, "retention_rate": 0.53125, "retention_by_kind": {"person": 0.5, "date": 0.3, "amount": 0.6666666666666666, "status": 0.9166666666666666, "place": 0.16666666666666666, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1364}}}, "reference_key_id": "keys/synthetic/syn_03006_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "05248504387de244bb26ede8192b5c4b34d448e31f7043a95c04d8a18c00902a", "output_hash": "cf42691eec772b9ec74d67643944de2e19e53724d7b772cef1e0f2d5ef3b4926"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "1a3140c59126460eceb41c0b", "attempt_id": "46f093d6496b", "run_group_id": "E4_core:A4|repairable|qwen3b:h8", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:04:18.983983Z", "task_family": "synthetic", "document_id": "syn_03006_2000", "document_hash": "89be5e77e0ca114963e9ffdf8dc38fe5610f13a78ac29e7a1956cd9124ce5217", "question_id": "q01_single_fact", "split": "test", "cluster_id": "syn_03006_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 8, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1890, "query_tokens": 600, "message_tokens": 512, "message_utf8_bytes": 1364, "cache_tensor_bytes": null, "serialized_payload_bytes": 1364, "side_information_bytes": 0, "fetched_bytes": 5859, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 177969.8518749792, "hops": {"1": 23133.45850002952, "2": 22054.83804200776, "3": 22289.791416958906, "4": 22041.22679203283, "5": 22040.885791997425, "6": 21990.91641593259, "7": 22282.63995796442, "8": 22136.094958055764}, "receiver_prefill_ms": 387.6050409162417, "ttft_ms": 99.33941706549376, "decode_ms": 108.42758300714195}, "total_wall_ms": 178565.93516701832, "ttft_ms": 99.33941706549376, "input_usage": {"lookups": 28}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 64, "facts_retained": 34, "retention_rate": 0.53125, "retention_by_kind": {"person": 0.5, "date": 0.3, "amount": 0.6666666666666666, "status": 0.9166666666666666, "place": 0.16666666666666666, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1364}, "ledger_by_hop": {"1": {"facts_total": 64, "facts_retained": 30, "retention_rate": 0.46875, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.6666666666666666, "status": 0.9166666666666666, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1087}, "2": {"facts_total": 64, "facts_retained": 30, "retention_rate": 0.46875, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.6666666666666666, "status": 0.9166666666666666, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1075}, "3": {"facts_total": 64, "facts_retained": 34, "retention_rate": 0.53125, "retention_by_kind": {"person": 0.5, "date": 0.3, "amount": 0.6666666666666666, "status": 0.9166666666666666, "place": 0.16666666666666666, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1364}, "4": {"facts_total": 64, "facts_retained": 34, "retention_rate": 0.53125, "retention_by_kind": {"person": 0.5, "date": 0.3, "amount": 0.6666666666666666, "status": 0.9166666666666666, "place": 0.16666666666666666, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1364}, "5": {"facts_total": 64, "facts_retained": 34, "retention_rate": 0.53125, "retention_by_kind": {"person": 0.5, "date": 0.3, "amount": 0.6666666666666666, "status": 0.9166666666666666, "place": 0.16666666666666666, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1364}, "6": {"facts_total": 64, "facts_retained": 34, "retention_rate": 0.53125, "retention_by_kind": {"person": 0.5, "date": 0.3, "amount": 0.6666666666666666, "status": 0.9166666666666666, "place": 0.16666666666666666, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1364}, "7": {"facts_total": 64, "facts_retained": 34, "retention_rate": 0.53125, "retention_by_kind": {"person": 0.5, "date": 0.3, "amount": 0.6666666666666666, "status": 0.9166666666666666, "place": 0.16666666666666666, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1364}, "8": {"facts_total": 64, "facts_retained": 34, "retention_rate": 0.53125, "retention_by_kind": {"person": 0.5, "date": 0.3, "amount": 0.6666666666666666, "status": 0.9166666666666666, "place": 0.16666666666666666, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1364}}}, "reference_key_id": "keys/synthetic/syn_03006_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "05248504387de244bb26ede8192b5c4b34d448e31f7043a95c04d8a18c00902a", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "4bff825a3413830680e9e1d0", "attempt_id": "3929a00dfbc2", "run_group_id": "E4_core:A4|repairable|qwen3b:h8", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:04:19.597532Z", "task_family": "synthetic", "document_id": "syn_03006_2000", "document_hash": "89be5e77e0ca114963e9ffdf8dc38fe5610f13a78ac29e7a1956cd9124ce5217", "question_id": "q02_exact_copy", "split": "test", "cluster_id": "syn_03006_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 8, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1890, "query_tokens": 599, "message_tokens": 512, "message_utf8_bytes": 1364, "cache_tensor_bytes": null, "serialized_payload_bytes": 1364, "side_information_bytes": 0, "fetched_bytes": 5859, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 177969.8518749792, "hops": {"1": 23133.45850002952, "2": 22054.83804200776, "3": 22289.791416958906, "4": 22041.22679203283, "5": 22040.885791997425, "6": 21990.91641593259, "7": 22282.63995796442, "8": 22136.094958055764}, "receiver_prefill_ms": 401.3481249567121, "ttft_ms": 40.401041973382235, "decode_ms": 26.552500086836517}, "total_wall_ms": 178438.82108398248, "ttft_ms": 40.401041973382235, "input_usage": {"lookups": 28}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 64, "facts_retained": 34, "retention_rate": 0.53125, "retention_by_kind": {"person": 0.5, "date": 0.3, "amount": 0.6666666666666666, "status": 0.9166666666666666, "place": 0.16666666666666666, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1364}, "ledger_by_hop": {"1": {"facts_total": 64, "facts_retained": 30, "retention_rate": 0.46875, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.6666666666666666, "status": 0.9166666666666666, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1087}, "2": {"facts_total": 64, "facts_retained": 30, "retention_rate": 0.46875, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.6666666666666666, "status": 0.9166666666666666, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1075}, "3": {"facts_total": 64, "facts_retained": 34, "retention_rate": 0.53125, "retention_by_kind": {"person": 0.5, "date": 0.3, "amount": 0.6666666666666666, "status": 0.9166666666666666, "place": 0.16666666666666666, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1364}, "4": {"facts_total": 64, "facts_retained": 34, "retention_rate": 0.53125, "retention_by_kind": {"person": 0.5, "date": 0.3, "amount": 0.6666666666666666, "status": 0.9166666666666666, "place": 0.16666666666666666, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1364}, "5": {"facts_total": 64, "facts_retained": 34, "retention_rate": 0.53125, "retention_by_kind": {"person": 0.5, "date": 0.3, "amount": 0.6666666666666666, "status": 0.9166666666666666, "place": 0.16666666666666666, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1364}, "6": {"facts_total": 64, "facts_retained": 34, "retention_rate": 0.53125, "retention_by_kind": {"person": 0.5, "date": 0.3, "amount": 0.6666666666666666, "status": 0.9166666666666666, "place": 0.16666666666666666, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1364}, "7": {"facts_total": 64, "facts_retained": 34, "retention_rate": 0.53125, "retention_by_kind": {"person": 0.5, "date": 0.3, "amount": 0.6666666666666666, "status": 0.9166666666666666, "place": 0.16666666666666666, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1364}, "8": {"facts_total": 64, "facts_retained": 34, "retention_rate": 0.53125, "retention_by_kind": {"person": 0.5, "date": 0.3, "amount": 0.6666666666666666, "status": 0.9166666666666666, "place": 0.16666666666666666, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1364}}}, "reference_key_id": "keys/synthetic/syn_03006_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "05248504387de244bb26ede8192b5c4b34d448e31f7043a95c04d8a18c00902a", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "c84f871d1c8c196fd6e540ff", "attempt_id": "3ac4ca5e8f5d", "run_group_id": "E4_core:A4|repairable|qwen3b:h8", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:04:20.083555Z", "task_family": "synthetic", "document_id": "syn_03006_2000", "document_hash": "89be5e77e0ca114963e9ffdf8dc38fe5610f13a78ac29e7a1956cd9124ce5217", "question_id": "q03_exact_copy", "split": "test", "cluster_id": "syn_03006_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 8, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1890, "query_tokens": 594, "message_tokens": 512, "message_utf8_bytes": 1364, "cache_tensor_bytes": null, "serialized_payload_bytes": 1364, "side_information_bytes": 0, "fetched_bytes": 5859, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 177969.8518749792, "hops": {"1": 23133.45850002952, "2": 22054.83804200776, "3": 22289.791416958906, "4": 22041.22679203283, "5": 22040.885791997425, "6": 21990.91641593259, "7": 22282.63995796442, "8": 22136.094958055764}, "receiver_prefill_ms": 402.3055409779772, "ttft_ms": 25.334666948765516, "decode_ms": 290.3715829597786}, "total_wall_ms": 178688.54716606438, "ttft_ms": 25.334666948765516, "input_usage": {"lookups": 28}, "output_usage": {"output_tokens": 8}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "Haruki Ouellet-Marchand", "evidence_artifact": {"ledger": {"facts_total": 64, "facts_retained": 34, "retention_rate": 0.53125, "retention_by_kind": {"person": 0.5, "date": 0.3, "amount": 0.6666666666666666, "status": 0.9166666666666666, "place": 0.16666666666666666, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1364}, "ledger_by_hop": {"1": {"facts_total": 64, "facts_retained": 30, "retention_rate": 0.46875, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.6666666666666666, "status": 0.9166666666666666, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1087}, "2": {"facts_total": 64, "facts_retained": 30, "retention_rate": 0.46875, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.6666666666666666, "status": 0.9166666666666666, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1075}, "3": {"facts_total": 64, "facts_retained": 34, "retention_rate": 0.53125, "retention_by_kind": {"person": 0.5, "date": 0.3, "amount": 0.6666666666666666, "status": 0.9166666666666666, "place": 0.16666666666666666, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1364}, "4": {"facts_total": 64, "facts_retained": 34, "retention_rate": 0.53125, "retention_by_kind": {"person": 0.5, "date": 0.3, "amount": 0.6666666666666666, "status": 0.9166666666666666, "place": 0.16666666666666666, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1364}, "5": {"facts_total": 64, "facts_retained": 34, "retention_rate": 0.53125, "retention_by_kind": {"person": 0.5, "date": 0.3, "amount": 0.6666666666666666, "status": 0.9166666666666666, "place": 0.16666666666666666, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1364}, "6": {"facts_total": 64, "facts_retained": 34, "retention_rate": 0.53125, "retention_by_kind": {"person": 0.5, "date": 0.3, "amount": 0.6666666666666666, "status": 0.9166666666666666, "place": 0.16666666666666666, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1364}, "7": {"facts_total": 64, "facts_retained": 34, "retention_rate": 0.53125, "retention_by_kind": {"person": 0.5, "date": 0.3, "amount": 0.6666666666666666, "status": 0.9166666666666666, "place": 0.16666666666666666, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1364}, "8": {"facts_total": 64, "facts_retained": 34, "retention_rate": 0.53125, "retention_by_kind": {"person": 0.5, "date": 0.3, "amount": 0.6666666666666666, "status": 0.9166666666666666, "place": 0.16666666666666666, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1364}}}, "reference_key_id": "keys/synthetic/syn_03006_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "05248504387de244bb26ede8192b5c4b34d448e31f7043a95c04d8a18c00902a", "output_hash": "a3b6e0a44d495387d8b8fc50ba6be0f2ed43121c674d3b3e962f8c4690e15b5c"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "a0f1043199509533d23a2c1c", "attempt_id": "e705f25a262f", "run_group_id": "E4_core:A4|repairable|qwen3b:h8", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:04:20.820432Z", "task_family": "synthetic", "document_id": "syn_03006_2000", "document_hash": "89be5e77e0ca114963e9ffdf8dc38fe5610f13a78ac29e7a1956cd9124ce5217", "question_id": "q04_multi_hop", "split": "test", "cluster_id": "syn_03006_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 8, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1890, "query_tokens": 603, "message_tokens": 512, "message_utf8_bytes": 1364, "cache_tensor_bytes": null, "serialized_payload_bytes": 1364, "side_information_bytes": 0, "fetched_bytes": 5859, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 177969.8518749792, "hops": {"1": 23133.45850002952, "2": 22054.83804200776, "3": 22289.791416958906, "4": 22041.22679203283, "5": 22040.885791997425, "6": 21990.91641593259, "7": 22282.63995796442, "8": 22136.094958055764}, "receiver_prefill_ms": 532.2695000795648, "ttft_ms": 35.321207949891686, "decode_ms": 37.48454200103879}, "total_wall_ms": 178575.59499994386, "ttft_ms": 35.321207949891686, "input_usage": {"lookups": 28}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 64, "facts_retained": 34, "retention_rate": 0.53125, "retention_by_kind": {"person": 0.5, "date": 0.3, "amount": 0.6666666666666666, "status": 0.9166666666666666, "place": 0.16666666666666666, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1364}, "ledger_by_hop": {"1": {"facts_total": 64, "facts_retained": 30, "retention_rate": 0.46875, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.6666666666666666, "status": 0.9166666666666666, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1087}, "2": {"facts_total": 64, "facts_retained": 30, "retention_rate": 0.46875, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.6666666666666666, "status": 0.9166666666666666, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1075}, "3": {"facts_total": 64, "facts_retained": 34, "retention_rate": 0.53125, "retention_by_kind": {"person": 0.5, "date": 0.3, "amount": 0.6666666666666666, "status": 0.9166666666666666, "place": 0.16666666666666666, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1364}, "4": {"facts_total": 64, "facts_retained": 34, "retention_rate": 0.53125, "retention_by_kind": {"person": 0.5, "date": 0.3, "amount": 0.6666666666666666, "status": 0.9166666666666666, "place": 0.16666666666666666, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1364}, "5": {"facts_total": 64, "facts_retained": 34, "retention_rate": 0.53125, "retention_by_kind": {"person": 0.5, "date": 0.3, "amount": 0.6666666666666666, "status": 0.9166666666666666, "place": 0.16666666666666666, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1364}, "6": {"facts_total": 64, "facts_retained": 34, "retention_rate": 0.53125, "retention_by_kind": {"person": 0.5, "date": 0.3, "amount": 0.6666666666666666, "status": 0.9166666666666666, "place": 0.16666666666666666, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1364}, "7": {"facts_total": 64, "facts_retained": 34, "retention_rate": 0.53125, "retention_by_kind": {"person": 0.5, "date": 0.3, "amount": 0.6666666666666666, "status": 0.9166666666666666, "place": 0.16666666666666666, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1364}, "8": {"facts_total": 64, "facts_retained": 34, "retention_rate": 0.53125, "retention_by_kind": {"person": 0.5, "date": 0.3, "amount": 0.6666666666666666, "status": 0.9166666666666666, "place": 0.16666666666666666, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1364}}}, "reference_key_id": "keys/synthetic/syn_03006_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "05248504387de244bb26ede8192b5c4b34d448e31f7043a95c04d8a18c00902a", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "7255af2cd648e7a1217c7df6", "attempt_id": "ec4bb915fd84", "run_group_id": "E4_core:A4|repairable|qwen3b:h8", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:04:21.443947Z", "task_family": "synthetic", "document_id": "syn_03006_2000", "document_hash": "89be5e77e0ca114963e9ffdf8dc38fe5610f13a78ac29e7a1956cd9124ce5217", "question_id": "q05_multi_hop", "split": "test", "cluster_id": "syn_03006_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 8, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1890, "query_tokens": 606, "message_tokens": 512, "message_utf8_bytes": 1364, "cache_tensor_bytes": null, "serialized_payload_bytes": 1364, "side_information_bytes": 0, "fetched_bytes": 5859, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 177969.8518749792, "hops": {"1": 23133.45850002952, "2": 22054.83804200776, "3": 22289.791416958906, "4": 22041.22679203283, "5": 22040.885791997425, "6": 21990.91641593259, "7": 22282.63995796442, "8": 22136.094958055764}, "receiver_prefill_ms": 402.603792026639, "ttft_ms": 21.58912504091859, "decode_ms": 322.41529203020036}, "total_wall_ms": 178717.3965839902, "ttft_ms": 21.58912504091859, "input_usage": {"lookups": 28}, "output_usage": {"output_tokens": 10}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "EP-52G-1330", "evidence_artifact": {"ledger": {"facts_total": 64, "facts_retained": 34, "retention_rate": 0.53125, "retention_by_kind": {"person": 0.5, "date": 0.3, "amount": 0.6666666666666666, "status": 0.9166666666666666, "place": 0.16666666666666666, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1364}, "ledger_by_hop": {"1": {"facts_total": 64, "facts_retained": 30, "retention_rate": 0.46875, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.6666666666666666, "status": 0.9166666666666666, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1087}, "2": {"facts_total": 64, "facts_retained": 30, "retention_rate": 0.46875, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.6666666666666666, "status": 0.9166666666666666, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1075}, "3": {"facts_total": 64, "facts_retained": 34, "retention_rate": 0.53125, "retention_by_kind": {"person": 0.5, "date": 0.3, "amount": 0.6666666666666666, "status": 0.9166666666666666, "place": 0.16666666666666666, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1364}, "4": {"facts_total": 64, "facts_retained": 34, "retention_rate": 0.53125, "retention_by_kind": {"person": 0.5, "date": 0.3, "amount": 0.6666666666666666, "status": 0.9166666666666666, "place": 0.16666666666666666, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1364}, "5": {"facts_total": 64, "facts_retained": 34, "retention_rate": 0.53125, "retention_by_kind": {"person": 0.5, "date": 0.3, "amount": 0.6666666666666666, "status": 0.9166666666666666, "place": 0.16666666666666666, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1364}, "6": {"facts_total": 64, "facts_retained": 34, "retention_rate": 0.53125, "retention_by_kind": {"person": 0.5, "date": 0.3, "amount": 0.6666666666666666, "status": 0.9166666666666666, "place": 0.16666666666666666, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1364}, "7": {"facts_total": 64, "facts_retained": 34, "retention_rate": 0.53125, "retention_by_kind": {"person": 0.5, "date": 0.3, "amount": 0.6666666666666666, "status": 0.9166666666666666, "place": 0.16666666666666666, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1364}, "8": {"facts_total": 64, "facts_retained": 34, "retention_rate": 0.53125, "retention_by_kind": {"person": 0.5, "date": 0.3, "amount": 0.6666666666666666, "status": 0.9166666666666666, "place": 0.16666666666666666, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1364}}}, "reference_key_id": "keys/synthetic/syn_03006_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "05248504387de244bb26ede8192b5c4b34d448e31f7043a95c04d8a18c00902a", "output_hash": "104db207937fd79e23ce16f9b3f10b1a9ed34f6dd0441a99be6ea1b2d0a09d99"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "385a52e11a6daf74a81cd665", "attempt_id": "ecb15d85e7a7", "run_group_id": "E4_core:A4|repairable|qwen3b:h8", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:04:22.209234Z", "task_family": "synthetic", "document_id": "syn_03006_2000", "document_hash": "89be5e77e0ca114963e9ffdf8dc38fe5610f13a78ac29e7a1956cd9124ce5217", "question_id": "q06_negation", "split": "test", "cluster_id": "syn_03006_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 8, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1890, "query_tokens": 594, "message_tokens": 512, "message_utf8_bytes": 1364, "cache_tensor_bytes": null, "serialized_payload_bytes": 1364, "side_information_bytes": 0, "fetched_bytes": 5859, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 177969.8518749792, "hops": {"1": 23133.45850002952, "2": 22054.83804200776, "3": 22289.791416958906, "4": 22041.22679203283, "5": 22040.885791997425, "6": 21990.91641593259, "7": 22282.63995796442, "8": 22136.094958055764}, "receiver_prefill_ms": 416.1559999920428, "ttft_ms": 41.79579205811024, "decode_ms": 90.88166593573987}, "total_wall_ms": 178519.37983289827, "ttft_ms": 41.79579205811024, "input_usage": {"lookups": 28}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 64, "facts_retained": 34, "retention_rate": 0.53125, "retention_by_kind": {"person": 0.5, "date": 0.3, "amount": 0.6666666666666666, "status": 0.9166666666666666, "place": 0.16666666666666666, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1364}, "ledger_by_hop": {"1": {"facts_total": 64, "facts_retained": 30, "retention_rate": 0.46875, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.6666666666666666, "status": 0.9166666666666666, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1087}, "2": {"facts_total": 64, "facts_retained": 30, "retention_rate": 0.46875, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.6666666666666666, "status": 0.9166666666666666, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1075}, "3": {"facts_total": 64, "facts_retained": 34, "retention_rate": 0.53125, "retention_by_kind": {"person": 0.5, "date": 0.3, "amount": 0.6666666666666666, "status": 0.9166666666666666, "place": 0.16666666666666666, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1364}, "4": {"facts_total": 64, "facts_retained": 34, "retention_rate": 0.53125, "retention_by_kind": {"person": 0.5, "date": 0.3, "amount": 0.6666666666666666, "status": 0.9166666666666666, "place": 0.16666666666666666, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1364}, "5": {"facts_total": 64, "facts_retained": 34, "retention_rate": 0.53125, "retention_by_kind": {"person": 0.5, "date": 0.3, "amount": 0.6666666666666666, "status": 0.9166666666666666, "place": 0.16666666666666666, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1364}, "6": {"facts_total": 64, "facts_retained": 34, "retention_rate": 0.53125, "retention_by_kind": {"person": 0.5, "date": 0.3, "amount": 0.6666666666666666, "status": 0.9166666666666666, "place": 0.16666666666666666, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1364}, "7": {"facts_total": 64, "facts_retained": 34, "retention_rate": 0.53125, "retention_by_kind": {"person": 0.5, "date": 0.3, "amount": 0.6666666666666666, "status": 0.9166666666666666, "place": 0.16666666666666666, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1364}, "8": {"facts_total": 64, "facts_retained": 34, "retention_rate": 0.53125, "retention_by_kind": {"person": 0.5, "date": 0.3, "amount": 0.6666666666666666, "status": 0.9166666666666666, "place": 0.16666666666666666, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1364}}}, "reference_key_id": "keys/synthetic/syn_03006_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "05248504387de244bb26ede8192b5c4b34d448e31f7043a95c04d8a18c00902a", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "6273070f37bba22d6efbabdf", "attempt_id": "f9f26861b7ec", "run_group_id": "E4_core:A4|repairable|qwen3b:h8", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:04:22.777580Z", "task_family": "synthetic", "document_id": "syn_03006_2000", "document_hash": "89be5e77e0ca114963e9ffdf8dc38fe5610f13a78ac29e7a1956cd9124ce5217", "question_id": "q07_negation", "split": "test", "cluster_id": "syn_03006_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 8, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1890, "query_tokens": 594, "message_tokens": 512, "message_utf8_bytes": 1364, "cache_tensor_bytes": null, "serialized_payload_bytes": 1364, "side_information_bytes": 0, "fetched_bytes": 5859, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 177969.8518749792, "hops": {"1": 23133.45850002952, "2": 22054.83804200776, "3": 22289.791416958906, "4": 22041.22679203283, "5": 22040.885791997425, "6": 21990.91641593259, "7": 22282.63995796442, "8": 22136.094958055764}, "receiver_prefill_ms": 485.9244169201702, "ttft_ms": 24.031250039115548, "decode_ms": 41.75679199397564}, "total_wall_ms": 178522.23562507425, "ttft_ms": 24.031250039115548, "input_usage": {"lookups": 28}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "no", "evidence_artifact": {"ledger": {"facts_total": 64, "facts_retained": 34, "retention_rate": 0.53125, "retention_by_kind": {"person": 0.5, "date": 0.3, "amount": 0.6666666666666666, "status": 0.9166666666666666, "place": 0.16666666666666666, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1364}, "ledger_by_hop": {"1": {"facts_total": 64, "facts_retained": 30, "retention_rate": 0.46875, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.6666666666666666, "status": 0.9166666666666666, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1087}, "2": {"facts_total": 64, "facts_retained": 30, "retention_rate": 0.46875, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.6666666666666666, "status": 0.9166666666666666, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1075}, "3": {"facts_total": 64, "facts_retained": 34, "retention_rate": 0.53125, "retention_by_kind": {"person": 0.5, "date": 0.3, "amount": 0.6666666666666666, "status": 0.9166666666666666, "place": 0.16666666666666666, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1364}, "4": {"facts_total": 64, "facts_retained": 34, "retention_rate": 0.53125, "retention_by_kind": {"person": 0.5, "date": 0.3, "amount": 0.6666666666666666, "status": 0.9166666666666666, "place": 0.16666666666666666, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1364}, "5": {"facts_total": 64, "facts_retained": 34, "retention_rate": 0.53125, "retention_by_kind": {"person": 0.5, "date": 0.3, "amount": 0.6666666666666666, "status": 0.9166666666666666, "place": 0.16666666666666666, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1364}, "6": {"facts_total": 64, "facts_retained": 34, "retention_rate": 0.53125, "retention_by_kind": {"person": 0.5, "date": 0.3, "amount": 0.6666666666666666, "status": 0.9166666666666666, "place": 0.16666666666666666, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1364}, "7": {"facts_total": 64, "facts_retained": 34, "retention_rate": 0.53125, "retention_by_kind": {"person": 0.5, "date": 0.3, "amount": 0.6666666666666666, "status": 0.9166666666666666, "place": 0.16666666666666666, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1364}, "8": {"facts_total": 64, "facts_retained": 34, "retention_rate": 0.53125, "retention_by_kind": {"person": 0.5, "date": 0.3, "amount": 0.6666666666666666, "status": 0.9166666666666666, "place": 0.16666666666666666, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1364}}}, "reference_key_id": "keys/synthetic/syn_03006_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "05248504387de244bb26ede8192b5c4b34d448e31f7043a95c04d8a18c00902a", "output_hash": "9390298f3fb0c5b160498935d79cb139aef28e1c47358b4bbba61862b9c26e59"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "7a4975e07d693d17ceead52c", "attempt_id": "efbe25eca3d8", "run_group_id": "E4_core:A4|repairable|qwen3b:h8", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:04:23.348795Z", "task_family": "synthetic", "document_id": "syn_03006_2000", "document_hash": "89be5e77e0ca114963e9ffdf8dc38fe5610f13a78ac29e7a1956cd9124ce5217", "question_id": "q08_revision", "split": "test", "cluster_id": "syn_03006_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 8, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1890, "query_tokens": 611, "message_tokens": 512, "message_utf8_bytes": 1364, "cache_tensor_bytes": null, "serialized_payload_bytes": 1364, "side_information_bytes": 0, "fetched_bytes": 5859, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 177969.8518749792, "hops": {"1": 23133.45850002952, "2": 22054.83804200776, "3": 22289.791416958906, "4": 22041.22679203283, "5": 22040.885791997425, "6": 21990.91641593259, "7": 22282.63995796442, "8": 22136.094958055764}, "receiver_prefill_ms": 414.8492499953136, "ttft_ms": 25.183625053614378, "decode_ms": 326.242332928814}, "total_wall_ms": 178736.81370797567, "ttft_ms": 25.183625053614378, "input_usage": {"lookups": 28}, "output_usage": {"output_tokens": 9}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "9,112,735", "evidence_artifact": {"ledger": {"facts_total": 64, "facts_retained": 34, "retention_rate": 0.53125, "retention_by_kind": {"person": 0.5, "date": 0.3, "amount": 0.6666666666666666, "status": 0.9166666666666666, "place": 0.16666666666666666, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1364}, "ledger_by_hop": {"1": {"facts_total": 64, "facts_retained": 30, "retention_rate": 0.46875, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.6666666666666666, "status": 0.9166666666666666, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1087}, "2": {"facts_total": 64, "facts_retained": 30, "retention_rate": 0.46875, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.6666666666666666, "status": 0.9166666666666666, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1075}, "3": {"facts_total": 64, "facts_retained": 34, "retention_rate": 0.53125, "retention_by_kind": {"person": 0.5, "date": 0.3, "amount": 0.6666666666666666, "status": 0.9166666666666666, "place": 0.16666666666666666, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1364}, "4": {"facts_total": 64, "facts_retained": 34, "retention_rate": 0.53125, "retention_by_kind": {"person": 0.5, "date": 0.3, "amount": 0.6666666666666666, "status": 0.9166666666666666, "place": 0.16666666666666666, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1364}, "5": {"facts_total": 64, "facts_retained": 34, "retention_rate": 0.53125, "retention_by_kind": {"person": 0.5, "date": 0.3, "amount": 0.6666666666666666, "status": 0.9166666666666666, "place": 0.16666666666666666, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1364}, "6": {"facts_total": 64, "facts_retained": 34, "retention_rate": 0.53125, "retention_by_kind": {"person": 0.5, "date": 0.3, "amount": 0.6666666666666666, "status": 0.9166666666666666, "place": 0.16666666666666666, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1364}, "7": {"facts_total": 64, "facts_retained": 34, "retention_rate": 0.53125, "retention_by_kind": {"person": 0.5, "date": 0.3, "amount": 0.6666666666666666, "status": 0.9166666666666666, "place": 0.16666666666666666, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1364}, "8": {"facts_total": 64, "facts_retained": 34, "retention_rate": 0.53125, "retention_by_kind": {"person": 0.5, "date": 0.3, "amount": 0.6666666666666666, "status": 0.9166666666666666, "place": 0.16666666666666666, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1364}}}, "reference_key_id": "keys/synthetic/syn_03006_2000.json", "task_scores": {"correct": 0.0, "abstained": 0.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "05248504387de244bb26ede8192b5c4b34d448e31f7043a95c04d8a18c00902a", "output_hash": "788fcda908c2029597420995b09d0cc919bfdbf7a28b68dcb0e331db7df7b342"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "45bc829b767acfd391dc1ba9", "attempt_id": "a04f8dc728ef", "run_group_id": "E4_core:A4|repairable|qwen3b:h8", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:04:24.132936Z", "task_family": "synthetic", "document_id": "syn_03006_2000", "document_hash": "89be5e77e0ca114963e9ffdf8dc38fe5610f13a78ac29e7a1956cd9124ce5217", "question_id": "q09_revision", "split": "test", "cluster_id": "syn_03006_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 8, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1890, "query_tokens": 604, "message_tokens": 512, "message_utf8_bytes": 1364, "cache_tensor_bytes": null, "serialized_payload_bytes": 1364, "side_information_bytes": 0, "fetched_bytes": 5859, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 177969.8518749792, "hops": {"1": 23133.45850002952, "2": 22054.83804200776, "3": 22289.791416958906, "4": 22041.22679203283, "5": 22040.885791997425, "6": 21990.91641593259, "7": 22282.63995796442, "8": 22136.094958055764}, "receiver_prefill_ms": 513.3357919985428, "ttft_ms": 42.03287500422448, "decode_ms": 299.56595902331173}, "total_wall_ms": 178825.5208749324, "ttft_ms": 42.03287500422448, "input_usage": {"lookups": 28}, "output_usage": {"output_tokens": 9}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "9,112,735", "evidence_artifact": {"ledger": {"facts_total": 64, "facts_retained": 34, "retention_rate": 0.53125, "retention_by_kind": {"person": 0.5, "date": 0.3, "amount": 0.6666666666666666, "status": 0.9166666666666666, "place": 0.16666666666666666, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1364}, "ledger_by_hop": {"1": {"facts_total": 64, "facts_retained": 30, "retention_rate": 0.46875, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.6666666666666666, "status": 0.9166666666666666, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1087}, "2": {"facts_total": 64, "facts_retained": 30, "retention_rate": 0.46875, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.6666666666666666, "status": 0.9166666666666666, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1075}, "3": {"facts_total": 64, "facts_retained": 34, "retention_rate": 0.53125, "retention_by_kind": {"person": 0.5, "date": 0.3, "amount": 0.6666666666666666, "status": 0.9166666666666666, "place": 0.16666666666666666, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1364}, "4": {"facts_total": 64, "facts_retained": 34, "retention_rate": 0.53125, "retention_by_kind": {"person": 0.5, "date": 0.3, "amount": 0.6666666666666666, "status": 0.9166666666666666, "place": 0.16666666666666666, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1364}, "5": {"facts_total": 64, "facts_retained": 34, "retention_rate": 0.53125, "retention_by_kind": {"person": 0.5, "date": 0.3, "amount": 0.6666666666666666, "status": 0.9166666666666666, "place": 0.16666666666666666, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1364}, "6": {"facts_total": 64, "facts_retained": 34, "retention_rate": 0.53125, "retention_by_kind": {"person": 0.5, "date": 0.3, "amount": 0.6666666666666666, "status": 0.9166666666666666, "place": 0.16666666666666666, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1364}, "7": {"facts_total": 64, "facts_retained": 34, "retention_rate": 0.53125, "retention_by_kind": {"person": 0.5, "date": 0.3, "amount": 0.6666666666666666, "status": 0.9166666666666666, "place": 0.16666666666666666, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1364}, "8": {"facts_total": 64, "facts_retained": 34, "retention_rate": 0.53125, "retention_by_kind": {"person": 0.5, "date": 0.3, "amount": 0.6666666666666666, "status": 0.9166666666666666, "place": 0.16666666666666666, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1364}}}, "reference_key_id": "keys/synthetic/syn_03006_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "05248504387de244bb26ede8192b5c4b34d448e31f7043a95c04d8a18c00902a", "output_hash": "788fcda908c2029597420995b09d0cc919bfdbf7a28b68dcb0e331db7df7b342"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "d6d9938796d9078d784eacb7", "attempt_id": "be6ac307dbe4", "run_group_id": "E4_core:A4|repairable|qwen3b:h8", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:04:25.007389Z", "task_family": "synthetic", "document_id": "syn_03006_2000", "document_hash": "89be5e77e0ca114963e9ffdf8dc38fe5610f13a78ac29e7a1956cd9124ce5217", "question_id": "q10_temporal", "split": "test", "cluster_id": "syn_03006_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 8, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1890, "query_tokens": 613, "message_tokens": 512, "message_utf8_bytes": 1364, "cache_tensor_bytes": null, "serialized_payload_bytes": 1364, "side_information_bytes": 0, "fetched_bytes": 5859, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 177969.8518749792, "hops": {"1": 23133.45850002952, "2": 22054.83804200776, "3": 22289.791416958906, "4": 22041.22679203283, "5": 22040.885791997425, "6": 21990.91641593259, "7": 22282.63995796442, "8": 22136.094958055764}, "receiver_prefill_ms": 415.7477079425007, "ttft_ms": 35.08920792955905, "decode_ms": 36.69337497558445}, "total_wall_ms": 178458.1072499277, "ttft_ms": 35.08920792955905, "input_usage": {"lookups": 28}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 64, "facts_retained": 34, "retention_rate": 0.53125, "retention_by_kind": {"person": 0.5, "date": 0.3, "amount": 0.6666666666666666, "status": 0.9166666666666666, "place": 0.16666666666666666, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1364}, "ledger_by_hop": {"1": {"facts_total": 64, "facts_retained": 30, "retention_rate": 0.46875, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.6666666666666666, "status": 0.9166666666666666, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1087}, "2": {"facts_total": 64, "facts_retained": 30, "retention_rate": 0.46875, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.6666666666666666, "status": 0.9166666666666666, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1075}, "3": {"facts_total": 64, "facts_retained": 34, "retention_rate": 0.53125, "retention_by_kind": {"person": 0.5, "date": 0.3, "amount": 0.6666666666666666, "status": 0.9166666666666666, "place": 0.16666666666666666, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1364}, "4": {"facts_total": 64, "facts_retained": 34, "retention_rate": 0.53125, "retention_by_kind": {"person": 0.5, "date": 0.3, "amount": 0.6666666666666666, "status": 0.9166666666666666, "place": 0.16666666666666666, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1364}, "5": {"facts_total": 64, "facts_retained": 34, "retention_rate": 0.53125, "retention_by_kind": {"person": 0.5, "date": 0.3, "amount": 0.6666666666666666, "status": 0.9166666666666666, "place": 0.16666666666666666, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1364}, "6": {"facts_total": 64, "facts_retained": 34, "retention_rate": 0.53125, "retention_by_kind": {"person": 0.5, "date": 0.3, "amount": 0.6666666666666666, "status": 0.9166666666666666, "place": 0.16666666666666666, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1364}, "7": {"facts_total": 64, "facts_retained": 34, "retention_rate": 0.53125, "retention_by_kind": {"person": 0.5, "date": 0.3, "amount": 0.6666666666666666, "status": 0.9166666666666666, "place": 0.16666666666666666, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1364}, "8": {"facts_total": 64, "facts_retained": 34, "retention_rate": 0.53125, "retention_by_kind": {"person": 0.5, "date": 0.3, "amount": 0.6666666666666666, "status": 0.9166666666666666, "place": 0.16666666666666666, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1364}}}, "reference_key_id": "keys/synthetic/syn_03006_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "05248504387de244bb26ede8192b5c4b34d448e31f7043a95c04d8a18c00902a", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "c35df4b32166976ad341d5d8", "attempt_id": "792f941fbdd7", "run_group_id": "E4_core:A4|repairable|qwen3b:h8", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:04:25.513878Z", "task_family": "synthetic", "document_id": "syn_03006_2000", "document_hash": "89be5e77e0ca114963e9ffdf8dc38fe5610f13a78ac29e7a1956cd9124ce5217", "question_id": "q11_aggregation", "split": "test", "cluster_id": "syn_03006_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 8, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1890, "query_tokens": 606, "message_tokens": 512, "message_utf8_bytes": 1364, "cache_tensor_bytes": null, "serialized_payload_bytes": 1364, "side_information_bytes": 0, "fetched_bytes": 5859, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 177969.8518749792, "hops": {"1": 23133.45850002952, "2": 22054.83804200776, "3": 22289.791416958906, "4": 22041.22679203283, "5": 22040.885791997425, "6": 21990.91641593259, "7": 22282.63995796442, "8": 22136.094958055764}, "receiver_prefill_ms": 375.5364589160308, "ttft_ms": 45.06525001488626, "decode_ms": 449.16850002482533}, "total_wall_ms": 178840.55800002534, "ttft_ms": 45.06525001488626, "input_usage": {"lookups": 28}, "output_usage": {"output_tokens": 7}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "8908553", "evidence_artifact": {"ledger": {"facts_total": 64, "facts_retained": 34, "retention_rate": 0.53125, "retention_by_kind": {"person": 0.5, "date": 0.3, "amount": 0.6666666666666666, "status": 0.9166666666666666, "place": 0.16666666666666666, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1364}, "ledger_by_hop": {"1": {"facts_total": 64, "facts_retained": 30, "retention_rate": 0.46875, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.6666666666666666, "status": 0.9166666666666666, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1087}, "2": {"facts_total": 64, "facts_retained": 30, "retention_rate": 0.46875, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.6666666666666666, "status": 0.9166666666666666, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1075}, "3": {"facts_total": 64, "facts_retained": 34, "retention_rate": 0.53125, "retention_by_kind": {"person": 0.5, "date": 0.3, "amount": 0.6666666666666666, "status": 0.9166666666666666, "place": 0.16666666666666666, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1364}, "4": {"facts_total": 64, "facts_retained": 34, "retention_rate": 0.53125, "retention_by_kind": {"person": 0.5, "date": 0.3, "amount": 0.6666666666666666, "status": 0.9166666666666666, "place": 0.16666666666666666, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1364}, "5": {"facts_total": 64, "facts_retained": 34, "retention_rate": 0.53125, "retention_by_kind": {"person": 0.5, "date": 0.3, "amount": 0.6666666666666666, "status": 0.9166666666666666, "place": 0.16666666666666666, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1364}, "6": {"facts_total": 64, "facts_retained": 34, "retention_rate": 0.53125, "retention_by_kind": {"person": 0.5, "date": 0.3, "amount": 0.6666666666666666, "status": 0.9166666666666666, "place": 0.16666666666666666, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1364}, "7": {"facts_total": 64, "facts_retained": 34, "retention_rate": 0.53125, "retention_by_kind": {"person": 0.5, "date": 0.3, "amount": 0.6666666666666666, "status": 0.9166666666666666, "place": 0.16666666666666666, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1364}, "8": {"facts_total": 64, "facts_retained": 34, "retention_rate": 0.53125, "retention_by_kind": {"person": 0.5, "date": 0.3, "amount": 0.6666666666666666, "status": 0.9166666666666666, "place": 0.16666666666666666, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1364}}}, "reference_key_id": "keys/synthetic/syn_03006_2000.json", "task_scores": {"correct": 0.0, "abstained": 0.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "05248504387de244bb26ede8192b5c4b34d448e31f7043a95c04d8a18c00902a", "output_hash": "5ad9e777dc3be36808c74badfc8aae217b41a486754d07adf1abda6f6bce9aca"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "796e8cedb89c9d3bf7ec6c16", "attempt_id": "0a75358b8ee0", "run_group_id": "E4_core:A4|repairable|qwen3b:h8", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:04:26.403888Z", "task_family": "synthetic", "document_id": "syn_03006_2000", "document_hash": "89be5e77e0ca114963e9ffdf8dc38fe5610f13a78ac29e7a1956cd9124ce5217", "question_id": "q12_unit_conversion", "split": "test", "cluster_id": "syn_03006_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 8, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1890, "query_tokens": 603, "message_tokens": 512, "message_utf8_bytes": 1364, "cache_tensor_bytes": null, "serialized_payload_bytes": 1364, "side_information_bytes": 0, "fetched_bytes": 5859, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 177969.8518749792, "hops": {"1": 23133.45850002952, "2": 22054.83804200776, "3": 22289.791416958906, "4": 22041.22679203283, "5": 22040.885791997425, "6": 21990.91641593259, "7": 22282.63995796442, "8": 22136.094958055764}, "receiver_prefill_ms": 390.3277500066906, "ttft_ms": 39.74883398041129, "decode_ms": 151.0541249299422}, "total_wall_ms": 178551.67954205535, "ttft_ms": 39.74883398041129, "input_usage": {"lookups": 28}, "output_usage": {"output_tokens": 5}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "39700", "evidence_artifact": {"ledger": {"facts_total": 64, "facts_retained": 34, "retention_rate": 0.53125, "retention_by_kind": {"person": 0.5, "date": 0.3, "amount": 0.6666666666666666, "status": 0.9166666666666666, "place": 0.16666666666666666, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1364}, "ledger_by_hop": {"1": {"facts_total": 64, "facts_retained": 30, "retention_rate": 0.46875, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.6666666666666666, "status": 0.9166666666666666, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1087}, "2": {"facts_total": 64, "facts_retained": 30, "retention_rate": 0.46875, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.6666666666666666, "status": 0.9166666666666666, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1075}, "3": {"facts_total": 64, "facts_retained": 34, "retention_rate": 0.53125, "retention_by_kind": {"person": 0.5, "date": 0.3, "amount": 0.6666666666666666, "status": 0.9166666666666666, "place": 0.16666666666666666, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1364}, "4": {"facts_total": 64, "facts_retained": 34, "retention_rate": 0.53125, "retention_by_kind": {"person": 0.5, "date": 0.3, "amount": 0.6666666666666666, "status": 0.9166666666666666, "place": 0.16666666666666666, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1364}, "5": {"facts_total": 64, "facts_retained": 34, "retention_rate": 0.53125, "retention_by_kind": {"person": 0.5, "date": 0.3, "amount": 0.6666666666666666, "status": 0.9166666666666666, "place": 0.16666666666666666, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1364}, "6": {"facts_total": 64, "facts_retained": 34, "retention_rate": 0.53125, "retention_by_kind": {"person": 0.5, "date": 0.3, "amount": 0.6666666666666666, "status": 0.9166666666666666, "place": 0.16666666666666666, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1364}, "7": {"facts_total": 64, "facts_retained": 34, "retention_rate": 0.53125, "retention_by_kind": {"person": 0.5, "date": 0.3, "amount": 0.6666666666666666, "status": 0.9166666666666666, "place": 0.16666666666666666, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1364}, "8": {"facts_total": 64, "facts_retained": 34, "retention_rate": 0.53125, "retention_by_kind": {"person": 0.5, "date": 0.3, "amount": 0.6666666666666666, "status": 0.9166666666666666, "place": 0.16666666666666666, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1364}}}, "reference_key_id": "keys/synthetic/syn_03006_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "05248504387de244bb26ede8192b5c4b34d448e31f7043a95c04d8a18c00902a", "output_hash": "d8f3f4a352330800cb9f3d838e47d965e9c710d166f53c6b3724fdeae73c9a5e"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "3c99689b392f92609454aea7", "attempt_id": "b5c021d1c3d2", "run_group_id": "E4_core:A4|repairable|qwen3b:h8", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:04:27.003210Z", "task_family": "synthetic", "document_id": "syn_03006_2000", "document_hash": "89be5e77e0ca114963e9ffdf8dc38fe5610f13a78ac29e7a1956cd9124ce5217", "question_id": "q13_unanswerable", "split": "test", "cluster_id": "syn_03006_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 8, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1890, "query_tokens": 607, "message_tokens": 512, "message_utf8_bytes": 1364, "cache_tensor_bytes": null, "serialized_payload_bytes": 1364, "side_information_bytes": 0, "fetched_bytes": 5859, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 177969.8518749792, "hops": {"1": 23133.45850002952, "2": 22054.83804200776, "3": 22289.791416958906, "4": 22041.22679203283, "5": 22040.885791997425, "6": 21990.91641593259, "7": 22282.63995796442, "8": 22136.094958055764}, "receiver_prefill_ms": 358.21225005201995, "ttft_ms": 50.21095904521644, "decode_ms": 50.980209023691714}, "total_wall_ms": 178429.94308297057, "ttft_ms": 50.21095904521644, "input_usage": {"lookups": 28}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 64, "facts_retained": 34, "retention_rate": 0.53125, "retention_by_kind": {"person": 0.5, "date": 0.3, "amount": 0.6666666666666666, "status": 0.9166666666666666, "place": 0.16666666666666666, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1364}, "ledger_by_hop": {"1": {"facts_total": 64, "facts_retained": 30, "retention_rate": 0.46875, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.6666666666666666, "status": 0.9166666666666666, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1087}, "2": {"facts_total": 64, "facts_retained": 30, "retention_rate": 0.46875, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.6666666666666666, "status": 0.9166666666666666, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1075}, "3": {"facts_total": 64, "facts_retained": 34, "retention_rate": 0.53125, "retention_by_kind": {"person": 0.5, "date": 0.3, "amount": 0.6666666666666666, "status": 0.9166666666666666, "place": 0.16666666666666666, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1364}, "4": {"facts_total": 64, "facts_retained": 34, "retention_rate": 0.53125, "retention_by_kind": {"person": 0.5, "date": 0.3, "amount": 0.6666666666666666, "status": 0.9166666666666666, "place": 0.16666666666666666, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1364}, "5": {"facts_total": 64, "facts_retained": 34, "retention_rate": 0.53125, "retention_by_kind": {"person": 0.5, "date": 0.3, "amount": 0.6666666666666666, "status": 0.9166666666666666, "place": 0.16666666666666666, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1364}, "6": {"facts_total": 64, "facts_retained": 34, "retention_rate": 0.53125, "retention_by_kind": {"person": 0.5, "date": 0.3, "amount": 0.6666666666666666, "status": 0.9166666666666666, "place": 0.16666666666666666, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1364}, "7": {"facts_total": 64, "facts_retained": 34, "retention_rate": 0.53125, "retention_by_kind": {"person": 0.5, "date": 0.3, "amount": 0.6666666666666666, "status": 0.9166666666666666, "place": 0.16666666666666666, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1364}, "8": {"facts_total": 64, "facts_retained": 34, "retention_rate": 0.53125, "retention_by_kind": {"person": 0.5, "date": 0.3, "amount": 0.6666666666666666, "status": 0.9166666666666666, "place": 0.16666666666666666, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1364}}}, "reference_key_id": "keys/synthetic/syn_03006_2000.json", "task_scores": {"correct": 1.0, "abstained": 1.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "05248504387de244bb26ede8192b5c4b34d448e31f7043a95c04d8a18c00902a", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "0450b480fc37ae373e979291", "attempt_id": "4baf9b4af547", "run_group_id": "E4_core:A4|repairable|qwen3b:h8", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:04:27.480920Z", "task_family": "synthetic", "document_id": "syn_03006_2000", "document_hash": "89be5e77e0ca114963e9ffdf8dc38fe5610f13a78ac29e7a1956cd9124ce5217", "question_id": "q14_unanswerable", "split": "test", "cluster_id": "syn_03006_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 8, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1890, "query_tokens": 608, "message_tokens": 512, "message_utf8_bytes": 1364, "cache_tensor_bytes": null, "serialized_payload_bytes": 1364, "side_information_bytes": 0, "fetched_bytes": 5859, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 177969.8518749792, "hops": {"1": 23133.45850002952, "2": 22054.83804200776, "3": 22289.791416958906, "4": 22041.22679203283, "5": 22040.885791997425, "6": 21990.91641593259, "7": 22282.63995796442, "8": 22136.094958055764}, "receiver_prefill_ms": 523.3383750310168, "ttft_ms": 31.771083013154566, "decode_ms": 40.048709022812545}, "total_wall_ms": 178565.702999942, "ttft_ms": 31.771083013154566, "input_usage": {"lookups": 28}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 64, "facts_retained": 34, "retention_rate": 0.53125, "retention_by_kind": {"person": 0.5, "date": 0.3, "amount": 0.6666666666666666, "status": 0.9166666666666666, "place": 0.16666666666666666, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1364}, "ledger_by_hop": {"1": {"facts_total": 64, "facts_retained": 30, "retention_rate": 0.46875, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.6666666666666666, "status": 0.9166666666666666, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1087}, "2": {"facts_total": 64, "facts_retained": 30, "retention_rate": 0.46875, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.6666666666666666, "status": 0.9166666666666666, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1075}, "3": {"facts_total": 64, "facts_retained": 34, "retention_rate": 0.53125, "retention_by_kind": {"person": 0.5, "date": 0.3, "amount": 0.6666666666666666, "status": 0.9166666666666666, "place": 0.16666666666666666, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1364}, "4": {"facts_total": 64, "facts_retained": 34, "retention_rate": 0.53125, "retention_by_kind": {"person": 0.5, "date": 0.3, "amount": 0.6666666666666666, "status": 0.9166666666666666, "place": 0.16666666666666666, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1364}, "5": {"facts_total": 64, "facts_retained": 34, "retention_rate": 0.53125, "retention_by_kind": {"person": 0.5, "date": 0.3, "amount": 0.6666666666666666, "status": 0.9166666666666666, "place": 0.16666666666666666, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1364}, "6": {"facts_total": 64, "facts_retained": 34, "retention_rate": 0.53125, "retention_by_kind": {"person": 0.5, "date": 0.3, "amount": 0.6666666666666666, "status": 0.9166666666666666, "place": 0.16666666666666666, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1364}, "7": {"facts_total": 64, "facts_retained": 34, "retention_rate": 0.53125, "retention_by_kind": {"person": 0.5, "date": 0.3, "amount": 0.6666666666666666, "status": 0.9166666666666666, "place": 0.16666666666666666, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1364}, "8": {"facts_total": 64, "facts_retained": 34, "retention_rate": 0.53125, "retention_by_kind": {"person": 0.5, "date": 0.3, "amount": 0.6666666666666666, "status": 0.9166666666666666, "place": 0.16666666666666666, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1364}}}, "reference_key_id": "keys/synthetic/syn_03006_2000.json", "task_scores": {"correct": 1.0, "abstained": 1.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "05248504387de244bb26ede8192b5c4b34d448e31f7043a95c04d8a18c00902a", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "dace660d24457059222c8783", "attempt_id": "bce6ba79dd55", "run_group_id": "E4_core:A5|repairable|qwen3b:h1", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:04:50.697486Z", "task_family": "synthetic", "document_id": "syn_03006_2000", "document_hash": "89be5e77e0ca114963e9ffdf8dc38fe5610f13a78ac29e7a1956cd9124ce5217", "question_id": "q00_single_fact", "split": "test", "cluster_id": "syn_03006_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 1, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1890, "query_tokens": 387, "message_tokens": 304, "message_utf8_bytes": 717, "cache_tensor_bytes": null, "serialized_payload_bytes": 717, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 22600.32433399465, "hops": {"1": 22600.32433399465}, "receiver_prefill_ms": 294.33208308182657, "ttft_ms": 25.103208026848733, "decode_ms": 17.85645808558911}, "total_wall_ms": 22938.128751004115, "ttft_ms": 25.103208026848733, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 64, "facts_retained": 10, "retention_rate": 0.15625, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.2, "amount": 0.16666666666666666, "status": 0.3333333333333333, "place": 0.0, "identifier": 0.0, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 669}, "ledger_by_hop": {"1": {"facts_total": 64, "facts_retained": 10, "retention_rate": 0.15625, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.2, "amount": 0.16666666666666666, "status": 0.3333333333333333, "place": 0.0, "identifier": 0.0, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 669}}}, "reference_key_id": "keys/synthetic/syn_03006_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "a7a61e059050ba2e9e391181c042fc9f413143ee1a9ef5f897c5cddf9f37fe4f", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "c66d92b31d20d9736035ce5a", "attempt_id": "cd3689c1af9f", "run_group_id": "E4_core:A5|repairable|qwen3b:h1", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:04:51.053200Z", "task_family": "synthetic", "document_id": "syn_03006_2000", "document_hash": "89be5e77e0ca114963e9ffdf8dc38fe5610f13a78ac29e7a1956cd9124ce5217", "question_id": "q01_single_fact", "split": "test", "cluster_id": "syn_03006_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 1, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1890, "query_tokens": 392, "message_tokens": 304, "message_utf8_bytes": 717, "cache_tensor_bytes": null, "serialized_payload_bytes": 717, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 22600.32433399465, "hops": {"1": 22600.32433399465}, "receiver_prefill_ms": 294.16808404494077, "ttft_ms": 46.638417057693005, "decode_ms": 70.05545904394239}, "total_wall_ms": 23011.702917981893, "ttft_ms": 46.638417057693005, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 64, "facts_retained": 10, "retention_rate": 0.15625, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.2, "amount": 0.16666666666666666, "status": 0.3333333333333333, "place": 0.0, "identifier": 0.0, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 669}, "ledger_by_hop": {"1": {"facts_total": 64, "facts_retained": 10, "retention_rate": 0.15625, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.2, "amount": 0.16666666666666666, "status": 0.3333333333333333, "place": 0.0, "identifier": 0.0, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 669}}}, "reference_key_id": "keys/synthetic/syn_03006_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "a7a61e059050ba2e9e391181c042fc9f413143ee1a9ef5f897c5cddf9f37fe4f", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "7189d5cb4033cd385ed8d702", "attempt_id": "52e33f86f242", "run_group_id": "E4_core:A5|repairable|qwen3b:h1", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:04:51.483974Z", "task_family": "synthetic", "document_id": "syn_03006_2000", "document_hash": "89be5e77e0ca114963e9ffdf8dc38fe5610f13a78ac29e7a1956cd9124ce5217", "question_id": "q02_exact_copy", "split": "test", "cluster_id": "syn_03006_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 1, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1890, "query_tokens": 391, "message_tokens": 304, "message_utf8_bytes": 717, "cache_tensor_bytes": null, "serialized_payload_bytes": 717, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 22600.32433399465, "hops": {"1": 22600.32433399465}, "receiver_prefill_ms": 411.9489169679582, "ttft_ms": 39.03229197021574, "decode_ms": 31.376583967357874}, "total_wall_ms": 23083.25479296036, "ttft_ms": 39.03229197021574, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 64, "facts_retained": 10, "retention_rate": 0.15625, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.2, "amount": 0.16666666666666666, "status": 0.3333333333333333, "place": 0.0, "identifier": 0.0, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 669}, "ledger_by_hop": {"1": {"facts_total": 64, "facts_retained": 10, "retention_rate": 0.15625, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.2, "amount": 0.16666666666666666, "status": 0.3333333333333333, "place": 0.0, "identifier": 0.0, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 669}}}, "reference_key_id": "keys/synthetic/syn_03006_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "a7a61e059050ba2e9e391181c042fc9f413143ee1a9ef5f897c5cddf9f37fe4f", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "512419f3cec711ab86c832f6", "attempt_id": "98b1a15e51bf", "run_group_id": "E4_core:A5|repairable|qwen3b:h1", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:04:51.985541Z", "task_family": "synthetic", "document_id": "syn_03006_2000", "document_hash": "89be5e77e0ca114963e9ffdf8dc38fe5610f13a78ac29e7a1956cd9124ce5217", "question_id": "q03_exact_copy", "split": "test", "cluster_id": "syn_03006_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 1, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1890, "query_tokens": 386, "message_tokens": 304, "message_utf8_bytes": 717, "cache_tensor_bytes": null, "serialized_payload_bytes": 717, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 22600.32433399465, "hops": {"1": 22600.32433399465}, "receiver_prefill_ms": 290.7774170162156, "ttft_ms": 42.99954103771597, "decode_ms": 33.75479206442833}, "total_wall_ms": 22968.35000102874, "ttft_ms": 42.99954103771597, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 64, "facts_retained": 10, "retention_rate": 0.15625, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.2, "amount": 0.16666666666666666, "status": 0.3333333333333333, "place": 0.0, "identifier": 0.0, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 669}, "ledger_by_hop": {"1": {"facts_total": 64, "facts_retained": 10, "retention_rate": 0.15625, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.2, "amount": 0.16666666666666666, "status": 0.3333333333333333, "place": 0.0, "identifier": 0.0, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 669}}}, "reference_key_id": "keys/synthetic/syn_03006_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "a7a61e059050ba2e9e391181c042fc9f413143ee1a9ef5f897c5cddf9f37fe4f", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "3ad648420b2a639496a94322", "attempt_id": "ed2f1c63741d", "run_group_id": "E4_core:A5|repairable|qwen3b:h1", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:04:52.370576Z", "task_family": "synthetic", "document_id": "syn_03006_2000", "document_hash": "89be5e77e0ca114963e9ffdf8dc38fe5610f13a78ac29e7a1956cd9124ce5217", "question_id": "q04_multi_hop", "split": "test", "cluster_id": "syn_03006_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 1, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1890, "query_tokens": 395, "message_tokens": 304, "message_utf8_bytes": 717, "cache_tensor_bytes": null, "serialized_payload_bytes": 717, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 22600.32433399465, "hops": {"1": 22600.32433399465}, "receiver_prefill_ms": 279.99487495981157, "ttft_ms": 27.155166026204824, "decode_ms": 21.520957932807505}, "total_wall_ms": 22929.447292000987, "ttft_ms": 27.155166026204824, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 64, "facts_retained": 10, "retention_rate": 0.15625, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.2, "amount": 0.16666666666666666, "status": 0.3333333333333333, "place": 0.0, "identifier": 0.0, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 669}, "ledger_by_hop": {"1": {"facts_total": 64, "facts_retained": 10, "retention_rate": 0.15625, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.2, "amount": 0.16666666666666666, "status": 0.3333333333333333, "place": 0.0, "identifier": 0.0, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 669}}}, "reference_key_id": "keys/synthetic/syn_03006_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "a7a61e059050ba2e9e391181c042fc9f413143ee1a9ef5f897c5cddf9f37fe4f", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "02848443f91466f03d42a829", "attempt_id": "7333d6a8bb7b", "run_group_id": "E4_core:A5|repairable|qwen3b:h1", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:04:52.716088Z", "task_family": "synthetic", "document_id": "syn_03006_2000", "document_hash": "89be5e77e0ca114963e9ffdf8dc38fe5610f13a78ac29e7a1956cd9124ce5217", "question_id": "q05_multi_hop", "split": "test", "cluster_id": "syn_03006_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 1, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1890, "query_tokens": 398, "message_tokens": 304, "message_utf8_bytes": 717, "cache_tensor_bytes": null, "serialized_payload_bytes": 717, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 22600.32433399465, "hops": {"1": 22600.32433399465}, "receiver_prefill_ms": 278.654458001256, "ttft_ms": 66.37379201129079, "decode_ms": 64.06162504572421}, "total_wall_ms": 23009.924584068358, "ttft_ms": 66.37379201129079, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 64, "facts_retained": 10, "retention_rate": 0.15625, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.2, "amount": 0.16666666666666666, "status": 0.3333333333333333, "place": 0.0, "identifier": 0.0, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 669}, "ledger_by_hop": {"1": {"facts_total": 64, "facts_retained": 10, "retention_rate": 0.15625, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.2, "amount": 0.16666666666666666, "status": 0.3333333333333333, "place": 0.0, "identifier": 0.0, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 669}}}, "reference_key_id": "keys/synthetic/syn_03006_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "a7a61e059050ba2e9e391181c042fc9f413143ee1a9ef5f897c5cddf9f37fe4f", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "ca14b165273d6f63da70f743", "attempt_id": "c26e362112d8", "run_group_id": "E4_core:A5|repairable|qwen3b:h1", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:04:53.144769Z", "task_family": "synthetic", "document_id": "syn_03006_2000", "document_hash": "89be5e77e0ca114963e9ffdf8dc38fe5610f13a78ac29e7a1956cd9124ce5217", "question_id": "q06_negation", "split": "test", "cluster_id": "syn_03006_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 1, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1890, "query_tokens": 386, "message_tokens": 304, "message_utf8_bytes": 717, "cache_tensor_bytes": null, "serialized_payload_bytes": 717, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 22600.32433399465, "hops": {"1": 22600.32433399465}, "receiver_prefill_ms": 420.6629160325974, "ttft_ms": 41.67720803525299, "decode_ms": 25.51945799496025}, "total_wall_ms": 23088.77329202369, "ttft_ms": 41.67720803525299, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 64, "facts_retained": 10, "retention_rate": 0.15625, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.2, "amount": 0.16666666666666666, "status": 0.3333333333333333, "place": 0.0, "identifier": 0.0, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 669}, "ledger_by_hop": {"1": {"facts_total": 64, "facts_retained": 10, "retention_rate": 0.15625, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.2, "amount": 0.16666666666666666, "status": 0.3333333333333333, "place": 0.0, "identifier": 0.0, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 669}}}, "reference_key_id": "keys/synthetic/syn_03006_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "a7a61e059050ba2e9e391181c042fc9f413143ee1a9ef5f897c5cddf9f37fe4f", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "c435b6444c767da94699e9c0", "attempt_id": "bae8d8f78f1f", "run_group_id": "E4_core:A5|repairable|qwen3b:h1", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:04:53.651227Z", "task_family": "synthetic", "document_id": "syn_03006_2000", "document_hash": "89be5e77e0ca114963e9ffdf8dc38fe5610f13a78ac29e7a1956cd9124ce5217", "question_id": "q07_negation", "split": "test", "cluster_id": "syn_03006_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 1, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1890, "query_tokens": 386, "message_tokens": 304, "message_utf8_bytes": 717, "cache_tensor_bytes": null, "serialized_payload_bytes": 717, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 22600.32433399465, "hops": {"1": 22600.32433399465}, "receiver_prefill_ms": 284.6437080297619, "ttft_ms": 25.106666958890855, "decode_ms": 37.1743340510875}, "total_wall_ms": 22947.761500021443, "ttft_ms": 25.106666958890855, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "no", "evidence_artifact": {"ledger": {"facts_total": 64, "facts_retained": 10, "retention_rate": 0.15625, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.2, "amount": 0.16666666666666666, "status": 0.3333333333333333, "place": 0.0, "identifier": 0.0, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 669}, "ledger_by_hop": {"1": {"facts_total": 64, "facts_retained": 10, "retention_rate": 0.15625, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.2, "amount": 0.16666666666666666, "status": 0.3333333333333333, "place": 0.0, "identifier": 0.0, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 669}}}, "reference_key_id": "keys/synthetic/syn_03006_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "a7a61e059050ba2e9e391181c042fc9f413143ee1a9ef5f897c5cddf9f37fe4f", "output_hash": "9390298f3fb0c5b160498935d79cb139aef28e1c47358b4bbba61862b9c26e59"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "918c767c9638e2d7b2673a2d", "attempt_id": "114375a2659d", "run_group_id": "E4_core:A5|repairable|qwen3b:h1", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:04:54.015933Z", "task_family": "synthetic", "document_id": "syn_03006_2000", "document_hash": "89be5e77e0ca114963e9ffdf8dc38fe5610f13a78ac29e7a1956cd9124ce5217", "question_id": "q08_revision", "split": "test", "cluster_id": "syn_03006_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 1, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1890, "query_tokens": 403, "message_tokens": 304, "message_utf8_bytes": 717, "cache_tensor_bytes": null, "serialized_payload_bytes": 717, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 22600.32433399465, "hops": {"1": 22600.32433399465}, "receiver_prefill_ms": 286.03199997451156, "ttft_ms": 37.257791962474585, "decode_ms": 381.29445910453796}, "total_wall_ms": 23305.46945903916, "ttft_ms": 37.257791962474585, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 8}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "10107545", "evidence_artifact": {"ledger": {"facts_total": 64, "facts_retained": 10, "retention_rate": 0.15625, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.2, "amount": 0.16666666666666666, "status": 0.3333333333333333, "place": 0.0, "identifier": 0.0, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 669}, "ledger_by_hop": {"1": {"facts_total": 64, "facts_retained": 10, "retention_rate": 0.15625, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.2, "amount": 0.16666666666666666, "status": 0.3333333333333333, "place": 0.0, "identifier": 0.0, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 669}}}, "reference_key_id": "keys/synthetic/syn_03006_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "a7a61e059050ba2e9e391181c042fc9f413143ee1a9ef5f897c5cddf9f37fe4f", "output_hash": "6faba9de4e29949bbdd8934f69fc2110f07076db6801b211aeca9351292537fe"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "37a440ac8e4aa31d15cf3601", "attempt_id": "2b82c97b1aad", "run_group_id": "E4_core:A5|repairable|qwen3b:h1", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:04:54.738542Z", "task_family": "synthetic", "document_id": "syn_03006_2000", "document_hash": "89be5e77e0ca114963e9ffdf8dc38fe5610f13a78ac29e7a1956cd9124ce5217", "question_id": "q09_revision", "split": "test", "cluster_id": "syn_03006_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 1, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1890, "query_tokens": 396, "message_tokens": 304, "message_utf8_bytes": 717, "cache_tensor_bytes": null, "serialized_payload_bytes": 717, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 22600.32433399465, "hops": {"1": 22600.32433399465}, "receiver_prefill_ms": 416.3103330647573, "ttft_ms": 38.48704101983458, "decode_ms": 259.30808403063565}, "total_wall_ms": 23314.956376096234, "ttft_ms": 38.48704101983458, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 7}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "9112735", "evidence_artifact": {"ledger": {"facts_total": 64, "facts_retained": 10, "retention_rate": 0.15625, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.2, "amount": 0.16666666666666666, "status": 0.3333333333333333, "place": 0.0, "identifier": 0.0, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 669}, "ledger_by_hop": {"1": {"facts_total": 64, "facts_retained": 10, "retention_rate": 0.15625, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.2, "amount": 0.16666666666666666, "status": 0.3333333333333333, "place": 0.0, "identifier": 0.0, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 669}}}, "reference_key_id": "keys/synthetic/syn_03006_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "a7a61e059050ba2e9e391181c042fc9f413143ee1a9ef5f897c5cddf9f37fe4f", "output_hash": "15a3f804636f0e62d4a3d9be7677f001a0ad6b1fb8ae7e8cdc63c44ec7bac31f"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "84fa00a4f4c569fe271ebdf2", "attempt_id": "d236732bd240", "run_group_id": "E4_core:A5|repairable|qwen3b:h1", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:04:55.471943Z", "task_family": "synthetic", "document_id": "syn_03006_2000", "document_hash": "89be5e77e0ca114963e9ffdf8dc38fe5610f13a78ac29e7a1956cd9124ce5217", "question_id": "q10_temporal", "split": "test", "cluster_id": "syn_03006_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 1, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1890, "query_tokens": 405, "message_tokens": 304, "message_utf8_bytes": 717, "cache_tensor_bytes": null, "serialized_payload_bytes": 717, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 22600.32433399465, "hops": {"1": 22600.32433399465}, "receiver_prefill_ms": 303.4562920220196, "ttft_ms": 37.35670796595514, "decode_ms": 32.00120897963643}, "total_wall_ms": 22973.563459003344, "ttft_ms": 37.35670796595514, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 64, "facts_retained": 10, "retention_rate": 0.15625, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.2, "amount": 0.16666666666666666, "status": 0.3333333333333333, "place": 0.0, "identifier": 0.0, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 669}, "ledger_by_hop": {"1": {"facts_total": 64, "facts_retained": 10, "retention_rate": 0.15625, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.2, "amount": 0.16666666666666666, "status": 0.3333333333333333, "place": 0.0, "identifier": 0.0, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 669}}}, "reference_key_id": "keys/synthetic/syn_03006_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "a7a61e059050ba2e9e391181c042fc9f413143ee1a9ef5f897c5cddf9f37fe4f", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "35895b857baea9b0493fc079", "attempt_id": "6cb00a686dd2", "run_group_id": "E4_core:A5|repairable|qwen3b:h1", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:04:55.862383Z", "task_family": "synthetic", "document_id": "syn_03006_2000", "document_hash": "89be5e77e0ca114963e9ffdf8dc38fe5610f13a78ac29e7a1956cd9124ce5217", "question_id": "q11_aggregation", "split": "test", "cluster_id": "syn_03006_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 1, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1890, "query_tokens": 398, "message_tokens": 304, "message_utf8_bytes": 717, "cache_tensor_bytes": null, "serialized_payload_bytes": 717, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 22600.32433399465, "hops": {"1": 22600.32433399465}, "receiver_prefill_ms": 299.7849580133334, "ttft_ms": 56.13195791374892, "decode_ms": 51.4935830142349}, "total_wall_ms": 23008.278875029646, "ttft_ms": 56.13195791374892, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 64, "facts_retained": 10, "retention_rate": 0.15625, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.2, "amount": 0.16666666666666666, "status": 0.3333333333333333, "place": 0.0, "identifier": 0.0, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 669}, "ledger_by_hop": {"1": {"facts_total": 64, "facts_retained": 10, "retention_rate": 0.15625, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.2, "amount": 0.16666666666666666, "status": 0.3333333333333333, "place": 0.0, "identifier": 0.0, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 669}}}, "reference_key_id": "keys/synthetic/syn_03006_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "a7a61e059050ba2e9e391181c042fc9f413143ee1a9ef5f897c5cddf9f37fe4f", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "10ea95d37ec661a07d2d028f", "attempt_id": "71e63e8e7fbb", "run_group_id": "E4_core:A5|repairable|qwen3b:h1", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:04:56.288382Z", "task_family": "synthetic", "document_id": "syn_03006_2000", "document_hash": "89be5e77e0ca114963e9ffdf8dc38fe5610f13a78ac29e7a1956cd9124ce5217", "question_id": "q12_unit_conversion", "split": "test", "cluster_id": "syn_03006_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 1, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1890, "query_tokens": 395, "message_tokens": 304, "message_utf8_bytes": 717, "cache_tensor_bytes": null, "serialized_payload_bytes": 717, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 22600.32433399465, "hops": {"1": 22600.32433399465}, "receiver_prefill_ms": 436.13408389501274, "ttft_ms": 42.437167023308575, "decode_ms": 308.45208396203816}, "total_wall_ms": 23387.80066696927, "ttft_ms": 42.437167023308575, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 8}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "11946000", "evidence_artifact": {"ledger": {"facts_total": 64, "facts_retained": 10, "retention_rate": 0.15625, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.2, "amount": 0.16666666666666666, "status": 0.3333333333333333, "place": 0.0, "identifier": 0.0, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 669}, "ledger_by_hop": {"1": {"facts_total": 64, "facts_retained": 10, "retention_rate": 0.15625, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.2, "amount": 0.16666666666666666, "status": 0.3333333333333333, "place": 0.0, "identifier": 0.0, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 669}}}, "reference_key_id": "keys/synthetic/syn_03006_2000.json", "task_scores": {"correct": 0.0, "abstained": 0.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "a7a61e059050ba2e9e391181c042fc9f413143ee1a9ef5f897c5cddf9f37fe4f", "output_hash": "edf27f197d1395ea4e7b86fae1feecd3591246e1299bba6a08655f5724a670ea"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "e483c20ca5146766927c8a83", "attempt_id": "7c20f69d5f33", "run_group_id": "E4_core:A5|repairable|qwen3b:h1", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:04:57.094653Z", "task_family": "synthetic", "document_id": "syn_03006_2000", "document_hash": "89be5e77e0ca114963e9ffdf8dc38fe5610f13a78ac29e7a1956cd9124ce5217", "question_id": "q13_unanswerable", "split": "test", "cluster_id": "syn_03006_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 1, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1890, "query_tokens": 399, "message_tokens": 304, "message_utf8_bytes": 717, "cache_tensor_bytes": null, "serialized_payload_bytes": 717, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 22600.32433399465, "hops": {"1": 22600.32433399465}, "receiver_prefill_ms": 278.11995800584555, "ttft_ms": 34.69187498558313, "decode_ms": 20.32916701864451}, "total_wall_ms": 22933.946167002432, "ttft_ms": 34.69187498558313, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 64, "facts_retained": 10, "retention_rate": 0.15625, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.2, "amount": 0.16666666666666666, "status": 0.3333333333333333, "place": 0.0, "identifier": 0.0, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 669}, "ledger_by_hop": {"1": {"facts_total": 64, "facts_retained": 10, "retention_rate": 0.15625, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.2, "amount": 0.16666666666666666, "status": 0.3333333333333333, "place": 0.0, "identifier": 0.0, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 669}}}, "reference_key_id": "keys/synthetic/syn_03006_2000.json", "task_scores": {"correct": 1.0, "abstained": 1.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "a7a61e059050ba2e9e391181c042fc9f413143ee1a9ef5f897c5cddf9f37fe4f", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "b56531ddf3a3801ba19bec43", "attempt_id": "ab2e6b0edc62", "run_group_id": "E4_core:A5|repairable|qwen3b:h1", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:04:57.445581Z", "task_family": "synthetic", "document_id": "syn_03006_2000", "document_hash": "89be5e77e0ca114963e9ffdf8dc38fe5610f13a78ac29e7a1956cd9124ce5217", "question_id": "q14_unanswerable", "split": "test", "cluster_id": "syn_03006_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 1, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1890, "query_tokens": 400, "message_tokens": 304, "message_utf8_bytes": 717, "cache_tensor_bytes": null, "serialized_payload_bytes": 717, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 22600.32433399465, "hops": {"1": 22600.32433399465}, "receiver_prefill_ms": 317.05804099328816, "ttft_ms": 36.67829092592001, "decode_ms": 68.94404196646065}, "total_wall_ms": 23023.566416930407, "ttft_ms": 36.67829092592001, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 64, "facts_retained": 10, "retention_rate": 0.15625, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.2, "amount": 0.16666666666666666, "status": 0.3333333333333333, "place": 0.0, "identifier": 0.0, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 669}, "ledger_by_hop": {"1": {"facts_total": 64, "facts_retained": 10, "retention_rate": 0.15625, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.2, "amount": 0.16666666666666666, "status": 0.3333333333333333, "place": 0.0, "identifier": 0.0, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 669}}}, "reference_key_id": "keys/synthetic/syn_03006_2000.json", "task_scores": {"correct": 1.0, "abstained": 1.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "a7a61e059050ba2e9e391181c042fc9f413143ee1a9ef5f897c5cddf9f37fe4f", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "62f48405528d085f68450da6", "attempt_id": "b47d6e89d6d2", "run_group_id": "E4_core:A5|repairable|qwen3b:h2", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:05:11.587560Z", "task_family": "synthetic", "document_id": "syn_03006_2000", "document_hash": "89be5e77e0ca114963e9ffdf8dc38fe5610f13a78ac29e7a1956cd9124ce5217", "question_id": "q00_single_fact", "split": "test", "cluster_id": "syn_03006_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 2, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1890, "query_tokens": 396, "message_tokens": 314, "message_utf8_bytes": 748, "cache_tensor_bytes": null, "serialized_payload_bytes": 748, "side_information_bytes": 0, "fetched_bytes": 600, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 36299.754125997424, "hops": {"1": 22600.32433399465, "2": 13699.429792002775}, "receiver_prefill_ms": 426.4782500686124, "ttft_ms": 38.81120902951807, "decode_ms": 27.608082978986204}, "total_wall_ms": 36793.292167945765, "ttft_ms": 38.81120902951807, "input_usage": {"lookups": 1}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 64, "facts_retained": 15, "retention_rate": 0.234375, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.2, "amount": 0.16666666666666666, "status": 0.75, "place": 0.0, "identifier": 0.0, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 700}, "ledger_by_hop": {"1": {"facts_total": 64, "facts_retained": 10, "retention_rate": 0.15625, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.2, "amount": 0.16666666666666666, "status": 0.3333333333333333, "place": 0.0, "identifier": 0.0, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 669}, "2": {"facts_total": 64, "facts_retained": 15, "retention_rate": 0.234375, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.2, "amount": 0.16666666666666666, "status": 0.75, "place": 0.0, "identifier": 0.0, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 700}}}, "reference_key_id": "keys/synthetic/syn_03006_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "0c54b5424952f31a24ef521de33c53e46d87191f991ea8eefeaec650a7d38f30", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "6040a008da59561d652ac09b", "attempt_id": "ec38e58ea668", "run_group_id": "E4_core:A5|repairable|qwen3b:h2", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:05:12.098909Z", "task_family": "synthetic", "document_id": "syn_03006_2000", "document_hash": "89be5e77e0ca114963e9ffdf8dc38fe5610f13a78ac29e7a1956cd9124ce5217", "question_id": "q01_single_fact", "split": "test", "cluster_id": "syn_03006_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 2, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1890, "query_tokens": 401, "message_tokens": 314, "message_utf8_bytes": 748, "cache_tensor_bytes": null, "serialized_payload_bytes": 748, "side_information_bytes": 0, "fetched_bytes": 600, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 36299.754125997424, "hops": {"1": 22600.32433399465, "2": 13699.429792002775}, "receiver_prefill_ms": 296.42362508457154, "ttft_ms": 32.17625000979751, "decode_ms": 32.731209066696465}, "total_wall_ms": 36661.715001100674, "ttft_ms": 32.17625000979751, "input_usage": {"lookups": 1}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 64, "facts_retained": 15, "retention_rate": 0.234375, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.2, "amount": 0.16666666666666666, "status": 0.75, "place": 0.0, "identifier": 0.0, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 700}, "ledger_by_hop": {"1": {"facts_total": 64, "facts_retained": 10, "retention_rate": 0.15625, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.2, "amount": 0.16666666666666666, "status": 0.3333333333333333, "place": 0.0, "identifier": 0.0, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 669}, "2": {"facts_total": 64, "facts_retained": 15, "retention_rate": 0.234375, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.2, "amount": 0.16666666666666666, "status": 0.75, "place": 0.0, "identifier": 0.0, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 700}}}, "reference_key_id": "keys/synthetic/syn_03006_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "0c54b5424952f31a24ef521de33c53e46d87191f991ea8eefeaec650a7d38f30", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "d9bfb4db8d8ce2436efda56a", "attempt_id": "842c36972652", "run_group_id": "E4_core:A5|repairable|qwen3b:h2", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:05:12.478541Z", "task_family": "synthetic", "document_id": "syn_03006_2000", "document_hash": "89be5e77e0ca114963e9ffdf8dc38fe5610f13a78ac29e7a1956cd9124ce5217", "question_id": "q02_exact_copy", "split": "test", "cluster_id": "syn_03006_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 2, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1890, "query_tokens": 400, "message_tokens": 314, "message_utf8_bytes": 748, "cache_tensor_bytes": null, "serialized_payload_bytes": 748, "side_information_bytes": 0, "fetched_bytes": 600, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 36299.754125997424, "hops": {"1": 22600.32433399465, "2": 13699.429792002775}, "receiver_prefill_ms": 298.59195894096047, "ttft_ms": 20.348541904240847, "decode_ms": 19.886916037648916}, "total_wall_ms": 36639.04387608636, "ttft_ms": 20.348541904240847, "input_usage": {"lookups": 1}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 64, "facts_retained": 15, "retention_rate": 0.234375, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.2, "amount": 0.16666666666666666, "status": 0.75, "place": 0.0, "identifier": 0.0, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 700}, "ledger_by_hop": {"1": {"facts_total": 64, "facts_retained": 10, "retention_rate": 0.15625, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.2, "amount": 0.16666666666666666, "status": 0.3333333333333333, "place": 0.0, "identifier": 0.0, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 669}, "2": {"facts_total": 64, "facts_retained": 15, "retention_rate": 0.234375, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.2, "amount": 0.16666666666666666, "status": 0.75, "place": 0.0, "identifier": 0.0, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 700}}}, "reference_key_id": "keys/synthetic/syn_03006_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "0c54b5424952f31a24ef521de33c53e46d87191f991ea8eefeaec650a7d38f30", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "c5e5a261b64a14814e382dca", "attempt_id": "79e92accd8ee", "run_group_id": "E4_core:A5|repairable|qwen3b:h2", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:05:12.835002Z", "task_family": "synthetic", "document_id": "syn_03006_2000", "document_hash": "89be5e77e0ca114963e9ffdf8dc38fe5610f13a78ac29e7a1956cd9124ce5217", "question_id": "q03_exact_copy", "split": "test", "cluster_id": "syn_03006_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 2, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1890, "query_tokens": 395, "message_tokens": 314, "message_utf8_bytes": 748, "cache_tensor_bytes": null, "serialized_payload_bytes": 748, "side_information_bytes": 0, "fetched_bytes": 600, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 36299.754125997424, "hops": {"1": 22600.32433399465, "2": 13699.429792002775}, "receiver_prefill_ms": 304.1442499961704, "ttft_ms": 47.30895801912993, "decode_ms": 43.89162501320243}, "total_wall_ms": 36695.90608403087, "ttft_ms": 47.30895801912993, "input_usage": {"lookups": 1}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 64, "facts_retained": 15, "retention_rate": 0.234375, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.2, "amount": 0.16666666666666666, "status": 0.75, "place": 0.0, "identifier": 0.0, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 700}, "ledger_by_hop": {"1": {"facts_total": 64, "facts_retained": 10, "retention_rate": 0.15625, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.2, "amount": 0.16666666666666666, "status": 0.3333333333333333, "place": 0.0, "identifier": 0.0, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 669}, "2": {"facts_total": 64, "facts_retained": 15, "retention_rate": 0.234375, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.2, "amount": 0.16666666666666666, "status": 0.75, "place": 0.0, "identifier": 0.0, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 700}}}, "reference_key_id": "keys/synthetic/syn_03006_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "0c54b5424952f31a24ef521de33c53e46d87191f991ea8eefeaec650a7d38f30", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "b3f1eb0d18013e492e82a5d3", "attempt_id": "4c8af6bb3521", "run_group_id": "E4_core:A5|repairable|qwen3b:h2", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:05:13.250824Z", "task_family": "synthetic", "document_id": "syn_03006_2000", "document_hash": "89be5e77e0ca114963e9ffdf8dc38fe5610f13a78ac29e7a1956cd9124ce5217", "question_id": "q04_multi_hop", "split": "test", "cluster_id": "syn_03006_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 2, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1890, "query_tokens": 404, "message_tokens": 314, "message_utf8_bytes": 748, "cache_tensor_bytes": null, "serialized_payload_bytes": 748, "side_information_bytes": 0, "fetched_bytes": 600, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 36299.754125997424, "hops": {"1": 22600.32433399465, "2": 13699.429792002775}, "receiver_prefill_ms": 427.87912499625236, "ttft_ms": 39.067375008016825, "decode_ms": 31.531583052128553}, "total_wall_ms": 36798.858667956665, "ttft_ms": 39.067375008016825, "input_usage": {"lookups": 1}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 64, "facts_retained": 15, "retention_rate": 0.234375, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.2, "amount": 0.16666666666666666, "status": 0.75, "place": 0.0, "identifier": 0.0, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 700}, "ledger_by_hop": {"1": {"facts_total": 64, "facts_retained": 10, "retention_rate": 0.15625, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.2, "amount": 0.16666666666666666, "status": 0.3333333333333333, "place": 0.0, "identifier": 0.0, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 669}, "2": {"facts_total": 64, "facts_retained": 15, "retention_rate": 0.234375, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.2, "amount": 0.16666666666666666, "status": 0.75, "place": 0.0, "identifier": 0.0, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 700}}}, "reference_key_id": "keys/synthetic/syn_03006_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "0c54b5424952f31a24ef521de33c53e46d87191f991ea8eefeaec650a7d38f30", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "e2d6c7769caa0be77470552d", "attempt_id": "44c1302fba9a", "run_group_id": "E4_core:A5|repairable|qwen3b:h2", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:05:13.767421Z", "task_family": "synthetic", "document_id": "syn_03006_2000", "document_hash": "89be5e77e0ca114963e9ffdf8dc38fe5610f13a78ac29e7a1956cd9124ce5217", "question_id": "q05_multi_hop", "split": "test", "cluster_id": "syn_03006_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 2, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1890, "query_tokens": 407, "message_tokens": 314, "message_utf8_bytes": 748, "cache_tensor_bytes": null, "serialized_payload_bytes": 748, "side_information_bytes": 0, "fetched_bytes": 600, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 36299.754125997424, "hops": {"1": 22600.32433399465, "2": 13699.429792002775}, "receiver_prefill_ms": 284.67708290554583, "ttft_ms": 35.426208982244134, "decode_ms": 40.974624920636415}, "total_wall_ms": 36661.46425099578, "ttft_ms": 35.426208982244134, "input_usage": {"lookups": 1}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 64, "facts_retained": 15, "retention_rate": 0.234375, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.2, "amount": 0.16666666666666666, "status": 0.75, "place": 0.0, "identifier": 0.0, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 700}, "ledger_by_hop": {"1": {"facts_total": 64, "facts_retained": 10, "retention_rate": 0.15625, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.2, "amount": 0.16666666666666666, "status": 0.3333333333333333, "place": 0.0, "identifier": 0.0, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 669}, "2": {"facts_total": 64, "facts_retained": 15, "retention_rate": 0.234375, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.2, "amount": 0.16666666666666666, "status": 0.75, "place": 0.0, "identifier": 0.0, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 700}}}, "reference_key_id": "keys/synthetic/syn_03006_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "0c54b5424952f31a24ef521de33c53e46d87191f991ea8eefeaec650a7d38f30", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "d6801ed145d802f0dccd5aaf", "attempt_id": "72a96548c12b", "run_group_id": "E4_core:A5|repairable|qwen3b:h2", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:05:14.146135Z", "task_family": "synthetic", "document_id": "syn_03006_2000", "document_hash": "89be5e77e0ca114963e9ffdf8dc38fe5610f13a78ac29e7a1956cd9124ce5217", "question_id": "q06_negation", "split": "test", "cluster_id": "syn_03006_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 2, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1890, "query_tokens": 395, "message_tokens": 314, "message_utf8_bytes": 748, "cache_tensor_bytes": null, "serialized_payload_bytes": 748, "side_information_bytes": 0, "fetched_bytes": 600, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 36299.754125997424, "hops": {"1": 22600.32433399465, "2": 13699.429792002775}, "receiver_prefill_ms": 290.3099999530241, "ttft_ms": 27.277334011159837, "decode_ms": 20.235624979250133}, "total_wall_ms": 36638.08216690086, "ttft_ms": 27.277334011159837, "input_usage": {"lookups": 1}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 64, "facts_retained": 15, "retention_rate": 0.234375, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.2, "amount": 0.16666666666666666, "status": 0.75, "place": 0.0, "identifier": 0.0, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 700}, "ledger_by_hop": {"1": {"facts_total": 64, "facts_retained": 10, "retention_rate": 0.15625, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.2, "amount": 0.16666666666666666, "status": 0.3333333333333333, "place": 0.0, "identifier": 0.0, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 669}, "2": {"facts_total": 64, "facts_retained": 15, "retention_rate": 0.234375, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.2, "amount": 0.16666666666666666, "status": 0.75, "place": 0.0, "identifier": 0.0, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 700}}}, "reference_key_id": "keys/synthetic/syn_03006_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "0c54b5424952f31a24ef521de33c53e46d87191f991ea8eefeaec650a7d38f30", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "12d7402b1e66aac926397151", "attempt_id": "41e44ffb6d7d", "run_group_id": "E4_core:A5|repairable|qwen3b:h2", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:05:14.501720Z", "task_family": "synthetic", "document_id": "syn_03006_2000", "document_hash": "89be5e77e0ca114963e9ffdf8dc38fe5610f13a78ac29e7a1956cd9124ce5217", "question_id": "q07_negation", "split": "test", "cluster_id": "syn_03006_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 2, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1890, "query_tokens": 395, "message_tokens": 314, "message_utf8_bytes": 748, "cache_tensor_bytes": null, "serialized_payload_bytes": 748, "side_information_bytes": 0, "fetched_bytes": 600, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 36299.754125997424, "hops": {"1": 22600.32433399465, "2": 13699.429792002775}, "receiver_prefill_ms": 327.7547500329092, "ttft_ms": 52.3798749782145, "decode_ms": 40.83266691304743}, "total_wall_ms": 36721.398459980264, "ttft_ms": 52.3798749782145, "input_usage": {"lookups": 1}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "no", "evidence_artifact": {"ledger": {"facts_total": 64, "facts_retained": 15, "retention_rate": 0.234375, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.2, "amount": 0.16666666666666666, "status": 0.75, "place": 0.0, "identifier": 0.0, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 700}, "ledger_by_hop": {"1": {"facts_total": 64, "facts_retained": 10, "retention_rate": 0.15625, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.2, "amount": 0.16666666666666666, "status": 0.3333333333333333, "place": 0.0, "identifier": 0.0, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 669}, "2": {"facts_total": 64, "facts_retained": 15, "retention_rate": 0.234375, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.2, "amount": 0.16666666666666666, "status": 0.75, "place": 0.0, "identifier": 0.0, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 700}}}, "reference_key_id": "keys/synthetic/syn_03006_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "0c54b5424952f31a24ef521de33c53e46d87191f991ea8eefeaec650a7d38f30", "output_hash": "9390298f3fb0c5b160498935d79cb139aef28e1c47358b4bbba61862b9c26e59"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "70e3cd4b25d8972d2ed0b0ad", "attempt_id": "a21c23946dd9", "run_group_id": "E4_core:A5|repairable|qwen3b:h2", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:05:14.943076Z", "task_family": "synthetic", "document_id": "syn_03006_2000", "document_hash": "89be5e77e0ca114963e9ffdf8dc38fe5610f13a78ac29e7a1956cd9124ce5217", "question_id": "q08_revision", "split": "test", "cluster_id": "syn_03006_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 2, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1890, "query_tokens": 412, "message_tokens": 314, "message_utf8_bytes": 748, "cache_tensor_bytes": null, "serialized_payload_bytes": 748, "side_information_bytes": 0, "fetched_bytes": 600, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 36299.754125997424, "hops": {"1": 22600.32433399465, "2": 13699.429792002775}, "receiver_prefill_ms": 416.85358295217156, "ttft_ms": 37.08825004287064, "decode_ms": 257.26666604168713}, "total_wall_ms": 37011.52187597472, "ttft_ms": 37.08825004287064, "input_usage": {"lookups": 1}, "output_usage": {"output_tokens": 8}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "10107545", "evidence_artifact": {"ledger": {"facts_total": 64, "facts_retained": 15, "retention_rate": 0.234375, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.2, "amount": 0.16666666666666666, "status": 0.75, "place": 0.0, "identifier": 0.0, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 700}, "ledger_by_hop": {"1": {"facts_total": 64, "facts_retained": 10, "retention_rate": 0.15625, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.2, "amount": 0.16666666666666666, "status": 0.3333333333333333, "place": 0.0, "identifier": 0.0, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 669}, "2": {"facts_total": 64, "facts_retained": 15, "retention_rate": 0.234375, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.2, "amount": 0.16666666666666666, "status": 0.75, "place": 0.0, "identifier": 0.0, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 700}}}, "reference_key_id": "keys/synthetic/syn_03006_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "0c54b5424952f31a24ef521de33c53e46d87191f991ea8eefeaec650a7d38f30", "output_hash": "6faba9de4e29949bbdd8934f69fc2110f07076db6801b211aeca9351292537fe"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "9d8cde7932348c7ca16bc286", "attempt_id": "7e5d56732519", "run_group_id": "E4_core:A5|repairable|qwen3b:h2", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:05:15.671779Z", "task_family": "synthetic", "document_id": "syn_03006_2000", "document_hash": "89be5e77e0ca114963e9ffdf8dc38fe5610f13a78ac29e7a1956cd9124ce5217", "question_id": "q09_revision", "split": "test", "cluster_id": "syn_03006_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 2, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1890, "query_tokens": 405, "message_tokens": 314, "message_utf8_bytes": 748, "cache_tensor_bytes": null, "serialized_payload_bytes": 748, "side_information_bytes": 0, "fetched_bytes": 600, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 36299.754125997424, "hops": {"1": 22600.32433399465, "2": 13699.429792002775}, "receiver_prefill_ms": 297.377249924466, "ttft_ms": 38.452040986157954, "decode_ms": 358.3853329764679}, "total_wall_ms": 36994.65712590609, "ttft_ms": 38.452040986157954, "input_usage": {"lookups": 1}, "output_usage": {"output_tokens": 9}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "9,112,735", "evidence_artifact": {"ledger": {"facts_total": 64, "facts_retained": 15, "retention_rate": 0.234375, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.2, "amount": 0.16666666666666666, "status": 0.75, "place": 0.0, "identifier": 0.0, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 700}, "ledger_by_hop": {"1": {"facts_total": 64, "facts_retained": 10, "retention_rate": 0.15625, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.2, "amount": 0.16666666666666666, "status": 0.3333333333333333, "place": 0.0, "identifier": 0.0, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 669}, "2": {"facts_total": 64, "facts_retained": 15, "retention_rate": 0.234375, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.2, "amount": 0.16666666666666666, "status": 0.75, "place": 0.0, "identifier": 0.0, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 700}}}, "reference_key_id": "keys/synthetic/syn_03006_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "0c54b5424952f31a24ef521de33c53e46d87191f991ea8eefeaec650a7d38f30", "output_hash": "788fcda908c2029597420995b09d0cc919bfdbf7a28b68dcb0e331db7df7b342"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "3a47191ae9e5d15ca245ea9d", "attempt_id": "7b5917d98747", "run_group_id": "E4_core:A5|repairable|qwen3b:h2", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:05:16.384446Z", "task_family": "synthetic", "document_id": "syn_03006_2000", "document_hash": "89be5e77e0ca114963e9ffdf8dc38fe5610f13a78ac29e7a1956cd9124ce5217", "question_id": "q10_temporal", "split": "test", "cluster_id": "syn_03006_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 2, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1890, "query_tokens": 414, "message_tokens": 314, "message_utf8_bytes": 748, "cache_tensor_bytes": null, "serialized_payload_bytes": 748, "side_information_bytes": 0, "fetched_bytes": 600, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 36299.754125997424, "hops": {"1": 22600.32433399465, "2": 13699.429792002775}, "receiver_prefill_ms": 439.3436670070514, "ttft_ms": 32.13533293455839, "decode_ms": 44.88262499216944}, "total_wall_ms": 36816.59737601876, "ttft_ms": 32.13533293455839, "input_usage": {"lookups": 1}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 64, "facts_retained": 15, "retention_rate": 0.234375, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.2, "amount": 0.16666666666666666, "status": 0.75, "place": 0.0, "identifier": 0.0, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 700}, "ledger_by_hop": {"1": {"facts_total": 64, "facts_retained": 10, "retention_rate": 0.15625, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.2, "amount": 0.16666666666666666, "status": 0.3333333333333333, "place": 0.0, "identifier": 0.0, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 669}, "2": {"facts_total": 64, "facts_retained": 15, "retention_rate": 0.234375, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.2, "amount": 0.16666666666666666, "status": 0.75, "place": 0.0, "identifier": 0.0, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 700}}}, "reference_key_id": "keys/synthetic/syn_03006_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "0c54b5424952f31a24ef521de33c53e46d87191f991ea8eefeaec650a7d38f30", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "c59e6a8378692452e83aad11", "attempt_id": "ab94374e0b7a", "run_group_id": "E4_core:A5|repairable|qwen3b:h2", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:05:16.918237Z", "task_family": "synthetic", "document_id": "syn_03006_2000", "document_hash": "89be5e77e0ca114963e9ffdf8dc38fe5610f13a78ac29e7a1956cd9124ce5217", "question_id": "q11_aggregation", "split": "test", "cluster_id": "syn_03006_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 2, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1890, "query_tokens": 407, "message_tokens": 314, "message_utf8_bytes": 748, "cache_tensor_bytes": null, "serialized_payload_bytes": 748, "side_information_bytes": 0, "fetched_bytes": 600, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 36299.754125997424, "hops": {"1": 22600.32433399465, "2": 13699.429792002775}, "receiver_prefill_ms": 290.9249580698088, "ttft_ms": 26.018333970569074, "decode_ms": 41.16949997842312}, "total_wall_ms": 36658.36279198993, "ttft_ms": 26.018333970569074, "input_usage": {"lookups": 1}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 64, "facts_retained": 15, "retention_rate": 0.234375, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.2, "amount": 0.16666666666666666, "status": 0.75, "place": 0.0, "identifier": 0.0, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 700}, "ledger_by_hop": {"1": {"facts_total": 64, "facts_retained": 10, "retention_rate": 0.15625, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.2, "amount": 0.16666666666666666, "status": 0.3333333333333333, "place": 0.0, "identifier": 0.0, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 669}, "2": {"facts_total": 64, "facts_retained": 15, "retention_rate": 0.234375, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.2, "amount": 0.16666666666666666, "status": 0.75, "place": 0.0, "identifier": 0.0, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 700}}}, "reference_key_id": "keys/synthetic/syn_03006_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "0c54b5424952f31a24ef521de33c53e46d87191f991ea8eefeaec650a7d38f30", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "d2f83054e14c862f5b4708af", "attempt_id": "67cf45cc0d33", "run_group_id": "E4_core:A5|repairable|qwen3b:h2", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:05:17.294431Z", "task_family": "synthetic", "document_id": "syn_03006_2000", "document_hash": "89be5e77e0ca114963e9ffdf8dc38fe5610f13a78ac29e7a1956cd9124ce5217", "question_id": "q12_unit_conversion", "split": "test", "cluster_id": "syn_03006_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 2, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1890, "query_tokens": 404, "message_tokens": 314, "message_utf8_bytes": 748, "cache_tensor_bytes": null, "serialized_payload_bytes": 748, "side_information_bytes": 0, "fetched_bytes": 600, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 36299.754125997424, "hops": {"1": 22600.32433399465, "2": 13699.429792002775}, "receiver_prefill_ms": 287.58008289150894, "ttft_ms": 33.3678750321269, "decode_ms": 395.81470808479935}, "total_wall_ms": 37017.03333505429, "ttft_ms": 33.3678750321269, "input_usage": {"lookups": 1}, "output_usage": {"output_tokens": 8}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "11945603", "evidence_artifact": {"ledger": {"facts_total": 64, "facts_retained": 15, "retention_rate": 0.234375, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.2, "amount": 0.16666666666666666, "status": 0.75, "place": 0.0, "identifier": 0.0, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 700}, "ledger_by_hop": {"1": {"facts_total": 64, "facts_retained": 10, "retention_rate": 0.15625, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.2, "amount": 0.16666666666666666, "status": 0.3333333333333333, "place": 0.0, "identifier": 0.0, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 669}, "2": {"facts_total": 64, "facts_retained": 15, "retention_rate": 0.234375, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.2, "amount": 0.16666666666666666, "status": 0.75, "place": 0.0, "identifier": 0.0, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 700}}}, "reference_key_id": "keys/synthetic/syn_03006_2000.json", "task_scores": {"correct": 0.0, "abstained": 0.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "0c54b5424952f31a24ef521de33c53e46d87191f991ea8eefeaec650a7d38f30", "output_hash": "dba5c8babee9a210bc726fd3be9edcb356e41b3d6d60bc3524acbb17605102d3"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "c967bd76d409cf2f087f5a74", "attempt_id": "d8636f33aabe", "run_group_id": "E4_core:A5|repairable|qwen3b:h2", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:05:18.028850Z", "task_family": "synthetic", "document_id": "syn_03006_2000", "document_hash": "89be5e77e0ca114963e9ffdf8dc38fe5610f13a78ac29e7a1956cd9124ce5217", "question_id": "q13_unanswerable", "split": "test", "cluster_id": "syn_03006_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 2, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1890, "query_tokens": 408, "message_tokens": 314, "message_utf8_bytes": 748, "cache_tensor_bytes": null, "serialized_payload_bytes": 748, "side_information_bytes": 0, "fetched_bytes": 600, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 36299.754125997424, "hops": {"1": 22600.32433399465, "2": 13699.429792002775}, "receiver_prefill_ms": 422.62779094744474, "ttft_ms": 30.460832989774644, "decode_ms": 41.12991702277213}, "total_wall_ms": 36794.50454295147, "ttft_ms": 30.460832989774644, "input_usage": {"lookups": 1}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 64, "facts_retained": 15, "retention_rate": 0.234375, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.2, "amount": 0.16666666666666666, "status": 0.75, "place": 0.0, "identifier": 0.0, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 700}, "ledger_by_hop": {"1": {"facts_total": 64, "facts_retained": 10, "retention_rate": 0.15625, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.2, "amount": 0.16666666666666666, "status": 0.3333333333333333, "place": 0.0, "identifier": 0.0, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 669}, "2": {"facts_total": 64, "facts_retained": 15, "retention_rate": 0.234375, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.2, "amount": 0.16666666666666666, "status": 0.75, "place": 0.0, "identifier": 0.0, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 700}}}, "reference_key_id": "keys/synthetic/syn_03006_2000.json", "task_scores": {"correct": 1.0, "abstained": 1.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "0c54b5424952f31a24ef521de33c53e46d87191f991ea8eefeaec650a7d38f30", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "528f7a4c6ab39d83162ba516", "attempt_id": "a704e3b784d6", "run_group_id": "E4_core:A5|repairable|qwen3b:h2", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:05:18.541121Z", "task_family": "synthetic", "document_id": "syn_03006_2000", "document_hash": "89be5e77e0ca114963e9ffdf8dc38fe5610f13a78ac29e7a1956cd9124ce5217", "question_id": "q14_unanswerable", "split": "test", "cluster_id": "syn_03006_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 2, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1890, "query_tokens": 409, "message_tokens": 314, "message_utf8_bytes": 748, "cache_tensor_bytes": null, "serialized_payload_bytes": 748, "side_information_bytes": 0, "fetched_bytes": 600, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 36299.754125997424, "hops": {"1": 22600.32433399465, "2": 13699.429792002775}, "receiver_prefill_ms": 309.57345897331834, "ttft_ms": 20.668125012889504, "decode_ms": 45.94362492207438}, "total_wall_ms": 36676.46066797897, "ttft_ms": 20.668125012889504, "input_usage": {"lookups": 1}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 64, "facts_retained": 15, "retention_rate": 0.234375, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.2, "amount": 0.16666666666666666, "status": 0.75, "place": 0.0, "identifier": 0.0, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 700}, "ledger_by_hop": {"1": {"facts_total": 64, "facts_retained": 10, "retention_rate": 0.15625, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.2, "amount": 0.16666666666666666, "status": 0.3333333333333333, "place": 0.0, "identifier": 0.0, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 669}, "2": {"facts_total": 64, "facts_retained": 15, "retention_rate": 0.234375, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.2, "amount": 0.16666666666666666, "status": 0.75, "place": 0.0, "identifier": 0.0, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 700}}}, "reference_key_id": "keys/synthetic/syn_03006_2000.json", "task_scores": {"correct": 1.0, "abstained": 1.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "0c54b5424952f31a24ef521de33c53e46d87191f991ea8eefeaec650a7d38f30", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "5c7057fc6839eecdd99d7cf2", "attempt_id": "a67b610159ec", "run_group_id": "E4_core:A5|repairable|qwen3b:h4", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:05:46.436935Z", "task_family": "synthetic", "document_id": "syn_03006_2000", "document_hash": "89be5e77e0ca114963e9ffdf8dc38fe5610f13a78ac29e7a1956cd9124ce5217", "question_id": "q00_single_fact", "split": "test", "cluster_id": "syn_03006_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 4, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1890, "query_tokens": 396, "message_tokens": 314, "message_utf8_bytes": 748, "cache_tensor_bytes": null, "serialized_payload_bytes": 748, "side_information_bytes": 0, "fetched_bytes": 1800, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 63798.83479292039, "hops": {"1": 22600.32433399465, "2": 13699.429792002775, "3": 13789.627166930586, "4": 13709.453499992378}, "receiver_prefill_ms": 262.9420410376042, "ttft_ms": 53.819999913685024, "decode_ms": 34.49062502477318}, "total_wall_ms": 64150.69466794375, "ttft_ms": 53.819999913685024, "input_usage": {"lookups": 3}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 64, "facts_retained": 15, "retention_rate": 0.234375, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.2, "amount": 0.16666666666666666, "status": 0.75, "place": 0.0, "identifier": 0.0, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 700}, "ledger_by_hop": {"1": {"facts_total": 64, "facts_retained": 10, "retention_rate": 0.15625, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.2, "amount": 0.16666666666666666, "status": 0.3333333333333333, "place": 0.0, "identifier": 0.0, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 669}, "2": {"facts_total": 64, "facts_retained": 15, "retention_rate": 0.234375, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.2, "amount": 0.16666666666666666, "status": 0.75, "place": 0.0, "identifier": 0.0, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 700}, "3": {"facts_total": 64, "facts_retained": 15, "retention_rate": 0.234375, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.2, "amount": 0.16666666666666666, "status": 0.75, "place": 0.0, "identifier": 0.0, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 700}, "4": {"facts_total": 64, "facts_retained": 15, "retention_rate": 0.234375, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.2, "amount": 0.16666666666666666, "status": 0.75, "place": 0.0, "identifier": 0.0, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 700}}}, "reference_key_id": "keys/synthetic/syn_03006_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "0c54b5424952f31a24ef521de33c53e46d87191f991ea8eefeaec650a7d38f30", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "455b3c3d774a1fd591a21e6d", "attempt_id": "4f8b7ce8d2cb", "run_group_id": "E4_core:A5|repairable|qwen3b:h4", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:05:46.807422Z", "task_family": "synthetic", "document_id": "syn_03006_2000", "document_hash": "89be5e77e0ca114963e9ffdf8dc38fe5610f13a78ac29e7a1956cd9124ce5217", "question_id": "q01_single_fact", "split": "test", "cluster_id": "syn_03006_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 4, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1890, "query_tokens": 401, "message_tokens": 314, "message_utf8_bytes": 748, "cache_tensor_bytes": null, "serialized_payload_bytes": 748, "side_information_bytes": 0, "fetched_bytes": 1800, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 63798.83479292039, "hops": {"1": 22600.32433399465, "2": 13699.429792002775, "3": 13789.627166930586, "4": 13709.453499992378}, "receiver_prefill_ms": 430.2793750539422, "ttft_ms": 69.88320802338421, "decode_ms": 33.747624955140054}, "total_wall_ms": 64333.28879287001, "ttft_ms": 69.88320802338421, "input_usage": {"lookups": 3}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 64, "facts_retained": 15, "retention_rate": 0.234375, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.2, "amount": 0.16666666666666666, "status": 0.75, "place": 0.0, "identifier": 0.0, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 700}, "ledger_by_hop": {"1": {"facts_total": 64, "facts_retained": 10, "retention_rate": 0.15625, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.2, "amount": 0.16666666666666666, "status": 0.3333333333333333, "place": 0.0, "identifier": 0.0, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 669}, "2": {"facts_total": 64, "facts_retained": 15, "retention_rate": 0.234375, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.2, "amount": 0.16666666666666666, "status": 0.75, "place": 0.0, "identifier": 0.0, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 700}, "3": {"facts_total": 64, "facts_retained": 15, "retention_rate": 0.234375, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.2, "amount": 0.16666666666666666, "status": 0.75, "place": 0.0, "identifier": 0.0, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 700}, "4": {"facts_total": 64, "facts_retained": 15, "retention_rate": 0.234375, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.2, "amount": 0.16666666666666666, "status": 0.75, "place": 0.0, "identifier": 0.0, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 700}}}, "reference_key_id": "keys/synthetic/syn_03006_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "0c54b5424952f31a24ef521de33c53e46d87191f991ea8eefeaec650a7d38f30", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "d4717e23ad7e60059992162b", "attempt_id": "4a50f537a5f0", "run_group_id": "E4_core:A5|repairable|qwen3b:h4", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:05:47.361233Z", "task_family": "synthetic", "document_id": "syn_03006_2000", "document_hash": "89be5e77e0ca114963e9ffdf8dc38fe5610f13a78ac29e7a1956cd9124ce5217", "question_id": "q02_exact_copy", "split": "test", "cluster_id": "syn_03006_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 4, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1890, "query_tokens": 400, "message_tokens": 314, "message_utf8_bytes": 748, "cache_tensor_bytes": null, "serialized_payload_bytes": 748, "side_information_bytes": 0, "fetched_bytes": 1800, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 63798.83479292039, "hops": {"1": 22600.32433399465, "2": 13699.429792002775, "3": 13789.627166930586, "4": 13709.453499992378}, "receiver_prefill_ms": 308.88558307196945, "ttft_ms": 33.9244999922812, "decode_ms": 37.93483297340572}, "total_wall_ms": 64180.038710008375, "ttft_ms": 33.9244999922812, "input_usage": {"lookups": 3}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 64, "facts_retained": 15, "retention_rate": 0.234375, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.2, "amount": 0.16666666666666666, "status": 0.75, "place": 0.0, "identifier": 0.0, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 700}, "ledger_by_hop": {"1": {"facts_total": 64, "facts_retained": 10, "retention_rate": 0.15625, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.2, "amount": 0.16666666666666666, "status": 0.3333333333333333, "place": 0.0, "identifier": 0.0, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 669}, "2": {"facts_total": 64, "facts_retained": 15, "retention_rate": 0.234375, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.2, "amount": 0.16666666666666666, "status": 0.75, "place": 0.0, "identifier": 0.0, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 700}, "3": {"facts_total": 64, "facts_retained": 15, "retention_rate": 0.234375, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.2, "amount": 0.16666666666666666, "status": 0.75, "place": 0.0, "identifier": 0.0, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 700}, "4": {"facts_total": 64, "facts_retained": 15, "retention_rate": 0.234375, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.2, "amount": 0.16666666666666666, "status": 0.75, "place": 0.0, "identifier": 0.0, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 700}}}, "reference_key_id": "keys/synthetic/syn_03006_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "0c54b5424952f31a24ef521de33c53e46d87191f991ea8eefeaec650a7d38f30", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "e16707cbdff8ed7e08ae67fb", "attempt_id": "9852c1f09829", "run_group_id": "E4_core:A5|repairable|qwen3b:h4", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:05:47.759352Z", "task_family": "synthetic", "document_id": "syn_03006_2000", "document_hash": "89be5e77e0ca114963e9ffdf8dc38fe5610f13a78ac29e7a1956cd9124ce5217", "question_id": "q03_exact_copy", "split": "test", "cluster_id": "syn_03006_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 4, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1890, "query_tokens": 395, "message_tokens": 314, "message_utf8_bytes": 748, "cache_tensor_bytes": null, "serialized_payload_bytes": 748, "side_information_bytes": 0, "fetched_bytes": 1800, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 63798.83479292039, "hops": {"1": 22600.32433399465, "2": 13699.429792002775, "3": 13789.627166930586, "4": 13709.453499992378}, "receiver_prefill_ms": 279.2332499520853, "ttft_ms": 40.74858396779746, "decode_ms": 29.778040945529938}, "total_wall_ms": 64149.14625091478, "ttft_ms": 40.74858396779746, "input_usage": {"lookups": 3}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 64, "facts_retained": 15, "retention_rate": 0.234375, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.2, "amount": 0.16666666666666666, "status": 0.75, "place": 0.0, "identifier": 0.0, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 700}, "ledger_by_hop": {"1": {"facts_total": 64, "facts_retained": 10, "retention_rate": 0.15625, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.2, "amount": 0.16666666666666666, "status": 0.3333333333333333, "place": 0.0, "identifier": 0.0, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 669}, "2": {"facts_total": 64, "facts_retained": 15, "retention_rate": 0.234375, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.2, "amount": 0.16666666666666666, "status": 0.75, "place": 0.0, "identifier": 0.0, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 700}, "3": {"facts_total": 64, "facts_retained": 15, "retention_rate": 0.234375, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.2, "amount": 0.16666666666666666, "status": 0.75, "place": 0.0, "identifier": 0.0, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 700}, "4": {"facts_total": 64, "facts_retained": 15, "retention_rate": 0.234375, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.2, "amount": 0.16666666666666666, "status": 0.75, "place": 0.0, "identifier": 0.0, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 700}}}, "reference_key_id": "keys/synthetic/syn_03006_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "0c54b5424952f31a24ef521de33c53e46d87191f991ea8eefeaec650a7d38f30", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "d7ab526e9a7ceb6b99557f3d", "attempt_id": "6b0590025f6e", "run_group_id": "E4_core:A5|repairable|qwen3b:h4", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:05:48.127411Z", "task_family": "synthetic", "document_id": "syn_03006_2000", "document_hash": "89be5e77e0ca114963e9ffdf8dc38fe5610f13a78ac29e7a1956cd9124ce5217", "question_id": "q04_multi_hop", "split": "test", "cluster_id": "syn_03006_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 4, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1890, "query_tokens": 404, "message_tokens": 314, "message_utf8_bytes": 748, "cache_tensor_bytes": null, "serialized_payload_bytes": 748, "side_information_bytes": 0, "fetched_bytes": 1800, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 63798.83479292039, "hops": {"1": 22600.32433399465, "2": 13699.429792002775, "3": 13789.627166930586, "4": 13709.453499992378}, "receiver_prefill_ms": 276.7131249420345, "ttft_ms": 37.81804093159735, "decode_ms": 59.45320799946785}, "total_wall_ms": 64173.57287602499, "ttft_ms": 37.81804093159735, "input_usage": {"lookups": 3}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 64, "facts_retained": 15, "retention_rate": 0.234375, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.2, "amount": 0.16666666666666666, "status": 0.75, "place": 0.0, "identifier": 0.0, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 700}, "ledger_by_hop": {"1": {"facts_total": 64, "facts_retained": 10, "retention_rate": 0.15625, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.2, "amount": 0.16666666666666666, "status": 0.3333333333333333, "place": 0.0, "identifier": 0.0, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 669}, "2": {"facts_total": 64, "facts_retained": 15, "retention_rate": 0.234375, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.2, "amount": 0.16666666666666666, "status": 0.75, "place": 0.0, "identifier": 0.0, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 700}, "3": {"facts_total": 64, "facts_retained": 15, "retention_rate": 0.234375, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.2, "amount": 0.16666666666666666, "status": 0.75, "place": 0.0, "identifier": 0.0, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 700}, "4": {"facts_total": 64, "facts_retained": 15, "retention_rate": 0.234375, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.2, "amount": 0.16666666666666666, "status": 0.75, "place": 0.0, "identifier": 0.0, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 700}}}, "reference_key_id": "keys/synthetic/syn_03006_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "0c54b5424952f31a24ef521de33c53e46d87191f991ea8eefeaec650a7d38f30", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "f89a93a4a11a82f977f04423", "attempt_id": "93f3414dd1e7", "run_group_id": "E4_core:A5|repairable|qwen3b:h4", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:05:48.518732Z", "task_family": "synthetic", "document_id": "syn_03006_2000", "document_hash": "89be5e77e0ca114963e9ffdf8dc38fe5610f13a78ac29e7a1956cd9124ce5217", "question_id": "q05_multi_hop", "split": "test", "cluster_id": "syn_03006_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 4, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1890, "query_tokens": 407, "message_tokens": 314, "message_utf8_bytes": 748, "cache_tensor_bytes": null, "serialized_payload_bytes": 748, "side_information_bytes": 0, "fetched_bytes": 1800, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 63798.83479292039, "hops": {"1": 22600.32433399465, "2": 13699.429792002775, "3": 13789.627166930586, "4": 13709.453499992378}, "receiver_prefill_ms": 454.8720410093665, "ttft_ms": 28.01075007300824, "decode_ms": 39.988124975934625}, "total_wall_ms": 64322.14683492202, "ttft_ms": 28.01075007300824, "input_usage": {"lookups": 3}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 64, "facts_retained": 15, "retention_rate": 0.234375, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.2, "amount": 0.16666666666666666, "status": 0.75, "place": 0.0, "identifier": 0.0, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 700}, "ledger_by_hop": {"1": {"facts_total": 64, "facts_retained": 10, "retention_rate": 0.15625, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.2, "amount": 0.16666666666666666, "status": 0.3333333333333333, "place": 0.0, "identifier": 0.0, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 669}, "2": {"facts_total": 64, "facts_retained": 15, "retention_rate": 0.234375, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.2, "amount": 0.16666666666666666, "status": 0.75, "place": 0.0, "identifier": 0.0, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 700}, "3": {"facts_total": 64, "facts_retained": 15, "retention_rate": 0.234375, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.2, "amount": 0.16666666666666666, "status": 0.75, "place": 0.0, "identifier": 0.0, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 700}, "4": {"facts_total": 64, "facts_retained": 15, "retention_rate": 0.234375, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.2, "amount": 0.16666666666666666, "status": 0.75, "place": 0.0, "identifier": 0.0, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 700}}}, "reference_key_id": "keys/synthetic/syn_03006_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "0c54b5424952f31a24ef521de33c53e46d87191f991ea8eefeaec650a7d38f30", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "5b08147e720ac478f8e75f88", "attempt_id": "a9e0b3c0e472", "run_group_id": "E4_core:A5|repairable|qwen3b:h4", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:05:49.059004Z", "task_family": "synthetic", "document_id": "syn_03006_2000", "document_hash": "89be5e77e0ca114963e9ffdf8dc38fe5610f13a78ac29e7a1956cd9124ce5217", "question_id": "q06_negation", "split": "test", "cluster_id": "syn_03006_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 4, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1890, "query_tokens": 395, "message_tokens": 314, "message_utf8_bytes": 748, "cache_tensor_bytes": null, "serialized_payload_bytes": 748, "side_information_bytes": 0, "fetched_bytes": 1800, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 63798.83479292039, "hops": {"1": 22600.32433399465, "2": 13699.429792002775, "3": 13789.627166930586, "4": 13709.453499992378}, "receiver_prefill_ms": 289.28462509065866, "ttft_ms": 35.086542018689215, "decode_ms": 38.610124960541725}, "total_wall_ms": 64162.40633488633, "ttft_ms": 35.086542018689215, "input_usage": {"lookups": 3}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 64, "facts_retained": 15, "retention_rate": 0.234375, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.2, "amount": 0.16666666666666666, "status": 0.75, "place": 0.0, "identifier": 0.0, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 700}, "ledger_by_hop": {"1": {"facts_total": 64, "facts_retained": 10, "retention_rate": 0.15625, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.2, "amount": 0.16666666666666666, "status": 0.3333333333333333, "place": 0.0, "identifier": 0.0, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 669}, "2": {"facts_total": 64, "facts_retained": 15, "retention_rate": 0.234375, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.2, "amount": 0.16666666666666666, "status": 0.75, "place": 0.0, "identifier": 0.0, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 700}, "3": {"facts_total": 64, "facts_retained": 15, "retention_rate": 0.234375, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.2, "amount": 0.16666666666666666, "status": 0.75, "place": 0.0, "identifier": 0.0, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 700}, "4": {"facts_total": 64, "facts_retained": 15, "retention_rate": 0.234375, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.2, "amount": 0.16666666666666666, "status": 0.75, "place": 0.0, "identifier": 0.0, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 700}}}, "reference_key_id": "keys/synthetic/syn_03006_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "0c54b5424952f31a24ef521de33c53e46d87191f991ea8eefeaec650a7d38f30", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "d725c737696e758ed30d12b4", "attempt_id": "7574a711cf43", "run_group_id": "E4_core:A5|repairable|qwen3b:h4", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:05:49.439643Z", "task_family": "synthetic", "document_id": "syn_03006_2000", "document_hash": "89be5e77e0ca114963e9ffdf8dc38fe5610f13a78ac29e7a1956cd9124ce5217", "question_id": "q07_negation", "split": "test", "cluster_id": "syn_03006_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 4, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1890, "query_tokens": 395, "message_tokens": 314, "message_utf8_bytes": 748, "cache_tensor_bytes": null, "serialized_payload_bytes": 748, "side_information_bytes": 0, "fetched_bytes": 1800, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 63798.83479292039, "hops": {"1": 22600.32433399465, "2": 13699.429792002775, "3": 13789.627166930586, "4": 13709.453499992378}, "receiver_prefill_ms": 298.8002080237493, "ttft_ms": 37.66300005372614, "decode_ms": 27.434916933998466}, "total_wall_ms": 64163.16025191918, "ttft_ms": 37.66300005372614, "input_usage": {"lookups": 3}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "no", "evidence_artifact": {"ledger": {"facts_total": 64, "facts_retained": 15, "retention_rate": 0.234375, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.2, "amount": 0.16666666666666666, "status": 0.75, "place": 0.0, "identifier": 0.0, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 700}, "ledger_by_hop": {"1": {"facts_total": 64, "facts_retained": 10, "retention_rate": 0.15625, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.2, "amount": 0.16666666666666666, "status": 0.3333333333333333, "place": 0.0, "identifier": 0.0, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 669}, "2": {"facts_total": 64, "facts_retained": 15, "retention_rate": 0.234375, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.2, "amount": 0.16666666666666666, "status": 0.75, "place": 0.0, "identifier": 0.0, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 700}, "3": {"facts_total": 64, "facts_retained": 15, "retention_rate": 0.234375, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.2, "amount": 0.16666666666666666, "status": 0.75, "place": 0.0, "identifier": 0.0, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 700}, "4": {"facts_total": 64, "facts_retained": 15, "retention_rate": 0.234375, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.2, "amount": 0.16666666666666666, "status": 0.75, "place": 0.0, "identifier": 0.0, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 700}}}, "reference_key_id": "keys/synthetic/syn_03006_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "0c54b5424952f31a24ef521de33c53e46d87191f991ea8eefeaec650a7d38f30", "output_hash": "9390298f3fb0c5b160498935d79cb139aef28e1c47358b4bbba61862b9c26e59"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "86dcc888cfe567950bcc8cc1", "attempt_id": "28f6ae01bc35", "run_group_id": "E4_core:A5|repairable|qwen3b:h4", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:05:49.822189Z", "task_family": "synthetic", "document_id": "syn_03006_2000", "document_hash": "89be5e77e0ca114963e9ffdf8dc38fe5610f13a78ac29e7a1956cd9124ce5217", "question_id": "q08_revision", "split": "test", "cluster_id": "syn_03006_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 4, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1890, "query_tokens": 412, "message_tokens": 314, "message_utf8_bytes": 748, "cache_tensor_bytes": null, "serialized_payload_bytes": 748, "side_information_bytes": 0, "fetched_bytes": 1800, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 63798.83479292039, "hops": {"1": 22600.32433399465, "2": 13699.429792002775, "3": 13789.627166930586, "4": 13709.453499992378}, "receiver_prefill_ms": 261.3213330041617, "ttft_ms": 57.67470900900662, "decode_ms": 505.3523340029642}, "total_wall_ms": 64623.72079293709, "ttft_ms": 57.67470900900662, "input_usage": {"lookups": 3}, "output_usage": {"output_tokens": 8}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "10107545", "evidence_artifact": {"ledger": {"facts_total": 64, "facts_retained": 15, "retention_rate": 0.234375, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.2, "amount": 0.16666666666666666, "status": 0.75, "place": 0.0, "identifier": 0.0, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 700}, "ledger_by_hop": {"1": {"facts_total": 64, "facts_retained": 10, "retention_rate": 0.15625, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.2, "amount": 0.16666666666666666, "status": 0.3333333333333333, "place": 0.0, "identifier": 0.0, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 669}, "2": {"facts_total": 64, "facts_retained": 15, "retention_rate": 0.234375, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.2, "amount": 0.16666666666666666, "status": 0.75, "place": 0.0, "identifier": 0.0, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 700}, "3": {"facts_total": 64, "facts_retained": 15, "retention_rate": 0.234375, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.2, "amount": 0.16666666666666666, "status": 0.75, "place": 0.0, "identifier": 0.0, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 700}, "4": {"facts_total": 64, "facts_retained": 15, "retention_rate": 0.234375, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.2, "amount": 0.16666666666666666, "status": 0.75, "place": 0.0, "identifier": 0.0, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 700}}}, "reference_key_id": "keys/synthetic/syn_03006_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "0c54b5424952f31a24ef521de33c53e46d87191f991ea8eefeaec650a7d38f30", "output_hash": "6faba9de4e29949bbdd8934f69fc2110f07076db6801b211aeca9351292537fe"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "9b1b34f67a10580356a18714", "attempt_id": "a8b57a8fe818", "run_group_id": "E4_core:A5|repairable|qwen3b:h4", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:05:50.665508Z", "task_family": "synthetic", "document_id": "syn_03006_2000", "document_hash": "89be5e77e0ca114963e9ffdf8dc38fe5610f13a78ac29e7a1956cd9124ce5217", "question_id": "q09_revision", "split": "test", "cluster_id": "syn_03006_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 4, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1890, "query_tokens": 405, "message_tokens": 314, "message_utf8_bytes": 748, "cache_tensor_bytes": null, "serialized_payload_bytes": 748, "side_information_bytes": 0, "fetched_bytes": 1800, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 63798.83479292039, "hops": {"1": 22600.32433399465, "2": 13699.429792002775, "3": 13789.627166930586, "4": 13709.453499992378}, "receiver_prefill_ms": 290.2377910213545, "ttft_ms": 38.69779105298221, "decode_ms": 296.79812490940094}, "total_wall_ms": 64425.03812594805, "ttft_ms": 38.69779105298221, "input_usage": {"lookups": 3}, "output_usage": {"output_tokens": 9}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "9,112,735", "evidence_artifact": {"ledger": {"facts_total": 64, "facts_retained": 15, "retention_rate": 0.234375, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.2, "amount": 0.16666666666666666, "status": 0.75, "place": 0.0, "identifier": 0.0, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 700}, "ledger_by_hop": {"1": {"facts_total": 64, "facts_retained": 10, "retention_rate": 0.15625, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.2, "amount": 0.16666666666666666, "status": 0.3333333333333333, "place": 0.0, "identifier": 0.0, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 669}, "2": {"facts_total": 64, "facts_retained": 15, "retention_rate": 0.234375, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.2, "amount": 0.16666666666666666, "status": 0.75, "place": 0.0, "identifier": 0.0, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 700}, "3": {"facts_total": 64, "facts_retained": 15, "retention_rate": 0.234375, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.2, "amount": 0.16666666666666666, "status": 0.75, "place": 0.0, "identifier": 0.0, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 700}, "4": {"facts_total": 64, "facts_retained": 15, "retention_rate": 0.234375, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.2, "amount": 0.16666666666666666, "status": 0.75, "place": 0.0, "identifier": 0.0, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 700}}}, "reference_key_id": "keys/synthetic/syn_03006_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "0c54b5424952f31a24ef521de33c53e46d87191f991ea8eefeaec650a7d38f30", "output_hash": "788fcda908c2029597420995b09d0cc919bfdbf7a28b68dcb0e331db7df7b342"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "9cd89ecac222528233fc26ad", "attempt_id": "966ee69c171e", "run_group_id": "E4_core:A5|repairable|qwen3b:h4", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:05:51.308435Z", "task_family": "synthetic", "document_id": "syn_03006_2000", "document_hash": "89be5e77e0ca114963e9ffdf8dc38fe5610f13a78ac29e7a1956cd9124ce5217", "question_id": "q10_temporal", "split": "test", "cluster_id": "syn_03006_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 4, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1890, "query_tokens": 414, "message_tokens": 314, "message_utf8_bytes": 748, "cache_tensor_bytes": null, "serialized_payload_bytes": 748, "side_information_bytes": 0, "fetched_bytes": 1800, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 63798.83479292039, "hops": {"1": 22600.32433399465, "2": 13699.429792002775, "3": 13789.627166930586, "4": 13709.453499992378}, "receiver_prefill_ms": 245.85920793469995, "ttft_ms": 53.96420799661428, "decode_ms": 41.52112500742078}, "total_wall_ms": 64141.039876965806, "ttft_ms": 53.96420799661428, "input_usage": {"lookups": 3}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 64, "facts_retained": 15, "retention_rate": 0.234375, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.2, "amount": 0.16666666666666666, "status": 0.75, "place": 0.0, "identifier": 0.0, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 700}, "ledger_by_hop": {"1": {"facts_total": 64, "facts_retained": 10, "retention_rate": 0.15625, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.2, "amount": 0.16666666666666666, "status": 0.3333333333333333, "place": 0.0, "identifier": 0.0, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 669}, "2": {"facts_total": 64, "facts_retained": 15, "retention_rate": 0.234375, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.2, "amount": 0.16666666666666666, "status": 0.75, "place": 0.0, "identifier": 0.0, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 700}, "3": {"facts_total": 64, "facts_retained": 15, "retention_rate": 0.234375, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.2, "amount": 0.16666666666666666, "status": 0.75, "place": 0.0, "identifier": 0.0, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 700}, "4": {"facts_total": 64, "facts_retained": 15, "retention_rate": 0.234375, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.2, "amount": 0.16666666666666666, "status": 0.75, "place": 0.0, "identifier": 0.0, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 700}}}, "reference_key_id": "keys/synthetic/syn_03006_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "0c54b5424952f31a24ef521de33c53e46d87191f991ea8eefeaec650a7d38f30", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "3d0caa197d3589cfbd641b40", "attempt_id": "8126aac9ada6", "run_group_id": "E4_core:A5|repairable|qwen3b:h4", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:05:51.668452Z", "task_family": "synthetic", "document_id": "syn_03006_2000", "document_hash": "89be5e77e0ca114963e9ffdf8dc38fe5610f13a78ac29e7a1956cd9124ce5217", "question_id": "q11_aggregation", "split": "test", "cluster_id": "syn_03006_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 4, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1890, "query_tokens": 407, "message_tokens": 314, "message_utf8_bytes": 748, "cache_tensor_bytes": null, "serialized_payload_bytes": 748, "side_information_bytes": 0, "fetched_bytes": 1800, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 63798.83479292039, "hops": {"1": 22600.32433399465, "2": 13699.429792002775, "3": 13789.627166930586, "4": 13709.453499992378}, "receiver_prefill_ms": 460.0380000192672, "ttft_ms": 35.50250001717359, "decode_ms": 42.006707983091474}, "total_wall_ms": 64336.89570892602, "ttft_ms": 35.50250001717359, "input_usage": {"lookups": 3}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 64, "facts_retained": 15, "retention_rate": 0.234375, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.2, "amount": 0.16666666666666666, "status": 0.75, "place": 0.0, "identifier": 0.0, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 700}, "ledger_by_hop": {"1": {"facts_total": 64, "facts_retained": 10, "retention_rate": 0.15625, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.2, "amount": 0.16666666666666666, "status": 0.3333333333333333, "place": 0.0, "identifier": 0.0, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 669}, "2": {"facts_total": 64, "facts_retained": 15, "retention_rate": 0.234375, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.2, "amount": 0.16666666666666666, "status": 0.75, "place": 0.0, "identifier": 0.0, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 700}, "3": {"facts_total": 64, "facts_retained": 15, "retention_rate": 0.234375, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.2, "amount": 0.16666666666666666, "status": 0.75, "place": 0.0, "identifier": 0.0, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 700}, "4": {"facts_total": 64, "facts_retained": 15, "retention_rate": 0.234375, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.2, "amount": 0.16666666666666666, "status": 0.75, "place": 0.0, "identifier": 0.0, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 700}}}, "reference_key_id": "keys/synthetic/syn_03006_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "0c54b5424952f31a24ef521de33c53e46d87191f991ea8eefeaec650a7d38f30", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "289025274dca2f7b9550d3ff", "attempt_id": "11cd1cb5bfbc", "run_group_id": "E4_core:A5|repairable|qwen3b:h4", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:05:52.224333Z", "task_family": "synthetic", "document_id": "syn_03006_2000", "document_hash": "89be5e77e0ca114963e9ffdf8dc38fe5610f13a78ac29e7a1956cd9124ce5217", "question_id": "q12_unit_conversion", "split": "test", "cluster_id": "syn_03006_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 4, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1890, "query_tokens": 404, "message_tokens": 314, "message_utf8_bytes": 748, "cache_tensor_bytes": null, "serialized_payload_bytes": 748, "side_information_bytes": 0, "fetched_bytes": 1800, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 63798.83479292039, "hops": {"1": 22600.32433399465, "2": 13699.429792002775, "3": 13789.627166930586, "4": 13709.453499992378}, "receiver_prefill_ms": 295.3469160711393, "ttft_ms": 40.893458062782884, "decode_ms": 283.1753749633208}, "total_wall_ms": 64418.74575195834, "ttft_ms": 40.893458062782884, "input_usage": {"lookups": 3}, "output_usage": {"output_tokens": 8}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "11940603", "evidence_artifact": {"ledger": {"facts_total": 64, "facts_retained": 15, "retention_rate": 0.234375, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.2, "amount": 0.16666666666666666, "status": 0.75, "place": 0.0, "identifier": 0.0, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 700}, "ledger_by_hop": {"1": {"facts_total": 64, "facts_retained": 10, "retention_rate": 0.15625, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.2, "amount": 0.16666666666666666, "status": 0.3333333333333333, "place": 0.0, "identifier": 0.0, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 669}, "2": {"facts_total": 64, "facts_retained": 15, "retention_rate": 0.234375, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.2, "amount": 0.16666666666666666, "status": 0.75, "place": 0.0, "identifier": 0.0, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 700}, "3": {"facts_total": 64, "facts_retained": 15, "retention_rate": 0.234375, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.2, "amount": 0.16666666666666666, "status": 0.75, "place": 0.0, "identifier": 0.0, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 700}, "4": {"facts_total": 64, "facts_retained": 15, "retention_rate": 0.234375, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.2, "amount": 0.16666666666666666, "status": 0.75, "place": 0.0, "identifier": 0.0, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 700}}}, "reference_key_id": "keys/synthetic/syn_03006_2000.json", "task_scores": {"correct": 0.0, "abstained": 0.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "0c54b5424952f31a24ef521de33c53e46d87191f991ea8eefeaec650a7d38f30", "output_hash": "5f8eb540c68b7e1d88f2b98f4e9d0d3f953dfa8debea39ac84f45aa2ee2bdf05"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "c1dd62827e9668768f3114cc", "attempt_id": "7b3bb5891bb6", "run_group_id": "E4_core:A5|repairable|qwen3b:h4", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:05:52.861507Z", "task_family": "synthetic", "document_id": "syn_03006_2000", "document_hash": "89be5e77e0ca114963e9ffdf8dc38fe5610f13a78ac29e7a1956cd9124ce5217", "question_id": "q13_unanswerable", "split": "test", "cluster_id": "syn_03006_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 4, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1890, "query_tokens": 408, "message_tokens": 314, "message_utf8_bytes": 748, "cache_tensor_bytes": null, "serialized_payload_bytes": 748, "side_information_bytes": 0, "fetched_bytes": 1800, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 63798.83479292039, "hops": {"1": 22600.32433399465, "2": 13699.429792002775, "3": 13789.627166930586, "4": 13709.453499992378}, "receiver_prefill_ms": 256.6029999870807, "ttft_ms": 65.62991696409881, "decode_ms": 62.028583022765815}, "total_wall_ms": 64183.65658482071, "ttft_ms": 65.62991696409881, "input_usage": {"lookups": 3}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 64, "facts_retained": 15, "retention_rate": 0.234375, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.2, "amount": 0.16666666666666666, "status": 0.75, "place": 0.0, "identifier": 0.0, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 700}, "ledger_by_hop": {"1": {"facts_total": 64, "facts_retained": 10, "retention_rate": 0.15625, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.2, "amount": 0.16666666666666666, "status": 0.3333333333333333, "place": 0.0, "identifier": 0.0, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 669}, "2": {"facts_total": 64, "facts_retained": 15, "retention_rate": 0.234375, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.2, "amount": 0.16666666666666666, "status": 0.75, "place": 0.0, "identifier": 0.0, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 700}, "3": {"facts_total": 64, "facts_retained": 15, "retention_rate": 0.234375, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.2, "amount": 0.16666666666666666, "status": 0.75, "place": 0.0, "identifier": 0.0, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 700}, "4": {"facts_total": 64, "facts_retained": 15, "retention_rate": 0.234375, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.2, "amount": 0.16666666666666666, "status": 0.75, "place": 0.0, "identifier": 0.0, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 700}}}, "reference_key_id": "keys/synthetic/syn_03006_2000.json", "task_scores": {"correct": 1.0, "abstained": 1.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "0c54b5424952f31a24ef521de33c53e46d87191f991ea8eefeaec650a7d38f30", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "21be76b9876115bbfcee35a1", "attempt_id": "cbb7e37a39dd", "run_group_id": "E4_core:A5|repairable|qwen3b:h4", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:05:53.263929Z", "task_family": "synthetic", "document_id": "syn_03006_2000", "document_hash": "89be5e77e0ca114963e9ffdf8dc38fe5610f13a78ac29e7a1956cd9124ce5217", "question_id": "q14_unanswerable", "split": "test", "cluster_id": "syn_03006_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 4, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1890, "query_tokens": 409, "message_tokens": 314, "message_utf8_bytes": 748, "cache_tensor_bytes": null, "serialized_payload_bytes": 748, "side_information_bytes": 0, "fetched_bytes": 1800, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 63798.83479292039, "hops": {"1": 22600.32433399465, "2": 13699.429792002775, "3": 13789.627166930586, "4": 13709.453499992378}, "receiver_prefill_ms": 453.363792039454, "ttft_ms": 24.426375050097704, "decode_ms": 37.07579092588276}, "total_wall_ms": 64314.199334941804, "ttft_ms": 24.426375050097704, "input_usage": {"lookups": 3}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 64, "facts_retained": 15, "retention_rate": 0.234375, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.2, "amount": 0.16666666666666666, "status": 0.75, "place": 0.0, "identifier": 0.0, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 700}, "ledger_by_hop": {"1": {"facts_total": 64, "facts_retained": 10, "retention_rate": 0.15625, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.2, "amount": 0.16666666666666666, "status": 0.3333333333333333, "place": 0.0, "identifier": 0.0, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 669}, "2": {"facts_total": 64, "facts_retained": 15, "retention_rate": 0.234375, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.2, "amount": 0.16666666666666666, "status": 0.75, "place": 0.0, "identifier": 0.0, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 700}, "3": {"facts_total": 64, "facts_retained": 15, "retention_rate": 0.234375, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.2, "amount": 0.16666666666666666, "status": 0.75, "place": 0.0, "identifier": 0.0, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 700}, "4": {"facts_total": 64, "facts_retained": 15, "retention_rate": 0.234375, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.2, "amount": 0.16666666666666666, "status": 0.75, "place": 0.0, "identifier": 0.0, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 700}}}, "reference_key_id": "keys/synthetic/syn_03006_2000.json", "task_scores": {"correct": 1.0, "abstained": 1.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "0c54b5424952f31a24ef521de33c53e46d87191f991ea8eefeaec650a7d38f30", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "0a703a037fb84fcf3b4a4ec5", "attempt_id": "4be4a0c8a242", "run_group_id": "E4_core:A5|repairable|qwen3b:h8", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:06:49.122531Z", "task_family": "synthetic", "document_id": "syn_03006_2000", "document_hash": "89be5e77e0ca114963e9ffdf8dc38fe5610f13a78ac29e7a1956cd9124ce5217", "question_id": "q00_single_fact", "split": "test", "cluster_id": "syn_03006_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 8, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1890, "query_tokens": 396, "message_tokens": 314, "message_utf8_bytes": 748, "cache_tensor_bytes": null, "serialized_payload_bytes": 748, "side_information_bytes": 0, "fetched_bytes": 4200, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 119119.30483486503, "hops": {"1": 22600.32433399465, "2": 13699.429792002775, "3": 13789.627166930586, "4": 13709.453499992378, "5": 13876.501917024143, "6": 13709.757833043113, "7": 13953.084124950692, "8": 13781.126166926697}, "receiver_prefill_ms": 255.687708966434, "ttft_ms": 43.03245909977704, "decode_ms": 59.479791088961065}, "total_wall_ms": 119478.07112685405, "ttft_ms": 43.03245909977704, "input_usage": {"lookups": 7}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 64, "facts_retained": 15, "retention_rate": 0.234375, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.2, "amount": 0.16666666666666666, "status": 0.75, "place": 0.0, "identifier": 0.0, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 700}, "ledger_by_hop": {"1": {"facts_total": 64, "facts_retained": 10, "retention_rate": 0.15625, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.2, "amount": 0.16666666666666666, "status": 0.3333333333333333, "place": 0.0, "identifier": 0.0, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 669}, "2": {"facts_total": 64, "facts_retained": 15, "retention_rate": 0.234375, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.2, "amount": 0.16666666666666666, "status": 0.75, "place": 0.0, "identifier": 0.0, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 700}, "3": {"facts_total": 64, "facts_retained": 15, "retention_rate": 0.234375, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.2, "amount": 0.16666666666666666, "status": 0.75, "place": 0.0, "identifier": 0.0, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 700}, "4": {"facts_total": 64, "facts_retained": 15, "retention_rate": 0.234375, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.2, "amount": 0.16666666666666666, "status": 0.75, "place": 0.0, "identifier": 0.0, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 700}, "5": {"facts_total": 64, "facts_retained": 15, "retention_rate": 0.234375, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.2, "amount": 0.16666666666666666, "status": 0.75, "place": 0.0, "identifier": 0.0, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 700}, "6": {"facts_total": 64, "facts_retained": 15, "retention_rate": 0.234375, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.2, "amount": 0.16666666666666666, "status": 0.75, "place": 0.0, "identifier": 0.0, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 700}, "7": {"facts_total": 64, "facts_retained": 15, "retention_rate": 0.234375, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.2, "amount": 0.16666666666666666, "status": 0.75, "place": 0.0, "identifier": 0.0, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 700}, "8": {"facts_total": 64, "facts_retained": 15, "retention_rate": 0.234375, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.2, "amount": 0.16666666666666666, "status": 0.75, "place": 0.0, "identifier": 0.0, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 700}}}, "reference_key_id": "keys/synthetic/syn_03006_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "0c54b5424952f31a24ef521de33c53e46d87191f991ea8eefeaec650a7d38f30", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "3b17515393b97953e582ba38", "attempt_id": "fb270000ab98", "run_group_id": "E4_core:A5|repairable|qwen3b:h8", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:06:49.499592Z", "task_family": "synthetic", "document_id": "syn_03006_2000", "document_hash": "89be5e77e0ca114963e9ffdf8dc38fe5610f13a78ac29e7a1956cd9124ce5217", "question_id": "q01_single_fact", "split": "test", "cluster_id": "syn_03006_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 8, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1890, "query_tokens": 401, "message_tokens": 314, "message_utf8_bytes": 748, "cache_tensor_bytes": null, "serialized_payload_bytes": 748, "side_information_bytes": 0, "fetched_bytes": 4200, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 119119.30483486503, "hops": {"1": 22600.32433399465, "2": 13699.429792002775, "3": 13789.627166930586, "4": 13709.453499992378, "5": 13876.501917024143, "6": 13709.757833043113, "7": 13953.084124950692, "8": 13781.126166926697}, "receiver_prefill_ms": 410.2992090629414, "ttft_ms": 41.86291596852243, "decode_ms": 24.21054197475314}, "total_wall_ms": 119596.3175437646, "ttft_ms": 41.86291596852243, "input_usage": {"lookups": 7}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 64, "facts_retained": 15, "retention_rate": 0.234375, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.2, "amount": 0.16666666666666666, "status": 0.75, "place": 0.0, "identifier": 0.0, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 700}, "ledger_by_hop": {"1": {"facts_total": 64, "facts_retained": 10, "retention_rate": 0.15625, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.2, "amount": 0.16666666666666666, "status": 0.3333333333333333, "place": 0.0, "identifier": 0.0, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 669}, "2": {"facts_total": 64, "facts_retained": 15, "retention_rate": 0.234375, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.2, "amount": 0.16666666666666666, "status": 0.75, "place": 0.0, "identifier": 0.0, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 700}, "3": {"facts_total": 64, "facts_retained": 15, "retention_rate": 0.234375, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.2, "amount": 0.16666666666666666, "status": 0.75, "place": 0.0, "identifier": 0.0, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 700}, "4": {"facts_total": 64, "facts_retained": 15, "retention_rate": 0.234375, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.2, "amount": 0.16666666666666666, "status": 0.75, "place": 0.0, "identifier": 0.0, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 700}, "5": {"facts_total": 64, "facts_retained": 15, "retention_rate": 0.234375, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.2, "amount": 0.16666666666666666, "status": 0.75, "place": 0.0, "identifier": 0.0, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 700}, "6": {"facts_total": 64, "facts_retained": 15, "retention_rate": 0.234375, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.2, "amount": 0.16666666666666666, "status": 0.75, "place": 0.0, "identifier": 0.0, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 700}, "7": {"facts_total": 64, "facts_retained": 15, "retention_rate": 0.234375, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.2, "amount": 0.16666666666666666, "status": 0.75, "place": 0.0, "identifier": 0.0, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 700}, "8": {"facts_total": 64, "facts_retained": 15, "retention_rate": 0.234375, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.2, "amount": 0.16666666666666666, "status": 0.75, "place": 0.0, "identifier": 0.0, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 700}}}, "reference_key_id": "keys/synthetic/syn_03006_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "0c54b5424952f31a24ef521de33c53e46d87191f991ea8eefeaec650a7d38f30", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "e018a56eae0d593fe5c2f30e", "attempt_id": "282dbc8f3f88", "run_group_id": "E4_core:A5|repairable|qwen3b:h8", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:06:49.994782Z", "task_family": "synthetic", "document_id": "syn_03006_2000", "document_hash": "89be5e77e0ca114963e9ffdf8dc38fe5610f13a78ac29e7a1956cd9124ce5217", "question_id": "q02_exact_copy", "split": "test", "cluster_id": "syn_03006_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 8, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1890, "query_tokens": 400, "message_tokens": 314, "message_utf8_bytes": 748, "cache_tensor_bytes": null, "serialized_payload_bytes": 748, "side_information_bytes": 0, "fetched_bytes": 4200, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 119119.30483486503, "hops": {"1": 22600.32433399465, "2": 13699.429792002775, "3": 13789.627166930586, "4": 13709.453499992378, "5": 13876.501917024143, "6": 13709.757833043113, "7": 13953.084124950692, "8": 13781.126166926697}, "receiver_prefill_ms": 290.82383296918124, "ttft_ms": 41.388334007933736, "decode_ms": 28.129042009823024}, "total_wall_ms": 119480.25312682148, "ttft_ms": 41.388334007933736, "input_usage": {"lookups": 7}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 64, "facts_retained": 15, "retention_rate": 0.234375, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.2, "amount": 0.16666666666666666, "status": 0.75, "place": 0.0, "identifier": 0.0, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 700}, "ledger_by_hop": {"1": {"facts_total": 64, "facts_retained": 10, "retention_rate": 0.15625, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.2, "amount": 0.16666666666666666, "status": 0.3333333333333333, "place": 0.0, "identifier": 0.0, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 669}, "2": {"facts_total": 64, "facts_retained": 15, "retention_rate": 0.234375, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.2, "amount": 0.16666666666666666, "status": 0.75, "place": 0.0, "identifier": 0.0, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 700}, "3": {"facts_total": 64, "facts_retained": 15, "retention_rate": 0.234375, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.2, "amount": 0.16666666666666666, "status": 0.75, "place": 0.0, "identifier": 0.0, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 700}, "4": {"facts_total": 64, "facts_retained": 15, "retention_rate": 0.234375, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.2, "amount": 0.16666666666666666, "status": 0.75, "place": 0.0, "identifier": 0.0, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 700}, "5": {"facts_total": 64, "facts_retained": 15, "retention_rate": 0.234375, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.2, "amount": 0.16666666666666666, "status": 0.75, "place": 0.0, "identifier": 0.0, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 700}, "6": {"facts_total": 64, "facts_retained": 15, "retention_rate": 0.234375, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.2, "amount": 0.16666666666666666, "status": 0.75, "place": 0.0, "identifier": 0.0, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 700}, "7": {"facts_total": 64, "facts_retained": 15, "retention_rate": 0.234375, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.2, "amount": 0.16666666666666666, "status": 0.75, "place": 0.0, "identifier": 0.0, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 700}, "8": {"facts_total": 64, "facts_retained": 15, "retention_rate": 0.234375, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.2, "amount": 0.16666666666666666, "status": 0.75, "place": 0.0, "identifier": 0.0, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 700}}}, "reference_key_id": "keys/synthetic/syn_03006_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "0c54b5424952f31a24ef521de33c53e46d87191f991ea8eefeaec650a7d38f30", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "165eecaac66d8c8a22560855", "attempt_id": "2318e9496f84", "run_group_id": "E4_core:A5|repairable|qwen3b:h8", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:06:50.374064Z", "task_family": "synthetic", "document_id": "syn_03006_2000", "document_hash": "89be5e77e0ca114963e9ffdf8dc38fe5610f13a78ac29e7a1956cd9124ce5217", "question_id": "q03_exact_copy", "split": "test", "cluster_id": "syn_03006_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 8, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1890, "query_tokens": 395, "message_tokens": 314, "message_utf8_bytes": 748, "cache_tensor_bytes": null, "serialized_payload_bytes": 748, "side_information_bytes": 0, "fetched_bytes": 4200, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 119119.30483486503, "hops": {"1": 22600.32433399465, "2": 13699.429792002775, "3": 13789.627166930586, "4": 13709.453499992378, "5": 13876.501917024143, "6": 13709.757833043113, "7": 13953.084124950692, "8": 13781.126166926697}, "receiver_prefill_ms": 286.63945803418756, "ttft_ms": 19.687250023707747, "decode_ms": 35.39558290503919}, "total_wall_ms": 119461.47504379041, "ttft_ms": 19.687250023707747, "input_usage": {"lookups": 7}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 64, "facts_retained": 15, "retention_rate": 0.234375, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.2, "amount": 0.16666666666666666, "status": 0.75, "place": 0.0, "identifier": 0.0, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 700}, "ledger_by_hop": {"1": {"facts_total": 64, "facts_retained": 10, "retention_rate": 0.15625, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.2, "amount": 0.16666666666666666, "status": 0.3333333333333333, "place": 0.0, "identifier": 0.0, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 669}, "2": {"facts_total": 64, "facts_retained": 15, "retention_rate": 0.234375, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.2, "amount": 0.16666666666666666, "status": 0.75, "place": 0.0, "identifier": 0.0, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 700}, "3": {"facts_total": 64, "facts_retained": 15, "retention_rate": 0.234375, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.2, "amount": 0.16666666666666666, "status": 0.75, "place": 0.0, "identifier": 0.0, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 700}, "4": {"facts_total": 64, "facts_retained": 15, "retention_rate": 0.234375, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.2, "amount": 0.16666666666666666, "status": 0.75, "place": 0.0, "identifier": 0.0, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 700}, "5": {"facts_total": 64, "facts_retained": 15, "retention_rate": 0.234375, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.2, "amount": 0.16666666666666666, "status": 0.75, "place": 0.0, "identifier": 0.0, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 700}, "6": {"facts_total": 64, "facts_retained": 15, "retention_rate": 0.234375, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.2, "amount": 0.16666666666666666, "status": 0.75, "place": 0.0, "identifier": 0.0, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 700}, "7": {"facts_total": 64, "facts_retained": 15, "retention_rate": 0.234375, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.2, "amount": 0.16666666666666666, "status": 0.75, "place": 0.0, "identifier": 0.0, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 700}, "8": {"facts_total": 64, "facts_retained": 15, "retention_rate": 0.234375, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.2, "amount": 0.16666666666666666, "status": 0.75, "place": 0.0, "identifier": 0.0, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 700}}}, "reference_key_id": "keys/synthetic/syn_03006_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "0c54b5424952f31a24ef521de33c53e46d87191f991ea8eefeaec650a7d38f30", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "b831e7925c248954687fc218", "attempt_id": "af908e307908", "run_group_id": "E4_core:A5|repairable|qwen3b:h8", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:06:50.733256Z", "task_family": "synthetic", "document_id": "syn_03006_2000", "document_hash": "89be5e77e0ca114963e9ffdf8dc38fe5610f13a78ac29e7a1956cd9124ce5217", "question_id": "q04_multi_hop", "split": "test", "cluster_id": "syn_03006_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 8, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1890, "query_tokens": 404, "message_tokens": 314, "message_utf8_bytes": 748, "cache_tensor_bytes": null, "serialized_payload_bytes": 748, "side_information_bytes": 0, "fetched_bytes": 4200, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 119119.30483486503, "hops": {"1": 22600.32433399465, "2": 13699.429792002775, "3": 13789.627166930586, "4": 13709.453499992378, "5": 13876.501917024143, "6": 13709.757833043113, "7": 13953.084124950692, "8": 13781.126166926697}, "receiver_prefill_ms": 289.63695804122835, "ttft_ms": 61.338083003647625, "decode_ms": 65.42716699186713}, "total_wall_ms": 119536.24779381789, "ttft_ms": 61.338083003647625, "input_usage": {"lookups": 7}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 64, "facts_retained": 15, "retention_rate": 0.234375, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.2, "amount": 0.16666666666666666, "status": 0.75, "place": 0.0, "identifier": 0.0, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 700}, "ledger_by_hop": {"1": {"facts_total": 64, "facts_retained": 10, "retention_rate": 0.15625, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.2, "amount": 0.16666666666666666, "status": 0.3333333333333333, "place": 0.0, "identifier": 0.0, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 669}, "2": {"facts_total": 64, "facts_retained": 15, "retention_rate": 0.234375, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.2, "amount": 0.16666666666666666, "status": 0.75, "place": 0.0, "identifier": 0.0, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 700}, "3": {"facts_total": 64, "facts_retained": 15, "retention_rate": 0.234375, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.2, "amount": 0.16666666666666666, "status": 0.75, "place": 0.0, "identifier": 0.0, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 700}, "4": {"facts_total": 64, "facts_retained": 15, "retention_rate": 0.234375, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.2, "amount": 0.16666666666666666, "status": 0.75, "place": 0.0, "identifier": 0.0, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 700}, "5": {"facts_total": 64, "facts_retained": 15, "retention_rate": 0.234375, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.2, "amount": 0.16666666666666666, "status": 0.75, "place": 0.0, "identifier": 0.0, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 700}, "6": {"facts_total": 64, "facts_retained": 15, "retention_rate": 0.234375, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.2, "amount": 0.16666666666666666, "status": 0.75, "place": 0.0, "identifier": 0.0, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 700}, "7": {"facts_total": 64, "facts_retained": 15, "retention_rate": 0.234375, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.2, "amount": 0.16666666666666666, "status": 0.75, "place": 0.0, "identifier": 0.0, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 700}, "8": {"facts_total": 64, "facts_retained": 15, "retention_rate": 0.234375, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.2, "amount": 0.16666666666666666, "status": 0.75, "place": 0.0, "identifier": 0.0, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 700}}}, "reference_key_id": "keys/synthetic/syn_03006_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "0c54b5424952f31a24ef521de33c53e46d87191f991ea8eefeaec650a7d38f30", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "4e6a6766f576fad24d743421", "attempt_id": "2dc81054d208", "run_group_id": "E4_core:A5|repairable|qwen3b:h8", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:06:51.167846Z", "task_family": "synthetic", "document_id": "syn_03006_2000", "document_hash": "89be5e77e0ca114963e9ffdf8dc38fe5610f13a78ac29e7a1956cd9124ce5217", "question_id": "q05_multi_hop", "split": "test", "cluster_id": "syn_03006_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 8, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1890, "query_tokens": 407, "message_tokens": 314, "message_utf8_bytes": 748, "cache_tensor_bytes": null, "serialized_payload_bytes": 748, "side_information_bytes": 0, "fetched_bytes": 4200, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 119119.30483486503, "hops": {"1": 22600.32433399465, "2": 13699.429792002775, "3": 13789.627166930586, "4": 13709.453499992378, "5": 13876.501917024143, "6": 13709.757833043113, "7": 13953.084124950692, "8": 13781.126166926697}, "receiver_prefill_ms": 445.0003750389442, "ttft_ms": 30.903459060937166, "decode_ms": 34.34679098427296}, "total_wall_ms": 119630.12158486526, "ttft_ms": 30.903459060937166, "input_usage": {"lookups": 7}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 64, "facts_retained": 15, "retention_rate": 0.234375, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.2, "amount": 0.16666666666666666, "status": 0.75, "place": 0.0, "identifier": 0.0, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 700}, "ledger_by_hop": {"1": {"facts_total": 64, "facts_retained": 10, "retention_rate": 0.15625, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.2, "amount": 0.16666666666666666, "status": 0.3333333333333333, "place": 0.0, "identifier": 0.0, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 669}, "2": {"facts_total": 64, "facts_retained": 15, "retention_rate": 0.234375, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.2, "amount": 0.16666666666666666, "status": 0.75, "place": 0.0, "identifier": 0.0, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 700}, "3": {"facts_total": 64, "facts_retained": 15, "retention_rate": 0.234375, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.2, "amount": 0.16666666666666666, "status": 0.75, "place": 0.0, "identifier": 0.0, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 700}, "4": {"facts_total": 64, "facts_retained": 15, "retention_rate": 0.234375, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.2, "amount": 0.16666666666666666, "status": 0.75, "place": 0.0, "identifier": 0.0, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 700}, "5": {"facts_total": 64, "facts_retained": 15, "retention_rate": 0.234375, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.2, "amount": 0.16666666666666666, "status": 0.75, "place": 0.0, "identifier": 0.0, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 700}, "6": {"facts_total": 64, "facts_retained": 15, "retention_rate": 0.234375, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.2, "amount": 0.16666666666666666, "status": 0.75, "place": 0.0, "identifier": 0.0, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 700}, "7": {"facts_total": 64, "facts_retained": 15, "retention_rate": 0.234375, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.2, "amount": 0.16666666666666666, "status": 0.75, "place": 0.0, "identifier": 0.0, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 700}, "8": {"facts_total": 64, "facts_retained": 15, "retention_rate": 0.234375, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.2, "amount": 0.16666666666666666, "status": 0.75, "place": 0.0, "identifier": 0.0, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 700}}}, "reference_key_id": "keys/synthetic/syn_03006_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "0c54b5424952f31a24ef521de33c53e46d87191f991ea8eefeaec650a7d38f30", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "96735446823034305a8a3bfd", "attempt_id": "547bca3c1d26", "run_group_id": "E4_core:A5|repairable|qwen3b:h8", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:06:51.697409Z", "task_family": "synthetic", "document_id": "syn_03006_2000", "document_hash": "89be5e77e0ca114963e9ffdf8dc38fe5610f13a78ac29e7a1956cd9124ce5217", "question_id": "q06_negation", "split": "test", "cluster_id": "syn_03006_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 8, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1890, "query_tokens": 395, "message_tokens": 314, "message_utf8_bytes": 748, "cache_tensor_bytes": null, "serialized_payload_bytes": 748, "side_information_bytes": 0, "fetched_bytes": 4200, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 119119.30483486503, "hops": {"1": 22600.32433399465, "2": 13699.429792002775, "3": 13789.627166930586, "4": 13709.453499992378, "5": 13876.501917024143, "6": 13709.757833043113, "7": 13953.084124950692, "8": 13781.126166926697}, "receiver_prefill_ms": 299.4697920512408, "ttft_ms": 42.03208303079009, "decode_ms": 29.493500012904406}, "total_wall_ms": 119490.7462928677, "ttft_ms": 42.03208303079009, "input_usage": {"lookups": 7}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 64, "facts_retained": 15, "retention_rate": 0.234375, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.2, "amount": 0.16666666666666666, "status": 0.75, "place": 0.0, "identifier": 0.0, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 700}, "ledger_by_hop": {"1": {"facts_total": 64, "facts_retained": 10, "retention_rate": 0.15625, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.2, "amount": 0.16666666666666666, "status": 0.3333333333333333, "place": 0.0, "identifier": 0.0, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 669}, "2": {"facts_total": 64, "facts_retained": 15, "retention_rate": 0.234375, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.2, "amount": 0.16666666666666666, "status": 0.75, "place": 0.0, "identifier": 0.0, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 700}, "3": {"facts_total": 64, "facts_retained": 15, "retention_rate": 0.234375, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.2, "amount": 0.16666666666666666, "status": 0.75, "place": 0.0, "identifier": 0.0, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 700}, "4": {"facts_total": 64, "facts_retained": 15, "retention_rate": 0.234375, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.2, "amount": 0.16666666666666666, "status": 0.75, "place": 0.0, "identifier": 0.0, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 700}, "5": {"facts_total": 64, "facts_retained": 15, "retention_rate": 0.234375, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.2, "amount": 0.16666666666666666, "status": 0.75, "place": 0.0, "identifier": 0.0, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 700}, "6": {"facts_total": 64, "facts_retained": 15, "retention_rate": 0.234375, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.2, "amount": 0.16666666666666666, "status": 0.75, "place": 0.0, "identifier": 0.0, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 700}, "7": {"facts_total": 64, "facts_retained": 15, "retention_rate": 0.234375, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.2, "amount": 0.16666666666666666, "status": 0.75, "place": 0.0, "identifier": 0.0, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 700}, "8": {"facts_total": 64, "facts_retained": 15, "retention_rate": 0.234375, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.2, "amount": 0.16666666666666666, "status": 0.75, "place": 0.0, "identifier": 0.0, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 700}}}, "reference_key_id": "keys/synthetic/syn_03006_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "0c54b5424952f31a24ef521de33c53e46d87191f991ea8eefeaec650a7d38f30", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "332779ffbf1f1273e47eb9bf", "attempt_id": "bf2b4ea16736", "run_group_id": "E4_core:A5|repairable|qwen3b:h8", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:06:52.086565Z", "task_family": "synthetic", "document_id": "syn_03006_2000", "document_hash": "89be5e77e0ca114963e9ffdf8dc38fe5610f13a78ac29e7a1956cd9124ce5217", "question_id": "q07_negation", "split": "test", "cluster_id": "syn_03006_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 8, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1890, "query_tokens": 395, "message_tokens": 314, "message_utf8_bytes": 748, "cache_tensor_bytes": null, "serialized_payload_bytes": 748, "side_information_bytes": 0, "fetched_bytes": 4200, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 119119.30483486503, "hops": {"1": 22600.32433399465, "2": 13699.429792002775, "3": 13789.627166930586, "4": 13709.453499992378, "5": 13876.501917024143, "6": 13709.757833043113, "7": 13953.084124950692, "8": 13781.126166926697}, "receiver_prefill_ms": 282.26358303800225, "ttft_ms": 29.016625019721687, "decode_ms": 41.10070795286447}, "total_wall_ms": 119472.22450189292, "ttft_ms": 29.016625019721687, "input_usage": {"lookups": 7}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "no", "evidence_artifact": {"ledger": {"facts_total": 64, "facts_retained": 15, "retention_rate": 0.234375, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.2, "amount": 0.16666666666666666, "status": 0.75, "place": 0.0, "identifier": 0.0, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 700}, "ledger_by_hop": {"1": {"facts_total": 64, "facts_retained": 10, "retention_rate": 0.15625, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.2, "amount": 0.16666666666666666, "status": 0.3333333333333333, "place": 0.0, "identifier": 0.0, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 669}, "2": {"facts_total": 64, "facts_retained": 15, "retention_rate": 0.234375, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.2, "amount": 0.16666666666666666, "status": 0.75, "place": 0.0, "identifier": 0.0, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 700}, "3": {"facts_total": 64, "facts_retained": 15, "retention_rate": 0.234375, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.2, "amount": 0.16666666666666666, "status": 0.75, "place": 0.0, "identifier": 0.0, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 700}, "4": {"facts_total": 64, "facts_retained": 15, "retention_rate": 0.234375, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.2, "amount": 0.16666666666666666, "status": 0.75, "place": 0.0, "identifier": 0.0, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 700}, "5": {"facts_total": 64, "facts_retained": 15, "retention_rate": 0.234375, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.2, "amount": 0.16666666666666666, "status": 0.75, "place": 0.0, "identifier": 0.0, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 700}, "6": {"facts_total": 64, "facts_retained": 15, "retention_rate": 0.234375, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.2, "amount": 0.16666666666666666, "status": 0.75, "place": 0.0, "identifier": 0.0, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 700}, "7": {"facts_total": 64, "facts_retained": 15, "retention_rate": 0.234375, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.2, "amount": 0.16666666666666666, "status": 0.75, "place": 0.0, "identifier": 0.0, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 700}, "8": {"facts_total": 64, "facts_retained": 15, "retention_rate": 0.234375, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.2, "amount": 0.16666666666666666, "status": 0.75, "place": 0.0, "identifier": 0.0, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 700}}}, "reference_key_id": "keys/synthetic/syn_03006_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "0c54b5424952f31a24ef521de33c53e46d87191f991ea8eefeaec650a7d38f30", "output_hash": "9390298f3fb0c5b160498935d79cb139aef28e1c47358b4bbba61862b9c26e59"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "de53842d452bb749221652a3", "attempt_id": "03e48501f09d", "run_group_id": "E4_core:A5|repairable|qwen3b:h8", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:06:52.456920Z", "task_family": "synthetic", "document_id": "syn_03006_2000", "document_hash": "89be5e77e0ca114963e9ffdf8dc38fe5610f13a78ac29e7a1956cd9124ce5217", "question_id": "q08_revision", "split": "test", "cluster_id": "syn_03006_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 8, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1890, "query_tokens": 412, "message_tokens": 314, "message_utf8_bytes": 748, "cache_tensor_bytes": null, "serialized_payload_bytes": 748, "side_information_bytes": 0, "fetched_bytes": 4200, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 119119.30483486503, "hops": {"1": 22600.32433399465, "2": 13699.429792002775, "3": 13789.627166930586, "4": 13709.453499992378, "5": 13876.501917024143, "6": 13709.757833043113, "7": 13953.084124950692, "8": 13781.126166926697}, "receiver_prefill_ms": 254.5762499794364, "ttft_ms": 60.49929198343307, "decode_ms": 487.1170000405982}, "total_wall_ms": 119922.27337684017, "ttft_ms": 60.49929198343307, "input_usage": {"lookups": 7}, "output_usage": {"output_tokens": 8}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "10107545", "evidence_artifact": {"ledger": {"facts_total": 64, "facts_retained": 15, "retention_rate": 0.234375, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.2, "amount": 0.16666666666666666, "status": 0.75, "place": 0.0, "identifier": 0.0, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 700}, "ledger_by_hop": {"1": {"facts_total": 64, "facts_retained": 10, "retention_rate": 0.15625, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.2, "amount": 0.16666666666666666, "status": 0.3333333333333333, "place": 0.0, "identifier": 0.0, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 669}, "2": {"facts_total": 64, "facts_retained": 15, "retention_rate": 0.234375, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.2, "amount": 0.16666666666666666, "status": 0.75, "place": 0.0, "identifier": 0.0, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 700}, "3": {"facts_total": 64, "facts_retained": 15, "retention_rate": 0.234375, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.2, "amount": 0.16666666666666666, "status": 0.75, "place": 0.0, "identifier": 0.0, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 700}, "4": {"facts_total": 64, "facts_retained": 15, "retention_rate": 0.234375, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.2, "amount": 0.16666666666666666, "status": 0.75, "place": 0.0, "identifier": 0.0, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 700}, "5": {"facts_total": 64, "facts_retained": 15, "retention_rate": 0.234375, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.2, "amount": 0.16666666666666666, "status": 0.75, "place": 0.0, "identifier": 0.0, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 700}, "6": {"facts_total": 64, "facts_retained": 15, "retention_rate": 0.234375, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.2, "amount": 0.16666666666666666, "status": 0.75, "place": 0.0, "identifier": 0.0, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 700}, "7": {"facts_total": 64, "facts_retained": 15, "retention_rate": 0.234375, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.2, "amount": 0.16666666666666666, "status": 0.75, "place": 0.0, "identifier": 0.0, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 700}, "8": {"facts_total": 64, "facts_retained": 15, "retention_rate": 0.234375, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.2, "amount": 0.16666666666666666, "status": 0.75, "place": 0.0, "identifier": 0.0, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 700}}}, "reference_key_id": "keys/synthetic/syn_03006_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "0c54b5424952f31a24ef521de33c53e46d87191f991ea8eefeaec650a7d38f30", "output_hash": "6faba9de4e29949bbdd8934f69fc2110f07076db6801b211aeca9351292537fe"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "37ca04952747d8a2b1b91f7d", "attempt_id": "20ec4c8a0e05", "run_group_id": "E4_core:A5|repairable|qwen3b:h8", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:06:53.276744Z", "task_family": "synthetic", "document_id": "syn_03006_2000", "document_hash": "89be5e77e0ca114963e9ffdf8dc38fe5610f13a78ac29e7a1956cd9124ce5217", "question_id": "q09_revision", "split": "test", "cluster_id": "syn_03006_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 8, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1890, "query_tokens": 405, "message_tokens": 314, "message_utf8_bytes": 748, "cache_tensor_bytes": null, "serialized_payload_bytes": 748, "side_information_bytes": 0, "fetched_bytes": 4200, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 119119.30483486503, "hops": {"1": 22600.32433399465, "2": 13699.429792002775, "3": 13789.627166930586, "4": 13709.453499992378, "5": 13876.501917024143, "6": 13709.757833043113, "7": 13953.084124950692, "8": 13781.126166926697}, "receiver_prefill_ms": 304.2779579991475, "ttft_ms": 27.062666951678693, "decode_ms": 326.9495000131428}, "total_wall_ms": 119778.10120990034, "ttft_ms": 27.062666951678693, "input_usage": {"lookups": 7}, "output_usage": {"output_tokens": 9}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "9,112,735", "evidence_artifact": {"ledger": {"facts_total": 64, "facts_retained": 15, "retention_rate": 0.234375, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.2, "amount": 0.16666666666666666, "status": 0.75, "place": 0.0, "identifier": 0.0, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 700}, "ledger_by_hop": {"1": {"facts_total": 64, "facts_retained": 10, "retention_rate": 0.15625, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.2, "amount": 0.16666666666666666, "status": 0.3333333333333333, "place": 0.0, "identifier": 0.0, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 669}, "2": {"facts_total": 64, "facts_retained": 15, "retention_rate": 0.234375, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.2, "amount": 0.16666666666666666, "status": 0.75, "place": 0.0, "identifier": 0.0, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 700}, "3": {"facts_total": 64, "facts_retained": 15, "retention_rate": 0.234375, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.2, "amount": 0.16666666666666666, "status": 0.75, "place": 0.0, "identifier": 0.0, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 700}, "4": {"facts_total": 64, "facts_retained": 15, "retention_rate": 0.234375, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.2, "amount": 0.16666666666666666, "status": 0.75, "place": 0.0, "identifier": 0.0, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 700}, "5": {"facts_total": 64, "facts_retained": 15, "retention_rate": 0.234375, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.2, "amount": 0.16666666666666666, "status": 0.75, "place": 0.0, "identifier": 0.0, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 700}, "6": {"facts_total": 64, "facts_retained": 15, "retention_rate": 0.234375, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.2, "amount": 0.16666666666666666, "status": 0.75, "place": 0.0, "identifier": 0.0, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 700}, "7": {"facts_total": 64, "facts_retained": 15, "retention_rate": 0.234375, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.2, "amount": 0.16666666666666666, "status": 0.75, "place": 0.0, "identifier": 0.0, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 700}, "8": {"facts_total": 64, "facts_retained": 15, "retention_rate": 0.234375, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.2, "amount": 0.16666666666666666, "status": 0.75, "place": 0.0, "identifier": 0.0, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 700}}}, "reference_key_id": "keys/synthetic/syn_03006_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "0c54b5424952f31a24ef521de33c53e46d87191f991ea8eefeaec650a7d38f30", "output_hash": "788fcda908c2029597420995b09d0cc919bfdbf7a28b68dcb0e331db7df7b342"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "e8957a635aac3ae0250d32c2", "attempt_id": "e6b1de171c24", "run_group_id": "E4_core:A5|repairable|qwen3b:h8", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:06:53.953230Z", "task_family": "synthetic", "document_id": "syn_03006_2000", "document_hash": "89be5e77e0ca114963e9ffdf8dc38fe5610f13a78ac29e7a1956cd9124ce5217", "question_id": "q10_temporal", "split": "test", "cluster_id": "syn_03006_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 8, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1890, "query_tokens": 414, "message_tokens": 314, "message_utf8_bytes": 748, "cache_tensor_bytes": null, "serialized_payload_bytes": 748, "side_information_bytes": 0, "fetched_bytes": 4200, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 119119.30483486503, "hops": {"1": 22600.32433399465, "2": 13699.429792002775, "3": 13789.627166930586, "4": 13709.453499992378, "5": 13876.501917024143, "6": 13709.757833043113, "7": 13953.084124950692, "8": 13781.126166926697}, "receiver_prefill_ms": 272.45054207742214, "ttft_ms": 52.7615420287475, "decode_ms": 53.174917004071176}, "total_wall_ms": 119498.51104291156, "ttft_ms": 52.7615420287475, "input_usage": {"lookups": 7}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 64, "facts_retained": 15, "retention_rate": 0.234375, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.2, "amount": 0.16666666666666666, "status": 0.75, "place": 0.0, "identifier": 0.0, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 700}, "ledger_by_hop": {"1": {"facts_total": 64, "facts_retained": 10, "retention_rate": 0.15625, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.2, "amount": 0.16666666666666666, "status": 0.3333333333333333, "place": 0.0, "identifier": 0.0, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 669}, "2": {"facts_total": 64, "facts_retained": 15, "retention_rate": 0.234375, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.2, "amount": 0.16666666666666666, "status": 0.75, "place": 0.0, "identifier": 0.0, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 700}, "3": {"facts_total": 64, "facts_retained": 15, "retention_rate": 0.234375, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.2, "amount": 0.16666666666666666, "status": 0.75, "place": 0.0, "identifier": 0.0, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 700}, "4": {"facts_total": 64, "facts_retained": 15, "retention_rate": 0.234375, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.2, "amount": 0.16666666666666666, "status": 0.75, "place": 0.0, "identifier": 0.0, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 700}, "5": {"facts_total": 64, "facts_retained": 15, "retention_rate": 0.234375, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.2, "amount": 0.16666666666666666, "status": 0.75, "place": 0.0, "identifier": 0.0, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 700}, "6": {"facts_total": 64, "facts_retained": 15, "retention_rate": 0.234375, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.2, "amount": 0.16666666666666666, "status": 0.75, "place": 0.0, "identifier": 0.0, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 700}, "7": {"facts_total": 64, "facts_retained": 15, "retention_rate": 0.234375, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.2, "amount": 0.16666666666666666, "status": 0.75, "place": 0.0, "identifier": 0.0, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 700}, "8": {"facts_total": 64, "facts_retained": 15, "retention_rate": 0.234375, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.2, "amount": 0.16666666666666666, "status": 0.75, "place": 0.0, "identifier": 0.0, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 700}}}, "reference_key_id": "keys/synthetic/syn_03006_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "0c54b5424952f31a24ef521de33c53e46d87191f991ea8eefeaec650a7d38f30", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "b75aa23ff82fe01ecb6b8eb3", "attempt_id": "68588b16d584", "run_group_id": "E4_core:A5|repairable|qwen3b:h8", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:06:54.350543Z", "task_family": "synthetic", "document_id": "syn_03006_2000", "document_hash": "89be5e77e0ca114963e9ffdf8dc38fe5610f13a78ac29e7a1956cd9124ce5217", "question_id": "q11_aggregation", "split": "test", "cluster_id": "syn_03006_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 8, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1890, "query_tokens": 407, "message_tokens": 314, "message_utf8_bytes": 748, "cache_tensor_bytes": null, "serialized_payload_bytes": 748, "side_information_bytes": 0, "fetched_bytes": 4200, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 119119.30483486503, "hops": {"1": 22600.32433399465, "2": 13699.429792002775, "3": 13789.627166930586, "4": 13709.453499992378, "5": 13876.501917024143, "6": 13709.757833043113, "7": 13953.084124950692, "8": 13781.126166926697}, "receiver_prefill_ms": 438.5642920387909, "ttft_ms": 38.16950006876141, "decode_ms": 34.57270807120949}, "total_wall_ms": 119631.11008494161, "ttft_ms": 38.16950006876141, "input_usage": {"lookups": 7}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 64, "facts_retained": 15, "retention_rate": 0.234375, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.2, "amount": 0.16666666666666666, "status": 0.75, "place": 0.0, "identifier": 0.0, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 700}, "ledger_by_hop": {"1": {"facts_total": 64, "facts_retained": 10, "retention_rate": 0.15625, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.2, "amount": 0.16666666666666666, "status": 0.3333333333333333, "place": 0.0, "identifier": 0.0, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 669}, "2": {"facts_total": 64, "facts_retained": 15, "retention_rate": 0.234375, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.2, "amount": 0.16666666666666666, "status": 0.75, "place": 0.0, "identifier": 0.0, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 700}, "3": {"facts_total": 64, "facts_retained": 15, "retention_rate": 0.234375, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.2, "amount": 0.16666666666666666, "status": 0.75, "place": 0.0, "identifier": 0.0, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 700}, "4": {"facts_total": 64, "facts_retained": 15, "retention_rate": 0.234375, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.2, "amount": 0.16666666666666666, "status": 0.75, "place": 0.0, "identifier": 0.0, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 700}, "5": {"facts_total": 64, "facts_retained": 15, "retention_rate": 0.234375, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.2, "amount": 0.16666666666666666, "status": 0.75, "place": 0.0, "identifier": 0.0, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 700}, "6": {"facts_total": 64, "facts_retained": 15, "retention_rate": 0.234375, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.2, "amount": 0.16666666666666666, "status": 0.75, "place": 0.0, "identifier": 0.0, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 700}, "7": {"facts_total": 64, "facts_retained": 15, "retention_rate": 0.234375, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.2, "amount": 0.16666666666666666, "status": 0.75, "place": 0.0, "identifier": 0.0, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 700}, "8": {"facts_total": 64, "facts_retained": 15, "retention_rate": 0.234375, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.2, "amount": 0.16666666666666666, "status": 0.75, "place": 0.0, "identifier": 0.0, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 700}}}, "reference_key_id": "keys/synthetic/syn_03006_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "0c54b5424952f31a24ef521de33c53e46d87191f991ea8eefeaec650a7d38f30", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "9230fc6544b4a41fc39c4017", "attempt_id": "3041274780ec", "run_group_id": "E4_core:A5|repairable|qwen3b:h8", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:06:54.879985Z", "task_family": "synthetic", "document_id": "syn_03006_2000", "document_hash": "89be5e77e0ca114963e9ffdf8dc38fe5610f13a78ac29e7a1956cd9124ce5217", "question_id": "q12_unit_conversion", "split": "test", "cluster_id": "syn_03006_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 8, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1890, "query_tokens": 404, "message_tokens": 314, "message_utf8_bytes": 748, "cache_tensor_bytes": null, "serialized_payload_bytes": 748, "side_information_bytes": 0, "fetched_bytes": 4200, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 119119.30483486503, "hops": {"1": 22600.32433399465, "2": 13699.429792002775, "3": 13789.627166930586, "4": 13709.453499992378, "5": 13876.501917024143, "6": 13709.757833043113, "7": 13953.084124950692, "8": 13781.126166926697}, "receiver_prefill_ms": 283.2313330145553, "ttft_ms": 43.23650000151247, "decode_ms": 275.96270805224776}, "total_wall_ms": 119722.28987689596, "ttft_ms": 43.23650000151247, "input_usage": {"lookups": 7}, "output_usage": {"output_tokens": 8}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "11946000", "evidence_artifact": {"ledger": {"facts_total": 64, "facts_retained": 15, "retention_rate": 0.234375, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.2, "amount": 0.16666666666666666, "status": 0.75, "place": 0.0, "identifier": 0.0, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 700}, "ledger_by_hop": {"1": {"facts_total": 64, "facts_retained": 10, "retention_rate": 0.15625, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.2, "amount": 0.16666666666666666, "status": 0.3333333333333333, "place": 0.0, "identifier": 0.0, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 669}, "2": {"facts_total": 64, "facts_retained": 15, "retention_rate": 0.234375, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.2, "amount": 0.16666666666666666, "status": 0.75, "place": 0.0, "identifier": 0.0, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 700}, "3": {"facts_total": 64, "facts_retained": 15, "retention_rate": 0.234375, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.2, "amount": 0.16666666666666666, "status": 0.75, "place": 0.0, "identifier": 0.0, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 700}, "4": {"facts_total": 64, "facts_retained": 15, "retention_rate": 0.234375, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.2, "amount": 0.16666666666666666, "status": 0.75, "place": 0.0, "identifier": 0.0, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 700}, "5": {"facts_total": 64, "facts_retained": 15, "retention_rate": 0.234375, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.2, "amount": 0.16666666666666666, "status": 0.75, "place": 0.0, "identifier": 0.0, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 700}, "6": {"facts_total": 64, "facts_retained": 15, "retention_rate": 0.234375, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.2, "amount": 0.16666666666666666, "status": 0.75, "place": 0.0, "identifier": 0.0, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 700}, "7": {"facts_total": 64, "facts_retained": 15, "retention_rate": 0.234375, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.2, "amount": 0.16666666666666666, "status": 0.75, "place": 0.0, "identifier": 0.0, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 700}, "8": {"facts_total": 64, "facts_retained": 15, "retention_rate": 0.234375, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.2, "amount": 0.16666666666666666, "status": 0.75, "place": 0.0, "identifier": 0.0, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 700}}}, "reference_key_id": "keys/synthetic/syn_03006_2000.json", "task_scores": {"correct": 0.0, "abstained": 0.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "0c54b5424952f31a24ef521de33c53e46d87191f991ea8eefeaec650a7d38f30", "output_hash": "edf27f197d1395ea4e7b86fae1feecd3591246e1299bba6a08655f5724a670ea"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "b4022ef1a7816121cc35fa3d", "attempt_id": "51af399590ce", "run_group_id": "E4_core:A5|repairable|qwen3b:h8", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:06:55.499724Z", "task_family": "synthetic", "document_id": "syn_03006_2000", "document_hash": "89be5e77e0ca114963e9ffdf8dc38fe5610f13a78ac29e7a1956cd9124ce5217", "question_id": "q13_unanswerable", "split": "test", "cluster_id": "syn_03006_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 8, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1890, "query_tokens": 408, "message_tokens": 314, "message_utf8_bytes": 748, "cache_tensor_bytes": null, "serialized_payload_bytes": 748, "side_information_bytes": 0, "fetched_bytes": 4200, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 119119.30483486503, "hops": {"1": 22600.32433399465, "2": 13699.429792002775, "3": 13789.627166930586, "4": 13709.453499992378, "5": 13876.501917024143, "6": 13709.757833043113, "7": 13953.084124950692, "8": 13781.126166926697}, "receiver_prefill_ms": 286.40720900148153, "ttft_ms": 37.3239170294255, "decode_ms": 62.507083057425916}, "total_wall_ms": 119506.19137694594, "ttft_ms": 37.3239170294255, "input_usage": {"lookups": 7}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 64, "facts_retained": 15, "retention_rate": 0.234375, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.2, "amount": 0.16666666666666666, "status": 0.75, "place": 0.0, "identifier": 0.0, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 700}, "ledger_by_hop": {"1": {"facts_total": 64, "facts_retained": 10, "retention_rate": 0.15625, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.2, "amount": 0.16666666666666666, "status": 0.3333333333333333, "place": 0.0, "identifier": 0.0, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 669}, "2": {"facts_total": 64, "facts_retained": 15, "retention_rate": 0.234375, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.2, "amount": 0.16666666666666666, "status": 0.75, "place": 0.0, "identifier": 0.0, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 700}, "3": {"facts_total": 64, "facts_retained": 15, "retention_rate": 0.234375, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.2, "amount": 0.16666666666666666, "status": 0.75, "place": 0.0, "identifier": 0.0, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 700}, "4": {"facts_total": 64, "facts_retained": 15, "retention_rate": 0.234375, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.2, "amount": 0.16666666666666666, "status": 0.75, "place": 0.0, "identifier": 0.0, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 700}, "5": {"facts_total": 64, "facts_retained": 15, "retention_rate": 0.234375, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.2, "amount": 0.16666666666666666, "status": 0.75, "place": 0.0, "identifier": 0.0, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 700}, "6": {"facts_total": 64, "facts_retained": 15, "retention_rate": 0.234375, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.2, "amount": 0.16666666666666666, "status": 0.75, "place": 0.0, "identifier": 0.0, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 700}, "7": {"facts_total": 64, "facts_retained": 15, "retention_rate": 0.234375, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.2, "amount": 0.16666666666666666, "status": 0.75, "place": 0.0, "identifier": 0.0, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 700}, "8": {"facts_total": 64, "facts_retained": 15, "retention_rate": 0.234375, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.2, "amount": 0.16666666666666666, "status": 0.75, "place": 0.0, "identifier": 0.0, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 700}}}, "reference_key_id": "keys/synthetic/syn_03006_2000.json", "task_scores": {"correct": 1.0, "abstained": 1.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "0c54b5424952f31a24ef521de33c53e46d87191f991ea8eefeaec650a7d38f30", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "d604cbcbf3ec182f7a5ebded", "attempt_id": "a18efadfdf49", "run_group_id": "E4_core:A5|repairable|qwen3b:h8", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:06:55.903546Z", "task_family": "synthetic", "document_id": "syn_03006_2000", "document_hash": "89be5e77e0ca114963e9ffdf8dc38fe5610f13a78ac29e7a1956cd9124ce5217", "question_id": "q14_unanswerable", "split": "test", "cluster_id": "syn_03006_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 8, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1890, "query_tokens": 409, "message_tokens": 314, "message_utf8_bytes": 748, "cache_tensor_bytes": null, "serialized_payload_bytes": 748, "side_information_bytes": 0, "fetched_bytes": 4200, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 119119.30483486503, "hops": {"1": 22600.32433399465, "2": 13699.429792002775, "3": 13789.627166930586, "4": 13709.453499992378, "5": 13876.501917024143, "6": 13709.757833043113, "7": 13953.084124950692, "8": 13781.126166926697}, "receiver_prefill_ms": 440.7275420380756, "ttft_ms": 30.783249996602535, "decode_ms": 42.34208399429917}, "total_wall_ms": 119633.84770986158, "ttft_ms": 30.783249996602535, "input_usage": {"lookups": 7}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 64, "facts_retained": 15, "retention_rate": 0.234375, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.2, "amount": 0.16666666666666666, "status": 0.75, "place": 0.0, "identifier": 0.0, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 700}, "ledger_by_hop": {"1": {"facts_total": 64, "facts_retained": 10, "retention_rate": 0.15625, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.2, "amount": 0.16666666666666666, "status": 0.3333333333333333, "place": 0.0, "identifier": 0.0, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 669}, "2": {"facts_total": 64, "facts_retained": 15, "retention_rate": 0.234375, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.2, "amount": 0.16666666666666666, "status": 0.75, "place": 0.0, "identifier": 0.0, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 700}, "3": {"facts_total": 64, "facts_retained": 15, "retention_rate": 0.234375, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.2, "amount": 0.16666666666666666, "status": 0.75, "place": 0.0, "identifier": 0.0, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 700}, "4": {"facts_total": 64, "facts_retained": 15, "retention_rate": 0.234375, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.2, "amount": 0.16666666666666666, "status": 0.75, "place": 0.0, "identifier": 0.0, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 700}, "5": {"facts_total": 64, "facts_retained": 15, "retention_rate": 0.234375, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.2, "amount": 0.16666666666666666, "status": 0.75, "place": 0.0, "identifier": 0.0, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 700}, "6": {"facts_total": 64, "facts_retained": 15, "retention_rate": 0.234375, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.2, "amount": 0.16666666666666666, "status": 0.75, "place": 0.0, "identifier": 0.0, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 700}, "7": {"facts_total": 64, "facts_retained": 15, "retention_rate": 0.234375, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.2, "amount": 0.16666666666666666, "status": 0.75, "place": 0.0, "identifier": 0.0, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 700}, "8": {"facts_total": 64, "facts_retained": 15, "retention_rate": 0.234375, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.2, "amount": 0.16666666666666666, "status": 0.75, "place": 0.0, "identifier": 0.0, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 700}}}, "reference_key_id": "keys/synthetic/syn_03006_2000.json", "task_scores": {"correct": 1.0, "abstained": 1.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "0c54b5424952f31a24ef521de33c53e46d87191f991ea8eefeaec650a7d38f30", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "50b1db927351de59d24122c1", "attempt_id": "5851e280786d", "run_group_id": "E4_core:A4|closed|qwen3b-llama3b:h1", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:07:18.929139Z", "task_family": "synthetic", "document_id": "syn_03006_2000", "document_hash": "89be5e77e0ca114963e9ffdf8dc38fe5610f13a78ac29e7a1956cd9124ce5217", "question_id": "q00_single_fact", "split": "test", "cluster_id": "syn_03006_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 1, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1890, "query_tokens": 595, "message_tokens": 512, "message_utf8_bytes": 1087, "cache_tensor_bytes": null, "serialized_payload_bytes": 1087, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 22491.57329101581, "hops": {"1": 22491.57329101581}, "receiver_prefill_ms": 399.2386660538614, "ttft_ms": 38.624332984909415, "decode_ms": 496.3607919635251}, "total_wall_ms": 23426.628040964715, "ttft_ms": 38.624332984909415, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 10}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "XM-76Q-2958", "evidence_artifact": {"ledger": {"facts_total": 64, "facts_retained": 30, "retention_rate": 0.46875, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.6666666666666666, "status": 0.9166666666666666, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1087}, "ledger_by_hop": {"1": {"facts_total": 64, "facts_retained": 30, "retention_rate": 0.46875, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.6666666666666666, "status": 0.9166666666666666, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1087}}}, "reference_key_id": "keys/synthetic/syn_03006_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "47f843b1989b354c7cbe406d456e62c59b6229490f55dde77bcdb5f70147ff28", "output_hash": "cf42691eec772b9ec74d67643944de2e19e53724d7b772cef1e0f2d5ef3b4926"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "7dca3d942e3e6b0f4e74fe4c", "attempt_id": "2b4c9252ab2b", "run_group_id": "E4_core:A4|closed|qwen3b-llama3b:h1", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:07:19.882311Z", "task_family": "synthetic", "document_id": "syn_03006_2000", "document_hash": "89be5e77e0ca114963e9ffdf8dc38fe5610f13a78ac29e7a1956cd9124ce5217", "question_id": "q01_single_fact", "split": "test", "cluster_id": "syn_03006_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 1, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1890, "query_tokens": 600, "message_tokens": 512, "message_utf8_bytes": 1087, "cache_tensor_bytes": null, "serialized_payload_bytes": 1087, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 22491.57329101581, "hops": {"1": 22491.57329101581}, "receiver_prefill_ms": 483.1836250377819, "ttft_ms": 27.735125040635467, "decode_ms": 37.128041964024305}, "total_wall_ms": 23040.274249971844, "ttft_ms": 27.735125040635467, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 64, "facts_retained": 30, "retention_rate": 0.46875, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.6666666666666666, "status": 0.9166666666666666, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1087}, "ledger_by_hop": {"1": {"facts_total": 64, "facts_retained": 30, "retention_rate": 0.46875, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.6666666666666666, "status": 0.9166666666666666, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1087}}}, "reference_key_id": "keys/synthetic/syn_03006_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "47f843b1989b354c7cbe406d456e62c59b6229490f55dde77bcdb5f70147ff28", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "78c6d16c5160bede1ac9c06b", "attempt_id": "753cfe6f2507", "run_group_id": "E4_core:A4|closed|qwen3b-llama3b:h1", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:07:20.451258Z", "task_family": "synthetic", "document_id": "syn_03006_2000", "document_hash": "89be5e77e0ca114963e9ffdf8dc38fe5610f13a78ac29e7a1956cd9124ce5217", "question_id": "q02_exact_copy", "split": "test", "cluster_id": "syn_03006_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 1, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1890, "query_tokens": 599, "message_tokens": 512, "message_utf8_bytes": 1087, "cache_tensor_bytes": null, "serialized_payload_bytes": 1087, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 22491.57329101581, "hops": {"1": 22491.57329101581}, "receiver_prefill_ms": 404.5649999752641, "ttft_ms": 18.913666950538754, "decode_ms": 38.92045805696398}, "total_wall_ms": 22954.59241606295, "ttft_ms": 18.913666950538754, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 64, "facts_retained": 30, "retention_rate": 0.46875, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.6666666666666666, "status": 0.9166666666666666, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1087}, "ledger_by_hop": {"1": {"facts_total": 64, "facts_retained": 30, "retention_rate": 0.46875, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.6666666666666666, "status": 0.9166666666666666, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1087}}}, "reference_key_id": "keys/synthetic/syn_03006_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "47f843b1989b354c7cbe406d456e62c59b6229490f55dde77bcdb5f70147ff28", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "c4987a40ad33f41d9b57eb29", "attempt_id": "e9b1b6ffbc16", "run_group_id": "E4_core:A4|closed|qwen3b-llama3b:h1", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:07:20.932106Z", "task_family": "synthetic", "document_id": "syn_03006_2000", "document_hash": "89be5e77e0ca114963e9ffdf8dc38fe5610f13a78ac29e7a1956cd9124ce5217", "question_id": "q03_exact_copy", "split": "test", "cluster_id": "syn_03006_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 1, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1890, "query_tokens": 594, "message_tokens": 512, "message_utf8_bytes": 1087, "cache_tensor_bytes": null, "serialized_payload_bytes": 1087, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 22491.57329101581, "hops": {"1": 22491.57329101581}, "receiver_prefill_ms": 361.6281249560416, "ttft_ms": 52.36474995035678, "decode_ms": 499.9549170024693}, "total_wall_ms": 23406.16866608616, "ttft_ms": 52.36474995035678, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 8}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "Haruki Ouellet-Marchand", "evidence_artifact": {"ledger": {"facts_total": 64, "facts_retained": 30, "retention_rate": 0.46875, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.6666666666666666, "status": 0.9166666666666666, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1087}, "ledger_by_hop": {"1": {"facts_total": 64, "facts_retained": 30, "retention_rate": 0.46875, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.6666666666666666, "status": 0.9166666666666666, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1087}}}, "reference_key_id": "keys/synthetic/syn_03006_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "47f843b1989b354c7cbe406d456e62c59b6229490f55dde77bcdb5f70147ff28", "output_hash": "a3b6e0a44d495387d8b8fc50ba6be0f2ed43121c674d3b3e962f8c4690e15b5c"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "6e5251595efb6193b44af111", "attempt_id": "a16ae61af09c", "run_group_id": "E4_core:A4|closed|qwen3b-llama3b:h1", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:07:21.863344Z", "task_family": "synthetic", "document_id": "syn_03006_2000", "document_hash": "89be5e77e0ca114963e9ffdf8dc38fe5610f13a78ac29e7a1956cd9124ce5217", "question_id": "q04_multi_hop", "split": "test", "cluster_id": "syn_03006_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 1, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1890, "query_tokens": 603, "message_tokens": 512, "message_utf8_bytes": 1087, "cache_tensor_bytes": null, "serialized_payload_bytes": 1087, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 22491.57329101581, "hops": {"1": 22491.57329101581}, "receiver_prefill_ms": 408.3703749347478, "ttft_ms": 27.516833040863276, "decode_ms": 38.1917500635609}, "total_wall_ms": 22966.28979104571, "ttft_ms": 27.516833040863276, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 64, "facts_retained": 30, "retention_rate": 0.46875, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.6666666666666666, "status": 0.9166666666666666, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1087}, "ledger_by_hop": {"1": {"facts_total": 64, "facts_retained": 30, "retention_rate": 0.46875, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.6666666666666666, "status": 0.9166666666666666, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1087}}}, "reference_key_id": "keys/synthetic/syn_03006_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "47f843b1989b354c7cbe406d456e62c59b6229490f55dde77bcdb5f70147ff28", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "6b5a66ae3de04c8eabbaf34c", "attempt_id": "bd24aaf069f7", "run_group_id": "E4_core:A4|closed|qwen3b-llama3b:h1", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:07:22.355033Z", "task_family": "synthetic", "document_id": "syn_03006_2000", "document_hash": "89be5e77e0ca114963e9ffdf8dc38fe5610f13a78ac29e7a1956cd9124ce5217", "question_id": "q05_multi_hop", "split": "test", "cluster_id": "syn_03006_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 1, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1890, "query_tokens": 606, "message_tokens": 512, "message_utf8_bytes": 1087, "cache_tensor_bytes": null, "serialized_payload_bytes": 1087, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 22491.57329101581, "hops": {"1": 22491.57329101581}, "receiver_prefill_ms": 382.7119169291109, "ttft_ms": 30.410457984544337, "decode_ms": 641.8092919047922}, "total_wall_ms": 23547.261874075048, "ttft_ms": 30.410457984544337, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 10}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "EP-52G-1330", "evidence_artifact": {"ledger": {"facts_total": 64, "facts_retained": 30, "retention_rate": 0.46875, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.6666666666666666, "status": 0.9166666666666666, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1087}, "ledger_by_hop": {"1": {"facts_total": 64, "facts_retained": 30, "retention_rate": 0.46875, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.6666666666666666, "status": 0.9166666666666666, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1087}}}, "reference_key_id": "keys/synthetic/syn_03006_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "47f843b1989b354c7cbe406d456e62c59b6229490f55dde77bcdb5f70147ff28", "output_hash": "104db207937fd79e23ce16f9b3f10b1a9ed34f6dd0441a99be6ea1b2d0a09d99"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "027cabafb99c97a7fe14e54f", "attempt_id": "856d8fb4a692", "run_group_id": "E4_core:A4|closed|qwen3b-llama3b:h1", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:07:23.428468Z", "task_family": "synthetic", "document_id": "syn_03006_2000", "document_hash": "89be5e77e0ca114963e9ffdf8dc38fe5610f13a78ac29e7a1956cd9124ce5217", "question_id": "q06_negation", "split": "test", "cluster_id": "syn_03006_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 1, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1890, "query_tokens": 594, "message_tokens": 512, "message_utf8_bytes": 1087, "cache_tensor_bytes": null, "serialized_payload_bytes": 1087, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 22491.57329101581, "hops": {"1": 22491.57329101581}, "receiver_prefill_ms": 410.49845900852233, "ttft_ms": 35.677041043527424, "decode_ms": 37.11262496653944}, "total_wall_ms": 22975.506749004126, "ttft_ms": 35.677041043527424, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "no", "evidence_artifact": {"ledger": {"facts_total": 64, "facts_retained": 30, "retention_rate": 0.46875, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.6666666666666666, "status": 0.9166666666666666, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1087}, "ledger_by_hop": {"1": {"facts_total": 64, "facts_retained": 30, "retention_rate": 0.46875, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.6666666666666666, "status": 0.9166666666666666, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1087}}}, "reference_key_id": "keys/synthetic/syn_03006_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "47f843b1989b354c7cbe406d456e62c59b6229490f55dde77bcdb5f70147ff28", "output_hash": "9390298f3fb0c5b160498935d79cb139aef28e1c47358b4bbba61862b9c26e59"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "98ab2bf247b9bb59bc6c646b", "attempt_id": "075714b861fb", "run_group_id": "E4_core:A4|closed|qwen3b-llama3b:h1", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:07:23.930435Z", "task_family": "synthetic", "document_id": "syn_03006_2000", "document_hash": "89be5e77e0ca114963e9ffdf8dc38fe5610f13a78ac29e7a1956cd9124ce5217", "question_id": "q07_negation", "split": "test", "cluster_id": "syn_03006_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 1, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1890, "query_tokens": 594, "message_tokens": 512, "message_utf8_bytes": 1087, "cache_tensor_bytes": null, "serialized_payload_bytes": 1087, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 22491.57329101581, "hops": {"1": 22491.57329101581}, "receiver_prefill_ms": 384.1534999664873, "ttft_ms": 29.752249945886433, "decode_ms": 48.4838749980554}, "total_wall_ms": 22954.63595702313, "ttft_ms": 29.752249945886433, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "no", "evidence_artifact": {"ledger": {"facts_total": 64, "facts_retained": 30, "retention_rate": 0.46875, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.6666666666666666, "status": 0.9166666666666666, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1087}, "ledger_by_hop": {"1": {"facts_total": 64, "facts_retained": 30, "retention_rate": 0.46875, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.6666666666666666, "status": 0.9166666666666666, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1087}}}, "reference_key_id": "keys/synthetic/syn_03006_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "47f843b1989b354c7cbe406d456e62c59b6229490f55dde77bcdb5f70147ff28", "output_hash": "9390298f3fb0c5b160498935d79cb139aef28e1c47358b4bbba61862b9c26e59"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "e1c4d43796f6270c29ef16d8", "attempt_id": "9e1887e8a768", "run_group_id": "E4_core:A4|closed|qwen3b-llama3b:h1", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:07:24.410658Z", "task_family": "synthetic", "document_id": "syn_03006_2000", "document_hash": "89be5e77e0ca114963e9ffdf8dc38fe5610f13a78ac29e7a1956cd9124ce5217", "question_id": "q08_revision", "split": "test", "cluster_id": "syn_03006_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 1, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1890, "query_tokens": 611, "message_tokens": 512, "message_utf8_bytes": 1087, "cache_tensor_bytes": null, "serialized_payload_bytes": 1087, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 22491.57329101581, "hops": {"1": 22491.57329101581}, "receiver_prefill_ms": 418.3420839253813, "ttft_ms": 84.40687507390976, "decode_ms": 316.2348329788074}, "total_wall_ms": 23311.25387409702, "ttft_ms": 84.40687507390976, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 7}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "9112735", "evidence_artifact": {"ledger": {"facts_total": 64, "facts_retained": 30, "retention_rate": 0.46875, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.6666666666666666, "status": 0.9166666666666666, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1087}, "ledger_by_hop": {"1": {"facts_total": 64, "facts_retained": 30, "retention_rate": 0.46875, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.6666666666666666, "status": 0.9166666666666666, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1087}}}, "reference_key_id": "keys/synthetic/syn_03006_2000.json", "task_scores": {"correct": 0.0, "abstained": 0.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "47f843b1989b354c7cbe406d456e62c59b6229490f55dde77bcdb5f70147ff28", "output_hash": "15a3f804636f0e62d4a3d9be7677f001a0ad6b1fb8ae7e8cdc63c44ec7bac31f"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "9b0acc539a6d51ea8038e36e", "attempt_id": "193e52319158", "run_group_id": "E4_core:A4|closed|qwen3b-llama3b:h1", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:07:25.248559Z", "task_family": "synthetic", "document_id": "syn_03006_2000", "document_hash": "89be5e77e0ca114963e9ffdf8dc38fe5610f13a78ac29e7a1956cd9124ce5217", "question_id": "q09_revision", "split": "test", "cluster_id": "syn_03006_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 1, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1890, "query_tokens": 604, "message_tokens": 512, "message_utf8_bytes": 1087, "cache_tensor_bytes": null, "serialized_payload_bytes": 1087, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 22491.57329101581, "hops": {"1": 22491.57329101581}, "receiver_prefill_ms": 397.1153749153018, "ttft_ms": 33.449833979830146, "decode_ms": 217.8467499325052}, "total_wall_ms": 23140.574665972963, "ttft_ms": 33.449833979830146, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 7}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "9112735", "evidence_artifact": {"ledger": {"facts_total": 64, "facts_retained": 30, "retention_rate": 0.46875, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.6666666666666666, "status": 0.9166666666666666, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1087}, "ledger_by_hop": {"1": {"facts_total": 64, "facts_retained": 30, "retention_rate": 0.46875, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.6666666666666666, "status": 0.9166666666666666, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1087}}}, "reference_key_id": "keys/synthetic/syn_03006_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "47f843b1989b354c7cbe406d456e62c59b6229490f55dde77bcdb5f70147ff28", "output_hash": "15a3f804636f0e62d4a3d9be7677f001a0ad6b1fb8ae7e8cdc63c44ec7bac31f"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "6fb2ba7fd7a1490b2f289b97", "attempt_id": "699a31452797", "run_group_id": "E4_core:A4|closed|qwen3b-llama3b:h1", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:07:25.914596Z", "task_family": "synthetic", "document_id": "syn_03006_2000", "document_hash": "89be5e77e0ca114963e9ffdf8dc38fe5610f13a78ac29e7a1956cd9124ce5217", "question_id": "q10_temporal", "split": "test", "cluster_id": "syn_03006_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 1, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1890, "query_tokens": 613, "message_tokens": 512, "message_utf8_bytes": 1087, "cache_tensor_bytes": null, "serialized_payload_bytes": 1087, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 22491.57329101581, "hops": {"1": 22491.57329101581}, "receiver_prefill_ms": 357.7468330040574, "ttft_ms": 36.314583965577185, "decode_ms": 51.811582990922034}, "total_wall_ms": 22938.175790943205, "ttft_ms": 36.314583965577185, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 64, "facts_retained": 30, "retention_rate": 0.46875, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.6666666666666666, "status": 0.9166666666666666, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1087}, "ledger_by_hop": {"1": {"facts_total": 64, "facts_retained": 30, "retention_rate": 0.46875, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.6666666666666666, "status": 0.9166666666666666, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1087}}}, "reference_key_id": "keys/synthetic/syn_03006_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "47f843b1989b354c7cbe406d456e62c59b6229490f55dde77bcdb5f70147ff28", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "5440402f78a69ca3accaefc4", "attempt_id": "bd9fe7bc28f8", "run_group_id": "E4_core:A4|closed|qwen3b-llama3b:h1", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:07:26.379823Z", "task_family": "synthetic", "document_id": "syn_03006_2000", "document_hash": "89be5e77e0ca114963e9ffdf8dc38fe5610f13a78ac29e7a1956cd9124ce5217", "question_id": "q11_aggregation", "split": "test", "cluster_id": "syn_03006_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 1, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1890, "query_tokens": 606, "message_tokens": 512, "message_utf8_bytes": 1087, "cache_tensor_bytes": null, "serialized_payload_bytes": 1087, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 22491.57329101581, "hops": {"1": 22491.57329101581}, "receiver_prefill_ms": 528.2210840377957, "ttft_ms": 41.63975000847131, "decode_ms": 234.78749999776483}, "total_wall_ms": 23297.114290995523, "ttft_ms": 41.63975000847131, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 7}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "8908553", "evidence_artifact": {"ledger": {"facts_total": 64, "facts_retained": 30, "retention_rate": 0.46875, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.6666666666666666, "status": 0.9166666666666666, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1087}, "ledger_by_hop": {"1": {"facts_total": 64, "facts_retained": 30, "retention_rate": 0.46875, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.6666666666666666, "status": 0.9166666666666666, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1087}}}, "reference_key_id": "keys/synthetic/syn_03006_2000.json", "task_scores": {"correct": 0.0, "abstained": 0.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "47f843b1989b354c7cbe406d456e62c59b6229490f55dde77bcdb5f70147ff28", "output_hash": "5ad9e777dc3be36808c74badfc8aae217b41a486754d07adf1abda6f6bce9aca"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "d4641d2e186dea647acebb4e", "attempt_id": "ba76ea58852a", "run_group_id": "E4_core:A4|closed|qwen3b-llama3b:h1", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:07:27.203811Z", "task_family": "synthetic", "document_id": "syn_03006_2000", "document_hash": "89be5e77e0ca114963e9ffdf8dc38fe5610f13a78ac29e7a1956cd9124ce5217", "question_id": "q12_unit_conversion", "split": "test", "cluster_id": "syn_03006_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 1, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1890, "query_tokens": 603, "message_tokens": 512, "message_utf8_bytes": 1087, "cache_tensor_bytes": null, "serialized_payload_bytes": 1087, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 22491.57329101581, "hops": {"1": 22491.57329101581}, "receiver_prefill_ms": 410.87654104921967, "ttft_ms": 30.07483296096325, "decode_ms": 222.9516670340672}, "total_wall_ms": 23156.40420804266, "ttft_ms": 30.07483296096325, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 5}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "39730", "evidence_artifact": {"ledger": {"facts_total": 64, "facts_retained": 30, "retention_rate": 0.46875, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.6666666666666666, "status": 0.9166666666666666, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1087}, "ledger_by_hop": {"1": {"facts_total": 64, "facts_retained": 30, "retention_rate": 0.46875, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.6666666666666666, "status": 0.9166666666666666, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1087}}}, "reference_key_id": "keys/synthetic/syn_03006_2000.json", "task_scores": {"correct": 0.0, "abstained": 0.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "47f843b1989b354c7cbe406d456e62c59b6229490f55dde77bcdb5f70147ff28", "output_hash": "91306ecd2f87f05ab1c9b3377d592e9ec7f50f091fa1f94b9799c87e2d2a0108"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "061ce50e83a20f24b9823f16", "attempt_id": "947c46fcd5b3", "run_group_id": "E4_core:A4|closed|qwen3b-llama3b:h1", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:07:27.887363Z", "task_family": "synthetic", "document_id": "syn_03006_2000", "document_hash": "89be5e77e0ca114963e9ffdf8dc38fe5610f13a78ac29e7a1956cd9124ce5217", "question_id": "q13_unanswerable", "split": "test", "cluster_id": "syn_03006_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 1, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1890, "query_tokens": 607, "message_tokens": 512, "message_utf8_bytes": 1087, "cache_tensor_bytes": null, "serialized_payload_bytes": 1087, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 22491.57329101581, "hops": {"1": 22491.57329101581}, "receiver_prefill_ms": 505.1129589555785, "ttft_ms": 38.52137504145503, "decode_ms": 31.79083310533315}, "total_wall_ms": 23067.715500015765, "ttft_ms": 38.52137504145503, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 64, "facts_retained": 30, "retention_rate": 0.46875, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.6666666666666666, "status": 0.9166666666666666, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1087}, "ledger_by_hop": {"1": {"facts_total": 64, "facts_retained": 30, "retention_rate": 0.46875, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.6666666666666666, "status": 0.9166666666666666, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1087}}}, "reference_key_id": "keys/synthetic/syn_03006_2000.json", "task_scores": {"correct": 1.0, "abstained": 1.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "47f843b1989b354c7cbe406d456e62c59b6229490f55dde77bcdb5f70147ff28", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "ea0da46820f3f022edd13c9c", "attempt_id": "a207c8d728c9", "run_group_id": "E4_core:A4|closed|qwen3b-llama3b:h1", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:07:28.480966Z", "task_family": "synthetic", "document_id": "syn_03006_2000", "document_hash": "89be5e77e0ca114963e9ffdf8dc38fe5610f13a78ac29e7a1956cd9124ce5217", "question_id": "q14_unanswerable", "split": "test", "cluster_id": "syn_03006_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 1, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1890, "query_tokens": 608, "message_tokens": 512, "message_utf8_bytes": 1087, "cache_tensor_bytes": null, "serialized_payload_bytes": 1087, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 22491.57329101581, "hops": {"1": 22491.57329101581}, "receiver_prefill_ms": 390.56708407588303, "ttft_ms": 40.46408401336521, "decode_ms": 24.786500027403235}, "total_wall_ms": 22948.149831965566, "ttft_ms": 40.46408401336521, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 64, "facts_retained": 30, "retention_rate": 0.46875, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.6666666666666666, "status": 0.9166666666666666, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1087}, "ledger_by_hop": {"1": {"facts_total": 64, "facts_retained": 30, "retention_rate": 0.46875, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.6666666666666666, "status": 0.9166666666666666, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1087}}}, "reference_key_id": "keys/synthetic/syn_03006_2000.json", "task_scores": {"correct": 1.0, "abstained": 1.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "47f843b1989b354c7cbe406d456e62c59b6229490f55dde77bcdb5f70147ff28", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "1010e6665a3814000f4488b3", "attempt_id": "7ca7d54ba970", "run_group_id": "E4_core:A4|closed|qwen3b-llama3b:h2", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:07:45.618274Z", "task_family": "synthetic", "document_id": "syn_03006_2000", "document_hash": "89be5e77e0ca114963e9ffdf8dc38fe5610f13a78ac29e7a1956cd9124ce5217", "question_id": "q00_single_fact", "split": "test", "cluster_id": "syn_03006_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "tokenizer_hashes": {"receiver": "b9d8872cf3a99da99e1fc4b0f71232a451632dc49ca4c6771549b4efa5c1c763"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 2, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1717, "query_tokens": 528, "message_tokens": 424, "message_utf8_bytes": 1071, "cache_tensor_bytes": null, "serialized_payload_bytes": 1071, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 39154.01008212939, "hops": {"1": 22491.57329101581, "2": 16662.436791113578}, "receiver_prefill_ms": 337.26358402054757, "ttft_ms": 32.146124984137714, "decode_ms": 240.8219170756638}, "total_wall_ms": 39764.90066514816, "ttft_ms": 32.146124984137714, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 7}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "XM-76Q-2958", "evidence_artifact": {"ledger": {"facts_total": 64, "facts_retained": 30, "retention_rate": 0.46875, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.6666666666666666, "status": 0.9166666666666666, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1071}, "ledger_by_hop": {"1": {"facts_total": 64, "facts_retained": 30, "retention_rate": 0.46875, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.6666666666666666, "status": 0.9166666666666666, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1087}, "2": {"facts_total": 64, "facts_retained": 30, "retention_rate": 0.46875, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.6666666666666666, "status": 0.9166666666666666, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1071}}}, "reference_key_id": "keys/synthetic/syn_03006_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "e003c9569a8d38982c5059829bf26098eb9667ddf2655207f86179e344a0acf4", "output_hash": "cf42691eec772b9ec74d67643944de2e19e53724d7b772cef1e0f2d5ef3b4926"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "057afec3ee239b42d6b671d2", "attempt_id": "9b87e721983b", "run_group_id": "E4_core:A4|closed|qwen3b-llama3b:h2", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:07:46.247546Z", "task_family": "synthetic", "document_id": "syn_03006_2000", "document_hash": "89be5e77e0ca114963e9ffdf8dc38fe5610f13a78ac29e7a1956cd9124ce5217", "question_id": "q01_single_fact", "split": "test", "cluster_id": "syn_03006_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "tokenizer_hashes": {"receiver": "b9d8872cf3a99da99e1fc4b0f71232a451632dc49ca4c6771549b4efa5c1c763"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 2, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1717, "query_tokens": 534, "message_tokens": 424, "message_utf8_bytes": 1071, "cache_tensor_bytes": null, "serialized_payload_bytes": 1071, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 39154.01008212939, "hops": {"1": 22491.57329101581, "2": 16662.436791113578}, "receiver_prefill_ms": 380.153915961273, "ttft_ms": 134.6272920491174, "decode_ms": 38.13487500883639}, "total_wall_ms": 39707.593249040656, "ttft_ms": 134.6272920491174, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 64, "facts_retained": 30, "retention_rate": 0.46875, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.6666666666666666, "status": 0.9166666666666666, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1071}, "ledger_by_hop": {"1": {"facts_total": 64, "facts_retained": 30, "retention_rate": 0.46875, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.6666666666666666, "status": 0.9166666666666666, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1087}, "2": {"facts_total": 64, "facts_retained": 30, "retention_rate": 0.46875, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.6666666666666666, "status": 0.9166666666666666, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1071}}}, "reference_key_id": "keys/synthetic/syn_03006_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "e003c9569a8d38982c5059829bf26098eb9667ddf2655207f86179e344a0acf4", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "447960785144e946d1a7d296", "attempt_id": "d1b5461720f2", "run_group_id": "E4_core:A4|closed|qwen3b-llama3b:h2", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:07:46.819130Z", "task_family": "synthetic", "document_id": "syn_03006_2000", "document_hash": "89be5e77e0ca114963e9ffdf8dc38fe5610f13a78ac29e7a1956cd9124ce5217", "question_id": "q02_exact_copy", "split": "test", "cluster_id": "syn_03006_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "tokenizer_hashes": {"receiver": "b9d8872cf3a99da99e1fc4b0f71232a451632dc49ca4c6771549b4efa5c1c763"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 2, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1717, "query_tokens": 533, "message_tokens": 424, "message_utf8_bytes": 1071, "cache_tensor_bytes": null, "serialized_payload_bytes": 1071, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 39154.01008212939, "hops": {"1": 22491.57329101581, "2": 16662.436791113578}, "receiver_prefill_ms": 345.5545420292765, "ttft_ms": 41.88358306419104, "decode_ms": 24.510874995030463}, "total_wall_ms": 39566.576498094946, "ttft_ms": 41.88358306419104, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 64, "facts_retained": 30, "retention_rate": 0.46875, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.6666666666666666, "status": 0.9166666666666666, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1071}, "ledger_by_hop": {"1": {"facts_total": 64, "facts_retained": 30, "retention_rate": 0.46875, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.6666666666666666, "status": 0.9166666666666666, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1087}, "2": {"facts_total": 64, "facts_retained": 30, "retention_rate": 0.46875, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.6666666666666666, "status": 0.9166666666666666, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1071}}}, "reference_key_id": "keys/synthetic/syn_03006_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "e003c9569a8d38982c5059829bf26098eb9667ddf2655207f86179e344a0acf4", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "e4e0fcd7cb85433ca9210182", "attempt_id": "b1788fa3618d", "run_group_id": "E4_core:A4|closed|qwen3b-llama3b:h2", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:07:47.250805Z", "task_family": "synthetic", "document_id": "syn_03006_2000", "document_hash": "89be5e77e0ca114963e9ffdf8dc38fe5610f13a78ac29e7a1956cd9124ce5217", "question_id": "q03_exact_copy", "split": "test", "cluster_id": "syn_03006_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "tokenizer_hashes": {"receiver": "b9d8872cf3a99da99e1fc4b0f71232a451632dc49ca4c6771549b4efa5c1c763"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 2, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1717, "query_tokens": 528, "message_tokens": 424, "message_utf8_bytes": 1071, "cache_tensor_bytes": null, "serialized_payload_bytes": 1071, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 39154.01008212939, "hops": {"1": 22491.57329101581, "2": 16662.436791113578}, "receiver_prefill_ms": 334.3408330110833, "ttft_ms": 32.14583394583315, "decode_ms": 193.68670892436057}, "total_wall_ms": 39714.866374153644, "ttft_ms": 32.14583394583315, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 8}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "Haruki Ouellet-Marchand", "evidence_artifact": {"ledger": {"facts_total": 64, "facts_retained": 30, "retention_rate": 0.46875, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.6666666666666666, "status": 0.9166666666666666, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1071}, "ledger_by_hop": {"1": {"facts_total": 64, "facts_retained": 30, "retention_rate": 0.46875, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.6666666666666666, "status": 0.9166666666666666, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1087}, "2": {"facts_total": 64, "facts_retained": 30, "retention_rate": 0.46875, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.6666666666666666, "status": 0.9166666666666666, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1071}}}, "reference_key_id": "keys/synthetic/syn_03006_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "e003c9569a8d38982c5059829bf26098eb9667ddf2655207f86179e344a0acf4", "output_hash": "a3b6e0a44d495387d8b8fc50ba6be0f2ed43121c674d3b3e962f8c4690e15b5c"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "ee443496c2665db2d56fec48", "attempt_id": "8091cd8a000e", "run_group_id": "E4_core:A4|closed|qwen3b-llama3b:h2", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:07:47.830189Z", "task_family": "synthetic", "document_id": "syn_03006_2000", "document_hash": "89be5e77e0ca114963e9ffdf8dc38fe5610f13a78ac29e7a1956cd9124ce5217", "question_id": "q04_multi_hop", "split": "test", "cluster_id": "syn_03006_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "tokenizer_hashes": {"receiver": "b9d8872cf3a99da99e1fc4b0f71232a451632dc49ca4c6771549b4efa5c1c763"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 2, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1717, "query_tokens": 537, "message_tokens": 424, "message_utf8_bytes": 1071, "cache_tensor_bytes": null, "serialized_payload_bytes": 1071, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 39154.01008212939, "hops": {"1": 22491.57329101581, "2": 16662.436791113578}, "receiver_prefill_ms": 317.88683391641825, "ttft_ms": 44.445459032431245, "decode_ms": 54.71258296165615}, "total_wall_ms": 39571.71770709101, "ttft_ms": 44.445459032431245, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 64, "facts_retained": 30, "retention_rate": 0.46875, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.6666666666666666, "status": 0.9166666666666666, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1071}, "ledger_by_hop": {"1": {"facts_total": 64, "facts_retained": 30, "retention_rate": 0.46875, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.6666666666666666, "status": 0.9166666666666666, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1087}, "2": {"facts_total": 64, "facts_retained": 30, "retention_rate": 0.46875, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.6666666666666666, "status": 0.9166666666666666, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1071}}}, "reference_key_id": "keys/synthetic/syn_03006_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "e003c9569a8d38982c5059829bf26098eb9667ddf2655207f86179e344a0acf4", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "e4e97b158251ba1eab4d40ea", "attempt_id": "33d4c69d1c15", "run_group_id": "E4_core:A4|closed|qwen3b-llama3b:h2", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:07:48.267345Z", "task_family": "synthetic", "document_id": "syn_03006_2000", "document_hash": "89be5e77e0ca114963e9ffdf8dc38fe5610f13a78ac29e7a1956cd9124ce5217", "question_id": "q05_multi_hop", "split": "test", "cluster_id": "syn_03006_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "tokenizer_hashes": {"receiver": "b9d8872cf3a99da99e1fc4b0f71232a451632dc49ca4c6771549b4efa5c1c763"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 2, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1717, "query_tokens": 538, "message_tokens": 424, "message_utf8_bytes": 1071, "cache_tensor_bytes": null, "serialized_payload_bytes": 1071, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 39154.01008212939, "hops": {"1": 22491.57329101581, "2": 16662.436791113578}, "receiver_prefill_ms": 473.01887499634176, "ttft_ms": 26.285875006578863, "decode_ms": 228.55729202274233}, "total_wall_ms": 39882.47087306809, "ttft_ms": 26.285875006578863, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 7}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "EP-52G-1330", "evidence_artifact": {"ledger": {"facts_total": 64, "facts_retained": 30, "retention_rate": 0.46875, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.6666666666666666, "status": 0.9166666666666666, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1071}, "ledger_by_hop": {"1": {"facts_total": 64, "facts_retained": 30, "retention_rate": 0.46875, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.6666666666666666, "status": 0.9166666666666666, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1087}, "2": {"facts_total": 64, "facts_retained": 30, "retention_rate": 0.46875, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.6666666666666666, "status": 0.9166666666666666, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1071}}}, "reference_key_id": "keys/synthetic/syn_03006_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "e003c9569a8d38982c5059829bf26098eb9667ddf2655207f86179e344a0acf4", "output_hash": "104db207937fd79e23ce16f9b3f10b1a9ed34f6dd0441a99be6ea1b2d0a09d99"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "0caa0fc21424abc681770da6", "attempt_id": "829fd7c0c6c0", "run_group_id": "E4_core:A4|closed|qwen3b-llama3b:h2", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:07:49.013827Z", "task_family": "synthetic", "document_id": "syn_03006_2000", "document_hash": "89be5e77e0ca114963e9ffdf8dc38fe5610f13a78ac29e7a1956cd9124ce5217", "question_id": "q06_negation", "split": "test", "cluster_id": "syn_03006_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "tokenizer_hashes": {"receiver": "b9d8872cf3a99da99e1fc4b0f71232a451632dc49ca4c6771549b4efa5c1c763"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 2, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1717, "query_tokens": 528, "message_tokens": 424, "message_utf8_bytes": 1071, "cache_tensor_bytes": null, "serialized_payload_bytes": 1071, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 39154.01008212939, "hops": {"1": 22491.57329101581, "2": 16662.436791113578}, "receiver_prefill_ms": 351.69550008140504, "ttft_ms": 24.88362498115748, "decode_ms": 52.94170801062137}, "total_wall_ms": 39584.11020715721, "ttft_ms": 24.88362498115748, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 2}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "No.", "evidence_artifact": {"ledger": {"facts_total": 64, "facts_retained": 30, "retention_rate": 0.46875, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.6666666666666666, "status": 0.9166666666666666, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1071}, "ledger_by_hop": {"1": {"facts_total": 64, "facts_retained": 30, "retention_rate": 0.46875, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.6666666666666666, "status": 0.9166666666666666, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1087}, "2": {"facts_total": 64, "facts_retained": 30, "retention_rate": 0.46875, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.6666666666666666, "status": 0.9166666666666666, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1071}}}, "reference_key_id": "keys/synthetic/syn_03006_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "e003c9569a8d38982c5059829bf26098eb9667ddf2655207f86179e344a0acf4", "output_hash": "38178a20b470cfd18299fb1593dd4ba706f1a83321dac051d01086efd8b7a96f"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "a93dbd89a2e1594ac98ad88a", "attempt_id": "c44796fa4a48", "run_group_id": "E4_core:A4|closed|qwen3b-llama3b:h2", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:07:49.461847Z", "task_family": "synthetic", "document_id": "syn_03006_2000", "document_hash": "89be5e77e0ca114963e9ffdf8dc38fe5610f13a78ac29e7a1956cd9124ce5217", "question_id": "q07_negation", "split": "test", "cluster_id": "syn_03006_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "tokenizer_hashes": {"receiver": "b9d8872cf3a99da99e1fc4b0f71232a451632dc49ca4c6771549b4efa5c1c763"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 2, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1717, "query_tokens": 528, "message_tokens": 424, "message_utf8_bytes": 1071, "cache_tensor_bytes": null, "serialized_payload_bytes": 1071, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 39154.01008212939, "hops": {"1": 22491.57329101581, "2": 16662.436791113578}, "receiver_prefill_ms": 316.45124999340624, "ttft_ms": 45.654417015612125, "decode_ms": 91.78741602227092}, "total_wall_ms": 39608.82529010996, "ttft_ms": 45.654417015612125, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 2}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "No.", "evidence_artifact": {"ledger": {"facts_total": 64, "facts_retained": 30, "retention_rate": 0.46875, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.6666666666666666, "status": 0.9166666666666666, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1071}, "ledger_by_hop": {"1": {"facts_total": 64, "facts_retained": 30, "retention_rate": 0.46875, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.6666666666666666, "status": 0.9166666666666666, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1087}, "2": {"facts_total": 64, "facts_retained": 30, "retention_rate": 0.46875, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.6666666666666666, "status": 0.9166666666666666, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1071}}}, "reference_key_id": "keys/synthetic/syn_03006_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "e003c9569a8d38982c5059829bf26098eb9667ddf2655207f86179e344a0acf4", "output_hash": "38178a20b470cfd18299fb1593dd4ba706f1a83321dac051d01086efd8b7a96f"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "7b4d291295c702c87146927e", "attempt_id": "14bf46e5ffe3", "run_group_id": "E4_core:A4|closed|qwen3b-llama3b:h2", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:07:49.935045Z", "task_family": "synthetic", "document_id": "syn_03006_2000", "document_hash": "89be5e77e0ca114963e9ffdf8dc38fe5610f13a78ac29e7a1956cd9124ce5217", "question_id": "q08_revision", "split": "test", "cluster_id": "syn_03006_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "tokenizer_hashes": {"receiver": "b9d8872cf3a99da99e1fc4b0f71232a451632dc49ca4c6771549b4efa5c1c763"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 2, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1717, "query_tokens": 544, "message_tokens": 424, "message_utf8_bytes": 1071, "cache_tensor_bytes": null, "serialized_payload_bytes": 1071, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 39154.01008212939, "hops": {"1": 22491.57329101581, "2": 16662.436791113578}, "receiver_prefill_ms": 461.74358401913196, "ttft_ms": 36.877083010040224, "decode_ms": 159.93820899166167}, "total_wall_ms": 39813.3189980872, "ttft_ms": 36.877083010040224, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 5}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "9,112,735", "evidence_artifact": {"ledger": {"facts_total": 64, "facts_retained": 30, "retention_rate": 0.46875, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.6666666666666666, "status": 0.9166666666666666, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1071}, "ledger_by_hop": {"1": {"facts_total": 64, "facts_retained": 30, "retention_rate": 0.46875, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.6666666666666666, "status": 0.9166666666666666, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1087}, "2": {"facts_total": 64, "facts_retained": 30, "retention_rate": 0.46875, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.6666666666666666, "status": 0.9166666666666666, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1071}}}, "reference_key_id": "keys/synthetic/syn_03006_2000.json", "task_scores": {"correct": 0.0, "abstained": 0.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "e003c9569a8d38982c5059829bf26098eb9667ddf2655207f86179e344a0acf4", "output_hash": "788fcda908c2029597420995b09d0cc919bfdbf7a28b68dcb0e331db7df7b342"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "b3e030ea37f78796fc52c831", "attempt_id": "9ec21b242503", "run_group_id": "E4_core:A4|closed|qwen3b-llama3b:h2", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:07:50.612502Z", "task_family": "synthetic", "document_id": "syn_03006_2000", "document_hash": "89be5e77e0ca114963e9ffdf8dc38fe5610f13a78ac29e7a1956cd9124ce5217", "question_id": "q09_revision", "split": "test", "cluster_id": "syn_03006_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "tokenizer_hashes": {"receiver": "b9d8872cf3a99da99e1fc4b0f71232a451632dc49ca4c6771549b4efa5c1c763"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 2, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1717, "query_tokens": 537, "message_tokens": 424, "message_utf8_bytes": 1071, "cache_tensor_bytes": null, "serialized_payload_bytes": 1071, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 39154.01008212939, "hops": {"1": 22491.57329101581, "2": 16662.436791113578}, "receiver_prefill_ms": 338.1000829394907, "ttft_ms": 36.68750007636845, "decode_ms": 141.6155000915751}, "total_wall_ms": 39671.10229108948, "ttft_ms": 36.68750007636845, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 5}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "9,112,735", "evidence_artifact": {"ledger": {"facts_total": 64, "facts_retained": 30, "retention_rate": 0.46875, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.6666666666666666, "status": 0.9166666666666666, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1071}, "ledger_by_hop": {"1": {"facts_total": 64, "facts_retained": 30, "retention_rate": 0.46875, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.6666666666666666, "status": 0.9166666666666666, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1087}, "2": {"facts_total": 64, "facts_retained": 30, "retention_rate": 0.46875, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.6666666666666666, "status": 0.9166666666666666, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1071}}}, "reference_key_id": "keys/synthetic/syn_03006_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "e003c9569a8d38982c5059829bf26098eb9667ddf2655207f86179e344a0acf4", "output_hash": "788fcda908c2029597420995b09d0cc919bfdbf7a28b68dcb0e331db7df7b342"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "9406c488042361f8f08f5189", "attempt_id": "a8e129435767", "run_group_id": "E4_core:A4|closed|qwen3b-llama3b:h2", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:07:51.146971Z", "task_family": "synthetic", "document_id": "syn_03006_2000", "document_hash": "89be5e77e0ca114963e9ffdf8dc38fe5610f13a78ac29e7a1956cd9124ce5217", "question_id": "q10_temporal", "split": "test", "cluster_id": "syn_03006_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "tokenizer_hashes": {"receiver": "b9d8872cf3a99da99e1fc4b0f71232a451632dc49ca4c6771549b4efa5c1c763"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 2, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1717, "query_tokens": 546, "message_tokens": 424, "message_utf8_bytes": 1071, "cache_tensor_bytes": null, "serialized_payload_bytes": 1071, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 39154.01008212939, "hops": {"1": 22491.57329101581, "2": 16662.436791113578}, "receiver_prefill_ms": 310.9325839905068, "ttft_ms": 34.92983407340944, "decode_ms": 35.50583298783749}, "total_wall_ms": 39536.148624145426, "ttft_ms": 34.92983407340944, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 64, "facts_retained": 30, "retention_rate": 0.46875, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.6666666666666666, "status": 0.9166666666666666, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1071}, "ledger_by_hop": {"1": {"facts_total": 64, "facts_retained": 30, "retention_rate": 0.46875, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.6666666666666666, "status": 0.9166666666666666, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1087}, "2": {"facts_total": 64, "facts_retained": 30, "retention_rate": 0.46875, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.6666666666666666, "status": 0.9166666666666666, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1071}}}, "reference_key_id": "keys/synthetic/syn_03006_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "e003c9569a8d38982c5059829bf26098eb9667ddf2655207f86179e344a0acf4", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "9243220d613c891c72495c9b", "attempt_id": "a9b44f2c71f5", "run_group_id": "E4_core:A4|closed|qwen3b-llama3b:h2", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:07:51.546782Z", "task_family": "synthetic", "document_id": "syn_03006_2000", "document_hash": "89be5e77e0ca114963e9ffdf8dc38fe5610f13a78ac29e7a1956cd9124ce5217", "question_id": "q11_aggregation", "split": "test", "cluster_id": "syn_03006_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "tokenizer_hashes": {"receiver": "b9d8872cf3a99da99e1fc4b0f71232a451632dc49ca4c6771549b4efa5c1c763"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 2, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1717, "query_tokens": 540, "message_tokens": 424, "message_utf8_bytes": 1071, "cache_tensor_bytes": null, "serialized_payload_bytes": 1071, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 39154.01008212939, "hops": {"1": 22491.57329101581, "2": 16662.436791113578}, "receiver_prefill_ms": 494.9257089756429, "ttft_ms": 32.33699989505112, "decode_ms": 158.8457499165088}, "total_wall_ms": 39840.81562317442, "ttft_ms": 32.33699989505112, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 5}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "8,908,553", "evidence_artifact": {"ledger": {"facts_total": 64, "facts_retained": 30, "retention_rate": 0.46875, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.6666666666666666, "status": 0.9166666666666666, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1071}, "ledger_by_hop": {"1": {"facts_total": 64, "facts_retained": 30, "retention_rate": 0.46875, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.6666666666666666, "status": 0.9166666666666666, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1087}, "2": {"facts_total": 64, "facts_retained": 30, "retention_rate": 0.46875, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.6666666666666666, "status": 0.9166666666666666, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1071}}}, "reference_key_id": "keys/synthetic/syn_03006_2000.json", "task_scores": {"correct": 0.0, "abstained": 0.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "e003c9569a8d38982c5059829bf26098eb9667ddf2655207f86179e344a0acf4", "output_hash": "cc434df580a6d88109305257f4ffd4ffb654a607abcff9be07a7507510825b63"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "635ac485a80c797d6ce4ab27", "attempt_id": "265c5fa57933", "run_group_id": "E4_core:A4|closed|qwen3b-llama3b:h2", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:07:52.251293Z", "task_family": "synthetic", "document_id": "syn_03006_2000", "document_hash": "89be5e77e0ca114963e9ffdf8dc38fe5610f13a78ac29e7a1956cd9124ce5217", "question_id": "q12_unit_conversion", "split": "test", "cluster_id": "syn_03006_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "tokenizer_hashes": {"receiver": "b9d8872cf3a99da99e1fc4b0f71232a451632dc49ca4c6771549b4efa5c1c763"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 2, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1717, "query_tokens": 536, "message_tokens": 424, "message_utf8_bytes": 1071, "cache_tensor_bytes": null, "serialized_payload_bytes": 1071, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 39154.01008212939, "hops": {"1": 22491.57329101581, "2": 16662.436791113578}, "receiver_prefill_ms": 337.40391698665917, "ttft_ms": 30.406042002141476, "decode_ms": 94.18629191350192}, "total_wall_ms": 39616.5396241704, "ttft_ms": 30.406042002141476, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 3}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "39,700", "evidence_artifact": {"ledger": {"facts_total": 64, "facts_retained": 30, "retention_rate": 0.46875, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.6666666666666666, "status": 0.9166666666666666, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1071}, "ledger_by_hop": {"1": {"facts_total": 64, "facts_retained": 30, "retention_rate": 0.46875, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.6666666666666666, "status": 0.9166666666666666, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1087}, "2": {"facts_total": 64, "facts_retained": 30, "retention_rate": 0.46875, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.6666666666666666, "status": 0.9166666666666666, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1071}}}, "reference_key_id": "keys/synthetic/syn_03006_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "e003c9569a8d38982c5059829bf26098eb9667ddf2655207f86179e344a0acf4", "output_hash": "5f62e7c00eed4af414e81f6c0d51672fbb9393169f5091e5137429e7a69e9a6d"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "fbb256fc79c3a99a33eaa107", "attempt_id": "641a4539b3fc", "run_group_id": "E4_core:A4|closed|qwen3b-llama3b:h2", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:07:52.732919Z", "task_family": "synthetic", "document_id": "syn_03006_2000", "document_hash": "89be5e77e0ca114963e9ffdf8dc38fe5610f13a78ac29e7a1956cd9124ce5217", "question_id": "q13_unanswerable", "split": "test", "cluster_id": "syn_03006_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "tokenizer_hashes": {"receiver": "b9d8872cf3a99da99e1fc4b0f71232a451632dc49ca4c6771549b4efa5c1c763"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 2, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1717, "query_tokens": 540, "message_tokens": 424, "message_utf8_bytes": 1071, "cache_tensor_bytes": null, "serialized_payload_bytes": 1071, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 39154.01008212939, "hops": {"1": 22491.57329101581, "2": 16662.436791113578}, "receiver_prefill_ms": 304.5865000458434, "ttft_ms": 42.67841705586761, "decode_ms": 365.18133291974664}, "total_wall_ms": 39867.26216610987, "ttft_ms": 42.67841705586761, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 7}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "XM-76Q-2958", "evidence_artifact": {"ledger": {"facts_total": 64, "facts_retained": 30, "retention_rate": 0.46875, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.6666666666666666, "status": 0.9166666666666666, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1071}, "ledger_by_hop": {"1": {"facts_total": 64, "facts_retained": 30, "retention_rate": 0.46875, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.6666666666666666, "status": 0.9166666666666666, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1087}, "2": {"facts_total": 64, "facts_retained": 30, "retention_rate": 0.46875, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.6666666666666666, "status": 0.9166666666666666, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1071}}}, "reference_key_id": "keys/synthetic/syn_03006_2000.json", "task_scores": {"correct": 0.0, "abstained": 0.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "e003c9569a8d38982c5059829bf26098eb9667ddf2655207f86179e344a0acf4", "output_hash": "cf42691eec772b9ec74d67643944de2e19e53724d7b772cef1e0f2d5ef3b4926"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "cd1b61ff515a3536c9602984", "attempt_id": "99f5bd7c5849", "run_group_id": "E4_core:A4|closed|qwen3b-llama3b:h2", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:07:53.463555Z", "task_family": "synthetic", "document_id": "syn_03006_2000", "document_hash": "89be5e77e0ca114963e9ffdf8dc38fe5610f13a78ac29e7a1956cd9124ce5217", "question_id": "q14_unanswerable", "split": "test", "cluster_id": "syn_03006_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "tokenizer_hashes": {"receiver": "b9d8872cf3a99da99e1fc4b0f71232a451632dc49ca4c6771549b4efa5c1c763"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 2, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1717, "query_tokens": 542, "message_tokens": 424, "message_utf8_bytes": 1071, "cache_tensor_bytes": null, "serialized_payload_bytes": 1071, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 39154.01008212939, "hops": {"1": 22491.57329101581, "2": 16662.436791113578}, "receiver_prefill_ms": 409.6249589929357, "ttft_ms": 36.18070890661329, "decode_ms": 24.38237506430596}, "total_wall_ms": 39624.77020721417, "ttft_ms": 36.18070890661329, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 64, "facts_retained": 30, "retention_rate": 0.46875, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.6666666666666666, "status": 0.9166666666666666, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1071}, "ledger_by_hop": {"1": {"facts_total": 64, "facts_retained": 30, "retention_rate": 0.46875, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.6666666666666666, "status": 0.9166666666666666, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1087}, "2": {"facts_total": 64, "facts_retained": 30, "retention_rate": 0.46875, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.6666666666666666, "status": 0.9166666666666666, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1071}}}, "reference_key_id": "keys/synthetic/syn_03006_2000.json", "task_scores": {"correct": 1.0, "abstained": 1.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "e003c9569a8d38982c5059829bf26098eb9667ddf2655207f86179e344a0acf4", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "a40f029a849c1cf20cfbd6b8", "attempt_id": "ecd3f41fddee", "run_group_id": "E4_core:A4|closed|qwen3b-llama3b:h4", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:08:28.531705Z", "task_family": "synthetic", "document_id": "syn_03006_2000", "document_hash": "89be5e77e0ca114963e9ffdf8dc38fe5610f13a78ac29e7a1956cd9124ce5217", "question_id": "q00_single_fact", "split": "test", "cluster_id": "syn_03006_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "tokenizer_hashes": {"receiver": "b9d8872cf3a99da99e1fc4b0f71232a451632dc49ca4c6771549b4efa5c1c763"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 4, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1717, "query_tokens": 509, "message_tokens": 406, "message_utf8_bytes": 1250, "cache_tensor_bytes": null, "serialized_payload_bytes": 1250, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 73730.44870712329, "hops": {"1": 22491.57329101581, "2": 16662.436791113578, "3": 18511.050833039917, "4": 16065.387791953981}, "receiver_prefill_ms": 352.72900003474206, "ttft_ms": 39.23687501810491, "decode_ms": 403.11762504279613}, "total_wall_ms": 74526.24554105569, "ttft_ms": 39.23687501810491, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 7}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "XM-76Q-2958", "evidence_artifact": {"ledger": {"facts_total": 64, "facts_retained": 30, "retention_rate": 0.46875, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.6666666666666666, "status": 0.9166666666666666, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1250}, "ledger_by_hop": {"1": {"facts_total": 64, "facts_retained": 30, "retention_rate": 0.46875, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.6666666666666666, "status": 0.9166666666666666, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1087}, "2": {"facts_total": 64, "facts_retained": 30, "retention_rate": 0.46875, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.6666666666666666, "status": 0.9166666666666666, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1071}, "3": {"facts_total": 64, "facts_retained": 30, "retention_rate": 0.46875, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.6666666666666666, "status": 0.9166666666666666, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 962}, "4": {"facts_total": 64, "facts_retained": 30, "retention_rate": 0.46875, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.6666666666666666, "status": 0.9166666666666666, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1250}}}, "reference_key_id": "keys/synthetic/syn_03006_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "a9ee5c99d37fccc8db66787ad20a94b821516c6fb6542156bc7cac20ae86eb2f", "output_hash": "cf42691eec772b9ec74d67643944de2e19e53724d7b772cef1e0f2d5ef3b4926"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "51534f53715363fdc4753227", "attempt_id": "eaafaae575cb", "run_group_id": "E4_core:A4|closed|qwen3b-llama3b:h4", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:08:29.345201Z", "task_family": "synthetic", "document_id": "syn_03006_2000", "document_hash": "89be5e77e0ca114963e9ffdf8dc38fe5610f13a78ac29e7a1956cd9124ce5217", "question_id": "q01_single_fact", "split": "test", "cluster_id": "syn_03006_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "tokenizer_hashes": {"receiver": "b9d8872cf3a99da99e1fc4b0f71232a451632dc49ca4c6771549b4efa5c1c763"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 4, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1717, "query_tokens": 515, "message_tokens": 406, "message_utf8_bytes": 1250, "cache_tensor_bytes": null, "serialized_payload_bytes": 1250, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 73730.44870712329, "hops": {"1": 22491.57329101581, "2": 16662.436791113578, "3": 18511.050833039917, "4": 16065.387791953981}, "receiver_prefill_ms": 410.8776670182124, "ttft_ms": 37.44929097592831, "decode_ms": 24.80187499895692}, "total_wall_ms": 74204.10041511059, "ttft_ms": 37.44929097592831, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 64, "facts_retained": 30, "retention_rate": 0.46875, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.6666666666666666, "status": 0.9166666666666666, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1250}, "ledger_by_hop": {"1": {"facts_total": 64, "facts_retained": 30, "retention_rate": 0.46875, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.6666666666666666, "status": 0.9166666666666666, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1087}, "2": {"facts_total": 64, "facts_retained": 30, "retention_rate": 0.46875, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.6666666666666666, "status": 0.9166666666666666, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1071}, "3": {"facts_total": 64, "facts_retained": 30, "retention_rate": 0.46875, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.6666666666666666, "status": 0.9166666666666666, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 962}, "4": {"facts_total": 64, "facts_retained": 30, "retention_rate": 0.46875, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.6666666666666666, "status": 0.9166666666666666, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1250}}}, "reference_key_id": "keys/synthetic/syn_03006_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "a9ee5c99d37fccc8db66787ad20a94b821516c6fb6542156bc7cac20ae86eb2f", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "44204257ef1a917fca2b263e", "attempt_id": "ea7bfcef99a5", "run_group_id": "E4_core:A4|closed|qwen3b-llama3b:h4", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:08:29.838270Z", "task_family": "synthetic", "document_id": "syn_03006_2000", "document_hash": "89be5e77e0ca114963e9ffdf8dc38fe5610f13a78ac29e7a1956cd9124ce5217", "question_id": "q02_exact_copy", "split": "test", "cluster_id": "syn_03006_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "tokenizer_hashes": {"receiver": "b9d8872cf3a99da99e1fc4b0f71232a451632dc49ca4c6771549b4efa5c1c763"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 4, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1717, "query_tokens": 514, "message_tokens": 406, "message_utf8_bytes": 1250, "cache_tensor_bytes": null, "serialized_payload_bytes": 1250, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 73730.44870712329, "hops": {"1": 22491.57329101581, "2": 16662.436791113578, "3": 18511.050833039917, "4": 16065.387791953981}, "receiver_prefill_ms": 338.59104209113866, "ttft_ms": 31.850541941821575, "decode_ms": 30.65033396705985}, "total_wall_ms": 74132.07245711237, "ttft_ms": 31.850541941821575, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 64, "facts_retained": 30, "retention_rate": 0.46875, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.6666666666666666, "status": 0.9166666666666666, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1250}, "ledger_by_hop": {"1": {"facts_total": 64, "facts_retained": 30, "retention_rate": 0.46875, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.6666666666666666, "status": 0.9166666666666666, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1087}, "2": {"facts_total": 64, "facts_retained": 30, "retention_rate": 0.46875, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.6666666666666666, "status": 0.9166666666666666, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1071}, "3": {"facts_total": 64, "facts_retained": 30, "retention_rate": 0.46875, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.6666666666666666, "status": 0.9166666666666666, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 962}, "4": {"facts_total": 64, "facts_retained": 30, "retention_rate": 0.46875, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.6666666666666666, "status": 0.9166666666666666, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1250}}}, "reference_key_id": "keys/synthetic/syn_03006_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "a9ee5c99d37fccc8db66787ad20a94b821516c6fb6542156bc7cac20ae86eb2f", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "bb3c7556c5c9f095846a64d7", "attempt_id": "264540f78ec6", "run_group_id": "E4_core:A4|closed|qwen3b-llama3b:h4", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:08:30.256426Z", "task_family": "synthetic", "document_id": "syn_03006_2000", "document_hash": "89be5e77e0ca114963e9ffdf8dc38fe5610f13a78ac29e7a1956cd9124ce5217", "question_id": "q03_exact_copy", "split": "test", "cluster_id": "syn_03006_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "tokenizer_hashes": {"receiver": "b9d8872cf3a99da99e1fc4b0f71232a451632dc49ca4c6771549b4efa5c1c763"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 4, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1717, "query_tokens": 509, "message_tokens": 406, "message_utf8_bytes": 1250, "cache_tensor_bytes": null, "serialized_payload_bytes": 1250, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 73730.44870712329, "hops": {"1": 22491.57329101581, "2": 16662.436791113578, "3": 18511.050833039917, "4": 16065.387791953981}, "receiver_prefill_ms": 340.34283296205103, "ttft_ms": 40.15116696245968, "decode_ms": 505.953999934718}, "total_wall_ms": 74617.48458212242, "ttft_ms": 40.15116696245968, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 8}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "Haruki Ouellet-Marchand", "evidence_artifact": {"ledger": {"facts_total": 64, "facts_retained": 30, "retention_rate": 0.46875, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.6666666666666666, "status": 0.9166666666666666, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1250}, "ledger_by_hop": {"1": {"facts_total": 64, "facts_retained": 30, "retention_rate": 0.46875, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.6666666666666666, "status": 0.9166666666666666, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1087}, "2": {"facts_total": 64, "facts_retained": 30, "retention_rate": 0.46875, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.6666666666666666, "status": 0.9166666666666666, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1071}, "3": {"facts_total": 64, "facts_retained": 30, "retention_rate": 0.46875, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.6666666666666666, "status": 0.9166666666666666, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 962}, "4": {"facts_total": 64, "facts_retained": 30, "retention_rate": 0.46875, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.6666666666666666, "status": 0.9166666666666666, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1250}}}, "reference_key_id": "keys/synthetic/syn_03006_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "a9ee5c99d37fccc8db66787ad20a94b821516c6fb6542156bc7cac20ae86eb2f", "output_hash": "a3b6e0a44d495387d8b8fc50ba6be0f2ed43121c674d3b3e962f8c4690e15b5c"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "b6eb185cf68ab401974b2bf7", "attempt_id": "6eab0b5b9c9d", "run_group_id": "E4_core:A4|closed|qwen3b-llama3b:h4", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:08:31.160723Z", "task_family": "synthetic", "document_id": "syn_03006_2000", "document_hash": "89be5e77e0ca114963e9ffdf8dc38fe5610f13a78ac29e7a1956cd9124ce5217", "question_id": "q04_multi_hop", "split": "test", "cluster_id": "syn_03006_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "tokenizer_hashes": {"receiver": "b9d8872cf3a99da99e1fc4b0f71232a451632dc49ca4c6771549b4efa5c1c763"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 4, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1717, "query_tokens": 518, "message_tokens": 406, "message_utf8_bytes": 1250, "cache_tensor_bytes": null, "serialized_payload_bytes": 1250, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 73730.44870712329, "hops": {"1": 22491.57329101581, "2": 16662.436791113578, "3": 18511.050833039917, "4": 16065.387791953981}, "receiver_prefill_ms": 331.298999954015, "ttft_ms": 40.37341696675867, "decode_ms": 24.100958020426333}, "total_wall_ms": 74126.83358206414, "ttft_ms": 40.37341696675867, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 64, "facts_retained": 30, "retention_rate": 0.46875, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.6666666666666666, "status": 0.9166666666666666, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1250}, "ledger_by_hop": {"1": {"facts_total": 64, "facts_retained": 30, "retention_rate": 0.46875, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.6666666666666666, "status": 0.9166666666666666, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1087}, "2": {"facts_total": 64, "facts_retained": 30, "retention_rate": 0.46875, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.6666666666666666, "status": 0.9166666666666666, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1071}, "3": {"facts_total": 64, "facts_retained": 30, "retention_rate": 0.46875, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.6666666666666666, "status": 0.9166666666666666, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 962}, "4": {"facts_total": 64, "facts_retained": 30, "retention_rate": 0.46875, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.6666666666666666, "status": 0.9166666666666666, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1250}}}, "reference_key_id": "keys/synthetic/syn_03006_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "a9ee5c99d37fccc8db66787ad20a94b821516c6fb6542156bc7cac20ae86eb2f", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "4a3c71aa594bf788b1c8f8dc", "attempt_id": "af5c3906b66f", "run_group_id": "E4_core:A4|closed|qwen3b-llama3b:h4", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:08:31.574611Z", "task_family": "synthetic", "document_id": "syn_03006_2000", "document_hash": "89be5e77e0ca114963e9ffdf8dc38fe5610f13a78ac29e7a1956cd9124ce5217", "question_id": "q05_multi_hop", "split": "test", "cluster_id": "syn_03006_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "tokenizer_hashes": {"receiver": "b9d8872cf3a99da99e1fc4b0f71232a451632dc49ca4c6771549b4efa5c1c763"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 4, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1717, "query_tokens": 519, "message_tokens": 406, "message_utf8_bytes": 1250, "cache_tensor_bytes": null, "serialized_payload_bytes": 1250, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 73730.44870712329, "hops": {"1": 22491.57329101581, "2": 16662.436791113578, "3": 18511.050833039917, "4": 16065.387791953981}, "receiver_prefill_ms": 336.9474159553647, "ttft_ms": 38.12599997036159, "decode_ms": 179.37370797153562}, "total_wall_ms": 74285.50199919846, "ttft_ms": 38.12599997036159, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 7}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "EP-52G-1330", "evidence_artifact": {"ledger": {"facts_total": 64, "facts_retained": 30, "retention_rate": 0.46875, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.6666666666666666, "status": 0.9166666666666666, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1250}, "ledger_by_hop": {"1": {"facts_total": 64, "facts_retained": 30, "retention_rate": 0.46875, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.6666666666666666, "status": 0.9166666666666666, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1087}, "2": {"facts_total": 64, "facts_retained": 30, "retention_rate": 0.46875, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.6666666666666666, "status": 0.9166666666666666, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1071}, "3": {"facts_total": 64, "facts_retained": 30, "retention_rate": 0.46875, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.6666666666666666, "status": 0.9166666666666666, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 962}, "4": {"facts_total": 64, "facts_retained": 30, "retention_rate": 0.46875, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.6666666666666666, "status": 0.9166666666666666, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1250}}}, "reference_key_id": "keys/synthetic/syn_03006_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "a9ee5c99d37fccc8db66787ad20a94b821516c6fb6542156bc7cac20ae86eb2f", "output_hash": "104db207937fd79e23ce16f9b3f10b1a9ed34f6dd0441a99be6ea1b2d0a09d99"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "a98601a0462000859121e2c8", "attempt_id": "071c965a05b6", "run_group_id": "E4_core:A4|closed|qwen3b-llama3b:h4", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:08:32.147258Z", "task_family": "synthetic", "document_id": "syn_03006_2000", "document_hash": "89be5e77e0ca114963e9ffdf8dc38fe5610f13a78ac29e7a1956cd9124ce5217", "question_id": "q06_negation", "split": "test", "cluster_id": "syn_03006_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "tokenizer_hashes": {"receiver": "b9d8872cf3a99da99e1fc4b0f71232a451632dc49ca4c6771549b4efa5c1c763"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 4, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1717, "query_tokens": 509, "message_tokens": 406, "message_utf8_bytes": 1250, "cache_tensor_bytes": null, "serialized_payload_bytes": 1250, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 73730.44870712329, "hops": {"1": 22491.57329101581, "2": 16662.436791113578, "3": 18511.050833039917, "4": 16065.387791953981}, "receiver_prefill_ms": 363.7777080293745, "ttft_ms": 67.95995903667063, "decode_ms": 238.68679103907198}, "total_wall_ms": 74401.60849911626, "ttft_ms": 67.95995903667063, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 2}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "No.", "evidence_artifact": {"ledger": {"facts_total": 64, "facts_retained": 30, "retention_rate": 0.46875, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.6666666666666666, "status": 0.9166666666666666, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1250}, "ledger_by_hop": {"1": {"facts_total": 64, "facts_retained": 30, "retention_rate": 0.46875, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.6666666666666666, "status": 0.9166666666666666, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1087}, "2": {"facts_total": 64, "facts_retained": 30, "retention_rate": 0.46875, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.6666666666666666, "status": 0.9166666666666666, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1071}, "3": {"facts_total": 64, "facts_retained": 30, "retention_rate": 0.46875, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.6666666666666666, "status": 0.9166666666666666, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 962}, "4": {"facts_total": 64, "facts_retained": 30, "retention_rate": 0.46875, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.6666666666666666, "status": 0.9166666666666666, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1250}}}, "reference_key_id": "keys/synthetic/syn_03006_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "a9ee5c99d37fccc8db66787ad20a94b821516c6fb6542156bc7cac20ae86eb2f", "output_hash": "38178a20b470cfd18299fb1593dd4ba706f1a83321dac051d01086efd8b7a96f"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "7dda858fecd67dc14d1d49c7", "attempt_id": "3c962b62bd38", "run_group_id": "E4_core:A4|closed|qwen3b-llama3b:h4", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:08:32.837549Z", "task_family": "synthetic", "document_id": "syn_03006_2000", "document_hash": "89be5e77e0ca114963e9ffdf8dc38fe5610f13a78ac29e7a1956cd9124ce5217", "question_id": "q07_negation", "split": "test", "cluster_id": "syn_03006_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "tokenizer_hashes": {"receiver": "b9d8872cf3a99da99e1fc4b0f71232a451632dc49ca4c6771549b4efa5c1c763"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 4, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1717, "query_tokens": 509, "message_tokens": 406, "message_utf8_bytes": 1250, "cache_tensor_bytes": null, "serialized_payload_bytes": 1250, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 73730.44870712329, "hops": {"1": 22491.57329101581, "2": 16662.436791113578, "3": 18511.050833039917, "4": 16065.387791953981}, "receiver_prefill_ms": 364.1203750157729, "ttft_ms": 26.062333956360817, "decode_ms": 78.76279193442315}, "total_wall_ms": 74200.00420708675, "ttft_ms": 26.062333956360817, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 2}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "No.", "evidence_artifact": {"ledger": {"facts_total": 64, "facts_retained": 30, "retention_rate": 0.46875, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.6666666666666666, "status": 0.9166666666666666, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1250}, "ledger_by_hop": {"1": {"facts_total": 64, "facts_retained": 30, "retention_rate": 0.46875, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.6666666666666666, "status": 0.9166666666666666, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1087}, "2": {"facts_total": 64, "facts_retained": 30, "retention_rate": 0.46875, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.6666666666666666, "status": 0.9166666666666666, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1071}, "3": {"facts_total": 64, "facts_retained": 30, "retention_rate": 0.46875, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.6666666666666666, "status": 0.9166666666666666, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 962}, "4": {"facts_total": 64, "facts_retained": 30, "retention_rate": 0.46875, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.6666666666666666, "status": 0.9166666666666666, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1250}}}, "reference_key_id": "keys/synthetic/syn_03006_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "a9ee5c99d37fccc8db66787ad20a94b821516c6fb6542156bc7cac20ae86eb2f", "output_hash": "38178a20b470cfd18299fb1593dd4ba706f1a83321dac051d01086efd8b7a96f"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "16a450e0a92c4bf776455445", "attempt_id": "1b6c51e469d3", "run_group_id": "E4_core:A4|closed|qwen3b-llama3b:h4", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:08:33.324187Z", "task_family": "synthetic", "document_id": "syn_03006_2000", "document_hash": "89be5e77e0ca114963e9ffdf8dc38fe5610f13a78ac29e7a1956cd9124ce5217", "question_id": "q08_revision", "split": "test", "cluster_id": "syn_03006_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "tokenizer_hashes": {"receiver": "b9d8872cf3a99da99e1fc4b0f71232a451632dc49ca4c6771549b4efa5c1c763"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 4, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1717, "query_tokens": 525, "message_tokens": 406, "message_utf8_bytes": 1250, "cache_tensor_bytes": null, "serialized_payload_bytes": 1250, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 73730.44870712329, "hops": {"1": 22491.57329101581, "2": 16662.436791113578, "3": 18511.050833039917, "4": 16065.387791953981}, "receiver_prefill_ms": 349.14399997796863, "ttft_ms": 25.7574999704957, "decode_ms": 160.86275002453476}, "total_wall_ms": 74266.98920712806, "ttft_ms": 25.7574999704957, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 5}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "10,107,545", "evidence_artifact": {"ledger": {"facts_total": 64, "facts_retained": 30, "retention_rate": 0.46875, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.6666666666666666, "status": 0.9166666666666666, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1250}, "ledger_by_hop": {"1": {"facts_total": 64, "facts_retained": 30, "retention_rate": 0.46875, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.6666666666666666, "status": 0.9166666666666666, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1087}, "2": {"facts_total": 64, "facts_retained": 30, "retention_rate": 0.46875, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.6666666666666666, "status": 0.9166666666666666, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1071}, "3": {"facts_total": 64, "facts_retained": 30, "retention_rate": 0.46875, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.6666666666666666, "status": 0.9166666666666666, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 962}, "4": {"facts_total": 64, "facts_retained": 30, "retention_rate": 0.46875, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.6666666666666666, "status": 0.9166666666666666, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1250}}}, "reference_key_id": "keys/synthetic/syn_03006_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "a9ee5c99d37fccc8db66787ad20a94b821516c6fb6542156bc7cac20ae86eb2f", "output_hash": "3d6a277763361ef55b8dd3f9791012605d24cd167ac5bdcdd0064ee16de161d4"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "3359a611db0650667a53fd4d", "attempt_id": "7a4a0e0f1d49", "run_group_id": "E4_core:A4|closed|qwen3b-llama3b:h4", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:08:33.878921Z", "task_family": "synthetic", "document_id": "syn_03006_2000", "document_hash": "89be5e77e0ca114963e9ffdf8dc38fe5610f13a78ac29e7a1956cd9124ce5217", "question_id": "q09_revision", "split": "test", "cluster_id": "syn_03006_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "tokenizer_hashes": {"receiver": "b9d8872cf3a99da99e1fc4b0f71232a451632dc49ca4c6771549b4efa5c1c763"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 4, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1717, "query_tokens": 518, "message_tokens": 406, "message_utf8_bytes": 1250, "cache_tensor_bytes": null, "serialized_payload_bytes": 1250, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 73730.44870712329, "hops": {"1": 22491.57329101581, "2": 16662.436791113578, "3": 18511.050833039917, "4": 16065.387791953981}, "receiver_prefill_ms": 317.9451250471175, "ttft_ms": 44.042458990588784, "decode_ms": 315.28420792892575}, "total_wall_ms": 74408.3311650902, "ttft_ms": 44.042458990588784, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 5}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "9,112,735", "evidence_artifact": {"ledger": {"facts_total": 64, "facts_retained": 30, "retention_rate": 0.46875, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.6666666666666666, "status": 0.9166666666666666, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1250}, "ledger_by_hop": {"1": {"facts_total": 64, "facts_retained": 30, "retention_rate": 0.46875, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.6666666666666666, "status": 0.9166666666666666, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1087}, "2": {"facts_total": 64, "facts_retained": 30, "retention_rate": 0.46875, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.6666666666666666, "status": 0.9166666666666666, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1071}, "3": {"facts_total": 64, "facts_retained": 30, "retention_rate": 0.46875, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.6666666666666666, "status": 0.9166666666666666, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 962}, "4": {"facts_total": 64, "facts_retained": 30, "retention_rate": 0.46875, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.6666666666666666, "status": 0.9166666666666666, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1250}}}, "reference_key_id": "keys/synthetic/syn_03006_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "a9ee5c99d37fccc8db66787ad20a94b821516c6fb6542156bc7cac20ae86eb2f", "output_hash": "788fcda908c2029597420995b09d0cc919bfdbf7a28b68dcb0e331db7df7b342"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "9a9979d6c55ce7751efa45cf", "attempt_id": "f62a68dd09cd", "run_group_id": "E4_core:A4|closed|qwen3b-llama3b:h4", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:08:34.574156Z", "task_family": "synthetic", "document_id": "syn_03006_2000", "document_hash": "89be5e77e0ca114963e9ffdf8dc38fe5610f13a78ac29e7a1956cd9124ce5217", "question_id": "q10_temporal", "split": "test", "cluster_id": "syn_03006_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "tokenizer_hashes": {"receiver": "b9d8872cf3a99da99e1fc4b0f71232a451632dc49ca4c6771549b4efa5c1c763"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 4, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1717, "query_tokens": 527, "message_tokens": 406, "message_utf8_bytes": 1250, "cache_tensor_bytes": null, "serialized_payload_bytes": 1250, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 73730.44870712329, "hops": {"1": 22491.57329101581, "2": 16662.436791113578, "3": 18511.050833039917, "4": 16065.387791953981}, "receiver_prefill_ms": 330.183457932435, "ttft_ms": 25.99808294326067, "decode_ms": 41.162459063343704}, "total_wall_ms": 74128.41483205557, "ttft_ms": 25.99808294326067, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 64, "facts_retained": 30, "retention_rate": 0.46875, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.6666666666666666, "status": 0.9166666666666666, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1250}, "ledger_by_hop": {"1": {"facts_total": 64, "facts_retained": 30, "retention_rate": 0.46875, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.6666666666666666, "status": 0.9166666666666666, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1087}, "2": {"facts_total": 64, "facts_retained": 30, "retention_rate": 0.46875, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.6666666666666666, "status": 0.9166666666666666, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1071}, "3": {"facts_total": 64, "facts_retained": 30, "retention_rate": 0.46875, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.6666666666666666, "status": 0.9166666666666666, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 962}, "4": {"facts_total": 64, "facts_retained": 30, "retention_rate": 0.46875, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.6666666666666666, "status": 0.9166666666666666, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1250}}}, "reference_key_id": "keys/synthetic/syn_03006_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "a9ee5c99d37fccc8db66787ad20a94b821516c6fb6542156bc7cac20ae86eb2f", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "5725584e08f10ff823a0127a", "attempt_id": "c2cf88f7b95d", "run_group_id": "E4_core:A4|closed|qwen3b-llama3b:h4", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:08:34.989479Z", "task_family": "synthetic", "document_id": "syn_03006_2000", "document_hash": "89be5e77e0ca114963e9ffdf8dc38fe5610f13a78ac29e7a1956cd9124ce5217", "question_id": "q11_aggregation", "split": "test", "cluster_id": "syn_03006_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "tokenizer_hashes": {"receiver": "b9d8872cf3a99da99e1fc4b0f71232a451632dc49ca4c6771549b4efa5c1c763"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 4, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1717, "query_tokens": 521, "message_tokens": 406, "message_utf8_bytes": 1250, "cache_tensor_bytes": null, "serialized_payload_bytes": 1250, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 73730.44870712329, "hops": {"1": 22491.57329101581, "2": 16662.436791113578, "3": 18511.050833039917, "4": 16065.387791953981}, "receiver_prefill_ms": 333.5657500429079, "ttft_ms": 25.282124988734722, "decode_ms": 132.64100009109825}, "total_wall_ms": 74222.57783205714, "ttft_ms": 25.282124988734722, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 5}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "9,973,385", "evidence_artifact": {"ledger": {"facts_total": 64, "facts_retained": 30, "retention_rate": 0.46875, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.6666666666666666, "status": 0.9166666666666666, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1250}, "ledger_by_hop": {"1": {"facts_total": 64, "facts_retained": 30, "retention_rate": 0.46875, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.6666666666666666, "status": 0.9166666666666666, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1087}, "2": {"facts_total": 64, "facts_retained": 30, "retention_rate": 0.46875, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.6666666666666666, "status": 0.9166666666666666, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1071}, "3": {"facts_total": 64, "facts_retained": 30, "retention_rate": 0.46875, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.6666666666666666, "status": 0.9166666666666666, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 962}, "4": {"facts_total": 64, "facts_retained": 30, "retention_rate": 0.46875, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.6666666666666666, "status": 0.9166666666666666, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1250}}}, "reference_key_id": "keys/synthetic/syn_03006_2000.json", "task_scores": {"correct": 0.0, "abstained": 0.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "a9ee5c99d37fccc8db66787ad20a94b821516c6fb6542156bc7cac20ae86eb2f", "output_hash": "297d7ffd9b46da3939e37bf9f4547d2df8ddcab1bac3b7dfe46741052b43552a"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "32efb21e3aee9530114d76d7", "attempt_id": "c1180efbb7b7", "run_group_id": "E4_core:A4|closed|qwen3b-llama3b:h4", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:08:35.498866Z", "task_family": "synthetic", "document_id": "syn_03006_2000", "document_hash": "89be5e77e0ca114963e9ffdf8dc38fe5610f13a78ac29e7a1956cd9124ce5217", "question_id": "q12_unit_conversion", "split": "test", "cluster_id": "syn_03006_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "tokenizer_hashes": {"receiver": "b9d8872cf3a99da99e1fc4b0f71232a451632dc49ca4c6771549b4efa5c1c763"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 4, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1717, "query_tokens": 517, "message_tokens": 406, "message_utf8_bytes": 1250, "cache_tensor_bytes": null, "serialized_payload_bytes": 1250, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 73730.44870712329, "hops": {"1": 22491.57329101581, "2": 16662.436791113578, "3": 18511.050833039917, "4": 16065.387791953981}, "receiver_prefill_ms": 316.15633401088417, "ttft_ms": 67.5690839998424, "decode_ms": 309.03520807623863}, "total_wall_ms": 74423.86112408713, "ttft_ms": 67.5690839998424, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 3}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "39.7", "evidence_artifact": {"ledger": {"facts_total": 64, "facts_retained": 30, "retention_rate": 0.46875, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.6666666666666666, "status": 0.9166666666666666, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1250}, "ledger_by_hop": {"1": {"facts_total": 64, "facts_retained": 30, "retention_rate": 0.46875, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.6666666666666666, "status": 0.9166666666666666, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1087}, "2": {"facts_total": 64, "facts_retained": 30, "retention_rate": 0.46875, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.6666666666666666, "status": 0.9166666666666666, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1071}, "3": {"facts_total": 64, "facts_retained": 30, "retention_rate": 0.46875, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.6666666666666666, "status": 0.9166666666666666, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 962}, "4": {"facts_total": 64, "facts_retained": 30, "retention_rate": 0.46875, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.6666666666666666, "status": 0.9166666666666666, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1250}}}, "reference_key_id": "keys/synthetic/syn_03006_2000.json", "task_scores": {"correct": 0.0, "abstained": 0.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "a9ee5c99d37fccc8db66787ad20a94b821516c6fb6542156bc7cac20ae86eb2f", "output_hash": "138c4b663aa3391624efc0a351c82fa6339a5f7ce2617bbf767a8c33c9c6c3e7"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "0ffbe12fc0da3b8e77ec3c49", "attempt_id": "3cef844e17d0", "run_group_id": "E4_core:A4|closed|qwen3b-llama3b:h4", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:08:36.210928Z", "task_family": "synthetic", "document_id": "syn_03006_2000", "document_hash": "89be5e77e0ca114963e9ffdf8dc38fe5610f13a78ac29e7a1956cd9124ce5217", "question_id": "q13_unanswerable", "split": "test", "cluster_id": "syn_03006_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "tokenizer_hashes": {"receiver": "b9d8872cf3a99da99e1fc4b0f71232a451632dc49ca4c6771549b4efa5c1c763"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 4, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1717, "query_tokens": 521, "message_tokens": 406, "message_utf8_bytes": 1250, "cache_tensor_bytes": null, "serialized_payload_bytes": 1250, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 73730.44870712329, "hops": {"1": 22491.57329101581, "2": 16662.436791113578, "3": 18511.050833039917, "4": 16065.387791953981}, "receiver_prefill_ms": 338.9326250180602, "ttft_ms": 34.736250061541796, "decode_ms": 215.65608400851488}, "total_wall_ms": 74320.35574910697, "ttft_ms": 34.736250061541796, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 7}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "XM-76Q-2958", "evidence_artifact": {"ledger": {"facts_total": 64, "facts_retained": 30, "retention_rate": 0.46875, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.6666666666666666, "status": 0.9166666666666666, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1250}, "ledger_by_hop": {"1": {"facts_total": 64, "facts_retained": 30, "retention_rate": 0.46875, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.6666666666666666, "status": 0.9166666666666666, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1087}, "2": {"facts_total": 64, "facts_retained": 30, "retention_rate": 0.46875, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.6666666666666666, "status": 0.9166666666666666, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1071}, "3": {"facts_total": 64, "facts_retained": 30, "retention_rate": 0.46875, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.6666666666666666, "status": 0.9166666666666666, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 962}, "4": {"facts_total": 64, "facts_retained": 30, "retention_rate": 0.46875, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.6666666666666666, "status": 0.9166666666666666, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1250}}}, "reference_key_id": "keys/synthetic/syn_03006_2000.json", "task_scores": {"correct": 0.0, "abstained": 0.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "a9ee5c99d37fccc8db66787ad20a94b821516c6fb6542156bc7cac20ae86eb2f", "output_hash": "cf42691eec772b9ec74d67643944de2e19e53724d7b772cef1e0f2d5ef3b4926"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "6983d3a8629fb132c3ee9481", "attempt_id": "5a4f06737ea8", "run_group_id": "E4_core:A4|closed|qwen3b-llama3b:h4", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:08:36.818334Z", "task_family": "synthetic", "document_id": "syn_03006_2000", "document_hash": "89be5e77e0ca114963e9ffdf8dc38fe5610f13a78ac29e7a1956cd9124ce5217", "question_id": "q14_unanswerable", "split": "test", "cluster_id": "syn_03006_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "tokenizer_hashes": {"receiver": "b9d8872cf3a99da99e1fc4b0f71232a451632dc49ca4c6771549b4efa5c1c763"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 4, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1717, "query_tokens": 523, "message_tokens": 406, "message_utf8_bytes": 1250, "cache_tensor_bytes": null, "serialized_payload_bytes": 1250, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 73730.44870712329, "hops": {"1": 22491.57329101581, "2": 16662.436791113578, "3": 18511.050833039917, "4": 16065.387791953981}, "receiver_prefill_ms": 332.86808303091675, "ttft_ms": 19.268249976448715, "decode_ms": 18.635791959241033}, "total_wall_ms": 74101.76770703401, "ttft_ms": 19.268249976448715, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 64, "facts_retained": 30, "retention_rate": 0.46875, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.6666666666666666, "status": 0.9166666666666666, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1250}, "ledger_by_hop": {"1": {"facts_total": 64, "facts_retained": 30, "retention_rate": 0.46875, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.6666666666666666, "status": 0.9166666666666666, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1087}, "2": {"facts_total": 64, "facts_retained": 30, "retention_rate": 0.46875, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.6666666666666666, "status": 0.9166666666666666, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1071}, "3": {"facts_total": 64, "facts_retained": 30, "retention_rate": 0.46875, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.6666666666666666, "status": 0.9166666666666666, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 962}, "4": {"facts_total": 64, "facts_retained": 30, "retention_rate": 0.46875, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.6666666666666666, "status": 0.9166666666666666, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1250}}}, "reference_key_id": "keys/synthetic/syn_03006_2000.json", "task_scores": {"correct": 1.0, "abstained": 1.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "a9ee5c99d37fccc8db66787ad20a94b821516c6fb6542156bc7cac20ae86eb2f", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "9baf89ca22829cd95db9ebfd", "attempt_id": "57edf1c16eb6", "run_group_id": "E4_core:A4|closed|qwen3b-llama3b:h8", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:09:47.993479Z", "task_family": "synthetic", "document_id": "syn_03006_2000", "document_hash": "89be5e77e0ca114963e9ffdf8dc38fe5610f13a78ac29e7a1956cd9124ce5217", "question_id": "q00_single_fact", "split": "test", "cluster_id": "syn_03006_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "tokenizer_hashes": {"receiver": "b9d8872cf3a99da99e1fc4b0f71232a451632dc49ca4c6771549b4efa5c1c763"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 8, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1717, "query_tokens": 487, "message_tokens": 384, "message_utf8_bytes": 1153, "cache_tensor_bytes": null, "serialized_payload_bytes": 1153, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 144511.98237319477, "hops": {"1": 22491.57329101581, "2": 16662.436791113578, "3": 18511.050833039917, "4": 16065.387791953981, "5": 20273.956040968187, "6": 15294.222000055015, "7": 20052.09716700483, "8": 15161.258458043449}, "receiver_prefill_ms": 371.9410829944536, "ttft_ms": 25.254999985918403, "decode_ms": 201.00312505383044}, "total_wall_ms": 145110.7901232317, "ttft_ms": 25.254999985918403, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 7}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "XM-76Q-2958", "evidence_artifact": {"ledger": {"facts_total": 64, "facts_retained": 28, "retention_rate": 0.4375, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.6666666666666666, "status": 0.9166666666666666, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1153}, "ledger_by_hop": {"1": {"facts_total": 64, "facts_retained": 30, "retention_rate": 0.46875, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.6666666666666666, "status": 0.9166666666666666, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1087}, "2": {"facts_total": 64, "facts_retained": 30, "retention_rate": 0.46875, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.6666666666666666, "status": 0.9166666666666666, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1071}, "3": {"facts_total": 64, "facts_retained": 30, "retention_rate": 0.46875, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.6666666666666666, "status": 0.9166666666666666, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 962}, "4": {"facts_total": 64, "facts_retained": 30, "retention_rate": 0.46875, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.6666666666666666, "status": 0.9166666666666666, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1250}, "5": {"facts_total": 64, "facts_retained": 30, "retention_rate": 0.46875, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.6666666666666666, "status": 0.9166666666666666, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1136}, "6": {"facts_total": 64, "facts_retained": 28, "retention_rate": 0.4375, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.6666666666666666, "status": 0.9166666666666666, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1182}, "7": {"facts_total": 64, "facts_retained": 28, "retention_rate": 0.4375, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.6666666666666666, "status": 0.9166666666666666, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1127}, "8": {"facts_total": 64, "facts_retained": 28, "retention_rate": 0.4375, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.6666666666666666, "status": 0.9166666666666666, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1153}}}, "reference_key_id": "keys/synthetic/syn_03006_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "a30b88cb731dc60bcffb9d27bc8ac6c2fede0f9c48d1a7115b443f92f9a96511", "output_hash": "cf42691eec772b9ec74d67643944de2e19e53724d7b772cef1e0f2d5ef3b4926"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "9636bfd2d92251c12376aaee", "attempt_id": "829d210927f2", "run_group_id": "E4_core:A4|closed|qwen3b-llama3b:h8", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:09:48.610133Z", "task_family": "synthetic", "document_id": "syn_03006_2000", "document_hash": "89be5e77e0ca114963e9ffdf8dc38fe5610f13a78ac29e7a1956cd9124ce5217", "question_id": "q01_single_fact", "split": "test", "cluster_id": "syn_03006_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "tokenizer_hashes": {"receiver": "b9d8872cf3a99da99e1fc4b0f71232a451632dc49ca4c6771549b4efa5c1c763"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 8, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1717, "query_tokens": 493, "message_tokens": 384, "message_utf8_bytes": 1153, "cache_tensor_bytes": null, "serialized_payload_bytes": 1153, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 144511.98237319477, "hops": {"1": 22491.57329101581, "2": 16662.436791113578, "3": 18511.050833039917, "4": 16065.387791953981, "5": 20273.956040968187, "6": 15294.222000055015, "7": 20052.09716700483, "8": 15161.258458043449}, "receiver_prefill_ms": 327.5463340105489, "ttft_ms": 41.53404198586941, "decode_ms": 34.105165977962315}, "total_wall_ms": 144915.8098732587, "ttft_ms": 41.53404198586941, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 64, "facts_retained": 28, "retention_rate": 0.4375, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.6666666666666666, "status": 0.9166666666666666, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1153}, "ledger_by_hop": {"1": {"facts_total": 64, "facts_retained": 30, "retention_rate": 0.46875, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.6666666666666666, "status": 0.9166666666666666, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1087}, "2": {"facts_total": 64, "facts_retained": 30, "retention_rate": 0.46875, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.6666666666666666, "status": 0.9166666666666666, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1071}, "3": {"facts_total": 64, "facts_retained": 30, "retention_rate": 0.46875, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.6666666666666666, "status": 0.9166666666666666, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 962}, "4": {"facts_total": 64, "facts_retained": 30, "retention_rate": 0.46875, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.6666666666666666, "status": 0.9166666666666666, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1250}, "5": {"facts_total": 64, "facts_retained": 30, "retention_rate": 0.46875, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.6666666666666666, "status": 0.9166666666666666, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1136}, "6": {"facts_total": 64, "facts_retained": 28, "retention_rate": 0.4375, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.6666666666666666, "status": 0.9166666666666666, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1182}, "7": {"facts_total": 64, "facts_retained": 28, "retention_rate": 0.4375, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.6666666666666666, "status": 0.9166666666666666, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1127}, "8": {"facts_total": 64, "facts_retained": 28, "retention_rate": 0.4375, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.6666666666666666, "status": 0.9166666666666666, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1153}}}, "reference_key_id": "keys/synthetic/syn_03006_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "a30b88cb731dc60bcffb9d27bc8ac6c2fede0f9c48d1a7115b443f92f9a96511", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "46000074a702be3a5f86b73e", "attempt_id": "c07cc9ef5896", "run_group_id": "E4_core:A4|closed|qwen3b-llama3b:h8", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:09:49.031562Z", "task_family": "synthetic", "document_id": "syn_03006_2000", "document_hash": "89be5e77e0ca114963e9ffdf8dc38fe5610f13a78ac29e7a1956cd9124ce5217", "question_id": "q02_exact_copy", "split": "test", "cluster_id": "syn_03006_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "tokenizer_hashes": {"receiver": "b9d8872cf3a99da99e1fc4b0f71232a451632dc49ca4c6771549b4efa5c1c763"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 8, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1717, "query_tokens": 492, "message_tokens": 384, "message_utf8_bytes": 1153, "cache_tensor_bytes": null, "serialized_payload_bytes": 1153, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 144511.98237319477, "hops": {"1": 22491.57329101581, "2": 16662.436791113578, "3": 18511.050833039917, "4": 16065.387791953981, "5": 20273.956040968187, "6": 15294.222000055015, "7": 20052.09716700483, "8": 15161.258458043449}, "receiver_prefill_ms": 467.75787498336285, "ttft_ms": 22.520540980622172, "decode_ms": 38.52212498895824}, "total_wall_ms": 145041.36212321464, "ttft_ms": 22.520540980622172, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 64, "facts_retained": 28, "retention_rate": 0.4375, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.6666666666666666, "status": 0.9166666666666666, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1153}, "ledger_by_hop": {"1": {"facts_total": 64, "facts_retained": 30, "retention_rate": 0.46875, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.6666666666666666, "status": 0.9166666666666666, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1087}, "2": {"facts_total": 64, "facts_retained": 30, "retention_rate": 0.46875, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.6666666666666666, "status": 0.9166666666666666, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1071}, "3": {"facts_total": 64, "facts_retained": 30, "retention_rate": 0.46875, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.6666666666666666, "status": 0.9166666666666666, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 962}, "4": {"facts_total": 64, "facts_retained": 30, "retention_rate": 0.46875, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.6666666666666666, "status": 0.9166666666666666, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1250}, "5": {"facts_total": 64, "facts_retained": 30, "retention_rate": 0.46875, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.6666666666666666, "status": 0.9166666666666666, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1136}, "6": {"facts_total": 64, "facts_retained": 28, "retention_rate": 0.4375, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.6666666666666666, "status": 0.9166666666666666, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1182}, "7": {"facts_total": 64, "facts_retained": 28, "retention_rate": 0.4375, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.6666666666666666, "status": 0.9166666666666666, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1127}, "8": {"facts_total": 64, "facts_retained": 28, "retention_rate": 0.4375, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.6666666666666666, "status": 0.9166666666666666, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1153}}}, "reference_key_id": "keys/synthetic/syn_03006_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "a30b88cb731dc60bcffb9d27bc8ac6c2fede0f9c48d1a7115b443f92f9a96511", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "3beb5e233c0c649c20255eac", "attempt_id": "c29b5751df2a", "run_group_id": "E4_core:A4|closed|qwen3b-llama3b:h8", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:09:49.578360Z", "task_family": "synthetic", "document_id": "syn_03006_2000", "document_hash": "89be5e77e0ca114963e9ffdf8dc38fe5610f13a78ac29e7a1956cd9124ce5217", "question_id": "q03_exact_copy", "split": "test", "cluster_id": "syn_03006_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "tokenizer_hashes": {"receiver": "b9d8872cf3a99da99e1fc4b0f71232a451632dc49ca4c6771549b4efa5c1c763"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 8, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1717, "query_tokens": 487, "message_tokens": 384, "message_utf8_bytes": 1153, "cache_tensor_bytes": null, "serialized_payload_bytes": 1153, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 144511.98237319477, "hops": {"1": 22491.57329101581, "2": 16662.436791113578, "3": 18511.050833039917, "4": 16065.387791953981, "5": 20273.956040968187, "6": 15294.222000055015, "7": 20052.09716700483, "8": 15161.258458043449}, "receiver_prefill_ms": 378.4876660211012, "ttft_ms": 21.41216699965298, "decode_ms": 246.62779190111905}, "total_wall_ms": 145159.09141523298, "ttft_ms": 21.41216699965298, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 8}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "Haruki Ouellet-Marchand", "evidence_artifact": {"ledger": {"facts_total": 64, "facts_retained": 28, "retention_rate": 0.4375, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.6666666666666666, "status": 0.9166666666666666, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1153}, "ledger_by_hop": {"1": {"facts_total": 64, "facts_retained": 30, "retention_rate": 0.46875, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.6666666666666666, "status": 0.9166666666666666, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1087}, "2": {"facts_total": 64, "facts_retained": 30, "retention_rate": 0.46875, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.6666666666666666, "status": 0.9166666666666666, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1071}, "3": {"facts_total": 64, "facts_retained": 30, "retention_rate": 0.46875, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.6666666666666666, "status": 0.9166666666666666, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 962}, "4": {"facts_total": 64, "facts_retained": 30, "retention_rate": 0.46875, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.6666666666666666, "status": 0.9166666666666666, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1250}, "5": {"facts_total": 64, "facts_retained": 30, "retention_rate": 0.46875, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.6666666666666666, "status": 0.9166666666666666, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1136}, "6": {"facts_total": 64, "facts_retained": 28, "retention_rate": 0.4375, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.6666666666666666, "status": 0.9166666666666666, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1182}, "7": {"facts_total": 64, "facts_retained": 28, "retention_rate": 0.4375, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.6666666666666666, "status": 0.9166666666666666, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1127}, "8": {"facts_total": 64, "facts_retained": 28, "retention_rate": 0.4375, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.6666666666666666, "status": 0.9166666666666666, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1153}}}, "reference_key_id": "keys/synthetic/syn_03006_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "a30b88cb731dc60bcffb9d27bc8ac6c2fede0f9c48d1a7115b443f92f9a96511", "output_hash": "a3b6e0a44d495387d8b8fc50ba6be0f2ed43121c674d3b3e962f8c4690e15b5c"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "06cd9dfc24e31a1b95d98c8d", "attempt_id": "7a8cbd1fa546", "run_group_id": "E4_core:A4|closed|qwen3b-llama3b:h8", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:09:50.243032Z", "task_family": "synthetic", "document_id": "syn_03006_2000", "document_hash": "89be5e77e0ca114963e9ffdf8dc38fe5610f13a78ac29e7a1956cd9124ce5217", "question_id": "q04_multi_hop", "split": "test", "cluster_id": "syn_03006_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "tokenizer_hashes": {"receiver": "b9d8872cf3a99da99e1fc4b0f71232a451632dc49ca4c6771549b4efa5c1c763"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 8, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1717, "query_tokens": 496, "message_tokens": 384, "message_utf8_bytes": 1153, "cache_tensor_bytes": null, "serialized_payload_bytes": 1153, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 144511.98237319477, "hops": {"1": 22491.57329101581, "2": 16662.436791113578, "3": 18511.050833039917, "4": 16065.387791953981, "5": 20273.956040968187, "6": 15294.222000055015, "7": 20052.09716700483, "8": 15161.258458043449}, "receiver_prefill_ms": 326.0037919972092, "ttft_ms": 34.55233294516802, "decode_ms": 47.97308298293501}, "total_wall_ms": 144921.14808224142, "ttft_ms": 34.55233294516802, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 64, "facts_retained": 28, "retention_rate": 0.4375, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.6666666666666666, "status": 0.9166666666666666, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1153}, "ledger_by_hop": {"1": {"facts_total": 64, "facts_retained": 30, "retention_rate": 0.46875, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.6666666666666666, "status": 0.9166666666666666, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1087}, "2": {"facts_total": 64, "facts_retained": 30, "retention_rate": 0.46875, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.6666666666666666, "status": 0.9166666666666666, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1071}, "3": {"facts_total": 64, "facts_retained": 30, "retention_rate": 0.46875, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.6666666666666666, "status": 0.9166666666666666, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 962}, "4": {"facts_total": 64, "facts_retained": 30, "retention_rate": 0.46875, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.6666666666666666, "status": 0.9166666666666666, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1250}, "5": {"facts_total": 64, "facts_retained": 30, "retention_rate": 0.46875, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.6666666666666666, "status": 0.9166666666666666, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1136}, "6": {"facts_total": 64, "facts_retained": 28, "retention_rate": 0.4375, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.6666666666666666, "status": 0.9166666666666666, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1182}, "7": {"facts_total": 64, "facts_retained": 28, "retention_rate": 0.4375, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.6666666666666666, "status": 0.9166666666666666, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1127}, "8": {"facts_total": 64, "facts_retained": 28, "retention_rate": 0.4375, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.6666666666666666, "status": 0.9166666666666666, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1153}}}, "reference_key_id": "keys/synthetic/syn_03006_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "a30b88cb731dc60bcffb9d27bc8ac6c2fede0f9c48d1a7115b443f92f9a96511", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "eac1a10f823f69c0d9ba6a41", "attempt_id": "e09786e84007", "run_group_id": "E4_core:A4|closed|qwen3b-llama3b:h8", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:09:50.670256Z", "task_family": "synthetic", "document_id": "syn_03006_2000", "document_hash": "89be5e77e0ca114963e9ffdf8dc38fe5610f13a78ac29e7a1956cd9124ce5217", "question_id": "q05_multi_hop", "split": "test", "cluster_id": "syn_03006_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "tokenizer_hashes": {"receiver": "b9d8872cf3a99da99e1fc4b0f71232a451632dc49ca4c6771549b4efa5c1c763"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 8, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1717, "query_tokens": 497, "message_tokens": 384, "message_utf8_bytes": 1153, "cache_tensor_bytes": null, "serialized_payload_bytes": 1153, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 144511.98237319477, "hops": {"1": 22491.57329101581, "2": 16662.436791113578, "3": 18511.050833039917, "4": 16065.387791953981, "5": 20273.956040968187, "6": 15294.222000055015, "7": 20052.09716700483, "8": 15161.258458043449}, "receiver_prefill_ms": 493.22483397554606, "ttft_ms": 35.9951660502702, "decode_ms": 213.38016702793539}, "total_wall_ms": 145255.12270617764, "ttft_ms": 35.9951660502702, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 7}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "EP-52G-1330", "evidence_artifact": {"ledger": {"facts_total": 64, "facts_retained": 28, "retention_rate": 0.4375, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.6666666666666666, "status": 0.9166666666666666, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1153}, "ledger_by_hop": {"1": {"facts_total": 64, "facts_retained": 30, "retention_rate": 0.46875, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.6666666666666666, "status": 0.9166666666666666, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1087}, "2": {"facts_total": 64, "facts_retained": 30, "retention_rate": 0.46875, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.6666666666666666, "status": 0.9166666666666666, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1071}, "3": {"facts_total": 64, "facts_retained": 30, "retention_rate": 0.46875, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.6666666666666666, "status": 0.9166666666666666, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 962}, "4": {"facts_total": 64, "facts_retained": 30, "retention_rate": 0.46875, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.6666666666666666, "status": 0.9166666666666666, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1250}, "5": {"facts_total": 64, "facts_retained": 30, "retention_rate": 0.46875, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.6666666666666666, "status": 0.9166666666666666, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1136}, "6": {"facts_total": 64, "facts_retained": 28, "retention_rate": 0.4375, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.6666666666666666, "status": 0.9166666666666666, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1182}, "7": {"facts_total": 64, "facts_retained": 28, "retention_rate": 0.4375, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.6666666666666666, "status": 0.9166666666666666, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1127}, "8": {"facts_total": 64, "facts_retained": 28, "retention_rate": 0.4375, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.6666666666666666, "status": 0.9166666666666666, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1153}}}, "reference_key_id": "keys/synthetic/syn_03006_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "a30b88cb731dc60bcffb9d27bc8ac6c2fede0f9c48d1a7115b443f92f9a96511", "output_hash": "104db207937fd79e23ce16f9b3f10b1a9ed34f6dd0441a99be6ea1b2d0a09d99"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "3f63a9b1df53a1ea1508aa99", "attempt_id": "2fe7caaa10b8", "run_group_id": "E4_core:A4|closed|qwen3b-llama3b:h8", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:09:51.432097Z", "task_family": "synthetic", "document_id": "syn_03006_2000", "document_hash": "89be5e77e0ca114963e9ffdf8dc38fe5610f13a78ac29e7a1956cd9124ce5217", "question_id": "q06_negation", "split": "test", "cluster_id": "syn_03006_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "tokenizer_hashes": {"receiver": "b9d8872cf3a99da99e1fc4b0f71232a451632dc49ca4c6771549b4efa5c1c763"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 8, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1717, "query_tokens": 487, "message_tokens": 384, "message_utf8_bytes": 1153, "cache_tensor_bytes": null, "serialized_payload_bytes": 1153, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 144511.98237319477, "hops": {"1": 22491.57329101581, "2": 16662.436791113578, "3": 18511.050833039917, "4": 16065.387791953981, "5": 20273.956040968187, "6": 15294.222000055015, "7": 20052.09716700483, "8": 15161.258458043449}, "receiver_prefill_ms": 369.6724169421941, "ttft_ms": 26.98075002990663, "decode_ms": 54.19825005810708}, "total_wall_ms": 144963.39574817102, "ttft_ms": 26.98075002990663, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 2}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "No.", "evidence_artifact": {"ledger": {"facts_total": 64, "facts_retained": 28, "retention_rate": 0.4375, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.6666666666666666, "status": 0.9166666666666666, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1153}, "ledger_by_hop": {"1": {"facts_total": 64, "facts_retained": 30, "retention_rate": 0.46875, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.6666666666666666, "status": 0.9166666666666666, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1087}, "2": {"facts_total": 64, "facts_retained": 30, "retention_rate": 0.46875, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.6666666666666666, "status": 0.9166666666666666, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1071}, "3": {"facts_total": 64, "facts_retained": 30, "retention_rate": 0.46875, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.6666666666666666, "status": 0.9166666666666666, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 962}, "4": {"facts_total": 64, "facts_retained": 30, "retention_rate": 0.46875, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.6666666666666666, "status": 0.9166666666666666, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1250}, "5": {"facts_total": 64, "facts_retained": 30, "retention_rate": 0.46875, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.6666666666666666, "status": 0.9166666666666666, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1136}, "6": {"facts_total": 64, "facts_retained": 28, "retention_rate": 0.4375, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.6666666666666666, "status": 0.9166666666666666, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1182}, "7": {"facts_total": 64, "facts_retained": 28, "retention_rate": 0.4375, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.6666666666666666, "status": 0.9166666666666666, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1127}, "8": {"facts_total": 64, "facts_retained": 28, "retention_rate": 0.4375, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.6666666666666666, "status": 0.9166666666666666, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1153}}}, "reference_key_id": "keys/synthetic/syn_03006_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "a30b88cb731dc60bcffb9d27bc8ac6c2fede0f9c48d1a7115b443f92f9a96511", "output_hash": "38178a20b470cfd18299fb1593dd4ba706f1a83321dac051d01086efd8b7a96f"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "4499f7229272c8d56f4fb773", "attempt_id": "cef0519e11dc", "run_group_id": "E4_core:A4|closed|qwen3b-llama3b:h8", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:09:51.900948Z", "task_family": "synthetic", "document_id": "syn_03006_2000", "document_hash": "89be5e77e0ca114963e9ffdf8dc38fe5610f13a78ac29e7a1956cd9124ce5217", "question_id": "q07_negation", "split": "test", "cluster_id": "syn_03006_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "tokenizer_hashes": {"receiver": "b9d8872cf3a99da99e1fc4b0f71232a451632dc49ca4c6771549b4efa5c1c763"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 8, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1717, "query_tokens": 487, "message_tokens": 384, "message_utf8_bytes": 1153, "cache_tensor_bytes": null, "serialized_payload_bytes": 1153, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 144511.98237319477, "hops": {"1": 22491.57329101581, "2": 16662.436791113578, "3": 18511.050833039917, "4": 16065.387791953981, "5": 20273.956040968187, "6": 15294.222000055015, "7": 20052.09716700483, "8": 15161.258458043449}, "receiver_prefill_ms": 356.6467909840867, "ttft_ms": 45.12158303987235, "decode_ms": 111.94591701496392}, "total_wall_ms": 145026.32695622742, "ttft_ms": 45.12158303987235, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 2}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "No.", "evidence_artifact": {"ledger": {"facts_total": 64, "facts_retained": 28, "retention_rate": 0.4375, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.6666666666666666, "status": 0.9166666666666666, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1153}, "ledger_by_hop": {"1": {"facts_total": 64, "facts_retained": 30, "retention_rate": 0.46875, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.6666666666666666, "status": 0.9166666666666666, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1087}, "2": {"facts_total": 64, "facts_retained": 30, "retention_rate": 0.46875, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.6666666666666666, "status": 0.9166666666666666, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1071}, "3": {"facts_total": 64, "facts_retained": 30, "retention_rate": 0.46875, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.6666666666666666, "status": 0.9166666666666666, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 962}, "4": {"facts_total": 64, "facts_retained": 30, "retention_rate": 0.46875, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.6666666666666666, "status": 0.9166666666666666, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1250}, "5": {"facts_total": 64, "facts_retained": 30, "retention_rate": 0.46875, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.6666666666666666, "status": 0.9166666666666666, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1136}, "6": {"facts_total": 64, "facts_retained": 28, "retention_rate": 0.4375, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.6666666666666666, "status": 0.9166666666666666, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1182}, "7": {"facts_total": 64, "facts_retained": 28, "retention_rate": 0.4375, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.6666666666666666, "status": 0.9166666666666666, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1127}, "8": {"facts_total": 64, "facts_retained": 28, "retention_rate": 0.4375, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.6666666666666666, "status": 0.9166666666666666, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1153}}}, "reference_key_id": "keys/synthetic/syn_03006_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "a30b88cb731dc60bcffb9d27bc8ac6c2fede0f9c48d1a7115b443f92f9a96511", "output_hash": "38178a20b470cfd18299fb1593dd4ba706f1a83321dac051d01086efd8b7a96f"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "2388d656de7379c6113c0af3", "attempt_id": "78e02a4578d4", "run_group_id": "E4_core:A4|closed|qwen3b-llama3b:h8", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:09:52.432111Z", "task_family": "synthetic", "document_id": "syn_03006_2000", "document_hash": "89be5e77e0ca114963e9ffdf8dc38fe5610f13a78ac29e7a1956cd9124ce5217", "question_id": "q08_revision", "split": "test", "cluster_id": "syn_03006_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "tokenizer_hashes": {"receiver": "b9d8872cf3a99da99e1fc4b0f71232a451632dc49ca4c6771549b4efa5c1c763"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 8, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1717, "query_tokens": 503, "message_tokens": 384, "message_utf8_bytes": 1153, "cache_tensor_bytes": null, "serialized_payload_bytes": 1153, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 144511.98237319477, "hops": {"1": 22491.57329101581, "2": 16662.436791113578, "3": 18511.050833039917, "4": 16065.387791953981, "5": 20273.956040968187, "6": 15294.222000055015, "7": 20052.09716700483, "8": 15161.258458043449}, "receiver_prefill_ms": 480.0554160028696, "ttft_ms": 35.52158398088068, "decode_ms": 157.92316698934883}, "total_wall_ms": 145186.01599812973, "ttft_ms": 35.52158398088068, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 5}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "10,107,545", "evidence_artifact": {"ledger": {"facts_total": 64, "facts_retained": 28, "retention_rate": 0.4375, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.6666666666666666, "status": 0.9166666666666666, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1153}, "ledger_by_hop": {"1": {"facts_total": 64, "facts_retained": 30, "retention_rate": 0.46875, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.6666666666666666, "status": 0.9166666666666666, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1087}, "2": {"facts_total": 64, "facts_retained": 30, "retention_rate": 0.46875, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.6666666666666666, "status": 0.9166666666666666, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1071}, "3": {"facts_total": 64, "facts_retained": 30, "retention_rate": 0.46875, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.6666666666666666, "status": 0.9166666666666666, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 962}, "4": {"facts_total": 64, "facts_retained": 30, "retention_rate": 0.46875, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.6666666666666666, "status": 0.9166666666666666, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1250}, "5": {"facts_total": 64, "facts_retained": 30, "retention_rate": 0.46875, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.6666666666666666, "status": 0.9166666666666666, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1136}, "6": {"facts_total": 64, "facts_retained": 28, "retention_rate": 0.4375, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.6666666666666666, "status": 0.9166666666666666, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1182}, "7": {"facts_total": 64, "facts_retained": 28, "retention_rate": 0.4375, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.6666666666666666, "status": 0.9166666666666666, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1127}, "8": {"facts_total": 64, "facts_retained": 28, "retention_rate": 0.4375, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.6666666666666666, "status": 0.9166666666666666, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1153}}}, "reference_key_id": "keys/synthetic/syn_03006_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "a30b88cb731dc60bcffb9d27bc8ac6c2fede0f9c48d1a7115b443f92f9a96511", "output_hash": "3d6a277763361ef55b8dd3f9791012605d24cd167ac5bdcdd0064ee16de161d4"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "468572c1ed072eab5e81a8ec", "attempt_id": "fca454b76296", "run_group_id": "E4_core:A4|closed|qwen3b-llama3b:h8", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:09:53.124914Z", "task_family": "synthetic", "document_id": "syn_03006_2000", "document_hash": "89be5e77e0ca114963e9ffdf8dc38fe5610f13a78ac29e7a1956cd9124ce5217", "question_id": "q09_revision", "split": "test", "cluster_id": "syn_03006_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "tokenizer_hashes": {"receiver": "b9d8872cf3a99da99e1fc4b0f71232a451632dc49ca4c6771549b4efa5c1c763"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 8, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1717, "query_tokens": 496, "message_tokens": 384, "message_utf8_bytes": 1153, "cache_tensor_bytes": null, "serialized_payload_bytes": 1153, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 144511.98237319477, "hops": {"1": 22491.57329101581, "2": 16662.436791113578, "3": 18511.050833039917, "4": 16065.387791953981, "5": 20273.956040968187, "6": 15294.222000055015, "7": 20052.09716700483, "8": 15161.258458043449}, "receiver_prefill_ms": 373.08195896912366, "ttft_ms": 32.772290986031294, "decode_ms": 151.46062499843538}, "total_wall_ms": 145070.0700811576, "ttft_ms": 32.772290986031294, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 5}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "9,112,735", "evidence_artifact": {"ledger": {"facts_total": 64, "facts_retained": 28, "retention_rate": 0.4375, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.6666666666666666, "status": 0.9166666666666666, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1153}, "ledger_by_hop": {"1": {"facts_total": 64, "facts_retained": 30, "retention_rate": 0.46875, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.6666666666666666, "status": 0.9166666666666666, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1087}, "2": {"facts_total": 64, "facts_retained": 30, "retention_rate": 0.46875, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.6666666666666666, "status": 0.9166666666666666, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1071}, "3": {"facts_total": 64, "facts_retained": 30, "retention_rate": 0.46875, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.6666666666666666, "status": 0.9166666666666666, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 962}, "4": {"facts_total": 64, "facts_retained": 30, "retention_rate": 0.46875, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.6666666666666666, "status": 0.9166666666666666, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1250}, "5": {"facts_total": 64, "facts_retained": 30, "retention_rate": 0.46875, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.6666666666666666, "status": 0.9166666666666666, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1136}, "6": {"facts_total": 64, "facts_retained": 28, "retention_rate": 0.4375, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.6666666666666666, "status": 0.9166666666666666, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1182}, "7": {"facts_total": 64, "facts_retained": 28, "retention_rate": 0.4375, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.6666666666666666, "status": 0.9166666666666666, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1127}, "8": {"facts_total": 64, "facts_retained": 28, "retention_rate": 0.4375, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.6666666666666666, "status": 0.9166666666666666, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1153}}}, "reference_key_id": "keys/synthetic/syn_03006_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "a30b88cb731dc60bcffb9d27bc8ac6c2fede0f9c48d1a7115b443f92f9a96511", "output_hash": "788fcda908c2029597420995b09d0cc919bfdbf7a28b68dcb0e331db7df7b342"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "9ceff0a454ea314bb68611a6", "attempt_id": "0fb4dc9b8d50", "run_group_id": "E4_core:A4|closed|qwen3b-llama3b:h8", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:09:53.700886Z", "task_family": "synthetic", "document_id": "syn_03006_2000", "document_hash": "89be5e77e0ca114963e9ffdf8dc38fe5610f13a78ac29e7a1956cd9124ce5217", "question_id": "q10_temporal", "split": "test", "cluster_id": "syn_03006_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "tokenizer_hashes": {"receiver": "b9d8872cf3a99da99e1fc4b0f71232a451632dc49ca4c6771549b4efa5c1c763"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 8, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1717, "query_tokens": 505, "message_tokens": 384, "message_utf8_bytes": 1153, "cache_tensor_bytes": null, "serialized_payload_bytes": 1153, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 144511.98237319477, "hops": {"1": 22491.57329101581, "2": 16662.436791113578, "3": 18511.050833039917, "4": 16065.387791953981, "5": 20273.956040968187, "6": 15294.222000055015, "7": 20052.09716700483, "8": 15161.258458043449}, "receiver_prefill_ms": 367.2684580087662, "ttft_ms": 90.09491698816419, "decode_ms": 136.05479209218174}, "total_wall_ms": 145105.97999824677, "ttft_ms": 90.09491698816419, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 64, "facts_retained": 28, "retention_rate": 0.4375, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.6666666666666666, "status": 0.9166666666666666, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1153}, "ledger_by_hop": {"1": {"facts_total": 64, "facts_retained": 30, "retention_rate": 0.46875, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.6666666666666666, "status": 0.9166666666666666, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1087}, "2": {"facts_total": 64, "facts_retained": 30, "retention_rate": 0.46875, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.6666666666666666, "status": 0.9166666666666666, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1071}, "3": {"facts_total": 64, "facts_retained": 30, "retention_rate": 0.46875, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.6666666666666666, "status": 0.9166666666666666, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 962}, "4": {"facts_total": 64, "facts_retained": 30, "retention_rate": 0.46875, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.6666666666666666, "status": 0.9166666666666666, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1250}, "5": {"facts_total": 64, "facts_retained": 30, "retention_rate": 0.46875, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.6666666666666666, "status": 0.9166666666666666, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1136}, "6": {"facts_total": 64, "facts_retained": 28, "retention_rate": 0.4375, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.6666666666666666, "status": 0.9166666666666666, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1182}, "7": {"facts_total": 64, "facts_retained": 28, "retention_rate": 0.4375, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.6666666666666666, "status": 0.9166666666666666, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1127}, "8": {"facts_total": 64, "facts_retained": 28, "retention_rate": 0.4375, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.6666666666666666, "status": 0.9166666666666666, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1153}}}, "reference_key_id": "keys/synthetic/syn_03006_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "a30b88cb731dc60bcffb9d27bc8ac6c2fede0f9c48d1a7115b443f92f9a96511", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "4a5fc845deb010edf4166242", "attempt_id": "ad0eec3e5cbc", "run_group_id": "E4_core:A4|closed|qwen3b-llama3b:h8", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:09:54.312332Z", "task_family": "synthetic", "document_id": "syn_03006_2000", "document_hash": "89be5e77e0ca114963e9ffdf8dc38fe5610f13a78ac29e7a1956cd9124ce5217", "question_id": "q11_aggregation", "split": "test", "cluster_id": "syn_03006_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "tokenizer_hashes": {"receiver": "b9d8872cf3a99da99e1fc4b0f71232a451632dc49ca4c6771549b4efa5c1c763"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 8, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1717, "query_tokens": 499, "message_tokens": 384, "message_utf8_bytes": 1153, "cache_tensor_bytes": null, "serialized_payload_bytes": 1153, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 144511.98237319477, "hops": {"1": 22491.57329101581, "2": 16662.436791113578, "3": 18511.050833039917, "4": 16065.387791953981, "5": 20273.956040968187, "6": 15294.222000055015, "7": 20052.09716700483, "8": 15161.258458043449}, "receiver_prefill_ms": 359.9707919638604, "ttft_ms": 23.481874959543347, "decode_ms": 175.04558293148875}, "total_wall_ms": 145071.0695402231, "ttft_ms": 23.481874959543347, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 5}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "9,973,385", "evidence_artifact": {"ledger": {"facts_total": 64, "facts_retained": 28, "retention_rate": 0.4375, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.6666666666666666, "status": 0.9166666666666666, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1153}, "ledger_by_hop": {"1": {"facts_total": 64, "facts_retained": 30, "retention_rate": 0.46875, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.6666666666666666, "status": 0.9166666666666666, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1087}, "2": {"facts_total": 64, "facts_retained": 30, "retention_rate": 0.46875, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.6666666666666666, "status": 0.9166666666666666, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1071}, "3": {"facts_total": 64, "facts_retained": 30, "retention_rate": 0.46875, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.6666666666666666, "status": 0.9166666666666666, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 962}, "4": {"facts_total": 64, "facts_retained": 30, "retention_rate": 0.46875, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.6666666666666666, "status": 0.9166666666666666, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1250}, "5": {"facts_total": 64, "facts_retained": 30, "retention_rate": 0.46875, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.6666666666666666, "status": 0.9166666666666666, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1136}, "6": {"facts_total": 64, "facts_retained": 28, "retention_rate": 0.4375, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.6666666666666666, "status": 0.9166666666666666, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1182}, "7": {"facts_total": 64, "facts_retained": 28, "retention_rate": 0.4375, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.6666666666666666, "status": 0.9166666666666666, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1127}, "8": {"facts_total": 64, "facts_retained": 28, "retention_rate": 0.4375, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.6666666666666666, "status": 0.9166666666666666, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1153}}}, "reference_key_id": "keys/synthetic/syn_03006_2000.json", "task_scores": {"correct": 0.0, "abstained": 0.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "a30b88cb731dc60bcffb9d27bc8ac6c2fede0f9c48d1a7115b443f92f9a96511", "output_hash": "297d7ffd9b46da3939e37bf9f4547d2df8ddcab1bac3b7dfe46741052b43552a"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "f09aa2d693e507646becbb94", "attempt_id": "31e5fef4c78f", "run_group_id": "E4_core:A4|closed|qwen3b-llama3b:h8", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:09:54.888066Z", "task_family": "synthetic", "document_id": "syn_03006_2000", "document_hash": "89be5e77e0ca114963e9ffdf8dc38fe5610f13a78ac29e7a1956cd9124ce5217", "question_id": "q12_unit_conversion", "split": "test", "cluster_id": "syn_03006_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "tokenizer_hashes": {"receiver": "b9d8872cf3a99da99e1fc4b0f71232a451632dc49ca4c6771549b4efa5c1c763"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 8, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1717, "query_tokens": 495, "message_tokens": 384, "message_utf8_bytes": 1153, "cache_tensor_bytes": null, "serialized_payload_bytes": 1153, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 144511.98237319477, "hops": {"1": 22491.57329101581, "2": 16662.436791113578, "3": 18511.050833039917, "4": 16065.387791953981, "5": 20273.956040968187, "6": 15294.222000055015, "7": 20052.09716700483, "8": 15161.258458043449}, "receiver_prefill_ms": 338.54316698852926, "ttft_ms": 19.390583038330078, "decode_ms": 160.723082954064}, "total_wall_ms": 145031.27158118878, "ttft_ms": 19.390583038330078, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 3}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "39,700", "evidence_artifact": {"ledger": {"facts_total": 64, "facts_retained": 28, "retention_rate": 0.4375, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.6666666666666666, "status": 0.9166666666666666, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1153}, "ledger_by_hop": {"1": {"facts_total": 64, "facts_retained": 30, "retention_rate": 0.46875, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.6666666666666666, "status": 0.9166666666666666, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1087}, "2": {"facts_total": 64, "facts_retained": 30, "retention_rate": 0.46875, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.6666666666666666, "status": 0.9166666666666666, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1071}, "3": {"facts_total": 64, "facts_retained": 30, "retention_rate": 0.46875, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.6666666666666666, "status": 0.9166666666666666, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 962}, "4": {"facts_total": 64, "facts_retained": 30, "retention_rate": 0.46875, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.6666666666666666, "status": 0.9166666666666666, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1250}, "5": {"facts_total": 64, "facts_retained": 30, "retention_rate": 0.46875, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.6666666666666666, "status": 0.9166666666666666, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1136}, "6": {"facts_total": 64, "facts_retained": 28, "retention_rate": 0.4375, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.6666666666666666, "status": 0.9166666666666666, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1182}, "7": {"facts_total": 64, "facts_retained": 28, "retention_rate": 0.4375, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.6666666666666666, "status": 0.9166666666666666, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1127}, "8": {"facts_total": 64, "facts_retained": 28, "retention_rate": 0.4375, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.6666666666666666, "status": 0.9166666666666666, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1153}}}, "reference_key_id": "keys/synthetic/syn_03006_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "a30b88cb731dc60bcffb9d27bc8ac6c2fede0f9c48d1a7115b443f92f9a96511", "output_hash": "5f62e7c00eed4af414e81f6c0d51672fbb9393169f5091e5137429e7a69e9a6d"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "36a05a7ec3a947047aa4a05b", "attempt_id": "ccbc13f5bc60", "run_group_id": "E4_core:A4|closed|qwen3b-llama3b:h8", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:09:55.424250Z", "task_family": "synthetic", "document_id": "syn_03006_2000", "document_hash": "89be5e77e0ca114963e9ffdf8dc38fe5610f13a78ac29e7a1956cd9124ce5217", "question_id": "q13_unanswerable", "split": "test", "cluster_id": "syn_03006_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "tokenizer_hashes": {"receiver": "b9d8872cf3a99da99e1fc4b0f71232a451632dc49ca4c6771549b4efa5c1c763"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 8, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1717, "query_tokens": 499, "message_tokens": 384, "message_utf8_bytes": 1153, "cache_tensor_bytes": null, "serialized_payload_bytes": 1153, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 144511.98237319477, "hops": {"1": 22491.57329101581, "2": 16662.436791113578, "3": 18511.050833039917, "4": 16065.387791953981, "5": 20273.956040968187, "6": 15294.222000055015, "7": 20052.09716700483, "8": 15161.258458043449}, "receiver_prefill_ms": 499.57337498199195, "ttft_ms": 40.82087497226894, "decode_ms": 25.06662509404123}, "total_wall_ms": 145078.05224822368, "ttft_ms": 40.82087497226894, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 64, "facts_retained": 28, "retention_rate": 0.4375, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.6666666666666666, "status": 0.9166666666666666, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1153}, "ledger_by_hop": {"1": {"facts_total": 64, "facts_retained": 30, "retention_rate": 0.46875, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.6666666666666666, "status": 0.9166666666666666, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1087}, "2": {"facts_total": 64, "facts_retained": 30, "retention_rate": 0.46875, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.6666666666666666, "status": 0.9166666666666666, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1071}, "3": {"facts_total": 64, "facts_retained": 30, "retention_rate": 0.46875, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.6666666666666666, "status": 0.9166666666666666, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 962}, "4": {"facts_total": 64, "facts_retained": 30, "retention_rate": 0.46875, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.6666666666666666, "status": 0.9166666666666666, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1250}, "5": {"facts_total": 64, "facts_retained": 30, "retention_rate": 0.46875, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.6666666666666666, "status": 0.9166666666666666, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1136}, "6": {"facts_total": 64, "facts_retained": 28, "retention_rate": 0.4375, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.6666666666666666, "status": 0.9166666666666666, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1182}, "7": {"facts_total": 64, "facts_retained": 28, "retention_rate": 0.4375, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.6666666666666666, "status": 0.9166666666666666, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1127}, "8": {"facts_total": 64, "facts_retained": 28, "retention_rate": 0.4375, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.6666666666666666, "status": 0.9166666666666666, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1153}}}, "reference_key_id": "keys/synthetic/syn_03006_2000.json", "task_scores": {"correct": 1.0, "abstained": 1.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "a30b88cb731dc60bcffb9d27bc8ac6c2fede0f9c48d1a7115b443f92f9a96511", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "d3ff8ed858c384858bcdc130", "attempt_id": "4895a426dea6", "run_group_id": "E4_core:A4|closed|qwen3b-llama3b:h8", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:09:56.009154Z", "task_family": "synthetic", "document_id": "syn_03006_2000", "document_hash": "89be5e77e0ca114963e9ffdf8dc38fe5610f13a78ac29e7a1956cd9124ce5217", "question_id": "q14_unanswerable", "split": "test", "cluster_id": "syn_03006_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "tokenizer_hashes": {"receiver": "b9d8872cf3a99da99e1fc4b0f71232a451632dc49ca4c6771549b4efa5c1c763"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 8, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1717, "query_tokens": 501, "message_tokens": 384, "message_utf8_bytes": 1153, "cache_tensor_bytes": null, "serialized_payload_bytes": 1153, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 144511.98237319477, "hops": {"1": 22491.57329101581, "2": 16662.436791113578, "3": 18511.050833039917, "4": 16065.387791953981, "5": 20273.956040968187, "6": 15294.222000055015, "7": 20052.09716700483, "8": 15161.258458043449}, "receiver_prefill_ms": 369.23520790878683, "ttft_ms": 35.00899998471141, "decode_ms": 27.367625036276877}, "total_wall_ms": 144944.15308220778, "ttft_ms": 35.00899998471141, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 64, "facts_retained": 28, "retention_rate": 0.4375, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.6666666666666666, "status": 0.9166666666666666, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1153}, "ledger_by_hop": {"1": {"facts_total": 64, "facts_retained": 30, "retention_rate": 0.46875, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.6666666666666666, "status": 0.9166666666666666, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1087}, "2": {"facts_total": 64, "facts_retained": 30, "retention_rate": 0.46875, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.6666666666666666, "status": 0.9166666666666666, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1071}, "3": {"facts_total": 64, "facts_retained": 30, "retention_rate": 0.46875, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.6666666666666666, "status": 0.9166666666666666, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 962}, "4": {"facts_total": 64, "facts_retained": 30, "retention_rate": 0.46875, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.6666666666666666, "status": 0.9166666666666666, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1250}, "5": {"facts_total": 64, "facts_retained": 30, "retention_rate": 0.46875, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.6666666666666666, "status": 0.9166666666666666, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1136}, "6": {"facts_total": 64, "facts_retained": 28, "retention_rate": 0.4375, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.6666666666666666, "status": 0.9166666666666666, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1182}, "7": {"facts_total": 64, "facts_retained": 28, "retention_rate": 0.4375, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.6666666666666666, "status": 0.9166666666666666, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1127}, "8": {"facts_total": 64, "facts_retained": 28, "retention_rate": 0.4375, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.6666666666666666, "status": 0.9166666666666666, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1153}}}, "reference_key_id": "keys/synthetic/syn_03006_2000.json", "task_scores": {"correct": 1.0, "abstained": 1.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "a30b88cb731dc60bcffb9d27bc8ac6c2fede0f9c48d1a7115b443f92f9a96511", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "a382356fd0e39f9b19c34c4f", "attempt_id": "a5e405a9abe4", "run_group_id": "E4_core:A5|closed|qwen3b-llama3b:h1", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:10:19.148899Z", "task_family": "synthetic", "document_id": "syn_03006_2000", "document_hash": "89be5e77e0ca114963e9ffdf8dc38fe5610f13a78ac29e7a1956cd9124ce5217", "question_id": "q00_single_fact", "split": "test", "cluster_id": "syn_03006_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 1, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1890, "query_tokens": 387, "message_tokens": 304, "message_utf8_bytes": 717, "cache_tensor_bytes": null, "serialized_payload_bytes": 717, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 22688.84387507569, "hops": {"1": 22688.84387507569}, "receiver_prefill_ms": 314.622791018337, "ttft_ms": 63.001667032949626, "decode_ms": 121.10941600985825}, "total_wall_ms": 23188.227542093955, "ttft_ms": 63.001667032949626, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 64, "facts_retained": 10, "retention_rate": 0.15625, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.2, "amount": 0.16666666666666666, "status": 0.3333333333333333, "place": 0.0, "identifier": 0.0, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 669}, "ledger_by_hop": {"1": {"facts_total": 64, "facts_retained": 10, "retention_rate": 0.15625, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.2, "amount": 0.16666666666666666, "status": 0.3333333333333333, "place": 0.0, "identifier": 0.0, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 669}}}, "reference_key_id": "keys/synthetic/syn_03006_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "a7a61e059050ba2e9e391181c042fc9f413143ee1a9ef5f897c5cddf9f37fe4f", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "2cdfe4077c9b193f604c4351", "attempt_id": "c928cd68c4af", "run_group_id": "E4_core:A5|closed|qwen3b-llama3b:h1", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:10:19.665837Z", "task_family": "synthetic", "document_id": "syn_03006_2000", "document_hash": "89be5e77e0ca114963e9ffdf8dc38fe5610f13a78ac29e7a1956cd9124ce5217", "question_id": "q01_single_fact", "split": "test", "cluster_id": "syn_03006_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 1, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1890, "query_tokens": 392, "message_tokens": 304, "message_utf8_bytes": 717, "cache_tensor_bytes": null, "serialized_payload_bytes": 717, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 22688.84387507569, "hops": {"1": 22688.84387507569}, "receiver_prefill_ms": 283.22154202032834, "ttft_ms": 38.05362503044307, "decode_ms": 24.042624980211258}, "total_wall_ms": 23034.65741709806, "ttft_ms": 38.05362503044307, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 64, "facts_retained": 10, "retention_rate": 0.15625, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.2, "amount": 0.16666666666666666, "status": 0.3333333333333333, "place": 0.0, "identifier": 0.0, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 669}, "ledger_by_hop": {"1": {"facts_total": 64, "facts_retained": 10, "retention_rate": 0.15625, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.2, "amount": 0.16666666666666666, "status": 0.3333333333333333, "place": 0.0, "identifier": 0.0, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 669}}}, "reference_key_id": "keys/synthetic/syn_03006_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "a7a61e059050ba2e9e391181c042fc9f413143ee1a9ef5f897c5cddf9f37fe4f", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "ee34ec6098a1704eac4fce3c", "attempt_id": "f1dc811a44fd", "run_group_id": "E4_core:A5|closed|qwen3b-llama3b:h1", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:10:20.030097Z", "task_family": "synthetic", "document_id": "syn_03006_2000", "document_hash": "89be5e77e0ca114963e9ffdf8dc38fe5610f13a78ac29e7a1956cd9124ce5217", "question_id": "q02_exact_copy", "split": "test", "cluster_id": "syn_03006_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 1, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1890, "query_tokens": 391, "message_tokens": 304, "message_utf8_bytes": 717, "cache_tensor_bytes": null, "serialized_payload_bytes": 717, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 22688.84387507569, "hops": {"1": 22688.84387507569}, "receiver_prefill_ms": 291.26258299220353, "ttft_ms": 27.24733401555568, "decode_ms": 37.140000029467046}, "total_wall_ms": 23044.916583108716, "ttft_ms": 27.24733401555568, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 64, "facts_retained": 10, "retention_rate": 0.15625, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.2, "amount": 0.16666666666666666, "status": 0.3333333333333333, "place": 0.0, "identifier": 0.0, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 669}, "ledger_by_hop": {"1": {"facts_total": 64, "facts_retained": 10, "retention_rate": 0.15625, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.2, "amount": 0.16666666666666666, "status": 0.3333333333333333, "place": 0.0, "identifier": 0.0, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 669}}}, "reference_key_id": "keys/synthetic/syn_03006_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "a7a61e059050ba2e9e391181c042fc9f413143ee1a9ef5f897c5cddf9f37fe4f", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "ff5bc1d96301914c1f6f4890", "attempt_id": "28a3846142eb", "run_group_id": "E4_core:A5|closed|qwen3b-llama3b:h1", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:10:20.402531Z", "task_family": "synthetic", "document_id": "syn_03006_2000", "document_hash": "89be5e77e0ca114963e9ffdf8dc38fe5610f13a78ac29e7a1956cd9124ce5217", "question_id": "q03_exact_copy", "split": "test", "cluster_id": "syn_03006_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 1, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1890, "query_tokens": 386, "message_tokens": 304, "message_utf8_bytes": 717, "cache_tensor_bytes": null, "serialized_payload_bytes": 717, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 22688.84387507569, "hops": {"1": 22688.84387507569}, "receiver_prefill_ms": 261.69408403802663, "ttft_ms": 56.72220792621374, "decode_ms": 60.83904206752777}, "total_wall_ms": 23068.627958069555, "ttft_ms": 56.72220792621374, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 64, "facts_retained": 10, "retention_rate": 0.15625, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.2, "amount": 0.16666666666666666, "status": 0.3333333333333333, "place": 0.0, "identifier": 0.0, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 669}, "ledger_by_hop": {"1": {"facts_total": 64, "facts_retained": 10, "retention_rate": 0.15625, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.2, "amount": 0.16666666666666666, "status": 0.3333333333333333, "place": 0.0, "identifier": 0.0, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 669}}}, "reference_key_id": "keys/synthetic/syn_03006_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "a7a61e059050ba2e9e391181c042fc9f413143ee1a9ef5f897c5cddf9f37fe4f", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "9f61071da36ec23ebb878a83", "attempt_id": "d8645fb103ec", "run_group_id": "E4_core:A5|closed|qwen3b-llama3b:h1", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:10:20.799211Z", "task_family": "synthetic", "document_id": "syn_03006_2000", "document_hash": "89be5e77e0ca114963e9ffdf8dc38fe5610f13a78ac29e7a1956cd9124ce5217", "question_id": "q04_multi_hop", "split": "test", "cluster_id": "syn_03006_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 1, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1890, "query_tokens": 395, "message_tokens": 304, "message_utf8_bytes": 717, "cache_tensor_bytes": null, "serialized_payload_bytes": 717, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 22688.84387507569, "hops": {"1": 22688.84387507569}, "receiver_prefill_ms": 334.38720903359354, "ttft_ms": 64.00091596879065, "decode_ms": 98.64441701211035}, "total_wall_ms": 23186.359458137304, "ttft_ms": 64.00091596879065, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 64, "facts_retained": 10, "retention_rate": 0.15625, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.2, "amount": 0.16666666666666666, "status": 0.3333333333333333, "place": 0.0, "identifier": 0.0, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 669}, "ledger_by_hop": {"1": {"facts_total": 64, "facts_retained": 10, "retention_rate": 0.15625, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.2, "amount": 0.16666666666666666, "status": 0.3333333333333333, "place": 0.0, "identifier": 0.0, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 669}}}, "reference_key_id": "keys/synthetic/syn_03006_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "a7a61e059050ba2e9e391181c042fc9f413143ee1a9ef5f897c5cddf9f37fe4f", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "87016c56457c5d2806f4c9fc", "attempt_id": "18e2a4c4420c", "run_group_id": "E4_core:A5|closed|qwen3b-llama3b:h1", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:10:21.315503Z", "task_family": "synthetic", "document_id": "syn_03006_2000", "document_hash": "89be5e77e0ca114963e9ffdf8dc38fe5610f13a78ac29e7a1956cd9124ce5217", "question_id": "q05_multi_hop", "split": "test", "cluster_id": "syn_03006_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 1, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1890, "query_tokens": 398, "message_tokens": 304, "message_utf8_bytes": 717, "cache_tensor_bytes": null, "serialized_payload_bytes": 717, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 22688.84387507569, "hops": {"1": 22688.84387507569}, "receiver_prefill_ms": 278.4962080186233, "ttft_ms": 25.623166002333164, "decode_ms": 41.80570808239281}, "total_wall_ms": 23035.230500041507, "ttft_ms": 25.623166002333164, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 64, "facts_retained": 10, "retention_rate": 0.15625, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.2, "amount": 0.16666666666666666, "status": 0.3333333333333333, "place": 0.0, "identifier": 0.0, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 669}, "ledger_by_hop": {"1": {"facts_total": 64, "facts_retained": 10, "retention_rate": 0.15625, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.2, "amount": 0.16666666666666666, "status": 0.3333333333333333, "place": 0.0, "identifier": 0.0, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 669}}}, "reference_key_id": "keys/synthetic/syn_03006_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "a7a61e059050ba2e9e391181c042fc9f413143ee1a9ef5f897c5cddf9f37fe4f", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "08fe3f52365293a8160204db", "attempt_id": "0ab24d75c75e", "run_group_id": "E4_core:A5|closed|qwen3b-llama3b:h1", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:10:21.679420Z", "task_family": "synthetic", "document_id": "syn_03006_2000", "document_hash": "89be5e77e0ca114963e9ffdf8dc38fe5610f13a78ac29e7a1956cd9124ce5217", "question_id": "q06_negation", "split": "test", "cluster_id": "syn_03006_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 1, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1890, "query_tokens": 386, "message_tokens": 304, "message_utf8_bytes": 717, "cache_tensor_bytes": null, "serialized_payload_bytes": 717, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 22688.84387507569, "hops": {"1": 22688.84387507569}, "receiver_prefill_ms": 295.2111669583246, "ttft_ms": 25.021541980095208, "decode_ms": 37.67504100687802}, "total_wall_ms": 23047.402542084455, "ttft_ms": 25.021541980095208, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 64, "facts_retained": 10, "retention_rate": 0.15625, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.2, "amount": 0.16666666666666666, "status": 0.3333333333333333, "place": 0.0, "identifier": 0.0, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 669}, "ledger_by_hop": {"1": {"facts_total": 64, "facts_retained": 10, "retention_rate": 0.15625, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.2, "amount": 0.16666666666666666, "status": 0.3333333333333333, "place": 0.0, "identifier": 0.0, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 669}}}, "reference_key_id": "keys/synthetic/syn_03006_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "a7a61e059050ba2e9e391181c042fc9f413143ee1a9ef5f897c5cddf9f37fe4f", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "fc4868d42f356735a7d50b67", "attempt_id": "d70f87271526", "run_group_id": "E4_core:A5|closed|qwen3b-llama3b:h1", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:10:22.057099Z", "task_family": "synthetic", "document_id": "syn_03006_2000", "document_hash": "89be5e77e0ca114963e9ffdf8dc38fe5610f13a78ac29e7a1956cd9124ce5217", "question_id": "q07_negation", "split": "test", "cluster_id": "syn_03006_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 1, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1890, "query_tokens": 386, "message_tokens": 304, "message_utf8_bytes": 717, "cache_tensor_bytes": null, "serialized_payload_bytes": 717, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 22688.84387507569, "hops": {"1": 22688.84387507569}, "receiver_prefill_ms": 256.0999169945717, "ttft_ms": 61.92550004925579, "decode_ms": 69.02116595301777}, "total_wall_ms": 23076.58204215113, "ttft_ms": 61.92550004925579, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "no", "evidence_artifact": {"ledger": {"facts_total": 64, "facts_retained": 10, "retention_rate": 0.15625, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.2, "amount": 0.16666666666666666, "status": 0.3333333333333333, "place": 0.0, "identifier": 0.0, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 669}, "ledger_by_hop": {"1": {"facts_total": 64, "facts_retained": 10, "retention_rate": 0.15625, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.2, "amount": 0.16666666666666666, "status": 0.3333333333333333, "place": 0.0, "identifier": 0.0, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 669}}}, "reference_key_id": "keys/synthetic/syn_03006_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "a7a61e059050ba2e9e391181c042fc9f413143ee1a9ef5f897c5cddf9f37fe4f", "output_hash": "9390298f3fb0c5b160498935d79cb139aef28e1c47358b4bbba61862b9c26e59"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "15375a7c6ff6d2e26eb09446", "attempt_id": "078882d392dc", "run_group_id": "E4_core:A5|closed|qwen3b-llama3b:h1", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:10:22.463276Z", "task_family": "synthetic", "document_id": "syn_03006_2000", "document_hash": "89be5e77e0ca114963e9ffdf8dc38fe5610f13a78ac29e7a1956cd9124ce5217", "question_id": "q08_revision", "split": "test", "cluster_id": "syn_03006_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 1, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1890, "query_tokens": 403, "message_tokens": 304, "message_utf8_bytes": 717, "cache_tensor_bytes": null, "serialized_payload_bytes": 717, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 22688.84387507569, "hops": {"1": 22688.84387507569}, "receiver_prefill_ms": 375.131499953568, "ttft_ms": 118.84229199495167, "decode_ms": 292.4161250703037}, "total_wall_ms": 23475.744166062213, "ttft_ms": 118.84229199495167, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 8}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "10107545", "evidence_artifact": {"ledger": {"facts_total": 64, "facts_retained": 10, "retention_rate": 0.15625, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.2, "amount": 0.16666666666666666, "status": 0.3333333333333333, "place": 0.0, "identifier": 0.0, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 669}, "ledger_by_hop": {"1": {"facts_total": 64, "facts_retained": 10, "retention_rate": 0.15625, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.2, "amount": 0.16666666666666666, "status": 0.3333333333333333, "place": 0.0, "identifier": 0.0, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 669}}}, "reference_key_id": "keys/synthetic/syn_03006_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "a7a61e059050ba2e9e391181c042fc9f413143ee1a9ef5f897c5cddf9f37fe4f", "output_hash": "6faba9de4e29949bbdd8934f69fc2110f07076db6801b211aeca9351292537fe"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "dccf2ea7999a7a0956bdcf2c", "attempt_id": "b5b037fd663a", "run_group_id": "E4_core:A5|closed|qwen3b-llama3b:h1", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:10:23.268796Z", "task_family": "synthetic", "document_id": "syn_03006_2000", "document_hash": "89be5e77e0ca114963e9ffdf8dc38fe5610f13a78ac29e7a1956cd9124ce5217", "question_id": "q09_revision", "split": "test", "cluster_id": "syn_03006_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 1, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1890, "query_tokens": 396, "message_tokens": 304, "message_utf8_bytes": 717, "cache_tensor_bytes": null, "serialized_payload_bytes": 717, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 22688.84387507569, "hops": {"1": 22688.84387507569}, "receiver_prefill_ms": 304.20520796906203, "ttft_ms": 32.14054100681096, "decode_ms": 197.27858307305723}, "total_wall_ms": 23222.99320809543, "ttft_ms": 32.14054100681096, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 7}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "9112735", "evidence_artifact": {"ledger": {"facts_total": 64, "facts_retained": 10, "retention_rate": 0.15625, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.2, "amount": 0.16666666666666666, "status": 0.3333333333333333, "place": 0.0, "identifier": 0.0, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 669}, "ledger_by_hop": {"1": {"facts_total": 64, "facts_retained": 10, "retention_rate": 0.15625, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.2, "amount": 0.16666666666666666, "status": 0.3333333333333333, "place": 0.0, "identifier": 0.0, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 669}}}, "reference_key_id": "keys/synthetic/syn_03006_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "a7a61e059050ba2e9e391181c042fc9f413143ee1a9ef5f897c5cddf9f37fe4f", "output_hash": "15a3f804636f0e62d4a3d9be7677f001a0ad6b1fb8ae7e8cdc63c44ec7bac31f"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "1c9be48a40f45249cf3073a3", "attempt_id": "c713b4b4e6d0", "run_group_id": "E4_core:A5|closed|qwen3b-llama3b:h1", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:10:23.820461Z", "task_family": "synthetic", "document_id": "syn_03006_2000", "document_hash": "89be5e77e0ca114963e9ffdf8dc38fe5610f13a78ac29e7a1956cd9124ce5217", "question_id": "q10_temporal", "split": "test", "cluster_id": "syn_03006_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 1, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1890, "query_tokens": 405, "message_tokens": 304, "message_utf8_bytes": 717, "cache_tensor_bytes": null, "serialized_payload_bytes": 717, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 22688.84387507569, "hops": {"1": 22688.84387507569}, "receiver_prefill_ms": 274.4848330039531, "ttft_ms": 61.50491605512798, "decode_ms": 76.81337499525398}, "total_wall_ms": 23102.2268331144, "ttft_ms": 61.50491605512798, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 64, "facts_retained": 10, "retention_rate": 0.15625, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.2, "amount": 0.16666666666666666, "status": 0.3333333333333333, "place": 0.0, "identifier": 0.0, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 669}, "ledger_by_hop": {"1": {"facts_total": 64, "facts_retained": 10, "retention_rate": 0.15625, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.2, "amount": 0.16666666666666666, "status": 0.3333333333333333, "place": 0.0, "identifier": 0.0, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 669}}}, "reference_key_id": "keys/synthetic/syn_03006_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "a7a61e059050ba2e9e391181c042fc9f413143ee1a9ef5f897c5cddf9f37fe4f", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "ec232f08c98ccbf5bfa270b5", "attempt_id": "1e2dec0ed6dc", "run_group_id": "E4_core:A5|closed|qwen3b-llama3b:h1", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:10:24.253334Z", "task_family": "synthetic", "document_id": "syn_03006_2000", "document_hash": "89be5e77e0ca114963e9ffdf8dc38fe5610f13a78ac29e7a1956cd9124ce5217", "question_id": "q11_aggregation", "split": "test", "cluster_id": "syn_03006_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 1, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1890, "query_tokens": 398, "message_tokens": 304, "message_utf8_bytes": 717, "cache_tensor_bytes": null, "serialized_payload_bytes": 717, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 22688.84387507569, "hops": {"1": 22688.84387507569}, "receiver_prefill_ms": 410.126292030327, "ttft_ms": 38.28300000168383, "decode_ms": 30.16733401454985}, "total_wall_ms": 23167.965625063516, "ttft_ms": 38.28300000168383, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 64, "facts_retained": 10, "retention_rate": 0.15625, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.2, "amount": 0.16666666666666666, "status": 0.3333333333333333, "place": 0.0, "identifier": 0.0, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 669}, "ledger_by_hop": {"1": {"facts_total": 64, "facts_retained": 10, "retention_rate": 0.15625, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.2, "amount": 0.16666666666666666, "status": 0.3333333333333333, "place": 0.0, "identifier": 0.0, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 669}}}, "reference_key_id": "keys/synthetic/syn_03006_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "a7a61e059050ba2e9e391181c042fc9f413143ee1a9ef5f897c5cddf9f37fe4f", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "162b9b1b29a728009cfdbee7", "attempt_id": "1ac5ba778b01", "run_group_id": "E4_core:A5|closed|qwen3b-llama3b:h1", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:10:24.751221Z", "task_family": "synthetic", "document_id": "syn_03006_2000", "document_hash": "89be5e77e0ca114963e9ffdf8dc38fe5610f13a78ac29e7a1956cd9124ce5217", "question_id": "q12_unit_conversion", "split": "test", "cluster_id": "syn_03006_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 1, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1890, "query_tokens": 395, "message_tokens": 304, "message_utf8_bytes": 717, "cache_tensor_bytes": null, "serialized_payload_bytes": 717, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 22688.84387507569, "hops": {"1": 22688.84387507569}, "receiver_prefill_ms": 284.2933749780059, "ttft_ms": 35.176875069737434, "decode_ms": 289.77158293128014}, "total_wall_ms": 23298.618750064634, "ttft_ms": 35.176875069737434, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 8}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "11946000", "evidence_artifact": {"ledger": {"facts_total": 64, "facts_retained": 10, "retention_rate": 0.15625, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.2, "amount": 0.16666666666666666, "status": 0.3333333333333333, "place": 0.0, "identifier": 0.0, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 669}, "ledger_by_hop": {"1": {"facts_total": 64, "facts_retained": 10, "retention_rate": 0.15625, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.2, "amount": 0.16666666666666666, "status": 0.3333333333333333, "place": 0.0, "identifier": 0.0, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 669}}}, "reference_key_id": "keys/synthetic/syn_03006_2000.json", "task_scores": {"correct": 0.0, "abstained": 0.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "a7a61e059050ba2e9e391181c042fc9f413143ee1a9ef5f897c5cddf9f37fe4f", "output_hash": "edf27f197d1395ea4e7b86fae1feecd3591246e1299bba6a08655f5724a670ea"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "f7d2ab287873e7de7117e4ac", "attempt_id": "2bce991f4b79", "run_group_id": "E4_core:A5|closed|qwen3b-llama3b:h1", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:10:25.378076Z", "task_family": "synthetic", "document_id": "syn_03006_2000", "document_hash": "89be5e77e0ca114963e9ffdf8dc38fe5610f13a78ac29e7a1956cd9124ce5217", "question_id": "q13_unanswerable", "split": "test", "cluster_id": "syn_03006_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 1, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1890, "query_tokens": 399, "message_tokens": 304, "message_utf8_bytes": 717, "cache_tensor_bytes": null, "serialized_payload_bytes": 717, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 22688.84387507569, "hops": {"1": 22688.84387507569}, "receiver_prefill_ms": 294.09766709432006, "ttft_ms": 54.38704101834446, "decode_ms": 42.30795905459672}, "total_wall_ms": 23080.44429216534, "ttft_ms": 54.38704101834446, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 64, "facts_retained": 10, "retention_rate": 0.15625, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.2, "amount": 0.16666666666666666, "status": 0.3333333333333333, "place": 0.0, "identifier": 0.0, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 669}, "ledger_by_hop": {"1": {"facts_total": 64, "facts_retained": 10, "retention_rate": 0.15625, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.2, "amount": 0.16666666666666666, "status": 0.3333333333333333, "place": 0.0, "identifier": 0.0, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 669}}}, "reference_key_id": "keys/synthetic/syn_03006_2000.json", "task_scores": {"correct": 1.0, "abstained": 1.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "a7a61e059050ba2e9e391181c042fc9f413143ee1a9ef5f897c5cddf9f37fe4f", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "51ed5891846776660529fea8", "attempt_id": "210c65efe0c5", "run_group_id": "E4_core:A5|closed|qwen3b-llama3b:h1", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:10:25.788741Z", "task_family": "synthetic", "document_id": "syn_03006_2000", "document_hash": "89be5e77e0ca114963e9ffdf8dc38fe5610f13a78ac29e7a1956cd9124ce5217", "question_id": "q14_unanswerable", "split": "test", "cluster_id": "syn_03006_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 1, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1890, "query_tokens": 400, "message_tokens": 304, "message_utf8_bytes": 717, "cache_tensor_bytes": null, "serialized_payload_bytes": 717, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 22688.84387507569, "hops": {"1": 22688.84387507569}, "receiver_prefill_ms": 439.3783750711009, "ttft_ms": 30.42479103896767, "decode_ms": 40.99520808085799}, "total_wall_ms": 23200.3617080627, "ttft_ms": 30.42479103896767, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 64, "facts_retained": 10, "retention_rate": 0.15625, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.2, "amount": 0.16666666666666666, "status": 0.3333333333333333, "place": 0.0, "identifier": 0.0, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 669}, "ledger_by_hop": {"1": {"facts_total": 64, "facts_retained": 10, "retention_rate": 0.15625, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.2, "amount": 0.16666666666666666, "status": 0.3333333333333333, "place": 0.0, "identifier": 0.0, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 669}}}, "reference_key_id": "keys/synthetic/syn_03006_2000.json", "task_scores": {"correct": 1.0, "abstained": 1.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "a7a61e059050ba2e9e391181c042fc9f413143ee1a9ef5f897c5cddf9f37fe4f", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "7ba7b7cb05c6f5d7d5f056e3", "attempt_id": "78facfe73359", "run_group_id": "E4_core:A5|closed|qwen3b-llama3b:h2", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:10:42.258339Z", "task_family": "synthetic", "document_id": "syn_03006_2000", "document_hash": "89be5e77e0ca114963e9ffdf8dc38fe5610f13a78ac29e7a1956cd9124ce5217", "question_id": "q00_single_fact", "split": "test", "cluster_id": "syn_03006_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "tokenizer_hashes": {"receiver": "b9d8872cf3a99da99e1fc4b0f71232a451632dc49ca4c6771549b4efa5c1c763"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 2, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1717, "query_tokens": 518, "message_tokens": 414, "message_utf8_bytes": 1205, "cache_tensor_bytes": null, "serialized_payload_bytes": 1205, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 38628.14470811281, "hops": {"1": 22688.84387507569, "2": 15939.300833037123}, "receiver_prefill_ms": 339.9933339096606, "ttft_ms": 35.394250066019595, "decode_ms": 89.75483290851116}, "total_wall_ms": 39094.01520807296, "ttft_ms": 35.394250066019595, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 64, "facts_retained": 15, "retention_rate": 0.234375, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.2, "amount": 0.16666666666666666, "status": 0.75, "place": 0.0, "identifier": 0.0, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1205}, "ledger_by_hop": {"1": {"facts_total": 64, "facts_retained": 10, "retention_rate": 0.15625, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.2, "amount": 0.16666666666666666, "status": 0.3333333333333333, "place": 0.0, "identifier": 0.0, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 669}, "2": {"facts_total": 64, "facts_retained": 15, "retention_rate": 0.234375, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.2, "amount": 0.16666666666666666, "status": 0.75, "place": 0.0, "identifier": 0.0, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1205}}}, "reference_key_id": "keys/synthetic/syn_03006_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "35bddc5bea5498891a8d0cc4e6438f645d89c36a525b6935b8ddcc57f193244a", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "5b752f76fa547a23e0860b4f", "attempt_id": "dc630d3a5b9a", "run_group_id": "E4_core:A5|closed|qwen3b-llama3b:h2", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:10:42.742571Z", "task_family": "synthetic", "document_id": "syn_03006_2000", "document_hash": "89be5e77e0ca114963e9ffdf8dc38fe5610f13a78ac29e7a1956cd9124ce5217", "question_id": "q01_single_fact", "split": "test", "cluster_id": "syn_03006_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "tokenizer_hashes": {"receiver": "b9d8872cf3a99da99e1fc4b0f71232a451632dc49ca4c6771549b4efa5c1c763"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 2, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1717, "query_tokens": 524, "message_tokens": 414, "message_utf8_bytes": 1205, "cache_tensor_bytes": null, "serialized_payload_bytes": 1205, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 38628.14470811281, "hops": {"1": 22688.84387507569, "2": 15939.300833037123}, "receiver_prefill_ms": 418.4347080299631, "ttft_ms": 37.65566600486636, "decode_ms": 24.905207916162908}, "total_wall_ms": 39109.82083308045, "ttft_ms": 37.65566600486636, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 64, "facts_retained": 15, "retention_rate": 0.234375, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.2, "amount": 0.16666666666666666, "status": 0.75, "place": 0.0, "identifier": 0.0, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1205}, "ledger_by_hop": {"1": {"facts_total": 64, "facts_retained": 10, "retention_rate": 0.15625, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.2, "amount": 0.16666666666666666, "status": 0.3333333333333333, "place": 0.0, "identifier": 0.0, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 669}, "2": {"facts_total": 64, "facts_retained": 15, "retention_rate": 0.234375, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.2, "amount": 0.16666666666666666, "status": 0.75, "place": 0.0, "identifier": 0.0, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1205}}}, "reference_key_id": "keys/synthetic/syn_03006_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "35bddc5bea5498891a8d0cc4e6438f645d89c36a525b6935b8ddcc57f193244a", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "b6184eee3d7662ef6c33cfb9", "attempt_id": "300766f71362", "run_group_id": "E4_core:A5|closed|qwen3b-llama3b:h2", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:10:43.243802Z", "task_family": "synthetic", "document_id": "syn_03006_2000", "document_hash": "89be5e77e0ca114963e9ffdf8dc38fe5610f13a78ac29e7a1956cd9124ce5217", "question_id": "q02_exact_copy", "split": "test", "cluster_id": "syn_03006_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "tokenizer_hashes": {"receiver": "b9d8872cf3a99da99e1fc4b0f71232a451632dc49ca4c6771549b4efa5c1c763"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 2, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1717, "query_tokens": 523, "message_tokens": 414, "message_utf8_bytes": 1205, "cache_tensor_bytes": null, "serialized_payload_bytes": 1205, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 38628.14470811281, "hops": {"1": 22688.84387507569, "2": 15939.300833037123}, "receiver_prefill_ms": 342.4763750517741, "ttft_ms": 33.79712498281151, "decode_ms": 22.09170802962035}, "total_wall_ms": 39027.2504161112, "ttft_ms": 33.79712498281151, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 64, "facts_retained": 15, "retention_rate": 0.234375, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.2, "amount": 0.16666666666666666, "status": 0.75, "place": 0.0, "identifier": 0.0, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1205}, "ledger_by_hop": {"1": {"facts_total": 64, "facts_retained": 10, "retention_rate": 0.15625, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.2, "amount": 0.16666666666666666, "status": 0.3333333333333333, "place": 0.0, "identifier": 0.0, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 669}, "2": {"facts_total": 64, "facts_retained": 15, "retention_rate": 0.234375, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.2, "amount": 0.16666666666666666, "status": 0.75, "place": 0.0, "identifier": 0.0, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1205}}}, "reference_key_id": "keys/synthetic/syn_03006_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "35bddc5bea5498891a8d0cc4e6438f645d89c36a525b6935b8ddcc57f193244a", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "ec0b3c1e3e10329392fbee2c", "attempt_id": "419eaf0cf89c", "run_group_id": "E4_core:A5|closed|qwen3b-llama3b:h2", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:10:43.659937Z", "task_family": "synthetic", "document_id": "syn_03006_2000", "document_hash": "89be5e77e0ca114963e9ffdf8dc38fe5610f13a78ac29e7a1956cd9124ce5217", "question_id": "q03_exact_copy", "split": "test", "cluster_id": "syn_03006_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "tokenizer_hashes": {"receiver": "b9d8872cf3a99da99e1fc4b0f71232a451632dc49ca4c6771549b4efa5c1c763"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 2, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1717, "query_tokens": 518, "message_tokens": 414, "message_utf8_bytes": 1205, "cache_tensor_bytes": null, "serialized_payload_bytes": 1205, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 38628.14470811281, "hops": {"1": 22688.84387507569, "2": 15939.300833037123}, "receiver_prefill_ms": 312.1827079448849, "ttft_ms": 11.689083999954164, "decode_ms": 11.536500067450106}, "total_wall_ms": 38964.18966620695, "ttft_ms": 11.689083999954164, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 64, "facts_retained": 15, "retention_rate": 0.234375, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.2, "amount": 0.16666666666666666, "status": 0.75, "place": 0.0, "identifier": 0.0, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1205}, "ledger_by_hop": {"1": {"facts_total": 64, "facts_retained": 10, "retention_rate": 0.15625, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.2, "amount": 0.16666666666666666, "status": 0.3333333333333333, "place": 0.0, "identifier": 0.0, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 669}, "2": {"facts_total": 64, "facts_retained": 15, "retention_rate": 0.234375, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.2, "amount": 0.16666666666666666, "status": 0.75, "place": 0.0, "identifier": 0.0, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1205}}}, "reference_key_id": "keys/synthetic/syn_03006_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "35bddc5bea5498891a8d0cc4e6438f645d89c36a525b6935b8ddcc57f193244a", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "1db445fd268f073a5089df47", "attempt_id": "6dc77a8dbda3", "run_group_id": "E4_core:A5|closed|qwen3b-llama3b:h2", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:10:44.013781Z", "task_family": "synthetic", "document_id": "syn_03006_2000", "document_hash": "89be5e77e0ca114963e9ffdf8dc38fe5610f13a78ac29e7a1956cd9124ce5217", "question_id": "q04_multi_hop", "split": "test", "cluster_id": "syn_03006_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "tokenizer_hashes": {"receiver": "b9d8872cf3a99da99e1fc4b0f71232a451632dc49ca4c6771549b4efa5c1c763"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 2, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1717, "query_tokens": 527, "message_tokens": 414, "message_utf8_bytes": 1205, "cache_tensor_bytes": null, "serialized_payload_bytes": 1205, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 38628.14470811281, "hops": {"1": 22688.84387507569, "2": 15939.300833037123}, "receiver_prefill_ms": 334.4412080477923, "ttft_ms": 54.53962495084852, "decode_ms": 56.15966708865017}, "total_wall_ms": 39074.090416077524, "ttft_ms": 54.53962495084852, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 64, "facts_retained": 15, "retention_rate": 0.234375, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.2, "amount": 0.16666666666666666, "status": 0.75, "place": 0.0, "identifier": 0.0, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1205}, "ledger_by_hop": {"1": {"facts_total": 64, "facts_retained": 10, "retention_rate": 0.15625, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.2, "amount": 0.16666666666666666, "status": 0.3333333333333333, "place": 0.0, "identifier": 0.0, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 669}, "2": {"facts_total": 64, "facts_retained": 15, "retention_rate": 0.234375, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.2, "amount": 0.16666666666666666, "status": 0.75, "place": 0.0, "identifier": 0.0, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1205}}}, "reference_key_id": "keys/synthetic/syn_03006_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "35bddc5bea5498891a8d0cc4e6438f645d89c36a525b6935b8ddcc57f193244a", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "1d295a1dfb5ed7c9e3059e94", "attempt_id": "049c881aaf2f", "run_group_id": "E4_core:A5|closed|qwen3b-llama3b:h2", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:10:44.480461Z", "task_family": "synthetic", "document_id": "syn_03006_2000", "document_hash": "89be5e77e0ca114963e9ffdf8dc38fe5610f13a78ac29e7a1956cd9124ce5217", "question_id": "q05_multi_hop", "split": "test", "cluster_id": "syn_03006_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "tokenizer_hashes": {"receiver": "b9d8872cf3a99da99e1fc4b0f71232a451632dc49ca4c6771549b4efa5c1c763"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 2, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1717, "query_tokens": 528, "message_tokens": 414, "message_utf8_bytes": 1205, "cache_tensor_bytes": null, "serialized_payload_bytes": 1205, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 38628.14470811281, "hops": {"1": 22688.84387507569, "2": 15939.300833037123}, "receiver_prefill_ms": 452.8940829914063, "ttft_ms": 37.77104103937745, "decode_ms": 20.781958010047674}, "total_wall_ms": 39140.16175013967, "ttft_ms": 37.77104103937745, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 64, "facts_retained": 15, "retention_rate": 0.234375, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.2, "amount": 0.16666666666666666, "status": 0.75, "place": 0.0, "identifier": 0.0, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1205}, "ledger_by_hop": {"1": {"facts_total": 64, "facts_retained": 10, "retention_rate": 0.15625, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.2, "amount": 0.16666666666666666, "status": 0.3333333333333333, "place": 0.0, "identifier": 0.0, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 669}, "2": {"facts_total": 64, "facts_retained": 15, "retention_rate": 0.234375, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.2, "amount": 0.16666666666666666, "status": 0.75, "place": 0.0, "identifier": 0.0, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1205}}}, "reference_key_id": "keys/synthetic/syn_03006_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "35bddc5bea5498891a8d0cc4e6438f645d89c36a525b6935b8ddcc57f193244a", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "a9d585380d8b2d7f4478d09f", "attempt_id": "3e8d2430c4b5", "run_group_id": "E4_core:A5|closed|qwen3b-llama3b:h2", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:10:45.010833Z", "task_family": "synthetic", "document_id": "syn_03006_2000", "document_hash": "89be5e77e0ca114963e9ffdf8dc38fe5610f13a78ac29e7a1956cd9124ce5217", "question_id": "q06_negation", "split": "test", "cluster_id": "syn_03006_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "tokenizer_hashes": {"receiver": "b9d8872cf3a99da99e1fc4b0f71232a451632dc49ca4c6771549b4efa5c1c763"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 2, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1717, "query_tokens": 518, "message_tokens": 414, "message_utf8_bytes": 1205, "cache_tensor_bytes": null, "serialized_payload_bytes": 1205, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 38628.14470811281, "hops": {"1": 22688.84387507569, "2": 15939.300833037123}, "receiver_prefill_ms": 334.94141697883606, "ttft_ms": 38.19904197007418, "decode_ms": 24.32441699784249}, "total_wall_ms": 39026.09983307775, "ttft_ms": 38.19904197007418, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 64, "facts_retained": 15, "retention_rate": 0.234375, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.2, "amount": 0.16666666666666666, "status": 0.75, "place": 0.0, "identifier": 0.0, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1205}, "ledger_by_hop": {"1": {"facts_total": 64, "facts_retained": 10, "retention_rate": 0.15625, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.2, "amount": 0.16666666666666666, "status": 0.3333333333333333, "place": 0.0, "identifier": 0.0, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 669}, "2": {"facts_total": 64, "facts_retained": 15, "retention_rate": 0.234375, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.2, "amount": 0.16666666666666666, "status": 0.75, "place": 0.0, "identifier": 0.0, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1205}}}, "reference_key_id": "keys/synthetic/syn_03006_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "35bddc5bea5498891a8d0cc4e6438f645d89c36a525b6935b8ddcc57f193244a", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "831bd19f615891a355850694", "attempt_id": "20fdf529e975", "run_group_id": "E4_core:A5|closed|qwen3b-llama3b:h2", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:10:45.426164Z", "task_family": "synthetic", "document_id": "syn_03006_2000", "document_hash": "89be5e77e0ca114963e9ffdf8dc38fe5610f13a78ac29e7a1956cd9124ce5217", "question_id": "q07_negation", "split": "test", "cluster_id": "syn_03006_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "tokenizer_hashes": {"receiver": "b9d8872cf3a99da99e1fc4b0f71232a451632dc49ca4c6771549b4efa5c1c763"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 2, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1717, "query_tokens": 518, "message_tokens": 414, "message_utf8_bytes": 1205, "cache_tensor_bytes": null, "serialized_payload_bytes": 1205, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 38628.14470811281, "hops": {"1": 22688.84387507569, "2": 15939.300833037123}, "receiver_prefill_ms": 335.99708299152553, "ttft_ms": 19.474916043691337, "decode_ms": 18.983832909725606}, "total_wall_ms": 39003.08324908838, "ttft_ms": 19.474916043691337, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 64, "facts_retained": 15, "retention_rate": 0.234375, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.2, "amount": 0.16666666666666666, "status": 0.75, "place": 0.0, "identifier": 0.0, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1205}, "ledger_by_hop": {"1": {"facts_total": 64, "facts_retained": 10, "retention_rate": 0.15625, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.2, "amount": 0.16666666666666666, "status": 0.3333333333333333, "place": 0.0, "identifier": 0.0, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 669}, "2": {"facts_total": 64, "facts_retained": 15, "retention_rate": 0.234375, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.2, "amount": 0.16666666666666666, "status": 0.75, "place": 0.0, "identifier": 0.0, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1205}}}, "reference_key_id": "keys/synthetic/syn_03006_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "35bddc5bea5498891a8d0cc4e6438f645d89c36a525b6935b8ddcc57f193244a", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "c7e755e113505ee189c4b0cb", "attempt_id": "7e94f5f350cd", "run_group_id": "E4_core:A5|closed|qwen3b-llama3b:h2", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:10:45.817891Z", "task_family": "synthetic", "document_id": "syn_03006_2000", "document_hash": "89be5e77e0ca114963e9ffdf8dc38fe5610f13a78ac29e7a1956cd9124ce5217", "question_id": "q08_revision", "split": "test", "cluster_id": "syn_03006_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "tokenizer_hashes": {"receiver": "b9d8872cf3a99da99e1fc4b0f71232a451632dc49ca4c6771549b4efa5c1c763"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 2, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1717, "query_tokens": 534, "message_tokens": 414, "message_utf8_bytes": 1205, "cache_tensor_bytes": null, "serialized_payload_bytes": 1205, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 38628.14470811281, "hops": {"1": 22688.84387507569, "2": 15939.300833037123}, "receiver_prefill_ms": 353.16025000065565, "ttft_ms": 42.15233295690268, "decode_ms": 349.1665000328794}, "total_wall_ms": 39373.42650012579, "ttft_ms": 42.15233295690268, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 5}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "10,107,545", "evidence_artifact": {"ledger": {"facts_total": 64, "facts_retained": 15, "retention_rate": 0.234375, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.2, "amount": 0.16666666666666666, "status": 0.75, "place": 0.0, "identifier": 0.0, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1205}, "ledger_by_hop": {"1": {"facts_total": 64, "facts_retained": 10, "retention_rate": 0.15625, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.2, "amount": 0.16666666666666666, "status": 0.3333333333333333, "place": 0.0, "identifier": 0.0, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 669}, "2": {"facts_total": 64, "facts_retained": 15, "retention_rate": 0.234375, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.2, "amount": 0.16666666666666666, "status": 0.75, "place": 0.0, "identifier": 0.0, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1205}}}, "reference_key_id": "keys/synthetic/syn_03006_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "35bddc5bea5498891a8d0cc4e6438f645d89c36a525b6935b8ddcc57f193244a", "output_hash": "3d6a277763361ef55b8dd3f9791012605d24cd167ac5bdcdd0064ee16de161d4"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "8748ab7703cd32b77f09a62e", "attempt_id": "9188667d8e0b", "run_group_id": "E4_core:A5|closed|qwen3b-llama3b:h2", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:10:46.583215Z", "task_family": "synthetic", "document_id": "syn_03006_2000", "document_hash": "89be5e77e0ca114963e9ffdf8dc38fe5610f13a78ac29e7a1956cd9124ce5217", "question_id": "q09_revision", "split": "test", "cluster_id": "syn_03006_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "tokenizer_hashes": {"receiver": "b9d8872cf3a99da99e1fc4b0f71232a451632dc49ca4c6771549b4efa5c1c763"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 2, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1717, "query_tokens": 527, "message_tokens": 414, "message_utf8_bytes": 1205, "cache_tensor_bytes": null, "serialized_payload_bytes": 1205, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 38628.14470811281, "hops": {"1": 22688.84387507569, "2": 15939.300833037123}, "receiver_prefill_ms": 336.02158399298787, "ttft_ms": 40.059999912045896, "decode_ms": 149.4368331041187}, "total_wall_ms": 39154.20970809646, "ttft_ms": 40.059999912045896, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 5}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "10,107,545", "evidence_artifact": {"ledger": {"facts_total": 64, "facts_retained": 15, "retention_rate": 0.234375, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.2, "amount": 0.16666666666666666, "status": 0.75, "place": 0.0, "identifier": 0.0, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1205}, "ledger_by_hop": {"1": {"facts_total": 64, "facts_retained": 10, "retention_rate": 0.15625, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.2, "amount": 0.16666666666666666, "status": 0.3333333333333333, "place": 0.0, "identifier": 0.0, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 669}, "2": {"facts_total": 64, "facts_retained": 15, "retention_rate": 0.234375, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.2, "amount": 0.16666666666666666, "status": 0.75, "place": 0.0, "identifier": 0.0, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1205}}}, "reference_key_id": "keys/synthetic/syn_03006_2000.json", "task_scores": {"correct": 0.0, "abstained": 0.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "35bddc5bea5498891a8d0cc4e6438f645d89c36a525b6935b8ddcc57f193244a", "output_hash": "3d6a277763361ef55b8dd3f9791012605d24cd167ac5bdcdd0064ee16de161d4"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "69f9bf62e79d01d993ae02e8", "attempt_id": "fd7863e3da96", "run_group_id": "E4_core:A5|closed|qwen3b-llama3b:h2", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:10:47.126581Z", "task_family": "synthetic", "document_id": "syn_03006_2000", "document_hash": "89be5e77e0ca114963e9ffdf8dc38fe5610f13a78ac29e7a1956cd9124ce5217", "question_id": "q10_temporal", "split": "test", "cluster_id": "syn_03006_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "tokenizer_hashes": {"receiver": "b9d8872cf3a99da99e1fc4b0f71232a451632dc49ca4c6771549b4efa5c1c763"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 2, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1717, "query_tokens": 536, "message_tokens": 414, "message_utf8_bytes": 1205, "cache_tensor_bytes": null, "serialized_payload_bytes": 1205, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 38628.14470811281, "hops": {"1": 22688.84387507569, "2": 15939.300833037123}, "receiver_prefill_ms": 337.8632499370724, "ttft_ms": 22.33854099176824, "decode_ms": 16.82429201900959}, "total_wall_ms": 39005.705874064006, "ttft_ms": 22.33854099176824, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 64, "facts_retained": 15, "retention_rate": 0.234375, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.2, "amount": 0.16666666666666666, "status": 0.75, "place": 0.0, "identifier": 0.0, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1205}, "ledger_by_hop": {"1": {"facts_total": 64, "facts_retained": 10, "retention_rate": 0.15625, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.2, "amount": 0.16666666666666666, "status": 0.3333333333333333, "place": 0.0, "identifier": 0.0, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 669}, "2": {"facts_total": 64, "facts_retained": 15, "retention_rate": 0.234375, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.2, "amount": 0.16666666666666666, "status": 0.75, "place": 0.0, "identifier": 0.0, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1205}}}, "reference_key_id": "keys/synthetic/syn_03006_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "35bddc5bea5498891a8d0cc4e6438f645d89c36a525b6935b8ddcc57f193244a", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "23438e684a27fd9efcebb99b", "attempt_id": "3a087a0bbf65", "run_group_id": "E4_core:A5|closed|qwen3b-llama3b:h2", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:10:47.521893Z", "task_family": "synthetic", "document_id": "syn_03006_2000", "document_hash": "89be5e77e0ca114963e9ffdf8dc38fe5610f13a78ac29e7a1956cd9124ce5217", "question_id": "q11_aggregation", "split": "test", "cluster_id": "syn_03006_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "tokenizer_hashes": {"receiver": "b9d8872cf3a99da99e1fc4b0f71232a451632dc49ca4c6771549b4efa5c1c763"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 2, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1717, "query_tokens": 530, "message_tokens": 414, "message_utf8_bytes": 1205, "cache_tensor_bytes": null, "serialized_payload_bytes": 1205, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 38628.14470811281, "hops": {"1": 22688.84387507569, "2": 15939.300833037123}, "receiver_prefill_ms": 349.0674579516053, "ttft_ms": 46.26137495506555, "decode_ms": 56.15870910696685}, "total_wall_ms": 39080.38275013678, "ttft_ms": 46.26137495506555, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 64, "facts_retained": 15, "retention_rate": 0.234375, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.2, "amount": 0.16666666666666666, "status": 0.75, "place": 0.0, "identifier": 0.0, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1205}, "ledger_by_hop": {"1": {"facts_total": 64, "facts_retained": 10, "retention_rate": 0.15625, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.2, "amount": 0.16666666666666666, "status": 0.3333333333333333, "place": 0.0, "identifier": 0.0, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 669}, "2": {"facts_total": 64, "facts_retained": 15, "retention_rate": 0.234375, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.2, "amount": 0.16666666666666666, "status": 0.75, "place": 0.0, "identifier": 0.0, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1205}}}, "reference_key_id": "keys/synthetic/syn_03006_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "35bddc5bea5498891a8d0cc4e6438f645d89c36a525b6935b8ddcc57f193244a", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "99067723f807257b79313ac8", "attempt_id": "52a76dbb9be2", "run_group_id": "E4_core:A5|closed|qwen3b-llama3b:h2", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:10:47.994552Z", "task_family": "synthetic", "document_id": "syn_03006_2000", "document_hash": "89be5e77e0ca114963e9ffdf8dc38fe5610f13a78ac29e7a1956cd9124ce5217", "question_id": "q12_unit_conversion", "split": "test", "cluster_id": "syn_03006_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "tokenizer_hashes": {"receiver": "b9d8872cf3a99da99e1fc4b0f71232a451632dc49ca4c6771549b4efa5c1c763"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 2, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1717, "query_tokens": 526, "message_tokens": 414, "message_utf8_bytes": 1205, "cache_tensor_bytes": null, "serialized_payload_bytes": 1205, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 38628.14470811281, "hops": {"1": 22688.84387507569, "2": 15939.300833037123}, "receiver_prefill_ms": 482.91008302476257, "ttft_ms": 18.405874958261847, "decode_ms": 94.1285410663113}, "total_wall_ms": 39224.182708072476, "ttft_ms": 18.405874958261847, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 3}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "39.7", "evidence_artifact": {"ledger": {"facts_total": 64, "facts_retained": 15, "retention_rate": 0.234375, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.2, "amount": 0.16666666666666666, "status": 0.75, "place": 0.0, "identifier": 0.0, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1205}, "ledger_by_hop": {"1": {"facts_total": 64, "facts_retained": 10, "retention_rate": 0.15625, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.2, "amount": 0.16666666666666666, "status": 0.3333333333333333, "place": 0.0, "identifier": 0.0, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 669}, "2": {"facts_total": 64, "facts_retained": 15, "retention_rate": 0.234375, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.2, "amount": 0.16666666666666666, "status": 0.75, "place": 0.0, "identifier": 0.0, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1205}}}, "reference_key_id": "keys/synthetic/syn_03006_2000.json", "task_scores": {"correct": 0.0, "abstained": 0.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "35bddc5bea5498891a8d0cc4e6438f645d89c36a525b6935b8ddcc57f193244a", "output_hash": "138c4b663aa3391624efc0a351c82fa6339a5f7ce2617bbf767a8c33c9c6c3e7"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "0c0fcfbc16af7e68090400c8", "attempt_id": "2a9d8f345d3f", "run_group_id": "E4_core:A5|closed|qwen3b-llama3b:h2", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:10:48.609866Z", "task_family": "synthetic", "document_id": "syn_03006_2000", "document_hash": "89be5e77e0ca114963e9ffdf8dc38fe5610f13a78ac29e7a1956cd9124ce5217", "question_id": "q13_unanswerable", "split": "test", "cluster_id": "syn_03006_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "tokenizer_hashes": {"receiver": "b9d8872cf3a99da99e1fc4b0f71232a451632dc49ca4c6771549b4efa5c1c763"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 2, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1717, "query_tokens": 530, "message_tokens": 414, "message_utf8_bytes": 1205, "cache_tensor_bytes": null, "serialized_payload_bytes": 1205, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 38628.14470811281, "hops": {"1": 22688.84387507569, "2": 15939.300833037123}, "receiver_prefill_ms": 351.4896660344675, "ttft_ms": 38.957167067565024, "decode_ms": 25.825874996371567}, "total_wall_ms": 39044.92416605353, "ttft_ms": 38.957167067565024, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 64, "facts_retained": 15, "retention_rate": 0.234375, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.2, "amount": 0.16666666666666666, "status": 0.75, "place": 0.0, "identifier": 0.0, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1205}, "ledger_by_hop": {"1": {"facts_total": 64, "facts_retained": 10, "retention_rate": 0.15625, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.2, "amount": 0.16666666666666666, "status": 0.3333333333333333, "place": 0.0, "identifier": 0.0, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 669}, "2": {"facts_total": 64, "facts_retained": 15, "retention_rate": 0.234375, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.2, "amount": 0.16666666666666666, "status": 0.75, "place": 0.0, "identifier": 0.0, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1205}}}, "reference_key_id": "keys/synthetic/syn_03006_2000.json", "task_scores": {"correct": 1.0, "abstained": 1.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "35bddc5bea5498891a8d0cc4e6438f645d89c36a525b6935b8ddcc57f193244a", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "c4fbb7ec0dfc7d9048db860e", "attempt_id": "643f82d8d9f7", "run_group_id": "E4_core:A5|closed|qwen3b-llama3b:h2", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:10:49.044932Z", "task_family": "synthetic", "document_id": "syn_03006_2000", "document_hash": "89be5e77e0ca114963e9ffdf8dc38fe5610f13a78ac29e7a1956cd9124ce5217", "question_id": "q14_unanswerable", "split": "test", "cluster_id": "syn_03006_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "tokenizer_hashes": {"receiver": "b9d8872cf3a99da99e1fc4b0f71232a451632dc49ca4c6771549b4efa5c1c763"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 2, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1717, "query_tokens": 532, "message_tokens": 414, "message_utf8_bytes": 1205, "cache_tensor_bytes": null, "serialized_payload_bytes": 1205, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 38628.14470811281, "hops": {"1": 22688.84387507569, "2": 15939.300833037123}, "receiver_prefill_ms": 301.6297920839861, "ttft_ms": 56.07299995608628, "decode_ms": 39.67566695064306}, "total_wall_ms": 39026.268375106156, "ttft_ms": 56.07299995608628, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 64, "facts_retained": 15, "retention_rate": 0.234375, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.2, "amount": 0.16666666666666666, "status": 0.75, "place": 0.0, "identifier": 0.0, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1205}, "ledger_by_hop": {"1": {"facts_total": 64, "facts_retained": 10, "retention_rate": 0.15625, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.2, "amount": 0.16666666666666666, "status": 0.3333333333333333, "place": 0.0, "identifier": 0.0, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 669}, "2": {"facts_total": 64, "facts_retained": 15, "retention_rate": 0.234375, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.2, "amount": 0.16666666666666666, "status": 0.75, "place": 0.0, "identifier": 0.0, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1205}}}, "reference_key_id": "keys/synthetic/syn_03006_2000.json", "task_scores": {"correct": 1.0, "abstained": 1.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "35bddc5bea5498891a8d0cc4e6438f645d89c36a525b6935b8ddcc57f193244a", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "dfa9f47b9e70e68572e6366a", "attempt_id": "2e47bd75f34c", "run_group_id": "E4_core:A5|closed|qwen3b-llama3b:h4", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:11:26.138965Z", "task_family": "synthetic", "document_id": "syn_03006_2000", "document_hash": "89be5e77e0ca114963e9ffdf8dc38fe5610f13a78ac29e7a1956cd9124ce5217", "question_id": "q00_single_fact", "split": "test", "cluster_id": "syn_03006_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "tokenizer_hashes": {"receiver": "b9d8872cf3a99da99e1fc4b0f71232a451632dc49ca4c6771549b4efa5c1c763"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 4, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1717, "query_tokens": 518, "message_tokens": 414, "message_utf8_bytes": 1205, "cache_tensor_bytes": null, "serialized_payload_bytes": 1205, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 75303.94629202783, "hops": {"1": 22688.84387507569, "2": 15939.300833037123, "3": 20253.31387494225, "4": 16422.487708972767}, "receiver_prefill_ms": 334.58541706204414, "ttft_ms": 17.318082973361015, "decode_ms": 34.697166993282735}, "total_wall_ms": 75691.16654212121, "ttft_ms": 17.318082973361015, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 64, "facts_retained": 15, "retention_rate": 0.234375, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.2, "amount": 0.16666666666666666, "status": 0.75, "place": 0.0, "identifier": 0.0, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1205}, "ledger_by_hop": {"1": {"facts_total": 64, "facts_retained": 10, "retention_rate": 0.15625, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.2, "amount": 0.16666666666666666, "status": 0.3333333333333333, "place": 0.0, "identifier": 0.0, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 669}, "2": {"facts_total": 64, "facts_retained": 15, "retention_rate": 0.234375, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.2, "amount": 0.16666666666666666, "status": 0.75, "place": 0.0, "identifier": 0.0, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1205}, "3": {"facts_total": 64, "facts_retained": 15, "retention_rate": 0.234375, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.2, "amount": 0.16666666666666666, "status": 0.75, "place": 0.0, "identifier": 0.0, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1205}, "4": {"facts_total": 64, "facts_retained": 15, "retention_rate": 0.234375, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.2, "amount": 0.16666666666666666, "status": 0.75, "place": 0.0, "identifier": 0.0, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1205}}}, "reference_key_id": "keys/synthetic/syn_03006_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "35bddc5bea5498891a8d0cc4e6438f645d89c36a525b6935b8ddcc57f193244a", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "6a9ae1ec4b0ca7e111471adb", "attempt_id": "35063da530cd", "run_group_id": "E4_core:A5|closed|qwen3b-llama3b:h4", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:11:26.544196Z", "task_family": "synthetic", "document_id": "syn_03006_2000", "document_hash": "89be5e77e0ca114963e9ffdf8dc38fe5610f13a78ac29e7a1956cd9124ce5217", "question_id": "q01_single_fact", "split": "test", "cluster_id": "syn_03006_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "tokenizer_hashes": {"receiver": "b9d8872cf3a99da99e1fc4b0f71232a451632dc49ca4c6771549b4efa5c1c763"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 4, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1717, "query_tokens": 524, "message_tokens": 414, "message_utf8_bytes": 1205, "cache_tensor_bytes": null, "serialized_payload_bytes": 1205, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 75303.94629202783, "hops": {"1": 22688.84387507569, "2": 15939.300833037123, "3": 20253.31387494225, "4": 16422.487708972767}, "receiver_prefill_ms": 362.61212499812245, "ttft_ms": 50.297207897529006, "decode_ms": 55.85491703823209}, "total_wall_ms": 75773.42087507714, "ttft_ms": 50.297207897529006, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 64, "facts_retained": 15, "retention_rate": 0.234375, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.2, "amount": 0.16666666666666666, "status": 0.75, "place": 0.0, "identifier": 0.0, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1205}, "ledger_by_hop": {"1": {"facts_total": 64, "facts_retained": 10, "retention_rate": 0.15625, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.2, "amount": 0.16666666666666666, "status": 0.3333333333333333, "place": 0.0, "identifier": 0.0, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 669}, "2": {"facts_total": 64, "facts_retained": 15, "retention_rate": 0.234375, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.2, "amount": 0.16666666666666666, "status": 0.75, "place": 0.0, "identifier": 0.0, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1205}, "3": {"facts_total": 64, "facts_retained": 15, "retention_rate": 0.234375, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.2, "amount": 0.16666666666666666, "status": 0.75, "place": 0.0, "identifier": 0.0, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1205}, "4": {"facts_total": 64, "facts_retained": 15, "retention_rate": 0.234375, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.2, "amount": 0.16666666666666666, "status": 0.75, "place": 0.0, "identifier": 0.0, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1205}}}, "reference_key_id": "keys/synthetic/syn_03006_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "35bddc5bea5498891a8d0cc4e6438f645d89c36a525b6935b8ddcc57f193244a", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "8f1f53dd95ba14873d9a6527", "attempt_id": "30431ce4fa45", "run_group_id": "E4_core:A5|closed|qwen3b-llama3b:h4", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:11:27.032472Z", "task_family": "synthetic", "document_id": "syn_03006_2000", "document_hash": "89be5e77e0ca114963e9ffdf8dc38fe5610f13a78ac29e7a1956cd9124ce5217", "question_id": "q02_exact_copy", "split": "test", "cluster_id": "syn_03006_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "tokenizer_hashes": {"receiver": "b9d8872cf3a99da99e1fc4b0f71232a451632dc49ca4c6771549b4efa5c1c763"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 4, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1717, "query_tokens": 523, "message_tokens": 414, "message_utf8_bytes": 1205, "cache_tensor_bytes": null, "serialized_payload_bytes": 1205, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 75303.94629202783, "hops": {"1": 22688.84387507569, "2": 15939.300833037123, "3": 20253.31387494225, "4": 16422.487708972767}, "receiver_prefill_ms": 467.23962493706495, "ttft_ms": 38.712667068466544, "decode_ms": 23.661374929361045}, "total_wall_ms": 75834.07558395993, "ttft_ms": 38.712667068466544, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 64, "facts_retained": 15, "retention_rate": 0.234375, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.2, "amount": 0.16666666666666666, "status": 0.75, "place": 0.0, "identifier": 0.0, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1205}, "ledger_by_hop": {"1": {"facts_total": 64, "facts_retained": 10, "retention_rate": 0.15625, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.2, "amount": 0.16666666666666666, "status": 0.3333333333333333, "place": 0.0, "identifier": 0.0, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 669}, "2": {"facts_total": 64, "facts_retained": 15, "retention_rate": 0.234375, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.2, "amount": 0.16666666666666666, "status": 0.75, "place": 0.0, "identifier": 0.0, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1205}, "3": {"facts_total": 64, "facts_retained": 15, "retention_rate": 0.234375, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.2, "amount": 0.16666666666666666, "status": 0.75, "place": 0.0, "identifier": 0.0, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1205}, "4": {"facts_total": 64, "facts_retained": 15, "retention_rate": 0.234375, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.2, "amount": 0.16666666666666666, "status": 0.75, "place": 0.0, "identifier": 0.0, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1205}}}, "reference_key_id": "keys/synthetic/syn_03006_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "35bddc5bea5498891a8d0cc4e6438f645d89c36a525b6935b8ddcc57f193244a", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "e86ae7db9c07452d5280af6c", "attempt_id": "fbe7a4a43337", "run_group_id": "E4_core:A5|closed|qwen3b-llama3b:h4", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:11:27.579858Z", "task_family": "synthetic", "document_id": "syn_03006_2000", "document_hash": "89be5e77e0ca114963e9ffdf8dc38fe5610f13a78ac29e7a1956cd9124ce5217", "question_id": "q03_exact_copy", "split": "test", "cluster_id": "syn_03006_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "tokenizer_hashes": {"receiver": "b9d8872cf3a99da99e1fc4b0f71232a451632dc49ca4c6771549b4efa5c1c763"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 4, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1717, "query_tokens": 518, "message_tokens": 414, "message_utf8_bytes": 1205, "cache_tensor_bytes": null, "serialized_payload_bytes": 1205, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 75303.94629202783, "hops": {"1": 22688.84387507569, "2": 15939.300833037123, "3": 20253.31387494225, "4": 16422.487708972767}, "receiver_prefill_ms": 337.85312506370246, "ttft_ms": 34.1797498986125, "decode_ms": 25.358290993608534}, "total_wall_ms": 75702.02016702387, "ttft_ms": 34.1797498986125, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 64, "facts_retained": 15, "retention_rate": 0.234375, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.2, "amount": 0.16666666666666666, "status": 0.75, "place": 0.0, "identifier": 0.0, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1205}, "ledger_by_hop": {"1": {"facts_total": 64, "facts_retained": 10, "retention_rate": 0.15625, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.2, "amount": 0.16666666666666666, "status": 0.3333333333333333, "place": 0.0, "identifier": 0.0, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 669}, "2": {"facts_total": 64, "facts_retained": 15, "retention_rate": 0.234375, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.2, "amount": 0.16666666666666666, "status": 0.75, "place": 0.0, "identifier": 0.0, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1205}, "3": {"facts_total": 64, "facts_retained": 15, "retention_rate": 0.234375, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.2, "amount": 0.16666666666666666, "status": 0.75, "place": 0.0, "identifier": 0.0, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1205}, "4": {"facts_total": 64, "facts_retained": 15, "retention_rate": 0.234375, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.2, "amount": 0.16666666666666666, "status": 0.75, "place": 0.0, "identifier": 0.0, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1205}}}, "reference_key_id": "keys/synthetic/syn_03006_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "35bddc5bea5498891a8d0cc4e6438f645d89c36a525b6935b8ddcc57f193244a", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "99ac9b593ce7f1534714b035", "attempt_id": "9987672e7c23", "run_group_id": "E4_core:A5|closed|qwen3b-llama3b:h4", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:11:27.995679Z", "task_family": "synthetic", "document_id": "syn_03006_2000", "document_hash": "89be5e77e0ca114963e9ffdf8dc38fe5610f13a78ac29e7a1956cd9124ce5217", "question_id": "q04_multi_hop", "split": "test", "cluster_id": "syn_03006_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "tokenizer_hashes": {"receiver": "b9d8872cf3a99da99e1fc4b0f71232a451632dc49ca4c6771549b4efa5c1c763"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 4, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1717, "query_tokens": 527, "message_tokens": 414, "message_utf8_bytes": 1205, "cache_tensor_bytes": null, "serialized_payload_bytes": 1205, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 75303.94629202783, "hops": {"1": 22688.84387507569, "2": 15939.300833037123, "3": 20253.31387494225, "4": 16422.487708972767}, "receiver_prefill_ms": 342.01637492515147, "ttft_ms": 21.782708005048335, "decode_ms": 16.33825001772493}, "total_wall_ms": 75684.7322920803, "ttft_ms": 21.782708005048335, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 64, "facts_retained": 15, "retention_rate": 0.234375, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.2, "amount": 0.16666666666666666, "status": 0.75, "place": 0.0, "identifier": 0.0, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1205}, "ledger_by_hop": {"1": {"facts_total": 64, "facts_retained": 10, "retention_rate": 0.15625, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.2, "amount": 0.16666666666666666, "status": 0.3333333333333333, "place": 0.0, "identifier": 0.0, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 669}, "2": {"facts_total": 64, "facts_retained": 15, "retention_rate": 0.234375, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.2, "amount": 0.16666666666666666, "status": 0.75, "place": 0.0, "identifier": 0.0, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1205}, "3": {"facts_total": 64, "facts_retained": 15, "retention_rate": 0.234375, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.2, "amount": 0.16666666666666666, "status": 0.75, "place": 0.0, "identifier": 0.0, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1205}, "4": {"facts_total": 64, "facts_retained": 15, "retention_rate": 0.234375, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.2, "amount": 0.16666666666666666, "status": 0.75, "place": 0.0, "identifier": 0.0, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1205}}}, "reference_key_id": "keys/synthetic/syn_03006_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "35bddc5bea5498891a8d0cc4e6438f645d89c36a525b6935b8ddcc57f193244a", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "3cb38e0a93a438c40168672e", "attempt_id": "2db1da37206c", "run_group_id": "E4_core:A5|closed|qwen3b-llama3b:h4", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:11:28.395902Z", "task_family": "synthetic", "document_id": "syn_03006_2000", "document_hash": "89be5e77e0ca114963e9ffdf8dc38fe5610f13a78ac29e7a1956cd9124ce5217", "question_id": "q05_multi_hop", "split": "test", "cluster_id": "syn_03006_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "tokenizer_hashes": {"receiver": "b9d8872cf3a99da99e1fc4b0f71232a451632dc49ca4c6771549b4efa5c1c763"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 4, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1717, "query_tokens": 528, "message_tokens": 414, "message_utf8_bytes": 1205, "cache_tensor_bytes": null, "serialized_payload_bytes": 1205, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 75303.94629202783, "hops": {"1": 22688.84387507569, "2": 15939.300833037123, "3": 20253.31387494225, "4": 16422.487708972767}, "receiver_prefill_ms": 331.69791707769036, "ttft_ms": 62.34075000975281, "decode_ms": 60.499208979308605}, "total_wall_ms": 75759.20870900154, "ttft_ms": 62.34075000975281, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 64, "facts_retained": 15, "retention_rate": 0.234375, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.2, "amount": 0.16666666666666666, "status": 0.75, "place": 0.0, "identifier": 0.0, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1205}, "ledger_by_hop": {"1": {"facts_total": 64, "facts_retained": 10, "retention_rate": 0.15625, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.2, "amount": 0.16666666666666666, "status": 0.3333333333333333, "place": 0.0, "identifier": 0.0, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 669}, "2": {"facts_total": 64, "facts_retained": 15, "retention_rate": 0.234375, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.2, "amount": 0.16666666666666666, "status": 0.75, "place": 0.0, "identifier": 0.0, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1205}, "3": {"facts_total": 64, "facts_retained": 15, "retention_rate": 0.234375, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.2, "amount": 0.16666666666666666, "status": 0.75, "place": 0.0, "identifier": 0.0, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1205}, "4": {"facts_total": 64, "facts_retained": 15, "retention_rate": 0.234375, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.2, "amount": 0.16666666666666666, "status": 0.75, "place": 0.0, "identifier": 0.0, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1205}}}, "reference_key_id": "keys/synthetic/syn_03006_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "35bddc5bea5498891a8d0cc4e6438f645d89c36a525b6935b8ddcc57f193244a", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "145243927ed91c7b6065f5a3", "attempt_id": "d9e3b31f8306", "run_group_id": "E4_core:A5|closed|qwen3b-llama3b:h4", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:11:28.871465Z", "task_family": "synthetic", "document_id": "syn_03006_2000", "document_hash": "89be5e77e0ca114963e9ffdf8dc38fe5610f13a78ac29e7a1956cd9124ce5217", "question_id": "q06_negation", "split": "test", "cluster_id": "syn_03006_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "tokenizer_hashes": {"receiver": "b9d8872cf3a99da99e1fc4b0f71232a451632dc49ca4c6771549b4efa5c1c763"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 4, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1717, "query_tokens": 518, "message_tokens": 414, "message_utf8_bytes": 1205, "cache_tensor_bytes": null, "serialized_payload_bytes": 1205, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 75303.94629202783, "hops": {"1": 22688.84387507569, "2": 15939.300833037123, "3": 20253.31387494225, "4": 16422.487708972767}, "receiver_prefill_ms": 476.138332975097, "ttft_ms": 20.769290975295007, "decode_ms": 37.67174994572997}, "total_wall_ms": 75839.18133401312, "ttft_ms": 20.769290975295007, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 64, "facts_retained": 15, "retention_rate": 0.234375, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.2, "amount": 0.16666666666666666, "status": 0.75, "place": 0.0, "identifier": 0.0, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1205}, "ledger_by_hop": {"1": {"facts_total": 64, "facts_retained": 10, "retention_rate": 0.15625, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.2, "amount": 0.16666666666666666, "status": 0.3333333333333333, "place": 0.0, "identifier": 0.0, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 669}, "2": {"facts_total": 64, "facts_retained": 15, "retention_rate": 0.234375, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.2, "amount": 0.16666666666666666, "status": 0.75, "place": 0.0, "identifier": 0.0, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1205}, "3": {"facts_total": 64, "facts_retained": 15, "retention_rate": 0.234375, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.2, "amount": 0.16666666666666666, "status": 0.75, "place": 0.0, "identifier": 0.0, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1205}, "4": {"facts_total": 64, "facts_retained": 15, "retention_rate": 0.234375, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.2, "amount": 0.16666666666666666, "status": 0.75, "place": 0.0, "identifier": 0.0, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1205}}}, "reference_key_id": "keys/synthetic/syn_03006_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "35bddc5bea5498891a8d0cc4e6438f645d89c36a525b6935b8ddcc57f193244a", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "6a25757270c3e8da325e2e4b", "attempt_id": "9481a2bb430d", "run_group_id": "E4_core:A5|closed|qwen3b-llama3b:h4", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:11:29.423894Z", "task_family": "synthetic", "document_id": "syn_03006_2000", "document_hash": "89be5e77e0ca114963e9ffdf8dc38fe5610f13a78ac29e7a1956cd9124ce5217", "question_id": "q07_negation", "split": "test", "cluster_id": "syn_03006_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "tokenizer_hashes": {"receiver": "b9d8872cf3a99da99e1fc4b0f71232a451632dc49ca4c6771549b4efa5c1c763"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 4, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1717, "query_tokens": 518, "message_tokens": 414, "message_utf8_bytes": 1205, "cache_tensor_bytes": null, "serialized_payload_bytes": 1205, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 75303.94629202783, "hops": {"1": 22688.84387507569, "2": 15939.300833037123, "3": 20253.31387494225, "4": 16422.487708972767}, "receiver_prefill_ms": 347.84804098308086, "ttft_ms": 24.065249948762357, "decode_ms": 32.46424999088049}, "total_wall_ms": 75708.84737605229, "ttft_ms": 24.065249948762357, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 64, "facts_retained": 15, "retention_rate": 0.234375, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.2, "amount": 0.16666666666666666, "status": 0.75, "place": 0.0, "identifier": 0.0, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1205}, "ledger_by_hop": {"1": {"facts_total": 64, "facts_retained": 10, "retention_rate": 0.15625, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.2, "amount": 0.16666666666666666, "status": 0.3333333333333333, "place": 0.0, "identifier": 0.0, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 669}, "2": {"facts_total": 64, "facts_retained": 15, "retention_rate": 0.234375, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.2, "amount": 0.16666666666666666, "status": 0.75, "place": 0.0, "identifier": 0.0, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1205}, "3": {"facts_total": 64, "facts_retained": 15, "retention_rate": 0.234375, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.2, "amount": 0.16666666666666666, "status": 0.75, "place": 0.0, "identifier": 0.0, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1205}, "4": {"facts_total": 64, "facts_retained": 15, "retention_rate": 0.234375, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.2, "amount": 0.16666666666666666, "status": 0.75, "place": 0.0, "identifier": 0.0, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1205}}}, "reference_key_id": "keys/synthetic/syn_03006_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "35bddc5bea5498891a8d0cc4e6438f645d89c36a525b6935b8ddcc57f193244a", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "c2c6983a15fce76602f9c40a", "attempt_id": "83e079759499", "run_group_id": "E4_core:A5|closed|qwen3b-llama3b:h4", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:11:29.846109Z", "task_family": "synthetic", "document_id": "syn_03006_2000", "document_hash": "89be5e77e0ca114963e9ffdf8dc38fe5610f13a78ac29e7a1956cd9124ce5217", "question_id": "q08_revision", "split": "test", "cluster_id": "syn_03006_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "tokenizer_hashes": {"receiver": "b9d8872cf3a99da99e1fc4b0f71232a451632dc49ca4c6771549b4efa5c1c763"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 4, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1717, "query_tokens": 534, "message_tokens": 414, "message_utf8_bytes": 1205, "cache_tensor_bytes": null, "serialized_payload_bytes": 1205, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 75303.94629202783, "hops": {"1": 22688.84387507569, "2": 15939.300833037123, "3": 20253.31387494225, "4": 16422.487708972767}, "receiver_prefill_ms": 317.63791595585644, "ttft_ms": 21.08979201875627, "decode_ms": 282.1225000079721}, "total_wall_ms": 75925.3899169853, "ttft_ms": 21.08979201875627, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 5}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "10,107,545", "evidence_artifact": {"ledger": {"facts_total": 64, "facts_retained": 15, "retention_rate": 0.234375, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.2, "amount": 0.16666666666666666, "status": 0.75, "place": 0.0, "identifier": 0.0, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1205}, "ledger_by_hop": {"1": {"facts_total": 64, "facts_retained": 10, "retention_rate": 0.15625, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.2, "amount": 0.16666666666666666, "status": 0.3333333333333333, "place": 0.0, "identifier": 0.0, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 669}, "2": {"facts_total": 64, "facts_retained": 15, "retention_rate": 0.234375, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.2, "amount": 0.16666666666666666, "status": 0.75, "place": 0.0, "identifier": 0.0, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1205}, "3": {"facts_total": 64, "facts_retained": 15, "retention_rate": 0.234375, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.2, "amount": 0.16666666666666666, "status": 0.75, "place": 0.0, "identifier": 0.0, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1205}, "4": {"facts_total": 64, "facts_retained": 15, "retention_rate": 0.234375, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.2, "amount": 0.16666666666666666, "status": 0.75, "place": 0.0, "identifier": 0.0, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1205}}}, "reference_key_id": "keys/synthetic/syn_03006_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "35bddc5bea5498891a8d0cc4e6438f645d89c36a525b6935b8ddcc57f193244a", "output_hash": "3d6a277763361ef55b8dd3f9791012605d24cd167ac5bdcdd0064ee16de161d4"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "3f16ed58c236b39a61b4efb0", "attempt_id": "479080ff8fdf", "run_group_id": "E4_core:A5|closed|qwen3b-llama3b:h4", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:11:30.486054Z", "task_family": "synthetic", "document_id": "syn_03006_2000", "document_hash": "89be5e77e0ca114963e9ffdf8dc38fe5610f13a78ac29e7a1956cd9124ce5217", "question_id": "q09_revision", "split": "test", "cluster_id": "syn_03006_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "tokenizer_hashes": {"receiver": "b9d8872cf3a99da99e1fc4b0f71232a451632dc49ca4c6771549b4efa5c1c763"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 4, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1717, "query_tokens": 527, "message_tokens": 414, "message_utf8_bytes": 1205, "cache_tensor_bytes": null, "serialized_payload_bytes": 1205, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 75303.94629202783, "hops": {"1": 22688.84387507569, "2": 15939.300833037123, "3": 20253.31387494225, "4": 16422.487708972767}, "receiver_prefill_ms": 454.6140410238877, "ttft_ms": 35.774749936535954, "decode_ms": 145.55862499400973}, "total_wall_ms": 75940.5859169783, "ttft_ms": 35.774749936535954, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 5}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "10,107,545", "evidence_artifact": {"ledger": {"facts_total": 64, "facts_retained": 15, "retention_rate": 0.234375, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.2, "amount": 0.16666666666666666, "status": 0.75, "place": 0.0, "identifier": 0.0, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1205}, "ledger_by_hop": {"1": {"facts_total": 64, "facts_retained": 10, "retention_rate": 0.15625, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.2, "amount": 0.16666666666666666, "status": 0.3333333333333333, "place": 0.0, "identifier": 0.0, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 669}, "2": {"facts_total": 64, "facts_retained": 15, "retention_rate": 0.234375, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.2, "amount": 0.16666666666666666, "status": 0.75, "place": 0.0, "identifier": 0.0, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1205}, "3": {"facts_total": 64, "facts_retained": 15, "retention_rate": 0.234375, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.2, "amount": 0.16666666666666666, "status": 0.75, "place": 0.0, "identifier": 0.0, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1205}, "4": {"facts_total": 64, "facts_retained": 15, "retention_rate": 0.234375, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.2, "amount": 0.16666666666666666, "status": 0.75, "place": 0.0, "identifier": 0.0, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1205}}}, "reference_key_id": "keys/synthetic/syn_03006_2000.json", "task_scores": {"correct": 0.0, "abstained": 0.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "35bddc5bea5498891a8d0cc4e6438f645d89c36a525b6935b8ddcc57f193244a", "output_hash": "3d6a277763361ef55b8dd3f9791012605d24cd167ac5bdcdd0064ee16de161d4"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "0b74f068d2927388ef6d17a0", "attempt_id": "9fbe487c13af", "run_group_id": "E4_core:A5|closed|qwen3b-llama3b:h4", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:11:31.140052Z", "task_family": "synthetic", "document_id": "syn_03006_2000", "document_hash": "89be5e77e0ca114963e9ffdf8dc38fe5610f13a78ac29e7a1956cd9124ce5217", "question_id": "q10_temporal", "split": "test", "cluster_id": "syn_03006_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "tokenizer_hashes": {"receiver": "b9d8872cf3a99da99e1fc4b0f71232a451632dc49ca4c6771549b4efa5c1c763"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 4, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1717, "query_tokens": 536, "message_tokens": 414, "message_utf8_bytes": 1205, "cache_tensor_bytes": null, "serialized_payload_bytes": 1205, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 75303.94629202783, "hops": {"1": 22688.84387507569, "2": 15939.300833037123, "3": 20253.31387494225, "4": 16422.487708972767}, "receiver_prefill_ms": 334.79991694912314, "ttft_ms": 40.04429199267179, "decode_ms": 24.41033406648785}, "total_wall_ms": 75703.68679205421, "ttft_ms": 40.04429199267179, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 64, "facts_retained": 15, "retention_rate": 0.234375, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.2, "amount": 0.16666666666666666, "status": 0.75, "place": 0.0, "identifier": 0.0, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1205}, "ledger_by_hop": {"1": {"facts_total": 64, "facts_retained": 10, "retention_rate": 0.15625, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.2, "amount": 0.16666666666666666, "status": 0.3333333333333333, "place": 0.0, "identifier": 0.0, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 669}, "2": {"facts_total": 64, "facts_retained": 15, "retention_rate": 0.234375, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.2, "amount": 0.16666666666666666, "status": 0.75, "place": 0.0, "identifier": 0.0, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1205}, "3": {"facts_total": 64, "facts_retained": 15, "retention_rate": 0.234375, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.2, "amount": 0.16666666666666666, "status": 0.75, "place": 0.0, "identifier": 0.0, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1205}, "4": {"facts_total": 64, "facts_retained": 15, "retention_rate": 0.234375, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.2, "amount": 0.16666666666666666, "status": 0.75, "place": 0.0, "identifier": 0.0, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1205}}}, "reference_key_id": "keys/synthetic/syn_03006_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "35bddc5bea5498891a8d0cc4e6438f645d89c36a525b6935b8ddcc57f193244a", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "d7fb1fb113ffd6461d21594d", "attempt_id": "b00e8e31d9dd", "run_group_id": "E4_core:A5|closed|qwen3b-llama3b:h4", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:11:31.557552Z", "task_family": "synthetic", "document_id": "syn_03006_2000", "document_hash": "89be5e77e0ca114963e9ffdf8dc38fe5610f13a78ac29e7a1956cd9124ce5217", "question_id": "q11_aggregation", "split": "test", "cluster_id": "syn_03006_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "tokenizer_hashes": {"receiver": "b9d8872cf3a99da99e1fc4b0f71232a451632dc49ca4c6771549b4efa5c1c763"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 4, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1717, "query_tokens": 530, "message_tokens": 414, "message_utf8_bytes": 1205, "cache_tensor_bytes": null, "serialized_payload_bytes": 1205, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 75303.94629202783, "hops": {"1": 22688.84387507569, "2": 15939.300833037123, "3": 20253.31387494225, "4": 16422.487708972767}, "receiver_prefill_ms": 317.63591698836535, "ttft_ms": 35.80512502230704, "decode_ms": 61.8114999961108}, "total_wall_ms": 75720.10391706135, "ttft_ms": 35.80512502230704, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 64, "facts_retained": 15, "retention_rate": 0.234375, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.2, "amount": 0.16666666666666666, "status": 0.75, "place": 0.0, "identifier": 0.0, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1205}, "ledger_by_hop": {"1": {"facts_total": 64, "facts_retained": 10, "retention_rate": 0.15625, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.2, "amount": 0.16666666666666666, "status": 0.3333333333333333, "place": 0.0, "identifier": 0.0, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 669}, "2": {"facts_total": 64, "facts_retained": 15, "retention_rate": 0.234375, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.2, "amount": 0.16666666666666666, "status": 0.75, "place": 0.0, "identifier": 0.0, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1205}, "3": {"facts_total": 64, "facts_retained": 15, "retention_rate": 0.234375, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.2, "amount": 0.16666666666666666, "status": 0.75, "place": 0.0, "identifier": 0.0, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1205}, "4": {"facts_total": 64, "facts_retained": 15, "retention_rate": 0.234375, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.2, "amount": 0.16666666666666666, "status": 0.75, "place": 0.0, "identifier": 0.0, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1205}}}, "reference_key_id": "keys/synthetic/syn_03006_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "35bddc5bea5498891a8d0cc4e6438f645d89c36a525b6935b8ddcc57f193244a", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "6b6c4fa9b61cc853bcdb1a19", "attempt_id": "79a34652580f", "run_group_id": "E4_core:A5|closed|qwen3b-llama3b:h4", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:11:31.991666Z", "task_family": "synthetic", "document_id": "syn_03006_2000", "document_hash": "89be5e77e0ca114963e9ffdf8dc38fe5610f13a78ac29e7a1956cd9124ce5217", "question_id": "q12_unit_conversion", "split": "test", "cluster_id": "syn_03006_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "tokenizer_hashes": {"receiver": "b9d8872cf3a99da99e1fc4b0f71232a451632dc49ca4c6771549b4efa5c1c763"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 4, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1717, "query_tokens": 526, "message_tokens": 414, "message_utf8_bytes": 1205, "cache_tensor_bytes": null, "serialized_payload_bytes": 1205, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 75303.94629202783, "hops": {"1": 22688.84387507569, "2": 15939.300833037123, "3": 20253.31387494225, "4": 16422.487708972767}, "receiver_prefill_ms": 356.90450004767627, "ttft_ms": 86.36925008613616, "decode_ms": 189.98954200651497}, "total_wall_ms": 75938.04712605197, "ttft_ms": 86.36925008613616, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 3}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "39.7", "evidence_artifact": {"ledger": {"facts_total": 64, "facts_retained": 15, "retention_rate": 0.234375, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.2, "amount": 0.16666666666666666, "status": 0.75, "place": 0.0, "identifier": 0.0, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1205}, "ledger_by_hop": {"1": {"facts_total": 64, "facts_retained": 10, "retention_rate": 0.15625, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.2, "amount": 0.16666666666666666, "status": 0.3333333333333333, "place": 0.0, "identifier": 0.0, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 669}, "2": {"facts_total": 64, "facts_retained": 15, "retention_rate": 0.234375, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.2, "amount": 0.16666666666666666, "status": 0.75, "place": 0.0, "identifier": 0.0, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1205}, "3": {"facts_total": 64, "facts_retained": 15, "retention_rate": 0.234375, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.2, "amount": 0.16666666666666666, "status": 0.75, "place": 0.0, "identifier": 0.0, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1205}, "4": {"facts_total": 64, "facts_retained": 15, "retention_rate": 0.234375, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.2, "amount": 0.16666666666666666, "status": 0.75, "place": 0.0, "identifier": 0.0, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1205}}}, "reference_key_id": "keys/synthetic/syn_03006_2000.json", "task_scores": {"correct": 0.0, "abstained": 0.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "35bddc5bea5498891a8d0cc4e6438f645d89c36a525b6935b8ddcc57f193244a", "output_hash": "138c4b663aa3391624efc0a351c82fa6339a5f7ce2617bbf767a8c33c9c6c3e7"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "ba8dd3c9aaca1c0fef800323", "attempt_id": "9711e98ae1d9", "run_group_id": "E4_core:A5|closed|qwen3b-llama3b:h4", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:11:32.643470Z", "task_family": "synthetic", "document_id": "syn_03006_2000", "document_hash": "89be5e77e0ca114963e9ffdf8dc38fe5610f13a78ac29e7a1956cd9124ce5217", "question_id": "q13_unanswerable", "split": "test", "cluster_id": "syn_03006_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "tokenizer_hashes": {"receiver": "b9d8872cf3a99da99e1fc4b0f71232a451632dc49ca4c6771549b4efa5c1c763"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 4, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1717, "query_tokens": 530, "message_tokens": 414, "message_utf8_bytes": 1205, "cache_tensor_bytes": null, "serialized_payload_bytes": 1205, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 75303.94629202783, "hops": {"1": 22688.84387507569, "2": 15939.300833037123, "3": 20253.31387494225, "4": 16422.487708972767}, "receiver_prefill_ms": 334.24512506462634, "ttft_ms": 38.01274998113513, "decode_ms": 24.05174996238202}, "total_wall_ms": 75700.86962601636, "ttft_ms": 38.01274998113513, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 64, "facts_retained": 15, "retention_rate": 0.234375, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.2, "amount": 0.16666666666666666, "status": 0.75, "place": 0.0, "identifier": 0.0, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1205}, "ledger_by_hop": {"1": {"facts_total": 64, "facts_retained": 10, "retention_rate": 0.15625, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.2, "amount": 0.16666666666666666, "status": 0.3333333333333333, "place": 0.0, "identifier": 0.0, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 669}, "2": {"facts_total": 64, "facts_retained": 15, "retention_rate": 0.234375, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.2, "amount": 0.16666666666666666, "status": 0.75, "place": 0.0, "identifier": 0.0, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1205}, "3": {"facts_total": 64, "facts_retained": 15, "retention_rate": 0.234375, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.2, "amount": 0.16666666666666666, "status": 0.75, "place": 0.0, "identifier": 0.0, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1205}, "4": {"facts_total": 64, "facts_retained": 15, "retention_rate": 0.234375, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.2, "amount": 0.16666666666666666, "status": 0.75, "place": 0.0, "identifier": 0.0, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1205}}}, "reference_key_id": "keys/synthetic/syn_03006_2000.json", "task_scores": {"correct": 1.0, "abstained": 1.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "35bddc5bea5498891a8d0cc4e6438f645d89c36a525b6935b8ddcc57f193244a", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "ac2cdaff949d16e848ac4b8a", "attempt_id": "04e827eb442c", "run_group_id": "E4_core:A5|closed|qwen3b-llama3b:h4", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:11:33.059372Z", "task_family": "synthetic", "document_id": "syn_03006_2000", "document_hash": "89be5e77e0ca114963e9ffdf8dc38fe5610f13a78ac29e7a1956cd9124ce5217", "question_id": "q14_unanswerable", "split": "test", "cluster_id": "syn_03006_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "tokenizer_hashes": {"receiver": "b9d8872cf3a99da99e1fc4b0f71232a451632dc49ca4c6771549b4efa5c1c763"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 4, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1717, "query_tokens": 532, "message_tokens": 414, "message_utf8_bytes": 1205, "cache_tensor_bytes": null, "serialized_payload_bytes": 1205, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 75303.94629202783, "hops": {"1": 22688.84387507569, "2": 15939.300833037123, "3": 20253.31387494225, "4": 16422.487708972767}, "receiver_prefill_ms": 335.34887502901256, "ttft_ms": 31.685792026109993, "decode_ms": 29.840791947208345}, "total_wall_ms": 75701.42229192425, "ttft_ms": 31.685792026109993, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 64, "facts_retained": 15, "retention_rate": 0.234375, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.2, "amount": 0.16666666666666666, "status": 0.75, "place": 0.0, "identifier": 0.0, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1205}, "ledger_by_hop": {"1": {"facts_total": 64, "facts_retained": 10, "retention_rate": 0.15625, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.2, "amount": 0.16666666666666666, "status": 0.3333333333333333, "place": 0.0, "identifier": 0.0, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 669}, "2": {"facts_total": 64, "facts_retained": 15, "retention_rate": 0.234375, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.2, "amount": 0.16666666666666666, "status": 0.75, "place": 0.0, "identifier": 0.0, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1205}, "3": {"facts_total": 64, "facts_retained": 15, "retention_rate": 0.234375, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.2, "amount": 0.16666666666666666, "status": 0.75, "place": 0.0, "identifier": 0.0, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1205}, "4": {"facts_total": 64, "facts_retained": 15, "retention_rate": 0.234375, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.2, "amount": 0.16666666666666666, "status": 0.75, "place": 0.0, "identifier": 0.0, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1205}}}, "reference_key_id": "keys/synthetic/syn_03006_2000.json", "task_scores": {"correct": 1.0, "abstained": 1.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "35bddc5bea5498891a8d0cc4e6438f645d89c36a525b6935b8ddcc57f193244a", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "6d962e4e501958e7069e5f03", "attempt_id": "73145267d62e", "run_group_id": "E4_core:A5|closed|qwen3b-llama3b:h8", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:12:49.062376Z", "task_family": "synthetic", "document_id": "syn_03006_2000", "document_hash": "89be5e77e0ca114963e9ffdf8dc38fe5610f13a78ac29e7a1956cd9124ce5217", "question_id": "q00_single_fact", "split": "test", "cluster_id": "syn_03006_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "tokenizer_hashes": {"receiver": "b9d8872cf3a99da99e1fc4b0f71232a451632dc49ca4c6771549b4efa5c1c763"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 8, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1717, "query_tokens": 518, "message_tokens": 414, "message_utf8_bytes": 1205, "cache_tensor_bytes": null, "serialized_payload_bytes": 1205, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 150886.93404209334, "hops": {"1": 22688.84387507569, "2": 15939.300833037123, "3": 20253.31387494225, "4": 16422.487708972767, "5": 20512.304083094932, "6": 16327.728666947223, "7": 20688.320417073555, "8": 18054.634582949802}, "receiver_prefill_ms": 397.85004197619855, "ttft_ms": 49.35991601087153, "decode_ms": 53.65445907227695}, "total_wall_ms": 151388.35366698913, "ttft_ms": 49.35991601087153, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 64, "facts_retained": 15, "retention_rate": 0.234375, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.2, "amount": 0.16666666666666666, "status": 0.75, "place": 0.0, "identifier": 0.0, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1205}, "ledger_by_hop": {"1": {"facts_total": 64, "facts_retained": 10, "retention_rate": 0.15625, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.2, "amount": 0.16666666666666666, "status": 0.3333333333333333, "place": 0.0, "identifier": 0.0, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 669}, "2": {"facts_total": 64, "facts_retained": 15, "retention_rate": 0.234375, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.2, "amount": 0.16666666666666666, "status": 0.75, "place": 0.0, "identifier": 0.0, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1205}, "3": {"facts_total": 64, "facts_retained": 15, "retention_rate": 0.234375, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.2, "amount": 0.16666666666666666, "status": 0.75, "place": 0.0, "identifier": 0.0, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1205}, "4": {"facts_total": 64, "facts_retained": 15, "retention_rate": 0.234375, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.2, "amount": 0.16666666666666666, "status": 0.75, "place": 0.0, "identifier": 0.0, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1205}, "5": {"facts_total": 64, "facts_retained": 15, "retention_rate": 0.234375, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.2, "amount": 0.16666666666666666, "status": 0.75, "place": 0.0, "identifier": 0.0, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1205}, "6": {"facts_total": 64, "facts_retained": 15, "retention_rate": 0.234375, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.2, "amount": 0.16666666666666666, "status": 0.75, "place": 0.0, "identifier": 0.0, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1205}, "7": {"facts_total": 64, "facts_retained": 15, "retention_rate": 0.234375, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.2, "amount": 0.16666666666666666, "status": 0.75, "place": 0.0, "identifier": 0.0, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1205}, "8": {"facts_total": 64, "facts_retained": 15, "retention_rate": 0.234375, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.2, "amount": 0.16666666666666666, "status": 0.75, "place": 0.0, "identifier": 0.0, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1205}}}, "reference_key_id": "keys/synthetic/syn_03006_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "35bddc5bea5498891a8d0cc4e6438f645d89c36a525b6935b8ddcc57f193244a", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "73e8f4a7c1624df56ff13af7", "attempt_id": "89bd10961ee9", "run_group_id": "E4_core:A5|closed|qwen3b-llama3b:h8", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:12:49.580893Z", "task_family": "synthetic", "document_id": "syn_03006_2000", "document_hash": "89be5e77e0ca114963e9ffdf8dc38fe5610f13a78ac29e7a1956cd9124ce5217", "question_id": "q01_single_fact", "split": "test", "cluster_id": "syn_03006_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "tokenizer_hashes": {"receiver": "b9d8872cf3a99da99e1fc4b0f71232a451632dc49ca4c6771549b4efa5c1c763"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 8, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1717, "query_tokens": 524, "message_tokens": 414, "message_utf8_bytes": 1205, "cache_tensor_bytes": null, "serialized_payload_bytes": 1205, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 150886.93404209334, "hops": {"1": 22688.84387507569, "2": 15939.300833037123, "3": 20253.31387494225, "4": 16422.487708972767, "5": 20512.304083094932, "6": 16327.728666947223, "7": 20688.320417073555, "8": 18054.634582949802}, "receiver_prefill_ms": 400.0657079741359, "ttft_ms": 47.25329205393791, "decode_ms": 49.26300002261996}, "total_wall_ms": 151384.0618750546, "ttft_ms": 47.25329205393791, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 64, "facts_retained": 15, "retention_rate": 0.234375, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.2, "amount": 0.16666666666666666, "status": 0.75, "place": 0.0, "identifier": 0.0, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1205}, "ledger_by_hop": {"1": {"facts_total": 64, "facts_retained": 10, "retention_rate": 0.15625, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.2, "amount": 0.16666666666666666, "status": 0.3333333333333333, "place": 0.0, "identifier": 0.0, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 669}, "2": {"facts_total": 64, "facts_retained": 15, "retention_rate": 0.234375, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.2, "amount": 0.16666666666666666, "status": 0.75, "place": 0.0, "identifier": 0.0, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1205}, "3": {"facts_total": 64, "facts_retained": 15, "retention_rate": 0.234375, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.2, "amount": 0.16666666666666666, "status": 0.75, "place": 0.0, "identifier": 0.0, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1205}, "4": {"facts_total": 64, "facts_retained": 15, "retention_rate": 0.234375, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.2, "amount": 0.16666666666666666, "status": 0.75, "place": 0.0, "identifier": 0.0, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1205}, "5": {"facts_total": 64, "facts_retained": 15, "retention_rate": 0.234375, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.2, "amount": 0.16666666666666666, "status": 0.75, "place": 0.0, "identifier": 0.0, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1205}, "6": {"facts_total": 64, "facts_retained": 15, "retention_rate": 0.234375, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.2, "amount": 0.16666666666666666, "status": 0.75, "place": 0.0, "identifier": 0.0, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1205}, "7": {"facts_total": 64, "facts_retained": 15, "retention_rate": 0.234375, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.2, "amount": 0.16666666666666666, "status": 0.75, "place": 0.0, "identifier": 0.0, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1205}, "8": {"facts_total": 64, "facts_retained": 15, "retention_rate": 0.234375, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.2, "amount": 0.16666666666666666, "status": 0.75, "place": 0.0, "identifier": 0.0, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1205}}}, "reference_key_id": "keys/synthetic/syn_03006_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "35bddc5bea5498891a8d0cc4e6438f645d89c36a525b6935b8ddcc57f193244a", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "d9111956823bcdb828b09d16", "attempt_id": "da7e7dca0b91", "run_group_id": "E4_core:A5|closed|qwen3b-llama3b:h8", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:12:50.095172Z", "task_family": "synthetic", "document_id": "syn_03006_2000", "document_hash": "89be5e77e0ca114963e9ffdf8dc38fe5610f13a78ac29e7a1956cd9124ce5217", "question_id": "q02_exact_copy", "split": "test", "cluster_id": "syn_03006_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "tokenizer_hashes": {"receiver": "b9d8872cf3a99da99e1fc4b0f71232a451632dc49ca4c6771549b4efa5c1c763"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 8, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1717, "query_tokens": 523, "message_tokens": 414, "message_utf8_bytes": 1205, "cache_tensor_bytes": null, "serialized_payload_bytes": 1205, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 150886.93404209334, "hops": {"1": 22688.84387507569, "2": 15939.300833037123, "3": 20253.31387494225, "4": 16422.487708972767, "5": 20512.304083094932, "6": 16327.728666947223, "7": 20688.320417073555, "8": 18054.634582949802}, "receiver_prefill_ms": 396.50320797227323, "ttft_ms": 55.900582927279174, "decode_ms": 89.44466698449105}, "total_wall_ms": 151429.3917500181, "ttft_ms": 55.900582927279174, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 64, "facts_retained": 15, "retention_rate": 0.234375, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.2, "amount": 0.16666666666666666, "status": 0.75, "place": 0.0, "identifier": 0.0, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1205}, "ledger_by_hop": {"1": {"facts_total": 64, "facts_retained": 10, "retention_rate": 0.15625, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.2, "amount": 0.16666666666666666, "status": 0.3333333333333333, "place": 0.0, "identifier": 0.0, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 669}, "2": {"facts_total": 64, "facts_retained": 15, "retention_rate": 0.234375, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.2, "amount": 0.16666666666666666, "status": 0.75, "place": 0.0, "identifier": 0.0, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1205}, "3": {"facts_total": 64, "facts_retained": 15, "retention_rate": 0.234375, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.2, "amount": 0.16666666666666666, "status": 0.75, "place": 0.0, "identifier": 0.0, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1205}, "4": {"facts_total": 64, "facts_retained": 15, "retention_rate": 0.234375, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.2, "amount": 0.16666666666666666, "status": 0.75, "place": 0.0, "identifier": 0.0, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1205}, "5": {"facts_total": 64, "facts_retained": 15, "retention_rate": 0.234375, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.2, "amount": 0.16666666666666666, "status": 0.75, "place": 0.0, "identifier": 0.0, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1205}, "6": {"facts_total": 64, "facts_retained": 15, "retention_rate": 0.234375, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.2, "amount": 0.16666666666666666, "status": 0.75, "place": 0.0, "identifier": 0.0, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1205}, "7": {"facts_total": 64, "facts_retained": 15, "retention_rate": 0.234375, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.2, "amount": 0.16666666666666666, "status": 0.75, "place": 0.0, "identifier": 0.0, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1205}, "8": {"facts_total": 64, "facts_retained": 15, "retention_rate": 0.234375, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.2, "amount": 0.16666666666666666, "status": 0.75, "place": 0.0, "identifier": 0.0, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1205}}}, "reference_key_id": "keys/synthetic/syn_03006_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "35bddc5bea5498891a8d0cc4e6438f645d89c36a525b6935b8ddcc57f193244a", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "1bd1b066b1e59830efd74d59", "attempt_id": "29d33d97f078", "run_group_id": "E4_core:A5|closed|qwen3b-llama3b:h8", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:12:50.654880Z", "task_family": "synthetic", "document_id": "syn_03006_2000", "document_hash": "89be5e77e0ca114963e9ffdf8dc38fe5610f13a78ac29e7a1956cd9124ce5217", "question_id": "q03_exact_copy", "split": "test", "cluster_id": "syn_03006_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "tokenizer_hashes": {"receiver": "b9d8872cf3a99da99e1fc4b0f71232a451632dc49ca4c6771549b4efa5c1c763"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 8, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1717, "query_tokens": 518, "message_tokens": 414, "message_utf8_bytes": 1205, "cache_tensor_bytes": null, "serialized_payload_bytes": 1205, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 150886.93404209334, "hops": {"1": 22688.84387507569, "2": 15939.300833037123, "3": 20253.31387494225, "4": 16422.487708972767, "5": 20512.304083094932, "6": 16327.728666947223, "7": 20688.320417073555, "8": 18054.634582949802}, "receiver_prefill_ms": 332.699624937959, "ttft_ms": 36.35220800060779, "decode_ms": 40.989625034853816}, "total_wall_ms": 151297.56300104782, "ttft_ms": 36.35220800060779, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 64, "facts_retained": 15, "retention_rate": 0.234375, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.2, "amount": 0.16666666666666666, "status": 0.75, "place": 0.0, "identifier": 0.0, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1205}, "ledger_by_hop": {"1": {"facts_total": 64, "facts_retained": 10, "retention_rate": 0.15625, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.2, "amount": 0.16666666666666666, "status": 0.3333333333333333, "place": 0.0, "identifier": 0.0, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 669}, "2": {"facts_total": 64, "facts_retained": 15, "retention_rate": 0.234375, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.2, "amount": 0.16666666666666666, "status": 0.75, "place": 0.0, "identifier": 0.0, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1205}, "3": {"facts_total": 64, "facts_retained": 15, "retention_rate": 0.234375, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.2, "amount": 0.16666666666666666, "status": 0.75, "place": 0.0, "identifier": 0.0, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1205}, "4": {"facts_total": 64, "facts_retained": 15, "retention_rate": 0.234375, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.2, "amount": 0.16666666666666666, "status": 0.75, "place": 0.0, "identifier": 0.0, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1205}, "5": {"facts_total": 64, "facts_retained": 15, "retention_rate": 0.234375, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.2, "amount": 0.16666666666666666, "status": 0.75, "place": 0.0, "identifier": 0.0, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1205}, "6": {"facts_total": 64, "facts_retained": 15, "retention_rate": 0.234375, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.2, "amount": 0.16666666666666666, "status": 0.75, "place": 0.0, "identifier": 0.0, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1205}, "7": {"facts_total": 64, "facts_retained": 15, "retention_rate": 0.234375, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.2, "amount": 0.16666666666666666, "status": 0.75, "place": 0.0, "identifier": 0.0, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1205}, "8": {"facts_total": 64, "facts_retained": 15, "retention_rate": 0.234375, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.2, "amount": 0.16666666666666666, "status": 0.75, "place": 0.0, "identifier": 0.0, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1205}}}, "reference_key_id": "keys/synthetic/syn_03006_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "35bddc5bea5498891a8d0cc4e6438f645d89c36a525b6935b8ddcc57f193244a", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "5453e040d5f0d26c6f50c994", "attempt_id": "51667df520ee", "run_group_id": "E4_core:A5|closed|qwen3b-llama3b:h8", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:12:51.085374Z", "task_family": "synthetic", "document_id": "syn_03006_2000", "document_hash": "89be5e77e0ca114963e9ffdf8dc38fe5610f13a78ac29e7a1956cd9124ce5217", "question_id": "q04_multi_hop", "split": "test", "cluster_id": "syn_03006_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "tokenizer_hashes": {"receiver": "b9d8872cf3a99da99e1fc4b0f71232a451632dc49ca4c6771549b4efa5c1c763"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 8, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1717, "query_tokens": 527, "message_tokens": 414, "message_utf8_bytes": 1205, "cache_tensor_bytes": null, "serialized_payload_bytes": 1205, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 150886.93404209334, "hops": {"1": 22688.84387507569, "2": 15939.300833037123, "3": 20253.31387494225, "4": 16422.487708972767, "5": 20512.304083094932, "6": 16327.728666947223, "7": 20688.320417073555, "8": 18054.634582949802}, "receiver_prefill_ms": 477.72287495899945, "ttft_ms": 19.96045804116875, "decode_ms": 35.697583109140396}, "total_wall_ms": 151420.92858301476, "ttft_ms": 19.96045804116875, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 64, "facts_retained": 15, "retention_rate": 0.234375, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.2, "amount": 0.16666666666666666, "status": 0.75, "place": 0.0, "identifier": 0.0, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1205}, "ledger_by_hop": {"1": {"facts_total": 64, "facts_retained": 10, "retention_rate": 0.15625, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.2, "amount": 0.16666666666666666, "status": 0.3333333333333333, "place": 0.0, "identifier": 0.0, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 669}, "2": {"facts_total": 64, "facts_retained": 15, "retention_rate": 0.234375, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.2, "amount": 0.16666666666666666, "status": 0.75, "place": 0.0, "identifier": 0.0, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1205}, "3": {"facts_total": 64, "facts_retained": 15, "retention_rate": 0.234375, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.2, "amount": 0.16666666666666666, "status": 0.75, "place": 0.0, "identifier": 0.0, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1205}, "4": {"facts_total": 64, "facts_retained": 15, "retention_rate": 0.234375, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.2, "amount": 0.16666666666666666, "status": 0.75, "place": 0.0, "identifier": 0.0, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1205}, "5": {"facts_total": 64, "facts_retained": 15, "retention_rate": 0.234375, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.2, "amount": 0.16666666666666666, "status": 0.75, "place": 0.0, "identifier": 0.0, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1205}, "6": {"facts_total": 64, "facts_retained": 15, "retention_rate": 0.234375, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.2, "amount": 0.16666666666666666, "status": 0.75, "place": 0.0, "identifier": 0.0, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1205}, "7": {"facts_total": 64, "facts_retained": 15, "retention_rate": 0.234375, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.2, "amount": 0.16666666666666666, "status": 0.75, "place": 0.0, "identifier": 0.0, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1205}, "8": {"facts_total": 64, "facts_retained": 15, "retention_rate": 0.234375, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.2, "amount": 0.16666666666666666, "status": 0.75, "place": 0.0, "identifier": 0.0, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1205}}}, "reference_key_id": "keys/synthetic/syn_03006_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "35bddc5bea5498891a8d0cc4e6438f645d89c36a525b6935b8ddcc57f193244a", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "80b464079e6106ba031817fc", "attempt_id": "333541a0829e", "run_group_id": "E4_core:A5|closed|qwen3b-llama3b:h8", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:12:51.637881Z", "task_family": "synthetic", "document_id": "syn_03006_2000", "document_hash": "89be5e77e0ca114963e9ffdf8dc38fe5610f13a78ac29e7a1956cd9124ce5217", "question_id": "q05_multi_hop", "split": "test", "cluster_id": "syn_03006_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "tokenizer_hashes": {"receiver": "b9d8872cf3a99da99e1fc4b0f71232a451632dc49ca4c6771549b4efa5c1c763"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 8, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1717, "query_tokens": 528, "message_tokens": 414, "message_utf8_bytes": 1205, "cache_tensor_bytes": null, "serialized_payload_bytes": 1205, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 150886.93404209334, "hops": {"1": 22688.84387507569, "2": 15939.300833037123, "3": 20253.31387494225, "4": 16422.487708972767, "5": 20512.304083094932, "6": 16327.728666947223, "7": 20688.320417073555, "8": 18054.634582949802}, "receiver_prefill_ms": 349.48737488593906, "ttft_ms": 27.155290939845145, "decode_ms": 35.955792060121894}, "total_wall_ms": 151300.0459591858, "ttft_ms": 27.155290939845145, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 64, "facts_retained": 15, "retention_rate": 0.234375, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.2, "amount": 0.16666666666666666, "status": 0.75, "place": 0.0, "identifier": 0.0, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1205}, "ledger_by_hop": {"1": {"facts_total": 64, "facts_retained": 10, "retention_rate": 0.15625, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.2, "amount": 0.16666666666666666, "status": 0.3333333333333333, "place": 0.0, "identifier": 0.0, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 669}, "2": {"facts_total": 64, "facts_retained": 15, "retention_rate": 0.234375, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.2, "amount": 0.16666666666666666, "status": 0.75, "place": 0.0, "identifier": 0.0, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1205}, "3": {"facts_total": 64, "facts_retained": 15, "retention_rate": 0.234375, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.2, "amount": 0.16666666666666666, "status": 0.75, "place": 0.0, "identifier": 0.0, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1205}, "4": {"facts_total": 64, "facts_retained": 15, "retention_rate": 0.234375, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.2, "amount": 0.16666666666666666, "status": 0.75, "place": 0.0, "identifier": 0.0, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1205}, "5": {"facts_total": 64, "facts_retained": 15, "retention_rate": 0.234375, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.2, "amount": 0.16666666666666666, "status": 0.75, "place": 0.0, "identifier": 0.0, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1205}, "6": {"facts_total": 64, "facts_retained": 15, "retention_rate": 0.234375, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.2, "amount": 0.16666666666666666, "status": 0.75, "place": 0.0, "identifier": 0.0, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1205}, "7": {"facts_total": 64, "facts_retained": 15, "retention_rate": 0.234375, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.2, "amount": 0.16666666666666666, "status": 0.75, "place": 0.0, "identifier": 0.0, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1205}, "8": {"facts_total": 64, "facts_retained": 15, "retention_rate": 0.234375, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.2, "amount": 0.16666666666666666, "status": 0.75, "place": 0.0, "identifier": 0.0, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1205}}}, "reference_key_id": "keys/synthetic/syn_03006_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "35bddc5bea5498891a8d0cc4e6438f645d89c36a525b6935b8ddcc57f193244a", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "c3878d28c34de9aa65486cfc", "attempt_id": "28b5893bb8c6", "run_group_id": "E4_core:A5|closed|qwen3b-llama3b:h8", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:12:52.068496Z", "task_family": "synthetic", "document_id": "syn_03006_2000", "document_hash": "89be5e77e0ca114963e9ffdf8dc38fe5610f13a78ac29e7a1956cd9124ce5217", "question_id": "q06_negation", "split": "test", "cluster_id": "syn_03006_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "tokenizer_hashes": {"receiver": "b9d8872cf3a99da99e1fc4b0f71232a451632dc49ca4c6771549b4efa5c1c763"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 8, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1717, "query_tokens": 518, "message_tokens": 414, "message_utf8_bytes": 1205, "cache_tensor_bytes": null, "serialized_payload_bytes": 1205, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 150886.93404209334, "hops": {"1": 22688.84387507569, "2": 15939.300833037123, "3": 20253.31387494225, "4": 16422.487708972767, "5": 20512.304083094932, "6": 16327.728666947223, "7": 20688.320417073555, "8": 18054.634582949802}, "receiver_prefill_ms": 339.2327920300886, "ttft_ms": 18.978249980136752, "decode_ms": 18.164874985814095}, "total_wall_ms": 151263.81050015334, "ttft_ms": 18.978249980136752, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 64, "facts_retained": 15, "retention_rate": 0.234375, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.2, "amount": 0.16666666666666666, "status": 0.75, "place": 0.0, "identifier": 0.0, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1205}, "ledger_by_hop": {"1": {"facts_total": 64, "facts_retained": 10, "retention_rate": 0.15625, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.2, "amount": 0.16666666666666666, "status": 0.3333333333333333, "place": 0.0, "identifier": 0.0, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 669}, "2": {"facts_total": 64, "facts_retained": 15, "retention_rate": 0.234375, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.2, "amount": 0.16666666666666666, "status": 0.75, "place": 0.0, "identifier": 0.0, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1205}, "3": {"facts_total": 64, "facts_retained": 15, "retention_rate": 0.234375, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.2, "amount": 0.16666666666666666, "status": 0.75, "place": 0.0, "identifier": 0.0, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1205}, "4": {"facts_total": 64, "facts_retained": 15, "retention_rate": 0.234375, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.2, "amount": 0.16666666666666666, "status": 0.75, "place": 0.0, "identifier": 0.0, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1205}, "5": {"facts_total": 64, "facts_retained": 15, "retention_rate": 0.234375, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.2, "amount": 0.16666666666666666, "status": 0.75, "place": 0.0, "identifier": 0.0, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1205}, "6": {"facts_total": 64, "facts_retained": 15, "retention_rate": 0.234375, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.2, "amount": 0.16666666666666666, "status": 0.75, "place": 0.0, "identifier": 0.0, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1205}, "7": {"facts_total": 64, "facts_retained": 15, "retention_rate": 0.234375, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.2, "amount": 0.16666666666666666, "status": 0.75, "place": 0.0, "identifier": 0.0, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1205}, "8": {"facts_total": 64, "facts_retained": 15, "retention_rate": 0.234375, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.2, "amount": 0.16666666666666666, "status": 0.75, "place": 0.0, "identifier": 0.0, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1205}}}, "reference_key_id": "keys/synthetic/syn_03006_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "35bddc5bea5498891a8d0cc4e6438f645d89c36a525b6935b8ddcc57f193244a", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "90fb50251977754c87af3479", "attempt_id": "cc4077546b4e", "run_group_id": "E4_core:A5|closed|qwen3b-llama3b:h8", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:12:52.462160Z", "task_family": "synthetic", "document_id": "syn_03006_2000", "document_hash": "89be5e77e0ca114963e9ffdf8dc38fe5610f13a78ac29e7a1956cd9124ce5217", "question_id": "q07_negation", "split": "test", "cluster_id": "syn_03006_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "tokenizer_hashes": {"receiver": "b9d8872cf3a99da99e1fc4b0f71232a451632dc49ca4c6771549b4efa5c1c763"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 8, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1717, "query_tokens": 518, "message_tokens": 414, "message_utf8_bytes": 1205, "cache_tensor_bytes": null, "serialized_payload_bytes": 1205, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 150886.93404209334, "hops": {"1": 22688.84387507569, "2": 15939.300833037123, "3": 20253.31387494225, "4": 16422.487708972767, "5": 20512.304083094932, "6": 16327.728666947223, "7": 20688.320417073555, "8": 18054.634582949802}, "receiver_prefill_ms": 331.8964580539614, "ttft_ms": 42.99200000241399, "decode_ms": 73.84412502869964}, "total_wall_ms": 151336.31116710603, "ttft_ms": 42.99200000241399, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 64, "facts_retained": 15, "retention_rate": 0.234375, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.2, "amount": 0.16666666666666666, "status": 0.75, "place": 0.0, "identifier": 0.0, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1205}, "ledger_by_hop": {"1": {"facts_total": 64, "facts_retained": 10, "retention_rate": 0.15625, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.2, "amount": 0.16666666666666666, "status": 0.3333333333333333, "place": 0.0, "identifier": 0.0, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 669}, "2": {"facts_total": 64, "facts_retained": 15, "retention_rate": 0.234375, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.2, "amount": 0.16666666666666666, "status": 0.75, "place": 0.0, "identifier": 0.0, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1205}, "3": {"facts_total": 64, "facts_retained": 15, "retention_rate": 0.234375, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.2, "amount": 0.16666666666666666, "status": 0.75, "place": 0.0, "identifier": 0.0, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1205}, "4": {"facts_total": 64, "facts_retained": 15, "retention_rate": 0.234375, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.2, "amount": 0.16666666666666666, "status": 0.75, "place": 0.0, "identifier": 0.0, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1205}, "5": {"facts_total": 64, "facts_retained": 15, "retention_rate": 0.234375, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.2, "amount": 0.16666666666666666, "status": 0.75, "place": 0.0, "identifier": 0.0, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1205}, "6": {"facts_total": 64, "facts_retained": 15, "retention_rate": 0.234375, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.2, "amount": 0.16666666666666666, "status": 0.75, "place": 0.0, "identifier": 0.0, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1205}, "7": {"facts_total": 64, "facts_retained": 15, "retention_rate": 0.234375, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.2, "amount": 0.16666666666666666, "status": 0.75, "place": 0.0, "identifier": 0.0, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1205}, "8": {"facts_total": 64, "facts_retained": 15, "retention_rate": 0.234375, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.2, "amount": 0.16666666666666666, "status": 0.75, "place": 0.0, "identifier": 0.0, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1205}}}, "reference_key_id": "keys/synthetic/syn_03006_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "35bddc5bea5498891a8d0cc4e6438f645d89c36a525b6935b8ddcc57f193244a", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "f99e18df8018c57dd1067223", "attempt_id": "c03b1ffa1d1f", "run_group_id": "E4_core:A5|closed|qwen3b-llama3b:h8", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:12:52.930695Z", "task_family": "synthetic", "document_id": "syn_03006_2000", "document_hash": "89be5e77e0ca114963e9ffdf8dc38fe5610f13a78ac29e7a1956cd9124ce5217", "question_id": "q08_revision", "split": "test", "cluster_id": "syn_03006_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "tokenizer_hashes": {"receiver": "b9d8872cf3a99da99e1fc4b0f71232a451632dc49ca4c6771549b4efa5c1c763"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 8, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1717, "query_tokens": 534, "message_tokens": 414, "message_utf8_bytes": 1205, "cache_tensor_bytes": null, "serialized_payload_bytes": 1205, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 150886.93404209334, "hops": {"1": 22688.84387507569, "2": 15939.300833037123, "3": 20253.31387494225, "4": 16422.487708972767, "5": 20512.304083094932, "6": 16327.728666947223, "7": 20688.320417073555, "8": 18054.634582949802}, "receiver_prefill_ms": 466.11675003077835, "ttft_ms": 32.30470803100616, "decode_ms": 144.15070798713714}, "total_wall_ms": 151530.2005840931, "ttft_ms": 32.30470803100616, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 5}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "10,107,545", "evidence_artifact": {"ledger": {"facts_total": 64, "facts_retained": 15, "retention_rate": 0.234375, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.2, "amount": 0.16666666666666666, "status": 0.75, "place": 0.0, "identifier": 0.0, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1205}, "ledger_by_hop": {"1": {"facts_total": 64, "facts_retained": 10, "retention_rate": 0.15625, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.2, "amount": 0.16666666666666666, "status": 0.3333333333333333, "place": 0.0, "identifier": 0.0, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 669}, "2": {"facts_total": 64, "facts_retained": 15, "retention_rate": 0.234375, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.2, "amount": 0.16666666666666666, "status": 0.75, "place": 0.0, "identifier": 0.0, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1205}, "3": {"facts_total": 64, "facts_retained": 15, "retention_rate": 0.234375, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.2, "amount": 0.16666666666666666, "status": 0.75, "place": 0.0, "identifier": 0.0, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1205}, "4": {"facts_total": 64, "facts_retained": 15, "retention_rate": 0.234375, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.2, "amount": 0.16666666666666666, "status": 0.75, "place": 0.0, "identifier": 0.0, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1205}, "5": {"facts_total": 64, "facts_retained": 15, "retention_rate": 0.234375, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.2, "amount": 0.16666666666666666, "status": 0.75, "place": 0.0, "identifier": 0.0, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1205}, "6": {"facts_total": 64, "facts_retained": 15, "retention_rate": 0.234375, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.2, "amount": 0.16666666666666666, "status": 0.75, "place": 0.0, "identifier": 0.0, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1205}, "7": {"facts_total": 64, "facts_retained": 15, "retention_rate": 0.234375, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.2, "amount": 0.16666666666666666, "status": 0.75, "place": 0.0, "identifier": 0.0, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1205}, "8": {"facts_total": 64, "facts_retained": 15, "retention_rate": 0.234375, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.2, "amount": 0.16666666666666666, "status": 0.75, "place": 0.0, "identifier": 0.0, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1205}}}, "reference_key_id": "keys/synthetic/syn_03006_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "35bddc5bea5498891a8d0cc4e6438f645d89c36a525b6935b8ddcc57f193244a", "output_hash": "3d6a277763361ef55b8dd3f9791012605d24cd167ac5bdcdd0064ee16de161d4"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "97903051bcab9a17cd55df21", "attempt_id": "2ee6ec6859c1", "run_group_id": "E4_core:A5|closed|qwen3b-llama3b:h8", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:12:53.592911Z", "task_family": "synthetic", "document_id": "syn_03006_2000", "document_hash": "89be5e77e0ca114963e9ffdf8dc38fe5610f13a78ac29e7a1956cd9124ce5217", "question_id": "q09_revision", "split": "test", "cluster_id": "syn_03006_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "tokenizer_hashes": {"receiver": "b9d8872cf3a99da99e1fc4b0f71232a451632dc49ca4c6771549b4efa5c1c763"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 8, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1717, "query_tokens": 527, "message_tokens": 414, "message_utf8_bytes": 1205, "cache_tensor_bytes": null, "serialized_payload_bytes": 1205, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 150886.93404209334, "hops": {"1": 22688.84387507569, "2": 15939.300833037123, "3": 20253.31387494225, "4": 16422.487708972767, "5": 20512.304083094932, "6": 16327.728666947223, "7": 20688.320417073555, "8": 18054.634582949802}, "receiver_prefill_ms": 335.12966707348824, "ttft_ms": 38.32575003616512, "decode_ms": 143.90166709199548}, "total_wall_ms": 151404.9181251321, "ttft_ms": 38.32575003616512, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 5}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "10,107,545", "evidence_artifact": {"ledger": {"facts_total": 64, "facts_retained": 15, "retention_rate": 0.234375, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.2, "amount": 0.16666666666666666, "status": 0.75, "place": 0.0, "identifier": 0.0, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1205}, "ledger_by_hop": {"1": {"facts_total": 64, "facts_retained": 10, "retention_rate": 0.15625, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.2, "amount": 0.16666666666666666, "status": 0.3333333333333333, "place": 0.0, "identifier": 0.0, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 669}, "2": {"facts_total": 64, "facts_retained": 15, "retention_rate": 0.234375, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.2, "amount": 0.16666666666666666, "status": 0.75, "place": 0.0, "identifier": 0.0, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1205}, "3": {"facts_total": 64, "facts_retained": 15, "retention_rate": 0.234375, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.2, "amount": 0.16666666666666666, "status": 0.75, "place": 0.0, "identifier": 0.0, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1205}, "4": {"facts_total": 64, "facts_retained": 15, "retention_rate": 0.234375, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.2, "amount": 0.16666666666666666, "status": 0.75, "place": 0.0, "identifier": 0.0, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1205}, "5": {"facts_total": 64, "facts_retained": 15, "retention_rate": 0.234375, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.2, "amount": 0.16666666666666666, "status": 0.75, "place": 0.0, "identifier": 0.0, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1205}, "6": {"facts_total": 64, "facts_retained": 15, "retention_rate": 0.234375, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.2, "amount": 0.16666666666666666, "status": 0.75, "place": 0.0, "identifier": 0.0, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1205}, "7": {"facts_total": 64, "facts_retained": 15, "retention_rate": 0.234375, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.2, "amount": 0.16666666666666666, "status": 0.75, "place": 0.0, "identifier": 0.0, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1205}, "8": {"facts_total": 64, "facts_retained": 15, "retention_rate": 0.234375, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.2, "amount": 0.16666666666666666, "status": 0.75, "place": 0.0, "identifier": 0.0, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1205}}}, "reference_key_id": "keys/synthetic/syn_03006_2000.json", "task_scores": {"correct": 0.0, "abstained": 0.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "35bddc5bea5498891a8d0cc4e6438f645d89c36a525b6935b8ddcc57f193244a", "output_hash": "3d6a277763361ef55b8dd3f9791012605d24cd167ac5bdcdd0064ee16de161d4"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "25275b9661f5eed970ef2ed6", "attempt_id": "7de106db924e", "run_group_id": "E4_core:A5|closed|qwen3b-llama3b:h8", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:12:54.128125Z", "task_family": "synthetic", "document_id": "syn_03006_2000", "document_hash": "89be5e77e0ca114963e9ffdf8dc38fe5610f13a78ac29e7a1956cd9124ce5217", "question_id": "q10_temporal", "split": "test", "cluster_id": "syn_03006_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "tokenizer_hashes": {"receiver": "b9d8872cf3a99da99e1fc4b0f71232a451632dc49ca4c6771549b4efa5c1c763"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 8, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1717, "query_tokens": 536, "message_tokens": 414, "message_utf8_bytes": 1205, "cache_tensor_bytes": null, "serialized_payload_bytes": 1205, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 150886.93404209334, "hops": {"1": 22688.84387507569, "2": 15939.300833037123, "3": 20253.31387494225, "4": 16422.487708972767, "5": 20512.304083094932, "6": 16327.728666947223, "7": 20688.320417073555, "8": 18054.634582949802}, "receiver_prefill_ms": 344.7957909666002, "ttft_ms": 70.21904201246798, "decode_ms": 48.05341700557619}, "total_wall_ms": 151350.6924591493, "ttft_ms": 70.21904201246798, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 64, "facts_retained": 15, "retention_rate": 0.234375, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.2, "amount": 0.16666666666666666, "status": 0.75, "place": 0.0, "identifier": 0.0, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1205}, "ledger_by_hop": {"1": {"facts_total": 64, "facts_retained": 10, "retention_rate": 0.15625, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.2, "amount": 0.16666666666666666, "status": 0.3333333333333333, "place": 0.0, "identifier": 0.0, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 669}, "2": {"facts_total": 64, "facts_retained": 15, "retention_rate": 0.234375, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.2, "amount": 0.16666666666666666, "status": 0.75, "place": 0.0, "identifier": 0.0, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1205}, "3": {"facts_total": 64, "facts_retained": 15, "retention_rate": 0.234375, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.2, "amount": 0.16666666666666666, "status": 0.75, "place": 0.0, "identifier": 0.0, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1205}, "4": {"facts_total": 64, "facts_retained": 15, "retention_rate": 0.234375, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.2, "amount": 0.16666666666666666, "status": 0.75, "place": 0.0, "identifier": 0.0, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1205}, "5": {"facts_total": 64, "facts_retained": 15, "retention_rate": 0.234375, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.2, "amount": 0.16666666666666666, "status": 0.75, "place": 0.0, "identifier": 0.0, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1205}, "6": {"facts_total": 64, "facts_retained": 15, "retention_rate": 0.234375, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.2, "amount": 0.16666666666666666, "status": 0.75, "place": 0.0, "identifier": 0.0, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1205}, "7": {"facts_total": 64, "facts_retained": 15, "retention_rate": 0.234375, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.2, "amount": 0.16666666666666666, "status": 0.75, "place": 0.0, "identifier": 0.0, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1205}, "8": {"facts_total": 64, "facts_retained": 15, "retention_rate": 0.234375, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.2, "amount": 0.16666666666666666, "status": 0.75, "place": 0.0, "identifier": 0.0, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1205}}}, "reference_key_id": "keys/synthetic/syn_03006_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "35bddc5bea5498891a8d0cc4e6438f645d89c36a525b6935b8ddcc57f193244a", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "6d23e2dcf5fe88c30a54d643", "attempt_id": "79310a435c80", "run_group_id": "E4_core:A5|closed|qwen3b-llama3b:h8", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:12:54.609879Z", "task_family": "synthetic", "document_id": "syn_03006_2000", "document_hash": "89be5e77e0ca114963e9ffdf8dc38fe5610f13a78ac29e7a1956cd9124ce5217", "question_id": "q11_aggregation", "split": "test", "cluster_id": "syn_03006_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "tokenizer_hashes": {"receiver": "b9d8872cf3a99da99e1fc4b0f71232a451632dc49ca4c6771549b4efa5c1c763"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 8, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1717, "query_tokens": 530, "message_tokens": 414, "message_utf8_bytes": 1205, "cache_tensor_bytes": null, "serialized_payload_bytes": 1205, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 150886.93404209334, "hops": {"1": 22688.84387507569, "2": 15939.300833037123, "3": 20253.31387494225, "4": 16422.487708972767, "5": 20512.304083094932, "6": 16327.728666947223, "7": 20688.320417073555, "8": 18054.634582949802}, "receiver_prefill_ms": 466.90350002609193, "ttft_ms": 36.56216699164361, "decode_ms": 27.59875007905066}, "total_wall_ms": 151418.6310841469, "ttft_ms": 36.56216699164361, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 64, "facts_retained": 15, "retention_rate": 0.234375, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.2, "amount": 0.16666666666666666, "status": 0.75, "place": 0.0, "identifier": 0.0, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1205}, "ledger_by_hop": {"1": {"facts_total": 64, "facts_retained": 10, "retention_rate": 0.15625, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.2, "amount": 0.16666666666666666, "status": 0.3333333333333333, "place": 0.0, "identifier": 0.0, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 669}, "2": {"facts_total": 64, "facts_retained": 15, "retention_rate": 0.234375, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.2, "amount": 0.16666666666666666, "status": 0.75, "place": 0.0, "identifier": 0.0, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1205}, "3": {"facts_total": 64, "facts_retained": 15, "retention_rate": 0.234375, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.2, "amount": 0.16666666666666666, "status": 0.75, "place": 0.0, "identifier": 0.0, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1205}, "4": {"facts_total": 64, "facts_retained": 15, "retention_rate": 0.234375, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.2, "amount": 0.16666666666666666, "status": 0.75, "place": 0.0, "identifier": 0.0, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1205}, "5": {"facts_total": 64, "facts_retained": 15, "retention_rate": 0.234375, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.2, "amount": 0.16666666666666666, "status": 0.75, "place": 0.0, "identifier": 0.0, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1205}, "6": {"facts_total": 64, "facts_retained": 15, "retention_rate": 0.234375, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.2, "amount": 0.16666666666666666, "status": 0.75, "place": 0.0, "identifier": 0.0, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1205}, "7": {"facts_total": 64, "facts_retained": 15, "retention_rate": 0.234375, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.2, "amount": 0.16666666666666666, "status": 0.75, "place": 0.0, "identifier": 0.0, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1205}, "8": {"facts_total": 64, "facts_retained": 15, "retention_rate": 0.234375, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.2, "amount": 0.16666666666666666, "status": 0.75, "place": 0.0, "identifier": 0.0, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1205}}}, "reference_key_id": "keys/synthetic/syn_03006_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "35bddc5bea5498891a8d0cc4e6438f645d89c36a525b6935b8ddcc57f193244a", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "c9e87c7d16b990dbef98cb15", "attempt_id": "db6bb0425212", "run_group_id": "E4_core:A5|closed|qwen3b-llama3b:h8", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:12:55.158284Z", "task_family": "synthetic", "document_id": "syn_03006_2000", "document_hash": "89be5e77e0ca114963e9ffdf8dc38fe5610f13a78ac29e7a1956cd9124ce5217", "question_id": "q12_unit_conversion", "split": "test", "cluster_id": "syn_03006_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "tokenizer_hashes": {"receiver": "b9d8872cf3a99da99e1fc4b0f71232a451632dc49ca4c6771549b4efa5c1c763"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 8, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1717, "query_tokens": 526, "message_tokens": 414, "message_utf8_bytes": 1205, "cache_tensor_bytes": null, "serialized_payload_bytes": 1205, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 150886.93404209334, "hops": {"1": 22688.84387507569, "2": 15939.300833037123, "3": 20253.31387494225, "4": 16422.487708972767, "5": 20512.304083094932, "6": 16327.728666947223, "7": 20688.320417073555, "8": 18054.634582949802}, "receiver_prefill_ms": 331.10533305443823, "ttft_ms": 27.526583056896925, "decode_ms": 103.4766249358654}, "total_wall_ms": 151349.73191702738, "ttft_ms": 27.526583056896925, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 3}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "39.7", "evidence_artifact": {"ledger": {"facts_total": 64, "facts_retained": 15, "retention_rate": 0.234375, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.2, "amount": 0.16666666666666666, "status": 0.75, "place": 0.0, "identifier": 0.0, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1205}, "ledger_by_hop": {"1": {"facts_total": 64, "facts_retained": 10, "retention_rate": 0.15625, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.2, "amount": 0.16666666666666666, "status": 0.3333333333333333, "place": 0.0, "identifier": 0.0, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 669}, "2": {"facts_total": 64, "facts_retained": 15, "retention_rate": 0.234375, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.2, "amount": 0.16666666666666666, "status": 0.75, "place": 0.0, "identifier": 0.0, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1205}, "3": {"facts_total": 64, "facts_retained": 15, "retention_rate": 0.234375, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.2, "amount": 0.16666666666666666, "status": 0.75, "place": 0.0, "identifier": 0.0, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1205}, "4": {"facts_total": 64, "facts_retained": 15, "retention_rate": 0.234375, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.2, "amount": 0.16666666666666666, "status": 0.75, "place": 0.0, "identifier": 0.0, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1205}, "5": {"facts_total": 64, "facts_retained": 15, "retention_rate": 0.234375, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.2, "amount": 0.16666666666666666, "status": 0.75, "place": 0.0, "identifier": 0.0, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1205}, "6": {"facts_total": 64, "facts_retained": 15, "retention_rate": 0.234375, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.2, "amount": 0.16666666666666666, "status": 0.75, "place": 0.0, "identifier": 0.0, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1205}, "7": {"facts_total": 64, "facts_retained": 15, "retention_rate": 0.234375, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.2, "amount": 0.16666666666666666, "status": 0.75, "place": 0.0, "identifier": 0.0, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1205}, "8": {"facts_total": 64, "facts_retained": 15, "retention_rate": 0.234375, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.2, "amount": 0.16666666666666666, "status": 0.75, "place": 0.0, "identifier": 0.0, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1205}}}, "reference_key_id": "keys/synthetic/syn_03006_2000.json", "task_scores": {"correct": 0.0, "abstained": 0.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "35bddc5bea5498891a8d0cc4e6438f645d89c36a525b6935b8ddcc57f193244a", "output_hash": "138c4b663aa3391624efc0a351c82fa6339a5f7ce2617bbf767a8c33c9c6c3e7"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "069ec7711725673cb60bca78", "attempt_id": "7c7308592108", "run_group_id": "E4_core:A5|closed|qwen3b-llama3b:h8", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:12:55.638555Z", "task_family": "synthetic", "document_id": "syn_03006_2000", "document_hash": "89be5e77e0ca114963e9ffdf8dc38fe5610f13a78ac29e7a1956cd9124ce5217", "question_id": "q13_unanswerable", "split": "test", "cluster_id": "syn_03006_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "tokenizer_hashes": {"receiver": "b9d8872cf3a99da99e1fc4b0f71232a451632dc49ca4c6771549b4efa5c1c763"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 8, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1717, "query_tokens": 530, "message_tokens": 414, "message_utf8_bytes": 1205, "cache_tensor_bytes": null, "serialized_payload_bytes": 1205, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 150886.93404209334, "hops": {"1": 22688.84387507569, "2": 15939.300833037123, "3": 20253.31387494225, "4": 16422.487708972767, "5": 20512.304083094932, "6": 16327.728666947223, "7": 20688.320417073555, "8": 18054.634582949802}, "receiver_prefill_ms": 307.579500018619, "ttft_ms": 14.843667042441666, "decode_ms": 11.52020797599107}, "total_wall_ms": 151221.35212516878, "ttft_ms": 14.843667042441666, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 64, "facts_retained": 15, "retention_rate": 0.234375, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.2, "amount": 0.16666666666666666, "status": 0.75, "place": 0.0, "identifier": 0.0, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1205}, "ledger_by_hop": {"1": {"facts_total": 64, "facts_retained": 10, "retention_rate": 0.15625, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.2, "amount": 0.16666666666666666, "status": 0.3333333333333333, "place": 0.0, "identifier": 0.0, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 669}, "2": {"facts_total": 64, "facts_retained": 15, "retention_rate": 0.234375, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.2, "amount": 0.16666666666666666, "status": 0.75, "place": 0.0, "identifier": 0.0, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1205}, "3": {"facts_total": 64, "facts_retained": 15, "retention_rate": 0.234375, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.2, "amount": 0.16666666666666666, "status": 0.75, "place": 0.0, "identifier": 0.0, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1205}, "4": {"facts_total": 64, "facts_retained": 15, "retention_rate": 0.234375, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.2, "amount": 0.16666666666666666, "status": 0.75, "place": 0.0, "identifier": 0.0, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1205}, "5": {"facts_total": 64, "facts_retained": 15, "retention_rate": 0.234375, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.2, "amount": 0.16666666666666666, "status": 0.75, "place": 0.0, "identifier": 0.0, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1205}, "6": {"facts_total": 64, "facts_retained": 15, "retention_rate": 0.234375, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.2, "amount": 0.16666666666666666, "status": 0.75, "place": 0.0, "identifier": 0.0, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1205}, "7": {"facts_total": 64, "facts_retained": 15, "retention_rate": 0.234375, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.2, "amount": 0.16666666666666666, "status": 0.75, "place": 0.0, "identifier": 0.0, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1205}, "8": {"facts_total": 64, "facts_retained": 15, "retention_rate": 0.234375, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.2, "amount": 0.16666666666666666, "status": 0.75, "place": 0.0, "identifier": 0.0, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1205}}}, "reference_key_id": "keys/synthetic/syn_03006_2000.json", "task_scores": {"correct": 1.0, "abstained": 1.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "35bddc5bea5498891a8d0cc4e6438f645d89c36a525b6935b8ddcc57f193244a", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "2cc1463f6a534be6c98fa824", "attempt_id": "0153db801973", "run_group_id": "E4_core:A5|closed|qwen3b-llama3b:h8", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:12:55.989862Z", "task_family": "synthetic", "document_id": "syn_03006_2000", "document_hash": "89be5e77e0ca114963e9ffdf8dc38fe5610f13a78ac29e7a1956cd9124ce5217", "question_id": "q14_unanswerable", "split": "test", "cluster_id": "syn_03006_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "tokenizer_hashes": {"receiver": "b9d8872cf3a99da99e1fc4b0f71232a451632dc49ca4c6771549b4efa5c1c763"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 8, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1717, "query_tokens": 532, "message_tokens": 414, "message_utf8_bytes": 1205, "cache_tensor_bytes": null, "serialized_payload_bytes": 1205, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 150886.93404209334, "hops": {"1": 22688.84387507569, "2": 15939.300833037123, "3": 20253.31387494225, "4": 16422.487708972767, "5": 20512.304083094932, "6": 16327.728666947223, "7": 20688.320417073555, "8": 18054.634582949802}, "receiver_prefill_ms": 336.53654193039984, "ttft_ms": 61.88033404760063, "decode_ms": 57.09420796483755}, "total_wall_ms": 151343.07183302008, "ttft_ms": 61.88033404760063, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 64, "facts_retained": 15, "retention_rate": 0.234375, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.2, "amount": 0.16666666666666666, "status": 0.75, "place": 0.0, "identifier": 0.0, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1205}, "ledger_by_hop": {"1": {"facts_total": 64, "facts_retained": 10, "retention_rate": 0.15625, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.2, "amount": 0.16666666666666666, "status": 0.3333333333333333, "place": 0.0, "identifier": 0.0, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 669}, "2": {"facts_total": 64, "facts_retained": 15, "retention_rate": 0.234375, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.2, "amount": 0.16666666666666666, "status": 0.75, "place": 0.0, "identifier": 0.0, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1205}, "3": {"facts_total": 64, "facts_retained": 15, "retention_rate": 0.234375, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.2, "amount": 0.16666666666666666, "status": 0.75, "place": 0.0, "identifier": 0.0, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1205}, "4": {"facts_total": 64, "facts_retained": 15, "retention_rate": 0.234375, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.2, "amount": 0.16666666666666666, "status": 0.75, "place": 0.0, "identifier": 0.0, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1205}, "5": {"facts_total": 64, "facts_retained": 15, "retention_rate": 0.234375, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.2, "amount": 0.16666666666666666, "status": 0.75, "place": 0.0, "identifier": 0.0, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1205}, "6": {"facts_total": 64, "facts_retained": 15, "retention_rate": 0.234375, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.2, "amount": 0.16666666666666666, "status": 0.75, "place": 0.0, "identifier": 0.0, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1205}, "7": {"facts_total": 64, "facts_retained": 15, "retention_rate": 0.234375, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.2, "amount": 0.16666666666666666, "status": 0.75, "place": 0.0, "identifier": 0.0, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1205}, "8": {"facts_total": 64, "facts_retained": 15, "retention_rate": 0.234375, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.2, "amount": 0.16666666666666666, "status": 0.75, "place": 0.0, "identifier": 0.0, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1205}}}, "reference_key_id": "keys/synthetic/syn_03006_2000.json", "task_scores": {"correct": 1.0, "abstained": 1.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "35bddc5bea5498891a8d0cc4e6438f645d89c36a525b6935b8ddcc57f193244a", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "23289d620eacdb4353787624", "attempt_id": "a41c4314959e", "run_group_id": "E4_core:A4|closed|qwen3b:h1", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:13:19.311809Z", "task_family": "synthetic", "document_id": "syn_03009_2000", "document_hash": "6ed94dcf728255bd7a7d11c97c362930728fb81a3e91f1906eb6eec86e0fedb7", "question_id": "q00_single_fact", "split": "test", "cluster_id": "syn_03009_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 1, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1976, "query_tokens": 595, "message_tokens": 512, "message_utf8_bytes": 1123, "cache_tensor_bytes": null, "serialized_payload_bytes": 1123, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 22747.736499994062, "hops": {"1": 22747.736499994062}, "receiver_prefill_ms": 397.419249988161, "ttft_ms": 24.769208976067603, "decode_ms": 633.0129590351135}, "total_wall_ms": 23803.933791001327, "ttft_ms": 24.769208976067603, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 10}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "PA-37F-6227", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 26, "retention_rate": 0.4, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.75, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1123}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 26, "retention_rate": 0.4, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.75, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1123}}}, "reference_key_id": "keys/synthetic/syn_03009_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "3bd2611f7bc59d6f4b6323c3ac1e9f7af93bcc7ed3e43397b183cdb4d6c6e058", "output_hash": "5b7645cf6e8140f4c0bf6ed4ad76121cf2924fe7e666d92161812527b86f4409"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "df06e4dcc66ee2314454a01d", "attempt_id": "5f3b4af15273", "run_group_id": "E4_core:A4|closed|qwen3b:h1", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:13:20.385975Z", "task_family": "synthetic", "document_id": "syn_03009_2000", "document_hash": "6ed94dcf728255bd7a7d11c97c362930728fb81a3e91f1906eb6eec86e0fedb7", "question_id": "q01_single_fact", "split": "test", "cluster_id": "syn_03009_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 1, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1976, "query_tokens": 600, "message_tokens": 512, "message_utf8_bytes": 1123, "cache_tensor_bytes": null, "serialized_payload_bytes": 1123, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 22747.736499994062, "hops": {"1": 22747.736499994062}, "receiver_prefill_ms": 415.8240839606151, "ttft_ms": 34.97124998830259, "decode_ms": 32.09829202387482}, "total_wall_ms": 23231.283166911453, "ttft_ms": 34.97124998830259, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 26, "retention_rate": 0.4, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.75, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1123}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 26, "retention_rate": 0.4, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.75, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1123}}}, "reference_key_id": "keys/synthetic/syn_03009_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "3bd2611f7bc59d6f4b6323c3ac1e9f7af93bcc7ed3e43397b183cdb4d6c6e058", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "cc7a645a26e026d61668bafa", "attempt_id": "509f1f2d7896", "run_group_id": "E4_core:A4|closed|qwen3b:h1", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:13:20.886628Z", "task_family": "synthetic", "document_id": "syn_03009_2000", "document_hash": "6ed94dcf728255bd7a7d11c97c362930728fb81a3e91f1906eb6eec86e0fedb7", "question_id": "q02_exact_copy", "split": "test", "cluster_id": "syn_03009_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 1, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1976, "query_tokens": 598, "message_tokens": 512, "message_utf8_bytes": 1123, "cache_tensor_bytes": null, "serialized_payload_bytes": 1123, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 22747.736499994062, "hops": {"1": 22747.736499994062}, "receiver_prefill_ms": 401.99662500526756, "ttft_ms": 20.686083007603884, "decode_ms": 20.33720794133842}, "total_wall_ms": 23191.452166996896, "ttft_ms": 20.686083007603884, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 26, "retention_rate": 0.4, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.75, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1123}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 26, "retention_rate": 0.4, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.75, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1123}}}, "reference_key_id": "keys/synthetic/syn_03009_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "3bd2611f7bc59d6f4b6323c3ac1e9f7af93bcc7ed3e43397b183cdb4d6c6e058", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "deb826fc5ee971ccbbf9216c", "attempt_id": "ad9bdfcbc15a", "run_group_id": "E4_core:A4|closed|qwen3b:h1", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:13:21.347512Z", "task_family": "synthetic", "document_id": "syn_03009_2000", "document_hash": "6ed94dcf728255bd7a7d11c97c362930728fb81a3e91f1906eb6eec86e0fedb7", "question_id": "q03_exact_copy", "split": "test", "cluster_id": "syn_03009_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 1, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1976, "query_tokens": 594, "message_tokens": 512, "message_utf8_bytes": 1123, "cache_tensor_bytes": null, "serialized_payload_bytes": 1123, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 22747.736499994062, "hops": {"1": 22747.736499994062}, "receiver_prefill_ms": 353.74508309178054, "ttft_ms": 48.7525420030579, "decode_ms": 377.1110000088811}, "total_wall_ms": 23528.07208406739, "ttft_ms": 48.7525420030579, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 5}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "Idris Bergström", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 26, "retention_rate": 0.4, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.75, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1123}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 26, "retention_rate": 0.4, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.75, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1123}}}, "reference_key_id": "keys/synthetic/syn_03009_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "3bd2611f7bc59d6f4b6323c3ac1e9f7af93bcc7ed3e43397b183cdb4d6c6e058", "output_hash": "fd88832c9fd0ab175818d76d875d70156225bd4b34ec401514546ef70f4f826c"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "c9685c30ffdb0b746cebde9a", "attempt_id": "dacdb4ba4ced", "run_group_id": "E4_core:A4|closed|qwen3b:h1", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:13:22.144150Z", "task_family": "synthetic", "document_id": "syn_03009_2000", "document_hash": "6ed94dcf728255bd7a7d11c97c362930728fb81a3e91f1906eb6eec86e0fedb7", "question_id": "q04_multi_hop", "split": "test", "cluster_id": "syn_03009_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 1, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1976, "query_tokens": 602, "message_tokens": 512, "message_utf8_bytes": 1123, "cache_tensor_bytes": null, "serialized_payload_bytes": 1123, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 22747.736499994062, "hops": {"1": 22747.736499994062}, "receiver_prefill_ms": 403.8512499537319, "ttft_ms": 26.29325003363192, "decode_ms": 230.0886670127511}, "total_wall_ms": 23408.678749925457, "ttft_ms": 26.29325003363192, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 7}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "4670383", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 26, "retention_rate": 0.4, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.75, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1123}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 26, "retention_rate": 0.4, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.75, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1123}}}, "reference_key_id": "keys/synthetic/syn_03009_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "3bd2611f7bc59d6f4b6323c3ac1e9f7af93bcc7ed3e43397b183cdb4d6c6e058", "output_hash": "19ae8c357e5500bf494a6163bfbd11380b2996e8ea532a2c64f4a599464820f7"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "44c01f53df7b57a579bc363a", "attempt_id": "01bb90ae701f", "run_group_id": "E4_core:A4|closed|qwen3b:h1", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:13:22.823014Z", "task_family": "synthetic", "document_id": "syn_03009_2000", "document_hash": "6ed94dcf728255bd7a7d11c97c362930728fb81a3e91f1906eb6eec86e0fedb7", "question_id": "q05_multi_hop", "split": "test", "cluster_id": "syn_03009_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 1, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1976, "query_tokens": 605, "message_tokens": 512, "message_utf8_bytes": 1123, "cache_tensor_bytes": null, "serialized_payload_bytes": 1123, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 22747.736499994062, "hops": {"1": 22747.736499994062}, "receiver_prefill_ms": 362.73020796943456, "ttft_ms": 44.348625000566244, "decode_ms": 55.15604093670845}, "total_wall_ms": 23210.845415946096, "ttft_ms": 44.348625000566244, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 26, "retention_rate": 0.4, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.75, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1123}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 26, "retention_rate": 0.4, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.75, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1123}}}, "reference_key_id": "keys/synthetic/syn_03009_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "3bd2611f7bc59d6f4b6323c3ac1e9f7af93bcc7ed3e43397b183cdb4d6c6e058", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "76f3a49bdd6af84ae9b5834b", "attempt_id": "940f5285e48c", "run_group_id": "E4_core:A4|closed|qwen3b:h1", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:13:23.303678Z", "task_family": "synthetic", "document_id": "syn_03009_2000", "document_hash": "6ed94dcf728255bd7a7d11c97c362930728fb81a3e91f1906eb6eec86e0fedb7", "question_id": "q06_negation", "split": "test", "cluster_id": "syn_03009_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 1, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1976, "query_tokens": 595, "message_tokens": 512, "message_utf8_bytes": 1123, "cache_tensor_bytes": null, "serialized_payload_bytes": 1123, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 22747.736499994062, "hops": {"1": 22747.736499994062}, "receiver_prefill_ms": 546.7990409815684, "ttft_ms": 39.52704195398837, "decode_ms": 24.22162506263703}, "total_wall_ms": 23359.046249999665, "ttft_ms": 39.52704195398837, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 26, "retention_rate": 0.4, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.75, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1123}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 26, "retention_rate": 0.4, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.75, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1123}}}, "reference_key_id": "keys/synthetic/syn_03009_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "3bd2611f7bc59d6f4b6323c3ac1e9f7af93bcc7ed3e43397b183cdb4d6c6e058", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "d2210ec8ed3f4f497f826d98", "attempt_id": "00314e1e97b1", "run_group_id": "E4_core:A4|closed|qwen3b:h1", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:13:23.932890Z", "task_family": "synthetic", "document_id": "syn_03009_2000", "document_hash": "6ed94dcf728255bd7a7d11c97c362930728fb81a3e91f1906eb6eec86e0fedb7", "question_id": "q07_negation", "split": "test", "cluster_id": "syn_03009_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 1, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1976, "query_tokens": 594, "message_tokens": 512, "message_utf8_bytes": 1123, "cache_tensor_bytes": null, "serialized_payload_bytes": 1123, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 22747.736499994062, "hops": {"1": 22747.736499994062}, "receiver_prefill_ms": 403.84325000923127, "ttft_ms": 42.691958020441234, "decode_ms": 32.39404200576246}, "total_wall_ms": 23227.336458978243, "ttft_ms": 42.691958020441234, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 26, "retention_rate": 0.4, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.75, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1123}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 26, "retention_rate": 0.4, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.75, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1123}}}, "reference_key_id": "keys/synthetic/syn_03009_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "3bd2611f7bc59d6f4b6323c3ac1e9f7af93bcc7ed3e43397b183cdb4d6c6e058", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "1396290a9cf09a95a6fec068", "attempt_id": "9bf87e164c57", "run_group_id": "E4_core:A4|closed|qwen3b:h1", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:13:24.430494Z", "task_family": "synthetic", "document_id": "syn_03009_2000", "document_hash": "6ed94dcf728255bd7a7d11c97c362930728fb81a3e91f1906eb6eec86e0fedb7", "question_id": "q08_revision", "split": "test", "cluster_id": "syn_03009_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 1, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1976, "query_tokens": 612, "message_tokens": 512, "message_utf8_bytes": 1123, "cache_tensor_bytes": null, "serialized_payload_bytes": 1123, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 22747.736499994062, "hops": {"1": 22747.736499994062}, "receiver_prefill_ms": 373.8666659919545, "ttft_ms": 21.449125022627413, "decode_ms": 564.2770830309018}, "total_wall_ms": 23708.12816696707, "ttft_ms": 21.449125022627413, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 8}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "10047105", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 26, "retention_rate": 0.4, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.75, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1123}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 26, "retention_rate": 0.4, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.75, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1123}}}, "reference_key_id": "keys/synthetic/syn_03009_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "3bd2611f7bc59d6f4b6323c3ac1e9f7af93bcc7ed3e43397b183cdb4d6c6e058", "output_hash": "bea99610d1e3e75a7f20939faf8e6824e02237b156dacd6442ba2a7e02617510"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "3cd1d54ee14e4e08e3a24419", "attempt_id": "cf1a9d4a676f", "run_group_id": "E4_core:A4|closed|qwen3b:h1", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:13:25.407391Z", "task_family": "synthetic", "document_id": "syn_03009_2000", "document_hash": "6ed94dcf728255bd7a7d11c97c362930728fb81a3e91f1906eb6eec86e0fedb7", "question_id": "q09_revision", "split": "test", "cluster_id": "syn_03009_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 1, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1976, "query_tokens": 605, "message_tokens": 512, "message_utf8_bytes": 1123, "cache_tensor_bytes": null, "serialized_payload_bytes": 1123, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 22747.736499994062, "hops": {"1": 22747.736499994062}, "receiver_prefill_ms": 389.1618330962956, "ttft_ms": 22.21199998166412, "decode_ms": 286.9401670759544}, "total_wall_ms": 23446.94758299738, "ttft_ms": 22.21199998166412, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 8}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "10047105", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 26, "retention_rate": 0.4, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.75, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1123}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 26, "retention_rate": 0.4, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.75, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1123}}}, "reference_key_id": "keys/synthetic/syn_03009_2000.json", "task_scores": {"correct": 0.0, "abstained": 0.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "3bd2611f7bc59d6f4b6323c3ac1e9f7af93bcc7ed3e43397b183cdb4d6c6e058", "output_hash": "bea99610d1e3e75a7f20939faf8e6824e02237b156dacd6442ba2a7e02617510"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "5216a702286ab9eb9ae11692", "attempt_id": "497df42d8f21", "run_group_id": "E4_core:A4|closed|qwen3b:h1", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:13:26.124104Z", "task_family": "synthetic", "document_id": "syn_03009_2000", "document_hash": "6ed94dcf728255bd7a7d11c97c362930728fb81a3e91f1906eb6eec86e0fedb7", "question_id": "q10_temporal", "split": "test", "cluster_id": "syn_03009_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 1, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1976, "query_tokens": 614, "message_tokens": 512, "message_utf8_bytes": 1123, "cache_tensor_bytes": null, "serialized_payload_bytes": 1123, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 22747.736499994062, "hops": {"1": 22747.736499994062}, "receiver_prefill_ms": 375.8632499957457, "ttft_ms": 61.6170420544222, "decode_ms": 44.34745793696493}, "total_wall_ms": 23230.390707962215, "ttft_ms": 61.6170420544222, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 26, "retention_rate": 0.4, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.75, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1123}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 26, "retention_rate": 0.4, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.75, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1123}}}, "reference_key_id": "keys/synthetic/syn_03009_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "3bd2611f7bc59d6f4b6323c3ac1e9f7af93bcc7ed3e43397b183cdb4d6c6e058", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "714fe894b2adeea0b951b5f9", "attempt_id": "1456e8bde02d", "run_group_id": "E4_core:A4|closed|qwen3b:h1", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:13:26.624291Z", "task_family": "synthetic", "document_id": "syn_03009_2000", "document_hash": "6ed94dcf728255bd7a7d11c97c362930728fb81a3e91f1906eb6eec86e0fedb7", "question_id": "q11_aggregation", "split": "test", "cluster_id": "syn_03009_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 1, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1976, "query_tokens": 605, "message_tokens": 512, "message_utf8_bytes": 1123, "cache_tensor_bytes": null, "serialized_payload_bytes": 1123, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 22747.736499994062, "hops": {"1": 22747.736499994062}, "receiver_prefill_ms": 537.7571249846369, "ttft_ms": 43.5149580007419, "decode_ms": 217.41004195064306}, "total_wall_ms": 23547.077832976356, "ttft_ms": 43.5149580007419, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 7}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "5781336", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 26, "retention_rate": 0.4, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.75, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1123}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 26, "retention_rate": 0.4, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.75, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1123}}}, "reference_key_id": "keys/synthetic/syn_03009_2000.json", "task_scores": {"correct": 0.0, "abstained": 0.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "3bd2611f7bc59d6f4b6323c3ac1e9f7af93bcc7ed3e43397b183cdb4d6c6e058", "output_hash": "a365e04f74ba5baeccdd9e30dc27a1c5ff916597b828d874e63884b2090a60e2"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "2cd215d9325a473af642a045", "attempt_id": "fc15d9196359", "run_group_id": "E4_core:A4|closed|qwen3b:h1", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:13:27.440402Z", "task_family": "synthetic", "document_id": "syn_03009_2000", "document_hash": "6ed94dcf728255bd7a7d11c97c362930728fb81a3e91f1906eb6eec86e0fedb7", "question_id": "q12_unit_conversion", "split": "test", "cluster_id": "syn_03009_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 1, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1976, "query_tokens": 603, "message_tokens": 512, "message_utf8_bytes": 1123, "cache_tensor_bytes": null, "serialized_payload_bytes": 1123, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 22747.736499994062, "hops": {"1": 22747.736499994062}, "receiver_prefill_ms": 397.95833302196115, "ttft_ms": 19.438875024206936, "decode_ms": 37.71662502549589}, "total_wall_ms": 23203.557041008025, "ttft_ms": 19.438875024206936, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 26, "retention_rate": 0.4, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.75, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1123}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 26, "retention_rate": 0.4, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.75, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1123}}}, "reference_key_id": "keys/synthetic/syn_03009_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "3bd2611f7bc59d6f4b6323c3ac1e9f7af93bcc7ed3e43397b183cdb4d6c6e058", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "a74f8f3fd1b1eb04218ef3d3", "attempt_id": "b17717d6f33d", "run_group_id": "E4_core:A4|closed|qwen3b:h1", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:13:27.913718Z", "task_family": "synthetic", "document_id": "syn_03009_2000", "document_hash": "6ed94dcf728255bd7a7d11c97c362930728fb81a3e91f1906eb6eec86e0fedb7", "question_id": "q13_unanswerable", "split": "test", "cluster_id": "syn_03009_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 1, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1976, "query_tokens": 607, "message_tokens": 512, "message_utf8_bytes": 1123, "cache_tensor_bytes": null, "serialized_payload_bytes": 1123, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 22747.736499994062, "hops": {"1": 22747.736499994062}, "receiver_prefill_ms": 378.87145893182606, "ttft_ms": 50.67262495867908, "decode_ms": 53.26499999500811}, "total_wall_ms": 23231.198707944714, "ttft_ms": 50.67262495867908, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 26, "retention_rate": 0.4, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.75, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1123}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 26, "retention_rate": 0.4, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.75, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1123}}}, "reference_key_id": "keys/synthetic/syn_03009_2000.json", "task_scores": {"correct": 1.0, "abstained": 1.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "3bd2611f7bc59d6f4b6323c3ac1e9f7af93bcc7ed3e43397b183cdb4d6c6e058", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "139ccb31e1d3832ae4b50459", "attempt_id": "5ab85841e32c", "run_group_id": "E4_core:A4|closed|qwen3b:h1", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:13:28.414044Z", "task_family": "synthetic", "document_id": "syn_03009_2000", "document_hash": "6ed94dcf728255bd7a7d11c97c362930728fb81a3e91f1906eb6eec86e0fedb7", "question_id": "q14_unanswerable", "split": "test", "cluster_id": "syn_03009_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 1, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1976, "query_tokens": 607, "message_tokens": 512, "message_utf8_bytes": 1123, "cache_tensor_bytes": null, "serialized_payload_bytes": 1123, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 22747.736499994062, "hops": {"1": 22747.736499994062}, "receiver_prefill_ms": 563.4827089961618, "ttft_ms": 30.36204108502716, "decode_ms": 42.58074995595962}, "total_wall_ms": 23385.02033299301, "ttft_ms": 30.36204108502716, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 26, "retention_rate": 0.4, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.75, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1123}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 26, "retention_rate": 0.4, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.75, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1123}}}, "reference_key_id": "keys/synthetic/syn_03009_2000.json", "task_scores": {"correct": 1.0, "abstained": 1.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "3bd2611f7bc59d6f4b6323c3ac1e9f7af93bcc7ed3e43397b183cdb4d6c6e058", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "173cd9d753e1815d5db000bc", "attempt_id": "ba88440b9b67", "run_group_id": "E4_core:A4|closed|qwen3b:h2", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:13:50.776269Z", "task_family": "synthetic", "document_id": "syn_03009_2000", "document_hash": "6ed94dcf728255bd7a7d11c97c362930728fb81a3e91f1906eb6eec86e0fedb7", "question_id": "q00_single_fact", "split": "test", "cluster_id": "syn_03009_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 2, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1976, "query_tokens": 590, "message_tokens": 506, "message_utf8_bytes": 1103, "cache_tensor_bytes": null, "serialized_payload_bytes": 1103, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 44453.804999939166, "hops": {"1": 22747.736499994062, "2": 21706.068499945104}, "receiver_prefill_ms": 415.94249999616295, "ttft_ms": 29.118500067852437, "decode_ms": 325.2239170251414}, "total_wall_ms": 45224.76841695607, "ttft_ms": 29.118500067852437, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 10}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "PA-37F-6227", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 26, "retention_rate": 0.4, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.75, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1103}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 26, "retention_rate": 0.4, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.75, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1123}, "2": {"facts_total": 65, "facts_retained": 26, "retention_rate": 0.4, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.75, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1103}}}, "reference_key_id": "keys/synthetic/syn_03009_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "92c6e2fac34ed63b0b281363062a4533def3743816eb87bae8df51c6c25e46c5", "output_hash": "5b7645cf6e8140f4c0bf6ed4ad76121cf2924fe7e666d92161812527b86f4409"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "67a223b405053d6e5af78c2e", "attempt_id": "c610ee1893d7", "run_group_id": "E4_core:A4|closed|qwen3b:h2", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:13:51.565735Z", "task_family": "synthetic", "document_id": "syn_03009_2000", "document_hash": "6ed94dcf728255bd7a7d11c97c362930728fb81a3e91f1906eb6eec86e0fedb7", "question_id": "q01_single_fact", "split": "test", "cluster_id": "syn_03009_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 2, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1976, "query_tokens": 595, "message_tokens": 506, "message_utf8_bytes": 1103, "cache_tensor_bytes": null, "serialized_payload_bytes": 1103, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 44453.804999939166, "hops": {"1": 22747.736499994062, "2": 21706.068499945104}, "receiver_prefill_ms": 391.5781669784337, "ttft_ms": 45.51900003571063, "decode_ms": 98.23858400341123}, "total_wall_ms": 44989.79858297389, "ttft_ms": 45.51900003571063, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 26, "retention_rate": 0.4, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.75, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1103}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 26, "retention_rate": 0.4, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.75, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1123}, "2": {"facts_total": 65, "facts_retained": 26, "retention_rate": 0.4, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.75, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1103}}}, "reference_key_id": "keys/synthetic/syn_03009_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "92c6e2fac34ed63b0b281363062a4533def3743816eb87bae8df51c6c25e46c5", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "8402b1a2221e069f91acf4dd", "attempt_id": "cc42c636f747", "run_group_id": "E4_core:A4|closed|qwen3b:h2", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:13:52.120027Z", "task_family": "synthetic", "document_id": "syn_03009_2000", "document_hash": "6ed94dcf728255bd7a7d11c97c362930728fb81a3e91f1906eb6eec86e0fedb7", "question_id": "q02_exact_copy", "split": "test", "cluster_id": "syn_03009_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 2, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1976, "query_tokens": 593, "message_tokens": 506, "message_utf8_bytes": 1103, "cache_tensor_bytes": null, "serialized_payload_bytes": 1103, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 44453.804999939166, "hops": {"1": 22747.736499994062, "2": 21706.068499945104}, "receiver_prefill_ms": 460.6097920332104, "ttft_ms": 37.29804209433496, "decode_ms": 22.07000006455928}, "total_wall_ms": 44974.39749992918, "ttft_ms": 37.29804209433496, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 26, "retention_rate": 0.4, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.75, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1103}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 26, "retention_rate": 0.4, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.75, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1123}, "2": {"facts_total": 65, "facts_retained": 26, "retention_rate": 0.4, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.75, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1103}}}, "reference_key_id": "keys/synthetic/syn_03009_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "92c6e2fac34ed63b0b281363062a4533def3743816eb87bae8df51c6c25e46c5", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "b396a96e3922dc31adca8ea2", "attempt_id": "592b945be582", "run_group_id": "E4_core:A4|closed|qwen3b:h2", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:13:52.657726Z", "task_family": "synthetic", "document_id": "syn_03009_2000", "document_hash": "6ed94dcf728255bd7a7d11c97c362930728fb81a3e91f1906eb6eec86e0fedb7", "question_id": "q03_exact_copy", "split": "test", "cluster_id": "syn_03009_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 2, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1976, "query_tokens": 589, "message_tokens": 506, "message_utf8_bytes": 1103, "cache_tensor_bytes": null, "serialized_payload_bytes": 1103, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 44453.804999939166, "hops": {"1": 22747.736499994062, "2": 21706.068499945104}, "receiver_prefill_ms": 397.25291600916535, "ttft_ms": 19.293209072202444, "decode_ms": 146.627250011079}, "total_wall_ms": 45017.57787493989, "ttft_ms": 19.293209072202444, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 5}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "Idris Bergström", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 26, "retention_rate": 0.4, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.75, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1103}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 26, "retention_rate": 0.4, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.75, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1123}, "2": {"facts_total": 65, "facts_retained": 26, "retention_rate": 0.4, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.75, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1103}}}, "reference_key_id": "keys/synthetic/syn_03009_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "92c6e2fac34ed63b0b281363062a4533def3743816eb87bae8df51c6c25e46c5", "output_hash": "fd88832c9fd0ab175818d76d875d70156225bd4b34ec401514546ef70f4f826c"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "ed941d534fc804ab1d40b97c", "attempt_id": "93335e25bcfd", "run_group_id": "E4_core:A4|closed|qwen3b:h2", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:13:53.237995Z", "task_family": "synthetic", "document_id": "syn_03009_2000", "document_hash": "6ed94dcf728255bd7a7d11c97c362930728fb81a3e91f1906eb6eec86e0fedb7", "question_id": "q04_multi_hop", "split": "test", "cluster_id": "syn_03009_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 2, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1976, "query_tokens": 597, "message_tokens": 506, "message_utf8_bytes": 1103, "cache_tensor_bytes": null, "serialized_payload_bytes": 1103, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 44453.804999939166, "hops": {"1": 22747.736499994062, "2": 21706.068499945104}, "receiver_prefill_ms": 430.26037502568215, "ttft_ms": 45.556292054243386, "decode_ms": 403.2930409302935}, "total_wall_ms": 45333.53391685523, "ttft_ms": 45.556292054243386, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 7}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "4670383", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 26, "retention_rate": 0.4, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.75, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1103}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 26, "retention_rate": 0.4, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.75, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1123}, "2": {"facts_total": 65, "facts_retained": 26, "retention_rate": 0.4, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.75, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1103}}}, "reference_key_id": "keys/synthetic/syn_03009_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "92c6e2fac34ed63b0b281363062a4533def3743816eb87bae8df51c6c25e46c5", "output_hash": "19ae8c357e5500bf494a6163bfbd11380b2996e8ea532a2c64f4a599464820f7"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "01902d2e65cbb19e67976451", "attempt_id": "ae1fc1e9a2ae", "run_group_id": "E4_core:A4|closed|qwen3b:h2", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:13:54.136953Z", "task_family": "synthetic", "document_id": "syn_03009_2000", "document_hash": "6ed94dcf728255bd7a7d11c97c362930728fb81a3e91f1906eb6eec86e0fedb7", "question_id": "q05_multi_hop", "split": "test", "cluster_id": "syn_03009_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 2, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1976, "query_tokens": 600, "message_tokens": 506, "message_utf8_bytes": 1103, "cache_tensor_bytes": null, "serialized_payload_bytes": 1103, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 44453.804999939166, "hops": {"1": 22747.736499994062, "2": 21706.068499945104}, "receiver_prefill_ms": 402.44816697668284, "ttft_ms": 34.71845795866102, "decode_ms": 30.729625024832785}, "total_wall_ms": 44922.27320896927, "ttft_ms": 34.71845795866102, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 26, "retention_rate": 0.4, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.75, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1103}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 26, "retention_rate": 0.4, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.75, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1123}, "2": {"facts_total": 65, "facts_retained": 26, "retention_rate": 0.4, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.75, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1103}}}, "reference_key_id": "keys/synthetic/syn_03009_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "92c6e2fac34ed63b0b281363062a4533def3743816eb87bae8df51c6c25e46c5", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "6aba022a5d1cb7700648a857", "attempt_id": "6cf82d22137e", "run_group_id": "E4_core:A4|closed|qwen3b:h2", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:13:54.622890Z", "task_family": "synthetic", "document_id": "syn_03009_2000", "document_hash": "6ed94dcf728255bd7a7d11c97c362930728fb81a3e91f1906eb6eec86e0fedb7", "question_id": "q06_negation", "split": "test", "cluster_id": "syn_03009_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 2, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1976, "query_tokens": 590, "message_tokens": 506, "message_utf8_bytes": 1103, "cache_tensor_bytes": null, "serialized_payload_bytes": 1103, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 44453.804999939166, "hops": {"1": 22747.736499994062, "2": 21706.068499945104}, "receiver_prefill_ms": 377.35233292914927, "ttft_ms": 60.958083951845765, "decode_ms": 50.5742080276832}, "total_wall_ms": 44943.77349992283, "ttft_ms": 60.958083951845765, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 26, "retention_rate": 0.4, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.75, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1103}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 26, "retention_rate": 0.4, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.75, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1123}, "2": {"facts_total": 65, "facts_retained": 26, "retention_rate": 0.4, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.75, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1103}}}, "reference_key_id": "keys/synthetic/syn_03009_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "92c6e2fac34ed63b0b281363062a4533def3743816eb87bae8df51c6c25e46c5", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "0baaf6cd19505a0f55721ba5", "attempt_id": "71d4c41360f9", "run_group_id": "E4_core:A4|closed|qwen3b:h2", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:13:55.131090Z", "task_family": "synthetic", "document_id": "syn_03009_2000", "document_hash": "6ed94dcf728255bd7a7d11c97c362930728fb81a3e91f1906eb6eec86e0fedb7", "question_id": "q07_negation", "split": "test", "cluster_id": "syn_03009_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 2, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1976, "query_tokens": 589, "message_tokens": 506, "message_utf8_bytes": 1103, "cache_tensor_bytes": null, "serialized_payload_bytes": 1103, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 44453.804999939166, "hops": {"1": 22747.736499994062, "2": 21706.068499945104}, "receiver_prefill_ms": 533.9793330058455, "ttft_ms": 35.773082985542715, "decode_ms": 20.331457955762744}, "total_wall_ms": 45044.57645898219, "ttft_ms": 35.773082985542715, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 26, "retention_rate": 0.4, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.75, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1103}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 26, "retention_rate": 0.4, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.75, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1123}, "2": {"facts_total": 65, "facts_retained": 26, "retention_rate": 0.4, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.75, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1103}}}, "reference_key_id": "keys/synthetic/syn_03009_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "92c6e2fac34ed63b0b281363062a4533def3743816eb87bae8df51c6c25e46c5", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "c4440f1f6fd34f277c8f28d0", "attempt_id": "f2b51912ce03", "run_group_id": "E4_core:A4|closed|qwen3b:h2", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:13:55.740019Z", "task_family": "synthetic", "document_id": "syn_03009_2000", "document_hash": "6ed94dcf728255bd7a7d11c97c362930728fb81a3e91f1906eb6eec86e0fedb7", "question_id": "q08_revision", "split": "test", "cluster_id": "syn_03009_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 2, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1976, "query_tokens": 607, "message_tokens": 506, "message_utf8_bytes": 1103, "cache_tensor_bytes": null, "serialized_payload_bytes": 1103, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 44453.804999939166, "hops": {"1": 22747.736499994062, "2": 21706.068499945104}, "receiver_prefill_ms": 398.9171660505235, "ttft_ms": 33.73841708526015, "decode_ms": 265.1266250759363}, "total_wall_ms": 45152.263707946986, "ttft_ms": 33.73841708526015, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 8}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "10047105", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 26, "retention_rate": 0.4, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.75, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1103}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 26, "retention_rate": 0.4, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.75, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1123}, "2": {"facts_total": 65, "facts_retained": 26, "retention_rate": 0.4, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.75, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1103}}}, "reference_key_id": "keys/synthetic/syn_03009_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "92c6e2fac34ed63b0b281363062a4533def3743816eb87bae8df51c6c25e46c5", "output_hash": "bea99610d1e3e75a7f20939faf8e6824e02237b156dacd6442ba2a7e02617510"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "9b028700e202dbb5f51a354d", "attempt_id": "335d8999866f", "run_group_id": "E4_core:A4|closed|qwen3b:h2", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:13:56.456145Z", "task_family": "synthetic", "document_id": "syn_03009_2000", "document_hash": "6ed94dcf728255bd7a7d11c97c362930728fb81a3e91f1906eb6eec86e0fedb7", "question_id": "q09_revision", "split": "test", "cluster_id": "syn_03009_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 2, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1976, "query_tokens": 600, "message_tokens": 506, "message_utf8_bytes": 1103, "cache_tensor_bytes": null, "serialized_payload_bytes": 1103, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 44453.804999939166, "hops": {"1": 22747.736499994062, "2": 21706.068499945104}, "receiver_prefill_ms": 361.51462490670383, "ttft_ms": 36.324792075902224, "decode_ms": 510.90379105880857}, "total_wall_ms": 45363.32749994472, "ttft_ms": 36.324792075902224, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 8}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "10047105", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 26, "retention_rate": 0.4, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.75, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1103}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 26, "retention_rate": 0.4, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.75, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1123}, "2": {"facts_total": 65, "facts_retained": 26, "retention_rate": 0.4, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.75, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1103}}}, "reference_key_id": "keys/synthetic/syn_03009_2000.json", "task_scores": {"correct": 0.0, "abstained": 0.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "92c6e2fac34ed63b0b281363062a4533def3743816eb87bae8df51c6c25e46c5", "output_hash": "bea99610d1e3e75a7f20939faf8e6824e02237b156dacd6442ba2a7e02617510"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "13f49d54160e682731963555", "attempt_id": "4c0c7e937212", "run_group_id": "E4_core:A4|closed|qwen3b:h2", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:13:57.383409Z", "task_family": "synthetic", "document_id": "syn_03009_2000", "document_hash": "6ed94dcf728255bd7a7d11c97c362930728fb81a3e91f1906eb6eec86e0fedb7", "question_id": "q10_temporal", "split": "test", "cluster_id": "syn_03009_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 2, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1976, "query_tokens": 609, "message_tokens": 506, "message_utf8_bytes": 1103, "cache_tensor_bytes": null, "serialized_payload_bytes": 1103, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 44453.804999939166, "hops": {"1": 22747.736499994062, "2": 21706.068499945104}, "receiver_prefill_ms": 414.6846670191735, "ttft_ms": 43.450707918964326, "decode_ms": 25.037667015567422}, "total_wall_ms": 44937.742416979745, "ttft_ms": 43.450707918964326, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 26, "retention_rate": 0.4, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.75, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1103}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 26, "retention_rate": 0.4, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.75, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1123}, "2": {"facts_total": 65, "facts_retained": 26, "retention_rate": 0.4, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.75, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1103}}}, "reference_key_id": "keys/synthetic/syn_03009_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "92c6e2fac34ed63b0b281363062a4533def3743816eb87bae8df51c6c25e46c5", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "3529a3205e93f2e8ed2663c4", "attempt_id": "fd5485e8f3f6", "run_group_id": "E4_core:A4|closed|qwen3b:h2", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:13:57.884755Z", "task_family": "synthetic", "document_id": "syn_03009_2000", "document_hash": "6ed94dcf728255bd7a7d11c97c362930728fb81a3e91f1906eb6eec86e0fedb7", "question_id": "q11_aggregation", "split": "test", "cluster_id": "syn_03009_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 2, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1976, "query_tokens": 600, "message_tokens": 506, "message_utf8_bytes": 1103, "cache_tensor_bytes": null, "serialized_payload_bytes": 1103, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 44453.804999939166, "hops": {"1": 22747.736499994062, "2": 21706.068499945104}, "receiver_prefill_ms": 371.57454201951623, "ttft_ms": 18.51375005207956, "decode_ms": 438.73829196672887}, "total_wall_ms": 45283.415957936086, "ttft_ms": 18.51375005207956, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 7}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "5247196", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 26, "retention_rate": 0.4, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.75, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1103}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 26, "retention_rate": 0.4, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.75, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1123}, "2": {"facts_total": 65, "facts_retained": 26, "retention_rate": 0.4, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.75, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1103}}}, "reference_key_id": "keys/synthetic/syn_03009_2000.json", "task_scores": {"correct": 0.0, "abstained": 0.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "92c6e2fac34ed63b0b281363062a4533def3743816eb87bae8df51c6c25e46c5", "output_hash": "3ac48b1390a5593419a76c6409f1c54c01affa22edaa1be1850800277310a17c"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "c96b613e2b2f65862db586a1", "attempt_id": "f61a47d448b1", "run_group_id": "E4_core:A4|closed|qwen3b:h2", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:13:58.732572Z", "task_family": "synthetic", "document_id": "syn_03009_2000", "document_hash": "6ed94dcf728255bd7a7d11c97c362930728fb81a3e91f1906eb6eec86e0fedb7", "question_id": "q12_unit_conversion", "split": "test", "cluster_id": "syn_03009_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 2, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1976, "query_tokens": 598, "message_tokens": 506, "message_utf8_bytes": 1103, "cache_tensor_bytes": null, "serialized_payload_bytes": 1103, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 44453.804999939166, "hops": {"1": 22747.736499994062, "2": 21706.068499945104}, "receiver_prefill_ms": 479.7141250455752, "ttft_ms": 43.10150002129376, "decode_ms": 32.23583404906094}, "total_wall_ms": 45009.4362499658, "ttft_ms": 43.10150002129376, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 26, "retention_rate": 0.4, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.75, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1103}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 26, "retention_rate": 0.4, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.75, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1123}, "2": {"facts_total": 65, "facts_retained": 26, "retention_rate": 0.4, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.75, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1103}}}, "reference_key_id": "keys/synthetic/syn_03009_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "92c6e2fac34ed63b0b281363062a4533def3743816eb87bae8df51c6c25e46c5", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "7736aa103d78ad2f88a0184d", "attempt_id": "3eb17cf91e5b", "run_group_id": "E4_core:A4|closed|qwen3b:h2", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:13:59.308183Z", "task_family": "synthetic", "document_id": "syn_03009_2000", "document_hash": "6ed94dcf728255bd7a7d11c97c362930728fb81a3e91f1906eb6eec86e0fedb7", "question_id": "q13_unanswerable", "split": "test", "cluster_id": "syn_03009_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 2, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1976, "query_tokens": 602, "message_tokens": 506, "message_utf8_bytes": 1103, "cache_tensor_bytes": null, "serialized_payload_bytes": 1103, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 44453.804999939166, "hops": {"1": 22747.736499994062, "2": 21706.068499945104}, "receiver_prefill_ms": 394.2864580312744, "ttft_ms": 38.29358296934515, "decode_ms": 24.15066701360047}, "total_wall_ms": 44911.14300000481, "ttft_ms": 38.29358296934515, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 26, "retention_rate": 0.4, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.75, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1103}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 26, "retention_rate": 0.4, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.75, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1123}, "2": {"facts_total": 65, "facts_retained": 26, "retention_rate": 0.4, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.75, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1103}}}, "reference_key_id": "keys/synthetic/syn_03009_2000.json", "task_scores": {"correct": 1.0, "abstained": 1.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "92c6e2fac34ed63b0b281363062a4533def3743816eb87bae8df51c6c25e46c5", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "26eb04f65546c974b980f9ec", "attempt_id": "75360d1ba937", "run_group_id": "E4_core:A4|closed|qwen3b:h2", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:13:59.782543Z", "task_family": "synthetic", "document_id": "syn_03009_2000", "document_hash": "6ed94dcf728255bd7a7d11c97c362930728fb81a3e91f1906eb6eec86e0fedb7", "question_id": "q14_unanswerable", "split": "test", "cluster_id": "syn_03009_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 2, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1976, "query_tokens": 602, "message_tokens": 506, "message_utf8_bytes": 1103, "cache_tensor_bytes": null, "serialized_payload_bytes": 1103, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 44453.804999939166, "hops": {"1": 22747.736499994062, "2": 21706.068499945104}, "receiver_prefill_ms": 344.90445896517485, "ttft_ms": 58.77704103477299, "decode_ms": 45.06916704121977}, "total_wall_ms": 44903.43087492511, "ttft_ms": 58.77704103477299, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 26, "retention_rate": 0.4, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.75, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1103}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 26, "retention_rate": 0.4, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.75, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1123}, "2": {"facts_total": 65, "facts_retained": 26, "retention_rate": 0.4, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.75, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1103}}}, "reference_key_id": "keys/synthetic/syn_03009_2000.json", "task_scores": {"correct": 1.0, "abstained": 1.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "92c6e2fac34ed63b0b281363062a4533def3743816eb87bae8df51c6c25e46c5", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "3927100820507e5b526648c9", "attempt_id": "9c571e5e0fed", "run_group_id": "E4_core:A4|closed|qwen3b:h4", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:14:43.632048Z", "task_family": "synthetic", "document_id": "syn_03009_2000", "document_hash": "6ed94dcf728255bd7a7d11c97c362930728fb81a3e91f1906eb6eec86e0fedb7", "question_id": "q00_single_fact", "split": "test", "cluster_id": "syn_03009_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 4, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1976, "query_tokens": 590, "message_tokens": 506, "message_utf8_bytes": 1103, "cache_tensor_bytes": null, "serialized_payload_bytes": 1103, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 87833.84841599036, "hops": {"1": 22747.736499994062, "2": 21706.068499945104, "3": 21679.54479099717, "4": 21700.498625054024}, "receiver_prefill_ms": 407.04724995885044, "ttft_ms": 95.73116595856845, "decode_ms": 442.4387910403311}, "total_wall_ms": 88779.92916607764, "ttft_ms": 95.73116595856845, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 10}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "PA-37F-6227", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 26, "retention_rate": 0.4, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.75, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1103}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 26, "retention_rate": 0.4, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.75, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1123}, "2": {"facts_total": 65, "facts_retained": 26, "retention_rate": 0.4, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.75, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1103}, "3": {"facts_total": 65, "facts_retained": 26, "retention_rate": 0.4, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.75, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1103}, "4": {"facts_total": 65, "facts_retained": 26, "retention_rate": 0.4, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.75, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1103}}}, "reference_key_id": "keys/synthetic/syn_03009_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "92c6e2fac34ed63b0b281363062a4533def3743816eb87bae8df51c6c25e46c5", "output_hash": "5b7645cf6e8140f4c0bf6ed4ad76121cf2924fe7e666d92161812527b86f4409"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "020c614f7aeec8ae73a26372", "attempt_id": "6c029a31230c", "run_group_id": "E4_core:A4|closed|qwen3b:h4", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:14:44.597089Z", "task_family": "synthetic", "document_id": "syn_03009_2000", "document_hash": "6ed94dcf728255bd7a7d11c97c362930728fb81a3e91f1906eb6eec86e0fedb7", "question_id": "q01_single_fact", "split": "test", "cluster_id": "syn_03009_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 4, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1976, "query_tokens": 595, "message_tokens": 506, "message_utf8_bytes": 1103, "cache_tensor_bytes": null, "serialized_payload_bytes": 1103, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 87833.84841599036, "hops": {"1": 22747.736499994062, "2": 21706.068499945104, "3": 21679.54479099717, "4": 21700.498625054024}, "receiver_prefill_ms": 399.4190839584917, "ttft_ms": 27.633833000436425, "decode_ms": 37.185165914706886}, "total_wall_ms": 88298.77554101404, "ttft_ms": 27.633833000436425, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 26, "retention_rate": 0.4, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.75, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1103}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 26, "retention_rate": 0.4, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.75, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1123}, "2": {"facts_total": 65, "facts_retained": 26, "retention_rate": 0.4, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.75, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1103}, "3": {"facts_total": 65, "facts_retained": 26, "retention_rate": 0.4, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.75, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1103}, "4": {"facts_total": 65, "facts_retained": 26, "retention_rate": 0.4, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.75, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1103}}}, "reference_key_id": "keys/synthetic/syn_03009_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "92c6e2fac34ed63b0b281363062a4533def3743816eb87bae8df51c6c25e46c5", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "3e2a38414b2707494293e730", "attempt_id": "f17fef625e8e", "run_group_id": "E4_core:A4|closed|qwen3b:h4", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:14:45.079901Z", "task_family": "synthetic", "document_id": "syn_03009_2000", "document_hash": "6ed94dcf728255bd7a7d11c97c362930728fb81a3e91f1906eb6eec86e0fedb7", "question_id": "q02_exact_copy", "split": "test", "cluster_id": "syn_03009_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 4, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1976, "query_tokens": 593, "message_tokens": 506, "message_utf8_bytes": 1103, "cache_tensor_bytes": null, "serialized_payload_bytes": 1103, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 87833.84841599036, "hops": {"1": 22747.736499994062, "2": 21706.068499945104, "3": 21679.54479099717, "4": 21700.498625054024}, "receiver_prefill_ms": 352.22612507641315, "ttft_ms": 35.82791704684496, "decode_ms": 62.54083302337676}, "total_wall_ms": 88285.23979091551, "ttft_ms": 35.82791704684496, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 26, "retention_rate": 0.4, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.75, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1103}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 26, "retention_rate": 0.4, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.75, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1123}, "2": {"facts_total": 65, "facts_retained": 26, "retention_rate": 0.4, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.75, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1103}, "3": {"facts_total": 65, "facts_retained": 26, "retention_rate": 0.4, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.75, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1103}, "4": {"facts_total": 65, "facts_retained": 26, "retention_rate": 0.4, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.75, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1103}}}, "reference_key_id": "keys/synthetic/syn_03009_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "92c6e2fac34ed63b0b281363062a4533def3743816eb87bae8df51c6c25e46c5", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "0dd703b767f3755381439497", "attempt_id": "2ec5f0e300b6", "run_group_id": "E4_core:A4|closed|qwen3b:h4", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:14:45.548887Z", "task_family": "synthetic", "document_id": "syn_03009_2000", "document_hash": "6ed94dcf728255bd7a7d11c97c362930728fb81a3e91f1906eb6eec86e0fedb7", "question_id": "q03_exact_copy", "split": "test", "cluster_id": "syn_03009_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 4, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1976, "query_tokens": 589, "message_tokens": 506, "message_utf8_bytes": 1103, "cache_tensor_bytes": null, "serialized_payload_bytes": 1103, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 87833.84841599036, "hops": {"1": 22747.736499994062, "2": 21706.068499945104, "3": 21679.54479099717, "4": 21700.498625054024}, "receiver_prefill_ms": 508.39400000404567, "ttft_ms": 26.718457927927375, "decode_ms": 166.4812919916585}, "total_wall_ms": 88536.1534570111, "ttft_ms": 26.718457927927375, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 5}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "Idris Bergström", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 26, "retention_rate": 0.4, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.75, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1103}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 26, "retention_rate": 0.4, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.75, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1123}, "2": {"facts_total": 65, "facts_retained": 26, "retention_rate": 0.4, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.75, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1103}, "3": {"facts_total": 65, "facts_retained": 26, "retention_rate": 0.4, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.75, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1103}, "4": {"facts_total": 65, "facts_retained": 26, "retention_rate": 0.4, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.75, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1103}}}, "reference_key_id": "keys/synthetic/syn_03009_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "92c6e2fac34ed63b0b281363062a4533def3743816eb87bae8df51c6c25e46c5", "output_hash": "fd88832c9fd0ab175818d76d875d70156225bd4b34ec401514546ef70f4f826c"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "5cbf03895bab556c72121895", "attempt_id": "2f22fbe9405f", "run_group_id": "E4_core:A4|closed|qwen3b:h4", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:14:46.268582Z", "task_family": "synthetic", "document_id": "syn_03009_2000", "document_hash": "6ed94dcf728255bd7a7d11c97c362930728fb81a3e91f1906eb6eec86e0fedb7", "question_id": "q04_multi_hop", "split": "test", "cluster_id": "syn_03009_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 4, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1976, "query_tokens": 597, "message_tokens": 506, "message_utf8_bytes": 1103, "cache_tensor_bytes": null, "serialized_payload_bytes": 1103, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 87833.84841599036, "hops": {"1": 22747.736499994062, "2": 21706.068499945104, "3": 21679.54479099717, "4": 21700.498625054024}, "receiver_prefill_ms": 404.5027500251308, "ttft_ms": 34.47470802348107, "decode_ms": 193.2235840940848}, "total_wall_ms": 88466.71524900012, "ttft_ms": 34.47470802348107, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 7}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "4670383", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 26, "retention_rate": 0.4, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.75, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1103}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 26, "retention_rate": 0.4, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.75, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1123}, "2": {"facts_total": 65, "facts_retained": 26, "retention_rate": 0.4, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.75, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1103}, "3": {"facts_total": 65, "facts_retained": 26, "retention_rate": 0.4, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.75, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1103}, "4": {"facts_total": 65, "facts_retained": 26, "retention_rate": 0.4, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.75, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1103}}}, "reference_key_id": "keys/synthetic/syn_03009_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "92c6e2fac34ed63b0b281363062a4533def3743816eb87bae8df51c6c25e46c5", "output_hash": "19ae8c357e5500bf494a6163bfbd11380b2996e8ea532a2c64f4a599464820f7"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "6229fcb36cf620a3f5e9af56", "attempt_id": "ddb83574cd16", "run_group_id": "E4_core:A4|closed|qwen3b:h4", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:14:46.918891Z", "task_family": "synthetic", "document_id": "syn_03009_2000", "document_hash": "6ed94dcf728255bd7a7d11c97c362930728fb81a3e91f1906eb6eec86e0fedb7", "question_id": "q05_multi_hop", "split": "test", "cluster_id": "syn_03009_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 4, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1976, "query_tokens": 600, "message_tokens": 506, "message_utf8_bytes": 1103, "cache_tensor_bytes": null, "serialized_payload_bytes": 1103, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 87833.84841599036, "hops": {"1": 22747.736499994062, "2": 21706.068499945104, "3": 21679.54479099717, "4": 21700.498625054024}, "receiver_prefill_ms": 390.221708919853, "ttft_ms": 52.5954159675166, "decode_ms": 35.66654201131314}, "total_wall_ms": 88313.19699902087, "ttft_ms": 52.5954159675166, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 26, "retention_rate": 0.4, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.75, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1103}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 26, "retention_rate": 0.4, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.75, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1123}, "2": {"facts_total": 65, "facts_retained": 26, "retention_rate": 0.4, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.75, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1103}, "3": {"facts_total": 65, "facts_retained": 26, "retention_rate": 0.4, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.75, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1103}, "4": {"facts_total": 65, "facts_retained": 26, "retention_rate": 0.4, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.75, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1103}}}, "reference_key_id": "keys/synthetic/syn_03009_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "92c6e2fac34ed63b0b281363062a4533def3743816eb87bae8df51c6c25e46c5", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "950d27c49019bbfda11d3f97", "attempt_id": "558615628b4d", "run_group_id": "E4_core:A4|closed|qwen3b:h4", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:14:47.417991Z", "task_family": "synthetic", "document_id": "syn_03009_2000", "document_hash": "6ed94dcf728255bd7a7d11c97c362930728fb81a3e91f1906eb6eec86e0fedb7", "question_id": "q06_negation", "split": "test", "cluster_id": "syn_03009_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 4, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1976, "query_tokens": 590, "message_tokens": 506, "message_utf8_bytes": 1103, "cache_tensor_bytes": null, "serialized_payload_bytes": 1103, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 87833.84841599036, "hops": {"1": 22747.736499994062, "2": 21706.068499945104, "3": 21679.54479099717, "4": 21700.498625054024}, "receiver_prefill_ms": 511.6297909989953, "ttft_ms": 38.23195805307478, "decode_ms": 24.054958019405603}, "total_wall_ms": 88408.3499580156, "ttft_ms": 38.23195805307478, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 26, "retention_rate": 0.4, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.75, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1103}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 26, "retention_rate": 0.4, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.75, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1123}, "2": {"facts_total": 65, "facts_retained": 26, "retention_rate": 0.4, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.75, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1103}, "3": {"facts_total": 65, "facts_retained": 26, "retention_rate": 0.4, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.75, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1103}, "4": {"facts_total": 65, "facts_retained": 26, "retention_rate": 0.4, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.75, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1103}}}, "reference_key_id": "keys/synthetic/syn_03009_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "92c6e2fac34ed63b0b281363062a4533def3743816eb87bae8df51c6c25e46c5", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "d2dcb2e316b8a001522e27fb", "attempt_id": "87843c85601d", "run_group_id": "E4_core:A4|closed|qwen3b:h4", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:14:48.009495Z", "task_family": "synthetic", "document_id": "syn_03009_2000", "document_hash": "6ed94dcf728255bd7a7d11c97c362930728fb81a3e91f1906eb6eec86e0fedb7", "question_id": "q07_negation", "split": "test", "cluster_id": "syn_03009_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 4, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1976, "query_tokens": 589, "message_tokens": 506, "message_utf8_bytes": 1103, "cache_tensor_bytes": null, "serialized_payload_bytes": 1103, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 87833.84841599036, "hops": {"1": 22747.736499994062, "2": 21706.068499945104, "3": 21679.54479099717, "4": 21700.498625054024}, "receiver_prefill_ms": 400.6030419841409, "ttft_ms": 33.98137493059039, "decode_ms": 29.7497920691967}, "total_wall_ms": 88298.76620799769, "ttft_ms": 33.98137493059039, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 26, "retention_rate": 0.4, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.75, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1103}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 26, "retention_rate": 0.4, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.75, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1123}, "2": {"facts_total": 65, "facts_retained": 26, "retention_rate": 0.4, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.75, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1103}, "3": {"facts_total": 65, "facts_retained": 26, "retention_rate": 0.4, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.75, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1103}, "4": {"facts_total": 65, "facts_retained": 26, "retention_rate": 0.4, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.75, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1103}}}, "reference_key_id": "keys/synthetic/syn_03009_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "92c6e2fac34ed63b0b281363062a4533def3743816eb87bae8df51c6c25e46c5", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "8949f71fbd5bb62e635cbd89", "attempt_id": "f535a82c3f98", "run_group_id": "E4_core:A4|closed|qwen3b:h4", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:14:48.491382Z", "task_family": "synthetic", "document_id": "syn_03009_2000", "document_hash": "6ed94dcf728255bd7a7d11c97c362930728fb81a3e91f1906eb6eec86e0fedb7", "question_id": "q08_revision", "split": "test", "cluster_id": "syn_03009_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 4, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1976, "query_tokens": 607, "message_tokens": 506, "message_utf8_bytes": 1103, "cache_tensor_bytes": null, "serialized_payload_bytes": 1103, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 87833.84841599036, "hops": {"1": 22747.736499994062, "2": 21706.068499945104, "3": 21679.54479099717, "4": 21700.498625054024}, "receiver_prefill_ms": 356.35670798365027, "ttft_ms": 51.25554197002202, "decode_ms": 531.6175410989672}, "total_wall_ms": 88773.86070799548, "ttft_ms": 51.25554197002202, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 8}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "10047105", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 26, "retention_rate": 0.4, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.75, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1103}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 26, "retention_rate": 0.4, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.75, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1123}, "2": {"facts_total": 65, "facts_retained": 26, "retention_rate": 0.4, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.75, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1103}, "3": {"facts_total": 65, "facts_retained": 26, "retention_rate": 0.4, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.75, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1103}, "4": {"facts_total": 65, "facts_retained": 26, "retention_rate": 0.4, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.75, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1103}}}, "reference_key_id": "keys/synthetic/syn_03009_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "92c6e2fac34ed63b0b281363062a4533def3743816eb87bae8df51c6c25e46c5", "output_hash": "bea99610d1e3e75a7f20939faf8e6824e02237b156dacd6442ba2a7e02617510"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "dd9aecda869efd2c11829c9b", "attempt_id": "75db1488cd5f", "run_group_id": "E4_core:A4|closed|qwen3b:h4", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:14:49.448597Z", "task_family": "synthetic", "document_id": "syn_03009_2000", "document_hash": "6ed94dcf728255bd7a7d11c97c362930728fb81a3e91f1906eb6eec86e0fedb7", "question_id": "q09_revision", "split": "test", "cluster_id": "syn_03009_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 4, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1976, "query_tokens": 600, "message_tokens": 506, "message_utf8_bytes": 1103, "cache_tensor_bytes": null, "serialized_payload_bytes": 1103, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 87833.84841599036, "hops": {"1": 22747.736499994062, "2": 21706.068499945104, "3": 21679.54479099717, "4": 21700.498625054024}, "receiver_prefill_ms": 388.5620830114931, "ttft_ms": 44.48558297008276, "decode_ms": 257.6406659791246}, "total_wall_ms": 88525.37049993407, "ttft_ms": 44.48558297008276, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 8}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "10047105", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 26, "retention_rate": 0.4, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.75, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1103}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 26, "retention_rate": 0.4, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.75, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1123}, "2": {"facts_total": 65, "facts_retained": 26, "retention_rate": 0.4, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.75, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1103}, "3": {"facts_total": 65, "facts_retained": 26, "retention_rate": 0.4, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.75, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1103}, "4": {"facts_total": 65, "facts_retained": 26, "retention_rate": 0.4, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.75, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1103}}}, "reference_key_id": "keys/synthetic/syn_03009_2000.json", "task_scores": {"correct": 0.0, "abstained": 0.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "92c6e2fac34ed63b0b281363062a4533def3743816eb87bae8df51c6c25e46c5", "output_hash": "bea99610d1e3e75a7f20939faf8e6824e02237b156dacd6442ba2a7e02617510"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "e8777307c31558d4a24de689", "attempt_id": "67db2f839c61", "run_group_id": "E4_core:A4|closed|qwen3b:h4", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:14:50.159135Z", "task_family": "synthetic", "document_id": "syn_03009_2000", "document_hash": "6ed94dcf728255bd7a7d11c97c362930728fb81a3e91f1906eb6eec86e0fedb7", "question_id": "q10_temporal", "split": "test", "cluster_id": "syn_03009_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 4, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1976, "query_tokens": 609, "message_tokens": 506, "message_utf8_bytes": 1103, "cache_tensor_bytes": null, "serialized_payload_bytes": 1103, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 87833.84841599036, "hops": {"1": 22747.736499994062, "2": 21706.068499945104, "3": 21679.54479099717, "4": 21700.498625054024}, "receiver_prefill_ms": 347.5087909027934, "ttft_ms": 43.467832962051034, "decode_ms": 56.01358402054757}, "total_wall_ms": 88281.62024996709, "ttft_ms": 43.467832962051034, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 26, "retention_rate": 0.4, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.75, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1103}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 26, "retention_rate": 0.4, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.75, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1123}, "2": {"facts_total": 65, "facts_retained": 26, "retention_rate": 0.4, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.75, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1103}, "3": {"facts_total": 65, "facts_retained": 26, "retention_rate": 0.4, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.75, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1103}, "4": {"facts_total": 65, "facts_retained": 26, "retention_rate": 0.4, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.75, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1103}}}, "reference_key_id": "keys/synthetic/syn_03009_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "92c6e2fac34ed63b0b281363062a4533def3743816eb87bae8df51c6c25e46c5", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "332630c90275a67cb4a2acec", "attempt_id": "c2304bb7b0ed", "run_group_id": "E4_core:A4|closed|qwen3b:h4", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:14:50.625437Z", "task_family": "synthetic", "document_id": "syn_03009_2000", "document_hash": "6ed94dcf728255bd7a7d11c97c362930728fb81a3e91f1906eb6eec86e0fedb7", "question_id": "q11_aggregation", "split": "test", "cluster_id": "syn_03009_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 4, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1976, "query_tokens": 600, "message_tokens": 506, "message_utf8_bytes": 1103, "cache_tensor_bytes": null, "serialized_payload_bytes": 1103, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 87833.84841599036, "hops": {"1": 22747.736499994062, "2": 21706.068499945104, "3": 21679.54479099717, "4": 21700.498625054024}, "receiver_prefill_ms": 521.3217079872265, "ttft_ms": 36.85599996242672, "decode_ms": 211.11750009004027}, "total_wall_ms": 88603.76058297697, "ttft_ms": 36.85599996242672, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 7}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "5781336", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 26, "retention_rate": 0.4, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.75, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1103}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 26, "retention_rate": 0.4, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.75, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1123}, "2": {"facts_total": 65, "facts_retained": 26, "retention_rate": 0.4, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.75, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1103}, "3": {"facts_total": 65, "facts_retained": 26, "retention_rate": 0.4, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.75, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1103}, "4": {"facts_total": 65, "facts_retained": 26, "retention_rate": 0.4, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.75, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1103}}}, "reference_key_id": "keys/synthetic/syn_03009_2000.json", "task_scores": {"correct": 0.0, "abstained": 0.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "92c6e2fac34ed63b0b281363062a4533def3743816eb87bae8df51c6c25e46c5", "output_hash": "a365e04f74ba5baeccdd9e30dc27a1c5ff916597b828d874e63884b2090a60e2"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "bc07951d3455a1a0c805242b", "attempt_id": "46047e7d95d7", "run_group_id": "E4_core:A4|closed|qwen3b:h4", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:14:51.413525Z", "task_family": "synthetic", "document_id": "syn_03009_2000", "document_hash": "6ed94dcf728255bd7a7d11c97c362930728fb81a3e91f1906eb6eec86e0fedb7", "question_id": "q12_unit_conversion", "split": "test", "cluster_id": "syn_03009_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 4, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1976, "query_tokens": 598, "message_tokens": 506, "message_utf8_bytes": 1103, "cache_tensor_bytes": null, "serialized_payload_bytes": 1103, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 87833.84841599036, "hops": {"1": 22747.736499994062, "2": 21706.068499945104, "3": 21679.54479099717, "4": 21700.498625054024}, "receiver_prefill_ms": 404.31962499860674, "ttft_ms": 33.65741693414748, "decode_ms": 30.26887495070696}, "total_wall_ms": 88302.66308202408, "ttft_ms": 33.65741693414748, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 26, "retention_rate": 0.4, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.75, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1103}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 26, "retention_rate": 0.4, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.75, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1123}, "2": {"facts_total": 65, "facts_retained": 26, "retention_rate": 0.4, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.75, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1103}, "3": {"facts_total": 65, "facts_retained": 26, "retention_rate": 0.4, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.75, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1103}, "4": {"facts_total": 65, "facts_retained": 26, "retention_rate": 0.4, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.75, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1103}}}, "reference_key_id": "keys/synthetic/syn_03009_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "92c6e2fac34ed63b0b281363062a4533def3743816eb87bae8df51c6c25e46c5", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "59a21c829996018ecc194804", "attempt_id": "9fb07535dd7c", "run_group_id": "E4_core:A4|closed|qwen3b:h4", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:14:51.899373Z", "task_family": "synthetic", "document_id": "syn_03009_2000", "document_hash": "6ed94dcf728255bd7a7d11c97c362930728fb81a3e91f1906eb6eec86e0fedb7", "question_id": "q13_unanswerable", "split": "test", "cluster_id": "syn_03009_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 4, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1976, "query_tokens": 602, "message_tokens": 506, "message_utf8_bytes": 1103, "cache_tensor_bytes": null, "serialized_payload_bytes": 1103, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 87833.84841599036, "hops": {"1": 22747.736499994062, "2": 21706.068499945104, "3": 21679.54479099717, "4": 21700.498625054024}, "receiver_prefill_ms": 350.885417079553, "ttft_ms": 39.73654203582555, "decode_ms": 63.92375007271767}, "total_wall_ms": 88289.04424898792, "ttft_ms": 39.73654203582555, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 26, "retention_rate": 0.4, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.75, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1103}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 26, "retention_rate": 0.4, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.75, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1123}, "2": {"facts_total": 65, "facts_retained": 26, "retention_rate": 0.4, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.75, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1103}, "3": {"facts_total": 65, "facts_retained": 26, "retention_rate": 0.4, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.75, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1103}, "4": {"facts_total": 65, "facts_retained": 26, "retention_rate": 0.4, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.75, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1103}}}, "reference_key_id": "keys/synthetic/syn_03009_2000.json", "task_scores": {"correct": 1.0, "abstained": 1.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "92c6e2fac34ed63b0b281363062a4533def3743816eb87bae8df51c6c25e46c5", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "baf5bf1ae2bdfa5fbb8bc385", "attempt_id": "e73b6cf17f39", "run_group_id": "E4_core:A4|closed|qwen3b:h4", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:14:52.371430Z", "task_family": "synthetic", "document_id": "syn_03009_2000", "document_hash": "6ed94dcf728255bd7a7d11c97c362930728fb81a3e91f1906eb6eec86e0fedb7", "question_id": "q14_unanswerable", "split": "test", "cluster_id": "syn_03009_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 4, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1976, "query_tokens": 602, "message_tokens": 506, "message_utf8_bytes": 1103, "cache_tensor_bytes": null, "serialized_payload_bytes": 1103, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 87833.84841599036, "hops": {"1": 22747.736499994062, "2": 21706.068499945104, "3": 21679.54479099717, "4": 21700.498625054024}, "receiver_prefill_ms": 518.9778750063851, "ttft_ms": 39.27512501832098, "decode_ms": 22.109749959781766}, "total_wall_ms": 88414.89329107571, "ttft_ms": 39.27512501832098, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 26, "retention_rate": 0.4, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.75, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1103}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 26, "retention_rate": 0.4, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.75, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1123}, "2": {"facts_total": 65, "facts_retained": 26, "retention_rate": 0.4, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.75, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1103}, "3": {"facts_total": 65, "facts_retained": 26, "retention_rate": 0.4, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.75, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1103}, "4": {"facts_total": 65, "facts_retained": 26, "retention_rate": 0.4, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.75, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1103}}}, "reference_key_id": "keys/synthetic/syn_03009_2000.json", "task_scores": {"correct": 1.0, "abstained": 1.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "92c6e2fac34ed63b0b281363062a4533def3743816eb87bae8df51c6c25e46c5", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "10664c72a23e00ce7b34957c", "attempt_id": "8e4bd9965ecd", "run_group_id": "E4_core:A4|closed|qwen3b:h8", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:16:20.219170Z", "task_family": "synthetic", "document_id": "syn_03009_2000", "document_hash": "6ed94dcf728255bd7a7d11c97c362930728fb81a3e91f1906eb6eec86e0fedb7", "question_id": "q00_single_fact", "split": "test", "cluster_id": "syn_03009_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 8, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1976, "query_tokens": 590, "message_tokens": 506, "message_utf8_bytes": 1103, "cache_tensor_bytes": null, "serialized_payload_bytes": 1103, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 175077.7335400926, "hops": {"1": 22747.736499994062, "2": 21706.068499945104, "3": 21679.54479099717, "4": 21700.498625054024, "5": 21832.789166015573, "6": 21813.687499961816, "7": 21884.764833026566, "8": 21712.64362509828}, "receiver_prefill_ms": 407.66274998895824, "ttft_ms": 34.80345802381635, "decode_ms": 423.45204192679375}, "total_wall_ms": 175944.34579019435, "ttft_ms": 34.80345802381635, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 10}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "PA-37F-6227", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 26, "retention_rate": 0.4, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.75, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1103}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 26, "retention_rate": 0.4, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.75, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1123}, "2": {"facts_total": 65, "facts_retained": 26, "retention_rate": 0.4, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.75, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1103}, "3": {"facts_total": 65, "facts_retained": 26, "retention_rate": 0.4, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.75, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1103}, "4": {"facts_total": 65, "facts_retained": 26, "retention_rate": 0.4, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.75, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1103}, "5": {"facts_total": 65, "facts_retained": 26, "retention_rate": 0.4, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.75, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1103}, "6": {"facts_total": 65, "facts_retained": 26, "retention_rate": 0.4, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.75, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1103}, "7": {"facts_total": 65, "facts_retained": 26, "retention_rate": 0.4, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.75, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1103}, "8": {"facts_total": 65, "facts_retained": 26, "retention_rate": 0.4, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.75, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1103}}}, "reference_key_id": "keys/synthetic/syn_03009_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "92c6e2fac34ed63b0b281363062a4533def3743816eb87bae8df51c6c25e46c5", "output_hash": "5b7645cf6e8140f4c0bf6ed4ad76121cf2924fe7e666d92161812527b86f4409"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "38be146320c36994527ab33e", "attempt_id": "71abeee3ab0a", "run_group_id": "E4_core:A4|closed|qwen3b:h8", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:16:21.103688Z", "task_family": "synthetic", "document_id": "syn_03009_2000", "document_hash": "6ed94dcf728255bd7a7d11c97c362930728fb81a3e91f1906eb6eec86e0fedb7", "question_id": "q01_single_fact", "split": "test", "cluster_id": "syn_03009_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 8, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1976, "query_tokens": 595, "message_tokens": 506, "message_utf8_bytes": 1103, "cache_tensor_bytes": null, "serialized_payload_bytes": 1103, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 175077.7335400926, "hops": {"1": 22747.736499994062, "2": 21706.068499945104, "3": 21679.54479099717, "4": 21700.498625054024, "5": 21832.789166015573, "6": 21813.687499961816, "7": 21884.764833026566, "8": 21712.64362509828}, "receiver_prefill_ms": 504.2896249797195, "ttft_ms": 26.297249947674572, "decode_ms": 40.70691706147045}, "total_wall_ms": 175649.7502900893, "ttft_ms": 26.297249947674572, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 26, "retention_rate": 0.4, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.75, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1103}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 26, "retention_rate": 0.4, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.75, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1123}, "2": {"facts_total": 65, "facts_retained": 26, "retention_rate": 0.4, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.75, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1103}, "3": {"facts_total": 65, "facts_retained": 26, "retention_rate": 0.4, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.75, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1103}, "4": {"facts_total": 65, "facts_retained": 26, "retention_rate": 0.4, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.75, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1103}, "5": {"facts_total": 65, "facts_retained": 26, "retention_rate": 0.4, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.75, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1103}, "6": {"facts_total": 65, "facts_retained": 26, "retention_rate": 0.4, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.75, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1103}, "7": {"facts_total": 65, "facts_retained": 26, "retention_rate": 0.4, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.75, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1103}, "8": {"facts_total": 65, "facts_retained": 26, "retention_rate": 0.4, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.75, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1103}}}, "reference_key_id": "keys/synthetic/syn_03009_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "92c6e2fac34ed63b0b281363062a4533def3743816eb87bae8df51c6c25e46c5", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "7171581c8aaa4ef30e6fb2f6", "attempt_id": "a3acd473c428", "run_group_id": "E4_core:A4|closed|qwen3b:h8", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:16:21.694096Z", "task_family": "synthetic", "document_id": "syn_03009_2000", "document_hash": "6ed94dcf728255bd7a7d11c97c362930728fb81a3e91f1906eb6eec86e0fedb7", "question_id": "q02_exact_copy", "split": "test", "cluster_id": "syn_03009_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 8, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1976, "query_tokens": 593, "message_tokens": 506, "message_utf8_bytes": 1103, "cache_tensor_bytes": null, "serialized_payload_bytes": 1103, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 175077.7335400926, "hops": {"1": 22747.736499994062, "2": 21706.068499945104, "3": 21679.54479099717, "4": 21700.498625054024, "5": 21832.789166015573, "6": 21813.687499961816, "7": 21884.764833026566, "8": 21712.64362509828}, "receiver_prefill_ms": 409.1677500400692, "ttft_ms": 41.197124985046685, "decode_ms": 33.43987499829382}, "total_wall_ms": 175562.15045705903, "ttft_ms": 41.197124985046685, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 26, "retention_rate": 0.4, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.75, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1103}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 26, "retention_rate": 0.4, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.75, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1123}, "2": {"facts_total": 65, "facts_retained": 26, "retention_rate": 0.4, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.75, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1103}, "3": {"facts_total": 65, "facts_retained": 26, "retention_rate": 0.4, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.75, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1103}, "4": {"facts_total": 65, "facts_retained": 26, "retention_rate": 0.4, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.75, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1103}, "5": {"facts_total": 65, "facts_retained": 26, "retention_rate": 0.4, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.75, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1103}, "6": {"facts_total": 65, "facts_retained": 26, "retention_rate": 0.4, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.75, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1103}, "7": {"facts_total": 65, "facts_retained": 26, "retention_rate": 0.4, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.75, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1103}, "8": {"facts_total": 65, "facts_retained": 26, "retention_rate": 0.4, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.75, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1103}}}, "reference_key_id": "keys/synthetic/syn_03009_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "92c6e2fac34ed63b0b281363062a4533def3743816eb87bae8df51c6c25e46c5", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "d7a019dce38a4479458476ce", "attempt_id": "0e6b6404c1f1", "run_group_id": "E4_core:A4|closed|qwen3b:h8", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:16:22.196097Z", "task_family": "synthetic", "document_id": "syn_03009_2000", "document_hash": "6ed94dcf728255bd7a7d11c97c362930728fb81a3e91f1906eb6eec86e0fedb7", "question_id": "q03_exact_copy", "split": "test", "cluster_id": "syn_03009_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 8, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1976, "query_tokens": 589, "message_tokens": 506, "message_utf8_bytes": 1103, "cache_tensor_bytes": null, "serialized_payload_bytes": 1103, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 175077.7335400926, "hops": {"1": 22747.736499994062, "2": 21706.068499945104, "3": 21679.54479099717, "4": 21700.498625054024, "5": 21832.789166015573, "6": 21813.687499961816, "7": 21884.764833026566, "8": 21712.64362509828}, "receiver_prefill_ms": 347.8298330446705, "ttft_ms": 45.4399170121178, "decode_ms": 345.5344579415396}, "total_wall_ms": 175817.2225401504, "ttft_ms": 45.4399170121178, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 5}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "Idris Bergström", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 26, "retention_rate": 0.4, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.75, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1103}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 26, "retention_rate": 0.4, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.75, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1123}, "2": {"facts_total": 65, "facts_retained": 26, "retention_rate": 0.4, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.75, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1103}, "3": {"facts_total": 65, "facts_retained": 26, "retention_rate": 0.4, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.75, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1103}, "4": {"facts_total": 65, "facts_retained": 26, "retention_rate": 0.4, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.75, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1103}, "5": {"facts_total": 65, "facts_retained": 26, "retention_rate": 0.4, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.75, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1103}, "6": {"facts_total": 65, "facts_retained": 26, "retention_rate": 0.4, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.75, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1103}, "7": {"facts_total": 65, "facts_retained": 26, "retention_rate": 0.4, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.75, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1103}, "8": {"facts_total": 65, "facts_retained": 26, "retention_rate": 0.4, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.75, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1103}}}, "reference_key_id": "keys/synthetic/syn_03009_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "92c6e2fac34ed63b0b281363062a4533def3743816eb87bae8df51c6c25e46c5", "output_hash": "fd88832c9fd0ab175818d76d875d70156225bd4b34ec401514546ef70f4f826c"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "f419dc7fc5fb3ed9c672f74a", "attempt_id": "879af95f1aec", "run_group_id": "E4_core:A4|closed|qwen3b:h8", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:16:22.952187Z", "task_family": "synthetic", "document_id": "syn_03009_2000", "document_hash": "6ed94dcf728255bd7a7d11c97c362930728fb81a3e91f1906eb6eec86e0fedb7", "question_id": "q04_multi_hop", "split": "test", "cluster_id": "syn_03009_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 8, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1976, "query_tokens": 597, "message_tokens": 506, "message_utf8_bytes": 1103, "cache_tensor_bytes": null, "serialized_payload_bytes": 1103, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 175077.7335400926, "hops": {"1": 22747.736499994062, "2": 21706.068499945104, "3": 21679.54479099717, "4": 21700.498625054024, "5": 21832.789166015573, "6": 21813.687499961816, "7": 21884.764833026566, "8": 21712.64362509828}, "receiver_prefill_ms": 460.8162089716643, "ttft_ms": 32.94137504417449, "decode_ms": 232.85045800730586}, "total_wall_ms": 175805.11233210564, "ttft_ms": 32.94137504417449, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 7}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "4670383", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 26, "retention_rate": 0.4, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.75, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1103}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 26, "retention_rate": 0.4, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.75, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1123}, "2": {"facts_total": 65, "facts_retained": 26, "retention_rate": 0.4, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.75, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1103}, "3": {"facts_total": 65, "facts_retained": 26, "retention_rate": 0.4, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.75, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1103}, "4": {"facts_total": 65, "facts_retained": 26, "retention_rate": 0.4, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.75, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1103}, "5": {"facts_total": 65, "facts_retained": 26, "retention_rate": 0.4, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.75, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1103}, "6": {"facts_total": 65, "facts_retained": 26, "retention_rate": 0.4, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.75, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1103}, "7": {"facts_total": 65, "facts_retained": 26, "retention_rate": 0.4, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.75, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1103}, "8": {"facts_total": 65, "facts_retained": 26, "retention_rate": 0.4, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.75, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1103}}}, "reference_key_id": "keys/synthetic/syn_03009_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "92c6e2fac34ed63b0b281363062a4533def3743816eb87bae8df51c6c25e46c5", "output_hash": "19ae8c357e5500bf494a6163bfbd11380b2996e8ea532a2c64f4a599464820f7"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "a933866d7ac6548b73f286f1", "attempt_id": "67ccf90443d9", "run_group_id": "E4_core:A4|closed|qwen3b:h8", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:16:23.697535Z", "task_family": "synthetic", "document_id": "syn_03009_2000", "document_hash": "6ed94dcf728255bd7a7d11c97c362930728fb81a3e91f1906eb6eec86e0fedb7", "question_id": "q05_multi_hop", "split": "test", "cluster_id": "syn_03009_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 8, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1976, "query_tokens": 600, "message_tokens": 506, "message_utf8_bytes": 1103, "cache_tensor_bytes": null, "serialized_payload_bytes": 1103, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 175077.7335400926, "hops": {"1": 22747.736499994062, "2": 21706.068499945104, "3": 21679.54479099717, "4": 21700.498625054024, "5": 21832.789166015573, "6": 21813.687499961816, "7": 21884.764833026566, "8": 21712.64362509828}, "receiver_prefill_ms": 372.60258302558213, "ttft_ms": 41.51608399115503, "decode_ms": 79.3677499750629}, "total_wall_ms": 175571.92024809774, "ttft_ms": 41.51608399115503, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 26, "retention_rate": 0.4, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.75, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1103}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 26, "retention_rate": 0.4, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.75, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1123}, "2": {"facts_total": 65, "facts_retained": 26, "retention_rate": 0.4, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.75, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1103}, "3": {"facts_total": 65, "facts_retained": 26, "retention_rate": 0.4, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.75, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1103}, "4": {"facts_total": 65, "facts_retained": 26, "retention_rate": 0.4, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.75, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1103}, "5": {"facts_total": 65, "facts_retained": 26, "retention_rate": 0.4, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.75, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1103}, "6": {"facts_total": 65, "facts_retained": 26, "retention_rate": 0.4, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.75, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1103}, "7": {"facts_total": 65, "facts_retained": 26, "retention_rate": 0.4, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.75, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1103}, "8": {"facts_total": 65, "facts_retained": 26, "retention_rate": 0.4, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.75, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1103}}}, "reference_key_id": "keys/synthetic/syn_03009_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "92c6e2fac34ed63b0b281363062a4533def3743816eb87bae8df51c6c25e46c5", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "7f506bd9e5927d3f21bdce6b", "attempt_id": "0480f5d36d83", "run_group_id": "E4_core:A4|closed|qwen3b:h8", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:16:24.209127Z", "task_family": "synthetic", "document_id": "syn_03009_2000", "document_hash": "6ed94dcf728255bd7a7d11c97c362930728fb81a3e91f1906eb6eec86e0fedb7", "question_id": "q06_negation", "split": "test", "cluster_id": "syn_03009_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 8, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1976, "query_tokens": 590, "message_tokens": 506, "message_utf8_bytes": 1103, "cache_tensor_bytes": null, "serialized_payload_bytes": 1103, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 175077.7335400926, "hops": {"1": 22747.736499994062, "2": 21706.068499945104, "3": 21679.54479099717, "4": 21700.498625054024, "5": 21832.789166015573, "6": 21813.687499961816, "7": 21884.764833026566, "8": 21712.64362509828}, "receiver_prefill_ms": 563.2275409298018, "ttft_ms": 22.736666025593877, "decode_ms": 38.014834048226476}, "total_wall_ms": 175702.44658202864, "ttft_ms": 22.736666025593877, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 26, "retention_rate": 0.4, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.75, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1103}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 26, "retention_rate": 0.4, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.75, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1123}, "2": {"facts_total": 65, "facts_retained": 26, "retention_rate": 0.4, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.75, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1103}, "3": {"facts_total": 65, "facts_retained": 26, "retention_rate": 0.4, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.75, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1103}, "4": {"facts_total": 65, "facts_retained": 26, "retention_rate": 0.4, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.75, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1103}, "5": {"facts_total": 65, "facts_retained": 26, "retention_rate": 0.4, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.75, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1103}, "6": {"facts_total": 65, "facts_retained": 26, "retention_rate": 0.4, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.75, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1103}, "7": {"facts_total": 65, "facts_retained": 26, "retention_rate": 0.4, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.75, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1103}, "8": {"facts_total": 65, "facts_retained": 26, "retention_rate": 0.4, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.75, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1103}}}, "reference_key_id": "keys/synthetic/syn_03009_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "92c6e2fac34ed63b0b281363062a4533def3743816eb87bae8df51c6c25e46c5", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "63107fc4ef045736fe2b28e5", "attempt_id": "1af7cbdea700", "run_group_id": "E4_core:A4|closed|qwen3b:h8", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:16:24.852610Z", "task_family": "synthetic", "document_id": "syn_03009_2000", "document_hash": "6ed94dcf728255bd7a7d11c97c362930728fb81a3e91f1906eb6eec86e0fedb7", "question_id": "q07_negation", "split": "test", "cluster_id": "syn_03009_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 8, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1976, "query_tokens": 589, "message_tokens": 506, "message_utf8_bytes": 1103, "cache_tensor_bytes": null, "serialized_payload_bytes": 1103, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 175077.7335400926, "hops": {"1": 22747.736499994062, "2": 21706.068499945104, "3": 21679.54479099717, "4": 21700.498625054024, "5": 21832.789166015573, "6": 21813.687499961816, "7": 21884.764833026566, "8": 21712.64362509828}, "receiver_prefill_ms": 414.28162495139986, "ttft_ms": 26.470209006220102, "decode_ms": 36.70170798432082}, "total_wall_ms": 175555.7645400986, "ttft_ms": 26.470209006220102, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 26, "retention_rate": 0.4, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.75, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1103}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 26, "retention_rate": 0.4, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.75, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1123}, "2": {"facts_total": 65, "facts_retained": 26, "retention_rate": 0.4, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.75, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1103}, "3": {"facts_total": 65, "facts_retained": 26, "retention_rate": 0.4, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.75, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1103}, "4": {"facts_total": 65, "facts_retained": 26, "retention_rate": 0.4, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.75, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1103}, "5": {"facts_total": 65, "facts_retained": 26, "retention_rate": 0.4, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.75, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1103}, "6": {"facts_total": 65, "facts_retained": 26, "retention_rate": 0.4, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.75, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1103}, "7": {"facts_total": 65, "facts_retained": 26, "retention_rate": 0.4, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.75, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1103}, "8": {"facts_total": 65, "facts_retained": 26, "retention_rate": 0.4, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.75, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1103}}}, "reference_key_id": "keys/synthetic/syn_03009_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "92c6e2fac34ed63b0b281363062a4533def3743816eb87bae8df51c6c25e46c5", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "dcb711932ff216b3a7a5fdd6", "attempt_id": "68383f22b3fa", "run_group_id": "E4_core:A4|closed|qwen3b:h8", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:16:25.348650Z", "task_family": "synthetic", "document_id": "syn_03009_2000", "document_hash": "6ed94dcf728255bd7a7d11c97c362930728fb81a3e91f1906eb6eec86e0fedb7", "question_id": "q08_revision", "split": "test", "cluster_id": "syn_03009_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 8, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1976, "query_tokens": 607, "message_tokens": 506, "message_utf8_bytes": 1103, "cache_tensor_bytes": null, "serialized_payload_bytes": 1103, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 175077.7335400926, "hops": {"1": 22747.736499994062, "2": 21706.068499945104, "3": 21679.54479099717, "4": 21700.498625054024, "5": 21832.789166015573, "6": 21813.687499961816, "7": 21884.764833026566, "8": 21712.64362509828}, "receiver_prefill_ms": 385.81333309412, "ttft_ms": 22.863625083118677, "decode_ms": 560.3419999824837}, "total_wall_ms": 176047.5948730018, "ttft_ms": 22.863625083118677, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 8}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "10047105", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 26, "retention_rate": 0.4, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.75, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1103}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 26, "retention_rate": 0.4, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.75, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1123}, "2": {"facts_total": 65, "facts_retained": 26, "retention_rate": 0.4, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.75, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1103}, "3": {"facts_total": 65, "facts_retained": 26, "retention_rate": 0.4, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.75, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1103}, "4": {"facts_total": 65, "facts_retained": 26, "retention_rate": 0.4, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.75, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1103}, "5": {"facts_total": 65, "facts_retained": 26, "retention_rate": 0.4, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.75, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1103}, "6": {"facts_total": 65, "facts_retained": 26, "retention_rate": 0.4, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.75, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1103}, "7": {"facts_total": 65, "facts_retained": 26, "retention_rate": 0.4, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.75, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1103}, "8": {"facts_total": 65, "facts_retained": 26, "retention_rate": 0.4, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.75, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1103}}}, "reference_key_id": "keys/synthetic/syn_03009_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "92c6e2fac34ed63b0b281363062a4533def3743816eb87bae8df51c6c25e46c5", "output_hash": "bea99610d1e3e75a7f20939faf8e6824e02237b156dacd6442ba2a7e02617510"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "79777b6cc7aa80e3fbce3e03", "attempt_id": "c47a488af537", "run_group_id": "E4_core:A4|closed|qwen3b:h8", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:16:26.335451Z", "task_family": "synthetic", "document_id": "syn_03009_2000", "document_hash": "6ed94dcf728255bd7a7d11c97c362930728fb81a3e91f1906eb6eec86e0fedb7", "question_id": "q09_revision", "split": "test", "cluster_id": "syn_03009_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 8, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1976, "query_tokens": 600, "message_tokens": 506, "message_utf8_bytes": 1103, "cache_tensor_bytes": null, "serialized_payload_bytes": 1103, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 175077.7335400926, "hops": {"1": 22747.736499994062, "2": 21706.068499945104, "3": 21679.54479099717, "4": 21700.498625054024, "5": 21832.789166015573, "6": 21813.687499961816, "7": 21884.764833026566, "8": 21712.64362509828}, "receiver_prefill_ms": 390.01766592264175, "ttft_ms": 22.908708080649376, "decode_ms": 277.2236670134589}, "total_wall_ms": 175768.5692490777, "ttft_ms": 22.908708080649376, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 8}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "10047105", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 26, "retention_rate": 0.4, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.75, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1103}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 26, "retention_rate": 0.4, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.75, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1123}, "2": {"facts_total": 65, "facts_retained": 26, "retention_rate": 0.4, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.75, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1103}, "3": {"facts_total": 65, "facts_retained": 26, "retention_rate": 0.4, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.75, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1103}, "4": {"facts_total": 65, "facts_retained": 26, "retention_rate": 0.4, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.75, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1103}, "5": {"facts_total": 65, "facts_retained": 26, "retention_rate": 0.4, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.75, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1103}, "6": {"facts_total": 65, "facts_retained": 26, "retention_rate": 0.4, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.75, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1103}, "7": {"facts_total": 65, "facts_retained": 26, "retention_rate": 0.4, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.75, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1103}, "8": {"facts_total": 65, "facts_retained": 26, "retention_rate": 0.4, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.75, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1103}}}, "reference_key_id": "keys/synthetic/syn_03009_2000.json", "task_scores": {"correct": 0.0, "abstained": 0.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "92c6e2fac34ed63b0b281363062a4533def3743816eb87bae8df51c6c25e46c5", "output_hash": "bea99610d1e3e75a7f20939faf8e6824e02237b156dacd6442ba2a7e02617510"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "5ce2ae52066a6c95ac92da39", "attempt_id": "88da4afbcb10", "run_group_id": "E4_core:A4|closed|qwen3b:h8", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:16:27.043922Z", "task_family": "synthetic", "document_id": "syn_03009_2000", "document_hash": "6ed94dcf728255bd7a7d11c97c362930728fb81a3e91f1906eb6eec86e0fedb7", "question_id": "q10_temporal", "split": "test", "cluster_id": "syn_03009_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 8, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1976, "query_tokens": 609, "message_tokens": 506, "message_utf8_bytes": 1103, "cache_tensor_bytes": null, "serialized_payload_bytes": 1103, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 175077.7335400926, "hops": {"1": 22747.736499994062, "2": 21706.068499945104, "3": 21679.54479099717, "4": 21700.498625054024, "5": 21832.789166015573, "6": 21813.687499961816, "7": 21884.764833026566, "8": 21712.64362509828}, "receiver_prefill_ms": 362.8030000254512, "ttft_ms": 52.09000001195818, "decode_ms": 42.35133295878768}, "total_wall_ms": 175535.90045613237, "ttft_ms": 52.09000001195818, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 26, "retention_rate": 0.4, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.75, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1103}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 26, "retention_rate": 0.4, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.75, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1123}, "2": {"facts_total": 65, "facts_retained": 26, "retention_rate": 0.4, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.75, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1103}, "3": {"facts_total": 65, "facts_retained": 26, "retention_rate": 0.4, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.75, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1103}, "4": {"facts_total": 65, "facts_retained": 26, "retention_rate": 0.4, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.75, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1103}, "5": {"facts_total": 65, "facts_retained": 26, "retention_rate": 0.4, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.75, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1103}, "6": {"facts_total": 65, "facts_retained": 26, "retention_rate": 0.4, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.75, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1103}, "7": {"facts_total": 65, "facts_retained": 26, "retention_rate": 0.4, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.75, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1103}, "8": {"facts_total": 65, "facts_retained": 26, "retention_rate": 0.4, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.75, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1103}}}, "reference_key_id": "keys/synthetic/syn_03009_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "92c6e2fac34ed63b0b281363062a4533def3743816eb87bae8df51c6c25e46c5", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "9f5279ead0b827027622e4d1", "attempt_id": "eeff03444a65", "run_group_id": "E4_core:A4|closed|qwen3b:h8", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:16:27.520416Z", "task_family": "synthetic", "document_id": "syn_03009_2000", "document_hash": "6ed94dcf728255bd7a7d11c97c362930728fb81a3e91f1906eb6eec86e0fedb7", "question_id": "q11_aggregation", "split": "test", "cluster_id": "syn_03009_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 8, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1976, "query_tokens": 600, "message_tokens": 506, "message_utf8_bytes": 1103, "cache_tensor_bytes": null, "serialized_payload_bytes": 1103, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 175077.7335400926, "hops": {"1": 22747.736499994062, "2": 21706.068499945104, "3": 21679.54479099717, "4": 21700.498625054024, "5": 21832.789166015573, "6": 21813.687499961816, "7": 21884.764833026566, "8": 21712.64362509828}, "receiver_prefill_ms": 411.78958292584866, "ttft_ms": 123.69354104157537, "decode_ms": 241.167499916628}, "total_wall_ms": 175855.02399806865, "ttft_ms": 123.69354104157537, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 7}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "5247196", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 26, "retention_rate": 0.4, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.75, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1103}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 26, "retention_rate": 0.4, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.75, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1123}, "2": {"facts_total": 65, "facts_retained": 26, "retention_rate": 0.4, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.75, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1103}, "3": {"facts_total": 65, "facts_retained": 26, "retention_rate": 0.4, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.75, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1103}, "4": {"facts_total": 65, "facts_retained": 26, "retention_rate": 0.4, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.75, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1103}, "5": {"facts_total": 65, "facts_retained": 26, "retention_rate": 0.4, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.75, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1103}, "6": {"facts_total": 65, "facts_retained": 26, "retention_rate": 0.4, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.75, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1103}, "7": {"facts_total": 65, "facts_retained": 26, "retention_rate": 0.4, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.75, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1103}, "8": {"facts_total": 65, "facts_retained": 26, "retention_rate": 0.4, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.75, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1103}}}, "reference_key_id": "keys/synthetic/syn_03009_2000.json", "task_scores": {"correct": 0.0, "abstained": 0.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "92c6e2fac34ed63b0b281363062a4533def3743816eb87bae8df51c6c25e46c5", "output_hash": "3ac48b1390a5593419a76c6409f1c54c01affa22edaa1be1850800277310a17c"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "b9c571090040f68a08578e55", "attempt_id": "d18da9482009", "run_group_id": "E4_core:A4|closed|qwen3b:h8", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:16:28.314851Z", "task_family": "synthetic", "document_id": "syn_03009_2000", "document_hash": "6ed94dcf728255bd7a7d11c97c362930728fb81a3e91f1906eb6eec86e0fedb7", "question_id": "q12_unit_conversion", "split": "test", "cluster_id": "syn_03009_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 8, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1976, "query_tokens": 598, "message_tokens": 506, "message_utf8_bytes": 1103, "cache_tensor_bytes": null, "serialized_payload_bytes": 1103, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 175077.7335400926, "hops": {"1": 22747.736499994062, "2": 21706.068499945104, "3": 21679.54479099717, "4": 21700.498625054024, "5": 21832.789166015573, "6": 21813.687499961816, "7": 21884.764833026566, "8": 21712.64362509828}, "receiver_prefill_ms": 399.09929200075567, "ttft_ms": 45.39187496993691, "decode_ms": 23.928957991302013}, "total_wall_ms": 175546.91841511521, "ttft_ms": 45.39187496993691, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 26, "retention_rate": 0.4, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.75, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1103}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 26, "retention_rate": 0.4, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.75, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1123}, "2": {"facts_total": 65, "facts_retained": 26, "retention_rate": 0.4, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.75, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1103}, "3": {"facts_total": 65, "facts_retained": 26, "retention_rate": 0.4, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.75, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1103}, "4": {"facts_total": 65, "facts_retained": 26, "retention_rate": 0.4, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.75, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1103}, "5": {"facts_total": 65, "facts_retained": 26, "retention_rate": 0.4, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.75, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1103}, "6": {"facts_total": 65, "facts_retained": 26, "retention_rate": 0.4, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.75, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1103}, "7": {"facts_total": 65, "facts_retained": 26, "retention_rate": 0.4, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.75, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1103}, "8": {"facts_total": 65, "facts_retained": 26, "retention_rate": 0.4, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.75, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1103}}}, "reference_key_id": "keys/synthetic/syn_03009_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "92c6e2fac34ed63b0b281363062a4533def3743816eb87bae8df51c6c25e46c5", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "97c51db7054a2cb0a3eddb71", "attempt_id": "3e582dd2d201", "run_group_id": "E4_core:A4|closed|qwen3b:h8", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:16:28.801679Z", "task_family": "synthetic", "document_id": "syn_03009_2000", "document_hash": "6ed94dcf728255bd7a7d11c97c362930728fb81a3e91f1906eb6eec86e0fedb7", "question_id": "q13_unanswerable", "split": "test", "cluster_id": "syn_03009_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 8, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1976, "query_tokens": 602, "message_tokens": 506, "message_utf8_bytes": 1103, "cache_tensor_bytes": null, "serialized_payload_bytes": 1103, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 175077.7335400926, "hops": {"1": 22747.736499994062, "2": 21706.068499945104, "3": 21679.54479099717, "4": 21700.498625054024, "5": 21832.789166015573, "6": 21813.687499961816, "7": 21884.764833026566, "8": 21712.64362509828}, "receiver_prefill_ms": 364.2659579636529, "ttft_ms": 49.613542039878666, "decode_ms": 48.34520793519914}, "total_wall_ms": 175540.75837309938, "ttft_ms": 49.613542039878666, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 26, "retention_rate": 0.4, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.75, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1103}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 26, "retention_rate": 0.4, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.75, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1123}, "2": {"facts_total": 65, "facts_retained": 26, "retention_rate": 0.4, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.75, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1103}, "3": {"facts_total": 65, "facts_retained": 26, "retention_rate": 0.4, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.75, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1103}, "4": {"facts_total": 65, "facts_retained": 26, "retention_rate": 0.4, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.75, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1103}, "5": {"facts_total": 65, "facts_retained": 26, "retention_rate": 0.4, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.75, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1103}, "6": {"facts_total": 65, "facts_retained": 26, "retention_rate": 0.4, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.75, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1103}, "7": {"facts_total": 65, "facts_retained": 26, "retention_rate": 0.4, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.75, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1103}, "8": {"facts_total": 65, "facts_retained": 26, "retention_rate": 0.4, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.75, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1103}}}, "reference_key_id": "keys/synthetic/syn_03009_2000.json", "task_scores": {"correct": 1.0, "abstained": 1.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "92c6e2fac34ed63b0b281363062a4533def3743816eb87bae8df51c6c25e46c5", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "9b181bc6b48c874e25d60a36", "attempt_id": "7463654fb62a", "run_group_id": "E4_core:A4|closed|qwen3b:h8", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:16:29.282193Z", "task_family": "synthetic", "document_id": "syn_03009_2000", "document_hash": "6ed94dcf728255bd7a7d11c97c362930728fb81a3e91f1906eb6eec86e0fedb7", "question_id": "q14_unanswerable", "split": "test", "cluster_id": "syn_03009_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 8, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1976, "query_tokens": 602, "message_tokens": 506, "message_utf8_bytes": 1103, "cache_tensor_bytes": null, "serialized_payload_bytes": 1103, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 175077.7335400926, "hops": {"1": 22747.736499994062, "2": 21706.068499945104, "3": 21679.54479099717, "4": 21700.498625054024, "5": 21832.789166015573, "6": 21813.687499961816, "7": 21884.764833026566, "8": 21712.64362509828}, "receiver_prefill_ms": 557.7695419779047, "ttft_ms": 35.47562495805323, "decode_ms": 23.877041065134108}, "total_wall_ms": 175695.6081649987, "ttft_ms": 35.47562495805323, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 26, "retention_rate": 0.4, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.75, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1103}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 26, "retention_rate": 0.4, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.75, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1123}, "2": {"facts_total": 65, "facts_retained": 26, "retention_rate": 0.4, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.75, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1103}, "3": {"facts_total": 65, "facts_retained": 26, "retention_rate": 0.4, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.75, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1103}, "4": {"facts_total": 65, "facts_retained": 26, "retention_rate": 0.4, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.75, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1103}, "5": {"facts_total": 65, "facts_retained": 26, "retention_rate": 0.4, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.75, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1103}, "6": {"facts_total": 65, "facts_retained": 26, "retention_rate": 0.4, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.75, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1103}, "7": {"facts_total": 65, "facts_retained": 26, "retention_rate": 0.4, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.75, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1103}, "8": {"facts_total": 65, "facts_retained": 26, "retention_rate": 0.4, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.75, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1103}}}, "reference_key_id": "keys/synthetic/syn_03009_2000.json", "task_scores": {"correct": 1.0, "abstained": 1.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "92c6e2fac34ed63b0b281363062a4533def3743816eb87bae8df51c6c25e46c5", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "af33cefca5c4c2d7ab6dafc8", "attempt_id": "82ee45348cc2", "run_group_id": "E4_core:A5|closed|qwen3b:h1", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:16:46.305879Z", "task_family": "synthetic", "document_id": "syn_03009_2000", "document_hash": "6ed94dcf728255bd7a7d11c97c362930728fb81a3e91f1906eb6eec86e0fedb7", "question_id": "q00_single_fact", "split": "test", "cluster_id": "syn_03009_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 1, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1976, "query_tokens": 318, "message_tokens": 234, "message_utf8_bytes": 698, "cache_tensor_bytes": null, "serialized_payload_bytes": 698, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 16386.116749956273, "hops": {"1": 16386.116749956273}, "receiver_prefill_ms": 390.859084087424, "ttft_ms": 30.823959037661552, "decode_ms": 180.415750015527}, "total_wall_ms": 16988.760374952108, "ttft_ms": 30.823959037661552, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 5}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "Milos Lindqvist", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 12, "retention_rate": 0.18461538461538463, "retention_by_kind": {"person": 1.0, "date": 0.0, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 698}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 12, "retention_rate": 0.18461538461538463, "retention_by_kind": {"person": 1.0, "date": 0.0, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 698}}}, "reference_key_id": "keys/synthetic/syn_03009_2000.json", "task_scores": {"correct": 0.0, "abstained": 0.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "b7aa2d9d5f9f3a06115e4e65c8ffaf4be2a07d6d067ceac89e49c6efeb6fdd5f", "output_hash": "f1a0ce3d66245c336fdc7ac1281d2b638395739df9d016b121be6f1b2512201e"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "3a5a443e5aeeb0561091b378", "attempt_id": "9fa39e51077b", "run_group_id": "E4_core:A5|closed|qwen3b:h1", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:16:46.926605Z", "task_family": "synthetic", "document_id": "syn_03009_2000", "document_hash": "6ed94dcf728255bd7a7d11c97c362930728fb81a3e91f1906eb6eec86e0fedb7", "question_id": "q01_single_fact", "split": "test", "cluster_id": "syn_03009_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 1, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1976, "query_tokens": 323, "message_tokens": 234, "message_utf8_bytes": 698, "cache_tensor_bytes": null, "serialized_payload_bytes": 698, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 16386.116749956273, "hops": {"1": 16386.116749956273}, "receiver_prefill_ms": 280.5695420829579, "ttft_ms": 45.510999974794686, "decode_ms": 22.823916980996728}, "total_wall_ms": 16735.509457997978, "ttft_ms": 45.510999974794686, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 12, "retention_rate": 0.18461538461538463, "retention_by_kind": {"person": 1.0, "date": 0.0, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 698}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 12, "retention_rate": 0.18461538461538463, "retention_by_kind": {"person": 1.0, "date": 0.0, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 698}}}, "reference_key_id": "keys/synthetic/syn_03009_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "b7aa2d9d5f9f3a06115e4e65c8ffaf4be2a07d6d067ceac89e49c6efeb6fdd5f", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "c0f3e34f1eaf0bd277f997f5", "attempt_id": "b702dfc08852", "run_group_id": "E4_core:A5|closed|qwen3b:h1", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:16:47.293645Z", "task_family": "synthetic", "document_id": "syn_03009_2000", "document_hash": "6ed94dcf728255bd7a7d11c97c362930728fb81a3e91f1906eb6eec86e0fedb7", "question_id": "q02_exact_copy", "split": "test", "cluster_id": "syn_03009_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 1, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1976, "query_tokens": 321, "message_tokens": 234, "message_utf8_bytes": 698, "cache_tensor_bytes": null, "serialized_payload_bytes": 698, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 16386.116749956273, "hops": {"1": 16386.116749956273}, "receiver_prefill_ms": 197.24354206118733, "ttft_ms": 37.425791029818356, "decode_ms": 59.76558290421963}, "total_wall_ms": 16681.130375014618, "ttft_ms": 37.425791029818356, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 12, "retention_rate": 0.18461538461538463, "retention_by_kind": {"person": 1.0, "date": 0.0, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 698}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 12, "retention_rate": 0.18461538461538463, "retention_by_kind": {"person": 1.0, "date": 0.0, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 698}}}, "reference_key_id": "keys/synthetic/syn_03009_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "b7aa2d9d5f9f3a06115e4e65c8ffaf4be2a07d6d067ceac89e49c6efeb6fdd5f", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "d9f7b3beb1b45b1f023fd7a4", "attempt_id": "d385bc3e5884", "run_group_id": "E4_core:A5|closed|qwen3b:h1", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:16:47.606050Z", "task_family": "synthetic", "document_id": "syn_03009_2000", "document_hash": "6ed94dcf728255bd7a7d11c97c362930728fb81a3e91f1906eb6eec86e0fedb7", "question_id": "q03_exact_copy", "split": "test", "cluster_id": "syn_03009_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 1, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1976, "query_tokens": 317, "message_tokens": 234, "message_utf8_bytes": 698, "cache_tensor_bytes": null, "serialized_payload_bytes": 698, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 16386.116749956273, "hops": {"1": 16386.116749956273}, "receiver_prefill_ms": 264.43704089615494, "ttft_ms": 90.90449998620898, "decode_ms": 272.62787509243935}, "total_wall_ms": 17014.646332943812, "ttft_ms": 90.90449998620898, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 5}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "Idris Bergström", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 12, "retention_rate": 0.18461538461538463, "retention_by_kind": {"person": 1.0, "date": 0.0, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 698}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 12, "retention_rate": 0.18461538461538463, "retention_by_kind": {"person": 1.0, "date": 0.0, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 698}}}, "reference_key_id": "keys/synthetic/syn_03009_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "b7aa2d9d5f9f3a06115e4e65c8ffaf4be2a07d6d067ceac89e49c6efeb6fdd5f", "output_hash": "fd88832c9fd0ab175818d76d875d70156225bd4b34ec401514546ef70f4f826c"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "9f15e4acaac45e823f7d723c", "attempt_id": "1adb0b707d00", "run_group_id": "E4_core:A5|closed|qwen3b:h1", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:16:48.251554Z", "task_family": "synthetic", "document_id": "syn_03009_2000", "document_hash": "6ed94dcf728255bd7a7d11c97c362930728fb81a3e91f1906eb6eec86e0fedb7", "question_id": "q04_multi_hop", "split": "test", "cluster_id": "syn_03009_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 1, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1976, "query_tokens": 325, "message_tokens": 234, "message_utf8_bytes": 698, "cache_tensor_bytes": null, "serialized_payload_bytes": 698, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 16386.116749956273, "hops": {"1": 16386.116749956273}, "receiver_prefill_ms": 275.73295892216265, "ttft_ms": 41.98158299550414, "decode_ms": 25.36116703413427}, "total_wall_ms": 16729.636374977417, "ttft_ms": 41.98158299550414, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 12, "retention_rate": 0.18461538461538463, "retention_by_kind": {"person": 1.0, "date": 0.0, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 698}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 12, "retention_rate": 0.18461538461538463, "retention_by_kind": {"person": 1.0, "date": 0.0, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 698}}}, "reference_key_id": "keys/synthetic/syn_03009_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "b7aa2d9d5f9f3a06115e4e65c8ffaf4be2a07d6d067ceac89e49c6efeb6fdd5f", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "68b0be2db01d2d1021d8808e", "attempt_id": "91a0592ce7f6", "run_group_id": "E4_core:A5|closed|qwen3b:h1", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:16:48.611702Z", "task_family": "synthetic", "document_id": "syn_03009_2000", "document_hash": "6ed94dcf728255bd7a7d11c97c362930728fb81a3e91f1906eb6eec86e0fedb7", "question_id": "q05_multi_hop", "split": "test", "cluster_id": "syn_03009_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 1, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1976, "query_tokens": 328, "message_tokens": 234, "message_utf8_bytes": 698, "cache_tensor_bytes": null, "serialized_payload_bytes": 698, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 16386.116749956273, "hops": {"1": 16386.116749956273}, "receiver_prefill_ms": 254.9006249755621, "ttft_ms": 29.669207986444235, "decode_ms": 25.316665996797383}, "total_wall_ms": 16696.449999930337, "ttft_ms": 29.669207986444235, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 12, "retention_rate": 0.18461538461538463, "retention_by_kind": {"person": 1.0, "date": 0.0, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 698}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 12, "retention_rate": 0.18461538461538463, "retention_by_kind": {"person": 1.0, "date": 0.0, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 698}}}, "reference_key_id": "keys/synthetic/syn_03009_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "b7aa2d9d5f9f3a06115e4e65c8ffaf4be2a07d6d067ceac89e49c6efeb6fdd5f", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "fd4e4d86a07c271148ab3216", "attempt_id": "12b01d1f7b8c", "run_group_id": "E4_core:A5|closed|qwen3b:h1", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:16:48.939637Z", "task_family": "synthetic", "document_id": "syn_03009_2000", "document_hash": "6ed94dcf728255bd7a7d11c97c362930728fb81a3e91f1906eb6eec86e0fedb7", "question_id": "q06_negation", "split": "test", "cluster_id": "syn_03009_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 1, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1976, "query_tokens": 318, "message_tokens": 234, "message_utf8_bytes": 698, "cache_tensor_bytes": null, "serialized_payload_bytes": 698, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 16386.116749956273, "hops": {"1": 16386.116749956273}, "receiver_prefill_ms": 207.32291706372052, "ttft_ms": 37.348082987591624, "decode_ms": 35.77895800117403}, "total_wall_ms": 16667.363665881567, "ttft_ms": 37.348082987591624, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "no", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 12, "retention_rate": 0.18461538461538463, "retention_by_kind": {"person": 1.0, "date": 0.0, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 698}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 12, "retention_rate": 0.18461538461538463, "retention_by_kind": {"person": 1.0, "date": 0.0, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 698}}}, "reference_key_id": "keys/synthetic/syn_03009_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "b7aa2d9d5f9f3a06115e4e65c8ffaf4be2a07d6d067ceac89e49c6efeb6fdd5f", "output_hash": "9390298f3fb0c5b160498935d79cb139aef28e1c47358b4bbba61862b9c26e59"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "82b6a04faa453511c553a5e1", "attempt_id": "195fdd2f4649", "run_group_id": "E4_core:A5|closed|qwen3b:h1", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:16:49.239137Z", "task_family": "synthetic", "document_id": "syn_03009_2000", "document_hash": "6ed94dcf728255bd7a7d11c97c362930728fb81a3e91f1906eb6eec86e0fedb7", "question_id": "q07_negation", "split": "test", "cluster_id": "syn_03009_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 1, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1976, "query_tokens": 317, "message_tokens": 234, "message_utf8_bytes": 698, "cache_tensor_bytes": null, "serialized_payload_bytes": 698, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 16386.116749956273, "hops": {"1": 16386.116749956273}, "receiver_prefill_ms": 228.95958402659744, "ttft_ms": 90.42758401483297, "decode_ms": 129.54583403188735}, "total_wall_ms": 16835.54924989585, "ttft_ms": 90.42758401483297, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "no", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 12, "retention_rate": 0.18461538461538463, "retention_by_kind": {"person": 1.0, "date": 0.0, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 698}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 12, "retention_rate": 0.18461538461538463, "retention_by_kind": {"person": 1.0, "date": 0.0, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 698}}}, "reference_key_id": "keys/synthetic/syn_03009_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "b7aa2d9d5f9f3a06115e4e65c8ffaf4be2a07d6d067ceac89e49c6efeb6fdd5f", "output_hash": "9390298f3fb0c5b160498935d79cb139aef28e1c47358b4bbba61862b9c26e59"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "14aae1d98f86c830bc451ef4", "attempt_id": "72a0d7527b11", "run_group_id": "E4_core:A5|closed|qwen3b:h1", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:16:49.705978Z", "task_family": "synthetic", "document_id": "syn_03009_2000", "document_hash": "6ed94dcf728255bd7a7d11c97c362930728fb81a3e91f1906eb6eec86e0fedb7", "question_id": "q08_revision", "split": "test", "cluster_id": "syn_03009_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 1, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1976, "query_tokens": 335, "message_tokens": 234, "message_utf8_bytes": 698, "cache_tensor_bytes": null, "serialized_payload_bytes": 698, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 16386.116749956273, "hops": {"1": 16386.116749956273}, "receiver_prefill_ms": 274.4670829270035, "ttft_ms": 41.875083930790424, "decode_ms": 24.89116706419736}, "total_wall_ms": 16728.141666040756, "ttft_ms": 41.875083930790424, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 12, "retention_rate": 0.18461538461538463, "retention_by_kind": {"person": 1.0, "date": 0.0, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 698}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 12, "retention_rate": 0.18461538461538463, "retention_by_kind": {"person": 1.0, "date": 0.0, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 698}}}, "reference_key_id": "keys/synthetic/syn_03009_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "b7aa2d9d5f9f3a06115e4e65c8ffaf4be2a07d6d067ceac89e49c6efeb6fdd5f", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "dfb0ccab7065525ccd068163", "attempt_id": "8c8941986f10", "run_group_id": "E4_core:A5|closed|qwen3b:h1", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:16:50.065227Z", "task_family": "synthetic", "document_id": "syn_03009_2000", "document_hash": "6ed94dcf728255bd7a7d11c97c362930728fb81a3e91f1906eb6eec86e0fedb7", "question_id": "q09_revision", "split": "test", "cluster_id": "syn_03009_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 1, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1976, "query_tokens": 328, "message_tokens": 234, "message_utf8_bytes": 698, "cache_tensor_bytes": null, "serialized_payload_bytes": 698, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 16386.116749956273, "hops": {"1": 16386.116749956273}, "receiver_prefill_ms": 272.3134170519188, "ttft_ms": 32.017959048971534, "decode_ms": 37.85254096146673}, "total_wall_ms": 16728.86120795738, "ttft_ms": 32.017959048971534, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 12, "retention_rate": 0.18461538461538463, "retention_by_kind": {"person": 1.0, "date": 0.0, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 698}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 12, "retention_rate": 0.18461538461538463, "retention_by_kind": {"person": 1.0, "date": 0.0, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 698}}}, "reference_key_id": "keys/synthetic/syn_03009_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "b7aa2d9d5f9f3a06115e4e65c8ffaf4be2a07d6d067ceac89e49c6efeb6fdd5f", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "3e419c2c471ccad9eec07aa8", "attempt_id": "81cc11fcc192", "run_group_id": "E4_core:A5|closed|qwen3b:h1", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:16:50.425978Z", "task_family": "synthetic", "document_id": "syn_03009_2000", "document_hash": "6ed94dcf728255bd7a7d11c97c362930728fb81a3e91f1906eb6eec86e0fedb7", "question_id": "q10_temporal", "split": "test", "cluster_id": "syn_03009_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 1, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1976, "query_tokens": 337, "message_tokens": 234, "message_utf8_bytes": 698, "cache_tensor_bytes": null, "serialized_payload_bytes": 698, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 16386.116749956273, "hops": {"1": 16386.116749956273}, "receiver_prefill_ms": 239.86283305566758, "ttft_ms": 42.59895801078528, "decode_ms": 501.7615830292925}, "total_wall_ms": 17170.8694169065, "ttft_ms": 42.59895801078528, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 7}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "Eshgamgam-11", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 12, "retention_rate": 0.18461538461538463, "retention_by_kind": {"person": 1.0, "date": 0.0, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 698}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 12, "retention_rate": 0.18461538461538463, "retention_by_kind": {"person": 1.0, "date": 0.0, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 698}}}, "reference_key_id": "keys/synthetic/syn_03009_2000.json", "task_scores": {"correct": 0.0, "abstained": 0.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "b7aa2d9d5f9f3a06115e4e65c8ffaf4be2a07d6d067ceac89e49c6efeb6fdd5f", "output_hash": "8c1313c53ab4dbffa5f336d721c810678133eecaa467241e53d8a92150601475"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "7f19bea5a8040437880d9e9b", "attempt_id": "909960e9400d", "run_group_id": "E4_core:A5|closed|qwen3b:h1", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:16:51.228709Z", "task_family": "synthetic", "document_id": "syn_03009_2000", "document_hash": "6ed94dcf728255bd7a7d11c97c362930728fb81a3e91f1906eb6eec86e0fedb7", "question_id": "q11_aggregation", "split": "test", "cluster_id": "syn_03009_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 1, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1976, "query_tokens": 328, "message_tokens": 234, "message_utf8_bytes": 698, "cache_tensor_bytes": null, "serialized_payload_bytes": 698, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 16386.116749956273, "hops": {"1": 16386.116749956273}, "receiver_prefill_ms": 273.57066702097654, "ttft_ms": 42.4166249576956, "decode_ms": 26.54179197270423}, "total_wall_ms": 16729.035332915373, "ttft_ms": 42.4166249576956, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 12, "retention_rate": 0.18461538461538463, "retention_by_kind": {"person": 1.0, "date": 0.0, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 698}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 12, "retention_rate": 0.18461538461538463, "retention_by_kind": {"person": 1.0, "date": 0.0, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 698}}}, "reference_key_id": "keys/synthetic/syn_03009_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "b7aa2d9d5f9f3a06115e4e65c8ffaf4be2a07d6d067ceac89e49c6efeb6fdd5f", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "b786079e1e38a20c73888f87", "attempt_id": "ebfdcf0ad46c", "run_group_id": "E4_core:A5|closed|qwen3b:h1", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:16:51.589311Z", "task_family": "synthetic", "document_id": "syn_03009_2000", "document_hash": "6ed94dcf728255bd7a7d11c97c362930728fb81a3e91f1906eb6eec86e0fedb7", "question_id": "q12_unit_conversion", "split": "test", "cluster_id": "syn_03009_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 1, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1976, "query_tokens": 326, "message_tokens": 234, "message_utf8_bytes": 698, "cache_tensor_bytes": null, "serialized_payload_bytes": 698, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 16386.116749956273, "hops": {"1": 16386.116749956273}, "receiver_prefill_ms": 271.2222500704229, "ttft_ms": 32.7167910290882, "decode_ms": 37.40870801266283}, "total_wall_ms": 16727.863583015278, "ttft_ms": 32.7167910290882, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 12, "retention_rate": 0.18461538461538463, "retention_by_kind": {"person": 1.0, "date": 0.0, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 698}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 12, "retention_rate": 0.18461538461538463, "retention_by_kind": {"person": 1.0, "date": 0.0, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 698}}}, "reference_key_id": "keys/synthetic/syn_03009_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "b7aa2d9d5f9f3a06115e4e65c8ffaf4be2a07d6d067ceac89e49c6efeb6fdd5f", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "3fb417aa4ceb30543e83a576", "attempt_id": "5880ee6b427d", "run_group_id": "E4_core:A5|closed|qwen3b:h1", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:16:51.948362Z", "task_family": "synthetic", "document_id": "syn_03009_2000", "document_hash": "6ed94dcf728255bd7a7d11c97c362930728fb81a3e91f1906eb6eec86e0fedb7", "question_id": "q13_unanswerable", "split": "test", "cluster_id": "syn_03009_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 1, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1976, "query_tokens": 330, "message_tokens": 234, "message_utf8_bytes": 698, "cache_tensor_bytes": null, "serialized_payload_bytes": 698, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 16386.116749956273, "hops": {"1": 16386.116749956273}, "receiver_prefill_ms": 241.82716698851436, "ttft_ms": 43.4743749210611, "decode_ms": 72.71079206839204}, "total_wall_ms": 16744.59674989339, "ttft_ms": 43.4743749210611, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 12, "retention_rate": 0.18461538461538463, "retention_by_kind": {"person": 1.0, "date": 0.0, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 698}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 12, "retention_rate": 0.18461538461538463, "retention_by_kind": {"person": 1.0, "date": 0.0, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 698}}}, "reference_key_id": "keys/synthetic/syn_03009_2000.json", "task_scores": {"correct": 1.0, "abstained": 1.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "b7aa2d9d5f9f3a06115e4e65c8ffaf4be2a07d6d067ceac89e49c6efeb6fdd5f", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "45ece59944a79fc394924d12", "attempt_id": "3ee903f8ca9b", "run_group_id": "E4_core:A5|closed|qwen3b:h1", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:16:52.323631Z", "task_family": "synthetic", "document_id": "syn_03009_2000", "document_hash": "6ed94dcf728255bd7a7d11c97c362930728fb81a3e91f1906eb6eec86e0fedb7", "question_id": "q14_unanswerable", "split": "test", "cluster_id": "syn_03009_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 1, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1976, "query_tokens": 330, "message_tokens": 234, "message_utf8_bytes": 698, "cache_tensor_bytes": null, "serialized_payload_bytes": 698, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 16386.116749956273, "hops": {"1": 16386.116749956273}, "receiver_prefill_ms": 266.19495905470103, "ttft_ms": 111.98233300819993, "decode_ms": 102.09791699890047}, "total_wall_ms": 16866.916291881353, "ttft_ms": 111.98233300819993, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 12, "retention_rate": 0.18461538461538463, "retention_by_kind": {"person": 1.0, "date": 0.0, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 698}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 12, "retention_rate": 0.18461538461538463, "retention_by_kind": {"person": 1.0, "date": 0.0, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 698}}}, "reference_key_id": "keys/synthetic/syn_03009_2000.json", "task_scores": {"correct": 1.0, "abstained": 1.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "b7aa2d9d5f9f3a06115e4e65c8ffaf4be2a07d6d067ceac89e49c6efeb6fdd5f", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "e0226b1bc51dfffbd16bbe1d", "attempt_id": "75e5181a5084", "run_group_id": "E4_core:A5|closed|qwen3b:h2", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:17:02.802835Z", "task_family": "synthetic", "document_id": "syn_03009_2000", "document_hash": "6ed94dcf728255bd7a7d11c97c362930728fb81a3e91f1906eb6eec86e0fedb7", "question_id": "q00_single_fact", "split": "test", "cluster_id": "syn_03009_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 2, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1976, "query_tokens": 318, "message_tokens": 234, "message_utf8_bytes": 698, "cache_tensor_bytes": null, "serialized_payload_bytes": 698, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 26364.528707927093, "hops": {"1": 16386.116749956273, "2": 9978.41195797082}, "receiver_prefill_ms": 237.84395796246827, "ttft_ms": 40.09345802478492, "decode_ms": 124.02729201130569}, "total_wall_ms": 26767.01999991201, "ttft_ms": 40.09345802478492, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 5}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "Milos Lindqvist", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 12, "retention_rate": 0.18461538461538463, "retention_by_kind": {"person": 1.0, "date": 0.0, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 698}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 12, "retention_rate": 0.18461538461538463, "retention_by_kind": {"person": 1.0, "date": 0.0, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 698}, "2": {"facts_total": 65, "facts_retained": 12, "retention_rate": 0.18461538461538463, "retention_by_kind": {"person": 1.0, "date": 0.0, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 698}}}, "reference_key_id": "keys/synthetic/syn_03009_2000.json", "task_scores": {"correct": 0.0, "abstained": 0.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "b7aa2d9d5f9f3a06115e4e65c8ffaf4be2a07d6d067ceac89e49c6efeb6fdd5f", "output_hash": "f1a0ce3d66245c336fdc7ac1281d2b638395739df9d016b121be6f1b2512201e"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "530e987cd8960745e48e0126", "attempt_id": "d0e9851b84de", "run_group_id": "E4_core:A5|closed|qwen3b:h2", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:17:03.223623Z", "task_family": "synthetic", "document_id": "syn_03009_2000", "document_hash": "6ed94dcf728255bd7a7d11c97c362930728fb81a3e91f1906eb6eec86e0fedb7", "question_id": "q01_single_fact", "split": "test", "cluster_id": "syn_03009_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 2, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1976, "query_tokens": 323, "message_tokens": 234, "message_utf8_bytes": 698, "cache_tensor_bytes": null, "serialized_payload_bytes": 698, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 26364.528707927093, "hops": {"1": 16386.116749956273, "2": 9978.41195797082}, "receiver_prefill_ms": 255.3165409481153, "ttft_ms": 49.982708995230496, "decode_ms": 77.96904200222343}, "total_wall_ms": 26748.291999916546, "ttft_ms": 49.982708995230496, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 12, "retention_rate": 0.18461538461538463, "retention_by_kind": {"person": 1.0, "date": 0.0, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 698}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 12, "retention_rate": 0.18461538461538463, "retention_by_kind": {"person": 1.0, "date": 0.0, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 698}, "2": {"facts_total": 65, "facts_retained": 12, "retention_rate": 0.18461538461538463, "retention_by_kind": {"person": 1.0, "date": 0.0, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 698}}}, "reference_key_id": "keys/synthetic/syn_03009_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "b7aa2d9d5f9f3a06115e4e65c8ffaf4be2a07d6d067ceac89e49c6efeb6fdd5f", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "ceb633e6cb3614ad9fa96929", "attempt_id": "b49323369756", "run_group_id": "E4_core:A5|closed|qwen3b:h2", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:17:03.624685Z", "task_family": "synthetic", "document_id": "syn_03009_2000", "document_hash": "6ed94dcf728255bd7a7d11c97c362930728fb81a3e91f1906eb6eec86e0fedb7", "question_id": "q02_exact_copy", "split": "test", "cluster_id": "syn_03009_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 2, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1976, "query_tokens": 321, "message_tokens": 234, "message_utf8_bytes": 698, "cache_tensor_bytes": null, "serialized_payload_bytes": 698, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 26364.528707927093, "hops": {"1": 16386.116749956273, "2": 9978.41195797082}, "receiver_prefill_ms": 371.12775002606213, "ttft_ms": 42.21262491773814, "decode_ms": 25.452291942201555}, "total_wall_ms": 26803.71704197023, "ttft_ms": 42.21262491773814, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 12, "retention_rate": 0.18461538461538463, "retention_by_kind": {"person": 1.0, "date": 0.0, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 698}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 12, "retention_rate": 0.18461538461538463, "retention_by_kind": {"person": 1.0, "date": 0.0, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 698}, "2": {"facts_total": 65, "facts_retained": 12, "retention_rate": 0.18461538461538463, "retention_by_kind": {"person": 1.0, "date": 0.0, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 698}}}, "reference_key_id": "keys/synthetic/syn_03009_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "b7aa2d9d5f9f3a06115e4e65c8ffaf4be2a07d6d067ceac89e49c6efeb6fdd5f", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "fa690b33df5070d8d3238174", "attempt_id": "2be8aa2ac2a1", "run_group_id": "E4_core:A5|closed|qwen3b:h2", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:17:04.083478Z", "task_family": "synthetic", "document_id": "syn_03009_2000", "document_hash": "6ed94dcf728255bd7a7d11c97c362930728fb81a3e91f1906eb6eec86e0fedb7", "question_id": "q03_exact_copy", "split": "test", "cluster_id": "syn_03009_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 2, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1976, "query_tokens": 317, "message_tokens": 234, "message_utf8_bytes": 698, "cache_tensor_bytes": null, "serialized_payload_bytes": 698, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 26364.528707927093, "hops": {"1": 16386.116749956273, "2": 9978.41195797082}, "receiver_prefill_ms": 232.96820803079754, "ttft_ms": 37.51391696278006, "decode_ms": 168.21375000290573}, "total_wall_ms": 26803.662457852624, "ttft_ms": 37.51391696278006, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 5}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "Idris Bergström", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 12, "retention_rate": 0.18461538461538463, "retention_by_kind": {"person": 1.0, "date": 0.0, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 698}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 12, "retention_rate": 0.18461538461538463, "retention_by_kind": {"person": 1.0, "date": 0.0, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 698}, "2": {"facts_total": 65, "facts_retained": 12, "retention_rate": 0.18461538461538463, "retention_by_kind": {"person": 1.0, "date": 0.0, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 698}}}, "reference_key_id": "keys/synthetic/syn_03009_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "b7aa2d9d5f9f3a06115e4e65c8ffaf4be2a07d6d067ceac89e49c6efeb6fdd5f", "output_hash": "fd88832c9fd0ab175818d76d875d70156225bd4b34ec401514546ef70f4f826c"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "5e1849d12bc8ad20d7f1d532", "attempt_id": "f0615791d99a", "run_group_id": "E4_core:A5|closed|qwen3b:h2", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:17:04.540285Z", "task_family": "synthetic", "document_id": "syn_03009_2000", "document_hash": "6ed94dcf728255bd7a7d11c97c362930728fb81a3e91f1906eb6eec86e0fedb7", "question_id": "q04_multi_hop", "split": "test", "cluster_id": "syn_03009_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 2, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1976, "query_tokens": 325, "message_tokens": 234, "message_utf8_bytes": 698, "cache_tensor_bytes": null, "serialized_payload_bytes": 698, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 26364.528707927093, "hops": {"1": 16386.116749956273, "2": 9978.41195797082}, "receiver_prefill_ms": 258.34758405108005, "ttft_ms": 69.63241600897163, "decode_ms": 45.51508300937712}, "total_wall_ms": 26738.479207851924, "ttft_ms": 69.63241600897163, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 12, "retention_rate": 0.18461538461538463, "retention_by_kind": {"person": 1.0, "date": 0.0, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 698}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 12, "retention_rate": 0.18461538461538463, "retention_by_kind": {"person": 1.0, "date": 0.0, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 698}, "2": {"facts_total": 65, "facts_retained": 12, "retention_rate": 0.18461538461538463, "retention_by_kind": {"person": 1.0, "date": 0.0, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 698}}}, "reference_key_id": "keys/synthetic/syn_03009_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "b7aa2d9d5f9f3a06115e4e65c8ffaf4be2a07d6d067ceac89e49c6efeb6fdd5f", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "5081e883201baf11af11729e", "attempt_id": "6e04e70d0b2b", "run_group_id": "E4_core:A5|closed|qwen3b:h2", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:17:04.931701Z", "task_family": "synthetic", "document_id": "syn_03009_2000", "document_hash": "6ed94dcf728255bd7a7d11c97c362930728fb81a3e91f1906eb6eec86e0fedb7", "question_id": "q05_multi_hop", "split": "test", "cluster_id": "syn_03009_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 2, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1976, "query_tokens": 328, "message_tokens": 234, "message_utf8_bytes": 698, "cache_tensor_bytes": null, "serialized_payload_bytes": 698, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 26364.528707927093, "hops": {"1": 16386.116749956273, "2": 9978.41195797082}, "receiver_prefill_ms": 305.7747499551624, "ttft_ms": 69.97108308132738, "decode_ms": 102.8326660161838}, "total_wall_ms": 26843.670457950793, "ttft_ms": 69.97108308132738, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 12, "retention_rate": 0.18461538461538463, "retention_by_kind": {"person": 1.0, "date": 0.0, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 698}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 12, "retention_rate": 0.18461538461538463, "retention_by_kind": {"person": 1.0, "date": 0.0, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 698}, "2": {"facts_total": 65, "facts_retained": 12, "retention_rate": 0.18461538461538463, "retention_by_kind": {"person": 1.0, "date": 0.0, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 698}}}, "reference_key_id": "keys/synthetic/syn_03009_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "b7aa2d9d5f9f3a06115e4e65c8ffaf4be2a07d6d067ceac89e49c6efeb6fdd5f", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "16b0e3a7809789a4f7aae857", "attempt_id": "7b1c9e4ac013", "run_group_id": "E4_core:A5|closed|qwen3b:h2", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:17:05.429752Z", "task_family": "synthetic", "document_id": "syn_03009_2000", "document_hash": "6ed94dcf728255bd7a7d11c97c362930728fb81a3e91f1906eb6eec86e0fedb7", "question_id": "q06_negation", "split": "test", "cluster_id": "syn_03009_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 2, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1976, "query_tokens": 318, "message_tokens": 234, "message_utf8_bytes": 698, "cache_tensor_bytes": null, "serialized_payload_bytes": 698, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 26364.528707927093, "hops": {"1": 16386.116749956273, "2": 9978.41195797082}, "receiver_prefill_ms": 233.7063750019297, "ttft_ms": 36.21425002347678, "decode_ms": 38.918333943001926}, "total_wall_ms": 26674.039082950912, "ttft_ms": 36.21425002347678, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "no", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 12, "retention_rate": 0.18461538461538463, "retention_by_kind": {"person": 1.0, "date": 0.0, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 698}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 12, "retention_rate": 0.18461538461538463, "retention_by_kind": {"person": 1.0, "date": 0.0, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 698}, "2": {"facts_total": 65, "facts_retained": 12, "retention_rate": 0.18461538461538463, "retention_by_kind": {"person": 1.0, "date": 0.0, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 698}}}, "reference_key_id": "keys/synthetic/syn_03009_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "b7aa2d9d5f9f3a06115e4e65c8ffaf4be2a07d6d067ceac89e49c6efeb6fdd5f", "output_hash": "9390298f3fb0c5b160498935d79cb139aef28e1c47358b4bbba61862b9c26e59"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "1888288b1a42ec8160a7d1e2", "attempt_id": "fe4a2cdc2b49", "run_group_id": "E4_core:A5|closed|qwen3b:h2", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:17:05.756724Z", "task_family": "synthetic", "document_id": "syn_03009_2000", "document_hash": "6ed94dcf728255bd7a7d11c97c362930728fb81a3e91f1906eb6eec86e0fedb7", "question_id": "q07_negation", "split": "test", "cluster_id": "syn_03009_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 2, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1976, "query_tokens": 317, "message_tokens": 234, "message_utf8_bytes": 698, "cache_tensor_bytes": null, "serialized_payload_bytes": 698, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 26364.528707927093, "hops": {"1": 16386.116749956273, "2": 9978.41195797082}, "receiver_prefill_ms": 236.52566596865654, "ttft_ms": 43.39254193473607, "decode_ms": 46.21616692747921}, "total_wall_ms": 26691.269957926124, "ttft_ms": 43.39254193473607, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "no", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 12, "retention_rate": 0.18461538461538463, "retention_by_kind": {"person": 1.0, "date": 0.0, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 698}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 12, "retention_rate": 0.18461538461538463, "retention_by_kind": {"person": 1.0, "date": 0.0, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 698}, "2": {"facts_total": 65, "facts_retained": 12, "retention_rate": 0.18461538461538463, "retention_by_kind": {"person": 1.0, "date": 0.0, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 698}}}, "reference_key_id": "keys/synthetic/syn_03009_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "b7aa2d9d5f9f3a06115e4e65c8ffaf4be2a07d6d067ceac89e49c6efeb6fdd5f", "output_hash": "9390298f3fb0c5b160498935d79cb139aef28e1c47358b4bbba61862b9c26e59"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "2ad9a52931c51d289da9833d", "attempt_id": "38991d7f5946", "run_group_id": "E4_core:A5|closed|qwen3b:h2", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:17:06.102899Z", "task_family": "synthetic", "document_id": "syn_03009_2000", "document_hash": "6ed94dcf728255bd7a7d11c97c362930728fb81a3e91f1906eb6eec86e0fedb7", "question_id": "q08_revision", "split": "test", "cluster_id": "syn_03009_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 2, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1976, "query_tokens": 335, "message_tokens": 234, "message_utf8_bytes": 698, "cache_tensor_bytes": null, "serialized_payload_bytes": 698, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 26364.528707927093, "hops": {"1": 16386.116749956273, "2": 9978.41195797082}, "receiver_prefill_ms": 231.3029170036316, "ttft_ms": 58.57266706880182, "decode_ms": 57.63783399015665}, "total_wall_ms": 26712.757082888857, "ttft_ms": 58.57266706880182, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 12, "retention_rate": 0.18461538461538463, "retention_by_kind": {"person": 1.0, "date": 0.0, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 698}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 12, "retention_rate": 0.18461538461538463, "retention_by_kind": {"person": 1.0, "date": 0.0, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 698}, "2": {"facts_total": 65, "facts_retained": 12, "retention_rate": 0.18461538461538463, "retention_by_kind": {"person": 1.0, "date": 0.0, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 698}}}, "reference_key_id": "keys/synthetic/syn_03009_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "b7aa2d9d5f9f3a06115e4e65c8ffaf4be2a07d6d067ceac89e49c6efeb6fdd5f", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "e2ada563ab7b0fa25876882e", "attempt_id": "bead2adbbdf9", "run_group_id": "E4_core:A5|closed|qwen3b:h2", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:17:06.469312Z", "task_family": "synthetic", "document_id": "syn_03009_2000", "document_hash": "6ed94dcf728255bd7a7d11c97c362930728fb81a3e91f1906eb6eec86e0fedb7", "question_id": "q09_revision", "split": "test", "cluster_id": "syn_03009_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 2, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1976, "query_tokens": 328, "message_tokens": 234, "message_utf8_bytes": 698, "cache_tensor_bytes": null, "serialized_payload_bytes": 698, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 26364.528707927093, "hops": {"1": 16386.116749956273, "2": 9978.41195797082}, "receiver_prefill_ms": 276.41166700050235, "ttft_ms": 92.89425006136298, "decode_ms": 136.60304201766849}, "total_wall_ms": 26871.046707965434, "ttft_ms": 92.89425006136298, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 12, "retention_rate": 0.18461538461538463, "retention_by_kind": {"person": 1.0, "date": 0.0, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 698}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 12, "retention_rate": 0.18461538461538463, "retention_by_kind": {"person": 1.0, "date": 0.0, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 698}, "2": {"facts_total": 65, "facts_retained": 12, "retention_rate": 0.18461538461538463, "retention_by_kind": {"person": 1.0, "date": 0.0, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 698}}}, "reference_key_id": "keys/synthetic/syn_03009_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "b7aa2d9d5f9f3a06115e4e65c8ffaf4be2a07d6d067ceac89e49c6efeb6fdd5f", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "6724fa496ad6702b98c7fc51", "attempt_id": "6ac7a6beb571", "run_group_id": "E4_core:A5|closed|qwen3b:h2", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:17:06.994493Z", "task_family": "synthetic", "document_id": "syn_03009_2000", "document_hash": "6ed94dcf728255bd7a7d11c97c362930728fb81a3e91f1906eb6eec86e0fedb7", "question_id": "q10_temporal", "split": "test", "cluster_id": "syn_03009_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 2, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1976, "query_tokens": 337, "message_tokens": 234, "message_utf8_bytes": 698, "cache_tensor_bytes": null, "serialized_payload_bytes": 698, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 26364.528707927093, "hops": {"1": 16386.116749956273, "2": 9978.41195797082}, "receiver_prefill_ms": 261.26087503507733, "ttft_ms": 28.85016694199294, "decode_ms": 254.98791597783566}, "total_wall_ms": 26910.0622079568, "ttft_ms": 28.85016694199294, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 7}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "Eshgamgam-11", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 12, "retention_rate": 0.18461538461538463, "retention_by_kind": {"person": 1.0, "date": 0.0, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 698}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 12, "retention_rate": 0.18461538461538463, "retention_by_kind": {"person": 1.0, "date": 0.0, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 698}, "2": {"facts_total": 65, "facts_retained": 12, "retention_rate": 0.18461538461538463, "retention_by_kind": {"person": 1.0, "date": 0.0, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 698}}}, "reference_key_id": "keys/synthetic/syn_03009_2000.json", "task_scores": {"correct": 0.0, "abstained": 0.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "b7aa2d9d5f9f3a06115e4e65c8ffaf4be2a07d6d067ceac89e49c6efeb6fdd5f", "output_hash": "8c1313c53ab4dbffa5f336d721c810678133eecaa467241e53d8a92150601475"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "76e9495fe97df288f8b31b8d", "attempt_id": "0f1795d5b7d2", "run_group_id": "E4_core:A5|closed|qwen3b:h2", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:17:07.557776Z", "task_family": "synthetic", "document_id": "syn_03009_2000", "document_hash": "6ed94dcf728255bd7a7d11c97c362930728fb81a3e91f1906eb6eec86e0fedb7", "question_id": "q11_aggregation", "split": "test", "cluster_id": "syn_03009_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 2, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1976, "query_tokens": 328, "message_tokens": 234, "message_utf8_bytes": 698, "cache_tensor_bytes": null, "serialized_payload_bytes": 698, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 26364.528707927093, "hops": {"1": 16386.116749956273, "2": 9978.41195797082}, "receiver_prefill_ms": 252.59495805948973, "ttft_ms": 24.095915956422687, "decode_ms": 20.89970896486193}, "total_wall_ms": 26662.68491698429, "ttft_ms": 24.095915956422687, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 12, "retention_rate": 0.18461538461538463, "retention_by_kind": {"person": 1.0, "date": 0.0, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 698}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 12, "retention_rate": 0.18461538461538463, "retention_by_kind": {"person": 1.0, "date": 0.0, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 698}, "2": {"facts_total": 65, "facts_retained": 12, "retention_rate": 0.18461538461538463, "retention_by_kind": {"person": 1.0, "date": 0.0, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 698}}}, "reference_key_id": "keys/synthetic/syn_03009_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "b7aa2d9d5f9f3a06115e4e65c8ffaf4be2a07d6d067ceac89e49c6efeb6fdd5f", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "10990c1a9c86c0b783aa6417", "attempt_id": "5d0850a5e0c8", "run_group_id": "E4_core:A5|closed|qwen3b:h2", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:17:07.873283Z", "task_family": "synthetic", "document_id": "syn_03009_2000", "document_hash": "6ed94dcf728255bd7a7d11c97c362930728fb81a3e91f1906eb6eec86e0fedb7", "question_id": "q12_unit_conversion", "split": "test", "cluster_id": "syn_03009_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 2, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1976, "query_tokens": 326, "message_tokens": 234, "message_utf8_bytes": 698, "cache_tensor_bytes": null, "serialized_payload_bytes": 698, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 26364.528707927093, "hops": {"1": 16386.116749956273, "2": 9978.41195797082}, "receiver_prefill_ms": 248.39566694572568, "ttft_ms": 72.13358394801617, "decode_ms": 58.15945798531175}, "total_wall_ms": 26743.7994988868, "ttft_ms": 72.13358394801617, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 12, "retention_rate": 0.18461538461538463, "retention_by_kind": {"person": 1.0, "date": 0.0, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 698}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 12, "retention_rate": 0.18461538461538463, "retention_by_kind": {"person": 1.0, "date": 0.0, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 698}, "2": {"facts_total": 65, "facts_retained": 12, "retention_rate": 0.18461538461538463, "retention_by_kind": {"person": 1.0, "date": 0.0, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 698}}}, "reference_key_id": "keys/synthetic/syn_03009_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "b7aa2d9d5f9f3a06115e4e65c8ffaf4be2a07d6d067ceac89e49c6efeb6fdd5f", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "263c5b061823670ef9c84b93", "attempt_id": "f5c4cb160669", "run_group_id": "E4_core:A5|closed|qwen3b:h2", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:17:08.270721Z", "task_family": "synthetic", "document_id": "syn_03009_2000", "document_hash": "6ed94dcf728255bd7a7d11c97c362930728fb81a3e91f1906eb6eec86e0fedb7", "question_id": "q13_unanswerable", "split": "test", "cluster_id": "syn_03009_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 2, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1976, "query_tokens": 330, "message_tokens": 234, "message_utf8_bytes": 698, "cache_tensor_bytes": null, "serialized_payload_bytes": 698, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 26364.528707927093, "hops": {"1": 16386.116749956273, "2": 9978.41195797082}, "receiver_prefill_ms": 393.2027081027627, "ttft_ms": 33.624415984377265, "decode_ms": 39.34487502556294}, "total_wall_ms": 26831.16970793344, "ttft_ms": 33.624415984377265, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 12, "retention_rate": 0.18461538461538463, "retention_by_kind": {"person": 1.0, "date": 0.0, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 698}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 12, "retention_rate": 0.18461538461538463, "retention_by_kind": {"person": 1.0, "date": 0.0, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 698}, "2": {"facts_total": 65, "facts_retained": 12, "retention_rate": 0.18461538461538463, "retention_by_kind": {"person": 1.0, "date": 0.0, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 698}}}, "reference_key_id": "keys/synthetic/syn_03009_2000.json", "task_scores": {"correct": 1.0, "abstained": 1.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "b7aa2d9d5f9f3a06115e4e65c8ffaf4be2a07d6d067ceac89e49c6efeb6fdd5f", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "8523d7815b47f4b9be481618", "attempt_id": "55030b11029b", "run_group_id": "E4_core:A5|closed|qwen3b:h2", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:17:08.755942Z", "task_family": "synthetic", "document_id": "syn_03009_2000", "document_hash": "6ed94dcf728255bd7a7d11c97c362930728fb81a3e91f1906eb6eec86e0fedb7", "question_id": "q14_unanswerable", "split": "test", "cluster_id": "syn_03009_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 2, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1976, "query_tokens": 330, "message_tokens": 234, "message_utf8_bytes": 698, "cache_tensor_bytes": null, "serialized_payload_bytes": 698, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 26364.528707927093, "hops": {"1": 16386.116749956273, "2": 9978.41195797082}, "receiver_prefill_ms": 261.33695791941136, "ttft_ms": 43.62916701938957, "decode_ms": 37.186000030487776}, "total_wall_ms": 26707.068166928366, "ttft_ms": 43.62916701938957, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 12, "retention_rate": 0.18461538461538463, "retention_by_kind": {"person": 1.0, "date": 0.0, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 698}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 12, "retention_rate": 0.18461538461538463, "retention_by_kind": {"person": 1.0, "date": 0.0, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 698}, "2": {"facts_total": 65, "facts_retained": 12, "retention_rate": 0.18461538461538463, "retention_by_kind": {"person": 1.0, "date": 0.0, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 698}}}, "reference_key_id": "keys/synthetic/syn_03009_2000.json", "task_scores": {"correct": 1.0, "abstained": 1.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "b7aa2d9d5f9f3a06115e4e65c8ffaf4be2a07d6d067ceac89e49c6efeb6fdd5f", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "78595419b0ec01cb9b82131e", "attempt_id": "6058dce0ec2b", "run_group_id": "E4_core:A5|closed|qwen3b:h4", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:17:29.362881Z", "task_family": "synthetic", "document_id": "syn_03009_2000", "document_hash": "6ed94dcf728255bd7a7d11c97c362930728fb81a3e91f1906eb6eec86e0fedb7", "question_id": "q00_single_fact", "split": "test", "cluster_id": "syn_03009_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 4, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1976, "query_tokens": 318, "message_tokens": 234, "message_utf8_bytes": 698, "cache_tensor_bytes": null, "serialized_payload_bytes": 698, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 46608.10408287216, "hops": {"1": 16386.116749956273, "2": 9978.41195797082, "3": 9960.819375002757, "4": 10282.75599994231}, "receiver_prefill_ms": 236.49279109667987, "ttft_ms": 31.513415975496173, "decode_ms": 165.29495804570615}, "total_wall_ms": 47041.84016678482, "ttft_ms": 31.513415975496173, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 5}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "Milos Lindqvist", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 12, "retention_rate": 0.18461538461538463, "retention_by_kind": {"person": 1.0, "date": 0.0, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 698}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 12, "retention_rate": 0.18461538461538463, "retention_by_kind": {"person": 1.0, "date": 0.0, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 698}, "2": {"facts_total": 65, "facts_retained": 12, "retention_rate": 0.18461538461538463, "retention_by_kind": {"person": 1.0, "date": 0.0, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 698}, "3": {"facts_total": 65, "facts_retained": 12, "retention_rate": 0.18461538461538463, "retention_by_kind": {"person": 1.0, "date": 0.0, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 698}, "4": {"facts_total": 65, "facts_retained": 12, "retention_rate": 0.18461538461538463, "retention_by_kind": {"person": 1.0, "date": 0.0, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 698}}}, "reference_key_id": "keys/synthetic/syn_03009_2000.json", "task_scores": {"correct": 0.0, "abstained": 0.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "b7aa2d9d5f9f3a06115e4e65c8ffaf4be2a07d6d067ceac89e49c6efeb6fdd5f", "output_hash": "f1a0ce3d66245c336fdc7ac1281d2b638395739df9d016b121be6f1b2512201e"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "7773b737d03e193d7f54af1f", "attempt_id": "47cc5683b23c", "run_group_id": "E4_core:A5|closed|qwen3b:h4", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:17:29.814333Z", "task_family": "synthetic", "document_id": "syn_03009_2000", "document_hash": "6ed94dcf728255bd7a7d11c97c362930728fb81a3e91f1906eb6eec86e0fedb7", "question_id": "q01_single_fact", "split": "test", "cluster_id": "syn_03009_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 4, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1976, "query_tokens": 323, "message_tokens": 234, "message_utf8_bytes": 698, "cache_tensor_bytes": null, "serialized_payload_bytes": 698, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 46608.10408287216, "hops": {"1": 16386.116749956273, "2": 9978.41195797082, "3": 9960.819375002757, "4": 10282.75599994231}, "receiver_prefill_ms": 231.58658400643617, "ttft_ms": 42.822540970519185, "decode_ms": 65.19337499048561}, "total_wall_ms": 46948.15804087557, "ttft_ms": 42.822540970519185, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 12, "retention_rate": 0.18461538461538463, "retention_by_kind": {"person": 1.0, "date": 0.0, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 698}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 12, "retention_rate": 0.18461538461538463, "retention_by_kind": {"person": 1.0, "date": 0.0, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 698}, "2": {"facts_total": 65, "facts_retained": 12, "retention_rate": 0.18461538461538463, "retention_by_kind": {"person": 1.0, "date": 0.0, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 698}, "3": {"facts_total": 65, "facts_retained": 12, "retention_rate": 0.18461538461538463, "retention_by_kind": {"person": 1.0, "date": 0.0, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 698}, "4": {"facts_total": 65, "facts_retained": 12, "retention_rate": 0.18461538461538463, "retention_by_kind": {"person": 1.0, "date": 0.0, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 698}}}, "reference_key_id": "keys/synthetic/syn_03009_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "b7aa2d9d5f9f3a06115e4e65c8ffaf4be2a07d6d067ceac89e49c6efeb6fdd5f", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "7f77aa3142d5b726318a4be9", "attempt_id": "6a5a559b21c1", "run_group_id": "E4_core:A5|closed|qwen3b:h4", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:17:30.172712Z", "task_family": "synthetic", "document_id": "syn_03009_2000", "document_hash": "6ed94dcf728255bd7a7d11c97c362930728fb81a3e91f1906eb6eec86e0fedb7", "question_id": "q02_exact_copy", "split": "test", "cluster_id": "syn_03009_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 4, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1976, "query_tokens": 321, "message_tokens": 234, "message_utf8_bytes": 698, "cache_tensor_bytes": null, "serialized_payload_bytes": 698, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 46608.10408287216, "hops": {"1": 16386.116749956273, "2": 9978.41195797082, "3": 9960.819375002757, "4": 10282.75599994231}, "receiver_prefill_ms": 224.66379194520414, "ttft_ms": 91.2853330373764, "decode_ms": 127.61766707990319}, "total_wall_ms": 47052.27629095316, "ttft_ms": 91.2853330373764, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 12, "retention_rate": 0.18461538461538463, "retention_by_kind": {"person": 1.0, "date": 0.0, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 698}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 12, "retention_rate": 0.18461538461538463, "retention_by_kind": {"person": 1.0, "date": 0.0, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 698}, "2": {"facts_total": 65, "facts_retained": 12, "retention_rate": 0.18461538461538463, "retention_by_kind": {"person": 1.0, "date": 0.0, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 698}, "3": {"facts_total": 65, "facts_retained": 12, "retention_rate": 0.18461538461538463, "retention_by_kind": {"person": 1.0, "date": 0.0, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 698}, "4": {"facts_total": 65, "facts_retained": 12, "retention_rate": 0.18461538461538463, "retention_by_kind": {"person": 1.0, "date": 0.0, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 698}}}, "reference_key_id": "keys/synthetic/syn_03009_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "b7aa2d9d5f9f3a06115e4e65c8ffaf4be2a07d6d067ceac89e49c6efeb6fdd5f", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "3e02d4f6aab40b275c71d5ad", "attempt_id": "7ae6ff79b9d8", "run_group_id": "E4_core:A5|closed|qwen3b:h4", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:17:30.634856Z", "task_family": "synthetic", "document_id": "syn_03009_2000", "document_hash": "6ed94dcf728255bd7a7d11c97c362930728fb81a3e91f1906eb6eec86e0fedb7", "question_id": "q03_exact_copy", "split": "test", "cluster_id": "syn_03009_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 4, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1976, "query_tokens": 317, "message_tokens": 234, "message_utf8_bytes": 698, "cache_tensor_bytes": null, "serialized_payload_bytes": 698, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 46608.10408287216, "hops": {"1": 16386.116749956273, "2": 9978.41195797082, "3": 9960.819375002757, "4": 10282.75599994231}, "receiver_prefill_ms": 253.41895897872746, "ttft_ms": 43.34395797923207, "decode_ms": 155.31737497076392}, "total_wall_ms": 47060.704707866535, "ttft_ms": 43.34395797923207, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 5}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "Idris Bergström", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 12, "retention_rate": 0.18461538461538463, "retention_by_kind": {"person": 1.0, "date": 0.0, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 698}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 12, "retention_rate": 0.18461538461538463, "retention_by_kind": {"person": 1.0, "date": 0.0, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 698}, "2": {"facts_total": 65, "facts_retained": 12, "retention_rate": 0.18461538461538463, "retention_by_kind": {"person": 1.0, "date": 0.0, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 698}, "3": {"facts_total": 65, "facts_retained": 12, "retention_rate": 0.18461538461538463, "retention_by_kind": {"person": 1.0, "date": 0.0, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 698}, "4": {"facts_total": 65, "facts_retained": 12, "retention_rate": 0.18461538461538463, "retention_by_kind": {"person": 1.0, "date": 0.0, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 698}}}, "reference_key_id": "keys/synthetic/syn_03009_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "b7aa2d9d5f9f3a06115e4e65c8ffaf4be2a07d6d067ceac89e49c6efeb6fdd5f", "output_hash": "fd88832c9fd0ab175818d76d875d70156225bd4b34ec401514546ef70f4f826c"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "b6d4fb29e6cdba0342cbaef3", "attempt_id": "0f7354f41c63", "run_group_id": "E4_core:A5|closed|qwen3b:h4", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:17:31.104863Z", "task_family": "synthetic", "document_id": "syn_03009_2000", "document_hash": "6ed94dcf728255bd7a7d11c97c362930728fb81a3e91f1906eb6eec86e0fedb7", "question_id": "q04_multi_hop", "split": "test", "cluster_id": "syn_03009_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 4, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1976, "query_tokens": 325, "message_tokens": 234, "message_utf8_bytes": 698, "cache_tensor_bytes": null, "serialized_payload_bytes": 698, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 46608.10408287216, "hops": {"1": 16386.116749956273, "2": 9978.41195797082, "3": 9960.819375002757, "4": 10282.75599994231}, "receiver_prefill_ms": 253.49379191175103, "ttft_ms": 29.76416703313589, "decode_ms": 34.096332965418696}, "total_wall_ms": 46925.861916854046, "ttft_ms": 29.76416703313589, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 12, "retention_rate": 0.18461538461538463, "retention_by_kind": {"person": 1.0, "date": 0.0, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 698}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 12, "retention_rate": 0.18461538461538463, "retention_by_kind": {"person": 1.0, "date": 0.0, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 698}, "2": {"facts_total": 65, "facts_retained": 12, "retention_rate": 0.18461538461538463, "retention_by_kind": {"person": 1.0, "date": 0.0, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 698}, "3": {"facts_total": 65, "facts_retained": 12, "retention_rate": 0.18461538461538463, "retention_by_kind": {"person": 1.0, "date": 0.0, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 698}, "4": {"facts_total": 65, "facts_retained": 12, "retention_rate": 0.18461538461538463, "retention_by_kind": {"person": 1.0, "date": 0.0, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 698}}}, "reference_key_id": "keys/synthetic/syn_03009_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "b7aa2d9d5f9f3a06115e4e65c8ffaf4be2a07d6d067ceac89e49c6efeb6fdd5f", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "139f1bd818a60fdc6778d9e5", "attempt_id": "cafae00ad615", "run_group_id": "E4_core:A5|closed|qwen3b:h4", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:17:31.439512Z", "task_family": "synthetic", "document_id": "syn_03009_2000", "document_hash": "6ed94dcf728255bd7a7d11c97c362930728fb81a3e91f1906eb6eec86e0fedb7", "question_id": "q05_multi_hop", "split": "test", "cluster_id": "syn_03009_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 4, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1976, "query_tokens": 328, "message_tokens": 234, "message_utf8_bytes": 698, "cache_tensor_bytes": null, "serialized_payload_bytes": 698, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 46608.10408287216, "hops": {"1": 16386.116749956273, "2": 9978.41195797082, "3": 9960.819375002757, "4": 10282.75599994231}, "receiver_prefill_ms": 222.7269159629941, "ttft_ms": 62.22937488928437, "decode_ms": 64.98183298390359}, "total_wall_ms": 46958.52729084436, "ttft_ms": 62.22937488928437, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 12, "retention_rate": 0.18461538461538463, "retention_by_kind": {"person": 1.0, "date": 0.0, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 698}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 12, "retention_rate": 0.18461538461538463, "retention_by_kind": {"person": 1.0, "date": 0.0, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 698}, "2": {"facts_total": 65, "facts_retained": 12, "retention_rate": 0.18461538461538463, "retention_by_kind": {"person": 1.0, "date": 0.0, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 698}, "3": {"facts_total": 65, "facts_retained": 12, "retention_rate": 0.18461538461538463, "retention_by_kind": {"person": 1.0, "date": 0.0, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 698}, "4": {"facts_total": 65, "facts_retained": 12, "retention_rate": 0.18461538461538463, "retention_by_kind": {"person": 1.0, "date": 0.0, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 698}}}, "reference_key_id": "keys/synthetic/syn_03009_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "b7aa2d9d5f9f3a06115e4e65c8ffaf4be2a07d6d067ceac89e49c6efeb6fdd5f", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "63984cc9194a2f7a56823a2e", "attempt_id": "ec6879f66ddc", "run_group_id": "E4_core:A5|closed|qwen3b:h4", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:17:31.806910Z", "task_family": "synthetic", "document_id": "syn_03009_2000", "document_hash": "6ed94dcf728255bd7a7d11c97c362930728fb81a3e91f1906eb6eec86e0fedb7", "question_id": "q06_negation", "split": "test", "cluster_id": "syn_03009_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 4, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1976, "query_tokens": 318, "message_tokens": 234, "message_utf8_bytes": 698, "cache_tensor_bytes": null, "serialized_payload_bytes": 698, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 46608.10408287216, "hops": {"1": 16386.116749956273, "2": 9978.41195797082, "3": 9960.819375002757, "4": 10282.75599994231}, "receiver_prefill_ms": 332.5656250817701, "ttft_ms": 70.38433302659541, "decode_ms": 31.656999955885112}, "total_wall_ms": 47043.11541689094, "ttft_ms": 70.38433302659541, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "no", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 12, "retention_rate": 0.18461538461538463, "retention_by_kind": {"person": 1.0, "date": 0.0, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 698}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 12, "retention_rate": 0.18461538461538463, "retention_by_kind": {"person": 1.0, "date": 0.0, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 698}, "2": {"facts_total": 65, "facts_retained": 12, "retention_rate": 0.18461538461538463, "retention_by_kind": {"person": 1.0, "date": 0.0, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 698}, "3": {"facts_total": 65, "facts_retained": 12, "retention_rate": 0.18461538461538463, "retention_by_kind": {"person": 1.0, "date": 0.0, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 698}, "4": {"facts_total": 65, "facts_retained": 12, "retention_rate": 0.18461538461538463, "retention_by_kind": {"person": 1.0, "date": 0.0, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 698}}}, "reference_key_id": "keys/synthetic/syn_03009_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "b7aa2d9d5f9f3a06115e4e65c8ffaf4be2a07d6d067ceac89e49c6efeb6fdd5f", "output_hash": "9390298f3fb0c5b160498935d79cb139aef28e1c47358b4bbba61862b9c26e59"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "3b7db26c3ba05a8b5104357f", "attempt_id": "d0c619f0954b", "run_group_id": "E4_core:A5|closed|qwen3b:h4", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:17:32.258957Z", "task_family": "synthetic", "document_id": "syn_03009_2000", "document_hash": "6ed94dcf728255bd7a7d11c97c362930728fb81a3e91f1906eb6eec86e0fedb7", "question_id": "q07_negation", "split": "test", "cluster_id": "syn_03009_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 4, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1976, "query_tokens": 317, "message_tokens": 234, "message_utf8_bytes": 698, "cache_tensor_bytes": null, "serialized_payload_bytes": 698, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 46608.10408287216, "hops": {"1": 16386.116749956273, "2": 9978.41195797082, "3": 9960.819375002757, "4": 10282.75599994231}, "receiver_prefill_ms": 239.20204199384898, "ttft_ms": 30.479207984171808, "decode_ms": 41.918625007383525}, "total_wall_ms": 46920.084415818565, "ttft_ms": 30.479207984171808, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "no", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 12, "retention_rate": 0.18461538461538463, "retention_by_kind": {"person": 1.0, "date": 0.0, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 698}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 12, "retention_rate": 0.18461538461538463, "retention_by_kind": {"person": 1.0, "date": 0.0, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 698}, "2": {"facts_total": 65, "facts_retained": 12, "retention_rate": 0.18461538461538463, "retention_by_kind": {"person": 1.0, "date": 0.0, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 698}, "3": {"facts_total": 65, "facts_retained": 12, "retention_rate": 0.18461538461538463, "retention_by_kind": {"person": 1.0, "date": 0.0, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 698}, "4": {"facts_total": 65, "facts_retained": 12, "retention_rate": 0.18461538461538463, "retention_by_kind": {"person": 1.0, "date": 0.0, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 698}}}, "reference_key_id": "keys/synthetic/syn_03009_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "b7aa2d9d5f9f3a06115e4e65c8ffaf4be2a07d6d067ceac89e49c6efeb6fdd5f", "output_hash": "9390298f3fb0c5b160498935d79cb139aef28e1c47358b4bbba61862b9c26e59"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "3f84c5d45b9ac974466cd3e5", "attempt_id": "6bb99e1b326f", "run_group_id": "E4_core:A5|closed|qwen3b:h4", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:17:32.588253Z", "task_family": "synthetic", "document_id": "syn_03009_2000", "document_hash": "6ed94dcf728255bd7a7d11c97c362930728fb81a3e91f1906eb6eec86e0fedb7", "question_id": "q08_revision", "split": "test", "cluster_id": "syn_03009_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 4, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1976, "query_tokens": 335, "message_tokens": 234, "message_utf8_bytes": 698, "cache_tensor_bytes": null, "serialized_payload_bytes": 698, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 46608.10408287216, "hops": {"1": 16386.116749956273, "2": 9978.41195797082, "3": 9960.819375002757, "4": 10282.75599994231}, "receiver_prefill_ms": 274.76558298803866, "ttft_ms": 30.694665969349444, "decode_ms": 25.015084072947502}, "total_wall_ms": 46939.11412486341, "ttft_ms": 30.694665969349444, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 12, "retention_rate": 0.18461538461538463, "retention_by_kind": {"person": 1.0, "date": 0.0, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 698}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 12, "retention_rate": 0.18461538461538463, "retention_by_kind": {"person": 1.0, "date": 0.0, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 698}, "2": {"facts_total": 65, "facts_retained": 12, "retention_rate": 0.18461538461538463, "retention_by_kind": {"person": 1.0, "date": 0.0, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 698}, "3": {"facts_total": 65, "facts_retained": 12, "retention_rate": 0.18461538461538463, "retention_by_kind": {"person": 1.0, "date": 0.0, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 698}, "4": {"facts_total": 65, "facts_retained": 12, "retention_rate": 0.18461538461538463, "retention_by_kind": {"person": 1.0, "date": 0.0, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 698}}}, "reference_key_id": "keys/synthetic/syn_03009_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "b7aa2d9d5f9f3a06115e4e65c8ffaf4be2a07d6d067ceac89e49c6efeb6fdd5f", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "e6bdfcacb64c2f557e13e3f7", "attempt_id": "b375ebe36b69", "run_group_id": "E4_core:A5|closed|qwen3b:h4", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:17:32.937087Z", "task_family": "synthetic", "document_id": "syn_03009_2000", "document_hash": "6ed94dcf728255bd7a7d11c97c362930728fb81a3e91f1906eb6eec86e0fedb7", "question_id": "q09_revision", "split": "test", "cluster_id": "syn_03009_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 4, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1976, "query_tokens": 328, "message_tokens": 234, "message_utf8_bytes": 698, "cache_tensor_bytes": null, "serialized_payload_bytes": 698, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 46608.10408287216, "hops": {"1": 16386.116749956273, "2": 9978.41195797082, "3": 9960.819375002757, "4": 10282.75599994231}, "receiver_prefill_ms": 230.07679102011025, "ttft_ms": 38.49304199684411, "decode_ms": 57.00633302330971}, "total_wall_ms": 46934.433374903165, "ttft_ms": 38.49304199684411, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 12, "retention_rate": 0.18461538461538463, "retention_by_kind": {"person": 1.0, "date": 0.0, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 698}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 12, "retention_rate": 0.18461538461538463, "retention_by_kind": {"person": 1.0, "date": 0.0, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 698}, "2": {"facts_total": 65, "facts_retained": 12, "retention_rate": 0.18461538461538463, "retention_by_kind": {"person": 1.0, "date": 0.0, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 698}, "3": {"facts_total": 65, "facts_retained": 12, "retention_rate": 0.18461538461538463, "retention_by_kind": {"person": 1.0, "date": 0.0, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 698}, "4": {"facts_total": 65, "facts_retained": 12, "retention_rate": 0.18461538461538463, "retention_by_kind": {"person": 1.0, "date": 0.0, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 698}}}, "reference_key_id": "keys/synthetic/syn_03009_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "b7aa2d9d5f9f3a06115e4e65c8ffaf4be2a07d6d067ceac89e49c6efeb6fdd5f", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "f4813917336612c14f465207", "attempt_id": "e62ea52c5938", "run_group_id": "E4_core:A5|closed|qwen3b:h4", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:17:33.279757Z", "task_family": "synthetic", "document_id": "syn_03009_2000", "document_hash": "6ed94dcf728255bd7a7d11c97c362930728fb81a3e91f1906eb6eec86e0fedb7", "question_id": "q10_temporal", "split": "test", "cluster_id": "syn_03009_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 4, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1976, "query_tokens": 337, "message_tokens": 234, "message_utf8_bytes": 698, "cache_tensor_bytes": null, "serialized_payload_bytes": 698, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 46608.10408287216, "hops": {"1": 16386.116749956273, "2": 9978.41195797082, "3": 9960.819375002757, "4": 10282.75599994231}, "receiver_prefill_ms": 325.1001660246402, "ttft_ms": 71.95729203522205, "decode_ms": 331.3941659871489}, "total_wall_ms": 47336.99024992529, "ttft_ms": 71.95729203522205, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 7}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "Eshgamgam-11", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 12, "retention_rate": 0.18461538461538463, "retention_by_kind": {"person": 1.0, "date": 0.0, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 698}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 12, "retention_rate": 0.18461538461538463, "retention_by_kind": {"person": 1.0, "date": 0.0, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 698}, "2": {"facts_total": 65, "facts_retained": 12, "retention_rate": 0.18461538461538463, "retention_by_kind": {"person": 1.0, "date": 0.0, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 698}, "3": {"facts_total": 65, "facts_retained": 12, "retention_rate": 0.18461538461538463, "retention_by_kind": {"person": 1.0, "date": 0.0, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 698}, "4": {"facts_total": 65, "facts_retained": 12, "retention_rate": 0.18461538461538463, "retention_by_kind": {"person": 1.0, "date": 0.0, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 698}}}, "reference_key_id": "keys/synthetic/syn_03009_2000.json", "task_scores": {"correct": 0.0, "abstained": 0.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "b7aa2d9d5f9f3a06115e4e65c8ffaf4be2a07d6d067ceac89e49c6efeb6fdd5f", "output_hash": "8c1313c53ab4dbffa5f336d721c810678133eecaa467241e53d8a92150601475"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "70eb633c7a6efb3bbc745d48", "attempt_id": "248ec5bcd6d9", "run_group_id": "E4_core:A5|closed|qwen3b:h4", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:17:34.025279Z", "task_family": "synthetic", "document_id": "syn_03009_2000", "document_hash": "6ed94dcf728255bd7a7d11c97c362930728fb81a3e91f1906eb6eec86e0fedb7", "question_id": "q11_aggregation", "split": "test", "cluster_id": "syn_03009_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 4, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1976, "query_tokens": 328, "message_tokens": 234, "message_utf8_bytes": 698, "cache_tensor_bytes": null, "serialized_payload_bytes": 698, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 46608.10408287216, "hops": {"1": 16386.116749956273, "2": 9978.41195797082, "3": 9960.819375002757, "4": 10282.75599994231}, "receiver_prefill_ms": 239.1870419960469, "ttft_ms": 39.85633293632418, "decode_ms": 30.948625062592328}, "total_wall_ms": 46918.45270781778, "ttft_ms": 39.85633293632418, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 12, "retention_rate": 0.18461538461538463, "retention_by_kind": {"person": 1.0, "date": 0.0, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 698}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 12, "retention_rate": 0.18461538461538463, "retention_by_kind": {"person": 1.0, "date": 0.0, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 698}, "2": {"facts_total": 65, "facts_retained": 12, "retention_rate": 0.18461538461538463, "retention_by_kind": {"person": 1.0, "date": 0.0, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 698}, "3": {"facts_total": 65, "facts_retained": 12, "retention_rate": 0.18461538461538463, "retention_by_kind": {"person": 1.0, "date": 0.0, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 698}, "4": {"facts_total": 65, "facts_retained": 12, "retention_rate": 0.18461538461538463, "retention_by_kind": {"person": 1.0, "date": 0.0, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 698}}}, "reference_key_id": "keys/synthetic/syn_03009_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "b7aa2d9d5f9f3a06115e4e65c8ffaf4be2a07d6d067ceac89e49c6efeb6fdd5f", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "a00cde834ede22e92245e677", "attempt_id": "05276ff067ea", "run_group_id": "E4_core:A5|closed|qwen3b:h4", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:17:34.353144Z", "task_family": "synthetic", "document_id": "syn_03009_2000", "document_hash": "6ed94dcf728255bd7a7d11c97c362930728fb81a3e91f1906eb6eec86e0fedb7", "question_id": "q12_unit_conversion", "split": "test", "cluster_id": "syn_03009_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 4, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1976, "query_tokens": 326, "message_tokens": 234, "message_utf8_bytes": 698, "cache_tensor_bytes": null, "serialized_payload_bytes": 698, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 46608.10408287216, "hops": {"1": 16386.116749956273, "2": 9978.41195797082, "3": 9960.819375002757, "4": 10282.75599994231}, "receiver_prefill_ms": 254.72341594286263, "ttft_ms": 31.80537500884384, "decode_ms": 19.926666980609298}, "total_wall_ms": 46914.97470787726, "ttft_ms": 31.80537500884384, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 12, "retention_rate": 0.18461538461538463, "retention_by_kind": {"person": 1.0, "date": 0.0, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 698}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 12, "retention_rate": 0.18461538461538463, "retention_by_kind": {"person": 1.0, "date": 0.0, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 698}, "2": {"facts_total": 65, "facts_retained": 12, "retention_rate": 0.18461538461538463, "retention_by_kind": {"person": 1.0, "date": 0.0, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 698}, "3": {"facts_total": 65, "facts_retained": 12, "retention_rate": 0.18461538461538463, "retention_by_kind": {"person": 1.0, "date": 0.0, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 698}, "4": {"facts_total": 65, "facts_retained": 12, "retention_rate": 0.18461538461538463, "retention_by_kind": {"person": 1.0, "date": 0.0, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 698}}}, "reference_key_id": "keys/synthetic/syn_03009_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "b7aa2d9d5f9f3a06115e4e65c8ffaf4be2a07d6d067ceac89e49c6efeb6fdd5f", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "50d5fd9bdf68c71645208bd4", "attempt_id": "c26df105b0d1", "run_group_id": "E4_core:A5|closed|qwen3b:h4", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:17:34.676484Z", "task_family": "synthetic", "document_id": "syn_03009_2000", "document_hash": "6ed94dcf728255bd7a7d11c97c362930728fb81a3e91f1906eb6eec86e0fedb7", "question_id": "q13_unanswerable", "split": "test", "cluster_id": "syn_03009_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 4, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1976, "query_tokens": 330, "message_tokens": 234, "message_utf8_bytes": 698, "cache_tensor_bytes": null, "serialized_payload_bytes": 698, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 46608.10408287216, "hops": {"1": 16386.116749956273, "2": 9978.41195797082, "3": 9960.819375002757, "4": 10282.75599994231}, "receiver_prefill_ms": 300.32979196403176, "ttft_ms": 51.80158303119242, "decode_ms": 50.57970795314759}, "total_wall_ms": 47011.36049989145, "ttft_ms": 51.80158303119242, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 12, "retention_rate": 0.18461538461538463, "retention_by_kind": {"person": 1.0, "date": 0.0, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 698}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 12, "retention_rate": 0.18461538461538463, "retention_by_kind": {"person": 1.0, "date": 0.0, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 698}, "2": {"facts_total": 65, "facts_retained": 12, "retention_rate": 0.18461538461538463, "retention_by_kind": {"person": 1.0, "date": 0.0, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 698}, "3": {"facts_total": 65, "facts_retained": 12, "retention_rate": 0.18461538461538463, "retention_by_kind": {"person": 1.0, "date": 0.0, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 698}, "4": {"facts_total": 65, "facts_retained": 12, "retention_rate": 0.18461538461538463, "retention_by_kind": {"person": 1.0, "date": 0.0, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 698}}}, "reference_key_id": "keys/synthetic/syn_03009_2000.json", "task_scores": {"correct": 1.0, "abstained": 1.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "b7aa2d9d5f9f3a06115e4e65c8ffaf4be2a07d6d067ceac89e49c6efeb6fdd5f", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "f75ed9fb0e719ed35db703b5", "attempt_id": "1c56a7c8f54a", "run_group_id": "E4_core:A5|closed|qwen3b:h4", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:17:35.099493Z", "task_family": "synthetic", "document_id": "syn_03009_2000", "document_hash": "6ed94dcf728255bd7a7d11c97c362930728fb81a3e91f1906eb6eec86e0fedb7", "question_id": "q14_unanswerable", "split": "test", "cluster_id": "syn_03009_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 4, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1976, "query_tokens": 330, "message_tokens": 234, "message_utf8_bytes": 698, "cache_tensor_bytes": null, "serialized_payload_bytes": 698, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 46608.10408287216, "hops": {"1": 16386.116749956273, "2": 9978.41195797082, "3": 9960.819375002757, "4": 10282.75599994231}, "receiver_prefill_ms": 393.91370792873204, "ttft_ms": 38.68333401624113, "decode_ms": 32.51891699619591}, "total_wall_ms": 47073.791373870336, "ttft_ms": 38.68333401624113, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 12, "retention_rate": 0.18461538461538463, "retention_by_kind": {"person": 1.0, "date": 0.0, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 698}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 12, "retention_rate": 0.18461538461538463, "retention_by_kind": {"person": 1.0, "date": 0.0, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 698}, "2": {"facts_total": 65, "facts_retained": 12, "retention_rate": 0.18461538461538463, "retention_by_kind": {"person": 1.0, "date": 0.0, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 698}, "3": {"facts_total": 65, "facts_retained": 12, "retention_rate": 0.18461538461538463, "retention_by_kind": {"person": 1.0, "date": 0.0, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 698}, "4": {"facts_total": 65, "facts_retained": 12, "retention_rate": 0.18461538461538463, "retention_by_kind": {"person": 1.0, "date": 0.0, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 698}}}, "reference_key_id": "keys/synthetic/syn_03009_2000.json", "task_scores": {"correct": 1.0, "abstained": 1.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "b7aa2d9d5f9f3a06115e4e65c8ffaf4be2a07d6d067ceac89e49c6efeb6fdd5f", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "a3dc7ea630b97333278920c9", "attempt_id": "7cb74db6f8fe", "run_group_id": "E4_core:A5|closed|qwen3b:h8", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:18:15.690130Z", "task_family": "synthetic", "document_id": "syn_03009_2000", "document_hash": "6ed94dcf728255bd7a7d11c97c362930728fb81a3e91f1906eb6eec86e0fedb7", "question_id": "q00_single_fact", "split": "test", "cluster_id": "syn_03009_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 8, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1976, "query_tokens": 318, "message_tokens": 234, "message_utf8_bytes": 698, "cache_tensor_bytes": null, "serialized_payload_bytes": 698, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 86710.96616482828, "hops": {"1": 16386.116749956273, "2": 9978.41195797082, "3": 9960.819375002757, "4": 10282.75599994231, "5": 9938.042040914297, "6": 10059.49283298105, "7": 10103.797167073935, "8": 10001.530040986836}, "receiver_prefill_ms": 201.34204102214426, "ttft_ms": 55.336124962195754, "decode_ms": 52.13141592685133}, "total_wall_ms": 87020.33762377687, "ttft_ms": 55.336124962195754, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 12, "retention_rate": 0.18461538461538463, "retention_by_kind": {"person": 1.0, "date": 0.0, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 698}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 12, "retention_rate": 0.18461538461538463, "retention_by_kind": {"person": 1.0, "date": 0.0, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 698}, "2": {"facts_total": 65, "facts_retained": 12, "retention_rate": 0.18461538461538463, "retention_by_kind": {"person": 1.0, "date": 0.0, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 698}, "3": {"facts_total": 65, "facts_retained": 12, "retention_rate": 0.18461538461538463, "retention_by_kind": {"person": 1.0, "date": 0.0, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 698}, "4": {"facts_total": 65, "facts_retained": 12, "retention_rate": 0.18461538461538463, "retention_by_kind": {"person": 1.0, "date": 0.0, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 698}, "5": {"facts_total": 65, "facts_retained": 12, "retention_rate": 0.18461538461538463, "retention_by_kind": {"person": 1.0, "date": 0.0, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 698}, "6": {"facts_total": 65, "facts_retained": 12, "retention_rate": 0.18461538461538463, "retention_by_kind": {"person": 1.0, "date": 0.0, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 698}, "7": {"facts_total": 65, "facts_retained": 12, "retention_rate": 0.18461538461538463, "retention_by_kind": {"person": 1.0, "date": 0.0, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 698}, "8": {"facts_total": 65, "facts_retained": 12, "retention_rate": 0.18461538461538463, "retention_by_kind": {"person": 1.0, "date": 0.0, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 698}}}, "reference_key_id": "keys/synthetic/syn_03009_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "b7aa2d9d5f9f3a06115e4e65c8ffaf4be2a07d6d067ceac89e49c6efeb6fdd5f", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "aa17994c3770fba63e4409c5", "attempt_id": "d7a639211afa", "run_group_id": "E4_core:A5|closed|qwen3b:h8", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:18:16.016588Z", "task_family": "synthetic", "document_id": "syn_03009_2000", "document_hash": "6ed94dcf728255bd7a7d11c97c362930728fb81a3e91f1906eb6eec86e0fedb7", "question_id": "q01_single_fact", "split": "test", "cluster_id": "syn_03009_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 8, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1976, "query_tokens": 323, "message_tokens": 234, "message_utf8_bytes": 698, "cache_tensor_bytes": null, "serialized_payload_bytes": 698, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 86710.96616482828, "hops": {"1": 16386.116749956273, "2": 9978.41195797082, "3": 9960.819375002757, "4": 10282.75599994231, "5": 9938.042040914297, "6": 10059.49283298105, "7": 10103.797167073935, "8": 10001.530040986836}, "receiver_prefill_ms": 433.52120905183256, "ttft_ms": 41.69383295811713, "decode_ms": 30.336625059135258}, "total_wall_ms": 87217.20474783797, "ttft_ms": 41.69383295811713, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 12, "retention_rate": 0.18461538461538463, "retention_by_kind": {"person": 1.0, "date": 0.0, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 698}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 12, "retention_rate": 0.18461538461538463, "retention_by_kind": {"person": 1.0, "date": 0.0, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 698}, "2": {"facts_total": 65, "facts_retained": 12, "retention_rate": 0.18461538461538463, "retention_by_kind": {"person": 1.0, "date": 0.0, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 698}, "3": {"facts_total": 65, "facts_retained": 12, "retention_rate": 0.18461538461538463, "retention_by_kind": {"person": 1.0, "date": 0.0, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 698}, "4": {"facts_total": 65, "facts_retained": 12, "retention_rate": 0.18461538461538463, "retention_by_kind": {"person": 1.0, "date": 0.0, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 698}, "5": {"facts_total": 65, "facts_retained": 12, "retention_rate": 0.18461538461538463, "retention_by_kind": {"person": 1.0, "date": 0.0, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 698}, "6": {"facts_total": 65, "facts_retained": 12, "retention_rate": 0.18461538461538463, "retention_by_kind": {"person": 1.0, "date": 0.0, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 698}, "7": {"facts_total": 65, "facts_retained": 12, "retention_rate": 0.18461538461538463, "retention_by_kind": {"person": 1.0, "date": 0.0, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 698}, "8": {"facts_total": 65, "facts_retained": 12, "retention_rate": 0.18461538461538463, "retention_by_kind": {"person": 1.0, "date": 0.0, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 698}}}, "reference_key_id": "keys/synthetic/syn_03009_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "b7aa2d9d5f9f3a06115e4e65c8ffaf4be2a07d6d067ceac89e49c6efeb6fdd5f", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "a4921c34ba15194dd85f10a4", "attempt_id": "b4b14a62c494", "run_group_id": "E4_core:A5|closed|qwen3b:h8", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:18:16.541552Z", "task_family": "synthetic", "document_id": "syn_03009_2000", "document_hash": "6ed94dcf728255bd7a7d11c97c362930728fb81a3e91f1906eb6eec86e0fedb7", "question_id": "q02_exact_copy", "split": "test", "cluster_id": "syn_03009_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 8, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1976, "query_tokens": 321, "message_tokens": 234, "message_utf8_bytes": 698, "cache_tensor_bytes": null, "serialized_payload_bytes": 698, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 86710.96616482828, "hops": {"1": 16386.116749956273, "2": 9978.41195797082, "3": 9960.819375002757, "4": 10282.75599994231, "5": 9938.042040914297, "6": 10059.49283298105, "7": 10103.797167073935, "8": 10001.530040986836}, "receiver_prefill_ms": 217.080874950625, "ttft_ms": 37.618916016072035, "decode_ms": 26.542792096734047}, "total_wall_ms": 86992.79599788133, "ttft_ms": 37.618916016072035, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 12, "retention_rate": 0.18461538461538463, "retention_by_kind": {"person": 1.0, "date": 0.0, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 698}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 12, "retention_rate": 0.18461538461538463, "retention_by_kind": {"person": 1.0, "date": 0.0, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 698}, "2": {"facts_total": 65, "facts_retained": 12, "retention_rate": 0.18461538461538463, "retention_by_kind": {"person": 1.0, "date": 0.0, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 698}, "3": {"facts_total": 65, "facts_retained": 12, "retention_rate": 0.18461538461538463, "retention_by_kind": {"person": 1.0, "date": 0.0, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 698}, "4": {"facts_total": 65, "facts_retained": 12, "retention_rate": 0.18461538461538463, "retention_by_kind": {"person": 1.0, "date": 0.0, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 698}, "5": {"facts_total": 65, "facts_retained": 12, "retention_rate": 0.18461538461538463, "retention_by_kind": {"person": 1.0, "date": 0.0, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 698}, "6": {"facts_total": 65, "facts_retained": 12, "retention_rate": 0.18461538461538463, "retention_by_kind": {"person": 1.0, "date": 0.0, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 698}, "7": {"facts_total": 65, "facts_retained": 12, "retention_rate": 0.18461538461538463, "retention_by_kind": {"person": 1.0, "date": 0.0, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 698}, "8": {"facts_total": 65, "facts_retained": 12, "retention_rate": 0.18461538461538463, "retention_by_kind": {"person": 1.0, "date": 0.0, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 698}}}, "reference_key_id": "keys/synthetic/syn_03009_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "b7aa2d9d5f9f3a06115e4e65c8ffaf4be2a07d6d067ceac89e49c6efeb6fdd5f", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "5bcb44b258d3b044f188b4e9", "attempt_id": "26623eeae1fb", "run_group_id": "E4_core:A5|closed|qwen3b:h8", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:18:16.840362Z", "task_family": "synthetic", "document_id": "syn_03009_2000", "document_hash": "6ed94dcf728255bd7a7d11c97c362930728fb81a3e91f1906eb6eec86e0fedb7", "question_id": "q03_exact_copy", "split": "test", "cluster_id": "syn_03009_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 8, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1976, "query_tokens": 317, "message_tokens": 234, "message_utf8_bytes": 698, "cache_tensor_bytes": null, "serialized_payload_bytes": 698, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 86710.96616482828, "hops": {"1": 16386.116749956273, "2": 9978.41195797082, "3": 9960.819375002757, "4": 10282.75599994231, "5": 9938.042040914297, "6": 10059.49283298105, "7": 10103.797167073935, "8": 10001.530040986836}, "receiver_prefill_ms": 234.8147500306368, "ttft_ms": 23.610999924130738, "decode_ms": 162.04550000838935}, "total_wall_ms": 87131.88791484572, "ttft_ms": 23.610999924130738, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 5}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "Idris Bergström", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 12, "retention_rate": 0.18461538461538463, "retention_by_kind": {"person": 1.0, "date": 0.0, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 698}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 12, "retention_rate": 0.18461538461538463, "retention_by_kind": {"person": 1.0, "date": 0.0, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 698}, "2": {"facts_total": 65, "facts_retained": 12, "retention_rate": 0.18461538461538463, "retention_by_kind": {"person": 1.0, "date": 0.0, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 698}, "3": {"facts_total": 65, "facts_retained": 12, "retention_rate": 0.18461538461538463, "retention_by_kind": {"person": 1.0, "date": 0.0, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 698}, "4": {"facts_total": 65, "facts_retained": 12, "retention_rate": 0.18461538461538463, "retention_by_kind": {"person": 1.0, "date": 0.0, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 698}, "5": {"facts_total": 65, "facts_retained": 12, "retention_rate": 0.18461538461538463, "retention_by_kind": {"person": 1.0, "date": 0.0, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 698}, "6": {"facts_total": 65, "facts_retained": 12, "retention_rate": 0.18461538461538463, "retention_by_kind": {"person": 1.0, "date": 0.0, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 698}, "7": {"facts_total": 65, "facts_retained": 12, "retention_rate": 0.18461538461538463, "retention_by_kind": {"person": 1.0, "date": 0.0, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 698}, "8": {"facts_total": 65, "facts_retained": 12, "retention_rate": 0.18461538461538463, "retention_by_kind": {"person": 1.0, "date": 0.0, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 698}}}, "reference_key_id": "keys/synthetic/syn_03009_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "b7aa2d9d5f9f3a06115e4e65c8ffaf4be2a07d6d067ceac89e49c6efeb6fdd5f", "output_hash": "fd88832c9fd0ab175818d76d875d70156225bd4b34ec401514546ef70f4f826c"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "5d85667d22fa381ee7e7da44", "attempt_id": "ecbca4b3a89f", "run_group_id": "E4_core:A5|closed|qwen3b:h8", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:18:17.278173Z", "task_family": "synthetic", "document_id": "syn_03009_2000", "document_hash": "6ed94dcf728255bd7a7d11c97c362930728fb81a3e91f1906eb6eec86e0fedb7", "question_id": "q04_multi_hop", "split": "test", "cluster_id": "syn_03009_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 8, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1976, "query_tokens": 325, "message_tokens": 234, "message_utf8_bytes": 698, "cache_tensor_bytes": null, "serialized_payload_bytes": 698, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 86710.96616482828, "hops": {"1": 16386.116749956273, "2": 9978.41195797082, "3": 9960.819375002757, "4": 10282.75599994231, "5": 9938.042040914297, "6": 10059.49283298105, "7": 10103.797167073935, "8": 10001.530040986836}, "receiver_prefill_ms": 278.73420901596546, "ttft_ms": 59.866833966225386, "decode_ms": 52.5285410694778}, "total_wall_ms": 87102.87958185654, "ttft_ms": 59.866833966225386, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 12, "retention_rate": 0.18461538461538463, "retention_by_kind": {"person": 1.0, "date": 0.0, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 698}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 12, "retention_rate": 0.18461538461538463, "retention_by_kind": {"person": 1.0, "date": 0.0, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 698}, "2": {"facts_total": 65, "facts_retained": 12, "retention_rate": 0.18461538461538463, "retention_by_kind": {"person": 1.0, "date": 0.0, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 698}, "3": {"facts_total": 65, "facts_retained": 12, "retention_rate": 0.18461538461538463, "retention_by_kind": {"person": 1.0, "date": 0.0, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 698}, "4": {"facts_total": 65, "facts_retained": 12, "retention_rate": 0.18461538461538463, "retention_by_kind": {"person": 1.0, "date": 0.0, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 698}, "5": {"facts_total": 65, "facts_retained": 12, "retention_rate": 0.18461538461538463, "retention_by_kind": {"person": 1.0, "date": 0.0, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 698}, "6": {"facts_total": 65, "facts_retained": 12, "retention_rate": 0.18461538461538463, "retention_by_kind": {"person": 1.0, "date": 0.0, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 698}, "7": {"facts_total": 65, "facts_retained": 12, "retention_rate": 0.18461538461538463, "retention_by_kind": {"person": 1.0, "date": 0.0, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 698}, "8": {"facts_total": 65, "facts_retained": 12, "retention_rate": 0.18461538461538463, "retention_by_kind": {"person": 1.0, "date": 0.0, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 698}}}, "reference_key_id": "keys/synthetic/syn_03009_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "b7aa2d9d5f9f3a06115e4e65c8ffaf4be2a07d6d067ceac89e49c6efeb6fdd5f", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "55246b0ed052a68ed3415f2d", "attempt_id": "093fc3e4215f", "run_group_id": "E4_core:A5|closed|qwen3b:h8", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:18:17.688116Z", "task_family": "synthetic", "document_id": "syn_03009_2000", "document_hash": "6ed94dcf728255bd7a7d11c97c362930728fb81a3e91f1906eb6eec86e0fedb7", "question_id": "q05_multi_hop", "split": "test", "cluster_id": "syn_03009_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 8, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1976, "query_tokens": 328, "message_tokens": 234, "message_utf8_bytes": 698, "cache_tensor_bytes": null, "serialized_payload_bytes": 698, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 86710.96616482828, "hops": {"1": 16386.116749956273, "2": 9978.41195797082, "3": 9960.819375002757, "4": 10282.75599994231, "5": 9938.042040914297, "6": 10059.49283298105, "7": 10103.797167073935, "8": 10001.530040986836}, "receiver_prefill_ms": 387.46737502515316, "ttft_ms": 39.4459170056507, "decode_ms": 32.019958016462624}, "total_wall_ms": 87170.4098728951, "ttft_ms": 39.4459170056507, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 12, "retention_rate": 0.18461538461538463, "retention_by_kind": {"person": 1.0, "date": 0.0, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 698}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 12, "retention_rate": 0.18461538461538463, "retention_by_kind": {"person": 1.0, "date": 0.0, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 698}, "2": {"facts_total": 65, "facts_retained": 12, "retention_rate": 0.18461538461538463, "retention_by_kind": {"person": 1.0, "date": 0.0, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 698}, "3": {"facts_total": 65, "facts_retained": 12, "retention_rate": 0.18461538461538463, "retention_by_kind": {"person": 1.0, "date": 0.0, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 698}, "4": {"facts_total": 65, "facts_retained": 12, "retention_rate": 0.18461538461538463, "retention_by_kind": {"person": 1.0, "date": 0.0, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 698}, "5": {"facts_total": 65, "facts_retained": 12, "retention_rate": 0.18461538461538463, "retention_by_kind": {"person": 1.0, "date": 0.0, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 698}, "6": {"facts_total": 65, "facts_retained": 12, "retention_rate": 0.18461538461538463, "retention_by_kind": {"person": 1.0, "date": 0.0, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 698}, "7": {"facts_total": 65, "facts_retained": 12, "retention_rate": 0.18461538461538463, "retention_by_kind": {"person": 1.0, "date": 0.0, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 698}, "8": {"facts_total": 65, "facts_retained": 12, "retention_rate": 0.18461538461538463, "retention_by_kind": {"person": 1.0, "date": 0.0, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 698}}}, "reference_key_id": "keys/synthetic/syn_03009_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "b7aa2d9d5f9f3a06115e4e65c8ffaf4be2a07d6d067ceac89e49c6efeb6fdd5f", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "12b66f59561dc613cdf7f8c5", "attempt_id": "57536c378266", "run_group_id": "E4_core:A5|closed|qwen3b:h8", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:18:18.164735Z", "task_family": "synthetic", "document_id": "syn_03009_2000", "document_hash": "6ed94dcf728255bd7a7d11c97c362930728fb81a3e91f1906eb6eec86e0fedb7", "question_id": "q06_negation", "split": "test", "cluster_id": "syn_03009_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 8, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1976, "query_tokens": 318, "message_tokens": 234, "message_utf8_bytes": 698, "cache_tensor_bytes": null, "serialized_payload_bytes": 698, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 86710.96616482828, "hops": {"1": 16386.116749956273, "2": 9978.41195797082, "3": 9960.819375002757, "4": 10282.75599994231, "5": 9938.042040914297, "6": 10059.49283298105, "7": 10103.797167073935, "8": 10001.530040986836}, "receiver_prefill_ms": 241.51833297219127, "ttft_ms": 31.26366692595184, "decode_ms": 47.19412501435727}, "total_wall_ms": 87031.37128986418, "ttft_ms": 31.26366692595184, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "no", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 12, "retention_rate": 0.18461538461538463, "retention_by_kind": {"person": 1.0, "date": 0.0, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 698}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 12, "retention_rate": 0.18461538461538463, "retention_by_kind": {"person": 1.0, "date": 0.0, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 698}, "2": {"facts_total": 65, "facts_retained": 12, "retention_rate": 0.18461538461538463, "retention_by_kind": {"person": 1.0, "date": 0.0, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 698}, "3": {"facts_total": 65, "facts_retained": 12, "retention_rate": 0.18461538461538463, "retention_by_kind": {"person": 1.0, "date": 0.0, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 698}, "4": {"facts_total": 65, "facts_retained": 12, "retention_rate": 0.18461538461538463, "retention_by_kind": {"person": 1.0, "date": 0.0, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 698}, "5": {"facts_total": 65, "facts_retained": 12, "retention_rate": 0.18461538461538463, "retention_by_kind": {"person": 1.0, "date": 0.0, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 698}, "6": {"facts_total": 65, "facts_retained": 12, "retention_rate": 0.18461538461538463, "retention_by_kind": {"person": 1.0, "date": 0.0, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 698}, "7": {"facts_total": 65, "facts_retained": 12, "retention_rate": 0.18461538461538463, "retention_by_kind": {"person": 1.0, "date": 0.0, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 698}, "8": {"facts_total": 65, "facts_retained": 12, "retention_rate": 0.18461538461538463, "retention_by_kind": {"person": 1.0, "date": 0.0, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 698}}}, "reference_key_id": "keys/synthetic/syn_03009_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "b7aa2d9d5f9f3a06115e4e65c8ffaf4be2a07d6d067ceac89e49c6efeb6fdd5f", "output_hash": "9390298f3fb0c5b160498935d79cb139aef28e1c47358b4bbba61862b9c26e59"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "549b8071faf6f617c0607f0a", "attempt_id": "3f60412502d3", "run_group_id": "E4_core:A5|closed|qwen3b:h8", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:18:18.502051Z", "task_family": "synthetic", "document_id": "syn_03009_2000", "document_hash": "6ed94dcf728255bd7a7d11c97c362930728fb81a3e91f1906eb6eec86e0fedb7", "question_id": "q07_negation", "split": "test", "cluster_id": "syn_03009_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 8, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1976, "query_tokens": 317, "message_tokens": 234, "message_utf8_bytes": 698, "cache_tensor_bytes": null, "serialized_payload_bytes": 698, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 86710.96616482828, "hops": {"1": 16386.116749956273, "2": 9978.41195797082, "3": 9960.819375002757, "4": 10282.75599994231, "5": 9938.042040914297, "6": 10059.49283298105, "7": 10103.797167073935, "8": 10001.530040986836}, "receiver_prefill_ms": 224.03358307201415, "ttft_ms": 36.575417034327984, "decode_ms": 23.837124928832054}, "total_wall_ms": 86995.81245682202, "ttft_ms": 36.575417034327984, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "no", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 12, "retention_rate": 0.18461538461538463, "retention_by_kind": {"person": 1.0, "date": 0.0, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 698}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 12, "retention_rate": 0.18461538461538463, "retention_by_kind": {"person": 1.0, "date": 0.0, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 698}, "2": {"facts_total": 65, "facts_retained": 12, "retention_rate": 0.18461538461538463, "retention_by_kind": {"person": 1.0, "date": 0.0, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 698}, "3": {"facts_total": 65, "facts_retained": 12, "retention_rate": 0.18461538461538463, "retention_by_kind": {"person": 1.0, "date": 0.0, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 698}, "4": {"facts_total": 65, "facts_retained": 12, "retention_rate": 0.18461538461538463, "retention_by_kind": {"person": 1.0, "date": 0.0, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 698}, "5": {"facts_total": 65, "facts_retained": 12, "retention_rate": 0.18461538461538463, "retention_by_kind": {"person": 1.0, "date": 0.0, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 698}, "6": {"facts_total": 65, "facts_retained": 12, "retention_rate": 0.18461538461538463, "retention_by_kind": {"person": 1.0, "date": 0.0, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 698}, "7": {"facts_total": 65, "facts_retained": 12, "retention_rate": 0.18461538461538463, "retention_by_kind": {"person": 1.0, "date": 0.0, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 698}, "8": {"facts_total": 65, "facts_retained": 12, "retention_rate": 0.18461538461538463, "retention_by_kind": {"person": 1.0, "date": 0.0, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 698}}}, "reference_key_id": "keys/synthetic/syn_03009_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "b7aa2d9d5f9f3a06115e4e65c8ffaf4be2a07d6d067ceac89e49c6efeb6fdd5f", "output_hash": "9390298f3fb0c5b160498935d79cb139aef28e1c47358b4bbba61862b9c26e59"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "707feb059de179cec0d8a298", "attempt_id": "2f63e6ac4100", "run_group_id": "E4_core:A5|closed|qwen3b:h8", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:18:18.803831Z", "task_family": "synthetic", "document_id": "syn_03009_2000", "document_hash": "6ed94dcf728255bd7a7d11c97c362930728fb81a3e91f1906eb6eec86e0fedb7", "question_id": "q08_revision", "split": "test", "cluster_id": "syn_03009_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 8, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1976, "query_tokens": 335, "message_tokens": 234, "message_utf8_bytes": 698, "cache_tensor_bytes": null, "serialized_payload_bytes": 698, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 86710.96616482828, "hops": {"1": 16386.116749956273, "2": 9978.41195797082, "3": 9960.819375002757, "4": 10282.75599994231, "5": 9938.042040914297, "6": 10059.49283298105, "7": 10103.797167073935, "8": 10001.530040986836}, "receiver_prefill_ms": 248.90554102603346, "ttft_ms": 58.71050001587719, "decode_ms": 55.164958932437}, "total_wall_ms": 87074.25178983249, "ttft_ms": 58.71050001587719, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 12, "retention_rate": 0.18461538461538463, "retention_by_kind": {"person": 1.0, "date": 0.0, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 698}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 12, "retention_rate": 0.18461538461538463, "retention_by_kind": {"person": 1.0, "date": 0.0, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 698}, "2": {"facts_total": 65, "facts_retained": 12, "retention_rate": 0.18461538461538463, "retention_by_kind": {"person": 1.0, "date": 0.0, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 698}, "3": {"facts_total": 65, "facts_retained": 12, "retention_rate": 0.18461538461538463, "retention_by_kind": {"person": 1.0, "date": 0.0, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 698}, "4": {"facts_total": 65, "facts_retained": 12, "retention_rate": 0.18461538461538463, "retention_by_kind": {"person": 1.0, "date": 0.0, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 698}, "5": {"facts_total": 65, "facts_retained": 12, "retention_rate": 0.18461538461538463, "retention_by_kind": {"person": 1.0, "date": 0.0, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 698}, "6": {"facts_total": 65, "facts_retained": 12, "retention_rate": 0.18461538461538463, "retention_by_kind": {"person": 1.0, "date": 0.0, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 698}, "7": {"facts_total": 65, "facts_retained": 12, "retention_rate": 0.18461538461538463, "retention_by_kind": {"person": 1.0, "date": 0.0, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 698}, "8": {"facts_total": 65, "facts_retained": 12, "retention_rate": 0.18461538461538463, "retention_by_kind": {"person": 1.0, "date": 0.0, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 698}}}, "reference_key_id": "keys/synthetic/syn_03009_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "b7aa2d9d5f9f3a06115e4e65c8ffaf4be2a07d6d067ceac89e49c6efeb6fdd5f", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "d41decd83b9c9f8e00f28cf9", "attempt_id": "e1cba6594a93", "run_group_id": "E4_core:A5|closed|qwen3b:h8", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:18:19.184177Z", "task_family": "synthetic", "document_id": "syn_03009_2000", "document_hash": "6ed94dcf728255bd7a7d11c97c362930728fb81a3e91f1906eb6eec86e0fedb7", "question_id": "q09_revision", "split": "test", "cluster_id": "syn_03009_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 8, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1976, "query_tokens": 328, "message_tokens": 234, "message_utf8_bytes": 698, "cache_tensor_bytes": null, "serialized_payload_bytes": 698, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 86710.96616482828, "hops": {"1": 16386.116749956273, "2": 9978.41195797082, "3": 9960.819375002757, "4": 10282.75599994231, "5": 9938.042040914297, "6": 10059.49283298105, "7": 10103.797167073935, "8": 10001.530040986836}, "receiver_prefill_ms": 315.06066699512303, "ttft_ms": 106.4465829404071, "decode_ms": 44.8947079712525}, "total_wall_ms": 87177.80041485094, "ttft_ms": 106.4465829404071, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 12, "retention_rate": 0.18461538461538463, "retention_by_kind": {"person": 1.0, "date": 0.0, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 698}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 12, "retention_rate": 0.18461538461538463, "retention_by_kind": {"person": 1.0, "date": 0.0, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 698}, "2": {"facts_total": 65, "facts_retained": 12, "retention_rate": 0.18461538461538463, "retention_by_kind": {"person": 1.0, "date": 0.0, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 698}, "3": {"facts_total": 65, "facts_retained": 12, "retention_rate": 0.18461538461538463, "retention_by_kind": {"person": 1.0, "date": 0.0, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 698}, "4": {"facts_total": 65, "facts_retained": 12, "retention_rate": 0.18461538461538463, "retention_by_kind": {"person": 1.0, "date": 0.0, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 698}, "5": {"facts_total": 65, "facts_retained": 12, "retention_rate": 0.18461538461538463, "retention_by_kind": {"person": 1.0, "date": 0.0, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 698}, "6": {"facts_total": 65, "facts_retained": 12, "retention_rate": 0.18461538461538463, "retention_by_kind": {"person": 1.0, "date": 0.0, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 698}, "7": {"facts_total": 65, "facts_retained": 12, "retention_rate": 0.18461538461538463, "retention_by_kind": {"person": 1.0, "date": 0.0, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 698}, "8": {"facts_total": 65, "facts_retained": 12, "retention_rate": 0.18461538461538463, "retention_by_kind": {"person": 1.0, "date": 0.0, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 698}}}, "reference_key_id": "keys/synthetic/syn_03009_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "b7aa2d9d5f9f3a06115e4e65c8ffaf4be2a07d6d067ceac89e49c6efeb6fdd5f", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "634da04b69c69f525188185d", "attempt_id": "39cf029a94d5", "run_group_id": "E4_core:A5|closed|qwen3b:h8", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:18:19.669290Z", "task_family": "synthetic", "document_id": "syn_03009_2000", "document_hash": "6ed94dcf728255bd7a7d11c97c362930728fb81a3e91f1906eb6eec86e0fedb7", "question_id": "q10_temporal", "split": "test", "cluster_id": "syn_03009_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 8, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1976, "query_tokens": 337, "message_tokens": 234, "message_utf8_bytes": 698, "cache_tensor_bytes": null, "serialized_payload_bytes": 698, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 86710.96616482828, "hops": {"1": 16386.116749956273, "2": 9978.41195797082, "3": 9960.819375002757, "4": 10282.75599994231, "5": 9938.042040914297, "6": 10059.49283298105, "7": 10103.797167073935, "8": 10001.530040986836}, "receiver_prefill_ms": 265.03562496509403, "ttft_ms": 24.94249993469566, "decode_ms": 275.36433399654925}, "total_wall_ms": 87276.90599882044, "ttft_ms": 24.94249993469566, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 7}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "Eshgamgam-11", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 12, "retention_rate": 0.18461538461538463, "retention_by_kind": {"person": 1.0, "date": 0.0, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 698}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 12, "retention_rate": 0.18461538461538463, "retention_by_kind": {"person": 1.0, "date": 0.0, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 698}, "2": {"facts_total": 65, "facts_retained": 12, "retention_rate": 0.18461538461538463, "retention_by_kind": {"person": 1.0, "date": 0.0, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 698}, "3": {"facts_total": 65, "facts_retained": 12, "retention_rate": 0.18461538461538463, "retention_by_kind": {"person": 1.0, "date": 0.0, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 698}, "4": {"facts_total": 65, "facts_retained": 12, "retention_rate": 0.18461538461538463, "retention_by_kind": {"person": 1.0, "date": 0.0, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 698}, "5": {"facts_total": 65, "facts_retained": 12, "retention_rate": 0.18461538461538463, "retention_by_kind": {"person": 1.0, "date": 0.0, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 698}, "6": {"facts_total": 65, "facts_retained": 12, "retention_rate": 0.18461538461538463, "retention_by_kind": {"person": 1.0, "date": 0.0, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 698}, "7": {"facts_total": 65, "facts_retained": 12, "retention_rate": 0.18461538461538463, "retention_by_kind": {"person": 1.0, "date": 0.0, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 698}, "8": {"facts_total": 65, "facts_retained": 12, "retention_rate": 0.18461538461538463, "retention_by_kind": {"person": 1.0, "date": 0.0, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 698}}}, "reference_key_id": "keys/synthetic/syn_03009_2000.json", "task_scores": {"correct": 0.0, "abstained": 0.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "b7aa2d9d5f9f3a06115e4e65c8ffaf4be2a07d6d067ceac89e49c6efeb6fdd5f", "output_hash": "8c1313c53ab4dbffa5f336d721c810678133eecaa467241e53d8a92150601475"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "ebe92b9aa76c56540d64cbee", "attempt_id": "294bddaef49e", "run_group_id": "E4_core:A5|closed|qwen3b:h8", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:18:20.252808Z", "task_family": "synthetic", "document_id": "syn_03009_2000", "document_hash": "6ed94dcf728255bd7a7d11c97c362930728fb81a3e91f1906eb6eec86e0fedb7", "question_id": "q11_aggregation", "split": "test", "cluster_id": "syn_03009_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 8, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1976, "query_tokens": 328, "message_tokens": 234, "message_utf8_bytes": 698, "cache_tensor_bytes": null, "serialized_payload_bytes": 698, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 86710.96616482828, "hops": {"1": 16386.116749956273, "2": 9978.41195797082, "3": 9960.819375002757, "4": 10282.75599994231, "5": 9938.042040914297, "6": 10059.49283298105, "7": 10103.797167073935, "8": 10001.530040986836}, "receiver_prefill_ms": 247.92487500235438, "ttft_ms": 53.70712501462549, "decode_ms": 74.19404201209545}, "total_wall_ms": 87087.22908177879, "ttft_ms": 53.70712501462549, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 12, "retention_rate": 0.18461538461538463, "retention_by_kind": {"person": 1.0, "date": 0.0, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 698}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 12, "retention_rate": 0.18461538461538463, "retention_by_kind": {"person": 1.0, "date": 0.0, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 698}, "2": {"facts_total": 65, "facts_retained": 12, "retention_rate": 0.18461538461538463, "retention_by_kind": {"person": 1.0, "date": 0.0, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 698}, "3": {"facts_total": 65, "facts_retained": 12, "retention_rate": 0.18461538461538463, "retention_by_kind": {"person": 1.0, "date": 0.0, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 698}, "4": {"facts_total": 65, "facts_retained": 12, "retention_rate": 0.18461538461538463, "retention_by_kind": {"person": 1.0, "date": 0.0, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 698}, "5": {"facts_total": 65, "facts_retained": 12, "retention_rate": 0.18461538461538463, "retention_by_kind": {"person": 1.0, "date": 0.0, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 698}, "6": {"facts_total": 65, "facts_retained": 12, "retention_rate": 0.18461538461538463, "retention_by_kind": {"person": 1.0, "date": 0.0, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 698}, "7": {"facts_total": 65, "facts_retained": 12, "retention_rate": 0.18461538461538463, "retention_by_kind": {"person": 1.0, "date": 0.0, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 698}, "8": {"facts_total": 65, "facts_retained": 12, "retention_rate": 0.18461538461538463, "retention_by_kind": {"person": 1.0, "date": 0.0, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 698}}}, "reference_key_id": "keys/synthetic/syn_03009_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "b7aa2d9d5f9f3a06115e4e65c8ffaf4be2a07d6d067ceac89e49c6efeb6fdd5f", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "f3efcd43fa50305567e7096f", "attempt_id": "b63c1e55da69", "run_group_id": "E4_core:A5|closed|qwen3b:h8", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:18:20.646913Z", "task_family": "synthetic", "document_id": "syn_03009_2000", "document_hash": "6ed94dcf728255bd7a7d11c97c362930728fb81a3e91f1906eb6eec86e0fedb7", "question_id": "q12_unit_conversion", "split": "test", "cluster_id": "syn_03009_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 8, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1976, "query_tokens": 326, "message_tokens": 234, "message_utf8_bytes": 698, "cache_tensor_bytes": null, "serialized_payload_bytes": 698, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 86710.96616482828, "hops": {"1": 16386.116749956273, "2": 9978.41195797082, "3": 9960.819375002757, "4": 10282.75599994231, "5": 9938.042040914297, "6": 10059.49283298105, "7": 10103.797167073935, "8": 10001.530040986836}, "receiver_prefill_ms": 257.119124988094, "ttft_ms": 115.06237497087568, "decode_ms": 105.2352919941768}, "total_wall_ms": 87188.92999785021, "ttft_ms": 115.06237497087568, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 12, "retention_rate": 0.18461538461538463, "retention_by_kind": {"person": 1.0, "date": 0.0, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 698}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 12, "retention_rate": 0.18461538461538463, "retention_by_kind": {"person": 1.0, "date": 0.0, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 698}, "2": {"facts_total": 65, "facts_retained": 12, "retention_rate": 0.18461538461538463, "retention_by_kind": {"person": 1.0, "date": 0.0, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 698}, "3": {"facts_total": 65, "facts_retained": 12, "retention_rate": 0.18461538461538463, "retention_by_kind": {"person": 1.0, "date": 0.0, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 698}, "4": {"facts_total": 65, "facts_retained": 12, "retention_rate": 0.18461538461538463, "retention_by_kind": {"person": 1.0, "date": 0.0, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 698}, "5": {"facts_total": 65, "facts_retained": 12, "retention_rate": 0.18461538461538463, "retention_by_kind": {"person": 1.0, "date": 0.0, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 698}, "6": {"facts_total": 65, "facts_retained": 12, "retention_rate": 0.18461538461538463, "retention_by_kind": {"person": 1.0, "date": 0.0, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 698}, "7": {"facts_total": 65, "facts_retained": 12, "retention_rate": 0.18461538461538463, "retention_by_kind": {"person": 1.0, "date": 0.0, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 698}, "8": {"facts_total": 65, "facts_retained": 12, "retention_rate": 0.18461538461538463, "retention_by_kind": {"person": 1.0, "date": 0.0, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 698}}}, "reference_key_id": "keys/synthetic/syn_03009_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "b7aa2d9d5f9f3a06115e4e65c8ffaf4be2a07d6d067ceac89e49c6efeb6fdd5f", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "df5caf6b8ada32ab8aa63559", "attempt_id": "664ad36d2008", "run_group_id": "E4_core:A5|closed|qwen3b:h8", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:18:21.143577Z", "task_family": "synthetic", "document_id": "syn_03009_2000", "document_hash": "6ed94dcf728255bd7a7d11c97c362930728fb81a3e91f1906eb6eec86e0fedb7", "question_id": "q13_unanswerable", "split": "test", "cluster_id": "syn_03009_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 8, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1976, "query_tokens": 330, "message_tokens": 234, "message_utf8_bytes": 698, "cache_tensor_bytes": null, "serialized_payload_bytes": 698, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 86710.96616482828, "hops": {"1": 16386.116749956273, "2": 9978.41195797082, "3": 9960.819375002757, "4": 10282.75599994231, "5": 9938.042040914297, "6": 10059.49283298105, "7": 10103.797167073935, "8": 10001.530040986836}, "receiver_prefill_ms": 253.11741698533297, "ttft_ms": 32.92929090093821, "decode_ms": 38.29645900987089}, "total_wall_ms": 87035.7750399271, "ttft_ms": 32.92929090093821, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 12, "retention_rate": 0.18461538461538463, "retention_by_kind": {"person": 1.0, "date": 0.0, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 698}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 12, "retention_rate": 0.18461538461538463, "retention_by_kind": {"person": 1.0, "date": 0.0, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 698}, "2": {"facts_total": 65, "facts_retained": 12, "retention_rate": 0.18461538461538463, "retention_by_kind": {"person": 1.0, "date": 0.0, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 698}, "3": {"facts_total": 65, "facts_retained": 12, "retention_rate": 0.18461538461538463, "retention_by_kind": {"person": 1.0, "date": 0.0, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 698}, "4": {"facts_total": 65, "facts_retained": 12, "retention_rate": 0.18461538461538463, "retention_by_kind": {"person": 1.0, "date": 0.0, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 698}, "5": {"facts_total": 65, "facts_retained": 12, "retention_rate": 0.18461538461538463, "retention_by_kind": {"person": 1.0, "date": 0.0, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 698}, "6": {"facts_total": 65, "facts_retained": 12, "retention_rate": 0.18461538461538463, "retention_by_kind": {"person": 1.0, "date": 0.0, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 698}, "7": {"facts_total": 65, "facts_retained": 12, "retention_rate": 0.18461538461538463, "retention_by_kind": {"person": 1.0, "date": 0.0, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 698}, "8": {"facts_total": 65, "facts_retained": 12, "retention_rate": 0.18461538461538463, "retention_by_kind": {"person": 1.0, "date": 0.0, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 698}}}, "reference_key_id": "keys/synthetic/syn_03009_2000.json", "task_scores": {"correct": 1.0, "abstained": 1.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "b7aa2d9d5f9f3a06115e4e65c8ffaf4be2a07d6d067ceac89e49c6efeb6fdd5f", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "f0029d29f2b337848063e72b", "attempt_id": "4b89408584e7", "run_group_id": "E4_core:A5|closed|qwen3b:h8", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:18:21.485252Z", "task_family": "synthetic", "document_id": "syn_03009_2000", "document_hash": "6ed94dcf728255bd7a7d11c97c362930728fb81a3e91f1906eb6eec86e0fedb7", "question_id": "q14_unanswerable", "split": "test", "cluster_id": "syn_03009_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 8, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1976, "query_tokens": 330, "message_tokens": 234, "message_utf8_bytes": 698, "cache_tensor_bytes": null, "serialized_payload_bytes": 698, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 86710.96616482828, "hops": {"1": 16386.116749956273, "2": 9978.41195797082, "3": 9960.819375002757, "4": 10282.75599994231, "5": 9938.042040914297, "6": 10059.49283298105, "7": 10103.797167073935, "8": 10001.530040986836}, "receiver_prefill_ms": 253.86775005608797, "ttft_ms": 37.092374986968935, "decode_ms": 28.079124982468784}, "total_wall_ms": 87030.39437276311, "ttft_ms": 37.092374986968935, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 12, "retention_rate": 0.18461538461538463, "retention_by_kind": {"person": 1.0, "date": 0.0, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 698}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 12, "retention_rate": 0.18461538461538463, "retention_by_kind": {"person": 1.0, "date": 0.0, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 698}, "2": {"facts_total": 65, "facts_retained": 12, "retention_rate": 0.18461538461538463, "retention_by_kind": {"person": 1.0, "date": 0.0, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 698}, "3": {"facts_total": 65, "facts_retained": 12, "retention_rate": 0.18461538461538463, "retention_by_kind": {"person": 1.0, "date": 0.0, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 698}, "4": {"facts_total": 65, "facts_retained": 12, "retention_rate": 0.18461538461538463, "retention_by_kind": {"person": 1.0, "date": 0.0, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 698}, "5": {"facts_total": 65, "facts_retained": 12, "retention_rate": 0.18461538461538463, "retention_by_kind": {"person": 1.0, "date": 0.0, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 698}, "6": {"facts_total": 65, "facts_retained": 12, "retention_rate": 0.18461538461538463, "retention_by_kind": {"person": 1.0, "date": 0.0, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 698}, "7": {"facts_total": 65, "facts_retained": 12, "retention_rate": 0.18461538461538463, "retention_by_kind": {"person": 1.0, "date": 0.0, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 698}, "8": {"facts_total": 65, "facts_retained": 12, "retention_rate": 0.18461538461538463, "retention_by_kind": {"person": 1.0, "date": 0.0, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 698}}}, "reference_key_id": "keys/synthetic/syn_03009_2000.json", "task_scores": {"correct": 1.0, "abstained": 1.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "b7aa2d9d5f9f3a06115e4e65c8ffaf4be2a07d6d067ceac89e49c6efeb6fdd5f", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "7f8f1065f38fe6dadb33f912", "attempt_id": "a994b8b587bd", "run_group_id": "E4_core:A4|repairable|qwen3b:h1", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:18:44.814541Z", "task_family": "synthetic", "document_id": "syn_03009_2000", "document_hash": "6ed94dcf728255bd7a7d11c97c362930728fb81a3e91f1906eb6eec86e0fedb7", "question_id": "q00_single_fact", "split": "test", "cluster_id": "syn_03009_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 1, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1976, "query_tokens": 595, "message_tokens": 512, "message_utf8_bytes": 1123, "cache_tensor_bytes": null, "serialized_payload_bytes": 1123, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 22990.705040981993, "hops": {"1": 22990.705040981993}, "receiver_prefill_ms": 415.2640829561278, "ttft_ms": 25.30291606672108, "decode_ms": 314.7803330793977}, "total_wall_ms": 23746.90379097592, "ttft_ms": 25.30291606672108, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 10}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "PA-37F-6227", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 26, "retention_rate": 0.4, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.75, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1123}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 26, "retention_rate": 0.4, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.75, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1123}}}, "reference_key_id": "keys/synthetic/syn_03009_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "3bd2611f7bc59d6f4b6323c3ac1e9f7af93bcc7ed3e43397b183cdb4d6c6e058", "output_hash": "5b7645cf6e8140f4c0bf6ed4ad76121cf2924fe7e666d92161812527b86f4409"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "2f741f8d5afbc41e55edf472", "attempt_id": "40e411d40482", "run_group_id": "E4_core:A4|repairable|qwen3b:h1", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:18:45.589194Z", "task_family": "synthetic", "document_id": "syn_03009_2000", "document_hash": "6ed94dcf728255bd7a7d11c97c362930728fb81a3e91f1906eb6eec86e0fedb7", "question_id": "q01_single_fact", "split": "test", "cluster_id": "syn_03009_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 1, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1976, "query_tokens": 600, "message_tokens": 512, "message_utf8_bytes": 1123, "cache_tensor_bytes": null, "serialized_payload_bytes": 1123, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 22990.705040981993, "hops": {"1": 22990.705040981993}, "receiver_prefill_ms": 352.9143329942599, "ttft_ms": 51.042333943769336, "decode_ms": 44.91295909974724}, "total_wall_ms": 23440.374498954043, "ttft_ms": 51.042333943769336, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 26, "retention_rate": 0.4, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.75, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1123}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 26, "retention_rate": 0.4, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.75, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1123}}}, "reference_key_id": "keys/synthetic/syn_03009_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "3bd2611f7bc59d6f4b6323c3ac1e9f7af93bcc7ed3e43397b183cdb4d6c6e058", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "1512ecd3226425ede497d62f", "attempt_id": "427922f2d0b1", "run_group_id": "E4_core:A4|repairable|qwen3b:h1", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:18:46.056539Z", "task_family": "synthetic", "document_id": "syn_03009_2000", "document_hash": "6ed94dcf728255bd7a7d11c97c362930728fb81a3e91f1906eb6eec86e0fedb7", "question_id": "q02_exact_copy", "split": "test", "cluster_id": "syn_03009_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 1, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1976, "query_tokens": 598, "message_tokens": 512, "message_utf8_bytes": 1123, "cache_tensor_bytes": null, "serialized_payload_bytes": 1123, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 22990.705040981993, "hops": {"1": 22990.705040981993}, "receiver_prefill_ms": 544.7408750187606, "ttft_ms": 25.312166078947484, "decode_ms": 40.07716698106378}, "total_wall_ms": 23601.63395793643, "ttft_ms": 25.312166078947484, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 26, "retention_rate": 0.4, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.75, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1123}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 26, "retention_rate": 0.4, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.75, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1123}}}, "reference_key_id": "keys/synthetic/syn_03009_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "3bd2611f7bc59d6f4b6323c3ac1e9f7af93bcc7ed3e43397b183cdb4d6c6e058", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "2316a2bdddbccde321bb657c", "attempt_id": "c939a1cf8b37", "run_group_id": "E4_core:A4|repairable|qwen3b:h1", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:18:46.688194Z", "task_family": "synthetic", "document_id": "syn_03009_2000", "document_hash": "6ed94dcf728255bd7a7d11c97c362930728fb81a3e91f1906eb6eec86e0fedb7", "question_id": "q03_exact_copy", "split": "test", "cluster_id": "syn_03009_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 1, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1976, "query_tokens": 594, "message_tokens": 512, "message_utf8_bytes": 1123, "cache_tensor_bytes": null, "serialized_payload_bytes": 1123, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 22990.705040981993, "hops": {"1": 22990.705040981993}, "receiver_prefill_ms": 406.16995794698596, "ttft_ms": 24.523083004169166, "decode_ms": 160.5069999350235}, "total_wall_ms": 23582.68349908758, "ttft_ms": 24.523083004169166, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 5}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "Idris Bergström", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 26, "retention_rate": 0.4, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.75, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1123}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 26, "retention_rate": 0.4, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.75, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1123}}}, "reference_key_id": "keys/synthetic/syn_03009_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "3bd2611f7bc59d6f4b6323c3ac1e9f7af93bcc7ed3e43397b183cdb4d6c6e058", "output_hash": "fd88832c9fd0ab175818d76d875d70156225bd4b34ec401514546ef70f4f826c"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "0fe89237ca9b6f14a73bfb24", "attempt_id": "681ac4c9666b", "run_group_id": "E4_core:A4|repairable|qwen3b:h1", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:18:47.298222Z", "task_family": "synthetic", "document_id": "syn_03009_2000", "document_hash": "6ed94dcf728255bd7a7d11c97c362930728fb81a3e91f1906eb6eec86e0fedb7", "question_id": "q04_multi_hop", "split": "test", "cluster_id": "syn_03009_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 1, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1976, "query_tokens": 602, "message_tokens": 512, "message_utf8_bytes": 1123, "cache_tensor_bytes": null, "serialized_payload_bytes": 1123, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 22990.705040981993, "hops": {"1": 22990.705040981993}, "receiver_prefill_ms": 360.24449998512864, "ttft_ms": 40.88291700463742, "decode_ms": 468.66608294658363}, "total_wall_ms": 23861.225040978752, "ttft_ms": 40.88291700463742, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 7}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "4670383", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 26, "retention_rate": 0.4, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.75, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1123}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 26, "retention_rate": 0.4, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.75, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1123}}}, "reference_key_id": "keys/synthetic/syn_03009_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "3bd2611f7bc59d6f4b6323c3ac1e9f7af93bcc7ed3e43397b183cdb4d6c6e058", "output_hash": "19ae8c357e5500bf494a6163bfbd11380b2996e8ea532a2c64f4a599464820f7"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "7ee77cb3e03c0a98c551067a", "attempt_id": "3e037317c401", "run_group_id": "E4_core:A4|repairable|qwen3b:h1", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:18:48.186819Z", "task_family": "synthetic", "document_id": "syn_03009_2000", "document_hash": "6ed94dcf728255bd7a7d11c97c362930728fb81a3e91f1906eb6eec86e0fedb7", "question_id": "q05_multi_hop", "split": "test", "cluster_id": "syn_03009_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 1, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1976, "query_tokens": 605, "message_tokens": 512, "message_utf8_bytes": 1123, "cache_tensor_bytes": null, "serialized_payload_bytes": 1123, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 22990.705040981993, "hops": {"1": 22990.705040981993}, "receiver_prefill_ms": 411.5750420605764, "ttft_ms": 41.67224990669638, "decode_ms": 21.87383291311562}, "total_wall_ms": 23466.5921239648, "ttft_ms": 41.67224990669638, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 26, "retention_rate": 0.4, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.75, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1123}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 26, "retention_rate": 0.4, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.75, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1123}}}, "reference_key_id": "keys/synthetic/syn_03009_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "3bd2611f7bc59d6f4b6323c3ac1e9f7af93bcc7ed3e43397b183cdb4d6c6e058", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "b9c8aa1984ccbe2a81f28d9d", "attempt_id": "bbda03a687b4", "run_group_id": "E4_core:A4|repairable|qwen3b:h1", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:18:48.680223Z", "task_family": "synthetic", "document_id": "syn_03009_2000", "document_hash": "6ed94dcf728255bd7a7d11c97c362930728fb81a3e91f1906eb6eec86e0fedb7", "question_id": "q06_negation", "split": "test", "cluster_id": "syn_03009_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 1, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1976, "query_tokens": 595, "message_tokens": 512, "message_utf8_bytes": 1123, "cache_tensor_bytes": null, "serialized_payload_bytes": 1123, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 22990.705040981993, "hops": {"1": 22990.705040981993}, "receiver_prefill_ms": 394.6552500128746, "ttft_ms": 23.737582960166037, "decode_ms": 19.249541917815804}, "total_wall_ms": 23429.054416017607, "ttft_ms": 23.737582960166037, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 26, "retention_rate": 0.4, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.75, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1123}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 26, "retention_rate": 0.4, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.75, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1123}}}, "reference_key_id": "keys/synthetic/syn_03009_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "3bd2611f7bc59d6f4b6323c3ac1e9f7af93bcc7ed3e43397b183cdb4d6c6e058", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "5390e7dadaa416829c351eb3", "attempt_id": "d2e25cb301f9", "run_group_id": "E4_core:A4|repairable|qwen3b:h1", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:18:49.136741Z", "task_family": "synthetic", "document_id": "syn_03009_2000", "document_hash": "6ed94dcf728255bd7a7d11c97c362930728fb81a3e91f1906eb6eec86e0fedb7", "question_id": "q07_negation", "split": "test", "cluster_id": "syn_03009_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 1, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1976, "query_tokens": 594, "message_tokens": 512, "message_utf8_bytes": 1123, "cache_tensor_bytes": null, "serialized_payload_bytes": 1123, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 22990.705040981993, "hops": {"1": 22990.705040981993}, "receiver_prefill_ms": 380.205333000049, "ttft_ms": 56.18637497536838, "decode_ms": 57.43108398746699}, "total_wall_ms": 23485.335457953624, "ttft_ms": 56.18637497536838, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 26, "retention_rate": 0.4, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.75, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1123}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 26, "retention_rate": 0.4, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.75, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1123}}}, "reference_key_id": "keys/synthetic/syn_03009_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "3bd2611f7bc59d6f4b6323c3ac1e9f7af93bcc7ed3e43397b183cdb4d6c6e058", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "e722e1a4db6c950122f0deb2", "attempt_id": "a1dd3172e75f", "run_group_id": "E4_core:A4|repairable|qwen3b:h1", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:18:49.651007Z", "task_family": "synthetic", "document_id": "syn_03009_2000", "document_hash": "6ed94dcf728255bd7a7d11c97c362930728fb81a3e91f1906eb6eec86e0fedb7", "question_id": "q08_revision", "split": "test", "cluster_id": "syn_03009_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 1, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1976, "query_tokens": 612, "message_tokens": 512, "message_utf8_bytes": 1123, "cache_tensor_bytes": null, "serialized_payload_bytes": 1123, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 22990.705040981993, "hops": {"1": 22990.705040981993}, "receiver_prefill_ms": 514.8997089127079, "ttft_ms": 34.10962503403425, "decode_ms": 294.12641702219844}, "total_wall_ms": 23834.79145693127, "ttft_ms": 34.10962503403425, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 8}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "10047105", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 26, "retention_rate": 0.4, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.75, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1123}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 26, "retention_rate": 0.4, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.75, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1123}}}, "reference_key_id": "keys/synthetic/syn_03009_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "3bd2611f7bc59d6f4b6323c3ac1e9f7af93bcc7ed3e43397b183cdb4d6c6e058", "output_hash": "bea99610d1e3e75a7f20939faf8e6824e02237b156dacd6442ba2a7e02617510"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "ee80f310d106932f6312a03b", "attempt_id": "f6e603fc48b1", "run_group_id": "E4_core:A4|repairable|qwen3b:h1", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:18:50.513070Z", "task_family": "synthetic", "document_id": "syn_03009_2000", "document_hash": "6ed94dcf728255bd7a7d11c97c362930728fb81a3e91f1906eb6eec86e0fedb7", "question_id": "q09_revision", "split": "test", "cluster_id": "syn_03009_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 1, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1976, "query_tokens": 605, "message_tokens": 512, "message_utf8_bytes": 1123, "cache_tensor_bytes": null, "serialized_payload_bytes": 1123, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 22990.705040981993, "hops": {"1": 22990.705040981993}, "receiver_prefill_ms": 371.58179201651365, "ttft_ms": 41.20295797474682, "decode_ms": 548.4827079344541}, "total_wall_ms": 23952.749499934725, "ttft_ms": 41.20295797474682, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 8}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "10047105", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 26, "retention_rate": 0.4, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.75, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1123}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 26, "retention_rate": 0.4, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.75, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1123}}}, "reference_key_id": "keys/synthetic/syn_03009_2000.json", "task_scores": {"correct": 0.0, "abstained": 0.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "3bd2611f7bc59d6f4b6323c3ac1e9f7af93bcc7ed3e43397b183cdb4d6c6e058", "output_hash": "bea99610d1e3e75a7f20939faf8e6824e02237b156dacd6442ba2a7e02617510"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "e212ee3966eba9f80159ec71", "attempt_id": "0969cc71cdb3", "run_group_id": "E4_core:A4|repairable|qwen3b:h1", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:18:51.492837Z", "task_family": "synthetic", "document_id": "syn_03009_2000", "document_hash": "6ed94dcf728255bd7a7d11c97c362930728fb81a3e91f1906eb6eec86e0fedb7", "question_id": "q10_temporal", "split": "test", "cluster_id": "syn_03009_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 1, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1976, "query_tokens": 614, "message_tokens": 512, "message_utf8_bytes": 1123, "cache_tensor_bytes": null, "serialized_payload_bytes": 1123, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 22990.705040981993, "hops": {"1": 22990.705040981993}, "receiver_prefill_ms": 414.8250420112163, "ttft_ms": 40.14595795888454, "decode_ms": 23.60629104077816}, "total_wall_ms": 23469.955374021083, "ttft_ms": 40.14595795888454, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 26, "retention_rate": 0.4, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.75, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1123}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 26, "retention_rate": 0.4, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.75, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1123}}}, "reference_key_id": "keys/synthetic/syn_03009_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "3bd2611f7bc59d6f4b6323c3ac1e9f7af93bcc7ed3e43397b183cdb4d6c6e058", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "f971234ab26712f85ef4f59d", "attempt_id": "f5dc6f80bc55", "run_group_id": "E4_core:A4|repairable|qwen3b:h1", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:18:51.989476Z", "task_family": "synthetic", "document_id": "syn_03009_2000", "document_hash": "6ed94dcf728255bd7a7d11c97c362930728fb81a3e91f1906eb6eec86e0fedb7", "question_id": "q11_aggregation", "split": "test", "cluster_id": "syn_03009_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 1, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1976, "query_tokens": 605, "message_tokens": 512, "message_utf8_bytes": 1123, "cache_tensor_bytes": null, "serialized_payload_bytes": 1123, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 22990.705040981993, "hops": {"1": 22990.705040981993}, "receiver_prefill_ms": 407.6667500194162, "ttft_ms": 21.076667006127536, "decode_ms": 352.541041909717}, "total_wall_ms": 23772.698207991198, "ttft_ms": 21.076667006127536, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 7}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "5781336", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 26, "retention_rate": 0.4, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.75, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1123}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 26, "retention_rate": 0.4, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.75, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1123}}}, "reference_key_id": "keys/synthetic/syn_03009_2000.json", "task_scores": {"correct": 0.0, "abstained": 0.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "3bd2611f7bc59d6f4b6323c3ac1e9f7af93bcc7ed3e43397b183cdb4d6c6e058", "output_hash": "a365e04f74ba5baeccdd9e30dc27a1c5ff916597b828d874e63884b2090a60e2"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "eaba3c2acaef56c6a1710c55", "attempt_id": "9469950dbe6b", "run_group_id": "E4_core:A4|repairable|qwen3b:h1", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:18:52.788176Z", "task_family": "synthetic", "document_id": "syn_03009_2000", "document_hash": "6ed94dcf728255bd7a7d11c97c362930728fb81a3e91f1906eb6eec86e0fedb7", "question_id": "q12_unit_conversion", "split": "test", "cluster_id": "syn_03009_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 1, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1976, "query_tokens": 603, "message_tokens": 512, "message_utf8_bytes": 1123, "cache_tensor_bytes": null, "serialized_payload_bytes": 1123, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 22990.705040981993, "hops": {"1": 22990.705040981993}, "receiver_prefill_ms": 520.6647919258103, "ttft_ms": 38.55062497314066, "decode_ms": 24.954083026386797}, "total_wall_ms": 23575.650373939425, "ttft_ms": 38.55062497314066, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 26, "retention_rate": 0.4, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.75, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1123}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 26, "retention_rate": 0.4, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.75, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1123}}}, "reference_key_id": "keys/synthetic/syn_03009_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "3bd2611f7bc59d6f4b6323c3ac1e9f7af93bcc7ed3e43397b183cdb4d6c6e058", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "c040fb59741a39fec462f1a3", "attempt_id": "c9343350af22", "run_group_id": "E4_core:A4|repairable|qwen3b:h1", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:18:53.392072Z", "task_family": "synthetic", "document_id": "syn_03009_2000", "document_hash": "6ed94dcf728255bd7a7d11c97c362930728fb81a3e91f1906eb6eec86e0fedb7", "question_id": "q13_unanswerable", "split": "test", "cluster_id": "syn_03009_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 1, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1976, "query_tokens": 607, "message_tokens": 512, "message_utf8_bytes": 1123, "cache_tensor_bytes": null, "serialized_payload_bytes": 1123, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 22990.705040981993, "hops": {"1": 22990.705040981993}, "receiver_prefill_ms": 403.50750007200986, "ttft_ms": 43.79558307118714, "decode_ms": 29.91862501949072}, "total_wall_ms": 23468.56916602701, "ttft_ms": 43.79558307118714, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 26, "retention_rate": 0.4, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.75, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1123}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 26, "retention_rate": 0.4, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.75, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1123}}}, "reference_key_id": "keys/synthetic/syn_03009_2000.json", "task_scores": {"correct": 1.0, "abstained": 1.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "3bd2611f7bc59d6f4b6323c3ac1e9f7af93bcc7ed3e43397b183cdb4d6c6e058", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "31a3827ab83d0dd66cf32980", "attempt_id": "9262d60d5b0c", "run_group_id": "E4_core:A4|repairable|qwen3b:h1", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:18:53.888107Z", "task_family": "synthetic", "document_id": "syn_03009_2000", "document_hash": "6ed94dcf728255bd7a7d11c97c362930728fb81a3e91f1906eb6eec86e0fedb7", "question_id": "q14_unanswerable", "split": "test", "cluster_id": "syn_03009_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 1, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1976, "query_tokens": 607, "message_tokens": 512, "message_utf8_bytes": 1123, "cache_tensor_bytes": null, "serialized_payload_bytes": 1123, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 22990.705040981993, "hops": {"1": 22990.705040981993}, "receiver_prefill_ms": 384.8057499853894, "ttft_ms": 55.02012500073761, "decode_ms": 57.10641702171415}, "total_wall_ms": 23488.29637398012, "ttft_ms": 55.02012500073761, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 26, "retention_rate": 0.4, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.75, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1123}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 26, "retention_rate": 0.4, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.75, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1123}}}, "reference_key_id": "keys/synthetic/syn_03009_2000.json", "task_scores": {"correct": 1.0, "abstained": 1.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "3bd2611f7bc59d6f4b6323c3ac1e9f7af93bcc7ed3e43397b183cdb4d6c6e058", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "4a41ed65b17b13e5b7584c01", "attempt_id": "76ce10c28173", "run_group_id": "E4_core:A4|repairable|qwen3b:h2", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:19:16.690992Z", "task_family": "synthetic", "document_id": "syn_03009_2000", "document_hash": "6ed94dcf728255bd7a7d11c97c362930728fb81a3e91f1906eb6eec86e0fedb7", "question_id": "q00_single_fact", "split": "test", "cluster_id": "syn_03009_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 2, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1976, "query_tokens": 595, "message_tokens": 512, "message_utf8_bytes": 1123, "cache_tensor_bytes": null, "serialized_payload_bytes": 1123, "side_information_bytes": 0, "fetched_bytes": 733, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 45278.01291598007, "hops": {"1": 22990.705040981993, "2": 22287.307874998078}, "receiver_prefill_ms": 408.63191708922386, "ttft_ms": 32.670416054315865, "decode_ms": 311.92495801951736}, "total_wall_ms": 46031.99091600254, "ttft_ms": 32.670416054315865, "input_usage": {"lookups": 4}, "output_usage": {"output_tokens": 10}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "PA-37F-6227", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 26, "retention_rate": 0.4, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.75, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1123}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 26, "retention_rate": 0.4, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.75, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1123}, "2": {"facts_total": 65, "facts_retained": 26, "retention_rate": 0.4, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.75, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1123}}}, "reference_key_id": "keys/synthetic/syn_03009_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "3bd2611f7bc59d6f4b6323c3ac1e9f7af93bcc7ed3e43397b183cdb4d6c6e058", "output_hash": "5b7645cf6e8140f4c0bf6ed4ad76121cf2924fe7e666d92161812527b86f4409"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "438d704d5a647ac4a5b073d5", "attempt_id": "242ad6016f3d", "run_group_id": "E4_core:A4|repairable|qwen3b:h2", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:19:17.462454Z", "task_family": "synthetic", "document_id": "syn_03009_2000", "document_hash": "6ed94dcf728255bd7a7d11c97c362930728fb81a3e91f1906eb6eec86e0fedb7", "question_id": "q01_single_fact", "split": "test", "cluster_id": "syn_03009_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 2, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1976, "query_tokens": 600, "message_tokens": 512, "message_utf8_bytes": 1123, "cache_tensor_bytes": null, "serialized_payload_bytes": 1123, "side_information_bytes": 0, "fetched_bytes": 733, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 45278.01291598007, "hops": {"1": 22990.705040981993, "2": 22287.307874998078}, "receiver_prefill_ms": 421.6401249868795, "ttft_ms": 55.77524995896965, "decode_ms": 119.97020803391933}, "total_wall_ms": 45876.15545792505, "ttft_ms": 55.77524995896965, "input_usage": {"lookups": 4}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 26, "retention_rate": 0.4, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.75, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1123}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 26, "retention_rate": 0.4, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.75, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1123}, "2": {"facts_total": 65, "facts_retained": 26, "retention_rate": 0.4, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.75, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1123}}}, "reference_key_id": "keys/synthetic/syn_03009_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "3bd2611f7bc59d6f4b6323c3ac1e9f7af93bcc7ed3e43397b183cdb4d6c6e058", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "130fdb5aae9977412beaa702", "attempt_id": "66cbce429912", "run_group_id": "E4_core:A4|repairable|qwen3b:h2", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:19:18.080318Z", "task_family": "synthetic", "document_id": "syn_03009_2000", "document_hash": "6ed94dcf728255bd7a7d11c97c362930728fb81a3e91f1906eb6eec86e0fedb7", "question_id": "q02_exact_copy", "split": "test", "cluster_id": "syn_03009_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 2, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1976, "query_tokens": 598, "message_tokens": 512, "message_utf8_bytes": 1123, "cache_tensor_bytes": null, "serialized_payload_bytes": 1123, "side_information_bytes": 0, "fetched_bytes": 733, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 45278.01291598007, "hops": {"1": 22990.705040981993, "2": 22287.307874998078}, "receiver_prefill_ms": 444.66879090759903, "ttft_ms": 39.52520899474621, "decode_ms": 23.92512501683086}, "total_wall_ms": 45786.96541604586, "ttft_ms": 39.52520899474621, "input_usage": {"lookups": 4}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 26, "retention_rate": 0.4, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.75, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1123}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 26, "retention_rate": 0.4, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.75, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1123}, "2": {"facts_total": 65, "facts_retained": 26, "retention_rate": 0.4, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.75, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1123}}}, "reference_key_id": "keys/synthetic/syn_03009_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "3bd2611f7bc59d6f4b6323c3ac1e9f7af93bcc7ed3e43397b183cdb4d6c6e058", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "6ed432c64336ac1cc4d36c3d", "attempt_id": "96c3568362a4", "run_group_id": "E4_core:A4|repairable|qwen3b:h2", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:19:18.607023Z", "task_family": "synthetic", "document_id": "syn_03009_2000", "document_hash": "6ed94dcf728255bd7a7d11c97c362930728fb81a3e91f1906eb6eec86e0fedb7", "question_id": "q03_exact_copy", "split": "test", "cluster_id": "syn_03009_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 2, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1976, "query_tokens": 594, "message_tokens": 512, "message_utf8_bytes": 1123, "cache_tensor_bytes": null, "serialized_payload_bytes": 1123, "side_information_bytes": 0, "fetched_bytes": 733, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 45278.01291598007, "hops": {"1": 22990.705040981993, "2": 22287.307874998078}, "receiver_prefill_ms": 400.6817499175668, "ttft_ms": 35.96266696695238, "decode_ms": 130.45616704039276}, "total_wall_ms": 45845.875249942765, "ttft_ms": 35.96266696695238, "input_usage": {"lookups": 4}, "output_usage": {"output_tokens": 5}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "Idris Bergström", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 26, "retention_rate": 0.4, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.75, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1123}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 26, "retention_rate": 0.4, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.75, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1123}, "2": {"facts_total": 65, "facts_retained": 26, "retention_rate": 0.4, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.75, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1123}}}, "reference_key_id": "keys/synthetic/syn_03009_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "3bd2611f7bc59d6f4b6323c3ac1e9f7af93bcc7ed3e43397b183cdb4d6c6e058", "output_hash": "fd88832c9fd0ab175818d76d875d70156225bd4b34ec401514546ef70f4f826c"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "c570d51450f5416297409c08", "attempt_id": "225312429dab", "run_group_id": "E4_core:A4|repairable|qwen3b:h2", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:19:19.192447Z", "task_family": "synthetic", "document_id": "syn_03009_2000", "document_hash": "6ed94dcf728255bd7a7d11c97c362930728fb81a3e91f1906eb6eec86e0fedb7", "question_id": "q04_multi_hop", "split": "test", "cluster_id": "syn_03009_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 2, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1976, "query_tokens": 602, "message_tokens": 512, "message_utf8_bytes": 1123, "cache_tensor_bytes": null, "serialized_payload_bytes": 1123, "side_information_bytes": 0, "fetched_bytes": 733, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 45278.01291598007, "hops": {"1": 22990.705040981993, "2": 22287.307874998078}, "receiver_prefill_ms": 434.8780829459429, "ttft_ms": 43.80629095248878, "decode_ms": 400.3288339590654}, "total_wall_ms": 46157.81420702115, "ttft_ms": 43.80629095248878, "input_usage": {"lookups": 4}, "output_usage": {"output_tokens": 7}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "4670383", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 26, "retention_rate": 0.4, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.75, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1123}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 26, "retention_rate": 0.4, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.75, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1123}, "2": {"facts_total": 65, "facts_retained": 26, "retention_rate": 0.4, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.75, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1123}}}, "reference_key_id": "keys/synthetic/syn_03009_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "3bd2611f7bc59d6f4b6323c3ac1e9f7af93bcc7ed3e43397b183cdb4d6c6e058", "output_hash": "19ae8c357e5500bf494a6163bfbd11380b2996e8ea532a2c64f4a599464820f7"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "cd2c033d147d831df19a5e81", "attempt_id": "99b06627b850", "run_group_id": "E4_core:A4|repairable|qwen3b:h2", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:19:20.092923Z", "task_family": "synthetic", "document_id": "syn_03009_2000", "document_hash": "6ed94dcf728255bd7a7d11c97c362930728fb81a3e91f1906eb6eec86e0fedb7", "question_id": "q05_multi_hop", "split": "test", "cluster_id": "syn_03009_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 2, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1976, "query_tokens": 605, "message_tokens": 512, "message_utf8_bytes": 1123, "cache_tensor_bytes": null, "serialized_payload_bytes": 1123, "side_information_bytes": 0, "fetched_bytes": 733, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 45278.01291598007, "hops": {"1": 22990.705040981993, "2": 22287.307874998078}, "receiver_prefill_ms": 393.6313330195844, "ttft_ms": 22.222624975256622, "decode_ms": 37.9425419960171}, "total_wall_ms": 45732.44399903342, "ttft_ms": 22.222624975256622, "input_usage": {"lookups": 4}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 26, "retention_rate": 0.4, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.75, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1123}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 26, "retention_rate": 0.4, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.75, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1123}, "2": {"facts_total": 65, "facts_retained": 26, "retention_rate": 0.4, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.75, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1123}}}, "reference_key_id": "keys/synthetic/syn_03009_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "3bd2611f7bc59d6f4b6323c3ac1e9f7af93bcc7ed3e43397b183cdb4d6c6e058", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "36c19982a9a1e917c69e1adc", "attempt_id": "7607205b8fbf", "run_group_id": "E4_core:A4|repairable|qwen3b:h2", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:19:20.564753Z", "task_family": "synthetic", "document_id": "syn_03009_2000", "document_hash": "6ed94dcf728255bd7a7d11c97c362930728fb81a3e91f1906eb6eec86e0fedb7", "question_id": "q06_negation", "split": "test", "cluster_id": "syn_03009_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 2, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1976, "query_tokens": 595, "message_tokens": 512, "message_utf8_bytes": 1123, "cache_tensor_bytes": null, "serialized_payload_bytes": 1123, "side_information_bytes": 0, "fetched_bytes": 733, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 45278.01291598007, "hops": {"1": 22990.705040981993, "2": 22287.307874998078}, "receiver_prefill_ms": 373.9637089893222, "ttft_ms": 46.26500001177192, "decode_ms": 55.344333057291806}, "total_wall_ms": 45754.48179105297, "ttft_ms": 46.26500001177192, "input_usage": {"lookups": 4}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 26, "retention_rate": 0.4, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.75, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1123}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 26, "retention_rate": 0.4, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.75, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1123}, "2": {"facts_total": 65, "facts_retained": 26, "retention_rate": 0.4, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.75, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1123}}}, "reference_key_id": "keys/synthetic/syn_03009_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "3bd2611f7bc59d6f4b6323c3ac1e9f7af93bcc7ed3e43397b183cdb4d6c6e058", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "210b024472d4115ac44c9a0c", "attempt_id": "afd39a69f2db", "run_group_id": "E4_core:A4|repairable|qwen3b:h2", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:19:21.058753Z", "task_family": "synthetic", "document_id": "syn_03009_2000", "document_hash": "6ed94dcf728255bd7a7d11c97c362930728fb81a3e91f1906eb6eec86e0fedb7", "question_id": "q07_negation", "split": "test", "cluster_id": "syn_03009_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 2, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1976, "query_tokens": 594, "message_tokens": 512, "message_utf8_bytes": 1123, "cache_tensor_bytes": null, "serialized_payload_bytes": 1123, "side_information_bytes": 0, "fetched_bytes": 733, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 45278.01291598007, "hops": {"1": 22990.705040981993, "2": 22287.307874998078}, "receiver_prefill_ms": 495.58070802595466, "ttft_ms": 82.4872920056805, "decode_ms": 31.727458001114428}, "total_wall_ms": 45888.509373995475, "ttft_ms": 82.4872920056805, "input_usage": {"lookups": 4}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 26, "retention_rate": 0.4, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.75, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1123}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 26, "retention_rate": 0.4, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.75, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1123}, "2": {"facts_total": 65, "facts_retained": 26, "retention_rate": 0.4, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.75, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1123}}}, "reference_key_id": "keys/synthetic/syn_03009_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "3bd2611f7bc59d6f4b6323c3ac1e9f7af93bcc7ed3e43397b183cdb4d6c6e058", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "3a9527a562debd0e8b6700b0", "attempt_id": "5c1059cbbd90", "run_group_id": "E4_core:A4|repairable|qwen3b:h2", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:19:21.686108Z", "task_family": "synthetic", "document_id": "syn_03009_2000", "document_hash": "6ed94dcf728255bd7a7d11c97c362930728fb81a3e91f1906eb6eec86e0fedb7", "question_id": "q08_revision", "split": "test", "cluster_id": "syn_03009_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 2, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1976, "query_tokens": 612, "message_tokens": 512, "message_utf8_bytes": 1123, "cache_tensor_bytes": null, "serialized_payload_bytes": 1123, "side_information_bytes": 0, "fetched_bytes": 733, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 45278.01291598007, "hops": {"1": 22990.705040981993, "2": 22287.307874998078}, "receiver_prefill_ms": 402.1789579419419, "ttft_ms": 34.26224994473159, "decode_ms": 260.502792079933}, "total_wall_ms": 45975.712832878344, "ttft_ms": 34.26224994473159, "input_usage": {"lookups": 4}, "output_usage": {"output_tokens": 8}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "10047105", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 26, "retention_rate": 0.4, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.75, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1123}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 26, "retention_rate": 0.4, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.75, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1123}, "2": {"facts_total": 65, "facts_retained": 26, "retention_rate": 0.4, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.75, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1123}}}, "reference_key_id": "keys/synthetic/syn_03009_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "3bd2611f7bc59d6f4b6323c3ac1e9f7af93bcc7ed3e43397b183cdb4d6c6e058", "output_hash": "bea99610d1e3e75a7f20939faf8e6824e02237b156dacd6442ba2a7e02617510"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "3246436601f3198c92e6d173", "attempt_id": "bbbaf3e8923f", "run_group_id": "E4_core:A4|repairable|qwen3b:h2", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:19:22.402301Z", "task_family": "synthetic", "document_id": "syn_03009_2000", "document_hash": "6ed94dcf728255bd7a7d11c97c362930728fb81a3e91f1906eb6eec86e0fedb7", "question_id": "q09_revision", "split": "test", "cluster_id": "syn_03009_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 2, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1976, "query_tokens": 605, "message_tokens": 512, "message_utf8_bytes": 1123, "cache_tensor_bytes": null, "serialized_payload_bytes": 1123, "side_information_bytes": 0, "fetched_bytes": 733, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 45278.01291598007, "hops": {"1": 22990.705040981993, "2": 22287.307874998078}, "receiver_prefill_ms": 354.68508303165436, "ttft_ms": 45.09204102214426, "decode_ms": 525.0479999231175}, "total_wall_ms": 46203.63620808348, "ttft_ms": 45.09204102214426, "input_usage": {"lookups": 4}, "output_usage": {"output_tokens": 8}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "10047105", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 26, "retention_rate": 0.4, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.75, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1123}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 26, "retention_rate": 0.4, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.75, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1123}, "2": {"facts_total": 65, "facts_retained": 26, "retention_rate": 0.4, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.75, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1123}}}, "reference_key_id": "keys/synthetic/syn_03009_2000.json", "task_scores": {"correct": 0.0, "abstained": 0.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "3bd2611f7bc59d6f4b6323c3ac1e9f7af93bcc7ed3e43397b183cdb4d6c6e058", "output_hash": "bea99610d1e3e75a7f20939faf8e6824e02237b156dacd6442ba2a7e02617510"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "a6ee27cdbaa2d6d1f7eec80c", "attempt_id": "5ca0621cfc4e", "run_group_id": "E4_core:A4|repairable|qwen3b:h2", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:19:23.346251Z", "task_family": "synthetic", "document_id": "syn_03009_2000", "document_hash": "6ed94dcf728255bd7a7d11c97c362930728fb81a3e91f1906eb6eec86e0fedb7", "question_id": "q10_temporal", "split": "test", "cluster_id": "syn_03009_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 2, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1976, "query_tokens": 614, "message_tokens": 512, "message_utf8_bytes": 1123, "cache_tensor_bytes": null, "serialized_payload_bytes": 1123, "side_information_bytes": 0, "fetched_bytes": 733, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 45278.01291598007, "hops": {"1": 22990.705040981993, "2": 22287.307874998078}, "receiver_prefill_ms": 400.8564999094233, "ttft_ms": 34.50199996586889, "decode_ms": 29.470541048794985}, "total_wall_ms": 45743.50424890872, "ttft_ms": 34.50199996586889, "input_usage": {"lookups": 4}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 26, "retention_rate": 0.4, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.75, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1123}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 26, "retention_rate": 0.4, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.75, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1123}, "2": {"facts_total": 65, "facts_retained": 26, "retention_rate": 0.4, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.75, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1123}}}, "reference_key_id": "keys/synthetic/syn_03009_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "3bd2611f7bc59d6f4b6323c3ac1e9f7af93bcc7ed3e43397b183cdb4d6c6e058", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "ea5597c0f7ac381324949e47", "attempt_id": "1f9a97fc265d", "run_group_id": "E4_core:A4|repairable|qwen3b:h2", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:19:23.829859Z", "task_family": "synthetic", "document_id": "syn_03009_2000", "document_hash": "6ed94dcf728255bd7a7d11c97c362930728fb81a3e91f1906eb6eec86e0fedb7", "question_id": "q11_aggregation", "split": "test", "cluster_id": "syn_03009_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 2, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1976, "query_tokens": 605, "message_tokens": 512, "message_utf8_bytes": 1123, "cache_tensor_bytes": null, "serialized_payload_bytes": 1123, "side_information_bytes": 0, "fetched_bytes": 733, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 45278.01291598007, "hops": {"1": 22990.705040981993, "2": 22287.307874998078}, "receiver_prefill_ms": 379.8192080575973, "ttft_ms": 19.58412502426654, "decode_ms": 352.5522090494633}, "total_wall_ms": 46031.01575002074, "ttft_ms": 19.58412502426654, "input_usage": {"lookups": 4}, "output_usage": {"output_tokens": 7}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "5781336", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 26, "retention_rate": 0.4, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.75, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1123}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 26, "retention_rate": 0.4, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.75, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1123}, "2": {"facts_total": 65, "facts_retained": 26, "retention_rate": 0.4, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.75, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1123}}}, "reference_key_id": "keys/synthetic/syn_03009_2000.json", "task_scores": {"correct": 0.0, "abstained": 0.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "3bd2611f7bc59d6f4b6323c3ac1e9f7af93bcc7ed3e43397b183cdb4d6c6e058", "output_hash": "a365e04f74ba5baeccdd9e30dc27a1c5ff916597b828d874e63884b2090a60e2"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "0c54f141801d381d44ba35b8", "attempt_id": "e66995b76b32", "run_group_id": "E4_core:A4|repairable|qwen3b:h2", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:19:24.600541Z", "task_family": "synthetic", "document_id": "syn_03009_2000", "document_hash": "6ed94dcf728255bd7a7d11c97c362930728fb81a3e91f1906eb6eec86e0fedb7", "question_id": "q12_unit_conversion", "split": "test", "cluster_id": "syn_03009_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 2, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1976, "query_tokens": 603, "message_tokens": 512, "message_utf8_bytes": 1123, "cache_tensor_bytes": null, "serialized_payload_bytes": 1123, "side_information_bytes": 0, "fetched_bytes": 733, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 45278.01291598007, "hops": {"1": 22990.705040981993, "2": 22287.307874998078}, "receiver_prefill_ms": 515.9553328994662, "ttft_ms": 43.14758302643895, "decode_ms": 45.26479099877179}, "total_wall_ms": 45883.20479099639, "ttft_ms": 43.14758302643895, "input_usage": {"lookups": 4}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 26, "retention_rate": 0.4, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.75, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1123}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 26, "retention_rate": 0.4, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.75, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1123}, "2": {"facts_total": 65, "facts_retained": 26, "retention_rate": 0.4, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.75, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1123}}}, "reference_key_id": "keys/synthetic/syn_03009_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "3bd2611f7bc59d6f4b6323c3ac1e9f7af93bcc7ed3e43397b183cdb4d6c6e058", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "af7a547291d3be92dba8b2d6", "attempt_id": "4c6f228770f5", "run_group_id": "E4_core:A4|repairable|qwen3b:h2", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:19:25.223504Z", "task_family": "synthetic", "document_id": "syn_03009_2000", "document_hash": "6ed94dcf728255bd7a7d11c97c362930728fb81a3e91f1906eb6eec86e0fedb7", "question_id": "q13_unanswerable", "split": "test", "cluster_id": "syn_03009_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 2, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1976, "query_tokens": 607, "message_tokens": 512, "message_utf8_bytes": 1123, "cache_tensor_bytes": null, "serialized_payload_bytes": 1123, "side_information_bytes": 0, "fetched_bytes": 733, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 45278.01291598007, "hops": {"1": 22990.705040981993, "2": 22287.307874998078}, "receiver_prefill_ms": 383.0150000285357, "ttft_ms": 40.56741704698652, "decode_ms": 21.532875020056963}, "total_wall_ms": 45723.90883299522, "ttft_ms": 40.56741704698652, "input_usage": {"lookups": 4}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 26, "retention_rate": 0.4, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.75, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1123}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 26, "retention_rate": 0.4, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.75, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1123}, "2": {"facts_total": 65, "facts_retained": 26, "retention_rate": 0.4, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.75, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1123}}}, "reference_key_id": "keys/synthetic/syn_03009_2000.json", "task_scores": {"correct": 1.0, "abstained": 1.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "3bd2611f7bc59d6f4b6323c3ac1e9f7af93bcc7ed3e43397b183cdb4d6c6e058", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "f170a198e9bd40c5054da5d8", "attempt_id": "389224c8bf31", "run_group_id": "E4_core:A4|repairable|qwen3b:h2", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:19:25.689016Z", "task_family": "synthetic", "document_id": "syn_03009_2000", "document_hash": "6ed94dcf728255bd7a7d11c97c362930728fb81a3e91f1906eb6eec86e0fedb7", "question_id": "q14_unanswerable", "split": "test", "cluster_id": "syn_03009_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 2, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1976, "query_tokens": 607, "message_tokens": 512, "message_utf8_bytes": 1123, "cache_tensor_bytes": null, "serialized_payload_bytes": 1123, "side_information_bytes": 0, "fetched_bytes": 733, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 45278.01291598007, "hops": {"1": 22990.705040981993, "2": 22287.307874998078}, "receiver_prefill_ms": 366.26366700511426, "ttft_ms": 39.07208296004683, "decode_ms": 50.67558400332928}, "total_wall_ms": 45734.84870698303, "ttft_ms": 39.07208296004683, "input_usage": {"lookups": 4}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 26, "retention_rate": 0.4, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.75, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1123}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 26, "retention_rate": 0.4, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.75, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1123}, "2": {"facts_total": 65, "facts_retained": 26, "retention_rate": 0.4, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.75, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1123}}}, "reference_key_id": "keys/synthetic/syn_03009_2000.json", "task_scores": {"correct": 1.0, "abstained": 1.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "3bd2611f7bc59d6f4b6323c3ac1e9f7af93bcc7ed3e43397b183cdb4d6c6e058", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "c834c289f75d913352d34aa8", "attempt_id": "21ed324326c5", "run_group_id": "E4_core:A4|repairable|qwen3b:h4", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:20:09.997752Z", "task_family": "synthetic", "document_id": "syn_03009_2000", "document_hash": "6ed94dcf728255bd7a7d11c97c362930728fb81a3e91f1906eb6eec86e0fedb7", "question_id": "q00_single_fact", "split": "test", "cluster_id": "syn_03009_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 4, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1976, "query_tokens": 595, "message_tokens": 512, "message_utf8_bytes": 1123, "cache_tensor_bytes": null, "serialized_payload_bytes": 1123, "side_information_bytes": 0, "fetched_bytes": 2199, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 89109.59145694505, "hops": {"1": 22990.705040981993, "2": 22287.307874998078, "3": 22002.059874939732, "4": 21829.51866602525}, "receiver_prefill_ms": 541.2793329451233, "ttft_ms": 38.2004160201177, "decode_ms": 392.7700830390677}, "total_wall_ms": 90082.63687288854, "ttft_ms": 38.2004160201177, "input_usage": {"lookups": 12}, "output_usage": {"output_tokens": 10}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "PA-37F-6227", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 26, "retention_rate": 0.4, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.75, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1123}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 26, "retention_rate": 0.4, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.75, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1123}, "2": {"facts_total": 65, "facts_retained": 26, "retention_rate": 0.4, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.75, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1123}, "3": {"facts_total": 65, "facts_retained": 26, "retention_rate": 0.4, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.75, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1123}, "4": {"facts_total": 65, "facts_retained": 26, "retention_rate": 0.4, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.75, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1123}}}, "reference_key_id": "keys/synthetic/syn_03009_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "3bd2611f7bc59d6f4b6323c3ac1e9f7af93bcc7ed3e43397b183cdb4d6c6e058", "output_hash": "5b7645cf6e8140f4c0bf6ed4ad76121cf2924fe7e666d92161812527b86f4409"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "696e68d77e5fda57b4c928cf", "attempt_id": "3cc236cf7a63", "run_group_id": "E4_core:A4|repairable|qwen3b:h4", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:20:10.990255Z", "task_family": "synthetic", "document_id": "syn_03009_2000", "document_hash": "6ed94dcf728255bd7a7d11c97c362930728fb81a3e91f1906eb6eec86e0fedb7", "question_id": "q01_single_fact", "split": "test", "cluster_id": "syn_03009_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 4, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1976, "query_tokens": 600, "message_tokens": 512, "message_utf8_bytes": 1123, "cache_tensor_bytes": null, "serialized_payload_bytes": 1123, "side_information_bytes": 0, "fetched_bytes": 2199, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 89109.59145694505, "hops": {"1": 22990.705040981993, "2": 22287.307874998078, "3": 22002.059874939732, "4": 21829.51866602525}, "receiver_prefill_ms": 369.7572919772938, "ttft_ms": 61.67633295990527, "decode_ms": 61.46825000178069}, "total_wall_ms": 89603.30449894536, "ttft_ms": 61.67633295990527, "input_usage": {"lookups": 12}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 26, "retention_rate": 0.4, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.75, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1123}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 26, "retention_rate": 0.4, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.75, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1123}, "2": {"facts_total": 65, "facts_retained": 26, "retention_rate": 0.4, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.75, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1123}, "3": {"facts_total": 65, "facts_retained": 26, "retention_rate": 0.4, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.75, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1123}, "4": {"facts_total": 65, "facts_retained": 26, "retention_rate": 0.4, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.75, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1123}}}, "reference_key_id": "keys/synthetic/syn_03009_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "3bd2611f7bc59d6f4b6323c3ac1e9f7af93bcc7ed3e43397b183cdb4d6c6e058", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "64f65c315a5eda75a9ccde15", "attempt_id": "625bef7b0593", "run_group_id": "E4_core:A4|repairable|qwen3b:h4", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:20:11.501722Z", "task_family": "synthetic", "document_id": "syn_03009_2000", "document_hash": "6ed94dcf728255bd7a7d11c97c362930728fb81a3e91f1906eb6eec86e0fedb7", "question_id": "q02_exact_copy", "split": "test", "cluster_id": "syn_03009_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 4, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1976, "query_tokens": 598, "message_tokens": 512, "message_utf8_bytes": 1123, "cache_tensor_bytes": null, "serialized_payload_bytes": 1123, "side_information_bytes": 0, "fetched_bytes": 2199, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 89109.59145694505, "hops": {"1": 22990.705040981993, "2": 22287.307874998078, "3": 22002.059874939732, "4": 21829.51866602525}, "receiver_prefill_ms": 562.5808340264484, "ttft_ms": 26.29549999255687, "decode_ms": 39.689583936706185}, "total_wall_ms": 89738.99179091677, "ttft_ms": 26.29549999255687, "input_usage": {"lookups": 12}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 26, "retention_rate": 0.4, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.75, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1123}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 26, "retention_rate": 0.4, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.75, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1123}, "2": {"facts_total": 65, "facts_retained": 26, "retention_rate": 0.4, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.75, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1123}, "3": {"facts_total": 65, "facts_retained": 26, "retention_rate": 0.4, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.75, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1123}, "4": {"facts_total": 65, "facts_retained": 26, "retention_rate": 0.4, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.75, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1123}}}, "reference_key_id": "keys/synthetic/syn_03009_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "3bd2611f7bc59d6f4b6323c3ac1e9f7af93bcc7ed3e43397b183cdb4d6c6e058", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "405f0ae0a0e92a981da8fcb4", "attempt_id": "37332c0e8a23", "run_group_id": "E4_core:A4|repairable|qwen3b:h4", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:20:12.150979Z", "task_family": "synthetic", "document_id": "syn_03009_2000", "document_hash": "6ed94dcf728255bd7a7d11c97c362930728fb81a3e91f1906eb6eec86e0fedb7", "question_id": "q03_exact_copy", "split": "test", "cluster_id": "syn_03009_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 4, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1976, "query_tokens": 594, "message_tokens": 512, "message_utf8_bytes": 1123, "cache_tensor_bytes": null, "serialized_payload_bytes": 1123, "side_information_bytes": 0, "fetched_bytes": 2199, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 89109.59145694505, "hops": {"1": 22990.705040981993, "2": 22287.307874998078, "3": 22002.059874939732, "4": 21829.51866602525}, "receiver_prefill_ms": 411.2509590340778, "ttft_ms": 25.403250008821487, "decode_ms": 166.22224997263402}, "total_wall_ms": 89713.27633201145, "ttft_ms": 25.403250008821487, "input_usage": {"lookups": 12}, "output_usage": {"output_tokens": 5}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "Idris Bergström", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 26, "retention_rate": 0.4, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.75, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1123}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 26, "retention_rate": 0.4, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.75, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1123}, "2": {"facts_total": 65, "facts_retained": 26, "retention_rate": 0.4, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.75, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1123}, "3": {"facts_total": 65, "facts_retained": 26, "retention_rate": 0.4, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.75, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1123}, "4": {"facts_total": 65, "facts_retained": 26, "retention_rate": 0.4, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.75, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1123}}}, "reference_key_id": "keys/synthetic/syn_03009_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "3bd2611f7bc59d6f4b6323c3ac1e9f7af93bcc7ed3e43397b183cdb4d6c6e058", "output_hash": "fd88832c9fd0ab175818d76d875d70156225bd4b34ec401514546ef70f4f826c"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "5009cb7f7ba116dd3cefba54", "attempt_id": "812f2e0729d2", "run_group_id": "E4_core:A4|repairable|qwen3b:h4", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:20:12.773371Z", "task_family": "synthetic", "document_id": "syn_03009_2000", "document_hash": "6ed94dcf728255bd7a7d11c97c362930728fb81a3e91f1906eb6eec86e0fedb7", "question_id": "q04_multi_hop", "split": "test", "cluster_id": "syn_03009_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 4, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1976, "query_tokens": 602, "message_tokens": 512, "message_utf8_bytes": 1123, "cache_tensor_bytes": null, "serialized_payload_bytes": 1123, "side_information_bytes": 0, "fetched_bytes": 2199, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 89109.59145694505, "hops": {"1": 22990.705040981993, "2": 22287.307874998078, "3": 22002.059874939732, "4": 21829.51866602525}, "receiver_prefill_ms": 374.1440000012517, "ttft_ms": 63.85087501257658, "decode_ms": 479.7987079946324}, "total_wall_ms": 90028.15208188258, "ttft_ms": 63.85087501257658, "input_usage": {"lookups": 12}, "output_usage": {"output_tokens": 7}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "4670383", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 26, "retention_rate": 0.4, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.75, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1123}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 26, "retention_rate": 0.4, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.75, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1123}, "2": {"facts_total": 65, "facts_retained": 26, "retention_rate": 0.4, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.75, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1123}, "3": {"facts_total": 65, "facts_retained": 26, "retention_rate": 0.4, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.75, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1123}, "4": {"facts_total": 65, "facts_retained": 26, "retention_rate": 0.4, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.75, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1123}}}, "reference_key_id": "keys/synthetic/syn_03009_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "3bd2611f7bc59d6f4b6323c3ac1e9f7af93bcc7ed3e43397b183cdb4d6c6e058", "output_hash": "19ae8c357e5500bf494a6163bfbd11380b2996e8ea532a2c64f4a599464820f7"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "5b87ec4f7f78d459caa7fb55", "attempt_id": "1814659aa35f", "run_group_id": "E4_core:A4|repairable|qwen3b:h4", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:20:13.708981Z", "task_family": "synthetic", "document_id": "syn_03009_2000", "document_hash": "6ed94dcf728255bd7a7d11c97c362930728fb81a3e91f1906eb6eec86e0fedb7", "question_id": "q05_multi_hop", "split": "test", "cluster_id": "syn_03009_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 4, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1976, "query_tokens": 605, "message_tokens": 512, "message_utf8_bytes": 1123, "cache_tensor_bytes": null, "serialized_payload_bytes": 1123, "side_information_bytes": 0, "fetched_bytes": 2199, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 89109.59145694505, "hops": {"1": 22990.705040981993, "2": 22287.307874998078, "3": 22002.059874939732, "4": 21829.51866602525}, "receiver_prefill_ms": 398.66141602396965, "ttft_ms": 32.833208912052214, "decode_ms": 33.05783390533179}, "total_wall_ms": 89575.01241494901, "ttft_ms": 32.833208912052214, "input_usage": {"lookups": 12}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 26, "retention_rate": 0.4, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.75, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1123}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 26, "retention_rate": 0.4, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.75, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1123}, "2": {"facts_total": 65, "facts_retained": 26, "retention_rate": 0.4, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.75, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1123}, "3": {"facts_total": 65, "facts_retained": 26, "retention_rate": 0.4, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.75, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1123}, "4": {"facts_total": 65, "facts_retained": 26, "retention_rate": 0.4, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.75, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1123}}}, "reference_key_id": "keys/synthetic/syn_03009_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "3bd2611f7bc59d6f4b6323c3ac1e9f7af93bcc7ed3e43397b183cdb4d6c6e058", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "4a5f88cdb22e101c5d83b699", "attempt_id": "e5f35280a451", "run_group_id": "E4_core:A4|repairable|qwen3b:h4", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:20:14.191819Z", "task_family": "synthetic", "document_id": "syn_03009_2000", "document_hash": "6ed94dcf728255bd7a7d11c97c362930728fb81a3e91f1906eb6eec86e0fedb7", "question_id": "q06_negation", "split": "test", "cluster_id": "syn_03009_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 4, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1976, "query_tokens": 595, "message_tokens": 512, "message_utf8_bytes": 1123, "cache_tensor_bytes": null, "serialized_payload_bytes": 1123, "side_information_bytes": 0, "fetched_bytes": 2199, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 89109.59145694505, "hops": {"1": 22990.705040981993, "2": 22287.307874998078, "3": 22002.059874939732, "4": 21829.51866602525}, "receiver_prefill_ms": 402.4305830243975, "ttft_ms": 27.345791924744844, "decode_ms": 17.103874939493835}, "total_wall_ms": 89557.06399888732, "ttft_ms": 27.345791924744844, "input_usage": {"lookups": 12}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 26, "retention_rate": 0.4, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.75, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1123}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 26, "retention_rate": 0.4, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.75, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1123}, "2": {"facts_total": 65, "facts_retained": 26, "retention_rate": 0.4, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.75, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1123}, "3": {"facts_total": 65, "facts_retained": 26, "retention_rate": 0.4, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.75, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1123}, "4": {"facts_total": 65, "facts_retained": 26, "retention_rate": 0.4, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.75, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1123}}}, "reference_key_id": "keys/synthetic/syn_03009_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "3bd2611f7bc59d6f4b6323c3ac1e9f7af93bcc7ed3e43397b183cdb4d6c6e058", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "1d7e2bd7781c64dc017e2a77", "attempt_id": "d1909a8bf4d7", "run_group_id": "E4_core:A4|repairable|qwen3b:h4", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:20:14.656510Z", "task_family": "synthetic", "document_id": "syn_03009_2000", "document_hash": "6ed94dcf728255bd7a7d11c97c362930728fb81a3e91f1906eb6eec86e0fedb7", "question_id": "q07_negation", "split": "test", "cluster_id": "syn_03009_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 4, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1976, "query_tokens": 594, "message_tokens": 512, "message_utf8_bytes": 1123, "cache_tensor_bytes": null, "serialized_payload_bytes": 1123, "side_information_bytes": 0, "fetched_bytes": 2199, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 89109.59145694505, "hops": {"1": 22990.705040981993, "2": 22287.307874998078, "3": 22002.059874939732, "4": 21829.51866602525}, "receiver_prefill_ms": 374.6467500459403, "ttft_ms": 67.40920897573233, "decode_ms": 59.6394999884069}, "total_wall_ms": 89612.10554093122, "ttft_ms": 67.40920897573233, "input_usage": {"lookups": 12}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 26, "retention_rate": 0.4, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.75, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1123}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 26, "retention_rate": 0.4, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.75, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1123}, "2": {"facts_total": 65, "facts_retained": 26, "retention_rate": 0.4, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.75, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1123}, "3": {"facts_total": 65, "facts_retained": 26, "retention_rate": 0.4, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.75, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1123}, "4": {"facts_total": 65, "facts_retained": 26, "retention_rate": 0.4, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.75, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1123}}}, "reference_key_id": "keys/synthetic/syn_03009_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "3bd2611f7bc59d6f4b6323c3ac1e9f7af93bcc7ed3e43397b183cdb4d6c6e058", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "3153d8ac2ba9c833a165b961", "attempt_id": "5f19bd620d61", "run_group_id": "E4_core:A4|repairable|qwen3b:h4", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:20:15.179215Z", "task_family": "synthetic", "document_id": "syn_03009_2000", "document_hash": "6ed94dcf728255bd7a7d11c97c362930728fb81a3e91f1906eb6eec86e0fedb7", "question_id": "q08_revision", "split": "test", "cluster_id": "syn_03009_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 4, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1976, "query_tokens": 612, "message_tokens": 512, "message_utf8_bytes": 1123, "cache_tensor_bytes": null, "serialized_payload_bytes": 1123, "side_information_bytes": 0, "fetched_bytes": 2199, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 89109.59145694505, "hops": {"1": 22990.705040981993, "2": 22287.307874998078, "3": 22002.059874939732, "4": 21829.51866602525}, "receiver_prefill_ms": 505.19950001034886, "ttft_ms": 37.71029191557318, "decode_ms": 257.848541950807}, "total_wall_ms": 89911.01733199321, "ttft_ms": 37.71029191557318, "input_usage": {"lookups": 12}, "output_usage": {"output_tokens": 8}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "10047105", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 26, "retention_rate": 0.4, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.75, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1123}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 26, "retention_rate": 0.4, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.75, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1123}, "2": {"facts_total": 65, "facts_retained": 26, "retention_rate": 0.4, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.75, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1123}, "3": {"facts_total": 65, "facts_retained": 26, "retention_rate": 0.4, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.75, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1123}, "4": {"facts_total": 65, "facts_retained": 26, "retention_rate": 0.4, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.75, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1123}}}, "reference_key_id": "keys/synthetic/syn_03009_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "3bd2611f7bc59d6f4b6323c3ac1e9f7af93bcc7ed3e43397b183cdb4d6c6e058", "output_hash": "bea99610d1e3e75a7f20939faf8e6824e02237b156dacd6442ba2a7e02617510"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "434b6e96ea496de722964ddd", "attempt_id": "6c6f9ddfa2b9", "run_group_id": "E4_core:A4|repairable|qwen3b:h4", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:20:15.998920Z", "task_family": "synthetic", "document_id": "syn_03009_2000", "document_hash": "6ed94dcf728255bd7a7d11c97c362930728fb81a3e91f1906eb6eec86e0fedb7", "question_id": "q09_revision", "split": "test", "cluster_id": "syn_03009_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 4, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1976, "query_tokens": 605, "message_tokens": 512, "message_utf8_bytes": 1123, "cache_tensor_bytes": null, "serialized_payload_bytes": 1123, "side_information_bytes": 0, "fetched_bytes": 2199, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 89109.59145694505, "hops": {"1": 22990.705040981993, "2": 22287.307874998078, "3": 22002.059874939732, "4": 21829.51866602525}, "receiver_prefill_ms": 376.76962499972433, "ttft_ms": 14.773499919101596, "decode_ms": 449.9700410524383}, "total_wall_ms": 89951.89191505779, "ttft_ms": 14.773499919101596, "input_usage": {"lookups": 12}, "output_usage": {"output_tokens": 8}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "10047105", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 26, "retention_rate": 0.4, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.75, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1123}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 26, "retention_rate": 0.4, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.75, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1123}, "2": {"facts_total": 65, "facts_retained": 26, "retention_rate": 0.4, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.75, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1123}, "3": {"facts_total": 65, "facts_retained": 26, "retention_rate": 0.4, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.75, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1123}, "4": {"facts_total": 65, "facts_retained": 26, "retention_rate": 0.4, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.75, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1123}}}, "reference_key_id": "keys/synthetic/syn_03009_2000.json", "task_scores": {"correct": 0.0, "abstained": 0.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "3bd2611f7bc59d6f4b6323c3ac1e9f7af93bcc7ed3e43397b183cdb4d6c6e058", "output_hash": "bea99610d1e3e75a7f20939faf8e6824e02237b156dacd6442ba2a7e02617510"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "13ef670997f12e01cbeac615", "attempt_id": "fc445a02794f", "run_group_id": "E4_core:A4|repairable|qwen3b:h4", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:20:16.859254Z", "task_family": "synthetic", "document_id": "syn_03009_2000", "document_hash": "6ed94dcf728255bd7a7d11c97c362930728fb81a3e91f1906eb6eec86e0fedb7", "question_id": "q10_temporal", "split": "test", "cluster_id": "syn_03009_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 4, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1976, "query_tokens": 614, "message_tokens": 512, "message_utf8_bytes": 1123, "cache_tensor_bytes": null, "serialized_payload_bytes": 1123, "side_information_bytes": 0, "fetched_bytes": 2199, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 89109.59145694505, "hops": {"1": 22990.705040981993, "2": 22287.307874998078, "3": 22002.059874939732, "4": 21829.51866602525}, "receiver_prefill_ms": 479.3967920122668, "ttft_ms": 38.57708303257823, "decode_ms": 25.352417025715113}, "total_wall_ms": 89653.61391589977, "ttft_ms": 38.57708303257823, "input_usage": {"lookups": 12}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 26, "retention_rate": 0.4, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.75, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1123}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 26, "retention_rate": 0.4, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.75, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1123}, "2": {"facts_total": 65, "facts_retained": 26, "retention_rate": 0.4, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.75, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1123}, "3": {"facts_total": 65, "facts_retained": 26, "retention_rate": 0.4, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.75, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1123}, "4": {"facts_total": 65, "facts_retained": 26, "retention_rate": 0.4, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.75, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1123}}}, "reference_key_id": "keys/synthetic/syn_03009_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "3bd2611f7bc59d6f4b6323c3ac1e9f7af93bcc7ed3e43397b183cdb4d6c6e058", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "691db9160ddd3542020a1a8f", "attempt_id": "bbb630f424bd", "run_group_id": "E4_core:A4|repairable|qwen3b:h4", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:20:17.423445Z", "task_family": "synthetic", "document_id": "syn_03009_2000", "document_hash": "6ed94dcf728255bd7a7d11c97c362930728fb81a3e91f1906eb6eec86e0fedb7", "question_id": "q11_aggregation", "split": "test", "cluster_id": "syn_03009_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 4, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1976, "query_tokens": 605, "message_tokens": 512, "message_utf8_bytes": 1123, "cache_tensor_bytes": null, "serialized_payload_bytes": 1123, "side_information_bytes": 0, "fetched_bytes": 2199, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 89109.59145694505, "hops": {"1": 22990.705040981993, "2": 22287.307874998078, "3": 22002.059874939732, "4": 21829.51866602525}, "receiver_prefill_ms": 404.73116701468825, "ttft_ms": 33.22354203555733, "decode_ms": 228.68475003633648}, "total_wall_ms": 89776.9820819376, "ttft_ms": 33.22354203555733, "input_usage": {"lookups": 12}, "output_usage": {"output_tokens": 7}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "5781936", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 26, "retention_rate": 0.4, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.75, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1123}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 26, "retention_rate": 0.4, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.75, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1123}, "2": {"facts_total": 65, "facts_retained": 26, "retention_rate": 0.4, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.75, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1123}, "3": {"facts_total": 65, "facts_retained": 26, "retention_rate": 0.4, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.75, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1123}, "4": {"facts_total": 65, "facts_retained": 26, "retention_rate": 0.4, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.75, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1123}}}, "reference_key_id": "keys/synthetic/syn_03009_2000.json", "task_scores": {"correct": 0.0, "abstained": 0.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "3bd2611f7bc59d6f4b6323c3ac1e9f7af93bcc7ed3e43397b183cdb4d6c6e058", "output_hash": "0281303171fb1ff03a7dba9f33d3ecbc477bff6e0cfb04aa1eb59f9c489a8a36"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "1d102ffdfc800c34a9d2bc67", "attempt_id": "622d97721d51", "run_group_id": "E4_core:A4|repairable|qwen3b:h4", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:20:18.109224Z", "task_family": "synthetic", "document_id": "syn_03009_2000", "document_hash": "6ed94dcf728255bd7a7d11c97c362930728fb81a3e91f1906eb6eec86e0fedb7", "question_id": "q12_unit_conversion", "split": "test", "cluster_id": "syn_03009_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 4, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1976, "query_tokens": 603, "message_tokens": 512, "message_utf8_bytes": 1123, "cache_tensor_bytes": null, "serialized_payload_bytes": 1123, "side_information_bytes": 0, "fetched_bytes": 2199, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 89109.59145694505, "hops": {"1": 22990.705040981993, "2": 22287.307874998078, "3": 22002.059874939732, "4": 21829.51866602525}, "receiver_prefill_ms": 522.9747910052538, "ttft_ms": 87.29687496088445, "decode_ms": 46.600583009421825}, "total_wall_ms": 89767.1736649936, "ttft_ms": 87.29687496088445, "input_usage": {"lookups": 12}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 26, "retention_rate": 0.4, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.75, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1123}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 26, "retention_rate": 0.4, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.75, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1123}, "2": {"facts_total": 65, "facts_retained": 26, "retention_rate": 0.4, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.75, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1123}, "3": {"facts_total": 65, "facts_retained": 26, "retention_rate": 0.4, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.75, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1123}, "4": {"facts_total": 65, "facts_retained": 26, "retention_rate": 0.4, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.75, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1123}}}, "reference_key_id": "keys/synthetic/syn_03009_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "3bd2611f7bc59d6f4b6323c3ac1e9f7af93bcc7ed3e43397b183cdb4d6c6e058", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "28845dd1936427a5e0bf035a", "attempt_id": "b884043c4175", "run_group_id": "E4_core:A4|repairable|qwen3b:h4", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:20:18.785757Z", "task_family": "synthetic", "document_id": "syn_03009_2000", "document_hash": "6ed94dcf728255bd7a7d11c97c362930728fb81a3e91f1906eb6eec86e0fedb7", "question_id": "q13_unanswerable", "split": "test", "cluster_id": "syn_03009_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 4, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1976, "query_tokens": 607, "message_tokens": 512, "message_utf8_bytes": 1123, "cache_tensor_bytes": null, "serialized_payload_bytes": 1123, "side_information_bytes": 0, "fetched_bytes": 2199, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 89109.59145694505, "hops": {"1": 22990.705040981993, "2": 22287.307874998078, "3": 22002.059874939732, "4": 21829.51866602525}, "receiver_prefill_ms": 391.14924997556955, "ttft_ms": 34.65516690630466, "decode_ms": 43.32529194653034}, "total_wall_ms": 89579.2970819166, "ttft_ms": 34.65516690630466, "input_usage": {"lookups": 12}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 26, "retention_rate": 0.4, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.75, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1123}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 26, "retention_rate": 0.4, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.75, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1123}, "2": {"facts_total": 65, "facts_retained": 26, "retention_rate": 0.4, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.75, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1123}, "3": {"facts_total": 65, "facts_retained": 26, "retention_rate": 0.4, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.75, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1123}, "4": {"facts_total": 65, "facts_retained": 26, "retention_rate": 0.4, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.75, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1123}}}, "reference_key_id": "keys/synthetic/syn_03009_2000.json", "task_scores": {"correct": 1.0, "abstained": 1.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "3bd2611f7bc59d6f4b6323c3ac1e9f7af93bcc7ed3e43397b183cdb4d6c6e058", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "4344d97aac0bad54c9291ceb", "attempt_id": "9f0e2022f02d", "run_group_id": "E4_core:A4|repairable|qwen3b:h4", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:20:19.272048Z", "task_family": "synthetic", "document_id": "syn_03009_2000", "document_hash": "6ed94dcf728255bd7a7d11c97c362930728fb81a3e91f1906eb6eec86e0fedb7", "question_id": "q14_unanswerable", "split": "test", "cluster_id": "syn_03009_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 4, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1976, "query_tokens": 607, "message_tokens": 512, "message_utf8_bytes": 1123, "cache_tensor_bytes": null, "serialized_payload_bytes": 1123, "side_information_bytes": 0, "fetched_bytes": 2199, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 89109.59145694505, "hops": {"1": 22990.705040981993, "2": 22287.307874998078, "3": 22002.059874939732, "4": 21829.51866602525}, "receiver_prefill_ms": 406.22587490361184, "ttft_ms": 23.136707954108715, "decode_ms": 19.882624968886375}, "total_wall_ms": 89559.46458200924, "ttft_ms": 23.136707954108715, "input_usage": {"lookups": 12}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 26, "retention_rate": 0.4, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.75, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1123}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 26, "retention_rate": 0.4, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.75, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1123}, "2": {"facts_total": 65, "facts_retained": 26, "retention_rate": 0.4, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.75, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1123}, "3": {"facts_total": 65, "facts_retained": 26, "retention_rate": 0.4, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.75, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1123}, "4": {"facts_total": 65, "facts_retained": 26, "retention_rate": 0.4, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.75, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1123}}}, "reference_key_id": "keys/synthetic/syn_03009_2000.json", "task_scores": {"correct": 1.0, "abstained": 1.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "3bd2611f7bc59d6f4b6323c3ac1e9f7af93bcc7ed3e43397b183cdb4d6c6e058", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "bb77b63a70d665a027a7fbc0", "attempt_id": "cb114e3fd70b", "run_group_id": "E4_core:A4|repairable|qwen3b:h8", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:21:47.816171Z", "task_family": "synthetic", "document_id": "syn_03009_2000", "document_hash": "6ed94dcf728255bd7a7d11c97c362930728fb81a3e91f1906eb6eec86e0fedb7", "question_id": "q00_single_fact", "split": "test", "cluster_id": "syn_03009_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 8, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1976, "query_tokens": 595, "message_tokens": 512, "message_utf8_bytes": 1123, "cache_tensor_bytes": null, "serialized_payload_bytes": 1123, "side_information_bytes": 0, "fetched_bytes": 5131, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 177180.95974787138, "hops": {"1": 22990.705040981993, "2": 22287.307874998078, "3": 22002.059874939732, "4": 21829.51866602525, "5": 21880.574999959208, "6": 21869.980000075884, "7": 22178.420749958605, "8": 22142.392540932633}, "receiver_prefill_ms": 395.6743750022724, "ttft_ms": 30.33824998419732, "decode_ms": 347.48016705270857}, "total_wall_ms": 177955.15637286007, "ttft_ms": 30.33824998419732, "input_usage": {"lookups": 28}, "output_usage": {"output_tokens": 10}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "PA-37F-6227", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 26, "retention_rate": 0.4, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.75, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1123}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 26, "retention_rate": 0.4, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.75, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1123}, "2": {"facts_total": 65, "facts_retained": 26, "retention_rate": 0.4, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.75, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1123}, "3": {"facts_total": 65, "facts_retained": 26, "retention_rate": 0.4, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.75, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1123}, "4": {"facts_total": 65, "facts_retained": 26, "retention_rate": 0.4, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.75, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1123}, "5": {"facts_total": 65, "facts_retained": 26, "retention_rate": 0.4, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.75, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1123}, "6": {"facts_total": 65, "facts_retained": 26, "retention_rate": 0.4, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.75, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1123}, "7": {"facts_total": 65, "facts_retained": 26, "retention_rate": 0.4, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.75, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1123}, "8": {"facts_total": 65, "facts_retained": 26, "retention_rate": 0.4, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.75, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1123}}}, "reference_key_id": "keys/synthetic/syn_03009_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "3bd2611f7bc59d6f4b6323c3ac1e9f7af93bcc7ed3e43397b183cdb4d6c6e058", "output_hash": "5b7645cf6e8140f4c0bf6ed4ad76121cf2924fe7e666d92161812527b86f4409"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "c2bf092c597cfa052bba03cf", "attempt_id": "f0a110334ca6", "run_group_id": "E4_core:A4|repairable|qwen3b:h8", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:21:48.607430Z", "task_family": "synthetic", "document_id": "syn_03009_2000", "document_hash": "6ed94dcf728255bd7a7d11c97c362930728fb81a3e91f1906eb6eec86e0fedb7", "question_id": "q01_single_fact", "split": "test", "cluster_id": "syn_03009_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 8, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1976, "query_tokens": 600, "message_tokens": 512, "message_utf8_bytes": 1123, "cache_tensor_bytes": null, "serialized_payload_bytes": 1123, "side_information_bytes": 0, "fetched_bytes": 5131, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 177180.95974787138, "hops": {"1": 22990.705040981993, "2": 22287.307874998078, "3": 22002.059874939732, "4": 21829.51866602525, "5": 21880.574999959208, "6": 21869.980000075884, "7": 22178.420749958605, "8": 22142.392540932633}, "receiver_prefill_ms": 350.0622910214588, "ttft_ms": 52.29170806705952, "decode_ms": 56.58462492283434}, "total_wall_ms": 177640.77299786732, "ttft_ms": 52.29170806705952, "input_usage": {"lookups": 28}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 26, "retention_rate": 0.4, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.75, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1123}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 26, "retention_rate": 0.4, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.75, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1123}, "2": {"facts_total": 65, "facts_retained": 26, "retention_rate": 0.4, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.75, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1123}, "3": {"facts_total": 65, "facts_retained": 26, "retention_rate": 0.4, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.75, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1123}, "4": {"facts_total": 65, "facts_retained": 26, "retention_rate": 0.4, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.75, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1123}, "5": {"facts_total": 65, "facts_retained": 26, "retention_rate": 0.4, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.75, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1123}, "6": {"facts_total": 65, "facts_retained": 26, "retention_rate": 0.4, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.75, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1123}, "7": {"facts_total": 65, "facts_retained": 26, "retention_rate": 0.4, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.75, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1123}, "8": {"facts_total": 65, "facts_retained": 26, "retention_rate": 0.4, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.75, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1123}}}, "reference_key_id": "keys/synthetic/syn_03009_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "3bd2611f7bc59d6f4b6323c3ac1e9f7af93bcc7ed3e43397b183cdb4d6c6e058", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "8ed8af51ad335f55a648d792", "attempt_id": "0cd6187b3aab", "run_group_id": "E4_core:A4|repairable|qwen3b:h8", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:21:49.084838Z", "task_family": "synthetic", "document_id": "syn_03009_2000", "document_hash": "6ed94dcf728255bd7a7d11c97c362930728fb81a3e91f1906eb6eec86e0fedb7", "question_id": "q02_exact_copy", "split": "test", "cluster_id": "syn_03009_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 8, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1976, "query_tokens": 598, "message_tokens": 512, "message_utf8_bytes": 1123, "cache_tensor_bytes": null, "serialized_payload_bytes": 1123, "side_information_bytes": 0, "fetched_bytes": 5131, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 177180.95974787138, "hops": {"1": 22990.705040981993, "2": 22287.307874998078, "3": 22002.059874939732, "4": 21829.51866602525, "5": 21880.574999959208, "6": 21869.980000075884, "7": 22178.420749958605, "8": 22142.392540932633}, "receiver_prefill_ms": 599.110292037949, "ttft_ms": 39.742833003401756, "decode_ms": 27.085250010713935}, "total_wall_ms": 177847.51841484103, "ttft_ms": 39.742833003401756, "input_usage": {"lookups": 28}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 26, "retention_rate": 0.4, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.75, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1123}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 26, "retention_rate": 0.4, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.75, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1123}, "2": {"facts_total": 65, "facts_retained": 26, "retention_rate": 0.4, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.75, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1123}, "3": {"facts_total": 65, "facts_retained": 26, "retention_rate": 0.4, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.75, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1123}, "4": {"facts_total": 65, "facts_retained": 26, "retention_rate": 0.4, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.75, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1123}, "5": {"facts_total": 65, "facts_retained": 26, "retention_rate": 0.4, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.75, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1123}, "6": {"facts_total": 65, "facts_retained": 26, "retention_rate": 0.4, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.75, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1123}, "7": {"facts_total": 65, "facts_retained": 26, "retention_rate": 0.4, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.75, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1123}, "8": {"facts_total": 65, "facts_retained": 26, "retention_rate": 0.4, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.75, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1123}}}, "reference_key_id": "keys/synthetic/syn_03009_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "3bd2611f7bc59d6f4b6323c3ac1e9f7af93bcc7ed3e43397b183cdb4d6c6e058", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "01e72761d0721d36affefe26", "attempt_id": "1af5510ab35d", "run_group_id": "E4_core:A4|repairable|qwen3b:h8", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:21:49.769863Z", "task_family": "synthetic", "document_id": "syn_03009_2000", "document_hash": "6ed94dcf728255bd7a7d11c97c362930728fb81a3e91f1906eb6eec86e0fedb7", "question_id": "q03_exact_copy", "split": "test", "cluster_id": "syn_03009_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 8, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1976, "query_tokens": 594, "message_tokens": 512, "message_utf8_bytes": 1123, "cache_tensor_bytes": null, "serialized_payload_bytes": 1123, "side_information_bytes": 0, "fetched_bytes": 5131, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 177180.95974787138, "hops": {"1": 22990.705040981993, "2": 22287.307874998078, "3": 22002.059874939732, "4": 21829.51866602525, "5": 21880.574999959208, "6": 21869.980000075884, "7": 22178.420749958605, "8": 22142.392540932633}, "receiver_prefill_ms": 400.5851249676198, "ttft_ms": 39.71925005316734, "decode_ms": 160.5768749723211}, "total_wall_ms": 177782.6231229119, "ttft_ms": 39.71925005316734, "input_usage": {"lookups": 28}, "output_usage": {"output_tokens": 5}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "Idris Bergström", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 26, "retention_rate": 0.4, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.75, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1123}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 26, "retention_rate": 0.4, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.75, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1123}, "2": {"facts_total": 65, "facts_retained": 26, "retention_rate": 0.4, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.75, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1123}, "3": {"facts_total": 65, "facts_retained": 26, "retention_rate": 0.4, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.75, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1123}, "4": {"facts_total": 65, "facts_retained": 26, "retention_rate": 0.4, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.75, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1123}, "5": {"facts_total": 65, "facts_retained": 26, "retention_rate": 0.4, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.75, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1123}, "6": {"facts_total": 65, "facts_retained": 26, "retention_rate": 0.4, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.75, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1123}, "7": {"facts_total": 65, "facts_retained": 26, "retention_rate": 0.4, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.75, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1123}, "8": {"facts_total": 65, "facts_retained": 26, "retention_rate": 0.4, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.75, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1123}}}, "reference_key_id": "keys/synthetic/syn_03009_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "3bd2611f7bc59d6f4b6323c3ac1e9f7af93bcc7ed3e43397b183cdb4d6c6e058", "output_hash": "fd88832c9fd0ab175818d76d875d70156225bd4b34ec401514546ef70f4f826c"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "7e74d6e527bb6cc9caa25174", "attempt_id": "889056090fd6", "run_group_id": "E4_core:A4|repairable|qwen3b:h8", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:21:50.389012Z", "task_family": "synthetic", "document_id": "syn_03009_2000", "document_hash": "6ed94dcf728255bd7a7d11c97c362930728fb81a3e91f1906eb6eec86e0fedb7", "question_id": "q04_multi_hop", "split": "test", "cluster_id": "syn_03009_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 8, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1976, "query_tokens": 602, "message_tokens": 512, "message_utf8_bytes": 1123, "cache_tensor_bytes": null, "serialized_payload_bytes": 1123, "side_information_bytes": 0, "fetched_bytes": 5131, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 177180.95974787138, "hops": {"1": 22990.705040981993, "2": 22287.307874998078, "3": 22002.059874939732, "4": 21829.51866602525, "5": 21880.574999959208, "6": 21869.980000075884, "7": 22178.420749958605, "8": 22142.392540932633}, "receiver_prefill_ms": 353.25708298478276, "ttft_ms": 40.06533301435411, "decode_ms": 451.4374170685187}, "total_wall_ms": 178026.39962290414, "ttft_ms": 40.06533301435411, "input_usage": {"lookups": 28}, "output_usage": {"output_tokens": 7}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "4670383", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 26, "retention_rate": 0.4, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.75, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1123}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 26, "retention_rate": 0.4, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.75, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1123}, "2": {"facts_total": 65, "facts_retained": 26, "retention_rate": 0.4, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.75, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1123}, "3": {"facts_total": 65, "facts_retained": 26, "retention_rate": 0.4, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.75, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1123}, "4": {"facts_total": 65, "facts_retained": 26, "retention_rate": 0.4, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.75, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1123}, "5": {"facts_total": 65, "facts_retained": 26, "retention_rate": 0.4, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.75, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1123}, "6": {"facts_total": 65, "facts_retained": 26, "retention_rate": 0.4, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.75, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1123}, "7": {"facts_total": 65, "facts_retained": 26, "retention_rate": 0.4, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.75, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1123}, "8": {"facts_total": 65, "facts_retained": 26, "retention_rate": 0.4, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.75, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1123}}}, "reference_key_id": "keys/synthetic/syn_03009_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "3bd2611f7bc59d6f4b6323c3ac1e9f7af93bcc7ed3e43397b183cdb4d6c6e058", "output_hash": "19ae8c357e5500bf494a6163bfbd11380b2996e8ea532a2c64f4a599464820f7"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "78b030dd6704ade145f073ac", "attempt_id": "75e81bc4f6db", "run_group_id": "E4_core:A4|repairable|qwen3b:h8", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:21:51.253057Z", "task_family": "synthetic", "document_id": "syn_03009_2000", "document_hash": "6ed94dcf728255bd7a7d11c97c362930728fb81a3e91f1906eb6eec86e0fedb7", "question_id": "q05_multi_hop", "split": "test", "cluster_id": "syn_03009_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 8, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1976, "query_tokens": 605, "message_tokens": 512, "message_utf8_bytes": 1123, "cache_tensor_bytes": null, "serialized_payload_bytes": 1123, "side_information_bytes": 0, "fetched_bytes": 5131, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 177180.95974787138, "hops": {"1": 22990.705040981993, "2": 22287.307874998078, "3": 22002.059874939732, "4": 21829.51866602525, "5": 21880.574999959208, "6": 21869.980000075884, "7": 22178.420749958605, "8": 22142.392540932633}, "receiver_prefill_ms": 407.8661659732461, "ttft_ms": 36.276416038163006, "decode_ms": 30.153041006997228}, "total_wall_ms": 177655.9234559536, "ttft_ms": 36.276416038163006, "input_usage": {"lookups": 28}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 26, "retention_rate": 0.4, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.75, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1123}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 26, "retention_rate": 0.4, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.75, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1123}, "2": {"facts_total": 65, "facts_retained": 26, "retention_rate": 0.4, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.75, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1123}, "3": {"facts_total": 65, "facts_retained": 26, "retention_rate": 0.4, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.75, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1123}, "4": {"facts_total": 65, "facts_retained": 26, "retention_rate": 0.4, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.75, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1123}, "5": {"facts_total": 65, "facts_retained": 26, "retention_rate": 0.4, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.75, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1123}, "6": {"facts_total": 65, "facts_retained": 26, "retention_rate": 0.4, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.75, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1123}, "7": {"facts_total": 65, "facts_retained": 26, "retention_rate": 0.4, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.75, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1123}, "8": {"facts_total": 65, "facts_retained": 26, "retention_rate": 0.4, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.75, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1123}}}, "reference_key_id": "keys/synthetic/syn_03009_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "3bd2611f7bc59d6f4b6323c3ac1e9f7af93bcc7ed3e43397b183cdb4d6c6e058", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "e0a4f9ed5acf914776875ae0", "attempt_id": "ced7e7410985", "run_group_id": "E4_core:A4|repairable|qwen3b:h8", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:21:51.745668Z", "task_family": "synthetic", "document_id": "syn_03009_2000", "document_hash": "6ed94dcf728255bd7a7d11c97c362930728fb81a3e91f1906eb6eec86e0fedb7", "question_id": "q06_negation", "split": "test", "cluster_id": "syn_03009_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 8, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1976, "query_tokens": 595, "message_tokens": 512, "message_utf8_bytes": 1123, "cache_tensor_bytes": null, "serialized_payload_bytes": 1123, "side_information_bytes": 0, "fetched_bytes": 5131, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 177180.95974787138, "hops": {"1": 22990.705040981993, "2": 22287.307874998078, "3": 22002.059874939732, "4": 21829.51866602525, "5": 21880.574999959208, "6": 21869.980000075884, "7": 22178.420749958605, "8": 22142.392540932633}, "receiver_prefill_ms": 379.06299997121096, "ttft_ms": 20.864833961240947, "decode_ms": 36.088500055484474}, "total_wall_ms": 177617.89999785833, "ttft_ms": 20.864833961240947, "input_usage": {"lookups": 28}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 26, "retention_rate": 0.4, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.75, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1123}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 26, "retention_rate": 0.4, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.75, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1123}, "2": {"facts_total": 65, "facts_retained": 26, "retention_rate": 0.4, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.75, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1123}, "3": {"facts_total": 65, "facts_retained": 26, "retention_rate": 0.4, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.75, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1123}, "4": {"facts_total": 65, "facts_retained": 26, "retention_rate": 0.4, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.75, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1123}, "5": {"facts_total": 65, "facts_retained": 26, "retention_rate": 0.4, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.75, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1123}, "6": {"facts_total": 65, "facts_retained": 26, "retention_rate": 0.4, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.75, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1123}, "7": {"facts_total": 65, "facts_retained": 26, "retention_rate": 0.4, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.75, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1123}, "8": {"facts_total": 65, "facts_retained": 26, "retention_rate": 0.4, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.75, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1123}}}, "reference_key_id": "keys/synthetic/syn_03009_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "3bd2611f7bc59d6f4b6323c3ac1e9f7af93bcc7ed3e43397b183cdb4d6c6e058", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "d4cee8c7da8cb30d47b7553f", "attempt_id": "58b1ac443b3e", "run_group_id": "E4_core:A4|repairable|qwen3b:h8", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:21:52.200834Z", "task_family": "synthetic", "document_id": "syn_03009_2000", "document_hash": "6ed94dcf728255bd7a7d11c97c362930728fb81a3e91f1906eb6eec86e0fedb7", "question_id": "q07_negation", "split": "test", "cluster_id": "syn_03009_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 8, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1976, "query_tokens": 594, "message_tokens": 512, "message_utf8_bytes": 1123, "cache_tensor_bytes": null, "serialized_payload_bytes": 1123, "side_information_bytes": 0, "fetched_bytes": 5131, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 177180.95974787138, "hops": {"1": 22990.705040981993, "2": 22287.307874998078, "3": 22002.059874939732, "4": 21829.51866602525, "5": 21880.574999959208, "6": 21869.980000075884, "7": 22178.420749958605, "8": 22142.392540932633}, "receiver_prefill_ms": 394.1206659656018, "ttft_ms": 37.72883303463459, "decode_ms": 90.51662497222424}, "total_wall_ms": 177704.11499787588, "ttft_ms": 37.72883303463459, "input_usage": {"lookups": 28}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 26, "retention_rate": 0.4, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.75, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1123}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 26, "retention_rate": 0.4, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.75, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1123}, "2": {"facts_total": 65, "facts_retained": 26, "retention_rate": 0.4, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.75, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1123}, "3": {"facts_total": 65, "facts_retained": 26, "retention_rate": 0.4, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.75, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1123}, "4": {"facts_total": 65, "facts_retained": 26, "retention_rate": 0.4, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.75, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1123}, "5": {"facts_total": 65, "facts_retained": 26, "retention_rate": 0.4, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.75, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1123}, "6": {"facts_total": 65, "facts_retained": 26, "retention_rate": 0.4, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.75, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1123}, "7": {"facts_total": 65, "facts_retained": 26, "retention_rate": 0.4, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.75, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1123}, "8": {"facts_total": 65, "facts_retained": 26, "retention_rate": 0.4, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.75, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1123}}}, "reference_key_id": "keys/synthetic/syn_03009_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "3bd2611f7bc59d6f4b6323c3ac1e9f7af93bcc7ed3e43397b183cdb4d6c6e058", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "4b555d9fc8c1ab0d6f7e2999", "attempt_id": "e39980639f45", "run_group_id": "E4_core:A4|repairable|qwen3b:h8", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:21:52.741862Z", "task_family": "synthetic", "document_id": "syn_03009_2000", "document_hash": "6ed94dcf728255bd7a7d11c97c362930728fb81a3e91f1906eb6eec86e0fedb7", "question_id": "q08_revision", "split": "test", "cluster_id": "syn_03009_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 8, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1976, "query_tokens": 612, "message_tokens": 512, "message_utf8_bytes": 1123, "cache_tensor_bytes": null, "serialized_payload_bytes": 1123, "side_information_bytes": 0, "fetched_bytes": 5131, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 177180.95974787138, "hops": {"1": 22990.705040981993, "2": 22287.307874998078, "3": 22002.059874939732, "4": 21829.51866602525, "5": 21880.574999959208, "6": 21869.980000075884, "7": 22178.420749958605, "8": 22142.392540932633}, "receiver_prefill_ms": 478.64716697949916, "ttft_ms": 33.42470806092024, "decode_ms": 278.1113749369979}, "total_wall_ms": 177971.7715808656, "ttft_ms": 33.42470806092024, "input_usage": {"lookups": 28}, "output_usage": {"output_tokens": 8}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "10047105", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 26, "retention_rate": 0.4, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.75, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1123}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 26, "retention_rate": 0.4, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.75, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1123}, "2": {"facts_total": 65, "facts_retained": 26, "retention_rate": 0.4, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.75, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1123}, "3": {"facts_total": 65, "facts_retained": 26, "retention_rate": 0.4, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.75, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1123}, "4": {"facts_total": 65, "facts_retained": 26, "retention_rate": 0.4, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.75, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1123}, "5": {"facts_total": 65, "facts_retained": 26, "retention_rate": 0.4, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.75, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1123}, "6": {"facts_total": 65, "facts_retained": 26, "retention_rate": 0.4, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.75, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1123}, "7": {"facts_total": 65, "facts_retained": 26, "retention_rate": 0.4, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.75, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1123}, "8": {"facts_total": 65, "facts_retained": 26, "retention_rate": 0.4, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.75, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1123}}}, "reference_key_id": "keys/synthetic/syn_03009_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "3bd2611f7bc59d6f4b6323c3ac1e9f7af93bcc7ed3e43397b183cdb4d6c6e058", "output_hash": "bea99610d1e3e75a7f20939faf8e6824e02237b156dacd6442ba2a7e02617510"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "94acee5199c1fe0850b3ee3b", "attempt_id": "e4ff8dd52d70", "run_group_id": "E4_core:A4|repairable|qwen3b:h8", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:21:53.552138Z", "task_family": "synthetic", "document_id": "syn_03009_2000", "document_hash": "6ed94dcf728255bd7a7d11c97c362930728fb81a3e91f1906eb6eec86e0fedb7", "question_id": "q09_revision", "split": "test", "cluster_id": "syn_03009_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 8, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1976, "query_tokens": 605, "message_tokens": 512, "message_utf8_bytes": 1123, "cache_tensor_bytes": null, "serialized_payload_bytes": 1123, "side_information_bytes": 0, "fetched_bytes": 5131, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 177180.95974787138, "hops": {"1": 22990.705040981993, "2": 22287.307874998078, "3": 22002.059874939732, "4": 21829.51866602525, "5": 21880.574999959208, "6": 21869.980000075884, "7": 22178.420749958605, "8": 22142.392540932633}, "receiver_prefill_ms": 365.7003330299631, "ttft_ms": 42.81174996867776, "decode_ms": 545.6228329567239}, "total_wall_ms": 178135.7873728266, "ttft_ms": 42.81174996867776, "input_usage": {"lookups": 28}, "output_usage": {"output_tokens": 8}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "10047105", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 26, "retention_rate": 0.4, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.75, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1123}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 26, "retention_rate": 0.4, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.75, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1123}, "2": {"facts_total": 65, "facts_retained": 26, "retention_rate": 0.4, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.75, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1123}, "3": {"facts_total": 65, "facts_retained": 26, "retention_rate": 0.4, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.75, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1123}, "4": {"facts_total": 65, "facts_retained": 26, "retention_rate": 0.4, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.75, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1123}, "5": {"facts_total": 65, "facts_retained": 26, "retention_rate": 0.4, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.75, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1123}, "6": {"facts_total": 65, "facts_retained": 26, "retention_rate": 0.4, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.75, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1123}, "7": {"facts_total": 65, "facts_retained": 26, "retention_rate": 0.4, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.75, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1123}, "8": {"facts_total": 65, "facts_retained": 26, "retention_rate": 0.4, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.75, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1123}}}, "reference_key_id": "keys/synthetic/syn_03009_2000.json", "task_scores": {"correct": 0.0, "abstained": 0.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "3bd2611f7bc59d6f4b6323c3ac1e9f7af93bcc7ed3e43397b183cdb4d6c6e058", "output_hash": "bea99610d1e3e75a7f20939faf8e6824e02237b156dacd6442ba2a7e02617510"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "4df89be45a4e33968bb0439e", "attempt_id": "634bace669e8", "run_group_id": "E4_core:A4|repairable|qwen3b:h8", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:21:54.524265Z", "task_family": "synthetic", "document_id": "syn_03009_2000", "document_hash": "6ed94dcf728255bd7a7d11c97c362930728fb81a3e91f1906eb6eec86e0fedb7", "question_id": "q10_temporal", "split": "test", "cluster_id": "syn_03009_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 8, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1976, "query_tokens": 614, "message_tokens": 512, "message_utf8_bytes": 1123, "cache_tensor_bytes": null, "serialized_payload_bytes": 1123, "side_information_bytes": 0, "fetched_bytes": 5131, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 177180.95974787138, "hops": {"1": 22990.705040981993, "2": 22287.307874998078, "3": 22002.059874939732, "4": 21829.51866602525, "5": 21880.574999959208, "6": 21869.980000075884, "7": 22178.420749958605, "8": 22142.392540932633}, "receiver_prefill_ms": 399.9282500008121, "ttft_ms": 22.566166007891297, "decode_ms": 41.82604199741036}, "total_wall_ms": 177645.89553978294, "ttft_ms": 22.566166007891297, "input_usage": {"lookups": 28}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 26, "retention_rate": 0.4, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.75, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1123}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 26, "retention_rate": 0.4, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.75, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1123}, "2": {"facts_total": 65, "facts_retained": 26, "retention_rate": 0.4, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.75, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1123}, "3": {"facts_total": 65, "facts_retained": 26, "retention_rate": 0.4, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.75, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1123}, "4": {"facts_total": 65, "facts_retained": 26, "retention_rate": 0.4, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.75, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1123}, "5": {"facts_total": 65, "facts_retained": 26, "retention_rate": 0.4, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.75, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1123}, "6": {"facts_total": 65, "facts_retained": 26, "retention_rate": 0.4, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.75, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1123}, "7": {"facts_total": 65, "facts_retained": 26, "retention_rate": 0.4, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.75, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1123}, "8": {"facts_total": 65, "facts_retained": 26, "retention_rate": 0.4, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.75, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1123}}}, "reference_key_id": "keys/synthetic/syn_03009_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "3bd2611f7bc59d6f4b6323c3ac1e9f7af93bcc7ed3e43397b183cdb4d6c6e058", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "a7a999ce7c1c64c6835c337f", "attempt_id": "cfcec449db83", "run_group_id": "E4_core:A4|repairable|qwen3b:h8", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:21:55.006501Z", "task_family": "synthetic", "document_id": "syn_03009_2000", "document_hash": "6ed94dcf728255bd7a7d11c97c362930728fb81a3e91f1906eb6eec86e0fedb7", "question_id": "q11_aggregation", "split": "test", "cluster_id": "syn_03009_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 8, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1976, "query_tokens": 605, "message_tokens": 512, "message_utf8_bytes": 1123, "cache_tensor_bytes": null, "serialized_payload_bytes": 1123, "side_information_bytes": 0, "fetched_bytes": 5131, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 177180.95974787138, "hops": {"1": 22990.705040981993, "2": 22287.307874998078, "3": 22002.059874939732, "4": 21829.51866602525, "5": 21880.574999959208, "6": 21869.980000075884, "7": 22178.420749958605, "8": 22142.392540932633}, "receiver_prefill_ms": 416.27299995161593, "ttft_ms": 23.47658295184374, "decode_ms": 309.1564999194816}, "total_wall_ms": 177930.6030808948, "ttft_ms": 23.47658295184374, "input_usage": {"lookups": 28}, "output_usage": {"output_tokens": 7}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "5781336", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 26, "retention_rate": 0.4, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.75, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1123}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 26, "retention_rate": 0.4, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.75, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1123}, "2": {"facts_total": 65, "facts_retained": 26, "retention_rate": 0.4, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.75, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1123}, "3": {"facts_total": 65, "facts_retained": 26, "retention_rate": 0.4, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.75, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1123}, "4": {"facts_total": 65, "facts_retained": 26, "retention_rate": 0.4, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.75, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1123}, "5": {"facts_total": 65, "facts_retained": 26, "retention_rate": 0.4, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.75, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1123}, "6": {"facts_total": 65, "facts_retained": 26, "retention_rate": 0.4, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.75, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1123}, "7": {"facts_total": 65, "facts_retained": 26, "retention_rate": 0.4, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.75, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1123}, "8": {"facts_total": 65, "facts_retained": 26, "retention_rate": 0.4, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.75, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1123}}}, "reference_key_id": "keys/synthetic/syn_03009_2000.json", "task_scores": {"correct": 0.0, "abstained": 0.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "3bd2611f7bc59d6f4b6323c3ac1e9f7af93bcc7ed3e43397b183cdb4d6c6e058", "output_hash": "a365e04f74ba5baeccdd9e30dc27a1c5ff916597b828d874e63884b2090a60e2"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "da84009d9597f94854d4af5d", "attempt_id": "1c9af6758fc3", "run_group_id": "E4_core:A4|repairable|qwen3b:h8", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:21:55.774271Z", "task_family": "synthetic", "document_id": "syn_03009_2000", "document_hash": "6ed94dcf728255bd7a7d11c97c362930728fb81a3e91f1906eb6eec86e0fedb7", "question_id": "q12_unit_conversion", "split": "test", "cluster_id": "syn_03009_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 8, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1976, "query_tokens": 603, "message_tokens": 512, "message_utf8_bytes": 1123, "cache_tensor_bytes": null, "serialized_payload_bytes": 1123, "side_information_bytes": 0, "fetched_bytes": 5131, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 177180.95974787138, "hops": {"1": 22990.705040981993, "2": 22287.307874998078, "3": 22002.059874939732, "4": 21829.51866602525, "5": 21880.574999959208, "6": 21869.980000075884, "7": 22178.420749958605, "8": 22142.392540932633}, "receiver_prefill_ms": 508.1551659386605, "ttft_ms": 23.412209004163742, "decode_ms": 42.660500039346516}, "total_wall_ms": 177755.90020697564, "ttft_ms": 23.412209004163742, "input_usage": {"lookups": 28}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 26, "retention_rate": 0.4, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.75, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1123}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 26, "retention_rate": 0.4, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.75, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1123}, "2": {"facts_total": 65, "facts_retained": 26, "retention_rate": 0.4, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.75, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1123}, "3": {"facts_total": 65, "facts_retained": 26, "retention_rate": 0.4, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.75, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1123}, "4": {"facts_total": 65, "facts_retained": 26, "retention_rate": 0.4, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.75, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1123}, "5": {"facts_total": 65, "facts_retained": 26, "retention_rate": 0.4, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.75, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1123}, "6": {"facts_total": 65, "facts_retained": 26, "retention_rate": 0.4, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.75, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1123}, "7": {"facts_total": 65, "facts_retained": 26, "retention_rate": 0.4, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.75, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1123}, "8": {"facts_total": 65, "facts_retained": 26, "retention_rate": 0.4, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.75, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1123}}}, "reference_key_id": "keys/synthetic/syn_03009_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "3bd2611f7bc59d6f4b6323c3ac1e9f7af93bcc7ed3e43397b183cdb4d6c6e058", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "682a4daea88e7bde59ce7445", "attempt_id": "0921561f3cc7", "run_group_id": "E4_core:A4|repairable|qwen3b:h8", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:21:56.367337Z", "task_family": "synthetic", "document_id": "syn_03009_2000", "document_hash": "6ed94dcf728255bd7a7d11c97c362930728fb81a3e91f1906eb6eec86e0fedb7", "question_id": "q13_unanswerable", "split": "test", "cluster_id": "syn_03009_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 8, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1976, "query_tokens": 607, "message_tokens": 512, "message_utf8_bytes": 1123, "cache_tensor_bytes": null, "serialized_payload_bytes": 1123, "side_information_bytes": 0, "fetched_bytes": 5131, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 177180.95974787138, "hops": {"1": 22990.705040981993, "2": 22287.307874998078, "3": 22002.059874939732, "4": 21829.51866602525, "5": 21880.574999959208, "6": 21869.980000075884, "7": 22178.420749958605, "8": 22142.392540932633}, "receiver_prefill_ms": 406.4974579960108, "ttft_ms": 30.604875064454973, "decode_ms": 37.5565419672057}, "total_wall_ms": 177656.1875808984, "ttft_ms": 30.604875064454973, "input_usage": {"lookups": 28}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 26, "retention_rate": 0.4, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.75, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1123}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 26, "retention_rate": 0.4, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.75, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1123}, "2": {"facts_total": 65, "facts_retained": 26, "retention_rate": 0.4, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.75, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1123}, "3": {"facts_total": 65, "facts_retained": 26, "retention_rate": 0.4, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.75, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1123}, "4": {"facts_total": 65, "facts_retained": 26, "retention_rate": 0.4, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.75, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1123}, "5": {"facts_total": 65, "facts_retained": 26, "retention_rate": 0.4, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.75, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1123}, "6": {"facts_total": 65, "facts_retained": 26, "retention_rate": 0.4, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.75, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1123}, "7": {"facts_total": 65, "facts_retained": 26, "retention_rate": 0.4, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.75, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1123}, "8": {"facts_total": 65, "facts_retained": 26, "retention_rate": 0.4, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.75, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1123}}}, "reference_key_id": "keys/synthetic/syn_03009_2000.json", "task_scores": {"correct": 1.0, "abstained": 1.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "3bd2611f7bc59d6f4b6323c3ac1e9f7af93bcc7ed3e43397b183cdb4d6c6e058", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "0f1ec0ee5490069416647cbd", "attempt_id": "3da90688d1a7", "run_group_id": "E4_core:A4|repairable|qwen3b:h8", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:21:56.860067Z", "task_family": "synthetic", "document_id": "syn_03009_2000", "document_hash": "6ed94dcf728255bd7a7d11c97c362930728fb81a3e91f1906eb6eec86e0fedb7", "question_id": "q14_unanswerable", "split": "test", "cluster_id": "syn_03009_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 8, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1976, "query_tokens": 607, "message_tokens": 512, "message_utf8_bytes": 1123, "cache_tensor_bytes": null, "serialized_payload_bytes": 1123, "side_information_bytes": 0, "fetched_bytes": 5131, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 177180.95974787138, "hops": {"1": 22990.705040981993, "2": 22287.307874998078, "3": 22002.059874939732, "4": 21829.51866602525, "5": 21880.574999959208, "6": 21869.980000075884, "7": 22178.420749958605, "8": 22142.392540932633}, "receiver_prefill_ms": 369.69649989623576, "ttft_ms": 42.11354092694819, "decode_ms": 65.72929106187075}, "total_wall_ms": 177659.20228883624, "ttft_ms": 42.11354092694819, "input_usage": {"lookups": 28}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 26, "retention_rate": 0.4, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.75, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1123}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 26, "retention_rate": 0.4, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.75, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1123}, "2": {"facts_total": 65, "facts_retained": 26, "retention_rate": 0.4, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.75, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1123}, "3": {"facts_total": 65, "facts_retained": 26, "retention_rate": 0.4, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.75, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1123}, "4": {"facts_total": 65, "facts_retained": 26, "retention_rate": 0.4, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.75, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1123}, "5": {"facts_total": 65, "facts_retained": 26, "retention_rate": 0.4, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.75, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1123}, "6": {"facts_total": 65, "facts_retained": 26, "retention_rate": 0.4, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.75, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1123}, "7": {"facts_total": 65, "facts_retained": 26, "retention_rate": 0.4, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.75, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1123}, "8": {"facts_total": 65, "facts_retained": 26, "retention_rate": 0.4, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.75, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1123}}}, "reference_key_id": "keys/synthetic/syn_03009_2000.json", "task_scores": {"correct": 1.0, "abstained": 1.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "3bd2611f7bc59d6f4b6323c3ac1e9f7af93bcc7ed3e43397b183cdb4d6c6e058", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "7afe183ef0a093e8f8f60153", "attempt_id": "545a38d90762", "run_group_id": "E4_core:A5|repairable|qwen3b:h1", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:22:13.959703Z", "task_family": "synthetic", "document_id": "syn_03009_2000", "document_hash": "6ed94dcf728255bd7a7d11c97c362930728fb81a3e91f1906eb6eec86e0fedb7", "question_id": "q00_single_fact", "split": "test", "cluster_id": "syn_03009_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 1, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1976, "query_tokens": 318, "message_tokens": 234, "message_utf8_bytes": 698, "cache_tensor_bytes": null, "serialized_payload_bytes": 698, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 16602.87162498571, "hops": {"1": 16602.87162498571}, "receiver_prefill_ms": 240.29529106337577, "ttft_ms": 61.09700002707541, "decode_ms": 385.70720795542}, "total_wall_ms": 17290.717749972828, "ttft_ms": 61.09700002707541, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 5}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "Milos Lindqvist", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 12, "retention_rate": 0.18461538461538463, "retention_by_kind": {"person": 1.0, "date": 0.0, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 698}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 12, "retention_rate": 0.18461538461538463, "retention_by_kind": {"person": 1.0, "date": 0.0, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 698}}}, "reference_key_id": "keys/synthetic/syn_03009_2000.json", "task_scores": {"correct": 0.0, "abstained": 0.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "b7aa2d9d5f9f3a06115e4e65c8ffaf4be2a07d6d067ceac89e49c6efeb6fdd5f", "output_hash": "f1a0ce3d66245c336fdc7ac1281d2b638395739df9d016b121be6f1b2512201e"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "7bb1c56c2940bca9b730bafc", "attempt_id": "dff0386889ad", "run_group_id": "E4_core:A5|repairable|qwen3b:h1", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:22:14.666557Z", "task_family": "synthetic", "document_id": "syn_03009_2000", "document_hash": "6ed94dcf728255bd7a7d11c97c362930728fb81a3e91f1906eb6eec86e0fedb7", "question_id": "q01_single_fact", "split": "test", "cluster_id": "syn_03009_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 1, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1976, "query_tokens": 323, "message_tokens": 234, "message_utf8_bytes": 698, "cache_tensor_bytes": null, "serialized_payload_bytes": 698, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 16602.87162498571, "hops": {"1": 16602.87162498571}, "receiver_prefill_ms": 286.9436669861898, "ttft_ms": 42.430417030118406, "decode_ms": 24.629499996080995}, "total_wall_ms": 16957.537834066898, "ttft_ms": 42.430417030118406, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 12, "retention_rate": 0.18461538461538463, "retention_by_kind": {"person": 1.0, "date": 0.0, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 698}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 12, "retention_rate": 0.18461538461538463, "retention_by_kind": {"person": 1.0, "date": 0.0, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 698}}}, "reference_key_id": "keys/synthetic/syn_03009_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "b7aa2d9d5f9f3a06115e4e65c8ffaf4be2a07d6d067ceac89e49c6efeb6fdd5f", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "1ecb480c67c00281bcbf4beb", "attempt_id": "f197a5dcf1cc", "run_group_id": "E4_core:A5|repairable|qwen3b:h1", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:22:15.038828Z", "task_family": "synthetic", "document_id": "syn_03009_2000", "document_hash": "6ed94dcf728255bd7a7d11c97c362930728fb81a3e91f1906eb6eec86e0fedb7", "question_id": "q02_exact_copy", "split": "test", "cluster_id": "syn_03009_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 1, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1976, "query_tokens": 321, "message_tokens": 234, "message_utf8_bytes": 698, "cache_tensor_bytes": null, "serialized_payload_bytes": 698, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 16602.87162498571, "hops": {"1": 16602.87162498571}, "receiver_prefill_ms": 217.72341593168676, "ttft_ms": 37.6756249461323, "decode_ms": 23.426708066836}, "total_wall_ms": 16882.27091601584, "ttft_ms": 37.6756249461323, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 12, "retention_rate": 0.18461538461538463, "retention_by_kind": {"person": 1.0, "date": 0.0, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 698}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 12, "retention_rate": 0.18461538461538463, "retention_by_kind": {"person": 1.0, "date": 0.0, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 698}}}, "reference_key_id": "keys/synthetic/syn_03009_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "b7aa2d9d5f9f3a06115e4e65c8ffaf4be2a07d6d067ceac89e49c6efeb6fdd5f", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "9eee0b8240a24eec7ebf388b", "attempt_id": "cec9419408ee", "run_group_id": "E4_core:A5|repairable|qwen3b:h1", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:22:15.336682Z", "task_family": "synthetic", "document_id": "syn_03009_2000", "document_hash": "6ed94dcf728255bd7a7d11c97c362930728fb81a3e91f1906eb6eec86e0fedb7", "question_id": "q03_exact_copy", "split": "test", "cluster_id": "syn_03009_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 1, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1976, "query_tokens": 317, "message_tokens": 234, "message_utf8_bytes": 698, "cache_tensor_bytes": null, "serialized_payload_bytes": 698, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 16602.87162498571, "hops": {"1": 16602.87162498571}, "receiver_prefill_ms": 223.6813330091536, "ttft_ms": 37.18566696625203, "decode_ms": 279.4228340499103}, "total_wall_ms": 17144.030374940485, "ttft_ms": 37.18566696625203, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 5}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "Idris Bergström", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 12, "retention_rate": 0.18461538461538463, "retention_by_kind": {"person": 1.0, "date": 0.0, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 698}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 12, "retention_rate": 0.18461538461538463, "retention_by_kind": {"person": 1.0, "date": 0.0, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 698}}}, "reference_key_id": "keys/synthetic/syn_03009_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "b7aa2d9d5f9f3a06115e4e65c8ffaf4be2a07d6d067ceac89e49c6efeb6fdd5f", "output_hash": "fd88832c9fd0ab175818d76d875d70156225bd4b34ec401514546ef70f4f826c"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "a91482a1c20051c62a3ed4be", "attempt_id": "98264ccef60f", "run_group_id": "E4_core:A5|repairable|qwen3b:h1", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:22:15.895442Z", "task_family": "synthetic", "document_id": "syn_03009_2000", "document_hash": "6ed94dcf728255bd7a7d11c97c362930728fb81a3e91f1906eb6eec86e0fedb7", "question_id": "q04_multi_hop", "split": "test", "cluster_id": "syn_03009_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 1, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1976, "query_tokens": 325, "message_tokens": 234, "message_utf8_bytes": 698, "cache_tensor_bytes": null, "serialized_payload_bytes": 698, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 16602.87162498571, "hops": {"1": 16602.87162498571}, "receiver_prefill_ms": 385.135457967408, "ttft_ms": 41.06808302458376, "decode_ms": 25.32212494406849}, "total_wall_ms": 17054.859417025, "ttft_ms": 41.06808302458376, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 12, "retention_rate": 0.18461538461538463, "retention_by_kind": {"person": 1.0, "date": 0.0, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 698}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 12, "retention_rate": 0.18461538461538463, "retention_by_kind": {"person": 1.0, "date": 0.0, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 698}}}, "reference_key_id": "keys/synthetic/syn_03009_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "b7aa2d9d5f9f3a06115e4e65c8ffaf4be2a07d6d067ceac89e49c6efeb6fdd5f", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "ba98933193c31570829b6d54", "attempt_id": "8ef491ed9ca0", "run_group_id": "E4_core:A5|repairable|qwen3b:h1", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:22:16.364410Z", "task_family": "synthetic", "document_id": "syn_03009_2000", "document_hash": "6ed94dcf728255bd7a7d11c97c362930728fb81a3e91f1906eb6eec86e0fedb7", "question_id": "q05_multi_hop", "split": "test", "cluster_id": "syn_03009_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 1, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1976, "query_tokens": 328, "message_tokens": 234, "message_utf8_bytes": 698, "cache_tensor_bytes": null, "serialized_payload_bytes": 698, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 16602.87162498571, "hops": {"1": 16602.87162498571}, "receiver_prefill_ms": 275.5231249611825, "ttft_ms": 46.38037504628301, "decode_ms": 24.29562492761761}, "total_wall_ms": 16949.474124936387, "ttft_ms": 46.38037504628301, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 12, "retention_rate": 0.18461538461538463, "retention_by_kind": {"person": 1.0, "date": 0.0, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 698}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 12, "retention_rate": 0.18461538461538463, "retention_by_kind": {"person": 1.0, "date": 0.0, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 698}}}, "reference_key_id": "keys/synthetic/syn_03009_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "b7aa2d9d5f9f3a06115e4e65c8ffaf4be2a07d6d067ceac89e49c6efeb6fdd5f", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "097ee3dd99ba23419194cd20", "attempt_id": "ae661bbc8ff9", "run_group_id": "E4_core:A5|repairable|qwen3b:h1", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:22:16.727989Z", "task_family": "synthetic", "document_id": "syn_03009_2000", "document_hash": "6ed94dcf728255bd7a7d11c97c362930728fb81a3e91f1906eb6eec86e0fedb7", "question_id": "q06_negation", "split": "test", "cluster_id": "syn_03009_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 1, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1976, "query_tokens": 318, "message_tokens": 234, "message_utf8_bytes": 698, "cache_tensor_bytes": null, "serialized_payload_bytes": 698, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 16602.87162498571, "hops": {"1": 16602.87162498571}, "receiver_prefill_ms": 238.41166601050645, "ttft_ms": 35.941417096182704, "decode_ms": 22.643208038061857}, "total_wall_ms": 16900.512999971397, "ttft_ms": 35.941417096182704, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "no", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 12, "retention_rate": 0.18461538461538463, "retention_by_kind": {"person": 1.0, "date": 0.0, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 698}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 12, "retention_rate": 0.18461538461538463, "retention_by_kind": {"person": 1.0, "date": 0.0, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 698}}}, "reference_key_id": "keys/synthetic/syn_03009_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "b7aa2d9d5f9f3a06115e4e65c8ffaf4be2a07d6d067ceac89e49c6efeb6fdd5f", "output_hash": "9390298f3fb0c5b160498935d79cb139aef28e1c47358b4bbba61862b9c26e59"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "be58a1e8444f5495fe974880", "attempt_id": "2ccb21269050", "run_group_id": "E4_core:A5|repairable|qwen3b:h1", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:22:17.043067Z", "task_family": "synthetic", "document_id": "syn_03009_2000", "document_hash": "6ed94dcf728255bd7a7d11c97c362930728fb81a3e91f1906eb6eec86e0fedb7", "question_id": "q07_negation", "split": "test", "cluster_id": "syn_03009_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 1, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1976, "query_tokens": 317, "message_tokens": 234, "message_utf8_bytes": 698, "cache_tensor_bytes": null, "serialized_payload_bytes": 698, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 16602.87162498571, "hops": {"1": 16602.87162498571}, "receiver_prefill_ms": 253.05883400142193, "ttft_ms": 61.91308400593698, "decode_ms": 52.750415983609855}, "total_wall_ms": 16971.13195899874, "ttft_ms": 61.91308400593698, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "no", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 12, "retention_rate": 0.18461538461538463, "retention_by_kind": {"person": 1.0, "date": 0.0, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 698}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 12, "retention_rate": 0.18461538461538463, "retention_by_kind": {"person": 1.0, "date": 0.0, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 698}}}, "reference_key_id": "keys/synthetic/syn_03009_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "b7aa2d9d5f9f3a06115e4e65c8ffaf4be2a07d6d067ceac89e49c6efeb6fdd5f", "output_hash": "9390298f3fb0c5b160498935d79cb139aef28e1c47358b4bbba61862b9c26e59"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "4d392595d89d5285f57e06c2", "attempt_id": "be02dfc3c6e9", "run_group_id": "E4_core:A5|repairable|qwen3b:h1", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:22:17.432445Z", "task_family": "synthetic", "document_id": "syn_03009_2000", "document_hash": "6ed94dcf728255bd7a7d11c97c362930728fb81a3e91f1906eb6eec86e0fedb7", "question_id": "q08_revision", "split": "test", "cluster_id": "syn_03009_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 1, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1976, "query_tokens": 335, "message_tokens": 234, "message_utf8_bytes": 698, "cache_tensor_bytes": null, "serialized_payload_bytes": 698, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 16602.87162498571, "hops": {"1": 16602.87162498571}, "receiver_prefill_ms": 398.0198339559138, "ttft_ms": 32.25254197604954, "decode_ms": 41.235749959014356}, "total_wall_ms": 17075.12637495529, "ttft_ms": 32.25254197604954, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 12, "retention_rate": 0.18461538461538463, "retention_by_kind": {"person": 1.0, "date": 0.0, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 698}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 12, "retention_rate": 0.18461538461538463, "retention_by_kind": {"person": 1.0, "date": 0.0, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 698}}}, "reference_key_id": "keys/synthetic/syn_03009_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "b7aa2d9d5f9f3a06115e4e65c8ffaf4be2a07d6d067ceac89e49c6efeb6fdd5f", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "0eb1460e9bd10dba633a64f4", "attempt_id": "c2c1a9c0ede9", "run_group_id": "E4_core:A5|repairable|qwen3b:h1", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:22:17.922767Z", "task_family": "synthetic", "document_id": "syn_03009_2000", "document_hash": "6ed94dcf728255bd7a7d11c97c362930728fb81a3e91f1906eb6eec86e0fedb7", "question_id": "q09_revision", "split": "test", "cluster_id": "syn_03009_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 1, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1976, "query_tokens": 328, "message_tokens": 234, "message_utf8_bytes": 698, "cache_tensor_bytes": null, "serialized_payload_bytes": 698, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 16602.87162498571, "hops": {"1": 16602.87162498571}, "receiver_prefill_ms": 254.11712494678795, "ttft_ms": 38.268832955509424, "decode_ms": 38.32995903212577}, "total_wall_ms": 16934.144166996703, "ttft_ms": 38.268832955509424, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 12, "retention_rate": 0.18461538461538463, "retention_by_kind": {"person": 1.0, "date": 0.0, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 698}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 12, "retention_rate": 0.18461538461538463, "retention_by_kind": {"person": 1.0, "date": 0.0, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 698}}}, "reference_key_id": "keys/synthetic/syn_03009_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "b7aa2d9d5f9f3a06115e4e65c8ffaf4be2a07d6d067ceac89e49c6efeb6fdd5f", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "afcd5d378910d71c615f03a8", "attempt_id": "1aee47502779", "run_group_id": "E4_core:A5|repairable|qwen3b:h1", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:22:18.272150Z", "task_family": "synthetic", "document_id": "syn_03009_2000", "document_hash": "6ed94dcf728255bd7a7d11c97c362930728fb81a3e91f1906eb6eec86e0fedb7", "question_id": "q10_temporal", "split": "test", "cluster_id": "syn_03009_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 1, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1976, "query_tokens": 337, "message_tokens": 234, "message_utf8_bytes": 698, "cache_tensor_bytes": null, "serialized_payload_bytes": 698, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 16602.87162498571, "hops": {"1": 16602.87162498571}, "receiver_prefill_ms": 270.97804204095155, "ttft_ms": 32.84729097504169, "decode_ms": 318.7008749227971}, "total_wall_ms": 17225.86524998769, "ttft_ms": 32.84729097504169, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 7}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "Eshgamgam-11", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 12, "retention_rate": 0.18461538461538463, "retention_by_kind": {"person": 1.0, "date": 0.0, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 698}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 12, "retention_rate": 0.18461538461538463, "retention_by_kind": {"person": 1.0, "date": 0.0, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 698}}}, "reference_key_id": "keys/synthetic/syn_03009_2000.json", "task_scores": {"correct": 0.0, "abstained": 0.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "b7aa2d9d5f9f3a06115e4e65c8ffaf4be2a07d6d067ceac89e49c6efeb6fdd5f", "output_hash": "8c1313c53ab4dbffa5f336d721c810678133eecaa467241e53d8a92150601475"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "c682aca12fe3951c287a19a1", "attempt_id": "6fc857595210", "run_group_id": "E4_core:A5|repairable|qwen3b:h1", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:22:18.913194Z", "task_family": "synthetic", "document_id": "syn_03009_2000", "document_hash": "6ed94dcf728255bd7a7d11c97c362930728fb81a3e91f1906eb6eec86e0fedb7", "question_id": "q11_aggregation", "split": "test", "cluster_id": "syn_03009_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 1, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1976, "query_tokens": 328, "message_tokens": 234, "message_utf8_bytes": 698, "cache_tensor_bytes": null, "serialized_payload_bytes": 698, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 16602.87162498571, "hops": {"1": 16602.87162498571}, "receiver_prefill_ms": 398.73895898927003, "ttft_ms": 39.54795899335295, "decode_ms": 30.210625031031668}, "total_wall_ms": 17071.83662499301, "ttft_ms": 39.54795899335295, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 12, "retention_rate": 0.18461538461538463, "retention_by_kind": {"person": 1.0, "date": 0.0, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 698}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 12, "retention_rate": 0.18461538461538463, "retention_by_kind": {"person": 1.0, "date": 0.0, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 698}}}, "reference_key_id": "keys/synthetic/syn_03009_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "b7aa2d9d5f9f3a06115e4e65c8ffaf4be2a07d6d067ceac89e49c6efeb6fdd5f", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "b1a0521048c99ef5c9f88bd5", "attempt_id": "8746bbdb2957", "run_group_id": "E4_core:A5|repairable|qwen3b:h1", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:22:19.400842Z", "task_family": "synthetic", "document_id": "syn_03009_2000", "document_hash": "6ed94dcf728255bd7a7d11c97c362930728fb81a3e91f1906eb6eec86e0fedb7", "question_id": "q12_unit_conversion", "split": "test", "cluster_id": "syn_03009_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 1, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1976, "query_tokens": 326, "message_tokens": 234, "message_utf8_bytes": 698, "cache_tensor_bytes": null, "serialized_payload_bytes": 698, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 16602.87162498571, "hops": {"1": 16602.87162498571}, "receiver_prefill_ms": 239.92658394854516, "ttft_ms": 41.33179096970707, "decode_ms": 29.61474994663149}, "total_wall_ms": 16914.328332873993, "ttft_ms": 41.33179096970707, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 12, "retention_rate": 0.18461538461538463, "retention_by_kind": {"person": 1.0, "date": 0.0, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 698}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 12, "retention_rate": 0.18461538461538463, "retention_by_kind": {"person": 1.0, "date": 0.0, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 698}}}, "reference_key_id": "keys/synthetic/syn_03009_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "b7aa2d9d5f9f3a06115e4e65c8ffaf4be2a07d6d067ceac89e49c6efeb6fdd5f", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "a56cc348fc1102daa3f2a626", "attempt_id": "98172547d2f7", "run_group_id": "E4_core:A5|repairable|qwen3b:h1", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:22:19.729300Z", "task_family": "synthetic", "document_id": "syn_03009_2000", "document_hash": "6ed94dcf728255bd7a7d11c97c362930728fb81a3e91f1906eb6eec86e0fedb7", "question_id": "q13_unanswerable", "split": "test", "cluster_id": "syn_03009_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 1, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1976, "query_tokens": 330, "message_tokens": 234, "message_utf8_bytes": 698, "cache_tensor_bytes": null, "serialized_payload_bytes": 698, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 16602.87162498571, "hops": {"1": 16602.87162498571}, "receiver_prefill_ms": 256.60524994600564, "ttft_ms": 21.917790989391506, "decode_ms": 40.313708945177495}, "total_wall_ms": 16922.287499997765, "ttft_ms": 21.917790989391506, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 12, "retention_rate": 0.18461538461538463, "retention_by_kind": {"person": 1.0, "date": 0.0, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 698}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 12, "retention_rate": 0.18461538461538463, "retention_by_kind": {"person": 1.0, "date": 0.0, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 698}}}, "reference_key_id": "keys/synthetic/syn_03009_2000.json", "task_scores": {"correct": 1.0, "abstained": 1.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "b7aa2d9d5f9f3a06115e4e65c8ffaf4be2a07d6d067ceac89e49c6efeb6fdd5f", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "f52c08d50d2601b45e3d15ae", "attempt_id": "bd2e1cc2576c", "run_group_id": "E4_core:A5|repairable|qwen3b:h1", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:22:20.065575Z", "task_family": "synthetic", "document_id": "syn_03009_2000", "document_hash": "6ed94dcf728255bd7a7d11c97c362930728fb81a3e91f1906eb6eec86e0fedb7", "question_id": "q14_unanswerable", "split": "test", "cluster_id": "syn_03009_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 1, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1976, "query_tokens": 330, "message_tokens": 234, "message_utf8_bytes": 698, "cache_tensor_bytes": null, "serialized_payload_bytes": 698, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 16602.87162498571, "hops": {"1": 16602.87162498571}, "receiver_prefill_ms": 239.36208302620798, "ttft_ms": 72.03754188958555, "decode_ms": 46.88049992546439}, "total_wall_ms": 16961.658708984032, "ttft_ms": 72.03754188958555, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 12, "retention_rate": 0.18461538461538463, "retention_by_kind": {"person": 1.0, "date": 0.0, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 698}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 12, "retention_rate": 0.18461538461538463, "retention_by_kind": {"person": 1.0, "date": 0.0, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 698}}}, "reference_key_id": "keys/synthetic/syn_03009_2000.json", "task_scores": {"correct": 1.0, "abstained": 1.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "b7aa2d9d5f9f3a06115e4e65c8ffaf4be2a07d6d067ceac89e49c6efeb6fdd5f", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "2c55e1dc8d58a9adbbe974d1", "attempt_id": "ef429974919b", "run_group_id": "E4_core:A5|repairable|qwen3b:h2", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:22:25.121301Z", "task_family": "synthetic", "document_id": "syn_03009_2000", "document_hash": "6ed94dcf728255bd7a7d11c97c362930728fb81a3e91f1906eb6eec86e0fedb7", "question_id": "q00_single_fact", "split": "test", "cluster_id": "syn_03009_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 2, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1976, "query_tokens": 177, "message_tokens": 94, "message_utf8_bytes": 310, "cache_tensor_bytes": null, "serialized_payload_bytes": 310, "side_information_bytes": 0, "fetched_bytes": 1080, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 21280.66345793195, "hops": {"1": 16602.87162498571, "2": 4677.791832946241}, "receiver_prefill_ms": 189.26587502937764, "ttft_ms": 41.05274996254593, "decode_ms": 23.541250033304095}, "total_wall_ms": 21534.90929189138, "ttft_ms": 41.05274996254593, "input_usage": {"lookups": 2}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 8, "retention_rate": 0.12307692307692308, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.0, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 310}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 12, "retention_rate": 0.18461538461538463, "retention_by_kind": {"person": 1.0, "date": 0.0, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 698}, "2": {"facts_total": 65, "facts_retained": 8, "retention_rate": 0.12307692307692308, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.0, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 310}}}, "reference_key_id": "keys/synthetic/syn_03009_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "f58a7dc5658fb58db0a87934226394775c9d870973cde3e211186d96f2a1306d", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "b8391afdb6d1a2f127bfe7f5", "attempt_id": "07450931448a", "run_group_id": "E4_core:A5|repairable|qwen3b:h2", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:22:25.393322Z", "task_family": "synthetic", "document_id": "syn_03009_2000", "document_hash": "6ed94dcf728255bd7a7d11c97c362930728fb81a3e91f1906eb6eec86e0fedb7", "question_id": "q01_single_fact", "split": "test", "cluster_id": "syn_03009_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 2, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1976, "query_tokens": 182, "message_tokens": 94, "message_utf8_bytes": 310, "cache_tensor_bytes": null, "serialized_payload_bytes": 310, "side_information_bytes": 0, "fetched_bytes": 1080, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 21280.66345793195, "hops": {"1": 16602.87162498571, "2": 4677.791832946241}, "receiver_prefill_ms": 183.47362498752773, "ttft_ms": 39.08112505450845, "decode_ms": 25.014457991346717}, "total_wall_ms": 21528.745999909006, "ttft_ms": 39.08112505450845, "input_usage": {"lookups": 2}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 8, "retention_rate": 0.12307692307692308, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.0, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 310}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 12, "retention_rate": 0.18461538461538463, "retention_by_kind": {"person": 1.0, "date": 0.0, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 698}, "2": {"facts_total": 65, "facts_retained": 8, "retention_rate": 0.12307692307692308, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.0, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 310}}}, "reference_key_id": "keys/synthetic/syn_03009_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "f58a7dc5658fb58db0a87934226394775c9d870973cde3e211186d96f2a1306d", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "61dfeb8581260c29385345c5", "attempt_id": "623c669f41eb", "run_group_id": "E4_core:A5|repairable|qwen3b:h2", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:22:25.660128Z", "task_family": "synthetic", "document_id": "syn_03009_2000", "document_hash": "6ed94dcf728255bd7a7d11c97c362930728fb81a3e91f1906eb6eec86e0fedb7", "question_id": "q02_exact_copy", "split": "test", "cluster_id": "syn_03009_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 2, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1976, "query_tokens": 180, "message_tokens": 94, "message_utf8_bytes": 310, "cache_tensor_bytes": null, "serialized_payload_bytes": 310, "side_information_bytes": 0, "fetched_bytes": 1080, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 21280.66345793195, "hops": {"1": 16602.87162498571, "2": 4677.791832946241}, "receiver_prefill_ms": 193.6656249454245, "ttft_ms": 20.112041966058314, "decode_ms": 36.00695892237127}, "total_wall_ms": 21530.852249939926, "ttft_ms": 20.112041966058314, "input_usage": {"lookups": 2}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 8, "retention_rate": 0.12307692307692308, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.0, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 310}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 12, "retention_rate": 0.18461538461538463, "retention_by_kind": {"person": 1.0, "date": 0.0, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 698}, "2": {"facts_total": 65, "facts_retained": 8, "retention_rate": 0.12307692307692308, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.0, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 310}}}, "reference_key_id": "keys/synthetic/syn_03009_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "f58a7dc5658fb58db0a87934226394775c9d870973cde3e211186d96f2a1306d", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "bd72d9b5e812754d7963c5b7", "attempt_id": "39bb9393ce69", "run_group_id": "E4_core:A5|repairable|qwen3b:h2", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:22:25.928633Z", "task_family": "synthetic", "document_id": "syn_03009_2000", "document_hash": "6ed94dcf728255bd7a7d11c97c362930728fb81a3e91f1906eb6eec86e0fedb7", "question_id": "q03_exact_copy", "split": "test", "cluster_id": "syn_03009_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 2, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1976, "query_tokens": 176, "message_tokens": 94, "message_utf8_bytes": 310, "cache_tensor_bytes": null, "serialized_payload_bytes": 310, "side_information_bytes": 0, "fetched_bytes": 1080, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 21280.66345793195, "hops": {"1": 16602.87162498571, "2": 4677.791832946241}, "receiver_prefill_ms": 195.9594580112025, "ttft_ms": 41.4519589394331, "decode_ms": 379.26737498492}, "total_wall_ms": 21897.812748909928, "ttft_ms": 41.4519589394331, "input_usage": {"lookups": 2}, "output_usage": {"output_tokens": 5}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "Idris Bergström", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 8, "retention_rate": 0.12307692307692308, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.0, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 310}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 12, "retention_rate": 0.18461538461538463, "retention_by_kind": {"person": 1.0, "date": 0.0, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 698}, "2": {"facts_total": 65, "facts_retained": 8, "retention_rate": 0.12307692307692308, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.0, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 310}}}, "reference_key_id": "keys/synthetic/syn_03009_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "f58a7dc5658fb58db0a87934226394775c9d870973cde3e211186d96f2a1306d", "output_hash": "fd88832c9fd0ab175818d76d875d70156225bd4b34ec401514546ef70f4f826c"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "e9461e628a5f1835345e4beb", "attempt_id": "1eaf0255e03c", "run_group_id": "E4_core:A5|repairable|qwen3b:h2", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:22:26.565140Z", "task_family": "synthetic", "document_id": "syn_03009_2000", "document_hash": "6ed94dcf728255bd7a7d11c97c362930728fb81a3e91f1906eb6eec86e0fedb7", "question_id": "q04_multi_hop", "split": "test", "cluster_id": "syn_03009_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 2, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1976, "query_tokens": 184, "message_tokens": 94, "message_utf8_bytes": 310, "cache_tensor_bytes": null, "serialized_payload_bytes": 310, "side_information_bytes": 0, "fetched_bytes": 1080, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 21280.66345793195, "hops": {"1": 16602.87162498571, "2": 4677.791832946241}, "receiver_prefill_ms": 217.31025003828108, "ttft_ms": 21.81129192467779, "decode_ms": 40.320207946933806}, "total_wall_ms": 21560.455375001766, "ttft_ms": 21.81129192467779, "input_usage": {"lookups": 2}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 8, "retention_rate": 0.12307692307692308, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.0, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 310}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 12, "retention_rate": 0.18461538461538463, "retention_by_kind": {"person": 1.0, "date": 0.0, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 698}, "2": {"facts_total": 65, "facts_retained": 8, "retention_rate": 0.12307692307692308, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.0, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 310}}}, "reference_key_id": "keys/synthetic/syn_03009_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "f58a7dc5658fb58db0a87934226394775c9d870973cde3e211186d96f2a1306d", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "3464d11c3f81693ea1d988c1", "attempt_id": "045d4c985e63", "run_group_id": "E4_core:A5|repairable|qwen3b:h2", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:22:26.862387Z", "task_family": "synthetic", "document_id": "syn_03009_2000", "document_hash": "6ed94dcf728255bd7a7d11c97c362930728fb81a3e91f1906eb6eec86e0fedb7", "question_id": "q05_multi_hop", "split": "test", "cluster_id": "syn_03009_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 2, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1976, "query_tokens": 187, "message_tokens": 94, "message_utf8_bytes": 310, "cache_tensor_bytes": null, "serialized_payload_bytes": 310, "side_information_bytes": 0, "fetched_bytes": 1080, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 21280.66345793195, "hops": {"1": 16602.87162498571, "2": 4677.791832946241}, "receiver_prefill_ms": 194.10083291586488, "ttft_ms": 39.40025006886572, "decode_ms": 24.022917030379176}, "total_wall_ms": 21538.50199887529, "ttft_ms": 39.40025006886572, "input_usage": {"lookups": 2}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 8, "retention_rate": 0.12307692307692308, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.0, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 310}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 12, "retention_rate": 0.18461538461538463, "retention_by_kind": {"person": 1.0, "date": 0.0, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 698}, "2": {"facts_total": 65, "facts_retained": 8, "retention_rate": 0.12307692307692308, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.0, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 310}}}, "reference_key_id": "keys/synthetic/syn_03009_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "f58a7dc5658fb58db0a87934226394775c9d870973cde3e211186d96f2a1306d", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "a6455494cc2d8075f682f464", "attempt_id": "75c3273c032f", "run_group_id": "E4_core:A5|repairable|qwen3b:h2", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:22:27.136640Z", "task_family": "synthetic", "document_id": "syn_03009_2000", "document_hash": "6ed94dcf728255bd7a7d11c97c362930728fb81a3e91f1906eb6eec86e0fedb7", "question_id": "q06_negation", "split": "test", "cluster_id": "syn_03009_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 2, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1976, "query_tokens": 177, "message_tokens": 94, "message_utf8_bytes": 310, "cache_tensor_bytes": null, "serialized_payload_bytes": 310, "side_information_bytes": 0, "fetched_bytes": 1080, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 21280.66345793195, "hops": {"1": 16602.87162498571, "2": 4677.791832946241}, "receiver_prefill_ms": 167.6949589746073, "ttft_ms": 35.968500073067844, "decode_ms": 64.90554194897413}, "total_wall_ms": 21549.613416893408, "ttft_ms": 35.968500073067844, "input_usage": {"lookups": 2}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 8, "retention_rate": 0.12307692307692308, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.0, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 310}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 12, "retention_rate": 0.18461538461538463, "retention_by_kind": {"person": 1.0, "date": 0.0, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 698}, "2": {"facts_total": 65, "facts_retained": 8, "retention_rate": 0.12307692307692308, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.0, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 310}}}, "reference_key_id": "keys/synthetic/syn_03009_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "f58a7dc5658fb58db0a87934226394775c9d870973cde3e211186d96f2a1306d", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "453ae0e781d5327fb563ae23", "attempt_id": "915c0cf8d7c4", "run_group_id": "E4_core:A5|repairable|qwen3b:h2", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:22:27.422780Z", "task_family": "synthetic", "document_id": "syn_03009_2000", "document_hash": "6ed94dcf728255bd7a7d11c97c362930728fb81a3e91f1906eb6eec86e0fedb7", "question_id": "q07_negation", "split": "test", "cluster_id": "syn_03009_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 2, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1976, "query_tokens": 176, "message_tokens": 94, "message_utf8_bytes": 310, "cache_tensor_bytes": null, "serialized_payload_bytes": 310, "side_information_bytes": 0, "fetched_bytes": 1080, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 21280.66345793195, "hops": {"1": 16602.87162498571, "2": 4677.791832946241}, "receiver_prefill_ms": 220.74875002726912, "ttft_ms": 105.50420801155269, "decode_ms": 144.09004105255008}, "total_wall_ms": 21751.322998898104, "ttft_ms": 105.50420801155269, "input_usage": {"lookups": 2}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 8, "retention_rate": 0.12307692307692308, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.0, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 310}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 12, "retention_rate": 0.18461538461538463, "retention_by_kind": {"person": 1.0, "date": 0.0, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 698}, "2": {"facts_total": 65, "facts_retained": 8, "retention_rate": 0.12307692307692308, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.0, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 310}}}, "reference_key_id": "keys/synthetic/syn_03009_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "f58a7dc5658fb58db0a87934226394775c9d870973cde3e211186d96f2a1306d", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "db295b39c9a38fa13a1d6315", "attempt_id": "31e916d9d30b", "run_group_id": "E4_core:A5|repairable|qwen3b:h2", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:22:27.911576Z", "task_family": "synthetic", "document_id": "syn_03009_2000", "document_hash": "6ed94dcf728255bd7a7d11c97c362930728fb81a3e91f1906eb6eec86e0fedb7", "question_id": "q08_revision", "split": "test", "cluster_id": "syn_03009_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 2, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1976, "query_tokens": 194, "message_tokens": 94, "message_utf8_bytes": 310, "cache_tensor_bytes": null, "serialized_payload_bytes": 310, "side_information_bytes": 0, "fetched_bytes": 1080, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 21280.66345793195, "hops": {"1": 16602.87162498571, "2": 4677.791832946241}, "receiver_prefill_ms": 193.75608302652836, "ttft_ms": 38.95429102703929, "decode_ms": 25.506375008262694}, "total_wall_ms": 21539.212915929966, "ttft_ms": 38.95429102703929, "input_usage": {"lookups": 2}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 8, "retention_rate": 0.12307692307692308, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.0, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 310}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 12, "retention_rate": 0.18461538461538463, "retention_by_kind": {"person": 1.0, "date": 0.0, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 698}, "2": {"facts_total": 65, "facts_retained": 8, "retention_rate": 0.12307692307692308, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.0, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 310}}}, "reference_key_id": "keys/synthetic/syn_03009_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "f58a7dc5658fb58db0a87934226394775c9d870973cde3e211186d96f2a1306d", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "86de51d1c637daf9e3349abd", "attempt_id": "7f0b251cd87f", "run_group_id": "E4_core:A5|repairable|qwen3b:h2", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:22:28.186961Z", "task_family": "synthetic", "document_id": "syn_03009_2000", "document_hash": "6ed94dcf728255bd7a7d11c97c362930728fb81a3e91f1906eb6eec86e0fedb7", "question_id": "q09_revision", "split": "test", "cluster_id": "syn_03009_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 2, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1976, "query_tokens": 187, "message_tokens": 94, "message_utf8_bytes": 310, "cache_tensor_bytes": null, "serialized_payload_bytes": 310, "side_information_bytes": 0, "fetched_bytes": 1080, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 21280.66345793195, "hops": {"1": 16602.87162498571, "2": 4677.791832946241}, "receiver_prefill_ms": 221.48845891933888, "ttft_ms": 21.55012497678399, "decode_ms": 41.302958969026804}, "total_wall_ms": 21565.377415856346, "ttft_ms": 21.55012497678399, "input_usage": {"lookups": 2}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 8, "retention_rate": 0.12307692307692308, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.0, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 310}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 12, "retention_rate": 0.18461538461538463, "retention_by_kind": {"person": 1.0, "date": 0.0, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 698}, "2": {"facts_total": 65, "facts_retained": 8, "retention_rate": 0.12307692307692308, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.0, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 310}}}, "reference_key_id": "keys/synthetic/syn_03009_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "f58a7dc5658fb58db0a87934226394775c9d870973cde3e211186d96f2a1306d", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "52d748ac70c31c631591d792", "attempt_id": "88b6d409a1d1", "run_group_id": "E4_core:A5|repairable|qwen3b:h2", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:22:28.489219Z", "task_family": "synthetic", "document_id": "syn_03009_2000", "document_hash": "6ed94dcf728255bd7a7d11c97c362930728fb81a3e91f1906eb6eec86e0fedb7", "question_id": "q10_temporal", "split": "test", "cluster_id": "syn_03009_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 2, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1976, "query_tokens": 196, "message_tokens": 94, "message_utf8_bytes": 310, "cache_tensor_bytes": null, "serialized_payload_bytes": 310, "side_information_bytes": 0, "fetched_bytes": 1080, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 21280.66345793195, "hops": {"1": 16602.87162498571, "2": 4677.791832946241}, "receiver_prefill_ms": 216.32079198025167, "ttft_ms": 21.34762506466359, "decode_ms": 25.37908300291747}, "total_wall_ms": 21544.18754100334, "ttft_ms": 21.34762506466359, "input_usage": {"lookups": 2}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 8, "retention_rate": 0.12307692307692308, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.0, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 310}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 12, "retention_rate": 0.18461538461538463, "retention_by_kind": {"person": 1.0, "date": 0.0, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 698}, "2": {"facts_total": 65, "facts_retained": 8, "retention_rate": 0.12307692307692308, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.0, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 310}}}, "reference_key_id": "keys/synthetic/syn_03009_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "f58a7dc5658fb58db0a87934226394775c9d870973cde3e211186d96f2a1306d", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "571aa2b8f414501f7399c427", "attempt_id": "871c313f3e50", "run_group_id": "E4_core:A5|repairable|qwen3b:h2", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:22:28.770447Z", "task_family": "synthetic", "document_id": "syn_03009_2000", "document_hash": "6ed94dcf728255bd7a7d11c97c362930728fb81a3e91f1906eb6eec86e0fedb7", "question_id": "q11_aggregation", "split": "test", "cluster_id": "syn_03009_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 2, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1976, "query_tokens": 187, "message_tokens": 94, "message_utf8_bytes": 310, "cache_tensor_bytes": null, "serialized_payload_bytes": 310, "side_information_bytes": 0, "fetched_bytes": 1080, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 21280.66345793195, "hops": {"1": 16602.87162498571, "2": 4677.791832946241}, "receiver_prefill_ms": 183.47491708118469, "ttft_ms": 52.66187503002584, "decode_ms": 50.97041593398899}, "total_wall_ms": 21568.23029089719, "ttft_ms": 52.66187503002584, "input_usage": {"lookups": 2}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 8, "retention_rate": 0.12307692307692308, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.0, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 310}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 12, "retention_rate": 0.18461538461538463, "retention_by_kind": {"person": 1.0, "date": 0.0, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 698}, "2": {"facts_total": 65, "facts_retained": 8, "retention_rate": 0.12307692307692308, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.0, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 310}}}, "reference_key_id": "keys/synthetic/syn_03009_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "f58a7dc5658fb58db0a87934226394775c9d870973cde3e211186d96f2a1306d", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "3b4f9f235d283fcc037b6229", "attempt_id": "9dc13ec7c472", "run_group_id": "E4_core:A5|repairable|qwen3b:h2", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:22:29.075280Z", "task_family": "synthetic", "document_id": "syn_03009_2000", "document_hash": "6ed94dcf728255bd7a7d11c97c362930728fb81a3e91f1906eb6eec86e0fedb7", "question_id": "q12_unit_conversion", "split": "test", "cluster_id": "syn_03009_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 2, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1976, "query_tokens": 185, "message_tokens": 94, "message_utf8_bytes": 310, "cache_tensor_bytes": null, "serialized_payload_bytes": 310, "side_information_bytes": 0, "fetched_bytes": 1080, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 21280.66345793195, "hops": {"1": 16602.87162498571, "2": 4677.791832946241}, "receiver_prefill_ms": 337.5880840467289, "ttft_ms": 43.82312495727092, "decode_ms": 24.413791950792074}, "total_wall_ms": 21686.900249915197, "ttft_ms": 43.82312495727092, "input_usage": {"lookups": 2}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 8, "retention_rate": 0.12307692307692308, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.0, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 310}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 12, "retention_rate": 0.18461538461538463, "retention_by_kind": {"person": 1.0, "date": 0.0, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 698}, "2": {"facts_total": 65, "facts_retained": 8, "retention_rate": 0.12307692307692308, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.0, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 310}}}, "reference_key_id": "keys/synthetic/syn_03009_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "f58a7dc5658fb58db0a87934226394775c9d870973cde3e211186d96f2a1306d", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "da066e271678120a8061c717", "attempt_id": "fef2cb68a3ae", "run_group_id": "E4_core:A5|repairable|qwen3b:h2", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:22:29.498692Z", "task_family": "synthetic", "document_id": "syn_03009_2000", "document_hash": "6ed94dcf728255bd7a7d11c97c362930728fb81a3e91f1906eb6eec86e0fedb7", "question_id": "q13_unanswerable", "split": "test", "cluster_id": "syn_03009_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 2, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1976, "query_tokens": 189, "message_tokens": 94, "message_utf8_bytes": 310, "cache_tensor_bytes": null, "serialized_payload_bytes": 310, "side_information_bytes": 0, "fetched_bytes": 1080, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 21280.66345793195, "hops": {"1": 16602.87162498571, "2": 4677.791832946241}, "receiver_prefill_ms": 227.84545796457678, "ttft_ms": 28.0050840228796, "decode_ms": 36.91620798781514}, "total_wall_ms": 21573.709874995984, "ttft_ms": 28.0050840228796, "input_usage": {"lookups": 2}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 8, "retention_rate": 0.12307692307692308, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.0, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 310}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 12, "retention_rate": 0.18461538461538463, "retention_by_kind": {"person": 1.0, "date": 0.0, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 698}, "2": {"facts_total": 65, "facts_retained": 8, "retention_rate": 0.12307692307692308, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.0, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 310}}}, "reference_key_id": "keys/synthetic/syn_03009_2000.json", "task_scores": {"correct": 1.0, "abstained": 1.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "f58a7dc5658fb58db0a87934226394775c9d870973cde3e211186d96f2a1306d", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "726239517679210210f3fc12", "attempt_id": "6bb0d29c1d15", "run_group_id": "E4_core:A5|repairable|qwen3b:h2", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:22:29.808475Z", "task_family": "synthetic", "document_id": "syn_03009_2000", "document_hash": "6ed94dcf728255bd7a7d11c97c362930728fb81a3e91f1906eb6eec86e0fedb7", "question_id": "q14_unanswerable", "split": "test", "cluster_id": "syn_03009_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 2, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1976, "query_tokens": 189, "message_tokens": 94, "message_utf8_bytes": 310, "cache_tensor_bytes": null, "serialized_payload_bytes": 310, "side_information_bytes": 0, "fetched_bytes": 1080, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 21280.66345793195, "hops": {"1": 16602.87162498571, "2": 4677.791832946241}, "receiver_prefill_ms": 185.9322499949485, "ttft_ms": 25.154792005196214, "decode_ms": 35.625499906018376}, "total_wall_ms": 21527.72199991159, "ttft_ms": 25.154792005196214, "input_usage": {"lookups": 2}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 8, "retention_rate": 0.12307692307692308, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.0, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 310}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 12, "retention_rate": 0.18461538461538463, "retention_by_kind": {"person": 1.0, "date": 0.0, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 698}, "2": {"facts_total": 65, "facts_retained": 8, "retention_rate": 0.12307692307692308, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.0, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 310}}}, "reference_key_id": "keys/synthetic/syn_03009_2000.json", "task_scores": {"correct": 1.0, "abstained": 1.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "f58a7dc5658fb58db0a87934226394775c9d870973cde3e211186d96f2a1306d", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "f8eafb6bdaeab71681c63201", "attempt_id": "69c2f0e990e8", "run_group_id": "E4_core:A5|repairable|qwen3b:h4", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:22:56.190403Z", "task_family": "synthetic", "document_id": "syn_03009_2000", "document_hash": "6ed94dcf728255bd7a7d11c97c362930728fb81a3e91f1906eb6eec86e0fedb7", "question_id": "q00_single_fact", "split": "test", "cluster_id": "syn_03009_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 4, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1976, "query_tokens": 422, "message_tokens": 339, "message_utf8_bytes": 945, "cache_tensor_bytes": null, "serialized_payload_bytes": 945, "side_information_bytes": 0, "fetched_bytes": 3240, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 47396.3774999138, "hops": {"1": 16602.87162498571, "2": 4677.791832946241, "3": 11236.125083058141, "4": 14879.588958923705}, "receiver_prefill_ms": 258.99633299559355, "ttft_ms": 49.77558401878923, "decode_ms": 50.01341598108411}, "total_wall_ms": 47755.767791881226, "ttft_ms": 49.77558401878923, "input_usage": {"lookups": 6}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 12, "retention_rate": 0.18461538461538463, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.2, "amount": 0.15384615384615385, "status": 0.5, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 945}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 12, "retention_rate": 0.18461538461538463, "retention_by_kind": {"person": 1.0, "date": 0.0, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 698}, "2": {"facts_total": 65, "facts_retained": 8, "retention_rate": 0.12307692307692308, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.0, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 310}, "3": {"facts_total": 65, "facts_retained": 12, "retention_rate": 0.18461538461538463, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.2, "amount": 0.15384615384615385, "status": 0.5, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 725}, "4": {"facts_total": 65, "facts_retained": 12, "retention_rate": 0.18461538461538463, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.2, "amount": 0.15384615384615385, "status": 0.5, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 945}}}, "reference_key_id": "keys/synthetic/syn_03009_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "855b6cb08da5888e0939be3cf8d9b3d7844b2cf024991e3f1c53d6e4e3638d06", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "bbdaf3efe1492b44e4e8b73b", "attempt_id": "43d37a54e17b", "run_group_id": "E4_core:A5|repairable|qwen3b:h4", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:22:56.568328Z", "task_family": "synthetic", "document_id": "syn_03009_2000", "document_hash": "6ed94dcf728255bd7a7d11c97c362930728fb81a3e91f1906eb6eec86e0fedb7", "question_id": "q01_single_fact", "split": "test", "cluster_id": "syn_03009_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 4, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1976, "query_tokens": 427, "message_tokens": 339, "message_utf8_bytes": 945, "cache_tensor_bytes": null, "serialized_payload_bytes": 945, "side_information_bytes": 0, "fetched_bytes": 3240, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 47396.3774999138, "hops": {"1": 16602.87162498571, "2": 4677.791832946241, "3": 11236.125083058141, "4": 14879.588958923705}, "receiver_prefill_ms": 450.9685829980299, "ttft_ms": 42.48562501743436, "decode_ms": 30.357208102941513}, "total_wall_ms": 47920.78166687861, "ttft_ms": 42.48562501743436, "input_usage": {"lookups": 6}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 12, "retention_rate": 0.18461538461538463, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.2, "amount": 0.15384615384615385, "status": 0.5, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 945}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 12, "retention_rate": 0.18461538461538463, "retention_by_kind": {"person": 1.0, "date": 0.0, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 698}, "2": {"facts_total": 65, "facts_retained": 8, "retention_rate": 0.12307692307692308, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.0, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 310}, "3": {"facts_total": 65, "facts_retained": 12, "retention_rate": 0.18461538461538463, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.2, "amount": 0.15384615384615385, "status": 0.5, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 725}, "4": {"facts_total": 65, "facts_retained": 12, "retention_rate": 0.18461538461538463, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.2, "amount": 0.15384615384615385, "status": 0.5, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 945}}}, "reference_key_id": "keys/synthetic/syn_03009_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "855b6cb08da5888e0939be3cf8d9b3d7844b2cf024991e3f1c53d6e4e3638d06", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "f21054421a0ea3e695ec927b", "attempt_id": "f08ca113f776", "run_group_id": "E4_core:A5|repairable|qwen3b:h4", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:22:57.110956Z", "task_family": "synthetic", "document_id": "syn_03009_2000", "document_hash": "6ed94dcf728255bd7a7d11c97c362930728fb81a3e91f1906eb6eec86e0fedb7", "question_id": "q02_exact_copy", "split": "test", "cluster_id": "syn_03009_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 4, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1976, "query_tokens": 425, "message_tokens": 339, "message_utf8_bytes": 945, "cache_tensor_bytes": null, "serialized_payload_bytes": 945, "side_information_bytes": 0, "fetched_bytes": 3240, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 47396.3774999138, "hops": {"1": 16602.87162498571, "2": 4677.791832946241, "3": 11236.125083058141, "4": 14879.588958923705}, "receiver_prefill_ms": 291.7893329868093, "ttft_ms": 27.184000005945563, "decode_ms": 39.55412493087351}, "total_wall_ms": 47755.44274994172, "ttft_ms": 27.184000005945563, "input_usage": {"lookups": 6}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 12, "retention_rate": 0.18461538461538463, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.2, "amount": 0.15384615384615385, "status": 0.5, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 945}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 12, "retention_rate": 0.18461538461538463, "retention_by_kind": {"person": 1.0, "date": 0.0, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 698}, "2": {"facts_total": 65, "facts_retained": 8, "retention_rate": 0.12307692307692308, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.0, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 310}, "3": {"facts_total": 65, "facts_retained": 12, "retention_rate": 0.18461538461538463, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.2, "amount": 0.15384615384615385, "status": 0.5, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 725}, "4": {"facts_total": 65, "facts_retained": 12, "retention_rate": 0.18461538461538463, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.2, "amount": 0.15384615384615385, "status": 0.5, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 945}}}, "reference_key_id": "keys/synthetic/syn_03009_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "855b6cb08da5888e0939be3cf8d9b3d7844b2cf024991e3f1c53d6e4e3638d06", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "253719536a9ec20eba5e9554", "attempt_id": "161efb66ada8", "run_group_id": "E4_core:A5|repairable|qwen3b:h4", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:22:57.487229Z", "task_family": "synthetic", "document_id": "syn_03009_2000", "document_hash": "6ed94dcf728255bd7a7d11c97c362930728fb81a3e91f1906eb6eec86e0fedb7", "question_id": "q03_exact_copy", "split": "test", "cluster_id": "syn_03009_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 4, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1976, "query_tokens": 421, "message_tokens": 339, "message_utf8_bytes": 945, "cache_tensor_bytes": null, "serialized_payload_bytes": 945, "side_information_bytes": 0, "fetched_bytes": 3240, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 47396.3774999138, "hops": {"1": 16602.87162498571, "2": 4677.791832946241, "3": 11236.125083058141, "4": 14879.588958923705}, "receiver_prefill_ms": 289.7155419923365, "ttft_ms": 21.305250003933907, "decode_ms": 177.08729195874184}, "total_wall_ms": 47885.16083289869, "ttft_ms": 21.305250003933907, "input_usage": {"lookups": 6}, "output_usage": {"output_tokens": 5}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "Idris Bergström", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 12, "retention_rate": 0.18461538461538463, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.2, "amount": 0.15384615384615385, "status": 0.5, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 945}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 12, "retention_rate": 0.18461538461538463, "retention_by_kind": {"person": 1.0, "date": 0.0, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 698}, "2": {"facts_total": 65, "facts_retained": 8, "retention_rate": 0.12307692307692308, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.0, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 310}, "3": {"facts_total": 65, "facts_retained": 12, "retention_rate": 0.18461538461538463, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.2, "amount": 0.15384615384615385, "status": 0.5, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 725}, "4": {"facts_total": 65, "facts_retained": 12, "retention_rate": 0.18461538461538463, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.2, "amount": 0.15384615384615385, "status": 0.5, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 945}}}, "reference_key_id": "keys/synthetic/syn_03009_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "855b6cb08da5888e0939be3cf8d9b3d7844b2cf024991e3f1c53d6e4e3638d06", "output_hash": "fd88832c9fd0ab175818d76d875d70156225bd4b34ec401514546ef70f4f826c"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "51fed8949029c0ea17bcb28c", "attempt_id": "d808426738b0", "run_group_id": "E4_core:A5|repairable|qwen3b:h4", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:22:57.993620Z", "task_family": "synthetic", "document_id": "syn_03009_2000", "document_hash": "6ed94dcf728255bd7a7d11c97c362930728fb81a3e91f1906eb6eec86e0fedb7", "question_id": "q04_multi_hop", "split": "test", "cluster_id": "syn_03009_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 4, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1976, "query_tokens": 429, "message_tokens": 339, "message_utf8_bytes": 945, "cache_tensor_bytes": null, "serialized_payload_bytes": 945, "side_information_bytes": 0, "fetched_bytes": 3240, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 47396.3774999138, "hops": {"1": 16602.87162498571, "2": 4677.791832946241, "3": 11236.125083058141, "4": 14879.588958923705}, "receiver_prefill_ms": 285.94600001815706, "ttft_ms": 63.11416602693498, "decode_ms": 388.6248749913648}, "total_wall_ms": 48134.72812494729, "ttft_ms": 63.11416602693498, "input_usage": {"lookups": 6}, "output_usage": {"output_tokens": 7}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "4670383", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 12, "retention_rate": 0.18461538461538463, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.2, "amount": 0.15384615384615385, "status": 0.5, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 945}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 12, "retention_rate": 0.18461538461538463, "retention_by_kind": {"person": 1.0, "date": 0.0, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 698}, "2": {"facts_total": 65, "facts_retained": 8, "retention_rate": 0.12307692307692308, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.0, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 310}, "3": {"facts_total": 65, "facts_retained": 12, "retention_rate": 0.18461538461538463, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.2, "amount": 0.15384615384615385, "status": 0.5, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 725}, "4": {"facts_total": 65, "facts_retained": 12, "retention_rate": 0.18461538461538463, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.2, "amount": 0.15384615384615385, "status": 0.5, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 945}}}, "reference_key_id": "keys/synthetic/syn_03009_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "855b6cb08da5888e0939be3cf8d9b3d7844b2cf024991e3f1c53d6e4e3638d06", "output_hash": "19ae8c357e5500bf494a6163bfbd11380b2996e8ea532a2c64f4a599464820f7"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "cbbad245c3322594d418198c", "attempt_id": "cc2f60f43a1d", "run_group_id": "E4_core:A5|repairable|qwen3b:h4", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:22:58.751341Z", "task_family": "synthetic", "document_id": "syn_03009_2000", "document_hash": "6ed94dcf728255bd7a7d11c97c362930728fb81a3e91f1906eb6eec86e0fedb7", "question_id": "q05_multi_hop", "split": "test", "cluster_id": "syn_03009_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 4, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1976, "query_tokens": 432, "message_tokens": 339, "message_utf8_bytes": 945, "cache_tensor_bytes": null, "serialized_payload_bytes": 945, "side_information_bytes": 0, "fetched_bytes": 3240, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 47396.3774999138, "hops": {"1": 16602.87162498571, "2": 4677.791832946241, "3": 11236.125083058141, "4": 14879.588958923705}, "receiver_prefill_ms": 306.41933297738433, "ttft_ms": 31.313624931499362, "decode_ms": 33.911957987584174}, "total_wall_ms": 47768.543458892964, "ttft_ms": 31.313624931499362, "input_usage": {"lookups": 6}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 12, "retention_rate": 0.18461538461538463, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.2, "amount": 0.15384615384615385, "status": 0.5, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 945}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 12, "retention_rate": 0.18461538461538463, "retention_by_kind": {"person": 1.0, "date": 0.0, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 698}, "2": {"facts_total": 65, "facts_retained": 8, "retention_rate": 0.12307692307692308, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.0, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 310}, "3": {"facts_total": 65, "facts_retained": 12, "retention_rate": 0.18461538461538463, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.2, "amount": 0.15384615384615385, "status": 0.5, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 725}, "4": {"facts_total": 65, "facts_retained": 12, "retention_rate": 0.18461538461538463, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.2, "amount": 0.15384615384615385, "status": 0.5, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 945}}}, "reference_key_id": "keys/synthetic/syn_03009_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "855b6cb08da5888e0939be3cf8d9b3d7844b2cf024991e3f1c53d6e4e3638d06", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "8269dd8830d13578deb66448", "attempt_id": "d92af7c9959c", "run_group_id": "E4_core:A5|repairable|qwen3b:h4", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:22:59.140477Z", "task_family": "synthetic", "document_id": "syn_03009_2000", "document_hash": "6ed94dcf728255bd7a7d11c97c362930728fb81a3e91f1906eb6eec86e0fedb7", "question_id": "q06_negation", "split": "test", "cluster_id": "syn_03009_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 4, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1976, "query_tokens": 422, "message_tokens": 339, "message_utf8_bytes": 945, "cache_tensor_bytes": null, "serialized_payload_bytes": 945, "side_information_bytes": 0, "fetched_bytes": 3240, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 47396.3774999138, "hops": {"1": 16602.87162498571, "2": 4677.791832946241, "3": 11236.125083058141, "4": 14879.588958923705}, "receiver_prefill_ms": 292.28383291047066, "ttft_ms": 29.62287503760308, "decode_ms": 20.11208306066692}, "total_wall_ms": 47738.89708297793, "ttft_ms": 29.62287503760308, "input_usage": {"lookups": 6}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 12, "retention_rate": 0.18461538461538463, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.2, "amount": 0.15384615384615385, "status": 0.5, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 945}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 12, "retention_rate": 0.18461538461538463, "retention_by_kind": {"person": 1.0, "date": 0.0, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 698}, "2": {"facts_total": 65, "facts_retained": 8, "retention_rate": 0.12307692307692308, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.0, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 310}, "3": {"facts_total": 65, "facts_retained": 12, "retention_rate": 0.18461538461538463, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.2, "amount": 0.15384615384615385, "status": 0.5, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 725}, "4": {"facts_total": 65, "facts_retained": 12, "retention_rate": 0.18461538461538463, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.2, "amount": 0.15384615384615385, "status": 0.5, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 945}}}, "reference_key_id": "keys/synthetic/syn_03009_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "855b6cb08da5888e0939be3cf8d9b3d7844b2cf024991e3f1c53d6e4e3638d06", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "84b8f4a91fbf995523379f67", "attempt_id": "93a0a6ad4939", "run_group_id": "E4_core:A5|repairable|qwen3b:h4", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:22:59.500046Z", "task_family": "synthetic", "document_id": "syn_03009_2000", "document_hash": "6ed94dcf728255bd7a7d11c97c362930728fb81a3e91f1906eb6eec86e0fedb7", "question_id": "q07_negation", "split": "test", "cluster_id": "syn_03009_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 4, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1976, "query_tokens": 421, "message_tokens": 339, "message_utf8_bytes": 945, "cache_tensor_bytes": null, "serialized_payload_bytes": 945, "side_information_bytes": 0, "fetched_bytes": 3240, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 47396.3774999138, "hops": {"1": 16602.87162498571, "2": 4677.791832946241, "3": 11236.125083058141, "4": 14879.588958923705}, "receiver_prefill_ms": 279.1978749446571, "ttft_ms": 65.36912498995662, "decode_ms": 69.23145800828934}, "total_wall_ms": 47810.71099988185, "ttft_ms": 65.36912498995662, "input_usage": {"lookups": 6}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 12, "retention_rate": 0.18461538461538463, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.2, "amount": 0.15384615384615385, "status": 0.5, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 945}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 12, "retention_rate": 0.18461538461538463, "retention_by_kind": {"person": 1.0, "date": 0.0, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 698}, "2": {"facts_total": 65, "facts_retained": 8, "retention_rate": 0.12307692307692308, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.0, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 310}, "3": {"facts_total": 65, "facts_retained": 12, "retention_rate": 0.18461538461538463, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.2, "amount": 0.15384615384615385, "status": 0.5, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 725}, "4": {"facts_total": 65, "facts_retained": 12, "retention_rate": 0.18461538461538463, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.2, "amount": 0.15384615384615385, "status": 0.5, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 945}}}, "reference_key_id": "keys/synthetic/syn_03009_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "855b6cb08da5888e0939be3cf8d9b3d7844b2cf024991e3f1c53d6e4e3638d06", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "a71d34164badc9a1a5b3c370", "attempt_id": "660953d07be3", "run_group_id": "E4_core:A5|repairable|qwen3b:h4", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:22:59.934005Z", "task_family": "synthetic", "document_id": "syn_03009_2000", "document_hash": "6ed94dcf728255bd7a7d11c97c362930728fb81a3e91f1906eb6eec86e0fedb7", "question_id": "q08_revision", "split": "test", "cluster_id": "syn_03009_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 4, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1976, "query_tokens": 439, "message_tokens": 339, "message_utf8_bytes": 945, "cache_tensor_bytes": null, "serialized_payload_bytes": 945, "side_information_bytes": 0, "fetched_bytes": 3240, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 47396.3774999138, "hops": {"1": 16602.87162498571, "2": 4677.791832946241, "3": 11236.125083058141, "4": 14879.588958923705}, "receiver_prefill_ms": 415.4934170655906, "ttft_ms": 36.75249998923391, "decode_ms": 39.87762506585568}, "total_wall_ms": 47889.132707961835, "ttft_ms": 36.75249998923391, "input_usage": {"lookups": 6}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 12, "retention_rate": 0.18461538461538463, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.2, "amount": 0.15384615384615385, "status": 0.5, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 945}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 12, "retention_rate": 0.18461538461538463, "retention_by_kind": {"person": 1.0, "date": 0.0, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 698}, "2": {"facts_total": 65, "facts_retained": 8, "retention_rate": 0.12307692307692308, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.0, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 310}, "3": {"facts_total": 65, "facts_retained": 12, "retention_rate": 0.18461538461538463, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.2, "amount": 0.15384615384615385, "status": 0.5, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 725}, "4": {"facts_total": 65, "facts_retained": 12, "retention_rate": 0.18461538461538463, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.2, "amount": 0.15384615384615385, "status": 0.5, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 945}}}, "reference_key_id": "keys/synthetic/syn_03009_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "855b6cb08da5888e0939be3cf8d9b3d7844b2cf024991e3f1c53d6e4e3638d06", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "f87b8202aa92e8699fb54a40", "attempt_id": "5a2041d683ac", "run_group_id": "E4_core:A5|repairable|qwen3b:h4", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:23:00.446612Z", "task_family": "synthetic", "document_id": "syn_03009_2000", "document_hash": "6ed94dcf728255bd7a7d11c97c362930728fb81a3e91f1906eb6eec86e0fedb7", "question_id": "q09_revision", "split": "test", "cluster_id": "syn_03009_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 4, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1976, "query_tokens": 432, "message_tokens": 339, "message_utf8_bytes": 945, "cache_tensor_bytes": null, "serialized_payload_bytes": 945, "side_information_bytes": 0, "fetched_bytes": 3240, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 47396.3774999138, "hops": {"1": 16602.87162498571, "2": 4677.791832946241, "3": 11236.125083058141, "4": 14879.588958923705}, "receiver_prefill_ms": 285.72670894209296, "ttft_ms": 41.20708291884512, "decode_ms": 30.14420799445361}, "total_wall_ms": 47753.99554090109, "ttft_ms": 41.20708291884512, "input_usage": {"lookups": 6}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 12, "retention_rate": 0.18461538461538463, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.2, "amount": 0.15384615384615385, "status": 0.5, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 945}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 12, "retention_rate": 0.18461538461538463, "retention_by_kind": {"person": 1.0, "date": 0.0, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 698}, "2": {"facts_total": 65, "facts_retained": 8, "retention_rate": 0.12307692307692308, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.0, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 310}, "3": {"facts_total": 65, "facts_retained": 12, "retention_rate": 0.18461538461538463, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.2, "amount": 0.15384615384615385, "status": 0.5, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 725}, "4": {"facts_total": 65, "facts_retained": 12, "retention_rate": 0.18461538461538463, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.2, "amount": 0.15384615384615385, "status": 0.5, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 945}}}, "reference_key_id": "keys/synthetic/syn_03009_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "855b6cb08da5888e0939be3cf8d9b3d7844b2cf024991e3f1c53d6e4e3638d06", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "c4263d6bb5e0805f87213f53", "attempt_id": "136b013687ec", "run_group_id": "E4_core:A5|repairable|qwen3b:h4", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:23:00.821206Z", "task_family": "synthetic", "document_id": "syn_03009_2000", "document_hash": "6ed94dcf728255bd7a7d11c97c362930728fb81a3e91f1906eb6eec86e0fedb7", "question_id": "q10_temporal", "split": "test", "cluster_id": "syn_03009_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 4, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1976, "query_tokens": 441, "message_tokens": 339, "message_utf8_bytes": 945, "cache_tensor_bytes": null, "serialized_payload_bytes": 945, "side_information_bytes": 0, "fetched_bytes": 3240, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 47396.3774999138, "hops": {"1": 16602.87162498571, "2": 4677.791832946241, "3": 11236.125083058141, "4": 14879.588958923705}, "receiver_prefill_ms": 294.7969170054421, "ttft_ms": 21.436083945445716, "decode_ms": 19.979208940640092}, "total_wall_ms": 47733.1599999452, "ttft_ms": 21.436083945445716, "input_usage": {"lookups": 6}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 12, "retention_rate": 0.18461538461538463, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.2, "amount": 0.15384615384615385, "status": 0.5, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 945}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 12, "retention_rate": 0.18461538461538463, "retention_by_kind": {"person": 1.0, "date": 0.0, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 698}, "2": {"facts_total": 65, "facts_retained": 8, "retention_rate": 0.12307692307692308, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.0, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 310}, "3": {"facts_total": 65, "facts_retained": 12, "retention_rate": 0.18461538461538463, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.2, "amount": 0.15384615384615385, "status": 0.5, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 725}, "4": {"facts_total": 65, "facts_retained": 12, "retention_rate": 0.18461538461538463, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.2, "amount": 0.15384615384615385, "status": 0.5, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 945}}}, "reference_key_id": "keys/synthetic/syn_03009_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "855b6cb08da5888e0939be3cf8d9b3d7844b2cf024991e3f1c53d6e4e3638d06", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "761c53f6e10e933f9a05dd42", "attempt_id": "d8258949b37b", "run_group_id": "E4_core:A5|repairable|qwen3b:h4", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:23:01.174870Z", "task_family": "synthetic", "document_id": "syn_03009_2000", "document_hash": "6ed94dcf728255bd7a7d11c97c362930728fb81a3e91f1906eb6eec86e0fedb7", "question_id": "q11_aggregation", "split": "test", "cluster_id": "syn_03009_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 4, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1976, "query_tokens": 432, "message_tokens": 339, "message_utf8_bytes": 945, "cache_tensor_bytes": null, "serialized_payload_bytes": 945, "side_information_bytes": 0, "fetched_bytes": 3240, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 47396.3774999138, "hops": {"1": 16602.87162498571, "2": 4677.791832946241, "3": 11236.125083058141, "4": 14879.588958923705}, "receiver_prefill_ms": 302.2075000917539, "ttft_ms": 47.441708971746266, "decode_ms": 427.285416983068}, "total_wall_ms": 48173.92624984495, "ttft_ms": 47.441708971746266, "input_usage": {"lookups": 6}, "output_usage": {"output_tokens": 7}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "5247196", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 12, "retention_rate": 0.18461538461538463, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.2, "amount": 0.15384615384615385, "status": 0.5, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 945}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 12, "retention_rate": 0.18461538461538463, "retention_by_kind": {"person": 1.0, "date": 0.0, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 698}, "2": {"facts_total": 65, "facts_retained": 8, "retention_rate": 0.12307692307692308, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.0, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 310}, "3": {"facts_total": 65, "facts_retained": 12, "retention_rate": 0.18461538461538463, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.2, "amount": 0.15384615384615385, "status": 0.5, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 725}, "4": {"facts_total": 65, "facts_retained": 12, "retention_rate": 0.18461538461538463, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.2, "amount": 0.15384615384615385, "status": 0.5, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 945}}}, "reference_key_id": "keys/synthetic/syn_03009_2000.json", "task_scores": {"correct": 0.0, "abstained": 0.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "855b6cb08da5888e0939be3cf8d9b3d7844b2cf024991e3f1c53d6e4e3638d06", "output_hash": "3ac48b1390a5593419a76c6409f1c54c01affa22edaa1be1850800277310a17c"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "4359a4d1da51867d6f043f19", "attempt_id": "8a15b8bbf4af", "run_group_id": "E4_core:A5|repairable|qwen3b:h4", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:23:01.972644Z", "task_family": "synthetic", "document_id": "syn_03009_2000", "document_hash": "6ed94dcf728255bd7a7d11c97c362930728fb81a3e91f1906eb6eec86e0fedb7", "question_id": "q12_unit_conversion", "split": "test", "cluster_id": "syn_03009_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 4, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1976, "query_tokens": 430, "message_tokens": 339, "message_utf8_bytes": 945, "cache_tensor_bytes": null, "serialized_payload_bytes": 945, "side_information_bytes": 0, "fetched_bytes": 3240, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 47396.3774999138, "hops": {"1": 16602.87162498571, "2": 4677.791832946241, "3": 11236.125083058141, "4": 14879.588958923705}, "receiver_prefill_ms": 276.50137490127236, "ttft_ms": 21.714792004786432, "decode_ms": 38.61770802177489}, "total_wall_ms": 47733.8831248926, "ttft_ms": 21.714792004786432, "input_usage": {"lookups": 6}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 12, "retention_rate": 0.18461538461538463, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.2, "amount": 0.15384615384615385, "status": 0.5, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 945}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 12, "retention_rate": 0.18461538461538463, "retention_by_kind": {"person": 1.0, "date": 0.0, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 698}, "2": {"facts_total": 65, "facts_retained": 8, "retention_rate": 0.12307692307692308, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.0, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 310}, "3": {"facts_total": 65, "facts_retained": 12, "retention_rate": 0.18461538461538463, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.2, "amount": 0.15384615384615385, "status": 0.5, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 725}, "4": {"facts_total": 65, "facts_retained": 12, "retention_rate": 0.18461538461538463, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.2, "amount": 0.15384615384615385, "status": 0.5, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 945}}}, "reference_key_id": "keys/synthetic/syn_03009_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "855b6cb08da5888e0939be3cf8d9b3d7844b2cf024991e3f1c53d6e4e3638d06", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "373c72a430b98cc75d2116d0", "attempt_id": "a559f2831004", "run_group_id": "E4_core:A5|repairable|qwen3b:h4", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:23:02.328512Z", "task_family": "synthetic", "document_id": "syn_03009_2000", "document_hash": "6ed94dcf728255bd7a7d11c97c362930728fb81a3e91f1906eb6eec86e0fedb7", "question_id": "q13_unanswerable", "split": "test", "cluster_id": "syn_03009_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 4, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1976, "query_tokens": 434, "message_tokens": 339, "message_utf8_bytes": 945, "cache_tensor_bytes": null, "serialized_payload_bytes": 945, "side_information_bytes": 0, "fetched_bytes": 3240, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 47396.3774999138, "hops": {"1": 16602.87162498571, "2": 4677.791832946241, "3": 11236.125083058141, "4": 14879.588958923705}, "receiver_prefill_ms": 307.75954097043723, "ttft_ms": 22.66791695728898, "decode_ms": 41.58970795106143}, "total_wall_ms": 47769.14637489244, "ttft_ms": 22.66791695728898, "input_usage": {"lookups": 6}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 12, "retention_rate": 0.18461538461538463, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.2, "amount": 0.15384615384615385, "status": 0.5, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 945}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 12, "retention_rate": 0.18461538461538463, "retention_by_kind": {"person": 1.0, "date": 0.0, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 698}, "2": {"facts_total": 65, "facts_retained": 8, "retention_rate": 0.12307692307692308, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.0, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 310}, "3": {"facts_total": 65, "facts_retained": 12, "retention_rate": 0.18461538461538463, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.2, "amount": 0.15384615384615385, "status": 0.5, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 725}, "4": {"facts_total": 65, "facts_retained": 12, "retention_rate": 0.18461538461538463, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.2, "amount": 0.15384615384615385, "status": 0.5, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 945}}}, "reference_key_id": "keys/synthetic/syn_03009_2000.json", "task_scores": {"correct": 1.0, "abstained": 1.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "855b6cb08da5888e0939be3cf8d9b3d7844b2cf024991e3f1c53d6e4e3638d06", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "020ff16e305a8c36b7543e4d", "attempt_id": "d46c7ce009b1", "run_group_id": "E4_core:A5|repairable|qwen3b:h4", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:23:02.719318Z", "task_family": "synthetic", "document_id": "syn_03009_2000", "document_hash": "6ed94dcf728255bd7a7d11c97c362930728fb81a3e91f1906eb6eec86e0fedb7", "question_id": "q14_unanswerable", "split": "test", "cluster_id": "syn_03009_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 4, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1976, "query_tokens": 434, "message_tokens": 339, "message_utf8_bytes": 945, "cache_tensor_bytes": null, "serialized_payload_bytes": 945, "side_information_bytes": 0, "fetched_bytes": 3240, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 47396.3774999138, "hops": {"1": 16602.87162498571, "2": 4677.791832946241, "3": 11236.125083058141, "4": 14879.588958923705}, "receiver_prefill_ms": 256.98000006377697, "ttft_ms": 52.5473749730736, "decode_ms": 58.012459077872336}, "total_wall_ms": 47764.660041895695, "ttft_ms": 52.5473749730736, "input_usage": {"lookups": 6}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 12, "retention_rate": 0.18461538461538463, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.2, "amount": 0.15384615384615385, "status": 0.5, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 945}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 12, "retention_rate": 0.18461538461538463, "retention_by_kind": {"person": 1.0, "date": 0.0, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 698}, "2": {"facts_total": 65, "facts_retained": 8, "retention_rate": 0.12307692307692308, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.0, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 310}, "3": {"facts_total": 65, "facts_retained": 12, "retention_rate": 0.18461538461538463, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.2, "amount": 0.15384615384615385, "status": 0.5, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 725}, "4": {"facts_total": 65, "facts_retained": 12, "retention_rate": 0.18461538461538463, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.2, "amount": 0.15384615384615385, "status": 0.5, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 945}}}, "reference_key_id": "keys/synthetic/syn_03009_2000.json", "task_scores": {"correct": 1.0, "abstained": 1.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "855b6cb08da5888e0939be3cf8d9b3d7844b2cf024991e3f1c53d6e4e3638d06", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "2886937dc4ccda5774d15b1f", "attempt_id": "62d8ef2153d0", "run_group_id": "E4_core:A5|repairable|qwen3b:h8", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:24:02.666192Z", "task_family": "synthetic", "document_id": "syn_03009_2000", "document_hash": "6ed94dcf728255bd7a7d11c97c362930728fb81a3e91f1906eb6eec86e0fedb7", "question_id": "q00_single_fact", "split": "test", "cluster_id": "syn_03009_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 8, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1976, "query_tokens": 422, "message_tokens": 339, "message_utf8_bytes": 945, "cache_tensor_bytes": null, "serialized_payload_bytes": 945, "side_information_bytes": 0, "fetched_bytes": 7560, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 106952.59120792616, "hops": {"1": 16602.87162498571, "2": 4677.791832946241, "3": 11236.125083058141, "4": 14879.588958923705, "5": 14824.957874952815, "6": 14889.837124967016, "7": 14804.55670808442, "8": 15036.862000008114}, "receiver_prefill_ms": 302.7138339821249, "ttft_ms": 27.20095799304545, "decode_ms": 36.596457939594984}, "total_wall_ms": 107319.68662492, "ttft_ms": 27.20095799304545, "input_usage": {"lookups": 14}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 12, "retention_rate": 0.18461538461538463, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.2, "amount": 0.15384615384615385, "status": 0.5, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 945}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 12, "retention_rate": 0.18461538461538463, "retention_by_kind": {"person": 1.0, "date": 0.0, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 698}, "2": {"facts_total": 65, "facts_retained": 8, "retention_rate": 0.12307692307692308, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.0, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 310}, "3": {"facts_total": 65, "facts_retained": 12, "retention_rate": 0.18461538461538463, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.2, "amount": 0.15384615384615385, "status": 0.5, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 725}, "4": {"facts_total": 65, "facts_retained": 12, "retention_rate": 0.18461538461538463, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.2, "amount": 0.15384615384615385, "status": 0.5, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 945}, "5": {"facts_total": 65, "facts_retained": 12, "retention_rate": 0.18461538461538463, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.2, "amount": 0.15384615384615385, "status": 0.5, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 945}, "6": {"facts_total": 65, "facts_retained": 12, "retention_rate": 0.18461538461538463, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.2, "amount": 0.15384615384615385, "status": 0.5, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 945}, "7": {"facts_total": 65, "facts_retained": 12, "retention_rate": 0.18461538461538463, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.2, "amount": 0.15384615384615385, "status": 0.5, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 945}, "8": {"facts_total": 65, "facts_retained": 12, "retention_rate": 0.18461538461538463, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.2, "amount": 0.15384615384615385, "status": 0.5, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 945}}}, "reference_key_id": "keys/synthetic/syn_03009_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "855b6cb08da5888e0939be3cf8d9b3d7844b2cf024991e3f1c53d6e4e3638d06", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "821bf6c07bc237639e765d7d", "attempt_id": "42b568896872", "run_group_id": "E4_core:A5|repairable|qwen3b:h8", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:24:03.051275Z", "task_family": "synthetic", "document_id": "syn_03009_2000", "document_hash": "6ed94dcf728255bd7a7d11c97c362930728fb81a3e91f1906eb6eec86e0fedb7", "question_id": "q01_single_fact", "split": "test", "cluster_id": "syn_03009_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 8, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1976, "query_tokens": 427, "message_tokens": 339, "message_utf8_bytes": 945, "cache_tensor_bytes": null, "serialized_payload_bytes": 945, "side_information_bytes": 0, "fetched_bytes": 7560, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 106952.59120792616, "hops": {"1": 16602.87162498571, "2": 4677.791832946241, "3": 11236.125083058141, "4": 14879.588958923705, "5": 14824.957874952815, "6": 14889.837124967016, "7": 14804.55670808442, "8": 15036.862000008114}, "receiver_prefill_ms": 301.9429170526564, "ttft_ms": 20.251416019164026, "decode_ms": 19.845374976284802}, "total_wall_ms": 107295.22870783694, "ttft_ms": 20.251416019164026, "input_usage": {"lookups": 14}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 12, "retention_rate": 0.18461538461538463, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.2, "amount": 0.15384615384615385, "status": 0.5, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 945}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 12, "retention_rate": 0.18461538461538463, "retention_by_kind": {"person": 1.0, "date": 0.0, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 698}, "2": {"facts_total": 65, "facts_retained": 8, "retention_rate": 0.12307692307692308, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.0, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 310}, "3": {"facts_total": 65, "facts_retained": 12, "retention_rate": 0.18461538461538463, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.2, "amount": 0.15384615384615385, "status": 0.5, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 725}, "4": {"facts_total": 65, "facts_retained": 12, "retention_rate": 0.18461538461538463, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.2, "amount": 0.15384615384615385, "status": 0.5, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 945}, "5": {"facts_total": 65, "facts_retained": 12, "retention_rate": 0.18461538461538463, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.2, "amount": 0.15384615384615385, "status": 0.5, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 945}, "6": {"facts_total": 65, "facts_retained": 12, "retention_rate": 0.18461538461538463, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.2, "amount": 0.15384615384615385, "status": 0.5, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 945}, "7": {"facts_total": 65, "facts_retained": 12, "retention_rate": 0.18461538461538463, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.2, "amount": 0.15384615384615385, "status": 0.5, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 945}, "8": {"facts_total": 65, "facts_retained": 12, "retention_rate": 0.18461538461538463, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.2, "amount": 0.15384615384615385, "status": 0.5, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 945}}}, "reference_key_id": "keys/synthetic/syn_03009_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "855b6cb08da5888e0939be3cf8d9b3d7844b2cf024991e3f1c53d6e4e3638d06", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "889c4b8e214061fe604e9b30", "attempt_id": "b7786326f8c4", "run_group_id": "E4_core:A5|repairable|qwen3b:h8", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:24:03.412930Z", "task_family": "synthetic", "document_id": "syn_03009_2000", "document_hash": "6ed94dcf728255bd7a7d11c97c362930728fb81a3e91f1906eb6eec86e0fedb7", "question_id": "q02_exact_copy", "split": "test", "cluster_id": "syn_03009_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 8, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1976, "query_tokens": 425, "message_tokens": 339, "message_utf8_bytes": 945, "cache_tensor_bytes": null, "serialized_payload_bytes": 945, "side_information_bytes": 0, "fetched_bytes": 7560, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 106952.59120792616, "hops": {"1": 16602.87162498571, "2": 4677.791832946241, "3": 11236.125083058141, "4": 14879.588958923705, "5": 14824.957874952815, "6": 14889.837124967016, "7": 14804.55670808442, "8": 15036.862000008114}, "receiver_prefill_ms": 296.7778330203146, "ttft_ms": 61.44091591704637, "decode_ms": 74.82512504793704}, "total_wall_ms": 107386.18516584393, "ttft_ms": 61.44091591704637, "input_usage": {"lookups": 14}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 12, "retention_rate": 0.18461538461538463, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.2, "amount": 0.15384615384615385, "status": 0.5, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 945}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 12, "retention_rate": 0.18461538461538463, "retention_by_kind": {"person": 1.0, "date": 0.0, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 698}, "2": {"facts_total": 65, "facts_retained": 8, "retention_rate": 0.12307692307692308, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.0, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 310}, "3": {"facts_total": 65, "facts_retained": 12, "retention_rate": 0.18461538461538463, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.2, "amount": 0.15384615384615385, "status": 0.5, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 725}, "4": {"facts_total": 65, "facts_retained": 12, "retention_rate": 0.18461538461538463, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.2, "amount": 0.15384615384615385, "status": 0.5, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 945}, "5": {"facts_total": 65, "facts_retained": 12, "retention_rate": 0.18461538461538463, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.2, "amount": 0.15384615384615385, "status": 0.5, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 945}, "6": {"facts_total": 65, "facts_retained": 12, "retention_rate": 0.18461538461538463, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.2, "amount": 0.15384615384615385, "status": 0.5, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 945}, "7": {"facts_total": 65, "facts_retained": 12, "retention_rate": 0.18461538461538463, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.2, "amount": 0.15384615384615385, "status": 0.5, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 945}, "8": {"facts_total": 65, "facts_retained": 12, "retention_rate": 0.18461538461538463, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.2, "amount": 0.15384615384615385, "status": 0.5, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 945}}}, "reference_key_id": "keys/synthetic/syn_03009_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "855b6cb08da5888e0939be3cf8d9b3d7844b2cf024991e3f1c53d6e4e3638d06", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "71208ecfa01453ee9326598b", "attempt_id": "f25dee0a5949", "run_group_id": "E4_core:A5|repairable|qwen3b:h8", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:24:03.862903Z", "task_family": "synthetic", "document_id": "syn_03009_2000", "document_hash": "6ed94dcf728255bd7a7d11c97c362930728fb81a3e91f1906eb6eec86e0fedb7", "question_id": "q03_exact_copy", "split": "test", "cluster_id": "syn_03009_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 8, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1976, "query_tokens": 421, "message_tokens": 339, "message_utf8_bytes": 945, "cache_tensor_bytes": null, "serialized_payload_bytes": 945, "side_information_bytes": 0, "fetched_bytes": 7560, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 106952.59120792616, "hops": {"1": 16602.87162498571, "2": 4677.791832946241, "3": 11236.125083058141, "4": 14879.588958923705, "5": 14824.957874952815, "6": 14889.837124967016, "7": 14804.55670808442, "8": 15036.862000008114}, "receiver_prefill_ms": 411.80304100271314, "ttft_ms": 37.99404203891754, "decode_ms": 156.89199999906123}, "total_wall_ms": 107559.97616588138, "ttft_ms": 37.99404203891754, "input_usage": {"lookups": 14}, "output_usage": {"output_tokens": 5}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "Idris Bergström", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 12, "retention_rate": 0.18461538461538463, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.2, "amount": 0.15384615384615385, "status": 0.5, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 945}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 12, "retention_rate": 0.18461538461538463, "retention_by_kind": {"person": 1.0, "date": 0.0, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 698}, "2": {"facts_total": 65, "facts_retained": 8, "retention_rate": 0.12307692307692308, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.0, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 310}, "3": {"facts_total": 65, "facts_retained": 12, "retention_rate": 0.18461538461538463, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.2, "amount": 0.15384615384615385, "status": 0.5, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 725}, "4": {"facts_total": 65, "facts_retained": 12, "retention_rate": 0.18461538461538463, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.2, "amount": 0.15384615384615385, "status": 0.5, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 945}, "5": {"facts_total": 65, "facts_retained": 12, "retention_rate": 0.18461538461538463, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.2, "amount": 0.15384615384615385, "status": 0.5, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 945}, "6": {"facts_total": 65, "facts_retained": 12, "retention_rate": 0.18461538461538463, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.2, "amount": 0.15384615384615385, "status": 0.5, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 945}, "7": {"facts_total": 65, "facts_retained": 12, "retention_rate": 0.18461538461538463, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.2, "amount": 0.15384615384615385, "status": 0.5, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 945}, "8": {"facts_total": 65, "facts_retained": 12, "retention_rate": 0.18461538461538463, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.2, "amount": 0.15384615384615385, "status": 0.5, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 945}}}, "reference_key_id": "keys/synthetic/syn_03009_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "855b6cb08da5888e0939be3cf8d9b3d7844b2cf024991e3f1c53d6e4e3638d06", "output_hash": "fd88832c9fd0ab175818d76d875d70156225bd4b34ec401514546ef70f4f826c"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "51e3055cba2407724f3eb76c", "attempt_id": "b2e7898de0e6", "run_group_id": "E4_core:A5|repairable|qwen3b:h8", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:24:04.490164Z", "task_family": "synthetic", "document_id": "syn_03009_2000", "document_hash": "6ed94dcf728255bd7a7d11c97c362930728fb81a3e91f1906eb6eec86e0fedb7", "question_id": "q04_multi_hop", "split": "test", "cluster_id": "syn_03009_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 8, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1976, "query_tokens": 429, "message_tokens": 339, "message_utf8_bytes": 945, "cache_tensor_bytes": null, "serialized_payload_bytes": 945, "side_information_bytes": 0, "fetched_bytes": 7560, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 106952.59120792616, "hops": {"1": 16602.87162498571, "2": 4677.791832946241, "3": 11236.125083058141, "4": 14879.588958923705, "5": 14824.957874952815, "6": 14889.837124967016, "7": 14804.55670808442, "8": 15036.862000008114}, "receiver_prefill_ms": 292.7512919995934, "ttft_ms": 38.80675008986145, "decode_ms": 190.0090000126511}, "total_wall_ms": 107474.75941595621, "ttft_ms": 38.80675008986145, "input_usage": {"lookups": 14}, "output_usage": {"output_tokens": 7}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "4670383", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 12, "retention_rate": 0.18461538461538463, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.2, "amount": 0.15384615384615385, "status": 0.5, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 945}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 12, "retention_rate": 0.18461538461538463, "retention_by_kind": {"person": 1.0, "date": 0.0, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 698}, "2": {"facts_total": 65, "facts_retained": 8, "retention_rate": 0.12307692307692308, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.0, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 310}, "3": {"facts_total": 65, "facts_retained": 12, "retention_rate": 0.18461538461538463, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.2, "amount": 0.15384615384615385, "status": 0.5, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 725}, "4": {"facts_total": 65, "facts_retained": 12, "retention_rate": 0.18461538461538463, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.2, "amount": 0.15384615384615385, "status": 0.5, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 945}, "5": {"facts_total": 65, "facts_retained": 12, "retention_rate": 0.18461538461538463, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.2, "amount": 0.15384615384615385, "status": 0.5, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 945}, "6": {"facts_total": 65, "facts_retained": 12, "retention_rate": 0.18461538461538463, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.2, "amount": 0.15384615384615385, "status": 0.5, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 945}, "7": {"facts_total": 65, "facts_retained": 12, "retention_rate": 0.18461538461538463, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.2, "amount": 0.15384615384615385, "status": 0.5, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 945}, "8": {"facts_total": 65, "facts_retained": 12, "retention_rate": 0.18461538461538463, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.2, "amount": 0.15384615384615385, "status": 0.5, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 945}}}, "reference_key_id": "keys/synthetic/syn_03009_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "855b6cb08da5888e0939be3cf8d9b3d7844b2cf024991e3f1c53d6e4e3638d06", "output_hash": "19ae8c357e5500bf494a6163bfbd11380b2996e8ea532a2c64f4a599464820f7"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "eadf4f4f5945c1da5c01884c", "attempt_id": "4eca1d6103a9", "run_group_id": "E4_core:A5|repairable|qwen3b:h8", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:24:05.030675Z", "task_family": "synthetic", "document_id": "syn_03009_2000", "document_hash": "6ed94dcf728255bd7a7d11c97c362930728fb81a3e91f1906eb6eec86e0fedb7", "question_id": "q05_multi_hop", "split": "test", "cluster_id": "syn_03009_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 8, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1976, "query_tokens": 432, "message_tokens": 339, "message_utf8_bytes": 945, "cache_tensor_bytes": null, "serialized_payload_bytes": 945, "side_information_bytes": 0, "fetched_bytes": 7560, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 106952.59120792616, "hops": {"1": 16602.87162498571, "2": 4677.791832946241, "3": 11236.125083058141, "4": 14879.588958923705, "5": 14824.957874952815, "6": 14889.837124967016, "7": 14804.55670808442, "8": 15036.862000008114}, "receiver_prefill_ms": 296.61249998025596, "ttft_ms": 36.93495795596391, "decode_ms": 55.355916963890195}, "total_wall_ms": 107342.220124905, "ttft_ms": 36.93495795596391, "input_usage": {"lookups": 14}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 12, "retention_rate": 0.18461538461538463, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.2, "amount": 0.15384615384615385, "status": 0.5, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 945}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 12, "retention_rate": 0.18461538461538463, "retention_by_kind": {"person": 1.0, "date": 0.0, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 698}, "2": {"facts_total": 65, "facts_retained": 8, "retention_rate": 0.12307692307692308, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.0, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 310}, "3": {"facts_total": 65, "facts_retained": 12, "retention_rate": 0.18461538461538463, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.2, "amount": 0.15384615384615385, "status": 0.5, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 725}, "4": {"facts_total": 65, "facts_retained": 12, "retention_rate": 0.18461538461538463, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.2, "amount": 0.15384615384615385, "status": 0.5, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 945}, "5": {"facts_total": 65, "facts_retained": 12, "retention_rate": 0.18461538461538463, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.2, "amount": 0.15384615384615385, "status": 0.5, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 945}, "6": {"facts_total": 65, "facts_retained": 12, "retention_rate": 0.18461538461538463, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.2, "amount": 0.15384615384615385, "status": 0.5, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 945}, "7": {"facts_total": 65, "facts_retained": 12, "retention_rate": 0.18461538461538463, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.2, "amount": 0.15384615384615385, "status": 0.5, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 945}, "8": {"facts_total": 65, "facts_retained": 12, "retention_rate": 0.18461538461538463, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.2, "amount": 0.15384615384615385, "status": 0.5, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 945}}}, "reference_key_id": "keys/synthetic/syn_03009_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "855b6cb08da5888e0939be3cf8d9b3d7844b2cf024991e3f1c53d6e4e3638d06", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "ce1ee2d93d7ef718bc73e9a0", "attempt_id": "8853f231342c", "run_group_id": "E4_core:A5|repairable|qwen3b:h8", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:24:05.440066Z", "task_family": "synthetic", "document_id": "syn_03009_2000", "document_hash": "6ed94dcf728255bd7a7d11c97c362930728fb81a3e91f1906eb6eec86e0fedb7", "question_id": "q06_negation", "split": "test", "cluster_id": "syn_03009_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 8, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1976, "query_tokens": 422, "message_tokens": 339, "message_utf8_bytes": 945, "cache_tensor_bytes": null, "serialized_payload_bytes": 945, "side_information_bytes": 0, "fetched_bytes": 7560, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 106952.59120792616, "hops": {"1": 16602.87162498571, "2": 4677.791832946241, "3": 11236.125083058141, "4": 14879.588958923705, "5": 14824.957874952815, "6": 14889.837124967016, "7": 14804.55670808442, "8": 15036.862000008114}, "receiver_prefill_ms": 433.1467079464346, "ttft_ms": 41.42404207959771, "decode_ms": 24.039999931119382}, "total_wall_ms": 107451.73566590529, "ttft_ms": 41.42404207959771, "input_usage": {"lookups": 14}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 12, "retention_rate": 0.18461538461538463, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.2, "amount": 0.15384615384615385, "status": 0.5, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 945}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 12, "retention_rate": 0.18461538461538463, "retention_by_kind": {"person": 1.0, "date": 0.0, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 698}, "2": {"facts_total": 65, "facts_retained": 8, "retention_rate": 0.12307692307692308, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.0, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 310}, "3": {"facts_total": 65, "facts_retained": 12, "retention_rate": 0.18461538461538463, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.2, "amount": 0.15384615384615385, "status": 0.5, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 725}, "4": {"facts_total": 65, "facts_retained": 12, "retention_rate": 0.18461538461538463, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.2, "amount": 0.15384615384615385, "status": 0.5, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 945}, "5": {"facts_total": 65, "facts_retained": 12, "retention_rate": 0.18461538461538463, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.2, "amount": 0.15384615384615385, "status": 0.5, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 945}, "6": {"facts_total": 65, "facts_retained": 12, "retention_rate": 0.18461538461538463, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.2, "amount": 0.15384615384615385, "status": 0.5, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 945}, "7": {"facts_total": 65, "facts_retained": 12, "retention_rate": 0.18461538461538463, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.2, "amount": 0.15384615384615385, "status": 0.5, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 945}, "8": {"facts_total": 65, "facts_retained": 12, "retention_rate": 0.18461538461538463, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.2, "amount": 0.15384615384615385, "status": 0.5, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 945}}}, "reference_key_id": "keys/synthetic/syn_03009_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "855b6cb08da5888e0939be3cf8d9b3d7844b2cf024991e3f1c53d6e4e3638d06", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "a32efa1bc47ab7641e015bf6", "attempt_id": "f74f37f5445d", "run_group_id": "E4_core:A5|repairable|qwen3b:h8", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:24:05.956906Z", "task_family": "synthetic", "document_id": "syn_03009_2000", "document_hash": "6ed94dcf728255bd7a7d11c97c362930728fb81a3e91f1906eb6eec86e0fedb7", "question_id": "q07_negation", "split": "test", "cluster_id": "syn_03009_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 8, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1976, "query_tokens": 421, "message_tokens": 339, "message_utf8_bytes": 945, "cache_tensor_bytes": null, "serialized_payload_bytes": 945, "side_information_bytes": 0, "fetched_bytes": 7560, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 106952.59120792616, "hops": {"1": 16602.87162498571, "2": 4677.791832946241, "3": 11236.125083058141, "4": 14879.588958923705, "5": 14824.957874952815, "6": 14889.837124967016, "7": 14804.55670808442, "8": 15036.862000008114}, "receiver_prefill_ms": 282.1082079317421, "ttft_ms": 24.781417101621628, "decode_ms": 37.460749968886375}, "total_wall_ms": 107297.46508295648, "ttft_ms": 24.781417101621628, "input_usage": {"lookups": 14}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 12, "retention_rate": 0.18461538461538463, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.2, "amount": 0.15384615384615385, "status": 0.5, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 945}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 12, "retention_rate": 0.18461538461538463, "retention_by_kind": {"person": 1.0, "date": 0.0, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 698}, "2": {"facts_total": 65, "facts_retained": 8, "retention_rate": 0.12307692307692308, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.0, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 310}, "3": {"facts_total": 65, "facts_retained": 12, "retention_rate": 0.18461538461538463, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.2, "amount": 0.15384615384615385, "status": 0.5, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 725}, "4": {"facts_total": 65, "facts_retained": 12, "retention_rate": 0.18461538461538463, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.2, "amount": 0.15384615384615385, "status": 0.5, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 945}, "5": {"facts_total": 65, "facts_retained": 12, "retention_rate": 0.18461538461538463, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.2, "amount": 0.15384615384615385, "status": 0.5, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 945}, "6": {"facts_total": 65, "facts_retained": 12, "retention_rate": 0.18461538461538463, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.2, "amount": 0.15384615384615385, "status": 0.5, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 945}, "7": {"facts_total": 65, "facts_retained": 12, "retention_rate": 0.18461538461538463, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.2, "amount": 0.15384615384615385, "status": 0.5, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 945}, "8": {"facts_total": 65, "facts_retained": 12, "retention_rate": 0.18461538461538463, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.2, "amount": 0.15384615384615385, "status": 0.5, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 945}}}, "reference_key_id": "keys/synthetic/syn_03009_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "855b6cb08da5888e0939be3cf8d9b3d7844b2cf024991e3f1c53d6e4e3638d06", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "57c4d3625fbe0f63d3146d30", "attempt_id": "cc6cfe2fde78", "run_group_id": "E4_core:A5|repairable|qwen3b:h8", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:24:06.319340Z", "task_family": "synthetic", "document_id": "syn_03009_2000", "document_hash": "6ed94dcf728255bd7a7d11c97c362930728fb81a3e91f1906eb6eec86e0fedb7", "question_id": "q08_revision", "split": "test", "cluster_id": "syn_03009_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 8, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1976, "query_tokens": 439, "message_tokens": 339, "message_utf8_bytes": 945, "cache_tensor_bytes": null, "serialized_payload_bytes": 945, "side_information_bytes": 0, "fetched_bytes": 7560, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 106952.59120792616, "hops": {"1": 16602.87162498571, "2": 4677.791832946241, "3": 11236.125083058141, "4": 14879.588958923705, "5": 14824.957874952815, "6": 14889.837124967016, "7": 14804.55670808442, "8": 15036.862000008114}, "receiver_prefill_ms": 292.8358749486506, "ttft_ms": 32.39533305168152, "decode_ms": 23.33470794837922}, "total_wall_ms": 107301.66295787785, "ttft_ms": 32.39533305168152, "input_usage": {"lookups": 14}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 12, "retention_rate": 0.18461538461538463, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.2, "amount": 0.15384615384615385, "status": 0.5, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 945}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 12, "retention_rate": 0.18461538461538463, "retention_by_kind": {"person": 1.0, "date": 0.0, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 698}, "2": {"facts_total": 65, "facts_retained": 8, "retention_rate": 0.12307692307692308, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.0, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 310}, "3": {"facts_total": 65, "facts_retained": 12, "retention_rate": 0.18461538461538463, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.2, "amount": 0.15384615384615385, "status": 0.5, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 725}, "4": {"facts_total": 65, "facts_retained": 12, "retention_rate": 0.18461538461538463, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.2, "amount": 0.15384615384615385, "status": 0.5, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 945}, "5": {"facts_total": 65, "facts_retained": 12, "retention_rate": 0.18461538461538463, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.2, "amount": 0.15384615384615385, "status": 0.5, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 945}, "6": {"facts_total": 65, "facts_retained": 12, "retention_rate": 0.18461538461538463, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.2, "amount": 0.15384615384615385, "status": 0.5, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 945}, "7": {"facts_total": 65, "facts_retained": 12, "retention_rate": 0.18461538461538463, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.2, "amount": 0.15384615384615385, "status": 0.5, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 945}, "8": {"facts_total": 65, "facts_retained": 12, "retention_rate": 0.18461538461538463, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.2, "amount": 0.15384615384615385, "status": 0.5, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 945}}}, "reference_key_id": "keys/synthetic/syn_03009_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "855b6cb08da5888e0939be3cf8d9b3d7844b2cf024991e3f1c53d6e4e3638d06", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "6f08cbdd74c55e1a6ee052d8", "attempt_id": "73032f4ab725", "run_group_id": "E4_core:A5|repairable|qwen3b:h8", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:24:06.686170Z", "task_family": "synthetic", "document_id": "syn_03009_2000", "document_hash": "6ed94dcf728255bd7a7d11c97c362930728fb81a3e91f1906eb6eec86e0fedb7", "question_id": "q09_revision", "split": "test", "cluster_id": "syn_03009_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 8, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1976, "query_tokens": 432, "message_tokens": 339, "message_utf8_bytes": 945, "cache_tensor_bytes": null, "serialized_payload_bytes": 945, "side_information_bytes": 0, "fetched_bytes": 7560, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 106952.59120792616, "hops": {"1": 16602.87162498571, "2": 4677.791832946241, "3": 11236.125083058141, "4": 14879.588958923705, "5": 14824.957874952815, "6": 14889.837124967016, "7": 14804.55670808442, "8": 15036.862000008114}, "receiver_prefill_ms": 270.242583937943, "ttft_ms": 57.08283407147974, "decode_ms": 58.87966696172953}, "total_wall_ms": 107339.53424892388, "ttft_ms": 57.08283407147974, "input_usage": {"lookups": 14}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 12, "retention_rate": 0.18461538461538463, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.2, "amount": 0.15384615384615385, "status": 0.5, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 945}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 12, "retention_rate": 0.18461538461538463, "retention_by_kind": {"person": 1.0, "date": 0.0, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 698}, "2": {"facts_total": 65, "facts_retained": 8, "retention_rate": 0.12307692307692308, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.0, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 310}, "3": {"facts_total": 65, "facts_retained": 12, "retention_rate": 0.18461538461538463, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.2, "amount": 0.15384615384615385, "status": 0.5, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 725}, "4": {"facts_total": 65, "facts_retained": 12, "retention_rate": 0.18461538461538463, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.2, "amount": 0.15384615384615385, "status": 0.5, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 945}, "5": {"facts_total": 65, "facts_retained": 12, "retention_rate": 0.18461538461538463, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.2, "amount": 0.15384615384615385, "status": 0.5, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 945}, "6": {"facts_total": 65, "facts_retained": 12, "retention_rate": 0.18461538461538463, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.2, "amount": 0.15384615384615385, "status": 0.5, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 945}, "7": {"facts_total": 65, "facts_retained": 12, "retention_rate": 0.18461538461538463, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.2, "amount": 0.15384615384615385, "status": 0.5, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 945}, "8": {"facts_total": 65, "facts_retained": 12, "retention_rate": 0.18461538461538463, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.2, "amount": 0.15384615384615385, "status": 0.5, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 945}}}, "reference_key_id": "keys/synthetic/syn_03009_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "855b6cb08da5888e0939be3cf8d9b3d7844b2cf024991e3f1c53d6e4e3638d06", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "245b2164b204e8d78d272a27", "attempt_id": "5d7d1aa780f1", "run_group_id": "E4_core:A5|repairable|qwen3b:h8", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:24:07.090390Z", "task_family": "synthetic", "document_id": "syn_03009_2000", "document_hash": "6ed94dcf728255bd7a7d11c97c362930728fb81a3e91f1906eb6eec86e0fedb7", "question_id": "q10_temporal", "split": "test", "cluster_id": "syn_03009_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 8, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1976, "query_tokens": 441, "message_tokens": 339, "message_utf8_bytes": 945, "cache_tensor_bytes": null, "serialized_payload_bytes": 945, "side_information_bytes": 0, "fetched_bytes": 7560, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 106952.59120792616, "hops": {"1": 16602.87162498571, "2": 4677.791832946241, "3": 11236.125083058141, "4": 14879.588958923705, "5": 14824.957874952815, "6": 14889.837124967016, "7": 14804.55670808442, "8": 15036.862000008114}, "receiver_prefill_ms": 444.3055420415476, "ttft_ms": 24.146458017639816, "decode_ms": 39.19024998322129}, "total_wall_ms": 107460.75691585429, "ttft_ms": 24.146458017639816, "input_usage": {"lookups": 14}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 12, "retention_rate": 0.18461538461538463, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.2, "amount": 0.15384615384615385, "status": 0.5, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 945}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 12, "retention_rate": 0.18461538461538463, "retention_by_kind": {"person": 1.0, "date": 0.0, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 698}, "2": {"facts_total": 65, "facts_retained": 8, "retention_rate": 0.12307692307692308, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.0, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 310}, "3": {"facts_total": 65, "facts_retained": 12, "retention_rate": 0.18461538461538463, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.2, "amount": 0.15384615384615385, "status": 0.5, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 725}, "4": {"facts_total": 65, "facts_retained": 12, "retention_rate": 0.18461538461538463, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.2, "amount": 0.15384615384615385, "status": 0.5, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 945}, "5": {"facts_total": 65, "facts_retained": 12, "retention_rate": 0.18461538461538463, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.2, "amount": 0.15384615384615385, "status": 0.5, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 945}, "6": {"facts_total": 65, "facts_retained": 12, "retention_rate": 0.18461538461538463, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.2, "amount": 0.15384615384615385, "status": 0.5, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 945}, "7": {"facts_total": 65, "facts_retained": 12, "retention_rate": 0.18461538461538463, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.2, "amount": 0.15384615384615385, "status": 0.5, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 945}, "8": {"facts_total": 65, "facts_retained": 12, "retention_rate": 0.18461538461538463, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.2, "amount": 0.15384615384615385, "status": 0.5, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 945}}}, "reference_key_id": "keys/synthetic/syn_03009_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "855b6cb08da5888e0939be3cf8d9b3d7844b2cf024991e3f1c53d6e4e3638d06", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "637750a797263a40e78f23b9", "attempt_id": "9e0ee72b4c0d", "run_group_id": "E4_core:A5|repairable|qwen3b:h8", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:24:07.615802Z", "task_family": "synthetic", "document_id": "syn_03009_2000", "document_hash": "6ed94dcf728255bd7a7d11c97c362930728fb81a3e91f1906eb6eec86e0fedb7", "question_id": "q11_aggregation", "split": "test", "cluster_id": "syn_03009_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 8, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1976, "query_tokens": 432, "message_tokens": 339, "message_utf8_bytes": 945, "cache_tensor_bytes": null, "serialized_payload_bytes": 945, "side_information_bytes": 0, "fetched_bytes": 7560, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 106952.59120792616, "hops": {"1": 16602.87162498571, "2": 4677.791832946241, "3": 11236.125083058141, "4": 14879.588958923705, "5": 14824.957874952815, "6": 14889.837124967016, "7": 14804.55670808442, "8": 15036.862000008114}, "receiver_prefill_ms": 309.7661250503734, "ttft_ms": 27.143624960444868, "decode_ms": 223.0064159957692}, "total_wall_ms": 107513.03666690364, "ttft_ms": 27.143624960444868, "input_usage": {"lookups": 14}, "output_usage": {"output_tokens": 7}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "5247196", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 12, "retention_rate": 0.18461538461538463, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.2, "amount": 0.15384615384615385, "status": 0.5, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 945}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 12, "retention_rate": 0.18461538461538463, "retention_by_kind": {"person": 1.0, "date": 0.0, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 698}, "2": {"facts_total": 65, "facts_retained": 8, "retention_rate": 0.12307692307692308, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.0, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 310}, "3": {"facts_total": 65, "facts_retained": 12, "retention_rate": 0.18461538461538463, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.2, "amount": 0.15384615384615385, "status": 0.5, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 725}, "4": {"facts_total": 65, "facts_retained": 12, "retention_rate": 0.18461538461538463, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.2, "amount": 0.15384615384615385, "status": 0.5, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 945}, "5": {"facts_total": 65, "facts_retained": 12, "retention_rate": 0.18461538461538463, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.2, "amount": 0.15384615384615385, "status": 0.5, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 945}, "6": {"facts_total": 65, "facts_retained": 12, "retention_rate": 0.18461538461538463, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.2, "amount": 0.15384615384615385, "status": 0.5, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 945}, "7": {"facts_total": 65, "facts_retained": 12, "retention_rate": 0.18461538461538463, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.2, "amount": 0.15384615384615385, "status": 0.5, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 945}, "8": {"facts_total": 65, "facts_retained": 12, "retention_rate": 0.18461538461538463, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.2, "amount": 0.15384615384615385, "status": 0.5, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 945}}}, "reference_key_id": "keys/synthetic/syn_03009_2000.json", "task_scores": {"correct": 0.0, "abstained": 0.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "855b6cb08da5888e0939be3cf8d9b3d7844b2cf024991e3f1c53d6e4e3638d06", "output_hash": "3ac48b1390a5593419a76c6409f1c54c01affa22edaa1be1850800277310a17c"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "367d84d85e09d3df7be9637d", "attempt_id": "93e9e752f10b", "run_group_id": "E4_core:A5|repairable|qwen3b:h8", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:24:08.194648Z", "task_family": "synthetic", "document_id": "syn_03009_2000", "document_hash": "6ed94dcf728255bd7a7d11c97c362930728fb81a3e91f1906eb6eec86e0fedb7", "question_id": "q12_unit_conversion", "split": "test", "cluster_id": "syn_03009_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 8, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1976, "query_tokens": 430, "message_tokens": 339, "message_utf8_bytes": 945, "cache_tensor_bytes": null, "serialized_payload_bytes": 945, "side_information_bytes": 0, "fetched_bytes": 7560, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 106952.59120792616, "hops": {"1": 16602.87162498571, "2": 4677.791832946241, "3": 11236.125083058141, "4": 14879.588958923705, "5": 14824.957874952815, "6": 14889.837124967016, "7": 14804.55670808442, "8": 15036.862000008114}, "receiver_prefill_ms": 264.24612500704825, "ttft_ms": 46.20279197115451, "decode_ms": 41.661833063699305}, "total_wall_ms": 107305.39899994619, "ttft_ms": 46.20279197115451, "input_usage": {"lookups": 14}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 12, "retention_rate": 0.18461538461538463, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.2, "amount": 0.15384615384615385, "status": 0.5, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 945}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 12, "retention_rate": 0.18461538461538463, "retention_by_kind": {"person": 1.0, "date": 0.0, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 698}, "2": {"facts_total": 65, "facts_retained": 8, "retention_rate": 0.12307692307692308, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.0, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 310}, "3": {"facts_total": 65, "facts_retained": 12, "retention_rate": 0.18461538461538463, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.2, "amount": 0.15384615384615385, "status": 0.5, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 725}, "4": {"facts_total": 65, "facts_retained": 12, "retention_rate": 0.18461538461538463, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.2, "amount": 0.15384615384615385, "status": 0.5, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 945}, "5": {"facts_total": 65, "facts_retained": 12, "retention_rate": 0.18461538461538463, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.2, "amount": 0.15384615384615385, "status": 0.5, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 945}, "6": {"facts_total": 65, "facts_retained": 12, "retention_rate": 0.18461538461538463, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.2, "amount": 0.15384615384615385, "status": 0.5, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 945}, "7": {"facts_total": 65, "facts_retained": 12, "retention_rate": 0.18461538461538463, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.2, "amount": 0.15384615384615385, "status": 0.5, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 945}, "8": {"facts_total": 65, "facts_retained": 12, "retention_rate": 0.18461538461538463, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.2, "amount": 0.15384615384615385, "status": 0.5, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 945}}}, "reference_key_id": "keys/synthetic/syn_03009_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "855b6cb08da5888e0939be3cf8d9b3d7844b2cf024991e3f1c53d6e4e3638d06", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "f59449f76deb41ace339b940", "attempt_id": "a7be31d670f6", "run_group_id": "E4_core:A5|repairable|qwen3b:h8", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:24:08.564923Z", "task_family": "synthetic", "document_id": "syn_03009_2000", "document_hash": "6ed94dcf728255bd7a7d11c97c362930728fb81a3e91f1906eb6eec86e0fedb7", "question_id": "q13_unanswerable", "split": "test", "cluster_id": "syn_03009_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 8, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1976, "query_tokens": 434, "message_tokens": 339, "message_utf8_bytes": 945, "cache_tensor_bytes": null, "serialized_payload_bytes": 945, "side_information_bytes": 0, "fetched_bytes": 7560, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 106952.59120792616, "hops": {"1": 16602.87162498571, "2": 4677.791832946241, "3": 11236.125083058141, "4": 14879.588958923705, "5": 14824.957874952815, "6": 14889.837124967016, "7": 14804.55670808442, "8": 15036.862000008114}, "receiver_prefill_ms": 412.2335829306394, "ttft_ms": 78.70979094877839, "decode_ms": 23.33933301270008}, "total_wall_ms": 107467.42695791181, "ttft_ms": 78.70979094877839, "input_usage": {"lookups": 14}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 12, "retention_rate": 0.18461538461538463, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.2, "amount": 0.15384615384615385, "status": 0.5, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 945}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 12, "retention_rate": 0.18461538461538463, "retention_by_kind": {"person": 1.0, "date": 0.0, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 698}, "2": {"facts_total": 65, "facts_retained": 8, "retention_rate": 0.12307692307692308, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.0, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 310}, "3": {"facts_total": 65, "facts_retained": 12, "retention_rate": 0.18461538461538463, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.2, "amount": 0.15384615384615385, "status": 0.5, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 725}, "4": {"facts_total": 65, "facts_retained": 12, "retention_rate": 0.18461538461538463, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.2, "amount": 0.15384615384615385, "status": 0.5, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 945}, "5": {"facts_total": 65, "facts_retained": 12, "retention_rate": 0.18461538461538463, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.2, "amount": 0.15384615384615385, "status": 0.5, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 945}, "6": {"facts_total": 65, "facts_retained": 12, "retention_rate": 0.18461538461538463, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.2, "amount": 0.15384615384615385, "status": 0.5, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 945}, "7": {"facts_total": 65, "facts_retained": 12, "retention_rate": 0.18461538461538463, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.2, "amount": 0.15384615384615385, "status": 0.5, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 945}, "8": {"facts_total": 65, "facts_retained": 12, "retention_rate": 0.18461538461538463, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.2, "amount": 0.15384615384615385, "status": 0.5, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 945}}}, "reference_key_id": "keys/synthetic/syn_03009_2000.json", "task_scores": {"correct": 1.0, "abstained": 1.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "855b6cb08da5888e0939be3cf8d9b3d7844b2cf024991e3f1c53d6e4e3638d06", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "690a89c1376489eb4c0f5654", "attempt_id": "227ed31b8ff7", "run_group_id": "E4_core:A5|repairable|qwen3b:h8", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:24:09.098367Z", "task_family": "synthetic", "document_id": "syn_03009_2000", "document_hash": "6ed94dcf728255bd7a7d11c97c362930728fb81a3e91f1906eb6eec86e0fedb7", "question_id": "q14_unanswerable", "split": "test", "cluster_id": "syn_03009_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 8, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1976, "query_tokens": 434, "message_tokens": 339, "message_utf8_bytes": 945, "cache_tensor_bytes": null, "serialized_payload_bytes": 945, "side_information_bytes": 0, "fetched_bytes": 7560, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 106952.59120792616, "hops": {"1": 16602.87162498571, "2": 4677.791832946241, "3": 11236.125083058141, "4": 14879.588958923705, "5": 14824.957874952815, "6": 14889.837124967016, "7": 14804.55670808442, "8": 15036.862000008114}, "receiver_prefill_ms": 295.2859590295702, "ttft_ms": 37.661875016056, "decode_ms": 27.1052080206573}, "total_wall_ms": 107313.18412395194, "ttft_ms": 37.661875016056, "input_usage": {"lookups": 14}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 12, "retention_rate": 0.18461538461538463, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.2, "amount": 0.15384615384615385, "status": 0.5, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 945}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 12, "retention_rate": 0.18461538461538463, "retention_by_kind": {"person": 1.0, "date": 0.0, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 698}, "2": {"facts_total": 65, "facts_retained": 8, "retention_rate": 0.12307692307692308, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.0, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 310}, "3": {"facts_total": 65, "facts_retained": 12, "retention_rate": 0.18461538461538463, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.2, "amount": 0.15384615384615385, "status": 0.5, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 725}, "4": {"facts_total": 65, "facts_retained": 12, "retention_rate": 0.18461538461538463, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.2, "amount": 0.15384615384615385, "status": 0.5, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 945}, "5": {"facts_total": 65, "facts_retained": 12, "retention_rate": 0.18461538461538463, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.2, "amount": 0.15384615384615385, "status": 0.5, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 945}, "6": {"facts_total": 65, "facts_retained": 12, "retention_rate": 0.18461538461538463, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.2, "amount": 0.15384615384615385, "status": 0.5, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 945}, "7": {"facts_total": 65, "facts_retained": 12, "retention_rate": 0.18461538461538463, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.2, "amount": 0.15384615384615385, "status": 0.5, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 945}, "8": {"facts_total": 65, "facts_retained": 12, "retention_rate": 0.18461538461538463, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.2, "amount": 0.15384615384615385, "status": 0.5, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 945}}}, "reference_key_id": "keys/synthetic/syn_03009_2000.json", "task_scores": {"correct": 1.0, "abstained": 1.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "855b6cb08da5888e0939be3cf8d9b3d7844b2cf024991e3f1c53d6e4e3638d06", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "e3d726c2e4e2b46bb0d90ab3", "attempt_id": "e85c8c7e242d", "run_group_id": "E4_core:A4|closed|qwen3b-llama3b:h1", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:24:32.033278Z", "task_family": "synthetic", "document_id": "syn_03009_2000", "document_hash": "6ed94dcf728255bd7a7d11c97c362930728fb81a3e91f1906eb6eec86e0fedb7", "question_id": "q00_single_fact", "split": "test", "cluster_id": "syn_03009_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 1, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1976, "query_tokens": 595, "message_tokens": 512, "message_utf8_bytes": 1123, "cache_tensor_bytes": null, "serialized_payload_bytes": 1123, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 22554.907249985263, "hops": {"1": 22554.907249985263}, "receiver_prefill_ms": 393.9372500171885, "ttft_ms": 41.7250000173226, "decode_ms": 504.9299169331789}, "total_wall_ms": 23496.35033297818, "ttft_ms": 41.7250000173226, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 10}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "PA-37F-6227", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 26, "retention_rate": 0.4, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.75, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1123}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 26, "retention_rate": 0.4, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.75, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1123}}}, "reference_key_id": "keys/synthetic/syn_03009_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "3bd2611f7bc59d6f4b6323c3ac1e9f7af93bcc7ed3e43397b183cdb4d6c6e058", "output_hash": "5b7645cf6e8140f4c0bf6ed4ad76121cf2924fe7e666d92161812527b86f4409"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "d8b10b402557fa9e4329f0c9", "attempt_id": "5610391289b8", "run_group_id": "E4_core:A4|closed|qwen3b-llama3b:h1", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:24:32.993364Z", "task_family": "synthetic", "document_id": "syn_03009_2000", "document_hash": "6ed94dcf728255bd7a7d11c97c362930728fb81a3e91f1906eb6eec86e0fedb7", "question_id": "q01_single_fact", "split": "test", "cluster_id": "syn_03009_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 1, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1976, "query_tokens": 600, "message_tokens": 512, "message_utf8_bytes": 1123, "cache_tensor_bytes": null, "serialized_payload_bytes": 1123, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 22554.907249985263, "hops": {"1": 22554.907249985263}, "receiver_prefill_ms": 396.0808329284191, "ttft_ms": 43.137166067026556, "decode_ms": 25.08875005878508}, "total_wall_ms": 23020.07924998179, "ttft_ms": 43.137166067026556, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 26, "retention_rate": 0.4, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.75, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1123}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 26, "retention_rate": 0.4, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.75, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1123}}}, "reference_key_id": "keys/synthetic/syn_03009_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "3bd2611f7bc59d6f4b6323c3ac1e9f7af93bcc7ed3e43397b183cdb4d6c6e058", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "ba6a66de448b9a381a108a1f", "attempt_id": "7d6c0a229127", "run_group_id": "E4_core:A4|closed|qwen3b-llama3b:h1", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:24:33.476769Z", "task_family": "synthetic", "document_id": "syn_03009_2000", "document_hash": "6ed94dcf728255bd7a7d11c97c362930728fb81a3e91f1906eb6eec86e0fedb7", "question_id": "q02_exact_copy", "split": "test", "cluster_id": "syn_03009_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 1, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1976, "query_tokens": 598, "message_tokens": 512, "message_utf8_bytes": 1123, "cache_tensor_bytes": null, "serialized_payload_bytes": 1123, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 22554.907249985263, "hops": {"1": 22554.907249985263}, "receiver_prefill_ms": 363.2935000350699, "ttft_ms": 57.23024997860193, "decode_ms": 53.40979201719165}, "total_wall_ms": 23029.77620891761, "ttft_ms": 57.23024997860193, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 26, "retention_rate": 0.4, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.75, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1123}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 26, "retention_rate": 0.4, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.75, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1123}}}, "reference_key_id": "keys/synthetic/syn_03009_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "3bd2611f7bc59d6f4b6323c3ac1e9f7af93bcc7ed3e43397b183cdb4d6c6e058", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "640c9564fe6f7c02f0a82018", "attempt_id": "f73be415c3dd", "run_group_id": "E4_core:A4|closed|qwen3b-llama3b:h1", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:24:33.970041Z", "task_family": "synthetic", "document_id": "syn_03009_2000", "document_hash": "6ed94dcf728255bd7a7d11c97c362930728fb81a3e91f1906eb6eec86e0fedb7", "question_id": "q03_exact_copy", "split": "test", "cluster_id": "syn_03009_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 1, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1976, "query_tokens": 594, "message_tokens": 512, "message_utf8_bytes": 1123, "cache_tensor_bytes": null, "serialized_payload_bytes": 1123, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 22554.907249985263, "hops": {"1": 22554.907249985263}, "receiver_prefill_ms": 576.4393750578165, "ttft_ms": 26.903833029791713, "decode_ms": 170.8090000320226}, "total_wall_ms": 23329.78562498465, "ttft_ms": 26.903833029791713, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 5}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "Idris Bergström", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 26, "retention_rate": 0.4, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.75, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1123}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 26, "retention_rate": 0.4, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.75, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1123}}}, "reference_key_id": "keys/synthetic/syn_03009_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "3bd2611f7bc59d6f4b6323c3ac1e9f7af93bcc7ed3e43397b183cdb4d6c6e058", "output_hash": "fd88832c9fd0ab175818d76d875d70156225bd4b34ec401514546ef70f4f826c"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "1bb20c747446969f4be3110f", "attempt_id": "5c6682e2b8aa", "run_group_id": "E4_core:A4|closed|qwen3b-llama3b:h1", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:24:34.765197Z", "task_family": "synthetic", "document_id": "syn_03009_2000", "document_hash": "6ed94dcf728255bd7a7d11c97c362930728fb81a3e91f1906eb6eec86e0fedb7", "question_id": "q04_multi_hop", "split": "test", "cluster_id": "syn_03009_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 1, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1976, "query_tokens": 602, "message_tokens": 512, "message_utf8_bytes": 1123, "cache_tensor_bytes": null, "serialized_payload_bytes": 1123, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 22554.907249985263, "hops": {"1": 22554.907249985263}, "receiver_prefill_ms": 402.8824999695644, "ttft_ms": 27.510375017300248, "decode_ms": 202.43245898745954}, "total_wall_ms": 23188.419915968552, "ttft_ms": 27.510375017300248, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 7}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "4670383", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 26, "retention_rate": 0.4, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.75, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1123}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 26, "retention_rate": 0.4, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.75, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1123}}}, "reference_key_id": "keys/synthetic/syn_03009_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "3bd2611f7bc59d6f4b6323c3ac1e9f7af93bcc7ed3e43397b183cdb4d6c6e058", "output_hash": "19ae8c357e5500bf494a6163bfbd11380b2996e8ea532a2c64f4a599464820f7"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "6f125f4c91a21411f17c77fe", "attempt_id": "673539f36767", "run_group_id": "E4_core:A4|closed|qwen3b-llama3b:h1", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:24:35.416498Z", "task_family": "synthetic", "document_id": "syn_03009_2000", "document_hash": "6ed94dcf728255bd7a7d11c97c362930728fb81a3e91f1906eb6eec86e0fedb7", "question_id": "q05_multi_hop", "split": "test", "cluster_id": "syn_03009_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 1, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1976, "query_tokens": 605, "message_tokens": 512, "message_utf8_bytes": 1123, "cache_tensor_bytes": null, "serialized_payload_bytes": 1123, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 22554.907249985263, "hops": {"1": 22554.907249985263}, "receiver_prefill_ms": 426.3156671077013, "ttft_ms": 44.2142499377951, "decode_ms": 118.49533300846815}, "total_wall_ms": 23144.58162500523, "ttft_ms": 44.2142499377951, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 26, "retention_rate": 0.4, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.75, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1123}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 26, "retention_rate": 0.4, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.75, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1123}}}, "reference_key_id": "keys/synthetic/syn_03009_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "3bd2611f7bc59d6f4b6323c3ac1e9f7af93bcc7ed3e43397b183cdb4d6c6e058", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "3e202135382ee2d8ebe9ca74", "attempt_id": "d9e73f2d58b6", "run_group_id": "E4_core:A4|closed|qwen3b-llama3b:h1", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:24:36.026065Z", "task_family": "synthetic", "document_id": "syn_03009_2000", "document_hash": "6ed94dcf728255bd7a7d11c97c362930728fb81a3e91f1906eb6eec86e0fedb7", "question_id": "q06_negation", "split": "test", "cluster_id": "syn_03009_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 1, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1976, "query_tokens": 595, "message_tokens": 512, "message_utf8_bytes": 1123, "cache_tensor_bytes": null, "serialized_payload_bytes": 1123, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 22554.907249985263, "hops": {"1": 22554.907249985263}, "receiver_prefill_ms": 487.4297910137102, "ttft_ms": 26.935749920085073, "decode_ms": 41.00641701370478}, "total_wall_ms": 23110.91074999422, "ttft_ms": 26.935749920085073, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 26, "retention_rate": 0.4, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.75, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1123}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 26, "retention_rate": 0.4, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.75, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1123}}}, "reference_key_id": "keys/synthetic/syn_03009_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "3bd2611f7bc59d6f4b6323c3ac1e9f7af93bcc7ed3e43397b183cdb4d6c6e058", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "3627e25c56de0a1c62e95c29", "attempt_id": "3e6415553714", "run_group_id": "E4_core:A4|closed|qwen3b-llama3b:h1", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:24:36.599330Z", "task_family": "synthetic", "document_id": "syn_03009_2000", "document_hash": "6ed94dcf728255bd7a7d11c97c362930728fb81a3e91f1906eb6eec86e0fedb7", "question_id": "q07_negation", "split": "test", "cluster_id": "syn_03009_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 1, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1976, "query_tokens": 594, "message_tokens": 512, "message_utf8_bytes": 1123, "cache_tensor_bytes": null, "serialized_payload_bytes": 1123, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 22554.907249985263, "hops": {"1": 22554.907249985263}, "receiver_prefill_ms": 392.4384160200134, "ttft_ms": 22.81191700603813, "decode_ms": 38.19600003771484}, "total_wall_ms": 23008.994332980365, "ttft_ms": 22.81191700603813, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 26, "retention_rate": 0.4, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.75, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1123}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 26, "retention_rate": 0.4, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.75, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1123}}}, "reference_key_id": "keys/synthetic/syn_03009_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "3bd2611f7bc59d6f4b6323c3ac1e9f7af93bcc7ed3e43397b183cdb4d6c6e058", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "835784761305352a5688854f", "attempt_id": "693a13673ee4", "run_group_id": "E4_core:A4|closed|qwen3b-llama3b:h1", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:24:37.070490Z", "task_family": "synthetic", "document_id": "syn_03009_2000", "document_hash": "6ed94dcf728255bd7a7d11c97c362930728fb81a3e91f1906eb6eec86e0fedb7", "question_id": "q08_revision", "split": "test", "cluster_id": "syn_03009_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 1, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1976, "query_tokens": 612, "message_tokens": 512, "message_utf8_bytes": 1123, "cache_tensor_bytes": null, "serialized_payload_bytes": 1123, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 22554.907249985263, "hops": {"1": 22554.907249985263}, "receiver_prefill_ms": 369.5096659939736, "ttft_ms": 35.67912499420345, "decode_ms": 500.7487089606002}, "total_wall_ms": 23461.591749917716, "ttft_ms": 35.67912499420345, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 8}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "10047105", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 26, "retention_rate": 0.4, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.75, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1123}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 26, "retention_rate": 0.4, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.75, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1123}}}, "reference_key_id": "keys/synthetic/syn_03009_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "3bd2611f7bc59d6f4b6323c3ac1e9f7af93bcc7ed3e43397b183cdb4d6c6e058", "output_hash": "bea99610d1e3e75a7f20939faf8e6824e02237b156dacd6442ba2a7e02617510"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "8cab7d0a963840134a4c98a2", "attempt_id": "1f43142e7d77", "run_group_id": "E4_core:A4|closed|qwen3b-llama3b:h1", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:24:37.994976Z", "task_family": "synthetic", "document_id": "syn_03009_2000", "document_hash": "6ed94dcf728255bd7a7d11c97c362930728fb81a3e91f1906eb6eec86e0fedb7", "question_id": "q09_revision", "split": "test", "cluster_id": "syn_03009_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 1, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1976, "query_tokens": 605, "message_tokens": 512, "message_utf8_bytes": 1123, "cache_tensor_bytes": null, "serialized_payload_bytes": 1123, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 22554.907249985263, "hops": {"1": 22554.907249985263}, "receiver_prefill_ms": 393.28962506260723, "ttft_ms": 37.08450007252395, "decode_ms": 254.4050420401618}, "total_wall_ms": 23240.262417006306, "ttft_ms": 37.08450007252395, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 8}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "10047105", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 26, "retention_rate": 0.4, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.75, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1123}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 26, "retention_rate": 0.4, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.75, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1123}}}, "reference_key_id": "keys/synthetic/syn_03009_2000.json", "task_scores": {"correct": 0.0, "abstained": 0.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "3bd2611f7bc59d6f4b6323c3ac1e9f7af93bcc7ed3e43397b183cdb4d6c6e058", "output_hash": "bea99610d1e3e75a7f20939faf8e6824e02237b156dacd6442ba2a7e02617510"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "b9ea673631db53c108d8e084", "attempt_id": "68a77935eff6", "run_group_id": "E4_core:A4|closed|qwen3b-llama3b:h1", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:24:38.698430Z", "task_family": "synthetic", "document_id": "syn_03009_2000", "document_hash": "6ed94dcf728255bd7a7d11c97c362930728fb81a3e91f1906eb6eec86e0fedb7", "question_id": "q10_temporal", "split": "test", "cluster_id": "syn_03009_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 1, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1976, "query_tokens": 614, "message_tokens": 512, "message_utf8_bytes": 1123, "cache_tensor_bytes": null, "serialized_payload_bytes": 1123, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 22554.907249985263, "hops": {"1": 22554.907249985263}, "receiver_prefill_ms": 470.233624917455, "ttft_ms": 47.69704199861735, "decode_ms": 65.21329202223569}, "total_wall_ms": 23138.741249917075, "ttft_ms": 47.69704199861735, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 26, "retention_rate": 0.4, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.75, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1123}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 26, "retention_rate": 0.4, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.75, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1123}}}, "reference_key_id": "keys/synthetic/syn_03009_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "3bd2611f7bc59d6f4b6323c3ac1e9f7af93bcc7ed3e43397b183cdb4d6c6e058", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "a002102b2df1207b23cdfc36", "attempt_id": "41816e902cb8", "run_group_id": "E4_core:A4|closed|qwen3b-llama3b:h1", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:24:39.300649Z", "task_family": "synthetic", "document_id": "syn_03009_2000", "document_hash": "6ed94dcf728255bd7a7d11c97c362930728fb81a3e91f1906eb6eec86e0fedb7", "question_id": "q11_aggregation", "split": "test", "cluster_id": "syn_03009_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 1, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1976, "query_tokens": 605, "message_tokens": 512, "message_utf8_bytes": 1123, "cache_tensor_bytes": null, "serialized_payload_bytes": 1123, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 22554.907249985263, "hops": {"1": 22554.907249985263}, "receiver_prefill_ms": 459.6968749538064, "ttft_ms": 49.75624999497086, "decode_ms": 437.70791694987565}, "total_wall_ms": 23502.794374944642, "ttft_ms": 49.75624999497086, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 7}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "5781336", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 26, "retention_rate": 0.4, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.75, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1123}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 26, "retention_rate": 0.4, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.75, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1123}}}, "reference_key_id": "keys/synthetic/syn_03009_2000.json", "task_scores": {"correct": 0.0, "abstained": 0.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "3bd2611f7bc59d6f4b6323c3ac1e9f7af93bcc7ed3e43397b183cdb4d6c6e058", "output_hash": "a365e04f74ba5baeccdd9e30dc27a1c5ff916597b828d874e63884b2090a60e2"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "7a96bfe572cbe7815e322eab", "attempt_id": "cc41fa71083b", "run_group_id": "E4_core:A4|closed|qwen3b-llama3b:h1", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:24:40.265959Z", "task_family": "synthetic", "document_id": "syn_03009_2000", "document_hash": "6ed94dcf728255bd7a7d11c97c362930728fb81a3e91f1906eb6eec86e0fedb7", "question_id": "q12_unit_conversion", "split": "test", "cluster_id": "syn_03009_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 1, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1976, "query_tokens": 603, "message_tokens": 512, "message_utf8_bytes": 1123, "cache_tensor_bytes": null, "serialized_payload_bytes": 1123, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 22554.907249985263, "hops": {"1": 22554.907249985263}, "receiver_prefill_ms": 459.528416977264, "ttft_ms": 87.31908292975277, "decode_ms": 61.26441701781005}, "total_wall_ms": 23163.661708007567, "ttft_ms": 87.31908292975277, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 26, "retention_rate": 0.4, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.75, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1123}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 26, "retention_rate": 0.4, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.75, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1123}}}, "reference_key_id": "keys/synthetic/syn_03009_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "3bd2611f7bc59d6f4b6323c3ac1e9f7af93bcc7ed3e43397b183cdb4d6c6e058", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "69d22299d4370e22f2099895", "attempt_id": "d03d3efe07d7", "run_group_id": "E4_core:A4|closed|qwen3b-llama3b:h1", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:24:40.891655Z", "task_family": "synthetic", "document_id": "syn_03009_2000", "document_hash": "6ed94dcf728255bd7a7d11c97c362930728fb81a3e91f1906eb6eec86e0fedb7", "question_id": "q13_unanswerable", "split": "test", "cluster_id": "syn_03009_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 1, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1976, "query_tokens": 607, "message_tokens": 512, "message_utf8_bytes": 1123, "cache_tensor_bytes": null, "serialized_payload_bytes": 1123, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 22554.907249985263, "hops": {"1": 22554.907249985263}, "receiver_prefill_ms": 477.9918330023065, "ttft_ms": 71.14449993241578, "decode_ms": 66.4560419972986}, "total_wall_ms": 23171.15049995482, "ttft_ms": 71.14449993241578, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 26, "retention_rate": 0.4, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.75, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1123}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 26, "retention_rate": 0.4, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.75, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1123}}}, "reference_key_id": "keys/synthetic/syn_03009_2000.json", "task_scores": {"correct": 1.0, "abstained": 1.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "3bd2611f7bc59d6f4b6323c3ac1e9f7af93bcc7ed3e43397b183cdb4d6c6e058", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "cc436e9e6ca8f9ded21558ed", "attempt_id": "13b34fbe1631", "run_group_id": "E4_core:A4|closed|qwen3b-llama3b:h1", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:24:41.525278Z", "task_family": "synthetic", "document_id": "syn_03009_2000", "document_hash": "6ed94dcf728255bd7a7d11c97c362930728fb81a3e91f1906eb6eec86e0fedb7", "question_id": "q14_unanswerable", "split": "test", "cluster_id": "syn_03009_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 1, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1976, "query_tokens": 607, "message_tokens": 512, "message_utf8_bytes": 1123, "cache_tensor_bytes": null, "serialized_payload_bytes": 1123, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 22554.907249985263, "hops": {"1": 22554.907249985263}, "receiver_prefill_ms": 461.35920798406005, "ttft_ms": 100.59879103209823, "decode_ms": 43.531541945412755}, "total_wall_ms": 23161.114874994382, "ttft_ms": 100.59879103209823, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 26, "retention_rate": 0.4, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.75, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1123}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 26, "retention_rate": 0.4, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.75, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1123}}}, "reference_key_id": "keys/synthetic/syn_03009_2000.json", "task_scores": {"correct": 1.0, "abstained": 1.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "3bd2611f7bc59d6f4b6323c3ac1e9f7af93bcc7ed3e43397b183cdb4d6c6e058", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "082950448296a5ec2e175d0f", "attempt_id": "6d6421fd6cfb", "run_group_id": "E4_core:A4|closed|qwen3b-llama3b:h2", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:24:59.460926Z", "task_family": "synthetic", "document_id": "syn_03009_2000", "document_hash": "6ed94dcf728255bd7a7d11c97c362930728fb81a3e91f1906eb6eec86e0fedb7", "question_id": "q00_single_fact", "split": "test", "cluster_id": "syn_03009_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "tokenizer_hashes": {"receiver": "b9d8872cf3a99da99e1fc4b0f71232a451632dc49ca4c6771549b4efa5c1c763"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 2, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1770, "query_tokens": 528, "message_tokens": 423, "message_utf8_bytes": 1113, "cache_tensor_bytes": null, "serialized_payload_bytes": 1113, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 39865.23141700309, "hops": {"1": 22554.907249985263, "2": 17310.324167017825}, "receiver_prefill_ms": 421.11674998886883, "ttft_ms": 32.44462492875755, "decode_ms": 207.515083020553}, "total_wall_ms": 40526.96583408397, "ttft_ms": 32.44462492875755, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 7}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "PA-37F-6227", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 26, "retention_rate": 0.4, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.75, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1113}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 26, "retention_rate": 0.4, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.75, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1123}, "2": {"facts_total": 65, "facts_retained": 26, "retention_rate": 0.4, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.75, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1113}}}, "reference_key_id": "keys/synthetic/syn_03009_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "6e47c1b22533e450537a2f87b59ca7a7f4059ee74c7a609ecf8fe329a68fcc9a", "output_hash": "5b7645cf6e8140f4c0bf6ed4ad76121cf2924fe7e666d92161812527b86f4409"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "fd17c9bd5cbe352f645330cd", "attempt_id": "a824a76a7374", "run_group_id": "E4_core:A4|closed|qwen3b-llama3b:h2", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:25:00.142986Z", "task_family": "synthetic", "document_id": "syn_03009_2000", "document_hash": "6ed94dcf728255bd7a7d11c97c362930728fb81a3e91f1906eb6eec86e0fedb7", "question_id": "q01_single_fact", "split": "test", "cluster_id": "syn_03009_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "tokenizer_hashes": {"receiver": "b9d8872cf3a99da99e1fc4b0f71232a451632dc49ca4c6771549b4efa5c1c763"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 2, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1770, "query_tokens": 533, "message_tokens": 423, "message_utf8_bytes": 1113, "cache_tensor_bytes": null, "serialized_payload_bytes": 1113, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 39865.23141700309, "hops": {"1": 22554.907249985263, "2": 17310.324167017825}, "receiver_prefill_ms": 338.3352920645848, "ttft_ms": 35.86512501351535, "decode_ms": 22.164874942973256}, "total_wall_ms": 40262.222082936205, "ttft_ms": 35.86512501351535, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 26, "retention_rate": 0.4, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.75, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1113}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 26, "retention_rate": 0.4, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.75, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1123}, "2": {"facts_total": 65, "facts_retained": 26, "retention_rate": 0.4, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.75, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1113}}}, "reference_key_id": "keys/synthetic/syn_03009_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "6e47c1b22533e450537a2f87b59ca7a7f4059ee74c7a609ecf8fe329a68fcc9a", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "e1bb0abc0bfe2fcbad1a55cc", "attempt_id": "697f5fa1f6a5", "run_group_id": "E4_core:A4|closed|qwen3b-llama3b:h2", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:25:00.558089Z", "task_family": "synthetic", "document_id": "syn_03009_2000", "document_hash": "6ed94dcf728255bd7a7d11c97c362930728fb81a3e91f1906eb6eec86e0fedb7", "question_id": "q02_exact_copy", "split": "test", "cluster_id": "syn_03009_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "tokenizer_hashes": {"receiver": "b9d8872cf3a99da99e1fc4b0f71232a451632dc49ca4c6771549b4efa5c1c763"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 2, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1770, "query_tokens": 532, "message_tokens": 423, "message_utf8_bytes": 1113, "cache_tensor_bytes": null, "serialized_payload_bytes": 1113, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 39865.23141700309, "hops": {"1": 22554.907249985263, "2": 17310.324167017825}, "receiver_prefill_ms": 303.5245419014245, "ttft_ms": 34.048665896989405, "decode_ms": 34.51195906382054}, "total_wall_ms": 40238.05695795454, "ttft_ms": 34.048665896989405, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 26, "retention_rate": 0.4, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.75, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1113}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 26, "retention_rate": 0.4, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.75, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1123}, "2": {"facts_total": 65, "facts_retained": 26, "retention_rate": 0.4, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.75, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1113}}}, "reference_key_id": "keys/synthetic/syn_03009_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "6e47c1b22533e450537a2f87b59ca7a7f4059ee74c7a609ecf8fe329a68fcc9a", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "eb4d51f7de0f565e4864fad1", "attempt_id": "e68d7dbac261", "run_group_id": "E4_core:A4|closed|qwen3b-llama3b:h2", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:25:00.949840Z", "task_family": "synthetic", "document_id": "syn_03009_2000", "document_hash": "6ed94dcf728255bd7a7d11c97c362930728fb81a3e91f1906eb6eec86e0fedb7", "question_id": "q03_exact_copy", "split": "test", "cluster_id": "syn_03009_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "tokenizer_hashes": {"receiver": "b9d8872cf3a99da99e1fc4b0f71232a451632dc49ca4c6771549b4efa5c1c763"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 2, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1770, "query_tokens": 527, "message_tokens": 423, "message_utf8_bytes": 1113, "cache_tensor_bytes": null, "serialized_payload_bytes": 1113, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 39865.23141700309, "hops": {"1": 22554.907249985263, "2": 17310.324167017825}, "receiver_prefill_ms": 461.50412503629923, "ttft_ms": 35.93599994201213, "decode_ms": 140.39212500210851}, "total_wall_ms": 40503.68795893155, "ttft_ms": 35.93599994201213, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 5}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "Idris Bergström", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 26, "retention_rate": 0.4, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.75, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1113}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 26, "retention_rate": 0.4, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.75, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1123}, "2": {"facts_total": 65, "facts_retained": 26, "retention_rate": 0.4, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.75, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1113}}}, "reference_key_id": "keys/synthetic/syn_03009_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "6e47c1b22533e450537a2f87b59ca7a7f4059ee74c7a609ecf8fe329a68fcc9a", "output_hash": "fd88832c9fd0ab175818d76d875d70156225bd4b34ec401514546ef70f4f826c"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "efc67ebf47210ee9d785bd79", "attempt_id": "aa9bd0bc9ec7", "run_group_id": "E4_core:A4|closed|qwen3b-llama3b:h2", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:25:01.605564Z", "task_family": "synthetic", "document_id": "syn_03009_2000", "document_hash": "6ed94dcf728255bd7a7d11c97c362930728fb81a3e91f1906eb6eec86e0fedb7", "question_id": "q04_multi_hop", "split": "test", "cluster_id": "syn_03009_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "tokenizer_hashes": {"receiver": "b9d8872cf3a99da99e1fc4b0f71232a451632dc49ca4c6771549b4efa5c1c763"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 2, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1770, "query_tokens": 536, "message_tokens": 423, "message_utf8_bytes": 1113, "cache_tensor_bytes": null, "serialized_payload_bytes": 1113, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 39865.23141700309, "hops": {"1": 22554.907249985263, "2": 17310.324167017825}, "receiver_prefill_ms": 334.84700007829815, "ttft_ms": 38.19337498862296, "decode_ms": 149.10325000528246}, "total_wall_ms": 40387.97241693828, "ttft_ms": 38.19337498862296, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 5}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "4,670,383", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 26, "retention_rate": 0.4, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.75, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1113}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 26, "retention_rate": 0.4, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.75, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1123}, "2": {"facts_total": 65, "facts_retained": 26, "retention_rate": 0.4, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.75, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1113}}}, "reference_key_id": "keys/synthetic/syn_03009_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "6e47c1b22533e450537a2f87b59ca7a7f4059ee74c7a609ecf8fe329a68fcc9a", "output_hash": "9510633f2e349a1cedecda578083181521646f70ca5ae08bae4b9188f3bd4a3b"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "2a6ff6f3f673bcc57a843a97", "attempt_id": "d776619ccd29", "run_group_id": "E4_core:A4|closed|qwen3b-llama3b:h2", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:25:02.145684Z", "task_family": "synthetic", "document_id": "syn_03009_2000", "document_hash": "6ed94dcf728255bd7a7d11c97c362930728fb81a3e91f1906eb6eec86e0fedb7", "question_id": "q05_multi_hop", "split": "test", "cluster_id": "syn_03009_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "tokenizer_hashes": {"receiver": "b9d8872cf3a99da99e1fc4b0f71232a451632dc49ca4c6771549b4efa5c1c763"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 2, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1770, "query_tokens": 537, "message_tokens": 423, "message_utf8_bytes": 1113, "cache_tensor_bytes": null, "serialized_payload_bytes": 1113, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 39865.23141700309, "hops": {"1": 22554.907249985263, "2": 17310.324167017825}, "receiver_prefill_ms": 319.98404196929187, "ttft_ms": 39.99808290973306, "decode_ms": 40.51950003486127}, "total_wall_ms": 40266.36825001333, "ttft_ms": 39.99808290973306, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 26, "retention_rate": 0.4, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.75, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1113}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 26, "retention_rate": 0.4, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.75, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1123}, "2": {"facts_total": 65, "facts_retained": 26, "retention_rate": 0.4, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.75, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1113}}}, "reference_key_id": "keys/synthetic/syn_03009_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "6e47c1b22533e450537a2f87b59ca7a7f4059ee74c7a609ecf8fe329a68fcc9a", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "57a4b8df199f35ad646c0e07", "attempt_id": "174c26303884", "run_group_id": "E4_core:A4|closed|qwen3b-llama3b:h2", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:25:02.564376Z", "task_family": "synthetic", "document_id": "syn_03009_2000", "document_hash": "6ed94dcf728255bd7a7d11c97c362930728fb81a3e91f1906eb6eec86e0fedb7", "question_id": "q06_negation", "split": "test", "cluster_id": "syn_03009_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "tokenizer_hashes": {"receiver": "b9d8872cf3a99da99e1fc4b0f71232a451632dc49ca4c6771549b4efa5c1c763"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 2, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1770, "query_tokens": 529, "message_tokens": 423, "message_utf8_bytes": 1113, "cache_tensor_bytes": null, "serialized_payload_bytes": 1113, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 39865.23141700309, "hops": {"1": 22554.907249985263, "2": 17310.324167017825}, "receiver_prefill_ms": 397.77962502557784, "ttft_ms": 135.49174997024238, "decode_ms": 59.81925001833588}, "total_wall_ms": 40458.955167094246, "ttft_ms": 135.49174997024238, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 2}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "No.", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 26, "retention_rate": 0.4, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.75, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1113}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 26, "retention_rate": 0.4, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.75, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1123}, "2": {"facts_total": 65, "facts_retained": 26, "retention_rate": 0.4, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.75, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1113}}}, "reference_key_id": "keys/synthetic/syn_03009_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "6e47c1b22533e450537a2f87b59ca7a7f4059ee74c7a609ecf8fe329a68fcc9a", "output_hash": "38178a20b470cfd18299fb1593dd4ba706f1a83321dac051d01086efd8b7a96f"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "77b61fa84439342043c155f7", "attempt_id": "d4a403edd669", "run_group_id": "E4_core:A4|closed|qwen3b-llama3b:h2", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:25:03.175908Z", "task_family": "synthetic", "document_id": "syn_03009_2000", "document_hash": "6ed94dcf728255bd7a7d11c97c362930728fb81a3e91f1906eb6eec86e0fedb7", "question_id": "q07_negation", "split": "test", "cluster_id": "syn_03009_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "tokenizer_hashes": {"receiver": "b9d8872cf3a99da99e1fc4b0f71232a451632dc49ca4c6771549b4efa5c1c763"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 2, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1770, "query_tokens": 528, "message_tokens": 423, "message_utf8_bytes": 1113, "cache_tensor_bytes": null, "serialized_payload_bytes": 1113, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 39865.23141700309, "hops": {"1": 22554.907249985263, "2": 17310.324167017825}, "receiver_prefill_ms": 334.840124938637, "ttft_ms": 31.329916906543076, "decode_ms": 70.5750830238685}, "total_wall_ms": 40302.68491699826, "ttft_ms": 31.329916906543076, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 2}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "No.", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 26, "retention_rate": 0.4, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.75, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1113}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 26, "retention_rate": 0.4, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.75, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1123}, "2": {"facts_total": 65, "facts_retained": 26, "retention_rate": 0.4, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.75, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1113}}}, "reference_key_id": "keys/synthetic/syn_03009_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "6e47c1b22533e450537a2f87b59ca7a7f4059ee74c7a609ecf8fe329a68fcc9a", "output_hash": "38178a20b470cfd18299fb1593dd4ba706f1a83321dac051d01086efd8b7a96f"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "a97af0a3405ab5599c2ebe52", "attempt_id": "8e4eacc23531", "run_group_id": "E4_core:A4|closed|qwen3b-llama3b:h2", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:25:03.631117Z", "task_family": "synthetic", "document_id": "syn_03009_2000", "document_hash": "6ed94dcf728255bd7a7d11c97c362930728fb81a3e91f1906eb6eec86e0fedb7", "question_id": "q08_revision", "split": "test", "cluster_id": "syn_03009_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "tokenizer_hashes": {"receiver": "b9d8872cf3a99da99e1fc4b0f71232a451632dc49ca4c6771549b4efa5c1c763"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 2, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1770, "query_tokens": 545, "message_tokens": 423, "message_utf8_bytes": 1113, "cache_tensor_bytes": null, "serialized_payload_bytes": 1113, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 39865.23141700309, "hops": {"1": 22554.907249985263, "2": 17310.324167017825}, "receiver_prefill_ms": 350.2776249079034, "ttft_ms": 32.74991700891405, "decode_ms": 151.1119999922812}, "total_wall_ms": 40400.12491692323, "ttft_ms": 32.74991700891405, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 5}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "10,047,105", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 26, "retention_rate": 0.4, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.75, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1113}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 26, "retention_rate": 0.4, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.75, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1123}, "2": {"facts_total": 65, "facts_retained": 26, "retention_rate": 0.4, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.75, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1113}}}, "reference_key_id": "keys/synthetic/syn_03009_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "6e47c1b22533e450537a2f87b59ca7a7f4059ee74c7a609ecf8fe329a68fcc9a", "output_hash": "45ed933586262d1470caf31c89c80a05759b79bd4e620737608286d4591bcde5"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "165bc508c4c815cdf683639c", "attempt_id": "b515212215c5", "run_group_id": "E4_core:A4|closed|qwen3b-llama3b:h2", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:25:04.184295Z", "task_family": "synthetic", "document_id": "syn_03009_2000", "document_hash": "6ed94dcf728255bd7a7d11c97c362930728fb81a3e91f1906eb6eec86e0fedb7", "question_id": "q09_revision", "split": "test", "cluster_id": "syn_03009_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "tokenizer_hashes": {"receiver": "b9d8872cf3a99da99e1fc4b0f71232a451632dc49ca4c6771549b4efa5c1c763"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 2, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1770, "query_tokens": 538, "message_tokens": 423, "message_utf8_bytes": 1113, "cache_tensor_bytes": null, "serialized_payload_bytes": 1113, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 39865.23141700309, "hops": {"1": 22554.907249985263, "2": 17310.324167017825}, "receiver_prefill_ms": 346.43879102077335, "ttft_ms": 35.69570800755173, "decode_ms": 320.1733329333365}, "total_wall_ms": 40568.27266700566, "ttft_ms": 35.69570800755173, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 5}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "8,227,110", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 26, "retention_rate": 0.4, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.75, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1113}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 26, "retention_rate": 0.4, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.75, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1123}, "2": {"facts_total": 65, "facts_retained": 26, "retention_rate": 0.4, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.75, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1113}}}, "reference_key_id": "keys/synthetic/syn_03009_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "6e47c1b22533e450537a2f87b59ca7a7f4059ee74c7a609ecf8fe329a68fcc9a", "output_hash": "c3aa599fbb00899728f00a77f6ec2d981fcc6387f12136cbb65878e6772f4889"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "370e022ac27d603a0a88ecf2", "attempt_id": "ab08aef5db02", "run_group_id": "E4_core:A4|closed|qwen3b-llama3b:h2", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:25:04.905258Z", "task_family": "synthetic", "document_id": "syn_03009_2000", "document_hash": "6ed94dcf728255bd7a7d11c97c362930728fb81a3e91f1906eb6eec86e0fedb7", "question_id": "q10_temporal", "split": "test", "cluster_id": "syn_03009_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "tokenizer_hashes": {"receiver": "b9d8872cf3a99da99e1fc4b0f71232a451632dc49ca4c6771549b4efa5c1c763"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 2, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1770, "query_tokens": 546, "message_tokens": 423, "message_utf8_bytes": 1113, "cache_tensor_bytes": null, "serialized_payload_bytes": 1113, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 39865.23141700309, "hops": {"1": 22554.907249985263, "2": 17310.324167017825}, "receiver_prefill_ms": 349.15987495332956, "ttft_ms": 25.71125002577901, "decode_ms": 33.42325007542968}, "total_wall_ms": 40274.375625071116, "ttft_ms": 25.71125002577901, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 26, "retention_rate": 0.4, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.75, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1113}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 26, "retention_rate": 0.4, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.75, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1123}, "2": {"facts_total": 65, "facts_retained": 26, "retention_rate": 0.4, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.75, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1113}}}, "reference_key_id": "keys/synthetic/syn_03009_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "6e47c1b22533e450537a2f87b59ca7a7f4059ee74c7a609ecf8fe329a68fcc9a", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "f0c54c6eced56f99623a2374", "attempt_id": "175a15358444", "run_group_id": "E4_core:A4|closed|qwen3b-llama3b:h2", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:25:05.332451Z", "task_family": "synthetic", "document_id": "syn_03009_2000", "document_hash": "6ed94dcf728255bd7a7d11c97c362930728fb81a3e91f1906eb6eec86e0fedb7", "question_id": "q11_aggregation", "split": "test", "cluster_id": "syn_03009_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "tokenizer_hashes": {"receiver": "b9d8872cf3a99da99e1fc4b0f71232a451632dc49ca4c6771549b4efa5c1c763"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 2, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1770, "query_tokens": 539, "message_tokens": 423, "message_utf8_bytes": 1113, "cache_tensor_bytes": null, "serialized_payload_bytes": 1113, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 39865.23141700309, "hops": {"1": 22554.907249985263, "2": 17310.324167017825}, "receiver_prefill_ms": 350.99891701247543, "ttft_ms": 36.67908301576972, "decode_ms": 162.87291597109288}, "total_wall_ms": 40416.44749999978, "ttft_ms": 36.67908301576972, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 5}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "5,247,196", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 26, "retention_rate": 0.4, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.75, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1113}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 26, "retention_rate": 0.4, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.75, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1123}, "2": {"facts_total": 65, "facts_retained": 26, "retention_rate": 0.4, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.75, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1113}}}, "reference_key_id": "keys/synthetic/syn_03009_2000.json", "task_scores": {"correct": 0.0, "abstained": 0.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "6e47c1b22533e450537a2f87b59ca7a7f4059ee74c7a609ecf8fe329a68fcc9a", "output_hash": "40e869c2b62e27bc00d9598728026ccfc12fd1b97b0864921ab5373cfc6a82f7"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "eee1c7539e3866fe6e2bd8ca", "attempt_id": "f6f71122da9b", "run_group_id": "E4_core:A4|closed|qwen3b-llama3b:h2", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:25:05.901817Z", "task_family": "synthetic", "document_id": "syn_03009_2000", "document_hash": "6ed94dcf728255bd7a7d11c97c362930728fb81a3e91f1906eb6eec86e0fedb7", "question_id": "q12_unit_conversion", "split": "test", "cluster_id": "syn_03009_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "tokenizer_hashes": {"receiver": "b9d8872cf3a99da99e1fc4b0f71232a451632dc49ca4c6771549b4efa5c1c763"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 2, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1770, "query_tokens": 536, "message_tokens": 423, "message_utf8_bytes": 1113, "cache_tensor_bytes": null, "serialized_payload_bytes": 1113, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 39865.23141700309, "hops": {"1": 22554.907249985263, "2": 17310.324167017825}, "receiver_prefill_ms": 401.98029205203056, "ttft_ms": 50.18483393359929, "decode_ms": 122.08299990743399}, "total_wall_ms": 40440.1554169599, "ttft_ms": 50.18483393359929, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 26, "retention_rate": 0.4, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.75, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1113}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 26, "retention_rate": 0.4, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.75, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1123}, "2": {"facts_total": 65, "facts_retained": 26, "retention_rate": 0.4, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.75, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1113}}}, "reference_key_id": "keys/synthetic/syn_03009_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "6e47c1b22533e450537a2f87b59ca7a7f4059ee74c7a609ecf8fe329a68fcc9a", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "6e6692bf64a95f83d8e6a1c1", "attempt_id": "858deefb3c79", "run_group_id": "E4_core:A4|closed|qwen3b-llama3b:h2", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:25:06.495644Z", "task_family": "synthetic", "document_id": "syn_03009_2000", "document_hash": "6ed94dcf728255bd7a7d11c97c362930728fb81a3e91f1906eb6eec86e0fedb7", "question_id": "q13_unanswerable", "split": "test", "cluster_id": "syn_03009_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "tokenizer_hashes": {"receiver": "b9d8872cf3a99da99e1fc4b0f71232a451632dc49ca4c6771549b4efa5c1c763"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 2, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1770, "query_tokens": 540, "message_tokens": 423, "message_utf8_bytes": 1113, "cache_tensor_bytes": null, "serialized_payload_bytes": 1113, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 39865.23141700309, "hops": {"1": 22554.907249985263, "2": 17310.324167017825}, "receiver_prefill_ms": 330.22725000046194, "ttft_ms": 38.953166105784476, "decode_ms": 214.6396249299869}, "total_wall_ms": 40449.79904196225, "ttft_ms": 38.953166105784476, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 7}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "ZN-56T-7644", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 26, "retention_rate": 0.4, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.75, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1113}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 26, "retention_rate": 0.4, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.75, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1123}, "2": {"facts_total": 65, "facts_retained": 26, "retention_rate": 0.4, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.75, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1113}}}, "reference_key_id": "keys/synthetic/syn_03009_2000.json", "task_scores": {"correct": 0.0, "abstained": 0.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "6e47c1b22533e450537a2f87b59ca7a7f4059ee74c7a609ecf8fe329a68fcc9a", "output_hash": "88fcc7902d9821f9d642d72ea09ae89e37399bee8006477f26f821b7675edd7d"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "b112252dcc51e6036668040b", "attempt_id": "c990f21ca107", "run_group_id": "E4_core:A4|closed|qwen3b-llama3b:h2", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:25:07.098024Z", "task_family": "synthetic", "document_id": "syn_03009_2000", "document_hash": "6ed94dcf728255bd7a7d11c97c362930728fb81a3e91f1906eb6eec86e0fedb7", "question_id": "q14_unanswerable", "split": "test", "cluster_id": "syn_03009_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "tokenizer_hashes": {"receiver": "b9d8872cf3a99da99e1fc4b0f71232a451632dc49ca4c6771549b4efa5c1c763"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 2, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1770, "query_tokens": 541, "message_tokens": 423, "message_utf8_bytes": 1113, "cache_tensor_bytes": null, "serialized_payload_bytes": 1113, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 39865.23141700309, "hops": {"1": 22554.907249985263, "2": 17310.324167017825}, "receiver_prefill_ms": 335.5975419981405, "ttft_ms": 19.749999977648258, "decode_ms": 19.405541010200977}, "total_wall_ms": 40240.636542090215, "ttft_ms": 19.749999977648258, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 26, "retention_rate": 0.4, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.75, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1113}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 26, "retention_rate": 0.4, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.75, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1123}, "2": {"facts_total": 65, "facts_retained": 26, "retention_rate": 0.4, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.75, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1113}}}, "reference_key_id": "keys/synthetic/syn_03009_2000.json", "task_scores": {"correct": 1.0, "abstained": 1.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "6e47c1b22533e450537a2f87b59ca7a7f4059ee74c7a609ecf8fe329a68fcc9a", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "a74b522801f12675bb2f4d1c", "attempt_id": "162cfd4808fe", "run_group_id": "E4_core:A4|closed|qwen3b-llama3b:h4", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:25:44.679089Z", "task_family": "synthetic", "document_id": "syn_03009_2000", "document_hash": "6ed94dcf728255bd7a7d11c97c362930728fb81a3e91f1906eb6eec86e0fedb7", "question_id": "q00_single_fact", "split": "test", "cluster_id": "syn_03009_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "tokenizer_hashes": {"receiver": "b9d8872cf3a99da99e1fc4b0f71232a451632dc49ca4c6771549b4efa5c1c763"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 4, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1770, "query_tokens": 498, "message_tokens": 393, "message_utf8_bytes": 1066, "cache_tensor_bytes": null, "serialized_payload_bytes": 1066, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 77048.86541806627, "hops": {"1": 22554.907249985263, "2": 17310.324167017825, "3": 21701.9699170487, "4": 15481.664084014483}, "receiver_prefill_ms": 423.330666963011, "ttft_ms": 127.28129199240357, "decode_ms": 228.78187499009073}, "total_wall_ms": 77829.08746006433, "ttft_ms": 127.28129199240357, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 7}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "PA-37F-6227", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 26, "retention_rate": 0.4, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.75, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1066}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 26, "retention_rate": 0.4, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.75, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1123}, "2": {"facts_total": 65, "facts_retained": 26, "retention_rate": 0.4, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.75, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1113}, "3": {"facts_total": 65, "facts_retained": 26, "retention_rate": 0.4, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.75, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1108}, "4": {"facts_total": 65, "facts_retained": 26, "retention_rate": 0.4, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.75, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1066}}}, "reference_key_id": "keys/synthetic/syn_03009_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "36c67ce106ada2d33fb25da80f134ed643482cf81159dbb244d8fe4be5e25d39", "output_hash": "5b7645cf6e8140f4c0bf6ed4ad76121cf2924fe7e666d92161812527b86f4409"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "2629ca4b2d4d07cabef23bad", "attempt_id": "6a02e4f41a78", "run_group_id": "E4_core:A4|closed|qwen3b-llama3b:h4", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:25:45.478475Z", "task_family": "synthetic", "document_id": "syn_03009_2000", "document_hash": "6ed94dcf728255bd7a7d11c97c362930728fb81a3e91f1906eb6eec86e0fedb7", "question_id": "q01_single_fact", "split": "test", "cluster_id": "syn_03009_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "tokenizer_hashes": {"receiver": "b9d8872cf3a99da99e1fc4b0f71232a451632dc49ca4c6771549b4efa5c1c763"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 4, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1770, "query_tokens": 503, "message_tokens": 393, "message_utf8_bytes": 1066, "cache_tensor_bytes": null, "serialized_payload_bytes": 1066, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 77048.86541806627, "hops": {"1": 22554.907249985263, "2": 17310.324167017825, "3": 21701.9699170487, "4": 15481.664084014483}, "receiver_prefill_ms": 383.1864999374375, "ttft_ms": 26.5969589818269, "decode_ms": 36.114374990575016}, "total_wall_ms": 77495.375626022, "ttft_ms": 26.5969589818269, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 26, "retention_rate": 0.4, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.75, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1066}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 26, "retention_rate": 0.4, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.75, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1123}, "2": {"facts_total": 65, "facts_retained": 26, "retention_rate": 0.4, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.75, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1113}, "3": {"facts_total": 65, "facts_retained": 26, "retention_rate": 0.4, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.75, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1108}, "4": {"facts_total": 65, "facts_retained": 26, "retention_rate": 0.4, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.75, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1066}}}, "reference_key_id": "keys/synthetic/syn_03009_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "36c67ce106ada2d33fb25da80f134ed643482cf81159dbb244d8fe4be5e25d39", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "815ac91fb100fe8cf241b467", "attempt_id": "40cdd42afb78", "run_group_id": "E4_core:A4|closed|qwen3b-llama3b:h4", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:25:45.942670Z", "task_family": "synthetic", "document_id": "syn_03009_2000", "document_hash": "6ed94dcf728255bd7a7d11c97c362930728fb81a3e91f1906eb6eec86e0fedb7", "question_id": "q02_exact_copy", "split": "test", "cluster_id": "syn_03009_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "tokenizer_hashes": {"receiver": "b9d8872cf3a99da99e1fc4b0f71232a451632dc49ca4c6771549b4efa5c1c763"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 4, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1770, "query_tokens": 502, "message_tokens": 393, "message_utf8_bytes": 1066, "cache_tensor_bytes": null, "serialized_payload_bytes": 1066, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 77048.86541806627, "hops": {"1": 22554.907249985263, "2": 17310.324167017825, "3": 21701.9699170487, "4": 15481.664084014483}, "receiver_prefill_ms": 347.27254195604473, "ttft_ms": 43.00858301576227, "decode_ms": 50.97149999346584}, "total_wall_ms": 77490.85575109348, "ttft_ms": 43.00858301576227, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 26, "retention_rate": 0.4, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.75, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1066}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 26, "retention_rate": 0.4, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.75, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1123}, "2": {"facts_total": 65, "facts_retained": 26, "retention_rate": 0.4, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.75, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1113}, "3": {"facts_total": 65, "facts_retained": 26, "retention_rate": 0.4, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.75, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1108}, "4": {"facts_total": 65, "facts_retained": 26, "retention_rate": 0.4, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.75, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1066}}}, "reference_key_id": "keys/synthetic/syn_03009_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "36c67ce106ada2d33fb25da80f134ed643482cf81159dbb244d8fe4be5e25d39", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "ac947847d40db31522596f81", "attempt_id": "1248c344414c", "run_group_id": "E4_core:A4|closed|qwen3b-llama3b:h4", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:25:46.402020Z", "task_family": "synthetic", "document_id": "syn_03009_2000", "document_hash": "6ed94dcf728255bd7a7d11c97c362930728fb81a3e91f1906eb6eec86e0fedb7", "question_id": "q03_exact_copy", "split": "test", "cluster_id": "syn_03009_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "tokenizer_hashes": {"receiver": "b9d8872cf3a99da99e1fc4b0f71232a451632dc49ca4c6771549b4efa5c1c763"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 4, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1770, "query_tokens": 497, "message_tokens": 393, "message_utf8_bytes": 1066, "cache_tensor_bytes": null, "serialized_payload_bytes": 1066, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 77048.86541806627, "hops": {"1": 22554.907249985263, "2": 17310.324167017825, "3": 21701.9699170487, "4": 15481.664084014483}, "receiver_prefill_ms": 456.9458329351619, "ttft_ms": 48.50954096764326, "decode_ms": 148.98791699670255}, "total_wall_ms": 77703.89033516403, "ttft_ms": 48.50954096764326, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 5}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "Idris Bergström", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 26, "retention_rate": 0.4, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.75, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1066}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 26, "retention_rate": 0.4, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.75, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1123}, "2": {"facts_total": 65, "facts_retained": 26, "retention_rate": 0.4, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.75, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1113}, "3": {"facts_total": 65, "facts_retained": 26, "retention_rate": 0.4, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.75, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1108}, "4": {"facts_total": 65, "facts_retained": 26, "retention_rate": 0.4, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.75, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1066}}}, "reference_key_id": "keys/synthetic/syn_03009_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "36c67ce106ada2d33fb25da80f134ed643482cf81159dbb244d8fe4be5e25d39", "output_hash": "fd88832c9fd0ab175818d76d875d70156225bd4b34ec401514546ef70f4f826c"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "84c64c8ab47e04d9d0aade5f", "attempt_id": "5ce490b8fa3e", "run_group_id": "E4_core:A4|closed|qwen3b-llama3b:h4", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:25:47.075360Z", "task_family": "synthetic", "document_id": "syn_03009_2000", "document_hash": "6ed94dcf728255bd7a7d11c97c362930728fb81a3e91f1906eb6eec86e0fedb7", "question_id": "q04_multi_hop", "split": "test", "cluster_id": "syn_03009_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "tokenizer_hashes": {"receiver": "b9d8872cf3a99da99e1fc4b0f71232a451632dc49ca4c6771549b4efa5c1c763"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 4, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1770, "query_tokens": 506, "message_tokens": 393, "message_utf8_bytes": 1066, "cache_tensor_bytes": null, "serialized_payload_bytes": 1066, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 77048.86541806627, "hops": {"1": 22554.907249985263, "2": 17310.324167017825, "3": 21701.9699170487, "4": 15481.664084014483}, "receiver_prefill_ms": 366.9449590379372, "ttft_ms": 21.715709008276463, "decode_ms": 159.61174992844462}, "total_wall_ms": 77597.74521016516, "ttft_ms": 21.715709008276463, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 5}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "4,670,383", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 26, "retention_rate": 0.4, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.75, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1066}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 26, "retention_rate": 0.4, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.75, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1123}, "2": {"facts_total": 65, "facts_retained": 26, "retention_rate": 0.4, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.75, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1113}, "3": {"facts_total": 65, "facts_retained": 26, "retention_rate": 0.4, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.75, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1108}, "4": {"facts_total": 65, "facts_retained": 26, "retention_rate": 0.4, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.75, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1066}}}, "reference_key_id": "keys/synthetic/syn_03009_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "36c67ce106ada2d33fb25da80f134ed643482cf81159dbb244d8fe4be5e25d39", "output_hash": "9510633f2e349a1cedecda578083181521646f70ca5ae08bae4b9188f3bd4a3b"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "530d9b1869f803341cbba7ec", "attempt_id": "4dbc39d60422", "run_group_id": "E4_core:A4|closed|qwen3b-llama3b:h4", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:25:47.641860Z", "task_family": "synthetic", "document_id": "syn_03009_2000", "document_hash": "6ed94dcf728255bd7a7d11c97c362930728fb81a3e91f1906eb6eec86e0fedb7", "question_id": "q05_multi_hop", "split": "test", "cluster_id": "syn_03009_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "tokenizer_hashes": {"receiver": "b9d8872cf3a99da99e1fc4b0f71232a451632dc49ca4c6771549b4efa5c1c763"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 4, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1770, "query_tokens": 507, "message_tokens": 393, "message_utf8_bytes": 1066, "cache_tensor_bytes": null, "serialized_payload_bytes": 1066, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 77048.86541806627, "hops": {"1": 22554.907249985263, "2": 17310.324167017825, "3": 21701.9699170487, "4": 15481.664084014483}, "receiver_prefill_ms": 357.1236670250073, "ttft_ms": 42.70545905455947, "decode_ms": 57.68549989443272}, "total_wall_ms": 77507.27175211068, "ttft_ms": 42.70545905455947, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "Unknown", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 26, "retention_rate": 0.4, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.75, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1066}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 26, "retention_rate": 0.4, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.75, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1123}, "2": {"facts_total": 65, "facts_retained": 26, "retention_rate": 0.4, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.75, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1113}, "3": {"facts_total": 65, "facts_retained": 26, "retention_rate": 0.4, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.75, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1108}, "4": {"facts_total": 65, "facts_retained": 26, "retention_rate": 0.4, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.75, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1066}}}, "reference_key_id": "keys/synthetic/syn_03009_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "36c67ce106ada2d33fb25da80f134ed643482cf81159dbb244d8fe4be5e25d39", "output_hash": "b764cdc0eab7137467211272fa539f1260d1bf2e71bcf6ff3bdc960f5c16aa14"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "164f48aa90757ce026d72a1c", "attempt_id": "20fea194810e", "run_group_id": "E4_core:A4|closed|qwen3b-llama3b:h4", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:25:48.118735Z", "task_family": "synthetic", "document_id": "syn_03009_2000", "document_hash": "6ed94dcf728255bd7a7d11c97c362930728fb81a3e91f1906eb6eec86e0fedb7", "question_id": "q06_negation", "split": "test", "cluster_id": "syn_03009_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "tokenizer_hashes": {"receiver": "b9d8872cf3a99da99e1fc4b0f71232a451632dc49ca4c6771549b4efa5c1c763"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 4, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1770, "query_tokens": 499, "message_tokens": 393, "message_utf8_bytes": 1066, "cache_tensor_bytes": null, "serialized_payload_bytes": 1066, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 77048.86541806627, "hops": {"1": 22554.907249985263, "2": 17310.324167017825, "3": 21701.9699170487, "4": 15481.664084014483}, "receiver_prefill_ms": 501.61799998022616, "ttft_ms": 39.80845806654543, "decode_ms": 66.5536659071222}, "total_wall_ms": 77657.44196006563, "ttft_ms": 39.80845806654543, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 2}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "No.", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 26, "retention_rate": 0.4, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.75, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1066}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 26, "retention_rate": 0.4, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.75, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1123}, "2": {"facts_total": 65, "facts_retained": 26, "retention_rate": 0.4, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.75, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1113}, "3": {"facts_total": 65, "facts_retained": 26, "retention_rate": 0.4, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.75, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1108}, "4": {"facts_total": 65, "facts_retained": 26, "retention_rate": 0.4, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.75, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1066}}}, "reference_key_id": "keys/synthetic/syn_03009_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "36c67ce106ada2d33fb25da80f134ed643482cf81159dbb244d8fe4be5e25d39", "output_hash": "38178a20b470cfd18299fb1593dd4ba706f1a83321dac051d01086efd8b7a96f"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "925f95953fee3056718de6b7", "attempt_id": "164dcf0b770d", "run_group_id": "E4_core:A4|closed|qwen3b-llama3b:h4", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:25:48.745134Z", "task_family": "synthetic", "document_id": "syn_03009_2000", "document_hash": "6ed94dcf728255bd7a7d11c97c362930728fb81a3e91f1906eb6eec86e0fedb7", "question_id": "q07_negation", "split": "test", "cluster_id": "syn_03009_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "tokenizer_hashes": {"receiver": "b9d8872cf3a99da99e1fc4b0f71232a451632dc49ca4c6771549b4efa5c1c763"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 4, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1770, "query_tokens": 498, "message_tokens": 393, "message_utf8_bytes": 1066, "cache_tensor_bytes": null, "serialized_payload_bytes": 1066, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 77048.86541806627, "hops": {"1": 22554.907249985263, "2": 17310.324167017825, "3": 21701.9699170487, "4": 15481.664084014483}, "receiver_prefill_ms": 360.04704202059656, "ttft_ms": 37.00708399992436, "decode_ms": 68.25170910451561}, "total_wall_ms": 77514.79416806251, "ttft_ms": 37.00708399992436, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 2}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "No.", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 26, "retention_rate": 0.4, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.75, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1066}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 26, "retention_rate": 0.4, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.75, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1123}, "2": {"facts_total": 65, "facts_retained": 26, "retention_rate": 0.4, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.75, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1113}, "3": {"facts_total": 65, "facts_retained": 26, "retention_rate": 0.4, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.75, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1108}, "4": {"facts_total": 65, "facts_retained": 26, "retention_rate": 0.4, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.75, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1066}}}, "reference_key_id": "keys/synthetic/syn_03009_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "36c67ce106ada2d33fb25da80f134ed643482cf81159dbb244d8fe4be5e25d39", "output_hash": "38178a20b470cfd18299fb1593dd4ba706f1a83321dac051d01086efd8b7a96f"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "803969a5f4b9b89b77e23ae2", "attempt_id": "228d33028896", "run_group_id": "E4_core:A4|closed|qwen3b-llama3b:h4", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:25:49.229251Z", "task_family": "synthetic", "document_id": "syn_03009_2000", "document_hash": "6ed94dcf728255bd7a7d11c97c362930728fb81a3e91f1906eb6eec86e0fedb7", "question_id": "q08_revision", "split": "test", "cluster_id": "syn_03009_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "tokenizer_hashes": {"receiver": "b9d8872cf3a99da99e1fc4b0f71232a451632dc49ca4c6771549b4efa5c1c763"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 4, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1770, "query_tokens": 515, "message_tokens": 393, "message_utf8_bytes": 1066, "cache_tensor_bytes": null, "serialized_payload_bytes": 1066, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 77048.86541806627, "hops": {"1": 22554.907249985263, "2": 17310.324167017825, "3": 21701.9699170487, "4": 15481.664084014483}, "receiver_prefill_ms": 324.12079197820276, "ttft_ms": 55.39716698694974, "decode_ms": 323.03608302026987}, "total_wall_ms": 77752.32271011919, "ttft_ms": 55.39716698694974, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 5}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "10,047,105", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 26, "retention_rate": 0.4, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.75, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1066}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 26, "retention_rate": 0.4, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.75, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1123}, "2": {"facts_total": 65, "facts_retained": 26, "retention_rate": 0.4, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.75, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1113}, "3": {"facts_total": 65, "facts_retained": 26, "retention_rate": 0.4, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.75, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1108}, "4": {"facts_total": 65, "facts_retained": 26, "retention_rate": 0.4, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.75, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1066}}}, "reference_key_id": "keys/synthetic/syn_03009_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "36c67ce106ada2d33fb25da80f134ed643482cf81159dbb244d8fe4be5e25d39", "output_hash": "45ed933586262d1470caf31c89c80a05759b79bd4e620737608286d4591bcde5"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "e4e23702e1a94645570414a0", "attempt_id": "2c92ce58989e", "run_group_id": "E4_core:A4|closed|qwen3b-llama3b:h4", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:25:49.950655Z", "task_family": "synthetic", "document_id": "syn_03009_2000", "document_hash": "6ed94dcf728255bd7a7d11c97c362930728fb81a3e91f1906eb6eec86e0fedb7", "question_id": "q09_revision", "split": "test", "cluster_id": "syn_03009_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "tokenizer_hashes": {"receiver": "b9d8872cf3a99da99e1fc4b0f71232a451632dc49ca4c6771549b4efa5c1c763"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 4, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1770, "query_tokens": 508, "message_tokens": 393, "message_utf8_bytes": 1066, "cache_tensor_bytes": null, "serialized_payload_bytes": 1066, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 77048.86541806627, "hops": {"1": 22554.907249985263, "2": 17310.324167017825, "3": 21701.9699170487, "4": 15481.664084014483}, "receiver_prefill_ms": 434.1127499938011, "ttft_ms": 31.515332986600697, "decode_ms": 159.31987506337464}, "total_wall_ms": 77674.38962613232, "ttft_ms": 31.515332986600697, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 5}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "8,227,110", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 26, "retention_rate": 0.4, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.75, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1066}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 26, "retention_rate": 0.4, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.75, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1123}, "2": {"facts_total": 65, "facts_retained": 26, "retention_rate": 0.4, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.75, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1113}, "3": {"facts_total": 65, "facts_retained": 26, "retention_rate": 0.4, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.75, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1108}, "4": {"facts_total": 65, "facts_retained": 26, "retention_rate": 0.4, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.75, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1066}}}, "reference_key_id": "keys/synthetic/syn_03009_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "36c67ce106ada2d33fb25da80f134ed643482cf81159dbb244d8fe4be5e25d39", "output_hash": "c3aa599fbb00899728f00a77f6ec2d981fcc6387f12136cbb65878e6772f4889"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "e7dd1b99ee419a9a215db39e", "attempt_id": "87ee9c98a1bc", "run_group_id": "E4_core:A4|closed|qwen3b-llama3b:h4", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:25:50.594238Z", "task_family": "synthetic", "document_id": "syn_03009_2000", "document_hash": "6ed94dcf728255bd7a7d11c97c362930728fb81a3e91f1906eb6eec86e0fedb7", "question_id": "q10_temporal", "split": "test", "cluster_id": "syn_03009_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "tokenizer_hashes": {"receiver": "b9d8872cf3a99da99e1fc4b0f71232a451632dc49ca4c6771549b4efa5c1c763"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 4, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1770, "query_tokens": 516, "message_tokens": 393, "message_utf8_bytes": 1066, "cache_tensor_bytes": null, "serialized_payload_bytes": 1066, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 77048.86541806627, "hops": {"1": 22554.907249985263, "2": 17310.324167017825, "3": 21701.9699170487, "4": 15481.664084014483}, "receiver_prefill_ms": 336.5305420011282, "ttft_ms": 31.62016603164375, "decode_ms": 28.445999952964485}, "total_wall_ms": 77445.9926681593, "ttft_ms": 31.62016603164375, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 26, "retention_rate": 0.4, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.75, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1066}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 26, "retention_rate": 0.4, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.75, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1123}, "2": {"facts_total": 65, "facts_retained": 26, "retention_rate": 0.4, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.75, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1113}, "3": {"facts_total": 65, "facts_retained": 26, "retention_rate": 0.4, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.75, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1108}, "4": {"facts_total": 65, "facts_retained": 26, "retention_rate": 0.4, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.75, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1066}}}, "reference_key_id": "keys/synthetic/syn_03009_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "36c67ce106ada2d33fb25da80f134ed643482cf81159dbb244d8fe4be5e25d39", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "71f3d3ab61836ea3e8c869ef", "attempt_id": "c4dab6664403", "run_group_id": "E4_core:A4|closed|qwen3b-llama3b:h4", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:25:51.009030Z", "task_family": "synthetic", "document_id": "syn_03009_2000", "document_hash": "6ed94dcf728255bd7a7d11c97c362930728fb81a3e91f1906eb6eec86e0fedb7", "question_id": "q11_aggregation", "split": "test", "cluster_id": "syn_03009_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "tokenizer_hashes": {"receiver": "b9d8872cf3a99da99e1fc4b0f71232a451632dc49ca4c6771549b4efa5c1c763"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 4, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1770, "query_tokens": 509, "message_tokens": 393, "message_utf8_bytes": 1066, "cache_tensor_bytes": null, "serialized_payload_bytes": 1066, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 77048.86541806627, "hops": {"1": 22554.907249985263, "2": 17310.324167017825, "3": 21701.9699170487, "4": 15481.664084014483}, "receiver_prefill_ms": 328.7174589931965, "ttft_ms": 53.27787494752556, "decode_ms": 355.9078749967739}, "total_wall_ms": 77787.52608504146, "ttft_ms": 53.27787494752556, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 5}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "5,247,196", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 26, "retention_rate": 0.4, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.75, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1066}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 26, "retention_rate": 0.4, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.75, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1123}, "2": {"facts_total": 65, "facts_retained": 26, "retention_rate": 0.4, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.75, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1113}, "3": {"facts_total": 65, "facts_retained": 26, "retention_rate": 0.4, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.75, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1108}, "4": {"facts_total": 65, "facts_retained": 26, "retention_rate": 0.4, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.75, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1066}}}, "reference_key_id": "keys/synthetic/syn_03009_2000.json", "task_scores": {"correct": 0.0, "abstained": 0.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "36c67ce106ada2d33fb25da80f134ed643482cf81159dbb244d8fe4be5e25d39", "output_hash": "40e869c2b62e27bc00d9598728026ccfc12fd1b97b0864921ab5373cfc6a82f7"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "05fcb8ae112366cb5d7a8862", "attempt_id": "1e1d686afc11", "run_group_id": "E4_core:A4|closed|qwen3b-llama3b:h4", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:25:51.765397Z", "task_family": "synthetic", "document_id": "syn_03009_2000", "document_hash": "6ed94dcf728255bd7a7d11c97c362930728fb81a3e91f1906eb6eec86e0fedb7", "question_id": "q12_unit_conversion", "split": "test", "cluster_id": "syn_03009_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "tokenizer_hashes": {"receiver": "b9d8872cf3a99da99e1fc4b0f71232a451632dc49ca4c6771549b4efa5c1c763"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 4, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1770, "query_tokens": 506, "message_tokens": 393, "message_utf8_bytes": 1066, "cache_tensor_bytes": null, "serialized_payload_bytes": 1066, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 77048.86541806627, "hops": {"1": 22554.907249985263, "2": 17310.324167017825, "3": 21701.9699170487, "4": 15481.664084014483}, "receiver_prefill_ms": 389.60866699926555, "ttft_ms": 39.15399999823421, "decode_ms": 24.330749991349876}, "total_wall_ms": 77502.52758408897, "ttft_ms": 39.15399999823421, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 26, "retention_rate": 0.4, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.75, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1066}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 26, "retention_rate": 0.4, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.75, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1123}, "2": {"facts_total": 65, "facts_retained": 26, "retention_rate": 0.4, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.75, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1113}, "3": {"facts_total": 65, "facts_retained": 26, "retention_rate": 0.4, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.75, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1108}, "4": {"facts_total": 65, "facts_retained": 26, "retention_rate": 0.4, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.75, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1066}}}, "reference_key_id": "keys/synthetic/syn_03009_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "36c67ce106ada2d33fb25da80f134ed643482cf81159dbb244d8fe4be5e25d39", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "177f826bbe3029af65390c48", "attempt_id": "eec31d4bb5c1", "run_group_id": "E4_core:A4|closed|qwen3b-llama3b:h4", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:25:52.237302Z", "task_family": "synthetic", "document_id": "syn_03009_2000", "document_hash": "6ed94dcf728255bd7a7d11c97c362930728fb81a3e91f1906eb6eec86e0fedb7", "question_id": "q13_unanswerable", "split": "test", "cluster_id": "syn_03009_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "tokenizer_hashes": {"receiver": "b9d8872cf3a99da99e1fc4b0f71232a451632dc49ca4c6771549b4efa5c1c763"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 4, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1770, "query_tokens": 510, "message_tokens": 393, "message_utf8_bytes": 1066, "cache_tensor_bytes": null, "serialized_payload_bytes": 1066, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 77048.86541806627, "hops": {"1": 22554.907249985263, "2": 17310.324167017825, "3": 21701.9699170487, "4": 15481.664084014483}, "receiver_prefill_ms": 373.3096669893712, "ttft_ms": 19.238834036514163, "decode_ms": 42.214915971271694}, "total_wall_ms": 77484.435126069, "ttft_ms": 19.238834036514163, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 2}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "Unknown.", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 26, "retention_rate": 0.4, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.75, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1066}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 26, "retention_rate": 0.4, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.75, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1123}, "2": {"facts_total": 65, "facts_retained": 26, "retention_rate": 0.4, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.75, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1113}, "3": {"facts_total": 65, "facts_retained": 26, "retention_rate": 0.4, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.75, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1108}, "4": {"facts_total": 65, "facts_retained": 26, "retention_rate": 0.4, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.75, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1066}}}, "reference_key_id": "keys/synthetic/syn_03009_2000.json", "task_scores": {"correct": 1.0, "abstained": 1.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "36c67ce106ada2d33fb25da80f134ed643482cf81159dbb244d8fe4be5e25d39", "output_hash": "9cf9f50925b2a7803c73826a305235d4215dc428721c731072a9f2c0f5d6ae5a"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "fa1c6d80a3bfef742dd8d808", "attempt_id": "6078eef94850", "run_group_id": "E4_core:A4|closed|qwen3b-llama3b:h4", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:25:52.690909Z", "task_family": "synthetic", "document_id": "syn_03009_2000", "document_hash": "6ed94dcf728255bd7a7d11c97c362930728fb81a3e91f1906eb6eec86e0fedb7", "question_id": "q14_unanswerable", "split": "test", "cluster_id": "syn_03009_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "tokenizer_hashes": {"receiver": "b9d8872cf3a99da99e1fc4b0f71232a451632dc49ca4c6771549b4efa5c1c763"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 4, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1770, "query_tokens": 511, "message_tokens": 393, "message_utf8_bytes": 1066, "cache_tensor_bytes": null, "serialized_payload_bytes": 1066, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 77048.86541806627, "hops": {"1": 22554.907249985263, "2": 17310.324167017825, "3": 21701.9699170487, "4": 15481.664084014483}, "receiver_prefill_ms": 342.03270892612636, "ttft_ms": 61.71287503093481, "decode_ms": 106.14745796192437}, "total_wall_ms": 77559.51333511621, "ttft_ms": 61.71287503093481, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 26, "retention_rate": 0.4, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.75, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1066}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 26, "retention_rate": 0.4, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.75, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1123}, "2": {"facts_total": 65, "facts_retained": 26, "retention_rate": 0.4, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.75, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1113}, "3": {"facts_total": 65, "facts_retained": 26, "retention_rate": 0.4, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.75, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1108}, "4": {"facts_total": 65, "facts_retained": 26, "retention_rate": 0.4, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.75, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1066}}}, "reference_key_id": "keys/synthetic/syn_03009_2000.json", "task_scores": {"correct": 1.0, "abstained": 1.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "36c67ce106ada2d33fb25da80f134ed643482cf81159dbb244d8fe4be5e25d39", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "1ab7d254a5acc39bc8138c4c", "attempt_id": "acb766dc0c88", "run_group_id": "E4_core:A4|closed|qwen3b-llama3b:h8", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:27:08.973274Z", "task_family": "synthetic", "document_id": "syn_03009_2000", "document_hash": "6ed94dcf728255bd7a7d11c97c362930728fb81a3e91f1906eb6eec86e0fedb7", "question_id": "q00_single_fact", "split": "test", "cluster_id": "syn_03009_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "tokenizer_hashes": {"receiver": "b9d8872cf3a99da99e1fc4b0f71232a451632dc49ca4c6771549b4efa5c1c763"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 8, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1770, "query_tokens": 545, "message_tokens": 441, "message_utf8_bytes": 1263, "cache_tensor_bytes": null, "serialized_payload_bytes": 1263, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 152797.98233509064, "hops": {"1": 22554.907249985263, "2": 17310.324167017825, "3": 21701.9699170487, "4": 15481.664084014483, "5": 20563.99449997116, "6": 17204.107792000286, "7": 20497.943166992627, "8": 17483.071458060294}, "receiver_prefill_ms": 329.5132500352338, "ttft_ms": 37.656333995983005, "decode_ms": 214.0373339643702}, "total_wall_ms": 153379.95221000165, "ttft_ms": 37.656333995983005, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 7}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "PA-37F-6227", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 26, "retention_rate": 0.4, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.75, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1263}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 26, "retention_rate": 0.4, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.75, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1123}, "2": {"facts_total": 65, "facts_retained": 26, "retention_rate": 0.4, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.75, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1113}, "3": {"facts_total": 65, "facts_retained": 26, "retention_rate": 0.4, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.75, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1108}, "4": {"facts_total": 65, "facts_retained": 26, "retention_rate": 0.4, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.75, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1066}, "5": {"facts_total": 65, "facts_retained": 26, "retention_rate": 0.4, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.75, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1015}, "6": {"facts_total": 65, "facts_retained": 26, "retention_rate": 0.4, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.75, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1263}, "7": {"facts_total": 65, "facts_retained": 26, "retention_rate": 0.4, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.75, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1015}, "8": {"facts_total": 65, "facts_retained": 26, "retention_rate": 0.4, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.75, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1263}}}, "reference_key_id": "keys/synthetic/syn_03009_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "bc2d36656ace4d952605cfdb84c43c2e1074a8f402c688455cec0fdcb67b2260", "output_hash": "5b7645cf6e8140f4c0bf6ed4ad76121cf2924fe7e666d92161812527b86f4409"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "5b87976c7249d138af161055", "attempt_id": "e643b24d8ceb", "run_group_id": "E4_core:A4|closed|qwen3b-llama3b:h8", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:27:09.573134Z", "task_family": "synthetic", "document_id": "syn_03009_2000", "document_hash": "6ed94dcf728255bd7a7d11c97c362930728fb81a3e91f1906eb6eec86e0fedb7", "question_id": "q01_single_fact", "split": "test", "cluster_id": "syn_03009_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "tokenizer_hashes": {"receiver": "b9d8872cf3a99da99e1fc4b0f71232a451632dc49ca4c6771549b4efa5c1c763"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 8, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1770, "query_tokens": 550, "message_tokens": 441, "message_utf8_bytes": 1263, "cache_tensor_bytes": null, "serialized_payload_bytes": 1263, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 152797.98233509064, "hops": {"1": 22554.907249985263, "2": 17310.324167017825, "3": 21701.9699170487, "4": 15481.664084014483, "5": 20563.99449997116, "6": 17204.107792000286, "7": 20497.943166992627, "8": 17483.071458060294}, "receiver_prefill_ms": 319.16195794474334, "ttft_ms": 34.35854194685817, "decode_ms": 34.33933400083333}, "total_wall_ms": 153186.80496013258, "ttft_ms": 34.35854194685817, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 26, "retention_rate": 0.4, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.75, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1263}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 26, "retention_rate": 0.4, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.75, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1123}, "2": {"facts_total": 65, "facts_retained": 26, "retention_rate": 0.4, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.75, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1113}, "3": {"facts_total": 65, "facts_retained": 26, "retention_rate": 0.4, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.75, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1108}, "4": {"facts_total": 65, "facts_retained": 26, "retention_rate": 0.4, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.75, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1066}, "5": {"facts_total": 65, "facts_retained": 26, "retention_rate": 0.4, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.75, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1015}, "6": {"facts_total": 65, "facts_retained": 26, "retention_rate": 0.4, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.75, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1263}, "7": {"facts_total": 65, "facts_retained": 26, "retention_rate": 0.4, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.75, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1015}, "8": {"facts_total": 65, "facts_retained": 26, "retention_rate": 0.4, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.75, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1263}}}, "reference_key_id": "keys/synthetic/syn_03009_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "bc2d36656ace4d952605cfdb84c43c2e1074a8f402c688455cec0fdcb67b2260", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "22eed080604aa54bb7c940ce", "attempt_id": "5e23994c2f27", "run_group_id": "E4_core:A4|closed|qwen3b-llama3b:h8", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:27:09.979969Z", "task_family": "synthetic", "document_id": "syn_03009_2000", "document_hash": "6ed94dcf728255bd7a7d11c97c362930728fb81a3e91f1906eb6eec86e0fedb7", "question_id": "q02_exact_copy", "split": "test", "cluster_id": "syn_03009_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "tokenizer_hashes": {"receiver": "b9d8872cf3a99da99e1fc4b0f71232a451632dc49ca4c6771549b4efa5c1c763"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 8, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1770, "query_tokens": 549, "message_tokens": 441, "message_utf8_bytes": 1263, "cache_tensor_bytes": null, "serialized_payload_bytes": 1263, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 152797.98233509064, "hops": {"1": 22554.907249985263, "2": 17310.324167017825, "3": 21701.9699170487, "4": 15481.664084014483, "5": 20563.99449997116, "6": 17204.107792000286, "7": 20497.943166992627, "8": 17483.071458060294}, "receiver_prefill_ms": 369.1609170055017, "ttft_ms": 93.37124996818602, "decode_ms": 138.11554096173495}, "total_wall_ms": 153399.27637611981, "ttft_ms": 93.37124996818602, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 26, "retention_rate": 0.4, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.75, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1263}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 26, "retention_rate": 0.4, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.75, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1123}, "2": {"facts_total": 65, "facts_retained": 26, "retention_rate": 0.4, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.75, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1113}, "3": {"facts_total": 65, "facts_retained": 26, "retention_rate": 0.4, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.75, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1108}, "4": {"facts_total": 65, "facts_retained": 26, "retention_rate": 0.4, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.75, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1066}, "5": {"facts_total": 65, "facts_retained": 26, "retention_rate": 0.4, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.75, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1015}, "6": {"facts_total": 65, "facts_retained": 26, "retention_rate": 0.4, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.75, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1263}, "7": {"facts_total": 65, "facts_retained": 26, "retention_rate": 0.4, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.75, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1015}, "8": {"facts_total": 65, "facts_retained": 26, "retention_rate": 0.4, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.75, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1263}}}, "reference_key_id": "keys/synthetic/syn_03009_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "bc2d36656ace4d952605cfdb84c43c2e1074a8f402c688455cec0fdcb67b2260", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "45c6ab199cd756c1bc653dd4", "attempt_id": "8f84ca272391", "run_group_id": "E4_core:A4|closed|qwen3b-llama3b:h8", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:27:10.600441Z", "task_family": "synthetic", "document_id": "syn_03009_2000", "document_hash": "6ed94dcf728255bd7a7d11c97c362930728fb81a3e91f1906eb6eec86e0fedb7", "question_id": "q03_exact_copy", "split": "test", "cluster_id": "syn_03009_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "tokenizer_hashes": {"receiver": "b9d8872cf3a99da99e1fc4b0f71232a451632dc49ca4c6771549b4efa5c1c763"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 8, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1770, "query_tokens": 544, "message_tokens": 441, "message_utf8_bytes": 1263, "cache_tensor_bytes": null, "serialized_payload_bytes": 1263, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 152797.98233509064, "hops": {"1": 22554.907249985263, "2": 17310.324167017825, "3": 21701.9699170487, "4": 15481.664084014483, "5": 20563.99449997116, "6": 17204.107792000286, "7": 20497.943166992627, "8": 17483.071458060294}, "receiver_prefill_ms": 332.32024998869747, "ttft_ms": 32.26983407512307, "decode_ms": 158.46729197073728}, "total_wall_ms": 153321.64196006488, "ttft_ms": 32.26983407512307, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 5}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "Idris Bergström", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 26, "retention_rate": 0.4, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.75, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1263}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 26, "retention_rate": 0.4, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.75, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1123}, "2": {"facts_total": 65, "facts_retained": 26, "retention_rate": 0.4, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.75, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1113}, "3": {"facts_total": 65, "facts_retained": 26, "retention_rate": 0.4, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.75, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1108}, "4": {"facts_total": 65, "facts_retained": 26, "retention_rate": 0.4, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.75, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1066}, "5": {"facts_total": 65, "facts_retained": 26, "retention_rate": 0.4, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.75, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1015}, "6": {"facts_total": 65, "facts_retained": 26, "retention_rate": 0.4, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.75, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1263}, "7": {"facts_total": 65, "facts_retained": 26, "retention_rate": 0.4, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.75, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1015}, "8": {"facts_total": 65, "facts_retained": 26, "retention_rate": 0.4, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.75, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1263}}}, "reference_key_id": "keys/synthetic/syn_03009_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "bc2d36656ace4d952605cfdb84c43c2e1074a8f402c688455cec0fdcb67b2260", "output_hash": "fd88832c9fd0ab175818d76d875d70156225bd4b34ec401514546ef70f4f826c"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "02a2b1399168bda7cc20ef7b", "attempt_id": "093d26948f6a", "run_group_id": "E4_core:A4|closed|qwen3b-llama3b:h8", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:27:11.141406Z", "task_family": "synthetic", "document_id": "syn_03009_2000", "document_hash": "6ed94dcf728255bd7a7d11c97c362930728fb81a3e91f1906eb6eec86e0fedb7", "question_id": "q04_multi_hop", "split": "test", "cluster_id": "syn_03009_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "tokenizer_hashes": {"receiver": "b9d8872cf3a99da99e1fc4b0f71232a451632dc49ca4c6771549b4efa5c1c763"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 8, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1770, "query_tokens": 553, "message_tokens": 441, "message_utf8_bytes": 1263, "cache_tensor_bytes": null, "serialized_payload_bytes": 1263, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 152797.98233509064, "hops": {"1": 22554.907249985263, "2": 17310.324167017825, "3": 21701.9699170487, "4": 15481.664084014483, "5": 20563.99449997116, "6": 17204.107792000286, "7": 20497.943166992627, "8": 17483.071458060294}, "receiver_prefill_ms": 336.2563339760527, "ttft_ms": 31.961333006620407, "decode_ms": 119.53791696578264}, "total_wall_ms": 153286.42996004783, "ttft_ms": 31.961333006620407, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 5}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "4,670,383", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 26, "retention_rate": 0.4, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.75, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1263}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 26, "retention_rate": 0.4, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.75, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1123}, "2": {"facts_total": 65, "facts_retained": 26, "retention_rate": 0.4, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.75, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1113}, "3": {"facts_total": 65, "facts_retained": 26, "retention_rate": 0.4, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.75, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1108}, "4": {"facts_total": 65, "facts_retained": 26, "retention_rate": 0.4, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.75, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1066}, "5": {"facts_total": 65, "facts_retained": 26, "retention_rate": 0.4, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.75, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1015}, "6": {"facts_total": 65, "facts_retained": 26, "retention_rate": 0.4, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.75, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1263}, "7": {"facts_total": 65, "facts_retained": 26, "retention_rate": 0.4, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.75, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1015}, "8": {"facts_total": 65, "facts_retained": 26, "retention_rate": 0.4, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.75, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1263}}}, "reference_key_id": "keys/synthetic/syn_03009_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "bc2d36656ace4d952605cfdb84c43c2e1074a8f402c688455cec0fdcb67b2260", "output_hash": "9510633f2e349a1cedecda578083181521646f70ca5ae08bae4b9188f3bd4a3b"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "61f42cc6d17e5b9f2c6283c2", "attempt_id": "78b754d178e2", "run_group_id": "E4_core:A4|closed|qwen3b-llama3b:h8", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:27:11.647414Z", "task_family": "synthetic", "document_id": "syn_03009_2000", "document_hash": "6ed94dcf728255bd7a7d11c97c362930728fb81a3e91f1906eb6eec86e0fedb7", "question_id": "q05_multi_hop", "split": "test", "cluster_id": "syn_03009_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "tokenizer_hashes": {"receiver": "b9d8872cf3a99da99e1fc4b0f71232a451632dc49ca4c6771549b4efa5c1c763"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 8, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1770, "query_tokens": 554, "message_tokens": 441, "message_utf8_bytes": 1263, "cache_tensor_bytes": null, "serialized_payload_bytes": 1263, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 152797.98233509064, "hops": {"1": 22554.907249985263, "2": 17310.324167017825, "3": 21701.9699170487, "4": 15481.664084014483, "5": 20563.99449997116, "6": 17204.107792000286, "7": 20497.943166992627, "8": 17483.071458060294}, "receiver_prefill_ms": 322.22145900595933, "ttft_ms": 39.682417060248554, "decode_ms": 55.92408403754234}, "total_wall_ms": 153216.53221000452, "ttft_ms": 39.682417060248554, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 26, "retention_rate": 0.4, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.75, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1263}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 26, "retention_rate": 0.4, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.75, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1123}, "2": {"facts_total": 65, "facts_retained": 26, "retention_rate": 0.4, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.75, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1113}, "3": {"facts_total": 65, "facts_retained": 26, "retention_rate": 0.4, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.75, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1108}, "4": {"facts_total": 65, "facts_retained": 26, "retention_rate": 0.4, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.75, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1066}, "5": {"facts_total": 65, "facts_retained": 26, "retention_rate": 0.4, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.75, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1015}, "6": {"facts_total": 65, "facts_retained": 26, "retention_rate": 0.4, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.75, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1263}, "7": {"facts_total": 65, "facts_retained": 26, "retention_rate": 0.4, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.75, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1015}, "8": {"facts_total": 65, "facts_retained": 26, "retention_rate": 0.4, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.75, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1263}}}, "reference_key_id": "keys/synthetic/syn_03009_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "bc2d36656ace4d952605cfdb84c43c2e1074a8f402c688455cec0fdcb67b2260", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "02b75d199145b4d73cf832dd", "attempt_id": "b857432015e0", "run_group_id": "E4_core:A4|closed|qwen3b-llama3b:h8", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:27:12.084119Z", "task_family": "synthetic", "document_id": "syn_03009_2000", "document_hash": "6ed94dcf728255bd7a7d11c97c362930728fb81a3e91f1906eb6eec86e0fedb7", "question_id": "q06_negation", "split": "test", "cluster_id": "syn_03009_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "tokenizer_hashes": {"receiver": "b9d8872cf3a99da99e1fc4b0f71232a451632dc49ca4c6771549b4efa5c1c763"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 8, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1770, "query_tokens": 546, "message_tokens": 441, "message_utf8_bytes": 1263, "cache_tensor_bytes": null, "serialized_payload_bytes": 1263, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 152797.98233509064, "hops": {"1": 22554.907249985263, "2": 17310.324167017825, "3": 21701.9699170487, "4": 15481.664084014483, "5": 20563.99449997116, "6": 17204.107792000286, "7": 20497.943166992627, "8": 17483.071458060294}, "receiver_prefill_ms": 479.5554580632597, "ttft_ms": 20.583749981597066, "decode_ms": 61.57908297609538}, "total_wall_ms": 153360.36004300695, "ttft_ms": 20.583749981597066, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 2}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "No.", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 26, "retention_rate": 0.4, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.75, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1263}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 26, "retention_rate": 0.4, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.75, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1123}, "2": {"facts_total": 65, "facts_retained": 26, "retention_rate": 0.4, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.75, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1113}, "3": {"facts_total": 65, "facts_retained": 26, "retention_rate": 0.4, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.75, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1108}, "4": {"facts_total": 65, "facts_retained": 26, "retention_rate": 0.4, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.75, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1066}, "5": {"facts_total": 65, "facts_retained": 26, "retention_rate": 0.4, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.75, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1015}, "6": {"facts_total": 65, "facts_retained": 26, "retention_rate": 0.4, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.75, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1263}, "7": {"facts_total": 65, "facts_retained": 26, "retention_rate": 0.4, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.75, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1015}, "8": {"facts_total": 65, "facts_retained": 26, "retention_rate": 0.4, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.75, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1263}}}, "reference_key_id": "keys/synthetic/syn_03009_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "bc2d36656ace4d952605cfdb84c43c2e1074a8f402c688455cec0fdcb67b2260", "output_hash": "38178a20b470cfd18299fb1593dd4ba706f1a83321dac051d01086efd8b7a96f"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "22ed49ead9ad81814f66e6e6", "attempt_id": "ae3f3989c4e3", "run_group_id": "E4_core:A4|closed|qwen3b-llama3b:h8", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:27:12.664977Z", "task_family": "synthetic", "document_id": "syn_03009_2000", "document_hash": "6ed94dcf728255bd7a7d11c97c362930728fb81a3e91f1906eb6eec86e0fedb7", "question_id": "q07_negation", "split": "test", "cluster_id": "syn_03009_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "tokenizer_hashes": {"receiver": "b9d8872cf3a99da99e1fc4b0f71232a451632dc49ca4c6771549b4efa5c1c763"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 8, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1770, "query_tokens": 545, "message_tokens": 441, "message_utf8_bytes": 1263, "cache_tensor_bytes": null, "serialized_payload_bytes": 1263, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 152797.98233509064, "hops": {"1": 22554.907249985263, "2": 17310.324167017825, "3": 21701.9699170487, "4": 15481.664084014483, "5": 20563.99449997116, "6": 17204.107792000286, "7": 20497.943166992627, "8": 17483.071458060294}, "receiver_prefill_ms": 341.5760829811916, "ttft_ms": 39.328041952103376, "decode_ms": 57.62329208664596}, "total_wall_ms": 153237.16800217517, "ttft_ms": 39.328041952103376, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 2}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "No.", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 26, "retention_rate": 0.4, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.75, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1263}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 26, "retention_rate": 0.4, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.75, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1123}, "2": {"facts_total": 65, "facts_retained": 26, "retention_rate": 0.4, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.75, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1113}, "3": {"facts_total": 65, "facts_retained": 26, "retention_rate": 0.4, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.75, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1108}, "4": {"facts_total": 65, "facts_retained": 26, "retention_rate": 0.4, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.75, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1066}, "5": {"facts_total": 65, "facts_retained": 26, "retention_rate": 0.4, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.75, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1015}, "6": {"facts_total": 65, "facts_retained": 26, "retention_rate": 0.4, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.75, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1263}, "7": {"facts_total": 65, "facts_retained": 26, "retention_rate": 0.4, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.75, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1015}, "8": {"facts_total": 65, "facts_retained": 26, "retention_rate": 0.4, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.75, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1263}}}, "reference_key_id": "keys/synthetic/syn_03009_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "bc2d36656ace4d952605cfdb84c43c2e1074a8f402c688455cec0fdcb67b2260", "output_hash": "38178a20b470cfd18299fb1593dd4ba706f1a83321dac051d01086efd8b7a96f"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "f6f5b564463e86ac41245496", "attempt_id": "b1b25a81b9bd", "run_group_id": "E4_core:A4|closed|qwen3b-llama3b:h8", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:27:13.121919Z", "task_family": "synthetic", "document_id": "syn_03009_2000", "document_hash": "6ed94dcf728255bd7a7d11c97c362930728fb81a3e91f1906eb6eec86e0fedb7", "question_id": "q08_revision", "split": "test", "cluster_id": "syn_03009_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "tokenizer_hashes": {"receiver": "b9d8872cf3a99da99e1fc4b0f71232a451632dc49ca4c6771549b4efa5c1c763"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 8, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1770, "query_tokens": 562, "message_tokens": 441, "message_utf8_bytes": 1263, "cache_tensor_bytes": null, "serialized_payload_bytes": 1263, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 152797.98233509064, "hops": {"1": 22554.907249985263, "2": 17310.324167017825, "3": 21701.9699170487, "4": 15481.664084014483, "5": 20563.99449997116, "6": 17204.107792000286, "7": 20497.943166992627, "8": 17483.071458060294}, "receiver_prefill_ms": 343.00270897801965, "ttft_ms": 52.19262500759214, "decode_ms": 315.7678749412298}, "total_wall_ms": 153509.77371016052, "ttft_ms": 52.19262500759214, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 5}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "10,047,105", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 26, "retention_rate": 0.4, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.75, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1263}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 26, "retention_rate": 0.4, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.75, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1123}, "2": {"facts_total": 65, "facts_retained": 26, "retention_rate": 0.4, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.75, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1113}, "3": {"facts_total": 65, "facts_retained": 26, "retention_rate": 0.4, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.75, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1108}, "4": {"facts_total": 65, "facts_retained": 26, "retention_rate": 0.4, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.75, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1066}, "5": {"facts_total": 65, "facts_retained": 26, "retention_rate": 0.4, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.75, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1015}, "6": {"facts_total": 65, "facts_retained": 26, "retention_rate": 0.4, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.75, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1263}, "7": {"facts_total": 65, "facts_retained": 26, "retention_rate": 0.4, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.75, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1015}, "8": {"facts_total": 65, "facts_retained": 26, "retention_rate": 0.4, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.75, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1263}}}, "reference_key_id": "keys/synthetic/syn_03009_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "bc2d36656ace4d952605cfdb84c43c2e1074a8f402c688455cec0fdcb67b2260", "output_hash": "45ed933586262d1470caf31c89c80a05759b79bd4e620737608286d4591bcde5"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "5e333df50335c1443b1e4f39", "attempt_id": "e33e6e41984e", "run_group_id": "E4_core:A4|closed|qwen3b-llama3b:h8", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:27:13.851957Z", "task_family": "synthetic", "document_id": "syn_03009_2000", "document_hash": "6ed94dcf728255bd7a7d11c97c362930728fb81a3e91f1906eb6eec86e0fedb7", "question_id": "q09_revision", "split": "test", "cluster_id": "syn_03009_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "tokenizer_hashes": {"receiver": "b9d8872cf3a99da99e1fc4b0f71232a451632dc49ca4c6771549b4efa5c1c763"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 8, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1770, "query_tokens": 555, "message_tokens": 441, "message_utf8_bytes": 1263, "cache_tensor_bytes": null, "serialized_payload_bytes": 1263, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 152797.98233509064, "hops": {"1": 22554.907249985263, "2": 17310.324167017825, "3": 21701.9699170487, "4": 15481.664084014483, "5": 20563.99449997116, "6": 17204.107792000286, "7": 20497.943166992627, "8": 17483.071458060294}, "receiver_prefill_ms": 414.51908310409635, "ttft_ms": 34.93183304090053, "decode_ms": 148.79258396103978}, "total_wall_ms": 153396.9507511938, "ttft_ms": 34.93183304090053, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 5}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "8,227,110", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 26, "retention_rate": 0.4, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.75, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1263}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 26, "retention_rate": 0.4, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.75, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1123}, "2": {"facts_total": 65, "facts_retained": 26, "retention_rate": 0.4, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.75, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1113}, "3": {"facts_total": 65, "facts_retained": 26, "retention_rate": 0.4, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.75, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1108}, "4": {"facts_total": 65, "facts_retained": 26, "retention_rate": 0.4, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.75, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1066}, "5": {"facts_total": 65, "facts_retained": 26, "retention_rate": 0.4, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.75, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1015}, "6": {"facts_total": 65, "facts_retained": 26, "retention_rate": 0.4, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.75, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1263}, "7": {"facts_total": 65, "facts_retained": 26, "retention_rate": 0.4, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.75, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1015}, "8": {"facts_total": 65, "facts_retained": 26, "retention_rate": 0.4, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.75, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1263}}}, "reference_key_id": "keys/synthetic/syn_03009_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "bc2d36656ace4d952605cfdb84c43c2e1074a8f402c688455cec0fdcb67b2260", "output_hash": "c3aa599fbb00899728f00a77f6ec2d981fcc6387f12136cbb65878e6772f4889"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "ec040c9ff1303f36cd50c081", "attempt_id": "b706a3e8bc0c", "run_group_id": "E4_core:A4|closed|qwen3b-llama3b:h8", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:27:14.471138Z", "task_family": "synthetic", "document_id": "syn_03009_2000", "document_hash": "6ed94dcf728255bd7a7d11c97c362930728fb81a3e91f1906eb6eec86e0fedb7", "question_id": "q10_temporal", "split": "test", "cluster_id": "syn_03009_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "tokenizer_hashes": {"receiver": "b9d8872cf3a99da99e1fc4b0f71232a451632dc49ca4c6771549b4efa5c1c763"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 8, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1770, "query_tokens": 563, "message_tokens": 441, "message_utf8_bytes": 1263, "cache_tensor_bytes": null, "serialized_payload_bytes": 1263, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 152797.98233509064, "hops": {"1": 22554.907249985263, "2": 17310.324167017825, "3": 21701.9699170487, "4": 15481.664084014483, "5": 20563.99449997116, "6": 17204.107792000286, "7": 20497.943166992627, "8": 17483.071458060294}, "receiver_prefill_ms": 339.2646249849349, "ttft_ms": 37.665833020582795, "decode_ms": 23.716125055216253}, "total_wall_ms": 153199.29366803262, "ttft_ms": 37.665833020582795, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 26, "retention_rate": 0.4, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.75, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1263}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 26, "retention_rate": 0.4, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.75, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1123}, "2": {"facts_total": 65, "facts_retained": 26, "retention_rate": 0.4, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.75, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1113}, "3": {"facts_total": 65, "facts_retained": 26, "retention_rate": 0.4, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.75, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1108}, "4": {"facts_total": 65, "facts_retained": 26, "retention_rate": 0.4, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.75, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1066}, "5": {"facts_total": 65, "facts_retained": 26, "retention_rate": 0.4, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.75, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1015}, "6": {"facts_total": 65, "facts_retained": 26, "retention_rate": 0.4, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.75, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1263}, "7": {"facts_total": 65, "facts_retained": 26, "retention_rate": 0.4, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.75, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1015}, "8": {"facts_total": 65, "facts_retained": 26, "retention_rate": 0.4, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.75, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1263}}}, "reference_key_id": "keys/synthetic/syn_03009_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "bc2d36656ace4d952605cfdb84c43c2e1074a8f402c688455cec0fdcb67b2260", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "ae595521da270acde277f432", "attempt_id": "51ea2d6cc554", "run_group_id": "E4_core:A4|closed|qwen3b-llama3b:h8", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:27:14.890652Z", "task_family": "synthetic", "document_id": "syn_03009_2000", "document_hash": "6ed94dcf728255bd7a7d11c97c362930728fb81a3e91f1906eb6eec86e0fedb7", "question_id": "q11_aggregation", "split": "test", "cluster_id": "syn_03009_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "tokenizer_hashes": {"receiver": "b9d8872cf3a99da99e1fc4b0f71232a451632dc49ca4c6771549b4efa5c1c763"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 8, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1770, "query_tokens": 556, "message_tokens": 441, "message_utf8_bytes": 1263, "cache_tensor_bytes": null, "serialized_payload_bytes": 1263, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 152797.98233509064, "hops": {"1": 22554.907249985263, "2": 17310.324167017825, "3": 21701.9699170487, "4": 15481.664084014483, "5": 20563.99449997116, "6": 17204.107792000286, "7": 20497.943166992627, "8": 17483.071458060294}, "receiver_prefill_ms": 307.1504170075059, "ttft_ms": 45.99870799575001, "decode_ms": 279.59887497127056}, "total_wall_ms": 153431.6677520983, "ttft_ms": 45.99870799575001, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 5}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "5,247,196", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 26, "retention_rate": 0.4, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.75, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1263}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 26, "retention_rate": 0.4, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.75, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1123}, "2": {"facts_total": 65, "facts_retained": 26, "retention_rate": 0.4, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.75, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1113}, "3": {"facts_total": 65, "facts_retained": 26, "retention_rate": 0.4, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.75, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1108}, "4": {"facts_total": 65, "facts_retained": 26, "retention_rate": 0.4, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.75, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1066}, "5": {"facts_total": 65, "facts_retained": 26, "retention_rate": 0.4, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.75, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1015}, "6": {"facts_total": 65, "facts_retained": 26, "retention_rate": 0.4, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.75, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1263}, "7": {"facts_total": 65, "facts_retained": 26, "retention_rate": 0.4, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.75, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1015}, "8": {"facts_total": 65, "facts_retained": 26, "retention_rate": 0.4, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.75, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1263}}}, "reference_key_id": "keys/synthetic/syn_03009_2000.json", "task_scores": {"correct": 0.0, "abstained": 0.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "bc2d36656ace4d952605cfdb84c43c2e1074a8f402c688455cec0fdcb67b2260", "output_hash": "40e869c2b62e27bc00d9598728026ccfc12fd1b97b0864921ab5373cfc6a82f7"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "d99057a8225eb3d0bda63427", "attempt_id": "e00b7db9d2c9", "run_group_id": "E4_core:A4|closed|qwen3b-llama3b:h8", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:27:15.542005Z", "task_family": "synthetic", "document_id": "syn_03009_2000", "document_hash": "6ed94dcf728255bd7a7d11c97c362930728fb81a3e91f1906eb6eec86e0fedb7", "question_id": "q12_unit_conversion", "split": "test", "cluster_id": "syn_03009_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "tokenizer_hashes": {"receiver": "b9d8872cf3a99da99e1fc4b0f71232a451632dc49ca4c6771549b4efa5c1c763"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 8, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1770, "query_tokens": 553, "message_tokens": 441, "message_utf8_bytes": 1263, "cache_tensor_bytes": null, "serialized_payload_bytes": 1263, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 152797.98233509064, "hops": {"1": 22554.907249985263, "2": 17310.324167017825, "3": 21701.9699170487, "4": 15481.664084014483, "5": 20563.99449997116, "6": 17204.107792000286, "7": 20497.943166992627, "8": 17483.071458060294}, "receiver_prefill_ms": 422.3509579896927, "ttft_ms": 37.443250068463385, "decode_ms": 27.520791976712644}, "total_wall_ms": 153286.00650210865, "ttft_ms": 37.443250068463385, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 26, "retention_rate": 0.4, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.75, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1263}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 26, "retention_rate": 0.4, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.75, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1123}, "2": {"facts_total": 65, "facts_retained": 26, "retention_rate": 0.4, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.75, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1113}, "3": {"facts_total": 65, "facts_retained": 26, "retention_rate": 0.4, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.75, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1108}, "4": {"facts_total": 65, "facts_retained": 26, "retention_rate": 0.4, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.75, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1066}, "5": {"facts_total": 65, "facts_retained": 26, "retention_rate": 0.4, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.75, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1015}, "6": {"facts_total": 65, "facts_retained": 26, "retention_rate": 0.4, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.75, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1263}, "7": {"facts_total": 65, "facts_retained": 26, "retention_rate": 0.4, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.75, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1015}, "8": {"facts_total": 65, "facts_retained": 26, "retention_rate": 0.4, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.75, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1263}}}, "reference_key_id": "keys/synthetic/syn_03009_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "bc2d36656ace4d952605cfdb84c43c2e1074a8f402c688455cec0fdcb67b2260", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "54ae6bef943cc5d8d523c2ed", "attempt_id": "d696206b3239", "run_group_id": "E4_core:A4|closed|qwen3b-llama3b:h8", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:27:16.050357Z", "task_family": "synthetic", "document_id": "syn_03009_2000", "document_hash": "6ed94dcf728255bd7a7d11c97c362930728fb81a3e91f1906eb6eec86e0fedb7", "question_id": "q13_unanswerable", "split": "test", "cluster_id": "syn_03009_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "tokenizer_hashes": {"receiver": "b9d8872cf3a99da99e1fc4b0f71232a451632dc49ca4c6771549b4efa5c1c763"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 8, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1770, "query_tokens": 557, "message_tokens": 441, "message_utf8_bytes": 1263, "cache_tensor_bytes": null, "serialized_payload_bytes": 1263, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 152797.98233509064, "hops": {"1": 22554.907249985263, "2": 17310.324167017825, "3": 21701.9699170487, "4": 15481.664084014483, "5": 20563.99449997116, "6": 17204.107792000286, "7": 20497.943166992627, "8": 17483.071458060294}, "receiver_prefill_ms": 355.95124994870275, "ttft_ms": 29.353666002862155, "decode_ms": 37.34245803207159}, "total_wall_ms": 153221.3284181198, "ttft_ms": 29.353666002862155, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 26, "retention_rate": 0.4, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.75, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1263}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 26, "retention_rate": 0.4, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.75, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1123}, "2": {"facts_total": 65, "facts_retained": 26, "retention_rate": 0.4, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.75, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1113}, "3": {"facts_total": 65, "facts_retained": 26, "retention_rate": 0.4, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.75, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1108}, "4": {"facts_total": 65, "facts_retained": 26, "retention_rate": 0.4, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.75, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1066}, "5": {"facts_total": 65, "facts_retained": 26, "retention_rate": 0.4, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.75, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1015}, "6": {"facts_total": 65, "facts_retained": 26, "retention_rate": 0.4, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.75, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1263}, "7": {"facts_total": 65, "facts_retained": 26, "retention_rate": 0.4, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.75, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1015}, "8": {"facts_total": 65, "facts_retained": 26, "retention_rate": 0.4, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.75, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1263}}}, "reference_key_id": "keys/synthetic/syn_03009_2000.json", "task_scores": {"correct": 1.0, "abstained": 1.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "bc2d36656ace4d952605cfdb84c43c2e1074a8f402c688455cec0fdcb67b2260", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "f3b03f4ce39e7a7dd8ff2da7", "attempt_id": "f0c95f230bc0", "run_group_id": "E4_core:A4|closed|qwen3b-llama3b:h8", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:27:16.490827Z", "task_family": "synthetic", "document_id": "syn_03009_2000", "document_hash": "6ed94dcf728255bd7a7d11c97c362930728fb81a3e91f1906eb6eec86e0fedb7", "question_id": "q14_unanswerable", "split": "test", "cluster_id": "syn_03009_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "tokenizer_hashes": {"receiver": "b9d8872cf3a99da99e1fc4b0f71232a451632dc49ca4c6771549b4efa5c1c763"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 8, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1770, "query_tokens": 558, "message_tokens": 441, "message_utf8_bytes": 1263, "cache_tensor_bytes": null, "serialized_payload_bytes": 1263, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 152797.98233509064, "hops": {"1": 22554.907249985263, "2": 17310.324167017825, "3": 21701.9699170487, "4": 15481.664084014483, "5": 20563.99449997116, "6": 17204.107792000286, "7": 20497.943166992627, "8": 17483.071458060294}, "receiver_prefill_ms": 323.5609169350937, "ttft_ms": 34.38187495339662, "decode_ms": 35.3554169414565}, "total_wall_ms": 153192.0430851169, "ttft_ms": 34.38187495339662, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 26, "retention_rate": 0.4, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.75, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1263}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 26, "retention_rate": 0.4, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.75, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1123}, "2": {"facts_total": 65, "facts_retained": 26, "retention_rate": 0.4, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.75, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1113}, "3": {"facts_total": 65, "facts_retained": 26, "retention_rate": 0.4, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.75, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1108}, "4": {"facts_total": 65, "facts_retained": 26, "retention_rate": 0.4, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.75, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1066}, "5": {"facts_total": 65, "facts_retained": 26, "retention_rate": 0.4, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.75, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1015}, "6": {"facts_total": 65, "facts_retained": 26, "retention_rate": 0.4, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.75, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1263}, "7": {"facts_total": 65, "facts_retained": 26, "retention_rate": 0.4, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.75, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1015}, "8": {"facts_total": 65, "facts_retained": 26, "retention_rate": 0.4, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.75, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1263}}}, "reference_key_id": "keys/synthetic/syn_03009_2000.json", "task_scores": {"correct": 1.0, "abstained": 1.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "bc2d36656ace4d952605cfdb84c43c2e1074a8f402c688455cec0fdcb67b2260", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "3ae502643f36f1fc7246ea2c", "attempt_id": "4ee848096370", "run_group_id": "E4_core:A5|closed|qwen3b-llama3b:h1", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:27:33.682275Z", "task_family": "synthetic", "document_id": "syn_03009_2000", "document_hash": "6ed94dcf728255bd7a7d11c97c362930728fb81a3e91f1906eb6eec86e0fedb7", "question_id": "q00_single_fact", "split": "test", "cluster_id": "syn_03009_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 1, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1976, "query_tokens": 318, "message_tokens": 234, "message_utf8_bytes": 698, "cache_tensor_bytes": null, "serialized_payload_bytes": 698, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 16777.417791076005, "hops": {"1": 16777.417791076005}, "receiver_prefill_ms": 236.74033395946026, "ttft_ms": 95.7806670339778, "decode_ms": 273.96237500943244}, "total_wall_ms": 17384.546374087222, "ttft_ms": 95.7806670339778, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 5}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "Milos Lindqvist", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 12, "retention_rate": 0.18461538461538463, "retention_by_kind": {"person": 1.0, "date": 0.0, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 698}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 12, "retention_rate": 0.18461538461538463, "retention_by_kind": {"person": 1.0, "date": 0.0, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 698}}}, "reference_key_id": "keys/synthetic/syn_03009_2000.json", "task_scores": {"correct": 0.0, "abstained": 0.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "b7aa2d9d5f9f3a06115e4e65c8ffaf4be2a07d6d067ceac89e49c6efeb6fdd5f", "output_hash": "f1a0ce3d66245c336fdc7ac1281d2b638395739df9d016b121be6f1b2512201e"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "8bc67003ec9381cc02cfaacc", "attempt_id": "71879031d088", "run_group_id": "E4_core:A5|closed|qwen3b-llama3b:h1", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:27:34.307926Z", "task_family": "synthetic", "document_id": "syn_03009_2000", "document_hash": "6ed94dcf728255bd7a7d11c97c362930728fb81a3e91f1906eb6eec86e0fedb7", "question_id": "q01_single_fact", "split": "test", "cluster_id": "syn_03009_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 1, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1976, "query_tokens": 323, "message_tokens": 234, "message_utf8_bytes": 698, "cache_tensor_bytes": null, "serialized_payload_bytes": 698, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 16777.417791076005, "hops": {"1": 16777.417791076005}, "receiver_prefill_ms": 267.8275000071153, "ttft_ms": 38.52154198102653, "decode_ms": 25.316291954368353}, "total_wall_ms": 17109.74058310967, "ttft_ms": 38.52154198102653, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 12, "retention_rate": 0.18461538461538463, "retention_by_kind": {"person": 1.0, "date": 0.0, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 698}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 12, "retention_rate": 0.18461538461538463, "retention_by_kind": {"person": 1.0, "date": 0.0, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 698}}}, "reference_key_id": "keys/synthetic/syn_03009_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "b7aa2d9d5f9f3a06115e4e65c8ffaf4be2a07d6d067ceac89e49c6efeb6fdd5f", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "ef57749fd5732556371e701c", "attempt_id": "0af9360a8c2b", "run_group_id": "E4_core:A5|closed|qwen3b-llama3b:h1", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:27:34.658787Z", "task_family": "synthetic", "document_id": "syn_03009_2000", "document_hash": "6ed94dcf728255bd7a7d11c97c362930728fb81a3e91f1906eb6eec86e0fedb7", "question_id": "q02_exact_copy", "split": "test", "cluster_id": "syn_03009_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 1, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1976, "query_tokens": 321, "message_tokens": 234, "message_utf8_bytes": 698, "cache_tensor_bytes": null, "serialized_payload_bytes": 698, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 16777.417791076005, "hops": {"1": 16777.417791076005}, "receiver_prefill_ms": 220.2273749280721, "ttft_ms": 38.43995800707489, "decode_ms": 34.59625004325062}, "total_wall_ms": 17071.219041128643, "ttft_ms": 38.43995800707489, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 12, "retention_rate": 0.18461538461538463, "retention_by_kind": {"person": 1.0, "date": 0.0, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 698}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 12, "retention_rate": 0.18461538461538463, "retention_by_kind": {"person": 1.0, "date": 0.0, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 698}}}, "reference_key_id": "keys/synthetic/syn_03009_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "b7aa2d9d5f9f3a06115e4e65c8ffaf4be2a07d6d067ceac89e49c6efeb6fdd5f", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "0d95e03da12454a0ea9987e5", "attempt_id": "23e4d780895e", "run_group_id": "E4_core:A5|closed|qwen3b-llama3b:h1", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:27:34.970615Z", "task_family": "synthetic", "document_id": "syn_03009_2000", "document_hash": "6ed94dcf728255bd7a7d11c97c362930728fb81a3e91f1906eb6eec86e0fedb7", "question_id": "q03_exact_copy", "split": "test", "cluster_id": "syn_03009_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 1, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1976, "query_tokens": 317, "message_tokens": 234, "message_utf8_bytes": 698, "cache_tensor_bytes": null, "serialized_payload_bytes": 698, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 16777.417791076005, "hops": {"1": 16777.417791076005}, "receiver_prefill_ms": 207.4150829575956, "ttft_ms": 43.91870799008757, "decode_ms": 299.6704999823123}, "total_wall_ms": 17329.07104108017, "ttft_ms": 43.91870799008757, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 5}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "Idris Bergström", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 12, "retention_rate": 0.18461538461538463, "retention_by_kind": {"person": 1.0, "date": 0.0, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 698}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 12, "retention_rate": 0.18461538461538463, "retention_by_kind": {"person": 1.0, "date": 0.0, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 698}}}, "reference_key_id": "keys/synthetic/syn_03009_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "b7aa2d9d5f9f3a06115e4e65c8ffaf4be2a07d6d067ceac89e49c6efeb6fdd5f", "output_hash": "fd88832c9fd0ab175818d76d875d70156225bd4b34ec401514546ef70f4f826c"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "eac841520190c59dd376c73e", "attempt_id": "a7fd92b46572", "run_group_id": "E4_core:A5|closed|qwen3b-llama3b:h1", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:27:35.540531Z", "task_family": "synthetic", "document_id": "syn_03009_2000", "document_hash": "6ed94dcf728255bd7a7d11c97c362930728fb81a3e91f1906eb6eec86e0fedb7", "question_id": "q04_multi_hop", "split": "test", "cluster_id": "syn_03009_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 1, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1976, "query_tokens": 325, "message_tokens": 234, "message_utf8_bytes": 698, "cache_tensor_bytes": null, "serialized_payload_bytes": 698, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 16777.417791076005, "hops": {"1": 16777.417791076005}, "receiver_prefill_ms": 351.2802090262994, "ttft_ms": 26.84637508355081, "decode_ms": 39.60533393546939}, "total_wall_ms": 17195.5236250069, "ttft_ms": 26.84637508355081, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 12, "retention_rate": 0.18461538461538463, "retention_by_kind": {"person": 1.0, "date": 0.0, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 698}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 12, "retention_rate": 0.18461538461538463, "retention_by_kind": {"person": 1.0, "date": 0.0, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 698}}}, "reference_key_id": "keys/synthetic/syn_03009_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "b7aa2d9d5f9f3a06115e4e65c8ffaf4be2a07d6d067ceac89e49c6efeb6fdd5f", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "c3d931a4308e40b912ef66e0", "attempt_id": "836820a85666", "run_group_id": "E4_core:A5|closed|qwen3b-llama3b:h1", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:27:35.978281Z", "task_family": "synthetic", "document_id": "syn_03009_2000", "document_hash": "6ed94dcf728255bd7a7d11c97c362930728fb81a3e91f1906eb6eec86e0fedb7", "question_id": "q05_multi_hop", "split": "test", "cluster_id": "syn_03009_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 1, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1976, "query_tokens": 328, "message_tokens": 234, "message_utf8_bytes": 698, "cache_tensor_bytes": null, "serialized_payload_bytes": 698, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 16777.417791076005, "hops": {"1": 16777.417791076005}, "receiver_prefill_ms": 274.3002079660073, "ttft_ms": 32.94708300381899, "decode_ms": 37.17370796948671}, "total_wall_ms": 17122.27908312343, "ttft_ms": 32.94708300381899, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 12, "retention_rate": 0.18461538461538463, "retention_by_kind": {"person": 1.0, "date": 0.0, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 698}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 12, "retention_rate": 0.18461538461538463, "retention_by_kind": {"person": 1.0, "date": 0.0, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 698}}}, "reference_key_id": "keys/synthetic/syn_03009_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "b7aa2d9d5f9f3a06115e4e65c8ffaf4be2a07d6d067ceac89e49c6efeb6fdd5f", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "9a4b7102ea6ec08dcb7f1195", "attempt_id": "8571b1d1c19b", "run_group_id": "E4_core:A5|closed|qwen3b-llama3b:h1", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:27:36.340287Z", "task_family": "synthetic", "document_id": "syn_03009_2000", "document_hash": "6ed94dcf728255bd7a7d11c97c362930728fb81a3e91f1906eb6eec86e0fedb7", "question_id": "q06_negation", "split": "test", "cluster_id": "syn_03009_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 1, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1976, "query_tokens": 318, "message_tokens": 234, "message_utf8_bytes": 698, "cache_tensor_bytes": null, "serialized_payload_bytes": 698, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 16777.417791076005, "hops": {"1": 16777.417791076005}, "receiver_prefill_ms": 232.29166702367365, "ttft_ms": 20.029166946187615, "decode_ms": 20.628249971196055}, "total_wall_ms": 17051.235166029073, "ttft_ms": 20.029166946187615, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "no", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 12, "retention_rate": 0.18461538461538463, "retention_by_kind": {"person": 1.0, "date": 0.0, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 698}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 12, "retention_rate": 0.18461538461538463, "retention_by_kind": {"person": 1.0, "date": 0.0, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 698}}}, "reference_key_id": "keys/synthetic/syn_03009_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "b7aa2d9d5f9f3a06115e4e65c8ffaf4be2a07d6d067ceac89e49c6efeb6fdd5f", "output_hash": "9390298f3fb0c5b160498935d79cb139aef28e1c47358b4bbba61862b9c26e59"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "43f9f3bf00bf3c41443e80d2", "attempt_id": "931d4be61364", "run_group_id": "E4_core:A5|closed|qwen3b-llama3b:h1", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:27:36.631883Z", "task_family": "synthetic", "document_id": "syn_03009_2000", "document_hash": "6ed94dcf728255bd7a7d11c97c362930728fb81a3e91f1906eb6eec86e0fedb7", "question_id": "q07_negation", "split": "test", "cluster_id": "syn_03009_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 1, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1976, "query_tokens": 317, "message_tokens": 234, "message_utf8_bytes": 698, "cache_tensor_bytes": null, "serialized_payload_bytes": 698, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 16777.417791076005, "hops": {"1": 16777.417791076005}, "receiver_prefill_ms": 267.01708300970495, "ttft_ms": 56.240207981318235, "decode_ms": 53.013207972981036}, "total_wall_ms": 17154.153333045542, "ttft_ms": 56.240207981318235, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "no", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 12, "retention_rate": 0.18461538461538463, "retention_by_kind": {"person": 1.0, "date": 0.0, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 698}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 12, "retention_rate": 0.18461538461538463, "retention_by_kind": {"person": 1.0, "date": 0.0, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 698}}}, "reference_key_id": "keys/synthetic/syn_03009_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "b7aa2d9d5f9f3a06115e4e65c8ffaf4be2a07d6d067ceac89e49c6efeb6fdd5f", "output_hash": "9390298f3fb0c5b160498935d79cb139aef28e1c47358b4bbba61862b9c26e59"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "3d6d0cda6db03f9a41b977d8", "attempt_id": "9f0d42d9c82e", "run_group_id": "E4_core:A5|closed|qwen3b-llama3b:h1", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:27:37.026565Z", "task_family": "synthetic", "document_id": "syn_03009_2000", "document_hash": "6ed94dcf728255bd7a7d11c97c362930728fb81a3e91f1906eb6eec86e0fedb7", "question_id": "q08_revision", "split": "test", "cluster_id": "syn_03009_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 1, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1976, "query_tokens": 335, "message_tokens": 234, "message_utf8_bytes": 698, "cache_tensor_bytes": null, "serialized_payload_bytes": 698, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 16777.417791076005, "hops": {"1": 16777.417791076005}, "receiver_prefill_ms": 398.6537909368053, "ttft_ms": 38.37049996946007, "decode_ms": 41.265666019171476}, "total_wall_ms": 17256.26029097475, "ttft_ms": 38.37049996946007, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 12, "retention_rate": 0.18461538461538463, "retention_by_kind": {"person": 1.0, "date": 0.0, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 698}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 12, "retention_rate": 0.18461538461538463, "retention_by_kind": {"person": 1.0, "date": 0.0, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 698}}}, "reference_key_id": "keys/synthetic/syn_03009_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "b7aa2d9d5f9f3a06115e4e65c8ffaf4be2a07d6d067ceac89e49c6efeb6fdd5f", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "cf7bddabc0ee116e938fcbc1", "attempt_id": "350bdd5cc4d2", "run_group_id": "E4_core:A5|closed|qwen3b-llama3b:h1", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:27:37.522690Z", "task_family": "synthetic", "document_id": "syn_03009_2000", "document_hash": "6ed94dcf728255bd7a7d11c97c362930728fb81a3e91f1906eb6eec86e0fedb7", "question_id": "q09_revision", "split": "test", "cluster_id": "syn_03009_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 1, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1976, "query_tokens": 328, "message_tokens": 234, "message_utf8_bytes": 698, "cache_tensor_bytes": null, "serialized_payload_bytes": 698, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 16777.417791076005, "hops": {"1": 16777.417791076005}, "receiver_prefill_ms": 254.05566697008908, "ttft_ms": 37.546750041656196, "decode_ms": 30.768916942179203}, "total_wall_ms": 17100.260124076158, "ttft_ms": 37.546750041656196, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 12, "retention_rate": 0.18461538461538463, "retention_by_kind": {"person": 1.0, "date": 0.0, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 698}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 12, "retention_rate": 0.18461538461538463, "retention_by_kind": {"person": 1.0, "date": 0.0, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 698}}}, "reference_key_id": "keys/synthetic/syn_03009_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "b7aa2d9d5f9f3a06115e4e65c8ffaf4be2a07d6d067ceac89e49c6efeb6fdd5f", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "cb9cf01e562be6a1ab3907c3", "attempt_id": "d0207ebdaf24", "run_group_id": "E4_core:A5|closed|qwen3b-llama3b:h1", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:27:37.862859Z", "task_family": "synthetic", "document_id": "syn_03009_2000", "document_hash": "6ed94dcf728255bd7a7d11c97c362930728fb81a3e91f1906eb6eec86e0fedb7", "question_id": "q10_temporal", "split": "test", "cluster_id": "syn_03009_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 1, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1976, "query_tokens": 337, "message_tokens": 234, "message_utf8_bytes": 698, "cache_tensor_bytes": null, "serialized_payload_bytes": 698, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 16777.417791076005, "hops": {"1": 16777.417791076005}, "receiver_prefill_ms": 254.21358307357877, "ttft_ms": 34.740124945528805, "decode_ms": 334.228582913056}, "total_wall_ms": 17401.09670709353, "ttft_ms": 34.740124945528805, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 7}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "Eshgamgam-11", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 12, "retention_rate": 0.18461538461538463, "retention_by_kind": {"person": 1.0, "date": 0.0, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 698}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 12, "retention_rate": 0.18461538461538463, "retention_by_kind": {"person": 1.0, "date": 0.0, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 698}}}, "reference_key_id": "keys/synthetic/syn_03009_2000.json", "task_scores": {"correct": 0.0, "abstained": 0.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "b7aa2d9d5f9f3a06115e4e65c8ffaf4be2a07d6d067ceac89e49c6efeb6fdd5f", "output_hash": "8c1313c53ab4dbffa5f336d721c810678133eecaa467241e53d8a92150601475"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "d838c3de9edead042aeb6663", "attempt_id": "3a45d98aa709", "run_group_id": "E4_core:A5|closed|qwen3b-llama3b:h1", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:27:38.503281Z", "task_family": "synthetic", "document_id": "syn_03009_2000", "document_hash": "6ed94dcf728255bd7a7d11c97c362930728fb81a3e91f1906eb6eec86e0fedb7", "question_id": "q11_aggregation", "split": "test", "cluster_id": "syn_03009_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 1, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1976, "query_tokens": 328, "message_tokens": 234, "message_utf8_bytes": 698, "cache_tensor_bytes": null, "serialized_payload_bytes": 698, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 16777.417791076005, "hops": {"1": 16777.417791076005}, "receiver_prefill_ms": 406.905582989566, "ttft_ms": 33.35345908999443, "decode_ms": 38.97145902737975}, "total_wall_ms": 17257.123416173272, "ttft_ms": 33.35345908999443, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 12, "retention_rate": 0.18461538461538463, "retention_by_kind": {"person": 1.0, "date": 0.0, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 698}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 12, "retention_rate": 0.18461538461538463, "retention_by_kind": {"person": 1.0, "date": 0.0, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 698}}}, "reference_key_id": "keys/synthetic/syn_03009_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "b7aa2d9d5f9f3a06115e4e65c8ffaf4be2a07d6d067ceac89e49c6efeb6fdd5f", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "9ac09baf1a914c0a86d08e2f", "attempt_id": "aa88b7d5fb45", "run_group_id": "E4_core:A5|closed|qwen3b-llama3b:h1", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:27:39.001227Z", "task_family": "synthetic", "document_id": "syn_03009_2000", "document_hash": "6ed94dcf728255bd7a7d11c97c362930728fb81a3e91f1906eb6eec86e0fedb7", "question_id": "q12_unit_conversion", "split": "test", "cluster_id": "syn_03009_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 1, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1976, "query_tokens": 326, "message_tokens": 234, "message_utf8_bytes": 698, "cache_tensor_bytes": null, "serialized_payload_bytes": 698, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 16777.417791076005, "hops": {"1": 16777.417791076005}, "receiver_prefill_ms": 291.0081250593066, "ttft_ms": 38.7304580071941, "decode_ms": 34.121957956813276}, "total_wall_ms": 17141.757791046984, "ttft_ms": 38.7304580071941, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 12, "retention_rate": 0.18461538461538463, "retention_by_kind": {"person": 1.0, "date": 0.0, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 698}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 12, "retention_rate": 0.18461538461538463, "retention_by_kind": {"person": 1.0, "date": 0.0, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 698}}}, "reference_key_id": "keys/synthetic/syn_03009_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "b7aa2d9d5f9f3a06115e4e65c8ffaf4be2a07d6d067ceac89e49c6efeb6fdd5f", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "152643bf1fcfcc6cbf68b164", "attempt_id": "a316e15898d1", "run_group_id": "E4_core:A5|closed|qwen3b-llama3b:h1", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:27:39.383423Z", "task_family": "synthetic", "document_id": "syn_03009_2000", "document_hash": "6ed94dcf728255bd7a7d11c97c362930728fb81a3e91f1906eb6eec86e0fedb7", "question_id": "q13_unanswerable", "split": "test", "cluster_id": "syn_03009_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 1, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1976, "query_tokens": 330, "message_tokens": 234, "message_utf8_bytes": 698, "cache_tensor_bytes": null, "serialized_payload_bytes": 698, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 16777.417791076005, "hops": {"1": 16777.417791076005}, "receiver_prefill_ms": 270.5927080241963, "ttft_ms": 21.505665965378284, "decode_ms": 21.34583401493728}, "total_wall_ms": 17091.294999117963, "ttft_ms": 21.505665965378284, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 12, "retention_rate": 0.18461538461538463, "retention_by_kind": {"person": 1.0, "date": 0.0, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 698}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 12, "retention_rate": 0.18461538461538463, "retention_by_kind": {"person": 1.0, "date": 0.0, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 698}}}, "reference_key_id": "keys/synthetic/syn_03009_2000.json", "task_scores": {"correct": 1.0, "abstained": 1.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "b7aa2d9d5f9f3a06115e4e65c8ffaf4be2a07d6d067ceac89e49c6efeb6fdd5f", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "d593471742ddd65fcf203b64", "attempt_id": "b03dd550209b", "run_group_id": "E4_core:A5|closed|qwen3b-llama3b:h1", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:27:39.714555Z", "task_family": "synthetic", "document_id": "syn_03009_2000", "document_hash": "6ed94dcf728255bd7a7d11c97c362930728fb81a3e91f1906eb6eec86e0fedb7", "question_id": "q14_unanswerable", "split": "test", "cluster_id": "syn_03009_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 1, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1976, "query_tokens": 330, "message_tokens": 234, "message_utf8_bytes": 698, "cache_tensor_bytes": null, "serialized_payload_bytes": 698, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 16777.417791076005, "hops": {"1": 16777.417791076005}, "receiver_prefill_ms": 280.07350000552833, "ttft_ms": 41.46316694095731, "decode_ms": 51.00637499708682}, "total_wall_ms": 17150.486041093245, "ttft_ms": 41.46316694095731, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 12, "retention_rate": 0.18461538461538463, "retention_by_kind": {"person": 1.0, "date": 0.0, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 698}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 12, "retention_rate": 0.18461538461538463, "retention_by_kind": {"person": 1.0, "date": 0.0, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 698}}}, "reference_key_id": "keys/synthetic/syn_03009_2000.json", "task_scores": {"correct": 1.0, "abstained": 1.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "b7aa2d9d5f9f3a06115e4e65c8ffaf4be2a07d6d067ceac89e49c6efeb6fdd5f", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "f9a49690bd1b97cf9bcb686e", "attempt_id": "49d694dcaacc", "run_group_id": "E4_core:A5|closed|qwen3b-llama3b:h2", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:27:50.379495Z", "task_family": "synthetic", "document_id": "syn_03009_2000", "document_hash": "6ed94dcf728255bd7a7d11c97c362930728fb81a3e91f1906eb6eec86e0fedb7", "question_id": "q00_single_fact", "split": "test", "cluster_id": "syn_03009_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "tokenizer_hashes": {"receiver": "b9d8872cf3a99da99e1fc4b0f71232a451632dc49ca4c6771549b4efa5c1c763"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 2, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1770, "query_tokens": 360, "message_tokens": 255, "message_utf8_bytes": 775, "cache_tensor_bytes": null, "serialized_payload_bytes": 775, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 27049.207000178285, "hops": {"1": 16777.417791076005, "2": 10271.78920910228}, "receiver_prefill_ms": 344.2932079778984, "ttft_ms": 21.559374989010394, "decode_ms": 160.62249999959022}, "total_wall_ms": 27576.184500241652, "ttft_ms": 21.559374989010394, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 6}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "Eshgamgam-11", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 12, "retention_rate": 0.18461538461538463, "retention_by_kind": {"person": 1.0, "date": 0.0, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 775}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 12, "retention_rate": 0.18461538461538463, "retention_by_kind": {"person": 1.0, "date": 0.0, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 698}, "2": {"facts_total": 65, "facts_retained": 12, "retention_rate": 0.18461538461538463, "retention_by_kind": {"person": 1.0, "date": 0.0, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 775}}}, "reference_key_id": "keys/synthetic/syn_03009_2000.json", "task_scores": {"correct": 0.0, "abstained": 0.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "2d43b54be891e5f4db3e2062da0ff5b0470ace3acf8bc02542fac6aa4045349b", "output_hash": "8c1313c53ab4dbffa5f336d721c810678133eecaa467241e53d8a92150601475"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "063d4e0382ce2ee482d93b53", "attempt_id": "25f74d76347d", "run_group_id": "E4_core:A5|closed|qwen3b-llama3b:h2", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:27:50.925175Z", "task_family": "synthetic", "document_id": "syn_03009_2000", "document_hash": "6ed94dcf728255bd7a7d11c97c362930728fb81a3e91f1906eb6eec86e0fedb7", "question_id": "q01_single_fact", "split": "test", "cluster_id": "syn_03009_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "tokenizer_hashes": {"receiver": "b9d8872cf3a99da99e1fc4b0f71232a451632dc49ca4c6771549b4efa5c1c763"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 2, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1770, "query_tokens": 365, "message_tokens": 255, "message_utf8_bytes": 775, "cache_tensor_bytes": null, "serialized_payload_bytes": 775, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 27049.207000178285, "hops": {"1": 16777.417791076005, "2": 10271.78920910228}, "receiver_prefill_ms": 316.1098749842495, "ttft_ms": 59.66312496457249, "decode_ms": 113.25612501241267}, "total_wall_ms": 27538.91575022135, "ttft_ms": 59.66312496457249, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 12, "retention_rate": 0.18461538461538463, "retention_by_kind": {"person": 1.0, "date": 0.0, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 775}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 12, "retention_rate": 0.18461538461538463, "retention_by_kind": {"person": 1.0, "date": 0.0, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 698}, "2": {"facts_total": 65, "facts_retained": 12, "retention_rate": 0.18461538461538463, "retention_by_kind": {"person": 1.0, "date": 0.0, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 775}}}, "reference_key_id": "keys/synthetic/syn_03009_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "2d43b54be891e5f4db3e2062da0ff5b0470ace3acf8bc02542fac6aa4045349b", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "52b6ea6a991bb2fee183a0fc", "attempt_id": "4a72fd587932", "run_group_id": "E4_core:A5|closed|qwen3b-llama3b:h2", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:27:51.435207Z", "task_family": "synthetic", "document_id": "syn_03009_2000", "document_hash": "6ed94dcf728255bd7a7d11c97c362930728fb81a3e91f1906eb6eec86e0fedb7", "question_id": "q02_exact_copy", "split": "test", "cluster_id": "syn_03009_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "tokenizer_hashes": {"receiver": "b9d8872cf3a99da99e1fc4b0f71232a451632dc49ca4c6771549b4efa5c1c763"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 2, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1770, "query_tokens": 364, "message_tokens": 255, "message_utf8_bytes": 775, "cache_tensor_bytes": null, "serialized_payload_bytes": 775, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 27049.207000178285, "hops": {"1": 16777.417791076005, "2": 10271.78920910228}, "receiver_prefill_ms": 413.60008297488093, "ttft_ms": 25.449041044339538, "decode_ms": 35.03849997650832}, "total_wall_ms": 27523.691167123616, "ttft_ms": 25.449041044339538, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 12, "retention_rate": 0.18461538461538463, "retention_by_kind": {"person": 1.0, "date": 0.0, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 775}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 12, "retention_rate": 0.18461538461538463, "retention_by_kind": {"person": 1.0, "date": 0.0, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 698}, "2": {"facts_total": 65, "facts_retained": 12, "retention_rate": 0.18461538461538463, "retention_by_kind": {"person": 1.0, "date": 0.0, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 775}}}, "reference_key_id": "keys/synthetic/syn_03009_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "2d43b54be891e5f4db3e2062da0ff5b0470ace3acf8bc02542fac6aa4045349b", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "e078f6248ef6e79d1c164304", "attempt_id": "e026f57e524b", "run_group_id": "E4_core:A5|closed|qwen3b-llama3b:h2", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:27:51.929408Z", "task_family": "synthetic", "document_id": "syn_03009_2000", "document_hash": "6ed94dcf728255bd7a7d11c97c362930728fb81a3e91f1906eb6eec86e0fedb7", "question_id": "q03_exact_copy", "split": "test", "cluster_id": "syn_03009_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "tokenizer_hashes": {"receiver": "b9d8872cf3a99da99e1fc4b0f71232a451632dc49ca4c6771549b4efa5c1c763"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 2, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1770, "query_tokens": 359, "message_tokens": 255, "message_utf8_bytes": 775, "cache_tensor_bytes": null, "serialized_payload_bytes": 775, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 27049.207000178285, "hops": {"1": 16777.417791076005, "2": 10271.78920910228}, "receiver_prefill_ms": 326.61799993366003, "ttft_ms": 26.340791955590248, "decode_ms": 139.55616601742804}, "total_wall_ms": 27542.234750231728, "ttft_ms": 26.340791955590248, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 5}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "Idris Bergström", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 12, "retention_rate": 0.18461538461538463, "retention_by_kind": {"person": 1.0, "date": 0.0, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 775}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 12, "retention_rate": 0.18461538461538463, "retention_by_kind": {"person": 1.0, "date": 0.0, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 698}, "2": {"facts_total": 65, "facts_retained": 12, "retention_rate": 0.18461538461538463, "retention_by_kind": {"person": 1.0, "date": 0.0, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 775}}}, "reference_key_id": "keys/synthetic/syn_03009_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "2d43b54be891e5f4db3e2062da0ff5b0470ace3acf8bc02542fac6aa4045349b", "output_hash": "fd88832c9fd0ab175818d76d875d70156225bd4b34ec401514546ef70f4f826c"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "13fae89a172f34133d7619c8", "attempt_id": "20293b98b6b8", "run_group_id": "E4_core:A5|closed|qwen3b-llama3b:h2", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:27:52.439523Z", "task_family": "synthetic", "document_id": "syn_03009_2000", "document_hash": "6ed94dcf728255bd7a7d11c97c362930728fb81a3e91f1906eb6eec86e0fedb7", "question_id": "q04_multi_hop", "split": "test", "cluster_id": "syn_03009_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "tokenizer_hashes": {"receiver": "b9d8872cf3a99da99e1fc4b0f71232a451632dc49ca4c6771549b4efa5c1c763"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 2, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1770, "query_tokens": 368, "message_tokens": 255, "message_utf8_bytes": 775, "cache_tensor_bytes": null, "serialized_payload_bytes": 775, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 27049.207000178285, "hops": {"1": 16777.417791076005, "2": 10271.78920910228}, "receiver_prefill_ms": 349.01583404280245, "ttft_ms": 41.43000009935349, "decode_ms": 34.758999943733215}, "total_wall_ms": 27475.102250115015, "ttft_ms": 41.43000009935349, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 12, "retention_rate": 0.18461538461538463, "retention_by_kind": {"person": 1.0, "date": 0.0, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 775}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 12, "retention_rate": 0.18461538461538463, "retention_by_kind": {"person": 1.0, "date": 0.0, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 698}, "2": {"facts_total": 65, "facts_retained": 12, "retention_rate": 0.18461538461538463, "retention_by_kind": {"person": 1.0, "date": 0.0, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 775}}}, "reference_key_id": "keys/synthetic/syn_03009_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "2d43b54be891e5f4db3e2062da0ff5b0470ace3acf8bc02542fac6aa4045349b", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "36b5eb273d6f4e5e7ffb14bd", "attempt_id": "7bb8971b59ae", "run_group_id": "E4_core:A5|closed|qwen3b-llama3b:h2", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:27:52.883852Z", "task_family": "synthetic", "document_id": "syn_03009_2000", "document_hash": "6ed94dcf728255bd7a7d11c97c362930728fb81a3e91f1906eb6eec86e0fedb7", "question_id": "q05_multi_hop", "split": "test", "cluster_id": "syn_03009_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "tokenizer_hashes": {"receiver": "b9d8872cf3a99da99e1fc4b0f71232a451632dc49ca4c6771549b4efa5c1c763"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 2, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1770, "query_tokens": 369, "message_tokens": 255, "message_utf8_bytes": 775, "cache_tensor_bytes": null, "serialized_payload_bytes": 775, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 27049.207000178285, "hops": {"1": 16777.417791076005, "2": 10271.78920910228}, "receiver_prefill_ms": 460.5724170105532, "ttft_ms": 33.35729206446558, "decode_ms": 21.310584037564695}, "total_wall_ms": 27564.997875131667, "ttft_ms": 33.35729206446558, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 12, "retention_rate": 0.18461538461538463, "retention_by_kind": {"person": 1.0, "date": 0.0, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 775}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 12, "retention_rate": 0.18461538461538463, "retention_by_kind": {"person": 1.0, "date": 0.0, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 698}, "2": {"facts_total": 65, "facts_retained": 12, "retention_rate": 0.18461538461538463, "retention_by_kind": {"person": 1.0, "date": 0.0, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 775}}}, "reference_key_id": "keys/synthetic/syn_03009_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "2d43b54be891e5f4db3e2062da0ff5b0470ace3acf8bc02542fac6aa4045349b", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "0745c1d0ec2ba264a2ff53c8", "attempt_id": "e7f34576576d", "run_group_id": "E4_core:A5|closed|qwen3b-llama3b:h2", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:27:53.417044Z", "task_family": "synthetic", "document_id": "syn_03009_2000", "document_hash": "6ed94dcf728255bd7a7d11c97c362930728fb81a3e91f1906eb6eec86e0fedb7", "question_id": "q06_negation", "split": "test", "cluster_id": "syn_03009_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "tokenizer_hashes": {"receiver": "b9d8872cf3a99da99e1fc4b0f71232a451632dc49ca4c6771549b4efa5c1c763"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 2, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1770, "query_tokens": 361, "message_tokens": 255, "message_utf8_bytes": 775, "cache_tensor_bytes": null, "serialized_payload_bytes": 775, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 27049.207000178285, "hops": {"1": 16777.417791076005, "2": 10271.78920910228}, "receiver_prefill_ms": 316.6376670124009, "ttft_ms": 22.741375025361776, "decode_ms": 38.170874933712184}, "total_wall_ms": 27427.179708261974, "ttft_ms": 22.741375025361776, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "No", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 12, "retention_rate": 0.18461538461538463, "retention_by_kind": {"person": 1.0, "date": 0.0, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 775}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 12, "retention_rate": 0.18461538461538463, "retention_by_kind": {"person": 1.0, "date": 0.0, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 698}, "2": {"facts_total": 65, "facts_retained": 12, "retention_rate": 0.18461538461538463, "retention_by_kind": {"person": 1.0, "date": 0.0, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 775}}}, "reference_key_id": "keys/synthetic/syn_03009_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "2d43b54be891e5f4db3e2062da0ff5b0470ace3acf8bc02542fac6aa4045349b", "output_hash": "1ea442a134b2a184bd5d40104401f2a37fbc09ccf3f4bc9da161c6099be3691d"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "e96fd7c22a7396f6ce1a8901", "attempt_id": "0de10b0d90c4", "run_group_id": "E4_core:A5|closed|qwen3b-llama3b:h2", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:27:53.812031Z", "task_family": "synthetic", "document_id": "syn_03009_2000", "document_hash": "6ed94dcf728255bd7a7d11c97c362930728fb81a3e91f1906eb6eec86e0fedb7", "question_id": "q07_negation", "split": "test", "cluster_id": "syn_03009_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "tokenizer_hashes": {"receiver": "b9d8872cf3a99da99e1fc4b0f71232a451632dc49ca4c6771549b4efa5c1c763"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 2, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1770, "query_tokens": 360, "message_tokens": 255, "message_utf8_bytes": 775, "cache_tensor_bytes": null, "serialized_payload_bytes": 775, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 27049.207000178285, "hops": {"1": 16777.417791076005, "2": 10271.78920910228}, "receiver_prefill_ms": 291.26720898784697, "ttft_ms": 38.84962492156774, "decode_ms": 34.17629096657038}, "total_wall_ms": 27414.079542155378, "ttft_ms": 38.84962492156774, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "No", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 12, "retention_rate": 0.18461538461538463, "retention_by_kind": {"person": 1.0, "date": 0.0, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 775}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 12, "retention_rate": 0.18461538461538463, "retention_by_kind": {"person": 1.0, "date": 0.0, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 698}, "2": {"facts_total": 65, "facts_retained": 12, "retention_rate": 0.18461538461538463, "retention_by_kind": {"person": 1.0, "date": 0.0, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 775}}}, "reference_key_id": "keys/synthetic/syn_03009_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "2d43b54be891e5f4db3e2062da0ff5b0470ace3acf8bc02542fac6aa4045349b", "output_hash": "1ea442a134b2a184bd5d40104401f2a37fbc09ccf3f4bc9da161c6099be3691d"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "177746081af2a4e88ed39fcd", "attempt_id": "6d18b90796ab", "run_group_id": "E4_core:A5|closed|qwen3b-llama3b:h2", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:27:54.195788Z", "task_family": "synthetic", "document_id": "syn_03009_2000", "document_hash": "6ed94dcf728255bd7a7d11c97c362930728fb81a3e91f1906eb6eec86e0fedb7", "question_id": "q08_revision", "split": "test", "cluster_id": "syn_03009_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "tokenizer_hashes": {"receiver": "b9d8872cf3a99da99e1fc4b0f71232a451632dc49ca4c6771549b4efa5c1c763"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 2, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1770, "query_tokens": 377, "message_tokens": 255, "message_utf8_bytes": 775, "cache_tensor_bytes": null, "serialized_payload_bytes": 775, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 27049.207000178285, "hops": {"1": 16777.417791076005, "2": 10271.78920910228}, "receiver_prefill_ms": 352.8698340523988, "ttft_ms": 140.81920904573053, "decode_ms": 37.3268750263378}, "total_wall_ms": 27580.818542162888, "ttft_ms": 140.81920904573053, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 12, "retention_rate": 0.18461538461538463, "retention_by_kind": {"person": 1.0, "date": 0.0, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 775}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 12, "retention_rate": 0.18461538461538463, "retention_by_kind": {"person": 1.0, "date": 0.0, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 698}, "2": {"facts_total": 65, "facts_retained": 12, "retention_rate": 0.18461538461538463, "retention_by_kind": {"person": 1.0, "date": 0.0, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 775}}}, "reference_key_id": "keys/synthetic/syn_03009_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "2d43b54be891e5f4db3e2062da0ff5b0470ace3acf8bc02542fac6aa4045349b", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "3f1a80e508235407c2e07730", "attempt_id": "6dbb728e4a27", "run_group_id": "E4_core:A5|closed|qwen3b-llama3b:h2", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:27:54.745264Z", "task_family": "synthetic", "document_id": "syn_03009_2000", "document_hash": "6ed94dcf728255bd7a7d11c97c362930728fb81a3e91f1906eb6eec86e0fedb7", "question_id": "q09_revision", "split": "test", "cluster_id": "syn_03009_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "tokenizer_hashes": {"receiver": "b9d8872cf3a99da99e1fc4b0f71232a451632dc49ca4c6771549b4efa5c1c763"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 2, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1770, "query_tokens": 370, "message_tokens": 255, "message_utf8_bytes": 775, "cache_tensor_bytes": null, "serialized_payload_bytes": 775, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 27049.207000178285, "hops": {"1": 16777.417791076005, "2": 10271.78920910228}, "receiver_prefill_ms": 318.62520903814584, "ttft_ms": 35.57691699825227, "decode_ms": 25.969958980567753}, "total_wall_ms": 27429.854916175827, "ttft_ms": 35.57691699825227, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 12, "retention_rate": 0.18461538461538463, "retention_by_kind": {"person": 1.0, "date": 0.0, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 775}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 12, "retention_rate": 0.18461538461538463, "retention_by_kind": {"person": 1.0, "date": 0.0, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 698}, "2": {"facts_total": 65, "facts_retained": 12, "retention_rate": 0.18461538461538463, "retention_by_kind": {"person": 1.0, "date": 0.0, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 775}}}, "reference_key_id": "keys/synthetic/syn_03009_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "2d43b54be891e5f4db3e2062da0ff5b0470ace3acf8bc02542fac6aa4045349b", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "b40f1160f4305a85c9ed1e10", "attempt_id": "e513bbaa5324", "run_group_id": "E4_core:A5|closed|qwen3b-llama3b:h2", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:27:55.144379Z", "task_family": "synthetic", "document_id": "syn_03009_2000", "document_hash": "6ed94dcf728255bd7a7d11c97c362930728fb81a3e91f1906eb6eec86e0fedb7", "question_id": "q10_temporal", "split": "test", "cluster_id": "syn_03009_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "tokenizer_hashes": {"receiver": "b9d8872cf3a99da99e1fc4b0f71232a451632dc49ca4c6771549b4efa5c1c763"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 2, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1770, "query_tokens": 378, "message_tokens": 255, "message_utf8_bytes": 775, "cache_tensor_bytes": null, "serialized_payload_bytes": 775, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 27049.207000178285, "hops": {"1": 16777.417791076005, "2": 10271.78920910228}, "receiver_prefill_ms": 322.6091250544414, "ttft_ms": 23.513374966569245, "decode_ms": 175.15145801007748}, "total_wall_ms": 27570.998667157255, "ttft_ms": 23.513374966569245, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 6}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "Eshgamgam-11", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 12, "retention_rate": 0.18461538461538463, "retention_by_kind": {"person": 1.0, "date": 0.0, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 775}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 12, "retention_rate": 0.18461538461538463, "retention_by_kind": {"person": 1.0, "date": 0.0, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 698}, "2": {"facts_total": 65, "facts_retained": 12, "retention_rate": 0.18461538461538463, "retention_by_kind": {"person": 1.0, "date": 0.0, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 775}}}, "reference_key_id": "keys/synthetic/syn_03009_2000.json", "task_scores": {"correct": 0.0, "abstained": 0.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "2d43b54be891e5f4db3e2062da0ff5b0470ace3acf8bc02542fac6aa4045349b", "output_hash": "8c1313c53ab4dbffa5f336d721c810678133eecaa467241e53d8a92150601475"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "5a480a5512cce2f1ee6398ae", "attempt_id": "fea9d7876963", "run_group_id": "E4_core:A5|closed|qwen3b-llama3b:h2", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:27:55.682924Z", "task_family": "synthetic", "document_id": "syn_03009_2000", "document_hash": "6ed94dcf728255bd7a7d11c97c362930728fb81a3e91f1906eb6eec86e0fedb7", "question_id": "q11_aggregation", "split": "test", "cluster_id": "syn_03009_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "tokenizer_hashes": {"receiver": "b9d8872cf3a99da99e1fc4b0f71232a451632dc49ca4c6771549b4efa5c1c763"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 2, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1770, "query_tokens": 371, "message_tokens": 255, "message_utf8_bytes": 775, "cache_tensor_bytes": null, "serialized_payload_bytes": 775, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 27049.207000178285, "hops": {"1": 16777.417791076005, "2": 10271.78920910228}, "receiver_prefill_ms": 316.87020789831877, "ttft_ms": 95.80066695343703, "decode_ms": 136.9559169979766}, "total_wall_ms": 27599.339500186034, "ttft_ms": 95.80066695343703, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 12, "retention_rate": 0.18461538461538463, "retention_by_kind": {"person": 1.0, "date": 0.0, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 775}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 12, "retention_rate": 0.18461538461538463, "retention_by_kind": {"person": 1.0, "date": 0.0, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 698}, "2": {"facts_total": 65, "facts_retained": 12, "retention_rate": 0.18461538461538463, "retention_by_kind": {"person": 1.0, "date": 0.0, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 775}}}, "reference_key_id": "keys/synthetic/syn_03009_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "2d43b54be891e5f4db3e2062da0ff5b0470ace3acf8bc02542fac6aa4045349b", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "64bc6b9c5722ac3f6783d9ba", "attempt_id": "6fab6c1e1de4", "run_group_id": "E4_core:A5|closed|qwen3b-llama3b:h2", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:27:56.251078Z", "task_family": "synthetic", "document_id": "syn_03009_2000", "document_hash": "6ed94dcf728255bd7a7d11c97c362930728fb81a3e91f1906eb6eec86e0fedb7", "question_id": "q12_unit_conversion", "split": "test", "cluster_id": "syn_03009_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "tokenizer_hashes": {"receiver": "b9d8872cf3a99da99e1fc4b0f71232a451632dc49ca4c6771549b4efa5c1c763"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 2, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1770, "query_tokens": 368, "message_tokens": 255, "message_utf8_bytes": 775, "cache_tensor_bytes": null, "serialized_payload_bytes": 775, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 27049.207000178285, "hops": {"1": 16777.417791076005, "2": 10271.78920910228}, "receiver_prefill_ms": 321.9797919737175, "ttft_ms": 21.097750053741038, "decode_ms": 41.61104199010879}, "total_wall_ms": 27434.37929218635, "ttft_ms": 21.097750053741038, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 12, "retention_rate": 0.18461538461538463, "retention_by_kind": {"person": 1.0, "date": 0.0, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 775}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 12, "retention_rate": 0.18461538461538463, "retention_by_kind": {"person": 1.0, "date": 0.0, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 698}, "2": {"facts_total": 65, "facts_retained": 12, "retention_rate": 0.18461538461538463, "retention_by_kind": {"person": 1.0, "date": 0.0, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 775}}}, "reference_key_id": "keys/synthetic/syn_03009_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "2d43b54be891e5f4db3e2062da0ff5b0470ace3acf8bc02542fac6aa4045349b", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "1f483990463b09711d48852e", "attempt_id": "df56b47e85c4", "run_group_id": "E4_core:A5|closed|qwen3b-llama3b:h2", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:27:56.653998Z", "task_family": "synthetic", "document_id": "syn_03009_2000", "document_hash": "6ed94dcf728255bd7a7d11c97c362930728fb81a3e91f1906eb6eec86e0fedb7", "question_id": "q13_unanswerable", "split": "test", "cluster_id": "syn_03009_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "tokenizer_hashes": {"receiver": "b9d8872cf3a99da99e1fc4b0f71232a451632dc49ca4c6771549b4efa5c1c763"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 2, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1770, "query_tokens": 372, "message_tokens": 255, "message_utf8_bytes": 775, "cache_tensor_bytes": null, "serialized_payload_bytes": 775, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 27049.207000178285, "hops": {"1": 16777.417791076005, "2": 10271.78920910228}, "receiver_prefill_ms": 336.2016660394147, "ttft_ms": 23.862582980655134, "decode_ms": 31.127208028919995}, "total_wall_ms": 27440.798500087112, "ttft_ms": 23.862582980655134, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 12, "retention_rate": 0.18461538461538463, "retention_by_kind": {"person": 1.0, "date": 0.0, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 775}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 12, "retention_rate": 0.18461538461538463, "retention_by_kind": {"person": 1.0, "date": 0.0, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 698}, "2": {"facts_total": 65, "facts_retained": 12, "retention_rate": 0.18461538461538463, "retention_by_kind": {"person": 1.0, "date": 0.0, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 775}}}, "reference_key_id": "keys/synthetic/syn_03009_2000.json", "task_scores": {"correct": 1.0, "abstained": 1.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "2d43b54be891e5f4db3e2062da0ff5b0470ace3acf8bc02542fac6aa4045349b", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "1d6fd93c8915fc1eb7284ca4", "attempt_id": "233141276968", "run_group_id": "E4_core:A5|closed|qwen3b-llama3b:h2", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:27:57.063150Z", "task_family": "synthetic", "document_id": "syn_03009_2000", "document_hash": "6ed94dcf728255bd7a7d11c97c362930728fb81a3e91f1906eb6eec86e0fedb7", "question_id": "q14_unanswerable", "split": "test", "cluster_id": "syn_03009_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "tokenizer_hashes": {"receiver": "b9d8872cf3a99da99e1fc4b0f71232a451632dc49ca4c6771549b4efa5c1c763"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 2, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1770, "query_tokens": 373, "message_tokens": 255, "message_utf8_bytes": 775, "cache_tensor_bytes": null, "serialized_payload_bytes": 775, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 27049.207000178285, "hops": {"1": 16777.417791076005, "2": 10271.78920910228}, "receiver_prefill_ms": 273.05637497920543, "ttft_ms": 44.627040973864496, "decode_ms": 39.35029194690287}, "total_wall_ms": 27406.73425013665, "ttft_ms": 44.627040973864496, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 12, "retention_rate": 0.18461538461538463, "retention_by_kind": {"person": 1.0, "date": 0.0, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 775}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 12, "retention_rate": 0.18461538461538463, "retention_by_kind": {"person": 1.0, "date": 0.0, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 698}, "2": {"facts_total": 65, "facts_retained": 12, "retention_rate": 0.18461538461538463, "retention_by_kind": {"person": 1.0, "date": 0.0, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 775}}}, "reference_key_id": "keys/synthetic/syn_03009_2000.json", "task_scores": {"correct": 1.0, "abstained": 1.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "2d43b54be891e5f4db3e2062da0ff5b0470ace3acf8bc02542fac6aa4045349b", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "005c4afd91d379304ea515e2", "attempt_id": "154f75b294b8", "run_group_id": "E4_core:A5|closed|qwen3b-llama3b:h4", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:28:18.690033Z", "task_family": "synthetic", "document_id": "syn_03009_2000", "document_hash": "6ed94dcf728255bd7a7d11c97c362930728fb81a3e91f1906eb6eec86e0fedb7", "question_id": "q00_single_fact", "split": "test", "cluster_id": "syn_03009_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "tokenizer_hashes": {"receiver": "b9d8872cf3a99da99e1fc4b0f71232a451632dc49ca4c6771549b4efa5c1c763"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 4, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1770, "query_tokens": 360, "message_tokens": 255, "message_utf8_bytes": 775, "cache_tensor_bytes": null, "serialized_payload_bytes": 775, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 48297.88445925806, "hops": {"1": 16777.417791076005, "2": 10271.78920910228, "3": 11040.38683406543, "4": 10208.290625014342}, "receiver_prefill_ms": 313.59816703479737, "ttft_ms": 29.29791691713035, "decode_ms": 180.25024991948158}, "total_wall_ms": 48821.51354232337, "ttft_ms": 29.29791691713035, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 6}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "Eshgamgam-11", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 12, "retention_rate": 0.18461538461538463, "retention_by_kind": {"person": 1.0, "date": 0.0, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 775}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 12, "retention_rate": 0.18461538461538463, "retention_by_kind": {"person": 1.0, "date": 0.0, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 698}, "2": {"facts_total": 65, "facts_retained": 12, "retention_rate": 0.18461538461538463, "retention_by_kind": {"person": 1.0, "date": 0.0, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 775}, "3": {"facts_total": 65, "facts_retained": 12, "retention_rate": 0.18461538461538463, "retention_by_kind": {"person": 1.0, "date": 0.0, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 775}, "4": {"facts_total": 65, "facts_retained": 12, "retention_rate": 0.18461538461538463, "retention_by_kind": {"person": 1.0, "date": 0.0, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 775}}}, "reference_key_id": "keys/synthetic/syn_03009_2000.json", "task_scores": {"correct": 0.0, "abstained": 0.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "2d43b54be891e5f4db3e2062da0ff5b0470ace3acf8bc02542fac6aa4045349b", "output_hash": "8c1313c53ab4dbffa5f336d721c810678133eecaa467241e53d8a92150601475"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "fbd7e171e28011ebc29440c3", "attempt_id": "14d4415abfe6", "run_group_id": "E4_core:A5|closed|qwen3b-llama3b:h4", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:28:19.231448Z", "task_family": "synthetic", "document_id": "syn_03009_2000", "document_hash": "6ed94dcf728255bd7a7d11c97c362930728fb81a3e91f1906eb6eec86e0fedb7", "question_id": "q01_single_fact", "split": "test", "cluster_id": "syn_03009_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "tokenizer_hashes": {"receiver": "b9d8872cf3a99da99e1fc4b0f71232a451632dc49ca4c6771549b4efa5c1c763"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 4, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1770, "query_tokens": 365, "message_tokens": 255, "message_utf8_bytes": 775, "cache_tensor_bytes": null, "serialized_payload_bytes": 775, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 48297.88445925806, "hops": {"1": 16777.417791076005, "2": 10271.78920910228, "3": 11040.38683406543, "4": 10208.290625014342}, "receiver_prefill_ms": 283.8717499980703, "ttft_ms": 46.283416100777686, "decode_ms": 64.9918339913711}, "total_wall_ms": 48693.60170932487, "ttft_ms": 46.283416100777686, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 12, "retention_rate": 0.18461538461538463, "retention_by_kind": {"person": 1.0, "date": 0.0, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 775}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 12, "retention_rate": 0.18461538461538463, "retention_by_kind": {"person": 1.0, "date": 0.0, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 698}, "2": {"facts_total": 65, "facts_retained": 12, "retention_rate": 0.18461538461538463, "retention_by_kind": {"person": 1.0, "date": 0.0, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 775}, "3": {"facts_total": 65, "facts_retained": 12, "retention_rate": 0.18461538461538463, "retention_by_kind": {"person": 1.0, "date": 0.0, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 775}, "4": {"facts_total": 65, "facts_retained": 12, "retention_rate": 0.18461538461538463, "retention_by_kind": {"person": 1.0, "date": 0.0, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 775}}}, "reference_key_id": "keys/synthetic/syn_03009_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "2d43b54be891e5f4db3e2062da0ff5b0470ace3acf8bc02542fac6aa4045349b", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "5103f3fb41e7abb85a015422", "attempt_id": "a3bb3cb1330b", "run_group_id": "E4_core:A5|closed|qwen3b-llama3b:h4", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:28:19.644795Z", "task_family": "synthetic", "document_id": "syn_03009_2000", "document_hash": "6ed94dcf728255bd7a7d11c97c362930728fb81a3e91f1906eb6eec86e0fedb7", "question_id": "q02_exact_copy", "split": "test", "cluster_id": "syn_03009_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "tokenizer_hashes": {"receiver": "b9d8872cf3a99da99e1fc4b0f71232a451632dc49ca4c6771549b4efa5c1c763"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 4, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1770, "query_tokens": 364, "message_tokens": 255, "message_utf8_bytes": 775, "cache_tensor_bytes": null, "serialized_payload_bytes": 775, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 48297.88445925806, "hops": {"1": 16777.417791076005, "2": 10271.78920910228, "3": 11040.38683406543, "4": 10208.290625014342}, "receiver_prefill_ms": 443.6497500864789, "ttft_ms": 38.938374957069755, "decode_ms": 23.763292003422976}, "total_wall_ms": 48804.7169172205, "ttft_ms": 38.938374957069755, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 12, "retention_rate": 0.18461538461538463, "retention_by_kind": {"person": 1.0, "date": 0.0, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 775}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 12, "retention_rate": 0.18461538461538463, "retention_by_kind": {"person": 1.0, "date": 0.0, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 698}, "2": {"facts_total": 65, "facts_retained": 12, "retention_rate": 0.18461538461538463, "retention_by_kind": {"person": 1.0, "date": 0.0, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 775}, "3": {"facts_total": 65, "facts_retained": 12, "retention_rate": 0.18461538461538463, "retention_by_kind": {"person": 1.0, "date": 0.0, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 775}, "4": {"facts_total": 65, "facts_retained": 12, "retention_rate": 0.18461538461538463, "retention_by_kind": {"person": 1.0, "date": 0.0, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 775}}}, "reference_key_id": "keys/synthetic/syn_03009_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "2d43b54be891e5f4db3e2062da0ff5b0470ace3acf8bc02542fac6aa4045349b", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "44def2020e9b2a2ec2d65cb2", "attempt_id": "ef24c2ec7bc1", "run_group_id": "E4_core:A5|closed|qwen3b-llama3b:h4", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:28:20.168804Z", "task_family": "synthetic", "document_id": "syn_03009_2000", "document_hash": "6ed94dcf728255bd7a7d11c97c362930728fb81a3e91f1906eb6eec86e0fedb7", "question_id": "q03_exact_copy", "split": "test", "cluster_id": "syn_03009_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "tokenizer_hashes": {"receiver": "b9d8872cf3a99da99e1fc4b0f71232a451632dc49ca4c6771549b4efa5c1c763"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 4, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1770, "query_tokens": 359, "message_tokens": 255, "message_utf8_bytes": 775, "cache_tensor_bytes": null, "serialized_payload_bytes": 775, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 48297.88445925806, "hops": {"1": 16777.417791076005, "2": 10271.78920910228, "3": 11040.38683406543, "4": 10208.290625014342}, "receiver_prefill_ms": 317.99929100088775, "ttft_ms": 29.26941600162536, "decode_ms": 152.2160420427099}, "total_wall_ms": 48797.89120925125, "ttft_ms": 29.26941600162536, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 5}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "Idris Bergström", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 12, "retention_rate": 0.18461538461538463, "retention_by_kind": {"person": 1.0, "date": 0.0, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 775}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 12, "retention_rate": 0.18461538461538463, "retention_by_kind": {"person": 1.0, "date": 0.0, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 698}, "2": {"facts_total": 65, "facts_retained": 12, "retention_rate": 0.18461538461538463, "retention_by_kind": {"person": 1.0, "date": 0.0, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 775}, "3": {"facts_total": 65, "facts_retained": 12, "retention_rate": 0.18461538461538463, "retention_by_kind": {"person": 1.0, "date": 0.0, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 775}, "4": {"facts_total": 65, "facts_retained": 12, "retention_rate": 0.18461538461538463, "retention_by_kind": {"person": 1.0, "date": 0.0, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 775}}}, "reference_key_id": "keys/synthetic/syn_03009_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "2d43b54be891e5f4db3e2062da0ff5b0470ace3acf8bc02542fac6aa4045349b", "output_hash": "fd88832c9fd0ab175818d76d875d70156225bd4b34ec401514546ef70f4f826c"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "2d9ae75123b75d5923597464", "attempt_id": "f191af3e3656", "run_group_id": "E4_core:A5|closed|qwen3b-llama3b:h4", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:28:20.686339Z", "task_family": "synthetic", "document_id": "syn_03009_2000", "document_hash": "6ed94dcf728255bd7a7d11c97c362930728fb81a3e91f1906eb6eec86e0fedb7", "question_id": "q04_multi_hop", "split": "test", "cluster_id": "syn_03009_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "tokenizer_hashes": {"receiver": "b9d8872cf3a99da99e1fc4b0f71232a451632dc49ca4c6771549b4efa5c1c763"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 4, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1770, "query_tokens": 368, "message_tokens": 255, "message_utf8_bytes": 775, "cache_tensor_bytes": null, "serialized_payload_bytes": 775, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 48297.88445925806, "hops": {"1": 16777.417791076005, "2": 10271.78920910228, "3": 11040.38683406543, "4": 10208.290625014342}, "receiver_prefill_ms": 266.9207089347765, "ttft_ms": 40.419291937723756, "decode_ms": 50.278292037546635}, "total_wall_ms": 48656.00487624761, "ttft_ms": 40.419291937723756, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 12, "retention_rate": 0.18461538461538463, "retention_by_kind": {"person": 1.0, "date": 0.0, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 775}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 12, "retention_rate": 0.18461538461538463, "retention_by_kind": {"person": 1.0, "date": 0.0, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 698}, "2": {"facts_total": 65, "facts_retained": 12, "retention_rate": 0.18461538461538463, "retention_by_kind": {"person": 1.0, "date": 0.0, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 775}, "3": {"facts_total": 65, "facts_retained": 12, "retention_rate": 0.18461538461538463, "retention_by_kind": {"person": 1.0, "date": 0.0, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 775}, "4": {"facts_total": 65, "facts_retained": 12, "retention_rate": 0.18461538461538463, "retention_by_kind": {"person": 1.0, "date": 0.0, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 775}}}, "reference_key_id": "keys/synthetic/syn_03009_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "2d43b54be891e5f4db3e2062da0ff5b0470ace3acf8bc02542fac6aa4045349b", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "31060603a03a07bfedc2f544", "attempt_id": "5b36f43f5bb4", "run_group_id": "E4_core:A5|closed|qwen3b-llama3b:h4", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:28:21.061615Z", "task_family": "synthetic", "document_id": "syn_03009_2000", "document_hash": "6ed94dcf728255bd7a7d11c97c362930728fb81a3e91f1906eb6eec86e0fedb7", "question_id": "q05_multi_hop", "split": "test", "cluster_id": "syn_03009_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "tokenizer_hashes": {"receiver": "b9d8872cf3a99da99e1fc4b0f71232a451632dc49ca4c6771549b4efa5c1c763"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 4, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1770, "query_tokens": 369, "message_tokens": 255, "message_utf8_bytes": 775, "cache_tensor_bytes": null, "serialized_payload_bytes": 775, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 48297.88445925806, "hops": {"1": 16777.417791076005, "2": 10271.78920910228, "3": 11040.38683406543, "4": 10208.290625014342}, "receiver_prefill_ms": 448.02808307576925, "ttft_ms": 54.60908403620124, "decode_ms": 26.049999985843897}, "total_wall_ms": 48827.0202933345, "ttft_ms": 54.60908403620124, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 12, "retention_rate": 0.18461538461538463, "retention_by_kind": {"person": 1.0, "date": 0.0, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 775}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 12, "retention_rate": 0.18461538461538463, "retention_by_kind": {"person": 1.0, "date": 0.0, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 698}, "2": {"facts_total": 65, "facts_retained": 12, "retention_rate": 0.18461538461538463, "retention_by_kind": {"person": 1.0, "date": 0.0, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 775}, "3": {"facts_total": 65, "facts_retained": 12, "retention_rate": 0.18461538461538463, "retention_by_kind": {"person": 1.0, "date": 0.0, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 775}, "4": {"facts_total": 65, "facts_retained": 12, "retention_rate": 0.18461538461538463, "retention_by_kind": {"person": 1.0, "date": 0.0, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 775}}}, "reference_key_id": "keys/synthetic/syn_03009_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "2d43b54be891e5f4db3e2062da0ff5b0470ace3acf8bc02542fac6aa4045349b", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "d6a708732750b8eb93acb435", "attempt_id": "a5beb5cbcad4", "run_group_id": "E4_core:A5|closed|qwen3b-llama3b:h4", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:28:21.609828Z", "task_family": "synthetic", "document_id": "syn_03009_2000", "document_hash": "6ed94dcf728255bd7a7d11c97c362930728fb81a3e91f1906eb6eec86e0fedb7", "question_id": "q06_negation", "split": "test", "cluster_id": "syn_03009_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "tokenizer_hashes": {"receiver": "b9d8872cf3a99da99e1fc4b0f71232a451632dc49ca4c6771549b4efa5c1c763"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 4, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1770, "query_tokens": 361, "message_tokens": 255, "message_utf8_bytes": 775, "cache_tensor_bytes": null, "serialized_payload_bytes": 775, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 48297.88445925806, "hops": {"1": 16777.417791076005, "2": 10271.78920910228, "3": 11040.38683406543, "4": 10208.290625014342}, "receiver_prefill_ms": 327.78991700615734, "ttft_ms": 44.726124964654446, "decode_ms": 25.144917075522244}, "total_wall_ms": 48695.99229225423, "ttft_ms": 44.726124964654446, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "No", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 12, "retention_rate": 0.18461538461538463, "retention_by_kind": {"person": 1.0, "date": 0.0, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 775}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 12, "retention_rate": 0.18461538461538463, "retention_by_kind": {"person": 1.0, "date": 0.0, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 698}, "2": {"facts_total": 65, "facts_retained": 12, "retention_rate": 0.18461538461538463, "retention_by_kind": {"person": 1.0, "date": 0.0, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 775}, "3": {"facts_total": 65, "facts_retained": 12, "retention_rate": 0.18461538461538463, "retention_by_kind": {"person": 1.0, "date": 0.0, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 775}, "4": {"facts_total": 65, "facts_retained": 12, "retention_rate": 0.18461538461538463, "retention_by_kind": {"person": 1.0, "date": 0.0, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 775}}}, "reference_key_id": "keys/synthetic/syn_03009_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "2d43b54be891e5f4db3e2062da0ff5b0470ace3acf8bc02542fac6aa4045349b", "output_hash": "1ea442a134b2a184bd5d40104401f2a37fbc09ccf3f4bc9da161c6099be3691d"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "6d186d53037f2dec43795c6f", "attempt_id": "337b07cfba78", "run_group_id": "E4_core:A5|closed|qwen3b-llama3b:h4", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:28:22.025061Z", "task_family": "synthetic", "document_id": "syn_03009_2000", "document_hash": "6ed94dcf728255bd7a7d11c97c362930728fb81a3e91f1906eb6eec86e0fedb7", "question_id": "q07_negation", "split": "test", "cluster_id": "syn_03009_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "tokenizer_hashes": {"receiver": "b9d8872cf3a99da99e1fc4b0f71232a451632dc49ca4c6771549b4efa5c1c763"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 4, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1770, "query_tokens": 360, "message_tokens": 255, "message_utf8_bytes": 775, "cache_tensor_bytes": null, "serialized_payload_bytes": 775, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 48297.88445925806, "hops": {"1": 16777.417791076005, "2": 10271.78920910228, "3": 11040.38683406543, "4": 10208.290625014342}, "receiver_prefill_ms": 307.0667079882696, "ttft_ms": 36.74816607963294, "decode_ms": 17.484999960288405}, "total_wall_ms": 48659.63170921896, "ttft_ms": 36.74816607963294, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "No", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 12, "retention_rate": 0.18461538461538463, "retention_by_kind": {"person": 1.0, "date": 0.0, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 775}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 12, "retention_rate": 0.18461538461538463, "retention_by_kind": {"person": 1.0, "date": 0.0, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 698}, "2": {"facts_total": 65, "facts_retained": 12, "retention_rate": 0.18461538461538463, "retention_by_kind": {"person": 1.0, "date": 0.0, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 775}, "3": {"facts_total": 65, "facts_retained": 12, "retention_rate": 0.18461538461538463, "retention_by_kind": {"person": 1.0, "date": 0.0, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 775}, "4": {"facts_total": 65, "facts_retained": 12, "retention_rate": 0.18461538461538463, "retention_by_kind": {"person": 1.0, "date": 0.0, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 775}}}, "reference_key_id": "keys/synthetic/syn_03009_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "2d43b54be891e5f4db3e2062da0ff5b0470ace3acf8bc02542fac6aa4045349b", "output_hash": "1ea442a134b2a184bd5d40104401f2a37fbc09ccf3f4bc9da161c6099be3691d"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "8376009ac1ab8f259c3b3153", "attempt_id": "3e85f7d12efc", "run_group_id": "E4_core:A5|closed|qwen3b-llama3b:h4", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:28:22.403970Z", "task_family": "synthetic", "document_id": "syn_03009_2000", "document_hash": "6ed94dcf728255bd7a7d11c97c362930728fb81a3e91f1906eb6eec86e0fedb7", "question_id": "q08_revision", "split": "test", "cluster_id": "syn_03009_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "tokenizer_hashes": {"receiver": "b9d8872cf3a99da99e1fc4b0f71232a451632dc49ca4c6771549b4efa5c1c763"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 4, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1770, "query_tokens": 377, "message_tokens": 255, "message_utf8_bytes": 775, "cache_tensor_bytes": null, "serialized_payload_bytes": 775, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 48297.88445925806, "hops": {"1": 16777.417791076005, "2": 10271.78920910228, "3": 11040.38683406543, "4": 10208.290625014342}, "receiver_prefill_ms": 327.3807499790564, "ttft_ms": 38.79904199857265, "decode_ms": 53.74191596638411}, "total_wall_ms": 48718.26216822956, "ttft_ms": 38.79904199857265, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 12, "retention_rate": 0.18461538461538463, "retention_by_kind": {"person": 1.0, "date": 0.0, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 775}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 12, "retention_rate": 0.18461538461538463, "retention_by_kind": {"person": 1.0, "date": 0.0, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 698}, "2": {"facts_total": 65, "facts_retained": 12, "retention_rate": 0.18461538461538463, "retention_by_kind": {"person": 1.0, "date": 0.0, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 775}, "3": {"facts_total": 65, "facts_retained": 12, "retention_rate": 0.18461538461538463, "retention_by_kind": {"person": 1.0, "date": 0.0, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 775}, "4": {"facts_total": 65, "facts_retained": 12, "retention_rate": 0.18461538461538463, "retention_by_kind": {"person": 1.0, "date": 0.0, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 775}}}, "reference_key_id": "keys/synthetic/syn_03009_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "2d43b54be891e5f4db3e2062da0ff5b0470ace3acf8bc02542fac6aa4045349b", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "f569a7caf0041203661c1ba3", "attempt_id": "6897554358a7", "run_group_id": "E4_core:A5|closed|qwen3b-llama3b:h4", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:28:22.842267Z", "task_family": "synthetic", "document_id": "syn_03009_2000", "document_hash": "6ed94dcf728255bd7a7d11c97c362930728fb81a3e91f1906eb6eec86e0fedb7", "question_id": "q09_revision", "split": "test", "cluster_id": "syn_03009_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "tokenizer_hashes": {"receiver": "b9d8872cf3a99da99e1fc4b0f71232a451632dc49ca4c6771549b4efa5c1c763"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 4, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1770, "query_tokens": 370, "message_tokens": 255, "message_utf8_bytes": 775, "cache_tensor_bytes": null, "serialized_payload_bytes": 775, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 48297.88445925806, "hops": {"1": 16777.417791076005, "2": 10271.78920910228, "3": 11040.38683406543, "4": 10208.290625014342}, "receiver_prefill_ms": 446.1116249440238, "ttft_ms": 25.792457978241146, "decode_ms": 40.293124970048666}, "total_wall_ms": 48810.61020924244, "ttft_ms": 25.792457978241146, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 12, "retention_rate": 0.18461538461538463, "retention_by_kind": {"person": 1.0, "date": 0.0, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 775}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 12, "retention_rate": 0.18461538461538463, "retention_by_kind": {"person": 1.0, "date": 0.0, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 698}, "2": {"facts_total": 65, "facts_retained": 12, "retention_rate": 0.18461538461538463, "retention_by_kind": {"person": 1.0, "date": 0.0, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 775}, "3": {"facts_total": 65, "facts_retained": 12, "retention_rate": 0.18461538461538463, "retention_by_kind": {"person": 1.0, "date": 0.0, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 775}, "4": {"facts_total": 65, "facts_retained": 12, "retention_rate": 0.18461538461538463, "retention_by_kind": {"person": 1.0, "date": 0.0, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 775}}}, "reference_key_id": "keys/synthetic/syn_03009_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "2d43b54be891e5f4db3e2062da0ff5b0470ace3acf8bc02542fac6aa4045349b", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "6628ac11782b7d8288087790", "attempt_id": "e3570222460c", "run_group_id": "E4_core:A5|closed|qwen3b-llama3b:h4", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:28:23.374388Z", "task_family": "synthetic", "document_id": "syn_03009_2000", "document_hash": "6ed94dcf728255bd7a7d11c97c362930728fb81a3e91f1906eb6eec86e0fedb7", "question_id": "q10_temporal", "split": "test", "cluster_id": "syn_03009_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "tokenizer_hashes": {"receiver": "b9d8872cf3a99da99e1fc4b0f71232a451632dc49ca4c6771549b4efa5c1c763"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 4, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1770, "query_tokens": 378, "message_tokens": 255, "message_utf8_bytes": 775, "cache_tensor_bytes": null, "serialized_payload_bytes": 775, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 48297.88445925806, "hops": {"1": 16777.417791076005, "2": 10271.78920910228, "3": 11040.38683406543, "4": 10208.290625014342}, "receiver_prefill_ms": 324.28675005212426, "ttft_ms": 24.93087504990399, "decode_ms": 183.108540950343}, "total_wall_ms": 48830.66900016274, "ttft_ms": 24.93087504990399, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 6}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "Eshgamgam-11", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 12, "retention_rate": 0.18461538461538463, "retention_by_kind": {"person": 1.0, "date": 0.0, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 775}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 12, "retention_rate": 0.18461538461538463, "retention_by_kind": {"person": 1.0, "date": 0.0, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 698}, "2": {"facts_total": 65, "facts_retained": 12, "retention_rate": 0.18461538461538463, "retention_by_kind": {"person": 1.0, "date": 0.0, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 775}, "3": {"facts_total": 65, "facts_retained": 12, "retention_rate": 0.18461538461538463, "retention_by_kind": {"person": 1.0, "date": 0.0, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 775}, "4": {"facts_total": 65, "facts_retained": 12, "retention_rate": 0.18461538461538463, "retention_by_kind": {"person": 1.0, "date": 0.0, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 775}}}, "reference_key_id": "keys/synthetic/syn_03009_2000.json", "task_scores": {"correct": 0.0, "abstained": 0.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "2d43b54be891e5f4db3e2062da0ff5b0470ace3acf8bc02542fac6aa4045349b", "output_hash": "8c1313c53ab4dbffa5f336d721c810678133eecaa467241e53d8a92150601475"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "d66c662322aa926a77664ab7", "attempt_id": "4321c618e8ae", "run_group_id": "E4_core:A5|closed|qwen3b-llama3b:h4", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:28:23.924516Z", "task_family": "synthetic", "document_id": "syn_03009_2000", "document_hash": "6ed94dcf728255bd7a7d11c97c362930728fb81a3e91f1906eb6eec86e0fedb7", "question_id": "q11_aggregation", "split": "test", "cluster_id": "syn_03009_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "tokenizer_hashes": {"receiver": "b9d8872cf3a99da99e1fc4b0f71232a451632dc49ca4c6771549b4efa5c1c763"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 4, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1770, "query_tokens": 371, "message_tokens": 255, "message_utf8_bytes": 775, "cache_tensor_bytes": null, "serialized_payload_bytes": 775, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 48297.88445925806, "hops": {"1": 16777.417791076005, "2": 10271.78920910228, "3": 11040.38683406543, "4": 10208.290625014342}, "receiver_prefill_ms": 284.32299999985844, "ttft_ms": 38.49500010255724, "decode_ms": 47.17620799783617}, "total_wall_ms": 48668.40587626211, "ttft_ms": 38.49500010255724, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 12, "retention_rate": 0.18461538461538463, "retention_by_kind": {"person": 1.0, "date": 0.0, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 775}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 12, "retention_rate": 0.18461538461538463, "retention_by_kind": {"person": 1.0, "date": 0.0, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 698}, "2": {"facts_total": 65, "facts_retained": 12, "retention_rate": 0.18461538461538463, "retention_by_kind": {"person": 1.0, "date": 0.0, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 775}, "3": {"facts_total": 65, "facts_retained": 12, "retention_rate": 0.18461538461538463, "retention_by_kind": {"person": 1.0, "date": 0.0, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 775}, "4": {"facts_total": 65, "facts_retained": 12, "retention_rate": 0.18461538461538463, "retention_by_kind": {"person": 1.0, "date": 0.0, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 775}}}, "reference_key_id": "keys/synthetic/syn_03009_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "2d43b54be891e5f4db3e2062da0ff5b0470ace3acf8bc02542fac6aa4045349b", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "737ae5ec8bcb86271fb2e2ef", "attempt_id": "be307bfdd188", "run_group_id": "E4_core:A5|closed|qwen3b-llama3b:h4", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:28:24.313267Z", "task_family": "synthetic", "document_id": "syn_03009_2000", "document_hash": "6ed94dcf728255bd7a7d11c97c362930728fb81a3e91f1906eb6eec86e0fedb7", "question_id": "q12_unit_conversion", "split": "test", "cluster_id": "syn_03009_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "tokenizer_hashes": {"receiver": "b9d8872cf3a99da99e1fc4b0f71232a451632dc49ca4c6771549b4efa5c1c763"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 4, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1770, "query_tokens": 368, "message_tokens": 255, "message_utf8_bytes": 775, "cache_tensor_bytes": null, "serialized_payload_bytes": 775, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 48297.88445925806, "hops": {"1": 16777.417791076005, "2": 10271.78920910228, "3": 11040.38683406543, "4": 10208.290625014342}, "receiver_prefill_ms": 454.3464579619467, "ttft_ms": 41.21737496461719, "decode_ms": 25.013625039719045}, "total_wall_ms": 48818.89712624252, "ttft_ms": 41.21737496461719, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 12, "retention_rate": 0.18461538461538463, "retention_by_kind": {"person": 1.0, "date": 0.0, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 775}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 12, "retention_rate": 0.18461538461538463, "retention_by_kind": {"person": 1.0, "date": 0.0, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 698}, "2": {"facts_total": 65, "facts_retained": 12, "retention_rate": 0.18461538461538463, "retention_by_kind": {"person": 1.0, "date": 0.0, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 775}, "3": {"facts_total": 65, "facts_retained": 12, "retention_rate": 0.18461538461538463, "retention_by_kind": {"person": 1.0, "date": 0.0, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 775}, "4": {"facts_total": 65, "facts_retained": 12, "retention_rate": 0.18461538461538463, "retention_by_kind": {"person": 1.0, "date": 0.0, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 775}}}, "reference_key_id": "keys/synthetic/syn_03009_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "2d43b54be891e5f4db3e2062da0ff5b0470ace3acf8bc02542fac6aa4045349b", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "bef5e2960593deb0b69e4b93", "attempt_id": "7cffd4263606", "run_group_id": "E4_core:A5|closed|qwen3b-llama3b:h4", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:28:24.853017Z", "task_family": "synthetic", "document_id": "syn_03009_2000", "document_hash": "6ed94dcf728255bd7a7d11c97c362930728fb81a3e91f1906eb6eec86e0fedb7", "question_id": "q13_unanswerable", "split": "test", "cluster_id": "syn_03009_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "tokenizer_hashes": {"receiver": "b9d8872cf3a99da99e1fc4b0f71232a451632dc49ca4c6771549b4efa5c1c763"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 4, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1770, "query_tokens": 372, "message_tokens": 255, "message_utf8_bytes": 775, "cache_tensor_bytes": null, "serialized_payload_bytes": 775, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 48297.88445925806, "hops": {"1": 16777.417791076005, "2": 10271.78920910228, "3": 11040.38683406543, "4": 10208.290625014342}, "receiver_prefill_ms": 326.0708749294281, "ttft_ms": 21.802583010867238, "decode_ms": 37.908250000327826}, "total_wall_ms": 48684.11195918452, "ttft_ms": 21.802583010867238, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 12, "retention_rate": 0.18461538461538463, "retention_by_kind": {"person": 1.0, "date": 0.0, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 775}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 12, "retention_rate": 0.18461538461538463, "retention_by_kind": {"person": 1.0, "date": 0.0, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 698}, "2": {"facts_total": 65, "facts_retained": 12, "retention_rate": 0.18461538461538463, "retention_by_kind": {"person": 1.0, "date": 0.0, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 775}, "3": {"facts_total": 65, "facts_retained": 12, "retention_rate": 0.18461538461538463, "retention_by_kind": {"person": 1.0, "date": 0.0, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 775}, "4": {"facts_total": 65, "facts_retained": 12, "retention_rate": 0.18461538461538463, "retention_by_kind": {"person": 1.0, "date": 0.0, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 775}}}, "reference_key_id": "keys/synthetic/syn_03009_2000.json", "task_scores": {"correct": 1.0, "abstained": 1.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "2d43b54be891e5f4db3e2062da0ff5b0470ace3acf8bc02542fac6aa4045349b", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "de51994381cf1037fb038107", "attempt_id": "0126dd24deee", "run_group_id": "E4_core:A5|closed|qwen3b-llama3b:h4", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:28:25.257537Z", "task_family": "synthetic", "document_id": "syn_03009_2000", "document_hash": "6ed94dcf728255bd7a7d11c97c362930728fb81a3e91f1906eb6eec86e0fedb7", "question_id": "q14_unanswerable", "split": "test", "cluster_id": "syn_03009_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "tokenizer_hashes": {"receiver": "b9d8872cf3a99da99e1fc4b0f71232a451632dc49ca4c6771549b4efa5c1c763"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 4, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1770, "query_tokens": 373, "message_tokens": 255, "message_utf8_bytes": 775, "cache_tensor_bytes": null, "serialized_payload_bytes": 775, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 48297.88445925806, "hops": {"1": 16777.417791076005, "2": 10271.78920910228, "3": 11040.38683406543, "4": 10208.290625014342}, "receiver_prefill_ms": 293.3645830489695, "ttft_ms": 38.87837496586144, "decode_ms": 56.87516706530005}, "total_wall_ms": 48687.496417202055, "ttft_ms": 38.87837496586144, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 12, "retention_rate": 0.18461538461538463, "retention_by_kind": {"person": 1.0, "date": 0.0, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 775}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 12, "retention_rate": 0.18461538461538463, "retention_by_kind": {"person": 1.0, "date": 0.0, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 698}, "2": {"facts_total": 65, "facts_retained": 12, "retention_rate": 0.18461538461538463, "retention_by_kind": {"person": 1.0, "date": 0.0, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 775}, "3": {"facts_total": 65, "facts_retained": 12, "retention_rate": 0.18461538461538463, "retention_by_kind": {"person": 1.0, "date": 0.0, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 775}, "4": {"facts_total": 65, "facts_retained": 12, "retention_rate": 0.18461538461538463, "retention_by_kind": {"person": 1.0, "date": 0.0, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 775}}}, "reference_key_id": "keys/synthetic/syn_03009_2000.json", "task_scores": {"correct": 1.0, "abstained": 1.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "2d43b54be891e5f4db3e2062da0ff5b0470ace3acf8bc02542fac6aa4045349b", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "863795109ba836a6400a8ffe", "attempt_id": "238889c1d98f", "run_group_id": "E4_core:A5|closed|qwen3b-llama3b:h8", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:29:08.374566Z", "task_family": "synthetic", "document_id": "syn_03009_2000", "document_hash": "6ed94dcf728255bd7a7d11c97c362930728fb81a3e91f1906eb6eec86e0fedb7", "question_id": "q00_single_fact", "split": "test", "cluster_id": "syn_03009_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "tokenizer_hashes": {"receiver": "b9d8872cf3a99da99e1fc4b0f71232a451632dc49ca4c6771549b4efa5c1c763"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 8, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1770, "query_tokens": 360, "message_tokens": 255, "message_utf8_bytes": 775, "cache_tensor_bytes": null, "serialized_payload_bytes": 775, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 91002.5866671931, "hops": {"1": 16777.417791076005, "2": 10271.78920910228, "3": 11040.38683406543, "4": 10208.290625014342, "5": 11365.20437488798, "6": 10238.278790959157, "7": 11114.223542041145, "8": 9986.99550004676}, "receiver_prefill_ms": 454.01924999896437, "ttft_ms": 40.22087494377047, "decode_ms": 185.6031659990549}, "total_wall_ms": 91682.90241714567, "ttft_ms": 40.22087494377047, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 6}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "Eshgamgam-11", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 12, "retention_rate": 0.18461538461538463, "retention_by_kind": {"person": 1.0, "date": 0.0, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 775}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 12, "retention_rate": 0.18461538461538463, "retention_by_kind": {"person": 1.0, "date": 0.0, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 698}, "2": {"facts_total": 65, "facts_retained": 12, "retention_rate": 0.18461538461538463, "retention_by_kind": {"person": 1.0, "date": 0.0, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 775}, "3": {"facts_total": 65, "facts_retained": 12, "retention_rate": 0.18461538461538463, "retention_by_kind": {"person": 1.0, "date": 0.0, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 775}, "4": {"facts_total": 65, "facts_retained": 12, "retention_rate": 0.18461538461538463, "retention_by_kind": {"person": 1.0, "date": 0.0, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 775}, "5": {"facts_total": 65, "facts_retained": 12, "retention_rate": 0.18461538461538463, "retention_by_kind": {"person": 1.0, "date": 0.0, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 775}, "6": {"facts_total": 65, "facts_retained": 12, "retention_rate": 0.18461538461538463, "retention_by_kind": {"person": 1.0, "date": 0.0, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 775}, "7": {"facts_total": 65, "facts_retained": 12, "retention_rate": 0.18461538461538463, "retention_by_kind": {"person": 1.0, "date": 0.0, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 775}, "8": {"facts_total": 65, "facts_retained": 12, "retention_rate": 0.18461538461538463, "retention_by_kind": {"person": 1.0, "date": 0.0, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 775}}}, "reference_key_id": "keys/synthetic/syn_03009_2000.json", "task_scores": {"correct": 0.0, "abstained": 0.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "2d43b54be891e5f4db3e2062da0ff5b0470ace3acf8bc02542fac6aa4045349b", "output_hash": "8c1313c53ab4dbffa5f336d721c810678133eecaa467241e53d8a92150601475"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "1bf370634304ba9dc652d5bb", "attempt_id": "9567530c2b59", "run_group_id": "E4_core:A5|closed|qwen3b-llama3b:h8", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:29:09.072676Z", "task_family": "synthetic", "document_id": "syn_03009_2000", "document_hash": "6ed94dcf728255bd7a7d11c97c362930728fb81a3e91f1906eb6eec86e0fedb7", "question_id": "q01_single_fact", "split": "test", "cluster_id": "syn_03009_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "tokenizer_hashes": {"receiver": "b9d8872cf3a99da99e1fc4b0f71232a451632dc49ca4c6771549b4efa5c1c763"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 8, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1770, "query_tokens": 365, "message_tokens": 255, "message_utf8_bytes": 775, "cache_tensor_bytes": null, "serialized_payload_bytes": 775, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 91002.5866671931, "hops": {"1": 16777.417791076005, "2": 10271.78920910228, "3": 11040.38683406543, "4": 10208.290625014342, "5": 11365.20437488798, "6": 10238.278790959157, "7": 11114.223542041145, "8": 9986.99550004676}, "receiver_prefill_ms": 309.5243340358138, "ttft_ms": 36.85616690199822, "decode_ms": 23.590083001181483}, "total_wall_ms": 91372.93929222506, "ttft_ms": 36.85616690199822, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 12, "retention_rate": 0.18461538461538463, "retention_by_kind": {"person": 1.0, "date": 0.0, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 775}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 12, "retention_rate": 0.18461538461538463, "retention_by_kind": {"person": 1.0, "date": 0.0, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 698}, "2": {"facts_total": 65, "facts_retained": 12, "retention_rate": 0.18461538461538463, "retention_by_kind": {"person": 1.0, "date": 0.0, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 775}, "3": {"facts_total": 65, "facts_retained": 12, "retention_rate": 0.18461538461538463, "retention_by_kind": {"person": 1.0, "date": 0.0, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 775}, "4": {"facts_total": 65, "facts_retained": 12, "retention_rate": 0.18461538461538463, "retention_by_kind": {"person": 1.0, "date": 0.0, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 775}, "5": {"facts_total": 65, "facts_retained": 12, "retention_rate": 0.18461538461538463, "retention_by_kind": {"person": 1.0, "date": 0.0, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 775}, "6": {"facts_total": 65, "facts_retained": 12, "retention_rate": 0.18461538461538463, "retention_by_kind": {"person": 1.0, "date": 0.0, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 775}, "7": {"facts_total": 65, "facts_retained": 12, "retention_rate": 0.18461538461538463, "retention_by_kind": {"person": 1.0, "date": 0.0, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 775}, "8": {"facts_total": 65, "facts_retained": 12, "retention_rate": 0.18461538461538463, "retention_by_kind": {"person": 1.0, "date": 0.0, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 775}}}, "reference_key_id": "keys/synthetic/syn_03009_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "2d43b54be891e5f4db3e2062da0ff5b0470ace3acf8bc02542fac6aa4045349b", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "1ff59dcace9499e673a6bddc", "attempt_id": "ecd31eb1ef96", "run_group_id": "E4_core:A5|closed|qwen3b-llama3b:h8", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:29:09.461206Z", "task_family": "synthetic", "document_id": "syn_03009_2000", "document_hash": "6ed94dcf728255bd7a7d11c97c362930728fb81a3e91f1906eb6eec86e0fedb7", "question_id": "q02_exact_copy", "split": "test", "cluster_id": "syn_03009_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "tokenizer_hashes": {"receiver": "b9d8872cf3a99da99e1fc4b0f71232a451632dc49ca4c6771549b4efa5c1c763"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 8, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1770, "query_tokens": 364, "message_tokens": 255, "message_utf8_bytes": 775, "cache_tensor_bytes": null, "serialized_payload_bytes": 775, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 91002.5866671931, "hops": {"1": 16777.417791076005, "2": 10271.78920910228, "3": 11040.38683406543, "4": 10208.290625014342, "5": 11365.20437488798, "6": 10238.278790959157, "7": 11114.223542041145, "8": 9986.99550004676}, "receiver_prefill_ms": 305.94908306375146, "ttft_ms": 62.78241705149412, "decode_ms": 56.06195796281099}, "total_wall_ms": 91427.96279222239, "ttft_ms": 62.78241705149412, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 12, "retention_rate": 0.18461538461538463, "retention_by_kind": {"person": 1.0, "date": 0.0, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 775}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 12, "retention_rate": 0.18461538461538463, "retention_by_kind": {"person": 1.0, "date": 0.0, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 698}, "2": {"facts_total": 65, "facts_retained": 12, "retention_rate": 0.18461538461538463, "retention_by_kind": {"person": 1.0, "date": 0.0, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 775}, "3": {"facts_total": 65, "facts_retained": 12, "retention_rate": 0.18461538461538463, "retention_by_kind": {"person": 1.0, "date": 0.0, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 775}, "4": {"facts_total": 65, "facts_retained": 12, "retention_rate": 0.18461538461538463, "retention_by_kind": {"person": 1.0, "date": 0.0, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 775}, "5": {"facts_total": 65, "facts_retained": 12, "retention_rate": 0.18461538461538463, "retention_by_kind": {"person": 1.0, "date": 0.0, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 775}, "6": {"facts_total": 65, "facts_retained": 12, "retention_rate": 0.18461538461538463, "retention_by_kind": {"person": 1.0, "date": 0.0, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 775}, "7": {"facts_total": 65, "facts_retained": 12, "retention_rate": 0.18461538461538463, "retention_by_kind": {"person": 1.0, "date": 0.0, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 775}, "8": {"facts_total": 65, "facts_retained": 12, "retention_rate": 0.18461538461538463, "retention_by_kind": {"person": 1.0, "date": 0.0, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 775}}}, "reference_key_id": "keys/synthetic/syn_03009_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "2d43b54be891e5f4db3e2062da0ff5b0470ace3acf8bc02542fac6aa4045349b", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "a131a03b25f6dec3b474d2dd", "attempt_id": "a9667e33a032", "run_group_id": "E4_core:A5|closed|qwen3b-llama3b:h8", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:29:09.903925Z", "task_family": "synthetic", "document_id": "syn_03009_2000", "document_hash": "6ed94dcf728255bd7a7d11c97c362930728fb81a3e91f1906eb6eec86e0fedb7", "question_id": "q03_exact_copy", "split": "test", "cluster_id": "syn_03009_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "tokenizer_hashes": {"receiver": "b9d8872cf3a99da99e1fc4b0f71232a451632dc49ca4c6771549b4efa5c1c763"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 8, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1770, "query_tokens": 359, "message_tokens": 255, "message_utf8_bytes": 775, "cache_tensor_bytes": null, "serialized_payload_bytes": 775, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 91002.5866671931, "hops": {"1": 16777.417791076005, "2": 10271.78920910228, "3": 11040.38683406543, "4": 10208.290625014342, "5": 11365.20437488798, "6": 10238.278790959157, "7": 11114.223542041145, "8": 9986.99550004676}, "receiver_prefill_ms": 367.5940000684932, "ttft_ms": 99.85758294351399, "decode_ms": 157.76466694660485}, "total_wall_ms": 91628.3951671794, "ttft_ms": 99.85758294351399, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 5}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "Idris Bergström", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 12, "retention_rate": 0.18461538461538463, "retention_by_kind": {"person": 1.0, "date": 0.0, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 775}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 12, "retention_rate": 0.18461538461538463, "retention_by_kind": {"person": 1.0, "date": 0.0, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 698}, "2": {"facts_total": 65, "facts_retained": 12, "retention_rate": 0.18461538461538463, "retention_by_kind": {"person": 1.0, "date": 0.0, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 775}, "3": {"facts_total": 65, "facts_retained": 12, "retention_rate": 0.18461538461538463, "retention_by_kind": {"person": 1.0, "date": 0.0, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 775}, "4": {"facts_total": 65, "facts_retained": 12, "retention_rate": 0.18461538461538463, "retention_by_kind": {"person": 1.0, "date": 0.0, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 775}, "5": {"facts_total": 65, "facts_retained": 12, "retention_rate": 0.18461538461538463, "retention_by_kind": {"person": 1.0, "date": 0.0, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 775}, "6": {"facts_total": 65, "facts_retained": 12, "retention_rate": 0.18461538461538463, "retention_by_kind": {"person": 1.0, "date": 0.0, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 775}, "7": {"facts_total": 65, "facts_retained": 12, "retention_rate": 0.18461538461538463, "retention_by_kind": {"person": 1.0, "date": 0.0, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 775}, "8": {"facts_total": 65, "facts_retained": 12, "retention_rate": 0.18461538461538463, "retention_by_kind": {"person": 1.0, "date": 0.0, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 775}}}, "reference_key_id": "keys/synthetic/syn_03009_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "2d43b54be891e5f4db3e2062da0ff5b0470ace3acf8bc02542fac6aa4045349b", "output_hash": "fd88832c9fd0ab175818d76d875d70156225bd4b34ec401514546ef70f4f826c"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "af97c4da5e532d0af4cf7e52", "attempt_id": "473ebc15346f", "run_group_id": "E4_core:A5|closed|qwen3b-llama3b:h8", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:29:10.548545Z", "task_family": "synthetic", "document_id": "syn_03009_2000", "document_hash": "6ed94dcf728255bd7a7d11c97c362930728fb81a3e91f1906eb6eec86e0fedb7", "question_id": "q04_multi_hop", "split": "test", "cluster_id": "syn_03009_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "tokenizer_hashes": {"receiver": "b9d8872cf3a99da99e1fc4b0f71232a451632dc49ca4c6771549b4efa5c1c763"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 8, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1770, "query_tokens": 368, "message_tokens": 255, "message_utf8_bytes": 775, "cache_tensor_bytes": null, "serialized_payload_bytes": 775, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 91002.5866671931, "hops": {"1": 16777.417791076005, "2": 10271.78920910228, "3": 11040.38683406543, "4": 10208.290625014342, "5": 11365.20437488798, "6": 10238.278790959157, "7": 11114.223542041145, "8": 9986.99550004676}, "receiver_prefill_ms": 325.703667011112, "ttft_ms": 37.4701248947531, "decode_ms": 36.21320892125368}, "total_wall_ms": 91402.64258312527, "ttft_ms": 37.4701248947531, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 12, "retention_rate": 0.18461538461538463, "retention_by_kind": {"person": 1.0, "date": 0.0, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 775}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 12, "retention_rate": 0.18461538461538463, "retention_by_kind": {"person": 1.0, "date": 0.0, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 698}, "2": {"facts_total": 65, "facts_retained": 12, "retention_rate": 0.18461538461538463, "retention_by_kind": {"person": 1.0, "date": 0.0, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 775}, "3": {"facts_total": 65, "facts_retained": 12, "retention_rate": 0.18461538461538463, "retention_by_kind": {"person": 1.0, "date": 0.0, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 775}, "4": {"facts_total": 65, "facts_retained": 12, "retention_rate": 0.18461538461538463, "retention_by_kind": {"person": 1.0, "date": 0.0, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 775}, "5": {"facts_total": 65, "facts_retained": 12, "retention_rate": 0.18461538461538463, "retention_by_kind": {"person": 1.0, "date": 0.0, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 775}, "6": {"facts_total": 65, "facts_retained": 12, "retention_rate": 0.18461538461538463, "retention_by_kind": {"person": 1.0, "date": 0.0, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 775}, "7": {"facts_total": 65, "facts_retained": 12, "retention_rate": 0.18461538461538463, "retention_by_kind": {"person": 1.0, "date": 0.0, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 775}, "8": {"facts_total": 65, "facts_retained": 12, "retention_rate": 0.18461538461538463, "retention_by_kind": {"person": 1.0, "date": 0.0, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 775}}}, "reference_key_id": "keys/synthetic/syn_03009_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "2d43b54be891e5f4db3e2062da0ff5b0470ace3acf8bc02542fac6aa4045349b", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "33cbbb61d8ad9811390a5105", "attempt_id": "95bc43cae4fa", "run_group_id": "E4_core:A5|closed|qwen3b-llama3b:h8", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:29:10.966489Z", "task_family": "synthetic", "document_id": "syn_03009_2000", "document_hash": "6ed94dcf728255bd7a7d11c97c362930728fb81a3e91f1906eb6eec86e0fedb7", "question_id": "q05_multi_hop", "split": "test", "cluster_id": "syn_03009_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "tokenizer_hashes": {"receiver": "b9d8872cf3a99da99e1fc4b0f71232a451632dc49ca4c6771549b4efa5c1c763"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 8, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1770, "query_tokens": 369, "message_tokens": 255, "message_utf8_bytes": 775, "cache_tensor_bytes": null, "serialized_payload_bytes": 775, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 91002.5866671931, "hops": {"1": 16777.417791076005, "2": 10271.78920910228, "3": 11040.38683406543, "4": 10208.290625014342, "5": 11365.20437488798, "6": 10238.278790959157, "7": 11114.223542041145, "8": 9986.99550004676}, "receiver_prefill_ms": 288.12116698827595, "ttft_ms": 40.22387496661395, "decode_ms": 44.646042049862444}, "total_wall_ms": 91376.04391714558, "ttft_ms": 40.22387496661395, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 12, "retention_rate": 0.18461538461538463, "retention_by_kind": {"person": 1.0, "date": 0.0, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 775}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 12, "retention_rate": 0.18461538461538463, "retention_by_kind": {"person": 1.0, "date": 0.0, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 698}, "2": {"facts_total": 65, "facts_retained": 12, "retention_rate": 0.18461538461538463, "retention_by_kind": {"person": 1.0, "date": 0.0, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 775}, "3": {"facts_total": 65, "facts_retained": 12, "retention_rate": 0.18461538461538463, "retention_by_kind": {"person": 1.0, "date": 0.0, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 775}, "4": {"facts_total": 65, "facts_retained": 12, "retention_rate": 0.18461538461538463, "retention_by_kind": {"person": 1.0, "date": 0.0, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 775}, "5": {"facts_total": 65, "facts_retained": 12, "retention_rate": 0.18461538461538463, "retention_by_kind": {"person": 1.0, "date": 0.0, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 775}, "6": {"facts_total": 65, "facts_retained": 12, "retention_rate": 0.18461538461538463, "retention_by_kind": {"person": 1.0, "date": 0.0, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 775}, "7": {"facts_total": 65, "facts_retained": 12, "retention_rate": 0.18461538461538463, "retention_by_kind": {"person": 1.0, "date": 0.0, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 775}, "8": {"facts_total": 65, "facts_retained": 12, "retention_rate": 0.18461538461538463, "retention_by_kind": {"person": 1.0, "date": 0.0, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 775}}}, "reference_key_id": "keys/synthetic/syn_03009_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "2d43b54be891e5f4db3e2062da0ff5b0470ace3acf8bc02542fac6aa4045349b", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "9ef63d92ba503ba44ef860cb", "attempt_id": "9a8281bed28d", "run_group_id": "E4_core:A5|closed|qwen3b-llama3b:h8", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:29:11.358268Z", "task_family": "synthetic", "document_id": "syn_03009_2000", "document_hash": "6ed94dcf728255bd7a7d11c97c362930728fb81a3e91f1906eb6eec86e0fedb7", "question_id": "q06_negation", "split": "test", "cluster_id": "syn_03009_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "tokenizer_hashes": {"receiver": "b9d8872cf3a99da99e1fc4b0f71232a451632dc49ca4c6771549b4efa5c1c763"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 8, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1770, "query_tokens": 361, "message_tokens": 255, "message_utf8_bytes": 775, "cache_tensor_bytes": null, "serialized_payload_bytes": 775, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 91002.5866671931, "hops": {"1": 16777.417791076005, "2": 10271.78920910228, "3": 11040.38683406543, "4": 10208.290625014342, "5": 11365.20437488798, "6": 10238.278790959157, "7": 11114.223542041145, "8": 9986.99550004676}, "receiver_prefill_ms": 321.4991659624502, "ttft_ms": 90.42570798192173, "decode_ms": 137.06591597292572}, "total_wall_ms": 91552.020875155, "ttft_ms": 90.42570798192173, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "No", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 12, "retention_rate": 0.18461538461538463, "retention_by_kind": {"person": 1.0, "date": 0.0, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 775}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 12, "retention_rate": 0.18461538461538463, "retention_by_kind": {"person": 1.0, "date": 0.0, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 698}, "2": {"facts_total": 65, "facts_retained": 12, "retention_rate": 0.18461538461538463, "retention_by_kind": {"person": 1.0, "date": 0.0, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 775}, "3": {"facts_total": 65, "facts_retained": 12, "retention_rate": 0.18461538461538463, "retention_by_kind": {"person": 1.0, "date": 0.0, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 775}, "4": {"facts_total": 65, "facts_retained": 12, "retention_rate": 0.18461538461538463, "retention_by_kind": {"person": 1.0, "date": 0.0, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 775}, "5": {"facts_total": 65, "facts_retained": 12, "retention_rate": 0.18461538461538463, "retention_by_kind": {"person": 1.0, "date": 0.0, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 775}, "6": {"facts_total": 65, "facts_retained": 12, "retention_rate": 0.18461538461538463, "retention_by_kind": {"person": 1.0, "date": 0.0, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 775}, "7": {"facts_total": 65, "facts_retained": 12, "retention_rate": 0.18461538461538463, "retention_by_kind": {"person": 1.0, "date": 0.0, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 775}, "8": {"facts_total": 65, "facts_retained": 12, "retention_rate": 0.18461538461538463, "retention_by_kind": {"person": 1.0, "date": 0.0, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 775}}}, "reference_key_id": "keys/synthetic/syn_03009_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "2d43b54be891e5f4db3e2062da0ff5b0470ace3acf8bc02542fac6aa4045349b", "output_hash": "1ea442a134b2a184bd5d40104401f2a37fbc09ccf3f4bc9da161c6099be3691d"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "0639823f383bac9b3766fd24", "attempt_id": "29ec30395ebe", "run_group_id": "E4_core:A5|closed|qwen3b-llama3b:h8", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:29:11.924983Z", "task_family": "synthetic", "document_id": "syn_03009_2000", "document_hash": "6ed94dcf728255bd7a7d11c97c362930728fb81a3e91f1906eb6eec86e0fedb7", "question_id": "q07_negation", "split": "test", "cluster_id": "syn_03009_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "tokenizer_hashes": {"receiver": "b9d8872cf3a99da99e1fc4b0f71232a451632dc49ca4c6771549b4efa5c1c763"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 8, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1770, "query_tokens": 360, "message_tokens": 255, "message_utf8_bytes": 775, "cache_tensor_bytes": null, "serialized_payload_bytes": 775, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 91002.5866671931, "hops": {"1": 16777.417791076005, "2": 10271.78920910228, "3": 11040.38683406543, "4": 10208.290625014342, "5": 11365.20437488798, "6": 10238.278790959157, "7": 11114.223542041145, "8": 9986.99550004676}, "receiver_prefill_ms": 321.00287498906255, "ttft_ms": 20.225208019837737, "decode_ms": 36.62416595034301}, "total_wall_ms": 91380.90758409817, "ttft_ms": 20.225208019837737, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "No", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 12, "retention_rate": 0.18461538461538463, "retention_by_kind": {"person": 1.0, "date": 0.0, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 775}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 12, "retention_rate": 0.18461538461538463, "retention_by_kind": {"person": 1.0, "date": 0.0, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 698}, "2": {"facts_total": 65, "facts_retained": 12, "retention_rate": 0.18461538461538463, "retention_by_kind": {"person": 1.0, "date": 0.0, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 775}, "3": {"facts_total": 65, "facts_retained": 12, "retention_rate": 0.18461538461538463, "retention_by_kind": {"person": 1.0, "date": 0.0, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 775}, "4": {"facts_total": 65, "facts_retained": 12, "retention_rate": 0.18461538461538463, "retention_by_kind": {"person": 1.0, "date": 0.0, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 775}, "5": {"facts_total": 65, "facts_retained": 12, "retention_rate": 0.18461538461538463, "retention_by_kind": {"person": 1.0, "date": 0.0, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 775}, "6": {"facts_total": 65, "facts_retained": 12, "retention_rate": 0.18461538461538463, "retention_by_kind": {"person": 1.0, "date": 0.0, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 775}, "7": {"facts_total": 65, "facts_retained": 12, "retention_rate": 0.18461538461538463, "retention_by_kind": {"person": 1.0, "date": 0.0, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 775}, "8": {"facts_total": 65, "facts_retained": 12, "retention_rate": 0.18461538461538463, "retention_by_kind": {"person": 1.0, "date": 0.0, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 775}}}, "reference_key_id": "keys/synthetic/syn_03009_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "2d43b54be891e5f4db3e2062da0ff5b0470ace3acf8bc02542fac6aa4045349b", "output_hash": "1ea442a134b2a184bd5d40104401f2a37fbc09ccf3f4bc9da161c6099be3691d"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "148fdfa431360e63cd670e07", "attempt_id": "ff88fc79a0d0", "run_group_id": "E4_core:A5|closed|qwen3b-llama3b:h8", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:29:12.320444Z", "task_family": "synthetic", "document_id": "syn_03009_2000", "document_hash": "6ed94dcf728255bd7a7d11c97c362930728fb81a3e91f1906eb6eec86e0fedb7", "question_id": "q08_revision", "split": "test", "cluster_id": "syn_03009_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "tokenizer_hashes": {"receiver": "b9d8872cf3a99da99e1fc4b0f71232a451632dc49ca4c6771549b4efa5c1c763"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 8, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1770, "query_tokens": 377, "message_tokens": 255, "message_utf8_bytes": 775, "cache_tensor_bytes": null, "serialized_payload_bytes": 775, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 91002.5866671931, "hops": {"1": 16777.417791076005, "2": 10271.78920910228, "3": 11040.38683406543, "4": 10208.290625014342, "5": 11365.20437488798, "6": 10238.278790959157, "7": 11114.223542041145, "8": 9986.99550004676}, "receiver_prefill_ms": 339.99004191718996, "ttft_ms": 23.71566602960229, "decode_ms": 31.460583908483386}, "total_wall_ms": 91398.20883318316, "ttft_ms": 23.71566602960229, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 12, "retention_rate": 0.18461538461538463, "retention_by_kind": {"person": 1.0, "date": 0.0, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 775}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 12, "retention_rate": 0.18461538461538463, "retention_by_kind": {"person": 1.0, "date": 0.0, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 698}, "2": {"facts_total": 65, "facts_retained": 12, "retention_rate": 0.18461538461538463, "retention_by_kind": {"person": 1.0, "date": 0.0, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 775}, "3": {"facts_total": 65, "facts_retained": 12, "retention_rate": 0.18461538461538463, "retention_by_kind": {"person": 1.0, "date": 0.0, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 775}, "4": {"facts_total": 65, "facts_retained": 12, "retention_rate": 0.18461538461538463, "retention_by_kind": {"person": 1.0, "date": 0.0, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 775}, "5": {"facts_total": 65, "facts_retained": 12, "retention_rate": 0.18461538461538463, "retention_by_kind": {"person": 1.0, "date": 0.0, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 775}, "6": {"facts_total": 65, "facts_retained": 12, "retention_rate": 0.18461538461538463, "retention_by_kind": {"person": 1.0, "date": 0.0, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 775}, "7": {"facts_total": 65, "facts_retained": 12, "retention_rate": 0.18461538461538463, "retention_by_kind": {"person": 1.0, "date": 0.0, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 775}, "8": {"facts_total": 65, "facts_retained": 12, "retention_rate": 0.18461538461538463, "retention_by_kind": {"person": 1.0, "date": 0.0, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 775}}}, "reference_key_id": "keys/synthetic/syn_03009_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "2d43b54be891e5f4db3e2062da0ff5b0470ace3acf8bc02542fac6aa4045349b", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "528ebc1d3331cb571d42f798", "attempt_id": "5afabfef8a30", "run_group_id": "E4_core:A5|closed|qwen3b-llama3b:h8", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:29:12.733856Z", "task_family": "synthetic", "document_id": "syn_03009_2000", "document_hash": "6ed94dcf728255bd7a7d11c97c362930728fb81a3e91f1906eb6eec86e0fedb7", "question_id": "q09_revision", "split": "test", "cluster_id": "syn_03009_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "tokenizer_hashes": {"receiver": "b9d8872cf3a99da99e1fc4b0f71232a451632dc49ca4c6771549b4efa5c1c763"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 8, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1770, "query_tokens": 370, "message_tokens": 255, "message_utf8_bytes": 775, "cache_tensor_bytes": null, "serialized_payload_bytes": 775, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 91002.5866671931, "hops": {"1": 16777.417791076005, "2": 10271.78920910228, "3": 11040.38683406543, "4": 10208.290625014342, "5": 11365.20437488798, "6": 10238.278790959157, "7": 11114.223542041145, "8": 9986.99550004676}, "receiver_prefill_ms": 278.8091669790447, "ttft_ms": 39.697290980257094, "decode_ms": 59.61295892484486}, "total_wall_ms": 91381.24904211145, "ttft_ms": 39.697290980257094, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 12, "retention_rate": 0.18461538461538463, "retention_by_kind": {"person": 1.0, "date": 0.0, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 775}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 12, "retention_rate": 0.18461538461538463, "retention_by_kind": {"person": 1.0, "date": 0.0, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 698}, "2": {"facts_total": 65, "facts_retained": 12, "retention_rate": 0.18461538461538463, "retention_by_kind": {"person": 1.0, "date": 0.0, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 775}, "3": {"facts_total": 65, "facts_retained": 12, "retention_rate": 0.18461538461538463, "retention_by_kind": {"person": 1.0, "date": 0.0, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 775}, "4": {"facts_total": 65, "facts_retained": 12, "retention_rate": 0.18461538461538463, "retention_by_kind": {"person": 1.0, "date": 0.0, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 775}, "5": {"facts_total": 65, "facts_retained": 12, "retention_rate": 0.18461538461538463, "retention_by_kind": {"person": 1.0, "date": 0.0, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 775}, "6": {"facts_total": 65, "facts_retained": 12, "retention_rate": 0.18461538461538463, "retention_by_kind": {"person": 1.0, "date": 0.0, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 775}, "7": {"facts_total": 65, "facts_retained": 12, "retention_rate": 0.18461538461538463, "retention_by_kind": {"person": 1.0, "date": 0.0, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 775}, "8": {"facts_total": 65, "facts_retained": 12, "retention_rate": 0.18461538461538463, "retention_by_kind": {"person": 1.0, "date": 0.0, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 775}}}, "reference_key_id": "keys/synthetic/syn_03009_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "2d43b54be891e5f4db3e2062da0ff5b0470ace3acf8bc02542fac6aa4045349b", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "4ed115af6009b5921b5b5d82", "attempt_id": "19f324a57d77", "run_group_id": "E4_core:A5|closed|qwen3b-llama3b:h8", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:29:13.130431Z", "task_family": "synthetic", "document_id": "syn_03009_2000", "document_hash": "6ed94dcf728255bd7a7d11c97c362930728fb81a3e91f1906eb6eec86e0fedb7", "question_id": "q10_temporal", "split": "test", "cluster_id": "syn_03009_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "tokenizer_hashes": {"receiver": "b9d8872cf3a99da99e1fc4b0f71232a451632dc49ca4c6771549b4efa5c1c763"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 8, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1770, "query_tokens": 378, "message_tokens": 255, "message_utf8_bytes": 775, "cache_tensor_bytes": null, "serialized_payload_bytes": 775, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 91002.5866671931, "hops": {"1": 16777.417791076005, "2": 10271.78920910228, "3": 11040.38683406543, "4": 10208.290625014342, "5": 11365.20437488798, "6": 10238.278790959157, "7": 11114.223542041145, "8": 9986.99550004676}, "receiver_prefill_ms": 427.89679195266217, "ttft_ms": 37.81041700858623, "decode_ms": 181.80116696748883}, "total_wall_ms": 91650.62345925253, "ttft_ms": 37.81041700858623, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 6}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "Eshgamgam-11", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 12, "retention_rate": 0.18461538461538463, "retention_by_kind": {"person": 1.0, "date": 0.0, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 775}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 12, "retention_rate": 0.18461538461538463, "retention_by_kind": {"person": 1.0, "date": 0.0, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 698}, "2": {"facts_total": 65, "facts_retained": 12, "retention_rate": 0.18461538461538463, "retention_by_kind": {"person": 1.0, "date": 0.0, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 775}, "3": {"facts_total": 65, "facts_retained": 12, "retention_rate": 0.18461538461538463, "retention_by_kind": {"person": 1.0, "date": 0.0, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 775}, "4": {"facts_total": 65, "facts_retained": 12, "retention_rate": 0.18461538461538463, "retention_by_kind": {"person": 1.0, "date": 0.0, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 775}, "5": {"facts_total": 65, "facts_retained": 12, "retention_rate": 0.18461538461538463, "retention_by_kind": {"person": 1.0, "date": 0.0, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 775}, "6": {"facts_total": 65, "facts_retained": 12, "retention_rate": 0.18461538461538463, "retention_by_kind": {"person": 1.0, "date": 0.0, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 775}, "7": {"facts_total": 65, "facts_retained": 12, "retention_rate": 0.18461538461538463, "retention_by_kind": {"person": 1.0, "date": 0.0, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 775}, "8": {"facts_total": 65, "facts_retained": 12, "retention_rate": 0.18461538461538463, "retention_by_kind": {"person": 1.0, "date": 0.0, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 775}}}, "reference_key_id": "keys/synthetic/syn_03009_2000.json", "task_scores": {"correct": 0.0, "abstained": 0.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "2d43b54be891e5f4db3e2062da0ff5b0470ace3acf8bc02542fac6aa4045349b", "output_hash": "8c1313c53ab4dbffa5f336d721c810678133eecaa467241e53d8a92150601475"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "2fa8f416c9faeec9dc75ad4d", "attempt_id": "582479b4531b", "run_group_id": "E4_core:A5|closed|qwen3b-llama3b:h8", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:29:13.796572Z", "task_family": "synthetic", "document_id": "syn_03009_2000", "document_hash": "6ed94dcf728255bd7a7d11c97c362930728fb81a3e91f1906eb6eec86e0fedb7", "question_id": "q11_aggregation", "split": "test", "cluster_id": "syn_03009_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "tokenizer_hashes": {"receiver": "b9d8872cf3a99da99e1fc4b0f71232a451632dc49ca4c6771549b4efa5c1c763"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 8, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1770, "query_tokens": 371, "message_tokens": 255, "message_utf8_bytes": 775, "cache_tensor_bytes": null, "serialized_payload_bytes": 775, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 91002.5866671931, "hops": {"1": 16777.417791076005, "2": 10271.78920910228, "3": 11040.38683406543, "4": 10208.290625014342, "5": 11365.20437488798, "6": 10238.278790959157, "7": 11114.223542041145, "8": 9986.99550004676}, "receiver_prefill_ms": 339.49858299456537, "ttft_ms": 24.255082942545414, "decode_ms": 31.35166596621275}, "total_wall_ms": 91398.2192921685, "ttft_ms": 24.255082942545414, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 12, "retention_rate": 0.18461538461538463, "retention_by_kind": {"person": 1.0, "date": 0.0, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 775}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 12, "retention_rate": 0.18461538461538463, "retention_by_kind": {"person": 1.0, "date": 0.0, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 698}, "2": {"facts_total": 65, "facts_retained": 12, "retention_rate": 0.18461538461538463, "retention_by_kind": {"person": 1.0, "date": 0.0, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 775}, "3": {"facts_total": 65, "facts_retained": 12, "retention_rate": 0.18461538461538463, "retention_by_kind": {"person": 1.0, "date": 0.0, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 775}, "4": {"facts_total": 65, "facts_retained": 12, "retention_rate": 0.18461538461538463, "retention_by_kind": {"person": 1.0, "date": 0.0, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 775}, "5": {"facts_total": 65, "facts_retained": 12, "retention_rate": 0.18461538461538463, "retention_by_kind": {"person": 1.0, "date": 0.0, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 775}, "6": {"facts_total": 65, "facts_retained": 12, "retention_rate": 0.18461538461538463, "retention_by_kind": {"person": 1.0, "date": 0.0, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 775}, "7": {"facts_total": 65, "facts_retained": 12, "retention_rate": 0.18461538461538463, "retention_by_kind": {"person": 1.0, "date": 0.0, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 775}, "8": {"facts_total": 65, "facts_retained": 12, "retention_rate": 0.18461538461538463, "retention_by_kind": {"person": 1.0, "date": 0.0, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 775}}}, "reference_key_id": "keys/synthetic/syn_03009_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "2d43b54be891e5f4db3e2062da0ff5b0470ace3acf8bc02542fac6aa4045349b", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "ee461a61d5d5be0d1cb83b2f", "attempt_id": "13a6240a6b56", "run_group_id": "E4_core:A5|closed|qwen3b-llama3b:h8", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:29:14.210165Z", "task_family": "synthetic", "document_id": "syn_03009_2000", "document_hash": "6ed94dcf728255bd7a7d11c97c362930728fb81a3e91f1906eb6eec86e0fedb7", "question_id": "q12_unit_conversion", "split": "test", "cluster_id": "syn_03009_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "tokenizer_hashes": {"receiver": "b9d8872cf3a99da99e1fc4b0f71232a451632dc49ca4c6771549b4efa5c1c763"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 8, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1770, "query_tokens": 368, "message_tokens": 255, "message_utf8_bytes": 775, "cache_tensor_bytes": null, "serialized_payload_bytes": 775, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 91002.5866671931, "hops": {"1": 16777.417791076005, "2": 10271.78920910228, "3": 11040.38683406543, "4": 10208.290625014342, "5": 11365.20437488798, "6": 10238.278790959157, "7": 11114.223542041145, "8": 9986.99550004676}, "receiver_prefill_ms": 294.13091694004834, "ttft_ms": 46.23850004281849, "decode_ms": 40.429792017675936}, "total_wall_ms": 91384.00325016119, "ttft_ms": 46.23850004281849, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 12, "retention_rate": 0.18461538461538463, "retention_by_kind": {"person": 1.0, "date": 0.0, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 775}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 12, "retention_rate": 0.18461538461538463, "retention_by_kind": {"person": 1.0, "date": 0.0, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 698}, "2": {"facts_total": 65, "facts_retained": 12, "retention_rate": 0.18461538461538463, "retention_by_kind": {"person": 1.0, "date": 0.0, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 775}, "3": {"facts_total": 65, "facts_retained": 12, "retention_rate": 0.18461538461538463, "retention_by_kind": {"person": 1.0, "date": 0.0, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 775}, "4": {"facts_total": 65, "facts_retained": 12, "retention_rate": 0.18461538461538463, "retention_by_kind": {"person": 1.0, "date": 0.0, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 775}, "5": {"facts_total": 65, "facts_retained": 12, "retention_rate": 0.18461538461538463, "retention_by_kind": {"person": 1.0, "date": 0.0, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 775}, "6": {"facts_total": 65, "facts_retained": 12, "retention_rate": 0.18461538461538463, "retention_by_kind": {"person": 1.0, "date": 0.0, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 775}, "7": {"facts_total": 65, "facts_retained": 12, "retention_rate": 0.18461538461538463, "retention_by_kind": {"person": 1.0, "date": 0.0, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 775}, "8": {"facts_total": 65, "facts_retained": 12, "retention_rate": 0.18461538461538463, "retention_by_kind": {"person": 1.0, "date": 0.0, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 775}}}, "reference_key_id": "keys/synthetic/syn_03009_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "2d43b54be891e5f4db3e2062da0ff5b0470ace3acf8bc02542fac6aa4045349b", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "1acfe6e9c9f79776fa3052ff", "attempt_id": "0a473b88619a", "run_group_id": "E4_core:A5|closed|qwen3b-llama3b:h8", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:29:14.609497Z", "task_family": "synthetic", "document_id": "syn_03009_2000", "document_hash": "6ed94dcf728255bd7a7d11c97c362930728fb81a3e91f1906eb6eec86e0fedb7", "question_id": "q13_unanswerable", "split": "test", "cluster_id": "syn_03009_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "tokenizer_hashes": {"receiver": "b9d8872cf3a99da99e1fc4b0f71232a451632dc49ca4c6771549b4efa5c1c763"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 8, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1770, "query_tokens": 372, "message_tokens": 255, "message_utf8_bytes": 775, "cache_tensor_bytes": null, "serialized_payload_bytes": 775, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 91002.5866671931, "hops": {"1": 16777.417791076005, "2": 10271.78920910228, "3": 11040.38683406543, "4": 10208.290625014342, "5": 11365.20437488798, "6": 10238.278790959157, "7": 11114.223542041145, "8": 9986.99550004676}, "receiver_prefill_ms": 465.53158399183303, "ttft_ms": 43.22795802727342, "decode_ms": 23.90820800792426}, "total_wall_ms": 91535.77883320395, "ttft_ms": 43.22795802727342, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 12, "retention_rate": 0.18461538461538463, "retention_by_kind": {"person": 1.0, "date": 0.0, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 775}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 12, "retention_rate": 0.18461538461538463, "retention_by_kind": {"person": 1.0, "date": 0.0, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 698}, "2": {"facts_total": 65, "facts_retained": 12, "retention_rate": 0.18461538461538463, "retention_by_kind": {"person": 1.0, "date": 0.0, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 775}, "3": {"facts_total": 65, "facts_retained": 12, "retention_rate": 0.18461538461538463, "retention_by_kind": {"person": 1.0, "date": 0.0, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 775}, "4": {"facts_total": 65, "facts_retained": 12, "retention_rate": 0.18461538461538463, "retention_by_kind": {"person": 1.0, "date": 0.0, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 775}, "5": {"facts_total": 65, "facts_retained": 12, "retention_rate": 0.18461538461538463, "retention_by_kind": {"person": 1.0, "date": 0.0, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 775}, "6": {"facts_total": 65, "facts_retained": 12, "retention_rate": 0.18461538461538463, "retention_by_kind": {"person": 1.0, "date": 0.0, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 775}, "7": {"facts_total": 65, "facts_retained": 12, "retention_rate": 0.18461538461538463, "retention_by_kind": {"person": 1.0, "date": 0.0, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 775}, "8": {"facts_total": 65, "facts_retained": 12, "retention_rate": 0.18461538461538463, "retention_by_kind": {"person": 1.0, "date": 0.0, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 775}}}, "reference_key_id": "keys/synthetic/syn_03009_2000.json", "task_scores": {"correct": 1.0, "abstained": 1.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "2d43b54be891e5f4db3e2062da0ff5b0470ace3acf8bc02542fac6aa4045349b", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "a4d5fbb7cbe3596261eac92b", "attempt_id": "bebf6539f5cd", "run_group_id": "E4_core:A5|closed|qwen3b-llama3b:h8", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:29:15.160874Z", "task_family": "synthetic", "document_id": "syn_03009_2000", "document_hash": "6ed94dcf728255bd7a7d11c97c362930728fb81a3e91f1906eb6eec86e0fedb7", "question_id": "q14_unanswerable", "split": "test", "cluster_id": "syn_03009_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "tokenizer_hashes": {"receiver": "b9d8872cf3a99da99e1fc4b0f71232a451632dc49ca4c6771549b4efa5c1c763"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 8, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1770, "query_tokens": 373, "message_tokens": 255, "message_utf8_bytes": 775, "cache_tensor_bytes": null, "serialized_payload_bytes": 775, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 91002.5866671931, "hops": {"1": 16777.417791076005, "2": 10271.78920910228, "3": 11040.38683406543, "4": 10208.290625014342, "5": 11365.20437488798, "6": 10238.278790959157, "7": 11114.223542041145, "8": 9986.99550004676}, "receiver_prefill_ms": 337.1102079981938, "ttft_ms": 42.86283301189542, "decode_ms": 24.631667067296803}, "total_wall_ms": 91407.69225021359, "ttft_ms": 42.86283301189542, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 12, "retention_rate": 0.18461538461538463, "retention_by_kind": {"person": 1.0, "date": 0.0, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 775}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 12, "retention_rate": 0.18461538461538463, "retention_by_kind": {"person": 1.0, "date": 0.0, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 698}, "2": {"facts_total": 65, "facts_retained": 12, "retention_rate": 0.18461538461538463, "retention_by_kind": {"person": 1.0, "date": 0.0, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 775}, "3": {"facts_total": 65, "facts_retained": 12, "retention_rate": 0.18461538461538463, "retention_by_kind": {"person": 1.0, "date": 0.0, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 775}, "4": {"facts_total": 65, "facts_retained": 12, "retention_rate": 0.18461538461538463, "retention_by_kind": {"person": 1.0, "date": 0.0, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 775}, "5": {"facts_total": 65, "facts_retained": 12, "retention_rate": 0.18461538461538463, "retention_by_kind": {"person": 1.0, "date": 0.0, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 775}, "6": {"facts_total": 65, "facts_retained": 12, "retention_rate": 0.18461538461538463, "retention_by_kind": {"person": 1.0, "date": 0.0, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 775}, "7": {"facts_total": 65, "facts_retained": 12, "retention_rate": 0.18461538461538463, "retention_by_kind": {"person": 1.0, "date": 0.0, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 775}, "8": {"facts_total": 65, "facts_retained": 12, "retention_rate": 0.18461538461538463, "retention_by_kind": {"person": 1.0, "date": 0.0, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 775}}}, "reference_key_id": "keys/synthetic/syn_03009_2000.json", "task_scores": {"correct": 1.0, "abstained": 1.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "2d43b54be891e5f4db3e2062da0ff5b0470ace3acf8bc02542fac6aa4045349b", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "9891b8cb4235cd53d7bd01bf", "attempt_id": "8d1442efe1c0", "run_group_id": "E4_core:A4|closed|qwen3b:h1", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:29:38.168071Z", "task_family": "synthetic", "document_id": "syn_03012_2000", "document_hash": "03e441e68d54e1af036f8ff905dfd6cbbccead62636279c1a8e179352ab08945", "question_id": "q00_single_fact", "split": "test", "cluster_id": "syn_03012_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 1, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1927, "query_tokens": 594, "message_tokens": 512, "message_utf8_bytes": 1213, "cache_tensor_bytes": null, "serialized_payload_bytes": 1213, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 22483.77687507309, "hops": {"1": 22483.77687507309}, "receiver_prefill_ms": 393.794541945681, "ttft_ms": 72.97629199456424, "decode_ms": 411.6689999355003}, "total_wall_ms": 23362.875625141896, "ttft_ms": 72.97629199456424, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 10}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "XX-71M-6213", "evidence_artifact": {"ledger": {"facts_total": 59, "facts_retained": 30, "retention_rate": 0.5084745762711864, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.5714285714285714, "amount": 0.6, "status": 0.5833333333333334, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1213}, "ledger_by_hop": {"1": {"facts_total": 59, "facts_retained": 30, "retention_rate": 0.5084745762711864, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.5714285714285714, "amount": 0.6, "status": 0.5833333333333334, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1213}}}, "reference_key_id": "keys/synthetic/syn_03012_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "7c9b236b590faff2030e02587d972cf755ac2a88072e709174edbf405498973d", "output_hash": "66ef6c5da682b38795e9029723bbd71cdde2c4dec70cf1fae6efe62ef243f10a"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "ed2dd49a182698b0b1fc3a09", "attempt_id": "ce3d24e52277", "run_group_id": "E4_core:A4|closed|qwen3b:h1", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:29:39.065075Z", "task_family": "synthetic", "document_id": "syn_03012_2000", "document_hash": "03e441e68d54e1af036f8ff905dfd6cbbccead62636279c1a8e179352ab08945", "question_id": "q01_single_fact", "split": "test", "cluster_id": "syn_03012_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 1, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1927, "query_tokens": 600, "message_tokens": 512, "message_utf8_bytes": 1213, "cache_tensor_bytes": null, "serialized_payload_bytes": 1213, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 22483.77687507309, "hops": {"1": 22483.77687507309}, "receiver_prefill_ms": 402.5754160247743, "ttft_ms": 34.635000047273934, "decode_ms": 29.258749913424253}, "total_wall_ms": 22950.832583126612, "ttft_ms": 34.635000047273934, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 59, "facts_retained": 30, "retention_rate": 0.5084745762711864, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.5714285714285714, "amount": 0.6, "status": 0.5833333333333334, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1213}, "ledger_by_hop": {"1": {"facts_total": 59, "facts_retained": 30, "retention_rate": 0.5084745762711864, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.5714285714285714, "amount": 0.6, "status": 0.5833333333333334, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1213}}}, "reference_key_id": "keys/synthetic/syn_03012_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "7c9b236b590faff2030e02587d972cf755ac2a88072e709174edbf405498973d", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "46ce810c2bbc2c7383f87bd8", "attempt_id": "56f46836bd07", "run_group_id": "E4_core:A4|closed|qwen3b:h1", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:29:39.549533Z", "task_family": "synthetic", "document_id": "syn_03012_2000", "document_hash": "03e441e68d54e1af036f8ff905dfd6cbbccead62636279c1a8e179352ab08945", "question_id": "q02_exact_copy", "split": "test", "cluster_id": "syn_03012_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 1, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1927, "query_tokens": 599, "message_tokens": 512, "message_utf8_bytes": 1213, "cache_tensor_bytes": null, "serialized_payload_bytes": 1213, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 22483.77687507309, "hops": {"1": 22483.77687507309}, "receiver_prefill_ms": 356.85016703791916, "ttft_ms": 52.177665987983346, "decode_ms": 470.5755830509588}, "total_wall_ms": 23364.149625063874, "ttft_ms": 52.177665987983346, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 7}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "13.4343", "evidence_artifact": {"ledger": {"facts_total": 59, "facts_retained": 30, "retention_rate": 0.5084745762711864, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.5714285714285714, "amount": 0.6, "status": 0.5833333333333334, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1213}, "ledger_by_hop": {"1": {"facts_total": 59, "facts_retained": 30, "retention_rate": 0.5084745762711864, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.5714285714285714, "amount": 0.6, "status": 0.5833333333333334, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1213}}}, "reference_key_id": "keys/synthetic/syn_03012_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "7c9b236b590faff2030e02587d972cf755ac2a88072e709174edbf405498973d", "output_hash": "599bc22e373b1cf7b4679208eb9e515ad0e8dcb88dde0c7f579637ec7d3f3014"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "3d25f5ca787b6719a546f378", "attempt_id": "2661a401fa51", "run_group_id": "E4_core:A4|closed|qwen3b:h1", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:29:40.447323Z", "task_family": "synthetic", "document_id": "syn_03012_2000", "document_hash": "03e441e68d54e1af036f8ff905dfd6cbbccead62636279c1a8e179352ab08945", "question_id": "q03_exact_copy", "split": "test", "cluster_id": "syn_03012_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 1, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1927, "query_tokens": 595, "message_tokens": 512, "message_utf8_bytes": 1213, "cache_tensor_bytes": null, "serialized_payload_bytes": 1213, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 22483.77687507309, "hops": {"1": 22483.77687507309}, "receiver_prefill_ms": 416.6893339715898, "ttft_ms": 20.336958928965032, "decode_ms": 39.56691699568182}, "total_wall_ms": 22961.15083305631, "ttft_ms": 20.336958928965032, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 59, "facts_retained": 30, "retention_rate": 0.5084745762711864, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.5714285714285714, "amount": 0.6, "status": 0.5833333333333334, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1213}, "ledger_by_hop": {"1": {"facts_total": 59, "facts_retained": 30, "retention_rate": 0.5084745762711864, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.5714285714285714, "amount": 0.6, "status": 0.5833333333333334, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1213}}}, "reference_key_id": "keys/synthetic/syn_03012_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "7c9b236b590faff2030e02587d972cf755ac2a88072e709174edbf405498973d", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "913f809427e396dadd24c3ae", "attempt_id": "86ec01f60ada", "run_group_id": "E4_core:A4|closed|qwen3b:h1", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:29:40.943034Z", "task_family": "synthetic", "document_id": "syn_03012_2000", "document_hash": "03e441e68d54e1af036f8ff905dfd6cbbccead62636279c1a8e179352ab08945", "question_id": "q04_multi_hop", "split": "test", "cluster_id": "syn_03012_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 1, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1927, "query_tokens": 600, "message_tokens": 512, "message_utf8_bytes": 1213, "cache_tensor_bytes": null, "serialized_payload_bytes": 1213, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 22483.77687507309, "hops": {"1": 22483.77687507309}, "receiver_prefill_ms": 387.8147089853883, "ttft_ms": 26.119375019334257, "decode_ms": 422.19945893157274}, "total_wall_ms": 23320.671458146535, "ttft_ms": 26.119375019334257, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 7}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "8631552", "evidence_artifact": {"ledger": {"facts_total": 59, "facts_retained": 30, "retention_rate": 0.5084745762711864, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.5714285714285714, "amount": 0.6, "status": 0.5833333333333334, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1213}, "ledger_by_hop": {"1": {"facts_total": 59, "facts_retained": 30, "retention_rate": 0.5084745762711864, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.5714285714285714, "amount": 0.6, "status": 0.5833333333333334, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1213}}}, "reference_key_id": "keys/synthetic/syn_03012_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "7c9b236b590faff2030e02587d972cf755ac2a88072e709174edbf405498973d", "output_hash": "147d02c0587a77b3b53f0424c8d3d1454aaf2639ce9fdaf69af20db33363e849"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "a99b2229f90de50d1af78bca", "attempt_id": "99177981a504", "run_group_id": "E4_core:A4|closed|qwen3b:h1", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:29:41.796618Z", "task_family": "synthetic", "document_id": "syn_03012_2000", "document_hash": "03e441e68d54e1af036f8ff905dfd6cbbccead62636279c1a8e179352ab08945", "question_id": "q05_multi_hop", "split": "test", "cluster_id": "syn_03012_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 1, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1927, "query_tokens": 603, "message_tokens": 512, "message_utf8_bytes": 1213, "cache_tensor_bytes": null, "serialized_payload_bytes": 1213, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 22483.77687507309, "hops": {"1": 22483.77687507309}, "receiver_prefill_ms": 475.71012505795807, "ttft_ms": 33.0112500814721, "decode_ms": 335.7997920829803}, "total_wall_ms": 23328.909292118624, "ttft_ms": 33.0112500814721, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 10}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "DG-40E-2777", "evidence_artifact": {"ledger": {"facts_total": 59, "facts_retained": 30, "retention_rate": 0.5084745762711864, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.5714285714285714, "amount": 0.6, "status": 0.5833333333333334, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1213}, "ledger_by_hop": {"1": {"facts_total": 59, "facts_retained": 30, "retention_rate": 0.5084745762711864, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.5714285714285714, "amount": 0.6, "status": 0.5833333333333334, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1213}}}, "reference_key_id": "keys/synthetic/syn_03012_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "7c9b236b590faff2030e02587d972cf755ac2a88072e709174edbf405498973d", "output_hash": "16a7e3f4a7cc4a9b66f86c2cdcfad3b35daf1da199d9406e0627c8af2d8d73f3"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "81c13a4d0c28e30a57b9c455", "attempt_id": "daf8390c8289", "run_group_id": "E4_core:A4|closed|qwen3b:h1", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:29:42.661154Z", "task_family": "synthetic", "document_id": "syn_03012_2000", "document_hash": "03e441e68d54e1af036f8ff905dfd6cbbccead62636279c1a8e179352ab08945", "question_id": "q06_negation", "split": "test", "cluster_id": "syn_03012_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 1, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1927, "query_tokens": 595, "message_tokens": 512, "message_utf8_bytes": 1213, "cache_tensor_bytes": null, "serialized_payload_bytes": 1213, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 22483.77687507309, "hops": {"1": 22483.77687507309}, "receiver_prefill_ms": 350.1358750509098, "ttft_ms": 39.48266594670713, "decode_ms": 56.0263340594247}, "total_wall_ms": 22930.326291010715, "ttft_ms": 39.48266594670713, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "no", "evidence_artifact": {"ledger": {"facts_total": 59, "facts_retained": 30, "retention_rate": 0.5084745762711864, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.5714285714285714, "amount": 0.6, "status": 0.5833333333333334, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1213}, "ledger_by_hop": {"1": {"facts_total": 59, "facts_retained": 30, "retention_rate": 0.5084745762711864, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.5714285714285714, "amount": 0.6, "status": 0.5833333333333334, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1213}}}, "reference_key_id": "keys/synthetic/syn_03012_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "7c9b236b590faff2030e02587d972cf755ac2a88072e709174edbf405498973d", "output_hash": "9390298f3fb0c5b160498935d79cb139aef28e1c47358b4bbba61862b9c26e59"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "93f27bbf52afdac8fd439664", "attempt_id": "54cff9361a53", "run_group_id": "E4_core:A4|closed|qwen3b:h1", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:29:43.125277Z", "task_family": "synthetic", "document_id": "syn_03012_2000", "document_hash": "03e441e68d54e1af036f8ff905dfd6cbbccead62636279c1a8e179352ab08945", "question_id": "q07_negation", "split": "test", "cluster_id": "syn_03012_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 1, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1927, "query_tokens": 595, "message_tokens": 512, "message_utf8_bytes": 1213, "cache_tensor_bytes": null, "serialized_payload_bytes": 1213, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 22483.77687507309, "hops": {"1": 22483.77687507309}, "receiver_prefill_ms": 408.652542042546, "ttft_ms": 122.84929200541228, "decode_ms": 38.70491706766188}, "total_wall_ms": 23054.77641604375, "ttft_ms": 122.84929200541228, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "no", "evidence_artifact": {"ledger": {"facts_total": 59, "facts_retained": 30, "retention_rate": 0.5084745762711864, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.5714285714285714, "amount": 0.6, "status": 0.5833333333333334, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1213}, "ledger_by_hop": {"1": {"facts_total": 59, "facts_retained": 30, "retention_rate": 0.5084745762711864, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.5714285714285714, "amount": 0.6, "status": 0.5833333333333334, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1213}}}, "reference_key_id": "keys/synthetic/syn_03012_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "7c9b236b590faff2030e02587d972cf755ac2a88072e709174edbf405498973d", "output_hash": "9390298f3fb0c5b160498935d79cb139aef28e1c47358b4bbba61862b9c26e59"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "b597e49732db76ed13938947", "attempt_id": "5a718af6ae4b", "run_group_id": "E4_core:A4|closed|qwen3b:h1", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:29:43.713208Z", "task_family": "synthetic", "document_id": "syn_03012_2000", "document_hash": "03e441e68d54e1af036f8ff905dfd6cbbccead62636279c1a8e179352ab08945", "question_id": "q08_revision", "split": "test", "cluster_id": "syn_03012_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 1, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1927, "query_tokens": 610, "message_tokens": 512, "message_utf8_bytes": 1213, "cache_tensor_bytes": null, "serialized_payload_bytes": 1213, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 22483.77687507309, "hops": {"1": 22483.77687507309}, "receiver_prefill_ms": 418.7583749881014, "ttft_ms": 39.9043750949204, "decode_ms": 213.60554196871817}, "total_wall_ms": 23156.82345815003, "ttft_ms": 39.9043750949204, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 7}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "6602245", "evidence_artifact": {"ledger": {"facts_total": 59, "facts_retained": 30, "retention_rate": 0.5084745762711864, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.5714285714285714, "amount": 0.6, "status": 0.5833333333333334, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1213}, "ledger_by_hop": {"1": {"facts_total": 59, "facts_retained": 30, "retention_rate": 0.5084745762711864, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.5714285714285714, "amount": 0.6, "status": 0.5833333333333334, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1213}}}, "reference_key_id": "keys/synthetic/syn_03012_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "7c9b236b590faff2030e02587d972cf755ac2a88072e709174edbf405498973d", "output_hash": "d117fa8197988f8d1e4fabc6c4cdaab9516018e54553b57c63b875b3de82b16f"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "b11db635d98da23a922f1a26", "attempt_id": "21b3eb0073e1", "run_group_id": "E4_core:A4|closed|qwen3b:h1", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:29:44.403201Z", "task_family": "synthetic", "document_id": "syn_03012_2000", "document_hash": "03e441e68d54e1af036f8ff905dfd6cbbccead62636279c1a8e179352ab08945", "question_id": "q09_revision", "split": "test", "cluster_id": "syn_03012_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 1, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1927, "query_tokens": 603, "message_tokens": 512, "message_utf8_bytes": 1213, "cache_tensor_bytes": null, "serialized_payload_bytes": 1213, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 22483.77687507309, "hops": {"1": 22483.77687507309}, "receiver_prefill_ms": 359.5907080452889, "ttft_ms": 47.433374915272, "decode_ms": 482.61679091956466}, "total_wall_ms": 23374.207374989055, "ttft_ms": 47.433374915272, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 7}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "5899340", "evidence_artifact": {"ledger": {"facts_total": 59, "facts_retained": 30, "retention_rate": 0.5084745762711864, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.5714285714285714, "amount": 0.6, "status": 0.5833333333333334, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1213}, "ledger_by_hop": {"1": {"facts_total": 59, "facts_retained": 30, "retention_rate": 0.5084745762711864, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.5714285714285714, "amount": 0.6, "status": 0.5833333333333334, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1213}}}, "reference_key_id": "keys/synthetic/syn_03012_2000.json", "task_scores": {"correct": 0.0, "abstained": 0.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "7c9b236b590faff2030e02587d972cf755ac2a88072e709174edbf405498973d", "output_hash": "c920323a9a02b1afb8619cf11bdde427f7cd1ea3289be69bd62932985895d717"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "a69ed4570ae526015f8b71bd", "attempt_id": "f8afaf0a37a2", "run_group_id": "E4_core:A4|closed|qwen3b:h1", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:29:45.311151Z", "task_family": "synthetic", "document_id": "syn_03012_2000", "document_hash": "03e441e68d54e1af036f8ff905dfd6cbbccead62636279c1a8e179352ab08945", "question_id": "q10_temporal", "split": "test", "cluster_id": "syn_03012_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 1, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1927, "query_tokens": 613, "message_tokens": 512, "message_utf8_bytes": 1213, "cache_tensor_bytes": null, "serialized_payload_bytes": 1213, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 22483.77687507309, "hops": {"1": 22483.77687507309}, "receiver_prefill_ms": 404.203417012468, "ttft_ms": 26.163666043430567, "decode_ms": 216.14249993581325}, "total_wall_ms": 23130.972790997475, "ttft_ms": 26.163666043430567, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 6}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "Felaldfel-9", "evidence_artifact": {"ledger": {"facts_total": 59, "facts_retained": 30, "retention_rate": 0.5084745762711864, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.5714285714285714, "amount": 0.6, "status": 0.5833333333333334, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1213}, "ledger_by_hop": {"1": {"facts_total": 59, "facts_retained": 30, "retention_rate": 0.5084745762711864, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.5714285714285714, "amount": 0.6, "status": 0.5833333333333334, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1213}}}, "reference_key_id": "keys/synthetic/syn_03012_2000.json", "task_scores": {"correct": 0.0, "abstained": 0.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "7c9b236b590faff2030e02587d972cf755ac2a88072e709174edbf405498973d", "output_hash": "09ab1578c99909d3c22a0c8042977ad718c576e7b4cce4e50abc683b5f7d9a64"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "44c6c0efafc9edf6f4929e71", "attempt_id": "5e921c80c804", "run_group_id": "E4_core:A4|closed|qwen3b:h1", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:29:45.975220Z", "task_family": "synthetic", "document_id": "syn_03012_2000", "document_hash": "03e441e68d54e1af036f8ff905dfd6cbbccead62636279c1a8e179352ab08945", "question_id": "q11_aggregation", "split": "test", "cluster_id": "syn_03012_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 1, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1927, "query_tokens": 605, "message_tokens": 512, "message_utf8_bytes": 1213, "cache_tensor_bytes": null, "serialized_payload_bytes": 1213, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 22483.77687507309, "hops": {"1": 22483.77687507309}, "receiver_prefill_ms": 394.1142500843853, "ttft_ms": 58.20154200773686, "decode_ms": 517.7648330572993}, "total_wall_ms": 23454.506916110404, "ttft_ms": 58.20154200773686, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 8}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "19704477", "evidence_artifact": {"ledger": {"facts_total": 59, "facts_retained": 30, "retention_rate": 0.5084745762711864, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.5714285714285714, "amount": 0.6, "status": 0.5833333333333334, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1213}, "ledger_by_hop": {"1": {"facts_total": 59, "facts_retained": 30, "retention_rate": 0.5084745762711864, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.5714285714285714, "amount": 0.6, "status": 0.5833333333333334, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1213}}}, "reference_key_id": "keys/synthetic/syn_03012_2000.json", "task_scores": {"correct": 0.0, "abstained": 0.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "7c9b236b590faff2030e02587d972cf755ac2a88072e709174edbf405498973d", "output_hash": "ba53bcc03c2a9b1df7909eec6e2d5de7384570e56fb592d48b95504bfd97157d"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "9fd7c7b0ef3759fe411f79c2", "attempt_id": "27969a69e165", "run_group_id": "E4_core:A4|closed|qwen3b:h1", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:29:46.963226Z", "task_family": "synthetic", "document_id": "syn_03012_2000", "document_hash": "03e441e68d54e1af036f8ff905dfd6cbbccead62636279c1a8e179352ab08945", "question_id": "q12_unit_conversion", "split": "test", "cluster_id": "syn_03012_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 1, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1927, "query_tokens": 602, "message_tokens": 512, "message_utf8_bytes": 1213, "cache_tensor_bytes": null, "serialized_payload_bytes": 1213, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 22483.77687507309, "hops": {"1": 22483.77687507309}, "receiver_prefill_ms": 395.8152919076383, "ttft_ms": 24.31308396626264, "decode_ms": 175.0514169689268}, "total_wall_ms": 23079.51162511017, "ttft_ms": 24.31308396626264, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 5}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "44620", "evidence_artifact": {"ledger": {"facts_total": 59, "facts_retained": 30, "retention_rate": 0.5084745762711864, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.5714285714285714, "amount": 0.6, "status": 0.5833333333333334, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1213}, "ledger_by_hop": {"1": {"facts_total": 59, "facts_retained": 30, "retention_rate": 0.5084745762711864, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.5714285714285714, "amount": 0.6, "status": 0.5833333333333334, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1213}}}, "reference_key_id": "keys/synthetic/syn_03012_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "7c9b236b590faff2030e02587d972cf755ac2a88072e709174edbf405498973d", "output_hash": "2bd125847727be7e7c0a656ee4c81f3bc6d0fef11f536f50ce7bc83688907075"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "f4425d10c13de638764ef054", "attempt_id": "84f710319d5f", "run_group_id": "E4_core:A4|closed|qwen3b:h1", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:29:47.576030Z", "task_family": "synthetic", "document_id": "syn_03012_2000", "document_hash": "03e441e68d54e1af036f8ff905dfd6cbbccead62636279c1a8e179352ab08945", "question_id": "q13_unanswerable", "split": "test", "cluster_id": "syn_03012_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 1, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1927, "query_tokens": 607, "message_tokens": 512, "message_utf8_bytes": 1213, "cache_tensor_bytes": null, "serialized_payload_bytes": 1213, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 22483.77687507309, "hops": {"1": 22483.77687507309}, "receiver_prefill_ms": 358.9837499894202, "ttft_ms": 52.100708009675145, "decode_ms": 38.115625036880374}, "total_wall_ms": 22933.787625050172, "ttft_ms": 52.100708009675145, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 59, "facts_retained": 30, "retention_rate": 0.5084745762711864, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.5714285714285714, "amount": 0.6, "status": 0.5833333333333334, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1213}, "ledger_by_hop": {"1": {"facts_total": 59, "facts_retained": 30, "retention_rate": 0.5084745762711864, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.5714285714285714, "amount": 0.6, "status": 0.5833333333333334, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1213}}}, "reference_key_id": "keys/synthetic/syn_03012_2000.json", "task_scores": {"correct": 1.0, "abstained": 1.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "7c9b236b590faff2030e02587d972cf755ac2a88072e709174edbf405498973d", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "eb847734eec9b0dd40ab2d08", "attempt_id": "8b984e4b1f81", "run_group_id": "E4_core:A4|closed|qwen3b:h1", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:29:48.043238Z", "task_family": "synthetic", "document_id": "syn_03012_2000", "document_hash": "03e441e68d54e1af036f8ff905dfd6cbbccead62636279c1a8e179352ab08945", "question_id": "q14_unanswerable", "split": "test", "cluster_id": "syn_03012_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 1, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1927, "query_tokens": 607, "message_tokens": 512, "message_utf8_bytes": 1213, "cache_tensor_bytes": null, "serialized_payload_bytes": 1213, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 22483.77687507309, "hops": {"1": 22483.77687507309}, "receiver_prefill_ms": 523.9789999322966, "ttft_ms": 39.70225004013628, "decode_ms": 24.003832950256765}, "total_wall_ms": 23072.056875098497, "ttft_ms": 39.70225004013628, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 59, "facts_retained": 30, "retention_rate": 0.5084745762711864, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.5714285714285714, "amount": 0.6, "status": 0.5833333333333334, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1213}, "ledger_by_hop": {"1": {"facts_total": 59, "facts_retained": 30, "retention_rate": 0.5084745762711864, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.5714285714285714, "amount": 0.6, "status": 0.5833333333333334, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1213}}}, "reference_key_id": "keys/synthetic/syn_03012_2000.json", "task_scores": {"correct": 1.0, "abstained": 1.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "7c9b236b590faff2030e02587d972cf755ac2a88072e709174edbf405498973d", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "e5727a5f28ba89ce73fdc2df", "attempt_id": "04e24506e9d3", "run_group_id": "E4_core:A4|closed|qwen3b:h2", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:30:10.680183Z", "task_family": "synthetic", "document_id": "syn_03012_2000", "document_hash": "03e441e68d54e1af036f8ff905dfd6cbbccead62636279c1a8e179352ab08945", "question_id": "q00_single_fact", "split": "test", "cluster_id": "syn_03012_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 2, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1927, "query_tokens": 593, "message_tokens": 511, "message_utf8_bytes": 1208, "cache_tensor_bytes": null, "serialized_payload_bytes": 1208, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 44513.674042071216, "hops": {"1": 22483.77687507309, "2": 22029.897166998126}, "receiver_prefill_ms": 406.7442090017721, "ttft_ms": 21.833374979905784, "decode_ms": 448.9325419999659}, "total_wall_ms": 45392.098334035836, "ttft_ms": 21.833374979905784, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 10}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "XX-71M-6213", "evidence_artifact": {"ledger": {"facts_total": 59, "facts_retained": 30, "retention_rate": 0.5084745762711864, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.5714285714285714, "amount": 0.6, "status": 0.5833333333333334, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1208}, "ledger_by_hop": {"1": {"facts_total": 59, "facts_retained": 30, "retention_rate": 0.5084745762711864, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.5714285714285714, "amount": 0.6, "status": 0.5833333333333334, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1213}, "2": {"facts_total": 59, "facts_retained": 30, "retention_rate": 0.5084745762711864, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.5714285714285714, "amount": 0.6, "status": 0.5833333333333334, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1208}}}, "reference_key_id": "keys/synthetic/syn_03012_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "667371c4cebf83517e9807a77df9aaa26b062652ac7a464bc497f8741c1303bf", "output_hash": "66ef6c5da682b38795e9029723bbd71cdde2c4dec70cf1fae6efe62ef243f10a"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "7afc0cc6ad04dcf438d2753a", "attempt_id": "1bb5e55c8cb5", "run_group_id": "E4_core:A4|closed|qwen3b:h2", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:30:11.576257Z", "task_family": "synthetic", "document_id": "syn_03012_2000", "document_hash": "03e441e68d54e1af036f8ff905dfd6cbbccead62636279c1a8e179352ab08945", "question_id": "q01_single_fact", "split": "test", "cluster_id": "syn_03012_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 2, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1927, "query_tokens": 599, "message_tokens": 511, "message_utf8_bytes": 1208, "cache_tensor_bytes": null, "serialized_payload_bytes": 1208, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 44513.674042071216, "hops": {"1": 22483.77687507309, "2": 22029.897166998126}, "receiver_prefill_ms": 515.6356659717858, "ttft_ms": 41.79183300584555, "decode_ms": 27.345500071533024}, "total_wall_ms": 45099.03725003824, "ttft_ms": 41.79183300584555, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 59, "facts_retained": 30, "retention_rate": 0.5084745762711864, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.5714285714285714, "amount": 0.6, "status": 0.5833333333333334, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1208}, "ledger_by_hop": {"1": {"facts_total": 59, "facts_retained": 30, "retention_rate": 0.5084745762711864, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.5714285714285714, "amount": 0.6, "status": 0.5833333333333334, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1213}, "2": {"facts_total": 59, "facts_retained": 30, "retention_rate": 0.5084745762711864, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.5714285714285714, "amount": 0.6, "status": 0.5833333333333334, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1208}}}, "reference_key_id": "keys/synthetic/syn_03012_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "667371c4cebf83517e9807a77df9aaa26b062652ac7a464bc497f8741c1303bf", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "1a6377a78bfedab647477f61", "attempt_id": "86cbbc95516a", "run_group_id": "E4_core:A4|closed|qwen3b:h2", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:30:12.179536Z", "task_family": "synthetic", "document_id": "syn_03012_2000", "document_hash": "03e441e68d54e1af036f8ff905dfd6cbbccead62636279c1a8e179352ab08945", "question_id": "q02_exact_copy", "split": "test", "cluster_id": "syn_03012_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 2, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1927, "query_tokens": 598, "message_tokens": 511, "message_utf8_bytes": 1208, "cache_tensor_bytes": null, "serialized_payload_bytes": 1208, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 44513.674042071216, "hops": {"1": 22483.77687507309, "2": 22029.897166998126}, "receiver_prefill_ms": 390.3613749425858, "ttft_ms": 34.16112496051937, "decode_ms": 196.01854193024337}, "total_wall_ms": 45134.82508412562, "ttft_ms": 34.16112496051937, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 7}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "13.4343", "evidence_artifact": {"ledger": {"facts_total": 59, "facts_retained": 30, "retention_rate": 0.5084745762711864, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.5714285714285714, "amount": 0.6, "status": 0.5833333333333334, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1208}, "ledger_by_hop": {"1": {"facts_total": 59, "facts_retained": 30, "retention_rate": 0.5084745762711864, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.5714285714285714, "amount": 0.6, "status": 0.5833333333333334, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1213}, "2": {"facts_total": 59, "facts_retained": 30, "retention_rate": 0.5084745762711864, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.5714285714285714, "amount": 0.6, "status": 0.5833333333333334, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1208}}}, "reference_key_id": "keys/synthetic/syn_03012_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "667371c4cebf83517e9807a77df9aaa26b062652ac7a464bc497f8741c1303bf", "output_hash": "599bc22e373b1cf7b4679208eb9e515ad0e8dcb88dde0c7f579637ec7d3f3014"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "24dc1cf2ec47db5e7157e726", "attempt_id": "244b04914492", "run_group_id": "E4_core:A4|closed|qwen3b:h2", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:30:12.817106Z", "task_family": "synthetic", "document_id": "syn_03012_2000", "document_hash": "03e441e68d54e1af036f8ff905dfd6cbbccead62636279c1a8e179352ab08945", "question_id": "q03_exact_copy", "split": "test", "cluster_id": "syn_03012_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 2, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1927, "query_tokens": 594, "message_tokens": 511, "message_utf8_bytes": 1208, "cache_tensor_bytes": null, "serialized_payload_bytes": 1208, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 44513.674042071216, "hops": {"1": 22483.77687507309, "2": 22029.897166998126}, "receiver_prefill_ms": 387.66295893583447, "ttft_ms": 70.43312501627952, "decode_ms": 101.14883305504918}, "total_wall_ms": 45073.708667070605, "ttft_ms": 70.43312501627952, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 59, "facts_retained": 30, "retention_rate": 0.5084745762711864, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.5714285714285714, "amount": 0.6, "status": 0.5833333333333334, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1208}, "ledger_by_hop": {"1": {"facts_total": 59, "facts_retained": 30, "retention_rate": 0.5084745762711864, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.5714285714285714, "amount": 0.6, "status": 0.5833333333333334, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1213}, "2": {"facts_total": 59, "facts_retained": 30, "retention_rate": 0.5084745762711864, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.5714285714285714, "amount": 0.6, "status": 0.5833333333333334, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1208}}}, "reference_key_id": "keys/synthetic/syn_03012_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "667371c4cebf83517e9807a77df9aaa26b062652ac7a464bc497f8741c1303bf", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "2412ce1f4c6533f6d5487129", "attempt_id": "6db80ed110cd", "run_group_id": "E4_core:A4|closed|qwen3b:h2", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:30:13.396711Z", "task_family": "synthetic", "document_id": "syn_03012_2000", "document_hash": "03e441e68d54e1af036f8ff905dfd6cbbccead62636279c1a8e179352ab08945", "question_id": "q04_multi_hop", "split": "test", "cluster_id": "syn_03012_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 2, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1927, "query_tokens": 599, "message_tokens": 511, "message_utf8_bytes": 1208, "cache_tensor_bytes": null, "serialized_payload_bytes": 1208, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 44513.674042071216, "hops": {"1": 22483.77687507309, "2": 22029.897166998126}, "receiver_prefill_ms": 469.7324999142438, "ttft_ms": 24.339625029824674, "decode_ms": 234.05441700015217}, "total_wall_ms": 45242.39087605383, "ttft_ms": 24.339625029824674, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 7}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "8631552", "evidence_artifact": {"ledger": {"facts_total": 59, "facts_retained": 30, "retention_rate": 0.5084745762711864, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.5714285714285714, "amount": 0.6, "status": 0.5833333333333334, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1208}, "ledger_by_hop": {"1": {"facts_total": 59, "facts_retained": 30, "retention_rate": 0.5084745762711864, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.5714285714285714, "amount": 0.6, "status": 0.5833333333333334, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1213}, "2": {"facts_total": 59, "facts_retained": 30, "retention_rate": 0.5084745762711864, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.5714285714285714, "amount": 0.6, "status": 0.5833333333333334, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1208}}}, "reference_key_id": "keys/synthetic/syn_03012_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "667371c4cebf83517e9807a77df9aaa26b062652ac7a464bc497f8741c1303bf", "output_hash": "147d02c0587a77b3b53f0424c8d3d1454aaf2639ce9fdaf69af20db33363e849"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "4d5efa8da5535f3c78972b2e", "attempt_id": "653e9e2458db", "run_group_id": "E4_core:A4|closed|qwen3b:h2", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:30:14.145052Z", "task_family": "synthetic", "document_id": "syn_03012_2000", "document_hash": "03e441e68d54e1af036f8ff905dfd6cbbccead62636279c1a8e179352ab08945", "question_id": "q05_multi_hop", "split": "test", "cluster_id": "syn_03012_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 2, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1927, "query_tokens": 602, "message_tokens": 511, "message_utf8_bytes": 1208, "cache_tensor_bytes": null, "serialized_payload_bytes": 1208, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 44513.674042071216, "hops": {"1": 22483.77687507309, "2": 22029.897166998126}, "receiver_prefill_ms": 379.5497081009671, "ttft_ms": 16.022499999962747, "decode_ms": 589.1558750299737}, "total_wall_ms": 45499.23291709274, "ttft_ms": 16.022499999962747, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 10}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "DG-40E-2777", "evidence_artifact": {"ledger": {"facts_total": 59, "facts_retained": 30, "retention_rate": 0.5084745762711864, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.5714285714285714, "amount": 0.6, "status": 0.5833333333333334, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1208}, "ledger_by_hop": {"1": {"facts_total": 59, "facts_retained": 30, "retention_rate": 0.5084745762711864, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.5714285714285714, "amount": 0.6, "status": 0.5833333333333334, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1213}, "2": {"facts_total": 59, "facts_retained": 30, "retention_rate": 0.5084745762711864, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.5714285714285714, "amount": 0.6, "status": 0.5833333333333334, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1208}}}, "reference_key_id": "keys/synthetic/syn_03012_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "667371c4cebf83517e9807a77df9aaa26b062652ac7a464bc497f8741c1303bf", "output_hash": "16a7e3f4a7cc4a9b66f86c2cdcfad3b35daf1da199d9406e0627c8af2d8d73f3"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "4b13b9cd86c223fed417f0f1", "attempt_id": "cffd262918fd", "run_group_id": "E4_core:A4|closed|qwen3b:h2", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:30:15.147966Z", "task_family": "synthetic", "document_id": "syn_03012_2000", "document_hash": "03e441e68d54e1af036f8ff905dfd6cbbccead62636279c1a8e179352ab08945", "question_id": "q06_negation", "split": "test", "cluster_id": "syn_03012_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 2, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1927, "query_tokens": 594, "message_tokens": 511, "message_utf8_bytes": 1208, "cache_tensor_bytes": null, "serialized_payload_bytes": 1208, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 44513.674042071216, "hops": {"1": 22483.77687507309, "2": 22029.897166998126}, "receiver_prefill_ms": 401.0747920256108, "ttft_ms": 40.36120802629739, "decode_ms": 30.594332958571613}, "total_wall_ms": 44986.47045902908, "ttft_ms": 40.36120802629739, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "no", "evidence_artifact": {"ledger": {"facts_total": 59, "facts_retained": 30, "retention_rate": 0.5084745762711864, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.5714285714285714, "amount": 0.6, "status": 0.5833333333333334, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1208}, "ledger_by_hop": {"1": {"facts_total": 59, "facts_retained": 30, "retention_rate": 0.5084745762711864, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.5714285714285714, "amount": 0.6, "status": 0.5833333333333334, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1213}, "2": {"facts_total": 59, "facts_retained": 30, "retention_rate": 0.5084745762711864, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.5714285714285714, "amount": 0.6, "status": 0.5833333333333334, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1208}}}, "reference_key_id": "keys/synthetic/syn_03012_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "667371c4cebf83517e9807a77df9aaa26b062652ac7a464bc497f8741c1303bf", "output_hash": "9390298f3fb0c5b160498935d79cb139aef28e1c47358b4bbba61862b9c26e59"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "5ee195e3c40c2e3dbbdcd32d", "attempt_id": "e6fb0f1f29b8", "run_group_id": "E4_core:A4|closed|qwen3b:h2", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:30:15.638262Z", "task_family": "synthetic", "document_id": "syn_03012_2000", "document_hash": "03e441e68d54e1af036f8ff905dfd6cbbccead62636279c1a8e179352ab08945", "question_id": "q07_negation", "split": "test", "cluster_id": "syn_03012_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 2, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1927, "query_tokens": 594, "message_tokens": 511, "message_utf8_bytes": 1208, "cache_tensor_bytes": null, "serialized_payload_bytes": 1208, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 44513.674042071216, "hops": {"1": 22483.77687507309, "2": 22029.897166998126}, "receiver_prefill_ms": 398.8936669193208, "ttft_ms": 27.65866694971919, "decode_ms": 36.20595799293369}, "total_wall_ms": 44977.06458403263, "ttft_ms": 27.65866694971919, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "no", "evidence_artifact": {"ledger": {"facts_total": 59, "facts_retained": 30, "retention_rate": 0.5084745762711864, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.5714285714285714, "amount": 0.6, "status": 0.5833333333333334, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1208}, "ledger_by_hop": {"1": {"facts_total": 59, "facts_retained": 30, "retention_rate": 0.5084745762711864, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.5714285714285714, "amount": 0.6, "status": 0.5833333333333334, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1213}, "2": {"facts_total": 59, "facts_retained": 30, "retention_rate": 0.5084745762711864, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.5714285714285714, "amount": 0.6, "status": 0.5833333333333334, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1208}}}, "reference_key_id": "keys/synthetic/syn_03012_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "667371c4cebf83517e9807a77df9aaa26b062652ac7a464bc497f8741c1303bf", "output_hash": "9390298f3fb0c5b160498935d79cb139aef28e1c47358b4bbba61862b9c26e59"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "a80d0e4905b5c2747513e88d", "attempt_id": "8447145cdab3", "run_group_id": "E4_core:A4|closed|qwen3b:h2", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:30:16.117895Z", "task_family": "synthetic", "document_id": "syn_03012_2000", "document_hash": "03e441e68d54e1af036f8ff905dfd6cbbccead62636279c1a8e179352ab08945", "question_id": "q08_revision", "split": "test", "cluster_id": "syn_03012_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 2, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1927, "query_tokens": 609, "message_tokens": 511, "message_utf8_bytes": 1208, "cache_tensor_bytes": null, "serialized_payload_bytes": 1208, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 44513.674042071216, "hops": {"1": 22483.77687507309, "2": 22029.897166998126}, "receiver_prefill_ms": 400.6599999265745, "ttft_ms": 50.94508302863687, "decode_ms": 432.8847920987755}, "total_wall_ms": 45398.95283407532, "ttft_ms": 50.94508302863687, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 7}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "6602245", "evidence_artifact": {"ledger": {"facts_total": 59, "facts_retained": 30, "retention_rate": 0.5084745762711864, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.5714285714285714, "amount": 0.6, "status": 0.5833333333333334, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1208}, "ledger_by_hop": {"1": {"facts_total": 59, "facts_retained": 30, "retention_rate": 0.5084745762711864, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.5714285714285714, "amount": 0.6, "status": 0.5833333333333334, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1213}, "2": {"facts_total": 59, "facts_retained": 30, "retention_rate": 0.5084745762711864, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.5714285714285714, "amount": 0.6, "status": 0.5833333333333334, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1208}}}, "reference_key_id": "keys/synthetic/syn_03012_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "667371c4cebf83517e9807a77df9aaa26b062652ac7a464bc497f8741c1303bf", "output_hash": "d117fa8197988f8d1e4fabc6c4cdaab9516018e54553b57c63b875b3de82b16f"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "0dbfba5027a5f46af1e3ebe2", "attempt_id": "6e476b03ea8a", "run_group_id": "E4_core:A4|closed|qwen3b:h2", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:30:17.020966Z", "task_family": "synthetic", "document_id": "syn_03012_2000", "document_hash": "03e441e68d54e1af036f8ff905dfd6cbbccead62636279c1a8e179352ab08945", "question_id": "q09_revision", "split": "test", "cluster_id": "syn_03012_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 2, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1927, "query_tokens": 602, "message_tokens": 511, "message_utf8_bytes": 1208, "cache_tensor_bytes": null, "serialized_payload_bytes": 1208, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 44513.674042071216, "hops": {"1": 22483.77687507309, "2": 22029.897166998126}, "receiver_prefill_ms": 400.8842499461025, "ttft_ms": 29.88787495996803, "decode_ms": 238.04933298379183}, "total_wall_ms": 45183.25279210694, "ttft_ms": 29.88787495996803, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 7}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "5899340", "evidence_artifact": {"ledger": {"facts_total": 59, "facts_retained": 30, "retention_rate": 0.5084745762711864, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.5714285714285714, "amount": 0.6, "status": 0.5833333333333334, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1208}, "ledger_by_hop": {"1": {"facts_total": 59, "facts_retained": 30, "retention_rate": 0.5084745762711864, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.5714285714285714, "amount": 0.6, "status": 0.5833333333333334, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1213}, "2": {"facts_total": 59, "facts_retained": 30, "retention_rate": 0.5084745762711864, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.5714285714285714, "amount": 0.6, "status": 0.5833333333333334, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1208}}}, "reference_key_id": "keys/synthetic/syn_03012_2000.json", "task_scores": {"correct": 0.0, "abstained": 0.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "667371c4cebf83517e9807a77df9aaa26b062652ac7a464bc497f8741c1303bf", "output_hash": "c920323a9a02b1afb8619cf11bdde427f7cd1ea3289be69bd62932985895d717"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "3c16d1d36270ef1f211eb899", "attempt_id": "ead323bf463b", "run_group_id": "E4_core:A4|closed|qwen3b:h2", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:30:17.708659Z", "task_family": "synthetic", "document_id": "syn_03012_2000", "document_hash": "03e441e68d54e1af036f8ff905dfd6cbbccead62636279c1a8e179352ab08945", "question_id": "q10_temporal", "split": "test", "cluster_id": "syn_03012_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 2, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1927, "query_tokens": 612, "message_tokens": 511, "message_utf8_bytes": 1208, "cache_tensor_bytes": null, "serialized_payload_bytes": 1208, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 44513.674042071216, "hops": {"1": 22483.77687507309, "2": 22029.897166998126}, "receiver_prefill_ms": 376.54579104855657, "ttft_ms": 48.53450006339699, "decode_ms": 36.27770906314254}, "total_wall_ms": 44975.88395909406, "ttft_ms": 48.53450006339699, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 59, "facts_retained": 30, "retention_rate": 0.5084745762711864, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.5714285714285714, "amount": 0.6, "status": 0.5833333333333334, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1208}, "ledger_by_hop": {"1": {"facts_total": 59, "facts_retained": 30, "retention_rate": 0.5084745762711864, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.5714285714285714, "amount": 0.6, "status": 0.5833333333333334, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1213}, "2": {"facts_total": 59, "facts_retained": 30, "retention_rate": 0.5084745762711864, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.5714285714285714, "amount": 0.6, "status": 0.5833333333333334, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1208}}}, "reference_key_id": "keys/synthetic/syn_03012_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "667371c4cebf83517e9807a77df9aaa26b062652ac7a464bc497f8741c1303bf", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "147f928f7f71e8a62f3e8f4e", "attempt_id": "be26d5351311", "run_group_id": "E4_core:A4|closed|qwen3b:h2", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:30:18.187922Z", "task_family": "synthetic", "document_id": "syn_03012_2000", "document_hash": "03e441e68d54e1af036f8ff905dfd6cbbccead62636279c1a8e179352ab08945", "question_id": "q11_aggregation", "split": "test", "cluster_id": "syn_03012_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 2, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1927, "query_tokens": 604, "message_tokens": 511, "message_utf8_bytes": 1208, "cache_tensor_bytes": null, "serialized_payload_bytes": 1208, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 44513.674042071216, "hops": {"1": 22483.77687507309, "2": 22029.897166998126}, "receiver_prefill_ms": 555.9896670747548, "ttft_ms": 35.38287500850856, "decode_ms": 324.5806669583544}, "total_wall_ms": 45430.33995805308, "ttft_ms": 35.38287500850856, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 8}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "19383807", "evidence_artifact": {"ledger": {"facts_total": 59, "facts_retained": 30, "retention_rate": 0.5084745762711864, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.5714285714285714, "amount": 0.6, "status": 0.5833333333333334, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1208}, "ledger_by_hop": {"1": {"facts_total": 59, "facts_retained": 30, "retention_rate": 0.5084745762711864, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.5714285714285714, "amount": 0.6, "status": 0.5833333333333334, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1213}, "2": {"facts_total": 59, "facts_retained": 30, "retention_rate": 0.5084745762711864, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.5714285714285714, "amount": 0.6, "status": 0.5833333333333334, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1208}}}, "reference_key_id": "keys/synthetic/syn_03012_2000.json", "task_scores": {"correct": 0.0, "abstained": 0.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "667371c4cebf83517e9807a77df9aaa26b062652ac7a464bc497f8741c1303bf", "output_hash": "159f7326260a001c796a72394c031cdce250b1587222f9017dfffe73aa9591ff"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "fe7f5c2f5569292cb72513be", "attempt_id": "3c8e622555b9", "run_group_id": "E4_core:A4|closed|qwen3b:h2", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:30:19.121550Z", "task_family": "synthetic", "document_id": "syn_03012_2000", "document_hash": "03e441e68d54e1af036f8ff905dfd6cbbccead62636279c1a8e179352ab08945", "question_id": "q12_unit_conversion", "split": "test", "cluster_id": "syn_03012_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 2, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1927, "query_tokens": 601, "message_tokens": 511, "message_utf8_bytes": 1208, "cache_tensor_bytes": null, "serialized_payload_bytes": 1208, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 44513.674042071216, "hops": {"1": 22483.77687507309, "2": 22029.897166998126}, "receiver_prefill_ms": 375.1474169548601, "ttft_ms": 35.91345890890807, "decode_ms": 257.258249912411}, "total_wall_ms": 45182.68650001846, "ttft_ms": 35.91345890890807, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 5}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "44620", "evidence_artifact": {"ledger": {"facts_total": 59, "facts_retained": 30, "retention_rate": 0.5084745762711864, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.5714285714285714, "amount": 0.6, "status": 0.5833333333333334, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1208}, "ledger_by_hop": {"1": {"facts_total": 59, "facts_retained": 30, "retention_rate": 0.5084745762711864, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.5714285714285714, "amount": 0.6, "status": 0.5833333333333334, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1213}, "2": {"facts_total": 59, "facts_retained": 30, "retention_rate": 0.5084745762711864, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.5714285714285714, "amount": 0.6, "status": 0.5833333333333334, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1208}}}, "reference_key_id": "keys/synthetic/syn_03012_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "667371c4cebf83517e9807a77df9aaa26b062652ac7a464bc497f8741c1303bf", "output_hash": "2bd125847727be7e7c0a656ee4c81f3bc6d0fef11f536f50ce7bc83688907075"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "51359e7b2fc4b742c1fdce70", "attempt_id": "c4f0316e2bfc", "run_group_id": "E4_core:A4|closed|qwen3b:h2", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:30:19.807469Z", "task_family": "synthetic", "document_id": "syn_03012_2000", "document_hash": "03e441e68d54e1af036f8ff905dfd6cbbccead62636279c1a8e179352ab08945", "question_id": "q13_unanswerable", "split": "test", "cluster_id": "syn_03012_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 2, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1927, "query_tokens": 606, "message_tokens": 511, "message_utf8_bytes": 1208, "cache_tensor_bytes": null, "serialized_payload_bytes": 1208, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 44513.674042071216, "hops": {"1": 22483.77687507309, "2": 22029.897166998126}, "receiver_prefill_ms": 531.3063330249861, "ttft_ms": 33.02199998870492, "decode_ms": 31.822624965570867}, "total_wall_ms": 45110.43962603435, "ttft_ms": 33.02199998870492, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 59, "facts_retained": 30, "retention_rate": 0.5084745762711864, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.5714285714285714, "amount": 0.6, "status": 0.5833333333333334, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1208}, "ledger_by_hop": {"1": {"facts_total": 59, "facts_retained": 30, "retention_rate": 0.5084745762711864, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.5714285714285714, "amount": 0.6, "status": 0.5833333333333334, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1213}, "2": {"facts_total": 59, "facts_retained": 30, "retention_rate": 0.5084745762711864, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.5714285714285714, "amount": 0.6, "status": 0.5833333333333334, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1208}}}, "reference_key_id": "keys/synthetic/syn_03012_2000.json", "task_scores": {"correct": 1.0, "abstained": 1.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "667371c4cebf83517e9807a77df9aaa26b062652ac7a464bc497f8741c1303bf", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "fbe57e6e63dc084a3780c86e", "attempt_id": "5e25967ec049", "run_group_id": "E4_core:A4|closed|qwen3b:h2", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:30:20.422295Z", "task_family": "synthetic", "document_id": "syn_03012_2000", "document_hash": "03e441e68d54e1af036f8ff905dfd6cbbccead62636279c1a8e179352ab08945", "question_id": "q14_unanswerable", "split": "test", "cluster_id": "syn_03012_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 2, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1927, "query_tokens": 606, "message_tokens": 511, "message_utf8_bytes": 1208, "cache_tensor_bytes": null, "serialized_payload_bytes": 1208, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 44513.674042071216, "hops": {"1": 22483.77687507309, "2": 22029.897166998126}, "receiver_prefill_ms": 404.7411249484867, "ttft_ms": 34.524291986599565, "decode_ms": 29.303999966941774}, "total_wall_ms": 44982.83787607215, "ttft_ms": 34.524291986599565, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 59, "facts_retained": 30, "retention_rate": 0.5084745762711864, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.5714285714285714, "amount": 0.6, "status": 0.5833333333333334, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1208}, "ledger_by_hop": {"1": {"facts_total": 59, "facts_retained": 30, "retention_rate": 0.5084745762711864, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.5714285714285714, "amount": 0.6, "status": 0.5833333333333334, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1213}, "2": {"facts_total": 59, "facts_retained": 30, "retention_rate": 0.5084745762711864, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.5714285714285714, "amount": 0.6, "status": 0.5833333333333334, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1208}}}, "reference_key_id": "keys/synthetic/syn_03012_2000.json", "task_scores": {"correct": 1.0, "abstained": 1.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "667371c4cebf83517e9807a77df9aaa26b062652ac7a464bc497f8741c1303bf", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "002a37d0ef303c823d1540df", "attempt_id": "dd1cc17995ab", "run_group_id": "E4_core:A4|closed|qwen3b:h4", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:31:05.050008Z", "task_family": "synthetic", "document_id": "syn_03012_2000", "document_hash": "03e441e68d54e1af036f8ff905dfd6cbbccead62636279c1a8e179352ab08945", "question_id": "q00_single_fact", "split": "test", "cluster_id": "syn_03012_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 4, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1927, "query_tokens": 593, "message_tokens": 511, "message_utf8_bytes": 1208, "cache_tensor_bytes": null, "serialized_payload_bytes": 1208, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 88652.4670420913, "hops": {"1": 22483.77687507309, "2": 22029.897166998126, "3": 21959.27741599735, "4": 22179.515584022738}, "receiver_prefill_ms": 462.3680419754237, "ttft_ms": 36.97545896284282, "decode_ms": 335.4939999990165}, "total_wall_ms": 89487.95179207809, "ttft_ms": 36.97545896284282, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 10}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "XX-71M-6213", "evidence_artifact": {"ledger": {"facts_total": 59, "facts_retained": 30, "retention_rate": 0.5084745762711864, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.5714285714285714, "amount": 0.6, "status": 0.5833333333333334, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1208}, "ledger_by_hop": {"1": {"facts_total": 59, "facts_retained": 30, "retention_rate": 0.5084745762711864, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.5714285714285714, "amount": 0.6, "status": 0.5833333333333334, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1213}, "2": {"facts_total": 59, "facts_retained": 30, "retention_rate": 0.5084745762711864, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.5714285714285714, "amount": 0.6, "status": 0.5833333333333334, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1208}, "3": {"facts_total": 59, "facts_retained": 30, "retention_rate": 0.5084745762711864, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.5714285714285714, "amount": 0.6, "status": 0.5833333333333334, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1208}, "4": {"facts_total": 59, "facts_retained": 30, "retention_rate": 0.5084745762711864, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.5714285714285714, "amount": 0.6, "status": 0.5833333333333334, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1208}}}, "reference_key_id": "keys/synthetic/syn_03012_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "667371c4cebf83517e9807a77df9aaa26b062652ac7a464bc497f8741c1303bf", "output_hash": "66ef6c5da682b38795e9029723bbd71cdde2c4dec70cf1fae6efe62ef243f10a"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "804204194c361d58ef2001b6", "attempt_id": "137633f90808", "run_group_id": "E4_core:A4|closed|qwen3b:h4", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:31:05.903261Z", "task_family": "synthetic", "document_id": "syn_03012_2000", "document_hash": "03e441e68d54e1af036f8ff905dfd6cbbccead62636279c1a8e179352ab08945", "question_id": "q01_single_fact", "split": "test", "cluster_id": "syn_03012_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 4, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1927, "query_tokens": 599, "message_tokens": 511, "message_utf8_bytes": 1208, "cache_tensor_bytes": null, "serialized_payload_bytes": 1208, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 88652.4670420913, "hops": {"1": 22483.77687507309, "2": 22029.897166998126, "3": 21959.27741599735, "4": 22179.515584022738}, "receiver_prefill_ms": 372.8331249440089, "ttft_ms": 49.12791703827679, "decode_ms": 69.23991697840393}, "total_wall_ms": 89144.49820900336, "ttft_ms": 49.12791703827679, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 59, "facts_retained": 30, "retention_rate": 0.5084745762711864, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.5714285714285714, "amount": 0.6, "status": 0.5833333333333334, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1208}, "ledger_by_hop": {"1": {"facts_total": 59, "facts_retained": 30, "retention_rate": 0.5084745762711864, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.5714285714285714, "amount": 0.6, "status": 0.5833333333333334, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1213}, "2": {"facts_total": 59, "facts_retained": 30, "retention_rate": 0.5084745762711864, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.5714285714285714, "amount": 0.6, "status": 0.5833333333333334, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1208}, "3": {"facts_total": 59, "facts_retained": 30, "retention_rate": 0.5084745762711864, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.5714285714285714, "amount": 0.6, "status": 0.5833333333333334, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1208}, "4": {"facts_total": 59, "facts_retained": 30, "retention_rate": 0.5084745762711864, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.5714285714285714, "amount": 0.6, "status": 0.5833333333333334, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1208}}}, "reference_key_id": "keys/synthetic/syn_03012_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "667371c4cebf83517e9807a77df9aaa26b062652ac7a464bc497f8741c1303bf", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "d0da90bf4df5ddd99b3cff07", "attempt_id": "2939d76ccb80", "run_group_id": "E4_core:A4|closed|qwen3b:h4", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:31:06.412074Z", "task_family": "synthetic", "document_id": "syn_03012_2000", "document_hash": "03e441e68d54e1af036f8ff905dfd6cbbccead62636279c1a8e179352ab08945", "question_id": "q02_exact_copy", "split": "test", "cluster_id": "syn_03012_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 4, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1927, "query_tokens": 598, "message_tokens": 511, "message_utf8_bytes": 1208, "cache_tensor_bytes": null, "serialized_payload_bytes": 1208, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 88652.4670420913, "hops": {"1": 22483.77687507309, "2": 22029.897166998126, "3": 21959.27741599735, "4": 22179.515584022738}, "receiver_prefill_ms": 508.10445798560977, "ttft_ms": 39.192874915897846, "decode_ms": 214.08579195849597}, "total_wall_ms": 89414.56100007053, "ttft_ms": 39.192874915897846, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 7}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "13.4343", "evidence_artifact": {"ledger": {"facts_total": 59, "facts_retained": 30, "retention_rate": 0.5084745762711864, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.5714285714285714, "amount": 0.6, "status": 0.5833333333333334, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1208}, "ledger_by_hop": {"1": {"facts_total": 59, "facts_retained": 30, "retention_rate": 0.5084745762711864, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.5714285714285714, "amount": 0.6, "status": 0.5833333333333334, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1213}, "2": {"facts_total": 59, "facts_retained": 30, "retention_rate": 0.5084745762711864, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.5714285714285714, "amount": 0.6, "status": 0.5833333333333334, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1208}, "3": {"facts_total": 59, "facts_retained": 30, "retention_rate": 0.5084745762711864, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.5714285714285714, "amount": 0.6, "status": 0.5833333333333334, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1208}, "4": {"facts_total": 59, "facts_retained": 30, "retention_rate": 0.5084745762711864, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.5714285714285714, "amount": 0.6, "status": 0.5833333333333334, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1208}}}, "reference_key_id": "keys/synthetic/syn_03012_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "667371c4cebf83517e9807a77df9aaa26b062652ac7a464bc497f8741c1303bf", "output_hash": "599bc22e373b1cf7b4679208eb9e515ad0e8dcb88dde0c7f579637ec7d3f3014"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "e454f5c2aca21634410ced07", "attempt_id": "a03483a24665", "run_group_id": "E4_core:A4|closed|qwen3b:h4", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:31:07.192597Z", "task_family": "synthetic", "document_id": "syn_03012_2000", "document_hash": "03e441e68d54e1af036f8ff905dfd6cbbccead62636279c1a8e179352ab08945", "question_id": "q03_exact_copy", "split": "test", "cluster_id": "syn_03012_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 4, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1927, "query_tokens": 594, "message_tokens": 511, "message_utf8_bytes": 1208, "cache_tensor_bytes": null, "serialized_payload_bytes": 1208, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 88652.4670420913, "hops": {"1": 22483.77687507309, "2": 22029.897166998126, "3": 21959.27741599735, "4": 22179.515584022738}, "receiver_prefill_ms": 401.22095798142254, "ttft_ms": 33.99312496185303, "decode_ms": 30.017000040970743}, "total_wall_ms": 89118.27225005254, "ttft_ms": 33.99312496185303, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 59, "facts_retained": 30, "retention_rate": 0.5084745762711864, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.5714285714285714, "amount": 0.6, "status": 0.5833333333333334, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1208}, "ledger_by_hop": {"1": {"facts_total": 59, "facts_retained": 30, "retention_rate": 0.5084745762711864, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.5714285714285714, "amount": 0.6, "status": 0.5833333333333334, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1213}, "2": {"facts_total": 59, "facts_retained": 30, "retention_rate": 0.5084745762711864, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.5714285714285714, "amount": 0.6, "status": 0.5833333333333334, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1208}, "3": {"facts_total": 59, "facts_retained": 30, "retention_rate": 0.5084745762711864, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.5714285714285714, "amount": 0.6, "status": 0.5833333333333334, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1208}, "4": {"facts_total": 59, "facts_retained": 30, "retention_rate": 0.5084745762711864, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.5714285714285714, "amount": 0.6, "status": 0.5833333333333334, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1208}}}, "reference_key_id": "keys/synthetic/syn_03012_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "667371c4cebf83517e9807a77df9aaa26b062652ac7a464bc497f8741c1303bf", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "8b33e71e9d109f96a5f227a1", "attempt_id": "a24f800e4c0b", "run_group_id": "E4_core:A4|closed|qwen3b:h4", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:31:07.675743Z", "task_family": "synthetic", "document_id": "syn_03012_2000", "document_hash": "03e441e68d54e1af036f8ff905dfd6cbbccead62636279c1a8e179352ab08945", "question_id": "q04_multi_hop", "split": "test", "cluster_id": "syn_03012_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 4, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1927, "query_tokens": 599, "message_tokens": 511, "message_utf8_bytes": 1208, "cache_tensor_bytes": null, "serialized_payload_bytes": 1208, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 88652.4670420913, "hops": {"1": 22483.77687507309, "2": 22029.897166998126, "3": 21959.27741599735, "4": 22179.515584022738}, "receiver_prefill_ms": 332.48995803296566, "ttft_ms": 54.294000030495226, "decode_ms": 461.0016660299152}, "total_wall_ms": 89501.17354211397, "ttft_ms": 54.294000030495226, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 7}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "8631552", "evidence_artifact": {"ledger": {"facts_total": 59, "facts_retained": 30, "retention_rate": 0.5084745762711864, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.5714285714285714, "amount": 0.6, "status": 0.5833333333333334, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1208}, "ledger_by_hop": {"1": {"facts_total": 59, "facts_retained": 30, "retention_rate": 0.5084745762711864, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.5714285714285714, "amount": 0.6, "status": 0.5833333333333334, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1213}, "2": {"facts_total": 59, "facts_retained": 30, "retention_rate": 0.5084745762711864, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.5714285714285714, "amount": 0.6, "status": 0.5833333333333334, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1208}, "3": {"facts_total": 59, "facts_retained": 30, "retention_rate": 0.5084745762711864, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.5714285714285714, "amount": 0.6, "status": 0.5833333333333334, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1208}, "4": {"facts_total": 59, "facts_retained": 30, "retention_rate": 0.5084745762711864, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.5714285714285714, "amount": 0.6, "status": 0.5833333333333334, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1208}}}, "reference_key_id": "keys/synthetic/syn_03012_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "667371c4cebf83517e9807a77df9aaa26b062652ac7a464bc497f8741c1303bf", "output_hash": "147d02c0587a77b3b53f0424c8d3d1454aaf2639ce9fdaf69af20db33363e849"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "578e6b5e105710c8b96e6bf9", "attempt_id": "b1acb0c035b8", "run_group_id": "E4_core:A4|closed|qwen3b:h4", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:31:08.541899Z", "task_family": "synthetic", "document_id": "syn_03012_2000", "document_hash": "03e441e68d54e1af036f8ff905dfd6cbbccead62636279c1a8e179352ab08945", "question_id": "q05_multi_hop", "split": "test", "cluster_id": "syn_03012_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 4, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1927, "query_tokens": 602, "message_tokens": 511, "message_utf8_bytes": 1208, "cache_tensor_bytes": null, "serialized_payload_bytes": 1208, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 88652.4670420913, "hops": {"1": 22483.77687507309, "2": 22029.897166998126, "3": 21959.27741599735, "4": 22179.515584022738}, "receiver_prefill_ms": 404.38066702336073, "ttft_ms": 25.324083981104195, "decode_ms": 311.17933394853026}, "total_wall_ms": 89394.06987500843, "ttft_ms": 25.324083981104195, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 10}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "DG-40E-2777", "evidence_artifact": {"ledger": {"facts_total": 59, "facts_retained": 30, "retention_rate": 0.5084745762711864, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.5714285714285714, "amount": 0.6, "status": 0.5833333333333334, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1208}, "ledger_by_hop": {"1": {"facts_total": 59, "facts_retained": 30, "retention_rate": 0.5084745762711864, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.5714285714285714, "amount": 0.6, "status": 0.5833333333333334, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1213}, "2": {"facts_total": 59, "facts_retained": 30, "retention_rate": 0.5084745762711864, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.5714285714285714, "amount": 0.6, "status": 0.5833333333333334, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1208}, "3": {"facts_total": 59, "facts_retained": 30, "retention_rate": 0.5084745762711864, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.5714285714285714, "amount": 0.6, "status": 0.5833333333333334, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1208}, "4": {"facts_total": 59, "facts_retained": 30, "retention_rate": 0.5084745762711864, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.5714285714285714, "amount": 0.6, "status": 0.5833333333333334, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1208}}}, "reference_key_id": "keys/synthetic/syn_03012_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "667371c4cebf83517e9807a77df9aaa26b062652ac7a464bc497f8741c1303bf", "output_hash": "16a7e3f4a7cc4a9b66f86c2cdcfad3b35daf1da199d9406e0627c8af2d8d73f3"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "22d167298ad985643f0a4492", "attempt_id": "072c7e6cfc44", "run_group_id": "E4_core:A4|closed|qwen3b:h4", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:31:09.300475Z", "task_family": "synthetic", "document_id": "syn_03012_2000", "document_hash": "03e441e68d54e1af036f8ff905dfd6cbbccead62636279c1a8e179352ab08945", "question_id": "q06_negation", "split": "test", "cluster_id": "syn_03012_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 4, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1927, "query_tokens": 594, "message_tokens": 511, "message_utf8_bytes": 1208, "cache_tensor_bytes": null, "serialized_payload_bytes": 1208, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 88652.4670420913, "hops": {"1": 22483.77687507309, "2": 22029.897166998126, "3": 21959.27741599735, "4": 22179.515584022738}, "receiver_prefill_ms": 347.2085000248626, "ttft_ms": 38.1269589997828, "decode_ms": 60.48204202670604}, "total_wall_ms": 89099.0973340813, "ttft_ms": 38.1269589997828, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "no", "evidence_artifact": {"ledger": {"facts_total": 59, "facts_retained": 30, "retention_rate": 0.5084745762711864, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.5714285714285714, "amount": 0.6, "status": 0.5833333333333334, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1208}, "ledger_by_hop": {"1": {"facts_total": 59, "facts_retained": 30, "retention_rate": 0.5084745762711864, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.5714285714285714, "amount": 0.6, "status": 0.5833333333333334, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1213}, "2": {"facts_total": 59, "facts_retained": 30, "retention_rate": 0.5084745762711864, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.5714285714285714, "amount": 0.6, "status": 0.5833333333333334, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1208}, "3": {"facts_total": 59, "facts_retained": 30, "retention_rate": 0.5084745762711864, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.5714285714285714, "amount": 0.6, "status": 0.5833333333333334, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1208}, "4": {"facts_total": 59, "facts_retained": 30, "retention_rate": 0.5084745762711864, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.5714285714285714, "amount": 0.6, "status": 0.5833333333333334, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1208}}}, "reference_key_id": "keys/synthetic/syn_03012_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "667371c4cebf83517e9807a77df9aaa26b062652ac7a464bc497f8741c1303bf", "output_hash": "9390298f3fb0c5b160498935d79cb139aef28e1c47358b4bbba61862b9c26e59"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "e002b9d6934b79e2435d09f3", "attempt_id": "7abcd7acfc38", "run_group_id": "E4_core:A4|closed|qwen3b:h4", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:31:09.764753Z", "task_family": "synthetic", "document_id": "syn_03012_2000", "document_hash": "03e441e68d54e1af036f8ff905dfd6cbbccead62636279c1a8e179352ab08945", "question_id": "q07_negation", "split": "test", "cluster_id": "syn_03012_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 4, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1927, "query_tokens": 594, "message_tokens": 511, "message_utf8_bytes": 1208, "cache_tensor_bytes": null, "serialized_payload_bytes": 1208, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 88652.4670420913, "hops": {"1": 22483.77687507309, "2": 22029.897166998126, "3": 21959.27741599735, "4": 22179.515584022738}, "receiver_prefill_ms": 555.5287089664489, "ttft_ms": 28.177208034321666, "decode_ms": 36.94704198278487}, "total_wall_ms": 89273.72345910408, "ttft_ms": 28.177208034321666, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "no", "evidence_artifact": {"ledger": {"facts_total": 59, "facts_retained": 30, "retention_rate": 0.5084745762711864, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.5714285714285714, "amount": 0.6, "status": 0.5833333333333334, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1208}, "ledger_by_hop": {"1": {"facts_total": 59, "facts_retained": 30, "retention_rate": 0.5084745762711864, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.5714285714285714, "amount": 0.6, "status": 0.5833333333333334, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1213}, "2": {"facts_total": 59, "facts_retained": 30, "retention_rate": 0.5084745762711864, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.5714285714285714, "amount": 0.6, "status": 0.5833333333333334, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1208}, "3": {"facts_total": 59, "facts_retained": 30, "retention_rate": 0.5084745762711864, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.5714285714285714, "amount": 0.6, "status": 0.5833333333333334, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1208}, "4": {"facts_total": 59, "facts_retained": 30, "retention_rate": 0.5084745762711864, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.5714285714285714, "amount": 0.6, "status": 0.5833333333333334, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1208}}}, "reference_key_id": "keys/synthetic/syn_03012_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "667371c4cebf83517e9807a77df9aaa26b062652ac7a464bc497f8741c1303bf", "output_hash": "9390298f3fb0c5b160498935d79cb139aef28e1c47358b4bbba61862b9c26e59"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "b2bf55d3e33e6c501c9939fe", "attempt_id": "0fa7d52310a3", "run_group_id": "E4_core:A4|closed|qwen3b:h4", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:31:10.402556Z", "task_family": "synthetic", "document_id": "syn_03012_2000", "document_hash": "03e441e68d54e1af036f8ff905dfd6cbbccead62636279c1a8e179352ab08945", "question_id": "q08_revision", "split": "test", "cluster_id": "syn_03012_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 4, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1927, "query_tokens": 609, "message_tokens": 511, "message_utf8_bytes": 1208, "cache_tensor_bytes": null, "serialized_payload_bytes": 1208, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 88652.4670420913, "hops": {"1": 22483.77687507309, "2": 22029.897166998126, "3": 21959.27741599735, "4": 22179.515584022738}, "receiver_prefill_ms": 406.4866660628468, "ttft_ms": 43.224125052802265, "decode_ms": 215.09358298499137}, "total_wall_ms": 89317.8567920113, "ttft_ms": 43.224125052802265, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 7}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "6602245", "evidence_artifact": {"ledger": {"facts_total": 59, "facts_retained": 30, "retention_rate": 0.5084745762711864, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.5714285714285714, "amount": 0.6, "status": 0.5833333333333334, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1208}, "ledger_by_hop": {"1": {"facts_total": 59, "facts_retained": 30, "retention_rate": 0.5084745762711864, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.5714285714285714, "amount": 0.6, "status": 0.5833333333333334, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1213}, "2": {"facts_total": 59, "facts_retained": 30, "retention_rate": 0.5084745762711864, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.5714285714285714, "amount": 0.6, "status": 0.5833333333333334, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1208}, "3": {"facts_total": 59, "facts_retained": 30, "retention_rate": 0.5084745762711864, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.5714285714285714, "amount": 0.6, "status": 0.5833333333333334, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1208}, "4": {"facts_total": 59, "facts_retained": 30, "retention_rate": 0.5084745762711864, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.5714285714285714, "amount": 0.6, "status": 0.5833333333333334, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1208}}}, "reference_key_id": "keys/synthetic/syn_03012_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "667371c4cebf83517e9807a77df9aaa26b062652ac7a464bc497f8741c1303bf", "output_hash": "d117fa8197988f8d1e4fabc6c4cdaab9516018e54553b57c63b875b3de82b16f"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "99bd6c9a9171114e109d32a5", "attempt_id": "9815c2c2e9cb", "run_group_id": "E4_core:A4|closed|qwen3b:h4", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:31:11.085467Z", "task_family": "synthetic", "document_id": "syn_03012_2000", "document_hash": "03e441e68d54e1af036f8ff905dfd6cbbccead62636279c1a8e179352ab08945", "question_id": "q09_revision", "split": "test", "cluster_id": "syn_03012_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 4, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1927, "query_tokens": 602, "message_tokens": 511, "message_utf8_bytes": 1208, "cache_tensor_bytes": null, "serialized_payload_bytes": 1208, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 88652.4670420913, "hops": {"1": 22483.77687507309, "2": 22029.897166998126, "3": 21959.27741599735, "4": 22179.515584022738}, "receiver_prefill_ms": 368.93716594204307, "ttft_ms": 42.388707981444895, "decode_ms": 428.58066700864583}, "total_wall_ms": 89493.07462607976, "ttft_ms": 42.388707981444895, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 7}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "5899340", "evidence_artifact": {"ledger": {"facts_total": 59, "facts_retained": 30, "retention_rate": 0.5084745762711864, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.5714285714285714, "amount": 0.6, "status": 0.5833333333333334, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1208}, "ledger_by_hop": {"1": {"facts_total": 59, "facts_retained": 30, "retention_rate": 0.5084745762711864, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.5714285714285714, "amount": 0.6, "status": 0.5833333333333334, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1213}, "2": {"facts_total": 59, "facts_retained": 30, "retention_rate": 0.5084745762711864, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.5714285714285714, "amount": 0.6, "status": 0.5833333333333334, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1208}, "3": {"facts_total": 59, "facts_retained": 30, "retention_rate": 0.5084745762711864, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.5714285714285714, "amount": 0.6, "status": 0.5833333333333334, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1208}, "4": {"facts_total": 59, "facts_retained": 30, "retention_rate": 0.5084745762711864, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.5714285714285714, "amount": 0.6, "status": 0.5833333333333334, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1208}}}, "reference_key_id": "keys/synthetic/syn_03012_2000.json", "task_scores": {"correct": 0.0, "abstained": 0.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "667371c4cebf83517e9807a77df9aaa26b062652ac7a464bc497f8741c1303bf", "output_hash": "c920323a9a02b1afb8619cf11bdde427f7cd1ea3289be69bd62932985895d717"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "766f2e534e98afffa3e69737", "attempt_id": "79e72d07375e", "run_group_id": "E4_core:A4|closed|qwen3b:h4", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:31:11.944782Z", "task_family": "synthetic", "document_id": "syn_03012_2000", "document_hash": "03e441e68d54e1af036f8ff905dfd6cbbccead62636279c1a8e179352ab08945", "question_id": "q10_temporal", "split": "test", "cluster_id": "syn_03012_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 4, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1927, "query_tokens": 612, "message_tokens": 511, "message_utf8_bytes": 1208, "cache_tensor_bytes": null, "serialized_payload_bytes": 1208, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 88652.4670420913, "hops": {"1": 22483.77687507309, "2": 22029.897166998126, "3": 21959.27741599735, "4": 22179.515584022738}, "receiver_prefill_ms": 410.6627500150353, "ttft_ms": 34.795665997080505, "decode_ms": 32.35383296851069}, "total_wall_ms": 89130.82633411977, "ttft_ms": 34.795665997080505, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 59, "facts_retained": 30, "retention_rate": 0.5084745762711864, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.5714285714285714, "amount": 0.6, "status": 0.5833333333333334, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1208}, "ledger_by_hop": {"1": {"facts_total": 59, "facts_retained": 30, "retention_rate": 0.5084745762711864, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.5714285714285714, "amount": 0.6, "status": 0.5833333333333334, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1213}, "2": {"facts_total": 59, "facts_retained": 30, "retention_rate": 0.5084745762711864, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.5714285714285714, "amount": 0.6, "status": 0.5833333333333334, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1208}, "3": {"facts_total": 59, "facts_retained": 30, "retention_rate": 0.5084745762711864, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.5714285714285714, "amount": 0.6, "status": 0.5833333333333334, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1208}, "4": {"facts_total": 59, "facts_retained": 30, "retention_rate": 0.5084745762711864, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.5714285714285714, "amount": 0.6, "status": 0.5833333333333334, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1208}}}, "reference_key_id": "keys/synthetic/syn_03012_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "667371c4cebf83517e9807a77df9aaa26b062652ac7a464bc497f8741c1303bf", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "9fd0293a4c08b46c4303241a", "attempt_id": "7df8a6ae0be2", "run_group_id": "E4_core:A4|closed|qwen3b:h4", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:31:12.440233Z", "task_family": "synthetic", "document_id": "syn_03012_2000", "document_hash": "03e441e68d54e1af036f8ff905dfd6cbbccead62636279c1a8e179352ab08945", "question_id": "q11_aggregation", "split": "test", "cluster_id": "syn_03012_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 4, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1927, "query_tokens": 604, "message_tokens": 511, "message_utf8_bytes": 1208, "cache_tensor_bytes": null, "serialized_payload_bytes": 1208, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 88652.4670420913, "hops": {"1": 22483.77687507309, "2": 22029.897166998126, "3": 21959.27741599735, "4": 22179.515584022738}, "receiver_prefill_ms": 379.6289999736473, "ttft_ms": 19.9703749967739, "decode_ms": 570.4962919699028}, "total_wall_ms": 89623.21112502832, "ttft_ms": 19.9703749967739, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 8}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "19704407", "evidence_artifact": {"ledger": {"facts_total": 59, "facts_retained": 30, "retention_rate": 0.5084745762711864, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.5714285714285714, "amount": 0.6, "status": 0.5833333333333334, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1208}, "ledger_by_hop": {"1": {"facts_total": 59, "facts_retained": 30, "retention_rate": 0.5084745762711864, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.5714285714285714, "amount": 0.6, "status": 0.5833333333333334, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1213}, "2": {"facts_total": 59, "facts_retained": 30, "retention_rate": 0.5084745762711864, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.5714285714285714, "amount": 0.6, "status": 0.5833333333333334, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1208}, "3": {"facts_total": 59, "facts_retained": 30, "retention_rate": 0.5084745762711864, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.5714285714285714, "amount": 0.6, "status": 0.5833333333333334, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1208}, "4": {"facts_total": 59, "facts_retained": 30, "retention_rate": 0.5084745762711864, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.5714285714285714, "amount": 0.6, "status": 0.5833333333333334, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1208}}}, "reference_key_id": "keys/synthetic/syn_03012_2000.json", "task_scores": {"correct": 0.0, "abstained": 0.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "667371c4cebf83517e9807a77df9aaa26b062652ac7a464bc497f8741c1303bf", "output_hash": "7d2f092d592efff17462066772d9cbf34a039e2cf284b0bec7cabb81fc0421eb"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "4d5461fe0995aa598dde1683", "attempt_id": "2c3facd027d5", "run_group_id": "E4_core:A4|closed|qwen3b:h4", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:31:13.428287Z", "task_family": "synthetic", "document_id": "syn_03012_2000", "document_hash": "03e441e68d54e1af036f8ff905dfd6cbbccead62636279c1a8e179352ab08945", "question_id": "q12_unit_conversion", "split": "test", "cluster_id": "syn_03012_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 4, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1927, "query_tokens": 601, "message_tokens": 511, "message_utf8_bytes": 1208, "cache_tensor_bytes": null, "serialized_payload_bytes": 1208, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 88652.4670420913, "hops": {"1": 22483.77687507309, "2": 22029.897166998126, "3": 21959.27741599735, "4": 22179.515584022738}, "receiver_prefill_ms": 401.81754098739475, "ttft_ms": 45.24570901412517, "decode_ms": 156.07133298180997}, "total_wall_ms": 89256.36804208625, "ttft_ms": 45.24570901412517, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 5}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "44620", "evidence_artifact": {"ledger": {"facts_total": 59, "facts_retained": 30, "retention_rate": 0.5084745762711864, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.5714285714285714, "amount": 0.6, "status": 0.5833333333333334, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1208}, "ledger_by_hop": {"1": {"facts_total": 59, "facts_retained": 30, "retention_rate": 0.5084745762711864, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.5714285714285714, "amount": 0.6, "status": 0.5833333333333334, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1213}, "2": {"facts_total": 59, "facts_retained": 30, "retention_rate": 0.5084745762711864, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.5714285714285714, "amount": 0.6, "status": 0.5833333333333334, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1208}, "3": {"facts_total": 59, "facts_retained": 30, "retention_rate": 0.5084745762711864, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.5714285714285714, "amount": 0.6, "status": 0.5833333333333334, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1208}, "4": {"facts_total": 59, "facts_retained": 30, "retention_rate": 0.5084745762711864, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.5714285714285714, "amount": 0.6, "status": 0.5833333333333334, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1208}}}, "reference_key_id": "keys/synthetic/syn_03012_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "667371c4cebf83517e9807a77df9aaa26b062652ac7a464bc497f8741c1303bf", "output_hash": "2bd125847727be7e7c0a656ee4c81f3bc6d0fef11f536f50ce7bc83688907075"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "b36be4869d350f8b01a99c70", "attempt_id": "9414644da5dd", "run_group_id": "E4_core:A4|closed|qwen3b:h4", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:31:14.049193Z", "task_family": "synthetic", "document_id": "syn_03012_2000", "document_hash": "03e441e68d54e1af036f8ff905dfd6cbbccead62636279c1a8e179352ab08945", "question_id": "q13_unanswerable", "split": "test", "cluster_id": "syn_03012_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 4, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1927, "query_tokens": 606, "message_tokens": 511, "message_utf8_bytes": 1208, "cache_tensor_bytes": null, "serialized_payload_bytes": 1208, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 88652.4670420913, "hops": {"1": 22483.77687507309, "2": 22029.897166998126, "3": 21959.27741599735, "4": 22179.515584022738}, "receiver_prefill_ms": 372.9344579624012, "ttft_ms": 21.79370797239244, "decode_ms": 67.67949997447431}, "total_wall_ms": 89115.5738341622, "ttft_ms": 21.79370797239244, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 59, "facts_retained": 30, "retention_rate": 0.5084745762711864, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.5714285714285714, "amount": 0.6, "status": 0.5833333333333334, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1208}, "ledger_by_hop": {"1": {"facts_total": 59, "facts_retained": 30, "retention_rate": 0.5084745762711864, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.5714285714285714, "amount": 0.6, "status": 0.5833333333333334, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1213}, "2": {"facts_total": 59, "facts_retained": 30, "retention_rate": 0.5084745762711864, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.5714285714285714, "amount": 0.6, "status": 0.5833333333333334, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1208}, "3": {"facts_total": 59, "facts_retained": 30, "retention_rate": 0.5084745762711864, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.5714285714285714, "amount": 0.6, "status": 0.5833333333333334, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1208}, "4": {"facts_total": 59, "facts_retained": 30, "retention_rate": 0.5084745762711864, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.5714285714285714, "amount": 0.6, "status": 0.5833333333333334, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1208}}}, "reference_key_id": "keys/synthetic/syn_03012_2000.json", "task_scores": {"correct": 1.0, "abstained": 1.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "667371c4cebf83517e9807a77df9aaa26b062652ac7a464bc497f8741c1303bf", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "8fc739bd4bc873e911374d67", "attempt_id": "bd4bcae4c817", "run_group_id": "E4_core:A4|closed|qwen3b:h4", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:31:14.529270Z", "task_family": "synthetic", "document_id": "syn_03012_2000", "document_hash": "03e441e68d54e1af036f8ff905dfd6cbbccead62636279c1a8e179352ab08945", "question_id": "q14_unanswerable", "split": "test", "cluster_id": "syn_03012_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 4, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1927, "query_tokens": 606, "message_tokens": 511, "message_utf8_bytes": 1208, "cache_tensor_bytes": null, "serialized_payload_bytes": 1208, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 88652.4670420913, "hops": {"1": 22483.77687507309, "2": 22029.897166998126, "3": 21959.27741599735, "4": 22179.515584022738}, "receiver_prefill_ms": 408.1088330131024, "ttft_ms": 111.98666703421623, "decode_ms": 100.4397920332849}, "total_wall_ms": 89273.72508309782, "ttft_ms": 111.98666703421623, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 59, "facts_retained": 30, "retention_rate": 0.5084745762711864, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.5714285714285714, "amount": 0.6, "status": 0.5833333333333334, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1208}, "ledger_by_hop": {"1": {"facts_total": 59, "facts_retained": 30, "retention_rate": 0.5084745762711864, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.5714285714285714, "amount": 0.6, "status": 0.5833333333333334, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1213}, "2": {"facts_total": 59, "facts_retained": 30, "retention_rate": 0.5084745762711864, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.5714285714285714, "amount": 0.6, "status": 0.5833333333333334, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1208}, "3": {"facts_total": 59, "facts_retained": 30, "retention_rate": 0.5084745762711864, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.5714285714285714, "amount": 0.6, "status": 0.5833333333333334, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1208}, "4": {"facts_total": 59, "facts_retained": 30, "retention_rate": 0.5084745762711864, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.5714285714285714, "amount": 0.6, "status": 0.5833333333333334, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1208}}}, "reference_key_id": "keys/synthetic/syn_03012_2000.json", "task_scores": {"correct": 1.0, "abstained": 1.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "667371c4cebf83517e9807a77df9aaa26b062652ac7a464bc497f8741c1303bf", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "cfeb22bfed6b2e24ee09342b", "attempt_id": "3e14dc2e9557", "run_group_id": "E4_core:A4|closed|qwen3b:h8", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:32:43.176651Z", "task_family": "synthetic", "document_id": "syn_03012_2000", "document_hash": "03e441e68d54e1af036f8ff905dfd6cbbccead62636279c1a8e179352ab08945", "question_id": "q00_single_fact", "split": "test", "cluster_id": "syn_03012_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 8, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1927, "query_tokens": 593, "message_tokens": 511, "message_utf8_bytes": 1208, "cache_tensor_bytes": null, "serialized_payload_bytes": 1208, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 176655.71554307826, "hops": {"1": 22483.77687507309, "2": 22029.897166998126, "3": 21959.27741599735, "4": 22179.515584022738, "5": 22028.93558400683, "6": 21918.76791708637, "7": 22043.121207971126, "8": 22012.42379192263}, "receiver_prefill_ms": 548.6037079244852, "ttft_ms": 26.592415990307927, "decode_ms": 351.5254579251632}, "total_wall_ms": 177583.1808759831, "ttft_ms": 26.592415990307927, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 10}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "XX-71M-6213", "evidence_artifact": {"ledger": {"facts_total": 59, "facts_retained": 30, "retention_rate": 0.5084745762711864, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.5714285714285714, "amount": 0.6, "status": 0.5833333333333334, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1208}, "ledger_by_hop": {"1": {"facts_total": 59, "facts_retained": 30, "retention_rate": 0.5084745762711864, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.5714285714285714, "amount": 0.6, "status": 0.5833333333333334, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1213}, "2": {"facts_total": 59, "facts_retained": 30, "retention_rate": 0.5084745762711864, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.5714285714285714, "amount": 0.6, "status": 0.5833333333333334, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1208}, "3": {"facts_total": 59, "facts_retained": 30, "retention_rate": 0.5084745762711864, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.5714285714285714, "amount": 0.6, "status": 0.5833333333333334, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1208}, "4": {"facts_total": 59, "facts_retained": 30, "retention_rate": 0.5084745762711864, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.5714285714285714, "amount": 0.6, "status": 0.5833333333333334, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1208}, "5": {"facts_total": 59, "facts_retained": 30, "retention_rate": 0.5084745762711864, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.5714285714285714, "amount": 0.6, "status": 0.5833333333333334, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1208}, "6": {"facts_total": 59, "facts_retained": 30, "retention_rate": 0.5084745762711864, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.5714285714285714, "amount": 0.6, "status": 0.5833333333333334, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1208}, "7": {"facts_total": 59, "facts_retained": 30, "retention_rate": 0.5084745762711864, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.5714285714285714, "amount": 0.6, "status": 0.5833333333333334, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1208}, "8": {"facts_total": 59, "facts_retained": 30, "retention_rate": 0.5084745762711864, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.5714285714285714, "amount": 0.6, "status": 0.5833333333333334, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1208}}}, "reference_key_id": "keys/synthetic/syn_03012_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "667371c4cebf83517e9807a77df9aaa26b062652ac7a464bc497f8741c1303bf", "output_hash": "66ef6c5da682b38795e9029723bbd71cdde2c4dec70cf1fae6efe62ef243f10a"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "408d3baa6d6aec4a7808cffa", "attempt_id": "bd87f09205b9", "run_group_id": "E4_core:A4|closed|qwen3b:h8", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:32:44.122922Z", "task_family": "synthetic", "document_id": "syn_03012_2000", "document_hash": "03e441e68d54e1af036f8ff905dfd6cbbccead62636279c1a8e179352ab08945", "question_id": "q01_single_fact", "split": "test", "cluster_id": "syn_03012_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 8, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1927, "query_tokens": 599, "message_tokens": 511, "message_utf8_bytes": 1208, "cache_tensor_bytes": null, "serialized_payload_bytes": 1208, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 176655.71554307826, "hops": {"1": 22483.77687507309, "2": 22029.897166998126, "3": 21959.27741599735, "4": 22179.515584022738, "5": 22028.93558400683, "6": 21918.76791708637, "7": 22043.121207971126, "8": 22012.42379192263}, "receiver_prefill_ms": 387.57091702427715, "ttft_ms": 23.732459056191146, "decode_ms": 66.16725004278123}, "total_wall_ms": 177133.93162598368, "ttft_ms": 23.732459056191146, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 59, "facts_retained": 30, "retention_rate": 0.5084745762711864, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.5714285714285714, "amount": 0.6, "status": 0.5833333333333334, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1208}, "ledger_by_hop": {"1": {"facts_total": 59, "facts_retained": 30, "retention_rate": 0.5084745762711864, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.5714285714285714, "amount": 0.6, "status": 0.5833333333333334, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1213}, "2": {"facts_total": 59, "facts_retained": 30, "retention_rate": 0.5084745762711864, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.5714285714285714, "amount": 0.6, "status": 0.5833333333333334, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1208}, "3": {"facts_total": 59, "facts_retained": 30, "retention_rate": 0.5084745762711864, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.5714285714285714, "amount": 0.6, "status": 0.5833333333333334, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1208}, "4": {"facts_total": 59, "facts_retained": 30, "retention_rate": 0.5084745762711864, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.5714285714285714, "amount": 0.6, "status": 0.5833333333333334, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1208}, "5": {"facts_total": 59, "facts_retained": 30, "retention_rate": 0.5084745762711864, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.5714285714285714, "amount": 0.6, "status": 0.5833333333333334, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1208}, "6": {"facts_total": 59, "facts_retained": 30, "retention_rate": 0.5084745762711864, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.5714285714285714, "amount": 0.6, "status": 0.5833333333333334, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1208}, "7": {"facts_total": 59, "facts_retained": 30, "retention_rate": 0.5084745762711864, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.5714285714285714, "amount": 0.6, "status": 0.5833333333333334, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1208}, "8": {"facts_total": 59, "facts_retained": 30, "retention_rate": 0.5084745762711864, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.5714285714285714, "amount": 0.6, "status": 0.5833333333333334, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1208}}}, "reference_key_id": "keys/synthetic/syn_03012_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "667371c4cebf83517e9807a77df9aaa26b062652ac7a464bc497f8741c1303bf", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "6b950dfe061aa3268850d05f", "attempt_id": "c945612d18bb", "run_group_id": "E4_core:A4|closed|qwen3b:h8", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:32:44.618834Z", "task_family": "synthetic", "document_id": "syn_03012_2000", "document_hash": "03e441e68d54e1af036f8ff905dfd6cbbccead62636279c1a8e179352ab08945", "question_id": "q02_exact_copy", "split": "test", "cluster_id": "syn_03012_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 8, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1927, "query_tokens": 598, "message_tokens": 511, "message_utf8_bytes": 1208, "cache_tensor_bytes": null, "serialized_payload_bytes": 1208, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 176655.71554307826, "hops": {"1": 22483.77687507309, "2": 22029.897166998126, "3": 21959.27741599735, "4": 22179.515584022738, "5": 22028.93558400683, "6": 21918.76791708637, "7": 22043.121207971126, "8": 22012.42379192263}, "receiver_prefill_ms": 388.93537502735853, "ttft_ms": 90.68054193630815, "decode_ms": 357.3124590329826}, "total_wall_ms": 177493.35570912808, "ttft_ms": 90.68054193630815, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 7}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "13.4343", "evidence_artifact": {"ledger": {"facts_total": 59, "facts_retained": 30, "retention_rate": 0.5084745762711864, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.5714285714285714, "amount": 0.6, "status": 0.5833333333333334, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1208}, "ledger_by_hop": {"1": {"facts_total": 59, "facts_retained": 30, "retention_rate": 0.5084745762711864, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.5714285714285714, "amount": 0.6, "status": 0.5833333333333334, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1213}, "2": {"facts_total": 59, "facts_retained": 30, "retention_rate": 0.5084745762711864, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.5714285714285714, "amount": 0.6, "status": 0.5833333333333334, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1208}, "3": {"facts_total": 59, "facts_retained": 30, "retention_rate": 0.5084745762711864, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.5714285714285714, "amount": 0.6, "status": 0.5833333333333334, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1208}, "4": {"facts_total": 59, "facts_retained": 30, "retention_rate": 0.5084745762711864, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.5714285714285714, "amount": 0.6, "status": 0.5833333333333334, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1208}, "5": {"facts_total": 59, "facts_retained": 30, "retention_rate": 0.5084745762711864, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.5714285714285714, "amount": 0.6, "status": 0.5833333333333334, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1208}, "6": {"facts_total": 59, "facts_retained": 30, "retention_rate": 0.5084745762711864, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.5714285714285714, "amount": 0.6, "status": 0.5833333333333334, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1208}, "7": {"facts_total": 59, "facts_retained": 30, "retention_rate": 0.5084745762711864, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.5714285714285714, "amount": 0.6, "status": 0.5833333333333334, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1208}, "8": {"facts_total": 59, "facts_retained": 30, "retention_rate": 0.5084745762711864, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.5714285714285714, "amount": 0.6, "status": 0.5833333333333334, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1208}}}, "reference_key_id": "keys/synthetic/syn_03012_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "667371c4cebf83517e9807a77df9aaa26b062652ac7a464bc497f8741c1303bf", "output_hash": "599bc22e373b1cf7b4679208eb9e515ad0e8dcb88dde0c7f579637ec7d3f3014"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "5fe7312fcf6429870364147f", "attempt_id": "ff666e560ae8", "run_group_id": "E4_core:A4|closed|qwen3b:h8", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:32:45.475600Z", "task_family": "synthetic", "document_id": "syn_03012_2000", "document_hash": "03e441e68d54e1af036f8ff905dfd6cbbccead62636279c1a8e179352ab08945", "question_id": "q03_exact_copy", "split": "test", "cluster_id": "syn_03012_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 8, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1927, "query_tokens": 594, "message_tokens": 511, "message_utf8_bytes": 1208, "cache_tensor_bytes": null, "serialized_payload_bytes": 1208, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 176655.71554307826, "hops": {"1": 22483.77687507309, "2": 22029.897166998126, "3": 21959.27741599735, "4": 22179.515584022738, "5": 22028.93558400683, "6": 21918.76791708637, "7": 22043.121207971126, "8": 22012.42379192263}, "receiver_prefill_ms": 412.3798339860514, "ttft_ms": 24.22070794273168, "decode_ms": 37.433624966070056}, "total_wall_ms": 177130.55058510508, "ttft_ms": 24.22070794273168, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 59, "facts_retained": 30, "retention_rate": 0.5084745762711864, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.5714285714285714, "amount": 0.6, "status": 0.5833333333333334, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1208}, "ledger_by_hop": {"1": {"facts_total": 59, "facts_retained": 30, "retention_rate": 0.5084745762711864, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.5714285714285714, "amount": 0.6, "status": 0.5833333333333334, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1213}, "2": {"facts_total": 59, "facts_retained": 30, "retention_rate": 0.5084745762711864, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.5714285714285714, "amount": 0.6, "status": 0.5833333333333334, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1208}, "3": {"facts_total": 59, "facts_retained": 30, "retention_rate": 0.5084745762711864, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.5714285714285714, "amount": 0.6, "status": 0.5833333333333334, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1208}, "4": {"facts_total": 59, "facts_retained": 30, "retention_rate": 0.5084745762711864, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.5714285714285714, "amount": 0.6, "status": 0.5833333333333334, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1208}, "5": {"facts_total": 59, "facts_retained": 30, "retention_rate": 0.5084745762711864, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.5714285714285714, "amount": 0.6, "status": 0.5833333333333334, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1208}, "6": {"facts_total": 59, "facts_retained": 30, "retention_rate": 0.5084745762711864, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.5714285714285714, "amount": 0.6, "status": 0.5833333333333334, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1208}, "7": {"facts_total": 59, "facts_retained": 30, "retention_rate": 0.5084745762711864, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.5714285714285714, "amount": 0.6, "status": 0.5833333333333334, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1208}, "8": {"facts_total": 59, "facts_retained": 30, "retention_rate": 0.5084745762711864, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.5714285714285714, "amount": 0.6, "status": 0.5833333333333334, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1208}}}, "reference_key_id": "keys/synthetic/syn_03012_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "667371c4cebf83517e9807a77df9aaa26b062652ac7a464bc497f8741c1303bf", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "37b2fbca71fd183abc9b72ef", "attempt_id": "b2448a706abf", "run_group_id": "E4_core:A4|closed|qwen3b:h8", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:32:45.968558Z", "task_family": "synthetic", "document_id": "syn_03012_2000", "document_hash": "03e441e68d54e1af036f8ff905dfd6cbbccead62636279c1a8e179352ab08945", "question_id": "q04_multi_hop", "split": "test", "cluster_id": "syn_03012_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 8, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1927, "query_tokens": 599, "message_tokens": 511, "message_utf8_bytes": 1208, "cache_tensor_bytes": null, "serialized_payload_bytes": 1208, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 176655.71554307826, "hops": {"1": 22483.77687507309, "2": 22029.897166998126, "3": 21959.27741599735, "4": 22179.515584022738, "5": 22028.93558400683, "6": 21918.76791708637, "7": 22043.121207971126, "8": 22012.42379192263}, "receiver_prefill_ms": 360.5628330260515, "ttft_ms": 36.772542051039636, "decode_ms": 475.7427910808474}, "total_wall_ms": 177529.48366804048, "ttft_ms": 36.772542051039636, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 7}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "8631552", "evidence_artifact": {"ledger": {"facts_total": 59, "facts_retained": 30, "retention_rate": 0.5084745762711864, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.5714285714285714, "amount": 0.6, "status": 0.5833333333333334, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1208}, "ledger_by_hop": {"1": {"facts_total": 59, "facts_retained": 30, "retention_rate": 0.5084745762711864, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.5714285714285714, "amount": 0.6, "status": 0.5833333333333334, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1213}, "2": {"facts_total": 59, "facts_retained": 30, "retention_rate": 0.5084745762711864, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.5714285714285714, "amount": 0.6, "status": 0.5833333333333334, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1208}, "3": {"facts_total": 59, "facts_retained": 30, "retention_rate": 0.5084745762711864, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.5714285714285714, "amount": 0.6, "status": 0.5833333333333334, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1208}, "4": {"facts_total": 59, "facts_retained": 30, "retention_rate": 0.5084745762711864, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.5714285714285714, "amount": 0.6, "status": 0.5833333333333334, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1208}, "5": {"facts_total": 59, "facts_retained": 30, "retention_rate": 0.5084745762711864, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.5714285714285714, "amount": 0.6, "status": 0.5833333333333334, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1208}, "6": {"facts_total": 59, "facts_retained": 30, "retention_rate": 0.5084745762711864, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.5714285714285714, "amount": 0.6, "status": 0.5833333333333334, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1208}, "7": {"facts_total": 59, "facts_retained": 30, "retention_rate": 0.5084745762711864, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.5714285714285714, "amount": 0.6, "status": 0.5833333333333334, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1208}, "8": {"facts_total": 59, "facts_retained": 30, "retention_rate": 0.5084745762711864, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.5714285714285714, "amount": 0.6, "status": 0.5833333333333334, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1208}}}, "reference_key_id": "keys/synthetic/syn_03012_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "667371c4cebf83517e9807a77df9aaa26b062652ac7a464bc497f8741c1303bf", "output_hash": "147d02c0587a77b3b53f0424c8d3d1454aaf2639ce9fdaf69af20db33363e849"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "fd9cfc0210b894774c6c4a80", "attempt_id": "7e78ce9de18c", "run_group_id": "E4_core:A4|closed|qwen3b:h8", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:32:46.860810Z", "task_family": "synthetic", "document_id": "syn_03012_2000", "document_hash": "03e441e68d54e1af036f8ff905dfd6cbbccead62636279c1a8e179352ab08945", "question_id": "q05_multi_hop", "split": "test", "cluster_id": "syn_03012_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 8, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1927, "query_tokens": 602, "message_tokens": 511, "message_utf8_bytes": 1208, "cache_tensor_bytes": null, "serialized_payload_bytes": 1208, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 176655.71554307826, "hops": {"1": 22483.77687507309, "2": 22029.897166998126, "3": 21959.27741599735, "4": 22179.515584022738, "5": 22028.93558400683, "6": 21918.76791708637, "7": 22043.121207971126, "8": 22012.42379192263}, "receiver_prefill_ms": 402.17824990395457, "ttft_ms": 39.6240419941023, "decode_ms": 312.9623339045793}, "total_wall_ms": 177411.12391802017, "ttft_ms": 39.6240419941023, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 10}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "DG-40E-2777", "evidence_artifact": {"ledger": {"facts_total": 59, "facts_retained": 30, "retention_rate": 0.5084745762711864, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.5714285714285714, "amount": 0.6, "status": 0.5833333333333334, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1208}, "ledger_by_hop": {"1": {"facts_total": 59, "facts_retained": 30, "retention_rate": 0.5084745762711864, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.5714285714285714, "amount": 0.6, "status": 0.5833333333333334, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1213}, "2": {"facts_total": 59, "facts_retained": 30, "retention_rate": 0.5084745762711864, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.5714285714285714, "amount": 0.6, "status": 0.5833333333333334, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1208}, "3": {"facts_total": 59, "facts_retained": 30, "retention_rate": 0.5084745762711864, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.5714285714285714, "amount": 0.6, "status": 0.5833333333333334, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1208}, "4": {"facts_total": 59, "facts_retained": 30, "retention_rate": 0.5084745762711864, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.5714285714285714, "amount": 0.6, "status": 0.5833333333333334, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1208}, "5": {"facts_total": 59, "facts_retained": 30, "retention_rate": 0.5084745762711864, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.5714285714285714, "amount": 0.6, "status": 0.5833333333333334, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1208}, "6": {"facts_total": 59, "facts_retained": 30, "retention_rate": 0.5084745762711864, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.5714285714285714, "amount": 0.6, "status": 0.5833333333333334, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1208}, "7": {"facts_total": 59, "facts_retained": 30, "retention_rate": 0.5084745762711864, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.5714285714285714, "amount": 0.6, "status": 0.5833333333333334, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1208}, "8": {"facts_total": 59, "facts_retained": 30, "retention_rate": 0.5084745762711864, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.5714285714285714, "amount": 0.6, "status": 0.5833333333333334, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1208}}}, "reference_key_id": "keys/synthetic/syn_03012_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "667371c4cebf83517e9807a77df9aaa26b062652ac7a464bc497f8741c1303bf", "output_hash": "16a7e3f4a7cc4a9b66f86c2cdcfad3b35daf1da199d9406e0627c8af2d8d73f3"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "d0f1b5a41dcd689d1cb7e9ce", "attempt_id": "5227e1f6cd90", "run_group_id": "E4_core:A4|closed|qwen3b:h8", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:32:47.633436Z", "task_family": "synthetic", "document_id": "syn_03012_2000", "document_hash": "03e441e68d54e1af036f8ff905dfd6cbbccead62636279c1a8e179352ab08945", "question_id": "q06_negation", "split": "test", "cluster_id": "syn_03012_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 8, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1927, "query_tokens": 594, "message_tokens": 511, "message_utf8_bytes": 1208, "cache_tensor_bytes": null, "serialized_payload_bytes": 1208, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 176655.71554307826, "hops": {"1": 22483.77687507309, "2": 22029.897166998126, "3": 21959.27741599735, "4": 22179.515584022738, "5": 22028.93558400683, "6": 21918.76791708637, "7": 22043.121207971126, "8": 22012.42379192263}, "receiver_prefill_ms": 355.2622910356149, "ttft_ms": 61.44758395384997, "decode_ms": 53.4423750359565}, "total_wall_ms": 177126.53500202578, "ttft_ms": 61.44758395384997, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "no", "evidence_artifact": {"ledger": {"facts_total": 59, "facts_retained": 30, "retention_rate": 0.5084745762711864, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.5714285714285714, "amount": 0.6, "status": 0.5833333333333334, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1208}, "ledger_by_hop": {"1": {"facts_total": 59, "facts_retained": 30, "retention_rate": 0.5084745762711864, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.5714285714285714, "amount": 0.6, "status": 0.5833333333333334, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1213}, "2": {"facts_total": 59, "facts_retained": 30, "retention_rate": 0.5084745762711864, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.5714285714285714, "amount": 0.6, "status": 0.5833333333333334, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1208}, "3": {"facts_total": 59, "facts_retained": 30, "retention_rate": 0.5084745762711864, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.5714285714285714, "amount": 0.6, "status": 0.5833333333333334, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1208}, "4": {"facts_total": 59, "facts_retained": 30, "retention_rate": 0.5084745762711864, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.5714285714285714, "amount": 0.6, "status": 0.5833333333333334, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1208}, "5": {"facts_total": 59, "facts_retained": 30, "retention_rate": 0.5084745762711864, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.5714285714285714, "amount": 0.6, "status": 0.5833333333333334, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1208}, "6": {"facts_total": 59, "facts_retained": 30, "retention_rate": 0.5084745762711864, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.5714285714285714, "amount": 0.6, "status": 0.5833333333333334, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1208}, "7": {"facts_total": 59, "facts_retained": 30, "retention_rate": 0.5084745762711864, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.5714285714285714, "amount": 0.6, "status": 0.5833333333333334, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1208}, "8": {"facts_total": 59, "facts_retained": 30, "retention_rate": 0.5084745762711864, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.5714285714285714, "amount": 0.6, "status": 0.5833333333333334, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1208}}}, "reference_key_id": "keys/synthetic/syn_03012_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "667371c4cebf83517e9807a77df9aaa26b062652ac7a464bc497f8741c1303bf", "output_hash": "9390298f3fb0c5b160498935d79cb139aef28e1c47358b4bbba61862b9c26e59"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "961d159d16c2b0e980d3fd0d", "attempt_id": "748f040f58fb", "run_group_id": "E4_core:A4|closed|qwen3b:h8", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:32:48.121930Z", "task_family": "synthetic", "document_id": "syn_03012_2000", "document_hash": "03e441e68d54e1af036f8ff905dfd6cbbccead62636279c1a8e179352ab08945", "question_id": "q07_negation", "split": "test", "cluster_id": "syn_03012_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 8, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1927, "query_tokens": 594, "message_tokens": 511, "message_utf8_bytes": 1208, "cache_tensor_bytes": null, "serialized_payload_bytes": 1208, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 176655.71554307826, "hops": {"1": 22483.77687507309, "2": 22029.897166998126, "3": 21959.27741599735, "4": 22179.515584022738, "5": 22028.93558400683, "6": 21918.76791708637, "7": 22043.121207971126, "8": 22012.42379192263}, "receiver_prefill_ms": 553.8036250509322, "ttft_ms": 22.415542043745518, "decode_ms": 40.688500041142106}, "total_wall_ms": 177273.51083501708, "ttft_ms": 22.415542043745518, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "no", "evidence_artifact": {"ledger": {"facts_total": 59, "facts_retained": 30, "retention_rate": 0.5084745762711864, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.5714285714285714, "amount": 0.6, "status": 0.5833333333333334, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1208}, "ledger_by_hop": {"1": {"facts_total": 59, "facts_retained": 30, "retention_rate": 0.5084745762711864, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.5714285714285714, "amount": 0.6, "status": 0.5833333333333334, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1213}, "2": {"facts_total": 59, "facts_retained": 30, "retention_rate": 0.5084745762711864, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.5714285714285714, "amount": 0.6, "status": 0.5833333333333334, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1208}, "3": {"facts_total": 59, "facts_retained": 30, "retention_rate": 0.5084745762711864, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.5714285714285714, "amount": 0.6, "status": 0.5833333333333334, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1208}, "4": {"facts_total": 59, "facts_retained": 30, "retention_rate": 0.5084745762711864, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.5714285714285714, "amount": 0.6, "status": 0.5833333333333334, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1208}, "5": {"facts_total": 59, "facts_retained": 30, "retention_rate": 0.5084745762711864, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.5714285714285714, "amount": 0.6, "status": 0.5833333333333334, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1208}, "6": {"facts_total": 59, "facts_retained": 30, "retention_rate": 0.5084745762711864, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.5714285714285714, "amount": 0.6, "status": 0.5833333333333334, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1208}, "7": {"facts_total": 59, "facts_retained": 30, "retention_rate": 0.5084745762711864, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.5714285714285714, "amount": 0.6, "status": 0.5833333333333334, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1208}, "8": {"facts_total": 59, "facts_retained": 30, "retention_rate": 0.5084745762711864, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.5714285714285714, "amount": 0.6, "status": 0.5833333333333334, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1208}}}, "reference_key_id": "keys/synthetic/syn_03012_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "667371c4cebf83517e9807a77df9aaa26b062652ac7a464bc497f8741c1303bf", "output_hash": "9390298f3fb0c5b160498935d79cb139aef28e1c47358b4bbba61862b9c26e59"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "7acd130ebc70a26bea827bad", "attempt_id": "3fb1b34b5eef", "run_group_id": "E4_core:A4|closed|qwen3b:h8", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:32:48.758281Z", "task_family": "synthetic", "document_id": "syn_03012_2000", "document_hash": "03e441e68d54e1af036f8ff905dfd6cbbccead62636279c1a8e179352ab08945", "question_id": "q08_revision", "split": "test", "cluster_id": "syn_03012_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 8, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1927, "query_tokens": 609, "message_tokens": 511, "message_utf8_bytes": 1208, "cache_tensor_bytes": null, "serialized_payload_bytes": 1208, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 176655.71554307826, "hops": {"1": 22483.77687507309, "2": 22029.897166998126, "3": 21959.27741599735, "4": 22179.515584022738, "5": 22028.93558400683, "6": 21918.76791708637, "7": 22043.121207971126, "8": 22012.42379192263}, "receiver_prefill_ms": 424.11345802247524, "ttft_ms": 40.98574991803616, "decode_ms": 206.32749993819743}, "total_wall_ms": 177327.78050098568, "ttft_ms": 40.98574991803616, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 7}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "6602245", "evidence_artifact": {"ledger": {"facts_total": 59, "facts_retained": 30, "retention_rate": 0.5084745762711864, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.5714285714285714, "amount": 0.6, "status": 0.5833333333333334, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1208}, "ledger_by_hop": {"1": {"facts_total": 59, "facts_retained": 30, "retention_rate": 0.5084745762711864, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.5714285714285714, "amount": 0.6, "status": 0.5833333333333334, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1213}, "2": {"facts_total": 59, "facts_retained": 30, "retention_rate": 0.5084745762711864, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.5714285714285714, "amount": 0.6, "status": 0.5833333333333334, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1208}, "3": {"facts_total": 59, "facts_retained": 30, "retention_rate": 0.5084745762711864, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.5714285714285714, "amount": 0.6, "status": 0.5833333333333334, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1208}, "4": {"facts_total": 59, "facts_retained": 30, "retention_rate": 0.5084745762711864, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.5714285714285714, "amount": 0.6, "status": 0.5833333333333334, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1208}, "5": {"facts_total": 59, "facts_retained": 30, "retention_rate": 0.5084745762711864, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.5714285714285714, "amount": 0.6, "status": 0.5833333333333334, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1208}, "6": {"facts_total": 59, "facts_retained": 30, "retention_rate": 0.5084745762711864, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.5714285714285714, "amount": 0.6, "status": 0.5833333333333334, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1208}, "7": {"facts_total": 59, "facts_retained": 30, "retention_rate": 0.5084745762711864, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.5714285714285714, "amount": 0.6, "status": 0.5833333333333334, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1208}, "8": {"facts_total": 59, "facts_retained": 30, "retention_rate": 0.5084745762711864, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.5714285714285714, "amount": 0.6, "status": 0.5833333333333334, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1208}}}, "reference_key_id": "keys/synthetic/syn_03012_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "667371c4cebf83517e9807a77df9aaa26b062652ac7a464bc497f8741c1303bf", "output_hash": "d117fa8197988f8d1e4fabc6c4cdaab9516018e54553b57c63b875b3de82b16f"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "c1aadbbff9ac817e8e54cdca", "attempt_id": "6c1edcbf4068", "run_group_id": "E4_core:A4|closed|qwen3b:h8", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:32:49.447086Z", "task_family": "synthetic", "document_id": "syn_03012_2000", "document_hash": "03e441e68d54e1af036f8ff905dfd6cbbccead62636279c1a8e179352ab08945", "question_id": "q09_revision", "split": "test", "cluster_id": "syn_03012_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 8, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1927, "query_tokens": 602, "message_tokens": 511, "message_utf8_bytes": 1208, "cache_tensor_bytes": null, "serialized_payload_bytes": 1208, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 176655.71554307826, "hops": {"1": 22483.77687507309, "2": 22029.897166998126, "3": 21959.27741599735, "4": 22179.515584022738, "5": 22028.93558400683, "6": 21918.76791708637, "7": 22043.121207971126, "8": 22012.42379192263}, "receiver_prefill_ms": 443.5735420556739, "ttft_ms": 51.8770000198856, "decode_ms": 393.85716593824327}, "total_wall_ms": 177546.0010850802, "ttft_ms": 51.8770000198856, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 7}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "5899340", "evidence_artifact": {"ledger": {"facts_total": 59, "facts_retained": 30, "retention_rate": 0.5084745762711864, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.5714285714285714, "amount": 0.6, "status": 0.5833333333333334, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1208}, "ledger_by_hop": {"1": {"facts_total": 59, "facts_retained": 30, "retention_rate": 0.5084745762711864, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.5714285714285714, "amount": 0.6, "status": 0.5833333333333334, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1213}, "2": {"facts_total": 59, "facts_retained": 30, "retention_rate": 0.5084745762711864, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.5714285714285714, "amount": 0.6, "status": 0.5833333333333334, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1208}, "3": {"facts_total": 59, "facts_retained": 30, "retention_rate": 0.5084745762711864, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.5714285714285714, "amount": 0.6, "status": 0.5833333333333334, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1208}, "4": {"facts_total": 59, "facts_retained": 30, "retention_rate": 0.5084745762711864, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.5714285714285714, "amount": 0.6, "status": 0.5833333333333334, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1208}, "5": {"facts_total": 59, "facts_retained": 30, "retention_rate": 0.5084745762711864, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.5714285714285714, "amount": 0.6, "status": 0.5833333333333334, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1208}, "6": {"facts_total": 59, "facts_retained": 30, "retention_rate": 0.5084745762711864, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.5714285714285714, "amount": 0.6, "status": 0.5833333333333334, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1208}, "7": {"facts_total": 59, "facts_retained": 30, "retention_rate": 0.5084745762711864, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.5714285714285714, "amount": 0.6, "status": 0.5833333333333334, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1208}, "8": {"facts_total": 59, "facts_retained": 30, "retention_rate": 0.5084745762711864, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.5714285714285714, "amount": 0.6, "status": 0.5833333333333334, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1208}}}, "reference_key_id": "keys/synthetic/syn_03012_2000.json", "task_scores": {"correct": 0.0, "abstained": 0.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "667371c4cebf83517e9807a77df9aaa26b062652ac7a464bc497f8741c1303bf", "output_hash": "c920323a9a02b1afb8619cf11bdde427f7cd1ea3289be69bd62932985895d717"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "b510593045790ddbe7a69d0f", "attempt_id": "d6ad8ff88238", "run_group_id": "E4_core:A4|closed|qwen3b:h8", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:32:50.356213Z", "task_family": "synthetic", "document_id": "syn_03012_2000", "document_hash": "03e441e68d54e1af036f8ff905dfd6cbbccead62636279c1a8e179352ab08945", "question_id": "q10_temporal", "split": "test", "cluster_id": "syn_03012_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 8, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1927, "query_tokens": 612, "message_tokens": 511, "message_utf8_bytes": 1208, "cache_tensor_bytes": null, "serialized_payload_bytes": 1208, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 176655.71554307826, "hops": {"1": 22483.77687507309, "2": 22029.897166998126, "3": 21959.27741599735, "4": 22179.515584022738, "5": 22028.93558400683, "6": 21918.76791708637, "7": 22043.121207971126, "8": 22012.42379192263}, "receiver_prefill_ms": 411.6696249693632, "ttft_ms": 27.697290992364287, "decode_ms": 36.98016598355025}, "total_wall_ms": 177132.73316808045, "ttft_ms": 27.697290992364287, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 59, "facts_retained": 30, "retention_rate": 0.5084745762711864, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.5714285714285714, "amount": 0.6, "status": 0.5833333333333334, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1208}, "ledger_by_hop": {"1": {"facts_total": 59, "facts_retained": 30, "retention_rate": 0.5084745762711864, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.5714285714285714, "amount": 0.6, "status": 0.5833333333333334, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1213}, "2": {"facts_total": 59, "facts_retained": 30, "retention_rate": 0.5084745762711864, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.5714285714285714, "amount": 0.6, "status": 0.5833333333333334, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1208}, "3": {"facts_total": 59, "facts_retained": 30, "retention_rate": 0.5084745762711864, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.5714285714285714, "amount": 0.6, "status": 0.5833333333333334, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1208}, "4": {"facts_total": 59, "facts_retained": 30, "retention_rate": 0.5084745762711864, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.5714285714285714, "amount": 0.6, "status": 0.5833333333333334, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1208}, "5": {"facts_total": 59, "facts_retained": 30, "retention_rate": 0.5084745762711864, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.5714285714285714, "amount": 0.6, "status": 0.5833333333333334, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1208}, "6": {"facts_total": 59, "facts_retained": 30, "retention_rate": 0.5084745762711864, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.5714285714285714, "amount": 0.6, "status": 0.5833333333333334, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1208}, "7": {"facts_total": 59, "facts_retained": 30, "retention_rate": 0.5084745762711864, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.5714285714285714, "amount": 0.6, "status": 0.5833333333333334, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1208}, "8": {"facts_total": 59, "facts_retained": 30, "retention_rate": 0.5084745762711864, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.5714285714285714, "amount": 0.6, "status": 0.5833333333333334, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1208}}}, "reference_key_id": "keys/synthetic/syn_03012_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "667371c4cebf83517e9807a77df9aaa26b062652ac7a464bc497f8741c1303bf", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "90626c84d52aa6f183ce277d", "attempt_id": "1fde426c3372", "run_group_id": "E4_core:A4|closed|qwen3b:h8", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:32:50.850719Z", "task_family": "synthetic", "document_id": "syn_03012_2000", "document_hash": "03e441e68d54e1af036f8ff905dfd6cbbccead62636279c1a8e179352ab08945", "question_id": "q11_aggregation", "split": "test", "cluster_id": "syn_03012_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 8, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1927, "query_tokens": 604, "message_tokens": 511, "message_utf8_bytes": 1208, "cache_tensor_bytes": null, "serialized_payload_bytes": 1208, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 176655.71554307826, "hops": {"1": 22483.77687507309, "2": 22029.897166998126, "3": 21959.27741599735, "4": 22179.515584022738, "5": 22028.93558400683, "6": 21918.76791708637, "7": 22043.121207971126, "8": 22012.42379192263}, "receiver_prefill_ms": 363.2097919471562, "ttft_ms": 46.87299998477101, "decode_ms": 542.0453329570591}, "total_wall_ms": 177608.58166811522, "ttft_ms": 46.87299998477101, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 8}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "19704407", "evidence_artifact": {"ledger": {"facts_total": 59, "facts_retained": 30, "retention_rate": 0.5084745762711864, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.5714285714285714, "amount": 0.6, "status": 0.5833333333333334, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1208}, "ledger_by_hop": {"1": {"facts_total": 59, "facts_retained": 30, "retention_rate": 0.5084745762711864, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.5714285714285714, "amount": 0.6, "status": 0.5833333333333334, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1213}, "2": {"facts_total": 59, "facts_retained": 30, "retention_rate": 0.5084745762711864, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.5714285714285714, "amount": 0.6, "status": 0.5833333333333334, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1208}, "3": {"facts_total": 59, "facts_retained": 30, "retention_rate": 0.5084745762711864, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.5714285714285714, "amount": 0.6, "status": 0.5833333333333334, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1208}, "4": {"facts_total": 59, "facts_retained": 30, "retention_rate": 0.5084745762711864, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.5714285714285714, "amount": 0.6, "status": 0.5833333333333334, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1208}, "5": {"facts_total": 59, "facts_retained": 30, "retention_rate": 0.5084745762711864, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.5714285714285714, "amount": 0.6, "status": 0.5833333333333334, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1208}, "6": {"facts_total": 59, "facts_retained": 30, "retention_rate": 0.5084745762711864, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.5714285714285714, "amount": 0.6, "status": 0.5833333333333334, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1208}, "7": {"facts_total": 59, "facts_retained": 30, "retention_rate": 0.5084745762711864, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.5714285714285714, "amount": 0.6, "status": 0.5833333333333334, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1208}, "8": {"facts_total": 59, "facts_retained": 30, "retention_rate": 0.5084745762711864, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.5714285714285714, "amount": 0.6, "status": 0.5833333333333334, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1208}}}, "reference_key_id": "keys/synthetic/syn_03012_2000.json", "task_scores": {"correct": 0.0, "abstained": 0.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "667371c4cebf83517e9807a77df9aaa26b062652ac7a464bc497f8741c1303bf", "output_hash": "7d2f092d592efff17462066772d9cbf34a039e2cf284b0bec7cabb81fc0421eb"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "9d3a21e81c346c4f8f297325", "attempt_id": "e91523e69fc3", "run_group_id": "E4_core:A4|closed|qwen3b:h8", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:32:51.821496Z", "task_family": "synthetic", "document_id": "syn_03012_2000", "document_hash": "03e441e68d54e1af036f8ff905dfd6cbbccead62636279c1a8e179352ab08945", "question_id": "q12_unit_conversion", "split": "test", "cluster_id": "syn_03012_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 8, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1927, "query_tokens": 601, "message_tokens": 511, "message_utf8_bytes": 1208, "cache_tensor_bytes": null, "serialized_payload_bytes": 1208, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 176655.71554307826, "hops": {"1": 22483.77687507309, "2": 22029.897166998126, "3": 21959.27741599735, "4": 22179.515584022738, "5": 22028.93558400683, "6": 21918.76791708637, "7": 22043.121207971126, "8": 22012.42379192263}, "receiver_prefill_ms": 414.53199996612966, "ttft_ms": 38.55083300732076, "decode_ms": 159.8737919703126}, "total_wall_ms": 177269.27491801325, "ttft_ms": 38.55083300732076, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 5}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "44620", "evidence_artifact": {"ledger": {"facts_total": 59, "facts_retained": 30, "retention_rate": 0.5084745762711864, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.5714285714285714, "amount": 0.6, "status": 0.5833333333333334, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1208}, "ledger_by_hop": {"1": {"facts_total": 59, "facts_retained": 30, "retention_rate": 0.5084745762711864, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.5714285714285714, "amount": 0.6, "status": 0.5833333333333334, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1213}, "2": {"facts_total": 59, "facts_retained": 30, "retention_rate": 0.5084745762711864, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.5714285714285714, "amount": 0.6, "status": 0.5833333333333334, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1208}, "3": {"facts_total": 59, "facts_retained": 30, "retention_rate": 0.5084745762711864, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.5714285714285714, "amount": 0.6, "status": 0.5833333333333334, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1208}, "4": {"facts_total": 59, "facts_retained": 30, "retention_rate": 0.5084745762711864, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.5714285714285714, "amount": 0.6, "status": 0.5833333333333334, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1208}, "5": {"facts_total": 59, "facts_retained": 30, "retention_rate": 0.5084745762711864, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.5714285714285714, "amount": 0.6, "status": 0.5833333333333334, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1208}, "6": {"facts_total": 59, "facts_retained": 30, "retention_rate": 0.5084745762711864, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.5714285714285714, "amount": 0.6, "status": 0.5833333333333334, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1208}, "7": {"facts_total": 59, "facts_retained": 30, "retention_rate": 0.5084745762711864, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.5714285714285714, "amount": 0.6, "status": 0.5833333333333334, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1208}, "8": {"facts_total": 59, "facts_retained": 30, "retention_rate": 0.5084745762711864, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.5714285714285714, "amount": 0.6, "status": 0.5833333333333334, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1208}}}, "reference_key_id": "keys/synthetic/syn_03012_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "667371c4cebf83517e9807a77df9aaa26b062652ac7a464bc497f8741c1303bf", "output_hash": "2bd125847727be7e7c0a656ee4c81f3bc6d0fef11f536f50ce7bc83688907075"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "e3f7b23860aa81255fa7f91a", "attempt_id": "23f76da60262", "run_group_id": "E4_core:A4|closed|qwen3b:h8", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:32:52.452594Z", "task_family": "synthetic", "document_id": "syn_03012_2000", "document_hash": "03e441e68d54e1af036f8ff905dfd6cbbccead62636279c1a8e179352ab08945", "question_id": "q13_unanswerable", "split": "test", "cluster_id": "syn_03012_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 8, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1927, "query_tokens": 606, "message_tokens": 511, "message_utf8_bytes": 1208, "cache_tensor_bytes": null, "serialized_payload_bytes": 1208, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 176655.71554307826, "hops": {"1": 22483.77687507309, "2": 22029.897166998126, "3": 21959.27741599735, "4": 22179.515584022738, "5": 22028.93558400683, "6": 21918.76791708637, "7": 22043.121207971126, "8": 22012.42379192263}, "receiver_prefill_ms": 355.17654207069427, "ttft_ms": 44.12700003013015, "decode_ms": 66.65779196191579}, "total_wall_ms": 177122.33512697276, "ttft_ms": 44.12700003013015, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 59, "facts_retained": 30, "retention_rate": 0.5084745762711864, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.5714285714285714, "amount": 0.6, "status": 0.5833333333333334, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1208}, "ledger_by_hop": {"1": {"facts_total": 59, "facts_retained": 30, "retention_rate": 0.5084745762711864, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.5714285714285714, "amount": 0.6, "status": 0.5833333333333334, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1213}, "2": {"facts_total": 59, "facts_retained": 30, "retention_rate": 0.5084745762711864, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.5714285714285714, "amount": 0.6, "status": 0.5833333333333334, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1208}, "3": {"facts_total": 59, "facts_retained": 30, "retention_rate": 0.5084745762711864, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.5714285714285714, "amount": 0.6, "status": 0.5833333333333334, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1208}, "4": {"facts_total": 59, "facts_retained": 30, "retention_rate": 0.5084745762711864, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.5714285714285714, "amount": 0.6, "status": 0.5833333333333334, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1208}, "5": {"facts_total": 59, "facts_retained": 30, "retention_rate": 0.5084745762711864, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.5714285714285714, "amount": 0.6, "status": 0.5833333333333334, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1208}, "6": {"facts_total": 59, "facts_retained": 30, "retention_rate": 0.5084745762711864, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.5714285714285714, "amount": 0.6, "status": 0.5833333333333334, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1208}, "7": {"facts_total": 59, "facts_retained": 30, "retention_rate": 0.5084745762711864, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.5714285714285714, "amount": 0.6, "status": 0.5833333333333334, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1208}, "8": {"facts_total": 59, "facts_retained": 30, "retention_rate": 0.5084745762711864, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.5714285714285714, "amount": 0.6, "status": 0.5833333333333334, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1208}}}, "reference_key_id": "keys/synthetic/syn_03012_2000.json", "task_scores": {"correct": 1.0, "abstained": 1.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "667371c4cebf83517e9807a77df9aaa26b062652ac7a464bc497f8741c1303bf", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "88e57fa3cb4476081267099e", "attempt_id": "5fd45114c7c5", "run_group_id": "E4_core:A4|closed|qwen3b:h8", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:32:52.935887Z", "task_family": "synthetic", "document_id": "syn_03012_2000", "document_hash": "03e441e68d54e1af036f8ff905dfd6cbbccead62636279c1a8e179352ab08945", "question_id": "q14_unanswerable", "split": "test", "cluster_id": "syn_03012_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 8, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1927, "query_tokens": 606, "message_tokens": 511, "message_utf8_bytes": 1208, "cache_tensor_bytes": null, "serialized_payload_bytes": 1208, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 176655.71554307826, "hops": {"1": 22483.77687507309, "2": 22029.897166998126, "3": 21959.27741599735, "4": 22179.515584022738, "5": 22028.93558400683, "6": 21918.76791708637, "7": 22043.121207971126, "8": 22012.42379192263}, "receiver_prefill_ms": 568.0283330148086, "ttft_ms": 24.630333064123988, "decode_ms": 38.30879193264991}, "total_wall_ms": 177287.40695910528, "ttft_ms": 24.630333064123988, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 59, "facts_retained": 30, "retention_rate": 0.5084745762711864, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.5714285714285714, "amount": 0.6, "status": 0.5833333333333334, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1208}, "ledger_by_hop": {"1": {"facts_total": 59, "facts_retained": 30, "retention_rate": 0.5084745762711864, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.5714285714285714, "amount": 0.6, "status": 0.5833333333333334, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1213}, "2": {"facts_total": 59, "facts_retained": 30, "retention_rate": 0.5084745762711864, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.5714285714285714, "amount": 0.6, "status": 0.5833333333333334, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1208}, "3": {"facts_total": 59, "facts_retained": 30, "retention_rate": 0.5084745762711864, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.5714285714285714, "amount": 0.6, "status": 0.5833333333333334, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1208}, "4": {"facts_total": 59, "facts_retained": 30, "retention_rate": 0.5084745762711864, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.5714285714285714, "amount": 0.6, "status": 0.5833333333333334, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1208}, "5": {"facts_total": 59, "facts_retained": 30, "retention_rate": 0.5084745762711864, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.5714285714285714, "amount": 0.6, "status": 0.5833333333333334, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1208}, "6": {"facts_total": 59, "facts_retained": 30, "retention_rate": 0.5084745762711864, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.5714285714285714, "amount": 0.6, "status": 0.5833333333333334, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1208}, "7": {"facts_total": 59, "facts_retained": 30, "retention_rate": 0.5084745762711864, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.5714285714285714, "amount": 0.6, "status": 0.5833333333333334, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1208}, "8": {"facts_total": 59, "facts_retained": 30, "retention_rate": 0.5084745762711864, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.5714285714285714, "amount": 0.6, "status": 0.5833333333333334, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1208}}}, "reference_key_id": "keys/synthetic/syn_03012_2000.json", "task_scores": {"correct": 1.0, "abstained": 1.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "667371c4cebf83517e9807a77df9aaa26b062652ac7a464bc497f8741c1303bf", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "0973bfca89eac4274bdd75dc", "attempt_id": "e7d8187ae24a", "run_group_id": "E4_core:A5|closed|qwen3b:h1", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:33:16.054606Z", "task_family": "synthetic", "document_id": "syn_03012_2000", "document_hash": "03e441e68d54e1af036f8ff905dfd6cbbccead62636279c1a8e179352ab08945", "question_id": "q00_single_fact", "split": "test", "cluster_id": "syn_03012_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 1, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1927, "query_tokens": 431, "message_tokens": 348, "message_utf8_bytes": 824, "cache_tensor_bytes": null, "serialized_payload_bytes": 824, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 22467.89562504273, "hops": {"1": 22467.89562504273}, "receiver_prefill_ms": 306.2139160465449, "ttft_ms": 37.2890830039978, "decode_ms": 420.20674992818385}, "total_wall_ms": 23232.50479111448, "ttft_ms": 37.2890830039978, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 10}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "XX-71M-6213", "evidence_artifact": {"ledger": {"facts_total": 59, "facts_retained": 11, "retention_rate": 0.1864406779661017, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.14285714285714285, "amount": 0.0, "status": 0.4166666666666667, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 760}, "ledger_by_hop": {"1": {"facts_total": 59, "facts_retained": 11, "retention_rate": 0.1864406779661017, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.14285714285714285, "amount": 0.0, "status": 0.4166666666666667, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 760}}}, "reference_key_id": "keys/synthetic/syn_03012_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "bb16af6fe25c6b4abc9bdb0ad8c8b3d8b7d3dc12769f5feefdb7a2e4bb915494", "output_hash": "66ef6c5da682b38795e9029723bbd71cdde2c4dec70cf1fae6efe62ef243f10a"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "433057a6b68d8a351a40fc08", "attempt_id": "058d7c8151e5", "run_group_id": "E4_core:A5|closed|qwen3b:h1", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:33:16.836645Z", "task_family": "synthetic", "document_id": "syn_03012_2000", "document_hash": "03e441e68d54e1af036f8ff905dfd6cbbccead62636279c1a8e179352ab08945", "question_id": "q01_single_fact", "split": "test", "cluster_id": "syn_03012_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 1, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1927, "query_tokens": 437, "message_tokens": 348, "message_utf8_bytes": 824, "cache_tensor_bytes": null, "serialized_payload_bytes": 824, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 22467.89562504273, "hops": {"1": 22467.89562504273}, "receiver_prefill_ms": 443.404582911171, "ttft_ms": 24.732875055633485, "decode_ms": 42.08766599185765}, "total_wall_ms": 22978.780167060904, "ttft_ms": 24.732875055633485, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 59, "facts_retained": 11, "retention_rate": 0.1864406779661017, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.14285714285714285, "amount": 0.0, "status": 0.4166666666666667, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 760}, "ledger_by_hop": {"1": {"facts_total": 59, "facts_retained": 11, "retention_rate": 0.1864406779661017, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.14285714285714285, "amount": 0.0, "status": 0.4166666666666667, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 760}}}, "reference_key_id": "keys/synthetic/syn_03012_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "bb16af6fe25c6b4abc9bdb0ad8c8b3d8b7d3dc12769f5feefdb7a2e4bb915494", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "1fe4626636f3b3b3106c1fc4", "attempt_id": "bd128ee015d9", "run_group_id": "E4_core:A5|closed|qwen3b:h1", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:33:17.365376Z", "task_family": "synthetic", "document_id": "syn_03012_2000", "document_hash": "03e441e68d54e1af036f8ff905dfd6cbbccead62636279c1a8e179352ab08945", "question_id": "q02_exact_copy", "split": "test", "cluster_id": "syn_03012_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 1, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1927, "query_tokens": 436, "message_tokens": 348, "message_utf8_bytes": 824, "cache_tensor_bytes": null, "serialized_payload_bytes": 824, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 22467.89562504273, "hops": {"1": 22467.89562504273}, "receiver_prefill_ms": 288.32116595003754, "ttft_ms": 34.27337505854666, "decode_ms": 38.329250062815845}, "total_wall_ms": 22829.329000087455, "ttft_ms": 34.27337505854666, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 59, "facts_retained": 11, "retention_rate": 0.1864406779661017, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.14285714285714285, "amount": 0.0, "status": 0.4166666666666667, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 760}, "ledger_by_hop": {"1": {"facts_total": 59, "facts_retained": 11, "retention_rate": 0.1864406779661017, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.14285714285714285, "amount": 0.0, "status": 0.4166666666666667, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 760}}}, "reference_key_id": "keys/synthetic/syn_03012_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "bb16af6fe25c6b4abc9bdb0ad8c8b3d8b7d3dc12769f5feefdb7a2e4bb915494", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "fe46e169af7ba5f41f9fa7a0", "attempt_id": "234cf5491abc", "run_group_id": "E4_core:A5|closed|qwen3b:h1", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:33:17.744127Z", "task_family": "synthetic", "document_id": "syn_03012_2000", "document_hash": "03e441e68d54e1af036f8ff905dfd6cbbccead62636279c1a8e179352ab08945", "question_id": "q03_exact_copy", "split": "test", "cluster_id": "syn_03012_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 1, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1927, "query_tokens": 432, "message_tokens": 348, "message_utf8_bytes": 824, "cache_tensor_bytes": null, "serialized_payload_bytes": 824, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 22467.89562504273, "hops": {"1": 22467.89562504273}, "receiver_prefill_ms": 286.69816709589213, "ttft_ms": 40.23404105100781, "decode_ms": 23.06995796971023}, "total_wall_ms": 22818.668208085, "ttft_ms": 40.23404105100781, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 59, "facts_retained": 11, "retention_rate": 0.1864406779661017, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.14285714285714285, "amount": 0.0, "status": 0.4166666666666667, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 760}, "ledger_by_hop": {"1": {"facts_total": 59, "facts_retained": 11, "retention_rate": 0.1864406779661017, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.14285714285714285, "amount": 0.0, "status": 0.4166666666666667, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 760}}}, "reference_key_id": "keys/synthetic/syn_03012_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "bb16af6fe25c6b4abc9bdb0ad8c8b3d8b7d3dc12769f5feefdb7a2e4bb915494", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "aed0cfc7d3db126626185567", "attempt_id": "eca64d8aa3fa", "run_group_id": "E4_core:A5|closed|qwen3b:h1", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:33:18.112492Z", "task_family": "synthetic", "document_id": "syn_03012_2000", "document_hash": "03e441e68d54e1af036f8ff905dfd6cbbccead62636279c1a8e179352ab08945", "question_id": "q04_multi_hop", "split": "test", "cluster_id": "syn_03012_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 1, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1927, "query_tokens": 437, "message_tokens": 348, "message_utf8_bytes": 824, "cache_tensor_bytes": null, "serialized_payload_bytes": 824, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 22467.89562504273, "hops": {"1": 22467.89562504273}, "receiver_prefill_ms": 279.82704201713204, "ttft_ms": 57.400833000428975, "decode_ms": 44.462916906923056}, "total_wall_ms": 22850.160208996385, "ttft_ms": 57.400833000428975, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 59, "facts_retained": 11, "retention_rate": 0.1864406779661017, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.14285714285714285, "amount": 0.0, "status": 0.4166666666666667, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 760}, "ledger_by_hop": {"1": {"facts_total": 59, "facts_retained": 11, "retention_rate": 0.1864406779661017, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.14285714285714285, "amount": 0.0, "status": 0.4166666666666667, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 760}}}, "reference_key_id": "keys/synthetic/syn_03012_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "bb16af6fe25c6b4abc9bdb0ad8c8b3d8b7d3dc12769f5feefdb7a2e4bb915494", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "792dc87f8d5a7cf25bfb820d", "attempt_id": "c0e32251722d", "run_group_id": "E4_core:A5|closed|qwen3b:h1", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:33:18.514415Z", "task_family": "synthetic", "document_id": "syn_03012_2000", "document_hash": "03e441e68d54e1af036f8ff905dfd6cbbccead62636279c1a8e179352ab08945", "question_id": "q05_multi_hop", "split": "test", "cluster_id": "syn_03012_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 1, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1927, "query_tokens": 440, "message_tokens": 348, "message_utf8_bytes": 824, "cache_tensor_bytes": null, "serialized_payload_bytes": 824, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 22467.89562504273, "hops": {"1": 22467.89562504273}, "receiver_prefill_ms": 426.23933299910277, "ttft_ms": 30.835125013254583, "decode_ms": 40.34616705030203}, "total_wall_ms": 22965.979458997026, "ttft_ms": 30.835125013254583, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 59, "facts_retained": 11, "retention_rate": 0.1864406779661017, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.14285714285714285, "amount": 0.0, "status": 0.4166666666666667, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 760}, "ledger_by_hop": {"1": {"facts_total": 59, "facts_retained": 11, "retention_rate": 0.1864406779661017, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.14285714285714285, "amount": 0.0, "status": 0.4166666666666667, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 760}}}, "reference_key_id": "keys/synthetic/syn_03012_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "bb16af6fe25c6b4abc9bdb0ad8c8b3d8b7d3dc12769f5feefdb7a2e4bb915494", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "97c6085ff30cc51ea318d5f8", "attempt_id": "4191f4edf520", "run_group_id": "E4_core:A5|closed|qwen3b:h1", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:33:19.030294Z", "task_family": "synthetic", "document_id": "syn_03012_2000", "document_hash": "03e441e68d54e1af036f8ff905dfd6cbbccead62636279c1a8e179352ab08945", "question_id": "q06_negation", "split": "test", "cluster_id": "syn_03012_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 1, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1927, "query_tokens": 432, "message_tokens": 348, "message_utf8_bytes": 824, "cache_tensor_bytes": null, "serialized_payload_bytes": 824, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 22467.89562504273, "hops": {"1": 22467.89562504273}, "receiver_prefill_ms": 302.5542909745127, "ttft_ms": 28.46020890865475, "decode_ms": 46.250290935859084}, "total_wall_ms": 22845.60079104267, "ttft_ms": 28.46020890865475, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 59, "facts_retained": 11, "retention_rate": 0.1864406779661017, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.14285714285714285, "amount": 0.0, "status": 0.4166666666666667, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 760}, "ledger_by_hop": {"1": {"facts_total": 59, "facts_retained": 11, "retention_rate": 0.1864406779661017, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.14285714285714285, "amount": 0.0, "status": 0.4166666666666667, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 760}}}, "reference_key_id": "keys/synthetic/syn_03012_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "bb16af6fe25c6b4abc9bdb0ad8c8b3d8b7d3dc12769f5feefdb7a2e4bb915494", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "d93754bfaa9cf8e156cfdf15", "attempt_id": "d7c6313a6a72", "run_group_id": "E4_core:A5|closed|qwen3b:h1", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:33:19.425489Z", "task_family": "synthetic", "document_id": "syn_03012_2000", "document_hash": "03e441e68d54e1af036f8ff905dfd6cbbccead62636279c1a8e179352ab08945", "question_id": "q07_negation", "split": "test", "cluster_id": "syn_03012_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 1, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1927, "query_tokens": 432, "message_tokens": 348, "message_utf8_bytes": 824, "cache_tensor_bytes": null, "serialized_payload_bytes": 824, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 22467.89562504273, "hops": {"1": 22467.89562504273}, "receiver_prefill_ms": 303.59050002880394, "ttft_ms": 20.240167039446533, "decode_ms": 19.776250002905726}, "total_wall_ms": 22811.928000068292, "ttft_ms": 20.240167039446533, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 59, "facts_retained": 11, "retention_rate": 0.1864406779661017, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.14285714285714285, "amount": 0.0, "status": 0.4166666666666667, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 760}, "ledger_by_hop": {"1": {"facts_total": 59, "facts_retained": 11, "retention_rate": 0.1864406779661017, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.14285714285714285, "amount": 0.0, "status": 0.4166666666666667, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 760}}}, "reference_key_id": "keys/synthetic/syn_03012_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "bb16af6fe25c6b4abc9bdb0ad8c8b3d8b7d3dc12769f5feefdb7a2e4bb915494", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "11c0dc4736cb663d4c0de57f", "attempt_id": "628ae27ce1b6", "run_group_id": "E4_core:A5|closed|qwen3b:h1", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:33:19.786055Z", "task_family": "synthetic", "document_id": "syn_03012_2000", "document_hash": "03e441e68d54e1af036f8ff905dfd6cbbccead62636279c1a8e179352ab08945", "question_id": "q08_revision", "split": "test", "cluster_id": "syn_03012_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 1, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1927, "query_tokens": 447, "message_tokens": 348, "message_utf8_bytes": 824, "cache_tensor_bytes": null, "serialized_payload_bytes": 824, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 22467.89562504273, "hops": {"1": 22467.89562504273}, "receiver_prefill_ms": 278.984708013013, "ttft_ms": 58.16033307928592, "decode_ms": 64.02074999641627}, "total_wall_ms": 22869.762459071353, "ttft_ms": 58.16033307928592, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 59, "facts_retained": 11, "retention_rate": 0.1864406779661017, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.14285714285714285, "amount": 0.0, "status": 0.4166666666666667, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 760}, "ledger_by_hop": {"1": {"facts_total": 59, "facts_retained": 11, "retention_rate": 0.1864406779661017, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.14285714285714285, "amount": 0.0, "status": 0.4166666666666667, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 760}}}, "reference_key_id": "keys/synthetic/syn_03012_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "bb16af6fe25c6b4abc9bdb0ad8c8b3d8b7d3dc12769f5feefdb7a2e4bb915494", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "b5bff60d9ad78e50aef755d6", "attempt_id": "e06ded5180ba", "run_group_id": "E4_core:A5|closed|qwen3b:h1", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:33:20.207801Z", "task_family": "synthetic", "document_id": "syn_03012_2000", "document_hash": "03e441e68d54e1af036f8ff905dfd6cbbccead62636279c1a8e179352ab08945", "question_id": "q09_revision", "split": "test", "cluster_id": "syn_03012_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 1, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1927, "query_tokens": 440, "message_tokens": 348, "message_utf8_bytes": 824, "cache_tensor_bytes": null, "serialized_payload_bytes": 824, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 22467.89562504273, "hops": {"1": 22467.89562504273}, "receiver_prefill_ms": 423.2864590594545, "ttft_ms": 43.966082972474396, "decode_ms": 37.52108302433044}, "total_wall_ms": 22973.177542095073, "ttft_ms": 43.966082972474396, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 59, "facts_retained": 11, "retention_rate": 0.1864406779661017, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.14285714285714285, "amount": 0.0, "status": 0.4166666666666667, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 760}, "ledger_by_hop": {"1": {"facts_total": 59, "facts_retained": 11, "retention_rate": 0.1864406779661017, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.14285714285714285, "amount": 0.0, "status": 0.4166666666666667, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 760}}}, "reference_key_id": "keys/synthetic/syn_03012_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "bb16af6fe25c6b4abc9bdb0ad8c8b3d8b7d3dc12769f5feefdb7a2e4bb915494", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "29253351583ee3d06c13d696", "attempt_id": "a1bad232ad66", "run_group_id": "E4_core:A5|closed|qwen3b:h1", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:33:20.731079Z", "task_family": "synthetic", "document_id": "syn_03012_2000", "document_hash": "03e441e68d54e1af036f8ff905dfd6cbbccead62636279c1a8e179352ab08945", "question_id": "q10_temporal", "split": "test", "cluster_id": "syn_03012_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 1, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1927, "query_tokens": 450, "message_tokens": 348, "message_utf8_bytes": 824, "cache_tensor_bytes": null, "serialized_payload_bytes": 824, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 22467.89562504273, "hops": {"1": 22467.89562504273}, "receiver_prefill_ms": 314.06629201956093, "ttft_ms": 42.63658297713846, "decode_ms": 29.039749992080033}, "total_wall_ms": 22854.246417060494, "ttft_ms": 42.63658297713846, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 59, "facts_retained": 11, "retention_rate": 0.1864406779661017, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.14285714285714285, "amount": 0.0, "status": 0.4166666666666667, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 760}, "ledger_by_hop": {"1": {"facts_total": 59, "facts_retained": 11, "retention_rate": 0.1864406779661017, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.14285714285714285, "amount": 0.0, "status": 0.4166666666666667, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 760}}}, "reference_key_id": "keys/synthetic/syn_03012_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "bb16af6fe25c6b4abc9bdb0ad8c8b3d8b7d3dc12769f5feefdb7a2e4bb915494", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "4e22f3d8b3bf12020e7de2d9", "attempt_id": "9ba0d5848ba1", "run_group_id": "E4_core:A5|closed|qwen3b:h1", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:33:21.134190Z", "task_family": "synthetic", "document_id": "syn_03012_2000", "document_hash": "03e441e68d54e1af036f8ff905dfd6cbbccead62636279c1a8e179352ab08945", "question_id": "q11_aggregation", "split": "test", "cluster_id": "syn_03012_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 1, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1927, "query_tokens": 442, "message_tokens": 348, "message_utf8_bytes": 824, "cache_tensor_bytes": null, "serialized_payload_bytes": 824, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 22467.89562504273, "hops": {"1": 22467.89562504273}, "receiver_prefill_ms": 284.97445792891085, "ttft_ms": 20.236333017237484, "decode_ms": 19.697667099535465}, "total_wall_ms": 22793.29908301588, "ttft_ms": 20.236333017237484, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 59, "facts_retained": 11, "retention_rate": 0.1864406779661017, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.14285714285714285, "amount": 0.0, "status": 0.4166666666666667, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 760}, "ledger_by_hop": {"1": {"facts_total": 59, "facts_retained": 11, "retention_rate": 0.1864406779661017, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.14285714285714285, "amount": 0.0, "status": 0.4166666666666667, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 760}}}, "reference_key_id": "keys/synthetic/syn_03012_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "bb16af6fe25c6b4abc9bdb0ad8c8b3d8b7d3dc12769f5feefdb7a2e4bb915494", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "fedf596ffc03914ded22530b", "attempt_id": "5101702606f0", "run_group_id": "E4_core:A5|closed|qwen3b:h1", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:33:21.475943Z", "task_family": "synthetic", "document_id": "syn_03012_2000", "document_hash": "03e441e68d54e1af036f8ff905dfd6cbbccead62636279c1a8e179352ab08945", "question_id": "q12_unit_conversion", "split": "test", "cluster_id": "syn_03012_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 1, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1927, "query_tokens": 439, "message_tokens": 348, "message_utf8_bytes": 824, "cache_tensor_bytes": null, "serialized_payload_bytes": 824, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 22467.89562504273, "hops": {"1": 22467.89562504273}, "receiver_prefill_ms": 279.7913750400767, "ttft_ms": 65.32595795579255, "decode_ms": 377.98320793081075}, "total_wall_ms": 23191.581374965608, "ttft_ms": 65.32595795579255, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 5}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "44620", "evidence_artifact": {"ledger": {"facts_total": 59, "facts_retained": 11, "retention_rate": 0.1864406779661017, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.14285714285714285, "amount": 0.0, "status": 0.4166666666666667, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 760}, "ledger_by_hop": {"1": {"facts_total": 59, "facts_retained": 11, "retention_rate": 0.1864406779661017, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.14285714285714285, "amount": 0.0, "status": 0.4166666666666667, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 760}}}, "reference_key_id": "keys/synthetic/syn_03012_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "bb16af6fe25c6b4abc9bdb0ad8c8b3d8b7d3dc12769f5feefdb7a2e4bb915494", "output_hash": "2bd125847727be7e7c0a656ee4c81f3bc6d0fef11f536f50ce7bc83688907075"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "81c1bbb3e9ddc072fa11dfc1", "attempt_id": "fb149dcf8909", "run_group_id": "E4_core:A5|closed|qwen3b:h1", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:33:22.219615Z", "task_family": "synthetic", "document_id": "syn_03012_2000", "document_hash": "03e441e68d54e1af036f8ff905dfd6cbbccead62636279c1a8e179352ab08945", "question_id": "q13_unanswerable", "split": "test", "cluster_id": "syn_03012_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 1, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1927, "query_tokens": 444, "message_tokens": 348, "message_utf8_bytes": 824, "cache_tensor_bytes": null, "serialized_payload_bytes": 824, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 22467.89562504273, "hops": {"1": 22467.89562504273}, "receiver_prefill_ms": 289.79425004217774, "ttft_ms": 42.13333292864263, "decode_ms": 34.7881669877097}, "total_wall_ms": 22835.19700006582, "ttft_ms": 42.13333292864263, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 59, "facts_retained": 11, "retention_rate": 0.1864406779661017, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.14285714285714285, "amount": 0.0, "status": 0.4166666666666667, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 760}, "ledger_by_hop": {"1": {"facts_total": 59, "facts_retained": 11, "retention_rate": 0.1864406779661017, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.14285714285714285, "amount": 0.0, "status": 0.4166666666666667, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 760}}}, "reference_key_id": "keys/synthetic/syn_03012_2000.json", "task_scores": {"correct": 1.0, "abstained": 1.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "bb16af6fe25c6b4abc9bdb0ad8c8b3d8b7d3dc12769f5feefdb7a2e4bb915494", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "211fdb95f1a8249d098f3ec3", "attempt_id": "475267a4e7be", "run_group_id": "E4_core:A5|closed|qwen3b:h1", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:33:22.604409Z", "task_family": "synthetic", "document_id": "syn_03012_2000", "document_hash": "03e441e68d54e1af036f8ff905dfd6cbbccead62636279c1a8e179352ab08945", "question_id": "q14_unanswerable", "split": "test", "cluster_id": "syn_03012_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 1, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1927, "query_tokens": 444, "message_tokens": 348, "message_utf8_bytes": 824, "cache_tensor_bytes": null, "serialized_payload_bytes": 824, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 22467.89562504273, "hops": {"1": 22467.89562504273}, "receiver_prefill_ms": 285.4959589894861, "ttft_ms": 33.239000011235476, "decode_ms": 33.38954201899469}, "total_wall_ms": 22820.60179207474, "ttft_ms": 33.239000011235476, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 59, "facts_retained": 11, "retention_rate": 0.1864406779661017, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.14285714285714285, "amount": 0.0, "status": 0.4166666666666667, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 760}, "ledger_by_hop": {"1": {"facts_total": 59, "facts_retained": 11, "retention_rate": 0.1864406779661017, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.14285714285714285, "amount": 0.0, "status": 0.4166666666666667, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 760}}}, "reference_key_id": "keys/synthetic/syn_03012_2000.json", "task_scores": {"correct": 1.0, "abstained": 1.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "bb16af6fe25c6b4abc9bdb0ad8c8b3d8b7d3dc12769f5feefdb7a2e4bb915494", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "601f89d12fbceba7453381be", "attempt_id": "12443643ae6e", "run_group_id": "E4_core:A5|closed|qwen3b:h2", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:33:38.694746Z", "task_family": "synthetic", "document_id": "syn_03012_2000", "document_hash": "03e441e68d54e1af036f8ff905dfd6cbbccead62636279c1a8e179352ab08945", "question_id": "q00_single_fact", "split": "test", "cluster_id": "syn_03012_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 2, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1927, "query_tokens": 431, "message_tokens": 348, "message_utf8_bytes": 824, "cache_tensor_bytes": null, "serialized_payload_bytes": 824, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 38187.22012510989, "hops": {"1": 22467.89562504273, "2": 15719.32450006716}, "receiver_prefill_ms": 305.89866591617465, "ttft_ms": 33.77412504050881, "decode_ms": 321.3921249844134}, "total_wall_ms": 38848.893833113834, "ttft_ms": 33.77412504050881, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 10}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "XX-71M-6213", "evidence_artifact": {"ledger": {"facts_total": 59, "facts_retained": 11, "retention_rate": 0.1864406779661017, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.14285714285714285, "amount": 0.0, "status": 0.4166666666666667, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 760}, "ledger_by_hop": {"1": {"facts_total": 59, "facts_retained": 11, "retention_rate": 0.1864406779661017, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.14285714285714285, "amount": 0.0, "status": 0.4166666666666667, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 760}, "2": {"facts_total": 59, "facts_retained": 11, "retention_rate": 0.1864406779661017, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.14285714285714285, "amount": 0.0, "status": 0.4166666666666667, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 760}}}, "reference_key_id": "keys/synthetic/syn_03012_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "bb16af6fe25c6b4abc9bdb0ad8c8b3d8b7d3dc12769f5feefdb7a2e4bb915494", "output_hash": "66ef6c5da682b38795e9029723bbd71cdde2c4dec70cf1fae6efe62ef243f10a"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "9f8a9a3d55fad673431680c2", "attempt_id": "4c7621475de1", "run_group_id": "E4_core:A5|closed|qwen3b:h2", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:33:39.374183Z", "task_family": "synthetic", "document_id": "syn_03012_2000", "document_hash": "03e441e68d54e1af036f8ff905dfd6cbbccead62636279c1a8e179352ab08945", "question_id": "q01_single_fact", "split": "test", "cluster_id": "syn_03012_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 2, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1927, "query_tokens": 437, "message_tokens": 348, "message_utf8_bytes": 824, "cache_tensor_bytes": null, "serialized_payload_bytes": 824, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 38187.22012510989, "hops": {"1": 22467.89562504273, "2": 15719.32450006716}, "receiver_prefill_ms": 257.95254192780703, "ttft_ms": 60.78241695649922, "decode_ms": 56.785875000059605}, "total_wall_ms": 38563.36841720622, "ttft_ms": 60.78241695649922, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 59, "facts_retained": 11, "retention_rate": 0.1864406779661017, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.14285714285714285, "amount": 0.0, "status": 0.4166666666666667, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 760}, "ledger_by_hop": {"1": {"facts_total": 59, "facts_retained": 11, "retention_rate": 0.1864406779661017, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.14285714285714285, "amount": 0.0, "status": 0.4166666666666667, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 760}, "2": {"facts_total": 59, "facts_retained": 11, "retention_rate": 0.1864406779661017, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.14285714285714285, "amount": 0.0, "status": 0.4166666666666667, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 760}}}, "reference_key_id": "keys/synthetic/syn_03012_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "bb16af6fe25c6b4abc9bdb0ad8c8b3d8b7d3dc12769f5feefdb7a2e4bb915494", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "a8d08f737e89c9c4fa7b1371", "attempt_id": "2673b1a1ddb0", "run_group_id": "E4_core:A5|closed|qwen3b:h2", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:33:39.768548Z", "task_family": "synthetic", "document_id": "syn_03012_2000", "document_hash": "03e441e68d54e1af036f8ff905dfd6cbbccead62636279c1a8e179352ab08945", "question_id": "q02_exact_copy", "split": "test", "cluster_id": "syn_03012_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 2, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1927, "query_tokens": 436, "message_tokens": 348, "message_utf8_bytes": 824, "cache_tensor_bytes": null, "serialized_payload_bytes": 824, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 38187.22012510989, "hops": {"1": 22467.89562504273, "2": 15719.32450006716}, "receiver_prefill_ms": 480.91620800551027, "ttft_ms": 32.44191699195653, "decode_ms": 41.228124988265336}, "total_wall_ms": 38742.56033310667, "ttft_ms": 32.44191699195653, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 59, "facts_retained": 11, "retention_rate": 0.1864406779661017, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.14285714285714285, "amount": 0.0, "status": 0.4166666666666667, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 760}, "ledger_by_hop": {"1": {"facts_total": 59, "facts_retained": 11, "retention_rate": 0.1864406779661017, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.14285714285714285, "amount": 0.0, "status": 0.4166666666666667, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 760}, "2": {"facts_total": 59, "facts_retained": 11, "retention_rate": 0.1864406779661017, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.14285714285714285, "amount": 0.0, "status": 0.4166666666666667, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 760}}}, "reference_key_id": "keys/synthetic/syn_03012_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "bb16af6fe25c6b4abc9bdb0ad8c8b3d8b7d3dc12769f5feefdb7a2e4bb915494", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "a222d77c15c137c0469fcfa0", "attempt_id": "a8012dfc9c2c", "run_group_id": "E4_core:A5|closed|qwen3b:h2", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:33:40.341951Z", "task_family": "synthetic", "document_id": "syn_03012_2000", "document_hash": "03e441e68d54e1af036f8ff905dfd6cbbccead62636279c1a8e179352ab08945", "question_id": "q03_exact_copy", "split": "test", "cluster_id": "syn_03012_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 2, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1927, "query_tokens": 432, "message_tokens": 348, "message_utf8_bytes": 824, "cache_tensor_bytes": null, "serialized_payload_bytes": 824, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 38187.22012510989, "hops": {"1": 22467.89562504273, "2": 15719.32450006716}, "receiver_prefill_ms": 286.8312910431996, "ttft_ms": 29.62608402594924, "decode_ms": 43.79891592543572}, "total_wall_ms": 38548.0027081212, "ttft_ms": 29.62608402594924, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 59, "facts_retained": 11, "retention_rate": 0.1864406779661017, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.14285714285714285, "amount": 0.0, "status": 0.4166666666666667, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 760}, "ledger_by_hop": {"1": {"facts_total": 59, "facts_retained": 11, "retention_rate": 0.1864406779661017, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.14285714285714285, "amount": 0.0, "status": 0.4166666666666667, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 760}, "2": {"facts_total": 59, "facts_retained": 11, "retention_rate": 0.1864406779661017, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.14285714285714285, "amount": 0.0, "status": 0.4166666666666667, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 760}}}, "reference_key_id": "keys/synthetic/syn_03012_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "bb16af6fe25c6b4abc9bdb0ad8c8b3d8b7d3dc12769f5feefdb7a2e4bb915494", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "a52b9267a77200b620f98924", "attempt_id": "efe2553ad664", "run_group_id": "E4_core:A5|closed|qwen3b:h2", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:33:40.720392Z", "task_family": "synthetic", "document_id": "syn_03012_2000", "document_hash": "03e441e68d54e1af036f8ff905dfd6cbbccead62636279c1a8e179352ab08945", "question_id": "q04_multi_hop", "split": "test", "cluster_id": "syn_03012_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 2, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1927, "query_tokens": 437, "message_tokens": 348, "message_utf8_bytes": 824, "cache_tensor_bytes": null, "serialized_payload_bytes": 824, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 38187.22012510989, "hops": {"1": 22467.89562504273, "2": 15719.32450006716}, "receiver_prefill_ms": 292.2379170777276, "ttft_ms": 28.738167020492256, "decode_ms": 42.0192499877885}, "total_wall_ms": 38550.81537517253, "ttft_ms": 28.738167020492256, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 59, "facts_retained": 11, "retention_rate": 0.1864406779661017, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.14285714285714285, "amount": 0.0, "status": 0.4166666666666667, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 760}, "ledger_by_hop": {"1": {"facts_total": 59, "facts_retained": 11, "retention_rate": 0.1864406779661017, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.14285714285714285, "amount": 0.0, "status": 0.4166666666666667, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 760}, "2": {"facts_total": 59, "facts_retained": 11, "retention_rate": 0.1864406779661017, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.14285714285714285, "amount": 0.0, "status": 0.4166666666666667, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 760}}}, "reference_key_id": "keys/synthetic/syn_03012_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "bb16af6fe25c6b4abc9bdb0ad8c8b3d8b7d3dc12769f5feefdb7a2e4bb915494", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "b8ae32d60bd4d01476fdaf57", "attempt_id": "33cdf195040f", "run_group_id": "E4_core:A5|closed|qwen3b:h2", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:33:41.101123Z", "task_family": "synthetic", "document_id": "syn_03012_2000", "document_hash": "03e441e68d54e1af036f8ff905dfd6cbbccead62636279c1a8e179352ab08945", "question_id": "q05_multi_hop", "split": "test", "cluster_id": "syn_03012_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 2, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1927, "query_tokens": 440, "message_tokens": 348, "message_utf8_bytes": 824, "cache_tensor_bytes": null, "serialized_payload_bytes": 824, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 38187.22012510989, "hops": {"1": 22467.89562504273, "2": 15719.32450006716}, "receiver_prefill_ms": 305.1022500731051, "ttft_ms": 53.922625025734305, "decode_ms": 41.323874960653484}, "total_wall_ms": 38588.089292054065, "ttft_ms": 53.922625025734305, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 59, "facts_retained": 11, "retention_rate": 0.1864406779661017, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.14285714285714285, "amount": 0.0, "status": 0.4166666666666667, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 760}, "ledger_by_hop": {"1": {"facts_total": 59, "facts_retained": 11, "retention_rate": 0.1864406779661017, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.14285714285714285, "amount": 0.0, "status": 0.4166666666666667, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 760}, "2": {"facts_total": 59, "facts_retained": 11, "retention_rate": 0.1864406779661017, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.14285714285714285, "amount": 0.0, "status": 0.4166666666666667, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 760}}}, "reference_key_id": "keys/synthetic/syn_03012_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "bb16af6fe25c6b4abc9bdb0ad8c8b3d8b7d3dc12769f5feefdb7a2e4bb915494", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "a01c450bfb4598f9abfc4d80", "attempt_id": "761e0b64a59a", "run_group_id": "E4_core:A5|closed|qwen3b:h2", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:33:41.520066Z", "task_family": "synthetic", "document_id": "syn_03012_2000", "document_hash": "03e441e68d54e1af036f8ff905dfd6cbbccead62636279c1a8e179352ab08945", "question_id": "q06_negation", "split": "test", "cluster_id": "syn_03012_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 2, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1927, "query_tokens": 432, "message_tokens": 348, "message_utf8_bytes": 824, "cache_tensor_bytes": null, "serialized_payload_bytes": 824, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 38187.22012510989, "hops": {"1": 22467.89562504273, "2": 15719.32450006716}, "receiver_prefill_ms": 449.9345830408856, "ttft_ms": 32.53595798742026, "decode_ms": 43.3657499961555}, "total_wall_ms": 38713.572708074935, "ttft_ms": 32.53595798742026, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 59, "facts_retained": 11, "retention_rate": 0.1864406779661017, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.14285714285714285, "amount": 0.0, "status": 0.4166666666666667, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 760}, "ledger_by_hop": {"1": {"facts_total": 59, "facts_retained": 11, "retention_rate": 0.1864406779661017, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.14285714285714285, "amount": 0.0, "status": 0.4166666666666667, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 760}, "2": {"facts_total": 59, "facts_retained": 11, "retention_rate": 0.1864406779661017, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.14285714285714285, "amount": 0.0, "status": 0.4166666666666667, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 760}}}, "reference_key_id": "keys/synthetic/syn_03012_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "bb16af6fe25c6b4abc9bdb0ad8c8b3d8b7d3dc12769f5feefdb7a2e4bb915494", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "80db566d266c32e9f9513c90", "attempt_id": "e7535ec77f4d", "run_group_id": "E4_core:A5|closed|qwen3b:h2", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:33:42.063715Z", "task_family": "synthetic", "document_id": "syn_03012_2000", "document_hash": "03e441e68d54e1af036f8ff905dfd6cbbccead62636279c1a8e179352ab08945", "question_id": "q07_negation", "split": "test", "cluster_id": "syn_03012_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 2, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1927, "query_tokens": 432, "message_tokens": 348, "message_utf8_bytes": 824, "cache_tensor_bytes": null, "serialized_payload_bytes": 824, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 38187.22012510989, "hops": {"1": 22467.89562504273, "2": 15719.32450006716}, "receiver_prefill_ms": 286.6857500048354, "ttft_ms": 22.14008301962167, "decode_ms": 40.419540950097144}, "total_wall_ms": 38537.00529213529, "ttft_ms": 22.14008301962167, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 59, "facts_retained": 11, "retention_rate": 0.1864406779661017, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.14285714285714285, "amount": 0.0, "status": 0.4166666666666667, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 760}, "ledger_by_hop": {"1": {"facts_total": 59, "facts_retained": 11, "retention_rate": 0.1864406779661017, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.14285714285714285, "amount": 0.0, "status": 0.4166666666666667, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 760}, "2": {"facts_total": 59, "facts_retained": 11, "retention_rate": 0.1864406779661017, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.14285714285714285, "amount": 0.0, "status": 0.4166666666666667, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 760}}}, "reference_key_id": "keys/synthetic/syn_03012_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "bb16af6fe25c6b4abc9bdb0ad8c8b3d8b7d3dc12769f5feefdb7a2e4bb915494", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "61efdd335d5474e7ee8ad21d", "attempt_id": "895395390b68", "run_group_id": "E4_core:A5|closed|qwen3b:h2", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:33:42.430661Z", "task_family": "synthetic", "document_id": "syn_03012_2000", "document_hash": "03e441e68d54e1af036f8ff905dfd6cbbccead62636279c1a8e179352ab08945", "question_id": "q08_revision", "split": "test", "cluster_id": "syn_03012_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 2, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1927, "query_tokens": 447, "message_tokens": 348, "message_utf8_bytes": 824, "cache_tensor_bytes": null, "serialized_payload_bytes": 824, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 38187.22012510989, "hops": {"1": 22467.89562504273, "2": 15719.32450006716}, "receiver_prefill_ms": 289.41275004763156, "ttft_ms": 29.94216699153185, "decode_ms": 20.304083009250462}, "total_wall_ms": 38527.45737507939, "ttft_ms": 29.94216699153185, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 59, "facts_retained": 11, "retention_rate": 0.1864406779661017, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.14285714285714285, "amount": 0.0, "status": 0.4166666666666667, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 760}, "ledger_by_hop": {"1": {"facts_total": 59, "facts_retained": 11, "retention_rate": 0.1864406779661017, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.14285714285714285, "amount": 0.0, "status": 0.4166666666666667, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 760}, "2": {"facts_total": 59, "facts_retained": 11, "retention_rate": 0.1864406779661017, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.14285714285714285, "amount": 0.0, "status": 0.4166666666666667, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 760}}}, "reference_key_id": "keys/synthetic/syn_03012_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "bb16af6fe25c6b4abc9bdb0ad8c8b3d8b7d3dc12769f5feefdb7a2e4bb915494", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "5e0e1357475fad921951b4b6", "attempt_id": "0ae3ac456e2f", "run_group_id": "E4_core:A5|closed|qwen3b:h2", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:33:42.788891Z", "task_family": "synthetic", "document_id": "syn_03012_2000", "document_hash": "03e441e68d54e1af036f8ff905dfd6cbbccead62636279c1a8e179352ab08945", "question_id": "q09_revision", "split": "test", "cluster_id": "syn_03012_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 2, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1927, "query_tokens": 440, "message_tokens": 348, "message_utf8_bytes": 824, "cache_tensor_bytes": null, "serialized_payload_bytes": 824, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 38187.22012510989, "hops": {"1": 22467.89562504273, "2": 15719.32450006716}, "receiver_prefill_ms": 305.5957079632208, "ttft_ms": 42.62979200575501, "decode_ms": 77.17920804861933}, "total_wall_ms": 38613.17108303774, "ttft_ms": 42.62979200575501, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 59, "facts_retained": 11, "retention_rate": 0.1864406779661017, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.14285714285714285, "amount": 0.0, "status": 0.4166666666666667, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 760}, "ledger_by_hop": {"1": {"facts_total": 59, "facts_retained": 11, "retention_rate": 0.1864406779661017, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.14285714285714285, "amount": 0.0, "status": 0.4166666666666667, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 760}, "2": {"facts_total": 59, "facts_retained": 11, "retention_rate": 0.1864406779661017, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.14285714285714285, "amount": 0.0, "status": 0.4166666666666667, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 760}}}, "reference_key_id": "keys/synthetic/syn_03012_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "bb16af6fe25c6b4abc9bdb0ad8c8b3d8b7d3dc12769f5feefdb7a2e4bb915494", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "a0f6f6fc9d14ec9b1036986c", "attempt_id": "11faa48c6760", "run_group_id": "E4_core:A5|closed|qwen3b:h2", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:33:43.232545Z", "task_family": "synthetic", "document_id": "syn_03012_2000", "document_hash": "03e441e68d54e1af036f8ff905dfd6cbbccead62636279c1a8e179352ab08945", "question_id": "q10_temporal", "split": "test", "cluster_id": "syn_03012_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 2, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1927, "query_tokens": 450, "message_tokens": 348, "message_utf8_bytes": 824, "cache_tensor_bytes": null, "serialized_payload_bytes": 824, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 38187.22012510989, "hops": {"1": 22467.89562504273, "2": 15719.32450006716}, "receiver_prefill_ms": 433.59858309850097, "ttft_ms": 38.869458017870784, "decode_ms": 37.33912494499236}, "total_wall_ms": 38697.49416608829, "ttft_ms": 38.869458017870784, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 59, "facts_retained": 11, "retention_rate": 0.1864406779661017, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.14285714285714285, "amount": 0.0, "status": 0.4166666666666667, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 760}, "ledger_by_hop": {"1": {"facts_total": 59, "facts_retained": 11, "retention_rate": 0.1864406779661017, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.14285714285714285, "amount": 0.0, "status": 0.4166666666666667, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 760}, "2": {"facts_total": 59, "facts_retained": 11, "retention_rate": 0.1864406779661017, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.14285714285714285, "amount": 0.0, "status": 0.4166666666666667, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 760}}}, "reference_key_id": "keys/synthetic/syn_03012_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "bb16af6fe25c6b4abc9bdb0ad8c8b3d8b7d3dc12769f5feefdb7a2e4bb915494", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "01abba35a9ac3bde02c2f602", "attempt_id": "792504be301f", "run_group_id": "E4_core:A5|closed|qwen3b:h2", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:33:43.761732Z", "task_family": "synthetic", "document_id": "syn_03012_2000", "document_hash": "03e441e68d54e1af036f8ff905dfd6cbbccead62636279c1a8e179352ab08945", "question_id": "q11_aggregation", "split": "test", "cluster_id": "syn_03012_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 2, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1927, "query_tokens": 442, "message_tokens": 348, "message_utf8_bytes": 824, "cache_tensor_bytes": null, "serialized_payload_bytes": 824, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 38187.22012510989, "hops": {"1": 22467.89562504273, "2": 15719.32450006716}, "receiver_prefill_ms": 287.9707500105724, "ttft_ms": 38.93933293875307, "decode_ms": 30.94583295751363}, "total_wall_ms": 38545.642459183, "ttft_ms": 38.93933293875307, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 59, "facts_retained": 11, "retention_rate": 0.1864406779661017, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.14285714285714285, "amount": 0.0, "status": 0.4166666666666667, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 760}, "ledger_by_hop": {"1": {"facts_total": 59, "facts_retained": 11, "retention_rate": 0.1864406779661017, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.14285714285714285, "amount": 0.0, "status": 0.4166666666666667, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 760}, "2": {"facts_total": 59, "facts_retained": 11, "retention_rate": 0.1864406779661017, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.14285714285714285, "amount": 0.0, "status": 0.4166666666666667, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 760}}}, "reference_key_id": "keys/synthetic/syn_03012_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "bb16af6fe25c6b4abc9bdb0ad8c8b3d8b7d3dc12769f5feefdb7a2e4bb915494", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "39fe9f1486c5e5406c440e13", "attempt_id": "1d5fecab0e20", "run_group_id": "E4_core:A5|closed|qwen3b:h2", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:33:44.136509Z", "task_family": "synthetic", "document_id": "syn_03012_2000", "document_hash": "03e441e68d54e1af036f8ff905dfd6cbbccead62636279c1a8e179352ab08945", "question_id": "q12_unit_conversion", "split": "test", "cluster_id": "syn_03012_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 2, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1927, "query_tokens": 439, "message_tokens": 348, "message_utf8_bytes": 824, "cache_tensor_bytes": null, "serialized_payload_bytes": 824, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 38187.22012510989, "hops": {"1": 22467.89562504273, "2": 15719.32450006716}, "receiver_prefill_ms": 284.8856250056997, "ttft_ms": 21.30475000012666, "decode_ms": 205.42975002899766}, "total_wall_ms": 38699.518375098705, "ttft_ms": 21.30475000012666, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 5}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "44620", "evidence_artifact": {"ledger": {"facts_total": 59, "facts_retained": 11, "retention_rate": 0.1864406779661017, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.14285714285714285, "amount": 0.0, "status": 0.4166666666666667, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 760}, "ledger_by_hop": {"1": {"facts_total": 59, "facts_retained": 11, "retention_rate": 0.1864406779661017, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.14285714285714285, "amount": 0.0, "status": 0.4166666666666667, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 760}, "2": {"facts_total": 59, "facts_retained": 11, "retention_rate": 0.1864406779661017, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.14285714285714285, "amount": 0.0, "status": 0.4166666666666667, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 760}}}, "reference_key_id": "keys/synthetic/syn_03012_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "bb16af6fe25c6b4abc9bdb0ad8c8b3d8b7d3dc12769f5feefdb7a2e4bb915494", "output_hash": "2bd125847727be7e7c0a656ee4c81f3bc6d0fef11f536f50ce7bc83688907075"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "0744e512cea789b8be29fdb6", "attempt_id": "dfb621704dda", "run_group_id": "E4_core:A5|closed|qwen3b:h2", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:33:44.665241Z", "task_family": "synthetic", "document_id": "syn_03012_2000", "document_hash": "03e441e68d54e1af036f8ff905dfd6cbbccead62636279c1a8e179352ab08945", "question_id": "q13_unanswerable", "split": "test", "cluster_id": "syn_03012_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 2, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1927, "query_tokens": 444, "message_tokens": 348, "message_utf8_bytes": 824, "cache_tensor_bytes": null, "serialized_payload_bytes": 824, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 38187.22012510989, "hops": {"1": 22467.89562504273, "2": 15719.32450006716}, "receiver_prefill_ms": 348.5516670625657, "ttft_ms": 129.4036250328645, "decode_ms": 41.3250409765169}, "total_wall_ms": 38707.076792023145, "ttft_ms": 129.4036250328645, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 59, "facts_retained": 11, "retention_rate": 0.1864406779661017, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.14285714285714285, "amount": 0.0, "status": 0.4166666666666667, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 760}, "ledger_by_hop": {"1": {"facts_total": 59, "facts_retained": 11, "retention_rate": 0.1864406779661017, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.14285714285714285, "amount": 0.0, "status": 0.4166666666666667, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 760}, "2": {"facts_total": 59, "facts_retained": 11, "retention_rate": 0.1864406779661017, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.14285714285714285, "amount": 0.0, "status": 0.4166666666666667, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 760}}}, "reference_key_id": "keys/synthetic/syn_03012_2000.json", "task_scores": {"correct": 1.0, "abstained": 1.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "bb16af6fe25c6b4abc9bdb0ad8c8b3d8b7d3dc12769f5feefdb7a2e4bb915494", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "fcf3a65664913c17e6c73b2f", "attempt_id": "29d42908c3af", "run_group_id": "E4_core:A5|closed|qwen3b:h2", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:33:45.202656Z", "task_family": "synthetic", "document_id": "syn_03012_2000", "document_hash": "03e441e68d54e1af036f8ff905dfd6cbbccead62636279c1a8e179352ab08945", "question_id": "q14_unanswerable", "split": "test", "cluster_id": "syn_03012_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 2, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1927, "query_tokens": 444, "message_tokens": 348, "message_utf8_bytes": 824, "cache_tensor_bytes": null, "serialized_payload_bytes": 824, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 38187.22012510989, "hops": {"1": 22467.89562504273, "2": 15719.32450006716}, "receiver_prefill_ms": 308.9942920487374, "ttft_ms": 40.159457945264876, "decode_ms": 25.86687495931983}, "total_wall_ms": 38562.70579213742, "ttft_ms": 40.159457945264876, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 59, "facts_retained": 11, "retention_rate": 0.1864406779661017, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.14285714285714285, "amount": 0.0, "status": 0.4166666666666667, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 760}, "ledger_by_hop": {"1": {"facts_total": 59, "facts_retained": 11, "retention_rate": 0.1864406779661017, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.14285714285714285, "amount": 0.0, "status": 0.4166666666666667, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 760}, "2": {"facts_total": 59, "facts_retained": 11, "retention_rate": 0.1864406779661017, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.14285714285714285, "amount": 0.0, "status": 0.4166666666666667, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 760}}}, "reference_key_id": "keys/synthetic/syn_03012_2000.json", "task_scores": {"correct": 1.0, "abstained": 1.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "bb16af6fe25c6b4abc9bdb0ad8c8b3d8b7d3dc12769f5feefdb7a2e4bb915494", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "5c4c3b13e863239c5f5f7492", "attempt_id": "e1405e690e0b", "run_group_id": "E4_core:A5|closed|qwen3b:h4", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:34:16.823202Z", "task_family": "synthetic", "document_id": "syn_03012_2000", "document_hash": "03e441e68d54e1af036f8ff905dfd6cbbccead62636279c1a8e179352ab08945", "question_id": "q00_single_fact", "split": "test", "cluster_id": "syn_03012_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 4, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1927, "query_tokens": 431, "message_tokens": 348, "message_utf8_bytes": 824, "cache_tensor_bytes": null, "serialized_payload_bytes": 824, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 69412.97741606832, "hops": {"1": 22467.89562504273, "2": 15719.32450006716, "3": 15511.978083057329, "4": 15713.779207901098}, "receiver_prefill_ms": 285.36579199135303, "ttft_ms": 24.539333069697022, "decode_ms": 290.57304200250655}, "total_wall_ms": 70014.20404098462, "ttft_ms": 24.539333069697022, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 10}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "XX-71M-6213", "evidence_artifact": {"ledger": {"facts_total": 59, "facts_retained": 11, "retention_rate": 0.1864406779661017, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.14285714285714285, "amount": 0.0, "status": 0.4166666666666667, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 760}, "ledger_by_hop": {"1": {"facts_total": 59, "facts_retained": 11, "retention_rate": 0.1864406779661017, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.14285714285714285, "amount": 0.0, "status": 0.4166666666666667, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 760}, "2": {"facts_total": 59, "facts_retained": 11, "retention_rate": 0.1864406779661017, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.14285714285714285, "amount": 0.0, "status": 0.4166666666666667, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 760}, "3": {"facts_total": 59, "facts_retained": 11, "retention_rate": 0.1864406779661017, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.14285714285714285, "amount": 0.0, "status": 0.4166666666666667, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 760}, "4": {"facts_total": 59, "facts_retained": 11, "retention_rate": 0.1864406779661017, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.14285714285714285, "amount": 0.0, "status": 0.4166666666666667, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 760}}}, "reference_key_id": "keys/synthetic/syn_03012_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "bb16af6fe25c6b4abc9bdb0ad8c8b3d8b7d3dc12769f5feefdb7a2e4bb915494", "output_hash": "66ef6c5da682b38795e9029723bbd71cdde2c4dec70cf1fae6efe62ef243f10a"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "1b269bcd0eba37c92bae6b59", "attempt_id": "a95cf9992666", "run_group_id": "E4_core:A5|closed|qwen3b:h4", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:34:17.442422Z", "task_family": "synthetic", "document_id": "syn_03012_2000", "document_hash": "03e441e68d54e1af036f8ff905dfd6cbbccead62636279c1a8e179352ab08945", "question_id": "q01_single_fact", "split": "test", "cluster_id": "syn_03012_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 4, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1927, "query_tokens": 437, "message_tokens": 348, "message_utf8_bytes": 824, "cache_tensor_bytes": null, "serialized_payload_bytes": 824, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 69412.97741606832, "hops": {"1": 22467.89562504273, "2": 15719.32450006716, "3": 15511.978083057329, "4": 15713.779207901098}, "receiver_prefill_ms": 344.98404199257493, "ttft_ms": 36.47545806597918, "decode_ms": 35.913625033572316}, "total_wall_ms": 69831.21454110369, "ttft_ms": 36.47545806597918, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 59, "facts_retained": 11, "retention_rate": 0.1864406779661017, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.14285714285714285, "amount": 0.0, "status": 0.4166666666666667, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 760}, "ledger_by_hop": {"1": {"facts_total": 59, "facts_retained": 11, "retention_rate": 0.1864406779661017, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.14285714285714285, "amount": 0.0, "status": 0.4166666666666667, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 760}, "2": {"facts_total": 59, "facts_retained": 11, "retention_rate": 0.1864406779661017, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.14285714285714285, "amount": 0.0, "status": 0.4166666666666667, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 760}, "3": {"facts_total": 59, "facts_retained": 11, "retention_rate": 0.1864406779661017, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.14285714285714285, "amount": 0.0, "status": 0.4166666666666667, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 760}, "4": {"facts_total": 59, "facts_retained": 11, "retention_rate": 0.1864406779661017, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.14285714285714285, "amount": 0.0, "status": 0.4166666666666667, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 760}}}, "reference_key_id": "keys/synthetic/syn_03012_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "bb16af6fe25c6b4abc9bdb0ad8c8b3d8b7d3dc12769f5feefdb7a2e4bb915494", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "e9c39ded11bf32202d3d3eef", "attempt_id": "b81b902005e3", "run_group_id": "E4_core:A5|closed|qwen3b:h4", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:34:17.879367Z", "task_family": "synthetic", "document_id": "syn_03012_2000", "document_hash": "03e441e68d54e1af036f8ff905dfd6cbbccead62636279c1a8e179352ab08945", "question_id": "q02_exact_copy", "split": "test", "cluster_id": "syn_03012_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 4, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1927, "query_tokens": 436, "message_tokens": 348, "message_utf8_bytes": 824, "cache_tensor_bytes": null, "serialized_payload_bytes": 824, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 69412.97741606832, "hops": {"1": 22467.89562504273, "2": 15719.32450006716, "3": 15511.978083057329, "4": 15713.779207901098}, "receiver_prefill_ms": 433.41770803090185, "ttft_ms": 30.940916971303523, "decode_ms": 43.97499992046505}, "total_wall_ms": 69921.74895806238, "ttft_ms": 30.940916971303523, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 59, "facts_retained": 11, "retention_rate": 0.1864406779661017, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.14285714285714285, "amount": 0.0, "status": 0.4166666666666667, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 760}, "ledger_by_hop": {"1": {"facts_total": 59, "facts_retained": 11, "retention_rate": 0.1864406779661017, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.14285714285714285, "amount": 0.0, "status": 0.4166666666666667, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 760}, "2": {"facts_total": 59, "facts_retained": 11, "retention_rate": 0.1864406779661017, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.14285714285714285, "amount": 0.0, "status": 0.4166666666666667, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 760}, "3": {"facts_total": 59, "facts_retained": 11, "retention_rate": 0.1864406779661017, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.14285714285714285, "amount": 0.0, "status": 0.4166666666666667, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 760}, "4": {"facts_total": 59, "facts_retained": 11, "retention_rate": 0.1864406779661017, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.14285714285714285, "amount": 0.0, "status": 0.4166666666666667, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 760}}}, "reference_key_id": "keys/synthetic/syn_03012_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "bb16af6fe25c6b4abc9bdb0ad8c8b3d8b7d3dc12769f5feefdb7a2e4bb915494", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "4a0b833135aa4e6e29dde591", "attempt_id": "7549ee0166f4", "run_group_id": "E4_core:A5|closed|qwen3b:h4", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:34:18.405335Z", "task_family": "synthetic", "document_id": "syn_03012_2000", "document_hash": "03e441e68d54e1af036f8ff905dfd6cbbccead62636279c1a8e179352ab08945", "question_id": "q03_exact_copy", "split": "test", "cluster_id": "syn_03012_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 4, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1927, "query_tokens": 432, "message_tokens": 348, "message_utf8_bytes": 824, "cache_tensor_bytes": null, "serialized_payload_bytes": 824, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 69412.97741606832, "hops": {"1": 22467.89562504273, "2": 15719.32450006716, "3": 15511.978083057329, "4": 15713.779207901098}, "receiver_prefill_ms": 291.84258298482746, "ttft_ms": 26.451083016581833, "decode_ms": 48.26945799868554}, "total_wall_ms": 69780.01170710195, "ttft_ms": 26.451083016581833, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 59, "facts_retained": 11, "retention_rate": 0.1864406779661017, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.14285714285714285, "amount": 0.0, "status": 0.4166666666666667, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 760}, "ledger_by_hop": {"1": {"facts_total": 59, "facts_retained": 11, "retention_rate": 0.1864406779661017, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.14285714285714285, "amount": 0.0, "status": 0.4166666666666667, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 760}, "2": {"facts_total": 59, "facts_retained": 11, "retention_rate": 0.1864406779661017, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.14285714285714285, "amount": 0.0, "status": 0.4166666666666667, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 760}, "3": {"facts_total": 59, "facts_retained": 11, "retention_rate": 0.1864406779661017, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.14285714285714285, "amount": 0.0, "status": 0.4166666666666667, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 760}, "4": {"facts_total": 59, "facts_retained": 11, "retention_rate": 0.1864406779661017, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.14285714285714285, "amount": 0.0, "status": 0.4166666666666667, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 760}}}, "reference_key_id": "keys/synthetic/syn_03012_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "bb16af6fe25c6b4abc9bdb0ad8c8b3d8b7d3dc12769f5feefdb7a2e4bb915494", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "ae5c661219f38294ba69db1e", "attempt_id": "da8a33193c63", "run_group_id": "E4_core:A5|closed|qwen3b:h4", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:34:18.789329Z", "task_family": "synthetic", "document_id": "syn_03012_2000", "document_hash": "03e441e68d54e1af036f8ff905dfd6cbbccead62636279c1a8e179352ab08945", "question_id": "q04_multi_hop", "split": "test", "cluster_id": "syn_03012_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 4, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1927, "query_tokens": 437, "message_tokens": 348, "message_utf8_bytes": 824, "cache_tensor_bytes": null, "serialized_payload_bytes": 824, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 69412.97741606832, "hops": {"1": 22467.89562504273, "2": 15719.32450006716, "3": 15511.978083057329, "4": 15713.779207901098}, "receiver_prefill_ms": 275.01108311116695, "ttft_ms": 17.449249979108572, "decode_ms": 53.451625048182905}, "total_wall_ms": 69759.40416601952, "ttft_ms": 17.449249979108572, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 59, "facts_retained": 11, "retention_rate": 0.1864406779661017, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.14285714285714285, "amount": 0.0, "status": 0.4166666666666667, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 760}, "ledger_by_hop": {"1": {"facts_total": 59, "facts_retained": 11, "retention_rate": 0.1864406779661017, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.14285714285714285, "amount": 0.0, "status": 0.4166666666666667, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 760}, "2": {"facts_total": 59, "facts_retained": 11, "retention_rate": 0.1864406779661017, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.14285714285714285, "amount": 0.0, "status": 0.4166666666666667, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 760}, "3": {"facts_total": 59, "facts_retained": 11, "retention_rate": 0.1864406779661017, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.14285714285714285, "amount": 0.0, "status": 0.4166666666666667, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 760}, "4": {"facts_total": 59, "facts_retained": 11, "retention_rate": 0.1864406779661017, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.14285714285714285, "amount": 0.0, "status": 0.4166666666666667, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 760}}}, "reference_key_id": "keys/synthetic/syn_03012_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "bb16af6fe25c6b4abc9bdb0ad8c8b3d8b7d3dc12769f5feefdb7a2e4bb915494", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "de87b7445bc3f28bead30769", "attempt_id": "9aa0e3260727", "run_group_id": "E4_core:A5|closed|qwen3b:h4", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:34:19.152084Z", "task_family": "synthetic", "document_id": "syn_03012_2000", "document_hash": "03e441e68d54e1af036f8ff905dfd6cbbccead62636279c1a8e179352ab08945", "question_id": "q05_multi_hop", "split": "test", "cluster_id": "syn_03012_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 4, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1927, "query_tokens": 440, "message_tokens": 348, "message_utf8_bytes": 824, "cache_tensor_bytes": null, "serialized_payload_bytes": 824, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 69412.97741606832, "hops": {"1": 22467.89562504273, "2": 15719.32450006716, "3": 15511.978083057329, "4": 15713.779207901098}, "receiver_prefill_ms": 297.3968329606578, "ttft_ms": 60.19608292263001, "decode_ms": 97.14533400256187}, "total_wall_ms": 69868.29437501729, "ttft_ms": 60.19608292263001, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 59, "facts_retained": 11, "retention_rate": 0.1864406779661017, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.14285714285714285, "amount": 0.0, "status": 0.4166666666666667, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 760}, "ledger_by_hop": {"1": {"facts_total": 59, "facts_retained": 11, "retention_rate": 0.1864406779661017, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.14285714285714285, "amount": 0.0, "status": 0.4166666666666667, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 760}, "2": {"facts_total": 59, "facts_retained": 11, "retention_rate": 0.1864406779661017, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.14285714285714285, "amount": 0.0, "status": 0.4166666666666667, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 760}, "3": {"facts_total": 59, "facts_retained": 11, "retention_rate": 0.1864406779661017, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.14285714285714285, "amount": 0.0, "status": 0.4166666666666667, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 760}, "4": {"facts_total": 59, "facts_retained": 11, "retention_rate": 0.1864406779661017, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.14285714285714285, "amount": 0.0, "status": 0.4166666666666667, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 760}}}, "reference_key_id": "keys/synthetic/syn_03012_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "bb16af6fe25c6b4abc9bdb0ad8c8b3d8b7d3dc12769f5feefdb7a2e4bb915494", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "a052227fb9f4f79ac2086f1e", "attempt_id": "2e2d3f8b5f93", "run_group_id": "E4_core:A5|closed|qwen3b:h4", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:34:19.625555Z", "task_family": "synthetic", "document_id": "syn_03012_2000", "document_hash": "03e441e68d54e1af036f8ff905dfd6cbbccead62636279c1a8e179352ab08945", "question_id": "q06_negation", "split": "test", "cluster_id": "syn_03012_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 4, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1927, "query_tokens": 432, "message_tokens": 348, "message_utf8_bytes": 824, "cache_tensor_bytes": null, "serialized_payload_bytes": 824, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 69412.97741606832, "hops": {"1": 22467.89562504273, "2": 15719.32450006716, "3": 15511.978083057329, "4": 15713.779207901098}, "receiver_prefill_ms": 381.6209579817951, "ttft_ms": 39.50299997814, "decode_ms": 33.508374937810004}, "total_wall_ms": 69868.25487401802, "ttft_ms": 39.50299997814, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 59, "facts_retained": 11, "retention_rate": 0.1864406779661017, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.14285714285714285, "amount": 0.0, "status": 0.4166666666666667, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 760}, "ledger_by_hop": {"1": {"facts_total": 59, "facts_retained": 11, "retention_rate": 0.1864406779661017, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.14285714285714285, "amount": 0.0, "status": 0.4166666666666667, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 760}, "2": {"facts_total": 59, "facts_retained": 11, "retention_rate": 0.1864406779661017, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.14285714285714285, "amount": 0.0, "status": 0.4166666666666667, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 760}, "3": {"facts_total": 59, "facts_retained": 11, "retention_rate": 0.1864406779661017, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.14285714285714285, "amount": 0.0, "status": 0.4166666666666667, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 760}, "4": {"facts_total": 59, "facts_retained": 11, "retention_rate": 0.1864406779661017, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.14285714285714285, "amount": 0.0, "status": 0.4166666666666667, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 760}}}, "reference_key_id": "keys/synthetic/syn_03012_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "bb16af6fe25c6b4abc9bdb0ad8c8b3d8b7d3dc12769f5feefdb7a2e4bb915494", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "cecde18cc34a79e968476d79", "attempt_id": "e48aaf0c87d8", "run_group_id": "E4_core:A5|closed|qwen3b:h4", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:34:20.100563Z", "task_family": "synthetic", "document_id": "syn_03012_2000", "document_hash": "03e441e68d54e1af036f8ff905dfd6cbbccead62636279c1a8e179352ab08945", "question_id": "q07_negation", "split": "test", "cluster_id": "syn_03012_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 4, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1927, "query_tokens": 432, "message_tokens": 348, "message_utf8_bytes": 824, "cache_tensor_bytes": null, "serialized_payload_bytes": 824, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 69412.97741606832, "hops": {"1": 22467.89562504273, "2": 15719.32450006716, "3": 15511.978083057329, "4": 15713.779207901098}, "receiver_prefill_ms": 282.6567499432713, "ttft_ms": 35.62341700308025, "decode_ms": 35.12395895086229}, "total_wall_ms": 69767.05366605893, "ttft_ms": 35.62341700308025, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 59, "facts_retained": 11, "retention_rate": 0.1864406779661017, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.14285714285714285, "amount": 0.0, "status": 0.4166666666666667, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 760}, "ledger_by_hop": {"1": {"facts_total": 59, "facts_retained": 11, "retention_rate": 0.1864406779661017, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.14285714285714285, "amount": 0.0, "status": 0.4166666666666667, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 760}, "2": {"facts_total": 59, "facts_retained": 11, "retention_rate": 0.1864406779661017, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.14285714285714285, "amount": 0.0, "status": 0.4166666666666667, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 760}, "3": {"facts_total": 59, "facts_retained": 11, "retention_rate": 0.1864406779661017, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.14285714285714285, "amount": 0.0, "status": 0.4166666666666667, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 760}, "4": {"facts_total": 59, "facts_retained": 11, "retention_rate": 0.1864406779661017, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.14285714285714285, "amount": 0.0, "status": 0.4166666666666667, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 760}}}, "reference_key_id": "keys/synthetic/syn_03012_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "bb16af6fe25c6b4abc9bdb0ad8c8b3d8b7d3dc12769f5feefdb7a2e4bb915494", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "a7c8693139985c13e9b0b2b1", "attempt_id": "7d820cc08202", "run_group_id": "E4_core:A5|closed|qwen3b:h4", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:34:20.470952Z", "task_family": "synthetic", "document_id": "syn_03012_2000", "document_hash": "03e441e68d54e1af036f8ff905dfd6cbbccead62636279c1a8e179352ab08945", "question_id": "q08_revision", "split": "test", "cluster_id": "syn_03012_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 4, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1927, "query_tokens": 447, "message_tokens": 348, "message_utf8_bytes": 824, "cache_tensor_bytes": null, "serialized_payload_bytes": 824, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 69412.97741606832, "hops": {"1": 22467.89562504273, "2": 15719.32450006716, "3": 15511.978083057329, "4": 15713.779207901098}, "receiver_prefill_ms": 268.0219589965418, "ttft_ms": 24.084709002636373, "decode_ms": 74.67279199045151}, "total_wall_ms": 69780.30399908312, "ttft_ms": 24.084709002636373, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 59, "facts_retained": 11, "retention_rate": 0.1864406779661017, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.14285714285714285, "amount": 0.0, "status": 0.4166666666666667, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 760}, "ledger_by_hop": {"1": {"facts_total": 59, "facts_retained": 11, "retention_rate": 0.1864406779661017, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.14285714285714285, "amount": 0.0, "status": 0.4166666666666667, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 760}, "2": {"facts_total": 59, "facts_retained": 11, "retention_rate": 0.1864406779661017, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.14285714285714285, "amount": 0.0, "status": 0.4166666666666667, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 760}, "3": {"facts_total": 59, "facts_retained": 11, "retention_rate": 0.1864406779661017, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.14285714285714285, "amount": 0.0, "status": 0.4166666666666667, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 760}, "4": {"facts_total": 59, "facts_retained": 11, "retention_rate": 0.1864406779661017, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.14285714285714285, "amount": 0.0, "status": 0.4166666666666667, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 760}}}, "reference_key_id": "keys/synthetic/syn_03012_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "bb16af6fe25c6b4abc9bdb0ad8c8b3d8b7d3dc12769f5feefdb7a2e4bb915494", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "b3c652a6496906f427b77403", "attempt_id": "27472afe65d8", "run_group_id": "E4_core:A5|closed|qwen3b:h4", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:34:20.855574Z", "task_family": "synthetic", "document_id": "syn_03012_2000", "document_hash": "03e441e68d54e1af036f8ff905dfd6cbbccead62636279c1a8e179352ab08945", "question_id": "q09_revision", "split": "test", "cluster_id": "syn_03012_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 4, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1927, "query_tokens": 440, "message_tokens": 348, "message_utf8_bytes": 824, "cache_tensor_bytes": null, "serialized_payload_bytes": 824, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 69412.97741606832, "hops": {"1": 22467.89562504273, "2": 15719.32450006716, "3": 15511.978083057329, "4": 15713.779207901098}, "receiver_prefill_ms": 279.4977920129895, "ttft_ms": 57.37329204566777, "decode_ms": 119.67124999500811}, "total_wall_ms": 69870.03666604869, "ttft_ms": 57.37329204566777, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 59, "facts_retained": 11, "retention_rate": 0.1864406779661017, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.14285714285714285, "amount": 0.0, "status": 0.4166666666666667, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 760}, "ledger_by_hop": {"1": {"facts_total": 59, "facts_retained": 11, "retention_rate": 0.1864406779661017, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.14285714285714285, "amount": 0.0, "status": 0.4166666666666667, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 760}, "2": {"facts_total": 59, "facts_retained": 11, "retention_rate": 0.1864406779661017, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.14285714285714285, "amount": 0.0, "status": 0.4166666666666667, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 760}, "3": {"facts_total": 59, "facts_retained": 11, "retention_rate": 0.1864406779661017, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.14285714285714285, "amount": 0.0, "status": 0.4166666666666667, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 760}, "4": {"facts_total": 59, "facts_retained": 11, "retention_rate": 0.1864406779661017, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.14285714285714285, "amount": 0.0, "status": 0.4166666666666667, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 760}}}, "reference_key_id": "keys/synthetic/syn_03012_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "bb16af6fe25c6b4abc9bdb0ad8c8b3d8b7d3dc12769f5feefdb7a2e4bb915494", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "f849fda8d4118f31e8403702", "attempt_id": "91f3d9d35ce4", "run_group_id": "E4_core:A5|closed|qwen3b:h4", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:34:21.330851Z", "task_family": "synthetic", "document_id": "syn_03012_2000", "document_hash": "03e441e68d54e1af036f8ff905dfd6cbbccead62636279c1a8e179352ab08945", "question_id": "q10_temporal", "split": "test", "cluster_id": "syn_03012_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 4, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1927, "query_tokens": 450, "message_tokens": 348, "message_utf8_bytes": 824, "cache_tensor_bytes": null, "serialized_payload_bytes": 824, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 69412.97741606832, "hops": {"1": 22467.89562504273, "2": 15719.32450006716, "3": 15511.978083057329, "4": 15713.779207901098}, "receiver_prefill_ms": 367.3314170446247, "ttft_ms": 40.51583295222372, "decode_ms": 29.56187492236495}, "total_wall_ms": 69850.83049896639, "ttft_ms": 40.51583295222372, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 59, "facts_retained": 11, "retention_rate": 0.1864406779661017, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.14285714285714285, "amount": 0.0, "status": 0.4166666666666667, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 760}, "ledger_by_hop": {"1": {"facts_total": 59, "facts_retained": 11, "retention_rate": 0.1864406779661017, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.14285714285714285, "amount": 0.0, "status": 0.4166666666666667, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 760}, "2": {"facts_total": 59, "facts_retained": 11, "retention_rate": 0.1864406779661017, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.14285714285714285, "amount": 0.0, "status": 0.4166666666666667, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 760}, "3": {"facts_total": 59, "facts_retained": 11, "retention_rate": 0.1864406779661017, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.14285714285714285, "amount": 0.0, "status": 0.4166666666666667, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 760}, "4": {"facts_total": 59, "facts_retained": 11, "retention_rate": 0.1864406779661017, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.14285714285714285, "amount": 0.0, "status": 0.4166666666666667, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 760}}}, "reference_key_id": "keys/synthetic/syn_03012_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "bb16af6fe25c6b4abc9bdb0ad8c8b3d8b7d3dc12769f5feefdb7a2e4bb915494", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "ea86ce4ddb76f2e474edf655", "attempt_id": "3cbe300515b2", "run_group_id": "E4_core:A5|closed|qwen3b:h4", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:34:21.785323Z", "task_family": "synthetic", "document_id": "syn_03012_2000", "document_hash": "03e441e68d54e1af036f8ff905dfd6cbbccead62636279c1a8e179352ab08945", "question_id": "q11_aggregation", "split": "test", "cluster_id": "syn_03012_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 4, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1927, "query_tokens": 442, "message_tokens": 348, "message_utf8_bytes": 824, "cache_tensor_bytes": null, "serialized_payload_bytes": 824, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 69412.97741606832, "hops": {"1": 22467.89562504273, "2": 15719.32450006716, "3": 15511.978083057329, "4": 15713.779207901098}, "receiver_prefill_ms": 284.8533330252394, "ttft_ms": 26.653000037185848, "decode_ms": 35.46050004661083}, "total_wall_ms": 69760.37554105278, "ttft_ms": 26.653000037185848, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 59, "facts_retained": 11, "retention_rate": 0.1864406779661017, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.14285714285714285, "amount": 0.0, "status": 0.4166666666666667, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 760}, "ledger_by_hop": {"1": {"facts_total": 59, "facts_retained": 11, "retention_rate": 0.1864406779661017, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.14285714285714285, "amount": 0.0, "status": 0.4166666666666667, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 760}, "2": {"facts_total": 59, "facts_retained": 11, "retention_rate": 0.1864406779661017, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.14285714285714285, "amount": 0.0, "status": 0.4166666666666667, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 760}, "3": {"facts_total": 59, "facts_retained": 11, "retention_rate": 0.1864406779661017, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.14285714285714285, "amount": 0.0, "status": 0.4166666666666667, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 760}, "4": {"facts_total": 59, "facts_retained": 11, "retention_rate": 0.1864406779661017, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.14285714285714285, "amount": 0.0, "status": 0.4166666666666667, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 760}}}, "reference_key_id": "keys/synthetic/syn_03012_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "bb16af6fe25c6b4abc9bdb0ad8c8b3d8b7d3dc12769f5feefdb7a2e4bb915494", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "7450e16914309a41d8fcd5c2", "attempt_id": "490476a05d87", "run_group_id": "E4_core:A5|closed|qwen3b:h4", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:34:22.149312Z", "task_family": "synthetic", "document_id": "syn_03012_2000", "document_hash": "03e441e68d54e1af036f8ff905dfd6cbbccead62636279c1a8e179352ab08945", "question_id": "q12_unit_conversion", "split": "test", "cluster_id": "syn_03012_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 4, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1927, "query_tokens": 439, "message_tokens": 348, "message_utf8_bytes": 824, "cache_tensor_bytes": null, "serialized_payload_bytes": 824, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 69412.97741606832, "hops": {"1": 22467.89562504273, "2": 15719.32450006716, "3": 15511.978083057329, "4": 15713.779207901098}, "receiver_prefill_ms": 281.86104202177376, "ttft_ms": 32.010915922001004, "decode_ms": 263.89950001612306}, "total_wall_ms": 69991.37141602114, "ttft_ms": 32.010915922001004, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 5}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "44620", "evidence_artifact": {"ledger": {"facts_total": 59, "facts_retained": 11, "retention_rate": 0.1864406779661017, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.14285714285714285, "amount": 0.0, "status": 0.4166666666666667, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 760}, "ledger_by_hop": {"1": {"facts_total": 59, "facts_retained": 11, "retention_rate": 0.1864406779661017, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.14285714285714285, "amount": 0.0, "status": 0.4166666666666667, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 760}, "2": {"facts_total": 59, "facts_retained": 11, "retention_rate": 0.1864406779661017, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.14285714285714285, "amount": 0.0, "status": 0.4166666666666667, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 760}, "3": {"facts_total": 59, "facts_retained": 11, "retention_rate": 0.1864406779661017, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.14285714285714285, "amount": 0.0, "status": 0.4166666666666667, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 760}, "4": {"facts_total": 59, "facts_retained": 11, "retention_rate": 0.1864406779661017, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.14285714285714285, "amount": 0.0, "status": 0.4166666666666667, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 760}}}, "reference_key_id": "keys/synthetic/syn_03012_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "bb16af6fe25c6b4abc9bdb0ad8c8b3d8b7d3dc12769f5feefdb7a2e4bb915494", "output_hash": "2bd125847727be7e7c0a656ee4c81f3bc6d0fef11f536f50ce7bc83688907075"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "5631aa201b5baf37e105f61c", "attempt_id": "51d967dd450b", "run_group_id": "E4_core:A5|closed|qwen3b:h4", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:34:22.745180Z", "task_family": "synthetic", "document_id": "syn_03012_2000", "document_hash": "03e441e68d54e1af036f8ff905dfd6cbbccead62636279c1a8e179352ab08945", "question_id": "q13_unanswerable", "split": "test", "cluster_id": "syn_03012_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 4, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1927, "query_tokens": 444, "message_tokens": 348, "message_utf8_bytes": 824, "cache_tensor_bytes": null, "serialized_payload_bytes": 824, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 69412.97741606832, "hops": {"1": 22467.89562504273, "2": 15719.32450006716, "3": 15511.978083057329, "4": 15713.779207901098}, "receiver_prefill_ms": 448.0460420018062, "ttft_ms": 22.04166690353304, "decode_ms": 40.601583081297576}, "total_wall_ms": 69924.21391606331, "ttft_ms": 22.04166690353304, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 59, "facts_retained": 11, "retention_rate": 0.1864406779661017, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.14285714285714285, "amount": 0.0, "status": 0.4166666666666667, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 760}, "ledger_by_hop": {"1": {"facts_total": 59, "facts_retained": 11, "retention_rate": 0.1864406779661017, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.14285714285714285, "amount": 0.0, "status": 0.4166666666666667, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 760}, "2": {"facts_total": 59, "facts_retained": 11, "retention_rate": 0.1864406779661017, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.14285714285714285, "amount": 0.0, "status": 0.4166666666666667, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 760}, "3": {"facts_total": 59, "facts_retained": 11, "retention_rate": 0.1864406779661017, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.14285714285714285, "amount": 0.0, "status": 0.4166666666666667, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 760}, "4": {"facts_total": 59, "facts_retained": 11, "retention_rate": 0.1864406779661017, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.14285714285714285, "amount": 0.0, "status": 0.4166666666666667, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 760}}}, "reference_key_id": "keys/synthetic/syn_03012_2000.json", "task_scores": {"correct": 1.0, "abstained": 1.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "bb16af6fe25c6b4abc9bdb0ad8c8b3d8b7d3dc12769f5feefdb7a2e4bb915494", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "190d9ca28b32ab0c7370f510", "attempt_id": "688ff3d1d0e4", "run_group_id": "E4_core:A5|closed|qwen3b:h4", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:34:23.273749Z", "task_family": "synthetic", "document_id": "syn_03012_2000", "document_hash": "03e441e68d54e1af036f8ff905dfd6cbbccead62636279c1a8e179352ab08945", "question_id": "q14_unanswerable", "split": "test", "cluster_id": "syn_03012_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 4, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1927, "query_tokens": 444, "message_tokens": 348, "message_utf8_bytes": 824, "cache_tensor_bytes": null, "serialized_payload_bytes": 824, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 69412.97741606832, "hops": {"1": 22467.89562504273, "2": 15719.32450006716, "3": 15511.978083057329, "4": 15713.779207901098}, "receiver_prefill_ms": 296.9218750949949, "ttft_ms": 25.148083921521902, "decode_ms": 48.2092919992283}, "total_wall_ms": 69783.78358308692, "ttft_ms": 25.148083921521902, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 59, "facts_retained": 11, "retention_rate": 0.1864406779661017, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.14285714285714285, "amount": 0.0, "status": 0.4166666666666667, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 760}, "ledger_by_hop": {"1": {"facts_total": 59, "facts_retained": 11, "retention_rate": 0.1864406779661017, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.14285714285714285, "amount": 0.0, "status": 0.4166666666666667, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 760}, "2": {"facts_total": 59, "facts_retained": 11, "retention_rate": 0.1864406779661017, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.14285714285714285, "amount": 0.0, "status": 0.4166666666666667, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 760}, "3": {"facts_total": 59, "facts_retained": 11, "retention_rate": 0.1864406779661017, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.14285714285714285, "amount": 0.0, "status": 0.4166666666666667, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 760}, "4": {"facts_total": 59, "facts_retained": 11, "retention_rate": 0.1864406779661017, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.14285714285714285, "amount": 0.0, "status": 0.4166666666666667, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 760}}}, "reference_key_id": "keys/synthetic/syn_03012_2000.json", "task_scores": {"correct": 1.0, "abstained": 1.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "bb16af6fe25c6b4abc9bdb0ad8c8b3d8b7d3dc12769f5feefdb7a2e4bb915494", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "89b55f9599243c81e8077810", "attempt_id": "37bd45c4206a", "run_group_id": "E4_core:A5|closed|qwen3b:h8", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:35:25.761050Z", "task_family": "synthetic", "document_id": "syn_03012_2000", "document_hash": "03e441e68d54e1af036f8ff905dfd6cbbccead62636279c1a8e179352ab08945", "question_id": "q00_single_fact", "split": "test", "cluster_id": "syn_03012_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 8, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1927, "query_tokens": 431, "message_tokens": 348, "message_utf8_bytes": 824, "cache_tensor_bytes": null, "serialized_payload_bytes": 824, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 131507.60937598534, "hops": {"1": 22467.89562504273, "2": 15719.32450006716, "3": 15511.978083057329, "4": 15713.779207901098, "5": 15311.877917032689, "6": 15750.822708941996, "7": 15396.881458931603, "8": 15635.049875010736}, "receiver_prefill_ms": 371.8123339349404, "ttft_ms": 42.05462499521673, "decode_ms": 320.7732089795172}, "total_wall_ms": 132242.9976259591, "ttft_ms": 42.05462499521673, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 10}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "XX-71M-6213", "evidence_artifact": {"ledger": {"facts_total": 59, "facts_retained": 11, "retention_rate": 0.1864406779661017, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.14285714285714285, "amount": 0.0, "status": 0.4166666666666667, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 760}, "ledger_by_hop": {"1": {"facts_total": 59, "facts_retained": 11, "retention_rate": 0.1864406779661017, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.14285714285714285, "amount": 0.0, "status": 0.4166666666666667, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 760}, "2": {"facts_total": 59, "facts_retained": 11, "retention_rate": 0.1864406779661017, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.14285714285714285, "amount": 0.0, "status": 0.4166666666666667, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 760}, "3": {"facts_total": 59, "facts_retained": 11, "retention_rate": 0.1864406779661017, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.14285714285714285, "amount": 0.0, "status": 0.4166666666666667, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 760}, "4": {"facts_total": 59, "facts_retained": 11, "retention_rate": 0.1864406779661017, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.14285714285714285, "amount": 0.0, "status": 0.4166666666666667, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 760}, "5": {"facts_total": 59, "facts_retained": 11, "retention_rate": 0.1864406779661017, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.14285714285714285, "amount": 0.0, "status": 0.4166666666666667, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 760}, "6": {"facts_total": 59, "facts_retained": 11, "retention_rate": 0.1864406779661017, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.14285714285714285, "amount": 0.0, "status": 0.4166666666666667, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 760}, "7": {"facts_total": 59, "facts_retained": 11, "retention_rate": 0.1864406779661017, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.14285714285714285, "amount": 0.0, "status": 0.4166666666666667, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 760}, "8": {"facts_total": 59, "facts_retained": 11, "retention_rate": 0.1864406779661017, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.14285714285714285, "amount": 0.0, "status": 0.4166666666666667, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 760}}}, "reference_key_id": "keys/synthetic/syn_03012_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "bb16af6fe25c6b4abc9bdb0ad8c8b3d8b7d3dc12769f5feefdb7a2e4bb915494", "output_hash": "66ef6c5da682b38795e9029723bbd71cdde2c4dec70cf1fae6efe62ef243f10a"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "1a9afa0bff752e7333a4ce33", "attempt_id": "9826d3dcdbb0", "run_group_id": "E4_core:A5|closed|qwen3b:h8", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:35:26.517458Z", "task_family": "synthetic", "document_id": "syn_03012_2000", "document_hash": "03e441e68d54e1af036f8ff905dfd6cbbccead62636279c1a8e179352ab08945", "question_id": "q01_single_fact", "split": "test", "cluster_id": "syn_03012_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 8, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1927, "query_tokens": 437, "message_tokens": 348, "message_utf8_bytes": 824, "cache_tensor_bytes": null, "serialized_payload_bytes": 824, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 131507.60937598534, "hops": {"1": 22467.89562504273, "2": 15719.32450006716, "3": 15511.978083057329, "4": 15713.779207901098, "5": 15311.877917032689, "6": 15750.822708941996, "7": 15396.881458931603, "8": 15635.049875010736}, "receiver_prefill_ms": 284.04175001196563, "ttft_ms": 37.79862495139241, "decode_ms": 43.71374996844679}, "total_wall_ms": 131873.84454300627, "ttft_ms": 37.79862495139241, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 59, "facts_retained": 11, "retention_rate": 0.1864406779661017, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.14285714285714285, "amount": 0.0, "status": 0.4166666666666667, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 760}, "ledger_by_hop": {"1": {"facts_total": 59, "facts_retained": 11, "retention_rate": 0.1864406779661017, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.14285714285714285, "amount": 0.0, "status": 0.4166666666666667, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 760}, "2": {"facts_total": 59, "facts_retained": 11, "retention_rate": 0.1864406779661017, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.14285714285714285, "amount": 0.0, "status": 0.4166666666666667, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 760}, "3": {"facts_total": 59, "facts_retained": 11, "retention_rate": 0.1864406779661017, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.14285714285714285, "amount": 0.0, "status": 0.4166666666666667, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 760}, "4": {"facts_total": 59, "facts_retained": 11, "retention_rate": 0.1864406779661017, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.14285714285714285, "amount": 0.0, "status": 0.4166666666666667, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 760}, "5": {"facts_total": 59, "facts_retained": 11, "retention_rate": 0.1864406779661017, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.14285714285714285, "amount": 0.0, "status": 0.4166666666666667, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 760}, "6": {"facts_total": 59, "facts_retained": 11, "retention_rate": 0.1864406779661017, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.14285714285714285, "amount": 0.0, "status": 0.4166666666666667, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 760}, "7": {"facts_total": 59, "facts_retained": 11, "retention_rate": 0.1864406779661017, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.14285714285714285, "amount": 0.0, "status": 0.4166666666666667, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 760}, "8": {"facts_total": 59, "facts_retained": 11, "retention_rate": 0.1864406779661017, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.14285714285714285, "amount": 0.0, "status": 0.4166666666666667, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 760}}}, "reference_key_id": "keys/synthetic/syn_03012_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "bb16af6fe25c6b4abc9bdb0ad8c8b3d8b7d3dc12769f5feefdb7a2e4bb915494", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "71ea29fbf19e4ecf625493fd", "attempt_id": "6f8c2bb40526", "run_group_id": "E4_core:A5|closed|qwen3b:h8", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:35:26.901516Z", "task_family": "synthetic", "document_id": "syn_03012_2000", "document_hash": "03e441e68d54e1af036f8ff905dfd6cbbccead62636279c1a8e179352ab08945", "question_id": "q02_exact_copy", "split": "test", "cluster_id": "syn_03012_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 8, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1927, "query_tokens": 436, "message_tokens": 348, "message_utf8_bytes": 824, "cache_tensor_bytes": null, "serialized_payload_bytes": 824, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 131507.60937598534, "hops": {"1": 22467.89562504273, "2": 15719.32450006716, "3": 15511.978083057329, "4": 15713.779207901098, "5": 15311.877917032689, "6": 15750.822708941996, "7": 15396.881458931603, "8": 15635.049875010736}, "receiver_prefill_ms": 320.5278749810532, "ttft_ms": 109.2718750005588, "decode_ms": 138.63866601604968}, "total_wall_ms": 132076.55966700986, "ttft_ms": 109.2718750005588, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 59, "facts_retained": 11, "retention_rate": 0.1864406779661017, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.14285714285714285, "amount": 0.0, "status": 0.4166666666666667, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 760}, "ledger_by_hop": {"1": {"facts_total": 59, "facts_retained": 11, "retention_rate": 0.1864406779661017, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.14285714285714285, "amount": 0.0, "status": 0.4166666666666667, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 760}, "2": {"facts_total": 59, "facts_retained": 11, "retention_rate": 0.1864406779661017, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.14285714285714285, "amount": 0.0, "status": 0.4166666666666667, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 760}, "3": {"facts_total": 59, "facts_retained": 11, "retention_rate": 0.1864406779661017, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.14285714285714285, "amount": 0.0, "status": 0.4166666666666667, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 760}, "4": {"facts_total": 59, "facts_retained": 11, "retention_rate": 0.1864406779661017, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.14285714285714285, "amount": 0.0, "status": 0.4166666666666667, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 760}, "5": {"facts_total": 59, "facts_retained": 11, "retention_rate": 0.1864406779661017, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.14285714285714285, "amount": 0.0, "status": 0.4166666666666667, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 760}, "6": {"facts_total": 59, "facts_retained": 11, "retention_rate": 0.1864406779661017, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.14285714285714285, "amount": 0.0, "status": 0.4166666666666667, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 760}, "7": {"facts_total": 59, "facts_retained": 11, "retention_rate": 0.1864406779661017, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.14285714285714285, "amount": 0.0, "status": 0.4166666666666667, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 760}, "8": {"facts_total": 59, "facts_retained": 11, "retention_rate": 0.1864406779661017, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.14285714285714285, "amount": 0.0, "status": 0.4166666666666667, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 760}}}, "reference_key_id": "keys/synthetic/syn_03012_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "bb16af6fe25c6b4abc9bdb0ad8c8b3d8b7d3dc12769f5feefdb7a2e4bb915494", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "e55644215fb67496cf9b9299", "attempt_id": "dd7c2060efbe", "run_group_id": "E4_core:A5|closed|qwen3b:h8", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:35:27.487814Z", "task_family": "synthetic", "document_id": "syn_03012_2000", "document_hash": "03e441e68d54e1af036f8ff905dfd6cbbccead62636279c1a8e179352ab08945", "question_id": "q03_exact_copy", "split": "test", "cluster_id": "syn_03012_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 8, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1927, "query_tokens": 432, "message_tokens": 348, "message_utf8_bytes": 824, "cache_tensor_bytes": null, "serialized_payload_bytes": 824, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 131507.60937598534, "hops": {"1": 22467.89562504273, "2": 15719.32450006716, "3": 15511.978083057329, "4": 15713.779207901098, "5": 15311.877917032689, "6": 15750.822708941996, "7": 15396.881458931603, "8": 15635.049875010736}, "receiver_prefill_ms": 294.8984589893371, "ttft_ms": 40.99895793478936, "decode_ms": 30.672334018163383}, "total_wall_ms": 131874.63487591594, "ttft_ms": 40.99895793478936, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 59, "facts_retained": 11, "retention_rate": 0.1864406779661017, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.14285714285714285, "amount": 0.0, "status": 0.4166666666666667, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 760}, "ledger_by_hop": {"1": {"facts_total": 59, "facts_retained": 11, "retention_rate": 0.1864406779661017, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.14285714285714285, "amount": 0.0, "status": 0.4166666666666667, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 760}, "2": {"facts_total": 59, "facts_retained": 11, "retention_rate": 0.1864406779661017, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.14285714285714285, "amount": 0.0, "status": 0.4166666666666667, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 760}, "3": {"facts_total": 59, "facts_retained": 11, "retention_rate": 0.1864406779661017, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.14285714285714285, "amount": 0.0, "status": 0.4166666666666667, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 760}, "4": {"facts_total": 59, "facts_retained": 11, "retention_rate": 0.1864406779661017, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.14285714285714285, "amount": 0.0, "status": 0.4166666666666667, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 760}, "5": {"facts_total": 59, "facts_retained": 11, "retention_rate": 0.1864406779661017, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.14285714285714285, "amount": 0.0, "status": 0.4166666666666667, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 760}, "6": {"facts_total": 59, "facts_retained": 11, "retention_rate": 0.1864406779661017, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.14285714285714285, "amount": 0.0, "status": 0.4166666666666667, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 760}, "7": {"facts_total": 59, "facts_retained": 11, "retention_rate": 0.1864406779661017, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.14285714285714285, "amount": 0.0, "status": 0.4166666666666667, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 760}, "8": {"facts_total": 59, "facts_retained": 11, "retention_rate": 0.1864406779661017, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.14285714285714285, "amount": 0.0, "status": 0.4166666666666667, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 760}}}, "reference_key_id": "keys/synthetic/syn_03012_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "bb16af6fe25c6b4abc9bdb0ad8c8b3d8b7d3dc12769f5feefdb7a2e4bb915494", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "933a3395bcdc2b46a6483942", "attempt_id": "54be768ad958", "run_group_id": "E4_core:A5|closed|qwen3b:h8", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:35:27.872130Z", "task_family": "synthetic", "document_id": "syn_03012_2000", "document_hash": "03e441e68d54e1af036f8ff905dfd6cbbccead62636279c1a8e179352ab08945", "question_id": "q04_multi_hop", "split": "test", "cluster_id": "syn_03012_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 8, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1927, "query_tokens": 437, "message_tokens": 348, "message_utf8_bytes": 824, "cache_tensor_bytes": null, "serialized_payload_bytes": 824, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 131507.60937598534, "hops": {"1": 22467.89562504273, "2": 15719.32450006716, "3": 15511.978083057329, "4": 15713.779207901098, "5": 15311.877917032689, "6": 15750.822708941996, "7": 15396.881458931603, "8": 15635.049875010736}, "receiver_prefill_ms": 291.0020829876885, "ttft_ms": 43.25641703326255, "decode_ms": 30.043375096283853}, "total_wall_ms": 131872.34366696794, "ttft_ms": 43.25641703326255, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 59, "facts_retained": 11, "retention_rate": 0.1864406779661017, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.14285714285714285, "amount": 0.0, "status": 0.4166666666666667, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 760}, "ledger_by_hop": {"1": {"facts_total": 59, "facts_retained": 11, "retention_rate": 0.1864406779661017, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.14285714285714285, "amount": 0.0, "status": 0.4166666666666667, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 760}, "2": {"facts_total": 59, "facts_retained": 11, "retention_rate": 0.1864406779661017, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.14285714285714285, "amount": 0.0, "status": 0.4166666666666667, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 760}, "3": {"facts_total": 59, "facts_retained": 11, "retention_rate": 0.1864406779661017, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.14285714285714285, "amount": 0.0, "status": 0.4166666666666667, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 760}, "4": {"facts_total": 59, "facts_retained": 11, "retention_rate": 0.1864406779661017, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.14285714285714285, "amount": 0.0, "status": 0.4166666666666667, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 760}, "5": {"facts_total": 59, "facts_retained": 11, "retention_rate": 0.1864406779661017, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.14285714285714285, "amount": 0.0, "status": 0.4166666666666667, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 760}, "6": {"facts_total": 59, "facts_retained": 11, "retention_rate": 0.1864406779661017, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.14285714285714285, "amount": 0.0, "status": 0.4166666666666667, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 760}, "7": {"facts_total": 59, "facts_retained": 11, "retention_rate": 0.1864406779661017, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.14285714285714285, "amount": 0.0, "status": 0.4166666666666667, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 760}, "8": {"facts_total": 59, "facts_retained": 11, "retention_rate": 0.1864406779661017, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.14285714285714285, "amount": 0.0, "status": 0.4166666666666667, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 760}}}, "reference_key_id": "keys/synthetic/syn_03012_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "bb16af6fe25c6b4abc9bdb0ad8c8b3d8b7d3dc12769f5feefdb7a2e4bb915494", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "db951f756f41507171fe940b", "attempt_id": "dca4b46f3d7c", "run_group_id": "E4_core:A5|closed|qwen3b:h8", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:35:28.254051Z", "task_family": "synthetic", "document_id": "syn_03012_2000", "document_hash": "03e441e68d54e1af036f8ff905dfd6cbbccead62636279c1a8e179352ab08945", "question_id": "q05_multi_hop", "split": "test", "cluster_id": "syn_03012_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 8, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1927, "query_tokens": 440, "message_tokens": 348, "message_utf8_bytes": 824, "cache_tensor_bytes": null, "serialized_payload_bytes": 824, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 131507.60937598534, "hops": {"1": 22467.89562504273, "2": 15719.32450006716, "3": 15511.978083057329, "4": 15713.779207901098, "5": 15311.877917032689, "6": 15750.822708941996, "7": 15396.881458931603, "8": 15635.049875010736}, "receiver_prefill_ms": 274.2312910268083, "ttft_ms": 64.640500000678, "decode_ms": 61.61479197908193}, "total_wall_ms": 131908.63325097598, "ttft_ms": 64.640500000678, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 59, "facts_retained": 11, "retention_rate": 0.1864406779661017, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.14285714285714285, "amount": 0.0, "status": 0.4166666666666667, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 760}, "ledger_by_hop": {"1": {"facts_total": 59, "facts_retained": 11, "retention_rate": 0.1864406779661017, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.14285714285714285, "amount": 0.0, "status": 0.4166666666666667, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 760}, "2": {"facts_total": 59, "facts_retained": 11, "retention_rate": 0.1864406779661017, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.14285714285714285, "amount": 0.0, "status": 0.4166666666666667, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 760}, "3": {"facts_total": 59, "facts_retained": 11, "retention_rate": 0.1864406779661017, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.14285714285714285, "amount": 0.0, "status": 0.4166666666666667, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 760}, "4": {"facts_total": 59, "facts_retained": 11, "retention_rate": 0.1864406779661017, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.14285714285714285, "amount": 0.0, "status": 0.4166666666666667, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 760}, "5": {"facts_total": 59, "facts_retained": 11, "retention_rate": 0.1864406779661017, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.14285714285714285, "amount": 0.0, "status": 0.4166666666666667, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 760}, "6": {"facts_total": 59, "facts_retained": 11, "retention_rate": 0.1864406779661017, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.14285714285714285, "amount": 0.0, "status": 0.4166666666666667, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 760}, "7": {"facts_total": 59, "facts_retained": 11, "retention_rate": 0.1864406779661017, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.14285714285714285, "amount": 0.0, "status": 0.4166666666666667, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 760}, "8": {"facts_total": 59, "facts_retained": 11, "retention_rate": 0.1864406779661017, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.14285714285714285, "amount": 0.0, "status": 0.4166666666666667, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 760}}}, "reference_key_id": "keys/synthetic/syn_03012_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "bb16af6fe25c6b4abc9bdb0ad8c8b3d8b7d3dc12769f5feefdb7a2e4bb915494", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "0a2236538096dd032a50a5e8", "attempt_id": "0893457538a9", "run_group_id": "E4_core:A5|closed|qwen3b:h8", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:35:28.671912Z", "task_family": "synthetic", "document_id": "syn_03012_2000", "document_hash": "03e441e68d54e1af036f8ff905dfd6cbbccead62636279c1a8e179352ab08945", "question_id": "q06_negation", "split": "test", "cluster_id": "syn_03012_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 8, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1927, "query_tokens": 432, "message_tokens": 348, "message_utf8_bytes": 824, "cache_tensor_bytes": null, "serialized_payload_bytes": 824, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 131507.60937598534, "hops": {"1": 22467.89562504273, "2": 15719.32450006716, "3": 15511.978083057329, "4": 15713.779207901098, "5": 15311.877917032689, "6": 15750.822708941996, "7": 15396.881458931603, "8": 15635.049875010736}, "receiver_prefill_ms": 332.58308295626193, "ttft_ms": 128.16466600634158, "decode_ms": 38.35245897062123}, "total_wall_ms": 132007.3147090152, "ttft_ms": 128.16466600634158, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 59, "facts_retained": 11, "retention_rate": 0.1864406779661017, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.14285714285714285, "amount": 0.0, "status": 0.4166666666666667, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 760}, "ledger_by_hop": {"1": {"facts_total": 59, "facts_retained": 11, "retention_rate": 0.1864406779661017, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.14285714285714285, "amount": 0.0, "status": 0.4166666666666667, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 760}, "2": {"facts_total": 59, "facts_retained": 11, "retention_rate": 0.1864406779661017, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.14285714285714285, "amount": 0.0, "status": 0.4166666666666667, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 760}, "3": {"facts_total": 59, "facts_retained": 11, "retention_rate": 0.1864406779661017, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.14285714285714285, "amount": 0.0, "status": 0.4166666666666667, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 760}, "4": {"facts_total": 59, "facts_retained": 11, "retention_rate": 0.1864406779661017, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.14285714285714285, "amount": 0.0, "status": 0.4166666666666667, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 760}, "5": {"facts_total": 59, "facts_retained": 11, "retention_rate": 0.1864406779661017, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.14285714285714285, "amount": 0.0, "status": 0.4166666666666667, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 760}, "6": {"facts_total": 59, "facts_retained": 11, "retention_rate": 0.1864406779661017, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.14285714285714285, "amount": 0.0, "status": 0.4166666666666667, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 760}, "7": {"facts_total": 59, "facts_retained": 11, "retention_rate": 0.1864406779661017, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.14285714285714285, "amount": 0.0, "status": 0.4166666666666667, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 760}, "8": {"facts_total": 59, "facts_retained": 11, "retention_rate": 0.1864406779661017, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.14285714285714285, "amount": 0.0, "status": 0.4166666666666667, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 760}}}, "reference_key_id": "keys/synthetic/syn_03012_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "bb16af6fe25c6b4abc9bdb0ad8c8b3d8b7d3dc12769f5feefdb7a2e4bb915494", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "1cf251ea7dbd592368d76c72", "attempt_id": "48849b0b68f8", "run_group_id": "E4_core:A5|closed|qwen3b:h8", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:35:29.190504Z", "task_family": "synthetic", "document_id": "syn_03012_2000", "document_hash": "03e441e68d54e1af036f8ff905dfd6cbbccead62636279c1a8e179352ab08945", "question_id": "q07_negation", "split": "test", "cluster_id": "syn_03012_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 8, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1927, "query_tokens": 432, "message_tokens": 348, "message_utf8_bytes": 824, "cache_tensor_bytes": null, "serialized_payload_bytes": 824, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 131507.60937598534, "hops": {"1": 22467.89562504273, "2": 15719.32450006716, "3": 15511.978083057329, "4": 15713.779207901098, "5": 15311.877917032689, "6": 15750.822708941996, "7": 15396.881458931603, "8": 15635.049875010736}, "receiver_prefill_ms": 285.13266693335027, "ttft_ms": 36.62383393384516, "decode_ms": 38.21995900943875}, "total_wall_ms": 131868.1204179302, "ttft_ms": 36.62383393384516, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 59, "facts_retained": 11, "retention_rate": 0.1864406779661017, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.14285714285714285, "amount": 0.0, "status": 0.4166666666666667, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 760}, "ledger_by_hop": {"1": {"facts_total": 59, "facts_retained": 11, "retention_rate": 0.1864406779661017, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.14285714285714285, "amount": 0.0, "status": 0.4166666666666667, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 760}, "2": {"facts_total": 59, "facts_retained": 11, "retention_rate": 0.1864406779661017, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.14285714285714285, "amount": 0.0, "status": 0.4166666666666667, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 760}, "3": {"facts_total": 59, "facts_retained": 11, "retention_rate": 0.1864406779661017, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.14285714285714285, "amount": 0.0, "status": 0.4166666666666667, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 760}, "4": {"facts_total": 59, "facts_retained": 11, "retention_rate": 0.1864406779661017, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.14285714285714285, "amount": 0.0, "status": 0.4166666666666667, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 760}, "5": {"facts_total": 59, "facts_retained": 11, "retention_rate": 0.1864406779661017, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.14285714285714285, "amount": 0.0, "status": 0.4166666666666667, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 760}, "6": {"facts_total": 59, "facts_retained": 11, "retention_rate": 0.1864406779661017, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.14285714285714285, "amount": 0.0, "status": 0.4166666666666667, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 760}, "7": {"facts_total": 59, "facts_retained": 11, "retention_rate": 0.1864406779661017, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.14285714285714285, "amount": 0.0, "status": 0.4166666666666667, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 760}, "8": {"facts_total": 59, "facts_retained": 11, "retention_rate": 0.1864406779661017, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.14285714285714285, "amount": 0.0, "status": 0.4166666666666667, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 760}}}, "reference_key_id": "keys/synthetic/syn_03012_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "bb16af6fe25c6b4abc9bdb0ad8c8b3d8b7d3dc12769f5feefdb7a2e4bb915494", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "cec1ee2ded4dffa3c1d3a729", "attempt_id": "0d7fade8132c", "run_group_id": "E4_core:A5|closed|qwen3b:h8", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:35:29.568487Z", "task_family": "synthetic", "document_id": "syn_03012_2000", "document_hash": "03e441e68d54e1af036f8ff905dfd6cbbccead62636279c1a8e179352ab08945", "question_id": "q08_revision", "split": "test", "cluster_id": "syn_03012_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 8, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1927, "query_tokens": 447, "message_tokens": 348, "message_utf8_bytes": 824, "cache_tensor_bytes": null, "serialized_payload_bytes": 824, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 131507.60937598534, "hops": {"1": 22467.89562504273, "2": 15719.32450006716, "3": 15511.978083057329, "4": 15713.779207901098, "5": 15311.877917032689, "6": 15750.822708941996, "7": 15396.881458931603, "8": 15635.049875010736}, "receiver_prefill_ms": 286.5218750666827, "ttft_ms": 36.750541999936104, "decode_ms": 36.06562502682209}, "total_wall_ms": 131867.40254296456, "ttft_ms": 36.750541999936104, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 59, "facts_retained": 11, "retention_rate": 0.1864406779661017, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.14285714285714285, "amount": 0.0, "status": 0.4166666666666667, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 760}, "ledger_by_hop": {"1": {"facts_total": 59, "facts_retained": 11, "retention_rate": 0.1864406779661017, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.14285714285714285, "amount": 0.0, "status": 0.4166666666666667, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 760}, "2": {"facts_total": 59, "facts_retained": 11, "retention_rate": 0.1864406779661017, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.14285714285714285, "amount": 0.0, "status": 0.4166666666666667, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 760}, "3": {"facts_total": 59, "facts_retained": 11, "retention_rate": 0.1864406779661017, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.14285714285714285, "amount": 0.0, "status": 0.4166666666666667, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 760}, "4": {"facts_total": 59, "facts_retained": 11, "retention_rate": 0.1864406779661017, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.14285714285714285, "amount": 0.0, "status": 0.4166666666666667, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 760}, "5": {"facts_total": 59, "facts_retained": 11, "retention_rate": 0.1864406779661017, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.14285714285714285, "amount": 0.0, "status": 0.4166666666666667, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 760}, "6": {"facts_total": 59, "facts_retained": 11, "retention_rate": 0.1864406779661017, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.14285714285714285, "amount": 0.0, "status": 0.4166666666666667, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 760}, "7": {"facts_total": 59, "facts_retained": 11, "retention_rate": 0.1864406779661017, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.14285714285714285, "amount": 0.0, "status": 0.4166666666666667, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 760}, "8": {"facts_total": 59, "facts_retained": 11, "retention_rate": 0.1864406779661017, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.14285714285714285, "amount": 0.0, "status": 0.4166666666666667, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 760}}}, "reference_key_id": "keys/synthetic/syn_03012_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "bb16af6fe25c6b4abc9bdb0ad8c8b3d8b7d3dc12769f5feefdb7a2e4bb915494", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "636666dd51673f6625d2eba2", "attempt_id": "47d3dba42924", "run_group_id": "E4_core:A5|closed|qwen3b:h8", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:35:29.945774Z", "task_family": "synthetic", "document_id": "syn_03012_2000", "document_hash": "03e441e68d54e1af036f8ff905dfd6cbbccead62636279c1a8e179352ab08945", "question_id": "q09_revision", "split": "test", "cluster_id": "syn_03012_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 8, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1927, "query_tokens": 440, "message_tokens": 348, "message_utf8_bytes": 824, "cache_tensor_bytes": null, "serialized_payload_bytes": 824, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 131507.60937598534, "hops": {"1": 22467.89562504273, "2": 15719.32450006716, "3": 15511.978083057329, "4": 15713.779207901098, "5": 15311.877917032689, "6": 15750.822708941996, "7": 15396.881458931603, "8": 15635.049875010736}, "receiver_prefill_ms": 276.9895839737728, "ttft_ms": 63.75116703566164, "decode_ms": 53.20533306803554}, "total_wall_ms": 131902.11700100917, "ttft_ms": 63.75116703566164, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 59, "facts_retained": 11, "retention_rate": 0.1864406779661017, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.14285714285714285, "amount": 0.0, "status": 0.4166666666666667, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 760}, "ledger_by_hop": {"1": {"facts_total": 59, "facts_retained": 11, "retention_rate": 0.1864406779661017, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.14285714285714285, "amount": 0.0, "status": 0.4166666666666667, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 760}, "2": {"facts_total": 59, "facts_retained": 11, "retention_rate": 0.1864406779661017, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.14285714285714285, "amount": 0.0, "status": 0.4166666666666667, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 760}, "3": {"facts_total": 59, "facts_retained": 11, "retention_rate": 0.1864406779661017, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.14285714285714285, "amount": 0.0, "status": 0.4166666666666667, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 760}, "4": {"facts_total": 59, "facts_retained": 11, "retention_rate": 0.1864406779661017, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.14285714285714285, "amount": 0.0, "status": 0.4166666666666667, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 760}, "5": {"facts_total": 59, "facts_retained": 11, "retention_rate": 0.1864406779661017, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.14285714285714285, "amount": 0.0, "status": 0.4166666666666667, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 760}, "6": {"facts_total": 59, "facts_retained": 11, "retention_rate": 0.1864406779661017, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.14285714285714285, "amount": 0.0, "status": 0.4166666666666667, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 760}, "7": {"facts_total": 59, "facts_retained": 11, "retention_rate": 0.1864406779661017, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.14285714285714285, "amount": 0.0, "status": 0.4166666666666667, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 760}, "8": {"facts_total": 59, "facts_retained": 11, "retention_rate": 0.1864406779661017, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.14285714285714285, "amount": 0.0, "status": 0.4166666666666667, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 760}}}, "reference_key_id": "keys/synthetic/syn_03012_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "bb16af6fe25c6b4abc9bdb0ad8c8b3d8b7d3dc12769f5feefdb7a2e4bb915494", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "304bbb5dcf8d5263fc5b56cd", "attempt_id": "48fa66fe48e4", "run_group_id": "E4_core:A5|closed|qwen3b:h8", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:35:30.358654Z", "task_family": "synthetic", "document_id": "syn_03012_2000", "document_hash": "03e441e68d54e1af036f8ff905dfd6cbbccead62636279c1a8e179352ab08945", "question_id": "q10_temporal", "split": "test", "cluster_id": "syn_03012_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 8, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1927, "query_tokens": 450, "message_tokens": 348, "message_utf8_bytes": 824, "cache_tensor_bytes": null, "serialized_payload_bytes": 824, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 131507.60937598534, "hops": {"1": 22467.89562504273, "2": 15719.32450006716, "3": 15511.978083057329, "4": 15713.779207901098, "5": 15311.877917032689, "6": 15750.822708941996, "7": 15396.881458931603, "8": 15635.049875010736}, "receiver_prefill_ms": 343.6065419809893, "ttft_ms": 123.23649995960295, "decode_ms": 47.75583301670849}, "total_wall_ms": 132022.9017090751, "ttft_ms": 123.23649995960295, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 59, "facts_retained": 11, "retention_rate": 0.1864406779661017, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.14285714285714285, "amount": 0.0, "status": 0.4166666666666667, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 760}, "ledger_by_hop": {"1": {"facts_total": 59, "facts_retained": 11, "retention_rate": 0.1864406779661017, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.14285714285714285, "amount": 0.0, "status": 0.4166666666666667, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 760}, "2": {"facts_total": 59, "facts_retained": 11, "retention_rate": 0.1864406779661017, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.14285714285714285, "amount": 0.0, "status": 0.4166666666666667, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 760}, "3": {"facts_total": 59, "facts_retained": 11, "retention_rate": 0.1864406779661017, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.14285714285714285, "amount": 0.0, "status": 0.4166666666666667, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 760}, "4": {"facts_total": 59, "facts_retained": 11, "retention_rate": 0.1864406779661017, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.14285714285714285, "amount": 0.0, "status": 0.4166666666666667, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 760}, "5": {"facts_total": 59, "facts_retained": 11, "retention_rate": 0.1864406779661017, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.14285714285714285, "amount": 0.0, "status": 0.4166666666666667, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 760}, "6": {"facts_total": 59, "facts_retained": 11, "retention_rate": 0.1864406779661017, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.14285714285714285, "amount": 0.0, "status": 0.4166666666666667, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 760}, "7": {"facts_total": 59, "facts_retained": 11, "retention_rate": 0.1864406779661017, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.14285714285714285, "amount": 0.0, "status": 0.4166666666666667, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 760}, "8": {"facts_total": 59, "facts_retained": 11, "retention_rate": 0.1864406779661017, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.14285714285714285, "amount": 0.0, "status": 0.4166666666666667, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 760}}}, "reference_key_id": "keys/synthetic/syn_03012_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "bb16af6fe25c6b4abc9bdb0ad8c8b3d8b7d3dc12769f5feefdb7a2e4bb915494", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "ad94d415818e2f8414aa4ee5", "attempt_id": "ddd70a94630a", "run_group_id": "E4_core:A5|closed|qwen3b:h8", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:35:30.892028Z", "task_family": "synthetic", "document_id": "syn_03012_2000", "document_hash": "03e441e68d54e1af036f8ff905dfd6cbbccead62636279c1a8e179352ab08945", "question_id": "q11_aggregation", "split": "test", "cluster_id": "syn_03012_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 8, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1927, "query_tokens": 442, "message_tokens": 348, "message_utf8_bytes": 824, "cache_tensor_bytes": null, "serialized_payload_bytes": 824, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 131507.60937598534, "hops": {"1": 22467.89562504273, "2": 15719.32450006716, "3": 15511.978083057329, "4": 15713.779207901098, "5": 15311.877917032689, "6": 15750.822708941996, "7": 15396.881458931603, "8": 15635.049875010736}, "receiver_prefill_ms": 282.91583398822695, "ttft_ms": 35.594332963228226, "decode_ms": 45.500332955271006}, "total_wall_ms": 131872.10054206662, "ttft_ms": 35.594332963228226, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 59, "facts_retained": 11, "retention_rate": 0.1864406779661017, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.14285714285714285, "amount": 0.0, "status": 0.4166666666666667, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 760}, "ledger_by_hop": {"1": {"facts_total": 59, "facts_retained": 11, "retention_rate": 0.1864406779661017, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.14285714285714285, "amount": 0.0, "status": 0.4166666666666667, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 760}, "2": {"facts_total": 59, "facts_retained": 11, "retention_rate": 0.1864406779661017, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.14285714285714285, "amount": 0.0, "status": 0.4166666666666667, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 760}, "3": {"facts_total": 59, "facts_retained": 11, "retention_rate": 0.1864406779661017, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.14285714285714285, "amount": 0.0, "status": 0.4166666666666667, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 760}, "4": {"facts_total": 59, "facts_retained": 11, "retention_rate": 0.1864406779661017, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.14285714285714285, "amount": 0.0, "status": 0.4166666666666667, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 760}, "5": {"facts_total": 59, "facts_retained": 11, "retention_rate": 0.1864406779661017, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.14285714285714285, "amount": 0.0, "status": 0.4166666666666667, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 760}, "6": {"facts_total": 59, "facts_retained": 11, "retention_rate": 0.1864406779661017, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.14285714285714285, "amount": 0.0, "status": 0.4166666666666667, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 760}, "7": {"facts_total": 59, "facts_retained": 11, "retention_rate": 0.1864406779661017, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.14285714285714285, "amount": 0.0, "status": 0.4166666666666667, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 760}, "8": {"facts_total": 59, "facts_retained": 11, "retention_rate": 0.1864406779661017, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.14285714285714285, "amount": 0.0, "status": 0.4166666666666667, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 760}}}, "reference_key_id": "keys/synthetic/syn_03012_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "bb16af6fe25c6b4abc9bdb0ad8c8b3d8b7d3dc12769f5feefdb7a2e4bb915494", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "4579d4b73759eda06d5d2d7c", "attempt_id": "817ccc09c543", "run_group_id": "E4_core:A5|closed|qwen3b:h8", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:35:31.273779Z", "task_family": "synthetic", "document_id": "syn_03012_2000", "document_hash": "03e441e68d54e1af036f8ff905dfd6cbbccead62636279c1a8e179352ab08945", "question_id": "q12_unit_conversion", "split": "test", "cluster_id": "syn_03012_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 8, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1927, "query_tokens": 439, "message_tokens": 348, "message_utf8_bytes": 824, "cache_tensor_bytes": null, "serialized_payload_bytes": 824, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 131507.60937598534, "hops": {"1": 22467.89562504273, "2": 15719.32450006716, "3": 15511.978083057329, "4": 15713.779207901098, "5": 15311.877917032689, "6": 15750.822708941996, "7": 15396.881458931603, "8": 15635.049875010736}, "receiver_prefill_ms": 287.220332887955, "ttft_ms": 35.535375005565584, "decode_ms": 142.02137500979006}, "total_wall_ms": 131972.85416792147, "ttft_ms": 35.535375005565584, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 5}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "44620", "evidence_artifact": {"ledger": {"facts_total": 59, "facts_retained": 11, "retention_rate": 0.1864406779661017, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.14285714285714285, "amount": 0.0, "status": 0.4166666666666667, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 760}, "ledger_by_hop": {"1": {"facts_total": 59, "facts_retained": 11, "retention_rate": 0.1864406779661017, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.14285714285714285, "amount": 0.0, "status": 0.4166666666666667, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 760}, "2": {"facts_total": 59, "facts_retained": 11, "retention_rate": 0.1864406779661017, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.14285714285714285, "amount": 0.0, "status": 0.4166666666666667, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 760}, "3": {"facts_total": 59, "facts_retained": 11, "retention_rate": 0.1864406779661017, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.14285714285714285, "amount": 0.0, "status": 0.4166666666666667, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 760}, "4": {"facts_total": 59, "facts_retained": 11, "retention_rate": 0.1864406779661017, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.14285714285714285, "amount": 0.0, "status": 0.4166666666666667, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 760}, "5": {"facts_total": 59, "facts_retained": 11, "retention_rate": 0.1864406779661017, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.14285714285714285, "amount": 0.0, "status": 0.4166666666666667, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 760}, "6": {"facts_total": 59, "facts_retained": 11, "retention_rate": 0.1864406779661017, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.14285714285714285, "amount": 0.0, "status": 0.4166666666666667, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 760}, "7": {"facts_total": 59, "facts_retained": 11, "retention_rate": 0.1864406779661017, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.14285714285714285, "amount": 0.0, "status": 0.4166666666666667, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 760}, "8": {"facts_total": 59, "facts_retained": 11, "retention_rate": 0.1864406779661017, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.14285714285714285, "amount": 0.0, "status": 0.4166666666666667, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 760}}}, "reference_key_id": "keys/synthetic/syn_03012_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "bb16af6fe25c6b4abc9bdb0ad8c8b3d8b7d3dc12769f5feefdb7a2e4bb915494", "output_hash": "2bd125847727be7e7c0a656ee4c81f3bc6d0fef11f536f50ce7bc83688907075"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "c2a800380c9c5d1bae273fd9", "attempt_id": "15fcefe00c63", "run_group_id": "E4_core:A5|closed|qwen3b:h8", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:35:31.755680Z", "task_family": "synthetic", "document_id": "syn_03012_2000", "document_hash": "03e441e68d54e1af036f8ff905dfd6cbbccead62636279c1a8e179352ab08945", "question_id": "q13_unanswerable", "split": "test", "cluster_id": "syn_03012_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 8, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1927, "query_tokens": 444, "message_tokens": 348, "message_utf8_bytes": 824, "cache_tensor_bytes": null, "serialized_payload_bytes": 824, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 131507.60937598534, "hops": {"1": 22467.89562504273, "2": 15719.32450006716, "3": 15511.978083057329, "4": 15713.779207901098, "5": 15311.877917032689, "6": 15750.822708941996, "7": 15396.881458931603, "8": 15635.049875010736}, "receiver_prefill_ms": 323.4367499826476, "ttft_ms": 52.99112503416836, "decode_ms": 44.915333040989935}, "total_wall_ms": 131929.86870999448, "ttft_ms": 52.99112503416836, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 59, "facts_retained": 11, "retention_rate": 0.1864406779661017, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.14285714285714285, "amount": 0.0, "status": 0.4166666666666667, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 760}, "ledger_by_hop": {"1": {"facts_total": 59, "facts_retained": 11, "retention_rate": 0.1864406779661017, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.14285714285714285, "amount": 0.0, "status": 0.4166666666666667, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 760}, "2": {"facts_total": 59, "facts_retained": 11, "retention_rate": 0.1864406779661017, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.14285714285714285, "amount": 0.0, "status": 0.4166666666666667, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 760}, "3": {"facts_total": 59, "facts_retained": 11, "retention_rate": 0.1864406779661017, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.14285714285714285, "amount": 0.0, "status": 0.4166666666666667, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 760}, "4": {"facts_total": 59, "facts_retained": 11, "retention_rate": 0.1864406779661017, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.14285714285714285, "amount": 0.0, "status": 0.4166666666666667, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 760}, "5": {"facts_total": 59, "facts_retained": 11, "retention_rate": 0.1864406779661017, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.14285714285714285, "amount": 0.0, "status": 0.4166666666666667, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 760}, "6": {"facts_total": 59, "facts_retained": 11, "retention_rate": 0.1864406779661017, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.14285714285714285, "amount": 0.0, "status": 0.4166666666666667, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 760}, "7": {"facts_total": 59, "facts_retained": 11, "retention_rate": 0.1864406779661017, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.14285714285714285, "amount": 0.0, "status": 0.4166666666666667, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 760}, "8": {"facts_total": 59, "facts_retained": 11, "retention_rate": 0.1864406779661017, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.14285714285714285, "amount": 0.0, "status": 0.4166666666666667, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 760}}}, "reference_key_id": "keys/synthetic/syn_03012_2000.json", "task_scores": {"correct": 1.0, "abstained": 1.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "bb16af6fe25c6b4abc9bdb0ad8c8b3d8b7d3dc12769f5feefdb7a2e4bb915494", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "d26b2f009c4914cc579895d1", "attempt_id": "9d07f02febb7", "run_group_id": "E4_core:A5|closed|qwen3b:h8", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:35:32.197747Z", "task_family": "synthetic", "document_id": "syn_03012_2000", "document_hash": "03e441e68d54e1af036f8ff905dfd6cbbccead62636279c1a8e179352ab08945", "question_id": "q14_unanswerable", "split": "test", "cluster_id": "syn_03012_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 8, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1927, "query_tokens": 444, "message_tokens": 348, "message_utf8_bytes": 824, "cache_tensor_bytes": null, "serialized_payload_bytes": 824, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 131507.60937598534, "hops": {"1": 22467.89562504273, "2": 15719.32450006716, "3": 15511.978083057329, "4": 15713.779207901098, "5": 15311.877917032689, "6": 15750.822708941996, "7": 15396.881458931603, "8": 15635.049875010736}, "receiver_prefill_ms": 425.8612919365987, "ttft_ms": 30.42370791081339, "decode_ms": 40.55475001223385}, "total_wall_ms": 132005.00116799958, "ttft_ms": 30.42370791081339, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 59, "facts_retained": 11, "retention_rate": 0.1864406779661017, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.14285714285714285, "amount": 0.0, "status": 0.4166666666666667, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 760}, "ledger_by_hop": {"1": {"facts_total": 59, "facts_retained": 11, "retention_rate": 0.1864406779661017, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.14285714285714285, "amount": 0.0, "status": 0.4166666666666667, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 760}, "2": {"facts_total": 59, "facts_retained": 11, "retention_rate": 0.1864406779661017, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.14285714285714285, "amount": 0.0, "status": 0.4166666666666667, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 760}, "3": {"facts_total": 59, "facts_retained": 11, "retention_rate": 0.1864406779661017, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.14285714285714285, "amount": 0.0, "status": 0.4166666666666667, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 760}, "4": {"facts_total": 59, "facts_retained": 11, "retention_rate": 0.1864406779661017, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.14285714285714285, "amount": 0.0, "status": 0.4166666666666667, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 760}, "5": {"facts_total": 59, "facts_retained": 11, "retention_rate": 0.1864406779661017, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.14285714285714285, "amount": 0.0, "status": 0.4166666666666667, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 760}, "6": {"facts_total": 59, "facts_retained": 11, "retention_rate": 0.1864406779661017, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.14285714285714285, "amount": 0.0, "status": 0.4166666666666667, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 760}, "7": {"facts_total": 59, "facts_retained": 11, "retention_rate": 0.1864406779661017, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.14285714285714285, "amount": 0.0, "status": 0.4166666666666667, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 760}, "8": {"facts_total": 59, "facts_retained": 11, "retention_rate": 0.1864406779661017, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.14285714285714285, "amount": 0.0, "status": 0.4166666666666667, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 760}}}, "reference_key_id": "keys/synthetic/syn_03012_2000.json", "task_scores": {"correct": 1.0, "abstained": 1.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "bb16af6fe25c6b4abc9bdb0ad8c8b3d8b7d3dc12769f5feefdb7a2e4bb915494", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "3d5c5e49e58690605e67e6dd", "attempt_id": "c4a32155a24c", "run_group_id": "E4_core:A4|repairable|qwen3b:h1", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:35:55.543987Z", "task_family": "synthetic", "document_id": "syn_03012_2000", "document_hash": "03e441e68d54e1af036f8ff905dfd6cbbccead62636279c1a8e179352ab08945", "question_id": "q00_single_fact", "split": "test", "cluster_id": "syn_03012_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 1, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1927, "query_tokens": 594, "message_tokens": 512, "message_utf8_bytes": 1213, "cache_tensor_bytes": null, "serialized_payload_bytes": 1213, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 22829.28358297795, "hops": {"1": 22829.28358297795}, "receiver_prefill_ms": 553.606666973792, "ttft_ms": 40.31699988991022, "decode_ms": 357.60541702620685}, "total_wall_ms": 23781.551666907035, "ttft_ms": 40.31699988991022, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 10}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "XX-71M-6213", "evidence_artifact": {"ledger": {"facts_total": 59, "facts_retained": 30, "retention_rate": 0.5084745762711864, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.5714285714285714, "amount": 0.6, "status": 0.5833333333333334, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1213}, "ledger_by_hop": {"1": {"facts_total": 59, "facts_retained": 30, "retention_rate": 0.5084745762711864, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.5714285714285714, "amount": 0.6, "status": 0.5833333333333334, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1213}}}, "reference_key_id": "keys/synthetic/syn_03012_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "7c9b236b590faff2030e02587d972cf755ac2a88072e709174edbf405498973d", "output_hash": "66ef6c5da682b38795e9029723bbd71cdde2c4dec70cf1fae6efe62ef243f10a"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "6d32106756e27a69df4930da", "attempt_id": "bf22157251be", "run_group_id": "E4_core:A4|repairable|qwen3b:h1", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:35:56.513880Z", "task_family": "synthetic", "document_id": "syn_03012_2000", "document_hash": "03e441e68d54e1af036f8ff905dfd6cbbccead62636279c1a8e179352ab08945", "question_id": "q01_single_fact", "split": "test", "cluster_id": "syn_03012_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 1, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1927, "query_tokens": 600, "message_tokens": 512, "message_utf8_bytes": 1213, "cache_tensor_bytes": null, "serialized_payload_bytes": 1213, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 22829.28358297795, "hops": {"1": 22829.28358297795}, "receiver_prefill_ms": 411.01633303333074, "ttft_ms": 26.38312499038875, "decode_ms": 20.20374999847263}, "total_wall_ms": 23287.523124017753, "ttft_ms": 26.38312499038875, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 59, "facts_retained": 30, "retention_rate": 0.5084745762711864, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.5714285714285714, "amount": 0.6, "status": 0.5833333333333334, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1213}, "ledger_by_hop": {"1": {"facts_total": 59, "facts_retained": 30, "retention_rate": 0.5084745762711864, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.5714285714285714, "amount": 0.6, "status": 0.5833333333333334, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1213}}}, "reference_key_id": "keys/synthetic/syn_03012_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "7c9b236b590faff2030e02587d972cf755ac2a88072e709174edbf405498973d", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "13a65f4deaac0c1f15a42905", "attempt_id": "e557dc876ece", "run_group_id": "E4_core:A4|repairable|qwen3b:h1", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:35:56.990238Z", "task_family": "synthetic", "document_id": "syn_03012_2000", "document_hash": "03e441e68d54e1af036f8ff905dfd6cbbccead62636279c1a8e179352ab08945", "question_id": "q02_exact_copy", "split": "test", "cluster_id": "syn_03012_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 1, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1927, "query_tokens": 599, "message_tokens": 512, "message_utf8_bytes": 1213, "cache_tensor_bytes": null, "serialized_payload_bytes": 1213, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 22829.28358297795, "hops": {"1": 22829.28358297795}, "receiver_prefill_ms": 393.8264169264585, "ttft_ms": 66.33937498554587, "decode_ms": 412.5542499823496}, "total_wall_ms": 23702.782916021533, "ttft_ms": 66.33937498554587, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 7}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "13.4343", "evidence_artifact": {"ledger": {"facts_total": 59, "facts_retained": 30, "retention_rate": 0.5084745762711864, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.5714285714285714, "amount": 0.6, "status": 0.5833333333333334, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1213}, "ledger_by_hop": {"1": {"facts_total": 59, "facts_retained": 30, "retention_rate": 0.5084745762711864, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.5714285714285714, "amount": 0.6, "status": 0.5833333333333334, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1213}}}, "reference_key_id": "keys/synthetic/syn_03012_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "7c9b236b590faff2030e02587d972cf755ac2a88072e709174edbf405498973d", "output_hash": "599bc22e373b1cf7b4679208eb9e515ad0e8dcb88dde0c7f579637ec7d3f3014"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "d8173c8984eda48230109173", "attempt_id": "a6b0716a7f0e", "run_group_id": "E4_core:A4|repairable|qwen3b:h1", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:35:57.883520Z", "task_family": "synthetic", "document_id": "syn_03012_2000", "document_hash": "03e441e68d54e1af036f8ff905dfd6cbbccead62636279c1a8e179352ab08945", "question_id": "q03_exact_copy", "split": "test", "cluster_id": "syn_03012_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 1, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1927, "query_tokens": 595, "message_tokens": 512, "message_utf8_bytes": 1213, "cache_tensor_bytes": null, "serialized_payload_bytes": 1213, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 22829.28358297795, "hops": {"1": 22829.28358297795}, "receiver_prefill_ms": 408.2940419903025, "ttft_ms": 37.186124944128096, "decode_ms": 30.00995807815343}, "total_wall_ms": 23305.398707976565, "ttft_ms": 37.186124944128096, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 59, "facts_retained": 30, "retention_rate": 0.5084745762711864, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.5714285714285714, "amount": 0.6, "status": 0.5833333333333334, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1213}, "ledger_by_hop": {"1": {"facts_total": 59, "facts_retained": 30, "retention_rate": 0.5084745762711864, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.5714285714285714, "amount": 0.6, "status": 0.5833333333333334, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1213}}}, "reference_key_id": "keys/synthetic/syn_03012_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "7c9b236b590faff2030e02587d972cf755ac2a88072e709174edbf405498973d", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "e93a22a6e7e80aedac877a09", "attempt_id": "45c441ae7d96", "run_group_id": "E4_core:A4|repairable|qwen3b:h1", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:35:58.376795Z", "task_family": "synthetic", "document_id": "syn_03012_2000", "document_hash": "03e441e68d54e1af036f8ff905dfd6cbbccead62636279c1a8e179352ab08945", "question_id": "q04_multi_hop", "split": "test", "cluster_id": "syn_03012_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 1, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1927, "query_tokens": 600, "message_tokens": 512, "message_utf8_bytes": 1213, "cache_tensor_bytes": null, "serialized_payload_bytes": 1213, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 22829.28358297795, "hops": {"1": 22829.28358297795}, "receiver_prefill_ms": 369.4954590173438, "ttft_ms": 36.14216693677008, "decode_ms": 504.50787495356053}, "total_wall_ms": 23740.110832965, "ttft_ms": 36.14216693677008, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 7}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "8631552", "evidence_artifact": {"ledger": {"facts_total": 59, "facts_retained": 30, "retention_rate": 0.5084745762711864, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.5714285714285714, "amount": 0.6, "status": 0.5833333333333334, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1213}, "ledger_by_hop": {"1": {"facts_total": 59, "facts_retained": 30, "retention_rate": 0.5084745762711864, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.5714285714285714, "amount": 0.6, "status": 0.5833333333333334, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1213}}}, "reference_key_id": "keys/synthetic/syn_03012_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "7c9b236b590faff2030e02587d972cf755ac2a88072e709174edbf405498973d", "output_hash": "147d02c0587a77b3b53f0424c8d3d1454aaf2639ce9fdaf69af20db33363e849"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "2222d36c7e86879c42f236c4", "attempt_id": "995ed986d1aa", "run_group_id": "E4_core:A4|repairable|qwen3b:h1", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:35:59.305059Z", "task_family": "synthetic", "document_id": "syn_03012_2000", "document_hash": "03e441e68d54e1af036f8ff905dfd6cbbccead62636279c1a8e179352ab08945", "question_id": "q05_multi_hop", "split": "test", "cluster_id": "syn_03012_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 1, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1927, "query_tokens": 603, "message_tokens": 512, "message_utf8_bytes": 1213, "cache_tensor_bytes": null, "serialized_payload_bytes": 1213, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 22829.28358297795, "hops": {"1": 22829.28358297795}, "receiver_prefill_ms": 392.8140830248594, "ttft_ms": 21.75924996845424, "decode_ms": 331.5203750971705}, "total_wall_ms": 23576.070290990174, "ttft_ms": 21.75924996845424, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 10}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "DG-40E-2777", "evidence_artifact": {"ledger": {"facts_total": 59, "facts_retained": 30, "retention_rate": 0.5084745762711864, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.5714285714285714, "amount": 0.6, "status": 0.5833333333333334, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1213}, "ledger_by_hop": {"1": {"facts_total": 59, "facts_retained": 30, "retention_rate": 0.5084745762711864, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.5714285714285714, "amount": 0.6, "status": 0.5833333333333334, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1213}}}, "reference_key_id": "keys/synthetic/syn_03012_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "7c9b236b590faff2030e02587d972cf755ac2a88072e709174edbf405498973d", "output_hash": "16a7e3f4a7cc4a9b66f86c2cdcfad3b35daf1da199d9406e0627c8af2d8d73f3"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "a8f9c69380e171fbcc938788", "attempt_id": "dda843c18013", "run_group_id": "E4_core:A4|repairable|qwen3b:h1", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:36:00.069345Z", "task_family": "synthetic", "document_id": "syn_03012_2000", "document_hash": "03e441e68d54e1af036f8ff905dfd6cbbccead62636279c1a8e179352ab08945", "question_id": "q06_negation", "split": "test", "cluster_id": "syn_03012_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 1, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1927, "query_tokens": 595, "message_tokens": 512, "message_utf8_bytes": 1213, "cache_tensor_bytes": null, "serialized_payload_bytes": 1213, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 22829.28358297795, "hops": {"1": 22829.28358297795}, "receiver_prefill_ms": 357.3586670681834, "ttft_ms": 50.341540947556496, "decode_ms": 38.31070789601654}, "total_wall_ms": 23276.357540977187, "ttft_ms": 50.341540947556496, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "no", "evidence_artifact": {"ledger": {"facts_total": 59, "facts_retained": 30, "retention_rate": 0.5084745762711864, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.5714285714285714, "amount": 0.6, "status": 0.5833333333333334, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1213}, "ledger_by_hop": {"1": {"facts_total": 59, "facts_retained": 30, "retention_rate": 0.5084745762711864, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.5714285714285714, "amount": 0.6, "status": 0.5833333333333334, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1213}}}, "reference_key_id": "keys/synthetic/syn_03012_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "7c9b236b590faff2030e02587d972cf755ac2a88072e709174edbf405498973d", "output_hash": "9390298f3fb0c5b160498935d79cb139aef28e1c47358b4bbba61862b9c26e59"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "d62e17fed6e323197c981341", "attempt_id": "320899a94505", "run_group_id": "E4_core:A4|repairable|qwen3b:h1", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:36:00.534717Z", "task_family": "synthetic", "document_id": "syn_03012_2000", "document_hash": "03e441e68d54e1af036f8ff905dfd6cbbccead62636279c1a8e179352ab08945", "question_id": "q07_negation", "split": "test", "cluster_id": "syn_03012_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 1, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1927, "query_tokens": 595, "message_tokens": 512, "message_utf8_bytes": 1213, "cache_tensor_bytes": null, "serialized_payload_bytes": 1213, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 22829.28358297795, "hops": {"1": 22829.28358297795}, "receiver_prefill_ms": 542.8633750416338, "ttft_ms": 33.42641610652208, "decode_ms": 29.911499936133623}, "total_wall_ms": 23436.14516593516, "ttft_ms": 33.42641610652208, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "no", "evidence_artifact": {"ledger": {"facts_total": 59, "facts_retained": 30, "retention_rate": 0.5084745762711864, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.5714285714285714, "amount": 0.6, "status": 0.5833333333333334, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1213}, "ledger_by_hop": {"1": {"facts_total": 59, "facts_retained": 30, "retention_rate": 0.5084745762711864, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.5714285714285714, "amount": 0.6, "status": 0.5833333333333334, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1213}}}, "reference_key_id": "keys/synthetic/syn_03012_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "7c9b236b590faff2030e02587d972cf755ac2a88072e709174edbf405498973d", "output_hash": "9390298f3fb0c5b160498935d79cb139aef28e1c47358b4bbba61862b9c26e59"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "75d5c4c142364ff5db027fe4", "attempt_id": "3778052172df", "run_group_id": "E4_core:A4|repairable|qwen3b:h1", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:36:01.158778Z", "task_family": "synthetic", "document_id": "syn_03012_2000", "document_hash": "03e441e68d54e1af036f8ff905dfd6cbbccead62636279c1a8e179352ab08945", "question_id": "q08_revision", "split": "test", "cluster_id": "syn_03012_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 1, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1927, "query_tokens": 610, "message_tokens": 512, "message_utf8_bytes": 1213, "cache_tensor_bytes": null, "serialized_payload_bytes": 1213, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 22829.28358297795, "hops": {"1": 22829.28358297795}, "receiver_prefill_ms": 402.37949998117983, "ttft_ms": 24.089250015094876, "decode_ms": 234.8184590227902}, "total_wall_ms": 23491.216874914244, "ttft_ms": 24.089250015094876, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 7}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "6602245", "evidence_artifact": {"ledger": {"facts_total": 59, "facts_retained": 30, "retention_rate": 0.5084745762711864, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.5714285714285714, "amount": 0.6, "status": 0.5833333333333334, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1213}, "ledger_by_hop": {"1": {"facts_total": 59, "facts_retained": 30, "retention_rate": 0.5084745762711864, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.5714285714285714, "amount": 0.6, "status": 0.5833333333333334, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1213}}}, "reference_key_id": "keys/synthetic/syn_03012_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "7c9b236b590faff2030e02587d972cf755ac2a88072e709174edbf405498973d", "output_hash": "d117fa8197988f8d1e4fabc6c4cdaab9516018e54553b57c63b875b3de82b16f"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "60b6b46b402670562805c56a", "attempt_id": "54ee3da1a245", "run_group_id": "E4_core:A4|repairable|qwen3b:h1", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:36:01.838012Z", "task_family": "synthetic", "document_id": "syn_03012_2000", "document_hash": "03e441e68d54e1af036f8ff905dfd6cbbccead62636279c1a8e179352ab08945", "question_id": "q09_revision", "split": "test", "cluster_id": "syn_03012_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 1, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1927, "query_tokens": 603, "message_tokens": 512, "message_utf8_bytes": 1213, "cache_tensor_bytes": null, "serialized_payload_bytes": 1213, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 22829.28358297795, "hops": {"1": 22829.28358297795}, "receiver_prefill_ms": 352.2840840741992, "ttft_ms": 52.314584027044475, "decode_ms": 465.277542010881}, "total_wall_ms": 23699.937416007742, "ttft_ms": 52.314584027044475, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 7}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "5899340", "evidence_artifact": {"ledger": {"facts_total": 59, "facts_retained": 30, "retention_rate": 0.5084745762711864, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.5714285714285714, "amount": 0.6, "status": 0.5833333333333334, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1213}, "ledger_by_hop": {"1": {"facts_total": 59, "facts_retained": 30, "retention_rate": 0.5084745762711864, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.5714285714285714, "amount": 0.6, "status": 0.5833333333333334, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1213}}}, "reference_key_id": "keys/synthetic/syn_03012_2000.json", "task_scores": {"correct": 0.0, "abstained": 0.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "7c9b236b590faff2030e02587d972cf755ac2a88072e709174edbf405498973d", "output_hash": "c920323a9a02b1afb8619cf11bdde427f7cd1ea3289be69bd62932985895d717"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "ac13f988c593b5fb2cb6d515", "attempt_id": "5f745b380f71", "run_group_id": "E4_core:A4|repairable|qwen3b:h1", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:36:02.727110Z", "task_family": "synthetic", "document_id": "syn_03012_2000", "document_hash": "03e441e68d54e1af036f8ff905dfd6cbbccead62636279c1a8e179352ab08945", "question_id": "q10_temporal", "split": "test", "cluster_id": "syn_03012_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 1, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1927, "query_tokens": 613, "message_tokens": 512, "message_utf8_bytes": 1213, "cache_tensor_bytes": null, "serialized_payload_bytes": 1213, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 22829.28358297795, "hops": {"1": 22829.28358297795}, "receiver_prefill_ms": 393.70412507560104, "ttft_ms": 40.328333037905395, "decode_ms": 189.32916701305658}, "total_wall_ms": 23453.39254103601, "ttft_ms": 40.328333037905395, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 6}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "Felaldfel-9", "evidence_artifact": {"ledger": {"facts_total": 59, "facts_retained": 30, "retention_rate": 0.5084745762711864, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.5714285714285714, "amount": 0.6, "status": 0.5833333333333334, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1213}, "ledger_by_hop": {"1": {"facts_total": 59, "facts_retained": 30, "retention_rate": 0.5084745762711864, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.5714285714285714, "amount": 0.6, "status": 0.5833333333333334, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1213}}}, "reference_key_id": "keys/synthetic/syn_03012_2000.json", "task_scores": {"correct": 0.0, "abstained": 0.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "7c9b236b590faff2030e02587d972cf755ac2a88072e709174edbf405498973d", "output_hash": "09ab1578c99909d3c22a0c8042977ad718c576e7b4cce4e50abc683b5f7d9a64"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "77c108ecf3c3bb5c1d2fbab8", "attempt_id": "aadb4d2290a0", "run_group_id": "E4_core:A4|repairable|qwen3b:h1", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:36:03.368274Z", "task_family": "synthetic", "document_id": "syn_03012_2000", "document_hash": "03e441e68d54e1af036f8ff905dfd6cbbccead62636279c1a8e179352ab08945", "question_id": "q11_aggregation", "split": "test", "cluster_id": "syn_03012_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 1, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1927, "query_tokens": 605, "message_tokens": 512, "message_utf8_bytes": 1213, "cache_tensor_bytes": null, "serialized_payload_bytes": 1213, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 22829.28358297795, "hops": {"1": 22829.28358297795}, "receiver_prefill_ms": 362.67154198139906, "ttft_ms": 54.02541707735509, "decode_ms": 523.3242500107735}, "total_wall_ms": 23769.985749968328, "ttft_ms": 54.02541707735509, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 8}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "19704477", "evidence_artifact": {"ledger": {"facts_total": 59, "facts_retained": 30, "retention_rate": 0.5084745762711864, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.5714285714285714, "amount": 0.6, "status": 0.5833333333333334, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1213}, "ledger_by_hop": {"1": {"facts_total": 59, "facts_retained": 30, "retention_rate": 0.5084745762711864, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.5714285714285714, "amount": 0.6, "status": 0.5833333333333334, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1213}}}, "reference_key_id": "keys/synthetic/syn_03012_2000.json", "task_scores": {"correct": 0.0, "abstained": 0.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "7c9b236b590faff2030e02587d972cf755ac2a88072e709174edbf405498973d", "output_hash": "ba53bcc03c2a9b1df7909eec6e2d5de7384570e56fb592d48b95504bfd97157d"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "7f270f82bcc1ff3703449344", "attempt_id": "dc37f0a9bf85", "run_group_id": "E4_core:A4|repairable|qwen3b:h1", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:36:04.327512Z", "task_family": "synthetic", "document_id": "syn_03012_2000", "document_hash": "03e441e68d54e1af036f8ff905dfd6cbbccead62636279c1a8e179352ab08945", "question_id": "q12_unit_conversion", "split": "test", "cluster_id": "syn_03012_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 1, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1927, "query_tokens": 602, "message_tokens": 512, "message_utf8_bytes": 1213, "cache_tensor_bytes": null, "serialized_payload_bytes": 1213, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 22829.28358297795, "hops": {"1": 22829.28358297795}, "receiver_prefill_ms": 392.7881669951603, "ttft_ms": 40.997999953106046, "decode_ms": 167.08304197527468}, "total_wall_ms": 23431.065624929033, "ttft_ms": 40.997999953106046, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 5}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "44620", "evidence_artifact": {"ledger": {"facts_total": 59, "facts_retained": 30, "retention_rate": 0.5084745762711864, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.5714285714285714, "amount": 0.6, "status": 0.5833333333333334, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1213}, "ledger_by_hop": {"1": {"facts_total": 59, "facts_retained": 30, "retention_rate": 0.5084745762711864, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.5714285714285714, "amount": 0.6, "status": 0.5833333333333334, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1213}}}, "reference_key_id": "keys/synthetic/syn_03012_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "7c9b236b590faff2030e02587d972cf755ac2a88072e709174edbf405498973d", "output_hash": "2bd125847727be7e7c0a656ee4c81f3bc6d0fef11f536f50ce7bc83688907075"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "8b11a4a5dbcf99f3833bd809", "attempt_id": "3b3e2c870703", "run_group_id": "E4_core:A4|repairable|qwen3b:h1", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:36:04.946536Z", "task_family": "synthetic", "document_id": "syn_03012_2000", "document_hash": "03e441e68d54e1af036f8ff905dfd6cbbccead62636279c1a8e179352ab08945", "question_id": "q13_unanswerable", "split": "test", "cluster_id": "syn_03012_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 1, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1927, "query_tokens": 607, "message_tokens": 512, "message_utf8_bytes": 1213, "cache_tensor_bytes": null, "serialized_payload_bytes": 1213, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 22829.28358297795, "hops": {"1": 22829.28358297795}, "receiver_prefill_ms": 374.8197080567479, "ttft_ms": 50.370958051644266, "decode_ms": 47.68075002357364}, "total_wall_ms": 23302.86779103335, "ttft_ms": 50.370958051644266, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 59, "facts_retained": 30, "retention_rate": 0.5084745762711864, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.5714285714285714, "amount": 0.6, "status": 0.5833333333333334, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1213}, "ledger_by_hop": {"1": {"facts_total": 59, "facts_retained": 30, "retention_rate": 0.5084745762711864, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.5714285714285714, "amount": 0.6, "status": 0.5833333333333334, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1213}}}, "reference_key_id": "keys/synthetic/syn_03012_2000.json", "task_scores": {"correct": 1.0, "abstained": 1.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "7c9b236b590faff2030e02587d972cf755ac2a88072e709174edbf405498973d", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "0a80f811ed9cadebd55ccd56", "attempt_id": "b1fdf8d26637", "run_group_id": "E4_core:A4|repairable|qwen3b:h1", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:36:05.437482Z", "task_family": "synthetic", "document_id": "syn_03012_2000", "document_hash": "03e441e68d54e1af036f8ff905dfd6cbbccead62636279c1a8e179352ab08945", "question_id": "q14_unanswerable", "split": "test", "cluster_id": "syn_03012_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 1, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1927, "query_tokens": 607, "message_tokens": 512, "message_utf8_bytes": 1213, "cache_tensor_bytes": null, "serialized_payload_bytes": 1213, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 22829.28358297795, "hops": {"1": 22829.28358297795}, "receiver_prefill_ms": 410.5375830549747, "ttft_ms": 125.2820419613272, "decode_ms": 43.9424590440467}, "total_wall_ms": 23410.014958004467, "ttft_ms": 125.2820419613272, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 59, "facts_retained": 30, "retention_rate": 0.5084745762711864, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.5714285714285714, "amount": 0.6, "status": 0.5833333333333334, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1213}, "ledger_by_hop": {"1": {"facts_total": 59, "facts_retained": 30, "retention_rate": 0.5084745762711864, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.5714285714285714, "amount": 0.6, "status": 0.5833333333333334, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1213}}}, "reference_key_id": "keys/synthetic/syn_03012_2000.json", "task_scores": {"correct": 1.0, "abstained": 1.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "7c9b236b590faff2030e02587d972cf755ac2a88072e709174edbf405498973d", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "245c3b1d7d48924f390e6dc6", "attempt_id": "1c6a29f5607c", "run_group_id": "E4_core:A4|repairable|qwen3b:h2", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:36:27.152777Z", "task_family": "synthetic", "document_id": "syn_03012_2000", "document_hash": "03e441e68d54e1af036f8ff905dfd6cbbccead62636279c1a8e179352ab08945", "question_id": "q00_single_fact", "split": "test", "cluster_id": "syn_03012_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 2, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1927, "query_tokens": 575, "message_tokens": 493, "message_utf8_bytes": 1105, "cache_tensor_bytes": null, "serialized_payload_bytes": 1105, "side_information_bytes": 0, "fetched_bytes": 731, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 43944.304332952015, "hops": {"1": 22829.28358297795, "2": 21115.020749974065}, "receiver_prefill_ms": 346.34887508582324, "ttft_ms": 35.67316697444767, "decode_ms": 573.8885829923674}, "total_wall_ms": 44901.124790892936, "ttft_ms": 35.67316697444767, "input_usage": {"lookups": 4}, "output_usage": {"output_tokens": 10}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "XX-71M-6213", "evidence_artifact": {"ledger": {"facts_total": 59, "facts_retained": 30, "retention_rate": 0.5084745762711864, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.5714285714285714, "amount": 0.6, "status": 0.5833333333333334, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1105}, "ledger_by_hop": {"1": {"facts_total": 59, "facts_retained": 30, "retention_rate": 0.5084745762711864, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.5714285714285714, "amount": 0.6, "status": 0.5833333333333334, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1213}, "2": {"facts_total": 59, "facts_retained": 30, "retention_rate": 0.5084745762711864, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.5714285714285714, "amount": 0.6, "status": 0.5833333333333334, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1105}}}, "reference_key_id": "keys/synthetic/syn_03012_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "a7b78bd60073f4ccb33f814bb453df3b570e2ce6661423f590ae92d973dc68c3", "output_hash": "66ef6c5da682b38795e9029723bbd71cdde2c4dec70cf1fae6efe62ef243f10a"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "b0780c5bf9c25cfe2459412d", "attempt_id": "8419fc6da5c1", "run_group_id": "E4_core:A4|repairable|qwen3b:h2", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:36:28.128114Z", "task_family": "synthetic", "document_id": "syn_03012_2000", "document_hash": "03e441e68d54e1af036f8ff905dfd6cbbccead62636279c1a8e179352ab08945", "question_id": "q01_single_fact", "split": "test", "cluster_id": "syn_03012_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 2, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1927, "query_tokens": 581, "message_tokens": 493, "message_utf8_bytes": 1105, "cache_tensor_bytes": null, "serialized_payload_bytes": 1105, "side_information_bytes": 0, "fetched_bytes": 731, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 43944.304332952015, "hops": {"1": 22829.28358297795, "2": 21115.020749974065}, "receiver_prefill_ms": 423.07220899965614, "ttft_ms": 41.28170805051923, "decode_ms": 30.99995804950595}, "total_wall_ms": 44440.33083296381, "ttft_ms": 41.28170805051923, "input_usage": {"lookups": 4}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 59, "facts_retained": 30, "retention_rate": 0.5084745762711864, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.5714285714285714, "amount": 0.6, "status": 0.5833333333333334, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1105}, "ledger_by_hop": {"1": {"facts_total": 59, "facts_retained": 30, "retention_rate": 0.5084745762711864, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.5714285714285714, "amount": 0.6, "status": 0.5833333333333334, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1213}, "2": {"facts_total": 59, "facts_retained": 30, "retention_rate": 0.5084745762711864, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.5714285714285714, "amount": 0.6, "status": 0.5833333333333334, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1105}}}, "reference_key_id": "keys/synthetic/syn_03012_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "a7b78bd60073f4ccb33f814bb453df3b570e2ce6661423f590ae92d973dc68c3", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "bd7801aa27180e3b9d4795d1", "attempt_id": "8540870ad77f", "run_group_id": "E4_core:A4|repairable|qwen3b:h2", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:36:28.641739Z", "task_family": "synthetic", "document_id": "syn_03012_2000", "document_hash": "03e441e68d54e1af036f8ff905dfd6cbbccead62636279c1a8e179352ab08945", "question_id": "q02_exact_copy", "split": "test", "cluster_id": "syn_03012_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 2, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1927, "query_tokens": 580, "message_tokens": 493, "message_utf8_bytes": 1105, "cache_tensor_bytes": null, "serialized_payload_bytes": 1105, "side_information_bytes": 0, "fetched_bytes": 731, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 43944.304332952015, "hops": {"1": 22829.28358297795, "2": 21115.020749974065}, "receiver_prefill_ms": 351.130791939795, "ttft_ms": 35.98641592543572, "decode_ms": 499.16633300017565}, "total_wall_ms": 44831.289082881995, "ttft_ms": 35.98641592543572, "input_usage": {"lookups": 4}, "output_usage": {"output_tokens": 7}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "13.4343", "evidence_artifact": {"ledger": {"facts_total": 59, "facts_retained": 30, "retention_rate": 0.5084745762711864, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.5714285714285714, "amount": 0.6, "status": 0.5833333333333334, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1105}, "ledger_by_hop": {"1": {"facts_total": 59, "facts_retained": 30, "retention_rate": 0.5084745762711864, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.5714285714285714, "amount": 0.6, "status": 0.5833333333333334, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1213}, "2": {"facts_total": 59, "facts_retained": 30, "retention_rate": 0.5084745762711864, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.5714285714285714, "amount": 0.6, "status": 0.5833333333333334, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1105}}}, "reference_key_id": "keys/synthetic/syn_03012_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "a7b78bd60073f4ccb33f814bb453df3b570e2ce6661423f590ae92d973dc68c3", "output_hash": "599bc22e373b1cf7b4679208eb9e515ad0e8dcb88dde0c7f579637ec7d3f3014"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "6b6b25228940e83aa9b8c499", "attempt_id": "68b0eb7e2b4c", "run_group_id": "E4_core:A4|repairable|qwen3b:h2", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:36:29.545523Z", "task_family": "synthetic", "document_id": "syn_03012_2000", "document_hash": "03e441e68d54e1af036f8ff905dfd6cbbccead62636279c1a8e179352ab08945", "question_id": "q03_exact_copy", "split": "test", "cluster_id": "syn_03012_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 2, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1927, "query_tokens": 576, "message_tokens": 493, "message_utf8_bytes": 1105, "cache_tensor_bytes": null, "serialized_payload_bytes": 1105, "side_information_bytes": 0, "fetched_bytes": 731, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 43944.304332952015, "hops": {"1": 22829.28358297795, "2": 21115.020749974065}, "receiver_prefill_ms": 384.5378749538213, "ttft_ms": 33.781500067561865, "decode_ms": 32.78437501285225}, "total_wall_ms": 44396.07495791279, "ttft_ms": 33.781500067561865, "input_usage": {"lookups": 4}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 59, "facts_retained": 30, "retention_rate": 0.5084745762711864, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.5714285714285714, "amount": 0.6, "status": 0.5833333333333334, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1105}, "ledger_by_hop": {"1": {"facts_total": 59, "facts_retained": 30, "retention_rate": 0.5084745762711864, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.5714285714285714, "amount": 0.6, "status": 0.5833333333333334, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1213}, "2": {"facts_total": 59, "facts_retained": 30, "retention_rate": 0.5084745762711864, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.5714285714285714, "amount": 0.6, "status": 0.5833333333333334, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1105}}}, "reference_key_id": "keys/synthetic/syn_03012_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "a7b78bd60073f4ccb33f814bb453df3b570e2ce6661423f590ae92d973dc68c3", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "2014ea94ee3a3fe3b9dbc8e1", "attempt_id": "190822e0887e", "run_group_id": "E4_core:A4|repairable|qwen3b:h2", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:36:30.015251Z", "task_family": "synthetic", "document_id": "syn_03012_2000", "document_hash": "03e441e68d54e1af036f8ff905dfd6cbbccead62636279c1a8e179352ab08945", "question_id": "q04_multi_hop", "split": "test", "cluster_id": "syn_03012_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 2, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1927, "query_tokens": 581, "message_tokens": 493, "message_utf8_bytes": 1105, "cache_tensor_bytes": null, "serialized_payload_bytes": 1105, "side_information_bytes": 0, "fetched_bytes": 731, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 43944.304332952015, "hops": {"1": 22829.28358297795, "2": 21115.020749974065}, "receiver_prefill_ms": 401.13712509628385, "ttft_ms": 34.83108291402459, "decode_ms": 262.23300001583993}, "total_wall_ms": 44643.21641589049, "ttft_ms": 34.83108291402459, "input_usage": {"lookups": 4}, "output_usage": {"output_tokens": 7}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "8631552", "evidence_artifact": {"ledger": {"facts_total": 59, "facts_retained": 30, "retention_rate": 0.5084745762711864, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.5714285714285714, "amount": 0.6, "status": 0.5833333333333334, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1105}, "ledger_by_hop": {"1": {"facts_total": 59, "facts_retained": 30, "retention_rate": 0.5084745762711864, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.5714285714285714, "amount": 0.6, "status": 0.5833333333333334, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1213}, "2": {"facts_total": 59, "facts_retained": 30, "retention_rate": 0.5084745762711864, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.5714285714285714, "amount": 0.6, "status": 0.5833333333333334, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1105}}}, "reference_key_id": "keys/synthetic/syn_03012_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "a7b78bd60073f4ccb33f814bb453df3b570e2ce6661423f590ae92d973dc68c3", "output_hash": "147d02c0587a77b3b53f0424c8d3d1454aaf2639ce9fdaf69af20db33363e849"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "0b647a5863587b02f3945236", "attempt_id": "8ab3a997fb8e", "run_group_id": "E4_core:A4|repairable|qwen3b:h2", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:36:30.733007Z", "task_family": "synthetic", "document_id": "syn_03012_2000", "document_hash": "03e441e68d54e1af036f8ff905dfd6cbbccead62636279c1a8e179352ab08945", "question_id": "q05_multi_hop", "split": "test", "cluster_id": "syn_03012_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 2, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1927, "query_tokens": 584, "message_tokens": 493, "message_utf8_bytes": 1105, "cache_tensor_bytes": null, "serialized_payload_bytes": 1105, "side_information_bytes": 0, "fetched_bytes": 731, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 43944.304332952015, "hops": {"1": 22829.28358297795, "2": 21115.020749974065}, "receiver_prefill_ms": 540.221874951385, "ttft_ms": 21.365542081184685, "decode_ms": 351.553792017512}, "total_wall_ms": 44858.37899998296, "ttft_ms": 21.365542081184685, "input_usage": {"lookups": 4}, "output_usage": {"output_tokens": 10}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "DG-40E-2777", "evidence_artifact": {"ledger": {"facts_total": 59, "facts_retained": 30, "retention_rate": 0.5084745762711864, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.5714285714285714, "amount": 0.6, "status": 0.5833333333333334, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1105}, "ledger_by_hop": {"1": {"facts_total": 59, "facts_retained": 30, "retention_rate": 0.5084745762711864, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.5714285714285714, "amount": 0.6, "status": 0.5833333333333334, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1213}, "2": {"facts_total": 59, "facts_retained": 30, "retention_rate": 0.5084745762711864, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.5714285714285714, "amount": 0.6, "status": 0.5833333333333334, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1105}}}, "reference_key_id": "keys/synthetic/syn_03012_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "a7b78bd60073f4ccb33f814bb453df3b570e2ce6661423f590ae92d973dc68c3", "output_hash": "16a7e3f4a7cc4a9b66f86c2cdcfad3b35daf1da199d9406e0627c8af2d8d73f3"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "9b70ebdfb5a17afc5661e3cd", "attempt_id": "249e0c2fcf0a", "run_group_id": "E4_core:A4|repairable|qwen3b:h2", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:36:31.666041Z", "task_family": "synthetic", "document_id": "syn_03012_2000", "document_hash": "03e441e68d54e1af036f8ff905dfd6cbbccead62636279c1a8e179352ab08945", "question_id": "q06_negation", "split": "test", "cluster_id": "syn_03012_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 2, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1927, "query_tokens": 576, "message_tokens": 493, "message_utf8_bytes": 1105, "cache_tensor_bytes": null, "serialized_payload_bytes": 1105, "side_information_bytes": 0, "fetched_bytes": 731, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 43944.304332952015, "hops": {"1": 22829.28358297795, "2": 21115.020749974065}, "receiver_prefill_ms": 390.2227079961449, "ttft_ms": 21.20595902670175, "decode_ms": 20.476541947573423}, "total_wall_ms": 44376.8344578566, "ttft_ms": 21.20595902670175, "input_usage": {"lookups": 4}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "no", "evidence_artifact": {"ledger": {"facts_total": 59, "facts_retained": 30, "retention_rate": 0.5084745762711864, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.5714285714285714, "amount": 0.6, "status": 0.5833333333333334, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1105}, "ledger_by_hop": {"1": {"facts_total": 59, "facts_retained": 30, "retention_rate": 0.5084745762711864, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.5714285714285714, "amount": 0.6, "status": 0.5833333333333334, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1213}, "2": {"facts_total": 59, "facts_retained": 30, "retention_rate": 0.5084745762711864, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.5714285714285714, "amount": 0.6, "status": 0.5833333333333334, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1105}}}, "reference_key_id": "keys/synthetic/syn_03012_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "a7b78bd60073f4ccb33f814bb453df3b570e2ce6661423f590ae92d973dc68c3", "output_hash": "9390298f3fb0c5b160498935d79cb139aef28e1c47358b4bbba61862b9c26e59"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "6aa5920e43cbfdb88a0cb730", "attempt_id": "506b33604e58", "run_group_id": "E4_core:A4|repairable|qwen3b:h2", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:36:32.115762Z", "task_family": "synthetic", "document_id": "syn_03012_2000", "document_hash": "03e441e68d54e1af036f8ff905dfd6cbbccead62636279c1a8e179352ab08945", "question_id": "q07_negation", "split": "test", "cluster_id": "syn_03012_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 2, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1927, "query_tokens": 576, "message_tokens": 493, "message_utf8_bytes": 1105, "cache_tensor_bytes": null, "serialized_payload_bytes": 1105, "side_information_bytes": 0, "fetched_bytes": 731, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 43944.304332952015, "hops": {"1": 22829.28358297795, "2": 21115.020749974065}, "receiver_prefill_ms": 385.69795805960894, "ttft_ms": 60.40716706775129, "decode_ms": 101.61337500903755}, "total_wall_ms": 44492.71866597701, "ttft_ms": 60.40716706775129, "input_usage": {"lookups": 4}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "no", "evidence_artifact": {"ledger": {"facts_total": 59, "facts_retained": 30, "retention_rate": 0.5084745762711864, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.5714285714285714, "amount": 0.6, "status": 0.5833333333333334, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1105}, "ledger_by_hop": {"1": {"facts_total": 59, "facts_retained": 30, "retention_rate": 0.5084745762711864, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.5714285714285714, "amount": 0.6, "status": 0.5833333333333334, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1213}, "2": {"facts_total": 59, "facts_retained": 30, "retention_rate": 0.5084745762711864, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.5714285714285714, "amount": 0.6, "status": 0.5833333333333334, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1105}}}, "reference_key_id": "keys/synthetic/syn_03012_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "a7b78bd60073f4ccb33f814bb453df3b570e2ce6661423f590ae92d973dc68c3", "output_hash": "9390298f3fb0c5b160498935d79cb139aef28e1c47358b4bbba61862b9c26e59"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "46e18a526019947caef5df28", "attempt_id": "5a7e31528ed2", "run_group_id": "E4_core:A4|repairable|qwen3b:h2", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:36:32.684028Z", "task_family": "synthetic", "document_id": "syn_03012_2000", "document_hash": "03e441e68d54e1af036f8ff905dfd6cbbccead62636279c1a8e179352ab08945", "question_id": "q08_revision", "split": "test", "cluster_id": "syn_03012_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 2, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1927, "query_tokens": 591, "message_tokens": 493, "message_utf8_bytes": 1105, "cache_tensor_bytes": null, "serialized_payload_bytes": 1105, "side_information_bytes": 0, "fetched_bytes": 731, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 43944.304332952015, "hops": {"1": 22829.28358297795, "2": 21115.020749974065}, "receiver_prefill_ms": 474.53666699584574, "ttft_ms": 42.42008295841515, "decode_ms": 222.4492090754211}, "total_wall_ms": 44684.321874869056, "ttft_ms": 42.42008295841515, "input_usage": {"lookups": 4}, "output_usage": {"output_tokens": 7}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "6602245", "evidence_artifact": {"ledger": {"facts_total": 59, "facts_retained": 30, "retention_rate": 0.5084745762711864, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.5714285714285714, "amount": 0.6, "status": 0.5833333333333334, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1105}, "ledger_by_hop": {"1": {"facts_total": 59, "facts_retained": 30, "retention_rate": 0.5084745762711864, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.5714285714285714, "amount": 0.6, "status": 0.5833333333333334, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1213}, "2": {"facts_total": 59, "facts_retained": 30, "retention_rate": 0.5084745762711864, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.5714285714285714, "amount": 0.6, "status": 0.5833333333333334, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1105}}}, "reference_key_id": "keys/synthetic/syn_03012_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "a7b78bd60073f4ccb33f814bb453df3b570e2ce6661423f590ae92d973dc68c3", "output_hash": "d117fa8197988f8d1e4fabc6c4cdaab9516018e54553b57c63b875b3de82b16f"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "a17c8560a5eb341507b9d0b4", "attempt_id": "c4f9f31d50fd", "run_group_id": "E4_core:A4|repairable|qwen3b:h2", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:36:33.443665Z", "task_family": "synthetic", "document_id": "syn_03012_2000", "document_hash": "03e441e68d54e1af036f8ff905dfd6cbbccead62636279c1a8e179352ab08945", "question_id": "q09_revision", "split": "test", "cluster_id": "syn_03012_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 2, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1927, "query_tokens": 584, "message_tokens": 493, "message_utf8_bytes": 1105, "cache_tensor_bytes": null, "serialized_payload_bytes": 1105, "side_information_bytes": 0, "fetched_bytes": 731, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 43944.304332952015, "hops": {"1": 22829.28358297795, "2": 21115.020749974065}, "receiver_prefill_ms": 370.4212090233341, "ttft_ms": 33.6920830886811, "decode_ms": 443.22133297100663}, "total_wall_ms": 44792.401498998515, "ttft_ms": 33.6920830886811, "input_usage": {"lookups": 4}, "output_usage": {"output_tokens": 7}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "5899340", "evidence_artifact": {"ledger": {"facts_total": 59, "facts_retained": 30, "retention_rate": 0.5084745762711864, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.5714285714285714, "amount": 0.6, "status": 0.5833333333333334, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1105}, "ledger_by_hop": {"1": {"facts_total": 59, "facts_retained": 30, "retention_rate": 0.5084745762711864, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.5714285714285714, "amount": 0.6, "status": 0.5833333333333334, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1213}, "2": {"facts_total": 59, "facts_retained": 30, "retention_rate": 0.5084745762711864, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.5714285714285714, "amount": 0.6, "status": 0.5833333333333334, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1105}}}, "reference_key_id": "keys/synthetic/syn_03012_2000.json", "task_scores": {"correct": 0.0, "abstained": 0.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "a7b78bd60073f4ccb33f814bb453df3b570e2ce6661423f590ae92d973dc68c3", "output_hash": "c920323a9a02b1afb8619cf11bdde427f7cd1ea3289be69bd62932985895d717"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "3d03d2666c2ddd9006c6c613", "attempt_id": "4f2f77be99dd", "run_group_id": "E4_core:A4|repairable|qwen3b:h2", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:36:34.311650Z", "task_family": "synthetic", "document_id": "syn_03012_2000", "document_hash": "03e441e68d54e1af036f8ff905dfd6cbbccead62636279c1a8e179352ab08945", "question_id": "q10_temporal", "split": "test", "cluster_id": "syn_03012_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 2, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1927, "query_tokens": 594, "message_tokens": 493, "message_utf8_bytes": 1105, "cache_tensor_bytes": null, "serialized_payload_bytes": 1105, "side_information_bytes": 0, "fetched_bytes": 731, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 43944.304332952015, "hops": {"1": 22829.28358297795, "2": 21115.020749974065}, "receiver_prefill_ms": 455.5460000410676, "ttft_ms": 49.78900006972253, "decode_ms": 56.65849999058992}, "total_wall_ms": 44506.90433289856, "ttft_ms": 49.78900006972253, "input_usage": {"lookups": 4}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 59, "facts_retained": 30, "retention_rate": 0.5084745762711864, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.5714285714285714, "amount": 0.6, "status": 0.5833333333333334, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1105}, "ledger_by_hop": {"1": {"facts_total": 59, "facts_retained": 30, "retention_rate": 0.5084745762711864, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.5714285714285714, "amount": 0.6, "status": 0.5833333333333334, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1213}, "2": {"facts_total": 59, "facts_retained": 30, "retention_rate": 0.5084745762711864, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.5714285714285714, "amount": 0.6, "status": 0.5833333333333334, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1105}}}, "reference_key_id": "keys/synthetic/syn_03012_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "a7b78bd60073f4ccb33f814bb453df3b570e2ce6661423f590ae92d973dc68c3", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "517bb5b26a2b0baf24172f05", "attempt_id": "be228f3c9dac", "run_group_id": "E4_core:A4|repairable|qwen3b:h2", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:36:34.891418Z", "task_family": "synthetic", "document_id": "syn_03012_2000", "document_hash": "03e441e68d54e1af036f8ff905dfd6cbbccead62636279c1a8e179352ab08945", "question_id": "q11_aggregation", "split": "test", "cluster_id": "syn_03012_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 2, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1927, "query_tokens": 586, "message_tokens": 493, "message_utf8_bytes": 1105, "cache_tensor_bytes": null, "serialized_payload_bytes": 1105, "side_information_bytes": 0, "fetched_bytes": 731, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 43944.304332952015, "hops": {"1": 22829.28358297795, "2": 21115.020749974065}, "receiver_prefill_ms": 446.86304102651775, "ttft_ms": 100.46624997630715, "decode_ms": 489.91641704924405}, "total_wall_ms": 44982.185791945085, "ttft_ms": 100.46624997630715, "input_usage": {"lookups": 4}, "output_usage": {"output_tokens": 8}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "14014487", "evidence_artifact": {"ledger": {"facts_total": 59, "facts_retained": 30, "retention_rate": 0.5084745762711864, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.5714285714285714, "amount": 0.6, "status": 0.5833333333333334, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1105}, "ledger_by_hop": {"1": {"facts_total": 59, "facts_retained": 30, "retention_rate": 0.5084745762711864, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.5714285714285714, "amount": 0.6, "status": 0.5833333333333334, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1213}, "2": {"facts_total": 59, "facts_retained": 30, "retention_rate": 0.5084745762711864, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.5714285714285714, "amount": 0.6, "status": 0.5833333333333334, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1105}}}, "reference_key_id": "keys/synthetic/syn_03012_2000.json", "task_scores": {"correct": 0.0, "abstained": 0.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "a7b78bd60073f4ccb33f814bb453df3b570e2ce6661423f590ae92d973dc68c3", "output_hash": "fe944c9839514bb6d761562d55571ee70262a59ab0933baad439c2081c1d31e1"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "37598d236f9f4af6f78f1ce9", "attempt_id": "1514d5cb95d7", "run_group_id": "E4_core:A4|repairable|qwen3b:h2", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:36:35.946373Z", "task_family": "synthetic", "document_id": "syn_03012_2000", "document_hash": "03e441e68d54e1af036f8ff905dfd6cbbccead62636279c1a8e179352ab08945", "question_id": "q12_unit_conversion", "split": "test", "cluster_id": "syn_03012_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 2, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1927, "query_tokens": 583, "message_tokens": 493, "message_utf8_bytes": 1105, "cache_tensor_bytes": null, "serialized_payload_bytes": 1105, "side_information_bytes": 0, "fetched_bytes": 731, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 43944.304332952015, "hops": {"1": 22829.28358297795, "2": 21115.020749974065}, "receiver_prefill_ms": 454.12754197604954, "ttft_ms": 51.32674996275455, "decode_ms": 337.18433300964534}, "total_wall_ms": 44787.61308290996, "ttft_ms": 51.32674996275455, "input_usage": {"lookups": 4}, "output_usage": {"output_tokens": 5}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "44620", "evidence_artifact": {"ledger": {"facts_total": 59, "facts_retained": 30, "retention_rate": 0.5084745762711864, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.5714285714285714, "amount": 0.6, "status": 0.5833333333333334, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1105}, "ledger_by_hop": {"1": {"facts_total": 59, "facts_retained": 30, "retention_rate": 0.5084745762711864, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.5714285714285714, "amount": 0.6, "status": 0.5833333333333334, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1213}, "2": {"facts_total": 59, "facts_retained": 30, "retention_rate": 0.5084745762711864, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.5714285714285714, "amount": 0.6, "status": 0.5833333333333334, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1105}}}, "reference_key_id": "keys/synthetic/syn_03012_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "a7b78bd60073f4ccb33f814bb453df3b570e2ce6661423f590ae92d973dc68c3", "output_hash": "2bd125847727be7e7c0a656ee4c81f3bc6d0fef11f536f50ce7bc83688907075"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "0b32738e72eaf8baddbd31d9", "attempt_id": "48d3f587e44c", "run_group_id": "E4_core:A4|repairable|qwen3b:h2", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:36:36.807238Z", "task_family": "synthetic", "document_id": "syn_03012_2000", "document_hash": "03e441e68d54e1af036f8ff905dfd6cbbccead62636279c1a8e179352ab08945", "question_id": "q13_unanswerable", "split": "test", "cluster_id": "syn_03012_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 2, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1927, "query_tokens": 588, "message_tokens": 493, "message_utf8_bytes": 1105, "cache_tensor_bytes": null, "serialized_payload_bytes": 1105, "side_information_bytes": 0, "fetched_bytes": 731, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 43944.304332952015, "hops": {"1": 22829.28358297795, "2": 21115.020749974065}, "receiver_prefill_ms": 487.093250034377, "ttft_ms": 89.21825001016259, "decode_ms": 62.42841691710055}, "total_wall_ms": 44583.63774896134, "ttft_ms": 89.21825001016259, "input_usage": {"lookups": 4}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 59, "facts_retained": 30, "retention_rate": 0.5084745762711864, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.5714285714285714, "amount": 0.6, "status": 0.5833333333333334, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1105}, "ledger_by_hop": {"1": {"facts_total": 59, "facts_retained": 30, "retention_rate": 0.5084745762711864, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.5714285714285714, "amount": 0.6, "status": 0.5833333333333334, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1213}, "2": {"facts_total": 59, "facts_retained": 30, "retention_rate": 0.5084745762711864, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.5714285714285714, "amount": 0.6, "status": 0.5833333333333334, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1105}}}, "reference_key_id": "keys/synthetic/syn_03012_2000.json", "task_scores": {"correct": 1.0, "abstained": 1.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "a7b78bd60073f4ccb33f814bb453df3b570e2ce6661423f590ae92d973dc68c3", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "465245c88c9e39c7deba65ed", "attempt_id": "9cc29ad32c62", "run_group_id": "E4_core:A4|repairable|qwen3b:h2", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:36:37.463896Z", "task_family": "synthetic", "document_id": "syn_03012_2000", "document_hash": "03e441e68d54e1af036f8ff905dfd6cbbccead62636279c1a8e179352ab08945", "question_id": "q14_unanswerable", "split": "test", "cluster_id": "syn_03012_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 2, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1927, "query_tokens": 588, "message_tokens": 493, "message_utf8_bytes": 1105, "cache_tensor_bytes": null, "serialized_payload_bytes": 1105, "side_information_bytes": 0, "fetched_bytes": 731, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 43944.304332952015, "hops": {"1": 22829.28358297795, "2": 21115.020749974065}, "receiver_prefill_ms": 480.96012498717755, "ttft_ms": 61.97649997193366, "decode_ms": 48.27166709583253}, "total_wall_ms": 44536.260999972, "ttft_ms": 61.97649997193366, "input_usage": {"lookups": 4}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 59, "facts_retained": 30, "retention_rate": 0.5084745762711864, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.5714285714285714, "amount": 0.6, "status": 0.5833333333333334, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1105}, "ledger_by_hop": {"1": {"facts_total": 59, "facts_retained": 30, "retention_rate": 0.5084745762711864, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.5714285714285714, "amount": 0.6, "status": 0.5833333333333334, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1213}, "2": {"facts_total": 59, "facts_retained": 30, "retention_rate": 0.5084745762711864, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.5714285714285714, "amount": 0.6, "status": 0.5833333333333334, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1105}}}, "reference_key_id": "keys/synthetic/syn_03012_2000.json", "task_scores": {"correct": 1.0, "abstained": 1.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "a7b78bd60073f4ccb33f814bb453df3b570e2ce6661423f590ae92d973dc68c3", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "df3af41a8b9d253946ffa2d6", "attempt_id": "e7af5f2686e0", "run_group_id": "E4_core:A4|repairable|qwen3b:h4", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:37:20.970119Z", "task_family": "synthetic", "document_id": "syn_03012_2000", "document_hash": "03e441e68d54e1af036f8ff905dfd6cbbccead62636279c1a8e179352ab08945", "question_id": "q00_single_fact", "split": "test", "cluster_id": "syn_03012_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 4, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1927, "query_tokens": 561, "message_tokens": 479, "message_utf8_bytes": 1073, "cache_tensor_bytes": null, "serialized_payload_bytes": 1073, "side_information_bytes": 0, "fetched_bytes": 2193, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 86838.3012090344, "hops": {"1": 22829.28358297795, "2": 21115.020749974065, "3": 22152.291209087707, "4": 20741.705666994676}, "receiver_prefill_ms": 394.9128750246018, "ttft_ms": 72.02641700860113, "decode_ms": 408.5975830676034}, "total_wall_ms": 87714.6235430846, "ttft_ms": 72.02641700860113, "input_usage": {"lookups": 12}, "output_usage": {"output_tokens": 10}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "XX-71M-6213", "evidence_artifact": {"ledger": {"facts_total": 59, "facts_retained": 30, "retention_rate": 0.5084745762711864, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.5714285714285714, "amount": 0.6, "status": 0.5833333333333334, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1073}, "ledger_by_hop": {"1": {"facts_total": 59, "facts_retained": 30, "retention_rate": 0.5084745762711864, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.5714285714285714, "amount": 0.6, "status": 0.5833333333333334, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1213}, "2": {"facts_total": 59, "facts_retained": 30, "retention_rate": 0.5084745762711864, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.5714285714285714, "amount": 0.6, "status": 0.5833333333333334, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1105}, "3": {"facts_total": 59, "facts_retained": 30, "retention_rate": 0.5084745762711864, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.5714285714285714, "amount": 0.6, "status": 0.5833333333333334, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1085}, "4": {"facts_total": 59, "facts_retained": 30, "retention_rate": 0.5084745762711864, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.5714285714285714, "amount": 0.6, "status": 0.5833333333333334, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1073}}}, "reference_key_id": "keys/synthetic/syn_03012_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "5aeedc1e4643b829686179704346397792278190d32a28471247e6a066095c87", "output_hash": "66ef6c5da682b38795e9029723bbd71cdde2c4dec70cf1fae6efe62ef243f10a"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "a28e9927da851b2aff488238", "attempt_id": "47a1c5051611", "run_group_id": "E4_core:A4|repairable|qwen3b:h4", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:37:21.864366Z", "task_family": "synthetic", "document_id": "syn_03012_2000", "document_hash": "03e441e68d54e1af036f8ff905dfd6cbbccead62636279c1a8e179352ab08945", "question_id": "q01_single_fact", "split": "test", "cluster_id": "syn_03012_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 4, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1927, "query_tokens": 567, "message_tokens": 479, "message_utf8_bytes": 1073, "cache_tensor_bytes": null, "serialized_payload_bytes": 1073, "side_information_bytes": 0, "fetched_bytes": 2193, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 86838.3012090344, "hops": {"1": 22829.28358297795, "2": 21115.020749974065, "3": 22152.291209087707, "4": 20741.705666994676}, "receiver_prefill_ms": 381.11850002314895, "ttft_ms": 36.590624949894845, "decode_ms": 29.569791979156435}, "total_wall_ms": 87286.16345894989, "ttft_ms": 36.590624949894845, "input_usage": {"lookups": 12}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 59, "facts_retained": 30, "retention_rate": 0.5084745762711864, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.5714285714285714, "amount": 0.6, "status": 0.5833333333333334, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1073}, "ledger_by_hop": {"1": {"facts_total": 59, "facts_retained": 30, "retention_rate": 0.5084745762711864, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.5714285714285714, "amount": 0.6, "status": 0.5833333333333334, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1213}, "2": {"facts_total": 59, "facts_retained": 30, "retention_rate": 0.5084745762711864, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.5714285714285714, "amount": 0.6, "status": 0.5833333333333334, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1105}, "3": {"facts_total": 59, "facts_retained": 30, "retention_rate": 0.5084745762711864, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.5714285714285714, "amount": 0.6, "status": 0.5833333333333334, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1085}, "4": {"facts_total": 59, "facts_retained": 30, "retention_rate": 0.5084745762711864, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.5714285714285714, "amount": 0.6, "status": 0.5833333333333334, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1073}}}, "reference_key_id": "keys/synthetic/syn_03012_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "5aeedc1e4643b829686179704346397792278190d32a28471247e6a066095c87", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "698a960369ef5e342b0741da", "attempt_id": "06dec4c6cdcc", "run_group_id": "E4_core:A4|repairable|qwen3b:h4", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:37:22.329695Z", "task_family": "synthetic", "document_id": "syn_03012_2000", "document_hash": "03e441e68d54e1af036f8ff905dfd6cbbccead62636279c1a8e179352ab08945", "question_id": "q02_exact_copy", "split": "test", "cluster_id": "syn_03012_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 4, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1927, "query_tokens": 566, "message_tokens": 479, "message_utf8_bytes": 1073, "cache_tensor_bytes": null, "serialized_payload_bytes": 1073, "side_information_bytes": 0, "fetched_bytes": 2193, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 86838.3012090344, "hops": {"1": 22829.28358297795, "2": 21115.020749974065, "3": 22152.291209087707, "4": 20741.705666994676}, "receiver_prefill_ms": 334.64091701898724, "ttft_ms": 35.98854201845825, "decode_ms": 455.9718749951571}, "total_wall_ms": 87665.54012498818, "ttft_ms": 35.98854201845825, "input_usage": {"lookups": 12}, "output_usage": {"output_tokens": 7}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "13.4343", "evidence_artifact": {"ledger": {"facts_total": 59, "facts_retained": 30, "retention_rate": 0.5084745762711864, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.5714285714285714, "amount": 0.6, "status": 0.5833333333333334, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1073}, "ledger_by_hop": {"1": {"facts_total": 59, "facts_retained": 30, "retention_rate": 0.5084745762711864, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.5714285714285714, "amount": 0.6, "status": 0.5833333333333334, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1213}, "2": {"facts_total": 59, "facts_retained": 30, "retention_rate": 0.5084745762711864, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.5714285714285714, "amount": 0.6, "status": 0.5833333333333334, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1105}, "3": {"facts_total": 59, "facts_retained": 30, "retention_rate": 0.5084745762711864, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.5714285714285714, "amount": 0.6, "status": 0.5833333333333334, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1085}, "4": {"facts_total": 59, "facts_retained": 30, "retention_rate": 0.5084745762711864, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.5714285714285714, "amount": 0.6, "status": 0.5833333333333334, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1073}}}, "reference_key_id": "keys/synthetic/syn_03012_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "5aeedc1e4643b829686179704346397792278190d32a28471247e6a066095c87", "output_hash": "599bc22e373b1cf7b4679208eb9e515ad0e8dcb88dde0c7f579637ec7d3f3014"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "40e01b3fe779fa5282e57fb8", "attempt_id": "01f5d127e7ff", "run_group_id": "E4_core:A4|repairable|qwen3b:h4", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:37:23.174392Z", "task_family": "synthetic", "document_id": "syn_03012_2000", "document_hash": "03e441e68d54e1af036f8ff905dfd6cbbccead62636279c1a8e179352ab08945", "question_id": "q03_exact_copy", "split": "test", "cluster_id": "syn_03012_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 4, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1927, "query_tokens": 562, "message_tokens": 479, "message_utf8_bytes": 1073, "cache_tensor_bytes": null, "serialized_payload_bytes": 1073, "side_information_bytes": 0, "fetched_bytes": 2193, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 86838.3012090344, "hops": {"1": 22829.28358297795, "2": 21115.020749974065, "3": 22152.291209087707, "4": 20741.705666994676}, "receiver_prefill_ms": 377.87416600622237, "ttft_ms": 27.163916965946555, "decode_ms": 37.758874939754605}, "total_wall_ms": 87281.69741702732, "ttft_ms": 27.163916965946555, "input_usage": {"lookups": 12}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 59, "facts_retained": 30, "retention_rate": 0.5084745762711864, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.5714285714285714, "amount": 0.6, "status": 0.5833333333333334, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1073}, "ledger_by_hop": {"1": {"facts_total": 59, "facts_retained": 30, "retention_rate": 0.5084745762711864, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.5714285714285714, "amount": 0.6, "status": 0.5833333333333334, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1213}, "2": {"facts_total": 59, "facts_retained": 30, "retention_rate": 0.5084745762711864, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.5714285714285714, "amount": 0.6, "status": 0.5833333333333334, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1105}, "3": {"facts_total": 59, "facts_retained": 30, "retention_rate": 0.5084745762711864, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.5714285714285714, "amount": 0.6, "status": 0.5833333333333334, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1085}, "4": {"facts_total": 59, "facts_retained": 30, "retention_rate": 0.5084745762711864, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.5714285714285714, "amount": 0.6, "status": 0.5833333333333334, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1073}}}, "reference_key_id": "keys/synthetic/syn_03012_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "5aeedc1e4643b829686179704346397792278190d32a28471247e6a066095c87", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "569188a2b428ae7512c285ae", "attempt_id": "b8d5ac26348f", "run_group_id": "E4_core:A4|repairable|qwen3b:h4", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:37:23.635383Z", "task_family": "synthetic", "document_id": "syn_03012_2000", "document_hash": "03e441e68d54e1af036f8ff905dfd6cbbccead62636279c1a8e179352ab08945", "question_id": "q04_multi_hop", "split": "test", "cluster_id": "syn_03012_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 4, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1927, "query_tokens": 567, "message_tokens": 479, "message_utf8_bytes": 1073, "cache_tensor_bytes": null, "serialized_payload_bytes": 1073, "side_information_bytes": 0, "fetched_bytes": 2193, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 86838.3012090344, "hops": {"1": 22829.28358297795, "2": 21115.020749974065, "3": 22152.291209087707, "4": 20741.705666994676}, "receiver_prefill_ms": 390.21054201293737, "ttft_ms": 29.856499982997775, "decode_ms": 467.5596250453964}, "total_wall_ms": 87726.70679213479, "ttft_ms": 29.856499982997775, "input_usage": {"lookups": 12}, "output_usage": {"output_tokens": 9}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "8,631,552", "evidence_artifact": {"ledger": {"facts_total": 59, "facts_retained": 30, "retention_rate": 0.5084745762711864, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.5714285714285714, "amount": 0.6, "status": 0.5833333333333334, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1073}, "ledger_by_hop": {"1": {"facts_total": 59, "facts_retained": 30, "retention_rate": 0.5084745762711864, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.5714285714285714, "amount": 0.6, "status": 0.5833333333333334, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1213}, "2": {"facts_total": 59, "facts_retained": 30, "retention_rate": 0.5084745762711864, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.5714285714285714, "amount": 0.6, "status": 0.5833333333333334, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1105}, "3": {"facts_total": 59, "facts_retained": 30, "retention_rate": 0.5084745762711864, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.5714285714285714, "amount": 0.6, "status": 0.5833333333333334, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1085}, "4": {"facts_total": 59, "facts_retained": 30, "retention_rate": 0.5084745762711864, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.5714285714285714, "amount": 0.6, "status": 0.5833333333333334, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1073}}}, "reference_key_id": "keys/synthetic/syn_03012_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "5aeedc1e4643b829686179704346397792278190d32a28471247e6a066095c87", "output_hash": "cecdbfc66b98662e79ca5deba218a2b11cf8d9e18b8f3d91c3032acf0c41fc29"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "346f5aaf8c5342df21276f66", "attempt_id": "d7b881dc77ef", "run_group_id": "E4_core:A4|repairable|qwen3b:h4", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:37:24.542167Z", "task_family": "synthetic", "document_id": "syn_03012_2000", "document_hash": "03e441e68d54e1af036f8ff905dfd6cbbccead62636279c1a8e179352ab08945", "question_id": "q05_multi_hop", "split": "test", "cluster_id": "syn_03012_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 4, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1927, "query_tokens": 570, "message_tokens": 479, "message_utf8_bytes": 1073, "cache_tensor_bytes": null, "serialized_payload_bytes": 1073, "side_information_bytes": 0, "fetched_bytes": 2193, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 86838.3012090344, "hops": {"1": 22829.28358297795, "2": 21115.020749974065, "3": 22152.291209087707, "4": 20741.705666994676}, "receiver_prefill_ms": 456.7992090014741, "ttft_ms": 42.54825005773455, "decode_ms": 321.63995795417577}, "total_wall_ms": 87659.95487605687, "ttft_ms": 42.54825005773455, "input_usage": {"lookups": 12}, "output_usage": {"output_tokens": 10}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "DG-40E-2777", "evidence_artifact": {"ledger": {"facts_total": 59, "facts_retained": 30, "retention_rate": 0.5084745762711864, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.5714285714285714, "amount": 0.6, "status": 0.5833333333333334, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1073}, "ledger_by_hop": {"1": {"facts_total": 59, "facts_retained": 30, "retention_rate": 0.5084745762711864, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.5714285714285714, "amount": 0.6, "status": 0.5833333333333334, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1213}, "2": {"facts_total": 59, "facts_retained": 30, "retention_rate": 0.5084745762711864, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.5714285714285714, "amount": 0.6, "status": 0.5833333333333334, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1105}, "3": {"facts_total": 59, "facts_retained": 30, "retention_rate": 0.5084745762711864, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.5714285714285714, "amount": 0.6, "status": 0.5833333333333334, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1085}, "4": {"facts_total": 59, "facts_retained": 30, "retention_rate": 0.5084745762711864, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.5714285714285714, "amount": 0.6, "status": 0.5833333333333334, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1073}}}, "reference_key_id": "keys/synthetic/syn_03012_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "5aeedc1e4643b829686179704346397792278190d32a28471247e6a066095c87", "output_hash": "16a7e3f4a7cc4a9b66f86c2cdcfad3b35daf1da199d9406e0627c8af2d8d73f3"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "91e5b41b128ac9ed619cdbe8", "attempt_id": "802af33cba3e", "run_group_id": "E4_core:A4|repairable|qwen3b:h4", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:37:25.383953Z", "task_family": "synthetic", "document_id": "syn_03012_2000", "document_hash": "03e441e68d54e1af036f8ff905dfd6cbbccead62636279c1a8e179352ab08945", "question_id": "q06_negation", "split": "test", "cluster_id": "syn_03012_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 4, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1927, "query_tokens": 562, "message_tokens": 479, "message_utf8_bytes": 1073, "cache_tensor_bytes": null, "serialized_payload_bytes": 1073, "side_information_bytes": 0, "fetched_bytes": 2193, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 86838.3012090344, "hops": {"1": 22829.28358297795, "2": 21115.020749974065, "3": 22152.291209087707, "4": 20741.705666994676}, "receiver_prefill_ms": 338.2355000358075, "ttft_ms": 35.809499910101295, "decode_ms": 49.48175000026822}, "total_wall_ms": 87262.52304203808, "ttft_ms": 35.809499910101295, "input_usage": {"lookups": 12}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "no", "evidence_artifact": {"ledger": {"facts_total": 59, "facts_retained": 30, "retention_rate": 0.5084745762711864, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.5714285714285714, "amount": 0.6, "status": 0.5833333333333334, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1073}, "ledger_by_hop": {"1": {"facts_total": 59, "facts_retained": 30, "retention_rate": 0.5084745762711864, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.5714285714285714, "amount": 0.6, "status": 0.5833333333333334, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1213}, "2": {"facts_total": 59, "facts_retained": 30, "retention_rate": 0.5084745762711864, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.5714285714285714, "amount": 0.6, "status": 0.5833333333333334, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1105}, "3": {"facts_total": 59, "facts_retained": 30, "retention_rate": 0.5084745762711864, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.5714285714285714, "amount": 0.6, "status": 0.5833333333333334, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1085}, "4": {"facts_total": 59, "facts_retained": 30, "retention_rate": 0.5084745762711864, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.5714285714285714, "amount": 0.6, "status": 0.5833333333333334, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1073}}}, "reference_key_id": "keys/synthetic/syn_03012_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "5aeedc1e4643b829686179704346397792278190d32a28471247e6a066095c87", "output_hash": "9390298f3fb0c5b160498935d79cb139aef28e1c47358b4bbba61862b9c26e59"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "e8e49cc5d12eedf76c137e7c", "attempt_id": "11da94107322", "run_group_id": "E4_core:A4|repairable|qwen3b:h4", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:37:25.825668Z", "task_family": "synthetic", "document_id": "syn_03012_2000", "document_hash": "03e441e68d54e1af036f8ff905dfd6cbbccead62636279c1a8e179352ab08945", "question_id": "q07_negation", "split": "test", "cluster_id": "syn_03012_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 4, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1927, "query_tokens": 562, "message_tokens": 479, "message_utf8_bytes": 1073, "cache_tensor_bytes": null, "serialized_payload_bytes": 1073, "side_information_bytes": 0, "fetched_bytes": 2193, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 86838.3012090344, "hops": {"1": 22829.28358297795, "2": 21115.020749974065, "3": 22152.291209087707, "4": 20741.705666994676}, "receiver_prefill_ms": 440.43070904444903, "ttft_ms": 89.33412504848093, "decode_ms": 39.96416600421071}, "total_wall_ms": 87408.66929211188, "ttft_ms": 89.33412504848093, "input_usage": {"lookups": 12}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "no", "evidence_artifact": {"ledger": {"facts_total": 59, "facts_retained": 30, "retention_rate": 0.5084745762711864, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.5714285714285714, "amount": 0.6, "status": 0.5833333333333334, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1073}, "ledger_by_hop": {"1": {"facts_total": 59, "facts_retained": 30, "retention_rate": 0.5084745762711864, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.5714285714285714, "amount": 0.6, "status": 0.5833333333333334, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1213}, "2": {"facts_total": 59, "facts_retained": 30, "retention_rate": 0.5084745762711864, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.5714285714285714, "amount": 0.6, "status": 0.5833333333333334, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1105}, "3": {"facts_total": 59, "facts_retained": 30, "retention_rate": 0.5084745762711864, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.5714285714285714, "amount": 0.6, "status": 0.5833333333333334, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1085}, "4": {"facts_total": 59, "facts_retained": 30, "retention_rate": 0.5084745762711864, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.5714285714285714, "amount": 0.6, "status": 0.5833333333333334, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1073}}}, "reference_key_id": "keys/synthetic/syn_03012_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "5aeedc1e4643b829686179704346397792278190d32a28471247e6a066095c87", "output_hash": "9390298f3fb0c5b160498935d79cb139aef28e1c47358b4bbba61862b9c26e59"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "b2d46bad5920af48cfeae3a7", "attempt_id": "dc91ac6e08a7", "run_group_id": "E4_core:A4|repairable|qwen3b:h4", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:37:26.414658Z", "task_family": "synthetic", "document_id": "syn_03012_2000", "document_hash": "03e441e68d54e1af036f8ff905dfd6cbbccead62636279c1a8e179352ab08945", "question_id": "q08_revision", "split": "test", "cluster_id": "syn_03012_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 4, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1927, "query_tokens": 577, "message_tokens": 479, "message_utf8_bytes": 1073, "cache_tensor_bytes": null, "serialized_payload_bytes": 1073, "side_information_bytes": 0, "fetched_bytes": 2193, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 86838.3012090344, "hops": {"1": 22829.28358297795, "2": 21115.020749974065, "3": 22152.291209087707, "4": 20741.705666994676}, "receiver_prefill_ms": 384.3592079356313, "ttft_ms": 21.241209004074335, "decode_ms": 231.60729196388274}, "total_wall_ms": 87476.26337502152, "ttft_ms": 21.241209004074335, "input_usage": {"lookups": 12}, "output_usage": {"output_tokens": 7}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "6602245", "evidence_artifact": {"ledger": {"facts_total": 59, "facts_retained": 30, "retention_rate": 0.5084745762711864, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.5714285714285714, "amount": 0.6, "status": 0.5833333333333334, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1073}, "ledger_by_hop": {"1": {"facts_total": 59, "facts_retained": 30, "retention_rate": 0.5084745762711864, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.5714285714285714, "amount": 0.6, "status": 0.5833333333333334, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1213}, "2": {"facts_total": 59, "facts_retained": 30, "retention_rate": 0.5084745762711864, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.5714285714285714, "amount": 0.6, "status": 0.5833333333333334, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1105}, "3": {"facts_total": 59, "facts_retained": 30, "retention_rate": 0.5084745762711864, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.5714285714285714, "amount": 0.6, "status": 0.5833333333333334, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1085}, "4": {"facts_total": 59, "facts_retained": 30, "retention_rate": 0.5084745762711864, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.5714285714285714, "amount": 0.6, "status": 0.5833333333333334, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1073}}}, "reference_key_id": "keys/synthetic/syn_03012_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "5aeedc1e4643b829686179704346397792278190d32a28471247e6a066095c87", "output_hash": "d117fa8197988f8d1e4fabc6c4cdaab9516018e54553b57c63b875b3de82b16f"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "aba8d2490332932d665aeaf2", "attempt_id": "c4685acd4763", "run_group_id": "E4_core:A4|repairable|qwen3b:h4", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:37:27.070071Z", "task_family": "synthetic", "document_id": "syn_03012_2000", "document_hash": "03e441e68d54e1af036f8ff905dfd6cbbccead62636279c1a8e179352ab08945", "question_id": "q09_revision", "split": "test", "cluster_id": "syn_03012_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 4, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1927, "query_tokens": 570, "message_tokens": 479, "message_utf8_bytes": 1073, "cache_tensor_bytes": null, "serialized_payload_bytes": 1073, "side_information_bytes": 0, "fetched_bytes": 2193, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 86838.3012090344, "hops": {"1": 22829.28358297795, "2": 21115.020749974065, "3": 22152.291209087707, "4": 20741.705666994676}, "receiver_prefill_ms": 351.53387498576194, "ttft_ms": 62.95387493446469, "decode_ms": 465.9875830402598}, "total_wall_ms": 87719.40920909401, "ttft_ms": 62.95387493446469, "input_usage": {"lookups": 12}, "output_usage": {"output_tokens": 7}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "5899340", "evidence_artifact": {"ledger": {"facts_total": 59, "facts_retained": 30, "retention_rate": 0.5084745762711864, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.5714285714285714, "amount": 0.6, "status": 0.5833333333333334, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1073}, "ledger_by_hop": {"1": {"facts_total": 59, "facts_retained": 30, "retention_rate": 0.5084745762711864, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.5714285714285714, "amount": 0.6, "status": 0.5833333333333334, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1213}, "2": {"facts_total": 59, "facts_retained": 30, "retention_rate": 0.5084745762711864, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.5714285714285714, "amount": 0.6, "status": 0.5833333333333334, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1105}, "3": {"facts_total": 59, "facts_retained": 30, "retention_rate": 0.5084745762711864, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.5714285714285714, "amount": 0.6, "status": 0.5833333333333334, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1085}, "4": {"facts_total": 59, "facts_retained": 30, "retention_rate": 0.5084745762711864, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.5714285714285714, "amount": 0.6, "status": 0.5833333333333334, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1073}}}, "reference_key_id": "keys/synthetic/syn_03012_2000.json", "task_scores": {"correct": 0.0, "abstained": 0.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "5aeedc1e4643b829686179704346397792278190d32a28471247e6a066095c87", "output_hash": "c920323a9a02b1afb8619cf11bdde427f7cd1ea3289be69bd62932985895d717"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "a8a41452ca9662023e3e9d93", "attempt_id": "ce307736dd66", "run_group_id": "E4_core:A4|repairable|qwen3b:h4", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:37:27.968337Z", "task_family": "synthetic", "document_id": "syn_03012_2000", "document_hash": "03e441e68d54e1af036f8ff905dfd6cbbccead62636279c1a8e179352ab08945", "question_id": "q10_temporal", "split": "test", "cluster_id": "syn_03012_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 4, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1927, "query_tokens": 580, "message_tokens": 479, "message_utf8_bytes": 1073, "cache_tensor_bytes": null, "serialized_payload_bytes": 1073, "side_information_bytes": 0, "fetched_bytes": 2193, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 86838.3012090344, "hops": {"1": 22829.28358297795, "2": 21115.020749974065, "3": 22152.291209087707, "4": 20741.705666994676}, "receiver_prefill_ms": 407.3276250855997, "ttft_ms": 26.908167055808008, "decode_ms": 36.25258302781731}, "total_wall_ms": 87309.37891697977, "ttft_ms": 26.908167055808008, "input_usage": {"lookups": 12}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 59, "facts_retained": 30, "retention_rate": 0.5084745762711864, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.5714285714285714, "amount": 0.6, "status": 0.5833333333333334, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1073}, "ledger_by_hop": {"1": {"facts_total": 59, "facts_retained": 30, "retention_rate": 0.5084745762711864, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.5714285714285714, "amount": 0.6, "status": 0.5833333333333334, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1213}, "2": {"facts_total": 59, "facts_retained": 30, "retention_rate": 0.5084745762711864, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.5714285714285714, "amount": 0.6, "status": 0.5833333333333334, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1105}, "3": {"facts_total": 59, "facts_retained": 30, "retention_rate": 0.5084745762711864, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.5714285714285714, "amount": 0.6, "status": 0.5833333333333334, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1085}, "4": {"facts_total": 59, "facts_retained": 30, "retention_rate": 0.5084745762711864, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.5714285714285714, "amount": 0.6, "status": 0.5833333333333334, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1073}}}, "reference_key_id": "keys/synthetic/syn_03012_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "5aeedc1e4643b829686179704346397792278190d32a28471247e6a066095c87", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "6910f5c531682a648ae6fa89", "attempt_id": "4575be2fe33d", "run_group_id": "E4_core:A4|repairable|qwen3b:h4", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:37:28.456542Z", "task_family": "synthetic", "document_id": "syn_03012_2000", "document_hash": "03e441e68d54e1af036f8ff905dfd6cbbccead62636279c1a8e179352ab08945", "question_id": "q11_aggregation", "split": "test", "cluster_id": "syn_03012_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 4, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1927, "query_tokens": 572, "message_tokens": 479, "message_utf8_bytes": 1073, "cache_tensor_bytes": null, "serialized_payload_bytes": 1073, "side_information_bytes": 0, "fetched_bytes": 2193, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 86838.3012090344, "hops": {"1": 22829.28358297795, "2": 21115.020749974065, "3": 22152.291209087707, "4": 20741.705666994676}, "receiver_prefill_ms": 377.92654195800424, "ttft_ms": 26.91954094916582, "decode_ms": 285.5886670295149}, "total_wall_ms": 87529.53037607949, "ttft_ms": 26.91954094916582, "input_usage": {"lookups": 12}, "output_usage": {"output_tokens": 8}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "19404677", "evidence_artifact": {"ledger": {"facts_total": 59, "facts_retained": 30, "retention_rate": 0.5084745762711864, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.5714285714285714, "amount": 0.6, "status": 0.5833333333333334, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1073}, "ledger_by_hop": {"1": {"facts_total": 59, "facts_retained": 30, "retention_rate": 0.5084745762711864, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.5714285714285714, "amount": 0.6, "status": 0.5833333333333334, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1213}, "2": {"facts_total": 59, "facts_retained": 30, "retention_rate": 0.5084745762711864, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.5714285714285714, "amount": 0.6, "status": 0.5833333333333334, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1105}, "3": {"facts_total": 59, "facts_retained": 30, "retention_rate": 0.5084745762711864, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.5714285714285714, "amount": 0.6, "status": 0.5833333333333334, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1085}, "4": {"facts_total": 59, "facts_retained": 30, "retention_rate": 0.5084745762711864, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.5714285714285714, "amount": 0.6, "status": 0.5833333333333334, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1073}}}, "reference_key_id": "keys/synthetic/syn_03012_2000.json", "task_scores": {"correct": 0.0, "abstained": 0.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "5aeedc1e4643b829686179704346397792278190d32a28471247e6a066095c87", "output_hash": "1c0b8d7c7978e679cd5aa402187714adeba47ec555f2e7b398b93a764a017984"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "53bf1ab85eab30844224e48a", "attempt_id": "3eb650430902", "run_group_id": "E4_core:A4|repairable|qwen3b:h4", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:37:29.165731Z", "task_family": "synthetic", "document_id": "syn_03012_2000", "document_hash": "03e441e68d54e1af036f8ff905dfd6cbbccead62636279c1a8e179352ab08945", "question_id": "q12_unit_conversion", "split": "test", "cluster_id": "syn_03012_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 4, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1927, "query_tokens": 569, "message_tokens": 479, "message_utf8_bytes": 1073, "cache_tensor_bytes": null, "serialized_payload_bytes": 1073, "side_information_bytes": 0, "fetched_bytes": 2193, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 86838.3012090344, "hops": {"1": 22829.28358297795, "2": 21115.020749974065, "3": 22152.291209087707, "4": 20741.705666994676}, "receiver_prefill_ms": 497.7218749700114, "ttft_ms": 38.57441700529307, "decode_ms": 150.9393750457093}, "total_wall_ms": 87526.27058397047, "ttft_ms": 38.57441700529307, "input_usage": {"lookups": 12}, "output_usage": {"output_tokens": 5}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "44620", "evidence_artifact": {"ledger": {"facts_total": 59, "facts_retained": 30, "retention_rate": 0.5084745762711864, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.5714285714285714, "amount": 0.6, "status": 0.5833333333333334, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1073}, "ledger_by_hop": {"1": {"facts_total": 59, "facts_retained": 30, "retention_rate": 0.5084745762711864, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.5714285714285714, "amount": 0.6, "status": 0.5833333333333334, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1213}, "2": {"facts_total": 59, "facts_retained": 30, "retention_rate": 0.5084745762711864, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.5714285714285714, "amount": 0.6, "status": 0.5833333333333334, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1105}, "3": {"facts_total": 59, "facts_retained": 30, "retention_rate": 0.5084745762711864, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.5714285714285714, "amount": 0.6, "status": 0.5833333333333334, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1085}, "4": {"facts_total": 59, "facts_retained": 30, "retention_rate": 0.5084745762711864, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.5714285714285714, "amount": 0.6, "status": 0.5833333333333334, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1073}}}, "reference_key_id": "keys/synthetic/syn_03012_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "5aeedc1e4643b829686179704346397792278190d32a28471247e6a066095c87", "output_hash": "2bd125847727be7e7c0a656ee4c81f3bc6d0fef11f536f50ce7bc83688907075"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "2b084f14e39b74f5c7093b26", "attempt_id": "7b913e299b96", "run_group_id": "E4_core:A4|repairable|qwen3b:h4", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:37:29.872158Z", "task_family": "synthetic", "document_id": "syn_03012_2000", "document_hash": "03e441e68d54e1af036f8ff905dfd6cbbccead62636279c1a8e179352ab08945", "question_id": "q13_unanswerable", "split": "test", "cluster_id": "syn_03012_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 4, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1927, "query_tokens": 574, "message_tokens": 479, "message_utf8_bytes": 1073, "cache_tensor_bytes": null, "serialized_payload_bytes": 1073, "side_information_bytes": 0, "fetched_bytes": 2193, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 86838.3012090344, "hops": {"1": 22829.28358297795, "2": 21115.020749974065, "3": 22152.291209087707, "4": 20741.705666994676}, "receiver_prefill_ms": 377.0999580156058, "ttft_ms": 26.28520899452269, "decode_ms": 39.933416061103344}, "total_wall_ms": 87282.60999999475, "ttft_ms": 26.28520899452269, "input_usage": {"lookups": 12}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 59, "facts_retained": 30, "retention_rate": 0.5084745762711864, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.5714285714285714, "amount": 0.6, "status": 0.5833333333333334, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1073}, "ledger_by_hop": {"1": {"facts_total": 59, "facts_retained": 30, "retention_rate": 0.5084745762711864, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.5714285714285714, "amount": 0.6, "status": 0.5833333333333334, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1213}, "2": {"facts_total": 59, "facts_retained": 30, "retention_rate": 0.5084745762711864, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.5714285714285714, "amount": 0.6, "status": 0.5833333333333334, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1105}, "3": {"facts_total": 59, "facts_retained": 30, "retention_rate": 0.5084745762711864, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.5714285714285714, "amount": 0.6, "status": 0.5833333333333334, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1085}, "4": {"facts_total": 59, "facts_retained": 30, "retention_rate": 0.5084745762711864, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.5714285714285714, "amount": 0.6, "status": 0.5833333333333334, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1073}}}, "reference_key_id": "keys/synthetic/syn_03012_2000.json", "task_scores": {"correct": 1.0, "abstained": 1.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "5aeedc1e4643b829686179704346397792278190d32a28471247e6a066095c87", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "311cfa2e54aeac66a938bb03", "attempt_id": "f6a613a032db", "run_group_id": "E4_core:A4|repairable|qwen3b:h4", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:37:30.335120Z", "task_family": "synthetic", "document_id": "syn_03012_2000", "document_hash": "03e441e68d54e1af036f8ff905dfd6cbbccead62636279c1a8e179352ab08945", "question_id": "q14_unanswerable", "split": "test", "cluster_id": "syn_03012_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 4, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1927, "query_tokens": 574, "message_tokens": 479, "message_utf8_bytes": 1073, "cache_tensor_bytes": null, "serialized_payload_bytes": 1073, "side_information_bytes": 0, "fetched_bytes": 2193, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 86838.3012090344, "hops": {"1": 22829.28358297795, "2": 21115.020749974065, "3": 22152.291209087707, "4": 20741.705666994676}, "receiver_prefill_ms": 341.67512506246567, "ttft_ms": 51.633625058457255, "decode_ms": 37.760416977107525}, "total_wall_ms": 87270.2658760827, "ttft_ms": 51.633625058457255, "input_usage": {"lookups": 12}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 59, "facts_retained": 30, "retention_rate": 0.5084745762711864, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.5714285714285714, "amount": 0.6, "status": 0.5833333333333334, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1073}, "ledger_by_hop": {"1": {"facts_total": 59, "facts_retained": 30, "retention_rate": 0.5084745762711864, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.5714285714285714, "amount": 0.6, "status": 0.5833333333333334, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1213}, "2": {"facts_total": 59, "facts_retained": 30, "retention_rate": 0.5084745762711864, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.5714285714285714, "amount": 0.6, "status": 0.5833333333333334, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1105}, "3": {"facts_total": 59, "facts_retained": 30, "retention_rate": 0.5084745762711864, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.5714285714285714, "amount": 0.6, "status": 0.5833333333333334, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1085}, "4": {"facts_total": 59, "facts_retained": 30, "retention_rate": 0.5084745762711864, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.5714285714285714, "amount": 0.6, "status": 0.5833333333333334, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1073}}}, "reference_key_id": "keys/synthetic/syn_03012_2000.json", "task_scores": {"correct": 1.0, "abstained": 1.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "5aeedc1e4643b829686179704346397792278190d32a28471247e6a066095c87", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "2b90afa481f76e13f32e22db", "attempt_id": "db4c734ff360", "run_group_id": "E4_core:A4|repairable|qwen3b:h8", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:38:53.733102Z", "task_family": "synthetic", "document_id": "syn_03012_2000", "document_hash": "03e441e68d54e1af036f8ff905dfd6cbbccead62636279c1a8e179352ab08945", "question_id": "q00_single_fact", "split": "test", "cluster_id": "syn_03012_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 8, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1927, "query_tokens": 561, "message_tokens": 479, "message_utf8_bytes": 1073, "cache_tensor_bytes": null, "serialized_payload_bytes": 1073, "side_information_bytes": 0, "fetched_bytes": 5117, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 169780.56020801887, "hops": {"1": 22829.28358297795, "2": 21115.020749974065, "3": 22152.291209087707, "4": 20741.705666994676, "5": 20553.900958038867, "6": 20653.28154200688, "7": 21087.469290941954, "8": 20647.60720799677}, "receiver_prefill_ms": 366.38545896857977, "ttft_ms": 26.174125028774142, "decode_ms": 630.4115409730002}, "total_wall_ms": 170804.29583298974, "ttft_ms": 26.174125028774142, "input_usage": {"lookups": 28}, "output_usage": {"output_tokens": 10}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "XX-71M-6213", "evidence_artifact": {"ledger": {"facts_total": 59, "facts_retained": 30, "retention_rate": 0.5084745762711864, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.5714285714285714, "amount": 0.6, "status": 0.5833333333333334, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1073}, "ledger_by_hop": {"1": {"facts_total": 59, "facts_retained": 30, "retention_rate": 0.5084745762711864, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.5714285714285714, "amount": 0.6, "status": 0.5833333333333334, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1213}, "2": {"facts_total": 59, "facts_retained": 30, "retention_rate": 0.5084745762711864, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.5714285714285714, "amount": 0.6, "status": 0.5833333333333334, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1105}, "3": {"facts_total": 59, "facts_retained": 30, "retention_rate": 0.5084745762711864, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.5714285714285714, "amount": 0.6, "status": 0.5833333333333334, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1085}, "4": {"facts_total": 59, "facts_retained": 30, "retention_rate": 0.5084745762711864, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.5714285714285714, "amount": 0.6, "status": 0.5833333333333334, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1073}, "5": {"facts_total": 59, "facts_retained": 30, "retention_rate": 0.5084745762711864, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.5714285714285714, "amount": 0.6, "status": 0.5833333333333334, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1073}, "6": {"facts_total": 59, "facts_retained": 30, "retention_rate": 0.5084745762711864, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.5714285714285714, "amount": 0.6, "status": 0.5833333333333334, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1073}, "7": {"facts_total": 59, "facts_retained": 30, "retention_rate": 0.5084745762711864, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.5714285714285714, "amount": 0.6, "status": 0.5833333333333334, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1073}, "8": {"facts_total": 59, "facts_retained": 30, "retention_rate": 0.5084745762711864, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.5714285714285714, "amount": 0.6, "status": 0.5833333333333334, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1073}}}, "reference_key_id": "keys/synthetic/syn_03012_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "5aeedc1e4643b829686179704346397792278190d32a28471247e6a066095c87", "output_hash": "66ef6c5da682b38795e9029723bbd71cdde2c4dec70cf1fae6efe62ef243f10a"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "a077dbd2932f9e2795bf1439", "attempt_id": "213fc57c07be", "run_group_id": "E4_core:A4|repairable|qwen3b:h8", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:38:54.774600Z", "task_family": "synthetic", "document_id": "syn_03012_2000", "document_hash": "03e441e68d54e1af036f8ff905dfd6cbbccead62636279c1a8e179352ab08945", "question_id": "q01_single_fact", "split": "test", "cluster_id": "syn_03012_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 8, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1927, "query_tokens": 567, "message_tokens": 479, "message_utf8_bytes": 1073, "cache_tensor_bytes": null, "serialized_payload_bytes": 1073, "side_information_bytes": 0, "fetched_bytes": 5117, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 169780.56020801887, "hops": {"1": 22829.28358297795, "2": 21115.020749974065, "3": 22152.291209087707, "4": 20741.705666994676, "5": 20553.900958038867, "6": 20653.28154200688, "7": 21087.469290941954, "8": 20647.60720799677}, "receiver_prefill_ms": 371.881666011177, "ttft_ms": 43.15941606182605, "decode_ms": 27.149375062435865}, "total_wall_ms": 170223.46845804714, "ttft_ms": 43.15941606182605, "input_usage": {"lookups": 28}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 59, "facts_retained": 30, "retention_rate": 0.5084745762711864, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.5714285714285714, "amount": 0.6, "status": 0.5833333333333334, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1073}, "ledger_by_hop": {"1": {"facts_total": 59, "facts_retained": 30, "retention_rate": 0.5084745762711864, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.5714285714285714, "amount": 0.6, "status": 0.5833333333333334, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1213}, "2": {"facts_total": 59, "facts_retained": 30, "retention_rate": 0.5084745762711864, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.5714285714285714, "amount": 0.6, "status": 0.5833333333333334, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1105}, "3": {"facts_total": 59, "facts_retained": 30, "retention_rate": 0.5084745762711864, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.5714285714285714, "amount": 0.6, "status": 0.5833333333333334, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1085}, "4": {"facts_total": 59, "facts_retained": 30, "retention_rate": 0.5084745762711864, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.5714285714285714, "amount": 0.6, "status": 0.5833333333333334, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1073}, "5": {"facts_total": 59, "facts_retained": 30, "retention_rate": 0.5084745762711864, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.5714285714285714, "amount": 0.6, "status": 0.5833333333333334, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1073}, "6": {"facts_total": 59, "facts_retained": 30, "retention_rate": 0.5084745762711864, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.5714285714285714, "amount": 0.6, "status": 0.5833333333333334, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1073}, "7": {"facts_total": 59, "facts_retained": 30, "retention_rate": 0.5084745762711864, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.5714285714285714, "amount": 0.6, "status": 0.5833333333333334, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1073}, "8": {"facts_total": 59, "facts_retained": 30, "retention_rate": 0.5084745762711864, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.5714285714285714, "amount": 0.6, "status": 0.5833333333333334, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1073}}}, "reference_key_id": "keys/synthetic/syn_03012_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "5aeedc1e4643b829686179704346397792278190d32a28471247e6a066095c87", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "d013697798d3c811fa49440a", "attempt_id": "17840dbf2fe4", "run_group_id": "E4_core:A4|repairable|qwen3b:h8", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:38:55.234769Z", "task_family": "synthetic", "document_id": "syn_03012_2000", "document_hash": "03e441e68d54e1af036f8ff905dfd6cbbccead62636279c1a8e179352ab08945", "question_id": "q02_exact_copy", "split": "test", "cluster_id": "syn_03012_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 8, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1927, "query_tokens": 566, "message_tokens": 479, "message_utf8_bytes": 1073, "cache_tensor_bytes": null, "serialized_payload_bytes": 1073, "side_information_bytes": 0, "fetched_bytes": 5117, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 169780.56020801887, "hops": {"1": 22829.28358297795, "2": 21115.020749974065, "3": 22152.291209087707, "4": 20741.705666994676, "5": 20553.900958038867, "6": 20653.28154200688, "7": 21087.469290941954, "8": 20647.60720799677}, "receiver_prefill_ms": 383.3994579035789, "ttft_ms": 25.47766698990017, "decode_ms": 367.8553330246359}, "total_wall_ms": 170558.0902080983, "ttft_ms": 25.47766698990017, "input_usage": {"lookups": 28}, "output_usage": {"output_tokens": 7}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "13.4343", "evidence_artifact": {"ledger": {"facts_total": 59, "facts_retained": 30, "retention_rate": 0.5084745762711864, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.5714285714285714, "amount": 0.6, "status": 0.5833333333333334, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1073}, "ledger_by_hop": {"1": {"facts_total": 59, "facts_retained": 30, "retention_rate": 0.5084745762711864, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.5714285714285714, "amount": 0.6, "status": 0.5833333333333334, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1213}, "2": {"facts_total": 59, "facts_retained": 30, "retention_rate": 0.5084745762711864, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.5714285714285714, "amount": 0.6, "status": 0.5833333333333334, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1105}, "3": {"facts_total": 59, "facts_retained": 30, "retention_rate": 0.5084745762711864, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.5714285714285714, "amount": 0.6, "status": 0.5833333333333334, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1085}, "4": {"facts_total": 59, "facts_retained": 30, "retention_rate": 0.5084745762711864, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.5714285714285714, "amount": 0.6, "status": 0.5833333333333334, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1073}, "5": {"facts_total": 59, "facts_retained": 30, "retention_rate": 0.5084745762711864, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.5714285714285714, "amount": 0.6, "status": 0.5833333333333334, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1073}, "6": {"facts_total": 59, "facts_retained": 30, "retention_rate": 0.5084745762711864, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.5714285714285714, "amount": 0.6, "status": 0.5833333333333334, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1073}, "7": {"facts_total": 59, "facts_retained": 30, "retention_rate": 0.5084745762711864, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.5714285714285714, "amount": 0.6, "status": 0.5833333333333334, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1073}, "8": {"facts_total": 59, "facts_retained": 30, "retention_rate": 0.5084745762711864, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.5714285714285714, "amount": 0.6, "status": 0.5833333333333334, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1073}}}, "reference_key_id": "keys/synthetic/syn_03012_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "5aeedc1e4643b829686179704346397792278190d32a28471247e6a066095c87", "output_hash": "599bc22e373b1cf7b4679208eb9e515ad0e8dcb88dde0c7f579637ec7d3f3014"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "e8a58197b5de63d736a1350f", "attempt_id": "e5cc861d6ac4", "run_group_id": "E4_core:A4|repairable|qwen3b:h8", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:38:56.030115Z", "task_family": "synthetic", "document_id": "syn_03012_2000", "document_hash": "03e441e68d54e1af036f8ff905dfd6cbbccead62636279c1a8e179352ab08945", "question_id": "q03_exact_copy", "split": "test", "cluster_id": "syn_03012_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 8, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1927, "query_tokens": 562, "message_tokens": 479, "message_utf8_bytes": 1073, "cache_tensor_bytes": null, "serialized_payload_bytes": 1073, "side_information_bytes": 0, "fetched_bytes": 5117, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 169780.56020801887, "hops": {"1": 22829.28358297795, "2": 21115.020749974065, "3": 22152.291209087707, "4": 20741.705666994676, "5": 20553.900958038867, "6": 20653.28154200688, "7": 21087.469290941954, "8": 20647.60720799677}, "receiver_prefill_ms": 494.15529204998165, "ttft_ms": 37.69291704520583, "decode_ms": 23.292207974009216}, "total_wall_ms": 170336.37629100122, "ttft_ms": 37.69291704520583, "input_usage": {"lookups": 28}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 59, "facts_retained": 30, "retention_rate": 0.5084745762711864, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.5714285714285714, "amount": 0.6, "status": 0.5833333333333334, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1073}, "ledger_by_hop": {"1": {"facts_total": 59, "facts_retained": 30, "retention_rate": 0.5084745762711864, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.5714285714285714, "amount": 0.6, "status": 0.5833333333333334, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1213}, "2": {"facts_total": 59, "facts_retained": 30, "retention_rate": 0.5084745762711864, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.5714285714285714, "amount": 0.6, "status": 0.5833333333333334, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1105}, "3": {"facts_total": 59, "facts_retained": 30, "retention_rate": 0.5084745762711864, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.5714285714285714, "amount": 0.6, "status": 0.5833333333333334, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1085}, "4": {"facts_total": 59, "facts_retained": 30, "retention_rate": 0.5084745762711864, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.5714285714285714, "amount": 0.6, "status": 0.5833333333333334, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1073}, "5": {"facts_total": 59, "facts_retained": 30, "retention_rate": 0.5084745762711864, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.5714285714285714, "amount": 0.6, "status": 0.5833333333333334, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1073}, "6": {"facts_total": 59, "facts_retained": 30, "retention_rate": 0.5084745762711864, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.5714285714285714, "amount": 0.6, "status": 0.5833333333333334, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1073}, "7": {"facts_total": 59, "facts_retained": 30, "retention_rate": 0.5084745762711864, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.5714285714285714, "amount": 0.6, "status": 0.5833333333333334, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1073}, "8": {"facts_total": 59, "facts_retained": 30, "retention_rate": 0.5084745762711864, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.5714285714285714, "amount": 0.6, "status": 0.5833333333333334, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1073}}}, "reference_key_id": "keys/synthetic/syn_03012_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "5aeedc1e4643b829686179704346397792278190d32a28471247e6a066095c87", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "9a0b5a074372066de6962e3e", "attempt_id": "eea1c819916e", "run_group_id": "E4_core:A4|repairable|qwen3b:h8", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:38:56.603709Z", "task_family": "synthetic", "document_id": "syn_03012_2000", "document_hash": "03e441e68d54e1af036f8ff905dfd6cbbccead62636279c1a8e179352ab08945", "question_id": "q04_multi_hop", "split": "test", "cluster_id": "syn_03012_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 8, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1927, "query_tokens": 567, "message_tokens": 479, "message_utf8_bytes": 1073, "cache_tensor_bytes": null, "serialized_payload_bytes": 1073, "side_information_bytes": 0, "fetched_bytes": 5117, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 169780.56020801887, "hops": {"1": 22829.28358297795, "2": 21115.020749974065, "3": 22152.291209087707, "4": 20741.705666994676, "5": 20553.900958038867, "6": 20653.28154200688, "7": 21087.469290941954, "8": 20647.60720799677}, "receiver_prefill_ms": 384.2217499623075, "ttft_ms": 38.976875017397106, "decode_ms": 266.9179999502376}, "total_wall_ms": 170471.34808299597, "ttft_ms": 38.976875017397106, "input_usage": {"lookups": 28}, "output_usage": {"output_tokens": 9}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "8,631,552", "evidence_artifact": {"ledger": {"facts_total": 59, "facts_retained": 30, "retention_rate": 0.5084745762711864, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.5714285714285714, "amount": 0.6, "status": 0.5833333333333334, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1073}, "ledger_by_hop": {"1": {"facts_total": 59, "facts_retained": 30, "retention_rate": 0.5084745762711864, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.5714285714285714, "amount": 0.6, "status": 0.5833333333333334, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1213}, "2": {"facts_total": 59, "facts_retained": 30, "retention_rate": 0.5084745762711864, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.5714285714285714, "amount": 0.6, "status": 0.5833333333333334, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1105}, "3": {"facts_total": 59, "facts_retained": 30, "retention_rate": 0.5084745762711864, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.5714285714285714, "amount": 0.6, "status": 0.5833333333333334, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1085}, "4": {"facts_total": 59, "facts_retained": 30, "retention_rate": 0.5084745762711864, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.5714285714285714, "amount": 0.6, "status": 0.5833333333333334, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1073}, "5": {"facts_total": 59, "facts_retained": 30, "retention_rate": 0.5084745762711864, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.5714285714285714, "amount": 0.6, "status": 0.5833333333333334, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1073}, "6": {"facts_total": 59, "facts_retained": 30, "retention_rate": 0.5084745762711864, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.5714285714285714, "amount": 0.6, "status": 0.5833333333333334, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1073}, "7": {"facts_total": 59, "facts_retained": 30, "retention_rate": 0.5084745762711864, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.5714285714285714, "amount": 0.6, "status": 0.5833333333333334, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1073}, "8": {"facts_total": 59, "facts_retained": 30, "retention_rate": 0.5084745762711864, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.5714285714285714, "amount": 0.6, "status": 0.5833333333333334, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1073}}}, "reference_key_id": "keys/synthetic/syn_03012_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "5aeedc1e4643b829686179704346397792278190d32a28471247e6a066095c87", "output_hash": "cecdbfc66b98662e79ca5deba218a2b11cf8d9e18b8f3d91c3032acf0c41fc29"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "fa4d4d3566028acb402ef012", "attempt_id": "2439bf68b171", "run_group_id": "E4_core:A4|repairable|qwen3b:h8", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:38:57.312494Z", "task_family": "synthetic", "document_id": "syn_03012_2000", "document_hash": "03e441e68d54e1af036f8ff905dfd6cbbccead62636279c1a8e179352ab08945", "question_id": "q05_multi_hop", "split": "test", "cluster_id": "syn_03012_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 8, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1927, "query_tokens": 570, "message_tokens": 479, "message_utf8_bytes": 1073, "cache_tensor_bytes": null, "serialized_payload_bytes": 1073, "side_information_bytes": 0, "fetched_bytes": 5117, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 169780.56020801887, "hops": {"1": 22829.28358297795, "2": 21115.020749974065, "3": 22152.291209087707, "4": 20741.705666994676, "5": 20553.900958038867, "6": 20653.28154200688, "7": 21087.469290941954, "8": 20647.60720799677}, "receiver_prefill_ms": 378.50045901723206, "ttft_ms": 95.50837508868426, "decode_ms": 439.02708298992366}, "total_wall_ms": 170694.41799994092, "ttft_ms": 95.50837508868426, "input_usage": {"lookups": 28}, "output_usage": {"output_tokens": 10}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "DG-40E-2777", "evidence_artifact": {"ledger": {"facts_total": 59, "facts_retained": 30, "retention_rate": 0.5084745762711864, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.5714285714285714, "amount": 0.6, "status": 0.5833333333333334, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1073}, "ledger_by_hop": {"1": {"facts_total": 59, "facts_retained": 30, "retention_rate": 0.5084745762711864, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.5714285714285714, "amount": 0.6, "status": 0.5833333333333334, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1213}, "2": {"facts_total": 59, "facts_retained": 30, "retention_rate": 0.5084745762711864, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.5714285714285714, "amount": 0.6, "status": 0.5833333333333334, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1105}, "3": {"facts_total": 59, "facts_retained": 30, "retention_rate": 0.5084745762711864, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.5714285714285714, "amount": 0.6, "status": 0.5833333333333334, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1085}, "4": {"facts_total": 59, "facts_retained": 30, "retention_rate": 0.5084745762711864, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.5714285714285714, "amount": 0.6, "status": 0.5833333333333334, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1073}, "5": {"facts_total": 59, "facts_retained": 30, "retention_rate": 0.5084745762711864, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.5714285714285714, "amount": 0.6, "status": 0.5833333333333334, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1073}, "6": {"facts_total": 59, "facts_retained": 30, "retention_rate": 0.5084745762711864, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.5714285714285714, "amount": 0.6, "status": 0.5833333333333334, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1073}, "7": {"facts_total": 59, "facts_retained": 30, "retention_rate": 0.5084745762711864, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.5714285714285714, "amount": 0.6, "status": 0.5833333333333334, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1073}, "8": {"facts_total": 59, "facts_retained": 30, "retention_rate": 0.5084745762711864, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.5714285714285714, "amount": 0.6, "status": 0.5833333333333334, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1073}}}, "reference_key_id": "keys/synthetic/syn_03012_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "5aeedc1e4643b829686179704346397792278190d32a28471247e6a066095c87", "output_hash": "16a7e3f4a7cc4a9b66f86c2cdcfad3b35daf1da199d9406e0627c8af2d8d73f3"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "33c5fb3161a90fdd111f1d35", "attempt_id": "2f9ca100eea2", "run_group_id": "E4_core:A4|repairable|qwen3b:h8", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:38:58.244187Z", "task_family": "synthetic", "document_id": "syn_03012_2000", "document_hash": "03e441e68d54e1af036f8ff905dfd6cbbccead62636279c1a8e179352ab08945", "question_id": "q06_negation", "split": "test", "cluster_id": "syn_03012_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 8, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1927, "query_tokens": 562, "message_tokens": 479, "message_utf8_bytes": 1073, "cache_tensor_bytes": null, "serialized_payload_bytes": 1073, "side_information_bytes": 0, "fetched_bytes": 5117, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 169780.56020801887, "hops": {"1": 22829.28358297795, "2": 21115.020749974065, "3": 22152.291209087707, "4": 20741.705666994676, "5": 20553.900958038867, "6": 20653.28154200688, "7": 21087.469290941954, "8": 20647.60720799677}, "receiver_prefill_ms": 383.5202920017764, "ttft_ms": 27.116167009808123, "decode_ms": 34.766500000841916}, "total_wall_ms": 170226.65595798753, "ttft_ms": 27.116167009808123, "input_usage": {"lookups": 28}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "no", "evidence_artifact": {"ledger": {"facts_total": 59, "facts_retained": 30, "retention_rate": 0.5084745762711864, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.5714285714285714, "amount": 0.6, "status": 0.5833333333333334, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1073}, "ledger_by_hop": {"1": {"facts_total": 59, "facts_retained": 30, "retention_rate": 0.5084745762711864, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.5714285714285714, "amount": 0.6, "status": 0.5833333333333334, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1213}, "2": {"facts_total": 59, "facts_retained": 30, "retention_rate": 0.5084745762711864, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.5714285714285714, "amount": 0.6, "status": 0.5833333333333334, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1105}, "3": {"facts_total": 59, "facts_retained": 30, "retention_rate": 0.5084745762711864, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.5714285714285714, "amount": 0.6, "status": 0.5833333333333334, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1085}, "4": {"facts_total": 59, "facts_retained": 30, "retention_rate": 0.5084745762711864, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.5714285714285714, "amount": 0.6, "status": 0.5833333333333334, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1073}, "5": {"facts_total": 59, "facts_retained": 30, "retention_rate": 0.5084745762711864, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.5714285714285714, "amount": 0.6, "status": 0.5833333333333334, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1073}, "6": {"facts_total": 59, "facts_retained": 30, "retention_rate": 0.5084745762711864, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.5714285714285714, "amount": 0.6, "status": 0.5833333333333334, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1073}, "7": {"facts_total": 59, "facts_retained": 30, "retention_rate": 0.5084745762711864, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.5714285714285714, "amount": 0.6, "status": 0.5833333333333334, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1073}, "8": {"facts_total": 59, "facts_retained": 30, "retention_rate": 0.5084745762711864, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.5714285714285714, "amount": 0.6, "status": 0.5833333333333334, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1073}}}, "reference_key_id": "keys/synthetic/syn_03012_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "5aeedc1e4643b829686179704346397792278190d32a28471247e6a066095c87", "output_hash": "9390298f3fb0c5b160498935d79cb139aef28e1c47358b4bbba61862b9c26e59"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "b9ba7c76fd5b8adcdaf462fd", "attempt_id": "18b36541165e", "run_group_id": "E4_core:A4|repairable|qwen3b:h8", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:38:58.708398Z", "task_family": "synthetic", "document_id": "syn_03012_2000", "document_hash": "03e441e68d54e1af036f8ff905dfd6cbbccead62636279c1a8e179352ab08945", "question_id": "q07_negation", "split": "test", "cluster_id": "syn_03012_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 8, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1927, "query_tokens": 562, "message_tokens": 479, "message_utf8_bytes": 1073, "cache_tensor_bytes": null, "serialized_payload_bytes": 1073, "side_information_bytes": 0, "fetched_bytes": 5117, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 169780.56020801887, "hops": {"1": 22829.28358297795, "2": 21115.020749974065, "3": 22152.291209087707, "4": 20741.705666994676, "5": 20553.900958038867, "6": 20653.28154200688, "7": 21087.469290941954, "8": 20647.60720799677}, "receiver_prefill_ms": 334.2432089848444, "ttft_ms": 43.2859999127686, "decode_ms": 61.323917005211115}, "total_wall_ms": 170220.08495801128, "ttft_ms": 43.2859999127686, "input_usage": {"lookups": 28}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "no", "evidence_artifact": {"ledger": {"facts_total": 59, "facts_retained": 30, "retention_rate": 0.5084745762711864, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.5714285714285714, "amount": 0.6, "status": 0.5833333333333334, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1073}, "ledger_by_hop": {"1": {"facts_total": 59, "facts_retained": 30, "retention_rate": 0.5084745762711864, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.5714285714285714, "amount": 0.6, "status": 0.5833333333333334, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1213}, "2": {"facts_total": 59, "facts_retained": 30, "retention_rate": 0.5084745762711864, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.5714285714285714, "amount": 0.6, "status": 0.5833333333333334, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1105}, "3": {"facts_total": 59, "facts_retained": 30, "retention_rate": 0.5084745762711864, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.5714285714285714, "amount": 0.6, "status": 0.5833333333333334, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1085}, "4": {"facts_total": 59, "facts_retained": 30, "retention_rate": 0.5084745762711864, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.5714285714285714, "amount": 0.6, "status": 0.5833333333333334, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1073}, "5": {"facts_total": 59, "facts_retained": 30, "retention_rate": 0.5084745762711864, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.5714285714285714, "amount": 0.6, "status": 0.5833333333333334, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1073}, "6": {"facts_total": 59, "facts_retained": 30, "retention_rate": 0.5084745762711864, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.5714285714285714, "amount": 0.6, "status": 0.5833333333333334, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1073}, "7": {"facts_total": 59, "facts_retained": 30, "retention_rate": 0.5084745762711864, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.5714285714285714, "amount": 0.6, "status": 0.5833333333333334, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1073}, "8": {"facts_total": 59, "facts_retained": 30, "retention_rate": 0.5084745762711864, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.5714285714285714, "amount": 0.6, "status": 0.5833333333333334, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1073}}}, "reference_key_id": "keys/synthetic/syn_03012_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "5aeedc1e4643b829686179704346397792278190d32a28471247e6a066095c87", "output_hash": "9390298f3fb0c5b160498935d79cb139aef28e1c47358b4bbba61862b9c26e59"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "ec408e3cb61a7b4208a36444", "attempt_id": "b5c1c02a5884", "run_group_id": "E4_core:A4|repairable|qwen3b:h8", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:38:59.165547Z", "task_family": "synthetic", "document_id": "syn_03012_2000", "document_hash": "03e441e68d54e1af036f8ff905dfd6cbbccead62636279c1a8e179352ab08945", "question_id": "q08_revision", "split": "test", "cluster_id": "syn_03012_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 8, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1927, "query_tokens": 577, "message_tokens": 479, "message_utf8_bytes": 1073, "cache_tensor_bytes": null, "serialized_payload_bytes": 1073, "side_information_bytes": 0, "fetched_bytes": 5117, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 169780.56020801887, "hops": {"1": 22829.28358297795, "2": 21115.020749974065, "3": 22152.291209087707, "4": 20741.705666994676, "5": 20553.900958038867, "6": 20653.28154200688, "7": 21087.469290941954, "8": 20647.60720799677}, "receiver_prefill_ms": 540.7455419190228, "ttft_ms": 40.41154193691909, "decode_ms": 218.75779097899795}, "total_wall_ms": 170581.11983304843, "ttft_ms": 40.41154193691909, "input_usage": {"lookups": 28}, "output_usage": {"output_tokens": 7}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "6602245", "evidence_artifact": {"ledger": {"facts_total": 59, "facts_retained": 30, "retention_rate": 0.5084745762711864, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.5714285714285714, "amount": 0.6, "status": 0.5833333333333334, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1073}, "ledger_by_hop": {"1": {"facts_total": 59, "facts_retained": 30, "retention_rate": 0.5084745762711864, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.5714285714285714, "amount": 0.6, "status": 0.5833333333333334, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1213}, "2": {"facts_total": 59, "facts_retained": 30, "retention_rate": 0.5084745762711864, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.5714285714285714, "amount": 0.6, "status": 0.5833333333333334, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1105}, "3": {"facts_total": 59, "facts_retained": 30, "retention_rate": 0.5084745762711864, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.5714285714285714, "amount": 0.6, "status": 0.5833333333333334, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1085}, "4": {"facts_total": 59, "facts_retained": 30, "retention_rate": 0.5084745762711864, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.5714285714285714, "amount": 0.6, "status": 0.5833333333333334, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1073}, "5": {"facts_total": 59, "facts_retained": 30, "retention_rate": 0.5084745762711864, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.5714285714285714, "amount": 0.6, "status": 0.5833333333333334, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1073}, "6": {"facts_total": 59, "facts_retained": 30, "retention_rate": 0.5084745762711864, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.5714285714285714, "amount": 0.6, "status": 0.5833333333333334, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1073}, "7": {"facts_total": 59, "facts_retained": 30, "retention_rate": 0.5084745762711864, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.5714285714285714, "amount": 0.6, "status": 0.5833333333333334, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1073}, "8": {"facts_total": 59, "facts_retained": 30, "retention_rate": 0.5084745762711864, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.5714285714285714, "amount": 0.6, "status": 0.5833333333333334, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1073}}}, "reference_key_id": "keys/synthetic/syn_03012_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "5aeedc1e4643b829686179704346397792278190d32a28471247e6a066095c87", "output_hash": "d117fa8197988f8d1e4fabc6c4cdaab9516018e54553b57c63b875b3de82b16f"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "ddc07f4df15ec88a161b9ed5", "attempt_id": "367d7bd28b45", "run_group_id": "E4_core:A4|repairable|qwen3b:h8", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:38:59.984617Z", "task_family": "synthetic", "document_id": "syn_03012_2000", "document_hash": "03e441e68d54e1af036f8ff905dfd6cbbccead62636279c1a8e179352ab08945", "question_id": "q09_revision", "split": "test", "cluster_id": "syn_03012_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 8, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1927, "query_tokens": 570, "message_tokens": 479, "message_utf8_bytes": 1073, "cache_tensor_bytes": null, "serialized_payload_bytes": 1073, "side_information_bytes": 0, "fetched_bytes": 5117, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 169780.56020801887, "hops": {"1": 22829.28358297795, "2": 21115.020749974065, "3": 22152.291209087707, "4": 20741.705666994676, "5": 20553.900958038867, "6": 20653.28154200688, "7": 21087.469290941954, "8": 20647.60720799677}, "receiver_prefill_ms": 381.5904159564525, "ttft_ms": 25.16724995803088, "decode_ms": 272.33950002118945}, "total_wall_ms": 170460.30833304394, "ttft_ms": 25.16724995803088, "input_usage": {"lookups": 28}, "output_usage": {"output_tokens": 7}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "5899340", "evidence_artifact": {"ledger": {"facts_total": 59, "facts_retained": 30, "retention_rate": 0.5084745762711864, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.5714285714285714, "amount": 0.6, "status": 0.5833333333333334, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1073}, "ledger_by_hop": {"1": {"facts_total": 59, "facts_retained": 30, "retention_rate": 0.5084745762711864, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.5714285714285714, "amount": 0.6, "status": 0.5833333333333334, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1213}, "2": {"facts_total": 59, "facts_retained": 30, "retention_rate": 0.5084745762711864, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.5714285714285714, "amount": 0.6, "status": 0.5833333333333334, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1105}, "3": {"facts_total": 59, "facts_retained": 30, "retention_rate": 0.5084745762711864, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.5714285714285714, "amount": 0.6, "status": 0.5833333333333334, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1085}, "4": {"facts_total": 59, "facts_retained": 30, "retention_rate": 0.5084745762711864, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.5714285714285714, "amount": 0.6, "status": 0.5833333333333334, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1073}, "5": {"facts_total": 59, "facts_retained": 30, "retention_rate": 0.5084745762711864, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.5714285714285714, "amount": 0.6, "status": 0.5833333333333334, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1073}, "6": {"facts_total": 59, "facts_retained": 30, "retention_rate": 0.5084745762711864, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.5714285714285714, "amount": 0.6, "status": 0.5833333333333334, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1073}, "7": {"facts_total": 59, "facts_retained": 30, "retention_rate": 0.5084745762711864, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.5714285714285714, "amount": 0.6, "status": 0.5833333333333334, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1073}, "8": {"facts_total": 59, "facts_retained": 30, "retention_rate": 0.5084745762711864, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.5714285714285714, "amount": 0.6, "status": 0.5833333333333334, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1073}}}, "reference_key_id": "keys/synthetic/syn_03012_2000.json", "task_scores": {"correct": 0.0, "abstained": 0.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "5aeedc1e4643b829686179704346397792278190d32a28471247e6a066095c87", "output_hash": "c920323a9a02b1afb8619cf11bdde427f7cd1ea3289be69bd62932985895d717"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "a88cfd3c09174ac6797d9fc4", "attempt_id": "43b192dd9a9f", "run_group_id": "E4_core:A4|repairable|qwen3b:h8", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:39:00.681346Z", "task_family": "synthetic", "document_id": "syn_03012_2000", "document_hash": "03e441e68d54e1af036f8ff905dfd6cbbccead62636279c1a8e179352ab08945", "question_id": "q10_temporal", "split": "test", "cluster_id": "syn_03012_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 8, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1927, "query_tokens": 580, "message_tokens": 479, "message_utf8_bytes": 1073, "cache_tensor_bytes": null, "serialized_payload_bytes": 1073, "side_information_bytes": 0, "fetched_bytes": 5117, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 169780.56020801887, "hops": {"1": 22829.28358297795, "2": 21115.020749974065, "3": 22152.291209087707, "4": 20741.705666994676, "5": 20553.900958038867, "6": 20653.28154200688, "7": 21087.469290941954, "8": 20647.60720799677}, "receiver_prefill_ms": 545.9942501038313, "ttft_ms": 27.565083000808954, "decode_ms": 37.03883406706154}, "total_wall_ms": 170391.79545803927, "ttft_ms": 27.565083000808954, "input_usage": {"lookups": 28}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 59, "facts_retained": 30, "retention_rate": 0.5084745762711864, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.5714285714285714, "amount": 0.6, "status": 0.5833333333333334, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1073}, "ledger_by_hop": {"1": {"facts_total": 59, "facts_retained": 30, "retention_rate": 0.5084745762711864, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.5714285714285714, "amount": 0.6, "status": 0.5833333333333334, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1213}, "2": {"facts_total": 59, "facts_retained": 30, "retention_rate": 0.5084745762711864, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.5714285714285714, "amount": 0.6, "status": 0.5833333333333334, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1105}, "3": {"facts_total": 59, "facts_retained": 30, "retention_rate": 0.5084745762711864, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.5714285714285714, "amount": 0.6, "status": 0.5833333333333334, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1085}, "4": {"facts_total": 59, "facts_retained": 30, "retention_rate": 0.5084745762711864, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.5714285714285714, "amount": 0.6, "status": 0.5833333333333334, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1073}, "5": {"facts_total": 59, "facts_retained": 30, "retention_rate": 0.5084745762711864, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.5714285714285714, "amount": 0.6, "status": 0.5833333333333334, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1073}, "6": {"facts_total": 59, "facts_retained": 30, "retention_rate": 0.5084745762711864, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.5714285714285714, "amount": 0.6, "status": 0.5833333333333334, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1073}, "7": {"facts_total": 59, "facts_retained": 30, "retention_rate": 0.5084745762711864, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.5714285714285714, "amount": 0.6, "status": 0.5833333333333334, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1073}, "8": {"facts_total": 59, "facts_retained": 30, "retention_rate": 0.5084745762711864, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.5714285714285714, "amount": 0.6, "status": 0.5833333333333334, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1073}}}, "reference_key_id": "keys/synthetic/syn_03012_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "5aeedc1e4643b829686179704346397792278190d32a28471247e6a066095c87", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "5da1c4ff1af362a62cd197bb", "attempt_id": "4891961e24ab", "run_group_id": "E4_core:A4|repairable|qwen3b:h8", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:39:01.311631Z", "task_family": "synthetic", "document_id": "syn_03012_2000", "document_hash": "03e441e68d54e1af036f8ff905dfd6cbbccead62636279c1a8e179352ab08945", "question_id": "q11_aggregation", "split": "test", "cluster_id": "syn_03012_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 8, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1927, "query_tokens": 572, "message_tokens": 479, "message_utf8_bytes": 1073, "cache_tensor_bytes": null, "serialized_payload_bytes": 1073, "side_information_bytes": 0, "fetched_bytes": 5117, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 169780.56020801887, "hops": {"1": 22829.28358297795, "2": 21115.020749974065, "3": 22152.291209087707, "4": 20741.705666994676, "5": 20553.900958038867, "6": 20653.28154200688, "7": 21087.469290941954, "8": 20647.60720799677}, "receiver_prefill_ms": 388.6029999703169, "ttft_ms": 35.81933293025941, "decode_ms": 261.7946669925004}, "total_wall_ms": 170467.49495796394, "ttft_ms": 35.81933293025941, "input_usage": {"lookups": 28}, "output_usage": {"output_tokens": 8}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "15014487", "evidence_artifact": {"ledger": {"facts_total": 59, "facts_retained": 30, "retention_rate": 0.5084745762711864, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.5714285714285714, "amount": 0.6, "status": 0.5833333333333334, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1073}, "ledger_by_hop": {"1": {"facts_total": 59, "facts_retained": 30, "retention_rate": 0.5084745762711864, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.5714285714285714, "amount": 0.6, "status": 0.5833333333333334, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1213}, "2": {"facts_total": 59, "facts_retained": 30, "retention_rate": 0.5084745762711864, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.5714285714285714, "amount": 0.6, "status": 0.5833333333333334, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1105}, "3": {"facts_total": 59, "facts_retained": 30, "retention_rate": 0.5084745762711864, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.5714285714285714, "amount": 0.6, "status": 0.5833333333333334, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1085}, "4": {"facts_total": 59, "facts_retained": 30, "retention_rate": 0.5084745762711864, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.5714285714285714, "amount": 0.6, "status": 0.5833333333333334, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1073}, "5": {"facts_total": 59, "facts_retained": 30, "retention_rate": 0.5084745762711864, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.5714285714285714, "amount": 0.6, "status": 0.5833333333333334, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1073}, "6": {"facts_total": 59, "facts_retained": 30, "retention_rate": 0.5084745762711864, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.5714285714285714, "amount": 0.6, "status": 0.5833333333333334, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1073}, "7": {"facts_total": 59, "facts_retained": 30, "retention_rate": 0.5084745762711864, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.5714285714285714, "amount": 0.6, "status": 0.5833333333333334, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1073}, "8": {"facts_total": 59, "facts_retained": 30, "retention_rate": 0.5084745762711864, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.5714285714285714, "amount": 0.6, "status": 0.5833333333333334, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1073}}}, "reference_key_id": "keys/synthetic/syn_03012_2000.json", "task_scores": {"correct": 0.0, "abstained": 0.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "5aeedc1e4643b829686179704346397792278190d32a28471247e6a066095c87", "output_hash": "7f803ac89469e407c90b9cf8bd6eddfc531f68179677b8d6f860c3ddb1aed324"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "d195cd3df12fba844e0ba1cf", "attempt_id": "5582640674b0", "run_group_id": "E4_core:A4|repairable|qwen3b:h8", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:39:02.015950Z", "task_family": "synthetic", "document_id": "syn_03012_2000", "document_hash": "03e441e68d54e1af036f8ff905dfd6cbbccead62636279c1a8e179352ab08945", "question_id": "q12_unit_conversion", "split": "test", "cluster_id": "syn_03012_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 8, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1927, "query_tokens": 569, "message_tokens": 479, "message_utf8_bytes": 1073, "cache_tensor_bytes": null, "serialized_payload_bytes": 1073, "side_information_bytes": 0, "fetched_bytes": 5117, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 169780.56020801887, "hops": {"1": 22829.28358297795, "2": 21115.020749974065, "3": 22152.291209087707, "4": 20741.705666994676, "5": 20553.900958038867, "6": 20653.28154200688, "7": 21087.469290941954, "8": 20647.60720799677}, "receiver_prefill_ms": 329.45212500635535, "ttft_ms": 53.37654193863273, "decode_ms": 393.96504091564566}, "total_wall_ms": 170558.30020806752, "ttft_ms": 53.37654193863273, "input_usage": {"lookups": 28}, "output_usage": {"output_tokens": 5}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "44620", "evidence_artifact": {"ledger": {"facts_total": 59, "facts_retained": 30, "retention_rate": 0.5084745762711864, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.5714285714285714, "amount": 0.6, "status": 0.5833333333333334, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1073}, "ledger_by_hop": {"1": {"facts_total": 59, "facts_retained": 30, "retention_rate": 0.5084745762711864, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.5714285714285714, "amount": 0.6, "status": 0.5833333333333334, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1213}, "2": {"facts_total": 59, "facts_retained": 30, "retention_rate": 0.5084745762711864, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.5714285714285714, "amount": 0.6, "status": 0.5833333333333334, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1105}, "3": {"facts_total": 59, "facts_retained": 30, "retention_rate": 0.5084745762711864, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.5714285714285714, "amount": 0.6, "status": 0.5833333333333334, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1085}, "4": {"facts_total": 59, "facts_retained": 30, "retention_rate": 0.5084745762711864, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.5714285714285714, "amount": 0.6, "status": 0.5833333333333334, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1073}, "5": {"facts_total": 59, "facts_retained": 30, "retention_rate": 0.5084745762711864, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.5714285714285714, "amount": 0.6, "status": 0.5833333333333334, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1073}, "6": {"facts_total": 59, "facts_retained": 30, "retention_rate": 0.5084745762711864, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.5714285714285714, "amount": 0.6, "status": 0.5833333333333334, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1073}, "7": {"facts_total": 59, "facts_retained": 30, "retention_rate": 0.5084745762711864, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.5714285714285714, "amount": 0.6, "status": 0.5833333333333334, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1073}, "8": {"facts_total": 59, "facts_retained": 30, "retention_rate": 0.5084745762711864, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.5714285714285714, "amount": 0.6, "status": 0.5833333333333334, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1073}}}, "reference_key_id": "keys/synthetic/syn_03012_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "5aeedc1e4643b829686179704346397792278190d32a28471247e6a066095c87", "output_hash": "2bd125847727be7e7c0a656ee4c81f3bc6d0fef11f536f50ce7bc83688907075"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "56e32c1aac508e1035ce9672", "attempt_id": "a6fcd6851be6", "run_group_id": "E4_core:A4|repairable|qwen3b:h8", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:39:02.812439Z", "task_family": "synthetic", "document_id": "syn_03012_2000", "document_hash": "03e441e68d54e1af036f8ff905dfd6cbbccead62636279c1a8e179352ab08945", "question_id": "q13_unanswerable", "split": "test", "cluster_id": "syn_03012_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 8, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1927, "query_tokens": 574, "message_tokens": 479, "message_utf8_bytes": 1073, "cache_tensor_bytes": null, "serialized_payload_bytes": 1073, "side_information_bytes": 0, "fetched_bytes": 5117, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 169780.56020801887, "hops": {"1": 22829.28358297795, "2": 21115.020749974065, "3": 22152.291209087707, "4": 20741.705666994676, "5": 20553.900958038867, "6": 20653.28154200688, "7": 21087.469290941954, "8": 20647.60720799677}, "receiver_prefill_ms": 398.35037500597537, "ttft_ms": 42.85629198420793, "decode_ms": 29.419958009384573}, "total_wall_ms": 170251.90712406766, "ttft_ms": 42.85629198420793, "input_usage": {"lookups": 28}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 59, "facts_retained": 30, "retention_rate": 0.5084745762711864, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.5714285714285714, "amount": 0.6, "status": 0.5833333333333334, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1073}, "ledger_by_hop": {"1": {"facts_total": 59, "facts_retained": 30, "retention_rate": 0.5084745762711864, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.5714285714285714, "amount": 0.6, "status": 0.5833333333333334, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1213}, "2": {"facts_total": 59, "facts_retained": 30, "retention_rate": 0.5084745762711864, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.5714285714285714, "amount": 0.6, "status": 0.5833333333333334, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1105}, "3": {"facts_total": 59, "facts_retained": 30, "retention_rate": 0.5084745762711864, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.5714285714285714, "amount": 0.6, "status": 0.5833333333333334, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1085}, "4": {"facts_total": 59, "facts_retained": 30, "retention_rate": 0.5084745762711864, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.5714285714285714, "amount": 0.6, "status": 0.5833333333333334, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1073}, "5": {"facts_total": 59, "facts_retained": 30, "retention_rate": 0.5084745762711864, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.5714285714285714, "amount": 0.6, "status": 0.5833333333333334, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1073}, "6": {"facts_total": 59, "facts_retained": 30, "retention_rate": 0.5084745762711864, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.5714285714285714, "amount": 0.6, "status": 0.5833333333333334, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1073}, "7": {"facts_total": 59, "facts_retained": 30, "retention_rate": 0.5084745762711864, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.5714285714285714, "amount": 0.6, "status": 0.5833333333333334, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1073}, "8": {"facts_total": 59, "facts_retained": 30, "retention_rate": 0.5084745762711864, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.5714285714285714, "amount": 0.6, "status": 0.5833333333333334, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1073}}}, "reference_key_id": "keys/synthetic/syn_03012_2000.json", "task_scores": {"correct": 1.0, "abstained": 1.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "5aeedc1e4643b829686179704346397792278190d32a28471247e6a066095c87", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "7df7c86200badb2e140be798", "attempt_id": "6e0d33f815e5", "run_group_id": "E4_core:A4|repairable|qwen3b:h8", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:39:03.301096Z", "task_family": "synthetic", "document_id": "syn_03012_2000", "document_hash": "03e441e68d54e1af036f8ff905dfd6cbbccead62636279c1a8e179352ab08945", "question_id": "q14_unanswerable", "split": "test", "cluster_id": "syn_03012_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 8, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1927, "query_tokens": 574, "message_tokens": 479, "message_utf8_bytes": 1073, "cache_tensor_bytes": null, "serialized_payload_bytes": 1073, "side_information_bytes": 0, "fetched_bytes": 5117, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 169780.56020801887, "hops": {"1": 22829.28358297795, "2": 21115.020749974065, "3": 22152.291209087707, "4": 20741.705666994676, "5": 20553.900958038867, "6": 20653.28154200688, "7": 21087.469290941954, "8": 20647.60720799677}, "receiver_prefill_ms": 377.69254099112004, "ttft_ms": 19.35924997087568, "decode_ms": 19.528959062881768}, "total_wall_ms": 170197.80262501445, "ttft_ms": 19.35924997087568, "input_usage": {"lookups": 28}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 59, "facts_retained": 30, "retention_rate": 0.5084745762711864, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.5714285714285714, "amount": 0.6, "status": 0.5833333333333334, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1073}, "ledger_by_hop": {"1": {"facts_total": 59, "facts_retained": 30, "retention_rate": 0.5084745762711864, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.5714285714285714, "amount": 0.6, "status": 0.5833333333333334, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1213}, "2": {"facts_total": 59, "facts_retained": 30, "retention_rate": 0.5084745762711864, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.5714285714285714, "amount": 0.6, "status": 0.5833333333333334, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1105}, "3": {"facts_total": 59, "facts_retained": 30, "retention_rate": 0.5084745762711864, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.5714285714285714, "amount": 0.6, "status": 0.5833333333333334, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1085}, "4": {"facts_total": 59, "facts_retained": 30, "retention_rate": 0.5084745762711864, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.5714285714285714, "amount": 0.6, "status": 0.5833333333333334, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1073}, "5": {"facts_total": 59, "facts_retained": 30, "retention_rate": 0.5084745762711864, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.5714285714285714, "amount": 0.6, "status": 0.5833333333333334, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1073}, "6": {"facts_total": 59, "facts_retained": 30, "retention_rate": 0.5084745762711864, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.5714285714285714, "amount": 0.6, "status": 0.5833333333333334, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1073}, "7": {"facts_total": 59, "facts_retained": 30, "retention_rate": 0.5084745762711864, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.5714285714285714, "amount": 0.6, "status": 0.5833333333333334, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1073}, "8": {"facts_total": 59, "facts_retained": 30, "retention_rate": 0.5084745762711864, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.5714285714285714, "amount": 0.6, "status": 0.5833333333333334, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1073}}}, "reference_key_id": "keys/synthetic/syn_03012_2000.json", "task_scores": {"correct": 1.0, "abstained": 1.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "5aeedc1e4643b829686179704346397792278190d32a28471247e6a066095c87", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "bf32c8a287183cb5ca4d9502", "attempt_id": "f80412da86a2", "run_group_id": "E4_core:A5|repairable|qwen3b:h1", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:39:26.800233Z", "task_family": "synthetic", "document_id": "syn_03012_2000", "document_hash": "03e441e68d54e1af036f8ff905dfd6cbbccead62636279c1a8e179352ab08945", "question_id": "q00_single_fact", "split": "test", "cluster_id": "syn_03012_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 1, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1927, "query_tokens": 431, "message_tokens": 348, "message_utf8_bytes": 824, "cache_tensor_bytes": null, "serialized_payload_bytes": 824, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 23063.031291007064, "hops": {"1": 23063.031291007064}, "receiver_prefill_ms": 291.96329205296934, "ttft_ms": 40.64575000666082, "decode_ms": 373.20441694464535}, "total_wall_ms": 23769.418665906414, "ttft_ms": 40.64575000666082, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 10}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "XX-71M-6213", "evidence_artifact": {"ledger": {"facts_total": 59, "facts_retained": 11, "retention_rate": 0.1864406779661017, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.14285714285714285, "amount": 0.0, "status": 0.4166666666666667, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 760}, "ledger_by_hop": {"1": {"facts_total": 59, "facts_retained": 11, "retention_rate": 0.1864406779661017, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.14285714285714285, "amount": 0.0, "status": 0.4166666666666667, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 760}}}, "reference_key_id": "keys/synthetic/syn_03012_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "bb16af6fe25c6b4abc9bdb0ad8c8b3d8b7d3dc12769f5feefdb7a2e4bb915494", "output_hash": "66ef6c5da682b38795e9029723bbd71cdde2c4dec70cf1fae6efe62ef243f10a"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "792a3e10dfb5534ab2da8c42", "attempt_id": "6430bae66bc0", "run_group_id": "E4_core:A5|repairable|qwen3b:h1", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:39:27.524352Z", "task_family": "synthetic", "document_id": "syn_03012_2000", "document_hash": "03e441e68d54e1af036f8ff905dfd6cbbccead62636279c1a8e179352ab08945", "question_id": "q01_single_fact", "split": "test", "cluster_id": "syn_03012_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 1, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1927, "query_tokens": 437, "message_tokens": 348, "message_utf8_bytes": 824, "cache_tensor_bytes": null, "serialized_payload_bytes": 824, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 23063.031291007064, "hops": {"1": 23063.031291007064}, "receiver_prefill_ms": 458.89750006608665, "ttft_ms": 42.67933405935764, "decode_ms": 44.485999969765544}, "total_wall_ms": 23609.7279159585, "ttft_ms": 42.67933405935764, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 59, "facts_retained": 11, "retention_rate": 0.1864406779661017, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.14285714285714285, "amount": 0.0, "status": 0.4166666666666667, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 760}, "ledger_by_hop": {"1": {"facts_total": 59, "facts_retained": 11, "retention_rate": 0.1864406779661017, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.14285714285714285, "amount": 0.0, "status": 0.4166666666666667, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 760}}}, "reference_key_id": "keys/synthetic/syn_03012_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "bb16af6fe25c6b4abc9bdb0ad8c8b3d8b7d3dc12769f5feefdb7a2e4bb915494", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "d55b5ac63a78ccdf07e68ff3", "attempt_id": "a878ff4c8b07", "run_group_id": "E4_core:A5|repairable|qwen3b:h1", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:39:28.090422Z", "task_family": "synthetic", "document_id": "syn_03012_2000", "document_hash": "03e441e68d54e1af036f8ff905dfd6cbbccead62636279c1a8e179352ab08945", "question_id": "q02_exact_copy", "split": "test", "cluster_id": "syn_03012_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 1, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1927, "query_tokens": 436, "message_tokens": 348, "message_utf8_bytes": 824, "cache_tensor_bytes": null, "serialized_payload_bytes": 824, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 23063.031291007064, "hops": {"1": 23063.031291007064}, "receiver_prefill_ms": 279.5525420224294, "ttft_ms": 38.98358391597867, "decode_ms": 32.146708923392}, "total_wall_ms": 23414.31887401268, "ttft_ms": 38.98358391597867, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 59, "facts_retained": 11, "retention_rate": 0.1864406779661017, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.14285714285714285, "amount": 0.0, "status": 0.4166666666666667, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 760}, "ledger_by_hop": {"1": {"facts_total": 59, "facts_retained": 11, "retention_rate": 0.1864406779661017, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.14285714285714285, "amount": 0.0, "status": 0.4166666666666667, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 760}}}, "reference_key_id": "keys/synthetic/syn_03012_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "bb16af6fe25c6b4abc9bdb0ad8c8b3d8b7d3dc12769f5feefdb7a2e4bb915494", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "4ea53779d791c17d06c99c39", "attempt_id": "f72f39054465", "run_group_id": "E4_core:A5|repairable|qwen3b:h1", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:39:28.459445Z", "task_family": "synthetic", "document_id": "syn_03012_2000", "document_hash": "03e441e68d54e1af036f8ff905dfd6cbbccead62636279c1a8e179352ab08945", "question_id": "q03_exact_copy", "split": "test", "cluster_id": "syn_03012_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 1, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1927, "query_tokens": 432, "message_tokens": 348, "message_utf8_bytes": 824, "cache_tensor_bytes": null, "serialized_payload_bytes": 824, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 23063.031291007064, "hops": {"1": 23063.031291007064}, "receiver_prefill_ms": 284.41750002093613, "ttft_ms": 31.323750037699938, "decode_ms": 35.37291695829481}, "total_wall_ms": 23414.654708001763, "ttft_ms": 31.323750037699938, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 59, "facts_retained": 11, "retention_rate": 0.1864406779661017, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.14285714285714285, "amount": 0.0, "status": 0.4166666666666667, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 760}, "ledger_by_hop": {"1": {"facts_total": 59, "facts_retained": 11, "retention_rate": 0.1864406779661017, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.14285714285714285, "amount": 0.0, "status": 0.4166666666666667, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 760}}}, "reference_key_id": "keys/synthetic/syn_03012_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "bb16af6fe25c6b4abc9bdb0ad8c8b3d8b7d3dc12769f5feefdb7a2e4bb915494", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "41cb34b4a19bcb643770d89f", "attempt_id": "db3cfe3d53a3", "run_group_id": "E4_core:A5|repairable|qwen3b:h1", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:39:28.827267Z", "task_family": "synthetic", "document_id": "syn_03012_2000", "document_hash": "03e441e68d54e1af036f8ff905dfd6cbbccead62636279c1a8e179352ab08945", "question_id": "q04_multi_hop", "split": "test", "cluster_id": "syn_03012_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 1, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1927, "query_tokens": 437, "message_tokens": 348, "message_utf8_bytes": 824, "cache_tensor_bytes": null, "serialized_payload_bytes": 824, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 23063.031291007064, "hops": {"1": 23063.031291007064}, "receiver_prefill_ms": 292.35337499994785, "ttft_ms": 46.74141702707857, "decode_ms": 56.94208398927003}, "total_wall_ms": 23459.643165930174, "ttft_ms": 46.74141702707857, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 59, "facts_retained": 11, "retention_rate": 0.1864406779661017, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.14285714285714285, "amount": 0.0, "status": 0.4166666666666667, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 760}, "ledger_by_hop": {"1": {"facts_total": 59, "facts_retained": 11, "retention_rate": 0.1864406779661017, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.14285714285714285, "amount": 0.0, "status": 0.4166666666666667, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 760}}}, "reference_key_id": "keys/synthetic/syn_03012_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "bb16af6fe25c6b4abc9bdb0ad8c8b3d8b7d3dc12769f5feefdb7a2e4bb915494", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "85deac8e27caaf8d1d1a66ef", "attempt_id": "e2d43d748d9a", "run_group_id": "E4_core:A5|repairable|qwen3b:h1", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:39:29.241272Z", "task_family": "synthetic", "document_id": "syn_03012_2000", "document_hash": "03e441e68d54e1af036f8ff905dfd6cbbccead62636279c1a8e179352ab08945", "question_id": "q05_multi_hop", "split": "test", "cluster_id": "syn_03012_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 1, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1927, "query_tokens": 440, "message_tokens": 348, "message_utf8_bytes": 824, "cache_tensor_bytes": null, "serialized_payload_bytes": 824, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 23063.031291007064, "hops": {"1": 23063.031291007064}, "receiver_prefill_ms": 442.6930829649791, "ttft_ms": 33.99920801166445, "decode_ms": 37.23079094197601}, "total_wall_ms": 23577.43899896741, "ttft_ms": 33.99920801166445, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 59, "facts_retained": 11, "retention_rate": 0.1864406779661017, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.14285714285714285, "amount": 0.0, "status": 0.4166666666666667, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 760}, "ledger_by_hop": {"1": {"facts_total": 59, "facts_retained": 11, "retention_rate": 0.1864406779661017, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.14285714285714285, "amount": 0.0, "status": 0.4166666666666667, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 760}}}, "reference_key_id": "keys/synthetic/syn_03012_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "bb16af6fe25c6b4abc9bdb0ad8c8b3d8b7d3dc12769f5feefdb7a2e4bb915494", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "3befe2463f027a30e7932ba5", "attempt_id": "b0ee5efbade2", "run_group_id": "E4_core:A5|repairable|qwen3b:h1", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:39:29.773227Z", "task_family": "synthetic", "document_id": "syn_03012_2000", "document_hash": "03e441e68d54e1af036f8ff905dfd6cbbccead62636279c1a8e179352ab08945", "question_id": "q06_negation", "split": "test", "cluster_id": "syn_03012_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 1, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1927, "query_tokens": 432, "message_tokens": 348, "message_utf8_bytes": 824, "cache_tensor_bytes": null, "serialized_payload_bytes": 824, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 23063.031291007064, "hops": {"1": 23063.031291007064}, "receiver_prefill_ms": 294.34845806099474, "ttft_ms": 34.472458995878696, "decode_ms": 38.208625046536326}, "total_wall_ms": 23430.504375020973, "ttft_ms": 34.472458995878696, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 59, "facts_retained": 11, "retention_rate": 0.1864406779661017, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.14285714285714285, "amount": 0.0, "status": 0.4166666666666667, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 760}, "ledger_by_hop": {"1": {"facts_total": 59, "facts_retained": 11, "retention_rate": 0.1864406779661017, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.14285714285714285, "amount": 0.0, "status": 0.4166666666666667, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 760}}}, "reference_key_id": "keys/synthetic/syn_03012_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "bb16af6fe25c6b4abc9bdb0ad8c8b3d8b7d3dc12769f5feefdb7a2e4bb915494", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "cbfd17f72cf586b6f90efffc", "attempt_id": "58166967eac8", "run_group_id": "E4_core:A5|repairable|qwen3b:h1", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:39:30.157494Z", "task_family": "synthetic", "document_id": "syn_03012_2000", "document_hash": "03e441e68d54e1af036f8ff905dfd6cbbccead62636279c1a8e179352ab08945", "question_id": "q07_negation", "split": "test", "cluster_id": "syn_03012_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 1, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1927, "query_tokens": 432, "message_tokens": 348, "message_utf8_bytes": 824, "cache_tensor_bytes": null, "serialized_payload_bytes": 824, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 23063.031291007064, "hops": {"1": 23063.031291007064}, "receiver_prefill_ms": 294.0881249960512, "ttft_ms": 33.98958302568644, "decode_ms": 36.78508405573666}, "total_wall_ms": 23428.307957947254, "ttft_ms": 33.98958302568644, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 59, "facts_retained": 11, "retention_rate": 0.1864406779661017, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.14285714285714285, "amount": 0.0, "status": 0.4166666666666667, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 760}, "ledger_by_hop": {"1": {"facts_total": 59, "facts_retained": 11, "retention_rate": 0.1864406779661017, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.14285714285714285, "amount": 0.0, "status": 0.4166666666666667, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 760}}}, "reference_key_id": "keys/synthetic/syn_03012_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "bb16af6fe25c6b4abc9bdb0ad8c8b3d8b7d3dc12769f5feefdb7a2e4bb915494", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "1cf5d44961f3e5da77262102", "attempt_id": "6e2c1528c641", "run_group_id": "E4_core:A5|repairable|qwen3b:h1", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:39:30.539139Z", "task_family": "synthetic", "document_id": "syn_03012_2000", "document_hash": "03e441e68d54e1af036f8ff905dfd6cbbccead62636279c1a8e179352ab08945", "question_id": "q08_revision", "split": "test", "cluster_id": "syn_03012_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 1, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1927, "query_tokens": 447, "message_tokens": 348, "message_utf8_bytes": 824, "cache_tensor_bytes": null, "serialized_payload_bytes": 824, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 23063.031291007064, "hops": {"1": 23063.031291007064}, "receiver_prefill_ms": 320.5646249698475, "ttft_ms": 57.76016600430012, "decode_ms": 50.52974994760007}, "total_wall_ms": 23492.58954101242, "ttft_ms": 57.76016600430012, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 59, "facts_retained": 11, "retention_rate": 0.1864406779661017, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.14285714285714285, "amount": 0.0, "status": 0.4166666666666667, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 760}, "ledger_by_hop": {"1": {"facts_total": 59, "facts_retained": 11, "retention_rate": 0.1864406779661017, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.14285714285714285, "amount": 0.0, "status": 0.4166666666666667, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 760}}}, "reference_key_id": "keys/synthetic/syn_03012_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "bb16af6fe25c6b4abc9bdb0ad8c8b3d8b7d3dc12769f5feefdb7a2e4bb915494", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "6aac9ee5023be27517dbe974", "attempt_id": "8f5dc5ea1a7d", "run_group_id": "E4_core:A5|repairable|qwen3b:h1", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:39:30.985750Z", "task_family": "synthetic", "document_id": "syn_03012_2000", "document_hash": "03e441e68d54e1af036f8ff905dfd6cbbccead62636279c1a8e179352ab08945", "question_id": "q09_revision", "split": "test", "cluster_id": "syn_03012_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 1, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1927, "query_tokens": 440, "message_tokens": 348, "message_utf8_bytes": 824, "cache_tensor_bytes": null, "serialized_payload_bytes": 824, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 23063.031291007064, "hops": {"1": 23063.031291007064}, "receiver_prefill_ms": 427.23845900036395, "ttft_ms": 33.53129210881889, "decode_ms": 34.824625006876886}, "total_wall_ms": 23559.123583021574, "ttft_ms": 33.53129210881889, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 59, "facts_retained": 11, "retention_rate": 0.1864406779661017, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.14285714285714285, "amount": 0.0, "status": 0.4166666666666667, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 760}, "ledger_by_hop": {"1": {"facts_total": 59, "facts_retained": 11, "retention_rate": 0.1864406779661017, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.14285714285714285, "amount": 0.0, "status": 0.4166666666666667, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 760}}}, "reference_key_id": "keys/synthetic/syn_03012_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "bb16af6fe25c6b4abc9bdb0ad8c8b3d8b7d3dc12769f5feefdb7a2e4bb915494", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "93ccfb79c19a22aa0d884e47", "attempt_id": "e4ba8a087960", "run_group_id": "E4_core:A5|repairable|qwen3b:h1", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:39:31.499232Z", "task_family": "synthetic", "document_id": "syn_03012_2000", "document_hash": "03e441e68d54e1af036f8ff905dfd6cbbccead62636279c1a8e179352ab08945", "question_id": "q10_temporal", "split": "test", "cluster_id": "syn_03012_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 1, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1927, "query_tokens": 450, "message_tokens": 348, "message_utf8_bytes": 824, "cache_tensor_bytes": null, "serialized_payload_bytes": 824, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 23063.031291007064, "hops": {"1": 23063.031291007064}, "receiver_prefill_ms": 326.2076659593731, "ttft_ms": 39.554167073220015, "decode_ms": 28.64049992058426}, "total_wall_ms": 23457.900665933266, "ttft_ms": 39.554167073220015, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 59, "facts_retained": 11, "retention_rate": 0.1864406779661017, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.14285714285714285, "amount": 0.0, "status": 0.4166666666666667, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 760}, "ledger_by_hop": {"1": {"facts_total": 59, "facts_retained": 11, "retention_rate": 0.1864406779661017, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.14285714285714285, "amount": 0.0, "status": 0.4166666666666667, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 760}}}, "reference_key_id": "keys/synthetic/syn_03012_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "bb16af6fe25c6b4abc9bdb0ad8c8b3d8b7d3dc12769f5feefdb7a2e4bb915494", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "17a595ae9e3abe348f88810f", "attempt_id": "59ef46f40294", "run_group_id": "E4_core:A5|repairable|qwen3b:h1", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:39:31.911529Z", "task_family": "synthetic", "document_id": "syn_03012_2000", "document_hash": "03e441e68d54e1af036f8ff905dfd6cbbccead62636279c1a8e179352ab08945", "question_id": "q11_aggregation", "split": "test", "cluster_id": "syn_03012_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 1, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1927, "query_tokens": 442, "message_tokens": 348, "message_utf8_bytes": 824, "cache_tensor_bytes": null, "serialized_payload_bytes": 824, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 23063.031291007064, "hops": {"1": 23063.031291007064}, "receiver_prefill_ms": 284.7854159772396, "ttft_ms": 22.115124971605837, "decode_ms": 20.15620900783688}, "total_wall_ms": 23390.56824904401, "ttft_ms": 22.115124971605837, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 59, "facts_retained": 11, "retention_rate": 0.1864406779661017, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.14285714285714285, "amount": 0.0, "status": 0.4166666666666667, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 760}, "ledger_by_hop": {"1": {"facts_total": 59, "facts_retained": 11, "retention_rate": 0.1864406779661017, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.14285714285714285, "amount": 0.0, "status": 0.4166666666666667, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 760}}}, "reference_key_id": "keys/synthetic/syn_03012_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "bb16af6fe25c6b4abc9bdb0ad8c8b3d8b7d3dc12769f5feefdb7a2e4bb915494", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "c35386d3f174da4d5b2b3d38", "attempt_id": "bf765a12d6f6", "run_group_id": "E4_core:A5|repairable|qwen3b:h1", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:39:32.255636Z", "task_family": "synthetic", "document_id": "syn_03012_2000", "document_hash": "03e441e68d54e1af036f8ff905dfd6cbbccead62636279c1a8e179352ab08945", "question_id": "q12_unit_conversion", "split": "test", "cluster_id": "syn_03012_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 1, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1927, "query_tokens": 439, "message_tokens": 348, "message_utf8_bytes": 824, "cache_tensor_bytes": null, "serialized_payload_bytes": 824, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 23063.031291007064, "hops": {"1": 23063.031291007064}, "receiver_prefill_ms": 319.8159580351785, "ttft_ms": 42.46054100804031, "decode_ms": 344.0049591008574}, "total_wall_ms": 23769.850206910633, "ttft_ms": 42.46054100804031, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 5}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "44620", "evidence_artifact": {"ledger": {"facts_total": 59, "facts_retained": 11, "retention_rate": 0.1864406779661017, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.14285714285714285, "amount": 0.0, "status": 0.4166666666666667, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 760}, "ledger_by_hop": {"1": {"facts_total": 59, "facts_retained": 11, "retention_rate": 0.1864406779661017, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.14285714285714285, "amount": 0.0, "status": 0.4166666666666667, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 760}}}, "reference_key_id": "keys/synthetic/syn_03012_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "bb16af6fe25c6b4abc9bdb0ad8c8b3d8b7d3dc12769f5feefdb7a2e4bb915494", "output_hash": "2bd125847727be7e7c0a656ee4c81f3bc6d0fef11f536f50ce7bc83688907075"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "594d5afd4e77735c0e057b8c", "attempt_id": "ca390ecb0c44", "run_group_id": "E4_core:A5|repairable|qwen3b:h1", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:39:32.981829Z", "task_family": "synthetic", "document_id": "syn_03012_2000", "document_hash": "03e441e68d54e1af036f8ff905dfd6cbbccead62636279c1a8e179352ab08945", "question_id": "q13_unanswerable", "split": "test", "cluster_id": "syn_03012_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 1, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1927, "query_tokens": 444, "message_tokens": 348, "message_utf8_bytes": 824, "cache_tensor_bytes": null, "serialized_payload_bytes": 824, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 23063.031291007064, "hops": {"1": 23063.031291007064}, "receiver_prefill_ms": 293.5911249369383, "ttft_ms": 35.103583009913564, "decode_ms": 33.644374925643206}, "total_wall_ms": 23425.851457985118, "ttft_ms": 35.103583009913564, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 59, "facts_retained": 11, "retention_rate": 0.1864406779661017, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.14285714285714285, "amount": 0.0, "status": 0.4166666666666667, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 760}, "ledger_by_hop": {"1": {"facts_total": 59, "facts_retained": 11, "retention_rate": 0.1864406779661017, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.14285714285714285, "amount": 0.0, "status": 0.4166666666666667, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 760}}}, "reference_key_id": "keys/synthetic/syn_03012_2000.json", "task_scores": {"correct": 1.0, "abstained": 1.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "bb16af6fe25c6b4abc9bdb0ad8c8b3d8b7d3dc12769f5feefdb7a2e4bb915494", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "9c7ea77ce263af325ccc5b04", "attempt_id": "32568c8f3a9e", "run_group_id": "E4_core:A5|repairable|qwen3b:h1", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:39:33.361804Z", "task_family": "synthetic", "document_id": "syn_03012_2000", "document_hash": "03e441e68d54e1af036f8ff905dfd6cbbccead62636279c1a8e179352ab08945", "question_id": "q14_unanswerable", "split": "test", "cluster_id": "syn_03012_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 1, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1927, "query_tokens": 444, "message_tokens": 348, "message_utf8_bytes": 824, "cache_tensor_bytes": null, "serialized_payload_bytes": 824, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 23063.031291007064, "hops": {"1": 23063.031291007064}, "receiver_prefill_ms": 295.70870799943805, "ttft_ms": 34.32229103054851, "decode_ms": 29.53795890789479}, "total_wall_ms": 23423.020041082054, "ttft_ms": 34.32229103054851, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 59, "facts_retained": 11, "retention_rate": 0.1864406779661017, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.14285714285714285, "amount": 0.0, "status": 0.4166666666666667, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 760}, "ledger_by_hop": {"1": {"facts_total": 59, "facts_retained": 11, "retention_rate": 0.1864406779661017, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.14285714285714285, "amount": 0.0, "status": 0.4166666666666667, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 760}}}, "reference_key_id": "keys/synthetic/syn_03012_2000.json", "task_scores": {"correct": 1.0, "abstained": 1.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "bb16af6fe25c6b4abc9bdb0ad8c8b3d8b7d3dc12769f5feefdb7a2e4bb915494", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "6b0d3218a6014cae5cdb81e9", "attempt_id": "3a6746ebff53", "run_group_id": "E4_core:A5|repairable|qwen3b:h2", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:39:49.156149Z", "task_family": "synthetic", "document_id": "syn_03012_2000", "document_hash": "03e441e68d54e1af036f8ff905dfd6cbbccead62636279c1a8e179352ab08945", "question_id": "q00_single_fact", "split": "test", "cluster_id": "syn_03012_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 2, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1927, "query_tokens": 440, "message_tokens": 358, "message_utf8_bytes": 853, "cache_tensor_bytes": null, "serialized_payload_bytes": 853, "side_information_bytes": 0, "fetched_bytes": 461, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 38478.24233304709, "hops": {"1": 23063.031291007064, "2": 15415.211042040028}, "receiver_prefill_ms": 467.72695798426867, "ttft_ms": 34.294792101718485, "decode_ms": 369.01824991218746}, "total_wall_ms": 39349.939250038005, "ttft_ms": 34.294792101718485, "input_usage": {"lookups": 1}, "output_usage": {"output_tokens": 10}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "XX-71M-6213", "evidence_artifact": {"ledger": {"facts_total": 59, "facts_retained": 13, "retention_rate": 0.22033898305084745, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.14285714285714285, "amount": 0.0, "status": 0.5833333333333334, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 789}, "ledger_by_hop": {"1": {"facts_total": 59, "facts_retained": 11, "retention_rate": 0.1864406779661017, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.14285714285714285, "amount": 0.0, "status": 0.4166666666666667, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 760}, "2": {"facts_total": 59, "facts_retained": 13, "retention_rate": 0.22033898305084745, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.14285714285714285, "amount": 0.0, "status": 0.5833333333333334, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 789}}}, "reference_key_id": "keys/synthetic/syn_03012_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "b7a5ccb481b692e93f148c0da2974299bed9aa510b975eeaf745e05e1a120812", "output_hash": "66ef6c5da682b38795e9029723bbd71cdde2c4dec70cf1fae6efe62ef243f10a"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "95ca965c70c937429cf47776", "attempt_id": "d0cf3614b8ba", "run_group_id": "E4_core:A5|repairable|qwen3b:h2", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:39:50.046576Z", "task_family": "synthetic", "document_id": "syn_03012_2000", "document_hash": "03e441e68d54e1af036f8ff905dfd6cbbccead62636279c1a8e179352ab08945", "question_id": "q01_single_fact", "split": "test", "cluster_id": "syn_03012_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 2, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1927, "query_tokens": 446, "message_tokens": 358, "message_utf8_bytes": 853, "cache_tensor_bytes": null, "serialized_payload_bytes": 853, "side_information_bytes": 0, "fetched_bytes": 461, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 38478.24233304709, "hops": {"1": 23063.031291007064, "2": 15415.211042040028}, "receiver_prefill_ms": 293.0793340783566, "ttft_ms": 18.890832900069654, "decode_ms": 675.3020839532837}, "total_wall_ms": 39466.13458311185, "ttft_ms": 18.890832900069654, "input_usage": {"lookups": 1}, "output_usage": {"output_tokens": 11}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "Lot 890-S in Mont-Laurier", "evidence_artifact": {"ledger": {"facts_total": 59, "facts_retained": 13, "retention_rate": 0.22033898305084745, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.14285714285714285, "amount": 0.0, "status": 0.5833333333333334, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 789}, "ledger_by_hop": {"1": {"facts_total": 59, "facts_retained": 11, "retention_rate": 0.1864406779661017, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.14285714285714285, "amount": 0.0, "status": 0.4166666666666667, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 760}, "2": {"facts_total": 59, "facts_retained": 13, "retention_rate": 0.22033898305084745, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.14285714285714285, "amount": 0.0, "status": 0.5833333333333334, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 789}}}, "reference_key_id": "keys/synthetic/syn_03012_2000.json", "task_scores": {"correct": 0.0, "abstained": 0.0, "f1": 0.5}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "b7a5ccb481b692e93f148c0da2974299bed9aa510b975eeaf745e05e1a120812", "output_hash": "2baffe26d5c1dd5b38462e2e406c5c37aa7822ef8eb0a09dd71903326b3c539a"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "f07c2658570bf551bcf53112", "attempt_id": "b34884bfe829", "run_group_id": "E4_core:A5|repairable|qwen3b:h2", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:39:51.053236Z", "task_family": "synthetic", "document_id": "syn_03012_2000", "document_hash": "03e441e68d54e1af036f8ff905dfd6cbbccead62636279c1a8e179352ab08945", "question_id": "q02_exact_copy", "split": "test", "cluster_id": "syn_03012_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 2, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1927, "query_tokens": 445, "message_tokens": 358, "message_utf8_bytes": 853, "cache_tensor_bytes": null, "serialized_payload_bytes": 853, "side_information_bytes": 0, "fetched_bytes": 461, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 38478.24233304709, "hops": {"1": 23063.031291007064, "2": 15415.211042040028}, "receiver_prefill_ms": 287.48312511015683, "ttft_ms": 38.1362090120092, "decode_ms": 25.419624987989664}, "total_wall_ms": 38829.8657911364, "ttft_ms": 38.1362090120092, "input_usage": {"lookups": 1}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 59, "facts_retained": 13, "retention_rate": 0.22033898305084745, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.14285714285714285, "amount": 0.0, "status": 0.5833333333333334, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 789}, "ledger_by_hop": {"1": {"facts_total": 59, "facts_retained": 11, "retention_rate": 0.1864406779661017, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.14285714285714285, "amount": 0.0, "status": 0.4166666666666667, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 760}, "2": {"facts_total": 59, "facts_retained": 13, "retention_rate": 0.22033898305084745, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.14285714285714285, "amount": 0.0, "status": 0.5833333333333334, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 789}}}, "reference_key_id": "keys/synthetic/syn_03012_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "b7a5ccb481b692e93f148c0da2974299bed9aa510b975eeaf745e05e1a120812", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "f3f27e92382da04001e8cae1", "attempt_id": "acfe17651570", "run_group_id": "E4_core:A5|repairable|qwen3b:h2", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:39:51.421972Z", "task_family": "synthetic", "document_id": "syn_03012_2000", "document_hash": "03e441e68d54e1af036f8ff905dfd6cbbccead62636279c1a8e179352ab08945", "question_id": "q03_exact_copy", "split": "test", "cluster_id": "syn_03012_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 2, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1927, "query_tokens": 441, "message_tokens": 358, "message_utf8_bytes": 853, "cache_tensor_bytes": null, "serialized_payload_bytes": 853, "side_information_bytes": 0, "fetched_bytes": 461, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 38478.24233304709, "hops": {"1": 23063.031291007064, "2": 15415.211042040028}, "receiver_prefill_ms": 296.1811249842867, "ttft_ms": 26.99974994175136, "decode_ms": 36.99783398769796}, "total_wall_ms": 38839.017916005105, "ttft_ms": 26.99974994175136, "input_usage": {"lookups": 1}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 59, "facts_retained": 13, "retention_rate": 0.22033898305084745, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.14285714285714285, "amount": 0.0, "status": 0.5833333333333334, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 789}, "ledger_by_hop": {"1": {"facts_total": 59, "facts_retained": 11, "retention_rate": 0.1864406779661017, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.14285714285714285, "amount": 0.0, "status": 0.4166666666666667, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 760}, "2": {"facts_total": 59, "facts_retained": 13, "retention_rate": 0.22033898305084745, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.14285714285714285, "amount": 0.0, "status": 0.5833333333333334, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 789}}}, "reference_key_id": "keys/synthetic/syn_03012_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "b7a5ccb481b692e93f148c0da2974299bed9aa510b975eeaf745e05e1a120812", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "d185c4db70b0736ccd7b2d50", "attempt_id": "cee88f49e1b3", "run_group_id": "E4_core:A5|repairable|qwen3b:h2", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:39:51.800115Z", "task_family": "synthetic", "document_id": "syn_03012_2000", "document_hash": "03e441e68d54e1af036f8ff905dfd6cbbccead62636279c1a8e179352ab08945", "question_id": "q04_multi_hop", "split": "test", "cluster_id": "syn_03012_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 2, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1927, "query_tokens": 446, "message_tokens": 358, "message_utf8_bytes": 853, "cache_tensor_bytes": null, "serialized_payload_bytes": 853, "side_information_bytes": 0, "fetched_bytes": 461, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 38478.24233304709, "hops": {"1": 23063.031291007064, "2": 15415.211042040028}, "receiver_prefill_ms": 267.33337494079024, "ttft_ms": 42.45170799549669, "decode_ms": 43.59416605439037}, "total_wall_ms": 38832.20025000628, "ttft_ms": 42.45170799549669, "input_usage": {"lookups": 1}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 59, "facts_retained": 13, "retention_rate": 0.22033898305084745, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.14285714285714285, "amount": 0.0, "status": 0.5833333333333334, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 789}, "ledger_by_hop": {"1": {"facts_total": 59, "facts_retained": 11, "retention_rate": 0.1864406779661017, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.14285714285714285, "amount": 0.0, "status": 0.4166666666666667, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 760}, "2": {"facts_total": 59, "facts_retained": 13, "retention_rate": 0.22033898305084745, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.14285714285714285, "amount": 0.0, "status": 0.5833333333333334, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 789}}}, "reference_key_id": "keys/synthetic/syn_03012_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "b7a5ccb481b692e93f148c0da2974299bed9aa510b975eeaf745e05e1a120812", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "890761a7cddb7339c4aaabbf", "attempt_id": "f4b6065cbad2", "run_group_id": "E4_core:A5|repairable|qwen3b:h2", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:39:52.172385Z", "task_family": "synthetic", "document_id": "syn_03012_2000", "document_hash": "03e441e68d54e1af036f8ff905dfd6cbbccead62636279c1a8e179352ab08945", "question_id": "q05_multi_hop", "split": "test", "cluster_id": "syn_03012_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 2, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1927, "query_tokens": 449, "message_tokens": 358, "message_utf8_bytes": 853, "cache_tensor_bytes": null, "serialized_payload_bytes": 853, "side_information_bytes": 0, "fetched_bytes": 461, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 38478.24233304709, "hops": {"1": 23063.031291007064, "2": 15415.211042040028}, "receiver_prefill_ms": 389.463874977082, "ttft_ms": 93.50816602818668, "decode_ms": 38.01625000778586}, "total_wall_ms": 38999.90716599859, "ttft_ms": 93.50816602818668, "input_usage": {"lookups": 1}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 59, "facts_retained": 13, "retention_rate": 0.22033898305084745, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.14285714285714285, "amount": 0.0, "status": 0.5833333333333334, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 789}, "ledger_by_hop": {"1": {"facts_total": 59, "facts_retained": 11, "retention_rate": 0.1864406779661017, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.14285714285714285, "amount": 0.0, "status": 0.4166666666666667, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 760}, "2": {"facts_total": 59, "facts_retained": 13, "retention_rate": 0.22033898305084745, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.14285714285714285, "amount": 0.0, "status": 0.5833333333333334, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 789}}}, "reference_key_id": "keys/synthetic/syn_03012_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "b7a5ccb481b692e93f148c0da2974299bed9aa510b975eeaf745e05e1a120812", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "1de69baf5568eb6b147e68e3", "attempt_id": "8e6fb0f128a3", "run_group_id": "E4_core:A5|repairable|qwen3b:h2", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:39:52.712000Z", "task_family": "synthetic", "document_id": "syn_03012_2000", "document_hash": "03e441e68d54e1af036f8ff905dfd6cbbccead62636279c1a8e179352ab08945", "question_id": "q06_negation", "split": "test", "cluster_id": "syn_03012_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 2, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1927, "query_tokens": 441, "message_tokens": 358, "message_utf8_bytes": 853, "cache_tensor_bytes": null, "serialized_payload_bytes": 853, "side_information_bytes": 0, "fetched_bytes": 461, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 38478.24233304709, "hops": {"1": 23063.031291007064, "2": 15415.211042040028}, "receiver_prefill_ms": 281.4855409087613, "ttft_ms": 30.719792004674673, "decode_ms": 43.95587497856468}, "total_wall_ms": 38834.85066599678, "ttft_ms": 30.719792004674673, "input_usage": {"lookups": 1}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 59, "facts_retained": 13, "retention_rate": 0.22033898305084745, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.14285714285714285, "amount": 0.0, "status": 0.5833333333333334, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 789}, "ledger_by_hop": {"1": {"facts_total": 59, "facts_retained": 11, "retention_rate": 0.1864406779661017, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.14285714285714285, "amount": 0.0, "status": 0.4166666666666667, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 760}, "2": {"facts_total": 59, "facts_retained": 13, "retention_rate": 0.22033898305084745, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.14285714285714285, "amount": 0.0, "status": 0.5833333333333334, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 789}}}, "reference_key_id": "keys/synthetic/syn_03012_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "b7a5ccb481b692e93f148c0da2974299bed9aa510b975eeaf745e05e1a120812", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "a16b30bd8e82fb29ec8414a9", "attempt_id": "8cb3c81224e7", "run_group_id": "E4_core:A5|repairable|qwen3b:h2", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:39:53.085232Z", "task_family": "synthetic", "document_id": "syn_03012_2000", "document_hash": "03e441e68d54e1af036f8ff905dfd6cbbccead62636279c1a8e179352ab08945", "question_id": "q07_negation", "split": "test", "cluster_id": "syn_03012_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 2, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1927, "query_tokens": 441, "message_tokens": 358, "message_utf8_bytes": 853, "cache_tensor_bytes": null, "serialized_payload_bytes": 853, "side_information_bytes": 0, "fetched_bytes": 461, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 38478.24233304709, "hops": {"1": 23063.031291007064, "2": 15415.211042040028}, "receiver_prefill_ms": 292.46716608759016, "ttft_ms": 27.07045793067664, "decode_ms": 40.15004204120487}, "total_wall_ms": 38838.59312499408, "ttft_ms": 27.07045793067664, "input_usage": {"lookups": 1}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 59, "facts_retained": 13, "retention_rate": 0.22033898305084745, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.14285714285714285, "amount": 0.0, "status": 0.5833333333333334, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 789}, "ledger_by_hop": {"1": {"facts_total": 59, "facts_retained": 11, "retention_rate": 0.1864406779661017, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.14285714285714285, "amount": 0.0, "status": 0.4166666666666667, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 760}, "2": {"facts_total": 59, "facts_retained": 13, "retention_rate": 0.22033898305084745, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.14285714285714285, "amount": 0.0, "status": 0.5833333333333334, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 789}}}, "reference_key_id": "keys/synthetic/syn_03012_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "b7a5ccb481b692e93f148c0da2974299bed9aa510b975eeaf745e05e1a120812", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "764b87b9fe651f0f8b5f9543", "attempt_id": "5ad7c405f4fe", "run_group_id": "E4_core:A5|repairable|qwen3b:h2", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:39:53.462752Z", "task_family": "synthetic", "document_id": "syn_03012_2000", "document_hash": "03e441e68d54e1af036f8ff905dfd6cbbccead62636279c1a8e179352ab08945", "question_id": "q08_revision", "split": "test", "cluster_id": "syn_03012_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 2, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1927, "query_tokens": 456, "message_tokens": 358, "message_utf8_bytes": 853, "cache_tensor_bytes": null, "serialized_payload_bytes": 853, "side_information_bytes": 0, "fetched_bytes": 461, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 38478.24233304709, "hops": {"1": 23063.031291007064, "2": 15415.211042040028}, "receiver_prefill_ms": 286.92000010050833, "ttft_ms": 49.16616703849286, "decode_ms": 74.54937498550862}, "total_wall_ms": 38889.548250008374, "ttft_ms": 49.16616703849286, "input_usage": {"lookups": 1}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 59, "facts_retained": 13, "retention_rate": 0.22033898305084745, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.14285714285714285, "amount": 0.0, "status": 0.5833333333333334, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 789}, "ledger_by_hop": {"1": {"facts_total": 59, "facts_retained": 11, "retention_rate": 0.1864406779661017, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.14285714285714285, "amount": 0.0, "status": 0.4166666666666667, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 760}, "2": {"facts_total": 59, "facts_retained": 13, "retention_rate": 0.22033898305084745, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.14285714285714285, "amount": 0.0, "status": 0.5833333333333334, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 789}}}, "reference_key_id": "keys/synthetic/syn_03012_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "b7a5ccb481b692e93f148c0da2974299bed9aa510b975eeaf745e05e1a120812", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "510a1136dbb4a517e0d0453e", "attempt_id": "2ab9f1fe941b", "run_group_id": "E4_core:A5|repairable|qwen3b:h2", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:39:53.892342Z", "task_family": "synthetic", "document_id": "syn_03012_2000", "document_hash": "03e441e68d54e1af036f8ff905dfd6cbbccead62636279c1a8e179352ab08945", "question_id": "q09_revision", "split": "test", "cluster_id": "syn_03012_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 2, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1927, "query_tokens": 449, "message_tokens": 358, "message_utf8_bytes": 853, "cache_tensor_bytes": null, "serialized_payload_bytes": 853, "side_information_bytes": 0, "fetched_bytes": 461, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 38478.24233304709, "hops": {"1": 23063.031291007064, "2": 15415.211042040028}, "receiver_prefill_ms": 329.7884159255773, "ttft_ms": 100.70645809173584, "decode_ms": 40.90379201807082}, "total_wall_ms": 38950.36766608246, "ttft_ms": 100.70645809173584, "input_usage": {"lookups": 1}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 59, "facts_retained": 13, "retention_rate": 0.22033898305084745, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.14285714285714285, "amount": 0.0, "status": 0.5833333333333334, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 789}, "ledger_by_hop": {"1": {"facts_total": 59, "facts_retained": 11, "retention_rate": 0.1864406779661017, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.14285714285714285, "amount": 0.0, "status": 0.4166666666666667, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 760}, "2": {"facts_total": 59, "facts_retained": 13, "retention_rate": 0.22033898305084745, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.14285714285714285, "amount": 0.0, "status": 0.5833333333333334, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 789}}}, "reference_key_id": "keys/synthetic/syn_03012_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "b7a5ccb481b692e93f148c0da2974299bed9aa510b975eeaf745e05e1a120812", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "d17eb8a593fc8480688fbb0a", "attempt_id": "8a9e3bf0cf41", "run_group_id": "E4_core:A5|repairable|qwen3b:h2", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:39:54.384131Z", "task_family": "synthetic", "document_id": "syn_03012_2000", "document_hash": "03e441e68d54e1af036f8ff905dfd6cbbccead62636279c1a8e179352ab08945", "question_id": "q10_temporal", "split": "test", "cluster_id": "syn_03012_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 2, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1927, "query_tokens": 459, "message_tokens": 358, "message_utf8_bytes": 853, "cache_tensor_bytes": null, "serialized_payload_bytes": 853, "side_information_bytes": 0, "fetched_bytes": 461, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 38478.24233304709, "hops": {"1": 23063.031291007064, "2": 15415.211042040028}, "receiver_prefill_ms": 314.2253749538213, "ttft_ms": 35.06270900834352, "decode_ms": 38.04383298847824}, "total_wall_ms": 38866.16658302955, "ttft_ms": 35.06270900834352, "input_usage": {"lookups": 1}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 59, "facts_retained": 13, "retention_rate": 0.22033898305084745, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.14285714285714285, "amount": 0.0, "status": 0.5833333333333334, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 789}, "ledger_by_hop": {"1": {"facts_total": 59, "facts_retained": 11, "retention_rate": 0.1864406779661017, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.14285714285714285, "amount": 0.0, "status": 0.4166666666666667, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 760}, "2": {"facts_total": 59, "facts_retained": 13, "retention_rate": 0.22033898305084745, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.14285714285714285, "amount": 0.0, "status": 0.5833333333333334, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 789}}}, "reference_key_id": "keys/synthetic/syn_03012_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "b7a5ccb481b692e93f148c0da2974299bed9aa510b975eeaf745e05e1a120812", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "155f425edd8547f14cd58f04", "attempt_id": "864e27526346", "run_group_id": "E4_core:A5|repairable|qwen3b:h2", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:39:54.789980Z", "task_family": "synthetic", "document_id": "syn_03012_2000", "document_hash": "03e441e68d54e1af036f8ff905dfd6cbbccead62636279c1a8e179352ab08945", "question_id": "q11_aggregation", "split": "test", "cluster_id": "syn_03012_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 2, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1927, "query_tokens": 451, "message_tokens": 358, "message_utf8_bytes": 853, "cache_tensor_bytes": null, "serialized_payload_bytes": 853, "side_information_bytes": 0, "fetched_bytes": 461, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 38478.24233304709, "hops": {"1": 23063.031291007064, "2": 15415.211042040028}, "receiver_prefill_ms": 318.4124999679625, "ttft_ms": 27.02375000808388, "decode_ms": 38.647959008812904}, "total_wall_ms": 38862.78754204977, "ttft_ms": 27.02375000808388, "input_usage": {"lookups": 1}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 59, "facts_retained": 13, "retention_rate": 0.22033898305084745, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.14285714285714285, "amount": 0.0, "status": 0.5833333333333334, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 789}, "ledger_by_hop": {"1": {"facts_total": 59, "facts_retained": 11, "retention_rate": 0.1864406779661017, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.14285714285714285, "amount": 0.0, "status": 0.4166666666666667, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 760}, "2": {"facts_total": 59, "facts_retained": 13, "retention_rate": 0.22033898305084745, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.14285714285714285, "amount": 0.0, "status": 0.5833333333333334, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 789}}}, "reference_key_id": "keys/synthetic/syn_03012_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "b7a5ccb481b692e93f148c0da2974299bed9aa510b975eeaf745e05e1a120812", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "e22c8a717c289b3dfd6b0854", "attempt_id": "f569d15704ac", "run_group_id": "E4_core:A5|repairable|qwen3b:h2", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:39:55.191888Z", "task_family": "synthetic", "document_id": "syn_03012_2000", "document_hash": "03e441e68d54e1af036f8ff905dfd6cbbccead62636279c1a8e179352ab08945", "question_id": "q12_unit_conversion", "split": "test", "cluster_id": "syn_03012_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 2, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1927, "query_tokens": 448, "message_tokens": 358, "message_utf8_bytes": 853, "cache_tensor_bytes": null, "serialized_payload_bytes": 853, "side_information_bytes": 0, "fetched_bytes": 461, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 38478.24233304709, "hops": {"1": 23063.031291007064, "2": 15415.211042040028}, "receiver_prefill_ms": 257.9436250962317, "ttft_ms": 56.37366697192192, "decode_ms": 428.18616703152657}, "total_wall_ms": 39221.29170806147, "ttft_ms": 56.37366697192192, "input_usage": {"lookups": 1}, "output_usage": {"output_tokens": 5}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "44620", "evidence_artifact": {"ledger": {"facts_total": 59, "facts_retained": 13, "retention_rate": 0.22033898305084745, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.14285714285714285, "amount": 0.0, "status": 0.5833333333333334, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 789}, "ledger_by_hop": {"1": {"facts_total": 59, "facts_retained": 11, "retention_rate": 0.1864406779661017, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.14285714285714285, "amount": 0.0, "status": 0.4166666666666667, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 760}, "2": {"facts_total": 59, "facts_retained": 13, "retention_rate": 0.22033898305084745, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.14285714285714285, "amount": 0.0, "status": 0.5833333333333334, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 789}}}, "reference_key_id": "keys/synthetic/syn_03012_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "b7a5ccb481b692e93f148c0da2974299bed9aa510b975eeaf745e05e1a120812", "output_hash": "2bd125847727be7e7c0a656ee4c81f3bc6d0fef11f536f50ce7bc83688907075"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "36c0dd93ce04fb303998694c", "attempt_id": "575c6dacdd36", "run_group_id": "E4_core:A5|repairable|qwen3b:h2", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:39:55.951976Z", "task_family": "synthetic", "document_id": "syn_03012_2000", "document_hash": "03e441e68d54e1af036f8ff905dfd6cbbccead62636279c1a8e179352ab08945", "question_id": "q13_unanswerable", "split": "test", "cluster_id": "syn_03012_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 2, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1927, "query_tokens": 453, "message_tokens": 358, "message_utf8_bytes": 853, "cache_tensor_bytes": null, "serialized_payload_bytes": 853, "side_information_bytes": 0, "fetched_bytes": 461, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 38478.24233304709, "hops": {"1": 23063.031291007064, "2": 15415.211042040028}, "receiver_prefill_ms": 313.01383289974183, "ttft_ms": 39.417667081579566, "decode_ms": 24.952500010840595}, "total_wall_ms": 38856.155541026965, "ttft_ms": 39.417667081579566, "input_usage": {"lookups": 1}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 59, "facts_retained": 13, "retention_rate": 0.22033898305084745, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.14285714285714285, "amount": 0.0, "status": 0.5833333333333334, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 789}, "ledger_by_hop": {"1": {"facts_total": 59, "facts_retained": 11, "retention_rate": 0.1864406779661017, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.14285714285714285, "amount": 0.0, "status": 0.4166666666666667, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 760}, "2": {"facts_total": 59, "facts_retained": 13, "retention_rate": 0.22033898305084745, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.14285714285714285, "amount": 0.0, "status": 0.5833333333333334, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 789}}}, "reference_key_id": "keys/synthetic/syn_03012_2000.json", "task_scores": {"correct": 1.0, "abstained": 1.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "b7a5ccb481b692e93f148c0da2974299bed9aa510b975eeaf745e05e1a120812", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "b1c5ffb8be28918386ee6dfc", "attempt_id": "f3cd44032c38", "run_group_id": "E4_core:A5|repairable|qwen3b:h2", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:39:56.347278Z", "task_family": "synthetic", "document_id": "syn_03012_2000", "document_hash": "03e441e68d54e1af036f8ff905dfd6cbbccead62636279c1a8e179352ab08945", "question_id": "q14_unanswerable", "split": "test", "cluster_id": "syn_03012_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 2, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1927, "query_tokens": 453, "message_tokens": 358, "message_utf8_bytes": 853, "cache_tensor_bytes": null, "serialized_payload_bytes": 853, "side_information_bytes": 0, "fetched_bytes": 461, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 38478.24233304709, "hops": {"1": 23063.031291007064, "2": 15415.211042040028}, "receiver_prefill_ms": 319.9374160030857, "ttft_ms": 38.391915964894, "decode_ms": 23.72645796276629}, "total_wall_ms": 38860.73466704693, "ttft_ms": 38.391915964894, "input_usage": {"lookups": 1}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 59, "facts_retained": 13, "retention_rate": 0.22033898305084745, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.14285714285714285, "amount": 0.0, "status": 0.5833333333333334, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 789}, "ledger_by_hop": {"1": {"facts_total": 59, "facts_retained": 11, "retention_rate": 0.1864406779661017, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.14285714285714285, "amount": 0.0, "status": 0.4166666666666667, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 760}, "2": {"facts_total": 59, "facts_retained": 13, "retention_rate": 0.22033898305084745, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.14285714285714285, "amount": 0.0, "status": 0.5833333333333334, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 789}}}, "reference_key_id": "keys/synthetic/syn_03012_2000.json", "task_scores": {"correct": 1.0, "abstained": 1.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "b7a5ccb481b692e93f148c0da2974299bed9aa510b975eeaf745e05e1a120812", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "67bc3c54bdaea837cfd77636", "attempt_id": "65aeac13b8d5", "run_group_id": "E4_core:A5|repairable|qwen3b:h4", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:40:28.210311Z", "task_family": "synthetic", "document_id": "syn_03012_2000", "document_hash": "03e441e68d54e1af036f8ff905dfd6cbbccead62636279c1a8e179352ab08945", "question_id": "q00_single_fact", "split": "test", "cluster_id": "syn_03012_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 4, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1927, "query_tokens": 440, "message_tokens": 358, "message_utf8_bytes": 853, "cache_tensor_bytes": null, "serialized_payload_bytes": 853, "side_information_bytes": 0, "fetched_bytes": 1383, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 69939.11387491971, "hops": {"1": 23063.031291007064, "2": 15415.211042040028, "3": 15675.964874913916, "4": 15784.906666958705}, "receiver_prefill_ms": 252.82512500416487, "ttft_ms": 37.81666699796915, "decode_ms": 564.003249979578}, "total_wall_ms": 70794.3132079672, "ttft_ms": 37.81666699796915, "input_usage": {"lookups": 3}, "output_usage": {"output_tokens": 10}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "XX-71M-6213", "evidence_artifact": {"ledger": {"facts_total": 59, "facts_retained": 13, "retention_rate": 0.22033898305084745, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.14285714285714285, "amount": 0.0, "status": 0.5833333333333334, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 789}, "ledger_by_hop": {"1": {"facts_total": 59, "facts_retained": 11, "retention_rate": 0.1864406779661017, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.14285714285714285, "amount": 0.0, "status": 0.4166666666666667, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 760}, "2": {"facts_total": 59, "facts_retained": 13, "retention_rate": 0.22033898305084745, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.14285714285714285, "amount": 0.0, "status": 0.5833333333333334, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 789}, "3": {"facts_total": 59, "facts_retained": 13, "retention_rate": 0.22033898305084745, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.14285714285714285, "amount": 0.0, "status": 0.5833333333333334, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 789}, "4": {"facts_total": 59, "facts_retained": 13, "retention_rate": 0.22033898305084745, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.14285714285714285, "amount": 0.0, "status": 0.5833333333333334, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 789}}}, "reference_key_id": "keys/synthetic/syn_03012_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "b7a5ccb481b692e93f148c0da2974299bed9aa510b975eeaf745e05e1a120812", "output_hash": "66ef6c5da682b38795e9029723bbd71cdde2c4dec70cf1fae6efe62ef243f10a"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "26d78c2d6aa3c6eaa9e66d67", "attempt_id": "04c870494aa2", "run_group_id": "E4_core:A5|repairable|qwen3b:h4", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:40:29.083125Z", "task_family": "synthetic", "document_id": "syn_03012_2000", "document_hash": "03e441e68d54e1af036f8ff905dfd6cbbccead62636279c1a8e179352ab08945", "question_id": "q01_single_fact", "split": "test", "cluster_id": "syn_03012_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 4, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1927, "query_tokens": 446, "message_tokens": 358, "message_utf8_bytes": 853, "cache_tensor_bytes": null, "serialized_payload_bytes": 853, "side_information_bytes": 0, "fetched_bytes": 1383, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 69939.11387491971, "hops": {"1": 23063.031291007064, "2": 15415.211042040028, "3": 15675.964874913916, "4": 15784.906666958705}, "receiver_prefill_ms": 294.9420000659302, "ttft_ms": 44.092083000577986, "decode_ms": 22.11670798715204}, "total_wall_ms": 70300.73820799589, "ttft_ms": 44.092083000577986, "input_usage": {"lookups": 3}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 59, "facts_retained": 13, "retention_rate": 0.22033898305084745, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.14285714285714285, "amount": 0.0, "status": 0.5833333333333334, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 789}, "ledger_by_hop": {"1": {"facts_total": 59, "facts_retained": 11, "retention_rate": 0.1864406779661017, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.14285714285714285, "amount": 0.0, "status": 0.4166666666666667, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 760}, "2": {"facts_total": 59, "facts_retained": 13, "retention_rate": 0.22033898305084745, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.14285714285714285, "amount": 0.0, "status": 0.5833333333333334, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 789}, "3": {"facts_total": 59, "facts_retained": 13, "retention_rate": 0.22033898305084745, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.14285714285714285, "amount": 0.0, "status": 0.5833333333333334, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 789}, "4": {"facts_total": 59, "facts_retained": 13, "retention_rate": 0.22033898305084745, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.14285714285714285, "amount": 0.0, "status": 0.5833333333333334, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 789}}}, "reference_key_id": "keys/synthetic/syn_03012_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "b7a5ccb481b692e93f148c0da2974299bed9aa510b975eeaf745e05e1a120812", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "ebe9b82a2ad62554ccf2bafe", "attempt_id": "0237b447f516", "run_group_id": "E4_core:A5|repairable|qwen3b:h4", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:40:29.461873Z", "task_family": "synthetic", "document_id": "syn_03012_2000", "document_hash": "03e441e68d54e1af036f8ff905dfd6cbbccead62636279c1a8e179352ab08945", "question_id": "q02_exact_copy", "split": "test", "cluster_id": "syn_03012_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 4, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1927, "query_tokens": 445, "message_tokens": 358, "message_utf8_bytes": 853, "cache_tensor_bytes": null, "serialized_payload_bytes": 853, "side_information_bytes": 0, "fetched_bytes": 1383, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 69939.11387491971, "hops": {"1": 23063.031291007064, "2": 15415.211042040028, "3": 15675.964874913916, "4": 15784.906666958705}, "receiver_prefill_ms": 296.00612504873425, "ttft_ms": 38.48383401054889, "decode_ms": 24.03008402325213}, "total_wall_ms": 70298.07779192924, "ttft_ms": 38.48383401054889, "input_usage": {"lookups": 3}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 59, "facts_retained": 13, "retention_rate": 0.22033898305084745, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.14285714285714285, "amount": 0.0, "status": 0.5833333333333334, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 789}, "ledger_by_hop": {"1": {"facts_total": 59, "facts_retained": 11, "retention_rate": 0.1864406779661017, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.14285714285714285, "amount": 0.0, "status": 0.4166666666666667, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 760}, "2": {"facts_total": 59, "facts_retained": 13, "retention_rate": 0.22033898305084745, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.14285714285714285, "amount": 0.0, "status": 0.5833333333333334, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 789}, "3": {"facts_total": 59, "facts_retained": 13, "retention_rate": 0.22033898305084745, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.14285714285714285, "amount": 0.0, "status": 0.5833333333333334, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 789}, "4": {"facts_total": 59, "facts_retained": 13, "retention_rate": 0.22033898305084745, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.14285714285714285, "amount": 0.0, "status": 0.5833333333333334, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 789}}}, "reference_key_id": "keys/synthetic/syn_03012_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "b7a5ccb481b692e93f148c0da2974299bed9aa510b975eeaf745e05e1a120812", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "7364ab293a107cf753b26137", "attempt_id": "036e8669673b", "run_group_id": "E4_core:A5|repairable|qwen3b:h4", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:40:29.837706Z", "task_family": "synthetic", "document_id": "syn_03012_2000", "document_hash": "03e441e68d54e1af036f8ff905dfd6cbbccead62636279c1a8e179352ab08945", "question_id": "q03_exact_copy", "split": "test", "cluster_id": "syn_03012_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 4, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1927, "query_tokens": 441, "message_tokens": 358, "message_utf8_bytes": 853, "cache_tensor_bytes": null, "serialized_payload_bytes": 853, "side_information_bytes": 0, "fetched_bytes": 1383, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 69939.11387491971, "hops": {"1": 23063.031291007064, "2": 15415.211042040028, "3": 15675.964874913916, "4": 15784.906666958705}, "receiver_prefill_ms": 289.65037502348423, "ttft_ms": 46.98908398859203, "decode_ms": 53.73400007374585}, "total_wall_ms": 70330.03420790192, "ttft_ms": 46.98908398859203, "input_usage": {"lookups": 3}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 59, "facts_retained": 13, "retention_rate": 0.22033898305084745, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.14285714285714285, "amount": 0.0, "status": 0.5833333333333334, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 789}, "ledger_by_hop": {"1": {"facts_total": 59, "facts_retained": 11, "retention_rate": 0.1864406779661017, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.14285714285714285, "amount": 0.0, "status": 0.4166666666666667, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 760}, "2": {"facts_total": 59, "facts_retained": 13, "retention_rate": 0.22033898305084745, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.14285714285714285, "amount": 0.0, "status": 0.5833333333333334, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 789}, "3": {"facts_total": 59, "facts_retained": 13, "retention_rate": 0.22033898305084745, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.14285714285714285, "amount": 0.0, "status": 0.5833333333333334, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 789}, "4": {"facts_total": 59, "facts_retained": 13, "retention_rate": 0.22033898305084745, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.14285714285714285, "amount": 0.0, "status": 0.5833333333333334, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 789}}}, "reference_key_id": "keys/synthetic/syn_03012_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "b7a5ccb481b692e93f148c0da2974299bed9aa510b975eeaf745e05e1a120812", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "59fb7b5ed2bebbac401969b2", "attempt_id": "501f11c0afa0", "run_group_id": "E4_core:A5|repairable|qwen3b:h4", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:40:30.247834Z", "task_family": "synthetic", "document_id": "syn_03012_2000", "document_hash": "03e441e68d54e1af036f8ff905dfd6cbbccead62636279c1a8e179352ab08945", "question_id": "q04_multi_hop", "split": "test", "cluster_id": "syn_03012_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 4, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1927, "query_tokens": 446, "message_tokens": 358, "message_utf8_bytes": 853, "cache_tensor_bytes": null, "serialized_payload_bytes": 853, "side_information_bytes": 0, "fetched_bytes": 1383, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 69939.11387491971, "hops": {"1": 23063.031291007064, "2": 15415.211042040028, "3": 15675.964874913916, "4": 15784.906666958705}, "receiver_prefill_ms": 447.551375022158, "ttft_ms": 32.174125080928206, "decode_ms": 38.5493750218302}, "total_wall_ms": 70457.92887487914, "ttft_ms": 32.174125080928206, "input_usage": {"lookups": 3}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 59, "facts_retained": 13, "retention_rate": 0.22033898305084745, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.14285714285714285, "amount": 0.0, "status": 0.5833333333333334, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 789}, "ledger_by_hop": {"1": {"facts_total": 59, "facts_retained": 11, "retention_rate": 0.1864406779661017, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.14285714285714285, "amount": 0.0, "status": 0.4166666666666667, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 760}, "2": {"facts_total": 59, "facts_retained": 13, "retention_rate": 0.22033898305084745, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.14285714285714285, "amount": 0.0, "status": 0.5833333333333334, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 789}, "3": {"facts_total": 59, "facts_retained": 13, "retention_rate": 0.22033898305084745, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.14285714285714285, "amount": 0.0, "status": 0.5833333333333334, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 789}, "4": {"facts_total": 59, "facts_retained": 13, "retention_rate": 0.22033898305084745, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.14285714285714285, "amount": 0.0, "status": 0.5833333333333334, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 789}}}, "reference_key_id": "keys/synthetic/syn_03012_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "b7a5ccb481b692e93f148c0da2974299bed9aa510b975eeaf745e05e1a120812", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "9b814333a57828869e524317", "attempt_id": "256c2bd90fbc", "run_group_id": "E4_core:A5|repairable|qwen3b:h4", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:40:30.783885Z", "task_family": "synthetic", "document_id": "syn_03012_2000", "document_hash": "03e441e68d54e1af036f8ff905dfd6cbbccead62636279c1a8e179352ab08945", "question_id": "q05_multi_hop", "split": "test", "cluster_id": "syn_03012_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 4, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1927, "query_tokens": 449, "message_tokens": 358, "message_utf8_bytes": 853, "cache_tensor_bytes": null, "serialized_payload_bytes": 853, "side_information_bytes": 0, "fetched_bytes": 1383, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 69939.11387491971, "hops": {"1": 23063.031291007064, "2": 15415.211042040028, "3": 15675.964874913916, "4": 15784.906666958705}, "receiver_prefill_ms": 297.51599999144673, "ttft_ms": 42.89408295881003, "decode_ms": 25.091832969337702}, "total_wall_ms": 70305.20858301315, "ttft_ms": 42.89408295881003, "input_usage": {"lookups": 3}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 59, "facts_retained": 13, "retention_rate": 0.22033898305084745, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.14285714285714285, "amount": 0.0, "status": 0.5833333333333334, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 789}, "ledger_by_hop": {"1": {"facts_total": 59, "facts_retained": 11, "retention_rate": 0.1864406779661017, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.14285714285714285, "amount": 0.0, "status": 0.4166666666666667, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 760}, "2": {"facts_total": 59, "facts_retained": 13, "retention_rate": 0.22033898305084745, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.14285714285714285, "amount": 0.0, "status": 0.5833333333333334, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 789}, "3": {"facts_total": 59, "facts_retained": 13, "retention_rate": 0.22033898305084745, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.14285714285714285, "amount": 0.0, "status": 0.5833333333333334, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 789}, "4": {"facts_total": 59, "facts_retained": 13, "retention_rate": 0.22033898305084745, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.14285714285714285, "amount": 0.0, "status": 0.5833333333333334, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 789}}}, "reference_key_id": "keys/synthetic/syn_03012_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "b7a5ccb481b692e93f148c0da2974299bed9aa510b975eeaf745e05e1a120812", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "4d3dafec2ce08fb80e5413ff", "attempt_id": "8b5723d5d9fe", "run_group_id": "E4_core:A5|repairable|qwen3b:h4", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:40:31.167855Z", "task_family": "synthetic", "document_id": "syn_03012_2000", "document_hash": "03e441e68d54e1af036f8ff905dfd6cbbccead62636279c1a8e179352ab08945", "question_id": "q06_negation", "split": "test", "cluster_id": "syn_03012_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 4, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1927, "query_tokens": 441, "message_tokens": 358, "message_utf8_bytes": 853, "cache_tensor_bytes": null, "serialized_payload_bytes": 853, "side_information_bytes": 0, "fetched_bytes": 1383, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 69939.11387491971, "hops": {"1": 23063.031291007064, "2": 15415.211042040028, "3": 15675.964874913916, "4": 15784.906666958705}, "receiver_prefill_ms": 301.25804198905826, "ttft_ms": 23.35454197600484, "decode_ms": 17.510042060166597}, "total_wall_ms": 70281.77979192697, "ttft_ms": 23.35454197600484, "input_usage": {"lookups": 3}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 59, "facts_retained": 13, "retention_rate": 0.22033898305084745, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.14285714285714285, "amount": 0.0, "status": 0.5833333333333334, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 789}, "ledger_by_hop": {"1": {"facts_total": 59, "facts_retained": 11, "retention_rate": 0.1864406779661017, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.14285714285714285, "amount": 0.0, "status": 0.4166666666666667, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 760}, "2": {"facts_total": 59, "facts_retained": 13, "retention_rate": 0.22033898305084745, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.14285714285714285, "amount": 0.0, "status": 0.5833333333333334, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 789}, "3": {"facts_total": 59, "facts_retained": 13, "retention_rate": 0.22033898305084745, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.14285714285714285, "amount": 0.0, "status": 0.5833333333333334, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 789}, "4": {"facts_total": 59, "facts_retained": 13, "retention_rate": 0.22033898305084745, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.14285714285714285, "amount": 0.0, "status": 0.5833333333333334, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 789}}}, "reference_key_id": "keys/synthetic/syn_03012_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "b7a5ccb481b692e93f148c0da2974299bed9aa510b975eeaf745e05e1a120812", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "eb52e49495cda8fe3cdd64e9", "attempt_id": "204754143709", "run_group_id": "E4_core:A5|repairable|qwen3b:h4", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:40:31.527598Z", "task_family": "synthetic", "document_id": "syn_03012_2000", "document_hash": "03e441e68d54e1af036f8ff905dfd6cbbccead62636279c1a8e179352ab08945", "question_id": "q07_negation", "split": "test", "cluster_id": "syn_03012_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 4, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1927, "query_tokens": 441, "message_tokens": 358, "message_utf8_bytes": 853, "cache_tensor_bytes": null, "serialized_payload_bytes": 853, "side_information_bytes": 0, "fetched_bytes": 1383, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 69939.11387491971, "hops": {"1": 23063.031291007064, "2": 15415.211042040028, "3": 15675.964874913916, "4": 15784.906666958705}, "receiver_prefill_ms": 275.26100003160536, "ttft_ms": 60.478791943751276, "decode_ms": 76.7247499898076}, "total_wall_ms": 70352.11737488862, "ttft_ms": 60.478791943751276, "input_usage": {"lookups": 3}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 59, "facts_retained": 13, "retention_rate": 0.22033898305084745, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.14285714285714285, "amount": 0.0, "status": 0.5833333333333334, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 789}, "ledger_by_hop": {"1": {"facts_total": 59, "facts_retained": 11, "retention_rate": 0.1864406779661017, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.14285714285714285, "amount": 0.0, "status": 0.4166666666666667, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 760}, "2": {"facts_total": 59, "facts_retained": 13, "retention_rate": 0.22033898305084745, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.14285714285714285, "amount": 0.0, "status": 0.5833333333333334, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 789}, "3": {"facts_total": 59, "facts_retained": 13, "retention_rate": 0.22033898305084745, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.14285714285714285, "amount": 0.0, "status": 0.5833333333333334, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 789}, "4": {"facts_total": 59, "facts_retained": 13, "retention_rate": 0.22033898305084745, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.14285714285714285, "amount": 0.0, "status": 0.5833333333333334, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 789}}}, "reference_key_id": "keys/synthetic/syn_03012_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "b7a5ccb481b692e93f148c0da2974299bed9aa510b975eeaf745e05e1a120812", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "f0d79bad51e3f6b7562f7ad0", "attempt_id": "3eccef40738b", "run_group_id": "E4_core:A5|repairable|qwen3b:h4", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:40:31.959870Z", "task_family": "synthetic", "document_id": "syn_03012_2000", "document_hash": "03e441e68d54e1af036f8ff905dfd6cbbccead62636279c1a8e179352ab08945", "question_id": "q08_revision", "split": "test", "cluster_id": "syn_03012_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 4, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1927, "query_tokens": 456, "message_tokens": 358, "message_utf8_bytes": 853, "cache_tensor_bytes": null, "serialized_payload_bytes": 853, "side_information_bytes": 0, "fetched_bytes": 1383, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 69939.11387491971, "hops": {"1": 23063.031291007064, "2": 15415.211042040028, "3": 15675.964874913916, "4": 15784.906666958705}, "receiver_prefill_ms": 432.68079101108015, "ttft_ms": 37.565666949376464, "decode_ms": 36.53020807541907}, "total_wall_ms": 70446.4225829579, "ttft_ms": 37.565666949376464, "input_usage": {"lookups": 3}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 59, "facts_retained": 13, "retention_rate": 0.22033898305084745, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.14285714285714285, "amount": 0.0, "status": 0.5833333333333334, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 789}, "ledger_by_hop": {"1": {"facts_total": 59, "facts_retained": 11, "retention_rate": 0.1864406779661017, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.14285714285714285, "amount": 0.0, "status": 0.4166666666666667, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 760}, "2": {"facts_total": 59, "facts_retained": 13, "retention_rate": 0.22033898305084745, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.14285714285714285, "amount": 0.0, "status": 0.5833333333333334, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 789}, "3": {"facts_total": 59, "facts_retained": 13, "retention_rate": 0.22033898305084745, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.14285714285714285, "amount": 0.0, "status": 0.5833333333333334, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 789}, "4": {"facts_total": 59, "facts_retained": 13, "retention_rate": 0.22033898305084745, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.14285714285714285, "amount": 0.0, "status": 0.5833333333333334, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 789}}}, "reference_key_id": "keys/synthetic/syn_03012_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "b7a5ccb481b692e93f148c0da2974299bed9aa510b975eeaf745e05e1a120812", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "3d1be62dbb2715c9faba02b3", "attempt_id": "8bf7c1df40ce", "run_group_id": "E4_core:A5|repairable|qwen3b:h4", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:40:32.484962Z", "task_family": "synthetic", "document_id": "syn_03012_2000", "document_hash": "03e441e68d54e1af036f8ff905dfd6cbbccead62636279c1a8e179352ab08945", "question_id": "q09_revision", "split": "test", "cluster_id": "syn_03012_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 4, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1927, "query_tokens": 449, "message_tokens": 358, "message_utf8_bytes": 853, "cache_tensor_bytes": null, "serialized_payload_bytes": 853, "side_information_bytes": 0, "fetched_bytes": 1383, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 69939.11387491971, "hops": {"1": 23063.031291007064, "2": 15415.211042040028, "3": 15675.964874913916, "4": 15784.906666958705}, "receiver_prefill_ms": 272.9569999501109, "ttft_ms": 28.190082986839116, "decode_ms": 42.92112495750189}, "total_wall_ms": 70283.80400000606, "ttft_ms": 28.190082986839116, "input_usage": {"lookups": 3}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 59, "facts_retained": 13, "retention_rate": 0.22033898305084745, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.14285714285714285, "amount": 0.0, "status": 0.5833333333333334, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 789}, "ledger_by_hop": {"1": {"facts_total": 59, "facts_retained": 11, "retention_rate": 0.1864406779661017, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.14285714285714285, "amount": 0.0, "status": 0.4166666666666667, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 760}, "2": {"facts_total": 59, "facts_retained": 13, "retention_rate": 0.22033898305084745, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.14285714285714285, "amount": 0.0, "status": 0.5833333333333334, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 789}, "3": {"facts_total": 59, "facts_retained": 13, "retention_rate": 0.22033898305084745, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.14285714285714285, "amount": 0.0, "status": 0.5833333333333334, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 789}, "4": {"facts_total": 59, "facts_retained": 13, "retention_rate": 0.22033898305084745, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.14285714285714285, "amount": 0.0, "status": 0.5833333333333334, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 789}}}, "reference_key_id": "keys/synthetic/syn_03012_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "b7a5ccb481b692e93f148c0da2974299bed9aa510b975eeaf745e05e1a120812", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "187b1ab270f5ec2a87da532d", "attempt_id": "0bc14ec2427a", "run_group_id": "E4_core:A5|repairable|qwen3b:h4", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:40:32.846722Z", "task_family": "synthetic", "document_id": "syn_03012_2000", "document_hash": "03e441e68d54e1af036f8ff905dfd6cbbccead62636279c1a8e179352ab08945", "question_id": "q10_temporal", "split": "test", "cluster_id": "syn_03012_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 4, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1927, "query_tokens": 459, "message_tokens": 358, "message_utf8_bytes": 853, "cache_tensor_bytes": null, "serialized_payload_bytes": 853, "side_information_bytes": 0, "fetched_bytes": 1383, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 69939.11387491971, "hops": {"1": 23063.031291007064, "2": 15415.211042040028, "3": 15675.964874913916, "4": 15784.906666958705}, "receiver_prefill_ms": 318.1129170116037, "ttft_ms": 21.81654202286154, "decode_ms": 22.586000035516918}, "total_wall_ms": 70302.12712485809, "ttft_ms": 21.81654202286154, "input_usage": {"lookups": 3}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 59, "facts_retained": 13, "retention_rate": 0.22033898305084745, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.14285714285714285, "amount": 0.0, "status": 0.5833333333333334, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 789}, "ledger_by_hop": {"1": {"facts_total": 59, "facts_retained": 11, "retention_rate": 0.1864406779661017, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.14285714285714285, "amount": 0.0, "status": 0.4166666666666667, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 760}, "2": {"facts_total": 59, "facts_retained": 13, "retention_rate": 0.22033898305084745, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.14285714285714285, "amount": 0.0, "status": 0.5833333333333334, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 789}, "3": {"facts_total": 59, "facts_retained": 13, "retention_rate": 0.22033898305084745, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.14285714285714285, "amount": 0.0, "status": 0.5833333333333334, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 789}, "4": {"facts_total": 59, "facts_retained": 13, "retention_rate": 0.22033898305084745, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.14285714285714285, "amount": 0.0, "status": 0.5833333333333334, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 789}}}, "reference_key_id": "keys/synthetic/syn_03012_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "b7a5ccb481b692e93f148c0da2974299bed9aa510b975eeaf745e05e1a120812", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "faa02396f1d8318984e509a7", "attempt_id": "f211c729c3f3", "run_group_id": "E4_core:A5|repairable|qwen3b:h4", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:40:33.226445Z", "task_family": "synthetic", "document_id": "syn_03012_2000", "document_hash": "03e441e68d54e1af036f8ff905dfd6cbbccead62636279c1a8e179352ab08945", "question_id": "q11_aggregation", "split": "test", "cluster_id": "syn_03012_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 4, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1927, "query_tokens": 451, "message_tokens": 358, "message_utf8_bytes": 853, "cache_tensor_bytes": null, "serialized_payload_bytes": 853, "side_information_bytes": 0, "fetched_bytes": 1383, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 69939.11387491971, "hops": {"1": 23063.031291007064, "2": 15415.211042040028, "3": 15675.964874913916, "4": 15784.906666958705}, "receiver_prefill_ms": 300.36487500183284, "ttft_ms": 54.64133294299245, "decode_ms": 67.00866704341024}, "total_wall_ms": 70361.75179085694, "ttft_ms": 54.64133294299245, "input_usage": {"lookups": 3}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 59, "facts_retained": 13, "retention_rate": 0.22033898305084745, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.14285714285714285, "amount": 0.0, "status": 0.5833333333333334, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 789}, "ledger_by_hop": {"1": {"facts_total": 59, "facts_retained": 11, "retention_rate": 0.1864406779661017, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.14285714285714285, "amount": 0.0, "status": 0.4166666666666667, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 760}, "2": {"facts_total": 59, "facts_retained": 13, "retention_rate": 0.22033898305084745, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.14285714285714285, "amount": 0.0, "status": 0.5833333333333334, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 789}, "3": {"facts_total": 59, "facts_retained": 13, "retention_rate": 0.22033898305084745, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.14285714285714285, "amount": 0.0, "status": 0.5833333333333334, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 789}, "4": {"facts_total": 59, "facts_retained": 13, "retention_rate": 0.22033898305084745, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.14285714285714285, "amount": 0.0, "status": 0.5833333333333334, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 789}}}, "reference_key_id": "keys/synthetic/syn_03012_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "b7a5ccb481b692e93f148c0da2974299bed9aa510b975eeaf745e05e1a120812", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "e4e123ec2d39d7cf7093d57e", "attempt_id": "7593d56fb8fd", "run_group_id": "E4_core:A5|repairable|qwen3b:h4", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:40:33.668372Z", "task_family": "synthetic", "document_id": "syn_03012_2000", "document_hash": "03e441e68d54e1af036f8ff905dfd6cbbccead62636279c1a8e179352ab08945", "question_id": "q12_unit_conversion", "split": "test", "cluster_id": "syn_03012_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 4, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1927, "query_tokens": 448, "message_tokens": 358, "message_utf8_bytes": 853, "cache_tensor_bytes": null, "serialized_payload_bytes": 853, "side_information_bytes": 0, "fetched_bytes": 1383, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 69939.11387491971, "hops": {"1": 23063.031291007064, "2": 15415.211042040028, "3": 15675.964874913916, "4": 15784.906666958705}, "receiver_prefill_ms": 423.00624994095415, "ttft_ms": 33.565124962478876, "decode_ms": 191.98845804203302}, "total_wall_ms": 70588.25495792553, "ttft_ms": 33.565124962478876, "input_usage": {"lookups": 3}, "output_usage": {"output_tokens": 5}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "44620", "evidence_artifact": {"ledger": {"facts_total": 59, "facts_retained": 13, "retention_rate": 0.22033898305084745, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.14285714285714285, "amount": 0.0, "status": 0.5833333333333334, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 789}, "ledger_by_hop": {"1": {"facts_total": 59, "facts_retained": 11, "retention_rate": 0.1864406779661017, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.14285714285714285, "amount": 0.0, "status": 0.4166666666666667, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 760}, "2": {"facts_total": 59, "facts_retained": 13, "retention_rate": 0.22033898305084745, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.14285714285714285, "amount": 0.0, "status": 0.5833333333333334, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 789}, "3": {"facts_total": 59, "facts_retained": 13, "retention_rate": 0.22033898305084745, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.14285714285714285, "amount": 0.0, "status": 0.5833333333333334, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 789}, "4": {"facts_total": 59, "facts_retained": 13, "retention_rate": 0.22033898305084745, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.14285714285714285, "amount": 0.0, "status": 0.5833333333333334, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 789}}}, "reference_key_id": "keys/synthetic/syn_03012_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "b7a5ccb481b692e93f148c0da2974299bed9aa510b975eeaf745e05e1a120812", "output_hash": "2bd125847727be7e7c0a656ee4c81f3bc6d0fef11f536f50ce7bc83688907075"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "b0936a4558b604e90c409e1d", "attempt_id": "94a932abe72d", "run_group_id": "E4_core:A5|repairable|qwen3b:h4", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:40:34.335827Z", "task_family": "synthetic", "document_id": "syn_03012_2000", "document_hash": "03e441e68d54e1af036f8ff905dfd6cbbccead62636279c1a8e179352ab08945", "question_id": "q13_unanswerable", "split": "test", "cluster_id": "syn_03012_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 4, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1927, "query_tokens": 453, "message_tokens": 358, "message_utf8_bytes": 853, "cache_tensor_bytes": null, "serialized_payload_bytes": 853, "side_information_bytes": 0, "fetched_bytes": 1383, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 69939.11387491971, "hops": {"1": 23063.031291007064, "2": 15415.211042040028, "3": 15675.964874913916, "4": 15784.906666958705}, "receiver_prefill_ms": 308.9884159853682, "ttft_ms": 30.74441698845476, "decode_ms": 33.088625059463084}, "total_wall_ms": 70312.54479195923, "ttft_ms": 30.74441698845476, "input_usage": {"lookups": 3}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 59, "facts_retained": 13, "retention_rate": 0.22033898305084745, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.14285714285714285, "amount": 0.0, "status": 0.5833333333333334, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 789}, "ledger_by_hop": {"1": {"facts_total": 59, "facts_retained": 11, "retention_rate": 0.1864406779661017, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.14285714285714285, "amount": 0.0, "status": 0.4166666666666667, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 760}, "2": {"facts_total": 59, "facts_retained": 13, "retention_rate": 0.22033898305084745, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.14285714285714285, "amount": 0.0, "status": 0.5833333333333334, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 789}, "3": {"facts_total": 59, "facts_retained": 13, "retention_rate": 0.22033898305084745, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.14285714285714285, "amount": 0.0, "status": 0.5833333333333334, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 789}, "4": {"facts_total": 59, "facts_retained": 13, "retention_rate": 0.22033898305084745, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.14285714285714285, "amount": 0.0, "status": 0.5833333333333334, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 789}}}, "reference_key_id": "keys/synthetic/syn_03012_2000.json", "task_scores": {"correct": 1.0, "abstained": 1.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "b7a5ccb481b692e93f148c0da2974299bed9aa510b975eeaf745e05e1a120812", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "9b76ee9dbfbc313432c806b9", "attempt_id": "866dacaf2243", "run_group_id": "E4_core:A5|repairable|qwen3b:h4", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:40:34.727496Z", "task_family": "synthetic", "document_id": "syn_03012_2000", "document_hash": "03e441e68d54e1af036f8ff905dfd6cbbccead62636279c1a8e179352ab08945", "question_id": "q14_unanswerable", "split": "test", "cluster_id": "syn_03012_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 4, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1927, "query_tokens": 453, "message_tokens": 358, "message_utf8_bytes": 853, "cache_tensor_bytes": null, "serialized_payload_bytes": 853, "side_information_bytes": 0, "fetched_bytes": 1383, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 69939.11387491971, "hops": {"1": 23063.031291007064, "2": 15415.211042040028, "3": 15675.964874913916, "4": 15784.906666958705}, "receiver_prefill_ms": 280.2883750991896, "ttft_ms": 48.483791993930936, "decode_ms": 69.22816694714129}, "total_wall_ms": 70337.76529191528, "ttft_ms": 48.483791993930936, "input_usage": {"lookups": 3}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 59, "facts_retained": 13, "retention_rate": 0.22033898305084745, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.14285714285714285, "amount": 0.0, "status": 0.5833333333333334, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 789}, "ledger_by_hop": {"1": {"facts_total": 59, "facts_retained": 11, "retention_rate": 0.1864406779661017, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.14285714285714285, "amount": 0.0, "status": 0.4166666666666667, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 760}, "2": {"facts_total": 59, "facts_retained": 13, "retention_rate": 0.22033898305084745, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.14285714285714285, "amount": 0.0, "status": 0.5833333333333334, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 789}, "3": {"facts_total": 59, "facts_retained": 13, "retention_rate": 0.22033898305084745, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.14285714285714285, "amount": 0.0, "status": 0.5833333333333334, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 789}, "4": {"facts_total": 59, "facts_retained": 13, "retention_rate": 0.22033898305084745, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.14285714285714285, "amount": 0.0, "status": 0.5833333333333334, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 789}}}, "reference_key_id": "keys/synthetic/syn_03012_2000.json", "task_scores": {"correct": 1.0, "abstained": 1.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "b7a5ccb481b692e93f148c0da2974299bed9aa510b975eeaf745e05e1a120812", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "6d48a123653d4664d4a69e81", "attempt_id": "4bdbef63da3a", "run_group_id": "E4_core:A5|repairable|qwen3b:h8", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:41:37.633383Z", "task_family": "synthetic", "document_id": "syn_03012_2000", "document_hash": "03e441e68d54e1af036f8ff905dfd6cbbccead62636279c1a8e179352ab08945", "question_id": "q00_single_fact", "split": "test", "cluster_id": "syn_03012_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 8, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1927, "query_tokens": 440, "message_tokens": 358, "message_utf8_bytes": 853, "cache_tensor_bytes": null, "serialized_payload_bytes": 853, "side_information_bytes": 0, "fetched_bytes": 3227, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 132424.26941602025, "hops": {"1": 23063.031291007064, "2": 15415.211042040028, "3": 15675.964874913916, "4": 15784.906666958705, "5": 15655.812000040896, "6": 15553.547790972516, "7": 15781.18929208722, "8": 15494.606457999907}, "receiver_prefill_ms": 287.1391660301015, "ttft_ms": 44.78520806878805, "decode_ms": 601.7802089918405}, "total_wall_ms": 133358.57566597406, "ttft_ms": 44.78520806878805, "input_usage": {"lookups": 7}, "output_usage": {"output_tokens": 10}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "XX-71M-6213", "evidence_artifact": {"ledger": {"facts_total": 59, "facts_retained": 13, "retention_rate": 0.22033898305084745, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.14285714285714285, "amount": 0.0, "status": 0.5833333333333334, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 789}, "ledger_by_hop": {"1": {"facts_total": 59, "facts_retained": 11, "retention_rate": 0.1864406779661017, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.14285714285714285, "amount": 0.0, "status": 0.4166666666666667, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 760}, "2": {"facts_total": 59, "facts_retained": 13, "retention_rate": 0.22033898305084745, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.14285714285714285, "amount": 0.0, "status": 0.5833333333333334, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 789}, "3": {"facts_total": 59, "facts_retained": 13, "retention_rate": 0.22033898305084745, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.14285714285714285, "amount": 0.0, "status": 0.5833333333333334, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 789}, "4": {"facts_total": 59, "facts_retained": 13, "retention_rate": 0.22033898305084745, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.14285714285714285, "amount": 0.0, "status": 0.5833333333333334, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 789}, "5": {"facts_total": 59, "facts_retained": 13, "retention_rate": 0.22033898305084745, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.14285714285714285, "amount": 0.0, "status": 0.5833333333333334, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 789}, "6": {"facts_total": 59, "facts_retained": 13, "retention_rate": 0.22033898305084745, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.14285714285714285, "amount": 0.0, "status": 0.5833333333333334, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 789}, "7": {"facts_total": 59, "facts_retained": 13, "retention_rate": 0.22033898305084745, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.14285714285714285, "amount": 0.0, "status": 0.5833333333333334, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 789}, "8": {"facts_total": 59, "facts_retained": 13, "retention_rate": 0.22033898305084745, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.14285714285714285, "amount": 0.0, "status": 0.5833333333333334, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 789}}}, "reference_key_id": "keys/synthetic/syn_03012_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "b7a5ccb481b692e93f148c0da2974299bed9aa510b975eeaf745e05e1a120812", "output_hash": "66ef6c5da682b38795e9029723bbd71cdde2c4dec70cf1fae6efe62ef243f10a"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "ef68f1020950a84cbcf99041", "attempt_id": "eb7898e66fdc", "run_group_id": "E4_core:A5|repairable|qwen3b:h8", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:41:38.585308Z", "task_family": "synthetic", "document_id": "syn_03012_2000", "document_hash": "03e441e68d54e1af036f8ff905dfd6cbbccead62636279c1a8e179352ab08945", "question_id": "q01_single_fact", "split": "test", "cluster_id": "syn_03012_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 8, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1927, "query_tokens": 446, "message_tokens": 358, "message_utf8_bytes": 853, "cache_tensor_bytes": null, "serialized_payload_bytes": 853, "side_information_bytes": 0, "fetched_bytes": 3227, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 132424.26941602025, "hops": {"1": 23063.031291007064, "2": 15415.211042040028, "3": 15675.964874913916, "4": 15784.906666958705, "5": 15655.812000040896, "6": 15553.547790972516, "7": 15781.18929208722, "8": 15494.606457999907}, "receiver_prefill_ms": 293.34024991840124, "ttft_ms": 38.42075006105006, "decode_ms": 24.126208038069308}, "total_wall_ms": 132780.6567910593, "ttft_ms": 38.42075006105006, "input_usage": {"lookups": 7}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 59, "facts_retained": 13, "retention_rate": 0.22033898305084745, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.14285714285714285, "amount": 0.0, "status": 0.5833333333333334, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 789}, "ledger_by_hop": {"1": {"facts_total": 59, "facts_retained": 11, "retention_rate": 0.1864406779661017, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.14285714285714285, "amount": 0.0, "status": 0.4166666666666667, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 760}, "2": {"facts_total": 59, "facts_retained": 13, "retention_rate": 0.22033898305084745, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.14285714285714285, "amount": 0.0, "status": 0.5833333333333334, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 789}, "3": {"facts_total": 59, "facts_retained": 13, "retention_rate": 0.22033898305084745, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.14285714285714285, "amount": 0.0, "status": 0.5833333333333334, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 789}, "4": {"facts_total": 59, "facts_retained": 13, "retention_rate": 0.22033898305084745, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.14285714285714285, "amount": 0.0, "status": 0.5833333333333334, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 789}, "5": {"facts_total": 59, "facts_retained": 13, "retention_rate": 0.22033898305084745, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.14285714285714285, "amount": 0.0, "status": 0.5833333333333334, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 789}, "6": {"facts_total": 59, "facts_retained": 13, "retention_rate": 0.22033898305084745, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.14285714285714285, "amount": 0.0, "status": 0.5833333333333334, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 789}, "7": {"facts_total": 59, "facts_retained": 13, "retention_rate": 0.22033898305084745, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.14285714285714285, "amount": 0.0, "status": 0.5833333333333334, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 789}, "8": {"facts_total": 59, "facts_retained": 13, "retention_rate": 0.22033898305084745, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.14285714285714285, "amount": 0.0, "status": 0.5833333333333334, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 789}}}, "reference_key_id": "keys/synthetic/syn_03012_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "b7a5ccb481b692e93f148c0da2974299bed9aa510b975eeaf745e05e1a120812", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "97c3407585198ebef9a53b2f", "attempt_id": "0a2769ab6d51", "run_group_id": "E4_core:A5|repairable|qwen3b:h8", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:41:38.958797Z", "task_family": "synthetic", "document_id": "syn_03012_2000", "document_hash": "03e441e68d54e1af036f8ff905dfd6cbbccead62636279c1a8e179352ab08945", "question_id": "q02_exact_copy", "split": "test", "cluster_id": "syn_03012_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 8, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1927, "query_tokens": 445, "message_tokens": 358, "message_utf8_bytes": 853, "cache_tensor_bytes": null, "serialized_payload_bytes": 853, "side_information_bytes": 0, "fetched_bytes": 3227, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 132424.26941602025, "hops": {"1": 23063.031291007064, "2": 15415.211042040028, "3": 15675.964874913916, "4": 15784.906666958705, "5": 15655.812000040896, "6": 15553.547790972516, "7": 15781.18929208722, "8": 15494.606457999907}, "receiver_prefill_ms": 295.3373750206083, "ttft_ms": 25.945208966732025, "decode_ms": 35.74574994854629}, "total_wall_ms": 132781.84345702175, "ttft_ms": 25.945208966732025, "input_usage": {"lookups": 7}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 59, "facts_retained": 13, "retention_rate": 0.22033898305084745, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.14285714285714285, "amount": 0.0, "status": 0.5833333333333334, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 789}, "ledger_by_hop": {"1": {"facts_total": 59, "facts_retained": 11, "retention_rate": 0.1864406779661017, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.14285714285714285, "amount": 0.0, "status": 0.4166666666666667, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 760}, "2": {"facts_total": 59, "facts_retained": 13, "retention_rate": 0.22033898305084745, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.14285714285714285, "amount": 0.0, "status": 0.5833333333333334, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 789}, "3": {"facts_total": 59, "facts_retained": 13, "retention_rate": 0.22033898305084745, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.14285714285714285, "amount": 0.0, "status": 0.5833333333333334, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 789}, "4": {"facts_total": 59, "facts_retained": 13, "retention_rate": 0.22033898305084745, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.14285714285714285, "amount": 0.0, "status": 0.5833333333333334, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 789}, "5": {"facts_total": 59, "facts_retained": 13, "retention_rate": 0.22033898305084745, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.14285714285714285, "amount": 0.0, "status": 0.5833333333333334, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 789}, "6": {"facts_total": 59, "facts_retained": 13, "retention_rate": 0.22033898305084745, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.14285714285714285, "amount": 0.0, "status": 0.5833333333333334, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 789}, "7": {"facts_total": 59, "facts_retained": 13, "retention_rate": 0.22033898305084745, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.14285714285714285, "amount": 0.0, "status": 0.5833333333333334, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 789}, "8": {"facts_total": 59, "facts_retained": 13, "retention_rate": 0.22033898305084745, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.14285714285714285, "amount": 0.0, "status": 0.5833333333333334, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 789}}}, "reference_key_id": "keys/synthetic/syn_03012_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "b7a5ccb481b692e93f148c0da2974299bed9aa510b975eeaf745e05e1a120812", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "107a55fe17d536aa13fd1ea6", "attempt_id": "62e0fc45f5da", "run_group_id": "E4_core:A5|repairable|qwen3b:h8", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:41:39.333415Z", "task_family": "synthetic", "document_id": "syn_03012_2000", "document_hash": "03e441e68d54e1af036f8ff905dfd6cbbccead62636279c1a8e179352ab08945", "question_id": "q03_exact_copy", "split": "test", "cluster_id": "syn_03012_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 8, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1927, "query_tokens": 441, "message_tokens": 358, "message_utf8_bytes": 853, "cache_tensor_bytes": null, "serialized_payload_bytes": 853, "side_information_bytes": 0, "fetched_bytes": 3227, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 132424.26941602025, "hops": {"1": 23063.031291007064, "2": 15415.211042040028, "3": 15675.964874913916, "4": 15784.906666958705, "5": 15655.812000040896, "6": 15553.547790972516, "7": 15781.18929208722, "8": 15494.606457999907}, "receiver_prefill_ms": 266.26262499485165, "ttft_ms": 42.021790985018015, "decode_ms": 53.80624998360872}, "total_wall_ms": 132787.2100820532, "ttft_ms": 42.021790985018015, "input_usage": {"lookups": 7}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 59, "facts_retained": 13, "retention_rate": 0.22033898305084745, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.14285714285714285, "amount": 0.0, "status": 0.5833333333333334, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 789}, "ledger_by_hop": {"1": {"facts_total": 59, "facts_retained": 11, "retention_rate": 0.1864406779661017, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.14285714285714285, "amount": 0.0, "status": 0.4166666666666667, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 760}, "2": {"facts_total": 59, "facts_retained": 13, "retention_rate": 0.22033898305084745, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.14285714285714285, "amount": 0.0, "status": 0.5833333333333334, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 789}, "3": {"facts_total": 59, "facts_retained": 13, "retention_rate": 0.22033898305084745, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.14285714285714285, "amount": 0.0, "status": 0.5833333333333334, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 789}, "4": {"facts_total": 59, "facts_retained": 13, "retention_rate": 0.22033898305084745, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.14285714285714285, "amount": 0.0, "status": 0.5833333333333334, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 789}, "5": {"facts_total": 59, "facts_retained": 13, "retention_rate": 0.22033898305084745, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.14285714285714285, "amount": 0.0, "status": 0.5833333333333334, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 789}, "6": {"facts_total": 59, "facts_retained": 13, "retention_rate": 0.22033898305084745, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.14285714285714285, "amount": 0.0, "status": 0.5833333333333334, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 789}, "7": {"facts_total": 59, "facts_retained": 13, "retention_rate": 0.22033898305084745, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.14285714285714285, "amount": 0.0, "status": 0.5833333333333334, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 789}, "8": {"facts_total": 59, "facts_retained": 13, "retention_rate": 0.22033898305084745, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.14285714285714285, "amount": 0.0, "status": 0.5833333333333334, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 789}}}, "reference_key_id": "keys/synthetic/syn_03012_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "b7a5ccb481b692e93f148c0da2974299bed9aa510b975eeaf745e05e1a120812", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "8f235896c440d556f4092ac4", "attempt_id": "508e92f1dd0b", "run_group_id": "E4_core:A5|repairable|qwen3b:h8", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:41:39.713430Z", "task_family": "synthetic", "document_id": "syn_03012_2000", "document_hash": "03e441e68d54e1af036f8ff905dfd6cbbccead62636279c1a8e179352ab08945", "question_id": "q04_multi_hop", "split": "test", "cluster_id": "syn_03012_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 8, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1927, "query_tokens": 446, "message_tokens": 358, "message_utf8_bytes": 853, "cache_tensor_bytes": null, "serialized_payload_bytes": 853, "side_information_bytes": 0, "fetched_bytes": 3227, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 132424.26941602025, "hops": {"1": 23063.031291007064, "2": 15415.211042040028, "3": 15675.964874913916, "4": 15784.906666958705, "5": 15655.812000040896, "6": 15553.547790972516, "7": 15781.18929208722, "8": 15494.606457999907}, "receiver_prefill_ms": 394.52899992465973, "ttft_ms": 74.5572920423001, "decode_ms": 39.357832982204854}, "total_wall_ms": 132933.31145704724, "ttft_ms": 74.5572920423001, "input_usage": {"lookups": 7}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 59, "facts_retained": 13, "retention_rate": 0.22033898305084745, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.14285714285714285, "amount": 0.0, "status": 0.5833333333333334, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 789}, "ledger_by_hop": {"1": {"facts_total": 59, "facts_retained": 11, "retention_rate": 0.1864406779661017, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.14285714285714285, "amount": 0.0, "status": 0.4166666666666667, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 760}, "2": {"facts_total": 59, "facts_retained": 13, "retention_rate": 0.22033898305084745, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.14285714285714285, "amount": 0.0, "status": 0.5833333333333334, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 789}, "3": {"facts_total": 59, "facts_retained": 13, "retention_rate": 0.22033898305084745, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.14285714285714285, "amount": 0.0, "status": 0.5833333333333334, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 789}, "4": {"facts_total": 59, "facts_retained": 13, "retention_rate": 0.22033898305084745, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.14285714285714285, "amount": 0.0, "status": 0.5833333333333334, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 789}, "5": {"facts_total": 59, "facts_retained": 13, "retention_rate": 0.22033898305084745, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.14285714285714285, "amount": 0.0, "status": 0.5833333333333334, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 789}, "6": {"facts_total": 59, "facts_retained": 13, "retention_rate": 0.22033898305084745, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.14285714285714285, "amount": 0.0, "status": 0.5833333333333334, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 789}, "7": {"facts_total": 59, "facts_retained": 13, "retention_rate": 0.22033898305084745, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.14285714285714285, "amount": 0.0, "status": 0.5833333333333334, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 789}, "8": {"facts_total": 59, "facts_retained": 13, "retention_rate": 0.22033898305084745, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.14285714285714285, "amount": 0.0, "status": 0.5833333333333334, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 789}}}, "reference_key_id": "keys/synthetic/syn_03012_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "b7a5ccb481b692e93f148c0da2974299bed9aa510b975eeaf745e05e1a120812", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "b4db8b65b2b68f9b2010f0d6", "attempt_id": "ad5727ba3dac", "run_group_id": "E4_core:A5|repairable|qwen3b:h8", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:41:40.241391Z", "task_family": "synthetic", "document_id": "syn_03012_2000", "document_hash": "03e441e68d54e1af036f8ff905dfd6cbbccead62636279c1a8e179352ab08945", "question_id": "q05_multi_hop", "split": "test", "cluster_id": "syn_03012_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 8, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1927, "query_tokens": 449, "message_tokens": 358, "message_utf8_bytes": 853, "cache_tensor_bytes": null, "serialized_payload_bytes": 853, "side_information_bytes": 0, "fetched_bytes": 3227, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 132424.26941602025, "hops": {"1": 23063.031291007064, "2": 15415.211042040028, "3": 15675.964874913916, "4": 15784.906666958705, "5": 15655.812000040896, "6": 15553.547790972516, "7": 15781.18929208722, "8": 15494.606457999907}, "receiver_prefill_ms": 290.91395798604935, "ttft_ms": 33.70791592169553, "decode_ms": 32.89554105140269}, "total_wall_ms": 132782.40299900062, "ttft_ms": 33.70791592169553, "input_usage": {"lookups": 7}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 59, "facts_retained": 13, "retention_rate": 0.22033898305084745, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.14285714285714285, "amount": 0.0, "status": 0.5833333333333334, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 789}, "ledger_by_hop": {"1": {"facts_total": 59, "facts_retained": 11, "retention_rate": 0.1864406779661017, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.14285714285714285, "amount": 0.0, "status": 0.4166666666666667, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 760}, "2": {"facts_total": 59, "facts_retained": 13, "retention_rate": 0.22033898305084745, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.14285714285714285, "amount": 0.0, "status": 0.5833333333333334, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 789}, "3": {"facts_total": 59, "facts_retained": 13, "retention_rate": 0.22033898305084745, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.14285714285714285, "amount": 0.0, "status": 0.5833333333333334, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 789}, "4": {"facts_total": 59, "facts_retained": 13, "retention_rate": 0.22033898305084745, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.14285714285714285, "amount": 0.0, "status": 0.5833333333333334, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 789}, "5": {"facts_total": 59, "facts_retained": 13, "retention_rate": 0.22033898305084745, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.14285714285714285, "amount": 0.0, "status": 0.5833333333333334, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 789}, "6": {"facts_total": 59, "facts_retained": 13, "retention_rate": 0.22033898305084745, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.14285714285714285, "amount": 0.0, "status": 0.5833333333333334, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 789}, "7": {"facts_total": 59, "facts_retained": 13, "retention_rate": 0.22033898305084745, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.14285714285714285, "amount": 0.0, "status": 0.5833333333333334, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 789}, "8": {"facts_total": 59, "facts_retained": 13, "retention_rate": 0.22033898305084745, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.14285714285714285, "amount": 0.0, "status": 0.5833333333333334, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 789}}}, "reference_key_id": "keys/synthetic/syn_03012_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "b7a5ccb481b692e93f148c0da2974299bed9aa510b975eeaf745e05e1a120812", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "30c1929165102060335b9a93", "attempt_id": "90fd569a11b6", "run_group_id": "E4_core:A5|repairable|qwen3b:h8", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:41:40.616917Z", "task_family": "synthetic", "document_id": "syn_03012_2000", "document_hash": "03e441e68d54e1af036f8ff905dfd6cbbccead62636279c1a8e179352ab08945", "question_id": "q06_negation", "split": "test", "cluster_id": "syn_03012_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 8, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1927, "query_tokens": 441, "message_tokens": 358, "message_utf8_bytes": 853, "cache_tensor_bytes": null, "serialized_payload_bytes": 853, "side_information_bytes": 0, "fetched_bytes": 3227, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 132424.26941602025, "hops": {"1": 23063.031291007064, "2": 15415.211042040028, "3": 15675.964874913916, "4": 15784.906666958705, "5": 15655.812000040896, "6": 15553.547790972516, "7": 15781.18929208722, "8": 15494.606457999907}, "receiver_prefill_ms": 299.19904191046953, "ttft_ms": 32.860292005352676, "decode_ms": 29.277749941684306}, "total_wall_ms": 132786.17316600867, "ttft_ms": 32.860292005352676, "input_usage": {"lookups": 7}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 59, "facts_retained": 13, "retention_rate": 0.22033898305084745, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.14285714285714285, "amount": 0.0, "status": 0.5833333333333334, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 789}, "ledger_by_hop": {"1": {"facts_total": 59, "facts_retained": 11, "retention_rate": 0.1864406779661017, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.14285714285714285, "amount": 0.0, "status": 0.4166666666666667, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 760}, "2": {"facts_total": 59, "facts_retained": 13, "retention_rate": 0.22033898305084745, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.14285714285714285, "amount": 0.0, "status": 0.5833333333333334, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 789}, "3": {"facts_total": 59, "facts_retained": 13, "retention_rate": 0.22033898305084745, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.14285714285714285, "amount": 0.0, "status": 0.5833333333333334, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 789}, "4": {"facts_total": 59, "facts_retained": 13, "retention_rate": 0.22033898305084745, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.14285714285714285, "amount": 0.0, "status": 0.5833333333333334, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 789}, "5": {"facts_total": 59, "facts_retained": 13, "retention_rate": 0.22033898305084745, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.14285714285714285, "amount": 0.0, "status": 0.5833333333333334, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 789}, "6": {"facts_total": 59, "facts_retained": 13, "retention_rate": 0.22033898305084745, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.14285714285714285, "amount": 0.0, "status": 0.5833333333333334, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 789}, "7": {"facts_total": 59, "facts_retained": 13, "retention_rate": 0.22033898305084745, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.14285714285714285, "amount": 0.0, "status": 0.5833333333333334, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 789}, "8": {"facts_total": 59, "facts_retained": 13, "retention_rate": 0.22033898305084745, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.14285714285714285, "amount": 0.0, "status": 0.5833333333333334, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 789}}}, "reference_key_id": "keys/synthetic/syn_03012_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "b7a5ccb481b692e93f148c0da2974299bed9aa510b975eeaf745e05e1a120812", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "c94853d75a55c9c8ae52736d", "attempt_id": "df662d3c06b6", "run_group_id": "E4_core:A5|repairable|qwen3b:h8", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:41:40.995989Z", "task_family": "synthetic", "document_id": "syn_03012_2000", "document_hash": "03e441e68d54e1af036f8ff905dfd6cbbccead62636279c1a8e179352ab08945", "question_id": "q07_negation", "split": "test", "cluster_id": "syn_03012_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 8, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1927, "query_tokens": 441, "message_tokens": 358, "message_utf8_bytes": 853, "cache_tensor_bytes": null, "serialized_payload_bytes": 853, "side_information_bytes": 0, "fetched_bytes": 3227, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 132424.26941602025, "hops": {"1": 23063.031291007064, "2": 15415.211042040028, "3": 15675.964874913916, "4": 15784.906666958705, "5": 15655.812000040896, "6": 15553.547790972516, "7": 15781.18929208722, "8": 15494.606457999907}, "receiver_prefill_ms": 262.76108308229595, "ttft_ms": 58.21008293423802, "decode_ms": 40.951124974526465}, "total_wall_ms": 132786.79770801682, "ttft_ms": 58.21008293423802, "input_usage": {"lookups": 7}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 59, "facts_retained": 13, "retention_rate": 0.22033898305084745, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.14285714285714285, "amount": 0.0, "status": 0.5833333333333334, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 789}, "ledger_by_hop": {"1": {"facts_total": 59, "facts_retained": 11, "retention_rate": 0.1864406779661017, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.14285714285714285, "amount": 0.0, "status": 0.4166666666666667, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 760}, "2": {"facts_total": 59, "facts_retained": 13, "retention_rate": 0.22033898305084745, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.14285714285714285, "amount": 0.0, "status": 0.5833333333333334, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 789}, "3": {"facts_total": 59, "facts_retained": 13, "retention_rate": 0.22033898305084745, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.14285714285714285, "amount": 0.0, "status": 0.5833333333333334, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 789}, "4": {"facts_total": 59, "facts_retained": 13, "retention_rate": 0.22033898305084745, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.14285714285714285, "amount": 0.0, "status": 0.5833333333333334, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 789}, "5": {"facts_total": 59, "facts_retained": 13, "retention_rate": 0.22033898305084745, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.14285714285714285, "amount": 0.0, "status": 0.5833333333333334, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 789}, "6": {"facts_total": 59, "facts_retained": 13, "retention_rate": 0.22033898305084745, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.14285714285714285, "amount": 0.0, "status": 0.5833333333333334, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 789}, "7": {"facts_total": 59, "facts_retained": 13, "retention_rate": 0.22033898305084745, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.14285714285714285, "amount": 0.0, "status": 0.5833333333333334, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 789}, "8": {"facts_total": 59, "facts_retained": 13, "retention_rate": 0.22033898305084745, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.14285714285714285, "amount": 0.0, "status": 0.5833333333333334, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 789}}}, "reference_key_id": "keys/synthetic/syn_03012_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "b7a5ccb481b692e93f148c0da2974299bed9aa510b975eeaf745e05e1a120812", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "f28d7f47d31eadf514dc185e", "attempt_id": "628c8cda48fa", "run_group_id": "E4_core:A5|repairable|qwen3b:h8", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:41:41.377440Z", "task_family": "synthetic", "document_id": "syn_03012_2000", "document_hash": "03e441e68d54e1af036f8ff905dfd6cbbccead62636279c1a8e179352ab08945", "question_id": "q08_revision", "split": "test", "cluster_id": "syn_03012_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 8, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1927, "query_tokens": 456, "message_tokens": 358, "message_utf8_bytes": 853, "cache_tensor_bytes": null, "serialized_payload_bytes": 853, "side_information_bytes": 0, "fetched_bytes": 3227, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 132424.26941602025, "hops": {"1": 23063.031291007064, "2": 15415.211042040028, "3": 15675.964874913916, "4": 15784.906666958705, "5": 15655.812000040896, "6": 15553.547790972516, "7": 15781.18929208722, "8": 15494.606457999907}, "receiver_prefill_ms": 466.1336250137538, "ttft_ms": 39.556915988214314, "decode_ms": 21.923957974649966}, "total_wall_ms": 132952.51862495206, "ttft_ms": 39.556915988214314, "input_usage": {"lookups": 7}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 59, "facts_retained": 13, "retention_rate": 0.22033898305084745, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.14285714285714285, "amount": 0.0, "status": 0.5833333333333334, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 789}, "ledger_by_hop": {"1": {"facts_total": 59, "facts_retained": 11, "retention_rate": 0.1864406779661017, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.14285714285714285, "amount": 0.0, "status": 0.4166666666666667, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 760}, "2": {"facts_total": 59, "facts_retained": 13, "retention_rate": 0.22033898305084745, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.14285714285714285, "amount": 0.0, "status": 0.5833333333333334, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 789}, "3": {"facts_total": 59, "facts_retained": 13, "retention_rate": 0.22033898305084745, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.14285714285714285, "amount": 0.0, "status": 0.5833333333333334, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 789}, "4": {"facts_total": 59, "facts_retained": 13, "retention_rate": 0.22033898305084745, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.14285714285714285, "amount": 0.0, "status": 0.5833333333333334, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 789}, "5": {"facts_total": 59, "facts_retained": 13, "retention_rate": 0.22033898305084745, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.14285714285714285, "amount": 0.0, "status": 0.5833333333333334, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 789}, "6": {"facts_total": 59, "facts_retained": 13, "retention_rate": 0.22033898305084745, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.14285714285714285, "amount": 0.0, "status": 0.5833333333333334, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 789}, "7": {"facts_total": 59, "facts_retained": 13, "retention_rate": 0.22033898305084745, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.14285714285714285, "amount": 0.0, "status": 0.5833333333333334, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 789}, "8": {"facts_total": 59, "facts_retained": 13, "retention_rate": 0.22033898305084745, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.14285714285714285, "amount": 0.0, "status": 0.5833333333333334, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 789}}}, "reference_key_id": "keys/synthetic/syn_03012_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "b7a5ccb481b692e93f148c0da2974299bed9aa510b975eeaf745e05e1a120812", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "8e3884e9e4f0172fb23406c7", "attempt_id": "09e0c9581b36", "run_group_id": "E4_core:A5|repairable|qwen3b:h8", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:41:41.923644Z", "task_family": "synthetic", "document_id": "syn_03012_2000", "document_hash": "03e441e68d54e1af036f8ff905dfd6cbbccead62636279c1a8e179352ab08945", "question_id": "q09_revision", "split": "test", "cluster_id": "syn_03012_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 8, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1927, "query_tokens": 449, "message_tokens": 358, "message_utf8_bytes": 853, "cache_tensor_bytes": null, "serialized_payload_bytes": 853, "side_information_bytes": 0, "fetched_bytes": 3227, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 132424.26941602025, "hops": {"1": 23063.031291007064, "2": 15415.211042040028, "3": 15675.964874913916, "4": 15784.906666958705, "5": 15655.812000040896, "6": 15553.547790972516, "7": 15781.18929208722, "8": 15494.606457999907}, "receiver_prefill_ms": 290.48704192973673, "ttft_ms": 42.571541969664395, "decode_ms": 31.036833068355918}, "total_wall_ms": 132788.94779109396, "ttft_ms": 42.571541969664395, "input_usage": {"lookups": 7}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 59, "facts_retained": 13, "retention_rate": 0.22033898305084745, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.14285714285714285, "amount": 0.0, "status": 0.5833333333333334, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 789}, "ledger_by_hop": {"1": {"facts_total": 59, "facts_retained": 11, "retention_rate": 0.1864406779661017, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.14285714285714285, "amount": 0.0, "status": 0.4166666666666667, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 760}, "2": {"facts_total": 59, "facts_retained": 13, "retention_rate": 0.22033898305084745, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.14285714285714285, "amount": 0.0, "status": 0.5833333333333334, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 789}, "3": {"facts_total": 59, "facts_retained": 13, "retention_rate": 0.22033898305084745, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.14285714285714285, "amount": 0.0, "status": 0.5833333333333334, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 789}, "4": {"facts_total": 59, "facts_retained": 13, "retention_rate": 0.22033898305084745, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.14285714285714285, "amount": 0.0, "status": 0.5833333333333334, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 789}, "5": {"facts_total": 59, "facts_retained": 13, "retention_rate": 0.22033898305084745, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.14285714285714285, "amount": 0.0, "status": 0.5833333333333334, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 789}, "6": {"facts_total": 59, "facts_retained": 13, "retention_rate": 0.22033898305084745, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.14285714285714285, "amount": 0.0, "status": 0.5833333333333334, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 789}, "7": {"facts_total": 59, "facts_retained": 13, "retention_rate": 0.22033898305084745, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.14285714285714285, "amount": 0.0, "status": 0.5833333333333334, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 789}, "8": {"facts_total": 59, "facts_retained": 13, "retention_rate": 0.22033898305084745, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.14285714285714285, "amount": 0.0, "status": 0.5833333333333334, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 789}}}, "reference_key_id": "keys/synthetic/syn_03012_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "b7a5ccb481b692e93f148c0da2974299bed9aa510b975eeaf745e05e1a120812", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "7cc77459398b20a62efe7fbc", "attempt_id": "090b7a5e9291", "run_group_id": "E4_core:A5|repairable|qwen3b:h8", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:41:42.305517Z", "task_family": "synthetic", "document_id": "syn_03012_2000", "document_hash": "03e441e68d54e1af036f8ff905dfd6cbbccead62636279c1a8e179352ab08945", "question_id": "q10_temporal", "split": "test", "cluster_id": "syn_03012_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 8, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1927, "query_tokens": 459, "message_tokens": 358, "message_utf8_bytes": 853, "cache_tensor_bytes": null, "serialized_payload_bytes": 853, "side_information_bytes": 0, "fetched_bytes": 3227, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 132424.26941602025, "hops": {"1": 23063.031291007064, "2": 15415.211042040028, "3": 15675.964874913916, "4": 15784.906666958705, "5": 15655.812000040896, "6": 15553.547790972516, "7": 15781.18929208722, "8": 15494.606457999907}, "receiver_prefill_ms": 319.3024169886485, "ttft_ms": 37.29887492954731, "decode_ms": 24.56979197449982}, "total_wall_ms": 132806.0268330155, "ttft_ms": 37.29887492954731, "input_usage": {"lookups": 7}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 59, "facts_retained": 13, "retention_rate": 0.22033898305084745, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.14285714285714285, "amount": 0.0, "status": 0.5833333333333334, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 789}, "ledger_by_hop": {"1": {"facts_total": 59, "facts_retained": 11, "retention_rate": 0.1864406779661017, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.14285714285714285, "amount": 0.0, "status": 0.4166666666666667, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 760}, "2": {"facts_total": 59, "facts_retained": 13, "retention_rate": 0.22033898305084745, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.14285714285714285, "amount": 0.0, "status": 0.5833333333333334, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 789}, "3": {"facts_total": 59, "facts_retained": 13, "retention_rate": 0.22033898305084745, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.14285714285714285, "amount": 0.0, "status": 0.5833333333333334, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 789}, "4": {"facts_total": 59, "facts_retained": 13, "retention_rate": 0.22033898305084745, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.14285714285714285, "amount": 0.0, "status": 0.5833333333333334, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 789}, "5": {"facts_total": 59, "facts_retained": 13, "retention_rate": 0.22033898305084745, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.14285714285714285, "amount": 0.0, "status": 0.5833333333333334, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 789}, "6": {"facts_total": 59, "facts_retained": 13, "retention_rate": 0.22033898305084745, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.14285714285714285, "amount": 0.0, "status": 0.5833333333333334, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 789}, "7": {"facts_total": 59, "facts_retained": 13, "retention_rate": 0.22033898305084745, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.14285714285714285, "amount": 0.0, "status": 0.5833333333333334, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 789}, "8": {"facts_total": 59, "facts_retained": 13, "retention_rate": 0.22033898305084745, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.14285714285714285, "amount": 0.0, "status": 0.5833333333333334, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 789}}}, "reference_key_id": "keys/synthetic/syn_03012_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "b7a5ccb481b692e93f148c0da2974299bed9aa510b975eeaf745e05e1a120812", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "e091b3eebd7f003cb5b506e5", "attempt_id": "a5e3a653879e", "run_group_id": "E4_core:A5|repairable|qwen3b:h8", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:41:42.703926Z", "task_family": "synthetic", "document_id": "syn_03012_2000", "document_hash": "03e441e68d54e1af036f8ff905dfd6cbbccead62636279c1a8e179352ab08945", "question_id": "q11_aggregation", "split": "test", "cluster_id": "syn_03012_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 8, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1927, "query_tokens": 451, "message_tokens": 358, "message_utf8_bytes": 853, "cache_tensor_bytes": null, "serialized_payload_bytes": 853, "side_information_bytes": 0, "fetched_bytes": 3227, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 132424.26941602025, "hops": {"1": 23063.031291007064, "2": 15415.211042040028, "3": 15675.964874913916, "4": 15784.906666958705, "5": 15655.812000040896, "6": 15553.547790972516, "7": 15781.18929208722, "8": 15494.606457999907}, "receiver_prefill_ms": 281.86716698110104, "ttft_ms": 55.59595802333206, "decode_ms": 39.36366701964289}, "total_wall_ms": 132801.8119580811, "ttft_ms": 55.59595802333206, "input_usage": {"lookups": 7}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 59, "facts_retained": 13, "retention_rate": 0.22033898305084745, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.14285714285714285, "amount": 0.0, "status": 0.5833333333333334, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 789}, "ledger_by_hop": {"1": {"facts_total": 59, "facts_retained": 11, "retention_rate": 0.1864406779661017, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.14285714285714285, "amount": 0.0, "status": 0.4166666666666667, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 760}, "2": {"facts_total": 59, "facts_retained": 13, "retention_rate": 0.22033898305084745, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.14285714285714285, "amount": 0.0, "status": 0.5833333333333334, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 789}, "3": {"facts_total": 59, "facts_retained": 13, "retention_rate": 0.22033898305084745, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.14285714285714285, "amount": 0.0, "status": 0.5833333333333334, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 789}, "4": {"facts_total": 59, "facts_retained": 13, "retention_rate": 0.22033898305084745, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.14285714285714285, "amount": 0.0, "status": 0.5833333333333334, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 789}, "5": {"facts_total": 59, "facts_retained": 13, "retention_rate": 0.22033898305084745, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.14285714285714285, "amount": 0.0, "status": 0.5833333333333334, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 789}, "6": {"facts_total": 59, "facts_retained": 13, "retention_rate": 0.22033898305084745, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.14285714285714285, "amount": 0.0, "status": 0.5833333333333334, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 789}, "7": {"facts_total": 59, "facts_retained": 13, "retention_rate": 0.22033898305084745, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.14285714285714285, "amount": 0.0, "status": 0.5833333333333334, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 789}, "8": {"facts_total": 59, "facts_retained": 13, "retention_rate": 0.22033898305084745, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.14285714285714285, "amount": 0.0, "status": 0.5833333333333334, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 789}}}, "reference_key_id": "keys/synthetic/syn_03012_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "b7a5ccb481b692e93f148c0da2974299bed9aa510b975eeaf745e05e1a120812", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "c589f509ea081ceb0f06f16d", "attempt_id": "5f76e768c6e8", "run_group_id": "E4_core:A5|repairable|qwen3b:h8", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:41:43.099316Z", "task_family": "synthetic", "document_id": "syn_03012_2000", "document_hash": "03e441e68d54e1af036f8ff905dfd6cbbccead62636279c1a8e179352ab08945", "question_id": "q12_unit_conversion", "split": "test", "cluster_id": "syn_03012_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 8, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1927, "query_tokens": 448, "message_tokens": 358, "message_utf8_bytes": 853, "cache_tensor_bytes": null, "serialized_payload_bytes": 853, "side_information_bytes": 0, "fetched_bytes": 3227, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 132424.26941602025, "hops": {"1": 23063.031291007064, "2": 15415.211042040028, "3": 15675.964874913916, "4": 15784.906666958705, "5": 15655.812000040896, "6": 15553.547790972516, "7": 15781.18929208722, "8": 15494.606457999907}, "receiver_prefill_ms": 354.27033400628716, "ttft_ms": 94.73862498998642, "decode_ms": 176.33083299733698}, "total_wall_ms": 133050.08929106407, "ttft_ms": 94.73862498998642, "input_usage": {"lookups": 7}, "output_usage": {"output_tokens": 5}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "44620", "evidence_artifact": {"ledger": {"facts_total": 59, "facts_retained": 13, "retention_rate": 0.22033898305084745, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.14285714285714285, "amount": 0.0, "status": 0.5833333333333334, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 789}, "ledger_by_hop": {"1": {"facts_total": 59, "facts_retained": 11, "retention_rate": 0.1864406779661017, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.14285714285714285, "amount": 0.0, "status": 0.4166666666666667, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 760}, "2": {"facts_total": 59, "facts_retained": 13, "retention_rate": 0.22033898305084745, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.14285714285714285, "amount": 0.0, "status": 0.5833333333333334, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 789}, "3": {"facts_total": 59, "facts_retained": 13, "retention_rate": 0.22033898305084745, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.14285714285714285, "amount": 0.0, "status": 0.5833333333333334, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 789}, "4": {"facts_total": 59, "facts_retained": 13, "retention_rate": 0.22033898305084745, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.14285714285714285, "amount": 0.0, "status": 0.5833333333333334, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 789}, "5": {"facts_total": 59, "facts_retained": 13, "retention_rate": 0.22033898305084745, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.14285714285714285, "amount": 0.0, "status": 0.5833333333333334, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 789}, "6": {"facts_total": 59, "facts_retained": 13, "retention_rate": 0.22033898305084745, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.14285714285714285, "amount": 0.0, "status": 0.5833333333333334, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 789}, "7": {"facts_total": 59, "facts_retained": 13, "retention_rate": 0.22033898305084745, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.14285714285714285, "amount": 0.0, "status": 0.5833333333333334, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 789}, "8": {"facts_total": 59, "facts_retained": 13, "retention_rate": 0.22033898305084745, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.14285714285714285, "amount": 0.0, "status": 0.5833333333333334, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 789}}}, "reference_key_id": "keys/synthetic/syn_03012_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "b7a5ccb481b692e93f148c0da2974299bed9aa510b975eeaf745e05e1a120812", "output_hash": "2bd125847727be7e7c0a656ee4c81f3bc6d0fef11f536f50ce7bc83688907075"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "cdbd0c5a796c26de0a88d83c", "attempt_id": "0e27b44ad914", "run_group_id": "E4_core:A5|repairable|qwen3b:h8", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:41:43.742278Z", "task_family": "synthetic", "document_id": "syn_03012_2000", "document_hash": "03e441e68d54e1af036f8ff905dfd6cbbccead62636279c1a8e179352ab08945", "question_id": "q13_unanswerable", "split": "test", "cluster_id": "syn_03012_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 8, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1927, "query_tokens": 453, "message_tokens": 358, "message_utf8_bytes": 853, "cache_tensor_bytes": null, "serialized_payload_bytes": 853, "side_information_bytes": 0, "fetched_bytes": 3227, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 132424.26941602025, "hops": {"1": 23063.031291007064, "2": 15415.211042040028, "3": 15675.964874913916, "4": 15784.906666958705, "5": 15655.812000040896, "6": 15553.547790972516, "7": 15781.18929208722, "8": 15494.606457999907}, "receiver_prefill_ms": 318.34170897491276, "ttft_ms": 38.694582995958626, "decode_ms": 27.317291009239852}, "total_wall_ms": 132809.10349998157, "ttft_ms": 38.694582995958626, "input_usage": {"lookups": 7}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 59, "facts_retained": 13, "retention_rate": 0.22033898305084745, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.14285714285714285, "amount": 0.0, "status": 0.5833333333333334, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 789}, "ledger_by_hop": {"1": {"facts_total": 59, "facts_retained": 11, "retention_rate": 0.1864406779661017, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.14285714285714285, "amount": 0.0, "status": 0.4166666666666667, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 760}, "2": {"facts_total": 59, "facts_retained": 13, "retention_rate": 0.22033898305084745, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.14285714285714285, "amount": 0.0, "status": 0.5833333333333334, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 789}, "3": {"facts_total": 59, "facts_retained": 13, "retention_rate": 0.22033898305084745, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.14285714285714285, "amount": 0.0, "status": 0.5833333333333334, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 789}, "4": {"facts_total": 59, "facts_retained": 13, "retention_rate": 0.22033898305084745, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.14285714285714285, "amount": 0.0, "status": 0.5833333333333334, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 789}, "5": {"facts_total": 59, "facts_retained": 13, "retention_rate": 0.22033898305084745, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.14285714285714285, "amount": 0.0, "status": 0.5833333333333334, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 789}, "6": {"facts_total": 59, "facts_retained": 13, "retention_rate": 0.22033898305084745, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.14285714285714285, "amount": 0.0, "status": 0.5833333333333334, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 789}, "7": {"facts_total": 59, "facts_retained": 13, "retention_rate": 0.22033898305084745, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.14285714285714285, "amount": 0.0, "status": 0.5833333333333334, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 789}, "8": {"facts_total": 59, "facts_retained": 13, "retention_rate": 0.22033898305084745, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.14285714285714285, "amount": 0.0, "status": 0.5833333333333334, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 789}}}, "reference_key_id": "keys/synthetic/syn_03012_2000.json", "task_scores": {"correct": 1.0, "abstained": 1.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "b7a5ccb481b692e93f148c0da2974299bed9aa510b975eeaf745e05e1a120812", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "c2cd0d3c5524c1e8c7593320", "attempt_id": "54abc8be84e4", "run_group_id": "E4_core:A5|repairable|qwen3b:h8", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:41:44.144547Z", "task_family": "synthetic", "document_id": "syn_03012_2000", "document_hash": "03e441e68d54e1af036f8ff905dfd6cbbccead62636279c1a8e179352ab08945", "question_id": "q14_unanswerable", "split": "test", "cluster_id": "syn_03012_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 8, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1927, "query_tokens": 453, "message_tokens": 358, "message_utf8_bytes": 853, "cache_tensor_bytes": null, "serialized_payload_bytes": 853, "side_information_bytes": 0, "fetched_bytes": 3227, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 132424.26941602025, "hops": {"1": 23063.031291007064, "2": 15415.211042040028, "3": 15675.964874913916, "4": 15784.906666958705, "5": 15655.812000040896, "6": 15553.547790972516, "7": 15781.18929208722, "8": 15494.606457999907}, "receiver_prefill_ms": 319.02870803605765, "ttft_ms": 22.507333080284297, "decode_ms": 18.972165999002755}, "total_wall_ms": 132785.57620709762, "ttft_ms": 22.507333080284297, "input_usage": {"lookups": 7}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 59, "facts_retained": 13, "retention_rate": 0.22033898305084745, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.14285714285714285, "amount": 0.0, "status": 0.5833333333333334, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 789}, "ledger_by_hop": {"1": {"facts_total": 59, "facts_retained": 11, "retention_rate": 0.1864406779661017, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.14285714285714285, "amount": 0.0, "status": 0.4166666666666667, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 760}, "2": {"facts_total": 59, "facts_retained": 13, "retention_rate": 0.22033898305084745, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.14285714285714285, "amount": 0.0, "status": 0.5833333333333334, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 789}, "3": {"facts_total": 59, "facts_retained": 13, "retention_rate": 0.22033898305084745, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.14285714285714285, "amount": 0.0, "status": 0.5833333333333334, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 789}, "4": {"facts_total": 59, "facts_retained": 13, "retention_rate": 0.22033898305084745, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.14285714285714285, "amount": 0.0, "status": 0.5833333333333334, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 789}, "5": {"facts_total": 59, "facts_retained": 13, "retention_rate": 0.22033898305084745, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.14285714285714285, "amount": 0.0, "status": 0.5833333333333334, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 789}, "6": {"facts_total": 59, "facts_retained": 13, "retention_rate": 0.22033898305084745, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.14285714285714285, "amount": 0.0, "status": 0.5833333333333334, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 789}, "7": {"facts_total": 59, "facts_retained": 13, "retention_rate": 0.22033898305084745, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.14285714285714285, "amount": 0.0, "status": 0.5833333333333334, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 789}, "8": {"facts_total": 59, "facts_retained": 13, "retention_rate": 0.22033898305084745, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.14285714285714285, "amount": 0.0, "status": 0.5833333333333334, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 789}}}, "reference_key_id": "keys/synthetic/syn_03012_2000.json", "task_scores": {"correct": 1.0, "abstained": 1.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "b7a5ccb481b692e93f148c0da2974299bed9aa510b975eeaf745e05e1a120812", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "003a5138433193ca3efbade5", "attempt_id": "d851cedf10f4", "run_group_id": "E4_core:A4|closed|qwen3b-llama3b:h1", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:42:07.685599Z", "task_family": "synthetic", "document_id": "syn_03012_2000", "document_hash": "03e441e68d54e1af036f8ff905dfd6cbbccead62636279c1a8e179352ab08945", "question_id": "q00_single_fact", "split": "test", "cluster_id": "syn_03012_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 1, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1927, "query_tokens": 594, "message_tokens": 512, "message_utf8_bytes": 1213, "cache_tensor_bytes": null, "serialized_payload_bytes": 1213, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 23159.84900004696, "hops": {"1": 23159.84900004696}, "receiver_prefill_ms": 388.51400010753423, "ttft_ms": 24.0307500353083, "decode_ms": 669.2864999640733}, "total_wall_ms": 24242.366875056177, "ttft_ms": 24.0307500353083, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 10}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "XX-71M-6213", "evidence_artifact": {"ledger": {"facts_total": 59, "facts_retained": 30, "retention_rate": 0.5084745762711864, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.5714285714285714, "amount": 0.6, "status": 0.5833333333333334, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1213}, "ledger_by_hop": {"1": {"facts_total": 59, "facts_retained": 30, "retention_rate": 0.5084745762711864, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.5714285714285714, "amount": 0.6, "status": 0.5833333333333334, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1213}}}, "reference_key_id": "keys/synthetic/syn_03012_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "7c9b236b590faff2030e02587d972cf755ac2a88072e709174edbf405498973d", "output_hash": "66ef6c5da682b38795e9029723bbd71cdde2c4dec70cf1fae6efe62ef243f10a"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "d0bbab9c4b81f98a032114f9", "attempt_id": "7fa24467e984", "run_group_id": "E4_core:A4|closed|qwen3b-llama3b:h1", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:42:08.785759Z", "task_family": "synthetic", "document_id": "syn_03012_2000", "document_hash": "03e441e68d54e1af036f8ff905dfd6cbbccead62636279c1a8e179352ab08945", "question_id": "q01_single_fact", "split": "test", "cluster_id": "syn_03012_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 1, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1927, "query_tokens": 600, "message_tokens": 512, "message_utf8_bytes": 1213, "cache_tensor_bytes": null, "serialized_payload_bytes": 1213, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 23159.84900004696, "hops": {"1": 23159.84900004696}, "receiver_prefill_ms": 407.6588749885559, "ttft_ms": 29.316750005818903, "decode_ms": 35.22083396092057}, "total_wall_ms": 23632.687999983318, "ttft_ms": 29.316750005818903, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 59, "facts_retained": 30, "retention_rate": 0.5084745762711864, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.5714285714285714, "amount": 0.6, "status": 0.5833333333333334, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1213}, "ledger_by_hop": {"1": {"facts_total": 59, "facts_retained": 30, "retention_rate": 0.5084745762711864, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.5714285714285714, "amount": 0.6, "status": 0.5833333333333334, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1213}}}, "reference_key_id": "keys/synthetic/syn_03012_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "7c9b236b590faff2030e02587d972cf755ac2a88072e709174edbf405498973d", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "0905f899e43bcf547e950286", "attempt_id": "ba8142ae3839", "run_group_id": "E4_core:A4|closed|qwen3b-llama3b:h1", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:42:09.276377Z", "task_family": "synthetic", "document_id": "syn_03012_2000", "document_hash": "03e441e68d54e1af036f8ff905dfd6cbbccead62636279c1a8e179352ab08945", "question_id": "q02_exact_copy", "split": "test", "cluster_id": "syn_03012_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 1, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1927, "query_tokens": 599, "message_tokens": 512, "message_utf8_bytes": 1213, "cache_tensor_bytes": null, "serialized_payload_bytes": 1213, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 23159.84900004696, "hops": {"1": 23159.84900004696}, "receiver_prefill_ms": 381.67591695673764, "ttft_ms": 14.831083011813462, "decode_ms": 319.6052910061553}, "total_wall_ms": 23876.692458055913, "ttft_ms": 14.831083011813462, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 7}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "13.4343", "evidence_artifact": {"ledger": {"facts_total": 59, "facts_retained": 30, "retention_rate": 0.5084745762711864, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.5714285714285714, "amount": 0.6, "status": 0.5833333333333334, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1213}, "ledger_by_hop": {"1": {"facts_total": 59, "facts_retained": 30, "retention_rate": 0.5084745762711864, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.5714285714285714, "amount": 0.6, "status": 0.5833333333333334, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1213}}}, "reference_key_id": "keys/synthetic/syn_03012_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "7c9b236b590faff2030e02587d972cf755ac2a88072e709174edbf405498973d", "output_hash": "599bc22e373b1cf7b4679208eb9e515ad0e8dcb88dde0c7f579637ec7d3f3014"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "be352e61d90b0fe4107c8406", "attempt_id": "b15047a17676", "run_group_id": "E4_core:A4|closed|qwen3b-llama3b:h1", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:42:10.010752Z", "task_family": "synthetic", "document_id": "syn_03012_2000", "document_hash": "03e441e68d54e1af036f8ff905dfd6cbbccead62636279c1a8e179352ab08945", "question_id": "q03_exact_copy", "split": "test", "cluster_id": "syn_03012_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 1, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1927, "query_tokens": 595, "message_tokens": 512, "message_utf8_bytes": 1213, "cache_tensor_bytes": null, "serialized_payload_bytes": 1213, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 23159.84900004696, "hops": {"1": 23159.84900004696}, "receiver_prefill_ms": 536.4594999700785, "ttft_ms": 25.4059589933604, "decode_ms": 40.32520903274417}, "total_wall_ms": 23762.739458004944, "ttft_ms": 25.4059589933604, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 59, "facts_retained": 30, "retention_rate": 0.5084745762711864, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.5714285714285714, "amount": 0.6, "status": 0.5833333333333334, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1213}, "ledger_by_hop": {"1": {"facts_total": 59, "facts_retained": 30, "retention_rate": 0.5084745762711864, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.5714285714285714, "amount": 0.6, "status": 0.5833333333333334, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1213}}}, "reference_key_id": "keys/synthetic/syn_03012_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "7c9b236b590faff2030e02587d972cf755ac2a88072e709174edbf405498973d", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "1d741123a43874cafc51f882", "attempt_id": "1d7d42a865a2", "run_group_id": "E4_core:A4|closed|qwen3b-llama3b:h1", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:42:10.632066Z", "task_family": "synthetic", "document_id": "syn_03012_2000", "document_hash": "03e441e68d54e1af036f8ff905dfd6cbbccead62636279c1a8e179352ab08945", "question_id": "q04_multi_hop", "split": "test", "cluster_id": "syn_03012_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 1, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1927, "query_tokens": 600, "message_tokens": 512, "message_utf8_bytes": 1213, "cache_tensor_bytes": null, "serialized_payload_bytes": 1213, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 23159.84900004696, "hops": {"1": 23159.84900004696}, "receiver_prefill_ms": 409.02954095508903, "ttft_ms": 22.46687503065914, "decode_ms": 208.91720801591873}, "total_wall_ms": 23800.878374953754, "ttft_ms": 22.46687503065914, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 7}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "8631552", "evidence_artifact": {"ledger": {"facts_total": 59, "facts_retained": 30, "retention_rate": 0.5084745762711864, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.5714285714285714, "amount": 0.6, "status": 0.5833333333333334, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1213}, "ledger_by_hop": {"1": {"facts_total": 59, "facts_retained": 30, "retention_rate": 0.5084745762711864, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.5714285714285714, "amount": 0.6, "status": 0.5833333333333334, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1213}}}, "reference_key_id": "keys/synthetic/syn_03012_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "7c9b236b590faff2030e02587d972cf755ac2a88072e709174edbf405498973d", "output_hash": "147d02c0587a77b3b53f0424c8d3d1454aaf2639ce9fdaf69af20db33363e849"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "3662227647860cd7f66fc88b", "attempt_id": "aacef136d21b", "run_group_id": "E4_core:A4|closed|qwen3b-llama3b:h1", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:42:11.291203Z", "task_family": "synthetic", "document_id": "syn_03012_2000", "document_hash": "03e441e68d54e1af036f8ff905dfd6cbbccead62636279c1a8e179352ab08945", "question_id": "q05_multi_hop", "split": "test", "cluster_id": "syn_03012_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 1, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1927, "query_tokens": 603, "message_tokens": 512, "message_utf8_bytes": 1213, "cache_tensor_bytes": null, "serialized_payload_bytes": 1213, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 23159.84900004696, "hops": {"1": 23159.84900004696}, "receiver_prefill_ms": 408.5809999378398, "ttft_ms": 57.84645804669708, "decode_ms": 519.1950830630958}, "total_wall_ms": 24146.277916966937, "ttft_ms": 57.84645804669708, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 10}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "DG-40E-2777", "evidence_artifact": {"ledger": {"facts_total": 59, "facts_retained": 30, "retention_rate": 0.5084745762711864, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.5714285714285714, "amount": 0.6, "status": 0.5833333333333334, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1213}, "ledger_by_hop": {"1": {"facts_total": 59, "facts_retained": 30, "retention_rate": 0.5084745762711864, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.5714285714285714, "amount": 0.6, "status": 0.5833333333333334, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1213}}}, "reference_key_id": "keys/synthetic/syn_03012_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "7c9b236b590faff2030e02587d972cf755ac2a88072e709174edbf405498973d", "output_hash": "16a7e3f4a7cc4a9b66f86c2cdcfad3b35daf1da199d9406e0627c8af2d8d73f3"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "10652708dc6c2a2fb7fd1af3", "attempt_id": "429d136148fc", "run_group_id": "E4_core:A4|closed|qwen3b-llama3b:h1", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:42:12.297560Z", "task_family": "synthetic", "document_id": "syn_03012_2000", "document_hash": "03e441e68d54e1af036f8ff905dfd6cbbccead62636279c1a8e179352ab08945", "question_id": "q06_negation", "split": "test", "cluster_id": "syn_03012_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 1, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1927, "query_tokens": 595, "message_tokens": 512, "message_utf8_bytes": 1213, "cache_tensor_bytes": null, "serialized_payload_bytes": 1213, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 23159.84900004696, "hops": {"1": 23159.84900004696}, "receiver_prefill_ms": 376.9257079111412, "ttft_ms": 36.4165409700945, "decode_ms": 25.74516599997878}, "total_wall_ms": 23599.555833032355, "ttft_ms": 36.4165409700945, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "no", "evidence_artifact": {"ledger": {"facts_total": 59, "facts_retained": 30, "retention_rate": 0.5084745762711864, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.5714285714285714, "amount": 0.6, "status": 0.5833333333333334, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1213}, "ledger_by_hop": {"1": {"facts_total": 59, "facts_retained": 30, "retention_rate": 0.5084745762711864, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.5714285714285714, "amount": 0.6, "status": 0.5833333333333334, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1213}}}, "reference_key_id": "keys/synthetic/syn_03012_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "7c9b236b590faff2030e02587d972cf755ac2a88072e709174edbf405498973d", "output_hash": "9390298f3fb0c5b160498935d79cb139aef28e1c47358b4bbba61862b9c26e59"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "57c804d3c4c3835bb7024bb7", "attempt_id": "71ff34d13234", "run_group_id": "E4_core:A4|closed|qwen3b-llama3b:h1", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:42:12.755974Z", "task_family": "synthetic", "document_id": "syn_03012_2000", "document_hash": "03e441e68d54e1af036f8ff905dfd6cbbccead62636279c1a8e179352ab08945", "question_id": "q07_negation", "split": "test", "cluster_id": "syn_03012_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 1, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1927, "query_tokens": 595, "message_tokens": 512, "message_utf8_bytes": 1213, "cache_tensor_bytes": null, "serialized_payload_bytes": 1213, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 23159.84900004696, "hops": {"1": 23159.84900004696}, "receiver_prefill_ms": 362.46533307712525, "ttft_ms": 54.52212505042553, "decode_ms": 57.55554209463298}, "total_wall_ms": 23635.061166947708, "ttft_ms": 54.52212505042553, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "no", "evidence_artifact": {"ledger": {"facts_total": 59, "facts_retained": 30, "retention_rate": 0.5084745762711864, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.5714285714285714, "amount": 0.6, "status": 0.5833333333333334, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1213}, "ledger_by_hop": {"1": {"facts_total": 59, "facts_retained": 30, "retention_rate": 0.5084745762711864, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.5714285714285714, "amount": 0.6, "status": 0.5833333333333334, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1213}}}, "reference_key_id": "keys/synthetic/syn_03012_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "7c9b236b590faff2030e02587d972cf755ac2a88072e709174edbf405498973d", "output_hash": "9390298f3fb0c5b160498935d79cb139aef28e1c47358b4bbba61862b9c26e59"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "131da6419e9cefc5bdc24f7e", "attempt_id": "144a5025dbeb", "run_group_id": "E4_core:A4|closed|qwen3b-llama3b:h1", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:42:13.248192Z", "task_family": "synthetic", "document_id": "syn_03012_2000", "document_hash": "03e441e68d54e1af036f8ff905dfd6cbbccead62636279c1a8e179352ab08945", "question_id": "q08_revision", "split": "test", "cluster_id": "syn_03012_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 1, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1927, "query_tokens": 610, "message_tokens": 512, "message_utf8_bytes": 1213, "cache_tensor_bytes": null, "serialized_payload_bytes": 1213, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 23159.84900004696, "hops": {"1": 23159.84900004696}, "receiver_prefill_ms": 554.774709045887, "ttft_ms": 37.2332080733031, "decode_ms": 253.40516702271998}, "total_wall_ms": 24006.027959054336, "ttft_ms": 37.2332080733031, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 7}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "6602245", "evidence_artifact": {"ledger": {"facts_total": 59, "facts_retained": 30, "retention_rate": 0.5084745762711864, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.5714285714285714, "amount": 0.6, "status": 0.5833333333333334, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1213}, "ledger_by_hop": {"1": {"facts_total": 59, "facts_retained": 30, "retention_rate": 0.5084745762711864, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.5714285714285714, "amount": 0.6, "status": 0.5833333333333334, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1213}}}, "reference_key_id": "keys/synthetic/syn_03012_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "7c9b236b590faff2030e02587d972cf755ac2a88072e709174edbf405498973d", "output_hash": "d117fa8197988f8d1e4fabc6c4cdaab9516018e54553b57c63b875b3de82b16f"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "07223ba137013fef56634011", "attempt_id": "2dcecb2ae685", "run_group_id": "E4_core:A4|closed|qwen3b-llama3b:h1", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:42:14.112682Z", "task_family": "synthetic", "document_id": "syn_03012_2000", "document_hash": "03e441e68d54e1af036f8ff905dfd6cbbccead62636279c1a8e179352ab08945", "question_id": "q09_revision", "split": "test", "cluster_id": "syn_03012_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 1, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1927, "query_tokens": 603, "message_tokens": 512, "message_utf8_bytes": 1213, "cache_tensor_bytes": null, "serialized_payload_bytes": 1213, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 23159.84900004696, "hops": {"1": 23159.84900004696}, "receiver_prefill_ms": 398.01162492949516, "ttft_ms": 19.92979203350842, "decode_ms": 294.4863330340013}, "total_wall_ms": 23872.97204101924, "ttft_ms": 19.92979203350842, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 7}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "5899340", "evidence_artifact": {"ledger": {"facts_total": 59, "facts_retained": 30, "retention_rate": 0.5084745762711864, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.5714285714285714, "amount": 0.6, "status": 0.5833333333333334, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1213}, "ledger_by_hop": {"1": {"facts_total": 59, "facts_retained": 30, "retention_rate": 0.5084745762711864, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.5714285714285714, "amount": 0.6, "status": 0.5833333333333334, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1213}}}, "reference_key_id": "keys/synthetic/syn_03012_2000.json", "task_scores": {"correct": 0.0, "abstained": 0.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "7c9b236b590faff2030e02587d972cf755ac2a88072e709174edbf405498973d", "output_hash": "c920323a9a02b1afb8619cf11bdde427f7cd1ea3289be69bd62932985895d717"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "6e6d56a95f4057263deb00f9", "attempt_id": "8f87f0b0bf67", "run_group_id": "E4_core:A4|closed|qwen3b-llama3b:h1", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:42:14.843351Z", "task_family": "synthetic", "document_id": "syn_03012_2000", "document_hash": "03e441e68d54e1af036f8ff905dfd6cbbccead62636279c1a8e179352ab08945", "question_id": "q10_temporal", "split": "test", "cluster_id": "syn_03012_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 1, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1927, "query_tokens": 613, "message_tokens": 512, "message_utf8_bytes": 1213, "cache_tensor_bytes": null, "serialized_payload_bytes": 1213, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 23159.84900004696, "hops": {"1": 23159.84900004696}, "receiver_prefill_ms": 499.88216697238386, "ttft_ms": 40.42316600680351, "decode_ms": 185.29937497805804}, "total_wall_ms": 23886.528708040714, "ttft_ms": 40.42316600680351, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 6}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "Felaldfel-9", "evidence_artifact": {"ledger": {"facts_total": 59, "facts_retained": 30, "retention_rate": 0.5084745762711864, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.5714285714285714, "amount": 0.6, "status": 0.5833333333333334, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1213}, "ledger_by_hop": {"1": {"facts_total": 59, "facts_retained": 30, "retention_rate": 0.5084745762711864, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.5714285714285714, "amount": 0.6, "status": 0.5833333333333334, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1213}}}, "reference_key_id": "keys/synthetic/syn_03012_2000.json", "task_scores": {"correct": 0.0, "abstained": 0.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "7c9b236b590faff2030e02587d972cf755ac2a88072e709174edbf405498973d", "output_hash": "09ab1578c99909d3c22a0c8042977ad718c576e7b4cce4e50abc683b5f7d9a64"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "f33592b09ab5513070a58a2f", "attempt_id": "2238707c0ccb", "run_group_id": "E4_core:A4|closed|qwen3b-llama3b:h1", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:42:15.588530Z", "task_family": "synthetic", "document_id": "syn_03012_2000", "document_hash": "03e441e68d54e1af036f8ff905dfd6cbbccead62636279c1a8e179352ab08945", "question_id": "q11_aggregation", "split": "test", "cluster_id": "syn_03012_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 1, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1927, "query_tokens": 605, "message_tokens": 512, "message_utf8_bytes": 1213, "cache_tensor_bytes": null, "serialized_payload_bytes": 1213, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 23159.84900004696, "hops": {"1": 23159.84900004696}, "receiver_prefill_ms": 403.670541010797, "ttft_ms": 23.14833295531571, "decode_ms": 229.46499998215586}, "total_wall_ms": 23817.155542084947, "ttft_ms": 23.14833295531571, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 8}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "19704477", "evidence_artifact": {"ledger": {"facts_total": 59, "facts_retained": 30, "retention_rate": 0.5084745762711864, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.5714285714285714, "amount": 0.6, "status": 0.5833333333333334, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1213}, "ledger_by_hop": {"1": {"facts_total": 59, "facts_retained": 30, "retention_rate": 0.5084745762711864, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.5714285714285714, "amount": 0.6, "status": 0.5833333333333334, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1213}}}, "reference_key_id": "keys/synthetic/syn_03012_2000.json", "task_scores": {"correct": 0.0, "abstained": 0.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "7c9b236b590faff2030e02587d972cf755ac2a88072e709174edbf405498973d", "output_hash": "ba53bcc03c2a9b1df7909eec6e2d5de7384570e56fb592d48b95504bfd97157d"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "8288367d799069cef8934056", "attempt_id": "99e6d52b2062", "run_group_id": "E4_core:A4|closed|qwen3b-llama3b:h1", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:42:16.264889Z", "task_family": "synthetic", "document_id": "syn_03012_2000", "document_hash": "03e441e68d54e1af036f8ff905dfd6cbbccead62636279c1a8e179352ab08945", "question_id": "q12_unit_conversion", "split": "test", "cluster_id": "syn_03012_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 1, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1927, "query_tokens": 602, "message_tokens": 512, "message_utf8_bytes": 1213, "cache_tensor_bytes": null, "serialized_payload_bytes": 1213, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 23159.84900004696, "hops": {"1": 23159.84900004696}, "receiver_prefill_ms": 442.982624983415, "ttft_ms": 101.64654196705669, "decode_ms": 266.25116693321615}, "total_wall_ms": 23971.58725000918, "ttft_ms": 101.64654196705669, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 5}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "44620", "evidence_artifact": {"ledger": {"facts_total": 59, "facts_retained": 30, "retention_rate": 0.5084745762711864, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.5714285714285714, "amount": 0.6, "status": 0.5833333333333334, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1213}, "ledger_by_hop": {"1": {"facts_total": 59, "facts_retained": 30, "retention_rate": 0.5084745762711864, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.5714285714285714, "amount": 0.6, "status": 0.5833333333333334, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1213}}}, "reference_key_id": "keys/synthetic/syn_03012_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "7c9b236b590faff2030e02587d972cf755ac2a88072e709174edbf405498973d", "output_hash": "2bd125847727be7e7c0a656ee4c81f3bc6d0fef11f536f50ce7bc83688907075"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "84999b19ef7dec54dbc9eab4", "attempt_id": "4cb4e036836f", "run_group_id": "E4_core:A4|closed|qwen3b-llama3b:h1", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:42:17.096483Z", "task_family": "synthetic", "document_id": "syn_03012_2000", "document_hash": "03e441e68d54e1af036f8ff905dfd6cbbccead62636279c1a8e179352ab08945", "question_id": "q13_unanswerable", "split": "test", "cluster_id": "syn_03012_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 1, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1927, "query_tokens": 607, "message_tokens": 512, "message_utf8_bytes": 1213, "cache_tensor_bytes": null, "serialized_payload_bytes": 1213, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 23159.84900004696, "hops": {"1": 23159.84900004696}, "receiver_prefill_ms": 398.1172089697793, "ttft_ms": 40.95754201989621, "decode_ms": 31.403666944243014}, "total_wall_ms": 23631.084542139433, "ttft_ms": 40.95754201989621, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 59, "facts_retained": 30, "retention_rate": 0.5084745762711864, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.5714285714285714, "amount": 0.6, "status": 0.5833333333333334, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1213}, "ledger_by_hop": {"1": {"facts_total": 59, "facts_retained": 30, "retention_rate": 0.5084745762711864, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.5714285714285714, "amount": 0.6, "status": 0.5833333333333334, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1213}}}, "reference_key_id": "keys/synthetic/syn_03012_2000.json", "task_scores": {"correct": 1.0, "abstained": 1.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "7c9b236b590faff2030e02587d972cf755ac2a88072e709174edbf405498973d", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "6fcc37c17e7920456d4611f2", "attempt_id": "479e185cca3f", "run_group_id": "E4_core:A4|closed|qwen3b-llama3b:h1", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:42:17.585270Z", "task_family": "synthetic", "document_id": "syn_03012_2000", "document_hash": "03e441e68d54e1af036f8ff905dfd6cbbccead62636279c1a8e179352ab08945", "question_id": "q14_unanswerable", "split": "test", "cluster_id": "syn_03012_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 1, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1927, "query_tokens": 607, "message_tokens": 512, "message_utf8_bytes": 1213, "cache_tensor_bytes": null, "serialized_payload_bytes": 1213, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 23159.84900004696, "hops": {"1": 23159.84900004696}, "receiver_prefill_ms": 373.30795801244676, "ttft_ms": 13.720583985559642, "decode_ms": 11.910874978639185}, "total_wall_ms": 23559.374833130278, "ttft_ms": 13.720583985559642, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 59, "facts_retained": 30, "retention_rate": 0.5084745762711864, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.5714285714285714, "amount": 0.6, "status": 0.5833333333333334, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1213}, "ledger_by_hop": {"1": {"facts_total": 59, "facts_retained": 30, "retention_rate": 0.5084745762711864, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.5714285714285714, "amount": 0.6, "status": 0.5833333333333334, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1213}}}, "reference_key_id": "keys/synthetic/syn_03012_2000.json", "task_scores": {"correct": 1.0, "abstained": 1.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "7c9b236b590faff2030e02587d972cf755ac2a88072e709174edbf405498973d", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "e96045d5efb5e81e0bdf900e", "attempt_id": "6144a687a3e6", "run_group_id": "E4_core:A4|closed|qwen3b-llama3b:h2", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:42:34.646612Z", "task_family": "synthetic", "document_id": "syn_03012_2000", "document_hash": "03e441e68d54e1af036f8ff905dfd6cbbccead62636279c1a8e179352ab08945", "question_id": "q00_single_fact", "split": "test", "cluster_id": "syn_03012_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "tokenizer_hashes": {"receiver": "b9d8872cf3a99da99e1fc4b0f71232a451632dc49ca4c6771549b4efa5c1c763"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 2, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1767, "query_tokens": 523, "message_tokens": 419, "message_utf8_bytes": 1137, "cache_tensor_bytes": null, "serialized_payload_bytes": 1137, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 39802.83679103013, "hops": {"1": 23159.84900004696, "2": 16642.98779098317}, "receiver_prefill_ms": 331.5938750747591, "ttft_ms": 37.70637488923967, "decode_ms": 435.13250001706183}, "total_wall_ms": 40607.94008302037, "ttft_ms": 37.70637488923967, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 7}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "XX-71M-6213", "evidence_artifact": {"ledger": {"facts_total": 59, "facts_retained": 30, "retention_rate": 0.5084745762711864, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.5714285714285714, "amount": 0.6, "status": 0.5833333333333334, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1137}, "ledger_by_hop": {"1": {"facts_total": 59, "facts_retained": 30, "retention_rate": 0.5084745762711864, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.5714285714285714, "amount": 0.6, "status": 0.5833333333333334, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1213}, "2": {"facts_total": 59, "facts_retained": 30, "retention_rate": 0.5084745762711864, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.5714285714285714, "amount": 0.6, "status": 0.5833333333333334, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1137}}}, "reference_key_id": "keys/synthetic/syn_03012_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "d1c14f4f9514cc9284b356cef000b1d2966bb027d2e4b70702d913f0be396a36", "output_hash": "66ef6c5da682b38795e9029723bbd71cdde2c4dec70cf1fae6efe62ef243f10a"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "7a8256c5706c5f60b72c5c61", "attempt_id": "a6a33373b283", "run_group_id": "E4_core:A4|closed|qwen3b-llama3b:h2", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:42:35.472219Z", "task_family": "synthetic", "document_id": "syn_03012_2000", "document_hash": "03e441e68d54e1af036f8ff905dfd6cbbccead62636279c1a8e179352ab08945", "question_id": "q01_single_fact", "split": "test", "cluster_id": "syn_03012_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "tokenizer_hashes": {"receiver": "b9d8872cf3a99da99e1fc4b0f71232a451632dc49ca4c6771549b4efa5c1c763"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 2, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1767, "query_tokens": 528, "message_tokens": 419, "message_utf8_bytes": 1137, "cache_tensor_bytes": null, "serialized_payload_bytes": 1137, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 39802.83679103013, "hops": {"1": 23159.84900004696, "2": 16642.98779098317}, "receiver_prefill_ms": 337.7302079461515, "ttft_ms": 44.24404201563448, "decode_ms": 22.951124934479594}, "total_wall_ms": 40208.40416604187, "ttft_ms": 44.24404201563448, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 59, "facts_retained": 30, "retention_rate": 0.5084745762711864, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.5714285714285714, "amount": 0.6, "status": 0.5833333333333334, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1137}, "ledger_by_hop": {"1": {"facts_total": 59, "facts_retained": 30, "retention_rate": 0.5084745762711864, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.5714285714285714, "amount": 0.6, "status": 0.5833333333333334, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1213}, "2": {"facts_total": 59, "facts_retained": 30, "retention_rate": 0.5084745762711864, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.5714285714285714, "amount": 0.6, "status": 0.5833333333333334, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1137}}}, "reference_key_id": "keys/synthetic/syn_03012_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "d1c14f4f9514cc9284b356cef000b1d2966bb027d2e4b70702d913f0be396a36", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "7179c67c0947dd6dbc73388e", "attempt_id": "64029c9061df", "run_group_id": "E4_core:A4|closed|qwen3b-llama3b:h2", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:42:35.896647Z", "task_family": "synthetic", "document_id": "syn_03012_2000", "document_hash": "03e441e68d54e1af036f8ff905dfd6cbbccead62636279c1a8e179352ab08945", "question_id": "q02_exact_copy", "split": "test", "cluster_id": "syn_03012_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "tokenizer_hashes": {"receiver": "b9d8872cf3a99da99e1fc4b0f71232a451632dc49ca4c6771549b4efa5c1c763"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 2, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1767, "query_tokens": 528, "message_tokens": 419, "message_utf8_bytes": 1137, "cache_tensor_bytes": null, "serialized_payload_bytes": 1137, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 39802.83679103013, "hops": {"1": 23159.84900004696, "2": 16642.98779098317}, "receiver_prefill_ms": 332.90287503041327, "ttft_ms": 37.18466591089964, "decode_ms": 74.69466596376151}, "total_wall_ms": 40248.22683201637, "ttft_ms": 37.18466591089964, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 4}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "13.4343", "evidence_artifact": {"ledger": {"facts_total": 59, "facts_retained": 30, "retention_rate": 0.5084745762711864, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.5714285714285714, "amount": 0.6, "status": 0.5833333333333334, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1137}, "ledger_by_hop": {"1": {"facts_total": 59, "facts_retained": 30, "retention_rate": 0.5084745762711864, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.5714285714285714, "amount": 0.6, "status": 0.5833333333333334, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1213}, "2": {"facts_total": 59, "facts_retained": 30, "retention_rate": 0.5084745762711864, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.5714285714285714, "amount": 0.6, "status": 0.5833333333333334, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1137}}}, "reference_key_id": "keys/synthetic/syn_03012_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "d1c14f4f9514cc9284b356cef000b1d2966bb027d2e4b70702d913f0be396a36", "output_hash": "599bc22e373b1cf7b4679208eb9e515ad0e8dcb88dde0c7f579637ec7d3f3014"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "7d9de1dc77d1e2cff38232fa", "attempt_id": "01d95e720bda", "run_group_id": "E4_core:A4|closed|qwen3b-llama3b:h2", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:42:36.359521Z", "task_family": "synthetic", "document_id": "syn_03012_2000", "document_hash": "03e441e68d54e1af036f8ff905dfd6cbbccead62636279c1a8e179352ab08945", "question_id": "q03_exact_copy", "split": "test", "cluster_id": "syn_03012_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "tokenizer_hashes": {"receiver": "b9d8872cf3a99da99e1fc4b0f71232a451632dc49ca4c6771549b4efa5c1c763"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 2, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1767, "query_tokens": 524, "message_tokens": 419, "message_utf8_bytes": 1137, "cache_tensor_bytes": null, "serialized_payload_bytes": 1137, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 39802.83679103013, "hops": {"1": 23159.84900004696, "2": 16642.98779098317}, "receiver_prefill_ms": 343.6856250045821, "ttft_ms": 46.37791693676263, "decode_ms": 40.60812504030764}, "total_wall_ms": 40234.16679096408, "ttft_ms": 46.37791693676263, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 59, "facts_retained": 30, "retention_rate": 0.5084745762711864, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.5714285714285714, "amount": 0.6, "status": 0.5833333333333334, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1137}, "ledger_by_hop": {"1": {"facts_total": 59, "facts_retained": 30, "retention_rate": 0.5084745762711864, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.5714285714285714, "amount": 0.6, "status": 0.5833333333333334, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1213}, "2": {"facts_total": 59, "facts_retained": 30, "retention_rate": 0.5084745762711864, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.5714285714285714, "amount": 0.6, "status": 0.5833333333333334, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1137}}}, "reference_key_id": "keys/synthetic/syn_03012_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "d1c14f4f9514cc9284b356cef000b1d2966bb027d2e4b70702d913f0be396a36", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "78553283faeddbff2bd654d8", "attempt_id": "e8feef31e961", "run_group_id": "E4_core:A4|closed|qwen3b-llama3b:h2", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:42:36.811190Z", "task_family": "synthetic", "document_id": "syn_03012_2000", "document_hash": "03e441e68d54e1af036f8ff905dfd6cbbccead62636279c1a8e179352ab08945", "question_id": "q04_multi_hop", "split": "test", "cluster_id": "syn_03012_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "tokenizer_hashes": {"receiver": "b9d8872cf3a99da99e1fc4b0f71232a451632dc49ca4c6771549b4efa5c1c763"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 2, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1767, "query_tokens": 529, "message_tokens": 419, "message_utf8_bytes": 1137, "cache_tensor_bytes": null, "serialized_payload_bytes": 1137, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 39802.83679103013, "hops": {"1": 23159.84900004696, "2": 16642.98779098317}, "receiver_prefill_ms": 462.9039589781314, "ttft_ms": 37.75149991270155, "decode_ms": 152.14304195251316}, "total_wall_ms": 40456.3930000877, "ttft_ms": 37.75149991270155, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 5}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "8,631,552", "evidence_artifact": {"ledger": {"facts_total": 59, "facts_retained": 30, "retention_rate": 0.5084745762711864, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.5714285714285714, "amount": 0.6, "status": 0.5833333333333334, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1137}, "ledger_by_hop": {"1": {"facts_total": 59, "facts_retained": 30, "retention_rate": 0.5084745762711864, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.5714285714285714, "amount": 0.6, "status": 0.5833333333333334, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1213}, "2": {"facts_total": 59, "facts_retained": 30, "retention_rate": 0.5084745762711864, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.5714285714285714, "amount": 0.6, "status": 0.5833333333333334, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1137}}}, "reference_key_id": "keys/synthetic/syn_03012_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "d1c14f4f9514cc9284b356cef000b1d2966bb027d2e4b70702d913f0be396a36", "output_hash": "cecdbfc66b98662e79ca5deba218a2b11cf8d9e18b8f3d91c3032acf0c41fc29"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "6e18feea201a87c85107d843", "attempt_id": "3a934dc69a3b", "run_group_id": "E4_core:A4|closed|qwen3b-llama3b:h2", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:42:37.483468Z", "task_family": "synthetic", "document_id": "syn_03012_2000", "document_hash": "03e441e68d54e1af036f8ff905dfd6cbbccead62636279c1a8e179352ab08945", "question_id": "q05_multi_hop", "split": "test", "cluster_id": "syn_03012_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "tokenizer_hashes": {"receiver": "b9d8872cf3a99da99e1fc4b0f71232a451632dc49ca4c6771549b4efa5c1c763"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 2, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1767, "query_tokens": 530, "message_tokens": 419, "message_utf8_bytes": 1137, "cache_tensor_bytes": null, "serialized_payload_bytes": 1137, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 39802.83679103013, "hops": {"1": 23159.84900004696, "2": 16642.98779098317}, "receiver_prefill_ms": 340.63429199159145, "ttft_ms": 37.41395799443126, "decode_ms": 181.21225002687424}, "total_wall_ms": 40362.79620695859, "ttft_ms": 37.41395799443126, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 7}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "DG-40E-2777", "evidence_artifact": {"ledger": {"facts_total": 59, "facts_retained": 30, "retention_rate": 0.5084745762711864, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.5714285714285714, "amount": 0.6, "status": 0.5833333333333334, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1137}, "ledger_by_hop": {"1": {"facts_total": 59, "facts_retained": 30, "retention_rate": 0.5084745762711864, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.5714285714285714, "amount": 0.6, "status": 0.5833333333333334, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1213}, "2": {"facts_total": 59, "facts_retained": 30, "retention_rate": 0.5084745762711864, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.5714285714285714, "amount": 0.6, "status": 0.5833333333333334, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1137}}}, "reference_key_id": "keys/synthetic/syn_03012_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "d1c14f4f9514cc9284b356cef000b1d2966bb027d2e4b70702d913f0be396a36", "output_hash": "16a7e3f4a7cc4a9b66f86c2cdcfad3b35daf1da199d9406e0627c8af2d8d73f3"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "e65610617b49c6c2afd5dcb0", "attempt_id": "49ff61a079e2", "run_group_id": "E4_core:A4|closed|qwen3b-llama3b:h2", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:42:38.061744Z", "task_family": "synthetic", "document_id": "syn_03012_2000", "document_hash": "03e441e68d54e1af036f8ff905dfd6cbbccead62636279c1a8e179352ab08945", "question_id": "q06_negation", "split": "test", "cluster_id": "syn_03012_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "tokenizer_hashes": {"receiver": "b9d8872cf3a99da99e1fc4b0f71232a451632dc49ca4c6771549b4efa5c1c763"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 2, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1767, "query_tokens": 523, "message_tokens": 419, "message_utf8_bytes": 1137, "cache_tensor_bytes": null, "serialized_payload_bytes": 1137, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 39802.83679103013, "hops": {"1": 23159.84900004696, "2": 16642.98779098317}, "receiver_prefill_ms": 355.7649169815704, "ttft_ms": 42.95750008895993, "decode_ms": 161.4076669793576}, "total_wall_ms": 40363.756333012134, "ttft_ms": 42.95750008895993, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 2}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "No.", "evidence_artifact": {"ledger": {"facts_total": 59, "facts_retained": 30, "retention_rate": 0.5084745762711864, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.5714285714285714, "amount": 0.6, "status": 0.5833333333333334, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1137}, "ledger_by_hop": {"1": {"facts_total": 59, "facts_retained": 30, "retention_rate": 0.5084745762711864, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.5714285714285714, "amount": 0.6, "status": 0.5833333333333334, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1213}, "2": {"facts_total": 59, "facts_retained": 30, "retention_rate": 0.5084745762711864, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.5714285714285714, "amount": 0.6, "status": 0.5833333333333334, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1137}}}, "reference_key_id": "keys/synthetic/syn_03012_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "d1c14f4f9514cc9284b356cef000b1d2966bb027d2e4b70702d913f0be396a36", "output_hash": "38178a20b470cfd18299fb1593dd4ba706f1a83321dac051d01086efd8b7a96f"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "9b7cc4fb0766c4eac10636f2", "attempt_id": "e8a36448a6d9", "run_group_id": "E4_core:A4|closed|qwen3b-llama3b:h2", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:42:38.642351Z", "task_family": "synthetic", "document_id": "syn_03012_2000", "document_hash": "03e441e68d54e1af036f8ff905dfd6cbbccead62636279c1a8e179352ab08945", "question_id": "q07_negation", "split": "test", "cluster_id": "syn_03012_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "tokenizer_hashes": {"receiver": "b9d8872cf3a99da99e1fc4b0f71232a451632dc49ca4c6771549b4efa5c1c763"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 2, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1767, "query_tokens": 524, "message_tokens": 419, "message_utf8_bytes": 1137, "cache_tensor_bytes": null, "serialized_payload_bytes": 1137, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 39802.83679103013, "hops": {"1": 23159.84900004696, "2": 16642.98779098317}, "receiver_prefill_ms": 402.96641597524285, "ttft_ms": 36.8208329891786, "decode_ms": 64.17258305009454}, "total_wall_ms": 40307.39412398543, "ttft_ms": 36.8208329891786, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 2}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "No.", "evidence_artifact": {"ledger": {"facts_total": 59, "facts_retained": 30, "retention_rate": 0.5084745762711864, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.5714285714285714, "amount": 0.6, "status": 0.5833333333333334, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1137}, "ledger_by_hop": {"1": {"facts_total": 59, "facts_retained": 30, "retention_rate": 0.5084745762711864, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.5714285714285714, "amount": 0.6, "status": 0.5833333333333334, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1213}, "2": {"facts_total": 59, "facts_retained": 30, "retention_rate": 0.5084745762711864, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.5714285714285714, "amount": 0.6, "status": 0.5833333333333334, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1137}}}, "reference_key_id": "keys/synthetic/syn_03012_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "d1c14f4f9514cc9284b356cef000b1d2966bb027d2e4b70702d913f0be396a36", "output_hash": "38178a20b470cfd18299fb1593dd4ba706f1a83321dac051d01086efd8b7a96f"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "29010b7abb5dbc402d0251c7", "attempt_id": "9322c9514bea", "run_group_id": "E4_core:A4|closed|qwen3b-llama3b:h2", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:42:39.163900Z", "task_family": "synthetic", "document_id": "syn_03012_2000", "document_hash": "03e441e68d54e1af036f8ff905dfd6cbbccead62636279c1a8e179352ab08945", "question_id": "q08_revision", "split": "test", "cluster_id": "syn_03012_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "tokenizer_hashes": {"receiver": "b9d8872cf3a99da99e1fc4b0f71232a451632dc49ca4c6771549b4efa5c1c763"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 2, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1767, "query_tokens": 539, "message_tokens": 419, "message_utf8_bytes": 1137, "cache_tensor_bytes": null, "serialized_payload_bytes": 1137, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 39802.83679103013, "hops": {"1": 23159.84900004696, "2": 16642.98779098317}, "receiver_prefill_ms": 346.8710840679705, "ttft_ms": 25.989833055064082, "decode_ms": 139.49412491638213}, "total_wall_ms": 40315.9478319576, "ttft_ms": 25.989833055064082, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 5}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "6,602,245", "evidence_artifact": {"ledger": {"facts_total": 59, "facts_retained": 30, "retention_rate": 0.5084745762711864, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.5714285714285714, "amount": 0.6, "status": 0.5833333333333334, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1137}, "ledger_by_hop": {"1": {"facts_total": 59, "facts_retained": 30, "retention_rate": 0.5084745762711864, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.5714285714285714, "amount": 0.6, "status": 0.5833333333333334, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1213}, "2": {"facts_total": 59, "facts_retained": 30, "retention_rate": 0.5084745762711864, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.5714285714285714, "amount": 0.6, "status": 0.5833333333333334, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1137}}}, "reference_key_id": "keys/synthetic/syn_03012_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "d1c14f4f9514cc9284b356cef000b1d2966bb027d2e4b70702d913f0be396a36", "output_hash": "f121947078d464ab91a83739861b98c336a10c001704e2df84cc2b041bb8f4de"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "9936b92fcfce0569a290cfd4", "attempt_id": "f4741ce857c6", "run_group_id": "E4_core:A4|closed|qwen3b-llama3b:h2", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:42:39.694017Z", "task_family": "synthetic", "document_id": "syn_03012_2000", "document_hash": "03e441e68d54e1af036f8ff905dfd6cbbccead62636279c1a8e179352ab08945", "question_id": "q09_revision", "split": "test", "cluster_id": "syn_03012_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "tokenizer_hashes": {"receiver": "b9d8872cf3a99da99e1fc4b0f71232a451632dc49ca4c6771549b4efa5c1c763"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 2, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1767, "query_tokens": 531, "message_tokens": 419, "message_utf8_bytes": 1137, "cache_tensor_bytes": null, "serialized_payload_bytes": 1137, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 39802.83679103013, "hops": {"1": 23159.84900004696, "2": 16642.98779098317}, "receiver_prefill_ms": 298.0555410031229, "ttft_ms": 38.69054198730737, "decode_ms": 37.91624994482845}, "total_wall_ms": 40178.644791129045, "ttft_ms": 38.69054198730737, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 59, "facts_retained": 30, "retention_rate": 0.5084745762711864, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.5714285714285714, "amount": 0.6, "status": 0.5833333333333334, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1137}, "ledger_by_hop": {"1": {"facts_total": 59, "facts_retained": 30, "retention_rate": 0.5084745762711864, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.5714285714285714, "amount": 0.6, "status": 0.5833333333333334, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1213}, "2": {"facts_total": 59, "facts_retained": 30, "retention_rate": 0.5084745762711864, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.5714285714285714, "amount": 0.6, "status": 0.5833333333333334, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1137}}}, "reference_key_id": "keys/synthetic/syn_03012_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "d1c14f4f9514cc9284b356cef000b1d2966bb027d2e4b70702d913f0be396a36", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "ba493694d104e9f397aaad4f", "attempt_id": "4602f85f447e", "run_group_id": "E4_core:A4|closed|qwen3b-llama3b:h2", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:42:40.088402Z", "task_family": "synthetic", "document_id": "syn_03012_2000", "document_hash": "03e441e68d54e1af036f8ff905dfd6cbbccead62636279c1a8e179352ab08945", "question_id": "q10_temporal", "split": "test", "cluster_id": "syn_03012_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "tokenizer_hashes": {"receiver": "b9d8872cf3a99da99e1fc4b0f71232a451632dc49ca4c6771549b4efa5c1c763"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 2, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1767, "query_tokens": 541, "message_tokens": 419, "message_utf8_bytes": 1137, "cache_tensor_bytes": null, "serialized_payload_bytes": 1137, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 39802.83679103013, "hops": {"1": 23159.84900004696, "2": 16642.98779098317}, "receiver_prefill_ms": 383.3734589861706, "ttft_ms": 109.3525419710204, "decode_ms": 31.308625009842217}, "total_wall_ms": 40327.54145807121, "ttft_ms": 109.3525419710204, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 59, "facts_retained": 30, "retention_rate": 0.5084745762711864, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.5714285714285714, "amount": 0.6, "status": 0.5833333333333334, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1137}, "ledger_by_hop": {"1": {"facts_total": 59, "facts_retained": 30, "retention_rate": 0.5084745762711864, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.5714285714285714, "amount": 0.6, "status": 0.5833333333333334, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1213}, "2": {"facts_total": 59, "facts_retained": 30, "retention_rate": 0.5084745762711864, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.5714285714285714, "amount": 0.6, "status": 0.5833333333333334, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1137}}}, "reference_key_id": "keys/synthetic/syn_03012_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "d1c14f4f9514cc9284b356cef000b1d2966bb027d2e4b70702d913f0be396a36", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "5ba62c722535738bd27f7b4b", "attempt_id": "2d20bd2547a4", "run_group_id": "E4_core:A4|closed|qwen3b-llama3b:h2", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:42:40.631120Z", "task_family": "synthetic", "document_id": "syn_03012_2000", "document_hash": "03e441e68d54e1af036f8ff905dfd6cbbccead62636279c1a8e179352ab08945", "question_id": "q11_aggregation", "split": "test", "cluster_id": "syn_03012_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "tokenizer_hashes": {"receiver": "b9d8872cf3a99da99e1fc4b0f71232a451632dc49ca4c6771549b4efa5c1c763"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 2, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1767, "query_tokens": 534, "message_tokens": 419, "message_utf8_bytes": 1137, "cache_tensor_bytes": null, "serialized_payload_bytes": 1137, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 39802.83679103013, "hops": {"1": 23159.84900004696, "2": 16642.98779098317}, "receiver_prefill_ms": 342.712959041819, "ttft_ms": 39.15920795407146, "decode_ms": 143.79341702442616}, "total_wall_ms": 40329.12041596137, "ttft_ms": 39.15920795407146, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 5}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "9,381,835", "evidence_artifact": {"ledger": {"facts_total": 59, "facts_retained": 30, "retention_rate": 0.5084745762711864, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.5714285714285714, "amount": 0.6, "status": 0.5833333333333334, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1137}, "ledger_by_hop": {"1": {"facts_total": 59, "facts_retained": 30, "retention_rate": 0.5084745762711864, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.5714285714285714, "amount": 0.6, "status": 0.5833333333333334, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1213}, "2": {"facts_total": 59, "facts_retained": 30, "retention_rate": 0.5084745762711864, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.5714285714285714, "amount": 0.6, "status": 0.5833333333333334, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1137}}}, "reference_key_id": "keys/synthetic/syn_03012_2000.json", "task_scores": {"correct": 0.0, "abstained": 0.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "d1c14f4f9514cc9284b356cef000b1d2966bb027d2e4b70702d913f0be396a36", "output_hash": "5f7b503199558e57bee95e5db837c17c224702f5159f2fcfb4effe6fed47fcba"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "b2dd61061658a490ea023243", "attempt_id": "df9eca121934", "run_group_id": "E4_core:A4|closed|qwen3b-llama3b:h2", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:42:41.174925Z", "task_family": "synthetic", "document_id": "syn_03012_2000", "document_hash": "03e441e68d54e1af036f8ff905dfd6cbbccead62636279c1a8e179352ab08945", "question_id": "q12_unit_conversion", "split": "test", "cluster_id": "syn_03012_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "tokenizer_hashes": {"receiver": "b9d8872cf3a99da99e1fc4b0f71232a451632dc49ca4c6771549b4efa5c1c763"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 2, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1767, "query_tokens": 531, "message_tokens": 419, "message_utf8_bytes": 1137, "cache_tensor_bytes": null, "serialized_payload_bytes": 1137, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 39802.83679103013, "hops": {"1": 23159.84900004696, "2": 16642.98779098317}, "receiver_prefill_ms": 335.7412910554558, "ttft_ms": 21.38391707558185, "decode_ms": 95.22733301855624}, "total_wall_ms": 40255.939249065705, "ttft_ms": 21.38391707558185, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 3}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "44.62", "evidence_artifact": {"ledger": {"facts_total": 59, "facts_retained": 30, "retention_rate": 0.5084745762711864, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.5714285714285714, "amount": 0.6, "status": 0.5833333333333334, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1137}, "ledger_by_hop": {"1": {"facts_total": 59, "facts_retained": 30, "retention_rate": 0.5084745762711864, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.5714285714285714, "amount": 0.6, "status": 0.5833333333333334, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1213}, "2": {"facts_total": 59, "facts_retained": 30, "retention_rate": 0.5084745762711864, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.5714285714285714, "amount": 0.6, "status": 0.5833333333333334, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1137}}}, "reference_key_id": "keys/synthetic/syn_03012_2000.json", "task_scores": {"correct": 0.0, "abstained": 0.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "d1c14f4f9514cc9284b356cef000b1d2966bb027d2e4b70702d913f0be396a36", "output_hash": "d9a3265d0a78db3d082616021f44cd388888ddddf0a7328ce458e95e7824f8e9"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "deaf45c76d023854585f89fc", "attempt_id": "9394c98e7678", "run_group_id": "E4_core:A4|closed|qwen3b-llama3b:h2", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:42:41.646317Z", "task_family": "synthetic", "document_id": "syn_03012_2000", "document_hash": "03e441e68d54e1af036f8ff905dfd6cbbccead62636279c1a8e179352ab08945", "question_id": "q13_unanswerable", "split": "test", "cluster_id": "syn_03012_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "tokenizer_hashes": {"receiver": "b9d8872cf3a99da99e1fc4b0f71232a451632dc49ca4c6771549b4efa5c1c763"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 2, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1767, "query_tokens": 535, "message_tokens": 419, "message_utf8_bytes": 1137, "cache_tensor_bytes": null, "serialized_payload_bytes": 1137, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 39802.83679103013, "hops": {"1": 23159.84900004696, "2": 16642.98779098317}, "receiver_prefill_ms": 319.56166704185307, "ttft_ms": 39.80733302887529, "decode_ms": 114.78083301335573}, "total_wall_ms": 40277.703374042176, "ttft_ms": 39.80733302887529, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 59, "facts_retained": 30, "retention_rate": 0.5084745762711864, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.5714285714285714, "amount": 0.6, "status": 0.5833333333333334, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1137}, "ledger_by_hop": {"1": {"facts_total": 59, "facts_retained": 30, "retention_rate": 0.5084745762711864, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.5714285714285714, "amount": 0.6, "status": 0.5833333333333334, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1213}, "2": {"facts_total": 59, "facts_retained": 30, "retention_rate": 0.5084745762711864, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.5714285714285714, "amount": 0.6, "status": 0.5833333333333334, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1137}}}, "reference_key_id": "keys/synthetic/syn_03012_2000.json", "task_scores": {"correct": 1.0, "abstained": 1.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "d1c14f4f9514cc9284b356cef000b1d2966bb027d2e4b70702d913f0be396a36", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "f23708b117c54eff94e73cd9", "attempt_id": "019e6d4d73b3", "run_group_id": "E4_core:A4|closed|qwen3b-llama3b:h2", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:42:42.139063Z", "task_family": "synthetic", "document_id": "syn_03012_2000", "document_hash": "03e441e68d54e1af036f8ff905dfd6cbbccead62636279c1a8e179352ab08945", "question_id": "q14_unanswerable", "split": "test", "cluster_id": "syn_03012_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "tokenizer_hashes": {"receiver": "b9d8872cf3a99da99e1fc4b0f71232a451632dc49ca4c6771549b4efa5c1c763"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 2, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1767, "query_tokens": 536, "message_tokens": 419, "message_utf8_bytes": 1137, "cache_tensor_bytes": null, "serialized_payload_bytes": 1137, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 39802.83679103013, "hops": {"1": 23159.84900004696, "2": 16642.98779098317}, "receiver_prefill_ms": 421.8140410957858, "ttft_ms": 31.351207988336682, "decode_ms": 19.435291993431747}, "total_wall_ms": 40276.17395797279, "ttft_ms": 31.351207988336682, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 59, "facts_retained": 30, "retention_rate": 0.5084745762711864, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.5714285714285714, "amount": 0.6, "status": 0.5833333333333334, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1137}, "ledger_by_hop": {"1": {"facts_total": 59, "facts_retained": 30, "retention_rate": 0.5084745762711864, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.5714285714285714, "amount": 0.6, "status": 0.5833333333333334, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1213}, "2": {"facts_total": 59, "facts_retained": 30, "retention_rate": 0.5084745762711864, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.5714285714285714, "amount": 0.6, "status": 0.5833333333333334, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1137}}}, "reference_key_id": "keys/synthetic/syn_03012_2000.json", "task_scores": {"correct": 1.0, "abstained": 1.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "d1c14f4f9514cc9284b356cef000b1d2966bb027d2e4b70702d913f0be396a36", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "a33bff84e32ace381d729485", "attempt_id": "81e2d5a71bb9", "run_group_id": "E4_core:A4|closed|qwen3b-llama3b:h4", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:43:20.961837Z", "task_family": "synthetic", "document_id": "syn_03012_2000", "document_hash": "03e441e68d54e1af036f8ff905dfd6cbbccead62636279c1a8e179352ab08945", "question_id": "q00_single_fact", "split": "test", "cluster_id": "syn_03012_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "tokenizer_hashes": {"receiver": "b9d8872cf3a99da99e1fc4b0f71232a451632dc49ca4c6771549b4efa5c1c763"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 4, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1767, "query_tokens": 524, "message_tokens": 420, "message_utf8_bytes": 1362, "cache_tensor_bytes": null, "serialized_payload_bytes": 1362, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 78129.69400000293, "hops": {"1": 23159.84900004696, "2": 16642.98779098317, "3": 21655.61054192949, "4": 16671.246667043306}, "receiver_prefill_ms": 345.80716595519334, "ttft_ms": 34.0406249742955, "decode_ms": 218.21470907889307}, "total_wall_ms": 78728.4520000685, "ttft_ms": 34.0406249742955, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 7}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "XX-71M-6213", "evidence_artifact": {"ledger": {"facts_total": 59, "facts_retained": 29, "retention_rate": 0.4915254237288136, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.5714285714285714, "amount": 0.5, "status": 0.5833333333333334, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1362}, "ledger_by_hop": {"1": {"facts_total": 59, "facts_retained": 30, "retention_rate": 0.5084745762711864, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.5714285714285714, "amount": 0.6, "status": 0.5833333333333334, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1213}, "2": {"facts_total": 59, "facts_retained": 30, "retention_rate": 0.5084745762711864, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.5714285714285714, "amount": 0.6, "status": 0.5833333333333334, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1137}, "3": {"facts_total": 59, "facts_retained": 30, "retention_rate": 0.5084745762711864, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.5714285714285714, "amount": 0.6, "status": 0.5833333333333334, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1137}, "4": {"facts_total": 59, "facts_retained": 29, "retention_rate": 0.4915254237288136, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.5714285714285714, "amount": 0.5, "status": 0.5833333333333334, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1362}}}, "reference_key_id": "keys/synthetic/syn_03012_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "420cad6bf9617e26059904798724742f1c2c3f1e7cf172f9d209dd8783419681", "output_hash": "66ef6c5da682b38795e9029723bbd71cdde2c4dec70cf1fae6efe62ef243f10a"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "13d8d0ebd531128ca845abc9", "attempt_id": "a2be9320e84c", "run_group_id": "E4_core:A4|closed|qwen3b-llama3b:h4", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:43:21.580100Z", "task_family": "synthetic", "document_id": "syn_03012_2000", "document_hash": "03e441e68d54e1af036f8ff905dfd6cbbccead62636279c1a8e179352ab08945", "question_id": "q01_single_fact", "split": "test", "cluster_id": "syn_03012_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "tokenizer_hashes": {"receiver": "b9d8872cf3a99da99e1fc4b0f71232a451632dc49ca4c6771549b4efa5c1c763"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 4, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1767, "query_tokens": 529, "message_tokens": 420, "message_utf8_bytes": 1362, "cache_tensor_bytes": null, "serialized_payload_bytes": 1362, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 78129.69400000293, "hops": {"1": 23159.84900004696, "2": 16642.98779098317, "3": 21655.61054192949, "4": 16671.246667043306}, "receiver_prefill_ms": 302.46783304028213, "ttft_ms": 35.81366699654609, "decode_ms": 39.88108295015991}, "total_wall_ms": 78508.52525001392, "ttft_ms": 35.81366699654609, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 59, "facts_retained": 29, "retention_rate": 0.4915254237288136, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.5714285714285714, "amount": 0.5, "status": 0.5833333333333334, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1362}, "ledger_by_hop": {"1": {"facts_total": 59, "facts_retained": 30, "retention_rate": 0.5084745762711864, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.5714285714285714, "amount": 0.6, "status": 0.5833333333333334, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1213}, "2": {"facts_total": 59, "facts_retained": 30, "retention_rate": 0.5084745762711864, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.5714285714285714, "amount": 0.6, "status": 0.5833333333333334, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1137}, "3": {"facts_total": 59, "facts_retained": 30, "retention_rate": 0.5084745762711864, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.5714285714285714, "amount": 0.6, "status": 0.5833333333333334, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1137}, "4": {"facts_total": 59, "facts_retained": 29, "retention_rate": 0.4915254237288136, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.5714285714285714, "amount": 0.5, "status": 0.5833333333333334, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1362}}}, "reference_key_id": "keys/synthetic/syn_03012_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "420cad6bf9617e26059904798724742f1c2c3f1e7cf172f9d209dd8783419681", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "03461538f15ca2946dd23283", "attempt_id": "4c70829d7872", "run_group_id": "E4_core:A4|closed|qwen3b-llama3b:h4", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:43:21.977328Z", "task_family": "synthetic", "document_id": "syn_03012_2000", "document_hash": "03e441e68d54e1af036f8ff905dfd6cbbccead62636279c1a8e179352ab08945", "question_id": "q02_exact_copy", "split": "test", "cluster_id": "syn_03012_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "tokenizer_hashes": {"receiver": "b9d8872cf3a99da99e1fc4b0f71232a451632dc49ca4c6771549b4efa5c1c763"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 4, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1767, "query_tokens": 529, "message_tokens": 420, "message_utf8_bytes": 1362, "cache_tensor_bytes": null, "serialized_payload_bytes": 1362, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 78129.69400000293, "hops": {"1": 23159.84900004696, "2": 16642.98779098317, "3": 21655.61054192949, "4": 16671.246667043306}, "receiver_prefill_ms": 372.2197499591857, "ttft_ms": 131.77783298306167, "decode_ms": 120.10604096576571}, "total_wall_ms": 78754.55470907036, "ttft_ms": 131.77783298306167, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 4}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "13.4343", "evidence_artifact": {"ledger": {"facts_total": 59, "facts_retained": 29, "retention_rate": 0.4915254237288136, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.5714285714285714, "amount": 0.5, "status": 0.5833333333333334, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1362}, "ledger_by_hop": {"1": {"facts_total": 59, "facts_retained": 30, "retention_rate": 0.5084745762711864, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.5714285714285714, "amount": 0.6, "status": 0.5833333333333334, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1213}, "2": {"facts_total": 59, "facts_retained": 30, "retention_rate": 0.5084745762711864, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.5714285714285714, "amount": 0.6, "status": 0.5833333333333334, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1137}, "3": {"facts_total": 59, "facts_retained": 30, "retention_rate": 0.5084745762711864, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.5714285714285714, "amount": 0.6, "status": 0.5833333333333334, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1137}, "4": {"facts_total": 59, "facts_retained": 29, "retention_rate": 0.4915254237288136, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.5714285714285714, "amount": 0.5, "status": 0.5833333333333334, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1362}}}, "reference_key_id": "keys/synthetic/syn_03012_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "420cad6bf9617e26059904798724742f1c2c3f1e7cf172f9d209dd8783419681", "output_hash": "599bc22e373b1cf7b4679208eb9e515ad0e8dcb88dde0c7f579637ec7d3f3014"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "ebd612c3e49ed1b3f9399cf2", "attempt_id": "c5ac99c8e16c", "run_group_id": "E4_core:A4|closed|qwen3b-llama3b:h4", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:43:22.620556Z", "task_family": "synthetic", "document_id": "syn_03012_2000", "document_hash": "03e441e68d54e1af036f8ff905dfd6cbbccead62636279c1a8e179352ab08945", "question_id": "q03_exact_copy", "split": "test", "cluster_id": "syn_03012_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "tokenizer_hashes": {"receiver": "b9d8872cf3a99da99e1fc4b0f71232a451632dc49ca4c6771549b4efa5c1c763"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 4, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1767, "query_tokens": 525, "message_tokens": 420, "message_utf8_bytes": 1362, "cache_tensor_bytes": null, "serialized_payload_bytes": 1362, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 78129.69400000293, "hops": {"1": 23159.84900004696, "2": 16642.98779098317, "3": 21655.61054192949, "4": 16671.246667043306}, "receiver_prefill_ms": 335.80120792612433, "ttft_ms": 37.96408302150667, "decode_ms": 23.753541987389326}, "total_wall_ms": 78527.87904196884, "ttft_ms": 37.96408302150667, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 59, "facts_retained": 29, "retention_rate": 0.4915254237288136, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.5714285714285714, "amount": 0.5, "status": 0.5833333333333334, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1362}, "ledger_by_hop": {"1": {"facts_total": 59, "facts_retained": 30, "retention_rate": 0.5084745762711864, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.5714285714285714, "amount": 0.6, "status": 0.5833333333333334, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1213}, "2": {"facts_total": 59, "facts_retained": 30, "retention_rate": 0.5084745762711864, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.5714285714285714, "amount": 0.6, "status": 0.5833333333333334, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1137}, "3": {"facts_total": 59, "facts_retained": 30, "retention_rate": 0.5084745762711864, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.5714285714285714, "amount": 0.6, "status": 0.5833333333333334, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1137}, "4": {"facts_total": 59, "facts_retained": 29, "retention_rate": 0.4915254237288136, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.5714285714285714, "amount": 0.5, "status": 0.5833333333333334, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1362}}}, "reference_key_id": "keys/synthetic/syn_03012_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "420cad6bf9617e26059904798724742f1c2c3f1e7cf172f9d209dd8783419681", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "b6d2293184336488ae0aa9f3", "attempt_id": "056088500450", "run_group_id": "E4_core:A4|closed|qwen3b-llama3b:h4", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:43:23.036705Z", "task_family": "synthetic", "document_id": "syn_03012_2000", "document_hash": "03e441e68d54e1af036f8ff905dfd6cbbccead62636279c1a8e179352ab08945", "question_id": "q04_multi_hop", "split": "test", "cluster_id": "syn_03012_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "tokenizer_hashes": {"receiver": "b9d8872cf3a99da99e1fc4b0f71232a451632dc49ca4c6771549b4efa5c1c763"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 4, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1767, "query_tokens": 530, "message_tokens": 420, "message_utf8_bytes": 1362, "cache_tensor_bytes": null, "serialized_payload_bytes": 1362, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 78129.69400000293, "hops": {"1": 23159.84900004696, "2": 16642.98779098317, "3": 21655.61054192949, "4": 16671.246667043306}, "receiver_prefill_ms": 335.8453329419717, "ttft_ms": 32.20354102086276, "decode_ms": 28.047541971318424}, "total_wall_ms": 78526.29574993625, "ttft_ms": 32.20354102086276, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 59, "facts_retained": 29, "retention_rate": 0.4915254237288136, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.5714285714285714, "amount": 0.5, "status": 0.5833333333333334, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1362}, "ledger_by_hop": {"1": {"facts_total": 59, "facts_retained": 30, "retention_rate": 0.5084745762711864, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.5714285714285714, "amount": 0.6, "status": 0.5833333333333334, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1213}, "2": {"facts_total": 59, "facts_retained": 30, "retention_rate": 0.5084745762711864, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.5714285714285714, "amount": 0.6, "status": 0.5833333333333334, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1137}, "3": {"facts_total": 59, "facts_retained": 30, "retention_rate": 0.5084745762711864, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.5714285714285714, "amount": 0.6, "status": 0.5833333333333334, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1137}, "4": {"facts_total": 59, "facts_retained": 29, "retention_rate": 0.4915254237288136, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.5714285714285714, "amount": 0.5, "status": 0.5833333333333334, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1362}}}, "reference_key_id": "keys/synthetic/syn_03012_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "420cad6bf9617e26059904798724742f1c2c3f1e7cf172f9d209dd8783419681", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "90008ba4bee6d4072df0b46f", "attempt_id": "4739f95213a3", "run_group_id": "E4_core:A4|closed|qwen3b-llama3b:h4", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:43:23.450527Z", "task_family": "synthetic", "document_id": "syn_03012_2000", "document_hash": "03e441e68d54e1af036f8ff905dfd6cbbccead62636279c1a8e179352ab08945", "question_id": "q05_multi_hop", "split": "test", "cluster_id": "syn_03012_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "tokenizer_hashes": {"receiver": "b9d8872cf3a99da99e1fc4b0f71232a451632dc49ca4c6771549b4efa5c1c763"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 4, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1767, "query_tokens": 531, "message_tokens": 420, "message_utf8_bytes": 1362, "cache_tensor_bytes": null, "serialized_payload_bytes": 1362, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 78129.69400000293, "hops": {"1": 23159.84900004696, "2": 16642.98779098317, "3": 21655.61054192949, "4": 16671.246667043306}, "receiver_prefill_ms": 306.4493340207264, "ttft_ms": 39.381915936246514, "decode_ms": 441.7737500043586}, "total_wall_ms": 78918.09429193381, "ttft_ms": 39.381915936246514, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 7}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "DG-40E-2777", "evidence_artifact": {"ledger": {"facts_total": 59, "facts_retained": 29, "retention_rate": 0.4915254237288136, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.5714285714285714, "amount": 0.5, "status": 0.5833333333333334, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1362}, "ledger_by_hop": {"1": {"facts_total": 59, "facts_retained": 30, "retention_rate": 0.5084745762711864, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.5714285714285714, "amount": 0.6, "status": 0.5833333333333334, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1213}, "2": {"facts_total": 59, "facts_retained": 30, "retention_rate": 0.5084745762711864, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.5714285714285714, "amount": 0.6, "status": 0.5833333333333334, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1137}, "3": {"facts_total": 59, "facts_retained": 30, "retention_rate": 0.5084745762711864, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.5714285714285714, "amount": 0.6, "status": 0.5833333333333334, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1137}, "4": {"facts_total": 59, "facts_retained": 29, "retention_rate": 0.4915254237288136, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.5714285714285714, "amount": 0.5, "status": 0.5833333333333334, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1362}}}, "reference_key_id": "keys/synthetic/syn_03012_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "420cad6bf9617e26059904798724742f1c2c3f1e7cf172f9d209dd8783419681", "output_hash": "16a7e3f4a7cc4a9b66f86c2cdcfad3b35daf1da199d9406e0627c8af2d8d73f3"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "68cd37b553a0444bc5c17f46", "attempt_id": "f622eb40152e", "run_group_id": "E4_core:A4|closed|qwen3b-llama3b:h4", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:43:24.256782Z", "task_family": "synthetic", "document_id": "syn_03012_2000", "document_hash": "03e441e68d54e1af036f8ff905dfd6cbbccead62636279c1a8e179352ab08945", "question_id": "q06_negation", "split": "test", "cluster_id": "syn_03012_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "tokenizer_hashes": {"receiver": "b9d8872cf3a99da99e1fc4b0f71232a451632dc49ca4c6771549b4efa5c1c763"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 4, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1767, "query_tokens": 524, "message_tokens": 420, "message_utf8_bytes": 1362, "cache_tensor_bytes": null, "serialized_payload_bytes": 1362, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 78129.69400000293, "hops": {"1": 23159.84900004696, "2": 16642.98779098317, "3": 21655.61054192949, "4": 16671.246667043306}, "receiver_prefill_ms": 328.19004100747406, "ttft_ms": 26.465333066880703, "decode_ms": 78.82554200477898}, "total_wall_ms": 78563.75412503257, "ttft_ms": 26.465333066880703, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 2}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "No.", "evidence_artifact": {"ledger": {"facts_total": 59, "facts_retained": 29, "retention_rate": 0.4915254237288136, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.5714285714285714, "amount": 0.5, "status": 0.5833333333333334, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1362}, "ledger_by_hop": {"1": {"facts_total": 59, "facts_retained": 30, "retention_rate": 0.5084745762711864, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.5714285714285714, "amount": 0.6, "status": 0.5833333333333334, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1213}, "2": {"facts_total": 59, "facts_retained": 30, "retention_rate": 0.5084745762711864, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.5714285714285714, "amount": 0.6, "status": 0.5833333333333334, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1137}, "3": {"facts_total": 59, "facts_retained": 30, "retention_rate": 0.5084745762711864, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.5714285714285714, "amount": 0.6, "status": 0.5833333333333334, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1137}, "4": {"facts_total": 59, "facts_retained": 29, "retention_rate": 0.4915254237288136, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.5714285714285714, "amount": 0.5, "status": 0.5833333333333334, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1362}}}, "reference_key_id": "keys/synthetic/syn_03012_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "420cad6bf9617e26059904798724742f1c2c3f1e7cf172f9d209dd8783419681", "output_hash": "38178a20b470cfd18299fb1593dd4ba706f1a83321dac051d01086efd8b7a96f"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "365b3308d96d5fb7af88e0f4", "attempt_id": "91b4b043dc8c", "run_group_id": "E4_core:A4|closed|qwen3b-llama3b:h4", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:43:24.708963Z", "task_family": "synthetic", "document_id": "syn_03012_2000", "document_hash": "03e441e68d54e1af036f8ff905dfd6cbbccead62636279c1a8e179352ab08945", "question_id": "q07_negation", "split": "test", "cluster_id": "syn_03012_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "tokenizer_hashes": {"receiver": "b9d8872cf3a99da99e1fc4b0f71232a451632dc49ca4c6771549b4efa5c1c763"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 4, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1767, "query_tokens": 525, "message_tokens": 420, "message_utf8_bytes": 1362, "cache_tensor_bytes": null, "serialized_payload_bytes": 1362, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 78129.69400000293, "hops": {"1": 23159.84900004696, "2": 16642.98779098317, "3": 21655.61054192949, "4": 16671.246667043306}, "receiver_prefill_ms": 349.761041929014, "ttft_ms": 25.737790972925723, "decode_ms": 49.55704195890576}, "total_wall_ms": 78555.36183295771, "ttft_ms": 25.737790972925723, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 2}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "No.", "evidence_artifact": {"ledger": {"facts_total": 59, "facts_retained": 29, "retention_rate": 0.4915254237288136, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.5714285714285714, "amount": 0.5, "status": 0.5833333333333334, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1362}, "ledger_by_hop": {"1": {"facts_total": 59, "facts_retained": 30, "retention_rate": 0.5084745762711864, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.5714285714285714, "amount": 0.6, "status": 0.5833333333333334, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1213}, "2": {"facts_total": 59, "facts_retained": 30, "retention_rate": 0.5084745762711864, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.5714285714285714, "amount": 0.6, "status": 0.5833333333333334, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1137}, "3": {"facts_total": 59, "facts_retained": 30, "retention_rate": 0.5084745762711864, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.5714285714285714, "amount": 0.6, "status": 0.5833333333333334, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1137}, "4": {"facts_total": 59, "facts_retained": 29, "retention_rate": 0.4915254237288136, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.5714285714285714, "amount": 0.5, "status": 0.5833333333333334, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1362}}}, "reference_key_id": "keys/synthetic/syn_03012_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "420cad6bf9617e26059904798724742f1c2c3f1e7cf172f9d209dd8783419681", "output_hash": "38178a20b470cfd18299fb1593dd4ba706f1a83321dac051d01086efd8b7a96f"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "815532f50e7440a6ae4086b8", "attempt_id": "2d03f01be522", "run_group_id": "E4_core:A4|closed|qwen3b-llama3b:h4", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:43:25.152485Z", "task_family": "synthetic", "document_id": "syn_03012_2000", "document_hash": "03e441e68d54e1af036f8ff905dfd6cbbccead62636279c1a8e179352ab08945", "question_id": "q08_revision", "split": "test", "cluster_id": "syn_03012_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "tokenizer_hashes": {"receiver": "b9d8872cf3a99da99e1fc4b0f71232a451632dc49ca4c6771549b4efa5c1c763"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 4, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1767, "query_tokens": 540, "message_tokens": 420, "message_utf8_bytes": 1362, "cache_tensor_bytes": null, "serialized_payload_bytes": 1362, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 78129.69400000293, "hops": {"1": 23159.84900004696, "2": 16642.98779098317, "3": 21655.61054192949, "4": 16671.246667043306}, "receiver_prefill_ms": 318.60270898323506, "ttft_ms": 41.64874996058643, "decode_ms": 356.330499984324}, "total_wall_ms": 78846.9547079876, "ttft_ms": 41.64874996058643, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 5}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "6,602,245", "evidence_artifact": {"ledger": {"facts_total": 59, "facts_retained": 29, "retention_rate": 0.4915254237288136, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.5714285714285714, "amount": 0.5, "status": 0.5833333333333334, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1362}, "ledger_by_hop": {"1": {"facts_total": 59, "facts_retained": 30, "retention_rate": 0.5084745762711864, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.5714285714285714, "amount": 0.6, "status": 0.5833333333333334, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1213}, "2": {"facts_total": 59, "facts_retained": 30, "retention_rate": 0.5084745762711864, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.5714285714285714, "amount": 0.6, "status": 0.5833333333333334, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1137}, "3": {"facts_total": 59, "facts_retained": 30, "retention_rate": 0.5084745762711864, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.5714285714285714, "amount": 0.6, "status": 0.5833333333333334, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1137}, "4": {"facts_total": 59, "facts_retained": 29, "retention_rate": 0.4915254237288136, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.5714285714285714, "amount": 0.5, "status": 0.5833333333333334, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1362}}}, "reference_key_id": "keys/synthetic/syn_03012_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "420cad6bf9617e26059904798724742f1c2c3f1e7cf172f9d209dd8783419681", "output_hash": "f121947078d464ab91a83739861b98c336a10c001704e2df84cc2b041bb8f4de"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "5b6eb2feb6abf88debe7c122", "attempt_id": "e1c0e366a267", "run_group_id": "E4_core:A4|closed|qwen3b-llama3b:h4", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:43:25.888941Z", "task_family": "synthetic", "document_id": "syn_03012_2000", "document_hash": "03e441e68d54e1af036f8ff905dfd6cbbccead62636279c1a8e179352ab08945", "question_id": "q09_revision", "split": "test", "cluster_id": "syn_03012_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "tokenizer_hashes": {"receiver": "b9d8872cf3a99da99e1fc4b0f71232a451632dc49ca4c6771549b4efa5c1c763"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 4, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1767, "query_tokens": 532, "message_tokens": 420, "message_utf8_bytes": 1362, "cache_tensor_bytes": null, "serialized_payload_bytes": 1362, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 78129.69400000293, "hops": {"1": 23159.84900004696, "2": 16642.98779098317, "3": 21655.61054192949, "4": 16671.246667043306}, "receiver_prefill_ms": 333.8912909384817, "ttft_ms": 36.01441695354879, "decode_ms": 22.313500056043267}, "total_wall_ms": 78522.50379195902, "ttft_ms": 36.01441695354879, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 59, "facts_retained": 29, "retention_rate": 0.4915254237288136, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.5714285714285714, "amount": 0.5, "status": 0.5833333333333334, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1362}, "ledger_by_hop": {"1": {"facts_total": 59, "facts_retained": 30, "retention_rate": 0.5084745762711864, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.5714285714285714, "amount": 0.6, "status": 0.5833333333333334, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1213}, "2": {"facts_total": 59, "facts_retained": 30, "retention_rate": 0.5084745762711864, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.5714285714285714, "amount": 0.6, "status": 0.5833333333333334, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1137}, "3": {"facts_total": 59, "facts_retained": 30, "retention_rate": 0.5084745762711864, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.5714285714285714, "amount": 0.6, "status": 0.5833333333333334, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1137}, "4": {"facts_total": 59, "facts_retained": 29, "retention_rate": 0.4915254237288136, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.5714285714285714, "amount": 0.5, "status": 0.5833333333333334, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1362}}}, "reference_key_id": "keys/synthetic/syn_03012_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "420cad6bf9617e26059904798724742f1c2c3f1e7cf172f9d209dd8783419681", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "f83f53ead1f3ac55a39f075e", "attempt_id": "7717513982d7", "run_group_id": "E4_core:A4|closed|qwen3b-llama3b:h4", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:43:26.298924Z", "task_family": "synthetic", "document_id": "syn_03012_2000", "document_hash": "03e441e68d54e1af036f8ff905dfd6cbbccead62636279c1a8e179352ab08945", "question_id": "q10_temporal", "split": "test", "cluster_id": "syn_03012_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "tokenizer_hashes": {"receiver": "b9d8872cf3a99da99e1fc4b0f71232a451632dc49ca4c6771549b4efa5c1c763"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 4, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1767, "query_tokens": 542, "message_tokens": 420, "message_utf8_bytes": 1362, "cache_tensor_bytes": null, "serialized_payload_bytes": 1362, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 78129.69400000293, "hops": {"1": 23159.84900004696, "2": 16642.98779098317, "3": 21655.61054192949, "4": 16671.246667043306}, "receiver_prefill_ms": 349.05125002842396, "ttft_ms": 31.932584010064602, "decode_ms": 20.1305840164423}, "total_wall_ms": 78531.40595799778, "ttft_ms": 31.932584010064602, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 59, "facts_retained": 29, "retention_rate": 0.4915254237288136, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.5714285714285714, "amount": 0.5, "status": 0.5833333333333334, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1362}, "ledger_by_hop": {"1": {"facts_total": 59, "facts_retained": 30, "retention_rate": 0.5084745762711864, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.5714285714285714, "amount": 0.6, "status": 0.5833333333333334, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1213}, "2": {"facts_total": 59, "facts_retained": 30, "retention_rate": 0.5084745762711864, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.5714285714285714, "amount": 0.6, "status": 0.5833333333333334, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1137}, "3": {"facts_total": 59, "facts_retained": 30, "retention_rate": 0.5084745762711864, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.5714285714285714, "amount": 0.6, "status": 0.5833333333333334, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1137}, "4": {"facts_total": 59, "facts_retained": 29, "retention_rate": 0.4915254237288136, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.5714285714285714, "amount": 0.5, "status": 0.5833333333333334, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1362}}}, "reference_key_id": "keys/synthetic/syn_03012_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "420cad6bf9617e26059904798724742f1c2c3f1e7cf172f9d209dd8783419681", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "b49f9e14b414113b88d7973b", "attempt_id": "5a6b2af83d69", "run_group_id": "E4_core:A4|closed|qwen3b-llama3b:h4", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:43:26.717959Z", "task_family": "synthetic", "document_id": "syn_03012_2000", "document_hash": "03e441e68d54e1af036f8ff905dfd6cbbccead62636279c1a8e179352ab08945", "question_id": "q11_aggregation", "split": "test", "cluster_id": "syn_03012_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "tokenizer_hashes": {"receiver": "b9d8872cf3a99da99e1fc4b0f71232a451632dc49ca4c6771549b4efa5c1c763"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 4, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1767, "query_tokens": 535, "message_tokens": 420, "message_utf8_bytes": 1362, "cache_tensor_bytes": null, "serialized_payload_bytes": 1362, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 78129.69400000293, "hops": {"1": 23159.84900004696, "2": 16642.98779098317, "3": 21655.61054192949, "4": 16671.246667043306}, "receiver_prefill_ms": 321.35024992749095, "ttft_ms": 46.08920798636973, "decode_ms": 321.99387496802956}, "total_wall_ms": 78819.89179097582, "ttft_ms": 46.08920798636973, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 5}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "9,381,835", "evidence_artifact": {"ledger": {"facts_total": 59, "facts_retained": 29, "retention_rate": 0.4915254237288136, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.5714285714285714, "amount": 0.5, "status": 0.5833333333333334, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1362}, "ledger_by_hop": {"1": {"facts_total": 59, "facts_retained": 30, "retention_rate": 0.5084745762711864, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.5714285714285714, "amount": 0.6, "status": 0.5833333333333334, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1213}, "2": {"facts_total": 59, "facts_retained": 30, "retention_rate": 0.5084745762711864, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.5714285714285714, "amount": 0.6, "status": 0.5833333333333334, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1137}, "3": {"facts_total": 59, "facts_retained": 30, "retention_rate": 0.5084745762711864, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.5714285714285714, "amount": 0.6, "status": 0.5833333333333334, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1137}, "4": {"facts_total": 59, "facts_retained": 29, "retention_rate": 0.4915254237288136, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.5714285714285714, "amount": 0.5, "status": 0.5833333333333334, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1362}}}, "reference_key_id": "keys/synthetic/syn_03012_2000.json", "task_scores": {"correct": 0.0, "abstained": 0.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "420cad6bf9617e26059904798724742f1c2c3f1e7cf172f9d209dd8783419681", "output_hash": "5f7b503199558e57bee95e5db837c17c224702f5159f2fcfb4effe6fed47fcba"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "1acabee653447e4c2e060a73", "attempt_id": "61970b27e575", "run_group_id": "E4_core:A4|closed|qwen3b-llama3b:h4", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:43:27.425747Z", "task_family": "synthetic", "document_id": "syn_03012_2000", "document_hash": "03e441e68d54e1af036f8ff905dfd6cbbccead62636279c1a8e179352ab08945", "question_id": "q12_unit_conversion", "split": "test", "cluster_id": "syn_03012_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "tokenizer_hashes": {"receiver": "b9d8872cf3a99da99e1fc4b0f71232a451632dc49ca4c6771549b4efa5c1c763"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 4, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1767, "query_tokens": 532, "message_tokens": 420, "message_utf8_bytes": 1362, "cache_tensor_bytes": null, "serialized_payload_bytes": 1362, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 78129.69400000293, "hops": {"1": 23159.84900004696, "2": 16642.98779098317, "3": 21655.61054192949, "4": 16671.246667043306}, "receiver_prefill_ms": 426.4595420099795, "ttft_ms": 37.004082929342985, "decode_ms": 105.876374989748}, "total_wall_ms": 78699.72508307546, "ttft_ms": 37.004082929342985, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 3}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "44.62", "evidence_artifact": {"ledger": {"facts_total": 59, "facts_retained": 29, "retention_rate": 0.4915254237288136, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.5714285714285714, "amount": 0.5, "status": 0.5833333333333334, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1362}, "ledger_by_hop": {"1": {"facts_total": 59, "facts_retained": 30, "retention_rate": 0.5084745762711864, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.5714285714285714, "amount": 0.6, "status": 0.5833333333333334, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1213}, "2": {"facts_total": 59, "facts_retained": 30, "retention_rate": 0.5084745762711864, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.5714285714285714, "amount": 0.6, "status": 0.5833333333333334, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1137}, "3": {"facts_total": 59, "facts_retained": 30, "retention_rate": 0.5084745762711864, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.5714285714285714, "amount": 0.6, "status": 0.5833333333333334, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1137}, "4": {"facts_total": 59, "facts_retained": 29, "retention_rate": 0.4915254237288136, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.5714285714285714, "amount": 0.5, "status": 0.5833333333333334, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1362}}}, "reference_key_id": "keys/synthetic/syn_03012_2000.json", "task_scores": {"correct": 0.0, "abstained": 0.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "420cad6bf9617e26059904798724742f1c2c3f1e7cf172f9d209dd8783419681", "output_hash": "d9a3265d0a78db3d082616021f44cd388888ddddf0a7328ce458e95e7824f8e9"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "e663661395bfdcf6f6c7f149", "attempt_id": "8cff6d12ebff", "run_group_id": "E4_core:A4|closed|qwen3b-llama3b:h4", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:43:28.015478Z", "task_family": "synthetic", "document_id": "syn_03012_2000", "document_hash": "03e441e68d54e1af036f8ff905dfd6cbbccead62636279c1a8e179352ab08945", "question_id": "q13_unanswerable", "split": "test", "cluster_id": "syn_03012_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "tokenizer_hashes": {"receiver": "b9d8872cf3a99da99e1fc4b0f71232a451632dc49ca4c6771549b4efa5c1c763"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 4, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1767, "query_tokens": 536, "message_tokens": 420, "message_utf8_bytes": 1362, "cache_tensor_bytes": null, "serialized_payload_bytes": 1362, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 78129.69400000293, "hops": {"1": 23159.84900004696, "2": 16642.98779098317, "3": 21655.61054192949, "4": 16671.246667043306}, "receiver_prefill_ms": 349.1747910156846, "ttft_ms": 26.845542015507817, "decode_ms": 33.075125073082745}, "total_wall_ms": 78539.55770900939, "ttft_ms": 26.845542015507817, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 59, "facts_retained": 29, "retention_rate": 0.4915254237288136, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.5714285714285714, "amount": 0.5, "status": 0.5833333333333334, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1362}, "ledger_by_hop": {"1": {"facts_total": 59, "facts_retained": 30, "retention_rate": 0.5084745762711864, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.5714285714285714, "amount": 0.6, "status": 0.5833333333333334, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1213}, "2": {"facts_total": 59, "facts_retained": 30, "retention_rate": 0.5084745762711864, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.5714285714285714, "amount": 0.6, "status": 0.5833333333333334, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1137}, "3": {"facts_total": 59, "facts_retained": 30, "retention_rate": 0.5084745762711864, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.5714285714285714, "amount": 0.6, "status": 0.5833333333333334, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1137}, "4": {"facts_total": 59, "facts_retained": 29, "retention_rate": 0.4915254237288136, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.5714285714285714, "amount": 0.5, "status": 0.5833333333333334, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1362}}}, "reference_key_id": "keys/synthetic/syn_03012_2000.json", "task_scores": {"correct": 1.0, "abstained": 1.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "420cad6bf9617e26059904798724742f1c2c3f1e7cf172f9d209dd8783419681", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "db1b35ab673ec8b52b6a5383", "attempt_id": "084084c67e04", "run_group_id": "E4_core:A4|closed|qwen3b-llama3b:h4", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:43:28.443458Z", "task_family": "synthetic", "document_id": "syn_03012_2000", "document_hash": "03e441e68d54e1af036f8ff905dfd6cbbccead62636279c1a8e179352ab08945", "question_id": "q14_unanswerable", "split": "test", "cluster_id": "syn_03012_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "tokenizer_hashes": {"receiver": "b9d8872cf3a99da99e1fc4b0f71232a451632dc49ca4c6771549b4efa5c1c763"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 4, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1767, "query_tokens": 537, "message_tokens": 420, "message_utf8_bytes": 1362, "cache_tensor_bytes": null, "serialized_payload_bytes": 1362, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 78129.69400000293, "hops": {"1": 23159.84900004696, "2": 16642.98779098317, "3": 21655.61054192949, "4": 16671.246667043306}, "receiver_prefill_ms": 334.96524998918176, "ttft_ms": 49.78204192593694, "decode_ms": 50.42841599788517}, "total_wall_ms": 78565.86700002663, "ttft_ms": 49.78204192593694, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 59, "facts_retained": 29, "retention_rate": 0.4915254237288136, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.5714285714285714, "amount": 0.5, "status": 0.5833333333333334, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1362}, "ledger_by_hop": {"1": {"facts_total": 59, "facts_retained": 30, "retention_rate": 0.5084745762711864, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.5714285714285714, "amount": 0.6, "status": 0.5833333333333334, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1213}, "2": {"facts_total": 59, "facts_retained": 30, "retention_rate": 0.5084745762711864, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.5714285714285714, "amount": 0.6, "status": 0.5833333333333334, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1137}, "3": {"facts_total": 59, "facts_retained": 30, "retention_rate": 0.5084745762711864, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.5714285714285714, "amount": 0.6, "status": 0.5833333333333334, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1137}, "4": {"facts_total": 59, "facts_retained": 29, "retention_rate": 0.4915254237288136, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.5714285714285714, "amount": 0.5, "status": 0.5833333333333334, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1362}}}, "reference_key_id": "keys/synthetic/syn_03012_2000.json", "task_scores": {"correct": 1.0, "abstained": 1.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "420cad6bf9617e26059904798724742f1c2c3f1e7cf172f9d209dd8783419681", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "f183cb838f5d29fda46aa5b0", "attempt_id": "1cb0c12f1666", "run_group_id": "E4_core:A4|closed|qwen3b-llama3b:h8", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:44:39.310263Z", "task_family": "synthetic", "document_id": "syn_03012_2000", "document_hash": "03e441e68d54e1af036f8ff905dfd6cbbccead62636279c1a8e179352ab08945", "question_id": "q00_single_fact", "split": "test", "cluster_id": "syn_03012_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "tokenizer_hashes": {"receiver": "b9d8872cf3a99da99e1fc4b0f71232a451632dc49ca4c6771549b4efa5c1c763"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 8, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1767, "query_tokens": 502, "message_tokens": 398, "message_utf8_bytes": 1341, "cache_tensor_bytes": null, "serialized_payload_bytes": 1341, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 148535.66862398293, "hops": {"1": 23159.84900004696, "2": 16642.98779098317, "3": 21655.61054192949, "4": 16671.246667043306, "5": 19499.049708014354, "6": 16652.313041035086, "7": 18373.46841697581, "8": 15881.14345795475}, "receiver_prefill_ms": 382.71262496709824, "ttft_ms": 33.29816600307822, "decode_ms": 219.6699579944834}, "total_wall_ms": 149171.96991597302, "ttft_ms": 33.29816600307822, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 7}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "XX-71M-6213", "evidence_artifact": {"ledger": {"facts_total": 59, "facts_retained": 29, "retention_rate": 0.4915254237288136, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.5714285714285714, "amount": 0.5, "status": 0.5833333333333334, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1341}, "ledger_by_hop": {"1": {"facts_total": 59, "facts_retained": 30, "retention_rate": 0.5084745762711864, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.5714285714285714, "amount": 0.6, "status": 0.5833333333333334, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1213}, "2": {"facts_total": 59, "facts_retained": 30, "retention_rate": 0.5084745762711864, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.5714285714285714, "amount": 0.6, "status": 0.5833333333333334, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1137}, "3": {"facts_total": 59, "facts_retained": 30, "retention_rate": 0.5084745762711864, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.5714285714285714, "amount": 0.6, "status": 0.5833333333333334, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1137}, "4": {"facts_total": 59, "facts_retained": 29, "retention_rate": 0.4915254237288136, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.5714285714285714, "amount": 0.5, "status": 0.5833333333333334, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1362}, "5": {"facts_total": 59, "facts_retained": 29, "retention_rate": 0.4915254237288136, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.5714285714285714, "amount": 0.5, "status": 0.5833333333333334, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1242}, "6": {"facts_total": 59, "facts_retained": 29, "retention_rate": 0.4915254237288136, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.5714285714285714, "amount": 0.5, "status": 0.5833333333333334, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1473}, "7": {"facts_total": 59, "facts_retained": 29, "retention_rate": 0.4915254237288136, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.5714285714285714, "amount": 0.5, "status": 0.5833333333333334, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1175}, "8": {"facts_total": 59, "facts_retained": 29, "retention_rate": 0.4915254237288136, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.5714285714285714, "amount": 0.5, "status": 0.5833333333333334, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1341}}}, "reference_key_id": "keys/synthetic/syn_03012_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "901314986d6cfc5eb94723973fc9ade34faba6815cd2a9c8d95879c4ad80b6ba", "output_hash": "66ef6c5da682b38795e9029723bbd71cdde2c4dec70cf1fae6efe62ef243f10a"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "8595c82252ea2afc0dc3968b", "attempt_id": "023077dfda6c", "run_group_id": "E4_core:A4|closed|qwen3b-llama3b:h8", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:44:39.964440Z", "task_family": "synthetic", "document_id": "syn_03012_2000", "document_hash": "03e441e68d54e1af036f8ff905dfd6cbbccead62636279c1a8e179352ab08945", "question_id": "q01_single_fact", "split": "test", "cluster_id": "syn_03012_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "tokenizer_hashes": {"receiver": "b9d8872cf3a99da99e1fc4b0f71232a451632dc49ca4c6771549b4efa5c1c763"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 8, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1767, "query_tokens": 507, "message_tokens": 398, "message_utf8_bytes": 1341, "cache_tensor_bytes": null, "serialized_payload_bytes": 1341, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 148535.66862398293, "hops": {"1": 23159.84900004696, "2": 16642.98779098317, "3": 21655.61054192949, "4": 16671.246667043306, "5": 19499.049708014354, "6": 16652.313041035086, "7": 18373.46841697581, "8": 15881.14345795475}, "receiver_prefill_ms": 338.7437079800293, "ttft_ms": 39.88483303692192, "decode_ms": 44.432374997995794}, "total_wall_ms": 148959.34658194892, "ttft_ms": 39.88483303692192, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 59, "facts_retained": 29, "retention_rate": 0.4915254237288136, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.5714285714285714, "amount": 0.5, "status": 0.5833333333333334, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1341}, "ledger_by_hop": {"1": {"facts_total": 59, "facts_retained": 30, "retention_rate": 0.5084745762711864, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.5714285714285714, "amount": 0.6, "status": 0.5833333333333334, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1213}, "2": {"facts_total": 59, "facts_retained": 30, "retention_rate": 0.5084745762711864, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.5714285714285714, "amount": 0.6, "status": 0.5833333333333334, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1137}, "3": {"facts_total": 59, "facts_retained": 30, "retention_rate": 0.5084745762711864, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.5714285714285714, "amount": 0.6, "status": 0.5833333333333334, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1137}, "4": {"facts_total": 59, "facts_retained": 29, "retention_rate": 0.4915254237288136, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.5714285714285714, "amount": 0.5, "status": 0.5833333333333334, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1362}, "5": {"facts_total": 59, "facts_retained": 29, "retention_rate": 0.4915254237288136, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.5714285714285714, "amount": 0.5, "status": 0.5833333333333334, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1242}, "6": {"facts_total": 59, "facts_retained": 29, "retention_rate": 0.4915254237288136, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.5714285714285714, "amount": 0.5, "status": 0.5833333333333334, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1473}, "7": {"facts_total": 59, "facts_retained": 29, "retention_rate": 0.4915254237288136, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.5714285714285714, "amount": 0.5, "status": 0.5833333333333334, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1175}, "8": {"facts_total": 59, "facts_retained": 29, "retention_rate": 0.4915254237288136, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.5714285714285714, "amount": 0.5, "status": 0.5833333333333334, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1341}}}, "reference_key_id": "keys/synthetic/syn_03012_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "901314986d6cfc5eb94723973fc9ade34faba6815cd2a9c8d95879c4ad80b6ba", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "bd5e567c690c45c4e7a2f46d", "attempt_id": "ce4ee63cd19c", "run_group_id": "E4_core:A4|closed|qwen3b-llama3b:h8", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:44:40.405779Z", "task_family": "synthetic", "document_id": "syn_03012_2000", "document_hash": "03e441e68d54e1af036f8ff905dfd6cbbccead62636279c1a8e179352ab08945", "question_id": "q02_exact_copy", "split": "test", "cluster_id": "syn_03012_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "tokenizer_hashes": {"receiver": "b9d8872cf3a99da99e1fc4b0f71232a451632dc49ca4c6771549b4efa5c1c763"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 8, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1767, "query_tokens": 507, "message_tokens": 398, "message_utf8_bytes": 1341, "cache_tensor_bytes": null, "serialized_payload_bytes": 1341, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 148535.66862398293, "hops": {"1": 23159.84900004696, "2": 16642.98779098317, "3": 21655.61054192949, "4": 16671.246667043306, "5": 19499.049708014354, "6": 16652.313041035086, "7": 18373.46841697581, "8": 15881.14345795475}, "receiver_prefill_ms": 521.6074590571225, "ttft_ms": 26.138958055526018, "decode_ms": 137.27095897775143}, "total_wall_ms": 149221.36595693883, "ttft_ms": 26.138958055526018, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 4}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "13.4343", "evidence_artifact": {"ledger": {"facts_total": 59, "facts_retained": 29, "retention_rate": 0.4915254237288136, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.5714285714285714, "amount": 0.5, "status": 0.5833333333333334, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1341}, "ledger_by_hop": {"1": {"facts_total": 59, "facts_retained": 30, "retention_rate": 0.5084745762711864, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.5714285714285714, "amount": 0.6, "status": 0.5833333333333334, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1213}, "2": {"facts_total": 59, "facts_retained": 30, "retention_rate": 0.5084745762711864, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.5714285714285714, "amount": 0.6, "status": 0.5833333333333334, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1137}, "3": {"facts_total": 59, "facts_retained": 30, "retention_rate": 0.5084745762711864, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.5714285714285714, "amount": 0.6, "status": 0.5833333333333334, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1137}, "4": {"facts_total": 59, "facts_retained": 29, "retention_rate": 0.4915254237288136, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.5714285714285714, "amount": 0.5, "status": 0.5833333333333334, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1362}, "5": {"facts_total": 59, "facts_retained": 29, "retention_rate": 0.4915254237288136, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.5714285714285714, "amount": 0.5, "status": 0.5833333333333334, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1242}, "6": {"facts_total": 59, "facts_retained": 29, "retention_rate": 0.4915254237288136, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.5714285714285714, "amount": 0.5, "status": 0.5833333333333334, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1473}, "7": {"facts_total": 59, "facts_retained": 29, "retention_rate": 0.4915254237288136, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.5714285714285714, "amount": 0.5, "status": 0.5833333333333334, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1175}, "8": {"facts_total": 59, "facts_retained": 29, "retention_rate": 0.4915254237288136, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.5714285714285714, "amount": 0.5, "status": 0.5833333333333334, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1341}}}, "reference_key_id": "keys/synthetic/syn_03012_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "901314986d6cfc5eb94723973fc9ade34faba6815cd2a9c8d95879c4ad80b6ba", "output_hash": "599bc22e373b1cf7b4679208eb9e515ad0e8dcb88dde0c7f579637ec7d3f3014"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "e75ec12d6636be146812fe08", "attempt_id": "baa0cc95a4bd", "run_group_id": "E4_core:A4|closed|qwen3b-llama3b:h8", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:44:41.110872Z", "task_family": "synthetic", "document_id": "syn_03012_2000", "document_hash": "03e441e68d54e1af036f8ff905dfd6cbbccead62636279c1a8e179352ab08945", "question_id": "q03_exact_copy", "split": "test", "cluster_id": "syn_03012_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "tokenizer_hashes": {"receiver": "b9d8872cf3a99da99e1fc4b0f71232a451632dc49ca4c6771549b4efa5c1c763"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 8, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1767, "query_tokens": 503, "message_tokens": 398, "message_utf8_bytes": 1341, "cache_tensor_bytes": null, "serialized_payload_bytes": 1341, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 148535.66862398293, "hops": {"1": 23159.84900004696, "2": 16642.98779098317, "3": 21655.61054192949, "4": 16671.246667043306, "5": 19499.049708014354, "6": 16652.313041035086, "7": 18373.46841697581, "8": 15881.14345795475}, "receiver_prefill_ms": 366.8627499137074, "ttft_ms": 38.04000001400709, "decode_ms": 26.884958031587303}, "total_wall_ms": 148968.06624904275, "ttft_ms": 38.04000001400709, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 59, "facts_retained": 29, "retention_rate": 0.4915254237288136, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.5714285714285714, "amount": 0.5, "status": 0.5833333333333334, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1341}, "ledger_by_hop": {"1": {"facts_total": 59, "facts_retained": 30, "retention_rate": 0.5084745762711864, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.5714285714285714, "amount": 0.6, "status": 0.5833333333333334, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1213}, "2": {"facts_total": 59, "facts_retained": 30, "retention_rate": 0.5084745762711864, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.5714285714285714, "amount": 0.6, "status": 0.5833333333333334, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1137}, "3": {"facts_total": 59, "facts_retained": 30, "retention_rate": 0.5084745762711864, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.5714285714285714, "amount": 0.6, "status": 0.5833333333333334, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1137}, "4": {"facts_total": 59, "facts_retained": 29, "retention_rate": 0.4915254237288136, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.5714285714285714, "amount": 0.5, "status": 0.5833333333333334, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1362}, "5": {"facts_total": 59, "facts_retained": 29, "retention_rate": 0.4915254237288136, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.5714285714285714, "amount": 0.5, "status": 0.5833333333333334, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1242}, "6": {"facts_total": 59, "facts_retained": 29, "retention_rate": 0.4915254237288136, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.5714285714285714, "amount": 0.5, "status": 0.5833333333333334, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1473}, "7": {"facts_total": 59, "facts_retained": 29, "retention_rate": 0.4915254237288136, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.5714285714285714, "amount": 0.5, "status": 0.5833333333333334, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1175}, "8": {"facts_total": 59, "facts_retained": 29, "retention_rate": 0.4915254237288136, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.5714285714285714, "amount": 0.5, "status": 0.5833333333333334, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1341}}}, "reference_key_id": "keys/synthetic/syn_03012_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "901314986d6cfc5eb94723973fc9ade34faba6815cd2a9c8d95879c4ad80b6ba", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "343e1cda86e57a15fcb51e09", "attempt_id": "590aa369e711", "run_group_id": "E4_core:A4|closed|qwen3b-llama3b:h8", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:44:41.561400Z", "task_family": "synthetic", "document_id": "syn_03012_2000", "document_hash": "03e441e68d54e1af036f8ff905dfd6cbbccead62636279c1a8e179352ab08945", "question_id": "q04_multi_hop", "split": "test", "cluster_id": "syn_03012_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "tokenizer_hashes": {"receiver": "b9d8872cf3a99da99e1fc4b0f71232a451632dc49ca4c6771549b4efa5c1c763"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 8, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1767, "query_tokens": 508, "message_tokens": 398, "message_utf8_bytes": 1341, "cache_tensor_bytes": null, "serialized_payload_bytes": 1341, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 148535.66862398293, "hops": {"1": 23159.84900004696, "2": 16642.98779098317, "3": 21655.61054192949, "4": 16671.246667043306, "5": 19499.049708014354, "6": 16652.313041035086, "7": 18373.46841697581, "8": 15881.14345795475}, "receiver_prefill_ms": 347.69908408634365, "ttft_ms": 37.347499979659915, "decode_ms": 58.41662501916289}, "total_wall_ms": 148979.83474901412, "ttft_ms": 37.347499979659915, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 59, "facts_retained": 29, "retention_rate": 0.4915254237288136, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.5714285714285714, "amount": 0.5, "status": 0.5833333333333334, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1341}, "ledger_by_hop": {"1": {"facts_total": 59, "facts_retained": 30, "retention_rate": 0.5084745762711864, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.5714285714285714, "amount": 0.6, "status": 0.5833333333333334, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1213}, "2": {"facts_total": 59, "facts_retained": 30, "retention_rate": 0.5084745762711864, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.5714285714285714, "amount": 0.6, "status": 0.5833333333333334, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1137}, "3": {"facts_total": 59, "facts_retained": 30, "retention_rate": 0.5084745762711864, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.5714285714285714, "amount": 0.6, "status": 0.5833333333333334, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1137}, "4": {"facts_total": 59, "facts_retained": 29, "retention_rate": 0.4915254237288136, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.5714285714285714, "amount": 0.5, "status": 0.5833333333333334, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1362}, "5": {"facts_total": 59, "facts_retained": 29, "retention_rate": 0.4915254237288136, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.5714285714285714, "amount": 0.5, "status": 0.5833333333333334, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1242}, "6": {"facts_total": 59, "facts_retained": 29, "retention_rate": 0.4915254237288136, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.5714285714285714, "amount": 0.5, "status": 0.5833333333333334, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1473}, "7": {"facts_total": 59, "facts_retained": 29, "retention_rate": 0.4915254237288136, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.5714285714285714, "amount": 0.5, "status": 0.5833333333333334, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1175}, "8": {"facts_total": 59, "facts_retained": 29, "retention_rate": 0.4915254237288136, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.5714285714285714, "amount": 0.5, "status": 0.5833333333333334, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1341}}}, "reference_key_id": "keys/synthetic/syn_03012_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "901314986d6cfc5eb94723973fc9ade34faba6815cd2a9c8d95879c4ad80b6ba", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "d9cc82e4f8e18d2fc8ae3954", "attempt_id": "9d217b308b87", "run_group_id": "E4_core:A4|closed|qwen3b-llama3b:h8", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:44:42.023113Z", "task_family": "synthetic", "document_id": "syn_03012_2000", "document_hash": "03e441e68d54e1af036f8ff905dfd6cbbccead62636279c1a8e179352ab08945", "question_id": "q05_multi_hop", "split": "test", "cluster_id": "syn_03012_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "tokenizer_hashes": {"receiver": "b9d8872cf3a99da99e1fc4b0f71232a451632dc49ca4c6771549b4efa5c1c763"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 8, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1767, "query_tokens": 509, "message_tokens": 398, "message_utf8_bytes": 1341, "cache_tensor_bytes": null, "serialized_payload_bytes": 1341, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 148535.66862398293, "hops": {"1": 23159.84900004696, "2": 16642.98779098317, "3": 21655.61054192949, "4": 16671.246667043306, "5": 19499.049708014354, "6": 16652.313041035086, "7": 18373.46841697581, "8": 15881.14345795475}, "receiver_prefill_ms": 523.2056670356542, "ttft_ms": 24.615874979645014, "decode_ms": 227.96712501440197}, "total_wall_ms": 149312.14345805347, "ttft_ms": 24.615874979645014, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 7}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "DG-40E-2777", "evidence_artifact": {"ledger": {"facts_total": 59, "facts_retained": 29, "retention_rate": 0.4915254237288136, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.5714285714285714, "amount": 0.5, "status": 0.5833333333333334, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1341}, "ledger_by_hop": {"1": {"facts_total": 59, "facts_retained": 30, "retention_rate": 0.5084745762711864, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.5714285714285714, "amount": 0.6, "status": 0.5833333333333334, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1213}, "2": {"facts_total": 59, "facts_retained": 30, "retention_rate": 0.5084745762711864, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.5714285714285714, "amount": 0.6, "status": 0.5833333333333334, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1137}, "3": {"facts_total": 59, "facts_retained": 30, "retention_rate": 0.5084745762711864, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.5714285714285714, "amount": 0.6, "status": 0.5833333333333334, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1137}, "4": {"facts_total": 59, "facts_retained": 29, "retention_rate": 0.4915254237288136, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.5714285714285714, "amount": 0.5, "status": 0.5833333333333334, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1362}, "5": {"facts_total": 59, "facts_retained": 29, "retention_rate": 0.4915254237288136, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.5714285714285714, "amount": 0.5, "status": 0.5833333333333334, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1242}, "6": {"facts_total": 59, "facts_retained": 29, "retention_rate": 0.4915254237288136, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.5714285714285714, "amount": 0.5, "status": 0.5833333333333334, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1473}, "7": {"facts_total": 59, "facts_retained": 29, "retention_rate": 0.4915254237288136, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.5714285714285714, "amount": 0.5, "status": 0.5833333333333334, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1175}, "8": {"facts_total": 59, "facts_retained": 29, "retention_rate": 0.4915254237288136, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.5714285714285714, "amount": 0.5, "status": 0.5833333333333334, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1341}}}, "reference_key_id": "keys/synthetic/syn_03012_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "901314986d6cfc5eb94723973fc9ade34faba6815cd2a9c8d95879c4ad80b6ba", "output_hash": "16a7e3f4a7cc4a9b66f86c2cdcfad3b35daf1da199d9406e0627c8af2d8d73f3"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "4f4f5b357d9821d8c9a873ac", "attempt_id": "c6e0ad6328ec", "run_group_id": "E4_core:A4|closed|qwen3b-llama3b:h8", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:44:42.817713Z", "task_family": "synthetic", "document_id": "syn_03012_2000", "document_hash": "03e441e68d54e1af036f8ff905dfd6cbbccead62636279c1a8e179352ab08945", "question_id": "q06_negation", "split": "test", "cluster_id": "syn_03012_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "tokenizer_hashes": {"receiver": "b9d8872cf3a99da99e1fc4b0f71232a451632dc49ca4c6771549b4efa5c1c763"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 8, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1767, "query_tokens": 502, "message_tokens": 398, "message_utf8_bytes": 1341, "cache_tensor_bytes": null, "serialized_payload_bytes": 1341, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 148535.66862398293, "hops": {"1": 23159.84900004696, "2": 16642.98779098317, "3": 21655.61054192949, "4": 16671.246667043306, "5": 19499.049708014354, "6": 16652.313041035086, "7": 18373.46841697581, "8": 15881.14345795475}, "receiver_prefill_ms": 377.27683305274695, "ttft_ms": 19.915084005333483, "decode_ms": 60.15316699631512}, "total_wall_ms": 148993.63479099702, "ttft_ms": 19.915084005333483, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 2}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "No.", "evidence_artifact": {"ledger": {"facts_total": 59, "facts_retained": 29, "retention_rate": 0.4915254237288136, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.5714285714285714, "amount": 0.5, "status": 0.5833333333333334, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1341}, "ledger_by_hop": {"1": {"facts_total": 59, "facts_retained": 30, "retention_rate": 0.5084745762711864, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.5714285714285714, "amount": 0.6, "status": 0.5833333333333334, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1213}, "2": {"facts_total": 59, "facts_retained": 30, "retention_rate": 0.5084745762711864, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.5714285714285714, "amount": 0.6, "status": 0.5833333333333334, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1137}, "3": {"facts_total": 59, "facts_retained": 30, "retention_rate": 0.5084745762711864, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.5714285714285714, "amount": 0.6, "status": 0.5833333333333334, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1137}, "4": {"facts_total": 59, "facts_retained": 29, "retention_rate": 0.4915254237288136, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.5714285714285714, "amount": 0.5, "status": 0.5833333333333334, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1362}, "5": {"facts_total": 59, "facts_retained": 29, "retention_rate": 0.4915254237288136, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.5714285714285714, "amount": 0.5, "status": 0.5833333333333334, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1242}, "6": {"facts_total": 59, "facts_retained": 29, "retention_rate": 0.4915254237288136, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.5714285714285714, "amount": 0.5, "status": 0.5833333333333334, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1473}, "7": {"facts_total": 59, "facts_retained": 29, "retention_rate": 0.4915254237288136, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.5714285714285714, "amount": 0.5, "status": 0.5833333333333334, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1175}, "8": {"facts_total": 59, "facts_retained": 29, "retention_rate": 0.4915254237288136, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.5714285714285714, "amount": 0.5, "status": 0.5833333333333334, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1341}}}, "reference_key_id": "keys/synthetic/syn_03012_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "901314986d6cfc5eb94723973fc9ade34faba6815cd2a9c8d95879c4ad80b6ba", "output_hash": "38178a20b470cfd18299fb1593dd4ba706f1a83321dac051d01086efd8b7a96f"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "d293102a02e8475efbaedc79", "attempt_id": "8a2b611ca0c5", "run_group_id": "E4_core:A4|closed|qwen3b-llama3b:h8", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:44:43.293277Z", "task_family": "synthetic", "document_id": "syn_03012_2000", "document_hash": "03e441e68d54e1af036f8ff905dfd6cbbccead62636279c1a8e179352ab08945", "question_id": "q07_negation", "split": "test", "cluster_id": "syn_03012_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "tokenizer_hashes": {"receiver": "b9d8872cf3a99da99e1fc4b0f71232a451632dc49ca4c6771549b4efa5c1c763"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 8, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1767, "query_tokens": 503, "message_tokens": 398, "message_utf8_bytes": 1341, "cache_tensor_bytes": null, "serialized_payload_bytes": 1341, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 148535.66862398293, "hops": {"1": 23159.84900004696, "2": 16642.98779098317, "3": 21655.61054192949, "4": 16671.246667043306, "5": 19499.049708014354, "6": 16652.313041035086, "7": 18373.46841697581, "8": 15881.14345795475}, "receiver_prefill_ms": 342.8816660307348, "ttft_ms": 38.59087498858571, "decode_ms": 120.9728749236092}, "total_wall_ms": 149038.75137399882, "ttft_ms": 38.59087498858571, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 2}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "No.", "evidence_artifact": {"ledger": {"facts_total": 59, "facts_retained": 29, "retention_rate": 0.4915254237288136, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.5714285714285714, "amount": 0.5, "status": 0.5833333333333334, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1341}, "ledger_by_hop": {"1": {"facts_total": 59, "facts_retained": 30, "retention_rate": 0.5084745762711864, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.5714285714285714, "amount": 0.6, "status": 0.5833333333333334, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1213}, "2": {"facts_total": 59, "facts_retained": 30, "retention_rate": 0.5084745762711864, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.5714285714285714, "amount": 0.6, "status": 0.5833333333333334, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1137}, "3": {"facts_total": 59, "facts_retained": 30, "retention_rate": 0.5084745762711864, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.5714285714285714, "amount": 0.6, "status": 0.5833333333333334, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1137}, "4": {"facts_total": 59, "facts_retained": 29, "retention_rate": 0.4915254237288136, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.5714285714285714, "amount": 0.5, "status": 0.5833333333333334, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1362}, "5": {"facts_total": 59, "facts_retained": 29, "retention_rate": 0.4915254237288136, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.5714285714285714, "amount": 0.5, "status": 0.5833333333333334, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1242}, "6": {"facts_total": 59, "facts_retained": 29, "retention_rate": 0.4915254237288136, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.5714285714285714, "amount": 0.5, "status": 0.5833333333333334, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1473}, "7": {"facts_total": 59, "facts_retained": 29, "retention_rate": 0.4915254237288136, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.5714285714285714, "amount": 0.5, "status": 0.5833333333333334, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1175}, "8": {"facts_total": 59, "facts_retained": 29, "retention_rate": 0.4915254237288136, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.5714285714285714, "amount": 0.5, "status": 0.5833333333333334, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1341}}}, "reference_key_id": "keys/synthetic/syn_03012_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "901314986d6cfc5eb94723973fc9ade34faba6815cd2a9c8d95879c4ad80b6ba", "output_hash": "38178a20b470cfd18299fb1593dd4ba706f1a83321dac051d01086efd8b7a96f"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "a4912ebd88fb17915999e2a5", "attempt_id": "2cfcaa0a8b72", "run_group_id": "E4_core:A4|closed|qwen3b-llama3b:h8", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:44:43.815459Z", "task_family": "synthetic", "document_id": "syn_03012_2000", "document_hash": "03e441e68d54e1af036f8ff905dfd6cbbccead62636279c1a8e179352ab08945", "question_id": "q08_revision", "split": "test", "cluster_id": "syn_03012_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "tokenizer_hashes": {"receiver": "b9d8872cf3a99da99e1fc4b0f71232a451632dc49ca4c6771549b4efa5c1c763"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 8, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1767, "query_tokens": 518, "message_tokens": 398, "message_utf8_bytes": 1341, "cache_tensor_bytes": null, "serialized_payload_bytes": 1341, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 148535.66862398293, "hops": {"1": 23159.84900004696, "2": 16642.98779098317, "3": 21655.61054192949, "4": 16671.246667043306, "5": 19499.049708014354, "6": 16652.313041035086, "7": 18373.46841697581, "8": 15881.14345795475}, "receiver_prefill_ms": 469.11341696977615, "ttft_ms": 38.20599999744445, "decode_ms": 151.92912507336587}, "total_wall_ms": 149195.66529092845, "ttft_ms": 38.20599999744445, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 5}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "6,602,245", "evidence_artifact": {"ledger": {"facts_total": 59, "facts_retained": 29, "retention_rate": 0.4915254237288136, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.5714285714285714, "amount": 0.5, "status": 0.5833333333333334, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1341}, "ledger_by_hop": {"1": {"facts_total": 59, "facts_retained": 30, "retention_rate": 0.5084745762711864, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.5714285714285714, "amount": 0.6, "status": 0.5833333333333334, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1213}, "2": {"facts_total": 59, "facts_retained": 30, "retention_rate": 0.5084745762711864, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.5714285714285714, "amount": 0.6, "status": 0.5833333333333334, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1137}, "3": {"facts_total": 59, "facts_retained": 30, "retention_rate": 0.5084745762711864, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.5714285714285714, "amount": 0.6, "status": 0.5833333333333334, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1137}, "4": {"facts_total": 59, "facts_retained": 29, "retention_rate": 0.4915254237288136, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.5714285714285714, "amount": 0.5, "status": 0.5833333333333334, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1362}, "5": {"facts_total": 59, "facts_retained": 29, "retention_rate": 0.4915254237288136, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.5714285714285714, "amount": 0.5, "status": 0.5833333333333334, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1242}, "6": {"facts_total": 59, "facts_retained": 29, "retention_rate": 0.4915254237288136, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.5714285714285714, "amount": 0.5, "status": 0.5833333333333334, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1473}, "7": {"facts_total": 59, "facts_retained": 29, "retention_rate": 0.4915254237288136, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.5714285714285714, "amount": 0.5, "status": 0.5833333333333334, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1175}, "8": {"facts_total": 59, "facts_retained": 29, "retention_rate": 0.4915254237288136, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.5714285714285714, "amount": 0.5, "status": 0.5833333333333334, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1341}}}, "reference_key_id": "keys/synthetic/syn_03012_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "901314986d6cfc5eb94723973fc9ade34faba6815cd2a9c8d95879c4ad80b6ba", "output_hash": "f121947078d464ab91a83739861b98c336a10c001704e2df84cc2b041bb8f4de"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "407c82ad0d58b907ef9e7ba5", "attempt_id": "d569015cd9f2", "run_group_id": "E4_core:A4|closed|qwen3b-llama3b:h8", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:44:44.493477Z", "task_family": "synthetic", "document_id": "syn_03012_2000", "document_hash": "03e441e68d54e1af036f8ff905dfd6cbbccead62636279c1a8e179352ab08945", "question_id": "q09_revision", "split": "test", "cluster_id": "syn_03012_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "tokenizer_hashes": {"receiver": "b9d8872cf3a99da99e1fc4b0f71232a451632dc49ca4c6771549b4efa5c1c763"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 8, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1767, "query_tokens": 510, "message_tokens": 398, "message_utf8_bytes": 1341, "cache_tensor_bytes": null, "serialized_payload_bytes": 1341, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 148535.66862398293, "hops": {"1": 23159.84900004696, "2": 16642.98779098317, "3": 21655.61054192949, "4": 16671.246667043306, "5": 19499.049708014354, "6": 16652.313041035086, "7": 18373.46841697581, "8": 15881.14345795475}, "receiver_prefill_ms": 373.7109999638051, "ttft_ms": 29.944875044748187, "decode_ms": 29.442082974128425}, "total_wall_ms": 148969.41012400202, "ttft_ms": 29.944875044748187, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 59, "facts_retained": 29, "retention_rate": 0.4915254237288136, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.5714285714285714, "amount": 0.5, "status": 0.5833333333333334, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1341}, "ledger_by_hop": {"1": {"facts_total": 59, "facts_retained": 30, "retention_rate": 0.5084745762711864, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.5714285714285714, "amount": 0.6, "status": 0.5833333333333334, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1213}, "2": {"facts_total": 59, "facts_retained": 30, "retention_rate": 0.5084745762711864, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.5714285714285714, "amount": 0.6, "status": 0.5833333333333334, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1137}, "3": {"facts_total": 59, "facts_retained": 30, "retention_rate": 0.5084745762711864, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.5714285714285714, "amount": 0.6, "status": 0.5833333333333334, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1137}, "4": {"facts_total": 59, "facts_retained": 29, "retention_rate": 0.4915254237288136, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.5714285714285714, "amount": 0.5, "status": 0.5833333333333334, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1362}, "5": {"facts_total": 59, "facts_retained": 29, "retention_rate": 0.4915254237288136, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.5714285714285714, "amount": 0.5, "status": 0.5833333333333334, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1242}, "6": {"facts_total": 59, "facts_retained": 29, "retention_rate": 0.4915254237288136, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.5714285714285714, "amount": 0.5, "status": 0.5833333333333334, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1473}, "7": {"facts_total": 59, "facts_retained": 29, "retention_rate": 0.4915254237288136, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.5714285714285714, "amount": 0.5, "status": 0.5833333333333334, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1175}, "8": {"facts_total": 59, "facts_retained": 29, "retention_rate": 0.4915254237288136, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.5714285714285714, "amount": 0.5, "status": 0.5833333333333334, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1341}}}, "reference_key_id": "keys/synthetic/syn_03012_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "901314986d6cfc5eb94723973fc9ade34faba6815cd2a9c8d95879c4ad80b6ba", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "94590639b3d117782fff794d", "attempt_id": "00bcc36164ab", "run_group_id": "E4_core:A4|closed|qwen3b-llama3b:h8", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:44:44.944606Z", "task_family": "synthetic", "document_id": "syn_03012_2000", "document_hash": "03e441e68d54e1af036f8ff905dfd6cbbccead62636279c1a8e179352ab08945", "question_id": "q10_temporal", "split": "test", "cluster_id": "syn_03012_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "tokenizer_hashes": {"receiver": "b9d8872cf3a99da99e1fc4b0f71232a451632dc49ca4c6771549b4efa5c1c763"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 8, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1767, "query_tokens": 520, "message_tokens": 398, "message_utf8_bytes": 1341, "cache_tensor_bytes": null, "serialized_payload_bytes": 1341, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 148535.66862398293, "hops": {"1": 23159.84900004696, "2": 16642.98779098317, "3": 21655.61054192949, "4": 16671.246667043306, "5": 19499.049708014354, "6": 16652.313041035086, "7": 18373.46841697581, "8": 15881.14345795475}, "receiver_prefill_ms": 323.6860829638317, "ttft_ms": 61.34812498930842, "decode_ms": 49.74083404522389}, "total_wall_ms": 148971.48274898063, "ttft_ms": 61.34812498930842, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 59, "facts_retained": 29, "retention_rate": 0.4915254237288136, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.5714285714285714, "amount": 0.5, "status": 0.5833333333333334, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1341}, "ledger_by_hop": {"1": {"facts_total": 59, "facts_retained": 30, "retention_rate": 0.5084745762711864, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.5714285714285714, "amount": 0.6, "status": 0.5833333333333334, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1213}, "2": {"facts_total": 59, "facts_retained": 30, "retention_rate": 0.5084745762711864, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.5714285714285714, "amount": 0.6, "status": 0.5833333333333334, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1137}, "3": {"facts_total": 59, "facts_retained": 30, "retention_rate": 0.5084745762711864, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.5714285714285714, "amount": 0.6, "status": 0.5833333333333334, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1137}, "4": {"facts_total": 59, "facts_retained": 29, "retention_rate": 0.4915254237288136, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.5714285714285714, "amount": 0.5, "status": 0.5833333333333334, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1362}, "5": {"facts_total": 59, "facts_retained": 29, "retention_rate": 0.4915254237288136, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.5714285714285714, "amount": 0.5, "status": 0.5833333333333334, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1242}, "6": {"facts_total": 59, "facts_retained": 29, "retention_rate": 0.4915254237288136, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.5714285714285714, "amount": 0.5, "status": 0.5833333333333334, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1473}, "7": {"facts_total": 59, "facts_retained": 29, "retention_rate": 0.4915254237288136, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.5714285714285714, "amount": 0.5, "status": 0.5833333333333334, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1175}, "8": {"facts_total": 59, "facts_retained": 29, "retention_rate": 0.4915254237288136, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.5714285714285714, "amount": 0.5, "status": 0.5833333333333334, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1341}}}, "reference_key_id": "keys/synthetic/syn_03012_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "901314986d6cfc5eb94723973fc9ade34faba6815cd2a9c8d95879c4ad80b6ba", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "8871136e41591384679b4a82", "attempt_id": "0bfa39e6a041", "run_group_id": "E4_core:A4|closed|qwen3b-llama3b:h8", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:44:45.397698Z", "task_family": "synthetic", "document_id": "syn_03012_2000", "document_hash": "03e441e68d54e1af036f8ff905dfd6cbbccead62636279c1a8e179352ab08945", "question_id": "q11_aggregation", "split": "test", "cluster_id": "syn_03012_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "tokenizer_hashes": {"receiver": "b9d8872cf3a99da99e1fc4b0f71232a451632dc49ca4c6771549b4efa5c1c763"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 8, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1767, "query_tokens": 513, "message_tokens": 398, "message_utf8_bytes": 1341, "cache_tensor_bytes": null, "serialized_payload_bytes": 1341, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 148535.66862398293, "hops": {"1": 23159.84900004696, "2": 16642.98779098317, "3": 21655.61054192949, "4": 16671.246667043306, "5": 19499.049708014354, "6": 16652.313041035086, "7": 18373.46841697581, "8": 15881.14345795475}, "receiver_prefill_ms": 457.8341250307858, "ttft_ms": 31.331082922406495, "decode_ms": 164.74837495479733}, "total_wall_ms": 149190.21349889226, "ttft_ms": 31.331082922406495, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 5}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "9,381,835", "evidence_artifact": {"ledger": {"facts_total": 59, "facts_retained": 29, "retention_rate": 0.4915254237288136, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.5714285714285714, "amount": 0.5, "status": 0.5833333333333334, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1341}, "ledger_by_hop": {"1": {"facts_total": 59, "facts_retained": 30, "retention_rate": 0.5084745762711864, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.5714285714285714, "amount": 0.6, "status": 0.5833333333333334, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1213}, "2": {"facts_total": 59, "facts_retained": 30, "retention_rate": 0.5084745762711864, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.5714285714285714, "amount": 0.6, "status": 0.5833333333333334, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1137}, "3": {"facts_total": 59, "facts_retained": 30, "retention_rate": 0.5084745762711864, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.5714285714285714, "amount": 0.6, "status": 0.5833333333333334, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1137}, "4": {"facts_total": 59, "facts_retained": 29, "retention_rate": 0.4915254237288136, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.5714285714285714, "amount": 0.5, "status": 0.5833333333333334, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1362}, "5": {"facts_total": 59, "facts_retained": 29, "retention_rate": 0.4915254237288136, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.5714285714285714, "amount": 0.5, "status": 0.5833333333333334, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1242}, "6": {"facts_total": 59, "facts_retained": 29, "retention_rate": 0.4915254237288136, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.5714285714285714, "amount": 0.5, "status": 0.5833333333333334, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1473}, "7": {"facts_total": 59, "facts_retained": 29, "retention_rate": 0.4915254237288136, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.5714285714285714, "amount": 0.5, "status": 0.5833333333333334, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1175}, "8": {"facts_total": 59, "facts_retained": 29, "retention_rate": 0.4915254237288136, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.5714285714285714, "amount": 0.5, "status": 0.5833333333333334, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1341}}}, "reference_key_id": "keys/synthetic/syn_03012_2000.json", "task_scores": {"correct": 0.0, "abstained": 0.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "901314986d6cfc5eb94723973fc9ade34faba6815cd2a9c8d95879c4ad80b6ba", "output_hash": "5f7b503199558e57bee95e5db837c17c224702f5159f2fcfb4effe6fed47fcba"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "8485478c1678fba6cd691b1e", "attempt_id": "a0e41b465e97", "run_group_id": "E4_core:A4|closed|qwen3b-llama3b:h8", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:44:46.069655Z", "task_family": "synthetic", "document_id": "syn_03012_2000", "document_hash": "03e441e68d54e1af036f8ff905dfd6cbbccead62636279c1a8e179352ab08945", "question_id": "q12_unit_conversion", "split": "test", "cluster_id": "syn_03012_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "tokenizer_hashes": {"receiver": "b9d8872cf3a99da99e1fc4b0f71232a451632dc49ca4c6771549b4efa5c1c763"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 8, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1767, "query_tokens": 510, "message_tokens": 398, "message_utf8_bytes": 1341, "cache_tensor_bytes": null, "serialized_payload_bytes": 1341, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 148535.66862398293, "hops": {"1": 23159.84900004696, "2": 16642.98779098317, "3": 21655.61054192949, "4": 16671.246667043306, "5": 19499.049708014354, "6": 16652.313041035086, "7": 18373.46841697581, "8": 15881.14345795475}, "receiver_prefill_ms": 388.8184579554945, "ttft_ms": 26.739708031527698, "decode_ms": 91.37029107660055}, "total_wall_ms": 149043.25141594745, "ttft_ms": 26.739708031527698, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 3}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "44.62", "evidence_artifact": {"ledger": {"facts_total": 59, "facts_retained": 29, "retention_rate": 0.4915254237288136, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.5714285714285714, "amount": 0.5, "status": 0.5833333333333334, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1341}, "ledger_by_hop": {"1": {"facts_total": 59, "facts_retained": 30, "retention_rate": 0.5084745762711864, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.5714285714285714, "amount": 0.6, "status": 0.5833333333333334, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1213}, "2": {"facts_total": 59, "facts_retained": 30, "retention_rate": 0.5084745762711864, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.5714285714285714, "amount": 0.6, "status": 0.5833333333333334, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1137}, "3": {"facts_total": 59, "facts_retained": 30, "retention_rate": 0.5084745762711864, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.5714285714285714, "amount": 0.6, "status": 0.5833333333333334, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1137}, "4": {"facts_total": 59, "facts_retained": 29, "retention_rate": 0.4915254237288136, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.5714285714285714, "amount": 0.5, "status": 0.5833333333333334, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1362}, "5": {"facts_total": 59, "facts_retained": 29, "retention_rate": 0.4915254237288136, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.5714285714285714, "amount": 0.5, "status": 0.5833333333333334, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1242}, "6": {"facts_total": 59, "facts_retained": 29, "retention_rate": 0.4915254237288136, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.5714285714285714, "amount": 0.5, "status": 0.5833333333333334, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1473}, "7": {"facts_total": 59, "facts_retained": 29, "retention_rate": 0.4915254237288136, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.5714285714285714, "amount": 0.5, "status": 0.5833333333333334, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1175}, "8": {"facts_total": 59, "facts_retained": 29, "retention_rate": 0.4915254237288136, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.5714285714285714, "amount": 0.5, "status": 0.5833333333333334, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1341}}}, "reference_key_id": "keys/synthetic/syn_03012_2000.json", "task_scores": {"correct": 0.0, "abstained": 0.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "901314986d6cfc5eb94723973fc9ade34faba6815cd2a9c8d95879c4ad80b6ba", "output_hash": "d9a3265d0a78db3d082616021f44cd388888ddddf0a7328ce458e95e7824f8e9"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "9813aebf35cc20f4ee0b7bf7", "attempt_id": "2e32934b3746", "run_group_id": "E4_core:A4|closed|qwen3b-llama3b:h8", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:44:46.595485Z", "task_family": "synthetic", "document_id": "syn_03012_2000", "document_hash": "03e441e68d54e1af036f8ff905dfd6cbbccead62636279c1a8e179352ab08945", "question_id": "q13_unanswerable", "split": "test", "cluster_id": "syn_03012_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "tokenizer_hashes": {"receiver": "b9d8872cf3a99da99e1fc4b0f71232a451632dc49ca4c6771549b4efa5c1c763"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 8, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1767, "query_tokens": 514, "message_tokens": 398, "message_utf8_bytes": 1341, "cache_tensor_bytes": null, "serialized_payload_bytes": 1341, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 148535.66862398293, "hops": {"1": 23159.84900004696, "2": 16642.98779098317, "3": 21655.61054192949, "4": 16671.246667043306, "5": 19499.049708014354, "6": 16652.313041035086, "7": 18373.46841697581, "8": 15881.14345795475}, "receiver_prefill_ms": 318.2702499907464, "ttft_ms": 57.068124995566905, "decode_ms": 60.71516592055559}, "total_wall_ms": 148972.64870698564, "ttft_ms": 57.068124995566905, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 59, "facts_retained": 29, "retention_rate": 0.4915254237288136, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.5714285714285714, "amount": 0.5, "status": 0.5833333333333334, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1341}, "ledger_by_hop": {"1": {"facts_total": 59, "facts_retained": 30, "retention_rate": 0.5084745762711864, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.5714285714285714, "amount": 0.6, "status": 0.5833333333333334, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1213}, "2": {"facts_total": 59, "facts_retained": 30, "retention_rate": 0.5084745762711864, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.5714285714285714, "amount": 0.6, "status": 0.5833333333333334, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1137}, "3": {"facts_total": 59, "facts_retained": 30, "retention_rate": 0.5084745762711864, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.5714285714285714, "amount": 0.6, "status": 0.5833333333333334, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1137}, "4": {"facts_total": 59, "facts_retained": 29, "retention_rate": 0.4915254237288136, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.5714285714285714, "amount": 0.5, "status": 0.5833333333333334, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1362}, "5": {"facts_total": 59, "facts_retained": 29, "retention_rate": 0.4915254237288136, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.5714285714285714, "amount": 0.5, "status": 0.5833333333333334, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1242}, "6": {"facts_total": 59, "facts_retained": 29, "retention_rate": 0.4915254237288136, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.5714285714285714, "amount": 0.5, "status": 0.5833333333333334, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1473}, "7": {"facts_total": 59, "facts_retained": 29, "retention_rate": 0.4915254237288136, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.5714285714285714, "amount": 0.5, "status": 0.5833333333333334, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1175}, "8": {"facts_total": 59, "facts_retained": 29, "retention_rate": 0.4915254237288136, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.5714285714285714, "amount": 0.5, "status": 0.5833333333333334, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1341}}}, "reference_key_id": "keys/synthetic/syn_03012_2000.json", "task_scores": {"correct": 1.0, "abstained": 1.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "901314986d6cfc5eb94723973fc9ade34faba6815cd2a9c8d95879c4ad80b6ba", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "aa5f3699f2dcab992bd92b9c", "attempt_id": "a33df393daa4", "run_group_id": "E4_core:A4|closed|qwen3b-llama3b:h8", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:44:47.050116Z", "task_family": "synthetic", "document_id": "syn_03012_2000", "document_hash": "03e441e68d54e1af036f8ff905dfd6cbbccead62636279c1a8e179352ab08945", "question_id": "q14_unanswerable", "split": "test", "cluster_id": "syn_03012_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "tokenizer_hashes": {"receiver": "b9d8872cf3a99da99e1fc4b0f71232a451632dc49ca4c6771549b4efa5c1c763"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 8, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1767, "query_tokens": 515, "message_tokens": 398, "message_utf8_bytes": 1341, "cache_tensor_bytes": null, "serialized_payload_bytes": 1341, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 148535.66862398293, "hops": {"1": 23159.84900004696, "2": 16642.98779098317, "3": 21655.61054192949, "4": 16671.246667043306, "5": 19499.049708014354, "6": 16652.313041035086, "7": 18373.46841697581, "8": 15881.14345795475}, "receiver_prefill_ms": 478.3567909616977, "ttft_ms": 23.002957925200462, "decode_ms": 37.429749965667725}, "total_wall_ms": 149075.14399895445, "ttft_ms": 23.002957925200462, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 59, "facts_retained": 29, "retention_rate": 0.4915254237288136, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.5714285714285714, "amount": 0.5, "status": 0.5833333333333334, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1341}, "ledger_by_hop": {"1": {"facts_total": 59, "facts_retained": 30, "retention_rate": 0.5084745762711864, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.5714285714285714, "amount": 0.6, "status": 0.5833333333333334, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1213}, "2": {"facts_total": 59, "facts_retained": 30, "retention_rate": 0.5084745762711864, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.5714285714285714, "amount": 0.6, "status": 0.5833333333333334, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1137}, "3": {"facts_total": 59, "facts_retained": 30, "retention_rate": 0.5084745762711864, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.5714285714285714, "amount": 0.6, "status": 0.5833333333333334, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1137}, "4": {"facts_total": 59, "facts_retained": 29, "retention_rate": 0.4915254237288136, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.5714285714285714, "amount": 0.5, "status": 0.5833333333333334, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1362}, "5": {"facts_total": 59, "facts_retained": 29, "retention_rate": 0.4915254237288136, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.5714285714285714, "amount": 0.5, "status": 0.5833333333333334, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1242}, "6": {"facts_total": 59, "facts_retained": 29, "retention_rate": 0.4915254237288136, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.5714285714285714, "amount": 0.5, "status": 0.5833333333333334, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1473}, "7": {"facts_total": 59, "facts_retained": 29, "retention_rate": 0.4915254237288136, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.5714285714285714, "amount": 0.5, "status": 0.5833333333333334, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1175}, "8": {"facts_total": 59, "facts_retained": 29, "retention_rate": 0.4915254237288136, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.5714285714285714, "amount": 0.5, "status": 0.5833333333333334, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1341}}}, "reference_key_id": "keys/synthetic/syn_03012_2000.json", "task_scores": {"correct": 1.0, "abstained": 1.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "901314986d6cfc5eb94723973fc9ade34faba6815cd2a9c8d95879c4ad80b6ba", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "80406b42ed399c3fb734bc8a", "attempt_id": "f4e2351b731d", "run_group_id": "E4_core:A5|closed|qwen3b-llama3b:h1", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:45:10.261472Z", "task_family": "synthetic", "document_id": "syn_03012_2000", "document_hash": "03e441e68d54e1af036f8ff905dfd6cbbccead62636279c1a8e179352ab08945", "question_id": "q00_single_fact", "split": "test", "cluster_id": "syn_03012_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 1, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1927, "query_tokens": 431, "message_tokens": 348, "message_utf8_bytes": 824, "cache_tensor_bytes": null, "serialized_payload_bytes": 824, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 22652.15183305554, "hops": {"1": 22652.15183305554}, "receiver_prefill_ms": 282.07145805936307, "ttft_ms": 26.316958013921976, "decode_ms": 628.9927910547704}, "total_wall_ms": 23590.06845811382, "ttft_ms": 26.316958013921976, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 10}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "XX-71M-6213", "evidence_artifact": {"ledger": {"facts_total": 59, "facts_retained": 11, "retention_rate": 0.1864406779661017, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.14285714285714285, "amount": 0.0, "status": 0.4166666666666667, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 760}, "ledger_by_hop": {"1": {"facts_total": 59, "facts_retained": 11, "retention_rate": 0.1864406779661017, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.14285714285714285, "amount": 0.0, "status": 0.4166666666666667, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 760}}}, "reference_key_id": "keys/synthetic/syn_03012_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "bb16af6fe25c6b4abc9bdb0ad8c8b3d8b7d3dc12769f5feefdb7a2e4bb915494", "output_hash": "66ef6c5da682b38795e9029723bbd71cdde2c4dec70cf1fae6efe62ef243f10a"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "e8d5cc99c12644f025356d4a", "attempt_id": "0474e1b40914", "run_group_id": "E4_core:A5|closed|qwen3b-llama3b:h1", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:45:11.217522Z", "task_family": "synthetic", "document_id": "syn_03012_2000", "document_hash": "03e441e68d54e1af036f8ff905dfd6cbbccead62636279c1a8e179352ab08945", "question_id": "q01_single_fact", "split": "test", "cluster_id": "syn_03012_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 1, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1927, "query_tokens": 437, "message_tokens": 348, "message_utf8_bytes": 824, "cache_tensor_bytes": null, "serialized_payload_bytes": 824, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 22652.15183305554, "hops": {"1": 22652.15183305554}, "receiver_prefill_ms": 286.0565420705825, "ttft_ms": 32.20312506891787, "decode_ms": 34.39329098910093}, "total_wall_ms": 23005.309666041285, "ttft_ms": 32.20312506891787, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 59, "facts_retained": 11, "retention_rate": 0.1864406779661017, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.14285714285714285, "amount": 0.0, "status": 0.4166666666666667, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 760}, "ledger_by_hop": {"1": {"facts_total": 59, "facts_retained": 11, "retention_rate": 0.1864406779661017, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.14285714285714285, "amount": 0.0, "status": 0.4166666666666667, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 760}}}, "reference_key_id": "keys/synthetic/syn_03012_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "bb16af6fe25c6b4abc9bdb0ad8c8b3d8b7d3dc12769f5feefdb7a2e4bb915494", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "eef39846719bd169a67f6c36", "attempt_id": "324b848f4d35", "run_group_id": "E4_core:A5|closed|qwen3b-llama3b:h1", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:45:11.587582Z", "task_family": "synthetic", "document_id": "syn_03012_2000", "document_hash": "03e441e68d54e1af036f8ff905dfd6cbbccead62636279c1a8e179352ab08945", "question_id": "q02_exact_copy", "split": "test", "cluster_id": "syn_03012_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 1, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1927, "query_tokens": 436, "message_tokens": 348, "message_utf8_bytes": 824, "cache_tensor_bytes": null, "serialized_payload_bytes": 824, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 22652.15183305554, "hops": {"1": 22652.15183305554}, "receiver_prefill_ms": 297.2080829786137, "ttft_ms": 33.90412498265505, "decode_ms": 29.56337493378669}, "total_wall_ms": 23013.285833061673, "ttft_ms": 33.90412498265505, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 59, "facts_retained": 11, "retention_rate": 0.1864406779661017, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.14285714285714285, "amount": 0.0, "status": 0.4166666666666667, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 760}, "ledger_by_hop": {"1": {"facts_total": 59, "facts_retained": 11, "retention_rate": 0.1864406779661017, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.14285714285714285, "amount": 0.0, "status": 0.4166666666666667, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 760}}}, "reference_key_id": "keys/synthetic/syn_03012_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "bb16af6fe25c6b4abc9bdb0ad8c8b3d8b7d3dc12769f5feefdb7a2e4bb915494", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "7fc99e09fd773bc0db3eaa58", "attempt_id": "f4a624fbd307", "run_group_id": "E4_core:A5|closed|qwen3b-llama3b:h1", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:45:11.966224Z", "task_family": "synthetic", "document_id": "syn_03012_2000", "document_hash": "03e441e68d54e1af036f8ff905dfd6cbbccead62636279c1a8e179352ab08945", "question_id": "q03_exact_copy", "split": "test", "cluster_id": "syn_03012_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 1, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1927, "query_tokens": 432, "message_tokens": 348, "message_utf8_bytes": 824, "cache_tensor_bytes": null, "serialized_payload_bytes": 824, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 22652.15183305554, "hops": {"1": 22652.15183305554}, "receiver_prefill_ms": 271.3910409947857, "ttft_ms": 76.94683305453509, "decode_ms": 61.5959579590708}, "total_wall_ms": 23062.648874125443, "ttft_ms": 76.94683305453509, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 59, "facts_retained": 11, "retention_rate": 0.1864406779661017, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.14285714285714285, "amount": 0.0, "status": 0.4166666666666667, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 760}, "ledger_by_hop": {"1": {"facts_total": 59, "facts_retained": 11, "retention_rate": 0.1864406779661017, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.14285714285714285, "amount": 0.0, "status": 0.4166666666666667, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 760}}}, "reference_key_id": "keys/synthetic/syn_03012_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "bb16af6fe25c6b4abc9bdb0ad8c8b3d8b7d3dc12769f5feefdb7a2e4bb915494", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "0a7044a04c44ae7a1cdce7e7", "attempt_id": "a8a7a38dae5a", "run_group_id": "E4_core:A5|closed|qwen3b-llama3b:h1", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:45:12.393973Z", "task_family": "synthetic", "document_id": "syn_03012_2000", "document_hash": "03e441e68d54e1af036f8ff905dfd6cbbccead62636279c1a8e179352ab08945", "question_id": "q04_multi_hop", "split": "test", "cluster_id": "syn_03012_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 1, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1927, "query_tokens": 437, "message_tokens": 348, "message_utf8_bytes": 824, "cache_tensor_bytes": null, "serialized_payload_bytes": 824, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 22652.15183305554, "hops": {"1": 22652.15183305554}, "receiver_prefill_ms": 431.45941698458046, "ttft_ms": 20.933124935254455, "decode_ms": 37.29425009805709}, "total_wall_ms": 23142.311666044407, "ttft_ms": 20.933124935254455, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 59, "facts_retained": 11, "retention_rate": 0.1864406779661017, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.14285714285714285, "amount": 0.0, "status": 0.4166666666666667, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 760}, "ledger_by_hop": {"1": {"facts_total": 59, "facts_retained": 11, "retention_rate": 0.1864406779661017, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.14285714285714285, "amount": 0.0, "status": 0.4166666666666667, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 760}}}, "reference_key_id": "keys/synthetic/syn_03012_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "bb16af6fe25c6b4abc9bdb0ad8c8b3d8b7d3dc12769f5feefdb7a2e4bb915494", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "0570136ed9363843948a9232", "attempt_id": "cf0ad18a0b4c", "run_group_id": "E4_core:A5|closed|qwen3b-llama3b:h1", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:45:12.903123Z", "task_family": "synthetic", "document_id": "syn_03012_2000", "document_hash": "03e441e68d54e1af036f8ff905dfd6cbbccead62636279c1a8e179352ab08945", "question_id": "q05_multi_hop", "split": "test", "cluster_id": "syn_03012_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 1, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1927, "query_tokens": 440, "message_tokens": 348, "message_utf8_bytes": 824, "cache_tensor_bytes": null, "serialized_payload_bytes": 824, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 22652.15183305554, "hops": {"1": 22652.15183305554}, "receiver_prefill_ms": 302.00874991714954, "ttft_ms": 21.78095793351531, "decode_ms": 44.39425002783537}, "total_wall_ms": 23021.053708042018, "ttft_ms": 21.78095793351531, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 59, "facts_retained": 11, "retention_rate": 0.1864406779661017, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.14285714285714285, "amount": 0.0, "status": 0.4166666666666667, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 760}, "ledger_by_hop": {"1": {"facts_total": 59, "facts_retained": 11, "retention_rate": 0.1864406779661017, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.14285714285714285, "amount": 0.0, "status": 0.4166666666666667, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 760}}}, "reference_key_id": "keys/synthetic/syn_03012_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "bb16af6fe25c6b4abc9bdb0ad8c8b3d8b7d3dc12769f5feefdb7a2e4bb915494", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "ec411809962692e5df1465f5", "attempt_id": "292bebd4c11e", "run_group_id": "E4_core:A5|closed|qwen3b-llama3b:h1", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:45:13.290105Z", "task_family": "synthetic", "document_id": "syn_03012_2000", "document_hash": "03e441e68d54e1af036f8ff905dfd6cbbccead62636279c1a8e179352ab08945", "question_id": "q06_negation", "split": "test", "cluster_id": "syn_03012_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 1, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1927, "query_tokens": 432, "message_tokens": 348, "message_utf8_bytes": 824, "cache_tensor_bytes": null, "serialized_payload_bytes": 824, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 22652.15183305554, "hops": {"1": 22652.15183305554}, "receiver_prefill_ms": 310.31862495001405, "ttft_ms": 38.44629193190485, "decode_ms": 30.04345903173089}, "total_wall_ms": 23031.582291121595, "ttft_ms": 38.44629193190485, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 59, "facts_retained": 11, "retention_rate": 0.1864406779661017, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.14285714285714285, "amount": 0.0, "status": 0.4166666666666667, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 760}, "ledger_by_hop": {"1": {"facts_total": 59, "facts_retained": 11, "retention_rate": 0.1864406779661017, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.14285714285714285, "amount": 0.0, "status": 0.4166666666666667, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 760}}}, "reference_key_id": "keys/synthetic/syn_03012_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "bb16af6fe25c6b4abc9bdb0ad8c8b3d8b7d3dc12769f5feefdb7a2e4bb915494", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "506bb7ff47c14814a9059f6e", "attempt_id": "f3b1e9cd684b", "run_group_id": "E4_core:A5|closed|qwen3b-llama3b:h1", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:45:13.686984Z", "task_family": "synthetic", "document_id": "syn_03012_2000", "document_hash": "03e441e68d54e1af036f8ff905dfd6cbbccead62636279c1a8e179352ab08945", "question_id": "q07_negation", "split": "test", "cluster_id": "syn_03012_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 1, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1927, "query_tokens": 432, "message_tokens": 348, "message_utf8_bytes": 824, "cache_tensor_bytes": null, "serialized_payload_bytes": 824, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 22652.15183305554, "hops": {"1": 22652.15183305554}, "receiver_prefill_ms": 262.92891695629805, "ttft_ms": 54.86416595522314, "decode_ms": 57.50279198400676}, "total_wall_ms": 23027.997000142932, "ttft_ms": 54.86416595522314, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 59, "facts_retained": 11, "retention_rate": 0.1864406779661017, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.14285714285714285, "amount": 0.0, "status": 0.4166666666666667, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 760}, "ledger_by_hop": {"1": {"facts_total": 59, "facts_retained": 11, "retention_rate": 0.1864406779661017, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.14285714285714285, "amount": 0.0, "status": 0.4166666666666667, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 760}}}, "reference_key_id": "keys/synthetic/syn_03012_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "bb16af6fe25c6b4abc9bdb0ad8c8b3d8b7d3dc12769f5feefdb7a2e4bb915494", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "eda7a9958bd808388459a728", "attempt_id": "88bd1c6e6bb2", "run_group_id": "E4_core:A5|closed|qwen3b-llama3b:h1", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:45:14.081792Z", "task_family": "synthetic", "document_id": "syn_03012_2000", "document_hash": "03e441e68d54e1af036f8ff905dfd6cbbccead62636279c1a8e179352ab08945", "question_id": "q08_revision", "split": "test", "cluster_id": "syn_03012_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 1, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1927, "query_tokens": 447, "message_tokens": 348, "message_utf8_bytes": 824, "cache_tensor_bytes": null, "serialized_payload_bytes": 824, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 22652.15183305554, "hops": {"1": 22652.15183305554}, "receiver_prefill_ms": 455.98504203371704, "ttft_ms": 31.549459090456367, "decode_ms": 42.095832992345095}, "total_wall_ms": 23182.68170801457, "ttft_ms": 31.549459090456367, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 59, "facts_retained": 11, "retention_rate": 0.1864406779661017, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.14285714285714285, "amount": 0.0, "status": 0.4166666666666667, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 760}, "ledger_by_hop": {"1": {"facts_total": 59, "facts_retained": 11, "retention_rate": 0.1864406779661017, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.14285714285714285, "amount": 0.0, "status": 0.4166666666666667, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 760}}}, "reference_key_id": "keys/synthetic/syn_03012_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "bb16af6fe25c6b4abc9bdb0ad8c8b3d8b7d3dc12769f5feefdb7a2e4bb915494", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "a203e5b8a6de583836e92013", "attempt_id": "a656c1fda28f", "run_group_id": "E4_core:A5|closed|qwen3b-llama3b:h1", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:45:14.630530Z", "task_family": "synthetic", "document_id": "syn_03012_2000", "document_hash": "03e441e68d54e1af036f8ff905dfd6cbbccead62636279c1a8e179352ab08945", "question_id": "q09_revision", "split": "test", "cluster_id": "syn_03012_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 1, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1927, "query_tokens": 440, "message_tokens": 348, "message_utf8_bytes": 824, "cache_tensor_bytes": null, "serialized_payload_bytes": 824, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 22652.15183305554, "hops": {"1": 22652.15183305554}, "receiver_prefill_ms": 297.17908310703933, "ttft_ms": 35.46908299904317, "decode_ms": 36.725500016473234}, "total_wall_ms": 23021.988208056428, "ttft_ms": 35.46908299904317, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 59, "facts_retained": 11, "retention_rate": 0.1864406779661017, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.14285714285714285, "amount": 0.0, "status": 0.4166666666666667, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 760}, "ledger_by_hop": {"1": {"facts_total": 59, "facts_retained": 11, "retention_rate": 0.1864406779661017, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.14285714285714285, "amount": 0.0, "status": 0.4166666666666667, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 760}}}, "reference_key_id": "keys/synthetic/syn_03012_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "bb16af6fe25c6b4abc9bdb0ad8c8b3d8b7d3dc12769f5feefdb7a2e4bb915494", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "a4535e6fd52eccaca7a23bd8", "attempt_id": "77bd69e1e860", "run_group_id": "E4_core:A5|closed|qwen3b-llama3b:h1", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:45:15.018758Z", "task_family": "synthetic", "document_id": "syn_03012_2000", "document_hash": "03e441e68d54e1af036f8ff905dfd6cbbccead62636279c1a8e179352ab08945", "question_id": "q10_temporal", "split": "test", "cluster_id": "syn_03012_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 1, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1927, "query_tokens": 450, "message_tokens": 348, "message_utf8_bytes": 824, "cache_tensor_bytes": null, "serialized_payload_bytes": 824, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 22652.15183305554, "hops": {"1": 22652.15183305554}, "receiver_prefill_ms": 318.4328330680728, "ttft_ms": 20.363375078886747, "decode_ms": 19.991916022263467}, "total_wall_ms": 23011.404083110392, "ttft_ms": 20.363375078886747, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 59, "facts_retained": 11, "retention_rate": 0.1864406779661017, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.14285714285714285, "amount": 0.0, "status": 0.4166666666666667, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 760}, "ledger_by_hop": {"1": {"facts_total": 59, "facts_retained": 11, "retention_rate": 0.1864406779661017, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.14285714285714285, "amount": 0.0, "status": 0.4166666666666667, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 760}}}, "reference_key_id": "keys/synthetic/syn_03012_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "bb16af6fe25c6b4abc9bdb0ad8c8b3d8b7d3dc12769f5feefdb7a2e4bb915494", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "535120b5a9cd90862f292e4b", "attempt_id": "5f54a6336024", "run_group_id": "E4_core:A5|closed|qwen3b-llama3b:h1", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:45:15.395419Z", "task_family": "synthetic", "document_id": "syn_03012_2000", "document_hash": "03e441e68d54e1af036f8ff905dfd6cbbccead62636279c1a8e179352ab08945", "question_id": "q11_aggregation", "split": "test", "cluster_id": "syn_03012_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 1, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1927, "query_tokens": 442, "message_tokens": 348, "message_utf8_bytes": 824, "cache_tensor_bytes": null, "serialized_payload_bytes": 824, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 22652.15183305554, "hops": {"1": 22652.15183305554}, "receiver_prefill_ms": 343.6899580992758, "ttft_ms": 46.83649993967265, "decode_ms": 112.11266706231982}, "total_wall_ms": 23155.46874899883, "ttft_ms": 46.83649993967265, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 59, "facts_retained": 11, "retention_rate": 0.1864406779661017, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.14285714285714285, "amount": 0.0, "status": 0.4166666666666667, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 760}, "ledger_by_hop": {"1": {"facts_total": 59, "facts_retained": 11, "retention_rate": 0.1864406779661017, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.14285714285714285, "amount": 0.0, "status": 0.4166666666666667, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 760}}}, "reference_key_id": "keys/synthetic/syn_03012_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "bb16af6fe25c6b4abc9bdb0ad8c8b3d8b7d3dc12769f5feefdb7a2e4bb915494", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "1f04719b39434acfd0cb7ccc", "attempt_id": "f5de6cd65e7c", "run_group_id": "E4_core:A5|closed|qwen3b-llama3b:h1", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:45:15.918124Z", "task_family": "synthetic", "document_id": "syn_03012_2000", "document_hash": "03e441e68d54e1af036f8ff905dfd6cbbccead62636279c1a8e179352ab08945", "question_id": "q12_unit_conversion", "split": "test", "cluster_id": "syn_03012_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 1, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1927, "query_tokens": 439, "message_tokens": 348, "message_utf8_bytes": 824, "cache_tensor_bytes": null, "serialized_payload_bytes": 824, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 22652.15183305554, "hops": {"1": 22652.15183305554}, "receiver_prefill_ms": 373.2212920440361, "ttft_ms": 39.90400000475347, "decode_ms": 183.45208303071558}, "total_wall_ms": 23249.24658308737, "ttft_ms": 39.90400000475347, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 5}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "44620", "evidence_artifact": {"ledger": {"facts_total": 59, "facts_retained": 11, "retention_rate": 0.1864406779661017, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.14285714285714285, "amount": 0.0, "status": 0.4166666666666667, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 760}, "ledger_by_hop": {"1": {"facts_total": 59, "facts_retained": 11, "retention_rate": 0.1864406779661017, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.14285714285714285, "amount": 0.0, "status": 0.4166666666666667, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 760}}}, "reference_key_id": "keys/synthetic/syn_03012_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "bb16af6fe25c6b4abc9bdb0ad8c8b3d8b7d3dc12769f5feefdb7a2e4bb915494", "output_hash": "2bd125847727be7e7c0a656ee4c81f3bc6d0fef11f536f50ce7bc83688907075"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "9ad02b1d15cf04539c23c7e6", "attempt_id": "0b012858a78d", "run_group_id": "E4_core:A5|closed|qwen3b-llama3b:h1", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:45:16.534953Z", "task_family": "synthetic", "document_id": "syn_03012_2000", "document_hash": "03e441e68d54e1af036f8ff905dfd6cbbccead62636279c1a8e179352ab08945", "question_id": "q13_unanswerable", "split": "test", "cluster_id": "syn_03012_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 1, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1927, "query_tokens": 444, "message_tokens": 348, "message_utf8_bytes": 824, "cache_tensor_bytes": null, "serialized_payload_bytes": 824, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 22652.15183305554, "hops": {"1": 22652.15183305554}, "receiver_prefill_ms": 293.3299579890445, "ttft_ms": 29.990124981850386, "decode_ms": 32.214832957834005}, "total_wall_ms": 23008.111208095215, "ttft_ms": 29.990124981850386, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 59, "facts_retained": 11, "retention_rate": 0.1864406779661017, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.14285714285714285, "amount": 0.0, "status": 0.4166666666666667, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 760}, "ledger_by_hop": {"1": {"facts_total": 59, "facts_retained": 11, "retention_rate": 0.1864406779661017, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.14285714285714285, "amount": 0.0, "status": 0.4166666666666667, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 760}}}, "reference_key_id": "keys/synthetic/syn_03012_2000.json", "task_scores": {"correct": 1.0, "abstained": 1.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "bb16af6fe25c6b4abc9bdb0ad8c8b3d8b7d3dc12769f5feefdb7a2e4bb915494", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "0c601642ab3f9512c25a0287", "attempt_id": "d456ed3e334a", "run_group_id": "E4_core:A5|closed|qwen3b-llama3b:h1", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:45:16.908057Z", "task_family": "synthetic", "document_id": "syn_03012_2000", "document_hash": "03e441e68d54e1af036f8ff905dfd6cbbccead62636279c1a8e179352ab08945", "question_id": "q14_unanswerable", "split": "test", "cluster_id": "syn_03012_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 1, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1927, "query_tokens": 444, "message_tokens": 348, "message_utf8_bytes": 824, "cache_tensor_bytes": null, "serialized_payload_bytes": 824, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 22652.15183305554, "hops": {"1": 22652.15183305554}, "receiver_prefill_ms": 273.25808396562934, "ttft_ms": 53.93837497103959, "decode_ms": 40.390208014287055}, "total_wall_ms": 23020.392708014697, "ttft_ms": 53.93837497103959, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 59, "facts_retained": 11, "retention_rate": 0.1864406779661017, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.14285714285714285, "amount": 0.0, "status": 0.4166666666666667, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 760}, "ledger_by_hop": {"1": {"facts_total": 59, "facts_retained": 11, "retention_rate": 0.1864406779661017, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.14285714285714285, "amount": 0.0, "status": 0.4166666666666667, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 760}}}, "reference_key_id": "keys/synthetic/syn_03012_2000.json", "task_scores": {"correct": 1.0, "abstained": 1.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "bb16af6fe25c6b4abc9bdb0ad8c8b3d8b7d3dc12769f5feefdb7a2e4bb915494", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "344f6602011bf61045b2a747", "attempt_id": "9a72afa94245", "run_group_id": "E4_core:A5|closed|qwen3b-llama3b:h2", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:45:32.567876Z", "task_family": "synthetic", "document_id": "syn_03012_2000", "document_hash": "03e441e68d54e1af036f8ff905dfd6cbbccead62636279c1a8e179352ab08945", "question_id": "q00_single_fact", "split": "test", "cluster_id": "syn_03012_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "tokenizer_hashes": {"receiver": "b9d8872cf3a99da99e1fc4b0f71232a451632dc49ca4c6771549b4efa5c1c763"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 2, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1767, "query_tokens": 492, "message_tokens": 388, "message_utf8_bytes": 1097, "cache_tensor_bytes": null, "serialized_payload_bytes": 1097, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 37924.419708084315, "hops": {"1": 22652.15183305554, "2": 15272.267875028774}, "receiver_prefill_ms": 498.429708997719, "ttft_ms": 35.99687502719462, "decode_ms": 233.07824996300042}, "total_wall_ms": 38692.56866700016, "ttft_ms": 35.99687502719462, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 7}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "XX-71M-6213", "evidence_artifact": {"ledger": {"facts_total": 59, "facts_retained": 13, "retention_rate": 0.22033898305084745, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.14285714285714285, "amount": 0.0, "status": 0.5833333333333334, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1097}, "ledger_by_hop": {"1": {"facts_total": 59, "facts_retained": 11, "retention_rate": 0.1864406779661017, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.14285714285714285, "amount": 0.0, "status": 0.4166666666666667, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 760}, "2": {"facts_total": 59, "facts_retained": 13, "retention_rate": 0.22033898305084745, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.14285714285714285, "amount": 0.0, "status": 0.5833333333333334, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1097}}}, "reference_key_id": "keys/synthetic/syn_03012_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "e9a21b2d20aaeefae66c356aa30a9c6e588cc37b8ff648442abfac81dba16da2", "output_hash": "66ef6c5da682b38795e9029723bbd71cdde2c4dec70cf1fae6efe62ef243f10a"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "2c45a016e3d5a88e5bc3e492", "attempt_id": "0c879d700667", "run_group_id": "E4_core:A5|closed|qwen3b-llama3b:h2", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:45:33.353920Z", "task_family": "synthetic", "document_id": "syn_03012_2000", "document_hash": "03e441e68d54e1af036f8ff905dfd6cbbccead62636279c1a8e179352ab08945", "question_id": "q01_single_fact", "split": "test", "cluster_id": "syn_03012_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "tokenizer_hashes": {"receiver": "b9d8872cf3a99da99e1fc4b0f71232a451632dc49ca4c6771549b4efa5c1c763"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 2, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1767, "query_tokens": 497, "message_tokens": 388, "message_utf8_bytes": 1097, "cache_tensor_bytes": null, "serialized_payload_bytes": 1097, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 37924.419708084315, "hops": {"1": 22652.15183305554, "2": 15272.267875028774}, "receiver_prefill_ms": 369.1821659449488, "ttft_ms": 32.824583002366126, "decode_ms": 31.027624965645373}, "total_wall_ms": 38358.24008309282, "ttft_ms": 32.824583002366126, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 59, "facts_retained": 13, "retention_rate": 0.22033898305084745, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.14285714285714285, "amount": 0.0, "status": 0.5833333333333334, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1097}, "ledger_by_hop": {"1": {"facts_total": 59, "facts_retained": 11, "retention_rate": 0.1864406779661017, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.14285714285714285, "amount": 0.0, "status": 0.4166666666666667, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 760}, "2": {"facts_total": 59, "facts_retained": 13, "retention_rate": 0.22033898305084745, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.14285714285714285, "amount": 0.0, "status": 0.5833333333333334, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1097}}}, "reference_key_id": "keys/synthetic/syn_03012_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "e9a21b2d20aaeefae66c356aa30a9c6e588cc37b8ff648442abfac81dba16da2", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "a49d6c1f400fd464268f9cec", "attempt_id": "81e791aa227c", "run_group_id": "E4_core:A5|closed|qwen3b-llama3b:h2", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:45:33.805735Z", "task_family": "synthetic", "document_id": "syn_03012_2000", "document_hash": "03e441e68d54e1af036f8ff905dfd6cbbccead62636279c1a8e179352ab08945", "question_id": "q02_exact_copy", "split": "test", "cluster_id": "syn_03012_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "tokenizer_hashes": {"receiver": "b9d8872cf3a99da99e1fc4b0f71232a451632dc49ca4c6771549b4efa5c1c763"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 2, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1767, "query_tokens": 497, "message_tokens": 388, "message_utf8_bytes": 1097, "cache_tensor_bytes": null, "serialized_payload_bytes": 1097, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 37924.419708084315, "hops": {"1": 22652.15183305554, "2": 15272.267875028774}, "receiver_prefill_ms": 336.5195420337841, "ttft_ms": 36.39487491454929, "decode_ms": 37.64350002165884}, "total_wall_ms": 38335.80145810265, "ttft_ms": 36.39487491454929, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 59, "facts_retained": 13, "retention_rate": 0.22033898305084745, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.14285714285714285, "amount": 0.0, "status": 0.5833333333333334, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1097}, "ledger_by_hop": {"1": {"facts_total": 59, "facts_retained": 11, "retention_rate": 0.1864406779661017, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.14285714285714285, "amount": 0.0, "status": 0.4166666666666667, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 760}, "2": {"facts_total": 59, "facts_retained": 13, "retention_rate": 0.22033898305084745, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.14285714285714285, "amount": 0.0, "status": 0.5833333333333334, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1097}}}, "reference_key_id": "keys/synthetic/syn_03012_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "e9a21b2d20aaeefae66c356aa30a9c6e588cc37b8ff648442abfac81dba16da2", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "51dacf4dc433f744c96654d7", "attempt_id": "81164b67a301", "run_group_id": "E4_core:A5|closed|qwen3b-llama3b:h2", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:45:34.235919Z", "task_family": "synthetic", "document_id": "syn_03012_2000", "document_hash": "03e441e68d54e1af036f8ff905dfd6cbbccead62636279c1a8e179352ab08945", "question_id": "q03_exact_copy", "split": "test", "cluster_id": "syn_03012_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "tokenizer_hashes": {"receiver": "b9d8872cf3a99da99e1fc4b0f71232a451632dc49ca4c6771549b4efa5c1c763"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 2, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1767, "query_tokens": 493, "message_tokens": 388, "message_utf8_bytes": 1097, "cache_tensor_bytes": null, "serialized_payload_bytes": 1097, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 37924.419708084315, "hops": {"1": 22652.15183305554, "2": 15272.267875028774}, "receiver_prefill_ms": 463.2286250125617, "ttft_ms": 35.067457938566804, "decode_ms": 23.925208020955324}, "total_wall_ms": 38447.185166180134, "ttft_ms": 35.067457938566804, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 59, "facts_retained": 13, "retention_rate": 0.22033898305084745, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.14285714285714285, "amount": 0.0, "status": 0.5833333333333334, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1097}, "ledger_by_hop": {"1": {"facts_total": 59, "facts_retained": 11, "retention_rate": 0.1864406779661017, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.14285714285714285, "amount": 0.0, "status": 0.4166666666666667, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 760}, "2": {"facts_total": 59, "facts_retained": 13, "retention_rate": 0.22033898305084745, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.14285714285714285, "amount": 0.0, "status": 0.5833333333333334, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1097}}}, "reference_key_id": "keys/synthetic/syn_03012_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "e9a21b2d20aaeefae66c356aa30a9c6e588cc37b8ff648442abfac81dba16da2", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "2ea4ea4e0ea8db995a76cf92", "attempt_id": "1a2d1759f432", "run_group_id": "E4_core:A5|closed|qwen3b-llama3b:h2", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:45:34.775813Z", "task_family": "synthetic", "document_id": "syn_03012_2000", "document_hash": "03e441e68d54e1af036f8ff905dfd6cbbccead62636279c1a8e179352ab08945", "question_id": "q04_multi_hop", "split": "test", "cluster_id": "syn_03012_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "tokenizer_hashes": {"receiver": "b9d8872cf3a99da99e1fc4b0f71232a451632dc49ca4c6771549b4efa5c1c763"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 2, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1767, "query_tokens": 498, "message_tokens": 388, "message_utf8_bytes": 1097, "cache_tensor_bytes": null, "serialized_payload_bytes": 1097, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 37924.419708084315, "hops": {"1": 22652.15183305554, "2": 15272.267875028774}, "receiver_prefill_ms": 371.6336670331657, "ttft_ms": 41.74729192163795, "decode_ms": 24.985000025480986}, "total_wall_ms": 38363.37862408254, "ttft_ms": 41.74729192163795, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 59, "facts_retained": 13, "retention_rate": 0.22033898305084745, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.14285714285714285, "amount": 0.0, "status": 0.5833333333333334, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1097}, "ledger_by_hop": {"1": {"facts_total": 59, "facts_retained": 11, "retention_rate": 0.1864406779661017, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.14285714285714285, "amount": 0.0, "status": 0.4166666666666667, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 760}, "2": {"facts_total": 59, "facts_retained": 13, "retention_rate": 0.22033898305084745, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.14285714285714285, "amount": 0.0, "status": 0.5833333333333334, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1097}}}, "reference_key_id": "keys/synthetic/syn_03012_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "e9a21b2d20aaeefae66c356aa30a9c6e588cc37b8ff648442abfac81dba16da2", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "4ae86c3fc0ba84a3c4df2378", "attempt_id": "ddd56a8703ba", "run_group_id": "E4_core:A5|closed|qwen3b-llama3b:h2", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:45:35.232148Z", "task_family": "synthetic", "document_id": "syn_03012_2000", "document_hash": "03e441e68d54e1af036f8ff905dfd6cbbccead62636279c1a8e179352ab08945", "question_id": "q05_multi_hop", "split": "test", "cluster_id": "syn_03012_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "tokenizer_hashes": {"receiver": "b9d8872cf3a99da99e1fc4b0f71232a451632dc49ca4c6771549b4efa5c1c763"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 2, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1767, "query_tokens": 499, "message_tokens": 388, "message_utf8_bytes": 1097, "cache_tensor_bytes": null, "serialized_payload_bytes": 1097, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 37924.419708084315, "hops": {"1": 22652.15183305554, "2": 15272.267875028774}, "receiver_prefill_ms": 348.66899996995926, "ttft_ms": 13.490915996953845, "decode_ms": 25.146374944597483}, "total_wall_ms": 38312.380292103626, "ttft_ms": 13.490915996953845, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 59, "facts_retained": 13, "retention_rate": 0.22033898305084745, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.14285714285714285, "amount": 0.0, "status": 0.5833333333333334, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1097}, "ledger_by_hop": {"1": {"facts_total": 59, "facts_retained": 11, "retention_rate": 0.1864406779661017, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.14285714285714285, "amount": 0.0, "status": 0.4166666666666667, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 760}, "2": {"facts_total": 59, "facts_retained": 13, "retention_rate": 0.22033898305084745, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.14285714285714285, "amount": 0.0, "status": 0.5833333333333334, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1097}}}, "reference_key_id": "keys/synthetic/syn_03012_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "e9a21b2d20aaeefae66c356aa30a9c6e588cc37b8ff648442abfac81dba16da2", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "cc993cd2ca275698ff358582", "attempt_id": "317c1be3e1dd", "run_group_id": "E4_core:A5|closed|qwen3b-llama3b:h2", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:45:35.637206Z", "task_family": "synthetic", "document_id": "syn_03012_2000", "document_hash": "03e441e68d54e1af036f8ff905dfd6cbbccead62636279c1a8e179352ab08945", "question_id": "q06_negation", "split": "test", "cluster_id": "syn_03012_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "tokenizer_hashes": {"receiver": "b9d8872cf3a99da99e1fc4b0f71232a451632dc49ca4c6771549b4efa5c1c763"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 2, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1767, "query_tokens": 492, "message_tokens": 388, "message_utf8_bytes": 1097, "cache_tensor_bytes": null, "serialized_payload_bytes": 1097, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 37924.419708084315, "hops": {"1": 22652.15183305554, "2": 15272.267875028774}, "receiver_prefill_ms": 358.2399160368368, "ttft_ms": 41.67533398140222, "decode_ms": 107.11674997583032}, "total_wall_ms": 38432.1455830941, "ttft_ms": 41.67533398140222, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "No", "evidence_artifact": {"ledger": {"facts_total": 59, "facts_retained": 13, "retention_rate": 0.22033898305084745, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.14285714285714285, "amount": 0.0, "status": 0.5833333333333334, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1097}, "ledger_by_hop": {"1": {"facts_total": 59, "facts_retained": 11, "retention_rate": 0.1864406779661017, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.14285714285714285, "amount": 0.0, "status": 0.4166666666666667, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 760}, "2": {"facts_total": 59, "facts_retained": 13, "retention_rate": 0.22033898305084745, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.14285714285714285, "amount": 0.0, "status": 0.5833333333333334, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1097}}}, "reference_key_id": "keys/synthetic/syn_03012_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "e9a21b2d20aaeefae66c356aa30a9c6e588cc37b8ff648442abfac81dba16da2", "output_hash": "1ea442a134b2a184bd5d40104401f2a37fbc09ccf3f4bc9da161c6099be3691d"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "90e16f8b14e3bca5f64aafae", "attempt_id": "3c182aae9039", "run_group_id": "E4_core:A5|closed|qwen3b-llama3b:h2", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:45:36.163545Z", "task_family": "synthetic", "document_id": "syn_03012_2000", "document_hash": "03e441e68d54e1af036f8ff905dfd6cbbccead62636279c1a8e179352ab08945", "question_id": "q07_negation", "split": "test", "cluster_id": "syn_03012_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "tokenizer_hashes": {"receiver": "b9d8872cf3a99da99e1fc4b0f71232a451632dc49ca4c6771549b4efa5c1c763"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 2, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1767, "query_tokens": 493, "message_tokens": 388, "message_utf8_bytes": 1097, "cache_tensor_bytes": null, "serialized_payload_bytes": 1097, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 37924.419708084315, "hops": {"1": 22652.15183305554, "2": 15272.267875028774}, "receiver_prefill_ms": 448.7235420383513, "ttft_ms": 38.899959065020084, "decode_ms": 23.782082949765027}, "total_wall_ms": 38436.36508309282, "ttft_ms": 38.899959065020084, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "No", "evidence_artifact": {"ledger": {"facts_total": 59, "facts_retained": 13, "retention_rate": 0.22033898305084745, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.14285714285714285, "amount": 0.0, "status": 0.5833333333333334, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1097}, "ledger_by_hop": {"1": {"facts_total": 59, "facts_retained": 11, "retention_rate": 0.1864406779661017, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.14285714285714285, "amount": 0.0, "status": 0.4166666666666667, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 760}, "2": {"facts_total": 59, "facts_retained": 13, "retention_rate": 0.22033898305084745, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.14285714285714285, "amount": 0.0, "status": 0.5833333333333334, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1097}}}, "reference_key_id": "keys/synthetic/syn_03012_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "e9a21b2d20aaeefae66c356aa30a9c6e588cc37b8ff648442abfac81dba16da2", "output_hash": "1ea442a134b2a184bd5d40104401f2a37fbc09ccf3f4bc9da161c6099be3691d"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "a3a5162567e97d53cbdd613a", "attempt_id": "142ae218b112", "run_group_id": "E4_core:A5|closed|qwen3b-llama3b:h2", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:45:36.695369Z", "task_family": "synthetic", "document_id": "syn_03012_2000", "document_hash": "03e441e68d54e1af036f8ff905dfd6cbbccead62636279c1a8e179352ab08945", "question_id": "q08_revision", "split": "test", "cluster_id": "syn_03012_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "tokenizer_hashes": {"receiver": "b9d8872cf3a99da99e1fc4b0f71232a451632dc49ca4c6771549b4efa5c1c763"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 2, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1767, "query_tokens": 508, "message_tokens": 388, "message_utf8_bytes": 1097, "cache_tensor_bytes": null, "serialized_payload_bytes": 1097, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 37924.419708084315, "hops": {"1": 22652.15183305554, "2": 15272.267875028774}, "receiver_prefill_ms": 376.5355829382315, "ttft_ms": 25.208333041518927, "decode_ms": 35.80612502992153}, "total_wall_ms": 38362.456999137066, "ttft_ms": 25.208333041518927, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 59, "facts_retained": 13, "retention_rate": 0.22033898305084745, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.14285714285714285, "amount": 0.0, "status": 0.5833333333333334, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1097}, "ledger_by_hop": {"1": {"facts_total": 59, "facts_retained": 11, "retention_rate": 0.1864406779661017, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.14285714285714285, "amount": 0.0, "status": 0.4166666666666667, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 760}, "2": {"facts_total": 59, "facts_retained": 13, "retention_rate": 0.22033898305084745, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.14285714285714285, "amount": 0.0, "status": 0.5833333333333334, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1097}}}, "reference_key_id": "keys/synthetic/syn_03012_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "e9a21b2d20aaeefae66c356aa30a9c6e588cc37b8ff648442abfac81dba16da2", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "0b4f7d7b810b449a7216202b", "attempt_id": "69b8035c2aa6", "run_group_id": "E4_core:A5|closed|qwen3b-llama3b:h2", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:45:37.150818Z", "task_family": "synthetic", "document_id": "syn_03012_2000", "document_hash": "03e441e68d54e1af036f8ff905dfd6cbbccead62636279c1a8e179352ab08945", "question_id": "q09_revision", "split": "test", "cluster_id": "syn_03012_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "tokenizer_hashes": {"receiver": "b9d8872cf3a99da99e1fc4b0f71232a451632dc49ca4c6771549b4efa5c1c763"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 2, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1767, "query_tokens": 500, "message_tokens": 388, "message_utf8_bytes": 1097, "cache_tensor_bytes": null, "serialized_payload_bytes": 1097, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 37924.419708084315, "hops": {"1": 22652.15183305554, "2": 15272.267875028774}, "receiver_prefill_ms": 343.36612501647323, "ttft_ms": 39.10233301576227, "decode_ms": 34.63324997574091}, "total_wall_ms": 38342.228791094385, "ttft_ms": 39.10233301576227, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 59, "facts_retained": 13, "retention_rate": 0.22033898305084745, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.14285714285714285, "amount": 0.0, "status": 0.5833333333333334, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1097}, "ledger_by_hop": {"1": {"facts_total": 59, "facts_retained": 11, "retention_rate": 0.1864406779661017, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.14285714285714285, "amount": 0.0, "status": 0.4166666666666667, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 760}, "2": {"facts_total": 59, "facts_retained": 13, "retention_rate": 0.22033898305084745, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.14285714285714285, "amount": 0.0, "status": 0.5833333333333334, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1097}}}, "reference_key_id": "keys/synthetic/syn_03012_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "e9a21b2d20aaeefae66c356aa30a9c6e588cc37b8ff648442abfac81dba16da2", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "acf0ed3f0dfa2a392c7eaa2f", "attempt_id": "cdf7a27fefaf", "run_group_id": "E4_core:A5|closed|qwen3b-llama3b:h2", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:45:37.585829Z", "task_family": "synthetic", "document_id": "syn_03012_2000", "document_hash": "03e441e68d54e1af036f8ff905dfd6cbbccead62636279c1a8e179352ab08945", "question_id": "q10_temporal", "split": "test", "cluster_id": "syn_03012_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "tokenizer_hashes": {"receiver": "b9d8872cf3a99da99e1fc4b0f71232a451632dc49ca4c6771549b4efa5c1c763"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 2, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1767, "query_tokens": 510, "message_tokens": 388, "message_utf8_bytes": 1097, "cache_tensor_bytes": null, "serialized_payload_bytes": 1097, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 37924.419708084315, "hops": {"1": 22652.15183305554, "2": 15272.267875028774}, "receiver_prefill_ms": 495.1957910088822, "ttft_ms": 38.7896669562906, "decode_ms": 21.615375066176057}, "total_wall_ms": 38480.69608316291, "ttft_ms": 38.7896669562906, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 59, "facts_retained": 13, "retention_rate": 0.22033898305084745, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.14285714285714285, "amount": 0.0, "status": 0.5833333333333334, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1097}, "ledger_by_hop": {"1": {"facts_total": 59, "facts_retained": 11, "retention_rate": 0.1864406779661017, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.14285714285714285, "amount": 0.0, "status": 0.4166666666666667, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 760}, "2": {"facts_total": 59, "facts_retained": 13, "retention_rate": 0.22033898305084745, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.14285714285714285, "amount": 0.0, "status": 0.5833333333333334, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1097}}}, "reference_key_id": "keys/synthetic/syn_03012_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "e9a21b2d20aaeefae66c356aa30a9c6e588cc37b8ff648442abfac81dba16da2", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "9f5596f881ec27c99cb4fc8a", "attempt_id": "718f999f4dfc", "run_group_id": "E4_core:A5|closed|qwen3b-llama3b:h2", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:45:38.159594Z", "task_family": "synthetic", "document_id": "syn_03012_2000", "document_hash": "03e441e68d54e1af036f8ff905dfd6cbbccead62636279c1a8e179352ab08945", "question_id": "q11_aggregation", "split": "test", "cluster_id": "syn_03012_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "tokenizer_hashes": {"receiver": "b9d8872cf3a99da99e1fc4b0f71232a451632dc49ca4c6771549b4efa5c1c763"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 2, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1767, "query_tokens": 503, "message_tokens": 388, "message_utf8_bytes": 1097, "cache_tensor_bytes": null, "serialized_payload_bytes": 1097, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 37924.419708084315, "hops": {"1": 22652.15183305554, "2": 15272.267875028774}, "receiver_prefill_ms": 370.6400420051068, "ttft_ms": 30.94566590152681, "decode_ms": 31.497666030190885}, "total_wall_ms": 38357.99570812378, "ttft_ms": 30.94566590152681, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 59, "facts_retained": 13, "retention_rate": 0.22033898305084745, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.14285714285714285, "amount": 0.0, "status": 0.5833333333333334, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1097}, "ledger_by_hop": {"1": {"facts_total": 59, "facts_retained": 11, "retention_rate": 0.1864406779661017, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.14285714285714285, "amount": 0.0, "status": 0.4166666666666667, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 760}, "2": {"facts_total": 59, "facts_retained": 13, "retention_rate": 0.22033898305084745, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.14285714285714285, "amount": 0.0, "status": 0.5833333333333334, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1097}}}, "reference_key_id": "keys/synthetic/syn_03012_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "e9a21b2d20aaeefae66c356aa30a9c6e588cc37b8ff648442abfac81dba16da2", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "edba7053cee7a3df1c5f0d81", "attempt_id": "90190c0285c7", "run_group_id": "E4_core:A5|closed|qwen3b-llama3b:h2", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:45:38.609882Z", "task_family": "synthetic", "document_id": "syn_03012_2000", "document_hash": "03e441e68d54e1af036f8ff905dfd6cbbccead62636279c1a8e179352ab08945", "question_id": "q12_unit_conversion", "split": "test", "cluster_id": "syn_03012_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "tokenizer_hashes": {"receiver": "b9d8872cf3a99da99e1fc4b0f71232a451632dc49ca4c6771549b4efa5c1c763"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 2, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1767, "query_tokens": 500, "message_tokens": 388, "message_utf8_bytes": 1097, "cache_tensor_bytes": null, "serialized_payload_bytes": 1097, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 37924.419708084315, "hops": {"1": 22652.15183305554, "2": 15272.267875028774}, "receiver_prefill_ms": 356.7475830204785, "ttft_ms": 44.6592919761315, "decode_ms": 163.96104195155203}, "total_wall_ms": 38490.38920807652, "ttft_ms": 44.6592919761315, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 3}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "44,620", "evidence_artifact": {"ledger": {"facts_total": 59, "facts_retained": 13, "retention_rate": 0.22033898305084745, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.14285714285714285, "amount": 0.0, "status": 0.5833333333333334, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1097}, "ledger_by_hop": {"1": {"facts_total": 59, "facts_retained": 11, "retention_rate": 0.1864406779661017, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.14285714285714285, "amount": 0.0, "status": 0.4166666666666667, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 760}, "2": {"facts_total": 59, "facts_retained": 13, "retention_rate": 0.22033898305084745, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.14285714285714285, "amount": 0.0, "status": 0.5833333333333334, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1097}}}, "reference_key_id": "keys/synthetic/syn_03012_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "e9a21b2d20aaeefae66c356aa30a9c6e588cc37b8ff648442abfac81dba16da2", "output_hash": "09a7740f04b91a7ab1ac87709e5c0c6d334d826341de977c51a3b65515114023"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "d8c032e37e3014e8fc3948c4", "attempt_id": "d6dce23bf35a", "run_group_id": "E4_core:A5|closed|qwen3b-llama3b:h2", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:45:39.192740Z", "task_family": "synthetic", "document_id": "syn_03012_2000", "document_hash": "03e441e68d54e1af036f8ff905dfd6cbbccead62636279c1a8e179352ab08945", "question_id": "q13_unanswerable", "split": "test", "cluster_id": "syn_03012_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "tokenizer_hashes": {"receiver": "b9d8872cf3a99da99e1fc4b0f71232a451632dc49ca4c6771549b4efa5c1c763"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 2, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1767, "query_tokens": 504, "message_tokens": 388, "message_utf8_bytes": 1097, "cache_tensor_bytes": null, "serialized_payload_bytes": 1097, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 37924.419708084315, "hops": {"1": 22652.15183305554, "2": 15272.267875028774}, "receiver_prefill_ms": 499.9727909453213, "ttft_ms": 26.130041107535362, "decode_ms": 35.420208936557174}, "total_wall_ms": 38486.55345803127, "ttft_ms": 26.130041107535362, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 59, "facts_retained": 13, "retention_rate": 0.22033898305084745, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.14285714285714285, "amount": 0.0, "status": 0.5833333333333334, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1097}, "ledger_by_hop": {"1": {"facts_total": 59, "facts_retained": 11, "retention_rate": 0.1864406779661017, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.14285714285714285, "amount": 0.0, "status": 0.4166666666666667, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 760}, "2": {"facts_total": 59, "facts_retained": 13, "retention_rate": 0.22033898305084745, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.14285714285714285, "amount": 0.0, "status": 0.5833333333333334, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1097}}}, "reference_key_id": "keys/synthetic/syn_03012_2000.json", "task_scores": {"correct": 1.0, "abstained": 1.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "e9a21b2d20aaeefae66c356aa30a9c6e588cc37b8ff648442abfac81dba16da2", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "6c19f3f86e38e7c96829bf62", "attempt_id": "cf207f2ae69b", "run_group_id": "E4_core:A5|closed|qwen3b-llama3b:h2", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:45:39.772902Z", "task_family": "synthetic", "document_id": "syn_03012_2000", "document_hash": "03e441e68d54e1af036f8ff905dfd6cbbccead62636279c1a8e179352ab08945", "question_id": "q14_unanswerable", "split": "test", "cluster_id": "syn_03012_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "tokenizer_hashes": {"receiver": "b9d8872cf3a99da99e1fc4b0f71232a451632dc49ca4c6771549b4efa5c1c763"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 2, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1767, "query_tokens": 505, "message_tokens": 388, "message_utf8_bytes": 1097, "cache_tensor_bytes": null, "serialized_payload_bytes": 1097, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 37924.419708084315, "hops": {"1": 22652.15183305554, "2": 15272.267875028774}, "receiver_prefill_ms": 368.9234579214826, "ttft_ms": 24.66450002975762, "decode_ms": 43.67333301343024}, "total_wall_ms": 38362.29075002484, "ttft_ms": 24.66450002975762, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 59, "facts_retained": 13, "retention_rate": 0.22033898305084745, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.14285714285714285, "amount": 0.0, "status": 0.5833333333333334, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1097}, "ledger_by_hop": {"1": {"facts_total": 59, "facts_retained": 11, "retention_rate": 0.1864406779661017, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.14285714285714285, "amount": 0.0, "status": 0.4166666666666667, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 760}, "2": {"facts_total": 59, "facts_retained": 13, "retention_rate": 0.22033898305084745, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.14285714285714285, "amount": 0.0, "status": 0.5833333333333334, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1097}}}, "reference_key_id": "keys/synthetic/syn_03012_2000.json", "task_scores": {"correct": 1.0, "abstained": 1.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "e9a21b2d20aaeefae66c356aa30a9c6e588cc37b8ff648442abfac81dba16da2", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "a07231deafcc9de45763e03d", "attempt_id": "f76d15bf453b", "run_group_id": "E4_core:A5|closed|qwen3b-llama3b:h4", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:46:13.524622Z", "task_family": "synthetic", "document_id": "syn_03012_2000", "document_hash": "03e441e68d54e1af036f8ff905dfd6cbbccead62636279c1a8e179352ab08945", "question_id": "q00_single_fact", "split": "test", "cluster_id": "syn_03012_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "tokenizer_hashes": {"receiver": "b9d8872cf3a99da99e1fc4b0f71232a451632dc49ca4c6771549b4efa5c1c763"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 4, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1767, "query_tokens": 492, "message_tokens": 388, "message_utf8_bytes": 1097, "cache_tensor_bytes": null, "serialized_payload_bytes": 1097, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 71218.57541613281, "hops": {"1": 22652.15183305554, "2": 15272.267875028774, "3": 18122.784957988188, "4": 15171.370750060305}, "receiver_prefill_ms": 391.3914579898119, "ttft_ms": 49.55708293709904, "decode_ms": 385.96479198895395}, "total_wall_ms": 72046.11637513153, "ttft_ms": 49.55708293709904, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 7}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "XX-71M-6213", "evidence_artifact": {"ledger": {"facts_total": 59, "facts_retained": 13, "retention_rate": 0.22033898305084745, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.14285714285714285, "amount": 0.0, "status": 0.5833333333333334, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1097}, "ledger_by_hop": {"1": {"facts_total": 59, "facts_retained": 11, "retention_rate": 0.1864406779661017, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.14285714285714285, "amount": 0.0, "status": 0.4166666666666667, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 760}, "2": {"facts_total": 59, "facts_retained": 13, "retention_rate": 0.22033898305084745, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.14285714285714285, "amount": 0.0, "status": 0.5833333333333334, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1097}, "3": {"facts_total": 59, "facts_retained": 13, "retention_rate": 0.22033898305084745, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.14285714285714285, "amount": 0.0, "status": 0.5833333333333334, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1097}, "4": {"facts_total": 59, "facts_retained": 13, "retention_rate": 0.22033898305084745, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.14285714285714285, "amount": 0.0, "status": 0.5833333333333334, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1097}}}, "reference_key_id": "keys/synthetic/syn_03012_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "e9a21b2d20aaeefae66c356aa30a9c6e588cc37b8ff648442abfac81dba16da2", "output_hash": "66ef6c5da682b38795e9029723bbd71cdde2c4dec70cf1fae6efe62ef243f10a"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "20ec5bd47edd4c2c9dcc86b0", "attempt_id": "5a5a15435913", "run_group_id": "E4_core:A5|closed|qwen3b-llama3b:h4", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:46:14.371928Z", "task_family": "synthetic", "document_id": "syn_03012_2000", "document_hash": "03e441e68d54e1af036f8ff905dfd6cbbccead62636279c1a8e179352ab08945", "question_id": "q01_single_fact", "split": "test", "cluster_id": "syn_03012_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "tokenizer_hashes": {"receiver": "b9d8872cf3a99da99e1fc4b0f71232a451632dc49ca4c6771549b4efa5c1c763"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 4, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1767, "query_tokens": 497, "message_tokens": 388, "message_utf8_bytes": 1097, "cache_tensor_bytes": null, "serialized_payload_bytes": 1097, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 71218.57541613281, "hops": {"1": 22652.15183305554, "2": 15272.267875028774, "3": 18122.784957988188, "4": 15171.370750060305}, "receiver_prefill_ms": 379.38129098620266, "ttft_ms": 28.11575005762279, "decode_ms": 39.31616700720042}, "total_wall_ms": 71666.03799920995, "ttft_ms": 28.11575005762279, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 59, "facts_retained": 13, "retention_rate": 0.22033898305084745, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.14285714285714285, "amount": 0.0, "status": 0.5833333333333334, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1097}, "ledger_by_hop": {"1": {"facts_total": 59, "facts_retained": 11, "retention_rate": 0.1864406779661017, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.14285714285714285, "amount": 0.0, "status": 0.4166666666666667, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 760}, "2": {"facts_total": 59, "facts_retained": 13, "retention_rate": 0.22033898305084745, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.14285714285714285, "amount": 0.0, "status": 0.5833333333333334, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1097}, "3": {"facts_total": 59, "facts_retained": 13, "retention_rate": 0.22033898305084745, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.14285714285714285, "amount": 0.0, "status": 0.5833333333333334, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1097}, "4": {"facts_total": 59, "facts_retained": 13, "retention_rate": 0.22033898305084745, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.14285714285714285, "amount": 0.0, "status": 0.5833333333333334, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1097}}}, "reference_key_id": "keys/synthetic/syn_03012_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "e9a21b2d20aaeefae66c356aa30a9c6e588cc37b8ff648442abfac81dba16da2", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "b5a36ac9c858bd3f3a0103cb", "attempt_id": "170dfb8ef4b4", "run_group_id": "E4_core:A5|closed|qwen3b-llama3b:h4", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:46:14.838108Z", "task_family": "synthetic", "document_id": "syn_03012_2000", "document_hash": "03e441e68d54e1af036f8ff905dfd6cbbccead62636279c1a8e179352ab08945", "question_id": "q02_exact_copy", "split": "test", "cluster_id": "syn_03012_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "tokenizer_hashes": {"receiver": "b9d8872cf3a99da99e1fc4b0f71232a451632dc49ca4c6771549b4efa5c1c763"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 4, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1767, "query_tokens": 497, "message_tokens": 388, "message_utf8_bytes": 1097, "cache_tensor_bytes": null, "serialized_payload_bytes": 1097, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 71218.57541613281, "hops": {"1": 22652.15183305554, "2": 15272.267875028774, "3": 18122.784957988188, "4": 15171.370750060305}, "receiver_prefill_ms": 343.03208300843835, "ttft_ms": 42.35445894300938, "decode_ms": 71.01512502413243}, "total_wall_ms": 71675.56062410586, "ttft_ms": 42.35445894300938, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 59, "facts_retained": 13, "retention_rate": 0.22033898305084745, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.14285714285714285, "amount": 0.0, "status": 0.5833333333333334, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1097}, "ledger_by_hop": {"1": {"facts_total": 59, "facts_retained": 11, "retention_rate": 0.1864406779661017, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.14285714285714285, "amount": 0.0, "status": 0.4166666666666667, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 760}, "2": {"facts_total": 59, "facts_retained": 13, "retention_rate": 0.22033898305084745, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.14285714285714285, "amount": 0.0, "status": 0.5833333333333334, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1097}, "3": {"facts_total": 59, "facts_retained": 13, "retention_rate": 0.22033898305084745, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.14285714285714285, "amount": 0.0, "status": 0.5833333333333334, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1097}, "4": {"facts_total": 59, "facts_retained": 13, "retention_rate": 0.22033898305084745, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.14285714285714285, "amount": 0.0, "status": 0.5833333333333334, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1097}}}, "reference_key_id": "keys/synthetic/syn_03012_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "e9a21b2d20aaeefae66c356aa30a9c6e588cc37b8ff648442abfac81dba16da2", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "cd48c2b608f4e6fc106f7502", "attempt_id": "ff7cc4e00f50", "run_group_id": "E4_core:A5|closed|qwen3b-llama3b:h4", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:46:15.312444Z", "task_family": "synthetic", "document_id": "syn_03012_2000", "document_hash": "03e441e68d54e1af036f8ff905dfd6cbbccead62636279c1a8e179352ab08945", "question_id": "q03_exact_copy", "split": "test", "cluster_id": "syn_03012_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "tokenizer_hashes": {"receiver": "b9d8872cf3a99da99e1fc4b0f71232a451632dc49ca4c6771549b4efa5c1c763"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 4, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1767, "query_tokens": 493, "message_tokens": 388, "message_utf8_bytes": 1097, "cache_tensor_bytes": null, "serialized_payload_bytes": 1097, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 71218.57541613281, "hops": {"1": 22652.15183305554, "2": 15272.267875028774, "3": 18122.784957988188, "4": 15171.370750060305}, "receiver_prefill_ms": 507.95491598546505, "ttft_ms": 40.97283398732543, "decode_ms": 21.590333082713187}, "total_wall_ms": 71789.733583224, "ttft_ms": 40.97283398732543, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 59, "facts_retained": 13, "retention_rate": 0.22033898305084745, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.14285714285714285, "amount": 0.0, "status": 0.5833333333333334, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1097}, "ledger_by_hop": {"1": {"facts_total": 59, "facts_retained": 11, "retention_rate": 0.1864406779661017, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.14285714285714285, "amount": 0.0, "status": 0.4166666666666667, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 760}, "2": {"facts_total": 59, "facts_retained": 13, "retention_rate": 0.22033898305084745, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.14285714285714285, "amount": 0.0, "status": 0.5833333333333334, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1097}, "3": {"facts_total": 59, "facts_retained": 13, "retention_rate": 0.22033898305084745, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.14285714285714285, "amount": 0.0, "status": 0.5833333333333334, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1097}, "4": {"facts_total": 59, "facts_retained": 13, "retention_rate": 0.22033898305084745, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.14285714285714285, "amount": 0.0, "status": 0.5833333333333334, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1097}}}, "reference_key_id": "keys/synthetic/syn_03012_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "e9a21b2d20aaeefae66c356aa30a9c6e588cc37b8ff648442abfac81dba16da2", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "e0a41a9f37159fb1933ce89d", "attempt_id": "eea1f7111cf4", "run_group_id": "E4_core:A5|closed|qwen3b-llama3b:h4", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:46:15.902122Z", "task_family": "synthetic", "document_id": "syn_03012_2000", "document_hash": "03e441e68d54e1af036f8ff905dfd6cbbccead62636279c1a8e179352ab08945", "question_id": "q04_multi_hop", "split": "test", "cluster_id": "syn_03012_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "tokenizer_hashes": {"receiver": "b9d8872cf3a99da99e1fc4b0f71232a451632dc49ca4c6771549b4efa5c1c763"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 4, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1767, "query_tokens": 498, "message_tokens": 388, "message_utf8_bytes": 1097, "cache_tensor_bytes": null, "serialized_payload_bytes": 1097, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 71218.57541613281, "hops": {"1": 22652.15183305554, "2": 15272.267875028774, "3": 18122.784957988188, "4": 15171.370750060305}, "receiver_prefill_ms": 371.6199159389362, "ttft_ms": 26.15095896180719, "decode_ms": 37.22341696266085}, "total_wall_ms": 71654.03591608629, "ttft_ms": 26.15095896180719, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 59, "facts_retained": 13, "retention_rate": 0.22033898305084745, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.14285714285714285, "amount": 0.0, "status": 0.5833333333333334, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1097}, "ledger_by_hop": {"1": {"facts_total": 59, "facts_retained": 11, "retention_rate": 0.1864406779661017, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.14285714285714285, "amount": 0.0, "status": 0.4166666666666667, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 760}, "2": {"facts_total": 59, "facts_retained": 13, "retention_rate": 0.22033898305084745, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.14285714285714285, "amount": 0.0, "status": 0.5833333333333334, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1097}, "3": {"facts_total": 59, "facts_retained": 13, "retention_rate": 0.22033898305084745, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.14285714285714285, "amount": 0.0, "status": 0.5833333333333334, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1097}, "4": {"facts_total": 59, "facts_retained": 13, "retention_rate": 0.22033898305084745, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.14285714285714285, "amount": 0.0, "status": 0.5833333333333334, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1097}}}, "reference_key_id": "keys/synthetic/syn_03012_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "e9a21b2d20aaeefae66c356aa30a9c6e588cc37b8ff648442abfac81dba16da2", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "e264ab4aa092d9619335a8b4", "attempt_id": "eab9657c5b1e", "run_group_id": "E4_core:A5|closed|qwen3b-llama3b:h4", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:46:16.355117Z", "task_family": "synthetic", "document_id": "syn_03012_2000", "document_hash": "03e441e68d54e1af036f8ff905dfd6cbbccead62636279c1a8e179352ab08945", "question_id": "q05_multi_hop", "split": "test", "cluster_id": "syn_03012_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "tokenizer_hashes": {"receiver": "b9d8872cf3a99da99e1fc4b0f71232a451632dc49ca4c6771549b4efa5c1c763"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 4, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1767, "query_tokens": 499, "message_tokens": 388, "message_utf8_bytes": 1097, "cache_tensor_bytes": null, "serialized_payload_bytes": 1097, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 71218.57541613281, "hops": {"1": 22652.15183305554, "2": 15272.267875028774, "3": 18122.784957988188, "4": 15171.370750060305}, "receiver_prefill_ms": 346.98716597631574, "ttft_ms": 36.36599995661527, "decode_ms": 21.10616595018655}, "total_wall_ms": 71623.4979161527, "ttft_ms": 36.36599995661527, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 59, "facts_retained": 13, "retention_rate": 0.22033898305084745, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.14285714285714285, "amount": 0.0, "status": 0.5833333333333334, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1097}, "ledger_by_hop": {"1": {"facts_total": 59, "facts_retained": 11, "retention_rate": 0.1864406779661017, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.14285714285714285, "amount": 0.0, "status": 0.4166666666666667, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 760}, "2": {"facts_total": 59, "facts_retained": 13, "retention_rate": 0.22033898305084745, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.14285714285714285, "amount": 0.0, "status": 0.5833333333333334, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1097}, "3": {"facts_total": 59, "facts_retained": 13, "retention_rate": 0.22033898305084745, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.14285714285714285, "amount": 0.0, "status": 0.5833333333333334, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1097}, "4": {"facts_total": 59, "facts_retained": 13, "retention_rate": 0.22033898305084745, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.14285714285714285, "amount": 0.0, "status": 0.5833333333333334, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1097}}}, "reference_key_id": "keys/synthetic/syn_03012_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "e9a21b2d20aaeefae66c356aa30a9c6e588cc37b8ff648442abfac81dba16da2", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "04efe30583434ad1153ad2d2", "attempt_id": "a85cce9ba681", "run_group_id": "E4_core:A5|closed|qwen3b-llama3b:h4", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:46:16.776918Z", "task_family": "synthetic", "document_id": "syn_03012_2000", "document_hash": "03e441e68d54e1af036f8ff905dfd6cbbccead62636279c1a8e179352ab08945", "question_id": "q06_negation", "split": "test", "cluster_id": "syn_03012_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "tokenizer_hashes": {"receiver": "b9d8872cf3a99da99e1fc4b0f71232a451632dc49ca4c6771549b4efa5c1c763"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 4, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1767, "query_tokens": 492, "message_tokens": 388, "message_utf8_bytes": 1097, "cache_tensor_bytes": null, "serialized_payload_bytes": 1097, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 71218.57541613281, "hops": {"1": 22652.15183305554, "2": 15272.267875028774, "3": 18122.784957988188, "4": 15171.370750060305}, "receiver_prefill_ms": 368.1285419734195, "ttft_ms": 46.1747080553323, "decode_ms": 47.794791986234486}, "total_wall_ms": 71681.44958210178, "ttft_ms": 46.1747080553323, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "No", "evidence_artifact": {"ledger": {"facts_total": 59, "facts_retained": 13, "retention_rate": 0.22033898305084745, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.14285714285714285, "amount": 0.0, "status": 0.5833333333333334, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1097}, "ledger_by_hop": {"1": {"facts_total": 59, "facts_retained": 11, "retention_rate": 0.1864406779661017, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.14285714285714285, "amount": 0.0, "status": 0.4166666666666667, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 760}, "2": {"facts_total": 59, "facts_retained": 13, "retention_rate": 0.22033898305084745, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.14285714285714285, "amount": 0.0, "status": 0.5833333333333334, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1097}, "3": {"facts_total": 59, "facts_retained": 13, "retention_rate": 0.22033898305084745, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.14285714285714285, "amount": 0.0, "status": 0.5833333333333334, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1097}, "4": {"facts_total": 59, "facts_retained": 13, "retention_rate": 0.22033898305084745, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.14285714285714285, "amount": 0.0, "status": 0.5833333333333334, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1097}}}, "reference_key_id": "keys/synthetic/syn_03012_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "e9a21b2d20aaeefae66c356aa30a9c6e588cc37b8ff648442abfac81dba16da2", "output_hash": "1ea442a134b2a184bd5d40104401f2a37fbc09ccf3f4bc9da161c6099be3691d"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "d2dea4994fb2d9cda52086a4", "attempt_id": "3de6edb697ea", "run_group_id": "E4_core:A5|closed|qwen3b-llama3b:h4", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:46:17.258191Z", "task_family": "synthetic", "document_id": "syn_03012_2000", "document_hash": "03e441e68d54e1af036f8ff905dfd6cbbccead62636279c1a8e179352ab08945", "question_id": "q07_negation", "split": "test", "cluster_id": "syn_03012_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "tokenizer_hashes": {"receiver": "b9d8872cf3a99da99e1fc4b0f71232a451632dc49ca4c6771549b4efa5c1c763"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 4, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1767, "query_tokens": 493, "message_tokens": 388, "message_utf8_bytes": 1097, "cache_tensor_bytes": null, "serialized_payload_bytes": 1097, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 71218.57541613281, "hops": {"1": 22652.15183305554, "2": 15272.267875028774, "3": 18122.784957988188, "4": 15171.370750060305}, "receiver_prefill_ms": 481.7270420026034, "ttft_ms": 35.83637508563697, "decode_ms": 23.424374987371266}, "total_wall_ms": 71760.1349571487, "ttft_ms": 35.83637508563697, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "No", "evidence_artifact": {"ledger": {"facts_total": 59, "facts_retained": 13, "retention_rate": 0.22033898305084745, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.14285714285714285, "amount": 0.0, "status": 0.5833333333333334, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1097}, "ledger_by_hop": {"1": {"facts_total": 59, "facts_retained": 11, "retention_rate": 0.1864406779661017, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.14285714285714285, "amount": 0.0, "status": 0.4166666666666667, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 760}, "2": {"facts_total": 59, "facts_retained": 13, "retention_rate": 0.22033898305084745, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.14285714285714285, "amount": 0.0, "status": 0.5833333333333334, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1097}, "3": {"facts_total": 59, "facts_retained": 13, "retention_rate": 0.22033898305084745, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.14285714285714285, "amount": 0.0, "status": 0.5833333333333334, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1097}, "4": {"facts_total": 59, "facts_retained": 13, "retention_rate": 0.22033898305084745, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.14285714285714285, "amount": 0.0, "status": 0.5833333333333334, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1097}}}, "reference_key_id": "keys/synthetic/syn_03012_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "e9a21b2d20aaeefae66c356aa30a9c6e588cc37b8ff648442abfac81dba16da2", "output_hash": "1ea442a134b2a184bd5d40104401f2a37fbc09ccf3f4bc9da161c6099be3691d"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "ced40935dbce1df80700a3c9", "attempt_id": "52e907162780", "run_group_id": "E4_core:A5|closed|qwen3b-llama3b:h4", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:46:17.817016Z", "task_family": "synthetic", "document_id": "syn_03012_2000", "document_hash": "03e441e68d54e1af036f8ff905dfd6cbbccead62636279c1a8e179352ab08945", "question_id": "q08_revision", "split": "test", "cluster_id": "syn_03012_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "tokenizer_hashes": {"receiver": "b9d8872cf3a99da99e1fc4b0f71232a451632dc49ca4c6771549b4efa5c1c763"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 4, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1767, "query_tokens": 508, "message_tokens": 388, "message_utf8_bytes": 1097, "cache_tensor_bytes": null, "serialized_payload_bytes": 1097, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 71218.57541613281, "hops": {"1": 22652.15183305554, "2": 15272.267875028774, "3": 18122.784957988188, "4": 15171.370750060305}, "receiver_prefill_ms": 371.1937910411507, "ttft_ms": 40.902042062953115, "decode_ms": 23.04720797110349}, "total_wall_ms": 71654.2831241386, "ttft_ms": 40.902042062953115, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 59, "facts_retained": 13, "retention_rate": 0.22033898305084745, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.14285714285714285, "amount": 0.0, "status": 0.5833333333333334, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1097}, "ledger_by_hop": {"1": {"facts_total": 59, "facts_retained": 11, "retention_rate": 0.1864406779661017, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.14285714285714285, "amount": 0.0, "status": 0.4166666666666667, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 760}, "2": {"facts_total": 59, "facts_retained": 13, "retention_rate": 0.22033898305084745, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.14285714285714285, "amount": 0.0, "status": 0.5833333333333334, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1097}, "3": {"facts_total": 59, "facts_retained": 13, "retention_rate": 0.22033898305084745, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.14285714285714285, "amount": 0.0, "status": 0.5833333333333334, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1097}, "4": {"facts_total": 59, "facts_retained": 13, "retention_rate": 0.22033898305084745, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.14285714285714285, "amount": 0.0, "status": 0.5833333333333334, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1097}}}, "reference_key_id": "keys/synthetic/syn_03012_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "e9a21b2d20aaeefae66c356aa30a9c6e588cc37b8ff648442abfac81dba16da2", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "ca1508267a0e9e61d15e4363", "attempt_id": "6298a5e167b6", "run_group_id": "E4_core:A5|closed|qwen3b-llama3b:h4", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:46:18.270114Z", "task_family": "synthetic", "document_id": "syn_03012_2000", "document_hash": "03e441e68d54e1af036f8ff905dfd6cbbccead62636279c1a8e179352ab08945", "question_id": "q09_revision", "split": "test", "cluster_id": "syn_03012_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "tokenizer_hashes": {"receiver": "b9d8872cf3a99da99e1fc4b0f71232a451632dc49ca4c6771549b4efa5c1c763"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 4, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1767, "query_tokens": 500, "message_tokens": 388, "message_utf8_bytes": 1097, "cache_tensor_bytes": null, "serialized_payload_bytes": 1097, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 71218.57541613281, "hops": {"1": 22652.15183305554, "2": 15272.267875028774, "3": 18122.784957988188, "4": 15171.370750060305}, "receiver_prefill_ms": 347.822375013493, "ttft_ms": 61.78787490352988, "decode_ms": 57.215332984924316}, "total_wall_ms": 71686.29587406758, "ttft_ms": 61.78787490352988, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 59, "facts_retained": 13, "retention_rate": 0.22033898305084745, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.14285714285714285, "amount": 0.0, "status": 0.5833333333333334, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1097}, "ledger_by_hop": {"1": {"facts_total": 59, "facts_retained": 11, "retention_rate": 0.1864406779661017, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.14285714285714285, "amount": 0.0, "status": 0.4166666666666667, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 760}, "2": {"facts_total": 59, "facts_retained": 13, "retention_rate": 0.22033898305084745, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.14285714285714285, "amount": 0.0, "status": 0.5833333333333334, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1097}, "3": {"facts_total": 59, "facts_retained": 13, "retention_rate": 0.22033898305084745, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.14285714285714285, "amount": 0.0, "status": 0.5833333333333334, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1097}, "4": {"facts_total": 59, "facts_retained": 13, "retention_rate": 0.22033898305084745, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.14285714285714285, "amount": 0.0, "status": 0.5833333333333334, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1097}}}, "reference_key_id": "keys/synthetic/syn_03012_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "e9a21b2d20aaeefae66c356aa30a9c6e588cc37b8ff648442abfac81dba16da2", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "7c635f5f8ad7415307823b26", "attempt_id": "7833029ea392", "run_group_id": "E4_core:A5|closed|qwen3b-llama3b:h4", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:46:18.755187Z", "task_family": "synthetic", "document_id": "syn_03012_2000", "document_hash": "03e441e68d54e1af036f8ff905dfd6cbbccead62636279c1a8e179352ab08945", "question_id": "q10_temporal", "split": "test", "cluster_id": "syn_03012_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "tokenizer_hashes": {"receiver": "b9d8872cf3a99da99e1fc4b0f71232a451632dc49ca4c6771549b4efa5c1c763"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 4, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1767, "query_tokens": 510, "message_tokens": 388, "message_utf8_bytes": 1097, "cache_tensor_bytes": null, "serialized_payload_bytes": 1097, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 71218.57541613281, "hops": {"1": 22652.15183305554, "2": 15272.267875028774, "3": 18122.784957988188, "4": 15171.370750060305}, "receiver_prefill_ms": 488.63287502899766, "ttft_ms": 37.076041917316616, "decode_ms": 20.791208022274077}, "total_wall_ms": 71765.66233311314, "ttft_ms": 37.076041917316616, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 59, "facts_retained": 13, "retention_rate": 0.22033898305084745, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.14285714285714285, "amount": 0.0, "status": 0.5833333333333334, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1097}, "ledger_by_hop": {"1": {"facts_total": 59, "facts_retained": 11, "retention_rate": 0.1864406779661017, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.14285714285714285, "amount": 0.0, "status": 0.4166666666666667, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 760}, "2": {"facts_total": 59, "facts_retained": 13, "retention_rate": 0.22033898305084745, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.14285714285714285, "amount": 0.0, "status": 0.5833333333333334, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1097}, "3": {"facts_total": 59, "facts_retained": 13, "retention_rate": 0.22033898305084745, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.14285714285714285, "amount": 0.0, "status": 0.5833333333333334, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1097}, "4": {"facts_total": 59, "facts_retained": 13, "retention_rate": 0.22033898305084745, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.14285714285714285, "amount": 0.0, "status": 0.5833333333333334, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1097}}}, "reference_key_id": "keys/synthetic/syn_03012_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "e9a21b2d20aaeefae66c356aa30a9c6e588cc37b8ff648442abfac81dba16da2", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "c9913a4b584c980c20bdb9c8", "attempt_id": "de6ee985fd5a", "run_group_id": "E4_core:A5|closed|qwen3b-llama3b:h4", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:46:19.319720Z", "task_family": "synthetic", "document_id": "syn_03012_2000", "document_hash": "03e441e68d54e1af036f8ff905dfd6cbbccead62636279c1a8e179352ab08945", "question_id": "q11_aggregation", "split": "test", "cluster_id": "syn_03012_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "tokenizer_hashes": {"receiver": "b9d8872cf3a99da99e1fc4b0f71232a451632dc49ca4c6771549b4efa5c1c763"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 4, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1767, "query_tokens": 503, "message_tokens": 388, "message_utf8_bytes": 1097, "cache_tensor_bytes": null, "serialized_payload_bytes": 1097, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 71218.57541613281, "hops": {"1": 22652.15183305554, "2": 15272.267875028774, "3": 18122.784957988188, "4": 15171.370750060305}, "receiver_prefill_ms": 371.1810839595273, "ttft_ms": 41.617332957684994, "decode_ms": 24.159874999895692}, "total_wall_ms": 71656.1015821062, "ttft_ms": 41.617332957684994, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 59, "facts_retained": 13, "retention_rate": 0.22033898305084745, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.14285714285714285, "amount": 0.0, "status": 0.5833333333333334, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1097}, "ledger_by_hop": {"1": {"facts_total": 59, "facts_retained": 11, "retention_rate": 0.1864406779661017, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.14285714285714285, "amount": 0.0, "status": 0.4166666666666667, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 760}, "2": {"facts_total": 59, "facts_retained": 13, "retention_rate": 0.22033898305084745, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.14285714285714285, "amount": 0.0, "status": 0.5833333333333334, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1097}, "3": {"facts_total": 59, "facts_retained": 13, "retention_rate": 0.22033898305084745, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.14285714285714285, "amount": 0.0, "status": 0.5833333333333334, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1097}, "4": {"facts_total": 59, "facts_retained": 13, "retention_rate": 0.22033898305084745, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.14285714285714285, "amount": 0.0, "status": 0.5833333333333334, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1097}}}, "reference_key_id": "keys/synthetic/syn_03012_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "e9a21b2d20aaeefae66c356aa30a9c6e588cc37b8ff648442abfac81dba16da2", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "8593a4c3d404262f5a3f7eac", "attempt_id": "0e1cb8e7e608", "run_group_id": "E4_core:A5|closed|qwen3b-llama3b:h4", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:46:19.775034Z", "task_family": "synthetic", "document_id": "syn_03012_2000", "document_hash": "03e441e68d54e1af036f8ff905dfd6cbbccead62636279c1a8e179352ab08945", "question_id": "q12_unit_conversion", "split": "test", "cluster_id": "syn_03012_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "tokenizer_hashes": {"receiver": "b9d8872cf3a99da99e1fc4b0f71232a451632dc49ca4c6771549b4efa5c1c763"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 4, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1767, "query_tokens": 500, "message_tokens": 388, "message_utf8_bytes": 1097, "cache_tensor_bytes": null, "serialized_payload_bytes": 1097, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 71218.57541613281, "hops": {"1": 22652.15183305554, "2": 15272.267875028774, "3": 18122.784957988188, "4": 15171.370750060305}, "receiver_prefill_ms": 350.7626249920577, "ttft_ms": 32.71083300933242, "decode_ms": 140.07295796182007}, "total_wall_ms": 71742.68566619139, "ttft_ms": 32.71083300933242, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 3}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "44,620", "evidence_artifact": {"ledger": {"facts_total": 59, "facts_retained": 13, "retention_rate": 0.22033898305084745, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.14285714285714285, "amount": 0.0, "status": 0.5833333333333334, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1097}, "ledger_by_hop": {"1": {"facts_total": 59, "facts_retained": 11, "retention_rate": 0.1864406779661017, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.14285714285714285, "amount": 0.0, "status": 0.4166666666666667, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 760}, "2": {"facts_total": 59, "facts_retained": 13, "retention_rate": 0.22033898305084745, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.14285714285714285, "amount": 0.0, "status": 0.5833333333333334, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1097}, "3": {"facts_total": 59, "facts_retained": 13, "retention_rate": 0.22033898305084745, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.14285714285714285, "amount": 0.0, "status": 0.5833333333333334, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1097}, "4": {"facts_total": 59, "facts_retained": 13, "retention_rate": 0.22033898305084745, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.14285714285714285, "amount": 0.0, "status": 0.5833333333333334, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1097}}}, "reference_key_id": "keys/synthetic/syn_03012_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "e9a21b2d20aaeefae66c356aa30a9c6e588cc37b8ff648442abfac81dba16da2", "output_hash": "09a7740f04b91a7ab1ac87709e5c0c6d334d826341de977c51a3b65515114023"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "5fa6267c857fd170fa71553e", "attempt_id": "f35acec2365e", "run_group_id": "E4_core:A5|closed|qwen3b-llama3b:h4", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:46:20.317364Z", "task_family": "synthetic", "document_id": "syn_03012_2000", "document_hash": "03e441e68d54e1af036f8ff905dfd6cbbccead62636279c1a8e179352ab08945", "question_id": "q13_unanswerable", "split": "test", "cluster_id": "syn_03012_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "tokenizer_hashes": {"receiver": "b9d8872cf3a99da99e1fc4b0f71232a451632dc49ca4c6771549b4efa5c1c763"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 4, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1767, "query_tokens": 504, "message_tokens": 388, "message_utf8_bytes": 1097, "cache_tensor_bytes": null, "serialized_payload_bytes": 1097, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 71218.57541613281, "hops": {"1": 22652.15183305554, "2": 15272.267875028774, "3": 18122.784957988188, "4": 15171.370750060305}, "receiver_prefill_ms": 434.17074996978045, "ttft_ms": 81.82691701222211, "decode_ms": 39.69383297953755}, "total_wall_ms": 71774.75962415338, "ttft_ms": 81.82691701222211, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 59, "facts_retained": 13, "retention_rate": 0.22033898305084745, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.14285714285714285, "amount": 0.0, "status": 0.5833333333333334, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1097}, "ledger_by_hop": {"1": {"facts_total": 59, "facts_retained": 11, "retention_rate": 0.1864406779661017, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.14285714285714285, "amount": 0.0, "status": 0.4166666666666667, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 760}, "2": {"facts_total": 59, "facts_retained": 13, "retention_rate": 0.22033898305084745, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.14285714285714285, "amount": 0.0, "status": 0.5833333333333334, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1097}, "3": {"facts_total": 59, "facts_retained": 13, "retention_rate": 0.22033898305084745, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.14285714285714285, "amount": 0.0, "status": 0.5833333333333334, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1097}, "4": {"facts_total": 59, "facts_retained": 13, "retention_rate": 0.22033898305084745, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.14285714285714285, "amount": 0.0, "status": 0.5833333333333334, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1097}}}, "reference_key_id": "keys/synthetic/syn_03012_2000.json", "task_scores": {"correct": 1.0, "abstained": 1.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "e9a21b2d20aaeefae66c356aa30a9c6e588cc37b8ff648442abfac81dba16da2", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "fdfe927407134eeeb31b1871", "attempt_id": "05fee8eb460c", "run_group_id": "E4_core:A5|closed|qwen3b-llama3b:h4", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:46:20.891043Z", "task_family": "synthetic", "document_id": "syn_03012_2000", "document_hash": "03e441e68d54e1af036f8ff905dfd6cbbccead62636279c1a8e179352ab08945", "question_id": "q14_unanswerable", "split": "test", "cluster_id": "syn_03012_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "tokenizer_hashes": {"receiver": "b9d8872cf3a99da99e1fc4b0f71232a451632dc49ca4c6771549b4efa5c1c763"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 4, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1767, "query_tokens": 505, "message_tokens": 388, "message_utf8_bytes": 1097, "cache_tensor_bytes": null, "serialized_payload_bytes": 1097, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 71218.57541613281, "hops": {"1": 22652.15183305554, "2": 15272.267875028774, "3": 18122.784957988188, "4": 15171.370750060305}, "receiver_prefill_ms": 375.4448329564184, "ttft_ms": 27.085333014838398, "decode_ms": 34.985249978490174}, "total_wall_ms": 71656.75770817325, "ttft_ms": 27.085333014838398, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 59, "facts_retained": 13, "retention_rate": 0.22033898305084745, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.14285714285714285, "amount": 0.0, "status": 0.5833333333333334, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1097}, "ledger_by_hop": {"1": {"facts_total": 59, "facts_retained": 11, "retention_rate": 0.1864406779661017, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.14285714285714285, "amount": 0.0, "status": 0.4166666666666667, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 760}, "2": {"facts_total": 59, "facts_retained": 13, "retention_rate": 0.22033898305084745, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.14285714285714285, "amount": 0.0, "status": 0.5833333333333334, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1097}, "3": {"facts_total": 59, "facts_retained": 13, "retention_rate": 0.22033898305084745, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.14285714285714285, "amount": 0.0, "status": 0.5833333333333334, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1097}, "4": {"facts_total": 59, "facts_retained": 13, "retention_rate": 0.22033898305084745, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.14285714285714285, "amount": 0.0, "status": 0.5833333333333334, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1097}}}, "reference_key_id": "keys/synthetic/syn_03012_2000.json", "task_scores": {"correct": 1.0, "abstained": 1.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "e9a21b2d20aaeefae66c356aa30a9c6e588cc37b8ff648442abfac81dba16da2", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "797a5f18424b8c9d486fc13d", "attempt_id": "f8e77c9c1f7c", "run_group_id": "E4_core:A5|closed|qwen3b-llama3b:h8", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:47:28.515882Z", "task_family": "synthetic", "document_id": "syn_03012_2000", "document_hash": "03e441e68d54e1af036f8ff905dfd6cbbccead62636279c1a8e179352ab08945", "question_id": "q00_single_fact", "split": "test", "cluster_id": "syn_03012_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "tokenizer_hashes": {"receiver": "b9d8872cf3a99da99e1fc4b0f71232a451632dc49ca4c6771549b4efa5c1c763"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 8, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1767, "query_tokens": 492, "message_tokens": 388, "message_utf8_bytes": 1097, "cache_tensor_bytes": null, "serialized_payload_bytes": 1097, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 138383.33570817485, "hops": {"1": 22652.15183305554, "2": 15272.267875028774, "3": 18122.784957988188, "4": 15171.370750060305, "5": 18196.480125072412, "6": 15206.121249939315, "7": 18390.757792047225, "8": 15371.401124983095}, "receiver_prefill_ms": 384.9354999838397, "ttft_ms": 19.83691705390811, "decode_ms": 229.56937493290752}, "total_wall_ms": 139018.27166613657, "ttft_ms": 19.83691705390811, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 7}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "XX-71M-6213", "evidence_artifact": {"ledger": {"facts_total": 59, "facts_retained": 13, "retention_rate": 0.22033898305084745, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.14285714285714285, "amount": 0.0, "status": 0.5833333333333334, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1097}, "ledger_by_hop": {"1": {"facts_total": 59, "facts_retained": 11, "retention_rate": 0.1864406779661017, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.14285714285714285, "amount": 0.0, "status": 0.4166666666666667, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 760}, "2": {"facts_total": 59, "facts_retained": 13, "retention_rate": 0.22033898305084745, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.14285714285714285, "amount": 0.0, "status": 0.5833333333333334, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1097}, "3": {"facts_total": 59, "facts_retained": 13, "retention_rate": 0.22033898305084745, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.14285714285714285, "amount": 0.0, "status": 0.5833333333333334, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1097}, "4": {"facts_total": 59, "facts_retained": 13, "retention_rate": 0.22033898305084745, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.14285714285714285, "amount": 0.0, "status": 0.5833333333333334, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1097}, "5": {"facts_total": 59, "facts_retained": 13, "retention_rate": 0.22033898305084745, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.14285714285714285, "amount": 0.0, "status": 0.5833333333333334, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1097}, "6": {"facts_total": 59, "facts_retained": 13, "retention_rate": 0.22033898305084745, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.14285714285714285, "amount": 0.0, "status": 0.5833333333333334, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1097}, "7": {"facts_total": 59, "facts_retained": 13, "retention_rate": 0.22033898305084745, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.14285714285714285, "amount": 0.0, "status": 0.5833333333333334, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1097}, "8": {"facts_total": 59, "facts_retained": 13, "retention_rate": 0.22033898305084745, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.14285714285714285, "amount": 0.0, "status": 0.5833333333333334, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1097}}}, "reference_key_id": "keys/synthetic/syn_03012_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "e9a21b2d20aaeefae66c356aa30a9c6e588cc37b8ff648442abfac81dba16da2", "output_hash": "66ef6c5da682b38795e9029723bbd71cdde2c4dec70cf1fae6efe62ef243f10a"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "38480688cea4e826b4997d8f", "attempt_id": "da3c0a1809dc", "run_group_id": "E4_core:A5|closed|qwen3b-llama3b:h8", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:47:29.169291Z", "task_family": "synthetic", "document_id": "syn_03012_2000", "document_hash": "03e441e68d54e1af036f8ff905dfd6cbbccead62636279c1a8e179352ab08945", "question_id": "q01_single_fact", "split": "test", "cluster_id": "syn_03012_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "tokenizer_hashes": {"receiver": "b9d8872cf3a99da99e1fc4b0f71232a451632dc49ca4c6771549b4efa5c1c763"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 8, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1767, "query_tokens": 497, "message_tokens": 388, "message_utf8_bytes": 1097, "cache_tensor_bytes": null, "serialized_payload_bytes": 1097, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 138383.33570817485, "hops": {"1": 22652.15183305554, "2": 15272.267875028774, "3": 18122.784957988188, "4": 15171.370750060305, "5": 18196.480125072412, "6": 15206.121249939315, "7": 18390.757792047225, "8": 15371.401124983095}, "receiver_prefill_ms": 389.9087079335004, "ttft_ms": 138.05454201065004, "decode_ms": 38.57137507293373}, "total_wall_ms": 138950.44037513435, "ttft_ms": 138.05454201065004, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 59, "facts_retained": 13, "retention_rate": 0.22033898305084745, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.14285714285714285, "amount": 0.0, "status": 0.5833333333333334, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1097}, "ledger_by_hop": {"1": {"facts_total": 59, "facts_retained": 11, "retention_rate": 0.1864406779661017, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.14285714285714285, "amount": 0.0, "status": 0.4166666666666667, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 760}, "2": {"facts_total": 59, "facts_retained": 13, "retention_rate": 0.22033898305084745, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.14285714285714285, "amount": 0.0, "status": 0.5833333333333334, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1097}, "3": {"facts_total": 59, "facts_retained": 13, "retention_rate": 0.22033898305084745, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.14285714285714285, "amount": 0.0, "status": 0.5833333333333334, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1097}, "4": {"facts_total": 59, "facts_retained": 13, "retention_rate": 0.22033898305084745, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.14285714285714285, "amount": 0.0, "status": 0.5833333333333334, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1097}, "5": {"facts_total": 59, "facts_retained": 13, "retention_rate": 0.22033898305084745, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.14285714285714285, "amount": 0.0, "status": 0.5833333333333334, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1097}, "6": {"facts_total": 59, "facts_retained": 13, "retention_rate": 0.22033898305084745, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.14285714285714285, "amount": 0.0, "status": 0.5833333333333334, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1097}, "7": {"facts_total": 59, "facts_retained": 13, "retention_rate": 0.22033898305084745, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.14285714285714285, "amount": 0.0, "status": 0.5833333333333334, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1097}, "8": {"facts_total": 59, "facts_retained": 13, "retention_rate": 0.22033898305084745, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.14285714285714285, "amount": 0.0, "status": 0.5833333333333334, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1097}}}, "reference_key_id": "keys/synthetic/syn_03012_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "e9a21b2d20aaeefae66c356aa30a9c6e588cc37b8ff648442abfac81dba16da2", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "aa4d5d44dfba45241a125088", "attempt_id": "0e73fbbb0a97", "run_group_id": "E4_core:A5|closed|qwen3b-llama3b:h8", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:47:29.754581Z", "task_family": "synthetic", "document_id": "syn_03012_2000", "document_hash": "03e441e68d54e1af036f8ff905dfd6cbbccead62636279c1a8e179352ab08945", "question_id": "q02_exact_copy", "split": "test", "cluster_id": "syn_03012_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "tokenizer_hashes": {"receiver": "b9d8872cf3a99da99e1fc4b0f71232a451632dc49ca4c6771549b4efa5c1c763"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 8, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1767, "query_tokens": 497, "message_tokens": 388, "message_utf8_bytes": 1097, "cache_tensor_bytes": null, "serialized_payload_bytes": 1097, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 138383.33570817485, "hops": {"1": 22652.15183305554, "2": 15272.267875028774, "3": 18122.784957988188, "4": 15171.370750060305, "5": 18196.480125072412, "6": 15206.121249939315, "7": 18390.757792047225, "8": 15371.401124983095}, "receiver_prefill_ms": 384.7897909581661, "ttft_ms": 24.472792050801218, "decode_ms": 40.04870797507465}, "total_wall_ms": 138833.1234161742, "ttft_ms": 24.472792050801218, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 59, "facts_retained": 13, "retention_rate": 0.22033898305084745, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.14285714285714285, "amount": 0.0, "status": 0.5833333333333334, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1097}, "ledger_by_hop": {"1": {"facts_total": 59, "facts_retained": 11, "retention_rate": 0.1864406779661017, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.14285714285714285, "amount": 0.0, "status": 0.4166666666666667, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 760}, "2": {"facts_total": 59, "facts_retained": 13, "retention_rate": 0.22033898305084745, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.14285714285714285, "amount": 0.0, "status": 0.5833333333333334, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1097}, "3": {"facts_total": 59, "facts_retained": 13, "retention_rate": 0.22033898305084745, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.14285714285714285, "amount": 0.0, "status": 0.5833333333333334, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1097}, "4": {"facts_total": 59, "facts_retained": 13, "retention_rate": 0.22033898305084745, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.14285714285714285, "amount": 0.0, "status": 0.5833333333333334, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1097}, "5": {"facts_total": 59, "facts_retained": 13, "retention_rate": 0.22033898305084745, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.14285714285714285, "amount": 0.0, "status": 0.5833333333333334, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1097}, "6": {"facts_total": 59, "facts_retained": 13, "retention_rate": 0.22033898305084745, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.14285714285714285, "amount": 0.0, "status": 0.5833333333333334, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1097}, "7": {"facts_total": 59, "facts_retained": 13, "retention_rate": 0.22033898305084745, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.14285714285714285, "amount": 0.0, "status": 0.5833333333333334, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1097}, "8": {"facts_total": 59, "facts_retained": 13, "retention_rate": 0.22033898305084745, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.14285714285714285, "amount": 0.0, "status": 0.5833333333333334, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1097}}}, "reference_key_id": "keys/synthetic/syn_03012_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "e9a21b2d20aaeefae66c356aa30a9c6e588cc37b8ff648442abfac81dba16da2", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "0cc5ff2e8ab011249895063a", "attempt_id": "a4d5e9eaaf2d", "run_group_id": "E4_core:A5|closed|qwen3b-llama3b:h8", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:47:30.221739Z", "task_family": "synthetic", "document_id": "syn_03012_2000", "document_hash": "03e441e68d54e1af036f8ff905dfd6cbbccead62636279c1a8e179352ab08945", "question_id": "q03_exact_copy", "split": "test", "cluster_id": "syn_03012_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "tokenizer_hashes": {"receiver": "b9d8872cf3a99da99e1fc4b0f71232a451632dc49ca4c6771549b4efa5c1c763"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 8, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1767, "query_tokens": 493, "message_tokens": 388, "message_utf8_bytes": 1097, "cache_tensor_bytes": null, "serialized_payload_bytes": 1097, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 138383.33570817485, "hops": {"1": 22652.15183305554, "2": 15272.267875028774, "3": 18122.784957988188, "4": 15171.370750060305, "5": 18196.480125072412, "6": 15206.121249939315, "7": 18390.757792047225, "8": 15371.401124983095}, "receiver_prefill_ms": 374.14799991529435, "ttft_ms": 19.330041017383337, "decode_ms": 18.61304196063429}, "total_wall_ms": 138795.93437514268, "ttft_ms": 19.330041017383337, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 59, "facts_retained": 13, "retention_rate": 0.22033898305084745, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.14285714285714285, "amount": 0.0, "status": 0.5833333333333334, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1097}, "ledger_by_hop": {"1": {"facts_total": 59, "facts_retained": 11, "retention_rate": 0.1864406779661017, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.14285714285714285, "amount": 0.0, "status": 0.4166666666666667, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 760}, "2": {"facts_total": 59, "facts_retained": 13, "retention_rate": 0.22033898305084745, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.14285714285714285, "amount": 0.0, "status": 0.5833333333333334, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1097}, "3": {"facts_total": 59, "facts_retained": 13, "retention_rate": 0.22033898305084745, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.14285714285714285, "amount": 0.0, "status": 0.5833333333333334, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1097}, "4": {"facts_total": 59, "facts_retained": 13, "retention_rate": 0.22033898305084745, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.14285714285714285, "amount": 0.0, "status": 0.5833333333333334, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1097}, "5": {"facts_total": 59, "facts_retained": 13, "retention_rate": 0.22033898305084745, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.14285714285714285, "amount": 0.0, "status": 0.5833333333333334, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1097}, "6": {"facts_total": 59, "facts_retained": 13, "retention_rate": 0.22033898305084745, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.14285714285714285, "amount": 0.0, "status": 0.5833333333333334, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1097}, "7": {"facts_total": 59, "facts_retained": 13, "retention_rate": 0.22033898305084745, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.14285714285714285, "amount": 0.0, "status": 0.5833333333333334, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1097}, "8": {"facts_total": 59, "facts_retained": 13, "retention_rate": 0.22033898305084745, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.14285714285714285, "amount": 0.0, "status": 0.5833333333333334, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1097}}}, "reference_key_id": "keys/synthetic/syn_03012_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "e9a21b2d20aaeefae66c356aa30a9c6e588cc37b8ff648442abfac81dba16da2", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "3a737121e23f6aa0ba2e6fca", "attempt_id": "391461e2e755", "run_group_id": "E4_core:A5|closed|qwen3b-llama3b:h8", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:47:30.651422Z", "task_family": "synthetic", "document_id": "syn_03012_2000", "document_hash": "03e441e68d54e1af036f8ff905dfd6cbbccead62636279c1a8e179352ab08945", "question_id": "q04_multi_hop", "split": "test", "cluster_id": "syn_03012_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "tokenizer_hashes": {"receiver": "b9d8872cf3a99da99e1fc4b0f71232a451632dc49ca4c6771549b4efa5c1c763"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 8, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1767, "query_tokens": 498, "message_tokens": 388, "message_utf8_bytes": 1097, "cache_tensor_bytes": null, "serialized_payload_bytes": 1097, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 138383.33570817485, "hops": {"1": 22652.15183305554, "2": 15272.267875028774, "3": 18122.784957988188, "4": 15171.370750060305, "5": 18196.480125072412, "6": 15206.121249939315, "7": 18390.757792047225, "8": 15371.401124983095}, "receiver_prefill_ms": 384.75724996533245, "ttft_ms": 58.865708066150546, "decode_ms": 89.69041705131531}, "total_wall_ms": 138917.26937517524, "ttft_ms": 58.865708066150546, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 59, "facts_retained": 13, "retention_rate": 0.22033898305084745, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.14285714285714285, "amount": 0.0, "status": 0.5833333333333334, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1097}, "ledger_by_hop": {"1": {"facts_total": 59, "facts_retained": 11, "retention_rate": 0.1864406779661017, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.14285714285714285, "amount": 0.0, "status": 0.4166666666666667, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 760}, "2": {"facts_total": 59, "facts_retained": 13, "retention_rate": 0.22033898305084745, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.14285714285714285, "amount": 0.0, "status": 0.5833333333333334, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1097}, "3": {"facts_total": 59, "facts_retained": 13, "retention_rate": 0.22033898305084745, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.14285714285714285, "amount": 0.0, "status": 0.5833333333333334, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1097}, "4": {"facts_total": 59, "facts_retained": 13, "retention_rate": 0.22033898305084745, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.14285714285714285, "amount": 0.0, "status": 0.5833333333333334, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1097}, "5": {"facts_total": 59, "facts_retained": 13, "retention_rate": 0.22033898305084745, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.14285714285714285, "amount": 0.0, "status": 0.5833333333333334, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1097}, "6": {"facts_total": 59, "facts_retained": 13, "retention_rate": 0.22033898305084745, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.14285714285714285, "amount": 0.0, "status": 0.5833333333333334, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1097}, "7": {"facts_total": 59, "facts_retained": 13, "retention_rate": 0.22033898305084745, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.14285714285714285, "amount": 0.0, "status": 0.5833333333333334, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1097}, "8": {"facts_total": 59, "facts_retained": 13, "retention_rate": 0.22033898305084745, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.14285714285714285, "amount": 0.0, "status": 0.5833333333333334, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1097}}}, "reference_key_id": "keys/synthetic/syn_03012_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "e9a21b2d20aaeefae66c356aa30a9c6e588cc37b8ff648442abfac81dba16da2", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "91babdca0e9b32461d88d6a6", "attempt_id": "5b8ddcdf0648", "run_group_id": "E4_core:A5|closed|qwen3b-llama3b:h8", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:47:31.205367Z", "task_family": "synthetic", "document_id": "syn_03012_2000", "document_hash": "03e441e68d54e1af036f8ff905dfd6cbbccead62636279c1a8e179352ab08945", "question_id": "q05_multi_hop", "split": "test", "cluster_id": "syn_03012_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "tokenizer_hashes": {"receiver": "b9d8872cf3a99da99e1fc4b0f71232a451632dc49ca4c6771549b4efa5c1c763"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 8, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1767, "query_tokens": 499, "message_tokens": 388, "message_utf8_bytes": 1097, "cache_tensor_bytes": null, "serialized_payload_bytes": 1097, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 138383.33570817485, "hops": {"1": 22652.15183305554, "2": 15272.267875028774, "3": 18122.784957988188, "4": 15171.370750060305, "5": 18196.480125072412, "6": 15206.121249939315, "7": 18390.757792047225, "8": 15371.401124983095}, "receiver_prefill_ms": 438.8945000246167, "ttft_ms": 25.908458046615124, "decode_ms": 36.28304100129753}, "total_wall_ms": 138884.91720810998, "ttft_ms": 25.908458046615124, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 59, "facts_retained": 13, "retention_rate": 0.22033898305084745, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.14285714285714285, "amount": 0.0, "status": 0.5833333333333334, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1097}, "ledger_by_hop": {"1": {"facts_total": 59, "facts_retained": 11, "retention_rate": 0.1864406779661017, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.14285714285714285, "amount": 0.0, "status": 0.4166666666666667, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 760}, "2": {"facts_total": 59, "facts_retained": 13, "retention_rate": 0.22033898305084745, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.14285714285714285, "amount": 0.0, "status": 0.5833333333333334, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1097}, "3": {"facts_total": 59, "facts_retained": 13, "retention_rate": 0.22033898305084745, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.14285714285714285, "amount": 0.0, "status": 0.5833333333333334, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1097}, "4": {"facts_total": 59, "facts_retained": 13, "retention_rate": 0.22033898305084745, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.14285714285714285, "amount": 0.0, "status": 0.5833333333333334, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1097}, "5": {"facts_total": 59, "facts_retained": 13, "retention_rate": 0.22033898305084745, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.14285714285714285, "amount": 0.0, "status": 0.5833333333333334, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1097}, "6": {"facts_total": 59, "facts_retained": 13, "retention_rate": 0.22033898305084745, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.14285714285714285, "amount": 0.0, "status": 0.5833333333333334, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1097}, "7": {"facts_total": 59, "facts_retained": 13, "retention_rate": 0.22033898305084745, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.14285714285714285, "amount": 0.0, "status": 0.5833333333333334, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1097}, "8": {"facts_total": 59, "facts_retained": 13, "retention_rate": 0.22033898305084745, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.14285714285714285, "amount": 0.0, "status": 0.5833333333333334, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1097}}}, "reference_key_id": "keys/synthetic/syn_03012_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "e9a21b2d20aaeefae66c356aa30a9c6e588cc37b8ff648442abfac81dba16da2", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "7f79fae75610e04380e6dc84", "attempt_id": "394e391ad686", "run_group_id": "E4_core:A5|closed|qwen3b-llama3b:h8", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:47:31.726209Z", "task_family": "synthetic", "document_id": "syn_03012_2000", "document_hash": "03e441e68d54e1af036f8ff905dfd6cbbccead62636279c1a8e179352ab08945", "question_id": "q06_negation", "split": "test", "cluster_id": "syn_03012_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "tokenizer_hashes": {"receiver": "b9d8872cf3a99da99e1fc4b0f71232a451632dc49ca4c6771549b4efa5c1c763"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 8, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1767, "query_tokens": 492, "message_tokens": 388, "message_utf8_bytes": 1097, "cache_tensor_bytes": null, "serialized_payload_bytes": 1097, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 138383.33570817485, "hops": {"1": 22652.15183305554, "2": 15272.267875028774, "3": 18122.784957988188, "4": 15171.370750060305, "5": 18196.480125072412, "6": 15206.121249939315, "7": 18390.757792047225, "8": 15371.401124983095}, "receiver_prefill_ms": 376.05587497819215, "ttft_ms": 32.30137506034225, "decode_ms": 28.59487500973046}, "total_wall_ms": 138820.83250023425, "ttft_ms": 32.30137506034225, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "No", "evidence_artifact": {"ledger": {"facts_total": 59, "facts_retained": 13, "retention_rate": 0.22033898305084745, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.14285714285714285, "amount": 0.0, "status": 0.5833333333333334, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1097}, "ledger_by_hop": {"1": {"facts_total": 59, "facts_retained": 11, "retention_rate": 0.1864406779661017, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.14285714285714285, "amount": 0.0, "status": 0.4166666666666667, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 760}, "2": {"facts_total": 59, "facts_retained": 13, "retention_rate": 0.22033898305084745, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.14285714285714285, "amount": 0.0, "status": 0.5833333333333334, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1097}, "3": {"facts_total": 59, "facts_retained": 13, "retention_rate": 0.22033898305084745, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.14285714285714285, "amount": 0.0, "status": 0.5833333333333334, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1097}, "4": {"facts_total": 59, "facts_retained": 13, "retention_rate": 0.22033898305084745, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.14285714285714285, "amount": 0.0, "status": 0.5833333333333334, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1097}, "5": {"facts_total": 59, "facts_retained": 13, "retention_rate": 0.22033898305084745, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.14285714285714285, "amount": 0.0, "status": 0.5833333333333334, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1097}, "6": {"facts_total": 59, "facts_retained": 13, "retention_rate": 0.22033898305084745, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.14285714285714285, "amount": 0.0, "status": 0.5833333333333334, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1097}, "7": {"facts_total": 59, "facts_retained": 13, "retention_rate": 0.22033898305084745, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.14285714285714285, "amount": 0.0, "status": 0.5833333333333334, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1097}, "8": {"facts_total": 59, "facts_retained": 13, "retention_rate": 0.22033898305084745, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.14285714285714285, "amount": 0.0, "status": 0.5833333333333334, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1097}}}, "reference_key_id": "keys/synthetic/syn_03012_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "e9a21b2d20aaeefae66c356aa30a9c6e588cc37b8ff648442abfac81dba16da2", "output_hash": "1ea442a134b2a184bd5d40104401f2a37fbc09ccf3f4bc9da161c6099be3691d"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "3d12e23f7bcec3adb9ff3596", "attempt_id": "2bad74dbc2fc", "run_group_id": "E4_core:A5|closed|qwen3b-llama3b:h8", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:47:32.181147Z", "task_family": "synthetic", "document_id": "syn_03012_2000", "document_hash": "03e441e68d54e1af036f8ff905dfd6cbbccead62636279c1a8e179352ab08945", "question_id": "q07_negation", "split": "test", "cluster_id": "syn_03012_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "tokenizer_hashes": {"receiver": "b9d8872cf3a99da99e1fc4b0f71232a451632dc49ca4c6771549b4efa5c1c763"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 8, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1767, "query_tokens": 493, "message_tokens": 388, "message_utf8_bytes": 1097, "cache_tensor_bytes": null, "serialized_payload_bytes": 1097, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 138383.33570817485, "hops": {"1": 22652.15183305554, "2": 15272.267875028774, "3": 18122.784957988188, "4": 15171.370750060305, "5": 18196.480125072412, "6": 15206.121249939315, "7": 18390.757792047225, "8": 15371.401124983095}, "receiver_prefill_ms": 322.6811670465395, "ttft_ms": 34.24866602290422, "decode_ms": 56.77299993112683}, "total_wall_ms": 138797.7325831307, "ttft_ms": 34.24866602290422, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "No", "evidence_artifact": {"ledger": {"facts_total": 59, "facts_retained": 13, "retention_rate": 0.22033898305084745, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.14285714285714285, "amount": 0.0, "status": 0.5833333333333334, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1097}, "ledger_by_hop": {"1": {"facts_total": 59, "facts_retained": 11, "retention_rate": 0.1864406779661017, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.14285714285714285, "amount": 0.0, "status": 0.4166666666666667, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 760}, "2": {"facts_total": 59, "facts_retained": 13, "retention_rate": 0.22033898305084745, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.14285714285714285, "amount": 0.0, "status": 0.5833333333333334, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1097}, "3": {"facts_total": 59, "facts_retained": 13, "retention_rate": 0.22033898305084745, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.14285714285714285, "amount": 0.0, "status": 0.5833333333333334, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1097}, "4": {"facts_total": 59, "facts_retained": 13, "retention_rate": 0.22033898305084745, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.14285714285714285, "amount": 0.0, "status": 0.5833333333333334, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1097}, "5": {"facts_total": 59, "facts_retained": 13, "retention_rate": 0.22033898305084745, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.14285714285714285, "amount": 0.0, "status": 0.5833333333333334, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1097}, "6": {"facts_total": 59, "facts_retained": 13, "retention_rate": 0.22033898305084745, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.14285714285714285, "amount": 0.0, "status": 0.5833333333333334, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1097}, "7": {"facts_total": 59, "facts_retained": 13, "retention_rate": 0.22033898305084745, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.14285714285714285, "amount": 0.0, "status": 0.5833333333333334, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1097}, "8": {"facts_total": 59, "facts_retained": 13, "retention_rate": 0.22033898305084745, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.14285714285714285, "amount": 0.0, "status": 0.5833333333333334, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1097}}}, "reference_key_id": "keys/synthetic/syn_03012_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "e9a21b2d20aaeefae66c356aa30a9c6e588cc37b8ff648442abfac81dba16da2", "output_hash": "1ea442a134b2a184bd5d40104401f2a37fbc09ccf3f4bc9da161c6099be3691d"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "d2e1afce391148969ec9e982", "attempt_id": "302c1e340396", "run_group_id": "E4_core:A5|closed|qwen3b-llama3b:h8", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:47:32.613791Z", "task_family": "synthetic", "document_id": "syn_03012_2000", "document_hash": "03e441e68d54e1af036f8ff905dfd6cbbccead62636279c1a8e179352ab08945", "question_id": "q08_revision", "split": "test", "cluster_id": "syn_03012_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "tokenizer_hashes": {"receiver": "b9d8872cf3a99da99e1fc4b0f71232a451632dc49ca4c6771549b4efa5c1c763"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 8, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1767, "query_tokens": 508, "message_tokens": 388, "message_utf8_bytes": 1097, "cache_tensor_bytes": null, "serialized_payload_bytes": 1097, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 138383.33570817485, "hops": {"1": 22652.15183305554, "2": 15272.267875028774, "3": 18122.784957988188, "4": 15171.370750060305, "5": 18196.480125072412, "6": 15206.121249939315, "7": 18390.757792047225, "8": 15371.401124983095}, "receiver_prefill_ms": 467.8040419239551, "ttft_ms": 21.893040975555778, "decode_ms": 37.876458023674786}, "total_wall_ms": 138911.52954113204, "ttft_ms": 21.893040975555778, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 59, "facts_retained": 13, "retention_rate": 0.22033898305084745, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.14285714285714285, "amount": 0.0, "status": 0.5833333333333334, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1097}, "ledger_by_hop": {"1": {"facts_total": 59, "facts_retained": 11, "retention_rate": 0.1864406779661017, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.14285714285714285, "amount": 0.0, "status": 0.4166666666666667, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 760}, "2": {"facts_total": 59, "facts_retained": 13, "retention_rate": 0.22033898305084745, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.14285714285714285, "amount": 0.0, "status": 0.5833333333333334, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1097}, "3": {"facts_total": 59, "facts_retained": 13, "retention_rate": 0.22033898305084745, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.14285714285714285, "amount": 0.0, "status": 0.5833333333333334, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1097}, "4": {"facts_total": 59, "facts_retained": 13, "retention_rate": 0.22033898305084745, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.14285714285714285, "amount": 0.0, "status": 0.5833333333333334, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1097}, "5": {"facts_total": 59, "facts_retained": 13, "retention_rate": 0.22033898305084745, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.14285714285714285, "amount": 0.0, "status": 0.5833333333333334, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1097}, "6": {"facts_total": 59, "facts_retained": 13, "retention_rate": 0.22033898305084745, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.14285714285714285, "amount": 0.0, "status": 0.5833333333333334, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1097}, "7": {"facts_total": 59, "facts_retained": 13, "retention_rate": 0.22033898305084745, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.14285714285714285, "amount": 0.0, "status": 0.5833333333333334, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1097}, "8": {"facts_total": 59, "facts_retained": 13, "retention_rate": 0.22033898305084745, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.14285714285714285, "amount": 0.0, "status": 0.5833333333333334, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1097}}}, "reference_key_id": "keys/synthetic/syn_03012_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "e9a21b2d20aaeefae66c356aa30a9c6e588cc37b8ff648442abfac81dba16da2", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "70d0029a917bd87ae08b1bf5", "attempt_id": "5e3b926f92a7", "run_group_id": "E4_core:A5|closed|qwen3b-llama3b:h8", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:47:33.159780Z", "task_family": "synthetic", "document_id": "syn_03012_2000", "document_hash": "03e441e68d54e1af036f8ff905dfd6cbbccead62636279c1a8e179352ab08945", "question_id": "q09_revision", "split": "test", "cluster_id": "syn_03012_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "tokenizer_hashes": {"receiver": "b9d8872cf3a99da99e1fc4b0f71232a451632dc49ca4c6771549b4efa5c1c763"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 8, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1767, "query_tokens": 500, "message_tokens": 388, "message_utf8_bytes": 1097, "cache_tensor_bytes": null, "serialized_payload_bytes": 1097, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 138383.33570817485, "hops": {"1": 22652.15183305554, "2": 15272.267875028774, "3": 18122.784957988188, "4": 15171.370750060305, "5": 18196.480125072412, "6": 15206.121249939315, "7": 18390.757792047225, "8": 15371.401124983095}, "receiver_prefill_ms": 385.55579097010195, "ttft_ms": 18.231875030323863, "decode_ms": 38.970416062511504}, "total_wall_ms": 138826.63379109, "ttft_ms": 18.231875030323863, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 59, "facts_retained": 13, "retention_rate": 0.22033898305084745, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.14285714285714285, "amount": 0.0, "status": 0.5833333333333334, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1097}, "ledger_by_hop": {"1": {"facts_total": 59, "facts_retained": 11, "retention_rate": 0.1864406779661017, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.14285714285714285, "amount": 0.0, "status": 0.4166666666666667, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 760}, "2": {"facts_total": 59, "facts_retained": 13, "retention_rate": 0.22033898305084745, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.14285714285714285, "amount": 0.0, "status": 0.5833333333333334, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1097}, "3": {"facts_total": 59, "facts_retained": 13, "retention_rate": 0.22033898305084745, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.14285714285714285, "amount": 0.0, "status": 0.5833333333333334, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1097}, "4": {"facts_total": 59, "facts_retained": 13, "retention_rate": 0.22033898305084745, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.14285714285714285, "amount": 0.0, "status": 0.5833333333333334, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1097}, "5": {"facts_total": 59, "facts_retained": 13, "retention_rate": 0.22033898305084745, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.14285714285714285, "amount": 0.0, "status": 0.5833333333333334, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1097}, "6": {"facts_total": 59, "facts_retained": 13, "retention_rate": 0.22033898305084745, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.14285714285714285, "amount": 0.0, "status": 0.5833333333333334, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1097}, "7": {"facts_total": 59, "facts_retained": 13, "retention_rate": 0.22033898305084745, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.14285714285714285, "amount": 0.0, "status": 0.5833333333333334, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1097}, "8": {"facts_total": 59, "facts_retained": 13, "retention_rate": 0.22033898305084745, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.14285714285714285, "amount": 0.0, "status": 0.5833333333333334, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1097}}}, "reference_key_id": "keys/synthetic/syn_03012_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "e9a21b2d20aaeefae66c356aa30a9c6e588cc37b8ff648442abfac81dba16da2", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "9a21d732b60a7c9212481980", "attempt_id": "6bd4c3174bfa", "run_group_id": "E4_core:A5|closed|qwen3b-llama3b:h8", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:47:33.621179Z", "task_family": "synthetic", "document_id": "syn_03012_2000", "document_hash": "03e441e68d54e1af036f8ff905dfd6cbbccead62636279c1a8e179352ab08945", "question_id": "q10_temporal", "split": "test", "cluster_id": "syn_03012_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "tokenizer_hashes": {"receiver": "b9d8872cf3a99da99e1fc4b0f71232a451632dc49ca4c6771549b4efa5c1c763"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 8, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1767, "query_tokens": 510, "message_tokens": 388, "message_utf8_bytes": 1097, "cache_tensor_bytes": null, "serialized_payload_bytes": 1097, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 138383.33570817485, "hops": {"1": 22652.15183305554, "2": 15272.267875028774, "3": 18122.784957988188, "4": 15171.370750060305, "5": 18196.480125072412, "6": 15206.121249939315, "7": 18390.757792047225, "8": 15371.401124983095}, "receiver_prefill_ms": 360.3400000138208, "ttft_ms": 14.291999978013337, "decode_ms": 11.50504092220217}, "total_wall_ms": 138769.95016622823, "ttft_ms": 14.291999978013337, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 59, "facts_retained": 13, "retention_rate": 0.22033898305084745, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.14285714285714285, "amount": 0.0, "status": 0.5833333333333334, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1097}, "ledger_by_hop": {"1": {"facts_total": 59, "facts_retained": 11, "retention_rate": 0.1864406779661017, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.14285714285714285, "amount": 0.0, "status": 0.4166666666666667, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 760}, "2": {"facts_total": 59, "facts_retained": 13, "retention_rate": 0.22033898305084745, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.14285714285714285, "amount": 0.0, "status": 0.5833333333333334, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1097}, "3": {"facts_total": 59, "facts_retained": 13, "retention_rate": 0.22033898305084745, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.14285714285714285, "amount": 0.0, "status": 0.5833333333333334, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1097}, "4": {"facts_total": 59, "facts_retained": 13, "retention_rate": 0.22033898305084745, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.14285714285714285, "amount": 0.0, "status": 0.5833333333333334, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1097}, "5": {"facts_total": 59, "facts_retained": 13, "retention_rate": 0.22033898305084745, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.14285714285714285, "amount": 0.0, "status": 0.5833333333333334, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1097}, "6": {"facts_total": 59, "facts_retained": 13, "retention_rate": 0.22033898305084745, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.14285714285714285, "amount": 0.0, "status": 0.5833333333333334, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1097}, "7": {"facts_total": 59, "facts_retained": 13, "retention_rate": 0.22033898305084745, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.14285714285714285, "amount": 0.0, "status": 0.5833333333333334, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1097}, "8": {"facts_total": 59, "facts_retained": 13, "retention_rate": 0.22033898305084745, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.14285714285714285, "amount": 0.0, "status": 0.5833333333333334, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1097}}}, "reference_key_id": "keys/synthetic/syn_03012_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "e9a21b2d20aaeefae66c356aa30a9c6e588cc37b8ff648442abfac81dba16da2", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "dc37652deb4367a58d95a2e3", "attempt_id": "2e1666bb5406", "run_group_id": "E4_core:A5|closed|qwen3b-llama3b:h8", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:47:34.024951Z", "task_family": "synthetic", "document_id": "syn_03012_2000", "document_hash": "03e441e68d54e1af036f8ff905dfd6cbbccead62636279c1a8e179352ab08945", "question_id": "q11_aggregation", "split": "test", "cluster_id": "syn_03012_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "tokenizer_hashes": {"receiver": "b9d8872cf3a99da99e1fc4b0f71232a451632dc49ca4c6771549b4efa5c1c763"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 8, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1767, "query_tokens": 503, "message_tokens": 388, "message_utf8_bytes": 1097, "cache_tensor_bytes": null, "serialized_payload_bytes": 1097, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 138383.33570817485, "hops": {"1": 22652.15183305554, "2": 15272.267875028774, "3": 18122.784957988188, "4": 15171.370750060305, "5": 18196.480125072412, "6": 15206.121249939315, "7": 18390.757792047225, "8": 15371.401124983095}, "receiver_prefill_ms": 370.9055830258876, "ttft_ms": 41.1918330937624, "decode_ms": 35.903124953620136}, "total_wall_ms": 138832.24370819516, "ttft_ms": 41.1918330937624, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 59, "facts_retained": 13, "retention_rate": 0.22033898305084745, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.14285714285714285, "amount": 0.0, "status": 0.5833333333333334, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1097}, "ledger_by_hop": {"1": {"facts_total": 59, "facts_retained": 11, "retention_rate": 0.1864406779661017, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.14285714285714285, "amount": 0.0, "status": 0.4166666666666667, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 760}, "2": {"facts_total": 59, "facts_retained": 13, "retention_rate": 0.22033898305084745, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.14285714285714285, "amount": 0.0, "status": 0.5833333333333334, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1097}, "3": {"facts_total": 59, "facts_retained": 13, "retention_rate": 0.22033898305084745, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.14285714285714285, "amount": 0.0, "status": 0.5833333333333334, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1097}, "4": {"facts_total": 59, "facts_retained": 13, "retention_rate": 0.22033898305084745, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.14285714285714285, "amount": 0.0, "status": 0.5833333333333334, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1097}, "5": {"facts_total": 59, "facts_retained": 13, "retention_rate": 0.22033898305084745, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.14285714285714285, "amount": 0.0, "status": 0.5833333333333334, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1097}, "6": {"facts_total": 59, "facts_retained": 13, "retention_rate": 0.22033898305084745, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.14285714285714285, "amount": 0.0, "status": 0.5833333333333334, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1097}, "7": {"facts_total": 59, "facts_retained": 13, "retention_rate": 0.22033898305084745, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.14285714285714285, "amount": 0.0, "status": 0.5833333333333334, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1097}, "8": {"facts_total": 59, "facts_retained": 13, "retention_rate": 0.22033898305084745, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.14285714285714285, "amount": 0.0, "status": 0.5833333333333334, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1097}}}, "reference_key_id": "keys/synthetic/syn_03012_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "e9a21b2d20aaeefae66c356aa30a9c6e588cc37b8ff648442abfac81dba16da2", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "e7214f59952867da7d6fba89", "attempt_id": "8a306095e9b1", "run_group_id": "E4_core:A5|closed|qwen3b-llama3b:h8", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:47:34.492851Z", "task_family": "synthetic", "document_id": "syn_03012_2000", "document_hash": "03e441e68d54e1af036f8ff905dfd6cbbccead62636279c1a8e179352ab08945", "question_id": "q12_unit_conversion", "split": "test", "cluster_id": "syn_03012_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "tokenizer_hashes": {"receiver": "b9d8872cf3a99da99e1fc4b0f71232a451632dc49ca4c6771549b4efa5c1c763"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 8, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1767, "query_tokens": 500, "message_tokens": 388, "message_utf8_bytes": 1097, "cache_tensor_bytes": null, "serialized_payload_bytes": 1097, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 138383.33570817485, "hops": {"1": 22652.15183305554, "2": 15272.267875028774, "3": 18122.784957988188, "4": 15171.370750060305, "5": 18196.480125072412, "6": 15206.121249939315, "7": 18390.757792047225, "8": 15371.401124983095}, "receiver_prefill_ms": 487.5949580455199, "ttft_ms": 35.28883296530694, "decode_ms": 83.86700006667525}, "total_wall_ms": 138990.632875124, "ttft_ms": 35.28883296530694, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 3}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "44,620", "evidence_artifact": {"ledger": {"facts_total": 59, "facts_retained": 13, "retention_rate": 0.22033898305084745, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.14285714285714285, "amount": 0.0, "status": 0.5833333333333334, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1097}, "ledger_by_hop": {"1": {"facts_total": 59, "facts_retained": 11, "retention_rate": 0.1864406779661017, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.14285714285714285, "amount": 0.0, "status": 0.4166666666666667, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 760}, "2": {"facts_total": 59, "facts_retained": 13, "retention_rate": 0.22033898305084745, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.14285714285714285, "amount": 0.0, "status": 0.5833333333333334, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1097}, "3": {"facts_total": 59, "facts_retained": 13, "retention_rate": 0.22033898305084745, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.14285714285714285, "amount": 0.0, "status": 0.5833333333333334, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1097}, "4": {"facts_total": 59, "facts_retained": 13, "retention_rate": 0.22033898305084745, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.14285714285714285, "amount": 0.0, "status": 0.5833333333333334, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1097}, "5": {"facts_total": 59, "facts_retained": 13, "retention_rate": 0.22033898305084745, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.14285714285714285, "amount": 0.0, "status": 0.5833333333333334, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1097}, "6": {"facts_total": 59, "facts_retained": 13, "retention_rate": 0.22033898305084745, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.14285714285714285, "amount": 0.0, "status": 0.5833333333333334, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1097}, "7": {"facts_total": 59, "facts_retained": 13, "retention_rate": 0.22033898305084745, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.14285714285714285, "amount": 0.0, "status": 0.5833333333333334, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1097}, "8": {"facts_total": 59, "facts_retained": 13, "retention_rate": 0.22033898305084745, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.14285714285714285, "amount": 0.0, "status": 0.5833333333333334, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1097}}}, "reference_key_id": "keys/synthetic/syn_03012_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "e9a21b2d20aaeefae66c356aa30a9c6e588cc37b8ff648442abfac81dba16da2", "output_hash": "09a7740f04b91a7ab1ac87709e5c0c6d334d826341de977c51a3b65515114023"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "1fc1a3405b5210d8f8d387b1", "attempt_id": "b1659ba7c20a", "run_group_id": "E4_core:A5|closed|qwen3b-llama3b:h8", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:47:35.118127Z", "task_family": "synthetic", "document_id": "syn_03012_2000", "document_hash": "03e441e68d54e1af036f8ff905dfd6cbbccead62636279c1a8e179352ab08945", "question_id": "q13_unanswerable", "split": "test", "cluster_id": "syn_03012_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "tokenizer_hashes": {"receiver": "b9d8872cf3a99da99e1fc4b0f71232a451632dc49ca4c6771549b4efa5c1c763"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 8, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1767, "query_tokens": 504, "message_tokens": 388, "message_utf8_bytes": 1097, "cache_tensor_bytes": null, "serialized_payload_bytes": 1097, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 138383.33570817485, "hops": {"1": 22652.15183305554, "2": 15272.267875028774, "3": 18122.784957988188, "4": 15171.370750060305, "5": 18196.480125072412, "6": 15206.121249939315, "7": 18390.757792047225, "8": 15371.401124983095}, "receiver_prefill_ms": 361.79529200308025, "ttft_ms": 26.162790949456394, "decode_ms": 36.835917038843036}, "total_wall_ms": 138808.67383314762, "ttft_ms": 26.162790949456394, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 59, "facts_retained": 13, "retention_rate": 0.22033898305084745, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.14285714285714285, "amount": 0.0, "status": 0.5833333333333334, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1097}, "ledger_by_hop": {"1": {"facts_total": 59, "facts_retained": 11, "retention_rate": 0.1864406779661017, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.14285714285714285, "amount": 0.0, "status": 0.4166666666666667, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 760}, "2": {"facts_total": 59, "facts_retained": 13, "retention_rate": 0.22033898305084745, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.14285714285714285, "amount": 0.0, "status": 0.5833333333333334, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1097}, "3": {"facts_total": 59, "facts_retained": 13, "retention_rate": 0.22033898305084745, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.14285714285714285, "amount": 0.0, "status": 0.5833333333333334, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1097}, "4": {"facts_total": 59, "facts_retained": 13, "retention_rate": 0.22033898305084745, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.14285714285714285, "amount": 0.0, "status": 0.5833333333333334, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1097}, "5": {"facts_total": 59, "facts_retained": 13, "retention_rate": 0.22033898305084745, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.14285714285714285, "amount": 0.0, "status": 0.5833333333333334, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1097}, "6": {"facts_total": 59, "facts_retained": 13, "retention_rate": 0.22033898305084745, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.14285714285714285, "amount": 0.0, "status": 0.5833333333333334, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1097}, "7": {"facts_total": 59, "facts_retained": 13, "retention_rate": 0.22033898305084745, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.14285714285714285, "amount": 0.0, "status": 0.5833333333333334, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1097}, "8": {"facts_total": 59, "facts_retained": 13, "retention_rate": 0.22033898305084745, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.14285714285714285, "amount": 0.0, "status": 0.5833333333333334, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1097}}}, "reference_key_id": "keys/synthetic/syn_03012_2000.json", "task_scores": {"correct": 1.0, "abstained": 1.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "e9a21b2d20aaeefae66c356aa30a9c6e588cc37b8ff648442abfac81dba16da2", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "f79e5f459ed8f7fa836b8eb7", "attempt_id": "758b7e6b704b", "run_group_id": "E4_core:A5|closed|qwen3b-llama3b:h8", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:47:35.560110Z", "task_family": "synthetic", "document_id": "syn_03012_2000", "document_hash": "03e441e68d54e1af036f8ff905dfd6cbbccead62636279c1a8e179352ab08945", "question_id": "q14_unanswerable", "split": "test", "cluster_id": "syn_03012_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "tokenizer_hashes": {"receiver": "b9d8872cf3a99da99e1fc4b0f71232a451632dc49ca4c6771549b4efa5c1c763"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 8, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1767, "query_tokens": 505, "message_tokens": 388, "message_utf8_bytes": 1097, "cache_tensor_bytes": null, "serialized_payload_bytes": 1097, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 138383.33570817485, "hops": {"1": 22652.15183305554, "2": 15272.267875028774, "3": 18122.784957988188, "4": 15171.370750060305, "5": 18196.480125072412, "6": 15206.121249939315, "7": 18390.757792047225, "8": 15371.401124983095}, "receiver_prefill_ms": 335.0220420397818, "ttft_ms": 64.80266700964421, "decode_ms": 50.324833020567894}, "total_wall_ms": 138834.1468331637, "ttft_ms": 64.80266700964421, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 59, "facts_retained": 13, "retention_rate": 0.22033898305084745, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.14285714285714285, "amount": 0.0, "status": 0.5833333333333334, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1097}, "ledger_by_hop": {"1": {"facts_total": 59, "facts_retained": 11, "retention_rate": 0.1864406779661017, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.14285714285714285, "amount": 0.0, "status": 0.4166666666666667, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 760}, "2": {"facts_total": 59, "facts_retained": 13, "retention_rate": 0.22033898305084745, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.14285714285714285, "amount": 0.0, "status": 0.5833333333333334, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1097}, "3": {"facts_total": 59, "facts_retained": 13, "retention_rate": 0.22033898305084745, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.14285714285714285, "amount": 0.0, "status": 0.5833333333333334, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1097}, "4": {"facts_total": 59, "facts_retained": 13, "retention_rate": 0.22033898305084745, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.14285714285714285, "amount": 0.0, "status": 0.5833333333333334, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1097}, "5": {"facts_total": 59, "facts_retained": 13, "retention_rate": 0.22033898305084745, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.14285714285714285, "amount": 0.0, "status": 0.5833333333333334, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1097}, "6": {"facts_total": 59, "facts_retained": 13, "retention_rate": 0.22033898305084745, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.14285714285714285, "amount": 0.0, "status": 0.5833333333333334, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1097}, "7": {"facts_total": 59, "facts_retained": 13, "retention_rate": 0.22033898305084745, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.14285714285714285, "amount": 0.0, "status": 0.5833333333333334, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1097}, "8": {"facts_total": 59, "facts_retained": 13, "retention_rate": 0.22033898305084745, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.14285714285714285, "amount": 0.0, "status": 0.5833333333333334, "place": 0.16666666666666666, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1097}}}, "reference_key_id": "keys/synthetic/syn_03012_2000.json", "task_scores": {"correct": 1.0, "abstained": 1.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "e9a21b2d20aaeefae66c356aa30a9c6e588cc37b8ff648442abfac81dba16da2", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "add5fd5461679cb9fa3794be", "attempt_id": "c5a5f6af340c", "run_group_id": "E4_core:A4|closed|qwen3b:h1", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:47:58.829931Z", "task_family": "synthetic", "document_id": "syn_03015_2000", "document_hash": "71a7284b9bc3f681d69d91d18b9e8d45a4dda486cd14f4145395786ac6a68a89", "question_id": "q00_single_fact", "split": "test", "cluster_id": "syn_03015_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 1, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1922, "query_tokens": 595, "message_tokens": 512, "message_utf8_bytes": 1161, "cache_tensor_bytes": null, "serialized_payload_bytes": 1161, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 22697.06304196734, "hops": {"1": 22697.06304196734}, "receiver_prefill_ms": 410.4682080214843, "ttft_ms": 25.749500025995076, "decode_ms": 41.87212500255555}, "total_wall_ms": 23175.920666893944, "ttft_ms": 25.749500025995076, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 61, "facts_retained": 24, "retention_rate": 0.39344262295081966, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.4166666666666667, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1161}, "ledger_by_hop": {"1": {"facts_total": 61, "facts_retained": 24, "retention_rate": 0.39344262295081966, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.4166666666666667, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1161}}}, "reference_key_id": "keys/synthetic/syn_03015_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "52d2d6723cbb6f8139ce72290860dae098b28331488ef44459da0ce2ae9f9f72", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "f431b0be3e3430a3fd238896", "attempt_id": "8a15b7ba6f3a", "run_group_id": "E4_core:A4|closed|qwen3b:h1", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:47:59.327368Z", "task_family": "synthetic", "document_id": "syn_03015_2000", "document_hash": "71a7284b9bc3f681d69d91d18b9e8d45a4dda486cd14f4145395786ac6a68a89", "question_id": "q01_single_fact", "split": "test", "cluster_id": "syn_03015_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 1, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1922, "query_tokens": 600, "message_tokens": 512, "message_utf8_bytes": 1161, "cache_tensor_bytes": null, "serialized_payload_bytes": 1161, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 22697.06304196734, "hops": {"1": 22697.06304196734}, "receiver_prefill_ms": 385.1512080291286, "ttft_ms": 42.8841250250116, "decode_ms": 61.17420899681747}, "total_wall_ms": 23187.011208967306, "ttft_ms": 42.8841250250116, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 61, "facts_retained": 24, "retention_rate": 0.39344262295081966, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.4166666666666667, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1161}, "ledger_by_hop": {"1": {"facts_total": 61, "facts_retained": 24, "retention_rate": 0.39344262295081966, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.4166666666666667, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1161}}}, "reference_key_id": "keys/synthetic/syn_03015_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "52d2d6723cbb6f8139ce72290860dae098b28331488ef44459da0ce2ae9f9f72", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "4ddfb007b30f8f471bda7a6f", "attempt_id": "2b756cdff63b", "run_group_id": "E4_core:A4|closed|qwen3b:h1", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:47:59.834491Z", "task_family": "synthetic", "document_id": "syn_03015_2000", "document_hash": "71a7284b9bc3f681d69d91d18b9e8d45a4dda486cd14f4145395786ac6a68a89", "question_id": "q02_exact_copy", "split": "test", "cluster_id": "syn_03015_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 1, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1922, "query_tokens": 599, "message_tokens": 512, "message_utf8_bytes": 1161, "cache_tensor_bytes": null, "serialized_payload_bytes": 1161, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 22697.06304196734, "hops": {"1": 22697.06304196734}, "receiver_prefill_ms": 526.0762079851702, "ttft_ms": 26.793332886882126, "decode_ms": 236.74079193733633}, "total_wall_ms": 23487.390458933078, "ttft_ms": 26.793332886882126, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 7}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "89.3122", "evidence_artifact": {"ledger": {"facts_total": 61, "facts_retained": 24, "retention_rate": 0.39344262295081966, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.4166666666666667, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1161}, "ledger_by_hop": {"1": {"facts_total": 61, "facts_retained": 24, "retention_rate": 0.39344262295081966, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.4166666666666667, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1161}}}, "reference_key_id": "keys/synthetic/syn_03015_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "52d2d6723cbb6f8139ce72290860dae098b28331488ef44459da0ce2ae9f9f72", "output_hash": "2138888e6519b2f4b1b54142e6f9c541102d7f9ee5bcf8e4184c81a382601757"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "d993538feef811a8a6ab59b6", "attempt_id": "287723444f98", "run_group_id": "E4_core:A4|closed|qwen3b:h1", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:48:00.643078Z", "task_family": "synthetic", "document_id": "syn_03015_2000", "document_hash": "71a7284b9bc3f681d69d91d18b9e8d45a4dda486cd14f4145395786ac6a68a89", "question_id": "q03_exact_copy", "split": "test", "cluster_id": "syn_03015_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 1, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1922, "query_tokens": 595, "message_tokens": 512, "message_utf8_bytes": 1161, "cache_tensor_bytes": null, "serialized_payload_bytes": 1161, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 22697.06304196734, "hops": {"1": 22697.06304196734}, "receiver_prefill_ms": 397.8761249454692, "ttft_ms": 24.501875042915344, "decode_ms": 279.1237080236897}, "total_wall_ms": 23399.276876007207, "ttft_ms": 24.501875042915344, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 6}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "Wilhelmina Dubrovnik", "evidence_artifact": {"ledger": {"facts_total": 61, "facts_retained": 24, "retention_rate": 0.39344262295081966, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.4166666666666667, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1161}, "ledger_by_hop": {"1": {"facts_total": 61, "facts_retained": 24, "retention_rate": 0.39344262295081966, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.4166666666666667, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1161}}}, "reference_key_id": "keys/synthetic/syn_03015_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "52d2d6723cbb6f8139ce72290860dae098b28331488ef44459da0ce2ae9f9f72", "output_hash": "f42f05ebfc96f7258cbfbbb37cd65e114d3d15f869e68e6c95b71fc288c30a32"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "01d8efddb2097d7d395cc716", "attempt_id": "74f95aeb91ac", "run_group_id": "E4_core:A4|closed|qwen3b:h1", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:48:01.362912Z", "task_family": "synthetic", "document_id": "syn_03015_2000", "document_hash": "71a7284b9bc3f681d69d91d18b9e8d45a4dda486cd14f4145395786ac6a68a89", "question_id": "q04_multi_hop", "split": "test", "cluster_id": "syn_03015_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 1, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1922, "query_tokens": 604, "message_tokens": 512, "message_utf8_bytes": 1161, "cache_tensor_bytes": null, "serialized_payload_bytes": 1161, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 22697.06304196734, "hops": {"1": 22697.06304196734}, "receiver_prefill_ms": 498.44479199964553, "ttft_ms": 36.06333304196596, "decode_ms": 216.86783293262124}, "total_wall_ms": 23449.22470895108, "ttft_ms": 36.06333304196596, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 7}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "8421725", "evidence_artifact": {"ledger": {"facts_total": 61, "facts_retained": 24, "retention_rate": 0.39344262295081966, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.4166666666666667, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1161}, "ledger_by_hop": {"1": {"facts_total": 61, "facts_retained": 24, "retention_rate": 0.39344262295081966, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.4166666666666667, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1161}}}, "reference_key_id": "keys/synthetic/syn_03015_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "52d2d6723cbb6f8139ce72290860dae098b28331488ef44459da0ce2ae9f9f72", "output_hash": "e4949aba3b0b77de05a87edaab854329fa75df7bf68b29a3def71e985a06e5b5"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "e6de0d4281521e75a9105cf0", "attempt_id": "d0ef0038ea6b", "run_group_id": "E4_core:A4|closed|qwen3b:h1", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:48:02.133178Z", "task_family": "synthetic", "document_id": "syn_03015_2000", "document_hash": "71a7284b9bc3f681d69d91d18b9e8d45a4dda486cd14f4145395786ac6a68a89", "question_id": "q05_multi_hop", "split": "test", "cluster_id": "syn_03015_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 1, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1922, "query_tokens": 604, "message_tokens": 512, "message_utf8_bytes": 1161, "cache_tensor_bytes": null, "serialized_payload_bytes": 1161, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 22697.06304196734, "hops": {"1": 22697.06304196734}, "receiver_prefill_ms": 391.42441703006625, "ttft_ms": 20.855666953139007, "decode_ms": 38.376082899048924}, "total_wall_ms": 23148.495124885812, "ttft_ms": 20.855666953139007, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 61, "facts_retained": 24, "retention_rate": 0.39344262295081966, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.4166666666666667, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1161}, "ledger_by_hop": {"1": {"facts_total": 61, "facts_retained": 24, "retention_rate": 0.39344262295081966, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.4166666666666667, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1161}}}, "reference_key_id": "keys/synthetic/syn_03015_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "52d2d6723cbb6f8139ce72290860dae098b28331488ef44459da0ce2ae9f9f72", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "f20c545f2ce64ffee9883711", "attempt_id": "8b3e7428d208", "run_group_id": "E4_core:A4|closed|qwen3b:h1", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:48:02.602515Z", "task_family": "synthetic", "document_id": "syn_03015_2000", "document_hash": "71a7284b9bc3f681d69d91d18b9e8d45a4dda486cd14f4145395786ac6a68a89", "question_id": "q06_negation", "split": "test", "cluster_id": "syn_03015_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 1, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1922, "query_tokens": 596, "message_tokens": 512, "message_utf8_bytes": 1161, "cache_tensor_bytes": null, "serialized_payload_bytes": 1161, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 22697.06304196734, "hops": {"1": 22697.06304196734}, "receiver_prefill_ms": 360.7033749576658, "ttft_ms": 52.91025002952665, "decode_ms": 44.246917008422315}, "total_wall_ms": 23155.720541952178, "ttft_ms": 52.91025002952665, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 61, "facts_retained": 24, "retention_rate": 0.39344262295081966, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.4166666666666667, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1161}, "ledger_by_hop": {"1": {"facts_total": 61, "facts_retained": 24, "retention_rate": 0.39344262295081966, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.4166666666666667, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1161}}}, "reference_key_id": "keys/synthetic/syn_03015_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "52d2d6723cbb6f8139ce72290860dae098b28331488ef44459da0ce2ae9f9f72", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "763591ec058e134021a651b1", "attempt_id": "4fba16f434db", "run_group_id": "E4_core:A4|closed|qwen3b:h1", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:48:03.078619Z", "task_family": "synthetic", "document_id": "syn_03015_2000", "document_hash": "71a7284b9bc3f681d69d91d18b9e8d45a4dda486cd14f4145395786ac6a68a89", "question_id": "q07_negation", "split": "test", "cluster_id": "syn_03015_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 1, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1922, "query_tokens": 596, "message_tokens": 512, "message_utf8_bytes": 1161, "cache_tensor_bytes": null, "serialized_payload_bytes": 1161, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 22697.06304196734, "hops": {"1": 22697.06304196734}, "receiver_prefill_ms": 503.67841601837426, "ttft_ms": 35.07379093207419, "decode_ms": 37.0566250057891}, "total_wall_ms": 23273.47487595398, "ttft_ms": 35.07379093207419, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 61, "facts_retained": 24, "retention_rate": 0.39344262295081966, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.4166666666666667, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1161}, "ledger_by_hop": {"1": {"facts_total": 61, "facts_retained": 24, "retention_rate": 0.39344262295081966, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.4166666666666667, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1161}}}, "reference_key_id": "keys/synthetic/syn_03015_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "52d2d6723cbb6f8139ce72290860dae098b28331488ef44459da0ce2ae9f9f72", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "1c34b4b29c4b7d3d2fc6c8d4", "attempt_id": "33581adb4e67", "run_group_id": "E4_core:A4|closed|qwen3b:h1", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:48:03.671459Z", "task_family": "synthetic", "document_id": "syn_03015_2000", "document_hash": "71a7284b9bc3f681d69d91d18b9e8d45a4dda486cd14f4145395786ac6a68a89", "question_id": "q08_revision", "split": "test", "cluster_id": "syn_03015_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 1, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1922, "query_tokens": 610, "message_tokens": 512, "message_utf8_bytes": 1161, "cache_tensor_bytes": null, "serialized_payload_bytes": 1161, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 22697.06304196734, "hops": {"1": 22697.06304196734}, "receiver_prefill_ms": 417.146582971327, "ttft_ms": 37.89295896422118, "decode_ms": 25.334249949082732}, "total_wall_ms": 23178.257626015693, "ttft_ms": 37.89295896422118, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 61, "facts_retained": 24, "retention_rate": 0.39344262295081966, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.4166666666666667, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1161}, "ledger_by_hop": {"1": {"facts_total": 61, "facts_retained": 24, "retention_rate": 0.39344262295081966, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.4166666666666667, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1161}}}, "reference_key_id": "keys/synthetic/syn_03015_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "52d2d6723cbb6f8139ce72290860dae098b28331488ef44459da0ce2ae9f9f72", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "9c6241003b21b3c5b85fbec7", "attempt_id": "03dec41f478b", "run_group_id": "E4_core:A4|closed|qwen3b:h1", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:48:04.169695Z", "task_family": "synthetic", "document_id": "syn_03015_2000", "document_hash": "71a7284b9bc3f681d69d91d18b9e8d45a4dda486cd14f4145395786ac6a68a89", "question_id": "q09_revision", "split": "test", "cluster_id": "syn_03015_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 1, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1922, "query_tokens": 604, "message_tokens": 512, "message_utf8_bytes": 1161, "cache_tensor_bytes": null, "serialized_payload_bytes": 1161, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 22697.06304196734, "hops": {"1": 22697.06304196734}, "receiver_prefill_ms": 378.94891609903425, "ttft_ms": 39.55991705879569, "decode_ms": 585.6318330625072}, "total_wall_ms": 23701.954582938924, "ttft_ms": 39.55991705879569, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 9}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "8,377,480", "evidence_artifact": {"ledger": {"facts_total": 61, "facts_retained": 24, "retention_rate": 0.39344262295081966, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.4166666666666667, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1161}, "ledger_by_hop": {"1": {"facts_total": 61, "facts_retained": 24, "retention_rate": 0.39344262295081966, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.4166666666666667, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1161}}}, "reference_key_id": "keys/synthetic/syn_03015_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "52d2d6723cbb6f8139ce72290860dae098b28331488ef44459da0ce2ae9f9f72", "output_hash": "ec4c68b327a2d2427ed21821415a5f9ddf87134bd9f654bdd69c7d1301febcb0"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "c50a0cd2ef9c3eb00593ec58", "attempt_id": "6904de06c72a", "run_group_id": "E4_core:A4|closed|qwen3b:h1", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:48:05.192864Z", "task_family": "synthetic", "document_id": "syn_03015_2000", "document_hash": "71a7284b9bc3f681d69d91d18b9e8d45a4dda486cd14f4145395786ac6a68a89", "question_id": "q10_temporal", "split": "test", "cluster_id": "syn_03015_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 1, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1922, "query_tokens": 613, "message_tokens": 512, "message_utf8_bytes": 1161, "cache_tensor_bytes": null, "serialized_payload_bytes": 1161, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 22697.06304196734, "hops": {"1": 22697.06304196734}, "receiver_prefill_ms": 413.60854206141084, "ttft_ms": 27.193124988116324, "decode_ms": 36.15187504328787}, "total_wall_ms": 23174.6856670361, "ttft_ms": 27.193124988116324, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 61, "facts_retained": 24, "retention_rate": 0.39344262295081966, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.4166666666666667, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1161}, "ledger_by_hop": {"1": {"facts_total": 61, "facts_retained": 24, "retention_rate": 0.39344262295081966, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.4166666666666667, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1161}}}, "reference_key_id": "keys/synthetic/syn_03015_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "52d2d6723cbb6f8139ce72290860dae098b28331488ef44459da0ce2ae9f9f72", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "b4517db9eeafe357634d84d5", "attempt_id": "58f08feb4f00", "run_group_id": "E4_core:A4|closed|qwen3b:h1", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:48:05.688626Z", "task_family": "synthetic", "document_id": "syn_03015_2000", "document_hash": "71a7284b9bc3f681d69d91d18b9e8d45a4dda486cd14f4145395786ac6a68a89", "question_id": "q11_aggregation", "split": "test", "cluster_id": "syn_03015_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 1, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1922, "query_tokens": 606, "message_tokens": 512, "message_utf8_bytes": 1161, "cache_tensor_bytes": null, "serialized_payload_bytes": 1161, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 22697.06304196734, "hops": {"1": 22697.06304196734}, "receiver_prefill_ms": 408.74120802618563, "ttft_ms": 23.871333920396864, "decode_ms": 323.003416066058}, "total_wall_ms": 23453.336166916415, "ttft_ms": 23.871333920396864, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 7}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "1806025", "evidence_artifact": {"ledger": {"facts_total": 61, "facts_retained": 24, "retention_rate": 0.39344262295081966, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.4166666666666667, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1161}, "ledger_by_hop": {"1": {"facts_total": 61, "facts_retained": 24, "retention_rate": 0.39344262295081966, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.4166666666666667, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1161}}}, "reference_key_id": "keys/synthetic/syn_03015_2000.json", "task_scores": {"correct": 0.0, "abstained": 0.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "52d2d6723cbb6f8139ce72290860dae098b28331488ef44459da0ce2ae9f9f72", "output_hash": "9801d1840792d286693349934c9cc13fead44d82d480a42803cf6015c01431f6"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "b94f02aa56452e74ab37b8e1", "attempt_id": "fe72acc6c32e", "run_group_id": "E4_core:A4|closed|qwen3b:h1", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:48:06.461890Z", "task_family": "synthetic", "document_id": "syn_03015_2000", "document_hash": "71a7284b9bc3f681d69d91d18b9e8d45a4dda486cd14f4145395786ac6a68a89", "question_id": "q12_unit_conversion", "split": "test", "cluster_id": "syn_03015_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 1, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1922, "query_tokens": 603, "message_tokens": 512, "message_utf8_bytes": 1161, "cache_tensor_bytes": null, "serialized_payload_bytes": 1161, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 22697.06304196734, "hops": {"1": 22697.06304196734}, "receiver_prefill_ms": 508.2758329808712, "ttft_ms": 34.23491702415049, "decode_ms": 28.81895797327161}, "total_wall_ms": 23269.0961669432, "ttft_ms": 34.23491702415049, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 61, "facts_retained": 24, "retention_rate": 0.39344262295081966, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.4166666666666667, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1161}, "ledger_by_hop": {"1": {"facts_total": 61, "facts_retained": 24, "retention_rate": 0.39344262295081966, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.4166666666666667, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1161}}}, "reference_key_id": "keys/synthetic/syn_03015_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "52d2d6723cbb6f8139ce72290860dae098b28331488ef44459da0ce2ae9f9f72", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "40680d1171e067a6e05fd492", "attempt_id": "e2e562f85e85", "run_group_id": "E4_core:A4|closed|qwen3b:h1", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:48:07.051682Z", "task_family": "synthetic", "document_id": "syn_03015_2000", "document_hash": "71a7284b9bc3f681d69d91d18b9e8d45a4dda486cd14f4145395786ac6a68a89", "question_id": "q13_unanswerable", "split": "test", "cluster_id": "syn_03015_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 1, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1922, "query_tokens": 607, "message_tokens": 512, "message_utf8_bytes": 1161, "cache_tensor_bytes": null, "serialized_payload_bytes": 1161, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 22697.06304196734, "hops": {"1": 22697.06304196734}, "receiver_prefill_ms": 402.2636250592768, "ttft_ms": 23.33412494044751, "decode_ms": 40.96504102926701}, "total_wall_ms": 23164.387791999616, "ttft_ms": 23.33412494044751, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 61, "facts_retained": 24, "retention_rate": 0.39344262295081966, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.4166666666666667, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1161}, "ledger_by_hop": {"1": {"facts_total": 61, "facts_retained": 24, "retention_rate": 0.39344262295081966, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.4166666666666667, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1161}}}, "reference_key_id": "keys/synthetic/syn_03015_2000.json", "task_scores": {"correct": 1.0, "abstained": 1.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "52d2d6723cbb6f8139ce72290860dae098b28331488ef44459da0ce2ae9f9f72", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "940aeed5d431b0548c9e5346", "attempt_id": "34716b282ed1", "run_group_id": "E4_core:A4|closed|qwen3b:h1", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:48:07.536766Z", "task_family": "synthetic", "document_id": "syn_03015_2000", "document_hash": "71a7284b9bc3f681d69d91d18b9e8d45a4dda486cd14f4145395786ac6a68a89", "question_id": "q14_unanswerable", "split": "test", "cluster_id": "syn_03015_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 1, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1922, "query_tokens": 608, "message_tokens": 512, "message_utf8_bytes": 1161, "cache_tensor_bytes": null, "serialized_payload_bytes": 1161, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 22697.06304196734, "hops": {"1": 22697.06304196734}, "receiver_prefill_ms": 363.0760000087321, "ttft_ms": 51.09887500293553, "decode_ms": 64.9380829418078}, "total_wall_ms": 23176.866207970306, "ttft_ms": 51.09887500293553, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 61, "facts_retained": 24, "retention_rate": 0.39344262295081966, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.4166666666666667, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1161}, "ledger_by_hop": {"1": {"facts_total": 61, "facts_retained": 24, "retention_rate": 0.39344262295081966, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.4166666666666667, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1161}}}, "reference_key_id": "keys/synthetic/syn_03015_2000.json", "task_scores": {"correct": 1.0, "abstained": 1.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "52d2d6723cbb6f8139ce72290860dae098b28331488ef44459da0ce2ae9f9f72", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "36043a82fc024422c9065063", "attempt_id": "f17f62b45572", "run_group_id": "E4_core:A4|closed|qwen3b:h2", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:48:30.135747Z", "task_family": "synthetic", "document_id": "syn_03015_2000", "document_hash": "71a7284b9bc3f681d69d91d18b9e8d45a4dda486cd14f4145395786ac6a68a89", "question_id": "q00_single_fact", "split": "test", "cluster_id": "syn_03015_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 2, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1922, "query_tokens": 585, "message_tokens": 503, "message_utf8_bytes": 1142, "cache_tensor_bytes": null, "serialized_payload_bytes": 1142, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 44797.16570896562, "hops": {"1": 22697.06304196734, "2": 22100.102666998282}, "receiver_prefill_ms": 457.97650003805757, "ttft_ms": 56.15208391100168, "decode_ms": 56.137458072043955}, "total_wall_ms": 45368.238833965734, "ttft_ms": 56.15208391100168, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 61, "facts_retained": 24, "retention_rate": 0.39344262295081966, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.4166666666666667, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1142}, "ledger_by_hop": {"1": {"facts_total": 61, "facts_retained": 24, "retention_rate": 0.39344262295081966, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.4166666666666667, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1161}, "2": {"facts_total": 61, "facts_retained": 24, "retention_rate": 0.39344262295081966, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.4166666666666667, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1142}}}, "reference_key_id": "keys/synthetic/syn_03015_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "6606d2a2280a507a57312c3b4cdff39a97c80f61816d5bc41d13980cbef297bd", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "5fd596dbd18a6157d46357ed", "attempt_id": "5ae1c58c5439", "run_group_id": "E4_core:A4|closed|qwen3b:h2", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:48:30.724365Z", "task_family": "synthetic", "document_id": "syn_03015_2000", "document_hash": "71a7284b9bc3f681d69d91d18b9e8d45a4dda486cd14f4145395786ac6a68a89", "question_id": "q01_single_fact", "split": "test", "cluster_id": "syn_03015_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 2, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1922, "query_tokens": 590, "message_tokens": 503, "message_utf8_bytes": 1142, "cache_tensor_bytes": null, "serialized_payload_bytes": 1142, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 44797.16570896562, "hops": {"1": 22697.06304196734, "2": 22100.102666998282}, "receiver_prefill_ms": 460.3060829686001, "ttft_ms": 87.00199995655566, "decode_ms": 67.89433397352695}, "total_wall_ms": 45413.13179291319, "ttft_ms": 87.00199995655566, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 61, "facts_retained": 24, "retention_rate": 0.39344262295081966, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.4166666666666667, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1142}, "ledger_by_hop": {"1": {"facts_total": 61, "facts_retained": 24, "retention_rate": 0.39344262295081966, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.4166666666666667, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1161}, "2": {"facts_total": 61, "facts_retained": 24, "retention_rate": 0.39344262295081966, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.4166666666666667, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1142}}}, "reference_key_id": "keys/synthetic/syn_03015_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "6606d2a2280a507a57312c3b4cdff39a97c80f61816d5bc41d13980cbef297bd", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "53bd48f040e025f2b6fcaa7a", "attempt_id": "49009a52ad7e", "run_group_id": "E4_core:A4|closed|qwen3b:h2", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:48:31.357334Z", "task_family": "synthetic", "document_id": "syn_03015_2000", "document_hash": "71a7284b9bc3f681d69d91d18b9e8d45a4dda486cd14f4145395786ac6a68a89", "question_id": "q02_exact_copy", "split": "test", "cluster_id": "syn_03015_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 2, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1922, "query_tokens": 589, "message_tokens": 503, "message_utf8_bytes": 1142, "cache_tensor_bytes": null, "serialized_payload_bytes": 1142, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 44797.16570896562, "hops": {"1": 22697.06304196734, "2": 22100.102666998282}, "receiver_prefill_ms": 482.0587910944596, "ttft_ms": 56.89149990212172, "decode_ms": 501.2815840309486}, "total_wall_ms": 45838.02487596404, "ttft_ms": 56.89149990212172, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 7}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "89.3122", "evidence_artifact": {"ledger": {"facts_total": 61, "facts_retained": 24, "retention_rate": 0.39344262295081966, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.4166666666666667, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1142}, "ledger_by_hop": {"1": {"facts_total": 61, "facts_retained": 24, "retention_rate": 0.39344262295081966, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.4166666666666667, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1161}, "2": {"facts_total": 61, "facts_retained": 24, "retention_rate": 0.39344262295081966, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.4166666666666667, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1142}}}, "reference_key_id": "keys/synthetic/syn_03015_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "6606d2a2280a507a57312c3b4cdff39a97c80f61816d5bc41d13980cbef297bd", "output_hash": "2138888e6519b2f4b1b54142e6f9c541102d7f9ee5bcf8e4184c81a382601757"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "2cc78b5220f015781d4bc21b", "attempt_id": "85f943216885", "run_group_id": "E4_core:A4|closed|qwen3b:h2", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:48:32.415125Z", "task_family": "synthetic", "document_id": "syn_03015_2000", "document_hash": "71a7284b9bc3f681d69d91d18b9e8d45a4dda486cd14f4145395786ac6a68a89", "question_id": "q03_exact_copy", "split": "test", "cluster_id": "syn_03015_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 2, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1922, "query_tokens": 585, "message_tokens": 503, "message_utf8_bytes": 1142, "cache_tensor_bytes": null, "serialized_payload_bytes": 1142, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 44797.16570896562, "hops": {"1": 22697.06304196734, "2": 22100.102666998282}, "receiver_prefill_ms": 462.5575829995796, "ttft_ms": 51.39725003391504, "decode_ms": 386.3630419364199}, "total_wall_ms": 45698.33125092555, "ttft_ms": 51.39725003391504, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 6}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "Wilhelmina Dubrovnik", "evidence_artifact": {"ledger": {"facts_total": 61, "facts_retained": 24, "retention_rate": 0.39344262295081966, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.4166666666666667, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1142}, "ledger_by_hop": {"1": {"facts_total": 61, "facts_retained": 24, "retention_rate": 0.39344262295081966, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.4166666666666667, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1161}, "2": {"facts_total": 61, "facts_retained": 24, "retention_rate": 0.39344262295081966, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.4166666666666667, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1142}}}, "reference_key_id": "keys/synthetic/syn_03015_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "6606d2a2280a507a57312c3b4cdff39a97c80f61816d5bc41d13980cbef297bd", "output_hash": "f42f05ebfc96f7258cbfbbb37cd65e114d3d15f869e68e6c95b71fc288c30a32"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "d27a050c4c47c9684fec930b", "attempt_id": "dda3dc56ed86", "run_group_id": "E4_core:A4|closed|qwen3b:h2", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:48:33.333508Z", "task_family": "synthetic", "document_id": "syn_03015_2000", "document_hash": "71a7284b9bc3f681d69d91d18b9e8d45a4dda486cd14f4145395786ac6a68a89", "question_id": "q04_multi_hop", "split": "test", "cluster_id": "syn_03015_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 2, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1922, "query_tokens": 594, "message_tokens": 503, "message_utf8_bytes": 1142, "cache_tensor_bytes": null, "serialized_payload_bytes": 1142, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 44797.16570896562, "hops": {"1": 22697.06304196734, "2": 22100.102666998282}, "receiver_prefill_ms": 496.4328750502318, "ttft_ms": 49.68987498432398, "decode_ms": 454.5455409679562}, "total_wall_ms": 45798.55745891109, "ttft_ms": 49.68987498432398, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 7}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "8421725", "evidence_artifact": {"ledger": {"facts_total": 61, "facts_retained": 24, "retention_rate": 0.39344262295081966, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.4166666666666667, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1142}, "ledger_by_hop": {"1": {"facts_total": 61, "facts_retained": 24, "retention_rate": 0.39344262295081966, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.4166666666666667, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1161}, "2": {"facts_total": 61, "facts_retained": 24, "retention_rate": 0.39344262295081966, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.4166666666666667, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1142}}}, "reference_key_id": "keys/synthetic/syn_03015_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "6606d2a2280a507a57312c3b4cdff39a97c80f61816d5bc41d13980cbef297bd", "output_hash": "e4949aba3b0b77de05a87edaab854329fa75df7bf68b29a3def71e985a06e5b5"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "bec0fcaf73fa0007e190da37", "attempt_id": "8d9b7035cae9", "run_group_id": "E4_core:A4|closed|qwen3b:h2", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:48:34.352298Z", "task_family": "synthetic", "document_id": "syn_03015_2000", "document_hash": "71a7284b9bc3f681d69d91d18b9e8d45a4dda486cd14f4145395786ac6a68a89", "question_id": "q05_multi_hop", "split": "test", "cluster_id": "syn_03015_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 2, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1922, "query_tokens": 594, "message_tokens": 503, "message_utf8_bytes": 1142, "cache_tensor_bytes": null, "serialized_payload_bytes": 1142, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 44797.16570896562, "hops": {"1": 22697.06304196734, "2": 22100.102666998282}, "receiver_prefill_ms": 499.5872500585392, "ttft_ms": 64.96841693297029, "decode_ms": 52.34958289656788}, "total_wall_ms": 45414.66045891866, "ttft_ms": 64.96841693297029, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 61, "facts_retained": 24, "retention_rate": 0.39344262295081966, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.4166666666666667, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1142}, "ledger_by_hop": {"1": {"facts_total": 61, "facts_retained": 24, "retention_rate": 0.39344262295081966, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.4166666666666667, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1161}, "2": {"facts_total": 61, "facts_retained": 24, "retention_rate": 0.39344262295081966, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.4166666666666667, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1142}}}, "reference_key_id": "keys/synthetic/syn_03015_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "6606d2a2280a507a57312c3b4cdff39a97c80f61816d5bc41d13980cbef297bd", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "0ca184f191e6ec281a6f4764", "attempt_id": "f6ee63c8937c", "run_group_id": "E4_core:A4|closed|qwen3b:h2", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:48:34.987274Z", "task_family": "synthetic", "document_id": "syn_03015_2000", "document_hash": "71a7284b9bc3f681d69d91d18b9e8d45a4dda486cd14f4145395786ac6a68a89", "question_id": "q06_negation", "split": "test", "cluster_id": "syn_03015_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 2, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1922, "query_tokens": 586, "message_tokens": 503, "message_utf8_bytes": 1142, "cache_tensor_bytes": null, "serialized_payload_bytes": 1142, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 44797.16570896562, "hops": {"1": 22697.06304196734, "2": 22100.102666998282}, "receiver_prefill_ms": 457.177875097841, "ttft_ms": 59.65791700873524, "decode_ms": 104.93775003124028}, "total_wall_ms": 45419.551708968356, "ttft_ms": 59.65791700873524, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 61, "facts_retained": 24, "retention_rate": 0.39344262295081966, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.4166666666666667, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1142}, "ledger_by_hop": {"1": {"facts_total": 61, "facts_retained": 24, "retention_rate": 0.39344262295081966, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.4166666666666667, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1161}, "2": {"facts_total": 61, "facts_retained": 24, "retention_rate": 0.39344262295081966, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.4166666666666667, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1142}}}, "reference_key_id": "keys/synthetic/syn_03015_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "6606d2a2280a507a57312c3b4cdff39a97c80f61816d5bc41d13980cbef297bd", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "4daae97802704218dcf739f9", "attempt_id": "8cb154e75e65", "run_group_id": "E4_core:A4|closed|qwen3b:h2", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:48:35.626457Z", "task_family": "synthetic", "document_id": "syn_03015_2000", "document_hash": "71a7284b9bc3f681d69d91d18b9e8d45a4dda486cd14f4145395786ac6a68a89", "question_id": "q07_negation", "split": "test", "cluster_id": "syn_03015_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 2, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1922, "query_tokens": 586, "message_tokens": 503, "message_utf8_bytes": 1142, "cache_tensor_bytes": null, "serialized_payload_bytes": 1142, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 44797.16570896562, "hops": {"1": 22697.06304196734, "2": 22100.102666998282}, "receiver_prefill_ms": 462.9593329736963, "ttft_ms": 12.399540981277823, "decode_ms": 12.270125094801188}, "total_wall_ms": 45285.442958935164, "ttft_ms": 12.399540981277823, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 61, "facts_retained": 24, "retention_rate": 0.39344262295081966, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.4166666666666667, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1142}, "ledger_by_hop": {"1": {"facts_total": 61, "facts_retained": 24, "retention_rate": 0.39344262295081966, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.4166666666666667, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1161}, "2": {"facts_total": 61, "facts_retained": 24, "retention_rate": 0.39344262295081966, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.4166666666666667, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1142}}}, "reference_key_id": "keys/synthetic/syn_03015_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "6606d2a2280a507a57312c3b4cdff39a97c80f61816d5bc41d13980cbef297bd", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "ebdbad3df6fc98a7a853df82", "attempt_id": "9b9c2a574ec9", "run_group_id": "E4_core:A4|closed|qwen3b:h2", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:48:36.132274Z", "task_family": "synthetic", "document_id": "syn_03015_2000", "document_hash": "71a7284b9bc3f681d69d91d18b9e8d45a4dda486cd14f4145395786ac6a68a89", "question_id": "q08_revision", "split": "test", "cluster_id": "syn_03015_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 2, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1922, "query_tokens": 600, "message_tokens": 503, "message_utf8_bytes": 1142, "cache_tensor_bytes": null, "serialized_payload_bytes": 1142, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 44797.16570896562, "hops": {"1": 22697.06304196734, "2": 22100.102666998282}, "receiver_prefill_ms": 226.80762503296137, "ttft_ms": 12.268041959032416, "decode_ms": 17.393457936123013}, "total_wall_ms": 45054.38600096386, "ttft_ms": 12.268041959032416, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 61, "facts_retained": 24, "retention_rate": 0.39344262295081966, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.4166666666666667, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1142}, "ledger_by_hop": {"1": {"facts_total": 61, "facts_retained": 24, "retention_rate": 0.39344262295081966, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.4166666666666667, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1161}, "2": {"facts_total": 61, "facts_retained": 24, "retention_rate": 0.39344262295081966, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.4166666666666667, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1142}}}, "reference_key_id": "keys/synthetic/syn_03015_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "6606d2a2280a507a57312c3b4cdff39a97c80f61816d5bc41d13980cbef297bd", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "bf600397ff3de8d25f0dada9", "attempt_id": "b507d51109a4", "run_group_id": "E4_core:A4|closed|qwen3b:h2", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:48:36.408365Z", "task_family": "synthetic", "document_id": "syn_03015_2000", "document_hash": "71a7284b9bc3f681d69d91d18b9e8d45a4dda486cd14f4145395786ac6a68a89", "question_id": "q09_revision", "split": "test", "cluster_id": "syn_03015_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 2, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1922, "query_tokens": 594, "message_tokens": 503, "message_utf8_bytes": 1142, "cache_tensor_bytes": null, "serialized_payload_bytes": 1142, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 44797.16570896562, "hops": {"1": 22697.06304196734, "2": 22100.102666998282}, "receiver_prefill_ms": 370.4542909981683, "ttft_ms": 42.96208301093429, "decode_ms": 492.76533303782344}, "total_wall_ms": 45704.179417924024, "ttft_ms": 42.96208301093429, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 9}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "8,377,480", "evidence_artifact": {"ledger": {"facts_total": 61, "facts_retained": 24, "retention_rate": 0.39344262295081966, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.4166666666666667, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1142}, "ledger_by_hop": {"1": {"facts_total": 61, "facts_retained": 24, "retention_rate": 0.39344262295081966, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.4166666666666667, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1161}, "2": {"facts_total": 61, "facts_retained": 24, "retention_rate": 0.39344262295081966, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.4166666666666667, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1142}}}, "reference_key_id": "keys/synthetic/syn_03015_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "6606d2a2280a507a57312c3b4cdff39a97c80f61816d5bc41d13980cbef297bd", "output_hash": "ec4c68b327a2d2427ed21821415a5f9ddf87134bd9f654bdd69c7d1301febcb0"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "c43580304bd4bd42a0c24697", "attempt_id": "66f8b85f606f", "run_group_id": "E4_core:A4|closed|qwen3b:h2", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:48:37.334592Z", "task_family": "synthetic", "document_id": "syn_03015_2000", "document_hash": "71a7284b9bc3f681d69d91d18b9e8d45a4dda486cd14f4145395786ac6a68a89", "question_id": "q10_temporal", "split": "test", "cluster_id": "syn_03015_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 2, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1922, "query_tokens": 603, "message_tokens": 503, "message_utf8_bytes": 1142, "cache_tensor_bytes": null, "serialized_payload_bytes": 1142, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 44797.16570896562, "hops": {"1": 22697.06304196734, "2": 22100.102666998282}, "receiver_prefill_ms": 417.21462504938245, "ttft_ms": 42.37725003622472, "decode_ms": 25.190209038555622}, "total_wall_ms": 45282.58312505204, "ttft_ms": 42.37725003622472, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 61, "facts_retained": 24, "retention_rate": 0.39344262295081966, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.4166666666666667, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1142}, "ledger_by_hop": {"1": {"facts_total": 61, "facts_retained": 24, "retention_rate": 0.39344262295081966, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.4166666666666667, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1161}, "2": {"facts_total": 61, "facts_retained": 24, "retention_rate": 0.39344262295081966, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.4166666666666667, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1142}}}, "reference_key_id": "keys/synthetic/syn_03015_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "6606d2a2280a507a57312c3b4cdff39a97c80f61816d5bc41d13980cbef297bd", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "ad20108f0a64db9a5d1a11f6", "attempt_id": "3c71144aeb6a", "run_group_id": "E4_core:A4|closed|qwen3b:h2", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:48:37.838359Z", "task_family": "synthetic", "document_id": "syn_03015_2000", "document_hash": "71a7284b9bc3f681d69d91d18b9e8d45a4dda486cd14f4145395786ac6a68a89", "question_id": "q11_aggregation", "split": "test", "cluster_id": "syn_03015_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 2, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1922, "query_tokens": 596, "message_tokens": 503, "message_utf8_bytes": 1142, "cache_tensor_bytes": null, "serialized_payload_bytes": 1142, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 44797.16570896562, "hops": {"1": 22697.06304196734, "2": 22100.102666998282}, "receiver_prefill_ms": 367.3691670410335, "ttft_ms": 44.4581659976393, "decode_ms": 487.52120905555785}, "total_wall_ms": 45697.19108392019, "ttft_ms": 44.4581659976393, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 7}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "1806025", "evidence_artifact": {"ledger": {"facts_total": 61, "facts_retained": 24, "retention_rate": 0.39344262295081966, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.4166666666666667, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1142}, "ledger_by_hop": {"1": {"facts_total": 61, "facts_retained": 24, "retention_rate": 0.39344262295081966, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.4166666666666667, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1161}, "2": {"facts_total": 61, "facts_retained": 24, "retention_rate": 0.39344262295081966, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.4166666666666667, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1142}}}, "reference_key_id": "keys/synthetic/syn_03015_2000.json", "task_scores": {"correct": 0.0, "abstained": 0.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "6606d2a2280a507a57312c3b4cdff39a97c80f61816d5bc41d13980cbef297bd", "output_hash": "9801d1840792d286693349934c9cc13fead44d82d480a42803cf6015c01431f6"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "bf8cd95549fc0bb18684d1f1", "attempt_id": "fa1f17aab5bd", "run_group_id": "E4_core:A4|closed|qwen3b:h2", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:48:38.755429Z", "task_family": "synthetic", "document_id": "syn_03015_2000", "document_hash": "71a7284b9bc3f681d69d91d18b9e8d45a4dda486cd14f4145395786ac6a68a89", "question_id": "q12_unit_conversion", "split": "test", "cluster_id": "syn_03015_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 2, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1922, "query_tokens": 593, "message_tokens": 503, "message_utf8_bytes": 1142, "cache_tensor_bytes": null, "serialized_payload_bytes": 1142, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 44797.16570896562, "hops": {"1": 22697.06304196734, "2": 22100.102666998282}, "receiver_prefill_ms": 399.1947079775855, "ttft_ms": 24.795332923531532, "decode_ms": 44.8900000192225}, "total_wall_ms": 45266.719208913855, "ttft_ms": 24.795332923531532, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 61, "facts_retained": 24, "retention_rate": 0.39344262295081966, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.4166666666666667, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1142}, "ledger_by_hop": {"1": {"facts_total": 61, "facts_retained": 24, "retention_rate": 0.39344262295081966, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.4166666666666667, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1161}, "2": {"facts_total": 61, "facts_retained": 24, "retention_rate": 0.39344262295081966, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.4166666666666667, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1142}}}, "reference_key_id": "keys/synthetic/syn_03015_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "6606d2a2280a507a57312c3b4cdff39a97c80f61816d5bc41d13980cbef297bd", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "6e7572a51f9e45952048aaf8", "attempt_id": "4b8ea8c61b56", "run_group_id": "E4_core:A4|closed|qwen3b:h2", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:48:39.242402Z", "task_family": "synthetic", "document_id": "syn_03015_2000", "document_hash": "71a7284b9bc3f681d69d91d18b9e8d45a4dda486cd14f4145395786ac6a68a89", "question_id": "q13_unanswerable", "split": "test", "cluster_id": "syn_03015_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 2, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1922, "query_tokens": 597, "message_tokens": 503, "message_utf8_bytes": 1142, "cache_tensor_bytes": null, "serialized_payload_bytes": 1142, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 44797.16570896562, "hops": {"1": 22697.06304196734, "2": 22100.102666998282}, "receiver_prefill_ms": 397.3958750721067, "ttft_ms": 23.309290991164744, "decode_ms": 23.66912504658103}, "total_wall_ms": 45242.17525101267, "ttft_ms": 23.309290991164744, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 61, "facts_retained": 24, "retention_rate": 0.39344262295081966, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.4166666666666667, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1142}, "ledger_by_hop": {"1": {"facts_total": 61, "facts_retained": 24, "retention_rate": 0.39344262295081966, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.4166666666666667, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1161}, "2": {"facts_total": 61, "facts_retained": 24, "retention_rate": 0.39344262295081966, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.4166666666666667, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1142}}}, "reference_key_id": "keys/synthetic/syn_03015_2000.json", "task_scores": {"correct": 1.0, "abstained": 1.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "6606d2a2280a507a57312c3b4cdff39a97c80f61816d5bc41d13980cbef297bd", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "42e00211f911dea258ba67c1", "attempt_id": "c627d6593ca5", "run_group_id": "E4_core:A4|closed|qwen3b:h2", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:48:39.704589Z", "task_family": "synthetic", "document_id": "syn_03015_2000", "document_hash": "71a7284b9bc3f681d69d91d18b9e8d45a4dda486cd14f4145395786ac6a68a89", "question_id": "q14_unanswerable", "split": "test", "cluster_id": "syn_03015_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 2, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1922, "query_tokens": 598, "message_tokens": 503, "message_utf8_bytes": 1142, "cache_tensor_bytes": null, "serialized_payload_bytes": 1142, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 44797.16570896562, "hops": {"1": 22697.06304196734, "2": 22100.102666998282}, "receiver_prefill_ms": 434.0372079750523, "ttft_ms": 48.37479197885841, "decode_ms": 96.19120799470693}, "total_wall_ms": 45376.672875951044, "ttft_ms": 48.37479197885841, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 61, "facts_retained": 24, "retention_rate": 0.39344262295081966, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.4166666666666667, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1142}, "ledger_by_hop": {"1": {"facts_total": 61, "facts_retained": 24, "retention_rate": 0.39344262295081966, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.4166666666666667, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1161}, "2": {"facts_total": 61, "facts_retained": 24, "retention_rate": 0.39344262295081966, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.4166666666666667, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1142}}}, "reference_key_id": "keys/synthetic/syn_03015_2000.json", "task_scores": {"correct": 1.0, "abstained": 1.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "6606d2a2280a507a57312c3b4cdff39a97c80f61816d5bc41d13980cbef297bd", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "f490090fb4d0023c4fbd0eab", "attempt_id": "e9b826434eac", "run_group_id": "E4_core:A4|closed|qwen3b:h4", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:49:23.388876Z", "task_family": "synthetic", "document_id": "syn_03015_2000", "document_hash": "71a7284b9bc3f681d69d91d18b9e8d45a4dda486cd14f4145395786ac6a68a89", "question_id": "q00_single_fact", "split": "test", "cluster_id": "syn_03015_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 4, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1922, "query_tokens": 585, "message_tokens": 503, "message_utf8_bytes": 1142, "cache_tensor_bytes": null, "serialized_payload_bytes": 1142, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 87879.06637589913, "hops": {"1": 22697.06304196734, "2": 22100.102666998282, "3": 21632.063457975164, "4": 21449.837208958343}, "receiver_prefill_ms": 379.11479198373854, "ttft_ms": 38.57350000180304, "decode_ms": 47.30904207099229}, "total_wall_ms": 88344.76083389018, "ttft_ms": 38.57350000180304, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 61, "facts_retained": 24, "retention_rate": 0.39344262295081966, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.4166666666666667, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1142}, "ledger_by_hop": {"1": {"facts_total": 61, "facts_retained": 24, "retention_rate": 0.39344262295081966, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.4166666666666667, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1161}, "2": {"facts_total": 61, "facts_retained": 24, "retention_rate": 0.39344262295081966, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.4166666666666667, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1142}, "3": {"facts_total": 61, "facts_retained": 24, "retention_rate": 0.39344262295081966, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.4166666666666667, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1142}, "4": {"facts_total": 61, "facts_retained": 24, "retention_rate": 0.39344262295081966, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.4166666666666667, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1142}}}, "reference_key_id": "keys/synthetic/syn_03015_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "6606d2a2280a507a57312c3b4cdff39a97c80f61816d5bc41d13980cbef297bd", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "e930f65e0e58669bfd1e6f2e", "attempt_id": "20d885baca25", "run_group_id": "E4_core:A4|closed|qwen3b:h4", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:49:23.872744Z", "task_family": "synthetic", "document_id": "syn_03015_2000", "document_hash": "71a7284b9bc3f681d69d91d18b9e8d45a4dda486cd14f4145395786ac6a68a89", "question_id": "q01_single_fact", "split": "test", "cluster_id": "syn_03015_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 4, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1922, "query_tokens": 590, "message_tokens": 503, "message_utf8_bytes": 1142, "cache_tensor_bytes": null, "serialized_payload_bytes": 1142, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 87879.06637589913, "hops": {"1": 22697.06304196734, "2": 22100.102666998282, "3": 21632.063457975164, "4": 21449.837208958343}, "receiver_prefill_ms": 521.8947499524802, "ttft_ms": 42.87079209461808, "decode_ms": 23.34541699383408}, "total_wall_ms": 88468.11920893379, "ttft_ms": 42.87079209461808, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 61, "facts_retained": 24, "retention_rate": 0.39344262295081966, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.4166666666666667, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1142}, "ledger_by_hop": {"1": {"facts_total": 61, "facts_retained": 24, "retention_rate": 0.39344262295081966, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.4166666666666667, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1161}, "2": {"facts_total": 61, "facts_retained": 24, "retention_rate": 0.39344262295081966, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.4166666666666667, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1142}, "3": {"facts_total": 61, "facts_retained": 24, "retention_rate": 0.39344262295081966, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.4166666666666667, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1142}, "4": {"facts_total": 61, "facts_retained": 24, "retention_rate": 0.39344262295081966, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.4166666666666667, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1142}}}, "reference_key_id": "keys/synthetic/syn_03015_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "6606d2a2280a507a57312c3b4cdff39a97c80f61816d5bc41d13980cbef297bd", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "ac961d4eabaab8dec9818656", "attempt_id": "5fa1ee09a179", "run_group_id": "E4_core:A4|closed|qwen3b:h4", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:49:24.479338Z", "task_family": "synthetic", "document_id": "syn_03015_2000", "document_hash": "71a7284b9bc3f681d69d91d18b9e8d45a4dda486cd14f4145395786ac6a68a89", "question_id": "q02_exact_copy", "split": "test", "cluster_id": "syn_03015_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 4, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1922, "query_tokens": 589, "message_tokens": 503, "message_utf8_bytes": 1142, "cache_tensor_bytes": null, "serialized_payload_bytes": 1142, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 87879.06637589913, "hops": {"1": 22697.06304196734, "2": 22100.102666998282, "3": 21632.063457975164, "4": 21449.837208958343}, "receiver_prefill_ms": 395.0342920143157, "ttft_ms": 43.39058301411569, "decode_ms": 192.14854203164577}, "total_wall_ms": 88510.28137584217, "ttft_ms": 43.39058301411569, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 7}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "89.3122", "evidence_artifact": {"ledger": {"facts_total": 61, "facts_retained": 24, "retention_rate": 0.39344262295081966, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.4166666666666667, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1142}, "ledger_by_hop": {"1": {"facts_total": 61, "facts_retained": 24, "retention_rate": 0.39344262295081966, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.4166666666666667, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1161}, "2": {"facts_total": 61, "facts_retained": 24, "retention_rate": 0.39344262295081966, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.4166666666666667, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1142}, "3": {"facts_total": 61, "facts_retained": 24, "retention_rate": 0.39344262295081966, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.4166666666666667, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1142}, "4": {"facts_total": 61, "facts_retained": 24, "retention_rate": 0.39344262295081966, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.4166666666666667, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1142}}}, "reference_key_id": "keys/synthetic/syn_03015_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "6606d2a2280a507a57312c3b4cdff39a97c80f61816d5bc41d13980cbef297bd", "output_hash": "2138888e6519b2f4b1b54142e6f9c541102d7f9ee5bcf8e4184c81a382601757"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "accf3ee44a46640f8b6051cf", "attempt_id": "3232a0b6ff23", "run_group_id": "E4_core:A4|closed|qwen3b:h4", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:49:25.127952Z", "task_family": "synthetic", "document_id": "syn_03015_2000", "document_hash": "71a7284b9bc3f681d69d91d18b9e8d45a4dda486cd14f4145395786ac6a68a89", "question_id": "q03_exact_copy", "split": "test", "cluster_id": "syn_03015_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 4, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1922, "query_tokens": 585, "message_tokens": 503, "message_utf8_bytes": 1142, "cache_tensor_bytes": null, "serialized_payload_bytes": 1142, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 87879.06637589913, "hops": {"1": 22697.06304196734, "2": 22100.102666998282, "3": 21632.063457975164, "4": 21449.837208958343}, "receiver_prefill_ms": 372.58995801676065, "ttft_ms": 54.415374994277954, "decode_ms": 394.33429192285985}, "total_wall_ms": 88701.12204283942, "ttft_ms": 54.415374994277954, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 6}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "Wilhelmina Dubrovnik", "evidence_artifact": {"ledger": {"facts_total": 61, "facts_retained": 24, "retention_rate": 0.39344262295081966, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.4166666666666667, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1142}, "ledger_by_hop": {"1": {"facts_total": 61, "facts_retained": 24, "retention_rate": 0.39344262295081966, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.4166666666666667, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1161}, "2": {"facts_total": 61, "facts_retained": 24, "retention_rate": 0.39344262295081966, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.4166666666666667, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1142}, "3": {"facts_total": 61, "facts_retained": 24, "retention_rate": 0.39344262295081966, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.4166666666666667, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1142}, "4": {"facts_total": 61, "facts_retained": 24, "retention_rate": 0.39344262295081966, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.4166666666666667, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1142}}}, "reference_key_id": "keys/synthetic/syn_03015_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "6606d2a2280a507a57312c3b4cdff39a97c80f61816d5bc41d13980cbef297bd", "output_hash": "f42f05ebfc96f7258cbfbbb37cd65e114d3d15f869e68e6c95b71fc288c30a32"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "31b07e3f4e8fe7f557c3f655", "attempt_id": "7ce3e3fe14be", "run_group_id": "E4_core:A4|closed|qwen3b:h4", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:49:25.966362Z", "task_family": "synthetic", "document_id": "syn_03015_2000", "document_hash": "71a7284b9bc3f681d69d91d18b9e8d45a4dda486cd14f4145395786ac6a68a89", "question_id": "q04_multi_hop", "split": "test", "cluster_id": "syn_03015_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 4, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1922, "query_tokens": 594, "message_tokens": 503, "message_utf8_bytes": 1142, "cache_tensor_bytes": null, "serialized_payload_bytes": 1142, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 87879.06637589913, "hops": {"1": 22697.06304196734, "2": 22100.102666998282, "3": 21632.063457975164, "4": 21449.837208958343}, "receiver_prefill_ms": 406.8464169977233, "ttft_ms": 23.286042036488652, "decode_ms": 243.00125008448958}, "total_wall_ms": 88552.77633492369, "ttft_ms": 23.286042036488652, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 7}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "8421725", "evidence_artifact": {"ledger": {"facts_total": 61, "facts_retained": 24, "retention_rate": 0.39344262295081966, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.4166666666666667, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1142}, "ledger_by_hop": {"1": {"facts_total": 61, "facts_retained": 24, "retention_rate": 0.39344262295081966, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.4166666666666667, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1161}, "2": {"facts_total": 61, "facts_retained": 24, "retention_rate": 0.39344262295081966, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.4166666666666667, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1142}, "3": {"facts_total": 61, "facts_retained": 24, "retention_rate": 0.39344262295081966, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.4166666666666667, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1142}, "4": {"facts_total": 61, "facts_retained": 24, "retention_rate": 0.39344262295081966, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.4166666666666667, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1142}}}, "reference_key_id": "keys/synthetic/syn_03015_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "6606d2a2280a507a57312c3b4cdff39a97c80f61816d5bc41d13980cbef297bd", "output_hash": "e4949aba3b0b77de05a87edaab854329fa75df7bf68b29a3def71e985a06e5b5"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "31036455ed4f53eb8b1ac9bc", "attempt_id": "db37318e81b3", "run_group_id": "E4_core:A4|closed|qwen3b:h4", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:49:26.658430Z", "task_family": "synthetic", "document_id": "syn_03015_2000", "document_hash": "71a7284b9bc3f681d69d91d18b9e8d45a4dda486cd14f4145395786ac6a68a89", "question_id": "q05_multi_hop", "split": "test", "cluster_id": "syn_03015_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 4, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1922, "query_tokens": 594, "message_tokens": 503, "message_utf8_bytes": 1142, "cache_tensor_bytes": null, "serialized_payload_bytes": 1142, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 87879.06637589913, "hops": {"1": 22697.06304196734, "2": 22100.102666998282, "3": 21632.063457975164, "4": 21449.837208958343}, "receiver_prefill_ms": 380.49020804464817, "ttft_ms": 49.53395901247859, "decode_ms": 42.38683299627155}, "total_wall_ms": 88352.47766785324, "ttft_ms": 49.53395901247859, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 61, "facts_retained": 24, "retention_rate": 0.39344262295081966, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.4166666666666667, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1142}, "ledger_by_hop": {"1": {"facts_total": 61, "facts_retained": 24, "retention_rate": 0.39344262295081966, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.4166666666666667, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1161}, "2": {"facts_total": 61, "facts_retained": 24, "retention_rate": 0.39344262295081966, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.4166666666666667, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1142}, "3": {"facts_total": 61, "facts_retained": 24, "retention_rate": 0.39344262295081966, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.4166666666666667, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1142}, "4": {"facts_total": 61, "facts_retained": 24, "retention_rate": 0.39344262295081966, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.4166666666666667, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1142}}}, "reference_key_id": "keys/synthetic/syn_03015_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "6606d2a2280a507a57312c3b4cdff39a97c80f61816d5bc41d13980cbef297bd", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "3a5b6bb76ea878721abc9854", "attempt_id": "1a22b6d5f16b", "run_group_id": "E4_core:A4|closed|qwen3b:h4", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:49:27.149092Z", "task_family": "synthetic", "document_id": "syn_03015_2000", "document_hash": "71a7284b9bc3f681d69d91d18b9e8d45a4dda486cd14f4145395786ac6a68a89", "question_id": "q06_negation", "split": "test", "cluster_id": "syn_03015_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 4, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1922, "query_tokens": 586, "message_tokens": 503, "message_utf8_bytes": 1142, "cache_tensor_bytes": null, "serialized_payload_bytes": 1142, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 87879.06637589913, "hops": {"1": 22697.06304196734, "2": 22100.102666998282, "3": 21632.063457975164, "4": 21449.837208958343}, "receiver_prefill_ms": 545.0069579528645, "ttft_ms": 38.49299997091293, "decode_ms": 34.49049999471754}, "total_wall_ms": 88497.85146000795, "ttft_ms": 38.49299997091293, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 61, "facts_retained": 24, "retention_rate": 0.39344262295081966, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.4166666666666667, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1142}, "ledger_by_hop": {"1": {"facts_total": 61, "facts_retained": 24, "retention_rate": 0.39344262295081966, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.4166666666666667, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1161}, "2": {"facts_total": 61, "facts_retained": 24, "retention_rate": 0.39344262295081966, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.4166666666666667, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1142}, "3": {"facts_total": 61, "facts_retained": 24, "retention_rate": 0.39344262295081966, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.4166666666666667, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1142}, "4": {"facts_total": 61, "facts_retained": 24, "retention_rate": 0.39344262295081966, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.4166666666666667, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1142}}}, "reference_key_id": "keys/synthetic/syn_03015_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "6606d2a2280a507a57312c3b4cdff39a97c80f61816d5bc41d13980cbef297bd", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "9df71236e38e39f4c7ee5dc5", "attempt_id": "e8d500c16eb2", "run_group_id": "E4_core:A4|closed|qwen3b:h4", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:49:27.786317Z", "task_family": "synthetic", "document_id": "syn_03015_2000", "document_hash": "71a7284b9bc3f681d69d91d18b9e8d45a4dda486cd14f4145395786ac6a68a89", "question_id": "q07_negation", "split": "test", "cluster_id": "syn_03015_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 4, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1922, "query_tokens": 586, "message_tokens": 503, "message_utf8_bytes": 1142, "cache_tensor_bytes": null, "serialized_payload_bytes": 1142, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 87879.06637589913, "hops": {"1": 22697.06304196734, "2": 22100.102666998282, "3": 21632.063457975164, "4": 21449.837208958343}, "receiver_prefill_ms": 391.59804105293006, "ttft_ms": 40.21424998063594, "decode_ms": 23.82987504824996}, "total_wall_ms": 88335.48079279717, "ttft_ms": 40.21424998063594, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 61, "facts_retained": 24, "retention_rate": 0.39344262295081966, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.4166666666666667, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1142}, "ledger_by_hop": {"1": {"facts_total": 61, "facts_retained": 24, "retention_rate": 0.39344262295081966, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.4166666666666667, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1161}, "2": {"facts_total": 61, "facts_retained": 24, "retention_rate": 0.39344262295081966, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.4166666666666667, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1142}, "3": {"facts_total": 61, "facts_retained": 24, "retention_rate": 0.39344262295081966, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.4166666666666667, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1142}, "4": {"facts_total": 61, "facts_retained": 24, "retention_rate": 0.39344262295081966, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.4166666666666667, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1142}}}, "reference_key_id": "keys/synthetic/syn_03015_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "6606d2a2280a507a57312c3b4cdff39a97c80f61816d5bc41d13980cbef297bd", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "19f171f163d31539edcdb2db", "attempt_id": "8131019f5b72", "run_group_id": "E4_core:A4|closed|qwen3b:h4", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:49:28.260489Z", "task_family": "synthetic", "document_id": "syn_03015_2000", "document_hash": "71a7284b9bc3f681d69d91d18b9e8d45a4dda486cd14f4145395786ac6a68a89", "question_id": "q08_revision", "split": "test", "cluster_id": "syn_03015_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 4, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1922, "query_tokens": 600, "message_tokens": 503, "message_utf8_bytes": 1142, "cache_tensor_bytes": null, "serialized_payload_bytes": 1142, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 87879.06637589913, "hops": {"1": 22697.06304196734, "2": 22100.102666998282, "3": 21632.063457975164, "4": 21449.837208958343}, "receiver_prefill_ms": 365.7087080646306, "ttft_ms": 39.98545790091157, "decode_ms": 59.831000049598515}, "total_wall_ms": 88345.50941782072, "ttft_ms": 39.98545790091157, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 61, "facts_retained": 24, "retention_rate": 0.39344262295081966, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.4166666666666667, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1142}, "ledger_by_hop": {"1": {"facts_total": 61, "facts_retained": 24, "retention_rate": 0.39344262295081966, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.4166666666666667, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1161}, "2": {"facts_total": 61, "facts_retained": 24, "retention_rate": 0.39344262295081966, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.4166666666666667, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1142}, "3": {"facts_total": 61, "facts_retained": 24, "retention_rate": 0.39344262295081966, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.4166666666666667, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1142}, "4": {"facts_total": 61, "facts_retained": 24, "retention_rate": 0.39344262295081966, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.4166666666666667, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1142}}}, "reference_key_id": "keys/synthetic/syn_03015_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "6606d2a2280a507a57312c3b4cdff39a97c80f61816d5bc41d13980cbef297bd", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "38e97155161db6357b7cfcf3", "attempt_id": "11ff2c769b86", "run_group_id": "E4_core:A4|closed|qwen3b:h4", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:49:28.744648Z", "task_family": "synthetic", "document_id": "syn_03015_2000", "document_hash": "71a7284b9bc3f681d69d91d18b9e8d45a4dda486cd14f4145395786ac6a68a89", "question_id": "q09_revision", "split": "test", "cluster_id": "syn_03015_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 4, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1922, "query_tokens": 594, "message_tokens": 503, "message_utf8_bytes": 1142, "cache_tensor_bytes": null, "serialized_payload_bytes": 1142, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 87879.06637589913, "hops": {"1": 22697.06304196734, "2": 22100.102666998282, "3": 21632.063457975164, "4": 21449.837208958343}, "receiver_prefill_ms": 517.1423749998212, "ttft_ms": 63.393249991349876, "decode_ms": 296.62137501873076}, "total_wall_ms": 88756.93645887077, "ttft_ms": 63.393249991349876, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 9}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "8,377,480", "evidence_artifact": {"ledger": {"facts_total": 61, "facts_retained": 24, "retention_rate": 0.39344262295081966, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.4166666666666667, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1142}, "ledger_by_hop": {"1": {"facts_total": 61, "facts_retained": 24, "retention_rate": 0.39344262295081966, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.4166666666666667, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1161}, "2": {"facts_total": 61, "facts_retained": 24, "retention_rate": 0.39344262295081966, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.4166666666666667, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1142}, "3": {"facts_total": 61, "facts_retained": 24, "retention_rate": 0.39344262295081966, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.4166666666666667, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1142}, "4": {"facts_total": 61, "facts_retained": 24, "retention_rate": 0.39344262295081966, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.4166666666666667, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1142}}}, "reference_key_id": "keys/synthetic/syn_03015_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "6606d2a2280a507a57312c3b4cdff39a97c80f61816d5bc41d13980cbef297bd", "output_hash": "ec4c68b327a2d2427ed21821415a5f9ddf87134bd9f654bdd69c7d1301febcb0"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "93df9dd70c480737bf945d93", "attempt_id": "b90a0bfd98c4", "run_group_id": "E4_core:A4|closed|qwen3b:h4", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:49:29.639636Z", "task_family": "synthetic", "document_id": "syn_03015_2000", "document_hash": "71a7284b9bc3f681d69d91d18b9e8d45a4dda486cd14f4145395786ac6a68a89", "question_id": "q10_temporal", "split": "test", "cluster_id": "syn_03015_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 4, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1922, "query_tokens": 603, "message_tokens": 503, "message_utf8_bytes": 1142, "cache_tensor_bytes": null, "serialized_payload_bytes": 1142, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 87879.06637589913, "hops": {"1": 22697.06304196734, "2": 22100.102666998282, "3": 21632.063457975164, "4": 21449.837208958343}, "receiver_prefill_ms": 392.97658298164606, "ttft_ms": 32.3246669722721, "decode_ms": 30.248665949329734}, "total_wall_ms": 88335.28470888268, "ttft_ms": 32.3246669722721, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 61, "facts_retained": 24, "retention_rate": 0.39344262295081966, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.4166666666666667, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1142}, "ledger_by_hop": {"1": {"facts_total": 61, "facts_retained": 24, "retention_rate": 0.39344262295081966, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.4166666666666667, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1161}, "2": {"facts_total": 61, "facts_retained": 24, "retention_rate": 0.39344262295081966, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.4166666666666667, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1142}, "3": {"facts_total": 61, "facts_retained": 24, "retention_rate": 0.39344262295081966, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.4166666666666667, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1142}, "4": {"facts_total": 61, "facts_retained": 24, "retention_rate": 0.39344262295081966, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.4166666666666667, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1142}}}, "reference_key_id": "keys/synthetic/syn_03015_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "6606d2a2280a507a57312c3b4cdff39a97c80f61816d5bc41d13980cbef297bd", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "ffcbb6ed40c12b0a7866f3b7", "attempt_id": "4d1267453040", "run_group_id": "E4_core:A4|closed|qwen3b:h4", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:49:30.113467Z", "task_family": "synthetic", "document_id": "syn_03015_2000", "document_hash": "71a7284b9bc3f681d69d91d18b9e8d45a4dda486cd14f4145395786ac6a68a89", "question_id": "q11_aggregation", "split": "test", "cluster_id": "syn_03015_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 4, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1922, "query_tokens": 596, "message_tokens": 503, "message_utf8_bytes": 1142, "cache_tensor_bytes": null, "serialized_payload_bytes": 1142, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 87879.06637589913, "hops": {"1": 22697.06304196734, "2": 22100.102666998282, "3": 21632.063457975164, "4": 21449.837208958343}, "receiver_prefill_ms": 350.53875006269664, "ttft_ms": 43.124458054080606, "decode_ms": 467.35062496736646}, "total_wall_ms": 88740.83616689313, "ttft_ms": 43.124458054080606, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 7}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "1806025", "evidence_artifact": {"ledger": {"facts_total": 61, "facts_retained": 24, "retention_rate": 0.39344262295081966, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.4166666666666667, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1142}, "ledger_by_hop": {"1": {"facts_total": 61, "facts_retained": 24, "retention_rate": 0.39344262295081966, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.4166666666666667, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1161}, "2": {"facts_total": 61, "facts_retained": 24, "retention_rate": 0.39344262295081966, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.4166666666666667, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1142}, "3": {"facts_total": 61, "facts_retained": 24, "retention_rate": 0.39344262295081966, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.4166666666666667, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1142}, "4": {"facts_total": 61, "facts_retained": 24, "retention_rate": 0.39344262295081966, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.4166666666666667, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1142}}}, "reference_key_id": "keys/synthetic/syn_03015_2000.json", "task_scores": {"correct": 0.0, "abstained": 0.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "6606d2a2280a507a57312c3b4cdff39a97c80f61816d5bc41d13980cbef297bd", "output_hash": "9801d1840792d286693349934c9cc13fead44d82d480a42803cf6015c01431f6"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "364170e12375f3d29de1b03d", "attempt_id": "56a9f6b91475", "run_group_id": "E4_core:A4|closed|qwen3b:h4", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:49:30.992712Z", "task_family": "synthetic", "document_id": "syn_03015_2000", "document_hash": "71a7284b9bc3f681d69d91d18b9e8d45a4dda486cd14f4145395786ac6a68a89", "question_id": "q12_unit_conversion", "split": "test", "cluster_id": "syn_03015_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 4, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1922, "query_tokens": 593, "message_tokens": 503, "message_utf8_bytes": 1142, "cache_tensor_bytes": null, "serialized_payload_bytes": 1142, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 87879.06637589913, "hops": {"1": 22697.06304196734, "2": 22100.102666998282, "3": 21632.063457975164, "4": 21449.837208958343}, "receiver_prefill_ms": 396.66137495078146, "ttft_ms": 35.595959052443504, "decode_ms": 31.799125019460917}, "total_wall_ms": 88343.766958802, "ttft_ms": 35.595959052443504, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 61, "facts_retained": 24, "retention_rate": 0.39344262295081966, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.4166666666666667, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1142}, "ledger_by_hop": {"1": {"facts_total": 61, "facts_retained": 24, "retention_rate": 0.39344262295081966, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.4166666666666667, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1161}, "2": {"facts_total": 61, "facts_retained": 24, "retention_rate": 0.39344262295081966, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.4166666666666667, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1142}, "3": {"facts_total": 61, "facts_retained": 24, "retention_rate": 0.39344262295081966, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.4166666666666667, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1142}, "4": {"facts_total": 61, "facts_retained": 24, "retention_rate": 0.39344262295081966, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.4166666666666667, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1142}}}, "reference_key_id": "keys/synthetic/syn_03015_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "6606d2a2280a507a57312c3b4cdff39a97c80f61816d5bc41d13980cbef297bd", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "ee1d75fbd3391b59eedebcc9", "attempt_id": "b1616616936a", "run_group_id": "E4_core:A4|closed|qwen3b:h4", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:49:31.474424Z", "task_family": "synthetic", "document_id": "syn_03015_2000", "document_hash": "71a7284b9bc3f681d69d91d18b9e8d45a4dda486cd14f4145395786ac6a68a89", "question_id": "q13_unanswerable", "split": "test", "cluster_id": "syn_03015_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 4, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1922, "query_tokens": 597, "message_tokens": 503, "message_utf8_bytes": 1142, "cache_tensor_bytes": null, "serialized_payload_bytes": 1142, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 87879.06637589913, "hops": {"1": 22697.06304196734, "2": 22100.102666998282, "3": 21632.063457975164, "4": 21449.837208958343}, "receiver_prefill_ms": 402.80795900616795, "ttft_ms": 24.86087498255074, "decode_ms": 19.185375072993338}, "total_wall_ms": 88326.59795891959, "ttft_ms": 24.86087498255074, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 61, "facts_retained": 24, "retention_rate": 0.39344262295081966, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.4166666666666667, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1142}, "ledger_by_hop": {"1": {"facts_total": 61, "facts_retained": 24, "retention_rate": 0.39344262295081966, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.4166666666666667, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1161}, "2": {"facts_total": 61, "facts_retained": 24, "retention_rate": 0.39344262295081966, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.4166666666666667, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1142}, "3": {"facts_total": 61, "facts_retained": 24, "retention_rate": 0.39344262295081966, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.4166666666666667, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1142}, "4": {"facts_total": 61, "facts_retained": 24, "retention_rate": 0.39344262295081966, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.4166666666666667, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1142}}}, "reference_key_id": "keys/synthetic/syn_03015_2000.json", "task_scores": {"correct": 1.0, "abstained": 1.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "6606d2a2280a507a57312c3b4cdff39a97c80f61816d5bc41d13980cbef297bd", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "75c2e00b4619c0c357bf1f13", "attempt_id": "182a2e8bd626", "run_group_id": "E4_core:A4|closed|qwen3b:h4", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:49:31.938928Z", "task_family": "synthetic", "document_id": "syn_03015_2000", "document_hash": "71a7284b9bc3f681d69d91d18b9e8d45a4dda486cd14f4145395786ac6a68a89", "question_id": "q14_unanswerable", "split": "test", "cluster_id": "syn_03015_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 4, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1922, "query_tokens": 598, "message_tokens": 503, "message_utf8_bytes": 1142, "cache_tensor_bytes": null, "serialized_payload_bytes": 1142, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 87879.06637589913, "hops": {"1": 22697.06304196734, "2": 22100.102666998282, "3": 21632.063457975164, "4": 21449.837208958343}, "receiver_prefill_ms": 413.7940000509843, "ttft_ms": 57.76554194744676, "decode_ms": 90.01649997662753}, "total_wall_ms": 88441.44841795787, "ttft_ms": 57.76554194744676, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 61, "facts_retained": 24, "retention_rate": 0.39344262295081966, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.4166666666666667, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1142}, "ledger_by_hop": {"1": {"facts_total": 61, "facts_retained": 24, "retention_rate": 0.39344262295081966, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.4166666666666667, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1161}, "2": {"facts_total": 61, "facts_retained": 24, "retention_rate": 0.39344262295081966, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.4166666666666667, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1142}, "3": {"facts_total": 61, "facts_retained": 24, "retention_rate": 0.39344262295081966, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.4166666666666667, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1142}, "4": {"facts_total": 61, "facts_retained": 24, "retention_rate": 0.39344262295081966, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.4166666666666667, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1142}}}, "reference_key_id": "keys/synthetic/syn_03015_2000.json", "task_scores": {"correct": 1.0, "abstained": 1.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "6606d2a2280a507a57312c3b4cdff39a97c80f61816d5bc41d13980cbef297bd", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "a10bca7f1e4c1b88453870c5", "attempt_id": "2f4d1df06135", "run_group_id": "E4_core:A4|closed|qwen3b:h8", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:50:58.861031Z", "task_family": "synthetic", "document_id": "syn_03015_2000", "document_hash": "71a7284b9bc3f681d69d91d18b9e8d45a4dda486cd14f4145395786ac6a68a89", "question_id": "q00_single_fact", "split": "test", "cluster_id": "syn_03015_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 8, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1922, "query_tokens": 585, "message_tokens": 503, "message_utf8_bytes": 1142, "cache_tensor_bytes": null, "serialized_payload_bytes": 1142, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 174213.77120900434, "hops": {"1": 22697.06304196734, "2": 22100.102666998282, "3": 21632.063457975164, "4": 21449.837208958343, "5": 21619.770125020295, "6": 21653.251958079636, "7": 21466.14995796699, "8": 21595.532792038284}, "receiver_prefill_ms": 394.94804199784994, "ttft_ms": 20.972458994947374, "decode_ms": 20.521917031146586}, "total_wall_ms": 174650.81420901697, "ttft_ms": 20.972458994947374, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 61, "facts_retained": 24, "retention_rate": 0.39344262295081966, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.4166666666666667, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1142}, "ledger_by_hop": {"1": {"facts_total": 61, "facts_retained": 24, "retention_rate": 0.39344262295081966, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.4166666666666667, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1161}, "2": {"facts_total": 61, "facts_retained": 24, "retention_rate": 0.39344262295081966, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.4166666666666667, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1142}, "3": {"facts_total": 61, "facts_retained": 24, "retention_rate": 0.39344262295081966, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.4166666666666667, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1142}, "4": {"facts_total": 61, "facts_retained": 24, "retention_rate": 0.39344262295081966, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.4166666666666667, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1142}, "5": {"facts_total": 61, "facts_retained": 24, "retention_rate": 0.39344262295081966, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.4166666666666667, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1142}, "6": {"facts_total": 61, "facts_retained": 24, "retention_rate": 0.39344262295081966, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.4166666666666667, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1142}, "7": {"facts_total": 61, "facts_retained": 24, "retention_rate": 0.39344262295081966, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.4166666666666667, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1142}, "8": {"facts_total": 61, "facts_retained": 24, "retention_rate": 0.39344262295081966, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.4166666666666667, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1142}}}, "reference_key_id": "keys/synthetic/syn_03015_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "6606d2a2280a507a57312c3b4cdff39a97c80f61816d5bc41d13980cbef297bd", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "5010e7d81877d10c69386903", "attempt_id": "2221fe03a2e8", "run_group_id": "E4_core:A4|closed|qwen3b:h8", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:50:59.315850Z", "task_family": "synthetic", "document_id": "syn_03015_2000", "document_hash": "71a7284b9bc3f681d69d91d18b9e8d45a4dda486cd14f4145395786ac6a68a89", "question_id": "q01_single_fact", "split": "test", "cluster_id": "syn_03015_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 8, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1922, "query_tokens": 590, "message_tokens": 503, "message_utf8_bytes": 1142, "cache_tensor_bytes": null, "serialized_payload_bytes": 1142, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 174213.77120900434, "hops": {"1": 22697.06304196734, "2": 22100.102666998282, "3": 21632.063457975164, "4": 21449.837208958343, "5": 21619.770125020295, "6": 21653.251958079636, "7": 21466.14995796699, "8": 21595.532792038284}, "receiver_prefill_ms": 435.9164579072967, "ttft_ms": 46.139499987475574, "decode_ms": 95.70249996613711}, "total_wall_ms": 174792.4344589701, "ttft_ms": 46.139499987475574, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 61, "facts_retained": 24, "retention_rate": 0.39344262295081966, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.4166666666666667, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1142}, "ledger_by_hop": {"1": {"facts_total": 61, "facts_retained": 24, "retention_rate": 0.39344262295081966, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.4166666666666667, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1161}, "2": {"facts_total": 61, "facts_retained": 24, "retention_rate": 0.39344262295081966, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.4166666666666667, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1142}, "3": {"facts_total": 61, "facts_retained": 24, "retention_rate": 0.39344262295081966, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.4166666666666667, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1142}, "4": {"facts_total": 61, "facts_retained": 24, "retention_rate": 0.39344262295081966, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.4166666666666667, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1142}, "5": {"facts_total": 61, "facts_retained": 24, "retention_rate": 0.39344262295081966, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.4166666666666667, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1142}, "6": {"facts_total": 61, "facts_retained": 24, "retention_rate": 0.39344262295081966, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.4166666666666667, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1142}, "7": {"facts_total": 61, "facts_retained": 24, "retention_rate": 0.39344262295081966, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.4166666666666667, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1142}, "8": {"facts_total": 61, "facts_retained": 24, "retention_rate": 0.39344262295081966, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.4166666666666667, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1142}}}, "reference_key_id": "keys/synthetic/syn_03015_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "6606d2a2280a507a57312c3b4cdff39a97c80f61816d5bc41d13980cbef297bd", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "b4f0c2374d8bb0102be3aab1", "attempt_id": "a6b0fe5232a9", "run_group_id": "E4_core:A4|closed|qwen3b:h8", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:50:59.914443Z", "task_family": "synthetic", "document_id": "syn_03015_2000", "document_hash": "71a7284b9bc3f681d69d91d18b9e8d45a4dda486cd14f4145395786ac6a68a89", "question_id": "q02_exact_copy", "split": "test", "cluster_id": "syn_03015_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 8, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1922, "query_tokens": 589, "message_tokens": 503, "message_utf8_bytes": 1142, "cache_tensor_bytes": null, "serialized_payload_bytes": 1142, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 174213.77120900434, "hops": {"1": 22697.06304196734, "2": 22100.102666998282, "3": 21632.063457975164, "4": 21449.837208958343, "5": 21619.770125020295, "6": 21653.251958079636, "7": 21466.14995796699, "8": 21595.532792038284}, "receiver_prefill_ms": 468.51958299521357, "ttft_ms": 39.06354203354567, "decode_ms": 246.15666596218944}, "total_wall_ms": 174968.2065011002, "ttft_ms": 39.06354203354567, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 7}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "89.3122", "evidence_artifact": {"ledger": {"facts_total": 61, "facts_retained": 24, "retention_rate": 0.39344262295081966, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.4166666666666667, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1142}, "ledger_by_hop": {"1": {"facts_total": 61, "facts_retained": 24, "retention_rate": 0.39344262295081966, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.4166666666666667, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1161}, "2": {"facts_total": 61, "facts_retained": 24, "retention_rate": 0.39344262295081966, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.4166666666666667, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1142}, "3": {"facts_total": 61, "facts_retained": 24, "retention_rate": 0.39344262295081966, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.4166666666666667, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1142}, "4": {"facts_total": 61, "facts_retained": 24, "retention_rate": 0.39344262295081966, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.4166666666666667, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1142}, "5": {"facts_total": 61, "facts_retained": 24, "retention_rate": 0.39344262295081966, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.4166666666666667, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1142}, "6": {"facts_total": 61, "facts_retained": 24, "retention_rate": 0.39344262295081966, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.4166666666666667, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1142}, "7": {"facts_total": 61, "facts_retained": 24, "retention_rate": 0.39344262295081966, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.4166666666666667, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1142}, "8": {"facts_total": 61, "facts_retained": 24, "retention_rate": 0.39344262295081966, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.4166666666666667, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1142}}}, "reference_key_id": "keys/synthetic/syn_03015_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "6606d2a2280a507a57312c3b4cdff39a97c80f61816d5bc41d13980cbef297bd", "output_hash": "2138888e6519b2f4b1b54142e6f9c541102d7f9ee5bcf8e4184c81a382601757"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "c7990ceb2ff08c59c0376c6c", "attempt_id": "1029b8d7230b", "run_group_id": "E4_core:A4|closed|qwen3b:h8", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:51:00.686225Z", "task_family": "synthetic", "document_id": "syn_03015_2000", "document_hash": "71a7284b9bc3f681d69d91d18b9e8d45a4dda486cd14f4145395786ac6a68a89", "question_id": "q03_exact_copy", "split": "test", "cluster_id": "syn_03015_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 8, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1922, "query_tokens": 585, "message_tokens": 503, "message_utf8_bytes": 1142, "cache_tensor_bytes": null, "serialized_payload_bytes": 1142, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 174213.77120900434, "hops": {"1": 22697.06304196734, "2": 22100.102666998282, "3": 21632.063457975164, "4": 21449.837208958343, "5": 21619.770125020295, "6": 21653.251958079636, "7": 21466.14995796699, "8": 21595.532792038284}, "receiver_prefill_ms": 376.3207080774009, "ttft_ms": 48.9623750327155, "decode_ms": 375.2556670224294}, "total_wall_ms": 175015.1642509736, "ttft_ms": 48.9623750327155, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 6}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "Wilhelmina Dubrovnik", "evidence_artifact": {"ledger": {"facts_total": 61, "facts_retained": 24, "retention_rate": 0.39344262295081966, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.4166666666666667, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1142}, "ledger_by_hop": {"1": {"facts_total": 61, "facts_retained": 24, "retention_rate": 0.39344262295081966, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.4166666666666667, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1161}, "2": {"facts_total": 61, "facts_retained": 24, "retention_rate": 0.39344262295081966, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.4166666666666667, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1142}, "3": {"facts_total": 61, "facts_retained": 24, "retention_rate": 0.39344262295081966, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.4166666666666667, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1142}, "4": {"facts_total": 61, "facts_retained": 24, "retention_rate": 0.39344262295081966, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.4166666666666667, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1142}, "5": {"facts_total": 61, "facts_retained": 24, "retention_rate": 0.39344262295081966, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.4166666666666667, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1142}, "6": {"facts_total": 61, "facts_retained": 24, "retention_rate": 0.39344262295081966, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.4166666666666667, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1142}, "7": {"facts_total": 61, "facts_retained": 24, "retention_rate": 0.39344262295081966, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.4166666666666667, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1142}, "8": {"facts_total": 61, "facts_retained": 24, "retention_rate": 0.39344262295081966, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.4166666666666667, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1142}}}, "reference_key_id": "keys/synthetic/syn_03015_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "6606d2a2280a507a57312c3b4cdff39a97c80f61816d5bc41d13980cbef297bd", "output_hash": "f42f05ebfc96f7258cbfbbb37cd65e114d3d15f869e68e6c95b71fc288c30a32"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "7ed06d5144c2aa2675593130", "attempt_id": "759f9acb715f", "run_group_id": "E4_core:A4|closed|qwen3b:h8", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:51:01.505851Z", "task_family": "synthetic", "document_id": "syn_03015_2000", "document_hash": "71a7284b9bc3f681d69d91d18b9e8d45a4dda486cd14f4145395786ac6a68a89", "question_id": "q04_multi_hop", "split": "test", "cluster_id": "syn_03015_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 8, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1922, "query_tokens": 594, "message_tokens": 503, "message_utf8_bytes": 1142, "cache_tensor_bytes": null, "serialized_payload_bytes": 1142, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 174213.77120900434, "hops": {"1": 22697.06304196734, "2": 22100.102666998282, "3": 21632.063457975164, "4": 21449.837208958343, "5": 21619.770125020295, "6": 21653.251958079636, "7": 21466.14995796699, "8": 21595.532792038284}, "receiver_prefill_ms": 467.0606249710545, "ttft_ms": 38.31787500530481, "decode_ms": 215.84295807406306}, "total_wall_ms": 174935.63270906452, "ttft_ms": 38.31787500530481, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 7}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "8421725", "evidence_artifact": {"ledger": {"facts_total": 61, "facts_retained": 24, "retention_rate": 0.39344262295081966, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.4166666666666667, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1142}, "ledger_by_hop": {"1": {"facts_total": 61, "facts_retained": 24, "retention_rate": 0.39344262295081966, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.4166666666666667, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1161}, "2": {"facts_total": 61, "facts_retained": 24, "retention_rate": 0.39344262295081966, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.4166666666666667, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1142}, "3": {"facts_total": 61, "facts_retained": 24, "retention_rate": 0.39344262295081966, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.4166666666666667, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1142}, "4": {"facts_total": 61, "facts_retained": 24, "retention_rate": 0.39344262295081966, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.4166666666666667, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1142}, "5": {"facts_total": 61, "facts_retained": 24, "retention_rate": 0.39344262295081966, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.4166666666666667, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1142}, "6": {"facts_total": 61, "facts_retained": 24, "retention_rate": 0.39344262295081966, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.4166666666666667, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1142}, "7": {"facts_total": 61, "facts_retained": 24, "retention_rate": 0.39344262295081966, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.4166666666666667, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1142}, "8": {"facts_total": 61, "facts_retained": 24, "retention_rate": 0.39344262295081966, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.4166666666666667, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1142}}}, "reference_key_id": "keys/synthetic/syn_03015_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "6606d2a2280a507a57312c3b4cdff39a97c80f61816d5bc41d13980cbef297bd", "output_hash": "e4949aba3b0b77de05a87edaab854329fa75df7bf68b29a3def71e985a06e5b5"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "d5a73750c4efc2f0c21274d6", "attempt_id": "b16e2e598a9e", "run_group_id": "E4_core:A4|closed|qwen3b:h8", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:51:02.244354Z", "task_family": "synthetic", "document_id": "syn_03015_2000", "document_hash": "71a7284b9bc3f681d69d91d18b9e8d45a4dda486cd14f4145395786ac6a68a89", "question_id": "q05_multi_hop", "split": "test", "cluster_id": "syn_03015_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 8, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1922, "query_tokens": 594, "message_tokens": 503, "message_utf8_bytes": 1142, "cache_tensor_bytes": null, "serialized_payload_bytes": 1142, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 174213.77120900434, "hops": {"1": 22697.06304196734, "2": 22100.102666998282, "3": 21632.063457975164, "4": 21449.837208958343, "5": 21619.770125020295, "6": 21653.251958079636, "7": 21466.14995796699, "8": 21595.532792038284}, "receiver_prefill_ms": 377.46716698165983, "ttft_ms": 14.336541993543506, "decode_ms": 11.773166013881564}, "total_wall_ms": 174617.89400107227, "ttft_ms": 14.336541993543506, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 61, "facts_retained": 24, "retention_rate": 0.39344262295081966, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.4166666666666667, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1142}, "ledger_by_hop": {"1": {"facts_total": 61, "facts_retained": 24, "retention_rate": 0.39344262295081966, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.4166666666666667, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1161}, "2": {"facts_total": 61, "facts_retained": 24, "retention_rate": 0.39344262295081966, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.4166666666666667, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1142}, "3": {"facts_total": 61, "facts_retained": 24, "retention_rate": 0.39344262295081966, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.4166666666666667, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1142}, "4": {"facts_total": 61, "facts_retained": 24, "retention_rate": 0.39344262295081966, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.4166666666666667, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1142}, "5": {"facts_total": 61, "facts_retained": 24, "retention_rate": 0.39344262295081966, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.4166666666666667, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1142}, "6": {"facts_total": 61, "facts_retained": 24, "retention_rate": 0.39344262295081966, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.4166666666666667, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1142}, "7": {"facts_total": 61, "facts_retained": 24, "retention_rate": 0.39344262295081966, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.4166666666666667, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1142}, "8": {"facts_total": 61, "facts_retained": 24, "retention_rate": 0.39344262295081966, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.4166666666666667, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1142}}}, "reference_key_id": "keys/synthetic/syn_03015_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "6606d2a2280a507a57312c3b4cdff39a97c80f61816d5bc41d13980cbef297bd", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "7bcac47a68d09b68dd965c7f", "attempt_id": "8f066054faa8", "run_group_id": "E4_core:A4|closed|qwen3b:h8", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:51:02.666057Z", "task_family": "synthetic", "document_id": "syn_03015_2000", "document_hash": "71a7284b9bc3f681d69d91d18b9e8d45a4dda486cd14f4145395786ac6a68a89", "question_id": "q06_negation", "split": "test", "cluster_id": "syn_03015_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 8, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1922, "query_tokens": 586, "message_tokens": 503, "message_utf8_bytes": 1142, "cache_tensor_bytes": null, "serialized_payload_bytes": 1142, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 174213.77120900434, "hops": {"1": 22697.06304196734, "2": 22100.102666998282, "3": 21632.063457975164, "4": 21449.837208958343, "5": 21619.770125020295, "6": 21653.251958079636, "7": 21466.14995796699, "8": 21595.532792038284}, "receiver_prefill_ms": 392.80637504998595, "ttft_ms": 49.50054094661027, "decode_ms": 114.76454196963459}, "total_wall_ms": 174771.76570903976, "ttft_ms": 49.50054094661027, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 61, "facts_retained": 24, "retention_rate": 0.39344262295081966, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.4166666666666667, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1142}, "ledger_by_hop": {"1": {"facts_total": 61, "facts_retained": 24, "retention_rate": 0.39344262295081966, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.4166666666666667, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1161}, "2": {"facts_total": 61, "facts_retained": 24, "retention_rate": 0.39344262295081966, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.4166666666666667, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1142}, "3": {"facts_total": 61, "facts_retained": 24, "retention_rate": 0.39344262295081966, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.4166666666666667, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1142}, "4": {"facts_total": 61, "facts_retained": 24, "retention_rate": 0.39344262295081966, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.4166666666666667, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1142}, "5": {"facts_total": 61, "facts_retained": 24, "retention_rate": 0.39344262295081966, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.4166666666666667, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1142}, "6": {"facts_total": 61, "facts_retained": 24, "retention_rate": 0.39344262295081966, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.4166666666666667, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1142}, "7": {"facts_total": 61, "facts_retained": 24, "retention_rate": 0.39344262295081966, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.4166666666666667, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1142}, "8": {"facts_total": 61, "facts_retained": 24, "retention_rate": 0.39344262295081966, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.4166666666666667, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1142}}}, "reference_key_id": "keys/synthetic/syn_03015_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "6606d2a2280a507a57312c3b4cdff39a97c80f61816d5bc41d13980cbef297bd", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "64a865fa3b33b66c2f4e3cbf", "attempt_id": "56a832e5c25a", "run_group_id": "E4_core:A4|closed|qwen3b:h8", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:51:03.244621Z", "task_family": "synthetic", "document_id": "syn_03015_2000", "document_hash": "71a7284b9bc3f681d69d91d18b9e8d45a4dda486cd14f4145395786ac6a68a89", "question_id": "q07_negation", "split": "test", "cluster_id": "syn_03015_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 8, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1922, "query_tokens": 586, "message_tokens": 503, "message_utf8_bytes": 1142, "cache_tensor_bytes": null, "serialized_payload_bytes": 1142, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 174213.77120900434, "hops": {"1": 22697.06304196734, "2": 22100.102666998282, "3": 21632.063457975164, "4": 21449.837208958343, "5": 21619.770125020295, "6": 21653.251958079636, "7": 21466.14995796699, "8": 21595.532792038284}, "receiver_prefill_ms": 471.24108299613, "ttft_ms": 26.722333976067603, "decode_ms": 35.731707932427526}, "total_wall_ms": 174748.1662929058, "ttft_ms": 26.722333976067603, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 61, "facts_retained": 24, "retention_rate": 0.39344262295081966, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.4166666666666667, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1142}, "ledger_by_hop": {"1": {"facts_total": 61, "facts_retained": 24, "retention_rate": 0.39344262295081966, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.4166666666666667, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1161}, "2": {"facts_total": 61, "facts_retained": 24, "retention_rate": 0.39344262295081966, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.4166666666666667, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1142}, "3": {"facts_total": 61, "facts_retained": 24, "retention_rate": 0.39344262295081966, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.4166666666666667, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1142}, "4": {"facts_total": 61, "facts_retained": 24, "retention_rate": 0.39344262295081966, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.4166666666666667, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1142}, "5": {"facts_total": 61, "facts_retained": 24, "retention_rate": 0.39344262295081966, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.4166666666666667, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1142}, "6": {"facts_total": 61, "facts_retained": 24, "retention_rate": 0.39344262295081966, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.4166666666666667, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1142}, "7": {"facts_total": 61, "facts_retained": 24, "retention_rate": 0.39344262295081966, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.4166666666666667, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1142}, "8": {"facts_total": 61, "facts_retained": 24, "retention_rate": 0.39344262295081966, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.4166666666666667, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1142}}}, "reference_key_id": "keys/synthetic/syn_03015_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "6606d2a2280a507a57312c3b4cdff39a97c80f61816d5bc41d13980cbef297bd", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "7b1382a1051593efcb891015", "attempt_id": "3b8dbc033b1d", "run_group_id": "E4_core:A4|closed|qwen3b:h8", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:51:03.800168Z", "task_family": "synthetic", "document_id": "syn_03015_2000", "document_hash": "71a7284b9bc3f681d69d91d18b9e8d45a4dda486cd14f4145395786ac6a68a89", "question_id": "q08_revision", "split": "test", "cluster_id": "syn_03015_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 8, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1922, "query_tokens": 600, "message_tokens": 503, "message_utf8_bytes": 1142, "cache_tensor_bytes": null, "serialized_payload_bytes": 1142, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 174213.77120900434, "hops": {"1": 22697.06304196734, "2": 22100.102666998282, "3": 21632.063457975164, "4": 21449.837208958343, "5": 21619.770125020295, "6": 21653.251958079636, "7": 21466.14995796699, "8": 21595.532792038284}, "receiver_prefill_ms": 409.2559579294175, "ttft_ms": 20.75162506662309, "decode_ms": 37.619957933202386}, "total_wall_ms": 174682.0411250228, "ttft_ms": 20.75162506662309, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 61, "facts_retained": 24, "retention_rate": 0.39344262295081966, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.4166666666666667, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1142}, "ledger_by_hop": {"1": {"facts_total": 61, "facts_retained": 24, "retention_rate": 0.39344262295081966, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.4166666666666667, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1161}, "2": {"facts_total": 61, "facts_retained": 24, "retention_rate": 0.39344262295081966, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.4166666666666667, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1142}, "3": {"facts_total": 61, "facts_retained": 24, "retention_rate": 0.39344262295081966, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.4166666666666667, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1142}, "4": {"facts_total": 61, "facts_retained": 24, "retention_rate": 0.39344262295081966, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.4166666666666667, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1142}, "5": {"facts_total": 61, "facts_retained": 24, "retention_rate": 0.39344262295081966, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.4166666666666667, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1142}, "6": {"facts_total": 61, "facts_retained": 24, "retention_rate": 0.39344262295081966, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.4166666666666667, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1142}, "7": {"facts_total": 61, "facts_retained": 24, "retention_rate": 0.39344262295081966, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.4166666666666667, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1142}, "8": {"facts_total": 61, "facts_retained": 24, "retention_rate": 0.39344262295081966, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.4166666666666667, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1142}}}, "reference_key_id": "keys/synthetic/syn_03015_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "6606d2a2280a507a57312c3b4cdff39a97c80f61816d5bc41d13980cbef297bd", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "f1468a66973e79a723bb5ed0", "attempt_id": "df6e518dc77f", "run_group_id": "E4_core:A4|closed|qwen3b:h8", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:51:04.285988Z", "task_family": "synthetic", "document_id": "syn_03015_2000", "document_hash": "71a7284b9bc3f681d69d91d18b9e8d45a4dda486cd14f4145395786ac6a68a89", "question_id": "q09_revision", "split": "test", "cluster_id": "syn_03015_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 8, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1922, "query_tokens": 594, "message_tokens": 503, "message_utf8_bytes": 1142, "cache_tensor_bytes": null, "serialized_payload_bytes": 1142, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 174213.77120900434, "hops": {"1": 22697.06304196734, "2": 22100.102666998282, "3": 21632.063457975164, "4": 21449.837208958343, "5": 21619.770125020295, "6": 21653.251958079636, "7": 21466.14995796699, "8": 21595.532792038284}, "receiver_prefill_ms": 356.76533298101276, "ttft_ms": 41.06187494471669, "decode_ms": 557.6473750406876}, "total_wall_ms": 175169.95129198767, "ttft_ms": 41.06187494471669, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 9}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "8,377,480", "evidence_artifact": {"ledger": {"facts_total": 61, "facts_retained": 24, "retention_rate": 0.39344262295081966, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.4166666666666667, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1142}, "ledger_by_hop": {"1": {"facts_total": 61, "facts_retained": 24, "retention_rate": 0.39344262295081966, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.4166666666666667, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1161}, "2": {"facts_total": 61, "facts_retained": 24, "retention_rate": 0.39344262295081966, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.4166666666666667, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1142}, "3": {"facts_total": 61, "facts_retained": 24, "retention_rate": 0.39344262295081966, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.4166666666666667, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1142}, "4": {"facts_total": 61, "facts_retained": 24, "retention_rate": 0.39344262295081966, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.4166666666666667, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1142}, "5": {"facts_total": 61, "facts_retained": 24, "retention_rate": 0.39344262295081966, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.4166666666666667, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1142}, "6": {"facts_total": 61, "facts_retained": 24, "retention_rate": 0.39344262295081966, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.4166666666666667, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1142}, "7": {"facts_total": 61, "facts_retained": 24, "retention_rate": 0.39344262295081966, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.4166666666666667, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1142}, "8": {"facts_total": 61, "facts_retained": 24, "retention_rate": 0.39344262295081966, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.4166666666666667, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1142}}}, "reference_key_id": "keys/synthetic/syn_03015_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "6606d2a2280a507a57312c3b4cdff39a97c80f61816d5bc41d13980cbef297bd", "output_hash": "ec4c68b327a2d2427ed21821415a5f9ddf87134bd9f654bdd69c7d1301febcb0"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "cd6cf903051988c860fdea0a", "attempt_id": "a27772c1a0a4", "run_group_id": "E4_core:A4|closed|qwen3b:h8", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:51:05.258501Z", "task_family": "synthetic", "document_id": "syn_03015_2000", "document_hash": "71a7284b9bc3f681d69d91d18b9e8d45a4dda486cd14f4145395786ac6a68a89", "question_id": "q10_temporal", "split": "test", "cluster_id": "syn_03015_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 8, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1922, "query_tokens": 603, "message_tokens": 503, "message_utf8_bytes": 1142, "cache_tensor_bytes": null, "serialized_payload_bytes": 1142, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 174213.77120900434, "hops": {"1": 22697.06304196734, "2": 22100.102666998282, "3": 21632.063457975164, "4": 21449.837208958343, "5": 21619.770125020295, "6": 21653.251958079636, "7": 21466.14995796699, "8": 21595.532792038284}, "receiver_prefill_ms": 411.3556669326499, "ttft_ms": 22.62012497521937, "decode_ms": 38.9524589991197}, "total_wall_ms": 174687.31075001415, "ttft_ms": 22.62012497521937, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 61, "facts_retained": 24, "retention_rate": 0.39344262295081966, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.4166666666666667, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1142}, "ledger_by_hop": {"1": {"facts_total": 61, "facts_retained": 24, "retention_rate": 0.39344262295081966, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.4166666666666667, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1161}, "2": {"facts_total": 61, "facts_retained": 24, "retention_rate": 0.39344262295081966, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.4166666666666667, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1142}, "3": {"facts_total": 61, "facts_retained": 24, "retention_rate": 0.39344262295081966, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.4166666666666667, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1142}, "4": {"facts_total": 61, "facts_retained": 24, "retention_rate": 0.39344262295081966, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.4166666666666667, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1142}, "5": {"facts_total": 61, "facts_retained": 24, "retention_rate": 0.39344262295081966, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.4166666666666667, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1142}, "6": {"facts_total": 61, "facts_retained": 24, "retention_rate": 0.39344262295081966, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.4166666666666667, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1142}, "7": {"facts_total": 61, "facts_retained": 24, "retention_rate": 0.39344262295081966, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.4166666666666667, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1142}, "8": {"facts_total": 61, "facts_retained": 24, "retention_rate": 0.39344262295081966, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.4166666666666667, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1142}}}, "reference_key_id": "keys/synthetic/syn_03015_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "6606d2a2280a507a57312c3b4cdff39a97c80f61816d5bc41d13980cbef297bd", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "cf90c45d4cd6a93a47d8e08d", "attempt_id": "46d027a31f6b", "run_group_id": "E4_core:A4|closed|qwen3b:h8", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:51:05.750361Z", "task_family": "synthetic", "document_id": "syn_03015_2000", "document_hash": "71a7284b9bc3f681d69d91d18b9e8d45a4dda486cd14f4145395786ac6a68a89", "question_id": "q11_aggregation", "split": "test", "cluster_id": "syn_03015_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 8, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1922, "query_tokens": 596, "message_tokens": 503, "message_utf8_bytes": 1142, "cache_tensor_bytes": null, "serialized_payload_bytes": 1142, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 174213.77120900434, "hops": {"1": 22697.06304196734, "2": 22100.102666998282, "3": 21632.063457975164, "4": 21449.837208958343, "5": 21619.770125020295, "6": 21653.251958079636, "7": 21466.14995796699, "8": 21595.532792038284}, "receiver_prefill_ms": 380.47270802780986, "ttft_ms": 56.666792021133006, "decode_ms": 518.6663749627769}, "total_wall_ms": 175170.47500004992, "ttft_ms": 56.666792021133006, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 7}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "1806025", "evidence_artifact": {"ledger": {"facts_total": 61, "facts_retained": 24, "retention_rate": 0.39344262295081966, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.4166666666666667, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1142}, "ledger_by_hop": {"1": {"facts_total": 61, "facts_retained": 24, "retention_rate": 0.39344262295081966, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.4166666666666667, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1161}, "2": {"facts_total": 61, "facts_retained": 24, "retention_rate": 0.39344262295081966, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.4166666666666667, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1142}, "3": {"facts_total": 61, "facts_retained": 24, "retention_rate": 0.39344262295081966, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.4166666666666667, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1142}, "4": {"facts_total": 61, "facts_retained": 24, "retention_rate": 0.39344262295081966, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.4166666666666667, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1142}, "5": {"facts_total": 61, "facts_retained": 24, "retention_rate": 0.39344262295081966, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.4166666666666667, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1142}, "6": {"facts_total": 61, "facts_retained": 24, "retention_rate": 0.39344262295081966, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.4166666666666667, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1142}, "7": {"facts_total": 61, "facts_retained": 24, "retention_rate": 0.39344262295081966, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.4166666666666667, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1142}, "8": {"facts_total": 61, "facts_retained": 24, "retention_rate": 0.39344262295081966, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.4166666666666667, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1142}}}, "reference_key_id": "keys/synthetic/syn_03015_2000.json", "task_scores": {"correct": 0.0, "abstained": 0.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "6606d2a2280a507a57312c3b4cdff39a97c80f61816d5bc41d13980cbef297bd", "output_hash": "9801d1840792d286693349934c9cc13fead44d82d480a42803cf6015c01431f6"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "285abc80c56e6e4dc72f4a94", "attempt_id": "32177471de60", "run_group_id": "E4_core:A4|closed|qwen3b:h8", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:51:06.725691Z", "task_family": "synthetic", "document_id": "syn_03015_2000", "document_hash": "71a7284b9bc3f681d69d91d18b9e8d45a4dda486cd14f4145395786ac6a68a89", "question_id": "q12_unit_conversion", "split": "test", "cluster_id": "syn_03015_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 8, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1922, "query_tokens": 593, "message_tokens": 503, "message_utf8_bytes": 1142, "cache_tensor_bytes": null, "serialized_payload_bytes": 1142, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 174213.77120900434, "hops": {"1": 22697.06304196734, "2": 22100.102666998282, "3": 21632.063457975164, "4": 21449.837208958343, "5": 21619.770125020295, "6": 21653.251958079636, "7": 21466.14995796699, "8": 21595.532792038284}, "receiver_prefill_ms": 401.7225409625098, "ttft_ms": 23.636207915842533, "decode_ms": 39.062125026248395}, "total_wall_ms": 174678.8279589964, "ttft_ms": 23.636207915842533, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 61, "facts_retained": 24, "retention_rate": 0.39344262295081966, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.4166666666666667, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1142}, "ledger_by_hop": {"1": {"facts_total": 61, "facts_retained": 24, "retention_rate": 0.39344262295081966, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.4166666666666667, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1161}, "2": {"facts_total": 61, "facts_retained": 24, "retention_rate": 0.39344262295081966, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.4166666666666667, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1142}, "3": {"facts_total": 61, "facts_retained": 24, "retention_rate": 0.39344262295081966, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.4166666666666667, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1142}, "4": {"facts_total": 61, "facts_retained": 24, "retention_rate": 0.39344262295081966, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.4166666666666667, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1142}, "5": {"facts_total": 61, "facts_retained": 24, "retention_rate": 0.39344262295081966, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.4166666666666667, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1142}, "6": {"facts_total": 61, "facts_retained": 24, "retention_rate": 0.39344262295081966, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.4166666666666667, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1142}, "7": {"facts_total": 61, "facts_retained": 24, "retention_rate": 0.39344262295081966, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.4166666666666667, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1142}, "8": {"facts_total": 61, "facts_retained": 24, "retention_rate": 0.39344262295081966, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.4166666666666667, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1142}}}, "reference_key_id": "keys/synthetic/syn_03015_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "6606d2a2280a507a57312c3b4cdff39a97c80f61816d5bc41d13980cbef297bd", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "e9a4f048e951c094cd5ccf54", "attempt_id": "bcda738057fb", "run_group_id": "E4_core:A4|closed|qwen3b:h8", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:51:07.208269Z", "task_family": "synthetic", "document_id": "syn_03015_2000", "document_hash": "71a7284b9bc3f681d69d91d18b9e8d45a4dda486cd14f4145395786ac6a68a89", "question_id": "q13_unanswerable", "split": "test", "cluster_id": "syn_03015_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 8, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1922, "query_tokens": 597, "message_tokens": 503, "message_utf8_bytes": 1142, "cache_tensor_bytes": null, "serialized_payload_bytes": 1142, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 174213.77120900434, "hops": {"1": 22697.06304196734, "2": 22100.102666998282, "3": 21632.063457975164, "4": 21449.837208958343, "5": 21619.770125020295, "6": 21653.251958079636, "7": 21466.14995796699, "8": 21595.532792038284}, "receiver_prefill_ms": 408.3962499862537, "ttft_ms": 23.078458034433424, "decode_ms": 23.825958953239024}, "total_wall_ms": 174669.89483404905, "ttft_ms": 23.078458034433424, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 61, "facts_retained": 24, "retention_rate": 0.39344262295081966, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.4166666666666667, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1142}, "ledger_by_hop": {"1": {"facts_total": 61, "facts_retained": 24, "retention_rate": 0.39344262295081966, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.4166666666666667, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1161}, "2": {"facts_total": 61, "facts_retained": 24, "retention_rate": 0.39344262295081966, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.4166666666666667, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1142}, "3": {"facts_total": 61, "facts_retained": 24, "retention_rate": 0.39344262295081966, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.4166666666666667, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1142}, "4": {"facts_total": 61, "facts_retained": 24, "retention_rate": 0.39344262295081966, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.4166666666666667, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1142}, "5": {"facts_total": 61, "facts_retained": 24, "retention_rate": 0.39344262295081966, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.4166666666666667, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1142}, "6": {"facts_total": 61, "facts_retained": 24, "retention_rate": 0.39344262295081966, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.4166666666666667, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1142}, "7": {"facts_total": 61, "facts_retained": 24, "retention_rate": 0.39344262295081966, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.4166666666666667, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1142}, "8": {"facts_total": 61, "facts_retained": 24, "retention_rate": 0.39344262295081966, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.4166666666666667, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1142}}}, "reference_key_id": "keys/synthetic/syn_03015_2000.json", "task_scores": {"correct": 1.0, "abstained": 1.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "6606d2a2280a507a57312c3b4cdff39a97c80f61816d5bc41d13980cbef297bd", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "7a6171a5743f951a866fe226", "attempt_id": "007a509986d2", "run_group_id": "E4_core:A4|closed|qwen3b:h8", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:51:07.681963Z", "task_family": "synthetic", "document_id": "syn_03015_2000", "document_hash": "71a7284b9bc3f681d69d91d18b9e8d45a4dda486cd14f4145395786ac6a68a89", "question_id": "q14_unanswerable", "split": "test", "cluster_id": "syn_03015_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 8, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1922, "query_tokens": 598, "message_tokens": 503, "message_utf8_bytes": 1142, "cache_tensor_bytes": null, "serialized_payload_bytes": 1142, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 174213.77120900434, "hops": {"1": 22697.06304196734, "2": 22100.102666998282, "3": 21632.063457975164, "4": 21449.837208958343, "5": 21619.770125020295, "6": 21653.251958079636, "7": 21466.14995796699, "8": 21595.532792038284}, "receiver_prefill_ms": 370.33587496262044, "ttft_ms": 44.55408302601427, "decode_ms": 69.7497499641031}, "total_wall_ms": 174699.11625003442, "ttft_ms": 44.55408302601427, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 61, "facts_retained": 24, "retention_rate": 0.39344262295081966, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.4166666666666667, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1142}, "ledger_by_hop": {"1": {"facts_total": 61, "facts_retained": 24, "retention_rate": 0.39344262295081966, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.4166666666666667, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1161}, "2": {"facts_total": 61, "facts_retained": 24, "retention_rate": 0.39344262295081966, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.4166666666666667, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1142}, "3": {"facts_total": 61, "facts_retained": 24, "retention_rate": 0.39344262295081966, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.4166666666666667, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1142}, "4": {"facts_total": 61, "facts_retained": 24, "retention_rate": 0.39344262295081966, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.4166666666666667, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1142}, "5": {"facts_total": 61, "facts_retained": 24, "retention_rate": 0.39344262295081966, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.4166666666666667, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1142}, "6": {"facts_total": 61, "facts_retained": 24, "retention_rate": 0.39344262295081966, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.4166666666666667, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1142}, "7": {"facts_total": 61, "facts_retained": 24, "retention_rate": 0.39344262295081966, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.4166666666666667, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1142}, "8": {"facts_total": 61, "facts_retained": 24, "retention_rate": 0.39344262295081966, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.4166666666666667, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1142}}}, "reference_key_id": "keys/synthetic/syn_03015_2000.json", "task_scores": {"correct": 1.0, "abstained": 1.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "6606d2a2280a507a57312c3b4cdff39a97c80f61816d5bc41d13980cbef297bd", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "8d0288763f38db09622a181c", "attempt_id": "642ad0449fc0", "run_group_id": "E4_core:A5|closed|qwen3b:h1", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:51:31.065034Z", "task_family": "synthetic", "document_id": "syn_03015_2000", "document_hash": "71a7284b9bc3f681d69d91d18b9e8d45a4dda486cd14f4145395786ac6a68a89", "question_id": "q00_single_fact", "split": "test", "cluster_id": "syn_03015_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 1, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1922, "query_tokens": 411, "message_tokens": 328, "message_utf8_bytes": 754, "cache_tensor_bytes": null, "serialized_payload_bytes": 754, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 22878.284082980826, "hops": {"1": 22878.284082980826}, "receiver_prefill_ms": 320.1054579112679, "ttft_ms": 23.37095804978162, "decode_ms": 20.87358303833753}, "total_wall_ms": 23243.240415933542, "ttft_ms": 23.37095804978162, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 61, "facts_retained": 7, "retention_rate": 0.11475409836065574, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.3076923076923077, "status": 0.0, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 690}, "ledger_by_hop": {"1": {"facts_total": 61, "facts_retained": 7, "retention_rate": 0.11475409836065574, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.3076923076923077, "status": 0.0, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 690}}}, "reference_key_id": "keys/synthetic/syn_03015_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "e54cba3528ab78e09d5da03579b2a2035abb56934b1d0571c12acb32bf58b543", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "2b92ddbfe8d9cfac4fabba74", "attempt_id": "64293bd7b96a", "run_group_id": "E4_core:A5|closed|qwen3b:h1", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:51:31.448394Z", "task_family": "synthetic", "document_id": "syn_03015_2000", "document_hash": "71a7284b9bc3f681d69d91d18b9e8d45a4dda486cd14f4145395786ac6a68a89", "question_id": "q01_single_fact", "split": "test", "cluster_id": "syn_03015_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 1, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1922, "query_tokens": 416, "message_tokens": 328, "message_utf8_bytes": 754, "cache_tensor_bytes": null, "serialized_payload_bytes": 754, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 22878.284082980826, "hops": {"1": 22878.284082980826}, "receiver_prefill_ms": 280.99758399184793, "ttft_ms": 54.558000061661005, "decode_ms": 59.444457991048694}, "total_wall_ms": 23273.776874993928, "ttft_ms": 54.558000061661005, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 61, "facts_retained": 7, "retention_rate": 0.11475409836065574, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.3076923076923077, "status": 0.0, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 690}, "ledger_by_hop": {"1": {"facts_total": 61, "facts_retained": 7, "retention_rate": 0.11475409836065574, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.3076923076923077, "status": 0.0, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 690}}}, "reference_key_id": "keys/synthetic/syn_03015_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "e54cba3528ab78e09d5da03579b2a2035abb56934b1d0571c12acb32bf58b543", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "76832b224f4c26081740c8d6", "attempt_id": "dc4b10433a34", "run_group_id": "E4_core:A5|closed|qwen3b:h1", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:51:31.861683Z", "task_family": "synthetic", "document_id": "syn_03015_2000", "document_hash": "71a7284b9bc3f681d69d91d18b9e8d45a4dda486cd14f4145395786ac6a68a89", "question_id": "q02_exact_copy", "split": "test", "cluster_id": "syn_03015_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 1, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1922, "query_tokens": 415, "message_tokens": 328, "message_utf8_bytes": 754, "cache_tensor_bytes": null, "serialized_payload_bytes": 754, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 22878.284082980826, "hops": {"1": 22878.284082980826}, "receiver_prefill_ms": 422.89704096037894, "ttft_ms": 33.04404206573963, "decode_ms": 29.571499908342957}, "total_wall_ms": 23364.252249943092, "ttft_ms": 33.04404206573963, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 61, "facts_retained": 7, "retention_rate": 0.11475409836065574, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.3076923076923077, "status": 0.0, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 690}, "ledger_by_hop": {"1": {"facts_total": 61, "facts_retained": 7, "retention_rate": 0.11475409836065574, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.3076923076923077, "status": 0.0, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 690}}}, "reference_key_id": "keys/synthetic/syn_03015_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "e54cba3528ab78e09d5da03579b2a2035abb56934b1d0571c12acb32bf58b543", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "53f2a5c346dd066129504b97", "attempt_id": "2654d11df115", "run_group_id": "E4_core:A5|closed|qwen3b:h1", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:51:32.365313Z", "task_family": "synthetic", "document_id": "syn_03015_2000", "document_hash": "71a7284b9bc3f681d69d91d18b9e8d45a4dda486cd14f4145395786ac6a68a89", "question_id": "q03_exact_copy", "split": "test", "cluster_id": "syn_03015_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 1, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1922, "query_tokens": 411, "message_tokens": 328, "message_utf8_bytes": 754, "cache_tensor_bytes": null, "serialized_payload_bytes": 754, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 22878.284082980826, "hops": {"1": 22878.284082980826}, "receiver_prefill_ms": 300.9138749912381, "ttft_ms": 41.94533405825496, "decode_ms": 193.42320796567947}, "total_wall_ms": 23415.019873995334, "ttft_ms": 41.94533405825496, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 6}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "Wilhelmina Dubrovnik", "evidence_artifact": {"ledger": {"facts_total": 61, "facts_retained": 7, "retention_rate": 0.11475409836065574, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.3076923076923077, "status": 0.0, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 690}, "ledger_by_hop": {"1": {"facts_total": 61, "facts_retained": 7, "retention_rate": 0.11475409836065574, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.3076923076923077, "status": 0.0, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 690}}}, "reference_key_id": "keys/synthetic/syn_03015_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "e54cba3528ab78e09d5da03579b2a2035abb56934b1d0571c12acb32bf58b543", "output_hash": "f42f05ebfc96f7258cbfbbb37cd65e114d3d15f869e68e6c95b71fc288c30a32"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "9f6373aab554ab9c27330189", "attempt_id": "840efb29333c", "run_group_id": "E4_core:A5|closed|qwen3b:h1", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:51:32.919069Z", "task_family": "synthetic", "document_id": "syn_03015_2000", "document_hash": "71a7284b9bc3f681d69d91d18b9e8d45a4dda486cd14f4145395786ac6a68a89", "question_id": "q04_multi_hop", "split": "test", "cluster_id": "syn_03015_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 1, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1922, "query_tokens": 420, "message_tokens": 328, "message_utf8_bytes": 754, "cache_tensor_bytes": null, "serialized_payload_bytes": 754, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 22878.284082980826, "hops": {"1": 22878.284082980826}, "receiver_prefill_ms": 300.4982080310583, "ttft_ms": 42.55408304743469, "decode_ms": 511.09491696115583}, "total_wall_ms": 23733.231500023976, "ttft_ms": 42.55408304743469, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 8}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "8421725 EUR", "evidence_artifact": {"ledger": {"facts_total": 61, "facts_retained": 7, "retention_rate": 0.11475409836065574, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.3076923076923077, "status": 0.0, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 690}, "ledger_by_hop": {"1": {"facts_total": 61, "facts_retained": 7, "retention_rate": 0.11475409836065574, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.3076923076923077, "status": 0.0, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 690}}}, "reference_key_id": "keys/synthetic/syn_03015_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 0.6666666666666666}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "e54cba3528ab78e09d5da03579b2a2035abb56934b1d0571c12acb32bf58b543", "output_hash": "976c8be2c1a175a4fb5643c5785f93e27ad9205cebd9569a64b821e1ae51ac0e"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "aa1ff4334e391f3911ab5b2d", "attempt_id": "e7d92a7bfaee", "run_group_id": "E4_core:A5|closed|qwen3b:h1", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:51:33.791403Z", "task_family": "synthetic", "document_id": "syn_03015_2000", "document_hash": "71a7284b9bc3f681d69d91d18b9e8d45a4dda486cd14f4145395786ac6a68a89", "question_id": "q05_multi_hop", "split": "test", "cluster_id": "syn_03015_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 1, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1922, "query_tokens": 420, "message_tokens": 328, "message_utf8_bytes": 754, "cache_tensor_bytes": null, "serialized_payload_bytes": 754, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 22878.284082980826, "hops": {"1": 22878.284082980826}, "receiver_prefill_ms": 304.13570802193135, "ttft_ms": 42.69349994137883, "decode_ms": 24.28574999794364}, "total_wall_ms": 23249.917666078545, "ttft_ms": 42.69349994137883, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 61, "facts_retained": 7, "retention_rate": 0.11475409836065574, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.3076923076923077, "status": 0.0, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 690}, "ledger_by_hop": {"1": {"facts_total": 61, "facts_retained": 7, "retention_rate": 0.11475409836065574, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.3076923076923077, "status": 0.0, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 690}}}, "reference_key_id": "keys/synthetic/syn_03015_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "e54cba3528ab78e09d5da03579b2a2035abb56934b1d0571c12acb32bf58b543", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "402dd3b0dece1aa247df24a9", "attempt_id": "a1bcf6944c24", "run_group_id": "E4_core:A5|closed|qwen3b:h1", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:51:34.180244Z", "task_family": "synthetic", "document_id": "syn_03015_2000", "document_hash": "71a7284b9bc3f681d69d91d18b9e8d45a4dda486cd14f4145395786ac6a68a89", "question_id": "q06_negation", "split": "test", "cluster_id": "syn_03015_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 1, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1922, "query_tokens": 412, "message_tokens": 328, "message_utf8_bytes": 754, "cache_tensor_bytes": null, "serialized_payload_bytes": 754, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 22878.284082980826, "hops": {"1": 22878.284082980826}, "receiver_prefill_ms": 296.3689579628408, "ttft_ms": 26.560583035461605, "decode_ms": 37.127333926036954}, "total_wall_ms": 23238.84124995675, "ttft_ms": 26.560583035461605, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 61, "facts_retained": 7, "retention_rate": 0.11475409836065574, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.3076923076923077, "status": 0.0, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 690}, "ledger_by_hop": {"1": {"facts_total": 61, "facts_retained": 7, "retention_rate": 0.11475409836065574, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.3076923076923077, "status": 0.0, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 690}}}, "reference_key_id": "keys/synthetic/syn_03015_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "e54cba3528ab78e09d5da03579b2a2035abb56934b1d0571c12acb32bf58b543", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "1d77bf11a0811bcbe146a31f", "attempt_id": "da82a1f4823e", "run_group_id": "E4_core:A5|closed|qwen3b:h1", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:51:34.557878Z", "task_family": "synthetic", "document_id": "syn_03015_2000", "document_hash": "71a7284b9bc3f681d69d91d18b9e8d45a4dda486cd14f4145395786ac6a68a89", "question_id": "q07_negation", "split": "test", "cluster_id": "syn_03015_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 1, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1922, "query_tokens": 412, "message_tokens": 328, "message_utf8_bytes": 754, "cache_tensor_bytes": null, "serialized_payload_bytes": 754, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 22878.284082980826, "hops": {"1": 22878.284082980826}, "receiver_prefill_ms": 265.4342500027269, "ttft_ms": 43.94604102708399, "decode_ms": 50.164332962594926}, "total_wall_ms": 23238.370207953267, "ttft_ms": 43.94604102708399, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 61, "facts_retained": 7, "retention_rate": 0.11475409836065574, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.3076923076923077, "status": 0.0, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 690}, "ledger_by_hop": {"1": {"facts_total": 61, "facts_retained": 7, "retention_rate": 0.11475409836065574, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.3076923076923077, "status": 0.0, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 690}}}, "reference_key_id": "keys/synthetic/syn_03015_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "e54cba3528ab78e09d5da03579b2a2035abb56934b1d0571c12acb32bf58b543", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "a0451a9d1b9a1090b8110214", "attempt_id": "fc4dc0e92ae6", "run_group_id": "E4_core:A5|closed|qwen3b:h1", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:51:34.935841Z", "task_family": "synthetic", "document_id": "syn_03015_2000", "document_hash": "71a7284b9bc3f681d69d91d18b9e8d45a4dda486cd14f4145395786ac6a68a89", "question_id": "q08_revision", "split": "test", "cluster_id": "syn_03015_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 1, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1922, "query_tokens": 426, "message_tokens": 328, "message_utf8_bytes": 754, "cache_tensor_bytes": null, "serialized_payload_bytes": 754, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 22878.284082980826, "hops": {"1": 22878.284082980826}, "receiver_prefill_ms": 443.9979159506038, "ttft_ms": 39.299458963796496, "decode_ms": 22.733834106475115}, "total_wall_ms": 23384.928832994774, "ttft_ms": 39.299458963796496, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 61, "facts_retained": 7, "retention_rate": 0.11475409836065574, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.3076923076923077, "status": 0.0, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 690}, "ledger_by_hop": {"1": {"facts_total": 61, "facts_retained": 7, "retention_rate": 0.11475409836065574, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.3076923076923077, "status": 0.0, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 690}}}, "reference_key_id": "keys/synthetic/syn_03015_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "e54cba3528ab78e09d5da03579b2a2035abb56934b1d0571c12acb32bf58b543", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "897a6cb9ed158da89549fabc", "attempt_id": "863fa2e5a4d9", "run_group_id": "E4_core:A5|closed|qwen3b:h1", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:51:35.461312Z", "task_family": "synthetic", "document_id": "syn_03015_2000", "document_hash": "71a7284b9bc3f681d69d91d18b9e8d45a4dda486cd14f4145395786ac6a68a89", "question_id": "q09_revision", "split": "test", "cluster_id": "syn_03015_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 1, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1922, "query_tokens": 420, "message_tokens": 328, "message_utf8_bytes": 754, "cache_tensor_bytes": null, "serialized_payload_bytes": 754, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 22878.284082980826, "hops": {"1": 22878.284082980826}, "receiver_prefill_ms": 298.05191699415445, "ttft_ms": 38.167124963365495, "decode_ms": 220.27450008317828}, "total_wall_ms": 23435.34224992618, "ttft_ms": 38.167124963365495, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 7}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "7028925", "evidence_artifact": {"ledger": {"facts_total": 61, "facts_retained": 7, "retention_rate": 0.11475409836065574, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.3076923076923077, "status": 0.0, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 690}, "ledger_by_hop": {"1": {"facts_total": 61, "facts_retained": 7, "retention_rate": 0.11475409836065574, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.3076923076923077, "status": 0.0, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 690}}}, "reference_key_id": "keys/synthetic/syn_03015_2000.json", "task_scores": {"correct": 0.0, "abstained": 0.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "e54cba3528ab78e09d5da03579b2a2035abb56934b1d0571c12acb32bf58b543", "output_hash": "e966a398855c104a86bce07742c3ce8108a432926bc704940ecfd07e44596774"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "2d844740927d8e351d79979b", "attempt_id": "b2a4bdaf8f67", "run_group_id": "E4_core:A5|closed|qwen3b:h1", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:51:36.035834Z", "task_family": "synthetic", "document_id": "syn_03015_2000", "document_hash": "71a7284b9bc3f681d69d91d18b9e8d45a4dda486cd14f4145395786ac6a68a89", "question_id": "q10_temporal", "split": "test", "cluster_id": "syn_03015_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 1, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1922, "query_tokens": 429, "message_tokens": 328, "message_utf8_bytes": 754, "cache_tensor_bytes": null, "serialized_payload_bytes": 754, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 22878.284082980826, "hops": {"1": 22878.284082980826}, "receiver_prefill_ms": 255.0347080687061, "ttft_ms": 38.17708289716393, "decode_ms": 64.96145809069276}, "total_wall_ms": 23237.013290985487, "ttft_ms": 38.17708289716393, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 61, "facts_retained": 7, "retention_rate": 0.11475409836065574, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.3076923076923077, "status": 0.0, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 690}, "ledger_by_hop": {"1": {"facts_total": 61, "facts_retained": 7, "retention_rate": 0.11475409836065574, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.3076923076923077, "status": 0.0, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 690}}}, "reference_key_id": "keys/synthetic/syn_03015_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "e54cba3528ab78e09d5da03579b2a2035abb56934b1d0571c12acb32bf58b543", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "2012620f8b953edb7ea704e4", "attempt_id": "ae1d85e51bd9", "run_group_id": "E4_core:A5|closed|qwen3b:h1", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:51:36.411391Z", "task_family": "synthetic", "document_id": "syn_03015_2000", "document_hash": "71a7284b9bc3f681d69d91d18b9e8d45a4dda486cd14f4145395786ac6a68a89", "question_id": "q11_aggregation", "split": "test", "cluster_id": "syn_03015_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 1, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1922, "query_tokens": 422, "message_tokens": 328, "message_utf8_bytes": 754, "cache_tensor_bytes": null, "serialized_payload_bytes": 754, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 22878.284082980826, "hops": {"1": 22878.284082980826}, "receiver_prefill_ms": 367.0918750576675, "ttft_ms": 130.12195797637105, "decode_ms": 39.30045792367309}, "total_wall_ms": 23415.568624972366, "ttft_ms": 130.12195797637105, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 61, "facts_retained": 7, "retention_rate": 0.11475409836065574, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.3076923076923077, "status": 0.0, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 690}, "ledger_by_hop": {"1": {"facts_total": 61, "facts_retained": 7, "retention_rate": 0.11475409836065574, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.3076923076923077, "status": 0.0, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 690}}}, "reference_key_id": "keys/synthetic/syn_03015_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "e54cba3528ab78e09d5da03579b2a2035abb56934b1d0571c12acb32bf58b543", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "e48155abea66f50c1cb6785e", "attempt_id": "23f61b56bcc4", "run_group_id": "E4_core:A5|closed|qwen3b:h1", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:51:36.967090Z", "task_family": "synthetic", "document_id": "syn_03015_2000", "document_hash": "71a7284b9bc3f681d69d91d18b9e8d45a4dda486cd14f4145395786ac6a68a89", "question_id": "q12_unit_conversion", "split": "test", "cluster_id": "syn_03015_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 1, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1922, "query_tokens": 419, "message_tokens": 328, "message_utf8_bytes": 754, "cache_tensor_bytes": null, "serialized_payload_bytes": 754, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 22878.284082980826, "hops": {"1": 22878.284082980826}, "receiver_prefill_ms": 308.65999998059124, "ttft_ms": 42.56145900581032, "decode_ms": 25.440166937187314}, "total_wall_ms": 23255.392499035224, "ttft_ms": 42.56145900581032, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 61, "facts_retained": 7, "retention_rate": 0.11475409836065574, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.3076923076923077, "status": 0.0, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 690}, "ledger_by_hop": {"1": {"facts_total": 61, "facts_retained": 7, "retention_rate": 0.11475409836065574, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.3076923076923077, "status": 0.0, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 690}}}, "reference_key_id": "keys/synthetic/syn_03015_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "e54cba3528ab78e09d5da03579b2a2035abb56934b1d0571c12acb32bf58b543", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "20777c9786f216723e1c2f11", "attempt_id": "892ca28eec8f", "run_group_id": "E4_core:A5|closed|qwen3b:h1", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:51:37.361224Z", "task_family": "synthetic", "document_id": "syn_03015_2000", "document_hash": "71a7284b9bc3f681d69d91d18b9e8d45a4dda486cd14f4145395786ac6a68a89", "question_id": "q13_unanswerable", "split": "test", "cluster_id": "syn_03015_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 1, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1922, "query_tokens": 423, "message_tokens": 328, "message_utf8_bytes": 754, "cache_tensor_bytes": null, "serialized_payload_bytes": 754, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 22878.284082980826, "hops": {"1": 22878.284082980826}, "receiver_prefill_ms": 297.8863330790773, "ttft_ms": 32.94112498406321, "decode_ms": 30.796042061410844}, "total_wall_ms": 23240.338165895082, "ttft_ms": 32.94112498406321, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 61, "facts_retained": 7, "retention_rate": 0.11475409836065574, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.3076923076923077, "status": 0.0, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 690}, "ledger_by_hop": {"1": {"facts_total": 61, "facts_retained": 7, "retention_rate": 0.11475409836065574, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.3076923076923077, "status": 0.0, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 690}}}, "reference_key_id": "keys/synthetic/syn_03015_2000.json", "task_scores": {"correct": 1.0, "abstained": 1.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "e54cba3528ab78e09d5da03579b2a2035abb56934b1d0571c12acb32bf58b543", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "5c23f2742a859f9f3fc44dcb", "attempt_id": "c72d429063d7", "run_group_id": "E4_core:A5|closed|qwen3b:h1", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:51:37.740104Z", "task_family": "synthetic", "document_id": "syn_03015_2000", "document_hash": "71a7284b9bc3f681d69d91d18b9e8d45a4dda486cd14f4145395786ac6a68a89", "question_id": "q14_unanswerable", "split": "test", "cluster_id": "syn_03015_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 1, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1922, "query_tokens": 424, "message_tokens": 328, "message_utf8_bytes": 754, "cache_tensor_bytes": null, "serialized_payload_bytes": 754, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 22878.284082980826, "hops": {"1": 22878.284082980826}, "receiver_prefill_ms": 251.0571670718491, "ttft_ms": 53.49470803048462, "decode_ms": 56.82870896998793}, "total_wall_ms": 23240.244623972103, "ttft_ms": 53.49470803048462, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 61, "facts_retained": 7, "retention_rate": 0.11475409836065574, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.3076923076923077, "status": 0.0, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 690}, "ledger_by_hop": {"1": {"facts_total": 61, "facts_retained": 7, "retention_rate": 0.11475409836065574, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.3076923076923077, "status": 0.0, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 690}}}, "reference_key_id": "keys/synthetic/syn_03015_2000.json", "task_scores": {"correct": 1.0, "abstained": 1.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "e54cba3528ab78e09d5da03579b2a2035abb56934b1d0571c12acb32bf58b543", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "1567b1f4f5b57113d61207e4", "attempt_id": "65b0195258c8", "run_group_id": "E4_core:A5|closed|qwen3b:h2", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:51:52.845983Z", "task_family": "synthetic", "document_id": "syn_03015_2000", "document_hash": "71a7284b9bc3f681d69d91d18b9e8d45a4dda486cd14f4145395786ac6a68a89", "question_id": "q00_single_fact", "split": "test", "cluster_id": "syn_03015_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 2, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1922, "query_tokens": 411, "message_tokens": 328, "message_utf8_bytes": 754, "cache_tensor_bytes": null, "serialized_payload_bytes": 754, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 37603.902374976315, "hops": {"1": 22878.284082980826, "2": 14725.618291995488}, "receiver_prefill_ms": 291.96583305019885, "ttft_ms": 41.65945900604129, "decode_ms": 47.88295808248222}, "total_wall_ms": 37985.943499952555, "ttft_ms": 41.65945900604129, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 61, "facts_retained": 7, "retention_rate": 0.11475409836065574, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.3076923076923077, "status": 0.0, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 690}, "ledger_by_hop": {"1": {"facts_total": 61, "facts_retained": 7, "retention_rate": 0.11475409836065574, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.3076923076923077, "status": 0.0, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 690}, "2": {"facts_total": 61, "facts_retained": 7, "retention_rate": 0.11475409836065574, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.3076923076923077, "status": 0.0, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 690}}}, "reference_key_id": "keys/synthetic/syn_03015_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "e54cba3528ab78e09d5da03579b2a2035abb56934b1d0571c12acb32bf58b543", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "5a81d3b2ec12121acd8c01a1", "attempt_id": "8d5f38901b56", "run_group_id": "E4_core:A5|closed|qwen3b:h2", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:51:53.246711Z", "task_family": "synthetic", "document_id": "syn_03015_2000", "document_hash": "71a7284b9bc3f681d69d91d18b9e8d45a4dda486cd14f4145395786ac6a68a89", "question_id": "q01_single_fact", "split": "test", "cluster_id": "syn_03015_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 2, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1922, "query_tokens": 416, "message_tokens": 328, "message_utf8_bytes": 754, "cache_tensor_bytes": null, "serialized_payload_bytes": 754, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 37603.902374976315, "hops": {"1": 22878.284082980826, "2": 14725.618291995488}, "receiver_prefill_ms": 442.3972920048982, "ttft_ms": 31.342500005848706, "decode_ms": 43.31750003620982}, "total_wall_ms": 38121.463541989215, "ttft_ms": 31.342500005848706, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 61, "facts_retained": 7, "retention_rate": 0.11475409836065574, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.3076923076923077, "status": 0.0, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 690}, "ledger_by_hop": {"1": {"facts_total": 61, "facts_retained": 7, "retention_rate": 0.11475409836065574, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.3076923076923077, "status": 0.0, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 690}, "2": {"facts_total": 61, "facts_retained": 7, "retention_rate": 0.11475409836065574, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.3076923076923077, "status": 0.0, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 690}}}, "reference_key_id": "keys/synthetic/syn_03015_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "e54cba3528ab78e09d5da03579b2a2035abb56934b1d0571c12acb32bf58b543", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "3fc3bbe164341872d37a4c74", "attempt_id": "0dd4f33f10b1", "run_group_id": "E4_core:A5|closed|qwen3b:h2", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:51:53.783065Z", "task_family": "synthetic", "document_id": "syn_03015_2000", "document_hash": "71a7284b9bc3f681d69d91d18b9e8d45a4dda486cd14f4145395786ac6a68a89", "question_id": "q02_exact_copy", "split": "test", "cluster_id": "syn_03015_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 2, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1922, "query_tokens": 415, "message_tokens": 328, "message_utf8_bytes": 754, "cache_tensor_bytes": null, "serialized_payload_bytes": 754, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 37603.902374976315, "hops": {"1": 22878.284082980826, "2": 14725.618291995488}, "receiver_prefill_ms": 304.15300000458956, "ttft_ms": 21.812666906043887, "decode_ms": 48.05416602175683}, "total_wall_ms": 37978.39170903899, "ttft_ms": 21.812666906043887, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 61, "facts_retained": 7, "retention_rate": 0.11475409836065574, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.3076923076923077, "status": 0.0, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 690}, "ledger_by_hop": {"1": {"facts_total": 61, "facts_retained": 7, "retention_rate": 0.11475409836065574, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.3076923076923077, "status": 0.0, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 690}, "2": {"facts_total": 61, "facts_retained": 7, "retention_rate": 0.11475409836065574, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.3076923076923077, "status": 0.0, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 690}}}, "reference_key_id": "keys/synthetic/syn_03015_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "e54cba3528ab78e09d5da03579b2a2035abb56934b1d0571c12acb32bf58b543", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "2497157861c64a2fb1e6ae63", "attempt_id": "7645d6b7ee45", "run_group_id": "E4_core:A5|closed|qwen3b:h2", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:51:54.175631Z", "task_family": "synthetic", "document_id": "syn_03015_2000", "document_hash": "71a7284b9bc3f681d69d91d18b9e8d45a4dda486cd14f4145395786ac6a68a89", "question_id": "q03_exact_copy", "split": "test", "cluster_id": "syn_03015_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 2, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1922, "query_tokens": 411, "message_tokens": 328, "message_utf8_bytes": 754, "cache_tensor_bytes": null, "serialized_payload_bytes": 754, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 37603.902374976315, "hops": {"1": 22878.284082980826, "2": 14725.618291995488}, "receiver_prefill_ms": 303.18870896007866, "ttft_ms": 24.280040990561247, "decode_ms": 252.69679096527398}, "total_wall_ms": 38184.76254190318, "ttft_ms": 24.280040990561247, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 6}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "Wilhelmina Dubrovnik", "evidence_artifact": {"ledger": {"facts_total": 61, "facts_retained": 7, "retention_rate": 0.11475409836065574, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.3076923076923077, "status": 0.0, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 690}, "ledger_by_hop": {"1": {"facts_total": 61, "facts_retained": 7, "retention_rate": 0.11475409836065574, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.3076923076923077, "status": 0.0, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 690}, "2": {"facts_total": 61, "facts_retained": 7, "retention_rate": 0.11475409836065574, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.3076923076923077, "status": 0.0, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 690}}}, "reference_key_id": "keys/synthetic/syn_03015_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "e54cba3528ab78e09d5da03579b2a2035abb56934b1d0571c12acb32bf58b543", "output_hash": "f42f05ebfc96f7258cbfbbb37cd65e114d3d15f869e68e6c95b71fc288c30a32"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "9fa4181eb91631d3cf537f98", "attempt_id": "40c6bb75b0b7", "run_group_id": "E4_core:A5|closed|qwen3b:h2", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:51:54.774506Z", "task_family": "synthetic", "document_id": "syn_03015_2000", "document_hash": "71a7284b9bc3f681d69d91d18b9e8d45a4dda486cd14f4145395786ac6a68a89", "question_id": "q04_multi_hop", "split": "test", "cluster_id": "syn_03015_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 2, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1922, "query_tokens": 420, "message_tokens": 328, "message_utf8_bytes": 754, "cache_tensor_bytes": null, "serialized_payload_bytes": 754, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 37603.902374976315, "hops": {"1": 22878.284082980826, "2": 14725.618291995488}, "receiver_prefill_ms": 375.5325829843059, "ttft_ms": 70.74683299288154, "decode_ms": 254.78558300528675}, "total_wall_ms": 38305.625749984756, "ttft_ms": 70.74683299288154, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 8}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "8421725 EUR", "evidence_artifact": {"ledger": {"facts_total": 61, "facts_retained": 7, "retention_rate": 0.11475409836065574, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.3076923076923077, "status": 0.0, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 690}, "ledger_by_hop": {"1": {"facts_total": 61, "facts_retained": 7, "retention_rate": 0.11475409836065574, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.3076923076923077, "status": 0.0, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 690}, "2": {"facts_total": 61, "facts_retained": 7, "retention_rate": 0.11475409836065574, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.3076923076923077, "status": 0.0, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 690}}}, "reference_key_id": "keys/synthetic/syn_03015_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 0.6666666666666666}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "e54cba3528ab78e09d5da03579b2a2035abb56934b1d0571c12acb32bf58b543", "output_hash": "976c8be2c1a175a4fb5643c5785f93e27ad9205cebd9569a64b821e1ae51ac0e"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "ca397c2cdb1fdd3a0069ab42", "attempt_id": "bbe68dbb8f1d", "run_group_id": "E4_core:A5|closed|qwen3b:h2", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:51:55.494603Z", "task_family": "synthetic", "document_id": "syn_03015_2000", "document_hash": "71a7284b9bc3f681d69d91d18b9e8d45a4dda486cd14f4145395786ac6a68a89", "question_id": "q05_multi_hop", "split": "test", "cluster_id": "syn_03015_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 2, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1922, "query_tokens": 420, "message_tokens": 328, "message_utf8_bytes": 754, "cache_tensor_bytes": null, "serialized_payload_bytes": 754, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 37603.902374976315, "hops": {"1": 22878.284082980826, "2": 14725.618291995488}, "receiver_prefill_ms": 310.6541249435395, "ttft_ms": 41.10979102551937, "decode_ms": 24.665000033564866}, "total_wall_ms": 37980.882291914895, "ttft_ms": 41.10979102551937, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 61, "facts_retained": 7, "retention_rate": 0.11475409836065574, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.3076923076923077, "status": 0.0, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 690}, "ledger_by_hop": {"1": {"facts_total": 61, "facts_retained": 7, "retention_rate": 0.11475409836065574, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.3076923076923077, "status": 0.0, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 690}, "2": {"facts_total": 61, "facts_retained": 7, "retention_rate": 0.11475409836065574, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.3076923076923077, "status": 0.0, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 690}}}, "reference_key_id": "keys/synthetic/syn_03015_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "e54cba3528ab78e09d5da03579b2a2035abb56934b1d0571c12acb32bf58b543", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "82f4fc9f34e7f20711b4864a", "attempt_id": "ff08569780da", "run_group_id": "E4_core:A5|closed|qwen3b:h2", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:51:55.890389Z", "task_family": "synthetic", "document_id": "syn_03015_2000", "document_hash": "71a7284b9bc3f681d69d91d18b9e8d45a4dda486cd14f4145395786ac6a68a89", "question_id": "q06_negation", "split": "test", "cluster_id": "syn_03015_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 2, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1922, "query_tokens": 412, "message_tokens": 328, "message_utf8_bytes": 754, "cache_tensor_bytes": null, "serialized_payload_bytes": 754, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 37603.902374976315, "hops": {"1": 22878.284082980826, "2": 14725.618291995488}, "receiver_prefill_ms": 262.4810410197824, "ttft_ms": 42.271042009815574, "decode_ms": 55.38783292286098}, "total_wall_ms": 37964.74520897027, "ttft_ms": 42.271042009815574, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 61, "facts_retained": 7, "retention_rate": 0.11475409836065574, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.3076923076923077, "status": 0.0, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 690}, "ledger_by_hop": {"1": {"facts_total": 61, "facts_retained": 7, "retention_rate": 0.11475409836065574, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.3076923076923077, "status": 0.0, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 690}, "2": {"facts_total": 61, "facts_retained": 7, "retention_rate": 0.11475409836065574, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.3076923076923077, "status": 0.0, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 690}}}, "reference_key_id": "keys/synthetic/syn_03015_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "e54cba3528ab78e09d5da03579b2a2035abb56934b1d0571c12acb32bf58b543", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "9144cfb960e5ac6238f090cd", "attempt_id": "c091bdfd8b36", "run_group_id": "E4_core:A5|closed|qwen3b:h2", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:51:56.268681Z", "task_family": "synthetic", "document_id": "syn_03015_2000", "document_hash": "71a7284b9bc3f681d69d91d18b9e8d45a4dda486cd14f4145395786ac6a68a89", "question_id": "q07_negation", "split": "test", "cluster_id": "syn_03015_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 2, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1922, "query_tokens": 412, "message_tokens": 328, "message_utf8_bytes": 754, "cache_tensor_bytes": null, "serialized_payload_bytes": 754, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 37603.902374976315, "hops": {"1": 22878.284082980826, "2": 14725.618291995488}, "receiver_prefill_ms": 283.65300002042204, "ttft_ms": 90.98587499465793, "decode_ms": 137.05479190684855}, "total_wall_ms": 38116.12779204734, "ttft_ms": 90.98587499465793, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 61, "facts_retained": 7, "retention_rate": 0.11475409836065574, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.3076923076923077, "status": 0.0, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 690}, "ledger_by_hop": {"1": {"facts_total": 61, "facts_retained": 7, "retention_rate": 0.11475409836065574, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.3076923076923077, "status": 0.0, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 690}, "2": {"facts_total": 61, "facts_retained": 7, "retention_rate": 0.11475409836065574, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.3076923076923077, "status": 0.0, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 690}}}, "reference_key_id": "keys/synthetic/syn_03015_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "e54cba3528ab78e09d5da03579b2a2035abb56934b1d0571c12acb32bf58b543", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "c025f0fccf8454da225fc21a", "attempt_id": "bda90dc551d9", "run_group_id": "E4_core:A5|closed|qwen3b:h2", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:51:56.798848Z", "task_family": "synthetic", "document_id": "syn_03015_2000", "document_hash": "71a7284b9bc3f681d69d91d18b9e8d45a4dda486cd14f4145395786ac6a68a89", "question_id": "q08_revision", "split": "test", "cluster_id": "syn_03015_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 2, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1922, "query_tokens": 426, "message_tokens": 328, "message_utf8_bytes": 754, "cache_tensor_bytes": null, "serialized_payload_bytes": 754, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 37603.902374976315, "hops": {"1": 22878.284082980826, "2": 14725.618291995488}, "receiver_prefill_ms": 288.52349997032434, "ttft_ms": 21.774582914076746, "decode_ms": 36.28895804286003}, "total_wall_ms": 37950.96208399627, "ttft_ms": 21.774582914076746, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 61, "facts_retained": 7, "retention_rate": 0.11475409836065574, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.3076923076923077, "status": 0.0, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 690}, "ledger_by_hop": {"1": {"facts_total": 61, "facts_retained": 7, "retention_rate": 0.11475409836065574, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.3076923076923077, "status": 0.0, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 690}, "2": {"facts_total": 61, "facts_retained": 7, "retention_rate": 0.11475409836065574, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.3076923076923077, "status": 0.0, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 690}}}, "reference_key_id": "keys/synthetic/syn_03015_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "e54cba3528ab78e09d5da03579b2a2035abb56934b1d0571c12acb32bf58b543", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "ff79530aca173dbc7e238543", "attempt_id": "16ba3bea3c77", "run_group_id": "E4_core:A5|closed|qwen3b:h2", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:51:57.163677Z", "task_family": "synthetic", "document_id": "syn_03015_2000", "document_hash": "71a7284b9bc3f681d69d91d18b9e8d45a4dda486cd14f4145395786ac6a68a89", "question_id": "q09_revision", "split": "test", "cluster_id": "syn_03015_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 2, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1922, "query_tokens": 420, "message_tokens": 328, "message_utf8_bytes": 754, "cache_tensor_bytes": null, "serialized_payload_bytes": 754, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 37603.902374976315, "hops": {"1": 22878.284082980826, "2": 14725.618291995488}, "receiver_prefill_ms": 309.5505000092089, "ttft_ms": 38.06954110041261, "decode_ms": 190.12516702059656}, "total_wall_ms": 38142.14033295866, "ttft_ms": 38.06954110041261, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 7}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "7028925", "evidence_artifact": {"ledger": {"facts_total": 61, "facts_retained": 7, "retention_rate": 0.11475409836065574, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.3076923076923077, "status": 0.0, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 690}, "ledger_by_hop": {"1": {"facts_total": 61, "facts_retained": 7, "retention_rate": 0.11475409836065574, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.3076923076923077, "status": 0.0, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 690}, "2": {"facts_total": 61, "facts_retained": 7, "retention_rate": 0.11475409836065574, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.3076923076923077, "status": 0.0, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 690}}}, "reference_key_id": "keys/synthetic/syn_03015_2000.json", "task_scores": {"correct": 0.0, "abstained": 0.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "e54cba3528ab78e09d5da03579b2a2035abb56934b1d0571c12acb32bf58b543", "output_hash": "e966a398855c104a86bce07742c3ce8108a432926bc704940ecfd07e44596774"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "df8530f193f4a1c9ff52a86f", "attempt_id": "e1b4cef6638d", "run_group_id": "E4_core:A5|closed|qwen3b:h2", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:51:57.718868Z", "task_family": "synthetic", "document_id": "syn_03015_2000", "document_hash": "71a7284b9bc3f681d69d91d18b9e8d45a4dda486cd14f4145395786ac6a68a89", "question_id": "q10_temporal", "split": "test", "cluster_id": "syn_03015_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 2, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1922, "query_tokens": 429, "message_tokens": 328, "message_utf8_bytes": 754, "cache_tensor_bytes": null, "serialized_payload_bytes": 754, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 37603.902374976315, "hops": {"1": 22878.284082980826, "2": 14725.618291995488}, "receiver_prefill_ms": 288.40008401311934, "ttft_ms": 44.273208011873066, "decode_ms": 61.93091697059572}, "total_wall_ms": 37999.13537490647, "ttft_ms": 44.273208011873066, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 61, "facts_retained": 7, "retention_rate": 0.11475409836065574, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.3076923076923077, "status": 0.0, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 690}, "ledger_by_hop": {"1": {"facts_total": 61, "facts_retained": 7, "retention_rate": 0.11475409836065574, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.3076923076923077, "status": 0.0, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 690}, "2": {"facts_total": 61, "facts_retained": 7, "retention_rate": 0.11475409836065574, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.3076923076923077, "status": 0.0, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 690}}}, "reference_key_id": "keys/synthetic/syn_03015_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "e54cba3528ab78e09d5da03579b2a2035abb56934b1d0571c12acb32bf58b543", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "3ba6bcbbb3b827482b226276", "attempt_id": "af44fcd1a920", "run_group_id": "E4_core:A5|closed|qwen3b:h2", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:51:58.133875Z", "task_family": "synthetic", "document_id": "syn_03015_2000", "document_hash": "71a7284b9bc3f681d69d91d18b9e8d45a4dda486cd14f4145395786ac6a68a89", "question_id": "q11_aggregation", "split": "test", "cluster_id": "syn_03015_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 2, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1922, "query_tokens": 422, "message_tokens": 328, "message_utf8_bytes": 754, "cache_tensor_bytes": null, "serialized_payload_bytes": 754, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 37603.902374976315, "hops": {"1": 22878.284082980826, "2": 14725.618291995488}, "receiver_prefill_ms": 434.8127499688417, "ttft_ms": 46.20929097291082, "decode_ms": 238.0283330567181}, "total_wall_ms": 38323.639416950755, "ttft_ms": 46.20929097291082, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 7}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "8421725", "evidence_artifact": {"ledger": {"facts_total": 61, "facts_retained": 7, "retention_rate": 0.11475409836065574, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.3076923076923077, "status": 0.0, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 690}, "ledger_by_hop": {"1": {"facts_total": 61, "facts_retained": 7, "retention_rate": 0.11475409836065574, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.3076923076923077, "status": 0.0, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 690}, "2": {"facts_total": 61, "facts_retained": 7, "retention_rate": 0.11475409836065574, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.3076923076923077, "status": 0.0, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 690}}}, "reference_key_id": "keys/synthetic/syn_03015_2000.json", "task_scores": {"correct": 0.0, "abstained": 0.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "e54cba3528ab78e09d5da03579b2a2035abb56934b1d0571c12acb32bf58b543", "output_hash": "e4949aba3b0b77de05a87edaab854329fa75df7bf68b29a3def71e985a06e5b5"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "6e242f2b1458b321e698955e", "attempt_id": "faa91b11210f", "run_group_id": "E4_core:A5|closed|qwen3b:h2", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:51:58.871602Z", "task_family": "synthetic", "document_id": "syn_03015_2000", "document_hash": "71a7284b9bc3f681d69d91d18b9e8d45a4dda486cd14f4145395786ac6a68a89", "question_id": "q12_unit_conversion", "split": "test", "cluster_id": "syn_03015_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 2, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1922, "query_tokens": 419, "message_tokens": 328, "message_utf8_bytes": 754, "cache_tensor_bytes": null, "serialized_payload_bytes": 754, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 37603.902374976315, "hops": {"1": 22878.284082980826, "2": 14725.618291995488}, "receiver_prefill_ms": 307.9222079832107, "ttft_ms": 34.223000053316355, "decode_ms": 29.177417047321796}, "total_wall_ms": 37975.6316250423, "ttft_ms": 34.223000053316355, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 61, "facts_retained": 7, "retention_rate": 0.11475409836065574, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.3076923076923077, "status": 0.0, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 690}, "ledger_by_hop": {"1": {"facts_total": 61, "facts_retained": 7, "retention_rate": 0.11475409836065574, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.3076923076923077, "status": 0.0, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 690}, "2": {"facts_total": 61, "facts_retained": 7, "retention_rate": 0.11475409836065574, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.3076923076923077, "status": 0.0, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 690}}}, "reference_key_id": "keys/synthetic/syn_03015_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "e54cba3528ab78e09d5da03579b2a2035abb56934b1d0571c12acb32bf58b543", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "3f2f6dcd5f41661a3a320f84", "attempt_id": "5a5be67f3f5f", "run_group_id": "E4_core:A5|closed|qwen3b:h2", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:51:59.259633Z", "task_family": "synthetic", "document_id": "syn_03015_2000", "document_hash": "71a7284b9bc3f681d69d91d18b9e8d45a4dda486cd14f4145395786ac6a68a89", "question_id": "q13_unanswerable", "split": "test", "cluster_id": "syn_03015_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 2, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1922, "query_tokens": 423, "message_tokens": 328, "message_utf8_bytes": 754, "cache_tensor_bytes": null, "serialized_payload_bytes": 754, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 37603.902374976315, "hops": {"1": 22878.284082980826, "2": 14725.618291995488}, "receiver_prefill_ms": 272.29383401572704, "ttft_ms": 41.81916592642665, "decode_ms": 60.54191698785871}, "total_wall_ms": 37979.15341693442, "ttft_ms": 41.81916592642665, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 61, "facts_retained": 7, "retention_rate": 0.11475409836065574, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.3076923076923077, "status": 0.0, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 690}, "ledger_by_hop": {"1": {"facts_total": 61, "facts_retained": 7, "retention_rate": 0.11475409836065574, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.3076923076923077, "status": 0.0, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 690}, "2": {"facts_total": 61, "facts_retained": 7, "retention_rate": 0.11475409836065574, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.3076923076923077, "status": 0.0, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 690}}}, "reference_key_id": "keys/synthetic/syn_03015_2000.json", "task_scores": {"correct": 1.0, "abstained": 1.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "e54cba3528ab78e09d5da03579b2a2035abb56934b1d0571c12acb32bf58b543", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "478e111682aeedd3296f484e", "attempt_id": "3f2721f0b71c", "run_group_id": "E4_core:A5|closed|qwen3b:h2", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:51:59.652590Z", "task_family": "synthetic", "document_id": "syn_03015_2000", "document_hash": "71a7284b9bc3f681d69d91d18b9e8d45a4dda486cd14f4145395786ac6a68a89", "question_id": "q14_unanswerable", "split": "test", "cluster_id": "syn_03015_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 2, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1922, "query_tokens": 424, "message_tokens": 328, "message_utf8_bytes": 754, "cache_tensor_bytes": null, "serialized_payload_bytes": 754, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 37603.902374976315, "hops": {"1": 22878.284082980826, "2": 14725.618291995488}, "receiver_prefill_ms": 438.0523329600692, "ttft_ms": 40.109666995704174, "decode_ms": 32.765708048827946}, "total_wall_ms": 38115.59933307581, "ttft_ms": 40.109666995704174, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 61, "facts_retained": 7, "retention_rate": 0.11475409836065574, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.3076923076923077, "status": 0.0, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 690}, "ledger_by_hop": {"1": {"facts_total": 61, "facts_retained": 7, "retention_rate": 0.11475409836065574, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.3076923076923077, "status": 0.0, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 690}, "2": {"facts_total": 61, "facts_retained": 7, "retention_rate": 0.11475409836065574, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.3076923076923077, "status": 0.0, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 690}}}, "reference_key_id": "keys/synthetic/syn_03015_2000.json", "task_scores": {"correct": 1.0, "abstained": 1.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "e54cba3528ab78e09d5da03579b2a2035abb56934b1d0571c12acb32bf58b543", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "c98a09fef45eb3d26c219774", "attempt_id": "27f6f1dbd7d6", "run_group_id": "E4_core:A5|closed|qwen3b:h4", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:52:29.230599Z", "task_family": "synthetic", "document_id": "syn_03015_2000", "document_hash": "71a7284b9bc3f681d69d91d18b9e8d45a4dda486cd14f4145395786ac6a68a89", "question_id": "q00_single_fact", "split": "test", "cluster_id": "syn_03015_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 4, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1922, "query_tokens": 411, "message_tokens": 328, "message_utf8_bytes": 754, "cache_tensor_bytes": null, "serialized_payload_bytes": 754, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 66650.15262493398, "hops": {"1": 22878.284082980826, "2": 14725.618291995488, "3": 14445.828791940585, "4": 14600.421458017081}, "receiver_prefill_ms": 265.62274992465973, "ttft_ms": 20.055250031873584, "decode_ms": 76.18058403022587}, "total_wall_ms": 67012.53412500955, "ttft_ms": 20.055250031873584, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 61, "facts_retained": 7, "retention_rate": 0.11475409836065574, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.3076923076923077, "status": 0.0, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 690}, "ledger_by_hop": {"1": {"facts_total": 61, "facts_retained": 7, "retention_rate": 0.11475409836065574, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.3076923076923077, "status": 0.0, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 690}, "2": {"facts_total": 61, "facts_retained": 7, "retention_rate": 0.11475409836065574, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.3076923076923077, "status": 0.0, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 690}, "3": {"facts_total": 61, "facts_retained": 7, "retention_rate": 0.11475409836065574, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.3076923076923077, "status": 0.0, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 690}, "4": {"facts_total": 61, "facts_retained": 7, "retention_rate": 0.11475409836065574, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.3076923076923077, "status": 0.0, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 690}}}, "reference_key_id": "keys/synthetic/syn_03015_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "e54cba3528ab78e09d5da03579b2a2035abb56934b1d0571c12acb32bf58b543", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "bc0817680619c74f0ba7cf80", "attempt_id": "236eb2c34e4d", "run_group_id": "E4_core:A5|closed|qwen3b:h4", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:52:29.610496Z", "task_family": "synthetic", "document_id": "syn_03015_2000", "document_hash": "71a7284b9bc3f681d69d91d18b9e8d45a4dda486cd14f4145395786ac6a68a89", "question_id": "q01_single_fact", "split": "test", "cluster_id": "syn_03015_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 4, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1922, "query_tokens": 416, "message_tokens": 328, "message_utf8_bytes": 754, "cache_tensor_bytes": null, "serialized_payload_bytes": 754, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 66650.15262493398, "hops": {"1": 22878.284082980826, "2": 14725.618291995488, "3": 14445.828791940585, "4": 14600.421458017081}, "receiver_prefill_ms": 330.11366694699973, "ttft_ms": 89.47133307810873, "decode_ms": 136.94620796013623}, "total_wall_ms": 67207.17841596343, "ttft_ms": 89.47133307810873, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 61, "facts_retained": 7, "retention_rate": 0.11475409836065574, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.3076923076923077, "status": 0.0, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 690}, "ledger_by_hop": {"1": {"facts_total": 61, "facts_retained": 7, "retention_rate": 0.11475409836065574, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.3076923076923077, "status": 0.0, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 690}, "2": {"facts_total": 61, "facts_retained": 7, "retention_rate": 0.11475409836065574, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.3076923076923077, "status": 0.0, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 690}, "3": {"facts_total": 61, "facts_retained": 7, "retention_rate": 0.11475409836065574, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.3076923076923077, "status": 0.0, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 690}, "4": {"facts_total": 61, "facts_retained": 7, "retention_rate": 0.11475409836065574, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.3076923076923077, "status": 0.0, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 690}}}, "reference_key_id": "keys/synthetic/syn_03015_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "e54cba3528ab78e09d5da03579b2a2035abb56934b1d0571c12acb32bf58b543", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "5a826a5c5952f5562c8f5a00", "attempt_id": "8981ca5e2a05", "run_group_id": "E4_core:A5|closed|qwen3b:h4", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:52:30.186101Z", "task_family": "synthetic", "document_id": "syn_03015_2000", "document_hash": "71a7284b9bc3f681d69d91d18b9e8d45a4dda486cd14f4145395786ac6a68a89", "question_id": "q02_exact_copy", "split": "test", "cluster_id": "syn_03015_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 4, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1922, "query_tokens": 415, "message_tokens": 328, "message_utf8_bytes": 754, "cache_tensor_bytes": null, "serialized_payload_bytes": 754, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 66650.15262493398, "hops": {"1": 22878.284082980826, "2": 14725.618291995488, "3": 14445.828791940585, "4": 14600.421458017081}, "receiver_prefill_ms": 294.5248750038445, "ttft_ms": 41.29474994260818, "decode_ms": 24.449417018331587}, "total_wall_ms": 67010.95683395397, "ttft_ms": 41.29474994260818, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 61, "facts_retained": 7, "retention_rate": 0.11475409836065574, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.3076923076923077, "status": 0.0, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 690}, "ledger_by_hop": {"1": {"facts_total": 61, "facts_retained": 7, "retention_rate": 0.11475409836065574, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.3076923076923077, "status": 0.0, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 690}, "2": {"facts_total": 61, "facts_retained": 7, "retention_rate": 0.11475409836065574, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.3076923076923077, "status": 0.0, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 690}, "3": {"facts_total": 61, "facts_retained": 7, "retention_rate": 0.11475409836065574, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.3076923076923077, "status": 0.0, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 690}, "4": {"facts_total": 61, "facts_retained": 7, "retention_rate": 0.11475409836065574, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.3076923076923077, "status": 0.0, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 690}}}, "reference_key_id": "keys/synthetic/syn_03015_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "e54cba3528ab78e09d5da03579b2a2035abb56934b1d0571c12acb32bf58b543", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "340c9a5ae86f82ea0143766d", "attempt_id": "b5390ca48684", "run_group_id": "E4_core:A5|closed|qwen3b:h4", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:52:30.563949Z", "task_family": "synthetic", "document_id": "syn_03015_2000", "document_hash": "71a7284b9bc3f681d69d91d18b9e8d45a4dda486cd14f4145395786ac6a68a89", "question_id": "q03_exact_copy", "split": "test", "cluster_id": "syn_03015_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 4, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1922, "query_tokens": 411, "message_tokens": 328, "message_utf8_bytes": 754, "cache_tensor_bytes": null, "serialized_payload_bytes": 754, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 66650.15262493398, "hops": {"1": 22878.284082980826, "2": 14725.618291995488, "3": 14445.828791940585, "4": 14600.421458017081}, "receiver_prefill_ms": 299.5140419807285, "ttft_ms": 41.43720807041973, "decode_ms": 180.1711660809815}, "total_wall_ms": 67171.7519168742, "ttft_ms": 41.43720807041973, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 6}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "Wilhelmina Dubrovnik", "evidence_artifact": {"ledger": {"facts_total": 61, "facts_retained": 7, "retention_rate": 0.11475409836065574, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.3076923076923077, "status": 0.0, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 690}, "ledger_by_hop": {"1": {"facts_total": 61, "facts_retained": 7, "retention_rate": 0.11475409836065574, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.3076923076923077, "status": 0.0, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 690}, "2": {"facts_total": 61, "facts_retained": 7, "retention_rate": 0.11475409836065574, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.3076923076923077, "status": 0.0, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 690}, "3": {"facts_total": 61, "facts_retained": 7, "retention_rate": 0.11475409836065574, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.3076923076923077, "status": 0.0, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 690}, "4": {"facts_total": 61, "facts_retained": 7, "retention_rate": 0.11475409836065574, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.3076923076923077, "status": 0.0, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 690}}}, "reference_key_id": "keys/synthetic/syn_03015_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "e54cba3528ab78e09d5da03579b2a2035abb56934b1d0571c12acb32bf58b543", "output_hash": "f42f05ebfc96f7258cbfbbb37cd65e114d3d15f869e68e6c95b71fc288c30a32"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "8c9c813f3b15229d753b0e06", "attempt_id": "efb2f9e6bca1", "run_group_id": "E4_core:A5|closed|qwen3b:h4", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:52:31.103277Z", "task_family": "synthetic", "document_id": "syn_03015_2000", "document_hash": "71a7284b9bc3f681d69d91d18b9e8d45a4dda486cd14f4145395786ac6a68a89", "question_id": "q04_multi_hop", "split": "test", "cluster_id": "syn_03015_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 4, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1922, "query_tokens": 420, "message_tokens": 328, "message_utf8_bytes": 754, "cache_tensor_bytes": null, "serialized_payload_bytes": 754, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 66650.15262493398, "hops": {"1": 22878.284082980826, "2": 14725.618291995488, "3": 14445.828791940585, "4": 14600.421458017081}, "receiver_prefill_ms": 300.43062497861683, "ttft_ms": 53.378458018414676, "decode_ms": 465.02845908980817}, "total_wall_ms": 67469.47566594463, "ttft_ms": 53.378458018414676, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 8}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "8421725 EUR", "evidence_artifact": {"ledger": {"facts_total": 61, "facts_retained": 7, "retention_rate": 0.11475409836065574, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.3076923076923077, "status": 0.0, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 690}, "ledger_by_hop": {"1": {"facts_total": 61, "facts_retained": 7, "retention_rate": 0.11475409836065574, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.3076923076923077, "status": 0.0, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 690}, "2": {"facts_total": 61, "facts_retained": 7, "retention_rate": 0.11475409836065574, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.3076923076923077, "status": 0.0, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 690}, "3": {"facts_total": 61, "facts_retained": 7, "retention_rate": 0.11475409836065574, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.3076923076923077, "status": 0.0, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 690}, "4": {"facts_total": 61, "facts_retained": 7, "retention_rate": 0.11475409836065574, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.3076923076923077, "status": 0.0, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 690}}}, "reference_key_id": "keys/synthetic/syn_03015_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 0.6666666666666666}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "e54cba3528ab78e09d5da03579b2a2035abb56934b1d0571c12acb32bf58b543", "output_hash": "976c8be2c1a175a4fb5643c5785f93e27ad9205cebd9569a64b821e1ae51ac0e"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "5dc70095bd04f5549ce493d4", "attempt_id": "d3add073b251", "run_group_id": "E4_core:A5|closed|qwen3b:h4", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:52:31.941989Z", "task_family": "synthetic", "document_id": "syn_03015_2000", "document_hash": "71a7284b9bc3f681d69d91d18b9e8d45a4dda486cd14f4145395786ac6a68a89", "question_id": "q05_multi_hop", "split": "test", "cluster_id": "syn_03015_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 4, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1922, "query_tokens": 420, "message_tokens": 328, "message_utf8_bytes": 754, "cache_tensor_bytes": null, "serialized_payload_bytes": 754, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 66650.15262493398, "hops": {"1": 22878.284082980826, "2": 14725.618291995488, "3": 14445.828791940585, "4": 14600.421458017081}, "receiver_prefill_ms": 298.91791695263237, "ttft_ms": 19.668209017254412, "decode_ms": 40.24645802564919}, "total_wall_ms": 67009.79874993209, "ttft_ms": 19.668209017254412, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 61, "facts_retained": 7, "retention_rate": 0.11475409836065574, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.3076923076923077, "status": 0.0, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 690}, "ledger_by_hop": {"1": {"facts_total": 61, "facts_retained": 7, "retention_rate": 0.11475409836065574, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.3076923076923077, "status": 0.0, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 690}, "2": {"facts_total": 61, "facts_retained": 7, "retention_rate": 0.11475409836065574, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.3076923076923077, "status": 0.0, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 690}, "3": {"facts_total": 61, "facts_retained": 7, "retention_rate": 0.11475409836065574, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.3076923076923077, "status": 0.0, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 690}, "4": {"facts_total": 61, "facts_retained": 7, "retention_rate": 0.11475409836065574, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.3076923076923077, "status": 0.0, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 690}}}, "reference_key_id": "keys/synthetic/syn_03015_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "e54cba3528ab78e09d5da03579b2a2035abb56934b1d0571c12acb32bf58b543", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "5a2f1609c12d91ccf489f6f2", "attempt_id": "a288a22d61fc", "run_group_id": "E4_core:A5|closed|qwen3b:h4", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:52:32.319355Z", "task_family": "synthetic", "document_id": "syn_03015_2000", "document_hash": "71a7284b9bc3f681d69d91d18b9e8d45a4dda486cd14f4145395786ac6a68a89", "question_id": "q06_negation", "split": "test", "cluster_id": "syn_03015_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 4, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1922, "query_tokens": 412, "message_tokens": 328, "message_utf8_bytes": 754, "cache_tensor_bytes": null, "serialized_payload_bytes": 754, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 66650.15262493398, "hops": {"1": 22878.284082980826, "2": 14725.618291995488, "3": 14445.828791940585, "4": 14600.421458017081}, "receiver_prefill_ms": 302.46129201259464, "ttft_ms": 19.464374985545874, "decode_ms": 19.715582951903343}, "total_wall_ms": 66992.35724995378, "ttft_ms": 19.464374985545874, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 61, "facts_retained": 7, "retention_rate": 0.11475409836065574, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.3076923076923077, "status": 0.0, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 690}, "ledger_by_hop": {"1": {"facts_total": 61, "facts_retained": 7, "retention_rate": 0.11475409836065574, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.3076923076923077, "status": 0.0, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 690}, "2": {"facts_total": 61, "facts_retained": 7, "retention_rate": 0.11475409836065574, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.3076923076923077, "status": 0.0, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 690}, "3": {"facts_total": 61, "facts_retained": 7, "retention_rate": 0.11475409836065574, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.3076923076923077, "status": 0.0, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 690}, "4": {"facts_total": 61, "facts_retained": 7, "retention_rate": 0.11475409836065574, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.3076923076923077, "status": 0.0, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 690}}}, "reference_key_id": "keys/synthetic/syn_03015_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "e54cba3528ab78e09d5da03579b2a2035abb56934b1d0571c12acb32bf58b543", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "5a22c61050386c27f71dea59", "attempt_id": "ecacd2a91ff3", "run_group_id": "E4_core:A5|closed|qwen3b:h4", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:52:32.680190Z", "task_family": "synthetic", "document_id": "syn_03015_2000", "document_hash": "71a7284b9bc3f681d69d91d18b9e8d45a4dda486cd14f4145395786ac6a68a89", "question_id": "q07_negation", "split": "test", "cluster_id": "syn_03015_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 4, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1922, "query_tokens": 412, "message_tokens": 328, "message_utf8_bytes": 754, "cache_tensor_bytes": null, "serialized_payload_bytes": 754, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 66650.15262493398, "hops": {"1": 22878.284082980826, "2": 14725.618291995488, "3": 14445.828791940585, "4": 14600.421458017081}, "receiver_prefill_ms": 269.6417080005631, "ttft_ms": 58.689124998636544, "decode_ms": 59.40783303231001}, "total_wall_ms": 67038.52295794059, "ttft_ms": 58.689124998636544, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 61, "facts_retained": 7, "retention_rate": 0.11475409836065574, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.3076923076923077, "status": 0.0, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 690}, "ledger_by_hop": {"1": {"facts_total": 61, "facts_retained": 7, "retention_rate": 0.11475409836065574, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.3076923076923077, "status": 0.0, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 690}, "2": {"facts_total": 61, "facts_retained": 7, "retention_rate": 0.11475409836065574, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.3076923076923077, "status": 0.0, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 690}, "3": {"facts_total": 61, "facts_retained": 7, "retention_rate": 0.11475409836065574, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.3076923076923077, "status": 0.0, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 690}, "4": {"facts_total": 61, "facts_retained": 7, "retention_rate": 0.11475409836065574, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.3076923076923077, "status": 0.0, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 690}}}, "reference_key_id": "keys/synthetic/syn_03015_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "e54cba3528ab78e09d5da03579b2a2035abb56934b1d0571c12acb32bf58b543", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "d6de134550a5da96ecc8bd39", "attempt_id": "23176e6686e8", "run_group_id": "E4_core:A5|closed|qwen3b:h4", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:52:33.085280Z", "task_family": "synthetic", "document_id": "syn_03015_2000", "document_hash": "71a7284b9bc3f681d69d91d18b9e8d45a4dda486cd14f4145395786ac6a68a89", "question_id": "q08_revision", "split": "test", "cluster_id": "syn_03015_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 4, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1922, "query_tokens": 426, "message_tokens": 328, "message_utf8_bytes": 754, "cache_tensor_bytes": null, "serialized_payload_bytes": 754, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 66650.15262493398, "hops": {"1": 22878.284082980826, "2": 14725.618291995488, "3": 14445.828791940585, "4": 14600.421458017081}, "receiver_prefill_ms": 441.10425002872944, "ttft_ms": 22.074999986216426, "decode_ms": 39.44637509994209}, "total_wall_ms": 67153.28899992164, "ttft_ms": 22.074999986216426, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 61, "facts_retained": 7, "retention_rate": 0.11475409836065574, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.3076923076923077, "status": 0.0, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 690}, "ledger_by_hop": {"1": {"facts_total": 61, "facts_retained": 7, "retention_rate": 0.11475409836065574, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.3076923076923077, "status": 0.0, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 690}, "2": {"facts_total": 61, "facts_retained": 7, "retention_rate": 0.11475409836065574, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.3076923076923077, "status": 0.0, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 690}, "3": {"facts_total": 61, "facts_retained": 7, "retention_rate": 0.11475409836065574, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.3076923076923077, "status": 0.0, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 690}, "4": {"facts_total": 61, "facts_retained": 7, "retention_rate": 0.11475409836065574, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.3076923076923077, "status": 0.0, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 690}}}, "reference_key_id": "keys/synthetic/syn_03015_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "e54cba3528ab78e09d5da03579b2a2035abb56934b1d0571c12acb32bf58b543", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "7a56f79912f18bb11bf1ea03", "attempt_id": "acdac552d8a1", "run_group_id": "E4_core:A5|closed|qwen3b:h4", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:52:33.605505Z", "task_family": "synthetic", "document_id": "syn_03015_2000", "document_hash": "71a7284b9bc3f681d69d91d18b9e8d45a4dda486cd14f4145395786ac6a68a89", "question_id": "q09_revision", "split": "test", "cluster_id": "syn_03015_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 4, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1922, "query_tokens": 420, "message_tokens": 328, "message_utf8_bytes": 754, "cache_tensor_bytes": null, "serialized_payload_bytes": 754, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 66650.15262493398, "hops": {"1": 22878.284082980826, "2": 14725.618291995488, "3": 14445.828791940585, "4": 14600.421458017081}, "receiver_prefill_ms": 309.7161250188947, "ttft_ms": 42.30537498369813, "decode_ms": 217.71508397068828}, "total_wall_ms": 67220.49929189961, "ttft_ms": 42.30537498369813, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 7}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "7028925", "evidence_artifact": {"ledger": {"facts_total": 61, "facts_retained": 7, "retention_rate": 0.11475409836065574, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.3076923076923077, "status": 0.0, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 690}, "ledger_by_hop": {"1": {"facts_total": 61, "facts_retained": 7, "retention_rate": 0.11475409836065574, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.3076923076923077, "status": 0.0, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 690}, "2": {"facts_total": 61, "facts_retained": 7, "retention_rate": 0.11475409836065574, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.3076923076923077, "status": 0.0, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 690}, "3": {"facts_total": 61, "facts_retained": 7, "retention_rate": 0.11475409836065574, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.3076923076923077, "status": 0.0, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 690}, "4": {"facts_total": 61, "facts_retained": 7, "retention_rate": 0.11475409836065574, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.3076923076923077, "status": 0.0, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 690}}}, "reference_key_id": "keys/synthetic/syn_03015_2000.json", "task_scores": {"correct": 0.0, "abstained": 0.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "e54cba3528ab78e09d5da03579b2a2035abb56934b1d0571c12acb32bf58b543", "output_hash": "e966a398855c104a86bce07742c3ce8108a432926bc704940ecfd07e44596774"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "855a83e7c18429a1b7430e8f", "attempt_id": "a12d30d51159", "run_group_id": "E4_core:A5|closed|qwen3b:h4", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:52:34.194014Z", "task_family": "synthetic", "document_id": "syn_03015_2000", "document_hash": "71a7284b9bc3f681d69d91d18b9e8d45a4dda486cd14f4145395786ac6a68a89", "question_id": "q10_temporal", "split": "test", "cluster_id": "syn_03015_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 4, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1922, "query_tokens": 429, "message_tokens": 328, "message_utf8_bytes": 754, "cache_tensor_bytes": null, "serialized_payload_bytes": 754, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 66650.15262493398, "hops": {"1": 22878.284082980826, "2": 14725.618291995488, "3": 14445.828791940585, "4": 14600.421458017081}, "receiver_prefill_ms": 255.081707960926, "ttft_ms": 39.37591705471277, "decode_ms": 36.553499987348914}, "total_wall_ms": 66982.01329191215, "ttft_ms": 39.37591705471277, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 61, "facts_retained": 7, "retention_rate": 0.11475409836065574, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.3076923076923077, "status": 0.0, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 690}, "ledger_by_hop": {"1": {"facts_total": 61, "facts_retained": 7, "retention_rate": 0.11475409836065574, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.3076923076923077, "status": 0.0, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 690}, "2": {"facts_total": 61, "facts_retained": 7, "retention_rate": 0.11475409836065574, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.3076923076923077, "status": 0.0, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 690}, "3": {"facts_total": 61, "facts_retained": 7, "retention_rate": 0.11475409836065574, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.3076923076923077, "status": 0.0, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 690}, "4": {"facts_total": 61, "facts_retained": 7, "retention_rate": 0.11475409836065574, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.3076923076923077, "status": 0.0, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 690}}}, "reference_key_id": "keys/synthetic/syn_03015_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "e54cba3528ab78e09d5da03579b2a2035abb56934b1d0571c12acb32bf58b543", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "6274dedacdb03135be9dbbf3", "attempt_id": "5ebddd066479", "run_group_id": "E4_core:A5|closed|qwen3b:h4", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:52:34.543025Z", "task_family": "synthetic", "document_id": "syn_03015_2000", "document_hash": "71a7284b9bc3f681d69d91d18b9e8d45a4dda486cd14f4145395786ac6a68a89", "question_id": "q11_aggregation", "split": "test", "cluster_id": "syn_03015_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 4, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1922, "query_tokens": 422, "message_tokens": 328, "message_utf8_bytes": 754, "cache_tensor_bytes": null, "serialized_payload_bytes": 754, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 66650.15262493398, "hops": {"1": 22878.284082980826, "2": 14725.618291995488, "3": 14445.828791940585, "4": 14600.421458017081}, "receiver_prefill_ms": 418.5397910187021, "ttft_ms": 75.11187496129423, "decode_ms": 244.76212495937943}, "total_wall_ms": 67389.02908284217, "ttft_ms": 75.11187496129423, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 7}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "8421725", "evidence_artifact": {"ledger": {"facts_total": 61, "facts_retained": 7, "retention_rate": 0.11475409836065574, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.3076923076923077, "status": 0.0, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 690}, "ledger_by_hop": {"1": {"facts_total": 61, "facts_retained": 7, "retention_rate": 0.11475409836065574, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.3076923076923077, "status": 0.0, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 690}, "2": {"facts_total": 61, "facts_retained": 7, "retention_rate": 0.11475409836065574, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.3076923076923077, "status": 0.0, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 690}, "3": {"facts_total": 61, "facts_retained": 7, "retention_rate": 0.11475409836065574, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.3076923076923077, "status": 0.0, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 690}, "4": {"facts_total": 61, "facts_retained": 7, "retention_rate": 0.11475409836065574, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.3076923076923077, "status": 0.0, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 690}}}, "reference_key_id": "keys/synthetic/syn_03015_2000.json", "task_scores": {"correct": 0.0, "abstained": 0.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "e54cba3528ab78e09d5da03579b2a2035abb56934b1d0571c12acb32bf58b543", "output_hash": "e4949aba3b0b77de05a87edaab854329fa75df7bf68b29a3def71e985a06e5b5"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "e001085472abecad0bad6daf", "attempt_id": "07b9409744c5", "run_group_id": "E4_core:A5|closed|qwen3b:h4", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:52:35.299359Z", "task_family": "synthetic", "document_id": "syn_03015_2000", "document_hash": "71a7284b9bc3f681d69d91d18b9e8d45a4dda486cd14f4145395786ac6a68a89", "question_id": "q12_unit_conversion", "split": "test", "cluster_id": "syn_03015_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 4, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1922, "query_tokens": 419, "message_tokens": 328, "message_utf8_bytes": 754, "cache_tensor_bytes": null, "serialized_payload_bytes": 754, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 66650.15262493398, "hops": {"1": 22878.284082980826, "2": 14725.618291995488, "3": 14445.828791940585, "4": 14600.421458017081}, "receiver_prefill_ms": 280.36125004291534, "ttft_ms": 41.7960409540683, "decode_ms": 21.68716595042497}, "total_wall_ms": 66994.51154191047, "ttft_ms": 41.7960409540683, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 61, "facts_retained": 7, "retention_rate": 0.11475409836065574, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.3076923076923077, "status": 0.0, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 690}, "ledger_by_hop": {"1": {"facts_total": 61, "facts_retained": 7, "retention_rate": 0.11475409836065574, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.3076923076923077, "status": 0.0, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 690}, "2": {"facts_total": 61, "facts_retained": 7, "retention_rate": 0.11475409836065574, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.3076923076923077, "status": 0.0, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 690}, "3": {"facts_total": 61, "facts_retained": 7, "retention_rate": 0.11475409836065574, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.3076923076923077, "status": 0.0, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 690}, "4": {"facts_total": 61, "facts_retained": 7, "retention_rate": 0.11475409836065574, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.3076923076923077, "status": 0.0, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 690}}}, "reference_key_id": "keys/synthetic/syn_03015_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "e54cba3528ab78e09d5da03579b2a2035abb56934b1d0571c12acb32bf58b543", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "e9b936970775f6c275b4fef9", "attempt_id": "0b7997d167ea", "run_group_id": "E4_core:A5|closed|qwen3b:h4", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:52:35.661849Z", "task_family": "synthetic", "document_id": "syn_03015_2000", "document_hash": "71a7284b9bc3f681d69d91d18b9e8d45a4dda486cd14f4145395786ac6a68a89", "question_id": "q13_unanswerable", "split": "test", "cluster_id": "syn_03015_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 4, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1922, "query_tokens": 423, "message_tokens": 328, "message_utf8_bytes": 754, "cache_tensor_bytes": null, "serialized_payload_bytes": 754, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 66650.15262493398, "hops": {"1": 22878.284082980826, "2": 14725.618291995488, "3": 14445.828791940585, "4": 14600.421458017081}, "receiver_prefill_ms": 274.6084580430761, "ttft_ms": 11.916667106561363, "decode_ms": 16.663082991726696}, "total_wall_ms": 66953.95729190204, "ttft_ms": 11.916667106561363, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 61, "facts_retained": 7, "retention_rate": 0.11475409836065574, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.3076923076923077, "status": 0.0, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 690}, "ledger_by_hop": {"1": {"facts_total": 61, "facts_retained": 7, "retention_rate": 0.11475409836065574, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.3076923076923077, "status": 0.0, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 690}, "2": {"facts_total": 61, "facts_retained": 7, "retention_rate": 0.11475409836065574, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.3076923076923077, "status": 0.0, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 690}, "3": {"facts_total": 61, "facts_retained": 7, "retention_rate": 0.11475409836065574, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.3076923076923077, "status": 0.0, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 690}, "4": {"facts_total": 61, "facts_retained": 7, "retention_rate": 0.11475409836065574, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.3076923076923077, "status": 0.0, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 690}}}, "reference_key_id": "keys/synthetic/syn_03015_2000.json", "task_scores": {"correct": 1.0, "abstained": 1.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "e54cba3528ab78e09d5da03579b2a2035abb56934b1d0571c12acb32bf58b543", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "ca50c27117e67c22480ac194", "attempt_id": "43599582e98e", "run_group_id": "E4_core:A5|closed|qwen3b:h4", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:52:35.983211Z", "task_family": "synthetic", "document_id": "syn_03015_2000", "document_hash": "71a7284b9bc3f681d69d91d18b9e8d45a4dda486cd14f4145395786ac6a68a89", "question_id": "q14_unanswerable", "split": "test", "cluster_id": "syn_03015_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 4, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1922, "query_tokens": 424, "message_tokens": 328, "message_utf8_bytes": 754, "cache_tensor_bytes": null, "serialized_payload_bytes": 754, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 66650.15262493398, "hops": {"1": 22878.284082980826, "2": 14725.618291995488, "3": 14445.828791940585, "4": 14600.421458017081}, "receiver_prefill_ms": 279.0158750722185, "ttft_ms": 39.506625034846365, "decode_ms": 63.064207904972136}, "total_wall_ms": 67032.42487483658, "ttft_ms": 39.506625034846365, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 61, "facts_retained": 7, "retention_rate": 0.11475409836065574, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.3076923076923077, "status": 0.0, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 690}, "ledger_by_hop": {"1": {"facts_total": 61, "facts_retained": 7, "retention_rate": 0.11475409836065574, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.3076923076923077, "status": 0.0, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 690}, "2": {"facts_total": 61, "facts_retained": 7, "retention_rate": 0.11475409836065574, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.3076923076923077, "status": 0.0, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 690}, "3": {"facts_total": 61, "facts_retained": 7, "retention_rate": 0.11475409836065574, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.3076923076923077, "status": 0.0, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 690}, "4": {"facts_total": 61, "facts_retained": 7, "retention_rate": 0.11475409836065574, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.3076923076923077, "status": 0.0, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 690}}}, "reference_key_id": "keys/synthetic/syn_03015_2000.json", "task_scores": {"correct": 1.0, "abstained": 1.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "e54cba3528ab78e09d5da03579b2a2035abb56934b1d0571c12acb32bf58b543", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "36f3809a11abfd08ccb8fc03", "attempt_id": "cdfebed06ba3", "run_group_id": "E4_core:A5|closed|qwen3b:h8", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:53:35.239327Z", "task_family": "synthetic", "document_id": "syn_03015_2000", "document_hash": "71a7284b9bc3f681d69d91d18b9e8d45a4dda486cd14f4145395786ac6a68a89", "question_id": "q00_single_fact", "split": "test", "cluster_id": "syn_03015_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 8, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1922, "query_tokens": 411, "message_tokens": 328, "message_utf8_bytes": 754, "cache_tensor_bytes": null, "serialized_payload_bytes": 754, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 125500.71437389124, "hops": {"1": 22878.284082980826, "2": 14725.618291995488, "3": 14445.828791940585, "4": 14600.421458017081, "5": 14639.447499997914, "6": 14733.761791023426, "7": 14705.673499964178, "8": 14771.678957971744}, "receiver_prefill_ms": 291.82820790447295, "ttft_ms": 40.50908307544887, "decode_ms": 25.671791983768344}, "total_wall_ms": 125859.23462396022, "ttft_ms": 40.50908307544887, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 61, "facts_retained": 7, "retention_rate": 0.11475409836065574, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.3076923076923077, "status": 0.0, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 690}, "ledger_by_hop": {"1": {"facts_total": 61, "facts_retained": 7, "retention_rate": 0.11475409836065574, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.3076923076923077, "status": 0.0, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 690}, "2": {"facts_total": 61, "facts_retained": 7, "retention_rate": 0.11475409836065574, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.3076923076923077, "status": 0.0, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 690}, "3": {"facts_total": 61, "facts_retained": 7, "retention_rate": 0.11475409836065574, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.3076923076923077, "status": 0.0, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 690}, "4": {"facts_total": 61, "facts_retained": 7, "retention_rate": 0.11475409836065574, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.3076923076923077, "status": 0.0, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 690}, "5": {"facts_total": 61, "facts_retained": 7, "retention_rate": 0.11475409836065574, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.3076923076923077, "status": 0.0, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 690}, "6": {"facts_total": 61, "facts_retained": 7, "retention_rate": 0.11475409836065574, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.3076923076923077, "status": 0.0, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 690}, "7": {"facts_total": 61, "facts_retained": 7, "retention_rate": 0.11475409836065574, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.3076923076923077, "status": 0.0, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 690}, "8": {"facts_total": 61, "facts_retained": 7, "retention_rate": 0.11475409836065574, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.3076923076923077, "status": 0.0, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 690}}}, "reference_key_id": "keys/synthetic/syn_03015_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "e54cba3528ab78e09d5da03579b2a2035abb56934b1d0571c12acb32bf58b543", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "d3e97c6bdc1945a7e25df0b2", "attempt_id": "9ccc5a603bfa", "run_group_id": "E4_core:A5|closed|qwen3b:h8", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:53:35.615389Z", "task_family": "synthetic", "document_id": "syn_03015_2000", "document_hash": "71a7284b9bc3f681d69d91d18b9e8d45a4dda486cd14f4145395786ac6a68a89", "question_id": "q01_single_fact", "split": "test", "cluster_id": "syn_03015_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 8, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1922, "query_tokens": 416, "message_tokens": 328, "message_utf8_bytes": 754, "cache_tensor_bytes": null, "serialized_payload_bytes": 754, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 125500.71437389124, "hops": {"1": 22878.284082980826, "2": 14725.618291995488, "3": 14445.828791940585, "4": 14600.421458017081, "5": 14639.447499997914, "6": 14733.761791023426, "7": 14705.673499964178, "8": 14771.678957971744}, "receiver_prefill_ms": 299.8079580720514, "ttft_ms": 38.50575000979006, "decode_ms": 24.852042086422443}, "total_wall_ms": 125864.2944159219, "ttft_ms": 38.50575000979006, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 61, "facts_retained": 7, "retention_rate": 0.11475409836065574, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.3076923076923077, "status": 0.0, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 690}, "ledger_by_hop": {"1": {"facts_total": 61, "facts_retained": 7, "retention_rate": 0.11475409836065574, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.3076923076923077, "status": 0.0, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 690}, "2": {"facts_total": 61, "facts_retained": 7, "retention_rate": 0.11475409836065574, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.3076923076923077, "status": 0.0, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 690}, "3": {"facts_total": 61, "facts_retained": 7, "retention_rate": 0.11475409836065574, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.3076923076923077, "status": 0.0, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 690}, "4": {"facts_total": 61, "facts_retained": 7, "retention_rate": 0.11475409836065574, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.3076923076923077, "status": 0.0, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 690}, "5": {"facts_total": 61, "facts_retained": 7, "retention_rate": 0.11475409836065574, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.3076923076923077, "status": 0.0, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 690}, "6": {"facts_total": 61, "facts_retained": 7, "retention_rate": 0.11475409836065574, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.3076923076923077, "status": 0.0, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 690}, "7": {"facts_total": 61, "facts_retained": 7, "retention_rate": 0.11475409836065574, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.3076923076923077, "status": 0.0, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 690}, "8": {"facts_total": 61, "facts_retained": 7, "retention_rate": 0.11475409836065574, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.3076923076923077, "status": 0.0, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 690}}}, "reference_key_id": "keys/synthetic/syn_03015_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "e54cba3528ab78e09d5da03579b2a2035abb56934b1d0571c12acb32bf58b543", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "9eff9594681aa684e99766e9", "attempt_id": "76784199cacf", "run_group_id": "E4_core:A5|closed|qwen3b:h8", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:53:35.996716Z", "task_family": "synthetic", "document_id": "syn_03015_2000", "document_hash": "71a7284b9bc3f681d69d91d18b9e8d45a4dda486cd14f4145395786ac6a68a89", "question_id": "q02_exact_copy", "split": "test", "cluster_id": "syn_03015_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 8, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1922, "query_tokens": 415, "message_tokens": 328, "message_utf8_bytes": 754, "cache_tensor_bytes": null, "serialized_payload_bytes": 754, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 125500.71437389124, "hops": {"1": 22878.284082980826, "2": 14725.618291995488, "3": 14445.828791940585, "4": 14600.421458017081, "5": 14639.447499997914, "6": 14733.761791023426, "7": 14705.673499964178, "8": 14771.678957971744}, "receiver_prefill_ms": 265.29695792123675, "ttft_ms": 37.45708300266415, "decode_ms": 54.51729195192456}, "total_wall_ms": 125858.66420681123, "ttft_ms": 37.45708300266415, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 61, "facts_retained": 7, "retention_rate": 0.11475409836065574, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.3076923076923077, "status": 0.0, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 690}, "ledger_by_hop": {"1": {"facts_total": 61, "facts_retained": 7, "retention_rate": 0.11475409836065574, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.3076923076923077, "status": 0.0, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 690}, "2": {"facts_total": 61, "facts_retained": 7, "retention_rate": 0.11475409836065574, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.3076923076923077, "status": 0.0, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 690}, "3": {"facts_total": 61, "facts_retained": 7, "retention_rate": 0.11475409836065574, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.3076923076923077, "status": 0.0, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 690}, "4": {"facts_total": 61, "facts_retained": 7, "retention_rate": 0.11475409836065574, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.3076923076923077, "status": 0.0, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 690}, "5": {"facts_total": 61, "facts_retained": 7, "retention_rate": 0.11475409836065574, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.3076923076923077, "status": 0.0, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 690}, "6": {"facts_total": 61, "facts_retained": 7, "retention_rate": 0.11475409836065574, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.3076923076923077, "status": 0.0, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 690}, "7": {"facts_total": 61, "facts_retained": 7, "retention_rate": 0.11475409836065574, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.3076923076923077, "status": 0.0, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 690}, "8": {"facts_total": 61, "facts_retained": 7, "retention_rate": 0.11475409836065574, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.3076923076923077, "status": 0.0, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 690}}}, "reference_key_id": "keys/synthetic/syn_03015_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "e54cba3528ab78e09d5da03579b2a2035abb56934b1d0571c12acb32bf58b543", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "7c81dc5f9998b4540e5df78e", "attempt_id": "1226235c9af1", "run_group_id": "E4_core:A5|closed|qwen3b:h8", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:53:36.372119Z", "task_family": "synthetic", "document_id": "syn_03015_2000", "document_hash": "71a7284b9bc3f681d69d91d18b9e8d45a4dda486cd14f4145395786ac6a68a89", "question_id": "q03_exact_copy", "split": "test", "cluster_id": "syn_03015_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 8, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1922, "query_tokens": 411, "message_tokens": 328, "message_utf8_bytes": 754, "cache_tensor_bytes": null, "serialized_payload_bytes": 754, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 125500.71437389124, "hops": {"1": 22878.284082980826, "2": 14725.618291995488, "3": 14445.828791940585, "4": 14600.421458017081, "5": 14639.447499997914, "6": 14733.761791023426, "7": 14705.673499964178, "8": 14771.678957971744}, "receiver_prefill_ms": 473.192417062819, "ttft_ms": 23.399583995342255, "decode_ms": 186.91670906264335}, "total_wall_ms": 126184.85991482157, "ttft_ms": 23.399583995342255, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 6}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "Wilhelmina Dubrovnik", "evidence_artifact": {"ledger": {"facts_total": 61, "facts_retained": 7, "retention_rate": 0.11475409836065574, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.3076923076923077, "status": 0.0, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 690}, "ledger_by_hop": {"1": {"facts_total": 61, "facts_retained": 7, "retention_rate": 0.11475409836065574, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.3076923076923077, "status": 0.0, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 690}, "2": {"facts_total": 61, "facts_retained": 7, "retention_rate": 0.11475409836065574, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.3076923076923077, "status": 0.0, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 690}, "3": {"facts_total": 61, "facts_retained": 7, "retention_rate": 0.11475409836065574, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.3076923076923077, "status": 0.0, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 690}, "4": {"facts_total": 61, "facts_retained": 7, "retention_rate": 0.11475409836065574, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.3076923076923077, "status": 0.0, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 690}, "5": {"facts_total": 61, "facts_retained": 7, "retention_rate": 0.11475409836065574, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.3076923076923077, "status": 0.0, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 690}, "6": {"facts_total": 61, "facts_retained": 7, "retention_rate": 0.11475409836065574, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.3076923076923077, "status": 0.0, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 690}, "7": {"facts_total": 61, "facts_retained": 7, "retention_rate": 0.11475409836065574, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.3076923076923077, "status": 0.0, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 690}, "8": {"facts_total": 61, "facts_retained": 7, "retention_rate": 0.11475409836065574, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.3076923076923077, "status": 0.0, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 690}}}, "reference_key_id": "keys/synthetic/syn_03015_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "e54cba3528ab78e09d5da03579b2a2035abb56934b1d0571c12acb32bf58b543", "output_hash": "f42f05ebfc96f7258cbfbbb37cd65e114d3d15f869e68e6c95b71fc288c30a32"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "dd5efe7a10a4667ef74707fd", "attempt_id": "1987a1c04769", "run_group_id": "E4_core:A5|closed|qwen3b:h8", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:53:37.074549Z", "task_family": "synthetic", "document_id": "syn_03015_2000", "document_hash": "71a7284b9bc3f681d69d91d18b9e8d45a4dda486cd14f4145395786ac6a68a89", "question_id": "q04_multi_hop", "split": "test", "cluster_id": "syn_03015_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 8, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1922, "query_tokens": 420, "message_tokens": 328, "message_utf8_bytes": 754, "cache_tensor_bytes": null, "serialized_payload_bytes": 754, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 125500.71437389124, "hops": {"1": 22878.284082980826, "2": 14725.618291995488, "3": 14445.828791940585, "4": 14600.421458017081, "5": 14639.447499997914, "6": 14733.761791023426, "7": 14705.673499964178, "8": 14771.678957971744}, "receiver_prefill_ms": 296.02795804385096, "ttft_ms": 35.36866698414087, "decode_ms": 246.89395900350064}, "total_wall_ms": 126079.50983289629, "ttft_ms": 35.36866698414087, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 8}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "8421725 EUR", "evidence_artifact": {"ledger": {"facts_total": 61, "facts_retained": 7, "retention_rate": 0.11475409836065574, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.3076923076923077, "status": 0.0, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 690}, "ledger_by_hop": {"1": {"facts_total": 61, "facts_retained": 7, "retention_rate": 0.11475409836065574, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.3076923076923077, "status": 0.0, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 690}, "2": {"facts_total": 61, "facts_retained": 7, "retention_rate": 0.11475409836065574, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.3076923076923077, "status": 0.0, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 690}, "3": {"facts_total": 61, "facts_retained": 7, "retention_rate": 0.11475409836065574, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.3076923076923077, "status": 0.0, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 690}, "4": {"facts_total": 61, "facts_retained": 7, "retention_rate": 0.11475409836065574, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.3076923076923077, "status": 0.0, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 690}, "5": {"facts_total": 61, "facts_retained": 7, "retention_rate": 0.11475409836065574, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.3076923076923077, "status": 0.0, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 690}, "6": {"facts_total": 61, "facts_retained": 7, "retention_rate": 0.11475409836065574, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.3076923076923077, "status": 0.0, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 690}, "7": {"facts_total": 61, "facts_retained": 7, "retention_rate": 0.11475409836065574, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.3076923076923077, "status": 0.0, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 690}, "8": {"facts_total": 61, "facts_retained": 7, "retention_rate": 0.11475409836065574, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.3076923076923077, "status": 0.0, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 690}}}, "reference_key_id": "keys/synthetic/syn_03015_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 0.6666666666666666}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "e54cba3528ab78e09d5da03579b2a2035abb56934b1d0571c12acb32bf58b543", "output_hash": "976c8be2c1a175a4fb5643c5785f93e27ad9205cebd9569a64b821e1ae51ac0e"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "26566fe4cce1ea8ab7d24a8e", "attempt_id": "57bca9520bba", "run_group_id": "E4_core:A5|closed|qwen3b:h8", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:53:37.670772Z", "task_family": "synthetic", "document_id": "syn_03015_2000", "document_hash": "71a7284b9bc3f681d69d91d18b9e8d45a4dda486cd14f4145395786ac6a68a89", "question_id": "q05_multi_hop", "split": "test", "cluster_id": "syn_03015_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 8, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1922, "query_tokens": 420, "message_tokens": 328, "message_utf8_bytes": 754, "cache_tensor_bytes": null, "serialized_payload_bytes": 754, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 125500.71437389124, "hops": {"1": 22878.284082980826, "2": 14725.618291995488, "3": 14445.828791940585, "4": 14600.421458017081, "5": 14639.447499997914, "6": 14733.761791023426, "7": 14705.673499964178, "8": 14771.678957971744}, "receiver_prefill_ms": 325.51100000273436, "ttft_ms": 43.354333960451186, "decode_ms": 55.45508302748203}, "total_wall_ms": 125925.73304090183, "ttft_ms": 43.354333960451186, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 61, "facts_retained": 7, "retention_rate": 0.11475409836065574, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.3076923076923077, "status": 0.0, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 690}, "ledger_by_hop": {"1": {"facts_total": 61, "facts_retained": 7, "retention_rate": 0.11475409836065574, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.3076923076923077, "status": 0.0, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 690}, "2": {"facts_total": 61, "facts_retained": 7, "retention_rate": 0.11475409836065574, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.3076923076923077, "status": 0.0, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 690}, "3": {"facts_total": 61, "facts_retained": 7, "retention_rate": 0.11475409836065574, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.3076923076923077, "status": 0.0, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 690}, "4": {"facts_total": 61, "facts_retained": 7, "retention_rate": 0.11475409836065574, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.3076923076923077, "status": 0.0, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 690}, "5": {"facts_total": 61, "facts_retained": 7, "retention_rate": 0.11475409836065574, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.3076923076923077, "status": 0.0, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 690}, "6": {"facts_total": 61, "facts_retained": 7, "retention_rate": 0.11475409836065574, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.3076923076923077, "status": 0.0, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 690}, "7": {"facts_total": 61, "facts_retained": 7, "retention_rate": 0.11475409836065574, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.3076923076923077, "status": 0.0, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 690}, "8": {"facts_total": 61, "facts_retained": 7, "retention_rate": 0.11475409836065574, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.3076923076923077, "status": 0.0, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 690}}}, "reference_key_id": "keys/synthetic/syn_03015_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "e54cba3528ab78e09d5da03579b2a2035abb56934b1d0571c12acb32bf58b543", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "e2021833dbed2830f29e01d7", "attempt_id": "63cf9cfe321f", "run_group_id": "E4_core:A5|closed|qwen3b:h8", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:53:38.117946Z", "task_family": "synthetic", "document_id": "syn_03015_2000", "document_hash": "71a7284b9bc3f681d69d91d18b9e8d45a4dda486cd14f4145395786ac6a68a89", "question_id": "q06_negation", "split": "test", "cluster_id": "syn_03015_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 8, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1922, "query_tokens": 412, "message_tokens": 328, "message_utf8_bytes": 754, "cache_tensor_bytes": null, "serialized_payload_bytes": 754, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 125500.71437389124, "hops": {"1": 22878.284082980826, "2": 14725.618291995488, "3": 14445.828791940585, "4": 14600.421458017081, "5": 14639.447499997914, "6": 14733.761791023426, "7": 14705.673499964178, "8": 14771.678957971744}, "receiver_prefill_ms": 408.27720798552036, "ttft_ms": 36.07762500178069, "decode_ms": 26.996167027391493}, "total_wall_ms": 125972.71233284846, "ttft_ms": 36.07762500178069, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 61, "facts_retained": 7, "retention_rate": 0.11475409836065574, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.3076923076923077, "status": 0.0, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 690}, "ledger_by_hop": {"1": {"facts_total": 61, "facts_retained": 7, "retention_rate": 0.11475409836065574, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.3076923076923077, "status": 0.0, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 690}, "2": {"facts_total": 61, "facts_retained": 7, "retention_rate": 0.11475409836065574, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.3076923076923077, "status": 0.0, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 690}, "3": {"facts_total": 61, "facts_retained": 7, "retention_rate": 0.11475409836065574, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.3076923076923077, "status": 0.0, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 690}, "4": {"facts_total": 61, "facts_retained": 7, "retention_rate": 0.11475409836065574, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.3076923076923077, "status": 0.0, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 690}, "5": {"facts_total": 61, "facts_retained": 7, "retention_rate": 0.11475409836065574, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.3076923076923077, "status": 0.0, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 690}, "6": {"facts_total": 61, "facts_retained": 7, "retention_rate": 0.11475409836065574, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.3076923076923077, "status": 0.0, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 690}, "7": {"facts_total": 61, "facts_retained": 7, "retention_rate": 0.11475409836065574, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.3076923076923077, "status": 0.0, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 690}, "8": {"facts_total": 61, "facts_retained": 7, "retention_rate": 0.11475409836065574, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.3076923076923077, "status": 0.0, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 690}}}, "reference_key_id": "keys/synthetic/syn_03015_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "e54cba3528ab78e09d5da03579b2a2035abb56934b1d0571c12acb32bf58b543", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "e77a037fadf96b96f9ef4643", "attempt_id": "dd5aa829b212", "run_group_id": "E4_core:A5|closed|qwen3b:h8", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:53:38.606903Z", "task_family": "synthetic", "document_id": "syn_03015_2000", "document_hash": "71a7284b9bc3f681d69d91d18b9e8d45a4dda486cd14f4145395786ac6a68a89", "question_id": "q07_negation", "split": "test", "cluster_id": "syn_03015_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 8, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1922, "query_tokens": 412, "message_tokens": 328, "message_utf8_bytes": 754, "cache_tensor_bytes": null, "serialized_payload_bytes": 754, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 125500.71437389124, "hops": {"1": 22878.284082980826, "2": 14725.618291995488, "3": 14445.828791940585, "4": 14600.421458017081, "5": 14639.447499997914, "6": 14733.761791023426, "7": 14705.673499964178, "8": 14771.678957971744}, "receiver_prefill_ms": 286.6574580548331, "ttft_ms": 25.65283398143947, "decode_ms": 44.52579189091921}, "total_wall_ms": 125858.02104091272, "ttft_ms": 25.65283398143947, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 61, "facts_retained": 7, "retention_rate": 0.11475409836065574, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.3076923076923077, "status": 0.0, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 690}, "ledger_by_hop": {"1": {"facts_total": 61, "facts_retained": 7, "retention_rate": 0.11475409836065574, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.3076923076923077, "status": 0.0, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 690}, "2": {"facts_total": 61, "facts_retained": 7, "retention_rate": 0.11475409836065574, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.3076923076923077, "status": 0.0, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 690}, "3": {"facts_total": 61, "facts_retained": 7, "retention_rate": 0.11475409836065574, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.3076923076923077, "status": 0.0, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 690}, "4": {"facts_total": 61, "facts_retained": 7, "retention_rate": 0.11475409836065574, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.3076923076923077, "status": 0.0, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 690}, "5": {"facts_total": 61, "facts_retained": 7, "retention_rate": 0.11475409836065574, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.3076923076923077, "status": 0.0, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 690}, "6": {"facts_total": 61, "facts_retained": 7, "retention_rate": 0.11475409836065574, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.3076923076923077, "status": 0.0, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 690}, "7": {"facts_total": 61, "facts_retained": 7, "retention_rate": 0.11475409836065574, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.3076923076923077, "status": 0.0, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 690}, "8": {"facts_total": 61, "facts_retained": 7, "retention_rate": 0.11475409836065574, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.3076923076923077, "status": 0.0, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 690}}}, "reference_key_id": "keys/synthetic/syn_03015_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "e54cba3528ab78e09d5da03579b2a2035abb56934b1d0571c12acb32bf58b543", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "b267bf692e656066f39c8fad", "attempt_id": "0f803b9cdada", "run_group_id": "E4_core:A5|closed|qwen3b:h8", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:53:38.981209Z", "task_family": "synthetic", "document_id": "syn_03015_2000", "document_hash": "71a7284b9bc3f681d69d91d18b9e8d45a4dda486cd14f4145395786ac6a68a89", "question_id": "q08_revision", "split": "test", "cluster_id": "syn_03015_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 8, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1922, "query_tokens": 426, "message_tokens": 328, "message_utf8_bytes": 754, "cache_tensor_bytes": null, "serialized_payload_bytes": 754, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 125500.71437389124, "hops": {"1": 22878.284082980826, "2": 14725.618291995488, "3": 14445.828791940585, "4": 14600.421458017081, "5": 14639.447499997914, "6": 14733.761791023426, "7": 14705.673499964178, "8": 14771.678957971744}, "receiver_prefill_ms": 302.7695839991793, "ttft_ms": 20.53708303719759, "decode_ms": 20.277875009924173}, "total_wall_ms": 125844.83695786912, "ttft_ms": 20.53708303719759, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 61, "facts_retained": 7, "retention_rate": 0.11475409836065574, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.3076923076923077, "status": 0.0, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 690}, "ledger_by_hop": {"1": {"facts_total": 61, "facts_retained": 7, "retention_rate": 0.11475409836065574, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.3076923076923077, "status": 0.0, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 690}, "2": {"facts_total": 61, "facts_retained": 7, "retention_rate": 0.11475409836065574, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.3076923076923077, "status": 0.0, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 690}, "3": {"facts_total": 61, "facts_retained": 7, "retention_rate": 0.11475409836065574, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.3076923076923077, "status": 0.0, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 690}, "4": {"facts_total": 61, "facts_retained": 7, "retention_rate": 0.11475409836065574, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.3076923076923077, "status": 0.0, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 690}, "5": {"facts_total": 61, "facts_retained": 7, "retention_rate": 0.11475409836065574, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.3076923076923077, "status": 0.0, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 690}, "6": {"facts_total": 61, "facts_retained": 7, "retention_rate": 0.11475409836065574, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.3076923076923077, "status": 0.0, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 690}, "7": {"facts_total": 61, "facts_retained": 7, "retention_rate": 0.11475409836065574, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.3076923076923077, "status": 0.0, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 690}, "8": {"facts_total": 61, "facts_retained": 7, "retention_rate": 0.11475409836065574, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.3076923076923077, "status": 0.0, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 690}}}, "reference_key_id": "keys/synthetic/syn_03015_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "e54cba3528ab78e09d5da03579b2a2035abb56934b1d0571c12acb32bf58b543", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "4ad150dea68bfa5c45c10cce", "attempt_id": "20969452d78b", "run_group_id": "E4_core:A5|closed|qwen3b:h8", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:53:39.342678Z", "task_family": "synthetic", "document_id": "syn_03015_2000", "document_hash": "71a7284b9bc3f681d69d91d18b9e8d45a4dda486cd14f4145395786ac6a68a89", "question_id": "q09_revision", "split": "test", "cluster_id": "syn_03015_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 8, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1922, "query_tokens": 420, "message_tokens": 328, "message_utf8_bytes": 754, "cache_tensor_bytes": null, "serialized_payload_bytes": 754, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 125500.71437389124, "hops": {"1": 22878.284082980826, "2": 14725.618291995488, "3": 14445.828791940585, "4": 14600.421458017081, "5": 14639.447499997914, "6": 14733.761791023426, "7": 14705.673499964178, "8": 14771.678957971744}, "receiver_prefill_ms": 277.63495792169124, "ttft_ms": 42.224957956932485, "decode_ms": 435.1280829869211}, "total_wall_ms": 126256.31104083732, "ttft_ms": 42.224957956932485, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 7}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "7028925", "evidence_artifact": {"ledger": {"facts_total": 61, "facts_retained": 7, "retention_rate": 0.11475409836065574, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.3076923076923077, "status": 0.0, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 690}, "ledger_by_hop": {"1": {"facts_total": 61, "facts_retained": 7, "retention_rate": 0.11475409836065574, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.3076923076923077, "status": 0.0, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 690}, "2": {"facts_total": 61, "facts_retained": 7, "retention_rate": 0.11475409836065574, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.3076923076923077, "status": 0.0, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 690}, "3": {"facts_total": 61, "facts_retained": 7, "retention_rate": 0.11475409836065574, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.3076923076923077, "status": 0.0, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 690}, "4": {"facts_total": 61, "facts_retained": 7, "retention_rate": 0.11475409836065574, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.3076923076923077, "status": 0.0, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 690}, "5": {"facts_total": 61, "facts_retained": 7, "retention_rate": 0.11475409836065574, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.3076923076923077, "status": 0.0, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 690}, "6": {"facts_total": 61, "facts_retained": 7, "retention_rate": 0.11475409836065574, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.3076923076923077, "status": 0.0, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 690}, "7": {"facts_total": 61, "facts_retained": 7, "retention_rate": 0.11475409836065574, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.3076923076923077, "status": 0.0, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 690}, "8": {"facts_total": 61, "facts_retained": 7, "retention_rate": 0.11475409836065574, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.3076923076923077, "status": 0.0, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 690}}}, "reference_key_id": "keys/synthetic/syn_03015_2000.json", "task_scores": {"correct": 0.0, "abstained": 0.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "e54cba3528ab78e09d5da03579b2a2035abb56934b1d0571c12acb32bf58b543", "output_hash": "e966a398855c104a86bce07742c3ce8108a432926bc704940ecfd07e44596774"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "ee51bb8db5a8265829b11746", "attempt_id": "ead61db61c6b", "run_group_id": "E4_core:A5|closed|qwen3b:h8", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:53:40.118130Z", "task_family": "synthetic", "document_id": "syn_03015_2000", "document_hash": "71a7284b9bc3f681d69d91d18b9e8d45a4dda486cd14f4145395786ac6a68a89", "question_id": "q10_temporal", "split": "test", "cluster_id": "syn_03015_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 8, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1922, "query_tokens": 429, "message_tokens": 328, "message_utf8_bytes": 754, "cache_tensor_bytes": null, "serialized_payload_bytes": 754, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 125500.71437389124, "hops": {"1": 22878.284082980826, "2": 14725.618291995488, "3": 14445.828791940585, "4": 14600.421458017081, "5": 14639.447499997914, "6": 14733.761791023426, "7": 14705.673499964178, "8": 14771.678957971744}, "receiver_prefill_ms": 306.96412501856685, "ttft_ms": 26.71350003220141, "decode_ms": 36.100583034567535}, "total_wall_ms": 125870.91495795175, "ttft_ms": 26.71350003220141, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 61, "facts_retained": 7, "retention_rate": 0.11475409836065574, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.3076923076923077, "status": 0.0, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 690}, "ledger_by_hop": {"1": {"facts_total": 61, "facts_retained": 7, "retention_rate": 0.11475409836065574, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.3076923076923077, "status": 0.0, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 690}, "2": {"facts_total": 61, "facts_retained": 7, "retention_rate": 0.11475409836065574, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.3076923076923077, "status": 0.0, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 690}, "3": {"facts_total": 61, "facts_retained": 7, "retention_rate": 0.11475409836065574, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.3076923076923077, "status": 0.0, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 690}, "4": {"facts_total": 61, "facts_retained": 7, "retention_rate": 0.11475409836065574, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.3076923076923077, "status": 0.0, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 690}, "5": {"facts_total": 61, "facts_retained": 7, "retention_rate": 0.11475409836065574, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.3076923076923077, "status": 0.0, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 690}, "6": {"facts_total": 61, "facts_retained": 7, "retention_rate": 0.11475409836065574, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.3076923076923077, "status": 0.0, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 690}, "7": {"facts_total": 61, "facts_retained": 7, "retention_rate": 0.11475409836065574, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.3076923076923077, "status": 0.0, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 690}, "8": {"facts_total": 61, "facts_retained": 7, "retention_rate": 0.11475409836065574, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.3076923076923077, "status": 0.0, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 690}}}, "reference_key_id": "keys/synthetic/syn_03015_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "e54cba3528ab78e09d5da03579b2a2035abb56934b1d0571c12acb32bf58b543", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "3674c60c901e8fcbc64fd544", "attempt_id": "e71d5d8d65e4", "run_group_id": "E4_core:A5|closed|qwen3b:h8", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:53:40.505248Z", "task_family": "synthetic", "document_id": "syn_03015_2000", "document_hash": "71a7284b9bc3f681d69d91d18b9e8d45a4dda486cd14f4145395786ac6a68a89", "question_id": "q11_aggregation", "split": "test", "cluster_id": "syn_03015_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 8, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1922, "query_tokens": 422, "message_tokens": 328, "message_utf8_bytes": 754, "cache_tensor_bytes": null, "serialized_payload_bytes": 754, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 125500.71437389124, "hops": {"1": 22878.284082980826, "2": 14725.618291995488, "3": 14445.828791940585, "4": 14600.421458017081, "5": 14639.447499997914, "6": 14733.761791023426, "7": 14705.673499964178, "8": 14771.678957971744}, "receiver_prefill_ms": 301.1267079273239, "ttft_ms": 19.115292001515627, "decode_ms": 37.40124998148531}, "total_wall_ms": 125858.95353986416, "ttft_ms": 19.115292001515627, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 61, "facts_retained": 7, "retention_rate": 0.11475409836065574, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.3076923076923077, "status": 0.0, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 690}, "ledger_by_hop": {"1": {"facts_total": 61, "facts_retained": 7, "retention_rate": 0.11475409836065574, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.3076923076923077, "status": 0.0, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 690}, "2": {"facts_total": 61, "facts_retained": 7, "retention_rate": 0.11475409836065574, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.3076923076923077, "status": 0.0, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 690}, "3": {"facts_total": 61, "facts_retained": 7, "retention_rate": 0.11475409836065574, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.3076923076923077, "status": 0.0, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 690}, "4": {"facts_total": 61, "facts_retained": 7, "retention_rate": 0.11475409836065574, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.3076923076923077, "status": 0.0, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 690}, "5": {"facts_total": 61, "facts_retained": 7, "retention_rate": 0.11475409836065574, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.3076923076923077, "status": 0.0, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 690}, "6": {"facts_total": 61, "facts_retained": 7, "retention_rate": 0.11475409836065574, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.3076923076923077, "status": 0.0, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 690}, "7": {"facts_total": 61, "facts_retained": 7, "retention_rate": 0.11475409836065574, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.3076923076923077, "status": 0.0, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 690}, "8": {"facts_total": 61, "facts_retained": 7, "retention_rate": 0.11475409836065574, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.3076923076923077, "status": 0.0, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 690}}}, "reference_key_id": "keys/synthetic/syn_03015_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "e54cba3528ab78e09d5da03579b2a2035abb56934b1d0571c12acb32bf58b543", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "294ae3e673fd912a249cc65c", "attempt_id": "365f2b718879", "run_group_id": "E4_core:A5|closed|qwen3b:h8", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:53:40.880929Z", "task_family": "synthetic", "document_id": "syn_03015_2000", "document_hash": "71a7284b9bc3f681d69d91d18b9e8d45a4dda486cd14f4145395786ac6a68a89", "question_id": "q12_unit_conversion", "split": "test", "cluster_id": "syn_03015_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 8, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1922, "query_tokens": 419, "message_tokens": 328, "message_utf8_bytes": 754, "cache_tensor_bytes": null, "serialized_payload_bytes": 754, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 125500.71437389124, "hops": {"1": 22878.284082980826, "2": 14725.618291995488, "3": 14445.828791940585, "4": 14600.421458017081, "5": 14639.447499997914, "6": 14733.761791023426, "7": 14705.673499964178, "8": 14771.678957971744}, "receiver_prefill_ms": 331.100125098601, "ttft_ms": 43.495040968991816, "decode_ms": 66.696000052616}, "total_wall_ms": 125942.62787385378, "ttft_ms": 43.495040968991816, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 61, "facts_retained": 7, "retention_rate": 0.11475409836065574, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.3076923076923077, "status": 0.0, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 690}, "ledger_by_hop": {"1": {"facts_total": 61, "facts_retained": 7, "retention_rate": 0.11475409836065574, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.3076923076923077, "status": 0.0, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 690}, "2": {"facts_total": 61, "facts_retained": 7, "retention_rate": 0.11475409836065574, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.3076923076923077, "status": 0.0, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 690}, "3": {"facts_total": 61, "facts_retained": 7, "retention_rate": 0.11475409836065574, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.3076923076923077, "status": 0.0, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 690}, "4": {"facts_total": 61, "facts_retained": 7, "retention_rate": 0.11475409836065574, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.3076923076923077, "status": 0.0, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 690}, "5": {"facts_total": 61, "facts_retained": 7, "retention_rate": 0.11475409836065574, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.3076923076923077, "status": 0.0, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 690}, "6": {"facts_total": 61, "facts_retained": 7, "retention_rate": 0.11475409836065574, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.3076923076923077, "status": 0.0, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 690}, "7": {"facts_total": 61, "facts_retained": 7, "retention_rate": 0.11475409836065574, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.3076923076923077, "status": 0.0, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 690}, "8": {"facts_total": 61, "facts_retained": 7, "retention_rate": 0.11475409836065574, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.3076923076923077, "status": 0.0, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 690}}}, "reference_key_id": "keys/synthetic/syn_03015_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "e54cba3528ab78e09d5da03579b2a2035abb56934b1d0571c12acb32bf58b543", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "139f5d89f648430e71da4224", "attempt_id": "8c610de2bd69", "run_group_id": "E4_core:A5|closed|qwen3b:h8", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:53:41.339256Z", "task_family": "synthetic", "document_id": "syn_03015_2000", "document_hash": "71a7284b9bc3f681d69d91d18b9e8d45a4dda486cd14f4145395786ac6a68a89", "question_id": "q13_unanswerable", "split": "test", "cluster_id": "syn_03015_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 8, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1922, "query_tokens": 423, "message_tokens": 328, "message_utf8_bytes": 754, "cache_tensor_bytes": null, "serialized_payload_bytes": 754, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 125500.71437389124, "hops": {"1": 22878.284082980826, "2": 14725.618291995488, "3": 14445.828791940585, "4": 14600.421458017081, "5": 14639.447499997914, "6": 14733.761791023426, "7": 14705.673499964178, "8": 14771.678957971744}, "receiver_prefill_ms": 421.7364169890061, "ttft_ms": 37.39016701001674, "decode_ms": 28.890624991618097}, "total_wall_ms": 125989.26499893423, "ttft_ms": 37.39016701001674, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 61, "facts_retained": 7, "retention_rate": 0.11475409836065574, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.3076923076923077, "status": 0.0, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 690}, "ledger_by_hop": {"1": {"facts_total": 61, "facts_retained": 7, "retention_rate": 0.11475409836065574, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.3076923076923077, "status": 0.0, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 690}, "2": {"facts_total": 61, "facts_retained": 7, "retention_rate": 0.11475409836065574, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.3076923076923077, "status": 0.0, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 690}, "3": {"facts_total": 61, "facts_retained": 7, "retention_rate": 0.11475409836065574, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.3076923076923077, "status": 0.0, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 690}, "4": {"facts_total": 61, "facts_retained": 7, "retention_rate": 0.11475409836065574, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.3076923076923077, "status": 0.0, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 690}, "5": {"facts_total": 61, "facts_retained": 7, "retention_rate": 0.11475409836065574, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.3076923076923077, "status": 0.0, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 690}, "6": {"facts_total": 61, "facts_retained": 7, "retention_rate": 0.11475409836065574, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.3076923076923077, "status": 0.0, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 690}, "7": {"facts_total": 61, "facts_retained": 7, "retention_rate": 0.11475409836065574, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.3076923076923077, "status": 0.0, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 690}, "8": {"facts_total": 61, "facts_retained": 7, "retention_rate": 0.11475409836065574, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.3076923076923077, "status": 0.0, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 690}}}, "reference_key_id": "keys/synthetic/syn_03015_2000.json", "task_scores": {"correct": 1.0, "abstained": 1.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "e54cba3528ab78e09d5da03579b2a2035abb56934b1d0571c12acb32bf58b543", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "5822085e29f5afccc022c9fb", "attempt_id": "9bdbaa365ae4", "run_group_id": "E4_core:A5|closed|qwen3b:h8", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:53:41.845898Z", "task_family": "synthetic", "document_id": "syn_03015_2000", "document_hash": "71a7284b9bc3f681d69d91d18b9e8d45a4dda486cd14f4145395786ac6a68a89", "question_id": "q14_unanswerable", "split": "test", "cluster_id": "syn_03015_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 8, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1922, "query_tokens": 424, "message_tokens": 328, "message_utf8_bytes": 754, "cache_tensor_bytes": null, "serialized_payload_bytes": 754, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 125500.71437389124, "hops": {"1": 22878.284082980826, "2": 14725.618291995488, "3": 14445.828791940585, "4": 14600.421458017081, "5": 14639.447499997914, "6": 14733.761791023426, "7": 14705.673499964178, "8": 14771.678957971744}, "receiver_prefill_ms": 286.5600000368431, "ttft_ms": 35.6331670191139, "decode_ms": 37.59458300191909}, "total_wall_ms": 125860.93133187387, "ttft_ms": 35.6331670191139, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 61, "facts_retained": 7, "retention_rate": 0.11475409836065574, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.3076923076923077, "status": 0.0, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 690}, "ledger_by_hop": {"1": {"facts_total": 61, "facts_retained": 7, "retention_rate": 0.11475409836065574, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.3076923076923077, "status": 0.0, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 690}, "2": {"facts_total": 61, "facts_retained": 7, "retention_rate": 0.11475409836065574, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.3076923076923077, "status": 0.0, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 690}, "3": {"facts_total": 61, "facts_retained": 7, "retention_rate": 0.11475409836065574, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.3076923076923077, "status": 0.0, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 690}, "4": {"facts_total": 61, "facts_retained": 7, "retention_rate": 0.11475409836065574, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.3076923076923077, "status": 0.0, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 690}, "5": {"facts_total": 61, "facts_retained": 7, "retention_rate": 0.11475409836065574, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.3076923076923077, "status": 0.0, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 690}, "6": {"facts_total": 61, "facts_retained": 7, "retention_rate": 0.11475409836065574, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.3076923076923077, "status": 0.0, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 690}, "7": {"facts_total": 61, "facts_retained": 7, "retention_rate": 0.11475409836065574, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.3076923076923077, "status": 0.0, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 690}, "8": {"facts_total": 61, "facts_retained": 7, "retention_rate": 0.11475409836065574, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.3076923076923077, "status": 0.0, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 690}}}, "reference_key_id": "keys/synthetic/syn_03015_2000.json", "task_scores": {"correct": 1.0, "abstained": 1.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "e54cba3528ab78e09d5da03579b2a2035abb56934b1d0571c12acb32bf58b543", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "cd990de65a495b1d9d6cb846", "attempt_id": "0e3c6f5a31ac", "run_group_id": "E4_core:A4|repairable|qwen3b:h1", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:54:04.963001Z", "task_family": "synthetic", "document_id": "syn_03015_2000", "document_hash": "71a7284b9bc3f681d69d91d18b9e8d45a4dda486cd14f4145395786ac6a68a89", "question_id": "q00_single_fact", "split": "test", "cluster_id": "syn_03015_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 1, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1922, "query_tokens": 595, "message_tokens": 512, "message_utf8_bytes": 1161, "cache_tensor_bytes": null, "serialized_payload_bytes": 1161, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 22737.68987506628, "hops": {"1": 22737.68987506628}, "receiver_prefill_ms": 538.0818330449983, "ttft_ms": 38.240291993133724, "decode_ms": 25.381000014021993}, "total_wall_ms": 23340.109458076768, "ttft_ms": 38.240291993133724, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 61, "facts_retained": 24, "retention_rate": 0.39344262295081966, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.4166666666666667, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1161}, "ledger_by_hop": {"1": {"facts_total": 61, "facts_retained": 24, "retention_rate": 0.39344262295081966, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.4166666666666667, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1161}}}, "reference_key_id": "keys/synthetic/syn_03015_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "52d2d6723cbb6f8139ce72290860dae098b28331488ef44459da0ce2ae9f9f72", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "68ab86f632659db1cabf6880", "attempt_id": "ab37bbcd80de", "run_group_id": "E4_core:A4|repairable|qwen3b:h1", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:54:05.583611Z", "task_family": "synthetic", "document_id": "syn_03015_2000", "document_hash": "71a7284b9bc3f681d69d91d18b9e8d45a4dda486cd14f4145395786ac6a68a89", "question_id": "q01_single_fact", "split": "test", "cluster_id": "syn_03015_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 1, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1922, "query_tokens": 600, "message_tokens": 512, "message_utf8_bytes": 1161, "cache_tensor_bytes": null, "serialized_payload_bytes": 1161, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 22737.68987506628, "hops": {"1": 22737.68987506628}, "receiver_prefill_ms": 404.68158398289233, "ttft_ms": 27.220457908697426, "decode_ms": 36.32345888763666}, "total_wall_ms": 23206.611833069474, "ttft_ms": 27.220457908697426, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 61, "facts_retained": 24, "retention_rate": 0.39344262295081966, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.4166666666666667, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1161}, "ledger_by_hop": {"1": {"facts_total": 61, "facts_retained": 24, "retention_rate": 0.39344262295081966, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.4166666666666667, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1161}}}, "reference_key_id": "keys/synthetic/syn_03015_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "52d2d6723cbb6f8139ce72290860dae098b28331488ef44459da0ce2ae9f9f72", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "cc65125ce598f276d90a43eb", "attempt_id": "936afe0ec7ec", "run_group_id": "E4_core:A4|repairable|qwen3b:h1", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:54:06.070001Z", "task_family": "synthetic", "document_id": "syn_03015_2000", "document_hash": "71a7284b9bc3f681d69d91d18b9e8d45a4dda486cd14f4145395786ac6a68a89", "question_id": "q02_exact_copy", "split": "test", "cluster_id": "syn_03015_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 1, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1922, "query_tokens": 599, "message_tokens": 512, "message_utf8_bytes": 1161, "cache_tensor_bytes": null, "serialized_payload_bytes": 1161, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 22737.68987506628, "hops": {"1": 22737.68987506628}, "receiver_prefill_ms": 345.83070897497237, "ttft_ms": 38.886790978722274, "decode_ms": 539.4316670717672}, "total_wall_ms": 23662.558790994808, "ttft_ms": 38.886790978722274, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 7}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "89.3122", "evidence_artifact": {"ledger": {"facts_total": 61, "facts_retained": 24, "retention_rate": 0.39344262295081966, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.4166666666666667, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1161}, "ledger_by_hop": {"1": {"facts_total": 61, "facts_retained": 24, "retention_rate": 0.39344262295081966, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.4166666666666667, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1161}}}, "reference_key_id": "keys/synthetic/syn_03015_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "52d2d6723cbb6f8139ce72290860dae098b28331488ef44459da0ce2ae9f9f72", "output_hash": "2138888e6519b2f4b1b54142e6f9c541102d7f9ee5bcf8e4184c81a382601757"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "694e4e0e47d0a2a83d8763ba", "attempt_id": "ac690a8eb206", "run_group_id": "E4_core:A4|repairable|qwen3b:h1", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:54:07.013170Z", "task_family": "synthetic", "document_id": "syn_03015_2000", "document_hash": "71a7284b9bc3f681d69d91d18b9e8d45a4dda486cd14f4145395786ac6a68a89", "question_id": "q03_exact_copy", "split": "test", "cluster_id": "syn_03015_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 1, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1922, "query_tokens": 595, "message_tokens": 512, "message_utf8_bytes": 1161, "cache_tensor_bytes": null, "serialized_payload_bytes": 1161, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 22737.68987506628, "hops": {"1": 22737.68987506628}, "receiver_prefill_ms": 397.4582910304889, "ttft_ms": 41.276374948211014, "decode_ms": 192.86033301614225}, "total_wall_ms": 23370.076459017582, "ttft_ms": 41.276374948211014, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 6}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "Wilhelmina Dubrovnik", "evidence_artifact": {"ledger": {"facts_total": 61, "facts_retained": 24, "retention_rate": 0.39344262295081966, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.4166666666666667, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1161}, "ledger_by_hop": {"1": {"facts_total": 61, "facts_retained": 24, "retention_rate": 0.39344262295081966, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.4166666666666667, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1161}}}, "reference_key_id": "keys/synthetic/syn_03015_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "52d2d6723cbb6f8139ce72290860dae098b28331488ef44459da0ce2ae9f9f72", "output_hash": "f42f05ebfc96f7258cbfbbb37cd65e114d3d15f869e68e6c95b71fc288c30a32"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "4036702c636ed27bb6805626", "attempt_id": "d746f2de2c6c", "run_group_id": "E4_core:A4|repairable|qwen3b:h1", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:54:07.662486Z", "task_family": "synthetic", "document_id": "syn_03015_2000", "document_hash": "71a7284b9bc3f681d69d91d18b9e8d45a4dda486cd14f4145395786ac6a68a89", "question_id": "q04_multi_hop", "split": "test", "cluster_id": "syn_03015_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 1, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1922, "query_tokens": 604, "message_tokens": 512, "message_utf8_bytes": 1161, "cache_tensor_bytes": null, "serialized_payload_bytes": 1161, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 22737.68987506628, "hops": {"1": 22737.68987506628}, "receiver_prefill_ms": 363.53312502615154, "ttft_ms": 53.72708407230675, "decode_ms": 506.29054103046656}, "total_wall_ms": 23661.95283306297, "ttft_ms": 53.72708407230675, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 7}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "8421725", "evidence_artifact": {"ledger": {"facts_total": 61, "facts_retained": 24, "retention_rate": 0.39344262295081966, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.4166666666666667, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1161}, "ledger_by_hop": {"1": {"facts_total": 61, "facts_retained": 24, "retention_rate": 0.39344262295081966, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.4166666666666667, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1161}}}, "reference_key_id": "keys/synthetic/syn_03015_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "52d2d6723cbb6f8139ce72290860dae098b28331488ef44459da0ce2ae9f9f72", "output_hash": "e4949aba3b0b77de05a87edaab854329fa75df7bf68b29a3def71e985a06e5b5"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "cb94cad2d97efcd529398635", "attempt_id": "b4e70bfa40f0", "run_group_id": "E4_core:A4|repairable|qwen3b:h1", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:54:08.605183Z", "task_family": "synthetic", "document_id": "syn_03015_2000", "document_hash": "71a7284b9bc3f681d69d91d18b9e8d45a4dda486cd14f4145395786ac6a68a89", "question_id": "q05_multi_hop", "split": "test", "cluster_id": "syn_03015_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 1, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1922, "query_tokens": 604, "message_tokens": 512, "message_utf8_bytes": 1161, "cache_tensor_bytes": null, "serialized_payload_bytes": 1161, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 22737.68987506628, "hops": {"1": 22737.68987506628}, "receiver_prefill_ms": 400.41833301074803, "ttft_ms": 40.954750031232834, "decode_ms": 23.904707981273532}, "total_wall_ms": 23203.539000125602, "ttft_ms": 40.954750031232834, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 61, "facts_retained": 24, "retention_rate": 0.39344262295081966, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.4166666666666667, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1161}, "ledger_by_hop": {"1": {"facts_total": 61, "facts_retained": 24, "retention_rate": 0.39344262295081966, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.4166666666666667, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1161}}}, "reference_key_id": "keys/synthetic/syn_03015_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "52d2d6723cbb6f8139ce72290860dae098b28331488ef44459da0ce2ae9f9f72", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "29ff8b68611d9c6db2e9bb65", "attempt_id": "548e8ef4c3b8", "run_group_id": "E4_core:A4|repairable|qwen3b:h1", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:54:09.087928Z", "task_family": "synthetic", "document_id": "syn_03015_2000", "document_hash": "71a7284b9bc3f681d69d91d18b9e8d45a4dda486cd14f4145395786ac6a68a89", "question_id": "q06_negation", "split": "test", "cluster_id": "syn_03015_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 1, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1922, "query_tokens": 596, "message_tokens": 512, "message_utf8_bytes": 1161, "cache_tensor_bytes": null, "serialized_payload_bytes": 1161, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 22737.68987506628, "hops": {"1": 22737.68987506628}, "receiver_prefill_ms": 394.40104202367365, "ttft_ms": 21.976249990984797, "decode_ms": 41.82675003539771}, "total_wall_ms": 23196.485624997877, "ttft_ms": 21.976249990984797, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 61, "facts_retained": 24, "retention_rate": 0.39344262295081966, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.4166666666666667, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1161}, "ledger_by_hop": {"1": {"facts_total": 61, "facts_retained": 24, "retention_rate": 0.39344262295081966, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.4166666666666667, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1161}}}, "reference_key_id": "keys/synthetic/syn_03015_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "52d2d6723cbb6f8139ce72290860dae098b28331488ef44459da0ce2ae9f9f72", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "befdfaf42cb6ad33315ebb3b", "attempt_id": "c864856603de", "run_group_id": "E4_core:A4|repairable|qwen3b:h1", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:54:09.564097Z", "task_family": "synthetic", "document_id": "syn_03015_2000", "document_hash": "71a7284b9bc3f681d69d91d18b9e8d45a4dda486cd14f4145395786ac6a68a89", "question_id": "q07_negation", "split": "test", "cluster_id": "syn_03015_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 1, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1922, "query_tokens": 596, "message_tokens": 512, "message_utf8_bytes": 1161, "cache_tensor_bytes": null, "serialized_payload_bytes": 1161, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 22737.68987506628, "hops": {"1": 22737.68987506628}, "receiver_prefill_ms": 369.1638340242207, "ttft_ms": 68.92175006214529, "decode_ms": 51.5722919953987}, "total_wall_ms": 23228.203625069, "ttft_ms": 68.92175006214529, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 61, "facts_retained": 24, "retention_rate": 0.39344262295081966, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.4166666666666667, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1161}, "ledger_by_hop": {"1": {"facts_total": 61, "facts_retained": 24, "retention_rate": 0.39344262295081966, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.4166666666666667, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1161}}}, "reference_key_id": "keys/synthetic/syn_03015_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "52d2d6723cbb6f8139ce72290860dae098b28331488ef44459da0ce2ae9f9f72", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "9a4bd47c2671e7a9a8421436", "attempt_id": "faea4e2585bb", "run_group_id": "E4_core:A4|repairable|qwen3b:h1", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:54:10.072401Z", "task_family": "synthetic", "document_id": "syn_03015_2000", "document_hash": "71a7284b9bc3f681d69d91d18b9e8d45a4dda486cd14f4145395786ac6a68a89", "question_id": "q08_revision", "split": "test", "cluster_id": "syn_03015_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 1, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1922, "query_tokens": 610, "message_tokens": 512, "message_utf8_bytes": 1161, "cache_tensor_bytes": null, "serialized_payload_bytes": 1161, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 22737.68987506628, "hops": {"1": 22737.68987506628}, "receiver_prefill_ms": 537.3046250315383, "ttft_ms": 38.858041050843894, "decode_ms": 26.705000083893538}, "total_wall_ms": 23341.267541982234, "ttft_ms": 38.858041050843894, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 61, "facts_retained": 24, "retention_rate": 0.39344262295081966, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.4166666666666667, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1161}, "ledger_by_hop": {"1": {"facts_total": 61, "facts_retained": 24, "retention_rate": 0.39344262295081966, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.4166666666666667, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1161}}}, "reference_key_id": "keys/synthetic/syn_03015_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "52d2d6723cbb6f8139ce72290860dae098b28331488ef44459da0ce2ae9f9f72", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "7521477a3cc6907ee755d603", "attempt_id": "37ab657fb5da", "run_group_id": "E4_core:A4|repairable|qwen3b:h1", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:54:10.693406Z", "task_family": "synthetic", "document_id": "syn_03015_2000", "document_hash": "71a7284b9bc3f681d69d91d18b9e8d45a4dda486cd14f4145395786ac6a68a89", "question_id": "q09_revision", "split": "test", "cluster_id": "syn_03015_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 1, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1922, "query_tokens": 604, "message_tokens": 512, "message_utf8_bytes": 1161, "cache_tensor_bytes": null, "serialized_payload_bytes": 1161, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 22737.68987506628, "hops": {"1": 22737.68987506628}, "receiver_prefill_ms": 403.092332999222, "ttft_ms": 43.84366690646857, "decode_ms": 320.73458295781165}, "total_wall_ms": 23506.13941613119, "ttft_ms": 43.84366690646857, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 9}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "8,377,480", "evidence_artifact": {"ledger": {"facts_total": 61, "facts_retained": 24, "retention_rate": 0.39344262295081966, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.4166666666666667, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1161}, "ledger_by_hop": {"1": {"facts_total": 61, "facts_retained": 24, "retention_rate": 0.39344262295081966, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.4166666666666667, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1161}}}, "reference_key_id": "keys/synthetic/syn_03015_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "52d2d6723cbb6f8139ce72290860dae098b28331488ef44459da0ce2ae9f9f72", "output_hash": "ec4c68b327a2d2427ed21821415a5f9ddf87134bd9f654bdd69c7d1301febcb0"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "d6f326894bd955742fe420ce", "attempt_id": "2bbc8db26f2d", "run_group_id": "E4_core:A4|repairable|qwen3b:h1", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:54:11.479792Z", "task_family": "synthetic", "document_id": "syn_03015_2000", "document_hash": "71a7284b9bc3f681d69d91d18b9e8d45a4dda486cd14f4145395786ac6a68a89", "question_id": "q10_temporal", "split": "test", "cluster_id": "syn_03015_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 1, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1922, "query_tokens": 613, "message_tokens": 512, "message_utf8_bytes": 1161, "cache_tensor_bytes": null, "serialized_payload_bytes": 1161, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 22737.68987506628, "hops": {"1": 22737.68987506628}, "receiver_prefill_ms": 400.3194581018761, "ttft_ms": 65.97920903004706, "decode_ms": 108.49733301438391}, "total_wall_ms": 23313.169542117976, "ttft_ms": 65.97920903004706, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 61, "facts_retained": 24, "retention_rate": 0.39344262295081966, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.4166666666666667, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1161}, "ledger_by_hop": {"1": {"facts_total": 61, "facts_retained": 24, "retention_rate": 0.39344262295081966, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.4166666666666667, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1161}}}, "reference_key_id": "keys/synthetic/syn_03015_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "52d2d6723cbb6f8139ce72290860dae098b28331488ef44459da0ce2ae9f9f72", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "147f4b2fd4590c1cceb001b5", "attempt_id": "5e8ca3a1cba5", "run_group_id": "E4_core:A4|repairable|qwen3b:h1", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:54:12.072738Z", "task_family": "synthetic", "document_id": "syn_03015_2000", "document_hash": "71a7284b9bc3f681d69d91d18b9e8d45a4dda486cd14f4145395786ac6a68a89", "question_id": "q11_aggregation", "split": "test", "cluster_id": "syn_03015_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 1, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1922, "query_tokens": 606, "message_tokens": 512, "message_utf8_bytes": 1161, "cache_tensor_bytes": null, "serialized_payload_bytes": 1161, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 22737.68987506628, "hops": {"1": 22737.68987506628}, "receiver_prefill_ms": 394.9310410534963, "ttft_ms": 26.323791011236608, "decode_ms": 232.094542006962}, "total_wall_ms": 23391.654542065226, "ttft_ms": 26.323791011236608, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 7}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "1806025", "evidence_artifact": {"ledger": {"facts_total": 61, "facts_retained": 24, "retention_rate": 0.39344262295081966, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.4166666666666667, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1161}, "ledger_by_hop": {"1": {"facts_total": 61, "facts_retained": 24, "retention_rate": 0.39344262295081966, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.4166666666666667, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1161}}}, "reference_key_id": "keys/synthetic/syn_03015_2000.json", "task_scores": {"correct": 0.0, "abstained": 0.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "52d2d6723cbb6f8139ce72290860dae098b28331488ef44459da0ce2ae9f9f72", "output_hash": "9801d1840792d286693349934c9cc13fead44d82d480a42803cf6015c01431f6"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "ac9200521d9ec0a803a410ac", "attempt_id": "58e88285fa18", "run_group_id": "E4_core:A4|repairable|qwen3b:h1", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:54:12.743813Z", "task_family": "synthetic", "document_id": "syn_03015_2000", "document_hash": "71a7284b9bc3f681d69d91d18b9e8d45a4dda486cd14f4145395786ac6a68a89", "question_id": "q12_unit_conversion", "split": "test", "cluster_id": "syn_03015_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 1, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1922, "query_tokens": 603, "message_tokens": 512, "message_utf8_bytes": 1161, "cache_tensor_bytes": null, "serialized_payload_bytes": 1161, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 22737.68987506628, "hops": {"1": 22737.68987506628}, "receiver_prefill_ms": 364.1763749765232, "ttft_ms": 46.293415944091976, "decode_ms": 52.79545800294727}, "total_wall_ms": 23201.790833030827, "ttft_ms": 46.293415944091976, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 61, "facts_retained": 24, "retention_rate": 0.39344262295081966, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.4166666666666667, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1161}, "ledger_by_hop": {"1": {"facts_total": 61, "facts_retained": 24, "retention_rate": 0.39344262295081966, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.4166666666666667, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1161}}}, "reference_key_id": "keys/synthetic/syn_03015_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "52d2d6723cbb6f8139ce72290860dae098b28331488ef44459da0ce2ae9f9f72", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "ccde4701f58360b2b0a23515", "attempt_id": "b490ea1afd94", "run_group_id": "E4_core:A4|repairable|qwen3b:h1", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:54:13.225638Z", "task_family": "synthetic", "document_id": "syn_03015_2000", "document_hash": "71a7284b9bc3f681d69d91d18b9e8d45a4dda486cd14f4145395786ac6a68a89", "question_id": "q13_unanswerable", "split": "test", "cluster_id": "syn_03015_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 1, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1922, "query_tokens": 607, "message_tokens": 512, "message_utf8_bytes": 1161, "cache_tensor_bytes": null, "serialized_payload_bytes": 1161, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 22737.68987506628, "hops": {"1": 22737.68987506628}, "receiver_prefill_ms": 416.07558296527714, "ttft_ms": 59.2416669242084, "decode_ms": 112.56708297878504}, "total_wall_ms": 23326.334666111507, "ttft_ms": 59.2416669242084, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 61, "facts_retained": 24, "retention_rate": 0.39344262295081966, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.4166666666666667, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1161}, "ledger_by_hop": {"1": {"facts_total": 61, "facts_retained": 24, "retention_rate": 0.39344262295081966, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.4166666666666667, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1161}}}, "reference_key_id": "keys/synthetic/syn_03015_2000.json", "task_scores": {"correct": 1.0, "abstained": 1.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "52d2d6723cbb6f8139ce72290860dae098b28331488ef44459da0ce2ae9f9f72", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "9b17e069155d135b8f45c561", "attempt_id": "7ee4ad49606a", "run_group_id": "E4_core:A4|repairable|qwen3b:h1", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:54:13.830929Z", "task_family": "synthetic", "document_id": "syn_03015_2000", "document_hash": "71a7284b9bc3f681d69d91d18b9e8d45a4dda486cd14f4145395786ac6a68a89", "question_id": "q14_unanswerable", "split": "test", "cluster_id": "syn_03015_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 1, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1922, "query_tokens": 608, "message_tokens": 512, "message_utf8_bytes": 1161, "cache_tensor_bytes": null, "serialized_payload_bytes": 1161, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 22737.68987506628, "hops": {"1": 22737.68987506628}, "receiver_prefill_ms": 391.5591250406578, "ttft_ms": 41.78683296777308, "decode_ms": 22.82695798203349}, "total_wall_ms": 23194.543958059512, "ttft_ms": 41.78683296777308, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 61, "facts_retained": 24, "retention_rate": 0.39344262295081966, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.4166666666666667, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1161}, "ledger_by_hop": {"1": {"facts_total": 61, "facts_retained": 24, "retention_rate": 0.39344262295081966, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.4166666666666667, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1161}}}, "reference_key_id": "keys/synthetic/syn_03015_2000.json", "task_scores": {"correct": 1.0, "abstained": 1.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "52d2d6723cbb6f8139ce72290860dae098b28331488ef44459da0ce2ae9f9f72", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "2f24ad36902e14f987a679b7", "attempt_id": "a5bf5c975588", "run_group_id": "E4_core:A4|repairable|qwen3b:h2", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:54:35.368286Z", "task_family": "synthetic", "document_id": "syn_03015_2000", "document_hash": "71a7284b9bc3f681d69d91d18b9e8d45a4dda486cd14f4145395786ac6a68a89", "question_id": "q00_single_fact", "split": "test", "cluster_id": "syn_03015_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 2, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1922, "query_tokens": 568, "message_tokens": 486, "message_utf8_bytes": 1045, "cache_tensor_bytes": null, "serialized_payload_bytes": 1045, "side_information_bytes": 0, "fetched_bytes": 888, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 43798.324792063795, "hops": {"1": 22737.68987506628, "2": 21060.634916997515}, "receiver_prefill_ms": 402.79966697562486, "ttft_ms": 137.09208311047405, "decode_ms": 41.8155409861356}, "total_wall_ms": 44380.91412500944, "ttft_ms": 137.09208311047405, "input_usage": {"lookups": 4}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 61, "facts_retained": 24, "retention_rate": 0.39344262295081966, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.4166666666666667, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1045}, "ledger_by_hop": {"1": {"facts_total": 61, "facts_retained": 24, "retention_rate": 0.39344262295081966, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.4166666666666667, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1161}, "2": {"facts_total": 61, "facts_retained": 24, "retention_rate": 0.39344262295081966, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.4166666666666667, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1045}}}, "reference_key_id": "keys/synthetic/syn_03015_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "e8b0cdaa0bfa2267f9b579e9fa68a2f8c2db666e848f0095eb6c897c6e7b372d", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "c1bc4847412abfdb9f5e3ee5", "attempt_id": "aa6d34ffbe54", "run_group_id": "E4_core:A4|repairable|qwen3b:h2", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:54:35.970114Z", "task_family": "synthetic", "document_id": "syn_03015_2000", "document_hash": "71a7284b9bc3f681d69d91d18b9e8d45a4dda486cd14f4145395786ac6a68a89", "question_id": "q01_single_fact", "split": "test", "cluster_id": "syn_03015_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 2, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1922, "query_tokens": 573, "message_tokens": 486, "message_utf8_bytes": 1045, "cache_tensor_bytes": null, "serialized_payload_bytes": 1045, "side_information_bytes": 0, "fetched_bytes": 888, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 43798.324792063795, "hops": {"1": 22737.68987506628, "2": 21060.634916997515}, "receiver_prefill_ms": 388.4817080106586, "ttft_ms": 31.76745807286352, "decode_ms": 36.31516604218632}, "total_wall_ms": 44255.52283413708, "ttft_ms": 31.76745807286352, "input_usage": {"lookups": 4}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 61, "facts_retained": 24, "retention_rate": 0.39344262295081966, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.4166666666666667, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1045}, "ledger_by_hop": {"1": {"facts_total": 61, "facts_retained": 24, "retention_rate": 0.39344262295081966, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.4166666666666667, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1161}, "2": {"facts_total": 61, "facts_retained": 24, "retention_rate": 0.39344262295081966, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.4166666666666667, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1045}}}, "reference_key_id": "keys/synthetic/syn_03015_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "e8b0cdaa0bfa2267f9b579e9fa68a2f8c2db666e848f0095eb6c897c6e7b372d", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "403957bb3c9a999b7ee73fee", "attempt_id": "20e386164c45", "run_group_id": "E4_core:A4|repairable|qwen3b:h2", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:54:36.444572Z", "task_family": "synthetic", "document_id": "syn_03015_2000", "document_hash": "71a7284b9bc3f681d69d91d18b9e8d45a4dda486cd14f4145395786ac6a68a89", "question_id": "q02_exact_copy", "split": "test", "cluster_id": "syn_03015_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 2, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1922, "query_tokens": 572, "message_tokens": 486, "message_utf8_bytes": 1045, "cache_tensor_bytes": null, "serialized_payload_bytes": 1045, "side_information_bytes": 0, "fetched_bytes": 888, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 43798.324792063795, "hops": {"1": 22737.68987506628, "2": 21060.634916997515}, "receiver_prefill_ms": 379.27304208278656, "ttft_ms": 22.286374936811626, "decode_ms": 336.2845409428701}, "total_wall_ms": 44536.82812512852, "ttft_ms": 22.286374936811626, "input_usage": {"lookups": 4}, "output_usage": {"output_tokens": 7}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "89.3122", "evidence_artifact": {"ledger": {"facts_total": 61, "facts_retained": 24, "retention_rate": 0.39344262295081966, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.4166666666666667, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1045}, "ledger_by_hop": {"1": {"facts_total": 61, "facts_retained": 24, "retention_rate": 0.39344262295081966, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.4166666666666667, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1161}, "2": {"facts_total": 61, "facts_retained": 24, "retention_rate": 0.39344262295081966, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.4166666666666667, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1045}}}, "reference_key_id": "keys/synthetic/syn_03015_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "e8b0cdaa0bfa2267f9b579e9fa68a2f8c2db666e848f0095eb6c897c6e7b372d", "output_hash": "2138888e6519b2f4b1b54142e6f9c541102d7f9ee5bcf8e4184c81a382601757"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "85a6eb3356bcf22fc5551977", "attempt_id": "61f1774c60d0", "run_group_id": "E4_core:A4|repairable|qwen3b:h2", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:54:37.200194Z", "task_family": "synthetic", "document_id": "syn_03015_2000", "document_hash": "71a7284b9bc3f681d69d91d18b9e8d45a4dda486cd14f4145395786ac6a68a89", "question_id": "q03_exact_copy", "split": "test", "cluster_id": "syn_03015_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 2, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1922, "query_tokens": 568, "message_tokens": 486, "message_utf8_bytes": 1045, "cache_tensor_bytes": null, "serialized_payload_bytes": 1045, "side_information_bytes": 0, "fetched_bytes": 888, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 43798.324792063795, "hops": {"1": 22737.68987506628, "2": 21060.634916997515}, "receiver_prefill_ms": 525.3118339460343, "ttft_ms": 24.170707911252975, "decode_ms": 186.02291704155505}, "total_wall_ms": 44534.526667091995, "ttft_ms": 24.170707911252975, "input_usage": {"lookups": 4}, "output_usage": {"output_tokens": 6}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "Wilhelmina Dubrovnik", "evidence_artifact": {"ledger": {"facts_total": 61, "facts_retained": 24, "retention_rate": 0.39344262295081966, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.4166666666666667, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1045}, "ledger_by_hop": {"1": {"facts_total": 61, "facts_retained": 24, "retention_rate": 0.39344262295081966, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.4166666666666667, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1161}, "2": {"facts_total": 61, "facts_retained": 24, "retention_rate": 0.39344262295081966, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.4166666666666667, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1045}}}, "reference_key_id": "keys/synthetic/syn_03015_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "e8b0cdaa0bfa2267f9b579e9fa68a2f8c2db666e848f0095eb6c897c6e7b372d", "output_hash": "f42f05ebfc96f7258cbfbbb37cd65e114d3d15f869e68e6c95b71fc288c30a32"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "a69f22850c0be08c372a7fc2", "attempt_id": "f2c3e5e8c5d6", "run_group_id": "E4_core:A4|repairable|qwen3b:h2", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:54:37.954313Z", "task_family": "synthetic", "document_id": "syn_03015_2000", "document_hash": "71a7284b9bc3f681d69d91d18b9e8d45a4dda486cd14f4145395786ac6a68a89", "question_id": "q04_multi_hop", "split": "test", "cluster_id": "syn_03015_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 2, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1922, "query_tokens": 577, "message_tokens": 486, "message_utf8_bytes": 1045, "cache_tensor_bytes": null, "serialized_payload_bytes": 1045, "side_information_bytes": 0, "fetched_bytes": 888, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 43798.324792063795, "hops": {"1": 22737.68987506628, "2": 21060.634916997515}, "receiver_prefill_ms": 375.532416976057, "ttft_ms": 20.915832952596247, "decode_ms": 296.14779201801866}, "total_wall_ms": 44491.56129208859, "ttft_ms": 20.915832952596247, "input_usage": {"lookups": 4}, "output_usage": {"output_tokens": 7}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "8421725", "evidence_artifact": {"ledger": {"facts_total": 61, "facts_retained": 24, "retention_rate": 0.39344262295081966, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.4166666666666667, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1045}, "ledger_by_hop": {"1": {"facts_total": 61, "facts_retained": 24, "retention_rate": 0.39344262295081966, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.4166666666666667, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1161}, "2": {"facts_total": 61, "facts_retained": 24, "retention_rate": 0.39344262295081966, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.4166666666666667, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1045}}}, "reference_key_id": "keys/synthetic/syn_03015_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "e8b0cdaa0bfa2267f9b579e9fa68a2f8c2db666e848f0095eb6c897c6e7b372d", "output_hash": "e4949aba3b0b77de05a87edaab854329fa75df7bf68b29a3def71e985a06e5b5"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "35f5509e92427b1484b7af12", "attempt_id": "b03684d2d2ac", "run_group_id": "E4_core:A4|repairable|qwen3b:h2", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:54:38.665292Z", "task_family": "synthetic", "document_id": "syn_03015_2000", "document_hash": "71a7284b9bc3f681d69d91d18b9e8d45a4dda486cd14f4145395786ac6a68a89", "question_id": "q05_multi_hop", "split": "test", "cluster_id": "syn_03015_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 2, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1922, "query_tokens": 577, "message_tokens": 486, "message_utf8_bytes": 1045, "cache_tensor_bytes": null, "serialized_payload_bytes": 1045, "side_information_bytes": 0, "fetched_bytes": 888, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 43798.324792063795, "hops": {"1": 22737.68987506628, "2": 21060.634916997515}, "receiver_prefill_ms": 513.9998749364167, "ttft_ms": 22.090457961894572, "decode_ms": 42.55879204720259}, "total_wall_ms": 44377.6565840235, "ttft_ms": 22.090457961894572, "input_usage": {"lookups": 4}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 61, "facts_retained": 24, "retention_rate": 0.39344262295081966, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.4166666666666667, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1045}, "ledger_by_hop": {"1": {"facts_total": 61, "facts_retained": 24, "retention_rate": 0.39344262295081966, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.4166666666666667, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1161}, "2": {"facts_total": 61, "facts_retained": 24, "retention_rate": 0.39344262295081966, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.4166666666666667, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1045}}}, "reference_key_id": "keys/synthetic/syn_03015_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "e8b0cdaa0bfa2267f9b579e9fa68a2f8c2db666e848f0095eb6c897c6e7b372d", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "95e58f66c10c2151719d4829", "attempt_id": "d6c86d1eff91", "run_group_id": "E4_core:A4|repairable|qwen3b:h2", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:54:39.262033Z", "task_family": "synthetic", "document_id": "syn_03015_2000", "document_hash": "71a7284b9bc3f681d69d91d18b9e8d45a4dda486cd14f4145395786ac6a68a89", "question_id": "q06_negation", "split": "test", "cluster_id": "syn_03015_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 2, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1922, "query_tokens": 569, "message_tokens": 486, "message_utf8_bytes": 1045, "cache_tensor_bytes": null, "serialized_payload_bytes": 1045, "side_information_bytes": 0, "fetched_bytes": 888, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 43798.324792063795, "hops": {"1": 22737.68987506628, "2": 21060.634916997515}, "receiver_prefill_ms": 365.3360840398818, "ttft_ms": 36.80845792405307, "decode_ms": 44.6487080771476}, "total_wall_ms": 44245.76279206667, "ttft_ms": 36.80845792405307, "input_usage": {"lookups": 4}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 61, "facts_retained": 24, "retention_rate": 0.39344262295081966, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.4166666666666667, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1045}, "ledger_by_hop": {"1": {"facts_total": 61, "facts_retained": 24, "retention_rate": 0.39344262295081966, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.4166666666666667, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1161}, "2": {"facts_total": 61, "facts_retained": 24, "retention_rate": 0.39344262295081966, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.4166666666666667, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1045}}}, "reference_key_id": "keys/synthetic/syn_03015_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "e8b0cdaa0bfa2267f9b579e9fa68a2f8c2db666e848f0095eb6c897c6e7b372d", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "dd42ab5ff9554350b9d3b66b", "attempt_id": "8cd1b951b979", "run_group_id": "E4_core:A4|repairable|qwen3b:h2", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:54:39.727146Z", "task_family": "synthetic", "document_id": "syn_03015_2000", "document_hash": "71a7284b9bc3f681d69d91d18b9e8d45a4dda486cd14f4145395786ac6a68a89", "question_id": "q07_negation", "split": "test", "cluster_id": "syn_03015_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 2, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1922, "query_tokens": 569, "message_tokens": 486, "message_utf8_bytes": 1045, "cache_tensor_bytes": null, "serialized_payload_bytes": 1045, "side_information_bytes": 0, "fetched_bytes": 888, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 43798.324792063795, "hops": {"1": 22737.68987506628, "2": 21060.634916997515}, "receiver_prefill_ms": 369.93391695432365, "ttft_ms": 20.506624947302043, "decode_ms": 38.84054196532816}, "total_wall_ms": 44228.49158407189, "ttft_ms": 20.506624947302043, "input_usage": {"lookups": 4}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 61, "facts_retained": 24, "retention_rate": 0.39344262295081966, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.4166666666666667, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1045}, "ledger_by_hop": {"1": {"facts_total": 61, "facts_retained": 24, "retention_rate": 0.39344262295081966, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.4166666666666667, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1161}, "2": {"facts_total": 61, "facts_retained": 24, "retention_rate": 0.39344262295081966, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.4166666666666667, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1045}}}, "reference_key_id": "keys/synthetic/syn_03015_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "e8b0cdaa0bfa2267f9b579e9fa68a2f8c2db666e848f0095eb6c897c6e7b372d", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "d040bb87d47cb39a85031696", "attempt_id": "a39206fbce36", "run_group_id": "E4_core:A4|repairable|qwen3b:h2", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:54:40.175378Z", "task_family": "synthetic", "document_id": "syn_03015_2000", "document_hash": "71a7284b9bc3f681d69d91d18b9e8d45a4dda486cd14f4145395786ac6a68a89", "question_id": "q08_revision", "split": "test", "cluster_id": "syn_03015_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 2, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1922, "query_tokens": 583, "message_tokens": 486, "message_utf8_bytes": 1045, "cache_tensor_bytes": null, "serialized_payload_bytes": 1045, "side_information_bytes": 0, "fetched_bytes": 888, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 43798.324792063795, "hops": {"1": 22737.68987506628, "2": 21060.634916997515}, "receiver_prefill_ms": 411.0318749444559, "ttft_ms": 57.577500003390014, "decode_ms": 88.4281670441851}, "total_wall_ms": 44356.04279208928, "ttft_ms": 57.577500003390014, "input_usage": {"lookups": 4}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 61, "facts_retained": 24, "retention_rate": 0.39344262295081966, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.4166666666666667, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1045}, "ledger_by_hop": {"1": {"facts_total": 61, "facts_retained": 24, "retention_rate": 0.39344262295081966, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.4166666666666667, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1161}, "2": {"facts_total": 61, "facts_retained": 24, "retention_rate": 0.39344262295081966, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.4166666666666667, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1045}}}, "reference_key_id": "keys/synthetic/syn_03015_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "e8b0cdaa0bfa2267f9b579e9fa68a2f8c2db666e848f0095eb6c897c6e7b372d", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "14e2a74f9725bfb7087aa159", "attempt_id": "a0618ac26ab5", "run_group_id": "E4_core:A4|repairable|qwen3b:h2", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:54:40.751608Z", "task_family": "synthetic", "document_id": "syn_03015_2000", "document_hash": "71a7284b9bc3f681d69d91d18b9e8d45a4dda486cd14f4145395786ac6a68a89", "question_id": "q09_revision", "split": "test", "cluster_id": "syn_03015_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 2, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1922, "query_tokens": 577, "message_tokens": 486, "message_utf8_bytes": 1045, "cache_tensor_bytes": null, "serialized_payload_bytes": 1045, "side_information_bytes": 0, "fetched_bytes": 888, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 43798.324792063795, "hops": {"1": 22737.68987506628, "2": 21060.634916997515}, "receiver_prefill_ms": 426.9413750153035, "ttft_ms": 43.53395791258663, "decode_ms": 275.6914159981534}, "total_wall_ms": 44545.167292119004, "ttft_ms": 43.53395791258663, "input_usage": {"lookups": 4}, "output_usage": {"output_tokens": 9}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "8,377,480", "evidence_artifact": {"ledger": {"facts_total": 61, "facts_retained": 24, "retention_rate": 0.39344262295081966, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.4166666666666667, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1045}, "ledger_by_hop": {"1": {"facts_total": 61, "facts_retained": 24, "retention_rate": 0.39344262295081966, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.4166666666666667, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1161}, "2": {"facts_total": 61, "facts_retained": 24, "retention_rate": 0.39344262295081966, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.4166666666666667, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1045}}}, "reference_key_id": "keys/synthetic/syn_03015_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "e8b0cdaa0bfa2267f9b579e9fa68a2f8c2db666e848f0095eb6c897c6e7b372d", "output_hash": "ec4c68b327a2d2427ed21821415a5f9ddf87134bd9f654bdd69c7d1301febcb0"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "dee0189e79ad2376a4c1d6fa", "attempt_id": "956fb7fa962c", "run_group_id": "E4_core:A4|repairable|qwen3b:h2", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:54:41.515023Z", "task_family": "synthetic", "document_id": "syn_03015_2000", "document_hash": "71a7284b9bc3f681d69d91d18b9e8d45a4dda486cd14f4145395786ac6a68a89", "question_id": "q10_temporal", "split": "test", "cluster_id": "syn_03015_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 2, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1922, "query_tokens": 586, "message_tokens": 486, "message_utf8_bytes": 1045, "cache_tensor_bytes": null, "serialized_payload_bytes": 1045, "side_information_bytes": 0, "fetched_bytes": 888, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 43798.324792063795, "hops": {"1": 22737.68987506628, "2": 21060.634916997515}, "receiver_prefill_ms": 358.9935830095783, "ttft_ms": 50.18304206896573, "decode_ms": 56.00412504281849}, "total_wall_ms": 44264.31229209993, "ttft_ms": 50.18304206896573, "input_usage": {"lookups": 4}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 61, "facts_retained": 24, "retention_rate": 0.39344262295081966, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.4166666666666667, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1045}, "ledger_by_hop": {"1": {"facts_total": 61, "facts_retained": 24, "retention_rate": 0.39344262295081966, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.4166666666666667, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1161}, "2": {"facts_total": 61, "facts_retained": 24, "retention_rate": 0.39344262295081966, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.4166666666666667, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1045}}}, "reference_key_id": "keys/synthetic/syn_03015_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "e8b0cdaa0bfa2267f9b579e9fa68a2f8c2db666e848f0095eb6c897c6e7b372d", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "344e08c6bccb113797f4447a", "attempt_id": "5388b28736c6", "run_group_id": "E4_core:A4|repairable|qwen3b:h2", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:54:42.000942Z", "task_family": "synthetic", "document_id": "syn_03015_2000", "document_hash": "71a7284b9bc3f681d69d91d18b9e8d45a4dda486cd14f4145395786ac6a68a89", "question_id": "q11_aggregation", "split": "test", "cluster_id": "syn_03015_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 2, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1922, "query_tokens": 579, "message_tokens": 486, "message_utf8_bytes": 1045, "cache_tensor_bytes": null, "serialized_payload_bytes": 1045, "side_information_bytes": 0, "fetched_bytes": 888, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 43798.324792063795, "hops": {"1": 22737.68987506628, "2": 21060.634916997515}, "receiver_prefill_ms": 533.3230409305543, "ttft_ms": 55.54424994625151, "decode_ms": 211.9084580335766}, "total_wall_ms": 44599.83983414713, "ttft_ms": 55.54424994625151, "input_usage": {"lookups": 4}, "output_usage": {"output_tokens": 7}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "1806025", "evidence_artifact": {"ledger": {"facts_total": 61, "facts_retained": 24, "retention_rate": 0.39344262295081966, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.4166666666666667, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1045}, "ledger_by_hop": {"1": {"facts_total": 61, "facts_retained": 24, "retention_rate": 0.39344262295081966, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.4166666666666667, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1161}, "2": {"facts_total": 61, "facts_retained": 24, "retention_rate": 0.39344262295081966, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.4166666666666667, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1045}}}, "reference_key_id": "keys/synthetic/syn_03015_2000.json", "task_scores": {"correct": 0.0, "abstained": 0.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "e8b0cdaa0bfa2267f9b579e9fa68a2f8c2db666e848f0095eb6c897c6e7b372d", "output_hash": "9801d1840792d286693349934c9cc13fead44d82d480a42803cf6015c01431f6"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "e17a8f32ef4892f0c982a33a", "attempt_id": "46f37c5219da", "run_group_id": "E4_core:A4|repairable|qwen3b:h2", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:54:42.819934Z", "task_family": "synthetic", "document_id": "syn_03015_2000", "document_hash": "71a7284b9bc3f681d69d91d18b9e8d45a4dda486cd14f4145395786ac6a68a89", "question_id": "q12_unit_conversion", "split": "test", "cluster_id": "syn_03015_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 2, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1922, "query_tokens": 576, "message_tokens": 486, "message_utf8_bytes": 1045, "cache_tensor_bytes": null, "serialized_payload_bytes": 1045, "side_information_bytes": 0, "fetched_bytes": 888, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 43798.324792063795, "hops": {"1": 22737.68987506628, "2": 21060.634916997515}, "receiver_prefill_ms": 378.4825000911951, "ttft_ms": 26.71358408406377, "decode_ms": 35.69900000002235}, "total_wall_ms": 44239.773500012234, "ttft_ms": 26.71358408406377, "input_usage": {"lookups": 4}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 61, "facts_retained": 24, "retention_rate": 0.39344262295081966, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.4166666666666667, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1045}, "ledger_by_hop": {"1": {"facts_total": 61, "facts_retained": 24, "retention_rate": 0.39344262295081966, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.4166666666666667, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1161}, "2": {"facts_total": 61, "facts_retained": 24, "retention_rate": 0.39344262295081966, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.4166666666666667, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1045}}}, "reference_key_id": "keys/synthetic/syn_03015_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "e8b0cdaa0bfa2267f9b579e9fa68a2f8c2db666e848f0095eb6c897c6e7b372d", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "ef1ade51c3bccd9069cfad19", "attempt_id": "a9b78a246fbf", "run_group_id": "E4_core:A4|repairable|qwen3b:h2", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:54:43.278366Z", "task_family": "synthetic", "document_id": "syn_03015_2000", "document_hash": "71a7284b9bc3f681d69d91d18b9e8d45a4dda486cd14f4145395786ac6a68a89", "question_id": "q13_unanswerable", "split": "test", "cluster_id": "syn_03015_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 2, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1922, "query_tokens": 580, "message_tokens": 486, "message_utf8_bytes": 1045, "cache_tensor_bytes": null, "serialized_payload_bytes": 1045, "side_information_bytes": 0, "fetched_bytes": 888, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 43798.324792063795, "hops": {"1": 22737.68987506628, "2": 21060.634916997515}, "receiver_prefill_ms": 368.98049991577864, "ttft_ms": 41.48641601204872, "decode_ms": 56.84875010047108}, "total_wall_ms": 44266.36979205068, "ttft_ms": 41.48641601204872, "input_usage": {"lookups": 4}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 61, "facts_retained": 24, "retention_rate": 0.39344262295081966, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.4166666666666667, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1045}, "ledger_by_hop": {"1": {"facts_total": 61, "facts_retained": 24, "retention_rate": 0.39344262295081966, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.4166666666666667, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1161}, "2": {"facts_total": 61, "facts_retained": 24, "retention_rate": 0.39344262295081966, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.4166666666666667, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1045}}}, "reference_key_id": "keys/synthetic/syn_03015_2000.json", "task_scores": {"correct": 1.0, "abstained": 1.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "e8b0cdaa0bfa2267f9b579e9fa68a2f8c2db666e848f0095eb6c897c6e7b372d", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "9020fc6056605dc602d7ea87", "attempt_id": "8cb4678f5f4a", "run_group_id": "E4_core:A4|repairable|qwen3b:h2", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:54:43.763926Z", "task_family": "synthetic", "document_id": "syn_03015_2000", "document_hash": "71a7284b9bc3f681d69d91d18b9e8d45a4dda486cd14f4145395786ac6a68a89", "question_id": "q14_unanswerable", "split": "test", "cluster_id": "syn_03015_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 2, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1922, "query_tokens": 581, "message_tokens": 486, "message_utf8_bytes": 1045, "cache_tensor_bytes": null, "serialized_payload_bytes": 1045, "side_information_bytes": 0, "fetched_bytes": 888, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 43798.324792063795, "hops": {"1": 22737.68987506628, "2": 21060.634916997515}, "receiver_prefill_ms": 557.9803329892457, "ttft_ms": 42.748874984681606, "decode_ms": 36.46545798983425}, "total_wall_ms": 44436.20645906776, "ttft_ms": 42.748874984681606, "input_usage": {"lookups": 4}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 61, "facts_retained": 24, "retention_rate": 0.39344262295081966, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.4166666666666667, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1045}, "ledger_by_hop": {"1": {"facts_total": 61, "facts_retained": 24, "retention_rate": 0.39344262295081966, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.4166666666666667, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1161}, "2": {"facts_total": 61, "facts_retained": 24, "retention_rate": 0.39344262295081966, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.4166666666666667, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1045}}}, "reference_key_id": "keys/synthetic/syn_03015_2000.json", "task_scores": {"correct": 1.0, "abstained": 1.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "e8b0cdaa0bfa2267f9b579e9fa68a2f8c2db666e848f0095eb6c897c6e7b372d", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "4133229fbbbc293604c156ca", "attempt_id": "ef917a71f3b7", "run_group_id": "E4_core:A4|repairable|qwen3b:h4", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:55:26.271442Z", "task_family": "synthetic", "document_id": "syn_03015_2000", "document_hash": "71a7284b9bc3f681d69d91d18b9e8d45a4dda486cd14f4145395786ac6a68a89", "question_id": "q00_single_fact", "split": "test", "cluster_id": "syn_03015_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 4, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1922, "query_tokens": 563, "message_tokens": 481, "message_utf8_bytes": 1029, "cache_tensor_bytes": null, "serialized_payload_bytes": 1029, "side_information_bytes": 0, "fetched_bytes": 2664, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 85647.05979300197, "hops": {"1": 22737.68987506628, "2": 21060.634916997515, "3": 21005.673917010427, "4": 20843.06108392775}, "receiver_prefill_ms": 479.83683401253074, "ttft_ms": 51.77887494210154, "decode_ms": 21.058999933302402}, "total_wall_ms": 86200.44341799803, "ttft_ms": 51.77887494210154, "input_usage": {"lookups": 12}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 61, "facts_retained": 24, "retention_rate": 0.39344262295081966, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.4166666666666667, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1029}, "ledger_by_hop": {"1": {"facts_total": 61, "facts_retained": 24, "retention_rate": 0.39344262295081966, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.4166666666666667, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1161}, "2": {"facts_total": 61, "facts_retained": 24, "retention_rate": 0.39344262295081966, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.4166666666666667, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1045}, "3": {"facts_total": 61, "facts_retained": 24, "retention_rate": 0.39344262295081966, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.4166666666666667, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1042}, "4": {"facts_total": 61, "facts_retained": 24, "retention_rate": 0.39344262295081966, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.4166666666666667, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1029}}}, "reference_key_id": "keys/synthetic/syn_03015_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "13f688cc0243307a3eb9bb4da5cbeb4b020d6f83175f53e3d75ced3472f82704", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "cd6be80b0608382b13340567", "attempt_id": "1e6110bff42c", "run_group_id": "E4_core:A4|repairable|qwen3b:h4", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:55:26.843828Z", "task_family": "synthetic", "document_id": "syn_03015_2000", "document_hash": "71a7284b9bc3f681d69d91d18b9e8d45a4dda486cd14f4145395786ac6a68a89", "question_id": "q01_single_fact", "split": "test", "cluster_id": "syn_03015_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 4, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1922, "query_tokens": 568, "message_tokens": 481, "message_utf8_bytes": 1029, "cache_tensor_bytes": null, "serialized_payload_bytes": 1029, "side_information_bytes": 0, "fetched_bytes": 2664, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 85647.05979300197, "hops": {"1": 22737.68987506628, "2": 21060.634916997515, "3": 21005.673917010427, "4": 20843.06108392775}, "receiver_prefill_ms": 382.30633293278515, "ttft_ms": 40.38179200142622, "decode_ms": 26.372540974989533}, "total_wall_ms": 86096.6865849914, "ttft_ms": 40.38179200142622, "input_usage": {"lookups": 12}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 61, "facts_retained": 24, "retention_rate": 0.39344262295081966, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.4166666666666667, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1029}, "ledger_by_hop": {"1": {"facts_total": 61, "facts_retained": 24, "retention_rate": 0.39344262295081966, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.4166666666666667, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1161}, "2": {"facts_total": 61, "facts_retained": 24, "retention_rate": 0.39344262295081966, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.4166666666666667, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1045}, "3": {"facts_total": 61, "facts_retained": 24, "retention_rate": 0.39344262295081966, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.4166666666666667, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1042}, "4": {"facts_total": 61, "facts_retained": 24, "retention_rate": 0.39344262295081966, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.4166666666666667, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1029}}}, "reference_key_id": "keys/synthetic/syn_03015_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "13f688cc0243307a3eb9bb4da5cbeb4b020d6f83175f53e3d75ced3472f82704", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "9c0d3f1e8080f129720323bc", "attempt_id": "4477a31d784a", "run_group_id": "E4_core:A4|repairable|qwen3b:h4", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:55:27.310772Z", "task_family": "synthetic", "document_id": "syn_03015_2000", "document_hash": "71a7284b9bc3f681d69d91d18b9e8d45a4dda486cd14f4145395786ac6a68a89", "question_id": "q02_exact_copy", "split": "test", "cluster_id": "syn_03015_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 4, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1922, "query_tokens": 567, "message_tokens": 481, "message_utf8_bytes": 1029, "cache_tensor_bytes": null, "serialized_payload_bytes": 1029, "side_information_bytes": 0, "fetched_bytes": 2664, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 85647.05979300197, "hops": {"1": 22737.68987506628, "2": 21060.634916997515, "3": 21005.673917010427, "4": 20843.06108392775}, "receiver_prefill_ms": 377.69741704687476, "ttft_ms": 24.650500039570034, "decode_ms": 19.40379198640585}, "total_wall_ms": 86069.4037519861, "ttft_ms": 24.650500039570034, "input_usage": {"lookups": 12}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 61, "facts_retained": 24, "retention_rate": 0.39344262295081966, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.4166666666666667, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1029}, "ledger_by_hop": {"1": {"facts_total": 61, "facts_retained": 24, "retention_rate": 0.39344262295081966, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.4166666666666667, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1161}, "2": {"facts_total": 61, "facts_retained": 24, "retention_rate": 0.39344262295081966, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.4166666666666667, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1045}, "3": {"facts_total": 61, "facts_retained": 24, "retention_rate": 0.39344262295081966, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.4166666666666667, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1042}, "4": {"facts_total": 61, "facts_retained": 24, "retention_rate": 0.39344262295081966, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.4166666666666667, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1029}}}, "reference_key_id": "keys/synthetic/syn_03015_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "13f688cc0243307a3eb9bb4da5cbeb4b020d6f83175f53e3d75ced3472f82704", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "b8cfc9a894ef31d571003bed", "attempt_id": "c860b034d5be", "run_group_id": "E4_core:A4|repairable|qwen3b:h4", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:55:27.750417Z", "task_family": "synthetic", "document_id": "syn_03015_2000", "document_hash": "71a7284b9bc3f681d69d91d18b9e8d45a4dda486cd14f4145395786ac6a68a89", "question_id": "q03_exact_copy", "split": "test", "cluster_id": "syn_03015_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 4, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1922, "query_tokens": 563, "message_tokens": 481, "message_utf8_bytes": 1029, "cache_tensor_bytes": null, "serialized_payload_bytes": 1029, "side_information_bytes": 0, "fetched_bytes": 2664, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 85647.05979300197, "hops": {"1": 22737.68987506628, "2": 21060.634916997515, "3": 21005.673917010427, "4": 20843.06108392775}, "receiver_prefill_ms": 367.9555830312893, "ttft_ms": 52.30958398897201, "decode_ms": 42.018249980174005}, "total_wall_ms": 86110.09541794192, "ttft_ms": 52.30958398897201, "input_usage": {"lookups": 12}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 61, "facts_retained": 24, "retention_rate": 0.39344262295081966, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.4166666666666667, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1029}, "ledger_by_hop": {"1": {"facts_total": 61, "facts_retained": 24, "retention_rate": 0.39344262295081966, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.4166666666666667, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1161}, "2": {"facts_total": 61, "facts_retained": 24, "retention_rate": 0.39344262295081966, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.4166666666666667, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1045}, "3": {"facts_total": 61, "facts_retained": 24, "retention_rate": 0.39344262295081966, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.4166666666666667, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1042}, "4": {"facts_total": 61, "facts_retained": 24, "retention_rate": 0.39344262295081966, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.4166666666666667, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1029}}}, "reference_key_id": "keys/synthetic/syn_03015_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "13f688cc0243307a3eb9bb4da5cbeb4b020d6f83175f53e3d75ced3472f82704", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "4eadaa800e867bcd6c7df777", "attempt_id": "f6ad4e444c10", "run_group_id": "E4_core:A4|repairable|qwen3b:h4", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:55:28.233766Z", "task_family": "synthetic", "document_id": "syn_03015_2000", "document_hash": "71a7284b9bc3f681d69d91d18b9e8d45a4dda486cd14f4145395786ac6a68a89", "question_id": "q04_multi_hop", "split": "test", "cluster_id": "syn_03015_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 4, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1922, "query_tokens": 572, "message_tokens": 481, "message_utf8_bytes": 1029, "cache_tensor_bytes": null, "serialized_payload_bytes": 1029, "side_information_bytes": 0, "fetched_bytes": 2664, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 85647.05979300197, "hops": {"1": 22737.68987506628, "2": 21060.634916997515, "3": 21005.673917010427, "4": 20843.06108392775}, "receiver_prefill_ms": 533.1526659429073, "ttft_ms": 45.313874958083034, "decode_ms": 218.76858395989984}, "total_wall_ms": 86445.04187698476, "ttft_ms": 45.313874958083034, "input_usage": {"lookups": 12}, "output_usage": {"output_tokens": 7}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "8421725", "evidence_artifact": {"ledger": {"facts_total": 61, "facts_retained": 24, "retention_rate": 0.39344262295081966, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.4166666666666667, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1029}, "ledger_by_hop": {"1": {"facts_total": 61, "facts_retained": 24, "retention_rate": 0.39344262295081966, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.4166666666666667, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1161}, "2": {"facts_total": 61, "facts_retained": 24, "retention_rate": 0.39344262295081966, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.4166666666666667, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1045}, "3": {"facts_total": 61, "facts_retained": 24, "retention_rate": 0.39344262295081966, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.4166666666666667, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1042}, "4": {"facts_total": 61, "facts_retained": 24, "retention_rate": 0.39344262295081966, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.4166666666666667, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1029}}}, "reference_key_id": "keys/synthetic/syn_03015_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "13f688cc0243307a3eb9bb4da5cbeb4b020d6f83175f53e3d75ced3472f82704", "output_hash": "e4949aba3b0b77de05a87edaab854329fa75df7bf68b29a3def71e985a06e5b5"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "edb0f79a07293d4bec951152", "attempt_id": "d016e9771139", "run_group_id": "E4_core:A4|repairable|qwen3b:h4", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:55:29.049467Z", "task_family": "synthetic", "document_id": "syn_03015_2000", "document_hash": "71a7284b9bc3f681d69d91d18b9e8d45a4dda486cd14f4145395786ac6a68a89", "question_id": "q05_multi_hop", "split": "test", "cluster_id": "syn_03015_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 4, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1922, "query_tokens": 572, "message_tokens": 481, "message_utf8_bytes": 1029, "cache_tensor_bytes": null, "serialized_payload_bytes": 1029, "side_information_bytes": 0, "fetched_bytes": 2664, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 85647.05979300197, "hops": {"1": 22737.68987506628, "2": 21060.634916997515, "3": 21005.673917010427, "4": 20843.06108392775}, "receiver_prefill_ms": 359.84745796304196, "ttft_ms": 46.097208047285676, "decode_ms": 67.6799169741571}, "total_wall_ms": 86121.32608401589, "ttft_ms": 46.097208047285676, "input_usage": {"lookups": 12}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 61, "facts_retained": 24, "retention_rate": 0.39344262295081966, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.4166666666666667, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1029}, "ledger_by_hop": {"1": {"facts_total": 61, "facts_retained": 24, "retention_rate": 0.39344262295081966, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.4166666666666667, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1161}, "2": {"facts_total": 61, "facts_retained": 24, "retention_rate": 0.39344262295081966, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.4166666666666667, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1045}, "3": {"facts_total": 61, "facts_retained": 24, "retention_rate": 0.39344262295081966, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.4166666666666667, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1042}, "4": {"facts_total": 61, "facts_retained": 24, "retention_rate": 0.39344262295081966, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.4166666666666667, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1029}}}, "reference_key_id": "keys/synthetic/syn_03015_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "13f688cc0243307a3eb9bb4da5cbeb4b020d6f83175f53e3d75ced3472f82704", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "dcb0600d2b8f021e6c0b9fce", "attempt_id": "d002e2a325e8", "run_group_id": "E4_core:A4|repairable|qwen3b:h4", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:55:29.541145Z", "task_family": "synthetic", "document_id": "syn_03015_2000", "document_hash": "71a7284b9bc3f681d69d91d18b9e8d45a4dda486cd14f4145395786ac6a68a89", "question_id": "q06_negation", "split": "test", "cluster_id": "syn_03015_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 4, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1922, "query_tokens": 564, "message_tokens": 481, "message_utf8_bytes": 1029, "cache_tensor_bytes": null, "serialized_payload_bytes": 1029, "side_information_bytes": 0, "fetched_bytes": 2664, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 85647.05979300197, "hops": {"1": 22737.68987506628, "2": 21060.634916997515, "3": 21005.673917010427, "4": 20843.06108392775}, "receiver_prefill_ms": 417.1405410161242, "ttft_ms": 58.71670797932893, "decode_ms": 96.2450000224635}, "total_wall_ms": 86220.01000097953, "ttft_ms": 58.71670797932893, "input_usage": {"lookups": 12}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 61, "facts_retained": 24, "retention_rate": 0.39344262295081966, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.4166666666666667, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1029}, "ledger_by_hop": {"1": {"facts_total": 61, "facts_retained": 24, "retention_rate": 0.39344262295081966, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.4166666666666667, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1161}, "2": {"facts_total": 61, "facts_retained": 24, "retention_rate": 0.39344262295081966, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.4166666666666667, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1045}, "3": {"facts_total": 61, "facts_retained": 24, "retention_rate": 0.39344262295081966, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.4166666666666667, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1042}, "4": {"facts_total": 61, "facts_retained": 24, "retention_rate": 0.39344262295081966, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.4166666666666667, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1029}}}, "reference_key_id": "keys/synthetic/syn_03015_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "13f688cc0243307a3eb9bb4da5cbeb4b020d6f83175f53e3d75ced3472f82704", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "ce1ea7698770d6d1c57a4fa1", "attempt_id": "0280a69e20bd", "run_group_id": "E4_core:A4|repairable|qwen3b:h4", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:55:30.132287Z", "task_family": "synthetic", "document_id": "syn_03015_2000", "document_hash": "71a7284b9bc3f681d69d91d18b9e8d45a4dda486cd14f4145395786ac6a68a89", "question_id": "q07_negation", "split": "test", "cluster_id": "syn_03015_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 4, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1922, "query_tokens": 564, "message_tokens": 481, "message_utf8_bytes": 1029, "cache_tensor_bytes": null, "serialized_payload_bytes": 1029, "side_information_bytes": 0, "fetched_bytes": 2664, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 85647.05979300197, "hops": {"1": 22737.68987506628, "2": 21060.634916997515, "3": 21005.673917010427, "4": 20843.06108392775}, "receiver_prefill_ms": 374.7162079671398, "ttft_ms": 27.14254194870591, "decode_ms": 41.15774994716048}, "total_wall_ms": 86090.69575194735, "ttft_ms": 27.14254194870591, "input_usage": {"lookups": 12}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 61, "facts_retained": 24, "retention_rate": 0.39344262295081966, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.4166666666666667, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1029}, "ledger_by_hop": {"1": {"facts_total": 61, "facts_retained": 24, "retention_rate": 0.39344262295081966, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.4166666666666667, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1161}, "2": {"facts_total": 61, "facts_retained": 24, "retention_rate": 0.39344262295081966, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.4166666666666667, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1045}, "3": {"facts_total": 61, "facts_retained": 24, "retention_rate": 0.39344262295081966, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.4166666666666667, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1042}, "4": {"facts_total": 61, "facts_retained": 24, "retention_rate": 0.39344262295081966, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.4166666666666667, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1029}}}, "reference_key_id": "keys/synthetic/syn_03015_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "13f688cc0243307a3eb9bb4da5cbeb4b020d6f83175f53e3d75ced3472f82704", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "8a6f0349b5bccf5dc29db720", "attempt_id": "855eb7c99819", "run_group_id": "E4_core:A4|repairable|qwen3b:h4", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:55:30.593381Z", "task_family": "synthetic", "document_id": "syn_03015_2000", "document_hash": "71a7284b9bc3f681d69d91d18b9e8d45a4dda486cd14f4145395786ac6a68a89", "question_id": "q08_revision", "split": "test", "cluster_id": "syn_03015_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 4, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1922, "query_tokens": 578, "message_tokens": 481, "message_utf8_bytes": 1029, "cache_tensor_bytes": null, "serialized_payload_bytes": 1029, "side_information_bytes": 0, "fetched_bytes": 2664, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 85647.05979300197, "hops": {"1": 22737.68987506628, "2": 21060.634916997515, "3": 21005.673917010427, "4": 20843.06108392775}, "receiver_prefill_ms": 400.055208010599, "ttft_ms": 34.936459036543965, "decode_ms": 29.515416943468153}, "total_wall_ms": 86112.1828770265, "ttft_ms": 34.936459036543965, "input_usage": {"lookups": 12}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 61, "facts_retained": 24, "retention_rate": 0.39344262295081966, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.4166666666666667, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1029}, "ledger_by_hop": {"1": {"facts_total": 61, "facts_retained": 24, "retention_rate": 0.39344262295081966, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.4166666666666667, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1161}, "2": {"facts_total": 61, "facts_retained": 24, "retention_rate": 0.39344262295081966, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.4166666666666667, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1045}, "3": {"facts_total": 61, "facts_retained": 24, "retention_rate": 0.39344262295081966, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.4166666666666667, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1042}, "4": {"facts_total": 61, "facts_retained": 24, "retention_rate": 0.39344262295081966, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.4166666666666667, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1029}}}, "reference_key_id": "keys/synthetic/syn_03015_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "13f688cc0243307a3eb9bb4da5cbeb4b020d6f83175f53e3d75ced3472f82704", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "49a5ecf3add0d23a38e506db", "attempt_id": "243dde678752", "run_group_id": "E4_core:A4|repairable|qwen3b:h4", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:55:31.075621Z", "task_family": "synthetic", "document_id": "syn_03015_2000", "document_hash": "71a7284b9bc3f681d69d91d18b9e8d45a4dda486cd14f4145395786ac6a68a89", "question_id": "q09_revision", "split": "test", "cluster_id": "syn_03015_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 4, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1922, "query_tokens": 572, "message_tokens": 481, "message_utf8_bytes": 1029, "cache_tensor_bytes": null, "serialized_payload_bytes": 1029, "side_information_bytes": 0, "fetched_bytes": 2664, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 85647.05979300197, "hops": {"1": 22737.68987506628, "2": 21060.634916997515, "3": 21005.673917010427, "4": 20843.06108392775}, "receiver_prefill_ms": 338.65245804190636, "ttft_ms": 37.892541964538395, "decode_ms": 529.2862079804763}, "total_wall_ms": 86553.53304301389, "ttft_ms": 37.892541964538395, "input_usage": {"lookups": 12}, "output_usage": {"output_tokens": 9}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "8,377,480", "evidence_artifact": {"ledger": {"facts_total": 61, "facts_retained": 24, "retention_rate": 0.39344262295081966, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.4166666666666667, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1029}, "ledger_by_hop": {"1": {"facts_total": 61, "facts_retained": 24, "retention_rate": 0.39344262295081966, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.4166666666666667, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1161}, "2": {"facts_total": 61, "facts_retained": 24, "retention_rate": 0.39344262295081966, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.4166666666666667, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1045}, "3": {"facts_total": 61, "facts_retained": 24, "retention_rate": 0.39344262295081966, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.4166666666666667, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1042}, "4": {"facts_total": 61, "facts_retained": 24, "retention_rate": 0.39344262295081966, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.4166666666666667, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1029}}}, "reference_key_id": "keys/synthetic/syn_03015_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "13f688cc0243307a3eb9bb4da5cbeb4b020d6f83175f53e3d75ced3472f82704", "output_hash": "ec4c68b327a2d2427ed21821415a5f9ddf87134bd9f654bdd69c7d1301febcb0"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "bebd479af4889ff83162e7b5", "attempt_id": "e1c6ccc48211", "run_group_id": "E4_core:A4|repairable|qwen3b:h4", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:55:32.000942Z", "task_family": "synthetic", "document_id": "syn_03015_2000", "document_hash": "71a7284b9bc3f681d69d91d18b9e8d45a4dda486cd14f4145395786ac6a68a89", "question_id": "q10_temporal", "split": "test", "cluster_id": "syn_03015_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 4, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1922, "query_tokens": 581, "message_tokens": 481, "message_utf8_bytes": 1029, "cache_tensor_bytes": null, "serialized_payload_bytes": 1029, "side_information_bytes": 0, "fetched_bytes": 2664, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 85647.05979300197, "hops": {"1": 22737.68987506628, "2": 21060.634916997515, "3": 21005.673917010427, "4": 20843.06108392775}, "receiver_prefill_ms": 399.8627499677241, "ttft_ms": 42.199791059829295, "decode_ms": 25.238291011191905}, "total_wall_ms": 86115.0042929221, "ttft_ms": 42.199791059829295, "input_usage": {"lookups": 12}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 61, "facts_retained": 24, "retention_rate": 0.39344262295081966, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.4166666666666667, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1029}, "ledger_by_hop": {"1": {"facts_total": 61, "facts_retained": 24, "retention_rate": 0.39344262295081966, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.4166666666666667, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1161}, "2": {"facts_total": 61, "facts_retained": 24, "retention_rate": 0.39344262295081966, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.4166666666666667, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1045}, "3": {"facts_total": 61, "facts_retained": 24, "retention_rate": 0.39344262295081966, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.4166666666666667, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1042}, "4": {"facts_total": 61, "facts_retained": 24, "retention_rate": 0.39344262295081966, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.4166666666666667, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1029}}}, "reference_key_id": "keys/synthetic/syn_03015_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "13f688cc0243307a3eb9bb4da5cbeb4b020d6f83175f53e3d75ced3472f82704", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "ec2123020f6546f70daa0ae7", "attempt_id": "ec4a6433e49d", "run_group_id": "E4_core:A4|repairable|qwen3b:h4", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:55:32.486168Z", "task_family": "synthetic", "document_id": "syn_03015_2000", "document_hash": "71a7284b9bc3f681d69d91d18b9e8d45a4dda486cd14f4145395786ac6a68a89", "question_id": "q11_aggregation", "split": "test", "cluster_id": "syn_03015_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 4, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1922, "query_tokens": 574, "message_tokens": 481, "message_utf8_bytes": 1029, "cache_tensor_bytes": null, "serialized_payload_bytes": 1029, "side_information_bytes": 0, "fetched_bytes": 2664, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 85647.05979300197, "hops": {"1": 22737.68987506628, "2": 21060.634916997515, "3": 21005.673917010427, "4": 20843.06108392775}, "receiver_prefill_ms": 339.2341660801321, "ttft_ms": 52.010583924129605, "decode_ms": 498.6140000401065}, "total_wall_ms": 86537.65833505895, "ttft_ms": 52.010583924129605, "input_usage": {"lookups": 12}, "output_usage": {"output_tokens": 7}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "1806025", "evidence_artifact": {"ledger": {"facts_total": 61, "facts_retained": 24, "retention_rate": 0.39344262295081966, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.4166666666666667, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1029}, "ledger_by_hop": {"1": {"facts_total": 61, "facts_retained": 24, "retention_rate": 0.39344262295081966, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.4166666666666667, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1161}, "2": {"facts_total": 61, "facts_retained": 24, "retention_rate": 0.39344262295081966, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.4166666666666667, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1045}, "3": {"facts_total": 61, "facts_retained": 24, "retention_rate": 0.39344262295081966, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.4166666666666667, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1042}, "4": {"facts_total": 61, "facts_retained": 24, "retention_rate": 0.39344262295081966, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.4166666666666667, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1029}}}, "reference_key_id": "keys/synthetic/syn_03015_2000.json", "task_scores": {"correct": 0.0, "abstained": 0.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "13f688cc0243307a3eb9bb4da5cbeb4b020d6f83175f53e3d75ced3472f82704", "output_hash": "9801d1840792d286693349934c9cc13fead44d82d480a42803cf6015c01431f6"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "ad70e0a7c085fa809762ec45", "attempt_id": "2c46ff7fafe8", "run_group_id": "E4_core:A4|repairable|qwen3b:h4", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:55:33.394638Z", "task_family": "synthetic", "document_id": "syn_03015_2000", "document_hash": "71a7284b9bc3f681d69d91d18b9e8d45a4dda486cd14f4145395786ac6a68a89", "question_id": "q12_unit_conversion", "split": "test", "cluster_id": "syn_03015_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 4, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1922, "query_tokens": 571, "message_tokens": 481, "message_utf8_bytes": 1029, "cache_tensor_bytes": null, "serialized_payload_bytes": 1029, "side_information_bytes": 0, "fetched_bytes": 2664, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 85647.05979300197, "hops": {"1": 22737.68987506628, "2": 21060.634916997515, "3": 21005.673917010427, "4": 20843.06108392775}, "receiver_prefill_ms": 364.7769170347601, "ttft_ms": 25.77175002079457, "decode_ms": 38.86516601778567}, "total_wall_ms": 86077.16566801537, "ttft_ms": 25.77175002079457, "input_usage": {"lookups": 12}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 61, "facts_retained": 24, "retention_rate": 0.39344262295081966, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.4166666666666667, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1029}, "ledger_by_hop": {"1": {"facts_total": 61, "facts_retained": 24, "retention_rate": 0.39344262295081966, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.4166666666666667, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1161}, "2": {"facts_total": 61, "facts_retained": 24, "retention_rate": 0.39344262295081966, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.4166666666666667, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1045}, "3": {"facts_total": 61, "facts_retained": 24, "retention_rate": 0.39344262295081966, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.4166666666666667, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1042}, "4": {"facts_total": 61, "facts_retained": 24, "retention_rate": 0.39344262295081966, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.4166666666666667, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1029}}}, "reference_key_id": "keys/synthetic/syn_03015_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "13f688cc0243307a3eb9bb4da5cbeb4b020d6f83175f53e3d75ced3472f82704", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "06a65dc26a1fcd5c493fdf29", "attempt_id": "16839172e22a", "run_group_id": "E4_core:A4|repairable|qwen3b:h4", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:55:33.842723Z", "task_family": "synthetic", "document_id": "syn_03015_2000", "document_hash": "71a7284b9bc3f681d69d91d18b9e8d45a4dda486cd14f4145395786ac6a68a89", "question_id": "q13_unanswerable", "split": "test", "cluster_id": "syn_03015_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 4, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1922, "query_tokens": 575, "message_tokens": 481, "message_utf8_bytes": 1029, "cache_tensor_bytes": null, "serialized_payload_bytes": 1029, "side_information_bytes": 0, "fetched_bytes": 2664, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 85647.05979300197, "hops": {"1": 22737.68987506628, "2": 21060.634916997515, "3": 21005.673917010427, "4": 20843.06108392775}, "receiver_prefill_ms": 394.10816703457385, "ttft_ms": 30.542458058334887, "decode_ms": 37.47837501578033}, "total_wall_ms": 86109.82770996634, "ttft_ms": 30.542458058334887, "input_usage": {"lookups": 12}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 61, "facts_retained": 24, "retention_rate": 0.39344262295081966, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.4166666666666667, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1029}, "ledger_by_hop": {"1": {"facts_total": 61, "facts_retained": 24, "retention_rate": 0.39344262295081966, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.4166666666666667, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1161}, "2": {"facts_total": 61, "facts_retained": 24, "retention_rate": 0.39344262295081966, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.4166666666666667, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1045}, "3": {"facts_total": 61, "facts_retained": 24, "retention_rate": 0.39344262295081966, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.4166666666666667, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1042}, "4": {"facts_total": 61, "facts_retained": 24, "retention_rate": 0.39344262295081966, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.4166666666666667, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1029}}}, "reference_key_id": "keys/synthetic/syn_03015_2000.json", "task_scores": {"correct": 1.0, "abstained": 1.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "13f688cc0243307a3eb9bb4da5cbeb4b020d6f83175f53e3d75ced3472f82704", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "a1bc9687ad552c1366703ef0", "attempt_id": "7af7dc866610", "run_group_id": "E4_core:A4|repairable|qwen3b:h4", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:55:34.324064Z", "task_family": "synthetic", "document_id": "syn_03015_2000", "document_hash": "71a7284b9bc3f681d69d91d18b9e8d45a4dda486cd14f4145395786ac6a68a89", "question_id": "q14_unanswerable", "split": "test", "cluster_id": "syn_03015_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 4, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1922, "query_tokens": 576, "message_tokens": 481, "message_utf8_bytes": 1029, "cache_tensor_bytes": null, "serialized_payload_bytes": 1029, "side_information_bytes": 0, "fetched_bytes": 2664, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 85647.05979300197, "hops": {"1": 22737.68987506628, "2": 21060.634916997515, "3": 21005.673917010427, "4": 20843.06108392775}, "receiver_prefill_ms": 340.295874979347, "ttft_ms": 39.398042019456625, "decode_ms": 56.02270795498043}, "total_wall_ms": 86083.63858505618, "ttft_ms": 39.398042019456625, "input_usage": {"lookups": 12}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 61, "facts_retained": 24, "retention_rate": 0.39344262295081966, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.4166666666666667, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1029}, "ledger_by_hop": {"1": {"facts_total": 61, "facts_retained": 24, "retention_rate": 0.39344262295081966, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.4166666666666667, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1161}, "2": {"facts_total": 61, "facts_retained": 24, "retention_rate": 0.39344262295081966, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.4166666666666667, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1045}, "3": {"facts_total": 61, "facts_retained": 24, "retention_rate": 0.39344262295081966, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.4166666666666667, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1042}, "4": {"facts_total": 61, "facts_retained": 24, "retention_rate": 0.39344262295081966, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.4166666666666667, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1029}}}, "reference_key_id": "keys/synthetic/syn_03015_2000.json", "task_scores": {"correct": 1.0, "abstained": 1.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "13f688cc0243307a3eb9bb4da5cbeb4b020d6f83175f53e3d75ced3472f82704", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "add9b5b1fcfb9091fc7feb99", "attempt_id": "c19279e9a6ab", "run_group_id": "E4_core:A4|repairable|qwen3b:h8", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:56:57.887380Z", "task_family": "synthetic", "document_id": "syn_03015_2000", "document_hash": "71a7284b9bc3f681d69d91d18b9e8d45a4dda486cd14f4145395786ac6a68a89", "question_id": "q00_single_fact", "split": "test", "cluster_id": "syn_03015_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 8, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1922, "query_tokens": 563, "message_tokens": 481, "message_utf8_bytes": 1029, "cache_tensor_bytes": null, "serialized_payload_bytes": 1029, "side_information_bytes": 0, "fetched_bytes": 6216, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 168804.19437598903, "hops": {"1": 22737.68987506628, "2": 21060.634916997515, "3": 21005.673917010427, "4": 20843.06108392775, "5": 20883.118791040033, "6": 20894.568374962546, "7": 20802.81575000845, "8": 20576.631666976027}, "receiver_prefill_ms": 355.62083299737424, "ttft_ms": 43.26758300885558, "decode_ms": 49.974167020991445}, "total_wall_ms": 169253.89658496715, "ttft_ms": 43.26758300885558, "input_usage": {"lookups": 28}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 61, "facts_retained": 24, "retention_rate": 0.39344262295081966, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.4166666666666667, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1029}, "ledger_by_hop": {"1": {"facts_total": 61, "facts_retained": 24, "retention_rate": 0.39344262295081966, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.4166666666666667, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1161}, "2": {"facts_total": 61, "facts_retained": 24, "retention_rate": 0.39344262295081966, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.4166666666666667, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1045}, "3": {"facts_total": 61, "facts_retained": 24, "retention_rate": 0.39344262295081966, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.4166666666666667, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1042}, "4": {"facts_total": 61, "facts_retained": 24, "retention_rate": 0.39344262295081966, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.4166666666666667, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1029}, "5": {"facts_total": 61, "facts_retained": 24, "retention_rate": 0.39344262295081966, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.4166666666666667, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1029}, "6": {"facts_total": 61, "facts_retained": 24, "retention_rate": 0.39344262295081966, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.4166666666666667, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1029}, "7": {"facts_total": 61, "facts_retained": 24, "retention_rate": 0.39344262295081966, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.4166666666666667, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1029}, "8": {"facts_total": 61, "facts_retained": 24, "retention_rate": 0.39344262295081966, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.4166666666666667, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1029}}}, "reference_key_id": "keys/synthetic/syn_03015_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "13f688cc0243307a3eb9bb4da5cbeb4b020d6f83175f53e3d75ced3472f82704", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "e17286b7e78c3acf410f0c2c", "attempt_id": "8bf8d9561fb6", "run_group_id": "E4_core:A4|repairable|qwen3b:h8", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:56:58.356481Z", "task_family": "synthetic", "document_id": "syn_03015_2000", "document_hash": "71a7284b9bc3f681d69d91d18b9e8d45a4dda486cd14f4145395786ac6a68a89", "question_id": "q01_single_fact", "split": "test", "cluster_id": "syn_03015_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 8, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1922, "query_tokens": 568, "message_tokens": 481, "message_utf8_bytes": 1029, "cache_tensor_bytes": null, "serialized_payload_bytes": 1029, "side_information_bytes": 0, "fetched_bytes": 6216, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 168804.19437598903, "hops": {"1": 22737.68987506628, "2": 21060.634916997515, "3": 21005.673917010427, "4": 20843.06108392775, "5": 20883.118791040033, "6": 20894.568374962546, "7": 20802.81575000845, "8": 20576.631666976027}, "receiver_prefill_ms": 492.97162506263703, "ttft_ms": 32.86862501408905, "decode_ms": 31.865541939623654}, "total_wall_ms": 169362.46650095563, "ttft_ms": 32.86862501408905, "input_usage": {"lookups": 28}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 61, "facts_retained": 24, "retention_rate": 0.39344262295081966, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.4166666666666667, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1029}, "ledger_by_hop": {"1": {"facts_total": 61, "facts_retained": 24, "retention_rate": 0.39344262295081966, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.4166666666666667, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1161}, "2": {"facts_total": 61, "facts_retained": 24, "retention_rate": 0.39344262295081966, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.4166666666666667, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1045}, "3": {"facts_total": 61, "facts_retained": 24, "retention_rate": 0.39344262295081966, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.4166666666666667, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1042}, "4": {"facts_total": 61, "facts_retained": 24, "retention_rate": 0.39344262295081966, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.4166666666666667, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1029}, "5": {"facts_total": 61, "facts_retained": 24, "retention_rate": 0.39344262295081966, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.4166666666666667, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1029}, "6": {"facts_total": 61, "facts_retained": 24, "retention_rate": 0.39344262295081966, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.4166666666666667, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1029}, "7": {"facts_total": 61, "facts_retained": 24, "retention_rate": 0.39344262295081966, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.4166666666666667, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1029}, "8": {"facts_total": 61, "facts_retained": 24, "retention_rate": 0.39344262295081966, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.4166666666666667, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1029}}}, "reference_key_id": "keys/synthetic/syn_03015_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "13f688cc0243307a3eb9bb4da5cbeb4b020d6f83175f53e3d75ced3472f82704", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "e7e594592dd871b732d08a5c", "attempt_id": "23be6e220ba7", "run_group_id": "E4_core:A4|repairable|qwen3b:h8", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:56:58.933260Z", "task_family": "synthetic", "document_id": "syn_03015_2000", "document_hash": "71a7284b9bc3f681d69d91d18b9e8d45a4dda486cd14f4145395786ac6a68a89", "question_id": "q02_exact_copy", "split": "test", "cluster_id": "syn_03015_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 8, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1922, "query_tokens": 567, "message_tokens": 481, "message_utf8_bytes": 1029, "cache_tensor_bytes": null, "serialized_payload_bytes": 1029, "side_information_bytes": 0, "fetched_bytes": 6216, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 168804.19437598903, "hops": {"1": 22737.68987506628, "2": 21060.634916997515, "3": 21005.673917010427, "4": 20843.06108392775, "5": 20883.118791040033, "6": 20894.568374962546, "7": 20802.81575000845, "8": 20576.631666976027}, "receiver_prefill_ms": 379.17166692204773, "ttft_ms": 21.593125071376562, "decode_ms": 38.02324994467199}, "total_wall_ms": 169243.54762595613, "ttft_ms": 21.593125071376562, "input_usage": {"lookups": 28}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 61, "facts_retained": 24, "retention_rate": 0.39344262295081966, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.4166666666666667, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1029}, "ledger_by_hop": {"1": {"facts_total": 61, "facts_retained": 24, "retention_rate": 0.39344262295081966, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.4166666666666667, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1161}, "2": {"facts_total": 61, "facts_retained": 24, "retention_rate": 0.39344262295081966, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.4166666666666667, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1045}, "3": {"facts_total": 61, "facts_retained": 24, "retention_rate": 0.39344262295081966, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.4166666666666667, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1042}, "4": {"facts_total": 61, "facts_retained": 24, "retention_rate": 0.39344262295081966, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.4166666666666667, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1029}, "5": {"facts_total": 61, "facts_retained": 24, "retention_rate": 0.39344262295081966, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.4166666666666667, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1029}, "6": {"facts_total": 61, "facts_retained": 24, "retention_rate": 0.39344262295081966, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.4166666666666667, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1029}, "7": {"facts_total": 61, "facts_retained": 24, "retention_rate": 0.39344262295081966, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.4166666666666667, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1029}, "8": {"facts_total": 61, "facts_retained": 24, "retention_rate": 0.39344262295081966, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.4166666666666667, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1029}}}, "reference_key_id": "keys/synthetic/syn_03015_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "13f688cc0243307a3eb9bb4da5cbeb4b020d6f83175f53e3d75ced3472f82704", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "e76049ab00147a3226e3ff88", "attempt_id": "6df1c2b92452", "run_group_id": "E4_core:A4|repairable|qwen3b:h8", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:56:59.390956Z", "task_family": "synthetic", "document_id": "syn_03015_2000", "document_hash": "71a7284b9bc3f681d69d91d18b9e8d45a4dda486cd14f4145395786ac6a68a89", "question_id": "q03_exact_copy", "split": "test", "cluster_id": "syn_03015_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 8, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1922, "query_tokens": 563, "message_tokens": 481, "message_utf8_bytes": 1029, "cache_tensor_bytes": null, "serialized_payload_bytes": 1029, "side_information_bytes": 0, "fetched_bytes": 6216, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 168804.19437598903, "hops": {"1": 22737.68987506628, "2": 21060.634916997515, "3": 21005.673917010427, "4": 20843.06108392775, "5": 20883.118791040033, "6": 20894.568374962546, "7": 20802.81575000845, "8": 20576.631666976027}, "receiver_prefill_ms": 373.4055410604924, "ttft_ms": 61.37120793573558, "decode_ms": 59.25258400384337}, "total_wall_ms": 169298.86108404025, "ttft_ms": 61.37120793573558, "input_usage": {"lookups": 28}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 61, "facts_retained": 24, "retention_rate": 0.39344262295081966, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.4166666666666667, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1029}, "ledger_by_hop": {"1": {"facts_total": 61, "facts_retained": 24, "retention_rate": 0.39344262295081966, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.4166666666666667, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1161}, "2": {"facts_total": 61, "facts_retained": 24, "retention_rate": 0.39344262295081966, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.4166666666666667, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1045}, "3": {"facts_total": 61, "facts_retained": 24, "retention_rate": 0.39344262295081966, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.4166666666666667, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1042}, "4": {"facts_total": 61, "facts_retained": 24, "retention_rate": 0.39344262295081966, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.4166666666666667, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1029}, "5": {"facts_total": 61, "facts_retained": 24, "retention_rate": 0.39344262295081966, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.4166666666666667, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1029}, "6": {"facts_total": 61, "facts_retained": 24, "retention_rate": 0.39344262295081966, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.4166666666666667, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1029}, "7": {"facts_total": 61, "facts_retained": 24, "retention_rate": 0.39344262295081966, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.4166666666666667, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1029}, "8": {"facts_total": 61, "facts_retained": 24, "retention_rate": 0.39344262295081966, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.4166666666666667, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1029}}}, "reference_key_id": "keys/synthetic/syn_03015_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "13f688cc0243307a3eb9bb4da5cbeb4b020d6f83175f53e3d75ced3472f82704", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "259e7a041b043c24a574d9c9", "attempt_id": "107cdb05daf4", "run_group_id": "E4_core:A4|repairable|qwen3b:h8", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:56:59.902999Z", "task_family": "synthetic", "document_id": "syn_03015_2000", "document_hash": "71a7284b9bc3f681d69d91d18b9e8d45a4dda486cd14f4145395786ac6a68a89", "question_id": "q04_multi_hop", "split": "test", "cluster_id": "syn_03015_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 8, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1922, "query_tokens": 572, "message_tokens": 481, "message_utf8_bytes": 1029, "cache_tensor_bytes": null, "serialized_payload_bytes": 1029, "side_information_bytes": 0, "fetched_bytes": 6216, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 168804.19437598903, "hops": {"1": 22737.68987506628, "2": 21060.634916997515, "3": 21005.673917010427, "4": 20843.06108392775, "5": 20883.118791040033, "6": 20894.568374962546, "7": 20802.81575000845, "8": 20576.631666976027}, "receiver_prefill_ms": 545.6249160924926, "ttft_ms": 39.46408396586776, "decode_ms": 216.72283299267292}, "total_wall_ms": 169606.6597510362, "ttft_ms": 39.46408396586776, "input_usage": {"lookups": 28}, "output_usage": {"output_tokens": 7}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "8421725", "evidence_artifact": {"ledger": {"facts_total": 61, "facts_retained": 24, "retention_rate": 0.39344262295081966, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.4166666666666667, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1029}, "ledger_by_hop": {"1": {"facts_total": 61, "facts_retained": 24, "retention_rate": 0.39344262295081966, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.4166666666666667, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1161}, "2": {"facts_total": 61, "facts_retained": 24, "retention_rate": 0.39344262295081966, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.4166666666666667, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1045}, "3": {"facts_total": 61, "facts_retained": 24, "retention_rate": 0.39344262295081966, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.4166666666666667, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1042}, "4": {"facts_total": 61, "facts_retained": 24, "retention_rate": 0.39344262295081966, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.4166666666666667, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1029}, "5": {"facts_total": 61, "facts_retained": 24, "retention_rate": 0.39344262295081966, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.4166666666666667, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1029}, "6": {"facts_total": 61, "facts_retained": 24, "retention_rate": 0.39344262295081966, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.4166666666666667, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1029}, "7": {"facts_total": 61, "facts_retained": 24, "retention_rate": 0.39344262295081966, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.4166666666666667, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1029}, "8": {"facts_total": 61, "facts_retained": 24, "retention_rate": 0.39344262295081966, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.4166666666666667, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1029}}}, "reference_key_id": "keys/synthetic/syn_03015_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "13f688cc0243307a3eb9bb4da5cbeb4b020d6f83175f53e3d75ced3472f82704", "output_hash": "e4949aba3b0b77de05a87edaab854329fa75df7bf68b29a3def71e985a06e5b5"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "0d0bfdb70ccf8b3742dfacce", "attempt_id": "f8fb15d4af05", "run_group_id": "E4_core:A4|repairable|qwen3b:h8", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:57:00.722761Z", "task_family": "synthetic", "document_id": "syn_03015_2000", "document_hash": "71a7284b9bc3f681d69d91d18b9e8d45a4dda486cd14f4145395786ac6a68a89", "question_id": "q05_multi_hop", "split": "test", "cluster_id": "syn_03015_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 8, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1922, "query_tokens": 572, "message_tokens": 481, "message_utf8_bytes": 1029, "cache_tensor_bytes": null, "serialized_payload_bytes": 1029, "side_information_bytes": 0, "fetched_bytes": 6216, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 168804.19437598903, "hops": {"1": 22737.68987506628, "2": 21060.634916997515, "3": 21005.673917010427, "4": 20843.06108392775, "5": 20883.118791040033, "6": 20894.568374962546, "7": 20802.81575000845, "8": 20576.631666976027}, "receiver_prefill_ms": 383.87958402745426, "ttft_ms": 34.56170903518796, "decode_ms": 30.940167023800313}, "total_wall_ms": 169254.07412601635, "ttft_ms": 34.56170903518796, "input_usage": {"lookups": 28}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 61, "facts_retained": 24, "retention_rate": 0.39344262295081966, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.4166666666666667, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1029}, "ledger_by_hop": {"1": {"facts_total": 61, "facts_retained": 24, "retention_rate": 0.39344262295081966, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.4166666666666667, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1161}, "2": {"facts_total": 61, "facts_retained": 24, "retention_rate": 0.39344262295081966, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.4166666666666667, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1045}, "3": {"facts_total": 61, "facts_retained": 24, "retention_rate": 0.39344262295081966, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.4166666666666667, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1042}, "4": {"facts_total": 61, "facts_retained": 24, "retention_rate": 0.39344262295081966, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.4166666666666667, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1029}, "5": {"facts_total": 61, "facts_retained": 24, "retention_rate": 0.39344262295081966, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.4166666666666667, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1029}, "6": {"facts_total": 61, "facts_retained": 24, "retention_rate": 0.39344262295081966, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.4166666666666667, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1029}, "7": {"facts_total": 61, "facts_retained": 24, "retention_rate": 0.39344262295081966, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.4166666666666667, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1029}, "8": {"facts_total": 61, "facts_retained": 24, "retention_rate": 0.39344262295081966, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.4166666666666667, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1029}}}, "reference_key_id": "keys/synthetic/syn_03015_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "13f688cc0243307a3eb9bb4da5cbeb4b020d6f83175f53e3d75ced3472f82704", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "632dbf8f587683d3a0fa4304", "attempt_id": "1af83359f2d1", "run_group_id": "E4_core:A4|repairable|qwen3b:h8", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:57:01.190262Z", "task_family": "synthetic", "document_id": "syn_03015_2000", "document_hash": "71a7284b9bc3f681d69d91d18b9e8d45a4dda486cd14f4145395786ac6a68a89", "question_id": "q06_negation", "split": "test", "cluster_id": "syn_03015_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 8, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1922, "query_tokens": 564, "message_tokens": 481, "message_utf8_bytes": 1029, "cache_tensor_bytes": null, "serialized_payload_bytes": 1029, "side_information_bytes": 0, "fetched_bytes": 6216, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 168804.19437598903, "hops": {"1": 22737.68987506628, "2": 21060.634916997515, "3": 21005.673917010427, "4": 20843.06108392775, "5": 20883.118791040033, "6": 20894.568374962546, "7": 20802.81575000845, "8": 20576.631666976027}, "receiver_prefill_ms": 343.9322919584811, "ttft_ms": 52.874207962304354, "decode_ms": 43.02837501745671}, "total_wall_ms": 169244.72141696606, "ttft_ms": 52.874207962304354, "input_usage": {"lookups": 28}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 61, "facts_retained": 24, "retention_rate": 0.39344262295081966, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.4166666666666667, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1029}, "ledger_by_hop": {"1": {"facts_total": 61, "facts_retained": 24, "retention_rate": 0.39344262295081966, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.4166666666666667, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1161}, "2": {"facts_total": 61, "facts_retained": 24, "retention_rate": 0.39344262295081966, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.4166666666666667, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1045}, "3": {"facts_total": 61, "facts_retained": 24, "retention_rate": 0.39344262295081966, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.4166666666666667, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1042}, "4": {"facts_total": 61, "facts_retained": 24, "retention_rate": 0.39344262295081966, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.4166666666666667, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1029}, "5": {"facts_total": 61, "facts_retained": 24, "retention_rate": 0.39344262295081966, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.4166666666666667, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1029}, "6": {"facts_total": 61, "facts_retained": 24, "retention_rate": 0.39344262295081966, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.4166666666666667, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1029}, "7": {"facts_total": 61, "facts_retained": 24, "retention_rate": 0.39344262295081966, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.4166666666666667, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1029}, "8": {"facts_total": 61, "facts_retained": 24, "retention_rate": 0.39344262295081966, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.4166666666666667, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1029}}}, "reference_key_id": "keys/synthetic/syn_03015_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "13f688cc0243307a3eb9bb4da5cbeb4b020d6f83175f53e3d75ced3472f82704", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "a4fcb0fd0458c2cf3e098d17", "attempt_id": "c4ec78f8c3af", "run_group_id": "E4_core:A4|repairable|qwen3b:h8", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:57:01.649532Z", "task_family": "synthetic", "document_id": "syn_03015_2000", "document_hash": "71a7284b9bc3f681d69d91d18b9e8d45a4dda486cd14f4145395786ac6a68a89", "question_id": "q07_negation", "split": "test", "cluster_id": "syn_03015_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 8, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1922, "query_tokens": 564, "message_tokens": 481, "message_utf8_bytes": 1029, "cache_tensor_bytes": null, "serialized_payload_bytes": 1029, "side_information_bytes": 0, "fetched_bytes": 6216, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 168804.19437598903, "hops": {"1": 22737.68987506628, "2": 21060.634916997515, "3": 21005.673917010427, "4": 20843.06108392775, "5": 20883.118791040033, "6": 20894.568374962546, "7": 20802.81575000845, "8": 20576.631666976027}, "receiver_prefill_ms": 509.1615420533344, "ttft_ms": 33.15741696860641, "decode_ms": 41.48587503004819}, "total_wall_ms": 169388.7162100291, "ttft_ms": 33.15741696860641, "input_usage": {"lookups": 28}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 61, "facts_retained": 24, "retention_rate": 0.39344262295081966, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.4166666666666667, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1029}, "ledger_by_hop": {"1": {"facts_total": 61, "facts_retained": 24, "retention_rate": 0.39344262295081966, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.4166666666666667, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1161}, "2": {"facts_total": 61, "facts_retained": 24, "retention_rate": 0.39344262295081966, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.4166666666666667, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1045}, "3": {"facts_total": 61, "facts_retained": 24, "retention_rate": 0.39344262295081966, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.4166666666666667, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1042}, "4": {"facts_total": 61, "facts_retained": 24, "retention_rate": 0.39344262295081966, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.4166666666666667, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1029}, "5": {"facts_total": 61, "facts_retained": 24, "retention_rate": 0.39344262295081966, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.4166666666666667, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1029}, "6": {"facts_total": 61, "facts_retained": 24, "retention_rate": 0.39344262295081966, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.4166666666666667, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1029}, "7": {"facts_total": 61, "facts_retained": 24, "retention_rate": 0.39344262295081966, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.4166666666666667, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1029}, "8": {"facts_total": 61, "facts_retained": 24, "retention_rate": 0.39344262295081966, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.4166666666666667, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1029}}}, "reference_key_id": "keys/synthetic/syn_03015_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "13f688cc0243307a3eb9bb4da5cbeb4b020d6f83175f53e3d75ced3472f82704", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "a5ca6bc0700c0ad4738d5bf8", "attempt_id": "42618a969b09", "run_group_id": "E4_core:A4|repairable|qwen3b:h8", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:57:02.252503Z", "task_family": "synthetic", "document_id": "syn_03015_2000", "document_hash": "71a7284b9bc3f681d69d91d18b9e8d45a4dda486cd14f4145395786ac6a68a89", "question_id": "q08_revision", "split": "test", "cluster_id": "syn_03015_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 8, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1922, "query_tokens": 578, "message_tokens": 481, "message_utf8_bytes": 1029, "cache_tensor_bytes": null, "serialized_payload_bytes": 1029, "side_information_bytes": 0, "fetched_bytes": 6216, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 168804.19437598903, "hops": {"1": 22737.68987506628, "2": 21060.634916997515, "3": 21005.673917010427, "4": 20843.06108392775, "5": 20883.118791040033, "6": 20894.568374962546, "7": 20802.81575000845, "8": 20576.631666976027}, "receiver_prefill_ms": 396.0695000132546, "ttft_ms": 31.76441602408886, "decode_ms": 40.02158308867365}, "total_wall_ms": 169272.6161259925, "ttft_ms": 31.76441602408886, "input_usage": {"lookups": 28}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 61, "facts_retained": 24, "retention_rate": 0.39344262295081966, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.4166666666666667, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1029}, "ledger_by_hop": {"1": {"facts_total": 61, "facts_retained": 24, "retention_rate": 0.39344262295081966, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.4166666666666667, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1161}, "2": {"facts_total": 61, "facts_retained": 24, "retention_rate": 0.39344262295081966, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.4166666666666667, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1045}, "3": {"facts_total": 61, "facts_retained": 24, "retention_rate": 0.39344262295081966, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.4166666666666667, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1042}, "4": {"facts_total": 61, "facts_retained": 24, "retention_rate": 0.39344262295081966, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.4166666666666667, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1029}, "5": {"facts_total": 61, "facts_retained": 24, "retention_rate": 0.39344262295081966, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.4166666666666667, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1029}, "6": {"facts_total": 61, "facts_retained": 24, "retention_rate": 0.39344262295081966, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.4166666666666667, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1029}, "7": {"facts_total": 61, "facts_retained": 24, "retention_rate": 0.39344262295081966, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.4166666666666667, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1029}, "8": {"facts_total": 61, "facts_retained": 24, "retention_rate": 0.39344262295081966, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.4166666666666667, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1029}}}, "reference_key_id": "keys/synthetic/syn_03015_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "13f688cc0243307a3eb9bb4da5cbeb4b020d6f83175f53e3d75ced3472f82704", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "28484f64067d06d3db3762bc", "attempt_id": "5ed505c0b9b6", "run_group_id": "E4_core:A4|repairable|qwen3b:h8", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:57:02.739041Z", "task_family": "synthetic", "document_id": "syn_03015_2000", "document_hash": "71a7284b9bc3f681d69d91d18b9e8d45a4dda486cd14f4145395786ac6a68a89", "question_id": "q09_revision", "split": "test", "cluster_id": "syn_03015_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 8, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1922, "query_tokens": 572, "message_tokens": 481, "message_utf8_bytes": 1029, "cache_tensor_bytes": null, "serialized_payload_bytes": 1029, "side_information_bytes": 0, "fetched_bytes": 6216, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 168804.19437598903, "hops": {"1": 22737.68987506628, "2": 21060.634916997515, "3": 21005.673917010427, "4": 20843.06108392775, "5": 20883.118791040033, "6": 20894.568374962546, "7": 20802.81575000845, "8": 20576.631666976027}, "receiver_prefill_ms": 344.7286660084501, "ttft_ms": 35.96554207615554, "decode_ms": 574.2784580215812}, "total_wall_ms": 169759.7960429266, "ttft_ms": 35.96554207615554, "input_usage": {"lookups": 28}, "output_usage": {"output_tokens": 9}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "8,377,480", "evidence_artifact": {"ledger": {"facts_total": 61, "facts_retained": 24, "retention_rate": 0.39344262295081966, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.4166666666666667, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1029}, "ledger_by_hop": {"1": {"facts_total": 61, "facts_retained": 24, "retention_rate": 0.39344262295081966, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.4166666666666667, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1161}, "2": {"facts_total": 61, "facts_retained": 24, "retention_rate": 0.39344262295081966, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.4166666666666667, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1045}, "3": {"facts_total": 61, "facts_retained": 24, "retention_rate": 0.39344262295081966, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.4166666666666667, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1042}, "4": {"facts_total": 61, "facts_retained": 24, "retention_rate": 0.39344262295081966, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.4166666666666667, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1029}, "5": {"facts_total": 61, "facts_retained": 24, "retention_rate": 0.39344262295081966, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.4166666666666667, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1029}, "6": {"facts_total": 61, "facts_retained": 24, "retention_rate": 0.39344262295081966, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.4166666666666667, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1029}, "7": {"facts_total": 61, "facts_retained": 24, "retention_rate": 0.39344262295081966, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.4166666666666667, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1029}, "8": {"facts_total": 61, "facts_retained": 24, "retention_rate": 0.39344262295081966, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.4166666666666667, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1029}}}, "reference_key_id": "keys/synthetic/syn_03015_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "13f688cc0243307a3eb9bb4da5cbeb4b020d6f83175f53e3d75ced3472f82704", "output_hash": "ec4c68b327a2d2427ed21821415a5f9ddf87134bd9f654bdd69c7d1301febcb0"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "1b2e4222ba864c29038d1989", "attempt_id": "4269eb687dbe", "run_group_id": "E4_core:A4|repairable|qwen3b:h8", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:57:03.712809Z", "task_family": "synthetic", "document_id": "syn_03015_2000", "document_hash": "71a7284b9bc3f681d69d91d18b9e8d45a4dda486cd14f4145395786ac6a68a89", "question_id": "q10_temporal", "split": "test", "cluster_id": "syn_03015_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 8, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1922, "query_tokens": 581, "message_tokens": 481, "message_utf8_bytes": 1029, "cache_tensor_bytes": null, "serialized_payload_bytes": 1029, "side_information_bytes": 0, "fetched_bytes": 6216, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 168804.19437598903, "hops": {"1": 22737.68987506628, "2": 21060.634916997515, "3": 21005.673917010427, "4": 20843.06108392775, "5": 20883.118791040033, "6": 20894.568374962546, "7": 20802.81575000845, "8": 20576.631666976027}, "receiver_prefill_ms": 402.72274997550994, "ttft_ms": 22.323540993966162, "decode_ms": 44.20229198876768}, "total_wall_ms": 169274.0608759923, "ttft_ms": 22.323540993966162, "input_usage": {"lookups": 28}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 61, "facts_retained": 24, "retention_rate": 0.39344262295081966, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.4166666666666667, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1029}, "ledger_by_hop": {"1": {"facts_total": 61, "facts_retained": 24, "retention_rate": 0.39344262295081966, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.4166666666666667, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1161}, "2": {"facts_total": 61, "facts_retained": 24, "retention_rate": 0.39344262295081966, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.4166666666666667, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1045}, "3": {"facts_total": 61, "facts_retained": 24, "retention_rate": 0.39344262295081966, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.4166666666666667, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1042}, "4": {"facts_total": 61, "facts_retained": 24, "retention_rate": 0.39344262295081966, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.4166666666666667, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1029}, "5": {"facts_total": 61, "facts_retained": 24, "retention_rate": 0.39344262295081966, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.4166666666666667, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1029}, "6": {"facts_total": 61, "facts_retained": 24, "retention_rate": 0.39344262295081966, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.4166666666666667, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1029}, "7": {"facts_total": 61, "facts_retained": 24, "retention_rate": 0.39344262295081966, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.4166666666666667, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1029}, "8": {"facts_total": 61, "facts_retained": 24, "retention_rate": 0.39344262295081966, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.4166666666666667, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1029}}}, "reference_key_id": "keys/synthetic/syn_03015_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "13f688cc0243307a3eb9bb4da5cbeb4b020d6f83175f53e3d75ced3472f82704", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "518624dda71520441cc7a848", "attempt_id": "39b6ec82dd7a", "run_group_id": "E4_core:A4|repairable|qwen3b:h8", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:57:04.200217Z", "task_family": "synthetic", "document_id": "syn_03015_2000", "document_hash": "71a7284b9bc3f681d69d91d18b9e8d45a4dda486cd14f4145395786ac6a68a89", "question_id": "q11_aggregation", "split": "test", "cluster_id": "syn_03015_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 8, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1922, "query_tokens": 574, "message_tokens": 481, "message_utf8_bytes": 1029, "cache_tensor_bytes": null, "serialized_payload_bytes": 1029, "side_information_bytes": 0, "fetched_bytes": 6216, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 168804.19437598903, "hops": {"1": 22737.68987506628, "2": 21060.634916997515, "3": 21005.673917010427, "4": 20843.06108392775, "5": 20883.118791040033, "6": 20894.568374962546, "7": 20802.81575000845, "8": 20576.631666976027}, "receiver_prefill_ms": 363.0964580224827, "ttft_ms": 17.994291032664478, "decode_ms": 317.55704106763005}, "total_wall_ms": 169503.5013339948, "ttft_ms": 17.994291032664478, "input_usage": {"lookups": 28}, "output_usage": {"output_tokens": 7}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "1806025", "evidence_artifact": {"ledger": {"facts_total": 61, "facts_retained": 24, "retention_rate": 0.39344262295081966, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.4166666666666667, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1029}, "ledger_by_hop": {"1": {"facts_total": 61, "facts_retained": 24, "retention_rate": 0.39344262295081966, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.4166666666666667, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1161}, "2": {"facts_total": 61, "facts_retained": 24, "retention_rate": 0.39344262295081966, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.4166666666666667, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1045}, "3": {"facts_total": 61, "facts_retained": 24, "retention_rate": 0.39344262295081966, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.4166666666666667, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1042}, "4": {"facts_total": 61, "facts_retained": 24, "retention_rate": 0.39344262295081966, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.4166666666666667, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1029}, "5": {"facts_total": 61, "facts_retained": 24, "retention_rate": 0.39344262295081966, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.4166666666666667, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1029}, "6": {"facts_total": 61, "facts_retained": 24, "retention_rate": 0.39344262295081966, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.4166666666666667, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1029}, "7": {"facts_total": 61, "facts_retained": 24, "retention_rate": 0.39344262295081966, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.4166666666666667, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1029}, "8": {"facts_total": 61, "facts_retained": 24, "retention_rate": 0.39344262295081966, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.4166666666666667, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1029}}}, "reference_key_id": "keys/synthetic/syn_03015_2000.json", "task_scores": {"correct": 0.0, "abstained": 0.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "13f688cc0243307a3eb9bb4da5cbeb4b020d6f83175f53e3d75ced3472f82704", "output_hash": "9801d1840792d286693349934c9cc13fead44d82d480a42803cf6015c01431f6"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "20666029b5d991ac6a4752ce", "attempt_id": "76ce82604af1", "run_group_id": "E4_core:A4|repairable|qwen3b:h8", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:57:04.916766Z", "task_family": "synthetic", "document_id": "syn_03015_2000", "document_hash": "71a7284b9bc3f681d69d91d18b9e8d45a4dda486cd14f4145395786ac6a68a89", "question_id": "q12_unit_conversion", "split": "test", "cluster_id": "syn_03015_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 8, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1922, "query_tokens": 571, "message_tokens": 481, "message_utf8_bytes": 1029, "cache_tensor_bytes": null, "serialized_payload_bytes": 1029, "side_information_bytes": 0, "fetched_bytes": 6216, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 168804.19437598903, "hops": {"1": 22737.68987506628, "2": 21060.634916997515, "3": 21005.673917010427, "4": 20843.06108392775, "5": 20883.118791040033, "6": 20894.568374962546, "7": 20802.81575000845, "8": 20576.631666976027}, "receiver_prefill_ms": 526.8174159573391, "ttft_ms": 31.45966597367078, "decode_ms": 42.90829098317772}, "total_wall_ms": 169406.1027088901, "ttft_ms": 31.45966597367078, "input_usage": {"lookups": 28}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 61, "facts_retained": 24, "retention_rate": 0.39344262295081966, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.4166666666666667, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1029}, "ledger_by_hop": {"1": {"facts_total": 61, "facts_retained": 24, "retention_rate": 0.39344262295081966, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.4166666666666667, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1161}, "2": {"facts_total": 61, "facts_retained": 24, "retention_rate": 0.39344262295081966, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.4166666666666667, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1045}, "3": {"facts_total": 61, "facts_retained": 24, "retention_rate": 0.39344262295081966, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.4166666666666667, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1042}, "4": {"facts_total": 61, "facts_retained": 24, "retention_rate": 0.39344262295081966, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.4166666666666667, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1029}, "5": {"facts_total": 61, "facts_retained": 24, "retention_rate": 0.39344262295081966, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.4166666666666667, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1029}, "6": {"facts_total": 61, "facts_retained": 24, "retention_rate": 0.39344262295081966, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.4166666666666667, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1029}, "7": {"facts_total": 61, "facts_retained": 24, "retention_rate": 0.39344262295081966, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.4166666666666667, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1029}, "8": {"facts_total": 61, "facts_retained": 24, "retention_rate": 0.39344262295081966, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.4166666666666667, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1029}}}, "reference_key_id": "keys/synthetic/syn_03015_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "13f688cc0243307a3eb9bb4da5cbeb4b020d6f83175f53e3d75ced3472f82704", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "2ab9085bbaafb6f837502606", "attempt_id": "7408d89511f1", "run_group_id": "E4_core:A4|repairable|qwen3b:h8", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:57:05.536296Z", "task_family": "synthetic", "document_id": "syn_03015_2000", "document_hash": "71a7284b9bc3f681d69d91d18b9e8d45a4dda486cd14f4145395786ac6a68a89", "question_id": "q13_unanswerable", "split": "test", "cluster_id": "syn_03015_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 8, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1922, "query_tokens": 575, "message_tokens": 481, "message_utf8_bytes": 1029, "cache_tensor_bytes": null, "serialized_payload_bytes": 1029, "side_information_bytes": 0, "fetched_bytes": 6216, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 168804.19437598903, "hops": {"1": 22737.68987506628, "2": 21060.634916997515, "3": 21005.673917010427, "4": 20843.06108392775, "5": 20883.118791040033, "6": 20894.568374962546, "7": 20802.81575000845, "8": 20576.631666976027}, "receiver_prefill_ms": 388.8722079573199, "ttft_ms": 27.72408409509808, "decode_ms": 36.55170800630003}, "total_wall_ms": 169257.9955010442, "ttft_ms": 27.72408409509808, "input_usage": {"lookups": 28}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 61, "facts_retained": 24, "retention_rate": 0.39344262295081966, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.4166666666666667, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1029}, "ledger_by_hop": {"1": {"facts_total": 61, "facts_retained": 24, "retention_rate": 0.39344262295081966, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.4166666666666667, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1161}, "2": {"facts_total": 61, "facts_retained": 24, "retention_rate": 0.39344262295081966, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.4166666666666667, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1045}, "3": {"facts_total": 61, "facts_retained": 24, "retention_rate": 0.39344262295081966, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.4166666666666667, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1042}, "4": {"facts_total": 61, "facts_retained": 24, "retention_rate": 0.39344262295081966, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.4166666666666667, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1029}, "5": {"facts_total": 61, "facts_retained": 24, "retention_rate": 0.39344262295081966, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.4166666666666667, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1029}, "6": {"facts_total": 61, "facts_retained": 24, "retention_rate": 0.39344262295081966, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.4166666666666667, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1029}, "7": {"facts_total": 61, "facts_retained": 24, "retention_rate": 0.39344262295081966, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.4166666666666667, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1029}, "8": {"facts_total": 61, "facts_retained": 24, "retention_rate": 0.39344262295081966, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.4166666666666667, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1029}}}, "reference_key_id": "keys/synthetic/syn_03015_2000.json", "task_scores": {"correct": 1.0, "abstained": 1.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "13f688cc0243307a3eb9bb4da5cbeb4b020d6f83175f53e3d75ced3472f82704", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "71e88ce5bf5d6ab4d9d2c929", "attempt_id": "7d254e87300e", "run_group_id": "E4_core:A4|repairable|qwen3b:h8", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:57:06.007362Z", "task_family": "synthetic", "document_id": "syn_03015_2000", "document_hash": "71a7284b9bc3f681d69d91d18b9e8d45a4dda486cd14f4145395786ac6a68a89", "question_id": "q14_unanswerable", "split": "test", "cluster_id": "syn_03015_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 8, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1922, "query_tokens": 576, "message_tokens": 481, "message_utf8_bytes": 1029, "cache_tensor_bytes": null, "serialized_payload_bytes": 1029, "side_information_bytes": 0, "fetched_bytes": 6216, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 168804.19437598903, "hops": {"1": 22737.68987506628, "2": 21060.634916997515, "3": 21005.673917010427, "4": 20843.06108392775, "5": 20883.118791040033, "6": 20894.568374962546, "7": 20802.81575000845, "8": 20576.631666976027}, "receiver_prefill_ms": 357.59774991311133, "ttft_ms": 43.617250048555434, "decode_ms": 49.495500046759844}, "total_wall_ms": 169255.6419589091, "ttft_ms": 43.617250048555434, "input_usage": {"lookups": 28}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 61, "facts_retained": 24, "retention_rate": 0.39344262295081966, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.4166666666666667, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1029}, "ledger_by_hop": {"1": {"facts_total": 61, "facts_retained": 24, "retention_rate": 0.39344262295081966, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.4166666666666667, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1161}, "2": {"facts_total": 61, "facts_retained": 24, "retention_rate": 0.39344262295081966, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.4166666666666667, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1045}, "3": {"facts_total": 61, "facts_retained": 24, "retention_rate": 0.39344262295081966, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.4166666666666667, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1042}, "4": {"facts_total": 61, "facts_retained": 24, "retention_rate": 0.39344262295081966, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.4166666666666667, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1029}, "5": {"facts_total": 61, "facts_retained": 24, "retention_rate": 0.39344262295081966, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.4166666666666667, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1029}, "6": {"facts_total": 61, "facts_retained": 24, "retention_rate": 0.39344262295081966, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.4166666666666667, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1029}, "7": {"facts_total": 61, "facts_retained": 24, "retention_rate": 0.39344262295081966, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.4166666666666667, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1029}, "8": {"facts_total": 61, "facts_retained": 24, "retention_rate": 0.39344262295081966, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.4166666666666667, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1029}}}, "reference_key_id": "keys/synthetic/syn_03015_2000.json", "task_scores": {"correct": 1.0, "abstained": 1.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "13f688cc0243307a3eb9bb4da5cbeb4b020d6f83175f53e3d75ced3472f82704", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "fb37665a685c5890a7242626", "attempt_id": "7de46e08655b", "run_group_id": "E4_core:A5|repairable|qwen3b:h1", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:57:29.469812Z", "task_family": "synthetic", "document_id": "syn_03015_2000", "document_hash": "71a7284b9bc3f681d69d91d18b9e8d45a4dda486cd14f4145395786ac6a68a89", "question_id": "q00_single_fact", "split": "test", "cluster_id": "syn_03015_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 1, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1922, "query_tokens": 411, "message_tokens": 328, "message_utf8_bytes": 754, "cache_tensor_bytes": null, "serialized_payload_bytes": 754, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 22991.516749956645, "hops": {"1": 22991.516749956645}, "receiver_prefill_ms": 294.52754196245223, "ttft_ms": 38.757957983762026, "decode_ms": 32.40383299998939}, "total_wall_ms": 23357.780874939635, "ttft_ms": 38.757957983762026, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 61, "facts_retained": 7, "retention_rate": 0.11475409836065574, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.3076923076923077, "status": 0.0, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 690}, "ledger_by_hop": {"1": {"facts_total": 61, "facts_retained": 7, "retention_rate": 0.11475409836065574, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.3076923076923077, "status": 0.0, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 690}}}, "reference_key_id": "keys/synthetic/syn_03015_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "e54cba3528ab78e09d5da03579b2a2035abb56934b1d0571c12acb32bf58b543", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "9589042c65feeeda46fddcfd", "attempt_id": "fd702b5b8bb7", "run_group_id": "E4_core:A5|repairable|qwen3b:h1", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:57:29.853583Z", "task_family": "synthetic", "document_id": "syn_03015_2000", "document_hash": "71a7284b9bc3f681d69d91d18b9e8d45a4dda486cd14f4145395786ac6a68a89", "question_id": "q01_single_fact", "split": "test", "cluster_id": "syn_03015_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 1, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1922, "query_tokens": 416, "message_tokens": 328, "message_utf8_bytes": 754, "cache_tensor_bytes": null, "serialized_payload_bytes": 754, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 22991.516749956645, "hops": {"1": 22991.516749956645}, "receiver_prefill_ms": 306.8722919560969, "ttft_ms": 57.75904096662998, "decode_ms": 65.62733300961554}, "total_wall_ms": 23422.39412490744, "ttft_ms": 57.75904096662998, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 61, "facts_retained": 7, "retention_rate": 0.11475409836065574, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.3076923076923077, "status": 0.0, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 690}, "ledger_by_hop": {"1": {"facts_total": 61, "facts_retained": 7, "retention_rate": 0.11475409836065574, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.3076923076923077, "status": 0.0, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 690}}}, "reference_key_id": "keys/synthetic/syn_03015_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "e54cba3528ab78e09d5da03579b2a2035abb56934b1d0571c12acb32bf58b543", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "65442777ae2214c4b6466930", "attempt_id": "98d25c02e33a", "run_group_id": "E4_core:A5|repairable|qwen3b:h1", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:57:30.303620Z", "task_family": "synthetic", "document_id": "syn_03015_2000", "document_hash": "71a7284b9bc3f681d69d91d18b9e8d45a4dda486cd14f4145395786ac6a68a89", "question_id": "q02_exact_copy", "split": "test", "cluster_id": "syn_03015_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 1, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1922, "query_tokens": 415, "message_tokens": 328, "message_utf8_bytes": 754, "cache_tensor_bytes": null, "serialized_payload_bytes": 754, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 22991.516749956645, "hops": {"1": 22991.516749956645}, "receiver_prefill_ms": 412.88137494120747, "ttft_ms": 41.32941598072648, "decode_ms": 34.53054197598249}, "total_wall_ms": 23480.78404192347, "ttft_ms": 41.32941598072648, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 61, "facts_retained": 7, "retention_rate": 0.11475409836065574, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.3076923076923077, "status": 0.0, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 690}, "ledger_by_hop": {"1": {"facts_total": 61, "facts_retained": 7, "retention_rate": 0.11475409836065574, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.3076923076923077, "status": 0.0, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 690}}}, "reference_key_id": "keys/synthetic/syn_03015_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "e54cba3528ab78e09d5da03579b2a2035abb56934b1d0571c12acb32bf58b543", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "f965862c674c242f44b18c1b", "attempt_id": "5c801a010eae", "run_group_id": "E4_core:A5|repairable|qwen3b:h1", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:57:30.813991Z", "task_family": "synthetic", "document_id": "syn_03015_2000", "document_hash": "71a7284b9bc3f681d69d91d18b9e8d45a4dda486cd14f4145395786ac6a68a89", "question_id": "q03_exact_copy", "split": "test", "cluster_id": "syn_03015_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 1, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1922, "query_tokens": 411, "message_tokens": 328, "message_utf8_bytes": 754, "cache_tensor_bytes": null, "serialized_payload_bytes": 754, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 22991.516749956645, "hops": {"1": 22991.516749956645}, "receiver_prefill_ms": 282.8732499619946, "ttft_ms": 32.47512492816895, "decode_ms": 221.31216700654477}, "total_wall_ms": 23528.754416038282, "ttft_ms": 32.47512492816895, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 6}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "Wilhelmina Dubrovnik", "evidence_artifact": {"ledger": {"facts_total": 61, "facts_retained": 7, "retention_rate": 0.11475409836065574, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.3076923076923077, "status": 0.0, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 690}, "ledger_by_hop": {"1": {"facts_total": 61, "facts_retained": 7, "retention_rate": 0.11475409836065574, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.3076923076923077, "status": 0.0, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 690}}}, "reference_key_id": "keys/synthetic/syn_03015_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "e54cba3528ab78e09d5da03579b2a2035abb56934b1d0571c12acb32bf58b543", "output_hash": "f42f05ebfc96f7258cbfbbb37cd65e114d3d15f869e68e6c95b71fc288c30a32"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "9ba77a65f0673ecd3d66e07d", "attempt_id": "f30838788a1b", "run_group_id": "E4_core:A5|repairable|qwen3b:h1", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:57:31.368127Z", "task_family": "synthetic", "document_id": "syn_03015_2000", "document_hash": "71a7284b9bc3f681d69d91d18b9e8d45a4dda486cd14f4145395786ac6a68a89", "question_id": "q04_multi_hop", "split": "test", "cluster_id": "syn_03015_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 1, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1922, "query_tokens": 420, "message_tokens": 328, "message_utf8_bytes": 754, "cache_tensor_bytes": null, "serialized_payload_bytes": 754, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 22991.516749956645, "hops": {"1": 22991.516749956645}, "receiver_prefill_ms": 274.7743329964578, "ttft_ms": 62.91604205034673, "decode_ms": 503.6843749694526}, "total_wall_ms": 23833.48216698505, "ttft_ms": 62.91604205034673, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 8}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "8421725 EUR", "evidence_artifact": {"ledger": {"facts_total": 61, "facts_retained": 7, "retention_rate": 0.11475409836065574, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.3076923076923077, "status": 0.0, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 690}, "ledger_by_hop": {"1": {"facts_total": 61, "facts_retained": 7, "retention_rate": 0.11475409836065574, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.3076923076923077, "status": 0.0, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 690}}}, "reference_key_id": "keys/synthetic/syn_03015_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 0.6666666666666666}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "e54cba3528ab78e09d5da03579b2a2035abb56934b1d0571c12acb32bf58b543", "output_hash": "976c8be2c1a175a4fb5643c5785f93e27ad9205cebd9569a64b821e1ae51ac0e"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "a1e1cd422dc0fb760e5b8a86", "attempt_id": "a9eb29edef18", "run_group_id": "E4_core:A5|repairable|qwen3b:h1", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:57:32.228327Z", "task_family": "synthetic", "document_id": "syn_03015_2000", "document_hash": "71a7284b9bc3f681d69d91d18b9e8d45a4dda486cd14f4145395786ac6a68a89", "question_id": "q05_multi_hop", "split": "test", "cluster_id": "syn_03015_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 1, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1922, "query_tokens": 420, "message_tokens": 328, "message_utf8_bytes": 754, "cache_tensor_bytes": null, "serialized_payload_bytes": 754, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 22991.516749956645, "hops": {"1": 22991.516749956645}, "receiver_prefill_ms": 309.85341710038483, "ttft_ms": 41.21920792385936, "decode_ms": 26.079750037752092}, "total_wall_ms": 23369.127124897204, "ttft_ms": 41.21920792385936, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 61, "facts_retained": 7, "retention_rate": 0.11475409836065574, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.3076923076923077, "status": 0.0, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 690}, "ledger_by_hop": {"1": {"facts_total": 61, "facts_retained": 7, "retention_rate": 0.11475409836065574, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.3076923076923077, "status": 0.0, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 690}}}, "reference_key_id": "keys/synthetic/syn_03015_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "e54cba3528ab78e09d5da03579b2a2035abb56934b1d0571c12acb32bf58b543", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "2ea9180a7c8d86ad69d9aa70", "attempt_id": "3ec66a52785d", "run_group_id": "E4_core:A5|repairable|qwen3b:h1", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:57:32.623141Z", "task_family": "synthetic", "document_id": "syn_03015_2000", "document_hash": "71a7284b9bc3f681d69d91d18b9e8d45a4dda486cd14f4145395786ac6a68a89", "question_id": "q06_negation", "split": "test", "cluster_id": "syn_03015_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 1, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1922, "query_tokens": 412, "message_tokens": 328, "message_utf8_bytes": 754, "cache_tensor_bytes": null, "serialized_payload_bytes": 754, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 22991.516749956645, "hops": {"1": 22991.516749956645}, "receiver_prefill_ms": 296.68908403255045, "ttft_ms": 26.3897500699386, "decode_ms": 38.418374955654144}, "total_wall_ms": 23353.44208299648, "ttft_ms": 26.3897500699386, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 61, "facts_retained": 7, "retention_rate": 0.11475409836065574, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.3076923076923077, "status": 0.0, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 690}, "ledger_by_hop": {"1": {"facts_total": 61, "facts_retained": 7, "retention_rate": 0.11475409836065574, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.3076923076923077, "status": 0.0, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 690}}}, "reference_key_id": "keys/synthetic/syn_03015_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "e54cba3528ab78e09d5da03579b2a2035abb56934b1d0571c12acb32bf58b543", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "e69e678106db4def13ae89b6", "attempt_id": "6fe9ef823377", "run_group_id": "E4_core:A5|repairable|qwen3b:h1", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:57:33.002504Z", "task_family": "synthetic", "document_id": "syn_03015_2000", "document_hash": "71a7284b9bc3f681d69d91d18b9e8d45a4dda486cd14f4145395786ac6a68a89", "question_id": "q07_negation", "split": "test", "cluster_id": "syn_03015_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 1, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1922, "query_tokens": 412, "message_tokens": 328, "message_utf8_bytes": 754, "cache_tensor_bytes": null, "serialized_payload_bytes": 754, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 22991.516749956645, "hops": {"1": 22991.516749956645}, "receiver_prefill_ms": 273.18529097829014, "ttft_ms": 56.82129203341901, "decode_ms": 62.44400003924966}, "total_wall_ms": 23384.687540936284, "ttft_ms": 56.82129203341901, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 61, "facts_retained": 7, "retention_rate": 0.11475409836065574, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.3076923076923077, "status": 0.0, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 690}, "ledger_by_hop": {"1": {"facts_total": 61, "facts_retained": 7, "retention_rate": 0.11475409836065574, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.3076923076923077, "status": 0.0, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 690}}}, "reference_key_id": "keys/synthetic/syn_03015_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "e54cba3528ab78e09d5da03579b2a2035abb56934b1d0571c12acb32bf58b543", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "4bb401aa7dff1851f8b5115b", "attempt_id": "206955717fab", "run_group_id": "E4_core:A5|repairable|qwen3b:h1", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:57:33.414412Z", "task_family": "synthetic", "document_id": "syn_03015_2000", "document_hash": "71a7284b9bc3f681d69d91d18b9e8d45a4dda486cd14f4145395786ac6a68a89", "question_id": "q08_revision", "split": "test", "cluster_id": "syn_03015_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 1, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1922, "query_tokens": 426, "message_tokens": 328, "message_utf8_bytes": 754, "cache_tensor_bytes": null, "serialized_payload_bytes": 754, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 22991.516749956645, "hops": {"1": 22991.516749956645}, "receiver_prefill_ms": 440.78816601540893, "ttft_ms": 24.671541061252356, "decode_ms": 39.478791994042695}, "total_wall_ms": 23496.997999958694, "ttft_ms": 24.671541061252356, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 61, "facts_retained": 7, "retention_rate": 0.11475409836065574, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.3076923076923077, "status": 0.0, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 690}, "ledger_by_hop": {"1": {"facts_total": 61, "facts_retained": 7, "retention_rate": 0.11475409836065574, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.3076923076923077, "status": 0.0, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 690}}}, "reference_key_id": "keys/synthetic/syn_03015_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "e54cba3528ab78e09d5da03579b2a2035abb56934b1d0571c12acb32bf58b543", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "9019e5b0d2e9116b8f99edc9", "attempt_id": "7c593849511b", "run_group_id": "E4_core:A5|repairable|qwen3b:h1", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:57:33.937141Z", "task_family": "synthetic", "document_id": "syn_03015_2000", "document_hash": "71a7284b9bc3f681d69d91d18b9e8d45a4dda486cd14f4145395786ac6a68a89", "question_id": "q09_revision", "split": "test", "cluster_id": "syn_03015_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 1, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1922, "query_tokens": 420, "message_tokens": 328, "message_utf8_bytes": 754, "cache_tensor_bytes": null, "serialized_payload_bytes": 754, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 22991.516749956645, "hops": {"1": 22991.516749956645}, "receiver_prefill_ms": 312.1711660642177, "ttft_ms": 38.31520804669708, "decode_ms": 217.85570902284235}, "total_wall_ms": 23560.402999981306, "ttft_ms": 38.31520804669708, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 7}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "7028925", "evidence_artifact": {"ledger": {"facts_total": 61, "facts_retained": 7, "retention_rate": 0.11475409836065574, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.3076923076923077, "status": 0.0, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 690}, "ledger_by_hop": {"1": {"facts_total": 61, "facts_retained": 7, "retention_rate": 0.11475409836065574, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.3076923076923077, "status": 0.0, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 690}}}, "reference_key_id": "keys/synthetic/syn_03015_2000.json", "task_scores": {"correct": 0.0, "abstained": 0.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "e54cba3528ab78e09d5da03579b2a2035abb56934b1d0571c12acb32bf58b543", "output_hash": "e966a398855c104a86bce07742c3ce8108a432926bc704940ecfd07e44596774"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "82f555625defe61fca1575ca", "attempt_id": "e2b6ee99d7b3", "run_group_id": "E4_core:A5|repairable|qwen3b:h1", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:57:34.524157Z", "task_family": "synthetic", "document_id": "syn_03015_2000", "document_hash": "71a7284b9bc3f681d69d91d18b9e8d45a4dda486cd14f4145395786ac6a68a89", "question_id": "q10_temporal", "split": "test", "cluster_id": "syn_03015_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 1, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1922, "query_tokens": 429, "message_tokens": 328, "message_utf8_bytes": 754, "cache_tensor_bytes": null, "serialized_payload_bytes": 754, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 22991.516749956645, "hops": {"1": 22991.516749956645}, "receiver_prefill_ms": 259.58425004500896, "ttft_ms": 44.45616703014821, "decode_ms": 51.13229202106595}, "total_wall_ms": 23347.394790966064, "ttft_ms": 44.45616703014821, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 61, "facts_retained": 7, "retention_rate": 0.11475409836065574, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.3076923076923077, "status": 0.0, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 690}, "ledger_by_hop": {"1": {"facts_total": 61, "facts_retained": 7, "retention_rate": 0.11475409836065574, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.3076923076923077, "status": 0.0, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 690}}}, "reference_key_id": "keys/synthetic/syn_03015_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "e54cba3528ab78e09d5da03579b2a2035abb56934b1d0571c12acb32bf58b543", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "0a1570b56fb1870e1b6e4513", "attempt_id": "5675a19a7a13", "run_group_id": "E4_core:A5|repairable|qwen3b:h1", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:57:34.897797Z", "task_family": "synthetic", "document_id": "syn_03015_2000", "document_hash": "71a7284b9bc3f681d69d91d18b9e8d45a4dda486cd14f4145395786ac6a68a89", "question_id": "q11_aggregation", "split": "test", "cluster_id": "syn_03015_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 1, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1922, "query_tokens": 422, "message_tokens": 328, "message_utf8_bytes": 754, "cache_tensor_bytes": null, "serialized_payload_bytes": 754, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 22991.516749956645, "hops": {"1": 22991.516749956645}, "receiver_prefill_ms": 462.99700008239597, "ttft_ms": 37.38875000271946, "decode_ms": 39.45420903619379}, "total_wall_ms": 23531.89066599589, "ttft_ms": 37.38875000271946, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 61, "facts_retained": 7, "retention_rate": 0.11475409836065574, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.3076923076923077, "status": 0.0, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 690}, "ledger_by_hop": {"1": {"facts_total": 61, "facts_retained": 7, "retention_rate": 0.11475409836065574, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.3076923076923077, "status": 0.0, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 690}}}, "reference_key_id": "keys/synthetic/syn_03015_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "e54cba3528ab78e09d5da03579b2a2035abb56934b1d0571c12acb32bf58b543", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "6b52d7823cc5a1d0dc790319", "attempt_id": "9ed23f6ea495", "run_group_id": "E4_core:A5|repairable|qwen3b:h1", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:57:35.456327Z", "task_family": "synthetic", "document_id": "syn_03015_2000", "document_hash": "71a7284b9bc3f681d69d91d18b9e8d45a4dda486cd14f4145395786ac6a68a89", "question_id": "q12_unit_conversion", "split": "test", "cluster_id": "syn_03015_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 1, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1922, "query_tokens": 419, "message_tokens": 328, "message_utf8_bytes": 754, "cache_tensor_bytes": null, "serialized_payload_bytes": 754, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 22991.516749956645, "hops": {"1": 22991.516749956645}, "receiver_prefill_ms": 289.02287501841784, "ttft_ms": 38.99304091464728, "decode_ms": 41.626333026215434}, "total_wall_ms": 23361.74191592727, "ttft_ms": 38.99304091464728, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 61, "facts_retained": 7, "retention_rate": 0.11475409836065574, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.3076923076923077, "status": 0.0, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 690}, "ledger_by_hop": {"1": {"facts_total": 61, "facts_retained": 7, "retention_rate": 0.11475409836065574, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.3076923076923077, "status": 0.0, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 690}}}, "reference_key_id": "keys/synthetic/syn_03015_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "e54cba3528ab78e09d5da03579b2a2035abb56934b1d0571c12acb32bf58b543", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "88a85842a38d21b61ab9ffaf", "attempt_id": "19e19a3e4937", "run_group_id": "E4_core:A5|repairable|qwen3b:h1", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:57:35.844786Z", "task_family": "synthetic", "document_id": "syn_03015_2000", "document_hash": "71a7284b9bc3f681d69d91d18b9e8d45a4dda486cd14f4145395786ac6a68a89", "question_id": "q13_unanswerable", "split": "test", "cluster_id": "syn_03015_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 1, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1922, "query_tokens": 423, "message_tokens": 328, "message_utf8_bytes": 754, "cache_tensor_bytes": null, "serialized_payload_bytes": 754, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 22991.516749956645, "hops": {"1": 22991.516749956645}, "receiver_prefill_ms": 294.93908397853374, "ttft_ms": 28.45458302181214, "decode_ms": 40.47062492463738}, "total_wall_ms": 23355.893416912295, "ttft_ms": 28.45458302181214, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 61, "facts_retained": 7, "retention_rate": 0.11475409836065574, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.3076923076923077, "status": 0.0, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 690}, "ledger_by_hop": {"1": {"facts_total": 61, "facts_retained": 7, "retention_rate": 0.11475409836065574, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.3076923076923077, "status": 0.0, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 690}}}, "reference_key_id": "keys/synthetic/syn_03015_2000.json", "task_scores": {"correct": 1.0, "abstained": 1.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "e54cba3528ab78e09d5da03579b2a2035abb56934b1d0571c12acb32bf58b543", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "8ae070ad933cde1f279472f1", "attempt_id": "3e3fc5b0bf10", "run_group_id": "E4_core:A5|repairable|qwen3b:h1", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:57:36.226001Z", "task_family": "synthetic", "document_id": "syn_03015_2000", "document_hash": "71a7284b9bc3f681d69d91d18b9e8d45a4dda486cd14f4145395786ac6a68a89", "question_id": "q14_unanswerable", "split": "test", "cluster_id": "syn_03015_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 1, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1922, "query_tokens": 424, "message_tokens": 328, "message_utf8_bytes": 754, "cache_tensor_bytes": null, "serialized_payload_bytes": 754, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 22991.516749956645, "hops": {"1": 22991.516749956645}, "receiver_prefill_ms": 299.5375000173226, "ttft_ms": 55.54891692008823, "decode_ms": 47.500957967713475}, "total_wall_ms": 23394.674666924402, "ttft_ms": 55.54891692008823, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 61, "facts_retained": 7, "retention_rate": 0.11475409836065574, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.3076923076923077, "status": 0.0, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 690}, "ledger_by_hop": {"1": {"facts_total": 61, "facts_retained": 7, "retention_rate": 0.11475409836065574, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.3076923076923077, "status": 0.0, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 690}}}, "reference_key_id": "keys/synthetic/syn_03015_2000.json", "task_scores": {"correct": 1.0, "abstained": 1.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "e54cba3528ab78e09d5da03579b2a2035abb56934b1d0571c12acb32bf58b543", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "6a85deb3b2e244bae98a0c77", "attempt_id": "54e8b7e8df8b", "run_group_id": "E4_core:A5|repairable|qwen3b:h2", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:57:49.814056Z", "task_family": "synthetic", "document_id": "syn_03015_2000", "document_hash": "71a7284b9bc3f681d69d91d18b9e8d45a4dda486cd14f4145395786ac6a68a89", "question_id": "q00_single_fact", "split": "test", "cluster_id": "syn_03015_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 2, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1922, "query_tokens": 376, "message_tokens": 294, "message_utf8_bytes": 678, "cache_tensor_bytes": null, "serialized_payload_bytes": 678, "side_information_bytes": 0, "fetched_bytes": 1207, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 36158.07445789687, "hops": {"1": 22991.516749956645, "2": 13166.557707940228}, "receiver_prefill_ms": 293.77604194451123, "ttft_ms": 21.046290989033878, "decode_ms": 20.215374999679625}, "total_wall_ms": 36493.55862487573, "ttft_ms": 21.046290989033878, "input_usage": {"lookups": 2}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 61, "facts_retained": 6, "retention_rate": 0.09836065573770492, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.0, "amount": 0.3076923076923077, "status": 0.0, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 624}, "ledger_by_hop": {"1": {"facts_total": 61, "facts_retained": 7, "retention_rate": 0.11475409836065574, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.3076923076923077, "status": 0.0, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 690}, "2": {"facts_total": 61, "facts_retained": 6, "retention_rate": 0.09836065573770492, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.0, "amount": 0.3076923076923077, "status": 0.0, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 624}}}, "reference_key_id": "keys/synthetic/syn_03015_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "ac61ed4025051e85e49f3f352d3d60c733554cb217faf29f390436fb9572a503", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "af4758a68eda86c12304798d", "attempt_id": "ff3cf1064840", "run_group_id": "E4_core:A5|repairable|qwen3b:h2", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:57:50.167350Z", "task_family": "synthetic", "document_id": "syn_03015_2000", "document_hash": "71a7284b9bc3f681d69d91d18b9e8d45a4dda486cd14f4145395786ac6a68a89", "question_id": "q01_single_fact", "split": "test", "cluster_id": "syn_03015_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 2, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1922, "query_tokens": 381, "message_tokens": 294, "message_utf8_bytes": 678, "cache_tensor_bytes": null, "serialized_payload_bytes": 678, "side_information_bytes": 0, "fetched_bytes": 1207, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 36158.07445789687, "hops": {"1": 22991.516749956645, "2": 13166.557707940228}, "receiver_prefill_ms": 285.1546669844538, "ttft_ms": 73.86624999344349, "decode_ms": 44.91516703274101}, "total_wall_ms": 36562.82379082404, "ttft_ms": 73.86624999344349, "input_usage": {"lookups": 2}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 61, "facts_retained": 6, "retention_rate": 0.09836065573770492, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.0, "amount": 0.3076923076923077, "status": 0.0, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 624}, "ledger_by_hop": {"1": {"facts_total": 61, "facts_retained": 7, "retention_rate": 0.11475409836065574, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.3076923076923077, "status": 0.0, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 690}, "2": {"facts_total": 61, "facts_retained": 6, "retention_rate": 0.09836065573770492, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.0, "amount": 0.3076923076923077, "status": 0.0, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 624}}}, "reference_key_id": "keys/synthetic/syn_03015_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "ac61ed4025051e85e49f3f352d3d60c733554cb217faf29f390436fb9572a503", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "903376151630b10204e9ed95", "attempt_id": "c74a1418a393", "run_group_id": "E4_core:A5|repairable|qwen3b:h2", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:57:50.592168Z", "task_family": "synthetic", "document_id": "syn_03015_2000", "document_hash": "71a7284b9bc3f681d69d91d18b9e8d45a4dda486cd14f4145395786ac6a68a89", "question_id": "q02_exact_copy", "split": "test", "cluster_id": "syn_03015_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 2, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1922, "query_tokens": 380, "message_tokens": 294, "message_utf8_bytes": 678, "cache_tensor_bytes": null, "serialized_payload_bytes": 678, "side_information_bytes": 0, "fetched_bytes": 1207, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 36158.07445789687, "hops": {"1": 22991.516749956645, "2": 13166.557707940228}, "receiver_prefill_ms": 423.7544590141624, "ttft_ms": 23.531083948910236, "decode_ms": 36.273582954891026}, "total_wall_ms": 36642.22537493333, "ttft_ms": 23.531083948910236, "input_usage": {"lookups": 2}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 61, "facts_retained": 6, "retention_rate": 0.09836065573770492, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.0, "amount": 0.3076923076923077, "status": 0.0, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 624}, "ledger_by_hop": {"1": {"facts_total": 61, "facts_retained": 7, "retention_rate": 0.11475409836065574, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.3076923076923077, "status": 0.0, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 690}, "2": {"facts_total": 61, "facts_retained": 6, "retention_rate": 0.09836065573770492, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.0, "amount": 0.3076923076923077, "status": 0.0, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 624}}}, "reference_key_id": "keys/synthetic/syn_03015_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "ac61ed4025051e85e49f3f352d3d60c733554cb217faf29f390436fb9572a503", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "63b251824732b4f7d49a2dfe", "attempt_id": "72ee5a06b429", "run_group_id": "E4_core:A5|repairable|qwen3b:h2", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:57:51.093882Z", "task_family": "synthetic", "document_id": "syn_03015_2000", "document_hash": "71a7284b9bc3f681d69d91d18b9e8d45a4dda486cd14f4145395786ac6a68a89", "question_id": "q03_exact_copy", "split": "test", "cluster_id": "syn_03015_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 2, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1922, "query_tokens": 376, "message_tokens": 294, "message_utf8_bytes": 678, "cache_tensor_bytes": null, "serialized_payload_bytes": 678, "side_information_bytes": 0, "fetched_bytes": 1207, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 36158.07445789687, "hops": {"1": 22991.516749956645, "2": 13166.557707940228}, "receiver_prefill_ms": 280.85212502628565, "ttft_ms": 38.19062502589077, "decode_ms": 206.56216703355312}, "total_wall_ms": 36684.2431249097, "ttft_ms": 38.19062502589077, "input_usage": {"lookups": 2}, "output_usage": {"output_tokens": 6}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "Wilhelmina Dubrovnik", "evidence_artifact": {"ledger": {"facts_total": 61, "facts_retained": 6, "retention_rate": 0.09836065573770492, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.0, "amount": 0.3076923076923077, "status": 0.0, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 624}, "ledger_by_hop": {"1": {"facts_total": 61, "facts_retained": 7, "retention_rate": 0.11475409836065574, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.3076923076923077, "status": 0.0, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 690}, "2": {"facts_total": 61, "facts_retained": 6, "retention_rate": 0.09836065573770492, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.0, "amount": 0.3076923076923077, "status": 0.0, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 624}}}, "reference_key_id": "keys/synthetic/syn_03015_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "ac61ed4025051e85e49f3f352d3d60c733554cb217faf29f390436fb9572a503", "output_hash": "f42f05ebfc96f7258cbfbbb37cd65e114d3d15f869e68e6c95b71fc288c30a32"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "78ce13a37656f6941cf7a18b", "attempt_id": "9912bb2c5c28", "run_group_id": "E4_core:A5|repairable|qwen3b:h2", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:57:51.637831Z", "task_family": "synthetic", "document_id": "syn_03015_2000", "document_hash": "71a7284b9bc3f681d69d91d18b9e8d45a4dda486cd14f4145395786ac6a68a89", "question_id": "q04_multi_hop", "split": "test", "cluster_id": "syn_03015_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 2, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1922, "query_tokens": 385, "message_tokens": 294, "message_utf8_bytes": 678, "cache_tensor_bytes": null, "serialized_payload_bytes": 678, "side_information_bytes": 0, "fetched_bytes": 1207, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 36158.07445789687, "hops": {"1": 22991.516749956645, "2": 13166.557707940228}, "receiver_prefill_ms": 238.4869169909507, "ttft_ms": 51.41200008802116, "decode_ms": 497.35512491315603}, "total_wall_ms": 36946.04879093822, "ttft_ms": 51.41200008802116, "input_usage": {"lookups": 2}, "output_usage": {"output_tokens": 8}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "8421725 EUR", "evidence_artifact": {"ledger": {"facts_total": 61, "facts_retained": 6, "retention_rate": 0.09836065573770492, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.0, "amount": 0.3076923076923077, "status": 0.0, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 624}, "ledger_by_hop": {"1": {"facts_total": 61, "facts_retained": 7, "retention_rate": 0.11475409836065574, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.3076923076923077, "status": 0.0, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 690}, "2": {"facts_total": 61, "facts_retained": 6, "retention_rate": 0.09836065573770492, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.0, "amount": 0.3076923076923077, "status": 0.0, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 624}}}, "reference_key_id": "keys/synthetic/syn_03015_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 0.6666666666666666}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "ac61ed4025051e85e49f3f352d3d60c733554cb217faf29f390436fb9572a503", "output_hash": "976c8be2c1a175a4fb5643c5785f93e27ad9205cebd9569a64b821e1ae51ac0e"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "fb1bfb278ee39d56afb55f1c", "attempt_id": "16e29628c665", "run_group_id": "E4_core:A5|repairable|qwen3b:h2", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:57:52.444368Z", "task_family": "synthetic", "document_id": "syn_03015_2000", "document_hash": "71a7284b9bc3f681d69d91d18b9e8d45a4dda486cd14f4145395786ac6a68a89", "question_id": "q05_multi_hop", "split": "test", "cluster_id": "syn_03015_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 2, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1922, "query_tokens": 385, "message_tokens": 294, "message_utf8_bytes": 678, "cache_tensor_bytes": null, "serialized_payload_bytes": 678, "side_information_bytes": 0, "fetched_bytes": 1207, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 36158.07445789687, "hops": {"1": 22991.516749956645, "2": 13166.557707940228}, "receiver_prefill_ms": 288.3245829725638, "ttft_ms": 43.17470802925527, "decode_ms": 26.582959108054638}, "total_wall_ms": 36516.63091592491, "ttft_ms": 43.17470802925527, "input_usage": {"lookups": 2}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 61, "facts_retained": 6, "retention_rate": 0.09836065573770492, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.0, "amount": 0.3076923076923077, "status": 0.0, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 624}, "ledger_by_hop": {"1": {"facts_total": 61, "facts_retained": 7, "retention_rate": 0.11475409836065574, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.3076923076923077, "status": 0.0, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 690}, "2": {"facts_total": 61, "facts_retained": 6, "retention_rate": 0.09836065573770492, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.0, "amount": 0.3076923076923077, "status": 0.0, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 624}}}, "reference_key_id": "keys/synthetic/syn_03015_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "ac61ed4025051e85e49f3f352d3d60c733554cb217faf29f390436fb9572a503", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "ca8ba0ad20ccb31726c81aa4", "attempt_id": "0890ab97455a", "run_group_id": "E4_core:A5|repairable|qwen3b:h2", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:57:52.820135Z", "task_family": "synthetic", "document_id": "syn_03015_2000", "document_hash": "71a7284b9bc3f681d69d91d18b9e8d45a4dda486cd14f4145395786ac6a68a89", "question_id": "q06_negation", "split": "test", "cluster_id": "syn_03015_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 2, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1922, "query_tokens": 377, "message_tokens": 294, "message_utf8_bytes": 678, "cache_tensor_bytes": null, "serialized_payload_bytes": 678, "side_information_bytes": 0, "fetched_bytes": 1207, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 36158.07445789687, "hops": {"1": 22991.516749956645, "2": 13166.557707940228}, "receiver_prefill_ms": 278.3542500110343, "ttft_ms": 25.64862498547882, "decode_ms": 36.75895801279694}, "total_wall_ms": 36499.29274991155, "ttft_ms": 25.64862498547882, "input_usage": {"lookups": 2}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 61, "facts_retained": 6, "retention_rate": 0.09836065573770492, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.0, "amount": 0.3076923076923077, "status": 0.0, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 624}, "ledger_by_hop": {"1": {"facts_total": 61, "facts_retained": 7, "retention_rate": 0.11475409836065574, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.3076923076923077, "status": 0.0, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 690}, "2": {"facts_total": 61, "facts_retained": 6, "retention_rate": 0.09836065573770492, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.0, "amount": 0.3076923076923077, "status": 0.0, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 624}}}, "reference_key_id": "keys/synthetic/syn_03015_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "ac61ed4025051e85e49f3f352d3d60c733554cb217faf29f390436fb9572a503", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "4452bd5af4cf35914d4b83fd", "attempt_id": "cc68c3b69656", "run_group_id": "E4_core:A5|repairable|qwen3b:h2", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:57:53.178539Z", "task_family": "synthetic", "document_id": "syn_03015_2000", "document_hash": "71a7284b9bc3f681d69d91d18b9e8d45a4dda486cd14f4145395786ac6a68a89", "question_id": "q07_negation", "split": "test", "cluster_id": "syn_03015_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 2, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1922, "query_tokens": 377, "message_tokens": 294, "message_utf8_bytes": 678, "cache_tensor_bytes": null, "serialized_payload_bytes": 678, "side_information_bytes": 0, "fetched_bytes": 1207, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 36158.07445789687, "hops": {"1": 22991.516749956645, "2": 13166.557707940228}, "receiver_prefill_ms": 260.87170897517353, "ttft_ms": 77.06383301410824, "decode_ms": 49.03504205867648}, "total_wall_ms": 36545.61995784752, "ttft_ms": 77.06383301410824, "input_usage": {"lookups": 2}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 61, "facts_retained": 6, "retention_rate": 0.09836065573770492, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.0, "amount": 0.3076923076923077, "status": 0.0, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 624}, "ledger_by_hop": {"1": {"facts_total": 61, "facts_retained": 7, "retention_rate": 0.11475409836065574, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.3076923076923077, "status": 0.0, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 690}, "2": {"facts_total": 61, "facts_retained": 6, "retention_rate": 0.09836065573770492, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.0, "amount": 0.3076923076923077, "status": 0.0, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 624}}}, "reference_key_id": "keys/synthetic/syn_03015_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "ac61ed4025051e85e49f3f352d3d60c733554cb217faf29f390436fb9572a503", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "e36f2a6bdc70f54d645324fe", "attempt_id": "947948105335", "run_group_id": "E4_core:A5|repairable|qwen3b:h2", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:57:53.583208Z", "task_family": "synthetic", "document_id": "syn_03015_2000", "document_hash": "71a7284b9bc3f681d69d91d18b9e8d45a4dda486cd14f4145395786ac6a68a89", "question_id": "q08_revision", "split": "test", "cluster_id": "syn_03015_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 2, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1922, "query_tokens": 391, "message_tokens": 294, "message_utf8_bytes": 678, "cache_tensor_bytes": null, "serialized_payload_bytes": 678, "side_information_bytes": 0, "fetched_bytes": 1207, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 36158.07445789687, "hops": {"1": 22991.516749956645, "2": 13166.557707940228}, "receiver_prefill_ms": 425.78587506432086, "ttft_ms": 33.79704197868705, "decode_ms": 38.872083998285234}, "total_wall_ms": 36657.13970793877, "ttft_ms": 33.79704197868705, "input_usage": {"lookups": 2}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 61, "facts_retained": 6, "retention_rate": 0.09836065573770492, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.0, "amount": 0.3076923076923077, "status": 0.0, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 624}, "ledger_by_hop": {"1": {"facts_total": 61, "facts_retained": 7, "retention_rate": 0.11475409836065574, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.3076923076923077, "status": 0.0, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 690}, "2": {"facts_total": 61, "facts_retained": 6, "retention_rate": 0.09836065573770492, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.0, "amount": 0.3076923076923077, "status": 0.0, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 624}}}, "reference_key_id": "keys/synthetic/syn_03015_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "ac61ed4025051e85e49f3f352d3d60c733554cb217faf29f390436fb9572a503", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "81243ed35b54d01d6f61819b", "attempt_id": "28642ce90525", "run_group_id": "E4_core:A5|repairable|qwen3b:h2", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:57:54.099473Z", "task_family": "synthetic", "document_id": "syn_03015_2000", "document_hash": "71a7284b9bc3f681d69d91d18b9e8d45a4dda486cd14f4145395786ac6a68a89", "question_id": "q09_revision", "split": "test", "cluster_id": "syn_03015_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 2, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1922, "query_tokens": 385, "message_tokens": 294, "message_utf8_bytes": 678, "cache_tensor_bytes": null, "serialized_payload_bytes": 678, "side_information_bytes": 0, "fetched_bytes": 1207, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 36158.07445789687, "hops": {"1": 22991.516749956645, "2": 13166.557707940228}, "receiver_prefill_ms": 285.623750067316, "ttft_ms": 28.410749975591898, "decode_ms": 233.0956660443917}, "total_wall_ms": 36705.70037479047, "ttft_ms": 28.410749975591898, "input_usage": {"lookups": 2}, "output_usage": {"output_tokens": 7}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "7028925", "evidence_artifact": {"ledger": {"facts_total": 61, "facts_retained": 6, "retention_rate": 0.09836065573770492, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.0, "amount": 0.3076923076923077, "status": 0.0, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 624}, "ledger_by_hop": {"1": {"facts_total": 61, "facts_retained": 7, "retention_rate": 0.11475409836065574, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.3076923076923077, "status": 0.0, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 690}, "2": {"facts_total": 61, "facts_retained": 6, "retention_rate": 0.09836065573770492, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.0, "amount": 0.3076923076923077, "status": 0.0, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 624}}}, "reference_key_id": "keys/synthetic/syn_03015_2000.json", "task_scores": {"correct": 0.0, "abstained": 0.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "ac61ed4025051e85e49f3f352d3d60c733554cb217faf29f390436fb9572a503", "output_hash": "e966a398855c104a86bce07742c3ce8108a432926bc704940ecfd07e44596774"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "4e6a3d8cf7f11b3de7fdf80a", "attempt_id": "30c9f20632ad", "run_group_id": "E4_core:A5|repairable|qwen3b:h2", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:57:54.665066Z", "task_family": "synthetic", "document_id": "syn_03015_2000", "document_hash": "71a7284b9bc3f681d69d91d18b9e8d45a4dda486cd14f4145395786ac6a68a89", "question_id": "q10_temporal", "split": "test", "cluster_id": "syn_03015_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 2, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1922, "query_tokens": 394, "message_tokens": 294, "message_utf8_bytes": 678, "cache_tensor_bytes": null, "serialized_payload_bytes": 678, "side_information_bytes": 0, "fetched_bytes": 1207, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 36158.07445789687, "hops": {"1": 22991.516749956645, "2": 13166.557707940228}, "receiver_prefill_ms": 267.29574997443706, "ttft_ms": 44.55937503371388, "decode_ms": 53.38116700295359}, "total_wall_ms": 36523.852790822275, "ttft_ms": 44.55937503371388, "input_usage": {"lookups": 2}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 61, "facts_retained": 6, "retention_rate": 0.09836065573770492, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.0, "amount": 0.3076923076923077, "status": 0.0, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 624}, "ledger_by_hop": {"1": {"facts_total": 61, "facts_retained": 7, "retention_rate": 0.11475409836065574, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.3076923076923077, "status": 0.0, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 690}, "2": {"facts_total": 61, "facts_retained": 6, "retention_rate": 0.09836065573770492, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.0, "amount": 0.3076923076923077, "status": 0.0, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 624}}}, "reference_key_id": "keys/synthetic/syn_03015_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "ac61ed4025051e85e49f3f352d3d60c733554cb217faf29f390436fb9572a503", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "1b9217faae2f6d0b6ea04d76", "attempt_id": "5fe5c67101db", "run_group_id": "E4_core:A5|repairable|qwen3b:h2", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:57:55.048142Z", "task_family": "synthetic", "document_id": "syn_03015_2000", "document_hash": "71a7284b9bc3f681d69d91d18b9e8d45a4dda486cd14f4145395786ac6a68a89", "question_id": "q11_aggregation", "split": "test", "cluster_id": "syn_03015_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 2, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1922, "query_tokens": 387, "message_tokens": 294, "message_utf8_bytes": 678, "cache_tensor_bytes": null, "serialized_payload_bytes": 678, "side_information_bytes": 0, "fetched_bytes": 1207, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 36158.07445789687, "hops": {"1": 22991.516749956645, "2": 13166.557707940228}, "receiver_prefill_ms": 454.32337501551956, "ttft_ms": 44.9943330604583, "decode_ms": 40.74124991893768}, "total_wall_ms": 36698.70208285283, "ttft_ms": 44.9943330604583, "input_usage": {"lookups": 2}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 61, "facts_retained": 6, "retention_rate": 0.09836065573770492, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.0, "amount": 0.3076923076923077, "status": 0.0, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 624}, "ledger_by_hop": {"1": {"facts_total": 61, "facts_retained": 7, "retention_rate": 0.11475409836065574, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.3076923076923077, "status": 0.0, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 690}, "2": {"facts_total": 61, "facts_retained": 6, "retention_rate": 0.09836065573770492, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.0, "amount": 0.3076923076923077, "status": 0.0, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 624}}}, "reference_key_id": "keys/synthetic/syn_03015_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "ac61ed4025051e85e49f3f352d3d60c733554cb217faf29f390436fb9572a503", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "af5c1f69fdc07b6f4af2a1f0", "attempt_id": "8726e2fb8e80", "run_group_id": "E4_core:A5|repairable|qwen3b:h2", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:57:55.607054Z", "task_family": "synthetic", "document_id": "syn_03015_2000", "document_hash": "71a7284b9bc3f681d69d91d18b9e8d45a4dda486cd14f4145395786ac6a68a89", "question_id": "q12_unit_conversion", "split": "test", "cluster_id": "syn_03015_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 2, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1922, "query_tokens": 384, "message_tokens": 294, "message_utf8_bytes": 678, "cache_tensor_bytes": null, "serialized_payload_bytes": 678, "side_information_bytes": 0, "fetched_bytes": 1207, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 36158.07445789687, "hops": {"1": 22991.516749956645, "2": 13166.557707940228}, "receiver_prefill_ms": 280.45020799618214, "ttft_ms": 38.624417036771774, "decode_ms": 37.15949994511902}, "total_wall_ms": 36514.70774994232, "ttft_ms": 38.624417036771774, "input_usage": {"lookups": 2}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 61, "facts_retained": 6, "retention_rate": 0.09836065573770492, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.0, "amount": 0.3076923076923077, "status": 0.0, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 624}, "ledger_by_hop": {"1": {"facts_total": 61, "facts_retained": 7, "retention_rate": 0.11475409836065574, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.3076923076923077, "status": 0.0, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 690}, "2": {"facts_total": 61, "facts_retained": 6, "retention_rate": 0.09836065573770492, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.0, "amount": 0.3076923076923077, "status": 0.0, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 624}}}, "reference_key_id": "keys/synthetic/syn_03015_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "ac61ed4025051e85e49f3f352d3d60c733554cb217faf29f390436fb9572a503", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "2860ca146955cef4853b5022", "attempt_id": "a180ec912fe1", "run_group_id": "E4_core:A5|repairable|qwen3b:h2", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:57:55.980650Z", "task_family": "synthetic", "document_id": "syn_03015_2000", "document_hash": "71a7284b9bc3f681d69d91d18b9e8d45a4dda486cd14f4145395786ac6a68a89", "question_id": "q13_unanswerable", "split": "test", "cluster_id": "syn_03015_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 2, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1922, "query_tokens": 388, "message_tokens": 294, "message_utf8_bytes": 678, "cache_tensor_bytes": null, "serialized_payload_bytes": 678, "side_information_bytes": 0, "fetched_bytes": 1207, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 36158.07445789687, "hops": {"1": 22991.516749956645, "2": 13166.557707940228}, "receiver_prefill_ms": 283.4072079276666, "ttft_ms": 36.00983403157443, "decode_ms": 40.74375005438924}, "total_wall_ms": 36518.7472908292, "ttft_ms": 36.00983403157443, "input_usage": {"lookups": 2}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 61, "facts_retained": 6, "retention_rate": 0.09836065573770492, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.0, "amount": 0.3076923076923077, "status": 0.0, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 624}, "ledger_by_hop": {"1": {"facts_total": 61, "facts_retained": 7, "retention_rate": 0.11475409836065574, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.3076923076923077, "status": 0.0, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 690}, "2": {"facts_total": 61, "facts_retained": 6, "retention_rate": 0.09836065573770492, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.0, "amount": 0.3076923076923077, "status": 0.0, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 624}}}, "reference_key_id": "keys/synthetic/syn_03015_2000.json", "task_scores": {"correct": 1.0, "abstained": 1.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "ac61ed4025051e85e49f3f352d3d60c733554cb217faf29f390436fb9572a503", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "22328eabe57ec7a49d15fcd7", "attempt_id": "1ce5921899b1", "run_group_id": "E4_core:A5|repairable|qwen3b:h2", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:57:56.358167Z", "task_family": "synthetic", "document_id": "syn_03015_2000", "document_hash": "71a7284b9bc3f681d69d91d18b9e8d45a4dda486cd14f4145395786ac6a68a89", "question_id": "q14_unanswerable", "split": "test", "cluster_id": "syn_03015_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 2, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1922, "query_tokens": 389, "message_tokens": 294, "message_utf8_bytes": 678, "cache_tensor_bytes": null, "serialized_payload_bytes": 678, "side_information_bytes": 0, "fetched_bytes": 1207, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 36158.07445789687, "hops": {"1": 22991.516749956645, "2": 13166.557707940228}, "receiver_prefill_ms": 301.50412500370294, "ttft_ms": 42.25445899646729, "decode_ms": 57.67283297609538}, "total_wall_ms": 36560.32908288762, "ttft_ms": 42.25445899646729, "input_usage": {"lookups": 2}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 61, "facts_retained": 6, "retention_rate": 0.09836065573770492, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.0, "amount": 0.3076923076923077, "status": 0.0, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 624}, "ledger_by_hop": {"1": {"facts_total": 61, "facts_retained": 7, "retention_rate": 0.11475409836065574, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.3076923076923077, "status": 0.0, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 690}, "2": {"facts_total": 61, "facts_retained": 6, "retention_rate": 0.09836065573770492, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.0, "amount": 0.3076923076923077, "status": 0.0, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 624}}}, "reference_key_id": "keys/synthetic/syn_03015_2000.json", "task_scores": {"correct": 1.0, "abstained": 1.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "ac61ed4025051e85e49f3f352d3d60c733554cb217faf29f390436fb9572a503", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "86959aa2a6a7a478079829f9", "attempt_id": "295ee4b6c7ff", "run_group_id": "E4_core:A5|repairable|qwen3b:h4", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:58:22.655851Z", "task_family": "synthetic", "document_id": "syn_03015_2000", "document_hash": "71a7284b9bc3f681d69d91d18b9e8d45a4dda486cd14f4145395786ac6a68a89", "question_id": "q00_single_fact", "split": "test", "cluster_id": "syn_03015_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 4, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1922, "query_tokens": 376, "message_tokens": 294, "message_utf8_bytes": 678, "cache_tensor_bytes": null, "serialized_payload_bytes": 678, "side_information_bytes": 0, "fetched_bytes": 3621, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 62033.760000020266, "hops": {"1": 22991.516749956645, "2": 13166.557707940228, "3": 12937.996500055306, "4": 12937.689042068087}, "receiver_prefill_ms": 277.6253749616444, "ttft_ms": 65.42441702913493, "decode_ms": 55.44837494380772}, "total_wall_ms": 62433.01454093307, "ttft_ms": 65.42441702913493, "input_usage": {"lookups": 6}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 61, "facts_retained": 6, "retention_rate": 0.09836065573770492, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.0, "amount": 0.3076923076923077, "status": 0.0, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 624}, "ledger_by_hop": {"1": {"facts_total": 61, "facts_retained": 7, "retention_rate": 0.11475409836065574, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.3076923076923077, "status": 0.0, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 690}, "2": {"facts_total": 61, "facts_retained": 6, "retention_rate": 0.09836065573770492, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.0, "amount": 0.3076923076923077, "status": 0.0, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 624}, "3": {"facts_total": 61, "facts_retained": 6, "retention_rate": 0.09836065573770492, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.0, "amount": 0.3076923076923077, "status": 0.0, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 624}, "4": {"facts_total": 61, "facts_retained": 6, "retention_rate": 0.09836065573770492, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.0, "amount": 0.3076923076923077, "status": 0.0, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 624}}}, "reference_key_id": "keys/synthetic/syn_03015_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "ac61ed4025051e85e49f3f352d3d60c733554cb217faf29f390436fb9572a503", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "0c0111f337c14c7350338a8c", "attempt_id": "dc374fa748c4", "run_group_id": "E4_core:A5|repairable|qwen3b:h4", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:58:23.074700Z", "task_family": "synthetic", "document_id": "syn_03015_2000", "document_hash": "71a7284b9bc3f681d69d91d18b9e8d45a4dda486cd14f4145395786ac6a68a89", "question_id": "q01_single_fact", "split": "test", "cluster_id": "syn_03015_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 4, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1922, "query_tokens": 381, "message_tokens": 294, "message_utf8_bytes": 678, "cache_tensor_bytes": null, "serialized_payload_bytes": 678, "side_information_bytes": 0, "fetched_bytes": 3621, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 62033.760000020266, "hops": {"1": 22991.516749956645, "2": 13166.557707940228, "3": 12937.996500055306, "4": 12937.689042068087}, "receiver_prefill_ms": 412.9709169501439, "ttft_ms": 42.76529198978096, "decode_ms": 25.061958003789186}, "total_wall_ms": 62515.18095808569, "ttft_ms": 42.76529198978096, "input_usage": {"lookups": 6}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 61, "facts_retained": 6, "retention_rate": 0.09836065573770492, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.0, "amount": 0.3076923076923077, "status": 0.0, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 624}, "ledger_by_hop": {"1": {"facts_total": 61, "facts_retained": 7, "retention_rate": 0.11475409836065574, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.3076923076923077, "status": 0.0, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 690}, "2": {"facts_total": 61, "facts_retained": 6, "retention_rate": 0.09836065573770492, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.0, "amount": 0.3076923076923077, "status": 0.0, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 624}, "3": {"facts_total": 61, "facts_retained": 6, "retention_rate": 0.09836065573770492, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.0, "amount": 0.3076923076923077, "status": 0.0, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 624}, "4": {"facts_total": 61, "facts_retained": 6, "retention_rate": 0.09836065573770492, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.0, "amount": 0.3076923076923077, "status": 0.0, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 624}}}, "reference_key_id": "keys/synthetic/syn_03015_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "ac61ed4025051e85e49f3f352d3d60c733554cb217faf29f390436fb9572a503", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "73a6bf0e9193152a0e10f8e1", "attempt_id": "caa85d2e9230", "run_group_id": "E4_core:A5|repairable|qwen3b:h4", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:58:23.573578Z", "task_family": "synthetic", "document_id": "syn_03015_2000", "document_hash": "71a7284b9bc3f681d69d91d18b9e8d45a4dda486cd14f4145395786ac6a68a89", "question_id": "q02_exact_copy", "split": "test", "cluster_id": "syn_03015_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 4, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1922, "query_tokens": 380, "message_tokens": 294, "message_utf8_bytes": 678, "cache_tensor_bytes": null, "serialized_payload_bytes": 678, "side_information_bytes": 0, "fetched_bytes": 3621, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 62033.760000020266, "hops": {"1": 22991.516749956645, "2": 13166.557707940228, "3": 12937.996500055306, "4": 12937.689042068087}, "receiver_prefill_ms": 278.1856249785051, "ttft_ms": 43.661708012223244, "decode_ms": 23.960208985954523}, "total_wall_ms": 62380.041250027716, "ttft_ms": 43.661708012223244, "input_usage": {"lookups": 6}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 61, "facts_retained": 6, "retention_rate": 0.09836065573770492, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.0, "amount": 0.3076923076923077, "status": 0.0, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 624}, "ledger_by_hop": {"1": {"facts_total": 61, "facts_retained": 7, "retention_rate": 0.11475409836065574, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.3076923076923077, "status": 0.0, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 690}, "2": {"facts_total": 61, "facts_retained": 6, "retention_rate": 0.09836065573770492, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.0, "amount": 0.3076923076923077, "status": 0.0, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 624}, "3": {"facts_total": 61, "facts_retained": 6, "retention_rate": 0.09836065573770492, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.0, "amount": 0.3076923076923077, "status": 0.0, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 624}, "4": {"facts_total": 61, "facts_retained": 6, "retention_rate": 0.09836065573770492, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.0, "amount": 0.3076923076923077, "status": 0.0, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 624}}}, "reference_key_id": "keys/synthetic/syn_03015_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "ac61ed4025051e85e49f3f352d3d60c733554cb217faf29f390436fb9572a503", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "05e8c3ccb91122e28ae0b67d", "attempt_id": "e13422769542", "run_group_id": "E4_core:A5|repairable|qwen3b:h4", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:58:23.936918Z", "task_family": "synthetic", "document_id": "syn_03015_2000", "document_hash": "71a7284b9bc3f681d69d91d18b9e8d45a4dda486cd14f4145395786ac6a68a89", "question_id": "q03_exact_copy", "split": "test", "cluster_id": "syn_03015_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 4, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1922, "query_tokens": 376, "message_tokens": 294, "message_utf8_bytes": 678, "cache_tensor_bytes": null, "serialized_payload_bytes": 678, "side_information_bytes": 0, "fetched_bytes": 3621, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 62033.760000020266, "hops": {"1": 22991.516749956645, "2": 13166.557707940228, "3": 12937.996500055306, "4": 12937.689042068087}, "receiver_prefill_ms": 280.98120901267976, "ttft_ms": 20.26991604361683, "decode_ms": 187.47212493326515}, "total_wall_ms": 62523.28141697217, "ttft_ms": 20.26991604361683, "input_usage": {"lookups": 6}, "output_usage": {"output_tokens": 6}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "Wilhelmina Dubrovnik", "evidence_artifact": {"ledger": {"facts_total": 61, "facts_retained": 6, "retention_rate": 0.09836065573770492, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.0, "amount": 0.3076923076923077, "status": 0.0, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 624}, "ledger_by_hop": {"1": {"facts_total": 61, "facts_retained": 7, "retention_rate": 0.11475409836065574, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.3076923076923077, "status": 0.0, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 690}, "2": {"facts_total": 61, "facts_retained": 6, "retention_rate": 0.09836065573770492, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.0, "amount": 0.3076923076923077, "status": 0.0, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 624}, "3": {"facts_total": 61, "facts_retained": 6, "retention_rate": 0.09836065573770492, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.0, "amount": 0.3076923076923077, "status": 0.0, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 624}, "4": {"facts_total": 61, "facts_retained": 6, "retention_rate": 0.09836065573770492, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.0, "amount": 0.3076923076923077, "status": 0.0, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 624}}}, "reference_key_id": "keys/synthetic/syn_03015_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "ac61ed4025051e85e49f3f352d3d60c733554cb217faf29f390436fb9572a503", "output_hash": "f42f05ebfc96f7258cbfbbb37cd65e114d3d15f869e68e6c95b71fc288c30a32"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "102e6b3b032233e60c1a0888", "attempt_id": "06a708a352d0", "run_group_id": "E4_core:A5|repairable|qwen3b:h4", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:58:24.443278Z", "task_family": "synthetic", "document_id": "syn_03015_2000", "document_hash": "71a7284b9bc3f681d69d91d18b9e8d45a4dda486cd14f4145395786ac6a68a89", "question_id": "q04_multi_hop", "split": "test", "cluster_id": "syn_03015_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 4, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1922, "query_tokens": 385, "message_tokens": 294, "message_utf8_bytes": 678, "cache_tensor_bytes": null, "serialized_payload_bytes": 678, "side_information_bytes": 0, "fetched_bytes": 3621, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 62033.760000020266, "hops": {"1": 22991.516749956645, "2": 13166.557707940228, "3": 12937.996500055306, "4": 12937.689042068087}, "receiver_prefill_ms": 259.0531249297783, "ttft_ms": 96.38687502592802, "decode_ms": 411.40695905778557}, "total_wall_ms": 62801.15950002801, "ttft_ms": 96.38687502592802, "input_usage": {"lookups": 6}, "output_usage": {"output_tokens": 8}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "8421725 EUR", "evidence_artifact": {"ledger": {"facts_total": 61, "facts_retained": 6, "retention_rate": 0.09836065573770492, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.0, "amount": 0.3076923076923077, "status": 0.0, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 624}, "ledger_by_hop": {"1": {"facts_total": 61, "facts_retained": 7, "retention_rate": 0.11475409836065574, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.3076923076923077, "status": 0.0, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 690}, "2": {"facts_total": 61, "facts_retained": 6, "retention_rate": 0.09836065573770492, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.0, "amount": 0.3076923076923077, "status": 0.0, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 624}, "3": {"facts_total": 61, "facts_retained": 6, "retention_rate": 0.09836065573770492, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.0, "amount": 0.3076923076923077, "status": 0.0, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 624}, "4": {"facts_total": 61, "facts_retained": 6, "retention_rate": 0.09836065573770492, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.0, "amount": 0.3076923076923077, "status": 0.0, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 624}}}, "reference_key_id": "keys/synthetic/syn_03015_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 0.6666666666666666}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "ac61ed4025051e85e49f3f352d3d60c733554cb217faf29f390436fb9572a503", "output_hash": "976c8be2c1a175a4fb5643c5785f93e27ad9205cebd9569a64b821e1ae51ac0e"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "6696002eb7d362329d78b0c4", "attempt_id": "69488db0d084", "run_group_id": "E4_core:A5|repairable|qwen3b:h4", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:58:25.227785Z", "task_family": "synthetic", "document_id": "syn_03015_2000", "document_hash": "71a7284b9bc3f681d69d91d18b9e8d45a4dda486cd14f4145395786ac6a68a89", "question_id": "q05_multi_hop", "split": "test", "cluster_id": "syn_03015_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 4, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1922, "query_tokens": 385, "message_tokens": 294, "message_utf8_bytes": 678, "cache_tensor_bytes": null, "serialized_payload_bytes": 678, "side_information_bytes": 0, "fetched_bytes": 3621, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 62033.760000020266, "hops": {"1": 22991.516749956645, "2": 13166.557707940228, "3": 12937.996500055306, "4": 12937.689042068087}, "receiver_prefill_ms": 262.5361249083653, "ttft_ms": 38.390708970837295, "decode_ms": 24.717999971471727}, "total_wall_ms": 62359.983291011304, "ttft_ms": 38.390708970837295, "input_usage": {"lookups": 6}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 61, "facts_retained": 6, "retention_rate": 0.09836065573770492, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.0, "amount": 0.3076923076923077, "status": 0.0, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 624}, "ledger_by_hop": {"1": {"facts_total": 61, "facts_retained": 7, "retention_rate": 0.11475409836065574, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.3076923076923077, "status": 0.0, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 690}, "2": {"facts_total": 61, "facts_retained": 6, "retention_rate": 0.09836065573770492, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.0, "amount": 0.3076923076923077, "status": 0.0, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 624}, "3": {"facts_total": 61, "facts_retained": 6, "retention_rate": 0.09836065573770492, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.0, "amount": 0.3076923076923077, "status": 0.0, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 624}, "4": {"facts_total": 61, "facts_retained": 6, "retention_rate": 0.09836065573770492, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.0, "amount": 0.3076923076923077, "status": 0.0, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 624}}}, "reference_key_id": "keys/synthetic/syn_03015_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "ac61ed4025051e85e49f3f352d3d60c733554cb217faf29f390436fb9572a503", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "f1e33eae3d918fdf93752246", "attempt_id": "c2b6be04a794", "run_group_id": "E4_core:A5|repairable|qwen3b:h4", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:58:25.572154Z", "task_family": "synthetic", "document_id": "syn_03015_2000", "document_hash": "71a7284b9bc3f681d69d91d18b9e8d45a4dda486cd14f4145395786ac6a68a89", "question_id": "q06_negation", "split": "test", "cluster_id": "syn_03015_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 4, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1922, "query_tokens": 377, "message_tokens": 294, "message_utf8_bytes": 678, "cache_tensor_bytes": null, "serialized_payload_bytes": 678, "side_information_bytes": 0, "fetched_bytes": 3621, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 62033.760000020266, "hops": {"1": 22991.516749956645, "2": 13166.557707940228, "3": 12937.996500055306, "4": 12937.689042068087}, "receiver_prefill_ms": 248.0417910264805, "ttft_ms": 27.752958005294204, "decode_ms": 74.24262503627688}, "total_wall_ms": 62384.34283295646, "ttft_ms": 27.752958005294204, "input_usage": {"lookups": 6}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 61, "facts_retained": 6, "retention_rate": 0.09836065573770492, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.0, "amount": 0.3076923076923077, "status": 0.0, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 624}, "ledger_by_hop": {"1": {"facts_total": 61, "facts_retained": 7, "retention_rate": 0.11475409836065574, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.3076923076923077, "status": 0.0, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 690}, "2": {"facts_total": 61, "facts_retained": 6, "retention_rate": 0.09836065573770492, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.0, "amount": 0.3076923076923077, "status": 0.0, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 624}, "3": {"facts_total": 61, "facts_retained": 6, "retention_rate": 0.09836065573770492, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.0, "amount": 0.3076923076923077, "status": 0.0, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 624}, "4": {"facts_total": 61, "facts_retained": 6, "retention_rate": 0.09836065573770492, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.0, "amount": 0.3076923076923077, "status": 0.0, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 624}}}, "reference_key_id": "keys/synthetic/syn_03015_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "ac61ed4025051e85e49f3f352d3d60c733554cb217faf29f390436fb9572a503", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "d0f3b13ea05e362f31de018a", "attempt_id": "5e32d6ecb1e9", "run_group_id": "E4_core:A5|repairable|qwen3b:h4", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:58:25.939533Z", "task_family": "synthetic", "document_id": "syn_03015_2000", "document_hash": "71a7284b9bc3f681d69d91d18b9e8d45a4dda486cd14f4145395786ac6a68a89", "question_id": "q07_negation", "split": "test", "cluster_id": "syn_03015_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 4, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1922, "query_tokens": 377, "message_tokens": 294, "message_utf8_bytes": 678, "cache_tensor_bytes": null, "serialized_payload_bytes": 678, "side_information_bytes": 0, "fetched_bytes": 3621, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 62033.760000020266, "hops": {"1": 22991.516749956645, "2": 13166.557707940228, "3": 12937.996500055306, "4": 12937.689042068087}, "receiver_prefill_ms": 279.8910830169916, "ttft_ms": 63.04645899217576, "decode_ms": 72.11520907003433}, "total_wall_ms": 62449.61999997031, "ttft_ms": 63.04645899217576, "input_usage": {"lookups": 6}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 61, "facts_retained": 6, "retention_rate": 0.09836065573770492, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.0, "amount": 0.3076923076923077, "status": 0.0, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 624}, "ledger_by_hop": {"1": {"facts_total": 61, "facts_retained": 7, "retention_rate": 0.11475409836065574, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.3076923076923077, "status": 0.0, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 690}, "2": {"facts_total": 61, "facts_retained": 6, "retention_rate": 0.09836065573770492, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.0, "amount": 0.3076923076923077, "status": 0.0, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 624}, "3": {"facts_total": 61, "facts_retained": 6, "retention_rate": 0.09836065573770492, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.0, "amount": 0.3076923076923077, "status": 0.0, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 624}, "4": {"facts_total": 61, "facts_retained": 6, "retention_rate": 0.09836065573770492, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.0, "amount": 0.3076923076923077, "status": 0.0, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 624}}}, "reference_key_id": "keys/synthetic/syn_03015_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "ac61ed4025051e85e49f3f352d3d60c733554cb217faf29f390436fb9572a503", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "b460c265b011701564e2d166", "attempt_id": "99bdc35d070f", "run_group_id": "E4_core:A5|repairable|qwen3b:h4", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:58:26.374602Z", "task_family": "synthetic", "document_id": "syn_03015_2000", "document_hash": "71a7284b9bc3f681d69d91d18b9e8d45a4dda486cd14f4145395786ac6a68a89", "question_id": "q08_revision", "split": "test", "cluster_id": "syn_03015_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 4, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1922, "query_tokens": 391, "message_tokens": 294, "message_utf8_bytes": 678, "cache_tensor_bytes": null, "serialized_payload_bytes": 678, "side_information_bytes": 0, "fetched_bytes": 3621, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 62033.760000020266, "hops": {"1": 22991.516749956645, "2": 13166.557707940228, "3": 12937.996500055306, "4": 12937.689042068087}, "receiver_prefill_ms": 359.89945905748755, "ttft_ms": 42.15516604017466, "decode_ms": 25.71449999231845}, "total_wall_ms": 62462.080875062384, "ttft_ms": 42.15516604017466, "input_usage": {"lookups": 6}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 61, "facts_retained": 6, "retention_rate": 0.09836065573770492, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.0, "amount": 0.3076923076923077, "status": 0.0, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 624}, "ledger_by_hop": {"1": {"facts_total": 61, "facts_retained": 7, "retention_rate": 0.11475409836065574, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.3076923076923077, "status": 0.0, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 690}, "2": {"facts_total": 61, "facts_retained": 6, "retention_rate": 0.09836065573770492, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.0, "amount": 0.3076923076923077, "status": 0.0, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 624}, "3": {"facts_total": 61, "facts_retained": 6, "retention_rate": 0.09836065573770492, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.0, "amount": 0.3076923076923077, "status": 0.0, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 624}, "4": {"facts_total": 61, "facts_retained": 6, "retention_rate": 0.09836065573770492, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.0, "amount": 0.3076923076923077, "status": 0.0, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 624}}}, "reference_key_id": "keys/synthetic/syn_03015_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "ac61ed4025051e85e49f3f352d3d60c733554cb217faf29f390436fb9572a503", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "0e0c95f13ecf74081bb1dad5", "attempt_id": "422e3e14143a", "run_group_id": "E4_core:A5|repairable|qwen3b:h4", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:58:26.819520Z", "task_family": "synthetic", "document_id": "syn_03015_2000", "document_hash": "71a7284b9bc3f681d69d91d18b9e8d45a4dda486cd14f4145395786ac6a68a89", "question_id": "q09_revision", "split": "test", "cluster_id": "syn_03015_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 4, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1922, "query_tokens": 385, "message_tokens": 294, "message_utf8_bytes": 678, "cache_tensor_bytes": null, "serialized_payload_bytes": 678, "side_information_bytes": 0, "fetched_bytes": 3621, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 62033.760000020266, "hops": {"1": 22991.516749956645, "2": 13166.557707940228, "3": 12937.996500055306, "4": 12937.689042068087}, "receiver_prefill_ms": 274.8500000452623, "ttft_ms": 35.61200003605336, "decode_ms": 220.0042090844363}, "total_wall_ms": 62564.81229211204, "ttft_ms": 35.61200003605336, "input_usage": {"lookups": 6}, "output_usage": {"output_tokens": 7}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "7028925", "evidence_artifact": {"ledger": {"facts_total": 61, "facts_retained": 6, "retention_rate": 0.09836065573770492, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.0, "amount": 0.3076923076923077, "status": 0.0, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 624}, "ledger_by_hop": {"1": {"facts_total": 61, "facts_retained": 7, "retention_rate": 0.11475409836065574, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.3076923076923077, "status": 0.0, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 690}, "2": {"facts_total": 61, "facts_retained": 6, "retention_rate": 0.09836065573770492, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.0, "amount": 0.3076923076923077, "status": 0.0, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 624}, "3": {"facts_total": 61, "facts_retained": 6, "retention_rate": 0.09836065573770492, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.0, "amount": 0.3076923076923077, "status": 0.0, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 624}, "4": {"facts_total": 61, "facts_retained": 6, "retention_rate": 0.09836065573770492, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.0, "amount": 0.3076923076923077, "status": 0.0, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 624}}}, "reference_key_id": "keys/synthetic/syn_03015_2000.json", "task_scores": {"correct": 0.0, "abstained": 0.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "ac61ed4025051e85e49f3f352d3d60c733554cb217faf29f390436fb9572a503", "output_hash": "e966a398855c104a86bce07742c3ce8108a432926bc704940ecfd07e44596774"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "7a7b1f1f92d53c1d4be97c8d", "attempt_id": "518587ebe44f", "run_group_id": "E4_core:A5|repairable|qwen3b:h4", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:58:27.368699Z", "task_family": "synthetic", "document_id": "syn_03015_2000", "document_hash": "71a7284b9bc3f681d69d91d18b9e8d45a4dda486cd14f4145395786ac6a68a89", "question_id": "q10_temporal", "split": "test", "cluster_id": "syn_03015_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 4, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1922, "query_tokens": 394, "message_tokens": 294, "message_utf8_bytes": 678, "cache_tensor_bytes": null, "serialized_payload_bytes": 678, "side_information_bytes": 0, "fetched_bytes": 3621, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 62033.760000020266, "hops": {"1": 22991.516749956645, "2": 13166.557707940228, "3": 12937.996500055306, "4": 12937.689042068087}, "receiver_prefill_ms": 267.7496250253171, "ttft_ms": 53.213082952424884, "decode_ms": 42.40712500177324}, "total_wall_ms": 62397.86899997853, "ttft_ms": 53.213082952424884, "input_usage": {"lookups": 6}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 61, "facts_retained": 6, "retention_rate": 0.09836065573770492, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.0, "amount": 0.3076923076923077, "status": 0.0, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 624}, "ledger_by_hop": {"1": {"facts_total": 61, "facts_retained": 7, "retention_rate": 0.11475409836065574, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.3076923076923077, "status": 0.0, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 690}, "2": {"facts_total": 61, "facts_retained": 6, "retention_rate": 0.09836065573770492, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.0, "amount": 0.3076923076923077, "status": 0.0, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 624}, "3": {"facts_total": 61, "facts_retained": 6, "retention_rate": 0.09836065573770492, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.0, "amount": 0.3076923076923077, "status": 0.0, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 624}, "4": {"facts_total": 61, "facts_retained": 6, "retention_rate": 0.09836065573770492, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.0, "amount": 0.3076923076923077, "status": 0.0, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 624}}}, "reference_key_id": "keys/synthetic/syn_03015_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "ac61ed4025051e85e49f3f352d3d60c733554cb217faf29f390436fb9572a503", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "a605123b21de0620c25c74fe", "attempt_id": "e25f4825d422", "run_group_id": "E4_core:A5|repairable|qwen3b:h4", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:58:27.751427Z", "task_family": "synthetic", "document_id": "syn_03015_2000", "document_hash": "71a7284b9bc3f681d69d91d18b9e8d45a4dda486cd14f4145395786ac6a68a89", "question_id": "q11_aggregation", "split": "test", "cluster_id": "syn_03015_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 4, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1922, "query_tokens": 387, "message_tokens": 294, "message_utf8_bytes": 678, "cache_tensor_bytes": null, "serialized_payload_bytes": 678, "side_information_bytes": 0, "fetched_bytes": 3621, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 62033.760000020266, "hops": {"1": 22991.516749956645, "2": 13166.557707940228, "3": 12937.996500055306, "4": 12937.689042068087}, "receiver_prefill_ms": 417.22420894075185, "ttft_ms": 21.998750045895576, "decode_ms": 44.945959001779556}, "total_wall_ms": 62518.55683303438, "ttft_ms": 21.998750045895576, "input_usage": {"lookups": 6}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 61, "facts_retained": 6, "retention_rate": 0.09836065573770492, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.0, "amount": 0.3076923076923077, "status": 0.0, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 624}, "ledger_by_hop": {"1": {"facts_total": 61, "facts_retained": 7, "retention_rate": 0.11475409836065574, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.3076923076923077, "status": 0.0, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 690}, "2": {"facts_total": 61, "facts_retained": 6, "retention_rate": 0.09836065573770492, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.0, "amount": 0.3076923076923077, "status": 0.0, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 624}, "3": {"facts_total": 61, "facts_retained": 6, "retention_rate": 0.09836065573770492, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.0, "amount": 0.3076923076923077, "status": 0.0, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 624}, "4": {"facts_total": 61, "facts_retained": 6, "retention_rate": 0.09836065573770492, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.0, "amount": 0.3076923076923077, "status": 0.0, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 624}}}, "reference_key_id": "keys/synthetic/syn_03015_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "ac61ed4025051e85e49f3f352d3d60c733554cb217faf29f390436fb9572a503", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "3244f58ef49162b7d544f516", "attempt_id": "433484b3f055", "run_group_id": "E4_core:A5|repairable|qwen3b:h4", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:58:28.253683Z", "task_family": "synthetic", "document_id": "syn_03015_2000", "document_hash": "71a7284b9bc3f681d69d91d18b9e8d45a4dda486cd14f4145395786ac6a68a89", "question_id": "q12_unit_conversion", "split": "test", "cluster_id": "syn_03015_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 4, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1922, "query_tokens": 384, "message_tokens": 294, "message_utf8_bytes": 678, "cache_tensor_bytes": null, "serialized_payload_bytes": 678, "side_information_bytes": 0, "fetched_bytes": 3621, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 62033.760000020266, "hops": {"1": 22991.516749956645, "2": 13166.557707940228, "3": 12937.996500055306, "4": 12937.689042068087}, "receiver_prefill_ms": 288.5745420353487, "ttft_ms": 27.554999920539558, "decode_ms": 36.27483290620148}, "total_wall_ms": 62386.601042118855, "ttft_ms": 27.554999920539558, "input_usage": {"lookups": 6}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 61, "facts_retained": 6, "retention_rate": 0.09836065573770492, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.0, "amount": 0.3076923076923077, "status": 0.0, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 624}, "ledger_by_hop": {"1": {"facts_total": 61, "facts_retained": 7, "retention_rate": 0.11475409836065574, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.3076923076923077, "status": 0.0, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 690}, "2": {"facts_total": 61, "facts_retained": 6, "retention_rate": 0.09836065573770492, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.0, "amount": 0.3076923076923077, "status": 0.0, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 624}, "3": {"facts_total": 61, "facts_retained": 6, "retention_rate": 0.09836065573770492, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.0, "amount": 0.3076923076923077, "status": 0.0, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 624}, "4": {"facts_total": 61, "facts_retained": 6, "retention_rate": 0.09836065573770492, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.0, "amount": 0.3076923076923077, "status": 0.0, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 624}}}, "reference_key_id": "keys/synthetic/syn_03015_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "ac61ed4025051e85e49f3f352d3d60c733554cb217faf29f390436fb9572a503", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "1da83c01f39436768bfb0eec", "attempt_id": "9b8c2d088289", "run_group_id": "E4_core:A5|repairable|qwen3b:h4", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:58:28.623529Z", "task_family": "synthetic", "document_id": "syn_03015_2000", "document_hash": "71a7284b9bc3f681d69d91d18b9e8d45a4dda486cd14f4145395786ac6a68a89", "question_id": "q13_unanswerable", "split": "test", "cluster_id": "syn_03015_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 4, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1922, "query_tokens": 388, "message_tokens": 294, "message_utf8_bytes": 678, "cache_tensor_bytes": null, "serialized_payload_bytes": 678, "side_information_bytes": 0, "fetched_bytes": 3621, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 62033.760000020266, "hops": {"1": 22991.516749956645, "2": 13166.557707940228, "3": 12937.996500055306, "4": 12937.689042068087}, "receiver_prefill_ms": 288.7237920658663, "ttft_ms": 30.007958062924445, "decode_ms": 38.452374981716275}, "total_wall_ms": 62391.48604206275, "ttft_ms": 30.007958062924445, "input_usage": {"lookups": 6}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 61, "facts_retained": 6, "retention_rate": 0.09836065573770492, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.0, "amount": 0.3076923076923077, "status": 0.0, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 624}, "ledger_by_hop": {"1": {"facts_total": 61, "facts_retained": 7, "retention_rate": 0.11475409836065574, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.3076923076923077, "status": 0.0, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 690}, "2": {"facts_total": 61, "facts_retained": 6, "retention_rate": 0.09836065573770492, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.0, "amount": 0.3076923076923077, "status": 0.0, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 624}, "3": {"facts_total": 61, "facts_retained": 6, "retention_rate": 0.09836065573770492, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.0, "amount": 0.3076923076923077, "status": 0.0, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 624}, "4": {"facts_total": 61, "facts_retained": 6, "retention_rate": 0.09836065573770492, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.0, "amount": 0.3076923076923077, "status": 0.0, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 624}}}, "reference_key_id": "keys/synthetic/syn_03015_2000.json", "task_scores": {"correct": 1.0, "abstained": 1.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "ac61ed4025051e85e49f3f352d3d60c733554cb217faf29f390436fb9572a503", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "73cb87c0b0950ceae5a791ea", "attempt_id": "d614e07f1f9a", "run_group_id": "E4_core:A5|repairable|qwen3b:h4", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:58:28.998731Z", "task_family": "synthetic", "document_id": "syn_03015_2000", "document_hash": "71a7284b9bc3f681d69d91d18b9e8d45a4dda486cd14f4145395786ac6a68a89", "question_id": "q14_unanswerable", "split": "test", "cluster_id": "syn_03015_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 4, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1922, "query_tokens": 389, "message_tokens": 294, "message_utf8_bytes": 678, "cache_tensor_bytes": null, "serialized_payload_bytes": 678, "side_information_bytes": 0, "fetched_bytes": 3621, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 62033.760000020266, "hops": {"1": 22991.516749956645, "2": 13166.557707940228, "3": 12937.996500055306, "4": 12937.689042068087}, "receiver_prefill_ms": 232.90420894045383, "ttft_ms": 54.00300002656877, "decode_ms": 40.3168749762699}, "total_wall_ms": 62361.70212505385, "ttft_ms": 54.00300002656877, "input_usage": {"lookups": 6}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 61, "facts_retained": 6, "retention_rate": 0.09836065573770492, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.0, "amount": 0.3076923076923077, "status": 0.0, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 624}, "ledger_by_hop": {"1": {"facts_total": 61, "facts_retained": 7, "retention_rate": 0.11475409836065574, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.3076923076923077, "status": 0.0, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 690}, "2": {"facts_total": 61, "facts_retained": 6, "retention_rate": 0.09836065573770492, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.0, "amount": 0.3076923076923077, "status": 0.0, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 624}, "3": {"facts_total": 61, "facts_retained": 6, "retention_rate": 0.09836065573770492, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.0, "amount": 0.3076923076923077, "status": 0.0, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 624}, "4": {"facts_total": 61, "facts_retained": 6, "retention_rate": 0.09836065573770492, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.0, "amount": 0.3076923076923077, "status": 0.0, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 624}}}, "reference_key_id": "keys/synthetic/syn_03015_2000.json", "task_scores": {"correct": 1.0, "abstained": 1.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "ac61ed4025051e85e49f3f352d3d60c733554cb217faf29f390436fb9572a503", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "0ad00c30dece48161a3d553b", "attempt_id": "b72f1fa4575a", "run_group_id": "E4_core:A5|repairable|qwen3b:h8", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:59:21.370618Z", "task_family": "synthetic", "document_id": "syn_03015_2000", "document_hash": "71a7284b9bc3f681d69d91d18b9e8d45a4dda486cd14f4145395786ac6a68a89", "question_id": "q00_single_fact", "split": "test", "cluster_id": "syn_03015_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 8, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1922, "query_tokens": 376, "message_tokens": 294, "message_utf8_bytes": 678, "cache_tensor_bytes": null, "serialized_payload_bytes": 678, "side_information_bytes": 0, "fetched_bytes": 8449, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 114057.2336249752, "hops": {"1": 22991.516749956645, "2": 13166.557707940228, "3": 12937.996500055306, "4": 12937.689042068087, "5": 13069.826667080633, "6": 13051.929207984358, "7": 12986.152207944542, "8": 12915.565541945398}, "receiver_prefill_ms": 253.87216603849083, "ttft_ms": 42.3647080315277, "decode_ms": 35.56908294558525}, "total_wall_ms": 114389.8297919659, "ttft_ms": 42.3647080315277, "input_usage": {"lookups": 14}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 61, "facts_retained": 6, "retention_rate": 0.09836065573770492, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.0, "amount": 0.3076923076923077, "status": 0.0, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 624}, "ledger_by_hop": {"1": {"facts_total": 61, "facts_retained": 7, "retention_rate": 0.11475409836065574, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.3076923076923077, "status": 0.0, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 690}, "2": {"facts_total": 61, "facts_retained": 6, "retention_rate": 0.09836065573770492, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.0, "amount": 0.3076923076923077, "status": 0.0, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 624}, "3": {"facts_total": 61, "facts_retained": 6, "retention_rate": 0.09836065573770492, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.0, "amount": 0.3076923076923077, "status": 0.0, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 624}, "4": {"facts_total": 61, "facts_retained": 6, "retention_rate": 0.09836065573770492, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.0, "amount": 0.3076923076923077, "status": 0.0, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 624}, "5": {"facts_total": 61, "facts_retained": 6, "retention_rate": 0.09836065573770492, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.0, "amount": 0.3076923076923077, "status": 0.0, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 624}, "6": {"facts_total": 61, "facts_retained": 6, "retention_rate": 0.09836065573770492, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.0, "amount": 0.3076923076923077, "status": 0.0, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 624}, "7": {"facts_total": 61, "facts_retained": 6, "retention_rate": 0.09836065573770492, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.0, "amount": 0.3076923076923077, "status": 0.0, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 624}, "8": {"facts_total": 61, "facts_retained": 6, "retention_rate": 0.09836065573770492, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.0, "amount": 0.3076923076923077, "status": 0.0, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 624}}}, "reference_key_id": "keys/synthetic/syn_03015_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "ac61ed4025051e85e49f3f352d3d60c733554cb217faf29f390436fb9572a503", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "e7070aaac3a12b061fa097c5", "attempt_id": "38bd99c08c60", "run_group_id": "E4_core:A5|repairable|qwen3b:h8", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:59:21.721785Z", "task_family": "synthetic", "document_id": "syn_03015_2000", "document_hash": "71a7284b9bc3f681d69d91d18b9e8d45a4dda486cd14f4145395786ac6a68a89", "question_id": "q01_single_fact", "split": "test", "cluster_id": "syn_03015_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 8, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1922, "query_tokens": 381, "message_tokens": 294, "message_utf8_bytes": 678, "cache_tensor_bytes": null, "serialized_payload_bytes": 678, "side_information_bytes": 0, "fetched_bytes": 8449, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 114057.2336249752, "hops": {"1": 22991.516749956645, "2": 13166.557707940228, "3": 12937.996500055306, "4": 12937.689042068087, "5": 13069.826667080633, "6": 13051.929207984358, "7": 12986.152207944542, "8": 12915.565541945398}, "receiver_prefill_ms": 324.06383310444653, "ttft_ms": 134.7742920042947, "decode_ms": 42.0387500198558}, "total_wall_ms": 114558.65745898336, "ttft_ms": 134.7742920042947, "input_usage": {"lookups": 14}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 61, "facts_retained": 6, "retention_rate": 0.09836065573770492, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.0, "amount": 0.3076923076923077, "status": 0.0, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 624}, "ledger_by_hop": {"1": {"facts_total": 61, "facts_retained": 7, "retention_rate": 0.11475409836065574, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.3076923076923077, "status": 0.0, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 690}, "2": {"facts_total": 61, "facts_retained": 6, "retention_rate": 0.09836065573770492, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.0, "amount": 0.3076923076923077, "status": 0.0, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 624}, "3": {"facts_total": 61, "facts_retained": 6, "retention_rate": 0.09836065573770492, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.0, "amount": 0.3076923076923077, "status": 0.0, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 624}, "4": {"facts_total": 61, "facts_retained": 6, "retention_rate": 0.09836065573770492, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.0, "amount": 0.3076923076923077, "status": 0.0, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 624}, "5": {"facts_total": 61, "facts_retained": 6, "retention_rate": 0.09836065573770492, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.0, "amount": 0.3076923076923077, "status": 0.0, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 624}, "6": {"facts_total": 61, "facts_retained": 6, "retention_rate": 0.09836065573770492, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.0, "amount": 0.3076923076923077, "status": 0.0, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 624}, "7": {"facts_total": 61, "facts_retained": 6, "retention_rate": 0.09836065573770492, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.0, "amount": 0.3076923076923077, "status": 0.0, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 624}, "8": {"facts_total": 61, "facts_retained": 6, "retention_rate": 0.09836065573770492, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.0, "amount": 0.3076923076923077, "status": 0.0, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 624}}}, "reference_key_id": "keys/synthetic/syn_03015_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "ac61ed4025051e85e49f3f352d3d60c733554cb217faf29f390436fb9572a503", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "258e379b55299679dcc50472", "attempt_id": "42b67392a5fd", "run_group_id": "E4_core:A5|repairable|qwen3b:h8", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:59:22.241118Z", "task_family": "synthetic", "document_id": "syn_03015_2000", "document_hash": "71a7284b9bc3f681d69d91d18b9e8d45a4dda486cd14f4145395786ac6a68a89", "question_id": "q02_exact_copy", "split": "test", "cluster_id": "syn_03015_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 8, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1922, "query_tokens": 380, "message_tokens": 294, "message_utf8_bytes": 678, "cache_tensor_bytes": null, "serialized_payload_bytes": 678, "side_information_bytes": 0, "fetched_bytes": 8449, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 114057.2336249752, "hops": {"1": 22991.516749956645, "2": 13166.557707940228, "3": 12937.996500055306, "4": 12937.689042068087, "5": 13069.826667080633, "6": 13051.929207984358, "7": 12986.152207944542, "8": 12915.565541945398}, "receiver_prefill_ms": 278.7827500142157, "ttft_ms": 24.10012506879866, "decode_ms": 38.07908296585083}, "total_wall_ms": 114398.638249957, "ttft_ms": 24.10012506879866, "input_usage": {"lookups": 14}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 61, "facts_retained": 6, "retention_rate": 0.09836065573770492, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.0, "amount": 0.3076923076923077, "status": 0.0, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 624}, "ledger_by_hop": {"1": {"facts_total": 61, "facts_retained": 7, "retention_rate": 0.11475409836065574, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.3076923076923077, "status": 0.0, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 690}, "2": {"facts_total": 61, "facts_retained": 6, "retention_rate": 0.09836065573770492, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.0, "amount": 0.3076923076923077, "status": 0.0, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 624}, "3": {"facts_total": 61, "facts_retained": 6, "retention_rate": 0.09836065573770492, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.0, "amount": 0.3076923076923077, "status": 0.0, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 624}, "4": {"facts_total": 61, "facts_retained": 6, "retention_rate": 0.09836065573770492, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.0, "amount": 0.3076923076923077, "status": 0.0, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 624}, "5": {"facts_total": 61, "facts_retained": 6, "retention_rate": 0.09836065573770492, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.0, "amount": 0.3076923076923077, "status": 0.0, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 624}, "6": {"facts_total": 61, "facts_retained": 6, "retention_rate": 0.09836065573770492, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.0, "amount": 0.3076923076923077, "status": 0.0, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 624}, "7": {"facts_total": 61, "facts_retained": 6, "retention_rate": 0.09836065573770492, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.0, "amount": 0.3076923076923077, "status": 0.0, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 624}, "8": {"facts_total": 61, "facts_retained": 6, "retention_rate": 0.09836065573770492, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.0, "amount": 0.3076923076923077, "status": 0.0, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 624}}}, "reference_key_id": "keys/synthetic/syn_03015_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "ac61ed4025051e85e49f3f352d3d60c733554cb217faf29f390436fb9572a503", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "e4d4d6e9d07d84c2e9b93a4f", "attempt_id": "4cd51a7136ef", "run_group_id": "E4_core:A5|repairable|qwen3b:h8", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:59:22.599856Z", "task_family": "synthetic", "document_id": "syn_03015_2000", "document_hash": "71a7284b9bc3f681d69d91d18b9e8d45a4dda486cd14f4145395786ac6a68a89", "question_id": "q03_exact_copy", "split": "test", "cluster_id": "syn_03015_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 8, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1922, "query_tokens": 376, "message_tokens": 294, "message_utf8_bytes": 678, "cache_tensor_bytes": null, "serialized_payload_bytes": 678, "side_information_bytes": 0, "fetched_bytes": 8449, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 114057.2336249752, "hops": {"1": 22991.516749956645, "2": 13166.557707940228, "3": 12937.996500055306, "4": 12937.689042068087, "5": 13069.826667080633, "6": 13051.929207984358, "7": 12986.152207944542, "8": 12915.565541945398}, "receiver_prefill_ms": 287.1701249387115, "ttft_ms": 21.774665918201208, "decode_ms": 168.37316600140184}, "total_wall_ms": 114535.08812491782, "ttft_ms": 21.774665918201208, "input_usage": {"lookups": 14}, "output_usage": {"output_tokens": 6}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "Wilhelmina Dubrovnik", "evidence_artifact": {"ledger": {"facts_total": 61, "facts_retained": 6, "retention_rate": 0.09836065573770492, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.0, "amount": 0.3076923076923077, "status": 0.0, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 624}, "ledger_by_hop": {"1": {"facts_total": 61, "facts_retained": 7, "retention_rate": 0.11475409836065574, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.3076923076923077, "status": 0.0, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 690}, "2": {"facts_total": 61, "facts_retained": 6, "retention_rate": 0.09836065573770492, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.0, "amount": 0.3076923076923077, "status": 0.0, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 624}, "3": {"facts_total": 61, "facts_retained": 6, "retention_rate": 0.09836065573770492, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.0, "amount": 0.3076923076923077, "status": 0.0, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 624}, "4": {"facts_total": 61, "facts_retained": 6, "retention_rate": 0.09836065573770492, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.0, "amount": 0.3076923076923077, "status": 0.0, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 624}, "5": {"facts_total": 61, "facts_retained": 6, "retention_rate": 0.09836065573770492, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.0, "amount": 0.3076923076923077, "status": 0.0, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 624}, "6": {"facts_total": 61, "facts_retained": 6, "retention_rate": 0.09836065573770492, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.0, "amount": 0.3076923076923077, "status": 0.0, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 624}, "7": {"facts_total": 61, "facts_retained": 6, "retention_rate": 0.09836065573770492, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.0, "amount": 0.3076923076923077, "status": 0.0, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 624}, "8": {"facts_total": 61, "facts_retained": 6, "retention_rate": 0.09836065573770492, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.0, "amount": 0.3076923076923077, "status": 0.0, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 624}}}, "reference_key_id": "keys/synthetic/syn_03015_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "ac61ed4025051e85e49f3f352d3d60c733554cb217faf29f390436fb9572a503", "output_hash": "f42f05ebfc96f7258cbfbbb37cd65e114d3d15f869e68e6c95b71fc288c30a32"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "f92c6de4be0e889b09a4054f", "attempt_id": "8320e37ff748", "run_group_id": "E4_core:A5|repairable|qwen3b:h8", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:59:23.095943Z", "task_family": "synthetic", "document_id": "syn_03015_2000", "document_hash": "71a7284b9bc3f681d69d91d18b9e8d45a4dda486cd14f4145395786ac6a68a89", "question_id": "q04_multi_hop", "split": "test", "cluster_id": "syn_03015_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 8, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1922, "query_tokens": 385, "message_tokens": 294, "message_utf8_bytes": 678, "cache_tensor_bytes": null, "serialized_payload_bytes": 678, "side_information_bytes": 0, "fetched_bytes": 8449, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 114057.2336249752, "hops": {"1": 22991.516749956645, "2": 13166.557707940228, "3": 12937.996500055306, "4": 12937.689042068087, "5": 13069.826667080633, "6": 13051.929207984358, "7": 12986.152207944542, "8": 12915.565541945398}, "receiver_prefill_ms": 260.2216249797493, "ttft_ms": 48.35658299271017, "decode_ms": 479.97737501282245}, "total_wall_ms": 114846.53891704511, "ttft_ms": 48.35658299271017, "input_usage": {"lookups": 14}, "output_usage": {"output_tokens": 8}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "8421725 EUR", "evidence_artifact": {"ledger": {"facts_total": 61, "facts_retained": 6, "retention_rate": 0.09836065573770492, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.0, "amount": 0.3076923076923077, "status": 0.0, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 624}, "ledger_by_hop": {"1": {"facts_total": 61, "facts_retained": 7, "retention_rate": 0.11475409836065574, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.3076923076923077, "status": 0.0, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 690}, "2": {"facts_total": 61, "facts_retained": 6, "retention_rate": 0.09836065573770492, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.0, "amount": 0.3076923076923077, "status": 0.0, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 624}, "3": {"facts_total": 61, "facts_retained": 6, "retention_rate": 0.09836065573770492, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.0, "amount": 0.3076923076923077, "status": 0.0, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 624}, "4": {"facts_total": 61, "facts_retained": 6, "retention_rate": 0.09836065573770492, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.0, "amount": 0.3076923076923077, "status": 0.0, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 624}, "5": {"facts_total": 61, "facts_retained": 6, "retention_rate": 0.09836065573770492, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.0, "amount": 0.3076923076923077, "status": 0.0, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 624}, "6": {"facts_total": 61, "facts_retained": 6, "retention_rate": 0.09836065573770492, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.0, "amount": 0.3076923076923077, "status": 0.0, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 624}, "7": {"facts_total": 61, "facts_retained": 6, "retention_rate": 0.09836065573770492, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.0, "amount": 0.3076923076923077, "status": 0.0, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 624}, "8": {"facts_total": 61, "facts_retained": 6, "retention_rate": 0.09836065573770492, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.0, "amount": 0.3076923076923077, "status": 0.0, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 624}}}, "reference_key_id": "keys/synthetic/syn_03015_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 0.6666666666666666}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "ac61ed4025051e85e49f3f352d3d60c733554cb217faf29f390436fb9572a503", "output_hash": "976c8be2c1a175a4fb5643c5785f93e27ad9205cebd9569a64b821e1ae51ac0e"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "fbb6e79bbfd031e349860ce5", "attempt_id": "07955c74e2d5", "run_group_id": "E4_core:A5|repairable|qwen3b:h8", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:59:23.903955Z", "task_family": "synthetic", "document_id": "syn_03015_2000", "document_hash": "71a7284b9bc3f681d69d91d18b9e8d45a4dda486cd14f4145395786ac6a68a89", "question_id": "q05_multi_hop", "split": "test", "cluster_id": "syn_03015_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 8, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1922, "query_tokens": 385, "message_tokens": 294, "message_utf8_bytes": 678, "cache_tensor_bytes": null, "serialized_payload_bytes": 678, "side_information_bytes": 0, "fetched_bytes": 8449, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 114057.2336249752, "hops": {"1": 22991.516749956645, "2": 13166.557707940228, "3": 12937.996500055306, "4": 12937.689042068087, "5": 13069.826667080633, "6": 13051.929207984358, "7": 12986.152207944542, "8": 12915.565541945398}, "receiver_prefill_ms": 279.5690420316532, "ttft_ms": 26.191084063611925, "decode_ms": 37.56766696460545}, "total_wall_ms": 114400.98554093856, "ttft_ms": 26.191084063611925, "input_usage": {"lookups": 14}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 61, "facts_retained": 6, "retention_rate": 0.09836065573770492, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.0, "amount": 0.3076923076923077, "status": 0.0, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 624}, "ledger_by_hop": {"1": {"facts_total": 61, "facts_retained": 7, "retention_rate": 0.11475409836065574, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.3076923076923077, "status": 0.0, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 690}, "2": {"facts_total": 61, "facts_retained": 6, "retention_rate": 0.09836065573770492, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.0, "amount": 0.3076923076923077, "status": 0.0, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 624}, "3": {"facts_total": 61, "facts_retained": 6, "retention_rate": 0.09836065573770492, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.0, "amount": 0.3076923076923077, "status": 0.0, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 624}, "4": {"facts_total": 61, "facts_retained": 6, "retention_rate": 0.09836065573770492, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.0, "amount": 0.3076923076923077, "status": 0.0, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 624}, "5": {"facts_total": 61, "facts_retained": 6, "retention_rate": 0.09836065573770492, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.0, "amount": 0.3076923076923077, "status": 0.0, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 624}, "6": {"facts_total": 61, "facts_retained": 6, "retention_rate": 0.09836065573770492, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.0, "amount": 0.3076923076923077, "status": 0.0, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 624}, "7": {"facts_total": 61, "facts_retained": 6, "retention_rate": 0.09836065573770492, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.0, "amount": 0.3076923076923077, "status": 0.0, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 624}, "8": {"facts_total": 61, "facts_retained": 6, "retention_rate": 0.09836065573770492, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.0, "amount": 0.3076923076923077, "status": 0.0, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 624}}}, "reference_key_id": "keys/synthetic/syn_03015_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "ac61ed4025051e85e49f3f352d3d60c733554cb217faf29f390436fb9572a503", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "eaa2567cbb927e4e0f60b26c", "attempt_id": "338a130d6900", "run_group_id": "E4_core:A5|repairable|qwen3b:h8", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:59:24.265180Z", "task_family": "synthetic", "document_id": "syn_03015_2000", "document_hash": "71a7284b9bc3f681d69d91d18b9e8d45a4dda486cd14f4145395786ac6a68a89", "question_id": "q06_negation", "split": "test", "cluster_id": "syn_03015_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 8, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1922, "query_tokens": 377, "message_tokens": 294, "message_utf8_bytes": 678, "cache_tensor_bytes": null, "serialized_payload_bytes": 678, "side_information_bytes": 0, "fetched_bytes": 8449, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 114057.2336249752, "hops": {"1": 22991.516749956645, "2": 13166.557707940228, "3": 12937.996500055306, "4": 12937.689042068087, "5": 13069.826667080633, "6": 13051.929207984358, "7": 12986.152207944542, "8": 12915.565541945398}, "receiver_prefill_ms": 284.41387496422976, "ttft_ms": 21.557792089879513, "decode_ms": 21.29733399488032}, "total_wall_ms": 114384.97662497684, "ttft_ms": 21.557792089879513, "input_usage": {"lookups": 14}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 61, "facts_retained": 6, "retention_rate": 0.09836065573770492, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.0, "amount": 0.3076923076923077, "status": 0.0, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 624}, "ledger_by_hop": {"1": {"facts_total": 61, "facts_retained": 7, "retention_rate": 0.11475409836065574, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.3076923076923077, "status": 0.0, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 690}, "2": {"facts_total": 61, "facts_retained": 6, "retention_rate": 0.09836065573770492, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.0, "amount": 0.3076923076923077, "status": 0.0, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 624}, "3": {"facts_total": 61, "facts_retained": 6, "retention_rate": 0.09836065573770492, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.0, "amount": 0.3076923076923077, "status": 0.0, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 624}, "4": {"facts_total": 61, "facts_retained": 6, "retention_rate": 0.09836065573770492, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.0, "amount": 0.3076923076923077, "status": 0.0, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 624}, "5": {"facts_total": 61, "facts_retained": 6, "retention_rate": 0.09836065573770492, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.0, "amount": 0.3076923076923077, "status": 0.0, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 624}, "6": {"facts_total": 61, "facts_retained": 6, "retention_rate": 0.09836065573770492, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.0, "amount": 0.3076923076923077, "status": 0.0, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 624}, "7": {"facts_total": 61, "facts_retained": 6, "retention_rate": 0.09836065573770492, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.0, "amount": 0.3076923076923077, "status": 0.0, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 624}, "8": {"facts_total": 61, "facts_retained": 6, "retention_rate": 0.09836065573770492, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.0, "amount": 0.3076923076923077, "status": 0.0, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 624}}}, "reference_key_id": "keys/synthetic/syn_03015_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "ac61ed4025051e85e49f3f352d3d60c733554cb217faf29f390436fb9572a503", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "3f8c35c0de5892415add13b4", "attempt_id": "30b0c684a178", "run_group_id": "E4_core:A5|repairable|qwen3b:h8", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:59:24.610177Z", "task_family": "synthetic", "document_id": "syn_03015_2000", "document_hash": "71a7284b9bc3f681d69d91d18b9e8d45a4dda486cd14f4145395786ac6a68a89", "question_id": "q07_negation", "split": "test", "cluster_id": "syn_03015_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 8, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1922, "query_tokens": 377, "message_tokens": 294, "message_utf8_bytes": 678, "cache_tensor_bytes": null, "serialized_payload_bytes": 678, "side_information_bytes": 0, "fetched_bytes": 8449, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 114057.2336249752, "hops": {"1": 22991.516749956645, "2": 13166.557707940228, "3": 12937.996500055306, "4": 12937.689042068087, "5": 13069.826667080633, "6": 13051.929207984358, "7": 12986.152207944542, "8": 12915.565541945398}, "receiver_prefill_ms": 290.42862507048994, "ttft_ms": 52.44895804207772, "decode_ms": 51.41379195265472}, "total_wall_ms": 114452.11849990301, "ttft_ms": 52.44895804207772, "input_usage": {"lookups": 14}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 61, "facts_retained": 6, "retention_rate": 0.09836065573770492, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.0, "amount": 0.3076923076923077, "status": 0.0, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 624}, "ledger_by_hop": {"1": {"facts_total": 61, "facts_retained": 7, "retention_rate": 0.11475409836065574, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.3076923076923077, "status": 0.0, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 690}, "2": {"facts_total": 61, "facts_retained": 6, "retention_rate": 0.09836065573770492, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.0, "amount": 0.3076923076923077, "status": 0.0, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 624}, "3": {"facts_total": 61, "facts_retained": 6, "retention_rate": 0.09836065573770492, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.0, "amount": 0.3076923076923077, "status": 0.0, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 624}, "4": {"facts_total": 61, "facts_retained": 6, "retention_rate": 0.09836065573770492, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.0, "amount": 0.3076923076923077, "status": 0.0, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 624}, "5": {"facts_total": 61, "facts_retained": 6, "retention_rate": 0.09836065573770492, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.0, "amount": 0.3076923076923077, "status": 0.0, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 624}, "6": {"facts_total": 61, "facts_retained": 6, "retention_rate": 0.09836065573770492, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.0, "amount": 0.3076923076923077, "status": 0.0, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 624}, "7": {"facts_total": 61, "facts_retained": 6, "retention_rate": 0.09836065573770492, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.0, "amount": 0.3076923076923077, "status": 0.0, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 624}, "8": {"facts_total": 61, "facts_retained": 6, "retention_rate": 0.09836065573770492, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.0, "amount": 0.3076923076923077, "status": 0.0, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 624}}}, "reference_key_id": "keys/synthetic/syn_03015_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "ac61ed4025051e85e49f3f352d3d60c733554cb217faf29f390436fb9572a503", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "7a6db3464930c04738bfd227", "attempt_id": "3e42f94aba77", "run_group_id": "E4_core:A5|repairable|qwen3b:h8", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:59:25.024488Z", "task_family": "synthetic", "document_id": "syn_03015_2000", "document_hash": "71a7284b9bc3f681d69d91d18b9e8d45a4dda486cd14f4145395786ac6a68a89", "question_id": "q08_revision", "split": "test", "cluster_id": "syn_03015_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 8, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1922, "query_tokens": 391, "message_tokens": 294, "message_utf8_bytes": 678, "cache_tensor_bytes": null, "serialized_payload_bytes": 678, "side_information_bytes": 0, "fetched_bytes": 8449, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 114057.2336249752, "hops": {"1": 22991.516749956645, "2": 13166.557707940228, "3": 12937.996500055306, "4": 12937.689042068087, "5": 13069.826667080633, "6": 13051.929207984358, "7": 12986.152207944542, "8": 12915.565541945398}, "receiver_prefill_ms": 418.7215840211138, "ttft_ms": 38.12275000382215, "decode_ms": 24.965125019662082}, "total_wall_ms": 114539.6615419304, "ttft_ms": 38.12275000382215, "input_usage": {"lookups": 14}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 61, "facts_retained": 6, "retention_rate": 0.09836065573770492, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.0, "amount": 0.3076923076923077, "status": 0.0, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 624}, "ledger_by_hop": {"1": {"facts_total": 61, "facts_retained": 7, "retention_rate": 0.11475409836065574, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.3076923076923077, "status": 0.0, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 690}, "2": {"facts_total": 61, "facts_retained": 6, "retention_rate": 0.09836065573770492, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.0, "amount": 0.3076923076923077, "status": 0.0, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 624}, "3": {"facts_total": 61, "facts_retained": 6, "retention_rate": 0.09836065573770492, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.0, "amount": 0.3076923076923077, "status": 0.0, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 624}, "4": {"facts_total": 61, "facts_retained": 6, "retention_rate": 0.09836065573770492, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.0, "amount": 0.3076923076923077, "status": 0.0, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 624}, "5": {"facts_total": 61, "facts_retained": 6, "retention_rate": 0.09836065573770492, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.0, "amount": 0.3076923076923077, "status": 0.0, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 624}, "6": {"facts_total": 61, "facts_retained": 6, "retention_rate": 0.09836065573770492, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.0, "amount": 0.3076923076923077, "status": 0.0, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 624}, "7": {"facts_total": 61, "facts_retained": 6, "retention_rate": 0.09836065573770492, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.0, "amount": 0.3076923076923077, "status": 0.0, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 624}, "8": {"facts_total": 61, "facts_retained": 6, "retention_rate": 0.09836065573770492, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.0, "amount": 0.3076923076923077, "status": 0.0, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 624}}}, "reference_key_id": "keys/synthetic/syn_03015_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "ac61ed4025051e85e49f3f352d3d60c733554cb217faf29f390436fb9572a503", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "2474d60becabfe59b97f026c", "attempt_id": "ef10aa3709e8", "run_group_id": "E4_core:A5|repairable|qwen3b:h8", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:59:25.527570Z", "task_family": "synthetic", "document_id": "syn_03015_2000", "document_hash": "71a7284b9bc3f681d69d91d18b9e8d45a4dda486cd14f4145395786ac6a68a89", "question_id": "q09_revision", "split": "test", "cluster_id": "syn_03015_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 8, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1922, "query_tokens": 385, "message_tokens": 294, "message_utf8_bytes": 678, "cache_tensor_bytes": null, "serialized_payload_bytes": 678, "side_information_bytes": 0, "fetched_bytes": 8449, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 114057.2336249752, "hops": {"1": 22991.516749956645, "2": 13166.557707940228, "3": 12937.996500055306, "4": 12937.689042068087, "5": 13069.826667080633, "6": 13051.929207984358, "7": 12986.152207944542, "8": 12915.565541945398}, "receiver_prefill_ms": 263.6124159907922, "ttft_ms": 25.694416952319443, "decode_ms": 224.38804095145315}, "total_wall_ms": 114571.42454199493, "ttft_ms": 25.694416952319443, "input_usage": {"lookups": 14}, "output_usage": {"output_tokens": 7}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "7028925", "evidence_artifact": {"ledger": {"facts_total": 61, "facts_retained": 6, "retention_rate": 0.09836065573770492, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.0, "amount": 0.3076923076923077, "status": 0.0, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 624}, "ledger_by_hop": {"1": {"facts_total": 61, "facts_retained": 7, "retention_rate": 0.11475409836065574, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.3076923076923077, "status": 0.0, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 690}, "2": {"facts_total": 61, "facts_retained": 6, "retention_rate": 0.09836065573770492, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.0, "amount": 0.3076923076923077, "status": 0.0, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 624}, "3": {"facts_total": 61, "facts_retained": 6, "retention_rate": 0.09836065573770492, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.0, "amount": 0.3076923076923077, "status": 0.0, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 624}, "4": {"facts_total": 61, "facts_retained": 6, "retention_rate": 0.09836065573770492, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.0, "amount": 0.3076923076923077, "status": 0.0, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 624}, "5": {"facts_total": 61, "facts_retained": 6, "retention_rate": 0.09836065573770492, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.0, "amount": 0.3076923076923077, "status": 0.0, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 624}, "6": {"facts_total": 61, "facts_retained": 6, "retention_rate": 0.09836065573770492, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.0, "amount": 0.3076923076923077, "status": 0.0, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 624}, "7": {"facts_total": 61, "facts_retained": 6, "retention_rate": 0.09836065573770492, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.0, "amount": 0.3076923076923077, "status": 0.0, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 624}, "8": {"facts_total": 61, "facts_retained": 6, "retention_rate": 0.09836065573770492, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.0, "amount": 0.3076923076923077, "status": 0.0, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 624}}}, "reference_key_id": "keys/synthetic/syn_03015_2000.json", "task_scores": {"correct": 0.0, "abstained": 0.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "ac61ed4025051e85e49f3f352d3d60c733554cb217faf29f390436fb9572a503", "output_hash": "e966a398855c104a86bce07742c3ce8108a432926bc704940ecfd07e44596774"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "4612334af427d174da9fe856", "attempt_id": "723a7a64aa45", "run_group_id": "E4_core:A5|repairable|qwen3b:h8", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:59:26.060822Z", "task_family": "synthetic", "document_id": "syn_03015_2000", "document_hash": "71a7284b9bc3f681d69d91d18b9e8d45a4dda486cd14f4145395786ac6a68a89", "question_id": "q10_temporal", "split": "test", "cluster_id": "syn_03015_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 8, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1922, "query_tokens": 394, "message_tokens": 294, "message_utf8_bytes": 678, "cache_tensor_bytes": null, "serialized_payload_bytes": 678, "side_information_bytes": 0, "fetched_bytes": 8449, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 114057.2336249752, "hops": {"1": 22991.516749956645, "2": 13166.557707940228, "3": 12937.996500055306, "4": 12937.689042068087, "5": 13069.826667080633, "6": 13051.929207984358, "7": 12986.152207944542, "8": 12915.565541945398}, "receiver_prefill_ms": 257.1739580016583, "ttft_ms": 38.5488331085071, "decode_ms": 64.34858404099941}, "total_wall_ms": 114417.87891695276, "ttft_ms": 38.5488331085071, "input_usage": {"lookups": 14}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 61, "facts_retained": 6, "retention_rate": 0.09836065573770492, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.0, "amount": 0.3076923076923077, "status": 0.0, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 624}, "ledger_by_hop": {"1": {"facts_total": 61, "facts_retained": 7, "retention_rate": 0.11475409836065574, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.3076923076923077, "status": 0.0, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 690}, "2": {"facts_total": 61, "facts_retained": 6, "retention_rate": 0.09836065573770492, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.0, "amount": 0.3076923076923077, "status": 0.0, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 624}, "3": {"facts_total": 61, "facts_retained": 6, "retention_rate": 0.09836065573770492, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.0, "amount": 0.3076923076923077, "status": 0.0, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 624}, "4": {"facts_total": 61, "facts_retained": 6, "retention_rate": 0.09836065573770492, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.0, "amount": 0.3076923076923077, "status": 0.0, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 624}, "5": {"facts_total": 61, "facts_retained": 6, "retention_rate": 0.09836065573770492, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.0, "amount": 0.3076923076923077, "status": 0.0, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 624}, "6": {"facts_total": 61, "facts_retained": 6, "retention_rate": 0.09836065573770492, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.0, "amount": 0.3076923076923077, "status": 0.0, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 624}, "7": {"facts_total": 61, "facts_retained": 6, "retention_rate": 0.09836065573770492, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.0, "amount": 0.3076923076923077, "status": 0.0, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 624}, "8": {"facts_total": 61, "facts_retained": 6, "retention_rate": 0.09836065573770492, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.0, "amount": 0.3076923076923077, "status": 0.0, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 624}}}, "reference_key_id": "keys/synthetic/syn_03015_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "ac61ed4025051e85e49f3f352d3d60c733554cb217faf29f390436fb9572a503", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "18083014bb1de41ffc49fa95", "attempt_id": "1a0d4a02c91e", "run_group_id": "E4_core:A5|repairable|qwen3b:h8", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:59:26.439599Z", "task_family": "synthetic", "document_id": "syn_03015_2000", "document_hash": "71a7284b9bc3f681d69d91d18b9e8d45a4dda486cd14f4145395786ac6a68a89", "question_id": "q11_aggregation", "split": "test", "cluster_id": "syn_03015_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 8, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1922, "query_tokens": 387, "message_tokens": 294, "message_utf8_bytes": 678, "cache_tensor_bytes": null, "serialized_payload_bytes": 678, "side_information_bytes": 0, "fetched_bytes": 8449, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 114057.2336249752, "hops": {"1": 22991.516749956645, "2": 13166.557707940228, "3": 12937.996500055306, "4": 12937.689042068087, "5": 13069.826667080633, "6": 13051.929207984358, "7": 12986.152207944542, "8": 12915.565541945398}, "receiver_prefill_ms": 324.11575003061444, "ttft_ms": 64.74308401811868, "decode_ms": 109.87624991685152}, "total_wall_ms": 114556.53662490658, "ttft_ms": 64.74308401811868, "input_usage": {"lookups": 14}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 61, "facts_retained": 6, "retention_rate": 0.09836065573770492, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.0, "amount": 0.3076923076923077, "status": 0.0, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 624}, "ledger_by_hop": {"1": {"facts_total": 61, "facts_retained": 7, "retention_rate": 0.11475409836065574, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.3076923076923077, "status": 0.0, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 690}, "2": {"facts_total": 61, "facts_retained": 6, "retention_rate": 0.09836065573770492, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.0, "amount": 0.3076923076923077, "status": 0.0, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 624}, "3": {"facts_total": 61, "facts_retained": 6, "retention_rate": 0.09836065573770492, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.0, "amount": 0.3076923076923077, "status": 0.0, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 624}, "4": {"facts_total": 61, "facts_retained": 6, "retention_rate": 0.09836065573770492, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.0, "amount": 0.3076923076923077, "status": 0.0, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 624}, "5": {"facts_total": 61, "facts_retained": 6, "retention_rate": 0.09836065573770492, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.0, "amount": 0.3076923076923077, "status": 0.0, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 624}, "6": {"facts_total": 61, "facts_retained": 6, "retention_rate": 0.09836065573770492, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.0, "amount": 0.3076923076923077, "status": 0.0, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 624}, "7": {"facts_total": 61, "facts_retained": 6, "retention_rate": 0.09836065573770492, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.0, "amount": 0.3076923076923077, "status": 0.0, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 624}, "8": {"facts_total": 61, "facts_retained": 6, "retention_rate": 0.09836065573770492, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.0, "amount": 0.3076923076923077, "status": 0.0, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 624}}}, "reference_key_id": "keys/synthetic/syn_03015_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "ac61ed4025051e85e49f3f352d3d60c733554cb217faf29f390436fb9572a503", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "e825462545b115690d8d41bd", "attempt_id": "277578e31e8e", "run_group_id": "E4_core:A5|repairable|qwen3b:h8", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:59:26.958472Z", "task_family": "synthetic", "document_id": "syn_03015_2000", "document_hash": "71a7284b9bc3f681d69d91d18b9e8d45a4dda486cd14f4145395786ac6a68a89", "question_id": "q12_unit_conversion", "split": "test", "cluster_id": "syn_03015_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 8, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1922, "query_tokens": 384, "message_tokens": 294, "message_utf8_bytes": 678, "cache_tensor_bytes": null, "serialized_payload_bytes": 678, "side_information_bytes": 0, "fetched_bytes": 8449, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 114057.2336249752, "hops": {"1": 22991.516749956645, "2": 13166.557707940228, "3": 12937.996500055306, "4": 12937.689042068087, "5": 13069.826667080633, "6": 13051.929207984358, "7": 12986.152207944542, "8": 12915.565541945398}, "receiver_prefill_ms": 286.8381249718368, "ttft_ms": 25.58524999767542, "decode_ms": 42.49645792879164}, "total_wall_ms": 114412.95566607732, "ttft_ms": 25.58524999767542, "input_usage": {"lookups": 14}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 61, "facts_retained": 6, "retention_rate": 0.09836065573770492, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.0, "amount": 0.3076923076923077, "status": 0.0, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 624}, "ledger_by_hop": {"1": {"facts_total": 61, "facts_retained": 7, "retention_rate": 0.11475409836065574, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.3076923076923077, "status": 0.0, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 690}, "2": {"facts_total": 61, "facts_retained": 6, "retention_rate": 0.09836065573770492, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.0, "amount": 0.3076923076923077, "status": 0.0, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 624}, "3": {"facts_total": 61, "facts_retained": 6, "retention_rate": 0.09836065573770492, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.0, "amount": 0.3076923076923077, "status": 0.0, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 624}, "4": {"facts_total": 61, "facts_retained": 6, "retention_rate": 0.09836065573770492, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.0, "amount": 0.3076923076923077, "status": 0.0, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 624}, "5": {"facts_total": 61, "facts_retained": 6, "retention_rate": 0.09836065573770492, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.0, "amount": 0.3076923076923077, "status": 0.0, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 624}, "6": {"facts_total": 61, "facts_retained": 6, "retention_rate": 0.09836065573770492, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.0, "amount": 0.3076923076923077, "status": 0.0, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 624}, "7": {"facts_total": 61, "facts_retained": 6, "retention_rate": 0.09836065573770492, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.0, "amount": 0.3076923076923077, "status": 0.0, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 624}, "8": {"facts_total": 61, "facts_retained": 6, "retention_rate": 0.09836065573770492, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.0, "amount": 0.3076923076923077, "status": 0.0, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 624}}}, "reference_key_id": "keys/synthetic/syn_03015_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "ac61ed4025051e85e49f3f352d3d60c733554cb217faf29f390436fb9572a503", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "d54d870b625e809a809e880b", "attempt_id": "12b17e2d855d", "run_group_id": "E4_core:A5|repairable|qwen3b:h8", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:59:27.332656Z", "task_family": "synthetic", "document_id": "syn_03015_2000", "document_hash": "71a7284b9bc3f681d69d91d18b9e8d45a4dda486cd14f4145395786ac6a68a89", "question_id": "q13_unanswerable", "split": "test", "cluster_id": "syn_03015_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 8, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1922, "query_tokens": 388, "message_tokens": 294, "message_utf8_bytes": 678, "cache_tensor_bytes": null, "serialized_payload_bytes": 678, "side_information_bytes": 0, "fetched_bytes": 8449, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 114057.2336249752, "hops": {"1": 22991.516749956645, "2": 13166.557707940228, "3": 12937.996500055306, "4": 12937.689042068087, "5": 13069.826667080633, "6": 13051.929207984358, "7": 12986.152207944542, "8": 12915.565541945398}, "receiver_prefill_ms": 301.6358339227736, "ttft_ms": 36.78520803805441, "decode_ms": 32.64416696038097}, "total_wall_ms": 114428.73358295765, "ttft_ms": 36.78520803805441, "input_usage": {"lookups": 14}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 61, "facts_retained": 6, "retention_rate": 0.09836065573770492, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.0, "amount": 0.3076923076923077, "status": 0.0, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 624}, "ledger_by_hop": {"1": {"facts_total": 61, "facts_retained": 7, "retention_rate": 0.11475409836065574, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.3076923076923077, "status": 0.0, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 690}, "2": {"facts_total": 61, "facts_retained": 6, "retention_rate": 0.09836065573770492, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.0, "amount": 0.3076923076923077, "status": 0.0, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 624}, "3": {"facts_total": 61, "facts_retained": 6, "retention_rate": 0.09836065573770492, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.0, "amount": 0.3076923076923077, "status": 0.0, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 624}, "4": {"facts_total": 61, "facts_retained": 6, "retention_rate": 0.09836065573770492, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.0, "amount": 0.3076923076923077, "status": 0.0, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 624}, "5": {"facts_total": 61, "facts_retained": 6, "retention_rate": 0.09836065573770492, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.0, "amount": 0.3076923076923077, "status": 0.0, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 624}, "6": {"facts_total": 61, "facts_retained": 6, "retention_rate": 0.09836065573770492, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.0, "amount": 0.3076923076923077, "status": 0.0, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 624}, "7": {"facts_total": 61, "facts_retained": 6, "retention_rate": 0.09836065573770492, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.0, "amount": 0.3076923076923077, "status": 0.0, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 624}, "8": {"facts_total": 61, "facts_retained": 6, "retention_rate": 0.09836065573770492, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.0, "amount": 0.3076923076923077, "status": 0.0, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 624}}}, "reference_key_id": "keys/synthetic/syn_03015_2000.json", "task_scores": {"correct": 1.0, "abstained": 1.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "ac61ed4025051e85e49f3f352d3d60c733554cb217faf29f390436fb9572a503", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "7f2ed9e539531c974d0bc239", "attempt_id": "617c9cd26e5b", "run_group_id": "E4_core:A5|repairable|qwen3b:h8", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:59:27.720854Z", "task_family": "synthetic", "document_id": "syn_03015_2000", "document_hash": "71a7284b9bc3f681d69d91d18b9e8d45a4dda486cd14f4145395786ac6a68a89", "question_id": "q14_unanswerable", "split": "test", "cluster_id": "syn_03015_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 8, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1922, "query_tokens": 389, "message_tokens": 294, "message_utf8_bytes": 678, "cache_tensor_bytes": null, "serialized_payload_bytes": 678, "side_information_bytes": 0, "fetched_bytes": 8449, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 114057.2336249752, "hops": {"1": 22991.516749956645, "2": 13166.557707940228, "3": 12937.996500055306, "4": 12937.689042068087, "5": 13069.826667080633, "6": 13051.929207984358, "7": 12986.152207944542, "8": 12915.565541945398}, "receiver_prefill_ms": 264.6162500604987, "ttft_ms": 38.76995900645852, "decode_ms": 52.922582952305675}, "total_wall_ms": 114414.12879095878, "ttft_ms": 38.76995900645852, "input_usage": {"lookups": 14}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 61, "facts_retained": 6, "retention_rate": 0.09836065573770492, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.0, "amount": 0.3076923076923077, "status": 0.0, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 624}, "ledger_by_hop": {"1": {"facts_total": 61, "facts_retained": 7, "retention_rate": 0.11475409836065574, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.3076923076923077, "status": 0.0, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 690}, "2": {"facts_total": 61, "facts_retained": 6, "retention_rate": 0.09836065573770492, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.0, "amount": 0.3076923076923077, "status": 0.0, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 624}, "3": {"facts_total": 61, "facts_retained": 6, "retention_rate": 0.09836065573770492, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.0, "amount": 0.3076923076923077, "status": 0.0, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 624}, "4": {"facts_total": 61, "facts_retained": 6, "retention_rate": 0.09836065573770492, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.0, "amount": 0.3076923076923077, "status": 0.0, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 624}, "5": {"facts_total": 61, "facts_retained": 6, "retention_rate": 0.09836065573770492, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.0, "amount": 0.3076923076923077, "status": 0.0, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 624}, "6": {"facts_total": 61, "facts_retained": 6, "retention_rate": 0.09836065573770492, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.0, "amount": 0.3076923076923077, "status": 0.0, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 624}, "7": {"facts_total": 61, "facts_retained": 6, "retention_rate": 0.09836065573770492, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.0, "amount": 0.3076923076923077, "status": 0.0, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 624}, "8": {"facts_total": 61, "facts_retained": 6, "retention_rate": 0.09836065573770492, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.0, "amount": 0.3076923076923077, "status": 0.0, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 624}}}, "reference_key_id": "keys/synthetic/syn_03015_2000.json", "task_scores": {"correct": 1.0, "abstained": 1.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "ac61ed4025051e85e49f3f352d3d60c733554cb217faf29f390436fb9572a503", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "4426de5356e3cdc7eaad597d", "attempt_id": "890f71026c7c", "run_group_id": "E4_core:A4|closed|qwen3b-llama3b:h1", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:59:51.022329Z", "task_family": "synthetic", "document_id": "syn_03015_2000", "document_hash": "71a7284b9bc3f681d69d91d18b9e8d45a4dda486cd14f4145395786ac6a68a89", "question_id": "q00_single_fact", "split": "test", "cluster_id": "syn_03015_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 1, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1922, "query_tokens": 595, "message_tokens": 512, "message_utf8_bytes": 1161, "cache_tensor_bytes": null, "serialized_payload_bytes": 1161, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 22925.759125035256, "hops": {"1": 22925.759125035256}, "receiver_prefill_ms": 400.5750840296969, "ttft_ms": 19.287916948087513, "decode_ms": 39.11691706161946}, "total_wall_ms": 23385.511875036173, "ttft_ms": 19.287916948087513, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 61, "facts_retained": 24, "retention_rate": 0.39344262295081966, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.4166666666666667, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1161}, "ledger_by_hop": {"1": {"facts_total": 61, "facts_retained": 24, "retention_rate": 0.39344262295081966, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.4166666666666667, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1161}}}, "reference_key_id": "keys/synthetic/syn_03015_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "52d2d6723cbb6f8139ce72290860dae098b28331488ef44459da0ce2ae9f9f72", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "a386bce7495da6cc1969a802", "attempt_id": "c867aa78955a", "run_group_id": "E4_core:A4|closed|qwen3b-llama3b:h1", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:59:51.500707Z", "task_family": "synthetic", "document_id": "syn_03015_2000", "document_hash": "71a7284b9bc3f681d69d91d18b9e8d45a4dda486cd14f4145395786ac6a68a89", "question_id": "q01_single_fact", "split": "test", "cluster_id": "syn_03015_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 1, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1922, "query_tokens": 600, "message_tokens": 512, "message_utf8_bytes": 1161, "cache_tensor_bytes": null, "serialized_payload_bytes": 1161, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 22925.759125035256, "hops": {"1": 22925.759125035256}, "receiver_prefill_ms": 357.4147910112515, "ttft_ms": 39.346624980680645, "decode_ms": 70.46650000847876}, "total_wall_ms": 23393.672208068892, "ttft_ms": 39.346624980680645, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 61, "facts_retained": 24, "retention_rate": 0.39344262295081966, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.4166666666666667, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1161}, "ledger_by_hop": {"1": {"facts_total": 61, "facts_retained": 24, "retention_rate": 0.39344262295081966, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.4166666666666667, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1161}}}, "reference_key_id": "keys/synthetic/syn_03015_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "52d2d6723cbb6f8139ce72290860dae098b28331488ef44459da0ce2ae9f9f72", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "8f7c2137fbeda20c2a19b5c8", "attempt_id": "ccd2975ffd05", "run_group_id": "E4_core:A4|closed|qwen3b-llama3b:h1", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:59:51.985973Z", "task_family": "synthetic", "document_id": "syn_03015_2000", "document_hash": "71a7284b9bc3f681d69d91d18b9e8d45a4dda486cd14f4145395786ac6a68a89", "question_id": "q02_exact_copy", "split": "test", "cluster_id": "syn_03015_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 1, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1922, "query_tokens": 599, "message_tokens": 512, "message_utf8_bytes": 1161, "cache_tensor_bytes": null, "serialized_payload_bytes": 1161, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 22925.759125035256, "hops": {"1": 22925.759125035256}, "receiver_prefill_ms": 537.3621670296416, "ttft_ms": 23.32608390133828, "decode_ms": 226.15616698749363}, "total_wall_ms": 23713.376083993353, "ttft_ms": 23.32608390133828, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 7}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "89.3122", "evidence_artifact": {"ledger": {"facts_total": 61, "facts_retained": 24, "retention_rate": 0.39344262295081966, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.4166666666666667, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1161}, "ledger_by_hop": {"1": {"facts_total": 61, "facts_retained": 24, "retention_rate": 0.39344262295081966, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.4166666666666667, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1161}}}, "reference_key_id": "keys/synthetic/syn_03015_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "52d2d6723cbb6f8139ce72290860dae098b28331488ef44459da0ce2ae9f9f72", "output_hash": "2138888e6519b2f4b1b54142e6f9c541102d7f9ee5bcf8e4184c81a382601757"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "d99d5da38dc92342b27a89e9", "attempt_id": "1f2468f99e45", "run_group_id": "E4_core:A4|closed|qwen3b-llama3b:h1", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:59:52.793136Z", "task_family": "synthetic", "document_id": "syn_03015_2000", "document_hash": "71a7284b9bc3f681d69d91d18b9e8d45a4dda486cd14f4145395786ac6a68a89", "question_id": "q03_exact_copy", "split": "test", "cluster_id": "syn_03015_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 1, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1922, "query_tokens": 595, "message_tokens": 512, "message_utf8_bytes": 1161, "cache_tensor_bytes": null, "serialized_payload_bytes": 1161, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 22925.759125035256, "hops": {"1": 22925.759125035256}, "receiver_prefill_ms": 419.9584589805454, "ttft_ms": 24.365541990846395, "decode_ms": 251.72920804470778}, "total_wall_ms": 23622.537291026674, "ttft_ms": 24.365541990846395, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 6}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "Wilhelmina Dubrovnik", "evidence_artifact": {"ledger": {"facts_total": 61, "facts_retained": 24, "retention_rate": 0.39344262295081966, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.4166666666666667, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1161}, "ledger_by_hop": {"1": {"facts_total": 61, "facts_retained": 24, "retention_rate": 0.39344262295081966, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.4166666666666667, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1161}}}, "reference_key_id": "keys/synthetic/syn_03015_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "52d2d6723cbb6f8139ce72290860dae098b28331488ef44459da0ce2ae9f9f72", "output_hash": "f42f05ebfc96f7258cbfbbb37cd65e114d3d15f869e68e6c95b71fc288c30a32"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "977b80b6da0f2cd213f65b1d", "attempt_id": "1ec98b3608ea", "run_group_id": "E4_core:A4|closed|qwen3b-llama3b:h1", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:59:53.509159Z", "task_family": "synthetic", "document_id": "syn_03015_2000", "document_hash": "71a7284b9bc3f681d69d91d18b9e8d45a4dda486cd14f4145395786ac6a68a89", "question_id": "q04_multi_hop", "split": "test", "cluster_id": "syn_03015_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 1, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1922, "query_tokens": 604, "message_tokens": 512, "message_utf8_bytes": 1161, "cache_tensor_bytes": null, "serialized_payload_bytes": 1161, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 22925.759125035256, "hops": {"1": 22925.759125035256}, "receiver_prefill_ms": 558.3667079918087, "ttft_ms": 21.91233297344297, "decode_ms": 224.4418750051409}, "total_wall_ms": 23731.188167002983, "ttft_ms": 21.91233297344297, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 7}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "8421725", "evidence_artifact": {"ledger": {"facts_total": 61, "facts_retained": 24, "retention_rate": 0.39344262295081966, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.4166666666666667, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1161}, "ledger_by_hop": {"1": {"facts_total": 61, "facts_retained": 24, "retention_rate": 0.39344262295081966, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.4166666666666667, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1161}}}, "reference_key_id": "keys/synthetic/syn_03015_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "52d2d6723cbb6f8139ce72290860dae098b28331488ef44459da0ce2ae9f9f72", "output_hash": "e4949aba3b0b77de05a87edaab854329fa75df7bf68b29a3def71e985a06e5b5"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "4dba8a6fe9c6aabd36a8672b", "attempt_id": "491d45ce31ec", "run_group_id": "E4_core:A4|closed|qwen3b-llama3b:h1", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:59:54.332464Z", "task_family": "synthetic", "document_id": "syn_03015_2000", "document_hash": "71a7284b9bc3f681d69d91d18b9e8d45a4dda486cd14f4145395786ac6a68a89", "question_id": "q05_multi_hop", "split": "test", "cluster_id": "syn_03015_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 1, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1922, "query_tokens": 604, "message_tokens": 512, "message_utf8_bytes": 1161, "cache_tensor_bytes": null, "serialized_payload_bytes": 1161, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 22925.759125035256, "hops": {"1": 22925.759125035256}, "receiver_prefill_ms": 407.48775005340576, "ttft_ms": 25.047458009794354, "decode_ms": 38.05037506390363}, "total_wall_ms": 23397.101667011157, "ttft_ms": 25.047458009794354, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 61, "facts_retained": 24, "retention_rate": 0.39344262295081966, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.4166666666666667, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1161}, "ledger_by_hop": {"1": {"facts_total": 61, "facts_retained": 24, "retention_rate": 0.39344262295081966, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.4166666666666667, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1161}}}, "reference_key_id": "keys/synthetic/syn_03015_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "52d2d6723cbb6f8139ce72290860dae098b28331488ef44459da0ce2ae9f9f72", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "c5e5ad9f3fa6a45517fc0cdf", "attempt_id": "ee6d377b04b0", "run_group_id": "E4_core:A4|closed|qwen3b-llama3b:h1", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:59:54.822197Z", "task_family": "synthetic", "document_id": "syn_03015_2000", "document_hash": "71a7284b9bc3f681d69d91d18b9e8d45a4dda486cd14f4145395786ac6a68a89", "question_id": "q06_negation", "split": "test", "cluster_id": "syn_03015_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 1, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1922, "query_tokens": 596, "message_tokens": 512, "message_utf8_bytes": 1161, "cache_tensor_bytes": null, "serialized_payload_bytes": 1161, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 22925.759125035256, "hops": {"1": 22925.759125035256}, "receiver_prefill_ms": 345.3142090002075, "ttft_ms": 51.62500008009374, "decode_ms": 42.42766590323299}, "total_wall_ms": 23366.09825002961, "ttft_ms": 51.62500008009374, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 61, "facts_retained": 24, "retention_rate": 0.39344262295081966, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.4166666666666667, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1161}, "ledger_by_hop": {"1": {"facts_total": 61, "facts_retained": 24, "retention_rate": 0.39344262295081966, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.4166666666666667, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1161}}}, "reference_key_id": "keys/synthetic/syn_03015_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "52d2d6723cbb6f8139ce72290860dae098b28331488ef44459da0ce2ae9f9f72", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "ec0720d14e22bcb3805477c8", "attempt_id": "9878e92d596f", "run_group_id": "E4_core:A4|closed|qwen3b-llama3b:h1", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:59:55.280898Z", "task_family": "synthetic", "document_id": "syn_03015_2000", "document_hash": "71a7284b9bc3f681d69d91d18b9e8d45a4dda486cd14f4145395786ac6a68a89", "question_id": "q07_negation", "split": "test", "cluster_id": "syn_03015_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 1, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1922, "query_tokens": 596, "message_tokens": 512, "message_utf8_bytes": 1161, "cache_tensor_bytes": null, "serialized_payload_bytes": 1161, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 22925.759125035256, "hops": {"1": 22925.759125035256}, "receiver_prefill_ms": 529.6176670817658, "ttft_ms": 30.512791010551155, "decode_ms": 39.056333946064115}, "total_wall_ms": 23525.5715000676, "ttft_ms": 30.512791010551155, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 61, "facts_retained": 24, "retention_rate": 0.39344262295081966, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.4166666666666667, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1161}, "ledger_by_hop": {"1": {"facts_total": 61, "facts_retained": 24, "retention_rate": 0.39344262295081966, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.4166666666666667, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1161}}}, "reference_key_id": "keys/synthetic/syn_03015_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "52d2d6723cbb6f8139ce72290860dae098b28331488ef44459da0ce2ae9f9f72", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "360ca6cfbf3051b804a2d54c", "attempt_id": "3d9e7e3dc8f1", "run_group_id": "E4_core:A4|closed|qwen3b-llama3b:h1", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:59:55.897939Z", "task_family": "synthetic", "document_id": "syn_03015_2000", "document_hash": "71a7284b9bc3f681d69d91d18b9e8d45a4dda486cd14f4145395786ac6a68a89", "question_id": "q08_revision", "split": "test", "cluster_id": "syn_03015_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 1, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1922, "query_tokens": 610, "message_tokens": 512, "message_utf8_bytes": 1161, "cache_tensor_bytes": null, "serialized_payload_bytes": 1161, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 22925.759125035256, "hops": {"1": 22925.759125035256}, "receiver_prefill_ms": 399.04779102653265, "ttft_ms": 23.40279205236584, "decode_ms": 37.99208300188184}, "total_wall_ms": 23386.87500008382, "ttft_ms": 23.40279205236584, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 61, "facts_retained": 24, "retention_rate": 0.39344262295081966, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.4166666666666667, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1161}, "ledger_by_hop": {"1": {"facts_total": 61, "facts_retained": 24, "retention_rate": 0.39344262295081966, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.4166666666666667, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1161}}}, "reference_key_id": "keys/synthetic/syn_03015_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "52d2d6723cbb6f8139ce72290860dae098b28331488ef44459da0ce2ae9f9f72", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "5d32d33e48291bcb3c96f24f", "attempt_id": "5660963d4ee5", "run_group_id": "E4_core:A4|closed|qwen3b-llama3b:h1", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:59:56.376851Z", "task_family": "synthetic", "document_id": "syn_03015_2000", "document_hash": "71a7284b9bc3f681d69d91d18b9e8d45a4dda486cd14f4145395786ac6a68a89", "question_id": "q09_revision", "split": "test", "cluster_id": "syn_03015_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 1, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1922, "query_tokens": 604, "message_tokens": 512, "message_utf8_bytes": 1161, "cache_tensor_bytes": null, "serialized_payload_bytes": 1161, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 22925.759125035256, "hops": {"1": 22925.759125035256}, "receiver_prefill_ms": 363.90020803082734, "ttft_ms": 46.990999951958656, "decode_ms": 586.5680000279099}, "total_wall_ms": 23924.255792051554, "ttft_ms": 46.990999951958656, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 9}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "8,377,480", "evidence_artifact": {"ledger": {"facts_total": 61, "facts_retained": 24, "retention_rate": 0.39344262295081966, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.4166666666666667, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1161}, "ledger_by_hop": {"1": {"facts_total": 61, "facts_retained": 24, "retention_rate": 0.39344262295081966, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.4166666666666667, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1161}}}, "reference_key_id": "keys/synthetic/syn_03015_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "52d2d6723cbb6f8139ce72290860dae098b28331488ef44459da0ce2ae9f9f72", "output_hash": "ec4c68b327a2d2427ed21821415a5f9ddf87134bd9f654bdd69c7d1301febcb0"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "e10d35c57c52018b2d4abec5", "attempt_id": "2bccec49d44c", "run_group_id": "E4_core:A4|closed|qwen3b-llama3b:h1", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:59:57.393737Z", "task_family": "synthetic", "document_id": "syn_03015_2000", "document_hash": "71a7284b9bc3f681d69d91d18b9e8d45a4dda486cd14f4145395786ac6a68a89", "question_id": "q10_temporal", "split": "test", "cluster_id": "syn_03015_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 1, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1922, "query_tokens": 613, "message_tokens": 512, "message_utf8_bytes": 1161, "cache_tensor_bytes": null, "serialized_payload_bytes": 1161, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 22925.759125035256, "hops": {"1": 22925.759125035256}, "receiver_prefill_ms": 395.7355839665979, "ttft_ms": 38.686083047650754, "decode_ms": 24.843916995450854}, "total_wall_ms": 23385.884750052355, "ttft_ms": 38.686083047650754, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 61, "facts_retained": 24, "retention_rate": 0.39344262295081966, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.4166666666666667, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1161}, "ledger_by_hop": {"1": {"facts_total": 61, "facts_retained": 24, "retention_rate": 0.39344262295081966, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.4166666666666667, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1161}}}, "reference_key_id": "keys/synthetic/syn_03015_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "52d2d6723cbb6f8139ce72290860dae098b28331488ef44459da0ce2ae9f9f72", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "45f9e6cd89e6e525283454c3", "attempt_id": "84b0170eb88a", "run_group_id": "E4_core:A4|closed|qwen3b-llama3b:h1", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:59:57.871014Z", "task_family": "synthetic", "document_id": "syn_03015_2000", "document_hash": "71a7284b9bc3f681d69d91d18b9e8d45a4dda486cd14f4145395786ac6a68a89", "question_id": "q11_aggregation", "split": "test", "cluster_id": "syn_03015_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 1, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1922, "query_tokens": 606, "message_tokens": 512, "message_utf8_bytes": 1161, "cache_tensor_bytes": null, "serialized_payload_bytes": 1161, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 22925.759125035256, "hops": {"1": 22925.759125035256}, "receiver_prefill_ms": 402.69741602241993, "ttft_ms": 25.211708969436586, "decode_ms": 313.61291697248816}, "total_wall_ms": 23668.080709059723, "ttft_ms": 25.211708969436586, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 7}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "1806025", "evidence_artifact": {"ledger": {"facts_total": 61, "facts_retained": 24, "retention_rate": 0.39344262295081966, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.4166666666666667, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1161}, "ledger_by_hop": {"1": {"facts_total": 61, "facts_retained": 24, "retention_rate": 0.39344262295081966, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.4166666666666667, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1161}}}, "reference_key_id": "keys/synthetic/syn_03015_2000.json", "task_scores": {"correct": 0.0, "abstained": 0.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "52d2d6723cbb6f8139ce72290860dae098b28331488ef44459da0ce2ae9f9f72", "output_hash": "9801d1840792d286693349934c9cc13fead44d82d480a42803cf6015c01431f6"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "b37619390d940e7653b1a5dd", "attempt_id": "e72acc0f33ab", "run_group_id": "E4_core:A4|closed|qwen3b-llama3b:h1", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:59:58.632108Z", "task_family": "synthetic", "document_id": "syn_03015_2000", "document_hash": "71a7284b9bc3f681d69d91d18b9e8d45a4dda486cd14f4145395786ac6a68a89", "question_id": "q12_unit_conversion", "split": "test", "cluster_id": "syn_03015_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 1, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1922, "query_tokens": 603, "message_tokens": 512, "message_utf8_bytes": 1161, "cache_tensor_bytes": null, "serialized_payload_bytes": 1161, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 22925.759125035256, "hops": {"1": 22925.759125035256}, "receiver_prefill_ms": 509.87679208628833, "ttft_ms": 34.00233399588615, "decode_ms": 30.556750018149614}, "total_wall_ms": 23500.812542042695, "ttft_ms": 34.00233399588615, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 61, "facts_retained": 24, "retention_rate": 0.39344262295081966, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.4166666666666667, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1161}, "ledger_by_hop": {"1": {"facts_total": 61, "facts_retained": 24, "retention_rate": 0.39344262295081966, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.4166666666666667, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1161}}}, "reference_key_id": "keys/synthetic/syn_03015_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "52d2d6723cbb6f8139ce72290860dae098b28331488ef44459da0ce2ae9f9f72", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "a363bd97ac3b28bbef32b725", "attempt_id": "6e48df944fb8", "run_group_id": "E4_core:A4|closed|qwen3b-llama3b:h1", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:59:59.225053Z", "task_family": "synthetic", "document_id": "syn_03015_2000", "document_hash": "71a7284b9bc3f681d69d91d18b9e8d45a4dda486cd14f4145395786ac6a68a89", "question_id": "q13_unanswerable", "split": "test", "cluster_id": "syn_03015_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 1, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1922, "query_tokens": 607, "message_tokens": 512, "message_utf8_bytes": 1161, "cache_tensor_bytes": null, "serialized_payload_bytes": 1161, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 22925.759125035256, "hops": {"1": 22925.759125035256}, "receiver_prefill_ms": 407.13466599117965, "ttft_ms": 38.95062499213964, "decode_ms": 25.452582980506122}, "total_wall_ms": 23397.98570901621, "ttft_ms": 38.95062499213964, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 61, "facts_retained": 24, "retention_rate": 0.39344262295081966, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.4166666666666667, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1161}, "ledger_by_hop": {"1": {"facts_total": 61, "facts_retained": 24, "retention_rate": 0.39344262295081966, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.4166666666666667, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1161}}}, "reference_key_id": "keys/synthetic/syn_03015_2000.json", "task_scores": {"correct": 1.0, "abstained": 1.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "52d2d6723cbb6f8139ce72290860dae098b28331488ef44459da0ce2ae9f9f72", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "7e671e7c91b80a9dcf061ed7", "attempt_id": "0f4b9a95d13b", "run_group_id": "E4_core:A4|closed|qwen3b-llama3b:h1", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T07:59:59.715069Z", "task_family": "synthetic", "document_id": "syn_03015_2000", "document_hash": "71a7284b9bc3f681d69d91d18b9e8d45a4dda486cd14f4145395786ac6a68a89", "question_id": "q14_unanswerable", "split": "test", "cluster_id": "syn_03015_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 1, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1922, "query_tokens": 608, "message_tokens": 512, "message_utf8_bytes": 1161, "cache_tensor_bytes": null, "serialized_payload_bytes": 1161, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 22925.759125035256, "hops": {"1": 22925.759125035256}, "receiver_prefill_ms": 369.25054108724, "ttft_ms": 53.22475009597838, "decode_ms": 61.32041697856039}, "total_wall_ms": 23410.279333009385, "ttft_ms": 53.22475009597838, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 61, "facts_retained": 24, "retention_rate": 0.39344262295081966, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.4166666666666667, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1161}, "ledger_by_hop": {"1": {"facts_total": 61, "facts_retained": 24, "retention_rate": 0.39344262295081966, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.4166666666666667, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1161}}}, "reference_key_id": "keys/synthetic/syn_03015_2000.json", "task_scores": {"correct": 1.0, "abstained": 1.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "52d2d6723cbb6f8139ce72290860dae098b28331488ef44459da0ce2ae9f9f72", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "274e7bf120af8fdd04548841", "attempt_id": "13fec790937a", "run_group_id": "E4_core:A4|closed|qwen3b-llama3b:h2", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:00:16.726982Z", "task_family": "synthetic", "document_id": "syn_03015_2000", "document_hash": "71a7284b9bc3f681d69d91d18b9e8d45a4dda486cd14f4145395786ac6a68a89", "question_id": "q00_single_fact", "split": "test", "cluster_id": "syn_03015_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "tokenizer_hashes": {"receiver": "b9d8872cf3a99da99e1fc4b0f71232a451632dc49ca4c6771549b4efa5c1c763"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 2, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1741, "query_tokens": 525, "message_tokens": 421, "message_utf8_bytes": 1123, "cache_tensor_bytes": null, "serialized_payload_bytes": 1123, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 39433.85270796716, "hops": {"1": 22925.759125035256, "2": 16508.093582931906}, "receiver_prefill_ms": 335.7034999644384, "ttft_ms": 35.11950001120567, "decode_ms": 53.579833009280264}, "total_wall_ms": 39858.93958294764, "ttft_ms": 35.11950001120567, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 61, "facts_retained": 24, "retention_rate": 0.39344262295081966, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.4166666666666667, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1123}, "ledger_by_hop": {"1": {"facts_total": 61, "facts_retained": 24, "retention_rate": 0.39344262295081966, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.4166666666666667, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1161}, "2": {"facts_total": 61, "facts_retained": 24, "retention_rate": 0.39344262295081966, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.4166666666666667, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1123}}}, "reference_key_id": "keys/synthetic/syn_03015_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "dd270f0df70dd6fa49ad5b20f7ef7fabbbd5bdc8e942a2b6f2a26e351c41b754", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "c3f81f56be786ef1d6e0d47f", "attempt_id": "275774917498", "run_group_id": "E4_core:A4|closed|qwen3b-llama3b:h2", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:00:17.171895Z", "task_family": "synthetic", "document_id": "syn_03015_2000", "document_hash": "71a7284b9bc3f681d69d91d18b9e8d45a4dda486cd14f4145395786ac6a68a89", "question_id": "q01_single_fact", "split": "test", "cluster_id": "syn_03015_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "tokenizer_hashes": {"receiver": "b9d8872cf3a99da99e1fc4b0f71232a451632dc49ca4c6771549b4efa5c1c763"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 2, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1741, "query_tokens": 529, "message_tokens": 421, "message_utf8_bytes": 1123, "cache_tensor_bytes": null, "serialized_payload_bytes": 1123, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 39433.85270796716, "hops": {"1": 22925.759125035256, "2": 16508.093582931906}, "receiver_prefill_ms": 474.08495796844363, "ttft_ms": 25.97241709008813, "decode_ms": 37.01358300168067}, "total_wall_ms": 39971.52029105928, "ttft_ms": 25.97241709008813, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 61, "facts_retained": 24, "retention_rate": 0.39344262295081966, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.4166666666666667, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1123}, "ledger_by_hop": {"1": {"facts_total": 61, "facts_retained": 24, "retention_rate": 0.39344262295081966, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.4166666666666667, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1161}, "2": {"facts_total": 61, "facts_retained": 24, "retention_rate": 0.39344262295081966, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.4166666666666667, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1123}}}, "reference_key_id": "keys/synthetic/syn_03015_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "dd270f0df70dd6fa49ad5b20f7ef7fabbbd5bdc8e942a2b6f2a26e351c41b754", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "c8a6fed638674d5d09958542", "attempt_id": "dfec997fb7b5", "run_group_id": "E4_core:A4|closed|qwen3b-llama3b:h2", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:00:17.728403Z", "task_family": "synthetic", "document_id": "syn_03015_2000", "document_hash": "71a7284b9bc3f681d69d91d18b9e8d45a4dda486cd14f4145395786ac6a68a89", "question_id": "q02_exact_copy", "split": "test", "cluster_id": "syn_03015_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "tokenizer_hashes": {"receiver": "b9d8872cf3a99da99e1fc4b0f71232a451632dc49ca4c6771549b4efa5c1c763"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 2, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1741, "query_tokens": 529, "message_tokens": 421, "message_utf8_bytes": 1123, "cache_tensor_bytes": null, "serialized_payload_bytes": 1123, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 39433.85270796716, "hops": {"1": 22925.759125035256, "2": 16508.093582931906}, "receiver_prefill_ms": 339.0515410574153, "ttft_ms": 39.81816698797047, "decode_ms": 128.02295805886388}, "total_wall_ms": 39941.4573329268, "ttft_ms": 39.81816698797047, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 4}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "89.3122", "evidence_artifact": {"ledger": {"facts_total": 61, "facts_retained": 24, "retention_rate": 0.39344262295081966, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.4166666666666667, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1123}, "ledger_by_hop": {"1": {"facts_total": 61, "facts_retained": 24, "retention_rate": 0.39344262295081966, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.4166666666666667, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1161}, "2": {"facts_total": 61, "facts_retained": 24, "retention_rate": 0.39344262295081966, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.4166666666666667, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1123}}}, "reference_key_id": "keys/synthetic/syn_03015_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "dd270f0df70dd6fa49ad5b20f7ef7fabbbd5bdc8e942a2b6f2a26e351c41b754", "output_hash": "2138888e6519b2f4b1b54142e6f9c541102d7f9ee5bcf8e4184c81a382601757"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "b757f82e25b26a9685fdc285", "attempt_id": "2c40a42f398c", "run_group_id": "E4_core:A4|closed|qwen3b-llama3b:h2", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:00:18.253805Z", "task_family": "synthetic", "document_id": "syn_03015_2000", "document_hash": "71a7284b9bc3f681d69d91d18b9e8d45a4dda486cd14f4145395786ac6a68a89", "question_id": "q03_exact_copy", "split": "test", "cluster_id": "syn_03015_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "tokenizer_hashes": {"receiver": "b9d8872cf3a99da99e1fc4b0f71232a451632dc49ca4c6771549b4efa5c1c763"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 2, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1741, "query_tokens": 525, "message_tokens": 421, "message_utf8_bytes": 1123, "cache_tensor_bytes": null, "serialized_payload_bytes": 1123, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 39433.85270796716, "hops": {"1": 22925.759125035256, "2": 16508.093582931906}, "receiver_prefill_ms": 333.03579199127853, "ttft_ms": 47.1873750211671, "decode_ms": 472.18787495512515}, "total_wall_ms": 40286.935290903784, "ttft_ms": 47.1873750211671, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 7}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "Wilhelmina Dubrovnik.", "evidence_artifact": {"ledger": {"facts_total": 61, "facts_retained": 24, "retention_rate": 0.39344262295081966, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.4166666666666667, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1123}, "ledger_by_hop": {"1": {"facts_total": 61, "facts_retained": 24, "retention_rate": 0.39344262295081966, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.4166666666666667, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1161}, "2": {"facts_total": 61, "facts_retained": 24, "retention_rate": 0.39344262295081966, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.4166666666666667, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1123}}}, "reference_key_id": "keys/synthetic/syn_03015_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "dd270f0df70dd6fa49ad5b20f7ef7fabbbd5bdc8e942a2b6f2a26e351c41b754", "output_hash": "79ba0181ee6c8e2b01e096d290463d1e2eb3a595aa1abc3242a857276f5ca877"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "af8a7362b5b33f256d81239c", "attempt_id": "ee71cb2b0793", "run_group_id": "E4_core:A4|closed|qwen3b-llama3b:h2", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:00:19.124449Z", "task_family": "synthetic", "document_id": "syn_03015_2000", "document_hash": "71a7284b9bc3f681d69d91d18b9e8d45a4dda486cd14f4145395786ac6a68a89", "question_id": "q04_multi_hop", "split": "test", "cluster_id": "syn_03015_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "tokenizer_hashes": {"receiver": "b9d8872cf3a99da99e1fc4b0f71232a451632dc49ca4c6771549b4efa5c1c763"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 2, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1741, "query_tokens": 534, "message_tokens": 421, "message_utf8_bytes": 1123, "cache_tensor_bytes": null, "serialized_payload_bytes": 1123, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 39433.85270796716, "hops": {"1": 22925.759125035256, "2": 16508.093582931906}, "receiver_prefill_ms": 334.49529204517603, "ttft_ms": 37.50333306379616, "decode_ms": 148.55995797552168}, "total_wall_ms": 39954.98870802112, "ttft_ms": 37.50333306379616, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 5}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "8,421,725", "evidence_artifact": {"ledger": {"facts_total": 61, "facts_retained": 24, "retention_rate": 0.39344262295081966, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.4166666666666667, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1123}, "ledger_by_hop": {"1": {"facts_total": 61, "facts_retained": 24, "retention_rate": 0.39344262295081966, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.4166666666666667, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1161}, "2": {"facts_total": 61, "facts_retained": 24, "retention_rate": 0.39344262295081966, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.4166666666666667, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1123}}}, "reference_key_id": "keys/synthetic/syn_03015_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "dd270f0df70dd6fa49ad5b20f7ef7fabbbd5bdc8e942a2b6f2a26e351c41b754", "output_hash": "6e31df6d4fb38417a5210ded39fc1e4c0e32f278ab9cba0da5b8e7585ddeb534"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "b4023d273713f93f27b7bb4a", "attempt_id": "31f7951e3940", "run_group_id": "E4_core:A4|closed|qwen3b-llama3b:h2", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:00:19.662971Z", "task_family": "synthetic", "document_id": "syn_03015_2000", "document_hash": "71a7284b9bc3f681d69d91d18b9e8d45a4dda486cd14f4145395786ac6a68a89", "question_id": "q05_multi_hop", "split": "test", "cluster_id": "syn_03015_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "tokenizer_hashes": {"receiver": "b9d8872cf3a99da99e1fc4b0f71232a451632dc49ca4c6771549b4efa5c1c763"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 2, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1741, "query_tokens": 532, "message_tokens": 421, "message_utf8_bytes": 1123, "cache_tensor_bytes": null, "serialized_payload_bytes": 1123, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 39433.85270796716, "hops": {"1": 22925.759125035256, "2": 16508.093582931906}, "receiver_prefill_ms": 336.5175840444863, "ttft_ms": 33.25666591990739, "decode_ms": 28.563124942593277}, "total_wall_ms": 39832.68891694024, "ttft_ms": 33.25666591990739, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 61, "facts_retained": 24, "retention_rate": 0.39344262295081966, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.4166666666666667, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1123}, "ledger_by_hop": {"1": {"facts_total": 61, "facts_retained": 24, "retention_rate": 0.39344262295081966, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.4166666666666667, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1161}, "2": {"facts_total": 61, "facts_retained": 24, "retention_rate": 0.39344262295081966, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.4166666666666667, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1123}}}, "reference_key_id": "keys/synthetic/syn_03015_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "dd270f0df70dd6fa49ad5b20f7ef7fabbbd5bdc8e942a2b6f2a26e351c41b754", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "7dcec325505d56d0baccbce1", "attempt_id": "51eaae32a0c4", "run_group_id": "E4_core:A4|closed|qwen3b-llama3b:h2", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:00:20.078575Z", "task_family": "synthetic", "document_id": "syn_03015_2000", "document_hash": "71a7284b9bc3f681d69d91d18b9e8d45a4dda486cd14f4145395786ac6a68a89", "question_id": "q06_negation", "split": "test", "cluster_id": "syn_03015_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "tokenizer_hashes": {"receiver": "b9d8872cf3a99da99e1fc4b0f71232a451632dc49ca4c6771549b4efa5c1c763"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 2, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1741, "query_tokens": 526, "message_tokens": 421, "message_utf8_bytes": 1123, "cache_tensor_bytes": null, "serialized_payload_bytes": 1123, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 39433.85270796716, "hops": {"1": 22925.759125035256, "2": 16508.093582931906}, "receiver_prefill_ms": 314.35066601261497, "ttft_ms": 53.87112498283386, "decode_ms": 69.10729198716581}, "total_wall_ms": 39872.19199992251, "ttft_ms": 53.87112498283386, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 2}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "No.", "evidence_artifact": {"ledger": {"facts_total": 61, "facts_retained": 24, "retention_rate": 0.39344262295081966, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.4166666666666667, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1123}, "ledger_by_hop": {"1": {"facts_total": 61, "facts_retained": 24, "retention_rate": 0.39344262295081966, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.4166666666666667, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1161}, "2": {"facts_total": 61, "facts_retained": 24, "retention_rate": 0.39344262295081966, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.4166666666666667, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1123}}}, "reference_key_id": "keys/synthetic/syn_03015_2000.json", "task_scores": {"correct": 0.0, "abstained": 0.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "dd270f0df70dd6fa49ad5b20f7ef7fabbbd5bdc8e942a2b6f2a26e351c41b754", "output_hash": "38178a20b470cfd18299fb1593dd4ba706f1a83321dac051d01086efd8b7a96f"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "72b5d81ba9344692d9df5ef7", "attempt_id": "fa5ffb1988f3", "run_group_id": "E4_core:A4|closed|qwen3b-llama3b:h2", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:00:20.535384Z", "task_family": "synthetic", "document_id": "syn_03015_2000", "document_hash": "71a7284b9bc3f681d69d91d18b9e8d45a4dda486cd14f4145395786ac6a68a89", "question_id": "q07_negation", "split": "test", "cluster_id": "syn_03015_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "tokenizer_hashes": {"receiver": "b9d8872cf3a99da99e1fc4b0f71232a451632dc49ca4c6771549b4efa5c1c763"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 2, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1741, "query_tokens": 526, "message_tokens": 421, "message_utf8_bytes": 1123, "cache_tensor_bytes": null, "serialized_payload_bytes": 1123, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 39433.85270796716, "hops": {"1": 22925.759125035256, "2": 16508.093582931906}, "receiver_prefill_ms": 478.6290409974754, "ttft_ms": 24.334791931323707, "decode_ms": 62.193207908421755}, "total_wall_ms": 39999.72612399142, "ttft_ms": 24.334791931323707, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 2}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "No.", "evidence_artifact": {"ledger": {"facts_total": 61, "facts_retained": 24, "retention_rate": 0.39344262295081966, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.4166666666666667, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1123}, "ledger_by_hop": {"1": {"facts_total": 61, "facts_retained": 24, "retention_rate": 0.39344262295081966, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.4166666666666667, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1161}, "2": {"facts_total": 61, "facts_retained": 24, "retention_rate": 0.39344262295081966, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.4166666666666667, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1123}}}, "reference_key_id": "keys/synthetic/syn_03015_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "dd270f0df70dd6fa49ad5b20f7ef7fabbbd5bdc8e942a2b6f2a26e351c41b754", "output_hash": "38178a20b470cfd18299fb1593dd4ba706f1a83321dac051d01086efd8b7a96f"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "984d04755bb2340be17e319d", "attempt_id": "137263e2935e", "run_group_id": "E4_core:A4|closed|qwen3b-llama3b:h2", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:00:21.119649Z", "task_family": "synthetic", "document_id": "syn_03015_2000", "document_hash": "71a7284b9bc3f681d69d91d18b9e8d45a4dda486cd14f4145395786ac6a68a89", "question_id": "q08_revision", "split": "test", "cluster_id": "syn_03015_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "tokenizer_hashes": {"receiver": "b9d8872cf3a99da99e1fc4b0f71232a451632dc49ca4c6771549b4efa5c1c763"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 2, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1741, "query_tokens": 539, "message_tokens": 421, "message_utf8_bytes": 1123, "cache_tensor_bytes": null, "serialized_payload_bytes": 1123, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 39433.85270796716, "hops": {"1": 22925.759125035256, "2": 16508.093582931906}, "receiver_prefill_ms": 344.3605829961598, "ttft_ms": 31.533083063550293, "decode_ms": 21.55879104975611}, "total_wall_ms": 39831.95595792495, "ttft_ms": 31.533083063550293, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 61, "facts_retained": 24, "retention_rate": 0.39344262295081966, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.4166666666666667, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1123}, "ledger_by_hop": {"1": {"facts_total": 61, "facts_retained": 24, "retention_rate": 0.39344262295081966, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.4166666666666667, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1161}, "2": {"facts_total": 61, "facts_retained": 24, "retention_rate": 0.39344262295081966, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.4166666666666667, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1123}}}, "reference_key_id": "keys/synthetic/syn_03015_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "dd270f0df70dd6fa49ad5b20f7ef7fabbbd5bdc8e942a2b6f2a26e351c41b754", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "e152f8fe259afbbe350f3fed", "attempt_id": "1a2cc16a61ab", "run_group_id": "E4_core:A4|closed|qwen3b-llama3b:h2", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:00:21.534825Z", "task_family": "synthetic", "document_id": "syn_03015_2000", "document_hash": "71a7284b9bc3f681d69d91d18b9e8d45a4dda486cd14f4145395786ac6a68a89", "question_id": "q09_revision", "split": "test", "cluster_id": "syn_03015_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "tokenizer_hashes": {"receiver": "b9d8872cf3a99da99e1fc4b0f71232a451632dc49ca4c6771549b4efa5c1c763"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 2, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1741, "query_tokens": 533, "message_tokens": 421, "message_utf8_bytes": 1123, "cache_tensor_bytes": null, "serialized_payload_bytes": 1123, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 39433.85270796716, "hops": {"1": 22925.759125035256, "2": 16508.093582931906}, "receiver_prefill_ms": 328.4824589500204, "ttft_ms": 18.808584078215063, "decode_ms": 121.79633404593915}, "total_wall_ms": 39903.7769169081, "ttft_ms": 18.808584078215063, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 5}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "7,028,925", "evidence_artifact": {"ledger": {"facts_total": 61, "facts_retained": 24, "retention_rate": 0.39344262295081966, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.4166666666666667, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1123}, "ledger_by_hop": {"1": {"facts_total": 61, "facts_retained": 24, "retention_rate": 0.39344262295081966, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.4166666666666667, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1161}, "2": {"facts_total": 61, "facts_retained": 24, "retention_rate": 0.39344262295081966, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.4166666666666667, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1123}}}, "reference_key_id": "keys/synthetic/syn_03015_2000.json", "task_scores": {"correct": 0.0, "abstained": 0.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "dd270f0df70dd6fa49ad5b20f7ef7fabbbd5bdc8e942a2b6f2a26e351c41b754", "output_hash": "e83edb3f38a69e604be343e63e2e84442156fcb2bb30f36ba9716575d5ffd57a"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "480cc6af0b794514b003f911", "attempt_id": "6d755512c8d9", "run_group_id": "E4_core:A4|closed|qwen3b-llama3b:h2", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:00:22.022567Z", "task_family": "synthetic", "document_id": "syn_03015_2000", "document_hash": "71a7284b9bc3f681d69d91d18b9e8d45a4dda486cd14f4145395786ac6a68a89", "question_id": "q10_temporal", "split": "test", "cluster_id": "syn_03015_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "tokenizer_hashes": {"receiver": "b9d8872cf3a99da99e1fc4b0f71232a451632dc49ca4c6771549b4efa5c1c763"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 2, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1741, "query_tokens": 542, "message_tokens": 421, "message_utf8_bytes": 1123, "cache_tensor_bytes": null, "serialized_payload_bytes": 1123, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 39433.85270796716, "hops": {"1": 22925.759125035256, "2": 16508.093582931906}, "receiver_prefill_ms": 321.36204198468477, "ttft_ms": 34.565959009341896, "decode_ms": 107.57479106541723}, "total_wall_ms": 39898.1114999624, "ttft_ms": 34.565959009341896, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 61, "facts_retained": 24, "retention_rate": 0.39344262295081966, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.4166666666666667, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1123}, "ledger_by_hop": {"1": {"facts_total": 61, "facts_retained": 24, "retention_rate": 0.39344262295081966, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.4166666666666667, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1161}, "2": {"facts_total": 61, "facts_retained": 24, "retention_rate": 0.39344262295081966, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.4166666666666667, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1123}}}, "reference_key_id": "keys/synthetic/syn_03015_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "dd270f0df70dd6fa49ad5b20f7ef7fabbbd5bdc8e942a2b6f2a26e351c41b754", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "ca11b02106df61d05063b8ab", "attempt_id": "a2d6820cb688", "run_group_id": "E4_core:A4|closed|qwen3b-llama3b:h2", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:00:22.504442Z", "task_family": "synthetic", "document_id": "syn_03015_2000", "document_hash": "71a7284b9bc3f681d69d91d18b9e8d45a4dda486cd14f4145395786ac6a68a89", "question_id": "q11_aggregation", "split": "test", "cluster_id": "syn_03015_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "tokenizer_hashes": {"receiver": "b9d8872cf3a99da99e1fc4b0f71232a451632dc49ca4c6771549b4efa5c1c763"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 2, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1741, "query_tokens": 536, "message_tokens": 421, "message_utf8_bytes": 1123, "cache_tensor_bytes": null, "serialized_payload_bytes": 1123, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 39433.85270796716, "hops": {"1": 22925.759125035256, "2": 16508.093582931906}, "receiver_prefill_ms": 421.9737920211628, "ttft_ms": 38.057208992540836, "decode_ms": 145.2821659622714}, "total_wall_ms": 40039.847790962085, "ttft_ms": 38.057208992540836, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 5}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "1,806,025", "evidence_artifact": {"ledger": {"facts_total": 61, "facts_retained": 24, "retention_rate": 0.39344262295081966, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.4166666666666667, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1123}, "ledger_by_hop": {"1": {"facts_total": 61, "facts_retained": 24, "retention_rate": 0.39344262295081966, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.4166666666666667, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1161}, "2": {"facts_total": 61, "facts_retained": 24, "retention_rate": 0.39344262295081966, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.4166666666666667, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1123}}}, "reference_key_id": "keys/synthetic/syn_03015_2000.json", "task_scores": {"correct": 0.0, "abstained": 0.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "dd270f0df70dd6fa49ad5b20f7ef7fabbbd5bdc8e942a2b6f2a26e351c41b754", "output_hash": "4de10b23990757532ff215b6a27f5f805aa6f751cc9757418279cc7ab85aa1a9"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "42c01b6f895ec2045d0e504f", "attempt_id": "2b382eaa82a1", "run_group_id": "E4_core:A4|closed|qwen3b-llama3b:h2", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:00:23.130609Z", "task_family": "synthetic", "document_id": "syn_03015_2000", "document_hash": "71a7284b9bc3f681d69d91d18b9e8d45a4dda486cd14f4145395786ac6a68a89", "question_id": "q12_unit_conversion", "split": "test", "cluster_id": "syn_03015_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "tokenizer_hashes": {"receiver": "b9d8872cf3a99da99e1fc4b0f71232a451632dc49ca4c6771549b4efa5c1c763"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 2, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1741, "query_tokens": 533, "message_tokens": 421, "message_utf8_bytes": 1123, "cache_tensor_bytes": null, "serialized_payload_bytes": 1123, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 39433.85270796716, "hops": {"1": 22925.759125035256, "2": 16508.093582931906}, "receiver_prefill_ms": 337.01299992389977, "ttft_ms": 34.74324999842793, "decode_ms": 25.79654101282358}, "total_wall_ms": 39832.054082886316, "ttft_ms": 34.74324999842793, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 61, "facts_retained": 24, "retention_rate": 0.39344262295081966, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.4166666666666667, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1123}, "ledger_by_hop": {"1": {"facts_total": 61, "facts_retained": 24, "retention_rate": 0.39344262295081966, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.4166666666666667, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1161}, "2": {"facts_total": 61, "facts_retained": 24, "retention_rate": 0.39344262295081966, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.4166666666666667, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1123}}}, "reference_key_id": "keys/synthetic/syn_03015_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "dd270f0df70dd6fa49ad5b20f7ef7fabbbd5bdc8e942a2b6f2a26e351c41b754", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "16e1740d174e4e012570116f", "attempt_id": "f77fc6bf9c6a", "run_group_id": "E4_core:A4|closed|qwen3b-llama3b:h2", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:00:23.547179Z", "task_family": "synthetic", "document_id": "syn_03015_2000", "document_hash": "71a7284b9bc3f681d69d91d18b9e8d45a4dda486cd14f4145395786ac6a68a89", "question_id": "q13_unanswerable", "split": "test", "cluster_id": "syn_03015_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "tokenizer_hashes": {"receiver": "b9d8872cf3a99da99e1fc4b0f71232a451632dc49ca4c6771549b4efa5c1c763"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 2, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1741, "query_tokens": 536, "message_tokens": 421, "message_utf8_bytes": 1123, "cache_tensor_bytes": null, "serialized_payload_bytes": 1123, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 39433.85270796716, "hops": {"1": 22925.759125035256, "2": 16508.093582931906}, "receiver_prefill_ms": 318.1566250277683, "ttft_ms": 47.39412502385676, "decode_ms": 107.35724994447082}, "total_wall_ms": 39907.463373965584, "ttft_ms": 47.39412502385676, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 2}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown.", "evidence_artifact": {"ledger": {"facts_total": 61, "facts_retained": 24, "retention_rate": 0.39344262295081966, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.4166666666666667, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1123}, "ledger_by_hop": {"1": {"facts_total": 61, "facts_retained": 24, "retention_rate": 0.39344262295081966, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.4166666666666667, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1161}, "2": {"facts_total": 61, "facts_retained": 24, "retention_rate": 0.39344262295081966, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.4166666666666667, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1123}}}, "reference_key_id": "keys/synthetic/syn_03015_2000.json", "task_scores": {"correct": 1.0, "abstained": 1.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "dd270f0df70dd6fa49ad5b20f7ef7fabbbd5bdc8e942a2b6f2a26e351c41b754", "output_hash": "4dd633b1f2ee444666fd70af2b36443331fc00e69f3e569847cf278ca625c041"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "b6cc424c5da78e928ae2c194", "attempt_id": "c68d60732fb5", "run_group_id": "E4_core:A4|closed|qwen3b-llama3b:h2", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:00:24.039155Z", "task_family": "synthetic", "document_id": "syn_03015_2000", "document_hash": "71a7284b9bc3f681d69d91d18b9e8d45a4dda486cd14f4145395786ac6a68a89", "question_id": "q14_unanswerable", "split": "test", "cluster_id": "syn_03015_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "tokenizer_hashes": {"receiver": "b9d8872cf3a99da99e1fc4b0f71232a451632dc49ca4c6771549b4efa5c1c763"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 2, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1741, "query_tokens": 538, "message_tokens": 421, "message_utf8_bytes": 1123, "cache_tensor_bytes": null, "serialized_payload_bytes": 1123, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 39433.85270796716, "hops": {"1": 22925.759125035256, "2": 16508.093582931906}, "receiver_prefill_ms": 397.01137493830174, "ttft_ms": 74.6606671018526, "decode_ms": 56.05695804115385}, "total_wall_ms": 39962.18637400307, "ttft_ms": 74.6606671018526, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 61, "facts_retained": 24, "retention_rate": 0.39344262295081966, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.4166666666666667, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1123}, "ledger_by_hop": {"1": {"facts_total": 61, "facts_retained": 24, "retention_rate": 0.39344262295081966, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.4166666666666667, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1161}, "2": {"facts_total": 61, "facts_retained": 24, "retention_rate": 0.39344262295081966, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.4166666666666667, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1123}}}, "reference_key_id": "keys/synthetic/syn_03015_2000.json", "task_scores": {"correct": 1.0, "abstained": 1.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "dd270f0df70dd6fa49ad5b20f7ef7fabbbd5bdc8e942a2b6f2a26e351c41b754", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "82781cc957a112541aa12734", "attempt_id": "b3a7f9c053d7", "run_group_id": "E4_core:A4|closed|qwen3b-llama3b:h4", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:01:04.776401Z", "task_family": "synthetic", "document_id": "syn_03015_2000", "document_hash": "71a7284b9bc3f681d69d91d18b9e8d45a4dda486cd14f4145395786ac6a68a89", "question_id": "q00_single_fact", "split": "test", "cluster_id": "syn_03015_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "tokenizer_hashes": {"receiver": "b9d8872cf3a99da99e1fc4b0f71232a451632dc49ca4c6771549b4efa5c1c763"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 4, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1741, "query_tokens": 523, "message_tokens": 419, "message_utf8_bytes": 1119, "cache_tensor_bytes": null, "serialized_payload_bytes": 1119, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 79624.5637909742, "hops": {"1": 22925.759125035256, "2": 16508.093582931906, "3": 23855.072916019708, "4": 16335.638166987337}, "receiver_prefill_ms": 302.3384170373902, "ttft_ms": 53.821290959604084, "decode_ms": 62.60649999603629}, "total_wall_ms": 80044.14941603318, "ttft_ms": 53.821290959604084, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 61, "facts_retained": 24, "retention_rate": 0.39344262295081966, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.4166666666666667, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1119}, "ledger_by_hop": {"1": {"facts_total": 61, "facts_retained": 24, "retention_rate": 0.39344262295081966, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.4166666666666667, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1161}, "2": {"facts_total": 61, "facts_retained": 24, "retention_rate": 0.39344262295081966, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.4166666666666667, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1123}, "3": {"facts_total": 61, "facts_retained": 24, "retention_rate": 0.39344262295081966, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.4166666666666667, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1123}, "4": {"facts_total": 61, "facts_retained": 24, "retention_rate": 0.39344262295081966, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.4166666666666667, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1119}}}, "reference_key_id": "keys/synthetic/syn_03015_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "9cff6b905fbff4d9c34852ce07efb835c3b85e73c292b038a4a7a512d29a8b69", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "91414f0ece6a6f1efe99ef42", "attempt_id": "97ad4c520145", "run_group_id": "E4_core:A4|closed|qwen3b-llama3b:h4", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:01:05.213939Z", "task_family": "synthetic", "document_id": "syn_03015_2000", "document_hash": "71a7284b9bc3f681d69d91d18b9e8d45a4dda486cd14f4145395786ac6a68a89", "question_id": "q01_single_fact", "split": "test", "cluster_id": "syn_03015_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "tokenizer_hashes": {"receiver": "b9d8872cf3a99da99e1fc4b0f71232a451632dc49ca4c6771549b4efa5c1c763"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 4, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1741, "query_tokens": 527, "message_tokens": 419, "message_utf8_bytes": 1119, "cache_tensor_bytes": null, "serialized_payload_bytes": 1119, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 79624.5637909742, "hops": {"1": 22925.759125035256, "2": 16508.093582931906, "3": 23855.072916019708, "4": 16335.638166987337}, "receiver_prefill_ms": 351.53504193294793, "ttft_ms": 45.30145798344165, "decode_ms": 124.60824998561293}, "total_wall_ms": 80146.64812502451, "ttft_ms": 45.30145798344165, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 61, "facts_retained": 24, "retention_rate": 0.39344262295081966, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.4166666666666667, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1119}, "ledger_by_hop": {"1": {"facts_total": 61, "facts_retained": 24, "retention_rate": 0.39344262295081966, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.4166666666666667, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1161}, "2": {"facts_total": 61, "facts_retained": 24, "retention_rate": 0.39344262295081966, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.4166666666666667, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1123}, "3": {"facts_total": 61, "facts_retained": 24, "retention_rate": 0.39344262295081966, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.4166666666666667, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1123}, "4": {"facts_total": 61, "facts_retained": 24, "retention_rate": 0.39344262295081966, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.4166666666666667, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1119}}}, "reference_key_id": "keys/synthetic/syn_03015_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "9cff6b905fbff4d9c34852ce07efb835c3b85e73c292b038a4a7a512d29a8b69", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "9f2c93dfca49d441db74ff94", "attempt_id": "fc501393ab7a", "run_group_id": "E4_core:A4|closed|qwen3b-llama3b:h4", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:01:05.753177Z", "task_family": "synthetic", "document_id": "syn_03015_2000", "document_hash": "71a7284b9bc3f681d69d91d18b9e8d45a4dda486cd14f4145395786ac6a68a89", "question_id": "q02_exact_copy", "split": "test", "cluster_id": "syn_03015_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "tokenizer_hashes": {"receiver": "b9d8872cf3a99da99e1fc4b0f71232a451632dc49ca4c6771549b4efa5c1c763"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 4, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1741, "query_tokens": 527, "message_tokens": 419, "message_utf8_bytes": 1119, "cache_tensor_bytes": null, "serialized_payload_bytes": 1119, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 79624.5637909742, "hops": {"1": 22925.759125035256, "2": 16508.093582931906, "3": 23855.072916019708, "4": 16335.638166987337}, "receiver_prefill_ms": 332.5755000114441, "ttft_ms": 38.30254194326699, "decode_ms": 126.11641699913889}, "total_wall_ms": 80122.19341599848, "ttft_ms": 38.30254194326699, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 4}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "89.3122", "evidence_artifact": {"ledger": {"facts_total": 61, "facts_retained": 24, "retention_rate": 0.39344262295081966, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.4166666666666667, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1119}, "ledger_by_hop": {"1": {"facts_total": 61, "facts_retained": 24, "retention_rate": 0.39344262295081966, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.4166666666666667, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1161}, "2": {"facts_total": 61, "facts_retained": 24, "retention_rate": 0.39344262295081966, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.4166666666666667, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1123}, "3": {"facts_total": 61, "facts_retained": 24, "retention_rate": 0.39344262295081966, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.4166666666666667, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1123}, "4": {"facts_total": 61, "facts_retained": 24, "retention_rate": 0.39344262295081966, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.4166666666666667, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1119}}}, "reference_key_id": "keys/synthetic/syn_03015_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "9cff6b905fbff4d9c34852ce07efb835c3b85e73c292b038a4a7a512d29a8b69", "output_hash": "2138888e6519b2f4b1b54142e6f9c541102d7f9ee5bcf8e4184c81a382601757"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "4b7318a1cb5f6c0c7350e4c2", "attempt_id": "d3d365c52bcb", "run_group_id": "E4_core:A4|closed|qwen3b-llama3b:h4", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:01:06.267548Z", "task_family": "synthetic", "document_id": "syn_03015_2000", "document_hash": "71a7284b9bc3f681d69d91d18b9e8d45a4dda486cd14f4145395786ac6a68a89", "question_id": "q03_exact_copy", "split": "test", "cluster_id": "syn_03015_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "tokenizer_hashes": {"receiver": "b9d8872cf3a99da99e1fc4b0f71232a451632dc49ca4c6771549b4efa5c1c763"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 4, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1741, "query_tokens": 523, "message_tokens": 419, "message_utf8_bytes": 1119, "cache_tensor_bytes": null, "serialized_payload_bytes": 1119, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 79624.5637909742, "hops": {"1": 22925.759125035256, "2": 16508.093582931906, "3": 23855.072916019708, "4": 16335.638166987337}, "receiver_prefill_ms": 350.07141600362957, "ttft_ms": 24.525374989025295, "decode_ms": 224.46133301127702}, "total_wall_ms": 80224.50854105409, "ttft_ms": 24.525374989025295, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 7}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "Wilhelmina Dubrovnik.", "evidence_artifact": {"ledger": {"facts_total": 61, "facts_retained": 24, "retention_rate": 0.39344262295081966, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.4166666666666667, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1119}, "ledger_by_hop": {"1": {"facts_total": 61, "facts_retained": 24, "retention_rate": 0.39344262295081966, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.4166666666666667, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1161}, "2": {"facts_total": 61, "facts_retained": 24, "retention_rate": 0.39344262295081966, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.4166666666666667, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1123}, "3": {"facts_total": 61, "facts_retained": 24, "retention_rate": 0.39344262295081966, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.4166666666666667, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1123}, "4": {"facts_total": 61, "facts_retained": 24, "retention_rate": 0.39344262295081966, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.4166666666666667, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1119}}}, "reference_key_id": "keys/synthetic/syn_03015_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "9cff6b905fbff4d9c34852ce07efb835c3b85e73c292b038a4a7a512d29a8b69", "output_hash": "79ba0181ee6c8e2b01e096d290463d1e2eb3a595aa1abc3242a857276f5ca877"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "046e43743fec9f7faed452b3", "attempt_id": "727d4bbfb64d", "run_group_id": "E4_core:A4|closed|qwen3b-llama3b:h4", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:01:06.885392Z", "task_family": "synthetic", "document_id": "syn_03015_2000", "document_hash": "71a7284b9bc3f681d69d91d18b9e8d45a4dda486cd14f4145395786ac6a68a89", "question_id": "q04_multi_hop", "split": "test", "cluster_id": "syn_03015_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "tokenizer_hashes": {"receiver": "b9d8872cf3a99da99e1fc4b0f71232a451632dc49ca4c6771549b4efa5c1c763"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 4, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1741, "query_tokens": 532, "message_tokens": 419, "message_utf8_bytes": 1119, "cache_tensor_bytes": null, "serialized_payload_bytes": 1119, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 79624.5637909742, "hops": {"1": 22925.759125035256, "2": 16508.093582931906, "3": 23855.072916019708, "4": 16335.638166987337}, "receiver_prefill_ms": 399.6447080280632, "ttft_ms": 144.36420798301697, "decode_ms": 160.6786249903962}, "total_wall_ms": 80330.03383292817, "ttft_ms": 144.36420798301697, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 5}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "8,421,725", "evidence_artifact": {"ledger": {"facts_total": 61, "facts_retained": 24, "retention_rate": 0.39344262295081966, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.4166666666666667, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1119}, "ledger_by_hop": {"1": {"facts_total": 61, "facts_retained": 24, "retention_rate": 0.39344262295081966, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.4166666666666667, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1161}, "2": {"facts_total": 61, "facts_retained": 24, "retention_rate": 0.39344262295081966, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.4166666666666667, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1123}, "3": {"facts_total": 61, "facts_retained": 24, "retention_rate": 0.39344262295081966, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.4166666666666667, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1123}, "4": {"facts_total": 61, "facts_retained": 24, "retention_rate": 0.39344262295081966, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.4166666666666667, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1119}}}, "reference_key_id": "keys/synthetic/syn_03015_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "9cff6b905fbff4d9c34852ce07efb835c3b85e73c292b038a4a7a512d29a8b69", "output_hash": "6e31df6d4fb38417a5210ded39fc1e4c0e32f278ab9cba0da5b8e7585ddeb534"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "3972c16437c36211eeb0f29a", "attempt_id": "23c794aa86cd", "run_group_id": "E4_core:A4|closed|qwen3b-llama3b:h4", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:01:07.608118Z", "task_family": "synthetic", "document_id": "syn_03015_2000", "document_hash": "71a7284b9bc3f681d69d91d18b9e8d45a4dda486cd14f4145395786ac6a68a89", "question_id": "q05_multi_hop", "split": "test", "cluster_id": "syn_03015_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "tokenizer_hashes": {"receiver": "b9d8872cf3a99da99e1fc4b0f71232a451632dc49ca4c6771549b4efa5c1c763"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 4, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1741, "query_tokens": 530, "message_tokens": 419, "message_utf8_bytes": 1119, "cache_tensor_bytes": null, "serialized_payload_bytes": 1119, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 79624.5637909742, "hops": {"1": 22925.759125035256, "2": 16508.093582931906, "3": 23855.072916019708, "4": 16335.638166987337}, "receiver_prefill_ms": 341.015417012386, "ttft_ms": 25.479875039309263, "decode_ms": 38.668125052936375}, "total_wall_ms": 80030.58949997649, "ttft_ms": 25.479875039309263, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 61, "facts_retained": 24, "retention_rate": 0.39344262295081966, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.4166666666666667, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1119}, "ledger_by_hop": {"1": {"facts_total": 61, "facts_retained": 24, "retention_rate": 0.39344262295081966, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.4166666666666667, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1161}, "2": {"facts_total": 61, "facts_retained": 24, "retention_rate": 0.39344262295081966, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.4166666666666667, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1123}, "3": {"facts_total": 61, "facts_retained": 24, "retention_rate": 0.39344262295081966, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.4166666666666667, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1123}, "4": {"facts_total": 61, "facts_retained": 24, "retention_rate": 0.39344262295081966, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.4166666666666667, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1119}}}, "reference_key_id": "keys/synthetic/syn_03015_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "9cff6b905fbff4d9c34852ce07efb835c3b85e73c292b038a4a7a512d29a8b69", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "a6c50c41449939a30d3bd8b4", "attempt_id": "f731fccd6a75", "run_group_id": "E4_core:A4|closed|qwen3b-llama3b:h4", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:01:08.033135Z", "task_family": "synthetic", "document_id": "syn_03015_2000", "document_hash": "71a7284b9bc3f681d69d91d18b9e8d45a4dda486cd14f4145395786ac6a68a89", "question_id": "q06_negation", "split": "test", "cluster_id": "syn_03015_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "tokenizer_hashes": {"receiver": "b9d8872cf3a99da99e1fc4b0f71232a451632dc49ca4c6771549b4efa5c1c763"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 4, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1741, "query_tokens": 524, "message_tokens": 419, "message_utf8_bytes": 1119, "cache_tensor_bytes": null, "serialized_payload_bytes": 1119, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 79624.5637909742, "hops": {"1": 22925.759125035256, "2": 16508.093582931906, "3": 23855.072916019708, "4": 16335.638166987337}, "receiver_prefill_ms": 343.7255419557914, "ttft_ms": 22.180584026500583, "decode_ms": 40.472625056281686}, "total_wall_ms": 80031.81399893947, "ttft_ms": 22.180584026500583, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 2}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "No.", "evidence_artifact": {"ledger": {"facts_total": 61, "facts_retained": 24, "retention_rate": 0.39344262295081966, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.4166666666666667, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1119}, "ledger_by_hop": {"1": {"facts_total": 61, "facts_retained": 24, "retention_rate": 0.39344262295081966, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.4166666666666667, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1161}, "2": {"facts_total": 61, "facts_retained": 24, "retention_rate": 0.39344262295081966, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.4166666666666667, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1123}, "3": {"facts_total": 61, "facts_retained": 24, "retention_rate": 0.39344262295081966, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.4166666666666667, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1123}, "4": {"facts_total": 61, "facts_retained": 24, "retention_rate": 0.39344262295081966, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.4166666666666667, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1119}}}, "reference_key_id": "keys/synthetic/syn_03015_2000.json", "task_scores": {"correct": 0.0, "abstained": 0.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "9cff6b905fbff4d9c34852ce07efb835c3b85e73c292b038a4a7a512d29a8b69", "output_hash": "38178a20b470cfd18299fb1593dd4ba706f1a83321dac051d01086efd8b7a96f"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "ef12e0c567fa64d6cd9bf7b0", "attempt_id": "e1136618a3da", "run_group_id": "E4_core:A4|closed|qwen3b-llama3b:h4", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:01:08.460557Z", "task_family": "synthetic", "document_id": "syn_03015_2000", "document_hash": "71a7284b9bc3f681d69d91d18b9e8d45a4dda486cd14f4145395786ac6a68a89", "question_id": "q07_negation", "split": "test", "cluster_id": "syn_03015_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "tokenizer_hashes": {"receiver": "b9d8872cf3a99da99e1fc4b0f71232a451632dc49ca4c6771549b4efa5c1c763"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 4, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1741, "query_tokens": 524, "message_tokens": 419, "message_utf8_bytes": 1119, "cache_tensor_bytes": null, "serialized_payload_bytes": 1119, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 79624.5637909742, "hops": {"1": 22925.759125035256, "2": 16508.093582931906, "3": 23855.072916019708, "4": 16335.638166987337}, "receiver_prefill_ms": 318.9456668915227, "ttft_ms": 52.28604201693088, "decode_ms": 158.91162492334843}, "total_wall_ms": 80155.45345807914, "ttft_ms": 52.28604201693088, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 2}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "No.", "evidence_artifact": {"ledger": {"facts_total": 61, "facts_retained": 24, "retention_rate": 0.39344262295081966, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.4166666666666667, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1119}, "ledger_by_hop": {"1": {"facts_total": 61, "facts_retained": 24, "retention_rate": 0.39344262295081966, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.4166666666666667, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1161}, "2": {"facts_total": 61, "facts_retained": 24, "retention_rate": 0.39344262295081966, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.4166666666666667, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1123}, "3": {"facts_total": 61, "facts_retained": 24, "retention_rate": 0.39344262295081966, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.4166666666666667, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1123}, "4": {"facts_total": 61, "facts_retained": 24, "retention_rate": 0.39344262295081966, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.4166666666666667, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1119}}}, "reference_key_id": "keys/synthetic/syn_03015_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "9cff6b905fbff4d9c34852ce07efb835c3b85e73c292b038a4a7a512d29a8b69", "output_hash": "38178a20b470cfd18299fb1593dd4ba706f1a83321dac051d01086efd8b7a96f"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "1f59a8b358000011964f07bf", "attempt_id": "57c27a6276b2", "run_group_id": "E4_core:A4|closed|qwen3b-llama3b:h4", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:01:09.009788Z", "task_family": "synthetic", "document_id": "syn_03015_2000", "document_hash": "71a7284b9bc3f681d69d91d18b9e8d45a4dda486cd14f4145395786ac6a68a89", "question_id": "q08_revision", "split": "test", "cluster_id": "syn_03015_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "tokenizer_hashes": {"receiver": "b9d8872cf3a99da99e1fc4b0f71232a451632dc49ca4c6771549b4efa5c1c763"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 4, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1741, "query_tokens": 537, "message_tokens": 419, "message_utf8_bytes": 1119, "cache_tensor_bytes": null, "serialized_payload_bytes": 1119, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 79624.5637909742, "hops": {"1": 22925.759125035256, "2": 16508.093582931906, "3": 23855.072916019708, "4": 16335.638166987337}, "receiver_prefill_ms": 416.748792049475, "ttft_ms": 34.16841593571007, "decode_ms": 24.706083931960166}, "total_wall_ms": 80100.8045830531, "ttft_ms": 34.16841593571007, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 61, "facts_retained": 24, "retention_rate": 0.39344262295081966, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.4166666666666667, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1119}, "ledger_by_hop": {"1": {"facts_total": 61, "facts_retained": 24, "retention_rate": 0.39344262295081966, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.4166666666666667, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1161}, "2": {"facts_total": 61, "facts_retained": 24, "retention_rate": 0.39344262295081966, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.4166666666666667, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1123}, "3": {"facts_total": 61, "facts_retained": 24, "retention_rate": 0.39344262295081966, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.4166666666666667, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1123}, "4": {"facts_total": 61, "facts_retained": 24, "retention_rate": 0.39344262295081966, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.4166666666666667, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1119}}}, "reference_key_id": "keys/synthetic/syn_03015_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "9cff6b905fbff4d9c34852ce07efb835c3b85e73c292b038a4a7a512d29a8b69", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "0db858ad27a1526b6858927b", "attempt_id": "bd93ca0467c0", "run_group_id": "E4_core:A4|closed|qwen3b-llama3b:h4", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:01:09.505717Z", "task_family": "synthetic", "document_id": "syn_03015_2000", "document_hash": "71a7284b9bc3f681d69d91d18b9e8d45a4dda486cd14f4145395786ac6a68a89", "question_id": "q09_revision", "split": "test", "cluster_id": "syn_03015_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "tokenizer_hashes": {"receiver": "b9d8872cf3a99da99e1fc4b0f71232a451632dc49ca4c6771549b4efa5c1c763"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 4, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1741, "query_tokens": 531, "message_tokens": 419, "message_utf8_bytes": 1119, "cache_tensor_bytes": null, "serialized_payload_bytes": 1119, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 79624.5637909742, "hops": {"1": 22925.759125035256, "2": 16508.093582931906, "3": 23855.072916019708, "4": 16335.638166987337}, "receiver_prefill_ms": 337.47699996456504, "ttft_ms": 38.45487500075251, "decode_ms": 148.29924993682653}, "total_wall_ms": 80149.38949898351, "ttft_ms": 38.45487500075251, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 5}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "7,028,925", "evidence_artifact": {"ledger": {"facts_total": 61, "facts_retained": 24, "retention_rate": 0.39344262295081966, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.4166666666666667, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1119}, "ledger_by_hop": {"1": {"facts_total": 61, "facts_retained": 24, "retention_rate": 0.39344262295081966, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.4166666666666667, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1161}, "2": {"facts_total": 61, "facts_retained": 24, "retention_rate": 0.39344262295081966, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.4166666666666667, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1123}, "3": {"facts_total": 61, "facts_retained": 24, "retention_rate": 0.39344262295081966, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.4166666666666667, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1123}, "4": {"facts_total": 61, "facts_retained": 24, "retention_rate": 0.39344262295081966, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.4166666666666667, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1119}}}, "reference_key_id": "keys/synthetic/syn_03015_2000.json", "task_scores": {"correct": 0.0, "abstained": 0.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "9cff6b905fbff4d9c34852ce07efb835c3b85e73c292b038a4a7a512d29a8b69", "output_hash": "e83edb3f38a69e604be343e63e2e84442156fcb2bb30f36ba9716575d5ffd57a"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "b2d00f0ce1c530671a7c7c4d", "attempt_id": "fc1cc8c8814e", "run_group_id": "E4_core:A4|closed|qwen3b-llama3b:h4", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:01:10.047760Z", "task_family": "synthetic", "document_id": "syn_03015_2000", "document_hash": "71a7284b9bc3f681d69d91d18b9e8d45a4dda486cd14f4145395786ac6a68a89", "question_id": "q10_temporal", "split": "test", "cluster_id": "syn_03015_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "tokenizer_hashes": {"receiver": "b9d8872cf3a99da99e1fc4b0f71232a451632dc49ca4c6771549b4efa5c1c763"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 4, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1741, "query_tokens": 540, "message_tokens": 419, "message_utf8_bytes": 1119, "cache_tensor_bytes": null, "serialized_payload_bytes": 1119, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 79624.5637909742, "hops": {"1": 22925.759125035256, "2": 16508.093582931906, "3": 23855.072916019708, "4": 16335.638166987337}, "receiver_prefill_ms": 291.6915829991922, "ttft_ms": 46.69083305634558, "decode_ms": 42.56041697226465}, "total_wall_ms": 80006.36091595516, "ttft_ms": 46.69083305634558, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 61, "facts_retained": 24, "retention_rate": 0.39344262295081966, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.4166666666666667, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1119}, "ledger_by_hop": {"1": {"facts_total": 61, "facts_retained": 24, "retention_rate": 0.39344262295081966, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.4166666666666667, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1161}, "2": {"facts_total": 61, "facts_retained": 24, "retention_rate": 0.39344262295081966, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.4166666666666667, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1123}, "3": {"facts_total": 61, "facts_retained": 24, "retention_rate": 0.39344262295081966, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.4166666666666667, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1123}, "4": {"facts_total": 61, "facts_retained": 24, "retention_rate": 0.39344262295081966, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.4166666666666667, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1119}}}, "reference_key_id": "keys/synthetic/syn_03015_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "9cff6b905fbff4d9c34852ce07efb835c3b85e73c292b038a4a7a512d29a8b69", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "aee997898aac19b83dd8afac", "attempt_id": "d2d2c88cbb10", "run_group_id": "E4_core:A4|closed|qwen3b-llama3b:h4", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:01:10.447310Z", "task_family": "synthetic", "document_id": "syn_03015_2000", "document_hash": "71a7284b9bc3f681d69d91d18b9e8d45a4dda486cd14f4145395786ac6a68a89", "question_id": "q11_aggregation", "split": "test", "cluster_id": "syn_03015_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "tokenizer_hashes": {"receiver": "b9d8872cf3a99da99e1fc4b0f71232a451632dc49ca4c6771549b4efa5c1c763"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 4, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1741, "query_tokens": 534, "message_tokens": 419, "message_utf8_bytes": 1119, "cache_tensor_bytes": null, "serialized_payload_bytes": 1119, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 79624.5637909742, "hops": {"1": 22925.759125035256, "2": 16508.093582931906, "3": 23855.072916019708, "4": 16335.638166987337}, "receiver_prefill_ms": 412.71487495396286, "ttft_ms": 99.8573750257492, "decode_ms": 157.67320792656392}, "total_wall_ms": 80295.51587393507, "ttft_ms": 99.8573750257492, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 5}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "1,806,025", "evidence_artifact": {"ledger": {"facts_total": 61, "facts_retained": 24, "retention_rate": 0.39344262295081966, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.4166666666666667, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1119}, "ledger_by_hop": {"1": {"facts_total": 61, "facts_retained": 24, "retention_rate": 0.39344262295081966, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.4166666666666667, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1161}, "2": {"facts_total": 61, "facts_retained": 24, "retention_rate": 0.39344262295081966, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.4166666666666667, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1123}, "3": {"facts_total": 61, "facts_retained": 24, "retention_rate": 0.39344262295081966, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.4166666666666667, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1123}, "4": {"facts_total": 61, "facts_retained": 24, "retention_rate": 0.39344262295081966, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.4166666666666667, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1119}}}, "reference_key_id": "keys/synthetic/syn_03015_2000.json", "task_scores": {"correct": 0.0, "abstained": 0.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "9cff6b905fbff4d9c34852ce07efb835c3b85e73c292b038a4a7a512d29a8b69", "output_hash": "4de10b23990757532ff215b6a27f5f805aa6f751cc9757418279cc7ab85aa1a9"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "1efe7f4600483f9d87bc9423", "attempt_id": "cd90729c4d9b", "run_group_id": "E4_core:A4|closed|qwen3b-llama3b:h4", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:01:11.135415Z", "task_family": "synthetic", "document_id": "syn_03015_2000", "document_hash": "71a7284b9bc3f681d69d91d18b9e8d45a4dda486cd14f4145395786ac6a68a89", "question_id": "q12_unit_conversion", "split": "test", "cluster_id": "syn_03015_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "tokenizer_hashes": {"receiver": "b9d8872cf3a99da99e1fc4b0f71232a451632dc49ca4c6771549b4efa5c1c763"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 4, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1741, "query_tokens": 531, "message_tokens": 419, "message_utf8_bytes": 1119, "cache_tensor_bytes": null, "serialized_payload_bytes": 1119, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 79624.5637909742, "hops": {"1": 22925.759125035256, "2": 16508.093582931906, "3": 23855.072916019708, "4": 16335.638166987337}, "receiver_prefill_ms": 349.4883750099689, "ttft_ms": 23.415750009007752, "decode_ms": 40.39866698440164}, "total_wall_ms": 80038.50220795721, "ttft_ms": 23.415750009007752, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 61, "facts_retained": 24, "retention_rate": 0.39344262295081966, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.4166666666666667, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1119}, "ledger_by_hop": {"1": {"facts_total": 61, "facts_retained": 24, "retention_rate": 0.39344262295081966, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.4166666666666667, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1161}, "2": {"facts_total": 61, "facts_retained": 24, "retention_rate": 0.39344262295081966, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.4166666666666667, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1123}, "3": {"facts_total": 61, "facts_retained": 24, "retention_rate": 0.39344262295081966, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.4166666666666667, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1123}, "4": {"facts_total": 61, "facts_retained": 24, "retention_rate": 0.39344262295081966, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.4166666666666667, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1119}}}, "reference_key_id": "keys/synthetic/syn_03015_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "9cff6b905fbff4d9c34852ce07efb835c3b85e73c292b038a4a7a512d29a8b69", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "1f95d7876fd6a39a0436ca5b", "attempt_id": "0ba094d90e56", "run_group_id": "E4_core:A4|closed|qwen3b-llama3b:h4", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:01:11.567008Z", "task_family": "synthetic", "document_id": "syn_03015_2000", "document_hash": "71a7284b9bc3f681d69d91d18b9e8d45a4dda486cd14f4145395786ac6a68a89", "question_id": "q13_unanswerable", "split": "test", "cluster_id": "syn_03015_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "tokenizer_hashes": {"receiver": "b9d8872cf3a99da99e1fc4b0f71232a451632dc49ca4c6771549b4efa5c1c763"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 4, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1741, "query_tokens": 534, "message_tokens": 419, "message_utf8_bytes": 1119, "cache_tensor_bytes": null, "serialized_payload_bytes": 1119, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 79624.5637909742, "hops": {"1": 22925.759125035256, "2": 16508.093582931906, "3": 23855.072916019708, "4": 16335.638166987337}, "receiver_prefill_ms": 338.76516693271697, "ttft_ms": 21.554374950937927, "decode_ms": 37.24595799576491}, "total_wall_ms": 80022.98741601408, "ttft_ms": 21.554374950937927, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 2}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown.", "evidence_artifact": {"ledger": {"facts_total": 61, "facts_retained": 24, "retention_rate": 0.39344262295081966, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.4166666666666667, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1119}, "ledger_by_hop": {"1": {"facts_total": 61, "facts_retained": 24, "retention_rate": 0.39344262295081966, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.4166666666666667, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1161}, "2": {"facts_total": 61, "facts_retained": 24, "retention_rate": 0.39344262295081966, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.4166666666666667, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1123}, "3": {"facts_total": 61, "facts_retained": 24, "retention_rate": 0.39344262295081966, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.4166666666666667, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1123}, "4": {"facts_total": 61, "facts_retained": 24, "retention_rate": 0.39344262295081966, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.4166666666666667, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1119}}}, "reference_key_id": "keys/synthetic/syn_03015_2000.json", "task_scores": {"correct": 1.0, "abstained": 1.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "9cff6b905fbff4d9c34852ce07efb835c3b85e73c292b038a4a7a512d29a8b69", "output_hash": "4dd633b1f2ee444666fd70af2b36443331fc00e69f3e569847cf278ca625c041"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "c19f6d61966cda6c5986d400", "attempt_id": "13bf9902b6c2", "run_group_id": "E4_core:A4|closed|qwen3b-llama3b:h4", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:01:11.983566Z", "task_family": "synthetic", "document_id": "syn_03015_2000", "document_hash": "71a7284b9bc3f681d69d91d18b9e8d45a4dda486cd14f4145395786ac6a68a89", "question_id": "q14_unanswerable", "split": "test", "cluster_id": "syn_03015_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "tokenizer_hashes": {"receiver": "b9d8872cf3a99da99e1fc4b0f71232a451632dc49ca4c6771549b4efa5c1c763"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 4, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1741, "query_tokens": 536, "message_tokens": 419, "message_utf8_bytes": 1119, "cache_tensor_bytes": null, "serialized_payload_bytes": 1119, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 79624.5637909742, "hops": {"1": 22925.759125035256, "2": 16508.093582931906, "3": 23855.072916019708, "4": 16335.638166987337}, "receiver_prefill_ms": 345.2578749274835, "ttft_ms": 49.957916024141014, "decode_ms": 49.988375045359135}, "total_wall_ms": 80070.65241597593, "ttft_ms": 49.957916024141014, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 61, "facts_retained": 24, "retention_rate": 0.39344262295081966, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.4166666666666667, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1119}, "ledger_by_hop": {"1": {"facts_total": 61, "facts_retained": 24, "retention_rate": 0.39344262295081966, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.4166666666666667, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1161}, "2": {"facts_total": 61, "facts_retained": 24, "retention_rate": 0.39344262295081966, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.4166666666666667, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1123}, "3": {"facts_total": 61, "facts_retained": 24, "retention_rate": 0.39344262295081966, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.4166666666666667, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1123}, "4": {"facts_total": 61, "facts_retained": 24, "retention_rate": 0.39344262295081966, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.4166666666666667, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1119}}}, "reference_key_id": "keys/synthetic/syn_03015_2000.json", "task_scores": {"correct": 1.0, "abstained": 1.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "9cff6b905fbff4d9c34852ce07efb835c3b85e73c292b038a4a7a512d29a8b69", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "a06f2e9ad07838e62384c267", "attempt_id": "a0729eaf7ee6", "run_group_id": "E4_core:A4|closed|qwen3b-llama3b:h8", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:02:29.496556Z", "task_family": "synthetic", "document_id": "syn_03015_2000", "document_hash": "71a7284b9bc3f681d69d91d18b9e8d45a4dda486cd14f4145395786ac6a68a89", "question_id": "q00_single_fact", "split": "test", "cluster_id": "syn_03015_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "tokenizer_hashes": {"receiver": "b9d8872cf3a99da99e1fc4b0f71232a451632dc49ca4c6771549b4efa5c1c763"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 8, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1741, "query_tokens": 523, "message_tokens": 419, "message_utf8_bytes": 1119, "cache_tensor_bytes": null, "serialized_payload_bytes": 1119, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 156668.96754212212, "hops": {"1": 22925.759125035256, "2": 16508.093582931906, "3": 23855.072916019708, "4": 16335.638166987337, "5": 21821.69304206036, "6": 16540.048209019005, "7": 22165.478167007677, "8": 16517.184333060868}, "receiver_prefill_ms": 334.58912500645965, "ttft_ms": 32.157166046090424, "decode_ms": 30.852583004161716}, "total_wall_ms": 157067.19575007446, "ttft_ms": 32.157166046090424, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 61, "facts_retained": 24, "retention_rate": 0.39344262295081966, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.4166666666666667, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1119}, "ledger_by_hop": {"1": {"facts_total": 61, "facts_retained": 24, "retention_rate": 0.39344262295081966, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.4166666666666667, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1161}, "2": {"facts_total": 61, "facts_retained": 24, "retention_rate": 0.39344262295081966, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.4166666666666667, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1123}, "3": {"facts_total": 61, "facts_retained": 24, "retention_rate": 0.39344262295081966, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.4166666666666667, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1123}, "4": {"facts_total": 61, "facts_retained": 24, "retention_rate": 0.39344262295081966, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.4166666666666667, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1119}, "5": {"facts_total": 61, "facts_retained": 24, "retention_rate": 0.39344262295081966, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.4166666666666667, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1119}, "6": {"facts_total": 61, "facts_retained": 24, "retention_rate": 0.39344262295081966, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.4166666666666667, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1119}, "7": {"facts_total": 61, "facts_retained": 24, "retention_rate": 0.39344262295081966, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.4166666666666667, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1119}, "8": {"facts_total": 61, "facts_retained": 24, "retention_rate": 0.39344262295081966, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.4166666666666667, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1119}}}, "reference_key_id": "keys/synthetic/syn_03015_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "9cff6b905fbff4d9c34852ce07efb835c3b85e73c292b038a4a7a512d29a8b69", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "2aac862d6871bf531e807cd4", "attempt_id": "c5ff163d01cd", "run_group_id": "E4_core:A4|closed|qwen3b-llama3b:h8", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:02:29.913045Z", "task_family": "synthetic", "document_id": "syn_03015_2000", "document_hash": "71a7284b9bc3f681d69d91d18b9e8d45a4dda486cd14f4145395786ac6a68a89", "question_id": "q01_single_fact", "split": "test", "cluster_id": "syn_03015_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "tokenizer_hashes": {"receiver": "b9d8872cf3a99da99e1fc4b0f71232a451632dc49ca4c6771549b4efa5c1c763"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 8, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1741, "query_tokens": 527, "message_tokens": 419, "message_utf8_bytes": 1119, "cache_tensor_bytes": null, "serialized_payload_bytes": 1119, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 156668.96754212212, "hops": {"1": 22925.759125035256, "2": 16508.093582931906, "3": 23855.072916019708, "4": 16335.638166987337, "5": 21821.69304206036, "6": 16540.048209019005, "7": 22165.478167007677, "8": 16517.184333060868}, "receiver_prefill_ms": 343.2081670034677, "ttft_ms": 37.08162496332079, "decode_ms": 20.123374997638166}, "total_wall_ms": 157069.95991710573, "ttft_ms": 37.08162496332079, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 61, "facts_retained": 24, "retention_rate": 0.39344262295081966, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.4166666666666667, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1119}, "ledger_by_hop": {"1": {"facts_total": 61, "facts_retained": 24, "retention_rate": 0.39344262295081966, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.4166666666666667, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1161}, "2": {"facts_total": 61, "facts_retained": 24, "retention_rate": 0.39344262295081966, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.4166666666666667, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1123}, "3": {"facts_total": 61, "facts_retained": 24, "retention_rate": 0.39344262295081966, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.4166666666666667, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1123}, "4": {"facts_total": 61, "facts_retained": 24, "retention_rate": 0.39344262295081966, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.4166666666666667, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1119}, "5": {"facts_total": 61, "facts_retained": 24, "retention_rate": 0.39344262295081966, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.4166666666666667, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1119}, "6": {"facts_total": 61, "facts_retained": 24, "retention_rate": 0.39344262295081966, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.4166666666666667, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1119}, "7": {"facts_total": 61, "facts_retained": 24, "retention_rate": 0.39344262295081966, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.4166666666666667, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1119}, "8": {"facts_total": 61, "facts_retained": 24, "retention_rate": 0.39344262295081966, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.4166666666666667, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1119}}}, "reference_key_id": "keys/synthetic/syn_03015_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "9cff6b905fbff4d9c34852ce07efb835c3b85e73c292b038a4a7a512d29a8b69", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "daf1a5119946bfb3b4e32db4", "attempt_id": "ac0a776a3434", "run_group_id": "E4_core:A4|closed|qwen3b-llama3b:h8", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:02:30.332007Z", "task_family": "synthetic", "document_id": "syn_03015_2000", "document_hash": "71a7284b9bc3f681d69d91d18b9e8d45a4dda486cd14f4145395786ac6a68a89", "question_id": "q02_exact_copy", "split": "test", "cluster_id": "syn_03015_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "tokenizer_hashes": {"receiver": "b9d8872cf3a99da99e1fc4b0f71232a451632dc49ca4c6771549b4efa5c1c763"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 8, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1741, "query_tokens": 527, "message_tokens": 419, "message_utf8_bytes": 1119, "cache_tensor_bytes": null, "serialized_payload_bytes": 1119, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 156668.96754212212, "hops": {"1": 22925.759125035256, "2": 16508.093582931906, "3": 23855.072916019708, "4": 16335.638166987337, "5": 21821.69304206036, "6": 16540.048209019005, "7": 22165.478167007677, "8": 16517.184333060868}, "receiver_prefill_ms": 319.1804999951273, "ttft_ms": 40.63179204240441, "decode_ms": 266.9024160131812}, "total_wall_ms": 157296.47800105158, "ttft_ms": 40.63179204240441, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 4}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "89.3122", "evidence_artifact": {"ledger": {"facts_total": 61, "facts_retained": 24, "retention_rate": 0.39344262295081966, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.4166666666666667, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1119}, "ledger_by_hop": {"1": {"facts_total": 61, "facts_retained": 24, "retention_rate": 0.39344262295081966, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.4166666666666667, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1161}, "2": {"facts_total": 61, "facts_retained": 24, "retention_rate": 0.39344262295081966, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.4166666666666667, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1123}, "3": {"facts_total": 61, "facts_retained": 24, "retention_rate": 0.39344262295081966, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.4166666666666667, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1123}, "4": {"facts_total": 61, "facts_retained": 24, "retention_rate": 0.39344262295081966, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.4166666666666667, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1119}, "5": {"facts_total": 61, "facts_retained": 24, "retention_rate": 0.39344262295081966, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.4166666666666667, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1119}, "6": {"facts_total": 61, "facts_retained": 24, "retention_rate": 0.39344262295081966, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.4166666666666667, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1119}, "7": {"facts_total": 61, "facts_retained": 24, "retention_rate": 0.39344262295081966, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.4166666666666667, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1119}, "8": {"facts_total": 61, "facts_retained": 24, "retention_rate": 0.39344262295081966, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.4166666666666667, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1119}}}, "reference_key_id": "keys/synthetic/syn_03015_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "9cff6b905fbff4d9c34852ce07efb835c3b85e73c292b038a4a7a512d29a8b69", "output_hash": "2138888e6519b2f4b1b54142e6f9c541102d7f9ee5bcf8e4184c81a382601757"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "fe2361e1b05ec38709522aeb", "attempt_id": "c5d9320fdb6e", "run_group_id": "E4_core:A4|closed|qwen3b-llama3b:h8", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:02:30.977024Z", "task_family": "synthetic", "document_id": "syn_03015_2000", "document_hash": "71a7284b9bc3f681d69d91d18b9e8d45a4dda486cd14f4145395786ac6a68a89", "question_id": "q03_exact_copy", "split": "test", "cluster_id": "syn_03015_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "tokenizer_hashes": {"receiver": "b9d8872cf3a99da99e1fc4b0f71232a451632dc49ca4c6771549b4efa5c1c763"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 8, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1741, "query_tokens": 523, "message_tokens": 419, "message_utf8_bytes": 1119, "cache_tensor_bytes": null, "serialized_payload_bytes": 1119, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 156668.96754212212, "hops": {"1": 22925.759125035256, "2": 16508.093582931906, "3": 23855.072916019708, "4": 16335.638166987337, "5": 21821.69304206036, "6": 16540.048209019005, "7": 22165.478167007677, "8": 16517.184333060868}, "receiver_prefill_ms": 414.788750000298, "ttft_ms": 38.39870903175324, "decode_ms": 213.69641600176692}, "total_wall_ms": 157336.39583410695, "ttft_ms": 38.39870903175324, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 7}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "Wilhelmina Dubrovnik.", "evidence_artifact": {"ledger": {"facts_total": 61, "facts_retained": 24, "retention_rate": 0.39344262295081966, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.4166666666666667, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1119}, "ledger_by_hop": {"1": {"facts_total": 61, "facts_retained": 24, "retention_rate": 0.39344262295081966, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.4166666666666667, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1161}, "2": {"facts_total": 61, "facts_retained": 24, "retention_rate": 0.39344262295081966, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.4166666666666667, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1123}, "3": {"facts_total": 61, "facts_retained": 24, "retention_rate": 0.39344262295081966, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.4166666666666667, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1123}, "4": {"facts_total": 61, "facts_retained": 24, "retention_rate": 0.39344262295081966, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.4166666666666667, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1119}, "5": {"facts_total": 61, "facts_retained": 24, "retention_rate": 0.39344262295081966, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.4166666666666667, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1119}, "6": {"facts_total": 61, "facts_retained": 24, "retention_rate": 0.39344262295081966, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.4166666666666667, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1119}, "7": {"facts_total": 61, "facts_retained": 24, "retention_rate": 0.39344262295081966, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.4166666666666667, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1119}, "8": {"facts_total": 61, "facts_retained": 24, "retention_rate": 0.39344262295081966, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.4166666666666667, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1119}}}, "reference_key_id": "keys/synthetic/syn_03015_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "9cff6b905fbff4d9c34852ce07efb835c3b85e73c292b038a4a7a512d29a8b69", "output_hash": "79ba0181ee6c8e2b01e096d290463d1e2eb3a595aa1abc3242a857276f5ca877"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "1d10c6b02670b1e12b076c5c", "attempt_id": "5d7451435d84", "run_group_id": "E4_core:A4|closed|qwen3b-llama3b:h8", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:02:31.664404Z", "task_family": "synthetic", "document_id": "syn_03015_2000", "document_hash": "71a7284b9bc3f681d69d91d18b9e8d45a4dda486cd14f4145395786ac6a68a89", "question_id": "q04_multi_hop", "split": "test", "cluster_id": "syn_03015_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "tokenizer_hashes": {"receiver": "b9d8872cf3a99da99e1fc4b0f71232a451632dc49ca4c6771549b4efa5c1c763"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 8, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1741, "query_tokens": 532, "message_tokens": 419, "message_utf8_bytes": 1119, "cache_tensor_bytes": null, "serialized_payload_bytes": 1119, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 156668.96754212212, "hops": {"1": 22925.759125035256, "2": 16508.093582931906, "3": 23855.072916019708, "4": 16335.638166987337, "5": 21821.69304206036, "6": 16540.048209019005, "7": 22165.478167007677, "8": 16517.184333060868}, "receiver_prefill_ms": 337.73716702125967, "ttft_ms": 37.220874917693436, "decode_ms": 86.70874999370426}, "total_wall_ms": 157131.28729211167, "ttft_ms": 37.220874917693436, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 5}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "8,421,725", "evidence_artifact": {"ledger": {"facts_total": 61, "facts_retained": 24, "retention_rate": 0.39344262295081966, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.4166666666666667, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1119}, "ledger_by_hop": {"1": {"facts_total": 61, "facts_retained": 24, "retention_rate": 0.39344262295081966, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.4166666666666667, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1161}, "2": {"facts_total": 61, "facts_retained": 24, "retention_rate": 0.39344262295081966, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.4166666666666667, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1123}, "3": {"facts_total": 61, "facts_retained": 24, "retention_rate": 0.39344262295081966, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.4166666666666667, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1123}, "4": {"facts_total": 61, "facts_retained": 24, "retention_rate": 0.39344262295081966, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.4166666666666667, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1119}, "5": {"facts_total": 61, "facts_retained": 24, "retention_rate": 0.39344262295081966, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.4166666666666667, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1119}, "6": {"facts_total": 61, "facts_retained": 24, "retention_rate": 0.39344262295081966, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.4166666666666667, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1119}, "7": {"facts_total": 61, "facts_retained": 24, "retention_rate": 0.39344262295081966, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.4166666666666667, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1119}, "8": {"facts_total": 61, "facts_retained": 24, "retention_rate": 0.39344262295081966, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.4166666666666667, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1119}}}, "reference_key_id": "keys/synthetic/syn_03015_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "9cff6b905fbff4d9c34852ce07efb835c3b85e73c292b038a4a7a512d29a8b69", "output_hash": "6e31df6d4fb38417a5210ded39fc1e4c0e32f278ab9cba0da5b8e7585ddeb534"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "b9cbcbd85af5ab465b733d4e", "attempt_id": "c416495a1204", "run_group_id": "E4_core:A4|closed|qwen3b-llama3b:h8", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:02:32.143838Z", "task_family": "synthetic", "document_id": "syn_03015_2000", "document_hash": "71a7284b9bc3f681d69d91d18b9e8d45a4dda486cd14f4145395786ac6a68a89", "question_id": "q05_multi_hop", "split": "test", "cluster_id": "syn_03015_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "tokenizer_hashes": {"receiver": "b9d8872cf3a99da99e1fc4b0f71232a451632dc49ca4c6771549b4efa5c1c763"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 8, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1741, "query_tokens": 530, "message_tokens": 419, "message_utf8_bytes": 1119, "cache_tensor_bytes": null, "serialized_payload_bytes": 1119, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 156668.96754212212, "hops": {"1": 22925.759125035256, "2": 16508.093582931906, "3": 23855.072916019708, "4": 16335.638166987337, "5": 21821.69304206036, "6": 16540.048209019005, "7": 22165.478167007677, "8": 16517.184333060868}, "receiver_prefill_ms": 321.32404204458, "ttft_ms": 56.157125043682754, "decode_ms": 62.89595807902515}, "total_wall_ms": 157110.27583410032, "ttft_ms": 56.157125043682754, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 61, "facts_retained": 24, "retention_rate": 0.39344262295081966, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.4166666666666667, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1119}, "ledger_by_hop": {"1": {"facts_total": 61, "facts_retained": 24, "retention_rate": 0.39344262295081966, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.4166666666666667, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1161}, "2": {"facts_total": 61, "facts_retained": 24, "retention_rate": 0.39344262295081966, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.4166666666666667, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1123}, "3": {"facts_total": 61, "facts_retained": 24, "retention_rate": 0.39344262295081966, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.4166666666666667, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1123}, "4": {"facts_total": 61, "facts_retained": 24, "retention_rate": 0.39344262295081966, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.4166666666666667, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1119}, "5": {"facts_total": 61, "facts_retained": 24, "retention_rate": 0.39344262295081966, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.4166666666666667, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1119}, "6": {"facts_total": 61, "facts_retained": 24, "retention_rate": 0.39344262295081966, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.4166666666666667, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1119}, "7": {"facts_total": 61, "facts_retained": 24, "retention_rate": 0.39344262295081966, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.4166666666666667, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1119}, "8": {"facts_total": 61, "facts_retained": 24, "retention_rate": 0.39344262295081966, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.4166666666666667, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1119}}}, "reference_key_id": "keys/synthetic/syn_03015_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "9cff6b905fbff4d9c34852ce07efb835c3b85e73c292b038a4a7a512d29a8b69", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "d4fb633fd6ef2031400fdabd", "attempt_id": "db187534b6b6", "run_group_id": "E4_core:A4|closed|qwen3b-llama3b:h8", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:02:32.603726Z", "task_family": "synthetic", "document_id": "syn_03015_2000", "document_hash": "71a7284b9bc3f681d69d91d18b9e8d45a4dda486cd14f4145395786ac6a68a89", "question_id": "q06_negation", "split": "test", "cluster_id": "syn_03015_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "tokenizer_hashes": {"receiver": "b9d8872cf3a99da99e1fc4b0f71232a451632dc49ca4c6771549b4efa5c1c763"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 8, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1741, "query_tokens": 524, "message_tokens": 419, "message_utf8_bytes": 1119, "cache_tensor_bytes": null, "serialized_payload_bytes": 1119, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 156668.96754212212, "hops": {"1": 22925.759125035256, "2": 16508.093582931906, "3": 23855.072916019708, "4": 16335.638166987337, "5": 21821.69304206036, "6": 16540.048209019005, "7": 22165.478167007677, "8": 16517.184333060868}, "receiver_prefill_ms": 469.76079198066145, "ttft_ms": 34.19029095675796, "decode_ms": 62.07337498199195}, "total_wall_ms": 157235.59700022452, "ttft_ms": 34.19029095675796, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 2}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "No.", "evidence_artifact": {"ledger": {"facts_total": 61, "facts_retained": 24, "retention_rate": 0.39344262295081966, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.4166666666666667, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1119}, "ledger_by_hop": {"1": {"facts_total": 61, "facts_retained": 24, "retention_rate": 0.39344262295081966, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.4166666666666667, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1161}, "2": {"facts_total": 61, "facts_retained": 24, "retention_rate": 0.39344262295081966, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.4166666666666667, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1123}, "3": {"facts_total": 61, "facts_retained": 24, "retention_rate": 0.39344262295081966, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.4166666666666667, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1123}, "4": {"facts_total": 61, "facts_retained": 24, "retention_rate": 0.39344262295081966, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.4166666666666667, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1119}, "5": {"facts_total": 61, "facts_retained": 24, "retention_rate": 0.39344262295081966, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.4166666666666667, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1119}, "6": {"facts_total": 61, "facts_retained": 24, "retention_rate": 0.39344262295081966, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.4166666666666667, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1119}, "7": {"facts_total": 61, "facts_retained": 24, "retention_rate": 0.39344262295081966, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.4166666666666667, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1119}, "8": {"facts_total": 61, "facts_retained": 24, "retention_rate": 0.39344262295081966, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.4166666666666667, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1119}}}, "reference_key_id": "keys/synthetic/syn_03015_2000.json", "task_scores": {"correct": 0.0, "abstained": 0.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "9cff6b905fbff4d9c34852ce07efb835c3b85e73c292b038a4a7a512d29a8b69", "output_hash": "38178a20b470cfd18299fb1593dd4ba706f1a83321dac051d01086efd8b7a96f"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "9edc582a88f9dde506ed6ab9", "attempt_id": "13f00ecdf854", "run_group_id": "E4_core:A4|closed|qwen3b-llama3b:h8", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:02:33.187731Z", "task_family": "synthetic", "document_id": "syn_03015_2000", "document_hash": "71a7284b9bc3f681d69d91d18b9e8d45a4dda486cd14f4145395786ac6a68a89", "question_id": "q07_negation", "split": "test", "cluster_id": "syn_03015_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "tokenizer_hashes": {"receiver": "b9d8872cf3a99da99e1fc4b0f71232a451632dc49ca4c6771549b4efa5c1c763"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 8, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1741, "query_tokens": 524, "message_tokens": 419, "message_utf8_bytes": 1119, "cache_tensor_bytes": null, "serialized_payload_bytes": 1119, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 156668.96754212212, "hops": {"1": 22925.759125035256, "2": 16508.093582931906, "3": 23855.072916019708, "4": 16335.638166987337, "5": 21821.69304206036, "6": 16540.048209019005, "7": 22165.478167007677, "8": 16517.184333060868}, "receiver_prefill_ms": 349.92329205852, "ttft_ms": 42.66412497963756, "decode_ms": 62.70795897580683}, "total_wall_ms": 157124.94520808104, "ttft_ms": 42.66412497963756, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 2}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "No.", "evidence_artifact": {"ledger": {"facts_total": 61, "facts_retained": 24, "retention_rate": 0.39344262295081966, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.4166666666666667, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1119}, "ledger_by_hop": {"1": {"facts_total": 61, "facts_retained": 24, "retention_rate": 0.39344262295081966, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.4166666666666667, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1161}, "2": {"facts_total": 61, "facts_retained": 24, "retention_rate": 0.39344262295081966, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.4166666666666667, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1123}, "3": {"facts_total": 61, "facts_retained": 24, "retention_rate": 0.39344262295081966, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.4166666666666667, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1123}, "4": {"facts_total": 61, "facts_retained": 24, "retention_rate": 0.39344262295081966, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.4166666666666667, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1119}, "5": {"facts_total": 61, "facts_retained": 24, "retention_rate": 0.39344262295081966, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.4166666666666667, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1119}, "6": {"facts_total": 61, "facts_retained": 24, "retention_rate": 0.39344262295081966, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.4166666666666667, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1119}, "7": {"facts_total": 61, "facts_retained": 24, "retention_rate": 0.39344262295081966, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.4166666666666667, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1119}, "8": {"facts_total": 61, "facts_retained": 24, "retention_rate": 0.39344262295081966, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.4166666666666667, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1119}}}, "reference_key_id": "keys/synthetic/syn_03015_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "9cff6b905fbff4d9c34852ce07efb835c3b85e73c292b038a4a7a512d29a8b69", "output_hash": "38178a20b470cfd18299fb1593dd4ba706f1a83321dac051d01086efd8b7a96f"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "4af4214d67051725b273a921", "attempt_id": "cd7167511fb4", "run_group_id": "E4_core:A4|closed|qwen3b-llama3b:h8", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:02:33.662101Z", "task_family": "synthetic", "document_id": "syn_03015_2000", "document_hash": "71a7284b9bc3f681d69d91d18b9e8d45a4dda486cd14f4145395786ac6a68a89", "question_id": "q08_revision", "split": "test", "cluster_id": "syn_03015_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "tokenizer_hashes": {"receiver": "b9d8872cf3a99da99e1fc4b0f71232a451632dc49ca4c6771549b4efa5c1c763"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 8, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1741, "query_tokens": 537, "message_tokens": 419, "message_utf8_bytes": 1119, "cache_tensor_bytes": null, "serialized_payload_bytes": 1119, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 156668.96754212212, "hops": {"1": 22925.759125035256, "2": 16508.093582931906, "3": 23855.072916019708, "4": 16335.638166987337, "5": 21821.69304206036, "6": 16540.048209019005, "7": 22165.478167007677, "8": 16517.184333060868}, "receiver_prefill_ms": 328.3788750413805, "ttft_ms": 55.96637504640967, "decode_ms": 56.98924988973886}, "total_wall_ms": 157111.17079202086, "ttft_ms": 55.96637504640967, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 61, "facts_retained": 24, "retention_rate": 0.39344262295081966, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.4166666666666667, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1119}, "ledger_by_hop": {"1": {"facts_total": 61, "facts_retained": 24, "retention_rate": 0.39344262295081966, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.4166666666666667, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1161}, "2": {"facts_total": 61, "facts_retained": 24, "retention_rate": 0.39344262295081966, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.4166666666666667, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1123}, "3": {"facts_total": 61, "facts_retained": 24, "retention_rate": 0.39344262295081966, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.4166666666666667, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1123}, "4": {"facts_total": 61, "facts_retained": 24, "retention_rate": 0.39344262295081966, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.4166666666666667, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1119}, "5": {"facts_total": 61, "facts_retained": 24, "retention_rate": 0.39344262295081966, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.4166666666666667, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1119}, "6": {"facts_total": 61, "facts_retained": 24, "retention_rate": 0.39344262295081966, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.4166666666666667, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1119}, "7": {"facts_total": 61, "facts_retained": 24, "retention_rate": 0.39344262295081966, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.4166666666666667, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1119}, "8": {"facts_total": 61, "facts_retained": 24, "retention_rate": 0.39344262295081966, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.4166666666666667, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1119}}}, "reference_key_id": "keys/synthetic/syn_03015_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "9cff6b905fbff4d9c34852ce07efb835c3b85e73c292b038a4a7a512d29a8b69", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "1826a801cddc868ba69c26f4", "attempt_id": "de6edfea2a05", "run_group_id": "E4_core:A4|closed|qwen3b-llama3b:h8", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:02:34.122737Z", "task_family": "synthetic", "document_id": "syn_03015_2000", "document_hash": "71a7284b9bc3f681d69d91d18b9e8d45a4dda486cd14f4145395786ac6a68a89", "question_id": "q09_revision", "split": "test", "cluster_id": "syn_03015_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "tokenizer_hashes": {"receiver": "b9d8872cf3a99da99e1fc4b0f71232a451632dc49ca4c6771549b4efa5c1c763"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 8, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1741, "query_tokens": 531, "message_tokens": 419, "message_utf8_bytes": 1119, "cache_tensor_bytes": null, "serialized_payload_bytes": 1119, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 156668.96754212212, "hops": {"1": 22925.759125035256, "2": 16508.093582931906, "3": 23855.072916019708, "4": 16335.638166987337, "5": 21821.69304206036, "6": 16540.048209019005, "7": 22165.478167007677, "8": 16517.184333060868}, "receiver_prefill_ms": 366.0657919244841, "ttft_ms": 145.20758297294378, "decode_ms": 160.05183407105505}, "total_wall_ms": 157341.07016713824, "ttft_ms": 145.20758297294378, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 5}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "7,028,925", "evidence_artifact": {"ledger": {"facts_total": 61, "facts_retained": 24, "retention_rate": 0.39344262295081966, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.4166666666666667, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1119}, "ledger_by_hop": {"1": {"facts_total": 61, "facts_retained": 24, "retention_rate": 0.39344262295081966, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.4166666666666667, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1161}, "2": {"facts_total": 61, "facts_retained": 24, "retention_rate": 0.39344262295081966, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.4166666666666667, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1123}, "3": {"facts_total": 61, "facts_retained": 24, "retention_rate": 0.39344262295081966, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.4166666666666667, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1123}, "4": {"facts_total": 61, "facts_retained": 24, "retention_rate": 0.39344262295081966, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.4166666666666667, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1119}, "5": {"facts_total": 61, "facts_retained": 24, "retention_rate": 0.39344262295081966, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.4166666666666667, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1119}, "6": {"facts_total": 61, "facts_retained": 24, "retention_rate": 0.39344262295081966, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.4166666666666667, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1119}, "7": {"facts_total": 61, "facts_retained": 24, "retention_rate": 0.39344262295081966, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.4166666666666667, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1119}, "8": {"facts_total": 61, "facts_retained": 24, "retention_rate": 0.39344262295081966, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.4166666666666667, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1119}}}, "reference_key_id": "keys/synthetic/syn_03015_2000.json", "task_scores": {"correct": 0.0, "abstained": 0.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "9cff6b905fbff4d9c34852ce07efb835c3b85e73c292b038a4a7a512d29a8b69", "output_hash": "e83edb3f38a69e604be343e63e2e84442156fcb2bb30f36ba9716575d5ffd57a"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "3d0c96b97e273dfc2a294e53", "attempt_id": "81a2ceea01e7", "run_group_id": "E4_core:A4|closed|qwen3b-llama3b:h8", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:02:34.811837Z", "task_family": "synthetic", "document_id": "syn_03015_2000", "document_hash": "71a7284b9bc3f681d69d91d18b9e8d45a4dda486cd14f4145395786ac6a68a89", "question_id": "q10_temporal", "split": "test", "cluster_id": "syn_03015_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "tokenizer_hashes": {"receiver": "b9d8872cf3a99da99e1fc4b0f71232a451632dc49ca4c6771549b4efa5c1c763"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 8, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1741, "query_tokens": 540, "message_tokens": 419, "message_utf8_bytes": 1119, "cache_tensor_bytes": null, "serialized_payload_bytes": 1119, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 156668.96754212212, "hops": {"1": 22925.759125035256, "2": 16508.093582931906, "3": 23855.072916019708, "4": 16335.638166987337, "5": 21821.69304206036, "6": 16540.048209019005, "7": 22165.478167007677, "8": 16517.184333060868}, "receiver_prefill_ms": 349.5162500767037, "ttft_ms": 32.856166013516486, "decode_ms": 31.132125062868}, "total_wall_ms": 157082.9821670195, "ttft_ms": 32.856166013516486, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 61, "facts_retained": 24, "retention_rate": 0.39344262295081966, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.4166666666666667, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1119}, "ledger_by_hop": {"1": {"facts_total": 61, "facts_retained": 24, "retention_rate": 0.39344262295081966, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.4166666666666667, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1161}, "2": {"facts_total": 61, "facts_retained": 24, "retention_rate": 0.39344262295081966, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.4166666666666667, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1123}, "3": {"facts_total": 61, "facts_retained": 24, "retention_rate": 0.39344262295081966, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.4166666666666667, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1123}, "4": {"facts_total": 61, "facts_retained": 24, "retention_rate": 0.39344262295081966, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.4166666666666667, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1119}, "5": {"facts_total": 61, "facts_retained": 24, "retention_rate": 0.39344262295081966, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.4166666666666667, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1119}, "6": {"facts_total": 61, "facts_retained": 24, "retention_rate": 0.39344262295081966, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.4166666666666667, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1119}, "7": {"facts_total": 61, "facts_retained": 24, "retention_rate": 0.39344262295081966, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.4166666666666667, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1119}, "8": {"facts_total": 61, "facts_retained": 24, "retention_rate": 0.39344262295081966, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.4166666666666667, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1119}}}, "reference_key_id": "keys/synthetic/syn_03015_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "9cff6b905fbff4d9c34852ce07efb835c3b85e73c292b038a4a7a512d29a8b69", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "c87af1931493c4d35f393f28", "attempt_id": "72afb15bc946", "run_group_id": "E4_core:A4|closed|qwen3b-llama3b:h8", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:02:35.243784Z", "task_family": "synthetic", "document_id": "syn_03015_2000", "document_hash": "71a7284b9bc3f681d69d91d18b9e8d45a4dda486cd14f4145395786ac6a68a89", "question_id": "q11_aggregation", "split": "test", "cluster_id": "syn_03015_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "tokenizer_hashes": {"receiver": "b9d8872cf3a99da99e1fc4b0f71232a451632dc49ca4c6771549b4efa5c1c763"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 8, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1741, "query_tokens": 534, "message_tokens": 419, "message_utf8_bytes": 1119, "cache_tensor_bytes": null, "serialized_payload_bytes": 1119, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 156668.96754212212, "hops": {"1": 22925.759125035256, "2": 16508.093582931906, "3": 23855.072916019708, "4": 16335.638166987337, "5": 21821.69304206036, "6": 16540.048209019005, "7": 22165.478167007677, "8": 16517.184333060868}, "receiver_prefill_ms": 337.44308399036527, "ttft_ms": 22.300042095594108, "decode_ms": 190.24762511253357}, "total_wall_ms": 157219.9105011532, "ttft_ms": 22.300042095594108, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 5}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "1,806,025", "evidence_artifact": {"ledger": {"facts_total": 61, "facts_retained": 24, "retention_rate": 0.39344262295081966, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.4166666666666667, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1119}, "ledger_by_hop": {"1": {"facts_total": 61, "facts_retained": 24, "retention_rate": 0.39344262295081966, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.4166666666666667, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1161}, "2": {"facts_total": 61, "facts_retained": 24, "retention_rate": 0.39344262295081966, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.4166666666666667, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1123}, "3": {"facts_total": 61, "facts_retained": 24, "retention_rate": 0.39344262295081966, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.4166666666666667, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1123}, "4": {"facts_total": 61, "facts_retained": 24, "retention_rate": 0.39344262295081966, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.4166666666666667, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1119}, "5": {"facts_total": 61, "facts_retained": 24, "retention_rate": 0.39344262295081966, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.4166666666666667, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1119}, "6": {"facts_total": 61, "facts_retained": 24, "retention_rate": 0.39344262295081966, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.4166666666666667, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1119}, "7": {"facts_total": 61, "facts_retained": 24, "retention_rate": 0.39344262295081966, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.4166666666666667, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1119}, "8": {"facts_total": 61, "facts_retained": 24, "retention_rate": 0.39344262295081966, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.4166666666666667, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1119}}}, "reference_key_id": "keys/synthetic/syn_03015_2000.json", "task_scores": {"correct": 0.0, "abstained": 0.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "9cff6b905fbff4d9c34852ce07efb835c3b85e73c292b038a4a7a512d29a8b69", "output_hash": "4de10b23990757532ff215b6a27f5f805aa6f751cc9757418279cc7ab85aa1a9"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "d6b1e82de72126fa7c309b27", "attempt_id": "d60163d98291", "run_group_id": "E4_core:A4|closed|qwen3b-llama3b:h8", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:02:35.812114Z", "task_family": "synthetic", "document_id": "syn_03015_2000", "document_hash": "71a7284b9bc3f681d69d91d18b9e8d45a4dda486cd14f4145395786ac6a68a89", "question_id": "q12_unit_conversion", "split": "test", "cluster_id": "syn_03015_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "tokenizer_hashes": {"receiver": "b9d8872cf3a99da99e1fc4b0f71232a451632dc49ca4c6771549b4efa5c1c763"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 8, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1741, "query_tokens": 531, "message_tokens": 419, "message_utf8_bytes": 1119, "cache_tensor_bytes": null, "serialized_payload_bytes": 1119, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 156668.96754212212, "hops": {"1": 22925.759125035256, "2": 16508.093582931906, "3": 23855.072916019708, "4": 16335.638166987337, "5": 21821.69304206036, "6": 16540.048209019005, "7": 22165.478167007677, "8": 16517.184333060868}, "receiver_prefill_ms": 369.3661659490317, "ttft_ms": 30.627166968770325, "decode_ms": 122.60366708505899}, "total_wall_ms": 157192.1331261983, "ttft_ms": 30.627166968770325, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 61, "facts_retained": 24, "retention_rate": 0.39344262295081966, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.4166666666666667, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1119}, "ledger_by_hop": {"1": {"facts_total": 61, "facts_retained": 24, "retention_rate": 0.39344262295081966, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.4166666666666667, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1161}, "2": {"facts_total": 61, "facts_retained": 24, "retention_rate": 0.39344262295081966, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.4166666666666667, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1123}, "3": {"facts_total": 61, "facts_retained": 24, "retention_rate": 0.39344262295081966, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.4166666666666667, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1123}, "4": {"facts_total": 61, "facts_retained": 24, "retention_rate": 0.39344262295081966, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.4166666666666667, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1119}, "5": {"facts_total": 61, "facts_retained": 24, "retention_rate": 0.39344262295081966, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.4166666666666667, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1119}, "6": {"facts_total": 61, "facts_retained": 24, "retention_rate": 0.39344262295081966, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.4166666666666667, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1119}, "7": {"facts_total": 61, "facts_retained": 24, "retention_rate": 0.39344262295081966, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.4166666666666667, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1119}, "8": {"facts_total": 61, "facts_retained": 24, "retention_rate": 0.39344262295081966, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.4166666666666667, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1119}}}, "reference_key_id": "keys/synthetic/syn_03015_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "9cff6b905fbff4d9c34852ce07efb835c3b85e73c292b038a4a7a512d29a8b69", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "8331b72782c0866a5a53a9c0", "attempt_id": "b26497a9ce4a", "run_group_id": "E4_core:A4|closed|qwen3b-llama3b:h8", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:02:36.352550Z", "task_family": "synthetic", "document_id": "syn_03015_2000", "document_hash": "71a7284b9bc3f681d69d91d18b9e8d45a4dda486cd14f4145395786ac6a68a89", "question_id": "q13_unanswerable", "split": "test", "cluster_id": "syn_03015_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "tokenizer_hashes": {"receiver": "b9d8872cf3a99da99e1fc4b0f71232a451632dc49ca4c6771549b4efa5c1c763"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 8, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1741, "query_tokens": 534, "message_tokens": 419, "message_utf8_bytes": 1119, "cache_tensor_bytes": null, "serialized_payload_bytes": 1119, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 156668.96754212212, "hops": {"1": 22925.759125035256, "2": 16508.093582931906, "3": 23855.072916019708, "4": 16335.638166987337, "5": 21821.69304206036, "6": 16540.048209019005, "7": 22165.478167007677, "8": 16517.184333060868}, "receiver_prefill_ms": 332.956500002183, "ttft_ms": 36.98579198680818, "decode_ms": 62.4381669331342}, "total_wall_ms": 157101.90375009552, "ttft_ms": 36.98579198680818, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 2}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown.", "evidence_artifact": {"ledger": {"facts_total": 61, "facts_retained": 24, "retention_rate": 0.39344262295081966, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.4166666666666667, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1119}, "ledger_by_hop": {"1": {"facts_total": 61, "facts_retained": 24, "retention_rate": 0.39344262295081966, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.4166666666666667, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1161}, "2": {"facts_total": 61, "facts_retained": 24, "retention_rate": 0.39344262295081966, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.4166666666666667, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1123}, "3": {"facts_total": 61, "facts_retained": 24, "retention_rate": 0.39344262295081966, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.4166666666666667, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1123}, "4": {"facts_total": 61, "facts_retained": 24, "retention_rate": 0.39344262295081966, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.4166666666666667, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1119}, "5": {"facts_total": 61, "facts_retained": 24, "retention_rate": 0.39344262295081966, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.4166666666666667, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1119}, "6": {"facts_total": 61, "facts_retained": 24, "retention_rate": 0.39344262295081966, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.4166666666666667, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1119}, "7": {"facts_total": 61, "facts_retained": 24, "retention_rate": 0.39344262295081966, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.4166666666666667, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1119}, "8": {"facts_total": 61, "facts_retained": 24, "retention_rate": 0.39344262295081966, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.4166666666666667, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1119}}}, "reference_key_id": "keys/synthetic/syn_03015_2000.json", "task_scores": {"correct": 1.0, "abstained": 1.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "9cff6b905fbff4d9c34852ce07efb835c3b85e73c292b038a4a7a512d29a8b69", "output_hash": "4dd633b1f2ee444666fd70af2b36443331fc00e69f3e569847cf278ca625c041"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "cf084a8a86bb85dd43586a7c", "attempt_id": "c09a851f062f", "run_group_id": "E4_core:A4|closed|qwen3b-llama3b:h8", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:02:36.802617Z", "task_family": "synthetic", "document_id": "syn_03015_2000", "document_hash": "71a7284b9bc3f681d69d91d18b9e8d45a4dda486cd14f4145395786ac6a68a89", "question_id": "q14_unanswerable", "split": "test", "cluster_id": "syn_03015_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "tokenizer_hashes": {"receiver": "b9d8872cf3a99da99e1fc4b0f71232a451632dc49ca4c6771549b4efa5c1c763"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 8, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1741, "query_tokens": 536, "message_tokens": 419, "message_utf8_bytes": 1119, "cache_tensor_bytes": null, "serialized_payload_bytes": 1119, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 156668.96754212212, "hops": {"1": 22925.759125035256, "2": 16508.093582931906, "3": 23855.072916019708, "4": 16335.638166987337, "5": 21821.69304206036, "6": 16540.048209019005, "7": 22165.478167007677, "8": 16517.184333060868}, "receiver_prefill_ms": 350.1413339981809, "ttft_ms": 19.90637497510761, "decode_ms": 35.21449991967529}, "total_wall_ms": 157074.79708408937, "ttft_ms": 19.90637497510761, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 61, "facts_retained": 24, "retention_rate": 0.39344262295081966, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.4166666666666667, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1119}, "ledger_by_hop": {"1": {"facts_total": 61, "facts_retained": 24, "retention_rate": 0.39344262295081966, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.4166666666666667, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1161}, "2": {"facts_total": 61, "facts_retained": 24, "retention_rate": 0.39344262295081966, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.4166666666666667, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1123}, "3": {"facts_total": 61, "facts_retained": 24, "retention_rate": 0.39344262295081966, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.4166666666666667, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1123}, "4": {"facts_total": 61, "facts_retained": 24, "retention_rate": 0.39344262295081966, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.4166666666666667, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1119}, "5": {"facts_total": 61, "facts_retained": 24, "retention_rate": 0.39344262295081966, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.4166666666666667, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1119}, "6": {"facts_total": 61, "facts_retained": 24, "retention_rate": 0.39344262295081966, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.4166666666666667, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1119}, "7": {"facts_total": 61, "facts_retained": 24, "retention_rate": 0.39344262295081966, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.4166666666666667, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1119}, "8": {"facts_total": 61, "facts_retained": 24, "retention_rate": 0.39344262295081966, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.5384615384615384, "status": 0.4166666666666667, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1119}}}, "reference_key_id": "keys/synthetic/syn_03015_2000.json", "task_scores": {"correct": 1.0, "abstained": 1.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "9cff6b905fbff4d9c34852ce07efb835c3b85e73c292b038a4a7a512d29a8b69", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "a359f4670fa8c3fc049c9c66", "attempt_id": "569ab4d64656", "run_group_id": "E4_core:A5|closed|qwen3b-llama3b:h1", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:03:00.015345Z", "task_family": "synthetic", "document_id": "syn_03015_2000", "document_hash": "71a7284b9bc3f681d69d91d18b9e8d45a4dda486cd14f4145395786ac6a68a89", "question_id": "q00_single_fact", "split": "test", "cluster_id": "syn_03015_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 1, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1922, "query_tokens": 411, "message_tokens": 328, "message_utf8_bytes": 754, "cache_tensor_bytes": null, "serialized_payload_bytes": 754, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 22787.491958006285, "hops": {"1": 22787.491958006285}, "receiver_prefill_ms": 344.2281670868397, "ttft_ms": 37.92466700542718, "decode_ms": 25.067374925129116}, "total_wall_ms": 23195.333291077986, "ttft_ms": 37.92466700542718, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 61, "facts_retained": 7, "retention_rate": 0.11475409836065574, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.3076923076923077, "status": 0.0, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 690}, "ledger_by_hop": {"1": {"facts_total": 61, "facts_retained": 7, "retention_rate": 0.11475409836065574, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.3076923076923077, "status": 0.0, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 690}}}, "reference_key_id": "keys/synthetic/syn_03015_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "e54cba3528ab78e09d5da03579b2a2035abb56934b1d0571c12acb32bf58b543", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "fe79fa225e01603257d93453", "attempt_id": "3ff8870ef201", "run_group_id": "E4_core:A5|closed|qwen3b-llama3b:h1", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:03:00.441071Z", "task_family": "synthetic", "document_id": "syn_03015_2000", "document_hash": "71a7284b9bc3f681d69d91d18b9e8d45a4dda486cd14f4145395786ac6a68a89", "question_id": "q01_single_fact", "split": "test", "cluster_id": "syn_03015_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 1, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1922, "query_tokens": 416, "message_tokens": 328, "message_utf8_bytes": 754, "cache_tensor_bytes": null, "serialized_payload_bytes": 754, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 22787.491958006285, "hops": {"1": 22787.491958006285}, "receiver_prefill_ms": 297.5750000914559, "ttft_ms": 40.32299993559718, "decode_ms": 24.953374988399446}, "total_wall_ms": 23150.785292033106, "ttft_ms": 40.32299993559718, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 61, "facts_retained": 7, "retention_rate": 0.11475409836065574, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.3076923076923077, "status": 0.0, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 690}, "ledger_by_hop": {"1": {"facts_total": 61, "facts_retained": 7, "retention_rate": 0.11475409836065574, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.3076923076923077, "status": 0.0, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 690}}}, "reference_key_id": "keys/synthetic/syn_03015_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "e54cba3528ab78e09d5da03579b2a2035abb56934b1d0571c12acb32bf58b543", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "ce4b48131f9ef2b620ac1101", "attempt_id": "adc5bda92816", "run_group_id": "E4_core:A5|closed|qwen3b-llama3b:h1", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:03:00.821663Z", "task_family": "synthetic", "document_id": "syn_03015_2000", "document_hash": "71a7284b9bc3f681d69d91d18b9e8d45a4dda486cd14f4145395786ac6a68a89", "question_id": "q02_exact_copy", "split": "test", "cluster_id": "syn_03015_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 1, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1922, "query_tokens": 415, "message_tokens": 328, "message_utf8_bytes": 754, "cache_tensor_bytes": null, "serialized_payload_bytes": 754, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 22787.491958006285, "hops": {"1": 22787.491958006285}, "receiver_prefill_ms": 280.5331669514999, "ttft_ms": 40.24770902469754, "decode_ms": 35.72420903947204}, "total_wall_ms": 23144.833791069686, "ttft_ms": 40.24770902469754, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 61, "facts_retained": 7, "retention_rate": 0.11475409836065574, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.3076923076923077, "status": 0.0, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 690}, "ledger_by_hop": {"1": {"facts_total": 61, "facts_retained": 7, "retention_rate": 0.11475409836065574, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.3076923076923077, "status": 0.0, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 690}}}, "reference_key_id": "keys/synthetic/syn_03015_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "e54cba3528ab78e09d5da03579b2a2035abb56934b1d0571c12acb32bf58b543", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "a901e067bbe050a22e9f0022", "attempt_id": "fee476dbaab1", "run_group_id": "E4_core:A5|closed|qwen3b-llama3b:h1", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:03:01.196013Z", "task_family": "synthetic", "document_id": "syn_03015_2000", "document_hash": "71a7284b9bc3f681d69d91d18b9e8d45a4dda486cd14f4145395786ac6a68a89", "question_id": "q03_exact_copy", "split": "test", "cluster_id": "syn_03015_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 1, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1922, "query_tokens": 411, "message_tokens": 328, "message_utf8_bytes": 754, "cache_tensor_bytes": null, "serialized_payload_bytes": 754, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 22787.491958006285, "hops": {"1": 22787.491958006285}, "receiver_prefill_ms": 304.1208341019228, "ttft_ms": 43.37233304977417, "decode_ms": 392.24220800679177}, "total_wall_ms": 23527.709083049558, "ttft_ms": 43.37233304977417, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 6}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "Wilhelmina Dubrovnik", "evidence_artifact": {"ledger": {"facts_total": 61, "facts_retained": 7, "retention_rate": 0.11475409836065574, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.3076923076923077, "status": 0.0, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 690}, "ledger_by_hop": {"1": {"facts_total": 61, "facts_retained": 7, "retention_rate": 0.11475409836065574, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.3076923076923077, "status": 0.0, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 690}}}, "reference_key_id": "keys/synthetic/syn_03015_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "e54cba3528ab78e09d5da03579b2a2035abb56934b1d0571c12acb32bf58b543", "output_hash": "f42f05ebfc96f7258cbfbbb37cd65e114d3d15f869e68e6c95b71fc288c30a32"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "486b18a0cf3fb004602c68e0", "attempt_id": "6cf0571eff8c", "run_group_id": "E4_core:A5|closed|qwen3b-llama3b:h1", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:03:01.953461Z", "task_family": "synthetic", "document_id": "syn_03015_2000", "document_hash": "71a7284b9bc3f681d69d91d18b9e8d45a4dda486cd14f4145395786ac6a68a89", "question_id": "q04_multi_hop", "split": "test", "cluster_id": "syn_03015_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 1, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1922, "query_tokens": 420, "message_tokens": 328, "message_utf8_bytes": 754, "cache_tensor_bytes": null, "serialized_payload_bytes": 754, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 22787.491958006285, "hops": {"1": 22787.491958006285}, "receiver_prefill_ms": 300.4436249611899, "ttft_ms": 42.90658293757588, "decode_ms": 264.9512920761481}, "total_wall_ms": 23396.243958035484, "ttft_ms": 42.90658293757588, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 8}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "8421725 EUR", "evidence_artifact": {"ledger": {"facts_total": 61, "facts_retained": 7, "retention_rate": 0.11475409836065574, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.3076923076923077, "status": 0.0, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 690}, "ledger_by_hop": {"1": {"facts_total": 61, "facts_retained": 7, "retention_rate": 0.11475409836065574, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.3076923076923077, "status": 0.0, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 690}}}, "reference_key_id": "keys/synthetic/syn_03015_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 0.6666666666666666}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "e54cba3528ab78e09d5da03579b2a2035abb56934b1d0571c12acb32bf58b543", "output_hash": "976c8be2c1a175a4fb5643c5785f93e27ad9205cebd9569a64b821e1ae51ac0e"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "832533f21396049dc8be6113", "attempt_id": "17b74606b5b0", "run_group_id": "E4_core:A5|closed|qwen3b-llama3b:h1", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:03:02.579273Z", "task_family": "synthetic", "document_id": "syn_03015_2000", "document_hash": "71a7284b9bc3f681d69d91d18b9e8d45a4dda486cd14f4145395786ac6a68a89", "question_id": "q05_multi_hop", "split": "test", "cluster_id": "syn_03015_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 1, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1922, "query_tokens": 420, "message_tokens": 328, "message_utf8_bytes": 754, "cache_tensor_bytes": null, "serialized_payload_bytes": 754, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 22787.491958006285, "hops": {"1": 22787.491958006285}, "receiver_prefill_ms": 267.4298749770969, "ttft_ms": 42.15274995658547, "decode_ms": 56.16495804861188}, "total_wall_ms": 23153.784999041818, "ttft_ms": 42.15274995658547, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 61, "facts_retained": 7, "retention_rate": 0.11475409836065574, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.3076923076923077, "status": 0.0, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 690}, "ledger_by_hop": {"1": {"facts_total": 61, "facts_retained": 7, "retention_rate": 0.11475409836065574, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.3076923076923077, "status": 0.0, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 690}}}, "reference_key_id": "keys/synthetic/syn_03015_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "e54cba3528ab78e09d5da03579b2a2035abb56934b1d0571c12acb32bf58b543", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "6e443bf6a52789ef16169546", "attempt_id": "3eec00e33c43", "run_group_id": "E4_core:A5|closed|qwen3b-llama3b:h1", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:03:02.963805Z", "task_family": "synthetic", "document_id": "syn_03015_2000", "document_hash": "71a7284b9bc3f681d69d91d18b9e8d45a4dda486cd14f4145395786ac6a68a89", "question_id": "q06_negation", "split": "test", "cluster_id": "syn_03015_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 1, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1922, "query_tokens": 412, "message_tokens": 328, "message_utf8_bytes": 754, "cache_tensor_bytes": null, "serialized_payload_bytes": 754, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 22787.491958006285, "hops": {"1": 22787.491958006285}, "receiver_prefill_ms": 413.36762497667223, "ttft_ms": 32.95337501913309, "decode_ms": 37.42404200602323}, "total_wall_ms": 23271.733958041295, "ttft_ms": 32.95337501913309, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 61, "facts_retained": 7, "retention_rate": 0.11475409836065574, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.3076923076923077, "status": 0.0, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 690}, "ledger_by_hop": {"1": {"facts_total": 61, "facts_retained": 7, "retention_rate": 0.11475409836065574, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.3076923076923077, "status": 0.0, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 690}}}, "reference_key_id": "keys/synthetic/syn_03015_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "e54cba3528ab78e09d5da03579b2a2035abb56934b1d0571c12acb32bf58b543", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "8c22bf478f3f6d571a3037e9", "attempt_id": "17e90d4cb86c", "run_group_id": "E4_core:A5|closed|qwen3b-llama3b:h1", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:03:03.465909Z", "task_family": "synthetic", "document_id": "syn_03015_2000", "document_hash": "71a7284b9bc3f681d69d91d18b9e8d45a4dda486cd14f4145395786ac6a68a89", "question_id": "q07_negation", "split": "test", "cluster_id": "syn_03015_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 1, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1922, "query_tokens": 412, "message_tokens": 328, "message_utf8_bytes": 754, "cache_tensor_bytes": null, "serialized_payload_bytes": 754, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 22787.491958006285, "hops": {"1": 22787.491958006285}, "receiver_prefill_ms": 305.59008300770074, "ttft_ms": 23.331708973273635, "decode_ms": 37.18237509019673}, "total_wall_ms": 23154.086333001032, "ttft_ms": 23.331708973273635, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 61, "facts_retained": 7, "retention_rate": 0.11475409836065574, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.3076923076923077, "status": 0.0, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 690}, "ledger_by_hop": {"1": {"facts_total": 61, "facts_retained": 7, "retention_rate": 0.11475409836065574, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.3076923076923077, "status": 0.0, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 690}}}, "reference_key_id": "keys/synthetic/syn_03015_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "e54cba3528ab78e09d5da03579b2a2035abb56934b1d0571c12acb32bf58b543", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "f29a416b94321ab1c51fd71c", "attempt_id": "a13754e1b830", "run_group_id": "E4_core:A5|closed|qwen3b-llama3b:h1", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:03:03.850848Z", "task_family": "synthetic", "document_id": "syn_03015_2000", "document_hash": "71a7284b9bc3f681d69d91d18b9e8d45a4dda486cd14f4145395786ac6a68a89", "question_id": "q08_revision", "split": "test", "cluster_id": "syn_03015_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 1, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1922, "query_tokens": 426, "message_tokens": 328, "message_utf8_bytes": 754, "cache_tensor_bytes": null, "serialized_payload_bytes": 754, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 22787.491958006285, "hops": {"1": 22787.491958006285}, "receiver_prefill_ms": 287.9949590424076, "ttft_ms": 31.695666955783963, "decode_ms": 26.10854199156165}, "total_wall_ms": 23134.022666024975, "ttft_ms": 31.695666955783963, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 61, "facts_retained": 7, "retention_rate": 0.11475409836065574, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.3076923076923077, "status": 0.0, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 690}, "ledger_by_hop": {"1": {"facts_total": 61, "facts_retained": 7, "retention_rate": 0.11475409836065574, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.3076923076923077, "status": 0.0, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 690}}}, "reference_key_id": "keys/synthetic/syn_03015_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "e54cba3528ab78e09d5da03579b2a2035abb56934b1d0571c12acb32bf58b543", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "6ca560fa068f0930bacd6eaa", "attempt_id": "1bcf8139e043", "run_group_id": "E4_core:A5|closed|qwen3b-llama3b:h1", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:03:04.215878Z", "task_family": "synthetic", "document_id": "syn_03015_2000", "document_hash": "71a7284b9bc3f681d69d91d18b9e8d45a4dda486cd14f4145395786ac6a68a89", "question_id": "q09_revision", "split": "test", "cluster_id": "syn_03015_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 1, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1922, "query_tokens": 420, "message_tokens": 328, "message_utf8_bytes": 754, "cache_tensor_bytes": null, "serialized_payload_bytes": 754, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 22787.491958006285, "hops": {"1": 22787.491958006285}, "receiver_prefill_ms": 230.55620805826038, "ttft_ms": 53.91358397901058, "decode_ms": 487.04858298879117}, "total_wall_ms": 23559.611917007715, "ttft_ms": 53.91358397901058, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 7}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "7028925", "evidence_artifact": {"ledger": {"facts_total": 61, "facts_retained": 7, "retention_rate": 0.11475409836065574, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.3076923076923077, "status": 0.0, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 690}, "ledger_by_hop": {"1": {"facts_total": 61, "facts_retained": 7, "retention_rate": 0.11475409836065574, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.3076923076923077, "status": 0.0, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 690}}}, "reference_key_id": "keys/synthetic/syn_03015_2000.json", "task_scores": {"correct": 0.0, "abstained": 0.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "e54cba3528ab78e09d5da03579b2a2035abb56934b1d0571c12acb32bf58b543", "output_hash": "e966a398855c104a86bce07742c3ce8108a432926bc704940ecfd07e44596774"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "f1f031ebc9c92ff7731af880", "attempt_id": "f3933e18281d", "run_group_id": "E4_core:A5|closed|qwen3b-llama3b:h1", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:03:05.005696Z", "task_family": "synthetic", "document_id": "syn_03015_2000", "document_hash": "71a7284b9bc3f681d69d91d18b9e8d45a4dda486cd14f4145395786ac6a68a89", "question_id": "q10_temporal", "split": "test", "cluster_id": "syn_03015_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 1, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1922, "query_tokens": 429, "message_tokens": 328, "message_utf8_bytes": 754, "cache_tensor_bytes": null, "serialized_payload_bytes": 754, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 22787.491958006285, "hops": {"1": 22787.491958006285}, "receiver_prefill_ms": 308.36004205048084, "ttft_ms": 34.698500065132976, "decode_ms": 33.809500047937036}, "total_wall_ms": 23164.92270794697, "ttft_ms": 34.698500065132976, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 61, "facts_retained": 7, "retention_rate": 0.11475409836065574, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.3076923076923077, "status": 0.0, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 690}, "ledger_by_hop": {"1": {"facts_total": 61, "facts_retained": 7, "retention_rate": 0.11475409836065574, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.3076923076923077, "status": 0.0, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 690}}}, "reference_key_id": "keys/synthetic/syn_03015_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "e54cba3528ab78e09d5da03579b2a2035abb56934b1d0571c12acb32bf58b543", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "3c92e2c235ded443e75d1c11", "attempt_id": "fdcd40cff258", "run_group_id": "E4_core:A5|closed|qwen3b-llama3b:h1", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:03:05.400496Z", "task_family": "synthetic", "document_id": "syn_03015_2000", "document_hash": "71a7284b9bc3f681d69d91d18b9e8d45a4dda486cd14f4145395786ac6a68a89", "question_id": "q11_aggregation", "split": "test", "cluster_id": "syn_03015_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 1, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1922, "query_tokens": 422, "message_tokens": 328, "message_utf8_bytes": 754, "cache_tensor_bytes": null, "serialized_payload_bytes": 754, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 22787.491958006285, "hops": {"1": 22787.491958006285}, "receiver_prefill_ms": 298.560458002612, "ttft_ms": 38.59216603450477, "decode_ms": 30.8410000288859}, "total_wall_ms": 23156.131708063185, "ttft_ms": 38.59216603450477, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 61, "facts_retained": 7, "retention_rate": 0.11475409836065574, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.3076923076923077, "status": 0.0, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 690}, "ledger_by_hop": {"1": {"facts_total": 61, "facts_retained": 7, "retention_rate": 0.11475409836065574, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.3076923076923077, "status": 0.0, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 690}}}, "reference_key_id": "keys/synthetic/syn_03015_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "e54cba3528ab78e09d5da03579b2a2035abb56934b1d0571c12acb32bf58b543", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "ea6a70adbdadcc5e6fbde268", "attempt_id": "e7c3f971b8c4", "run_group_id": "E4_core:A5|closed|qwen3b-llama3b:h1", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:03:05.786844Z", "task_family": "synthetic", "document_id": "syn_03015_2000", "document_hash": "71a7284b9bc3f681d69d91d18b9e8d45a4dda486cd14f4145395786ac6a68a89", "question_id": "q12_unit_conversion", "split": "test", "cluster_id": "syn_03015_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 1, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1922, "query_tokens": 419, "message_tokens": 328, "message_utf8_bytes": 754, "cache_tensor_bytes": null, "serialized_payload_bytes": 754, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 22787.491958006285, "hops": {"1": 22787.491958006285}, "receiver_prefill_ms": 277.7369589312002, "ttft_ms": 61.69312493875623, "decode_ms": 58.09158307965845}, "total_wall_ms": 23185.53220795002, "ttft_ms": 61.69312493875623, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 61, "facts_retained": 7, "retention_rate": 0.11475409836065574, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.3076923076923077, "status": 0.0, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 690}, "ledger_by_hop": {"1": {"facts_total": 61, "facts_retained": 7, "retention_rate": 0.11475409836065574, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.3076923076923077, "status": 0.0, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 690}}}, "reference_key_id": "keys/synthetic/syn_03015_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "e54cba3528ab78e09d5da03579b2a2035abb56934b1d0571c12acb32bf58b543", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "26b0b9dd46a404762dc7035c", "attempt_id": "bee7ef52bcfa", "run_group_id": "E4_core:A5|closed|qwen3b-llama3b:h1", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:03:06.202154Z", "task_family": "synthetic", "document_id": "syn_03015_2000", "document_hash": "71a7284b9bc3f681d69d91d18b9e8d45a4dda486cd14f4145395786ac6a68a89", "question_id": "q13_unanswerable", "split": "test", "cluster_id": "syn_03015_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 1, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1922, "query_tokens": 423, "message_tokens": 328, "message_utf8_bytes": 754, "cache_tensor_bytes": null, "serialized_payload_bytes": 754, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 22787.491958006285, "hops": {"1": 22787.491958006285}, "receiver_prefill_ms": 447.81958300154656, "ttft_ms": 38.87812502216548, "decode_ms": 24.980374961160123}, "total_wall_ms": 23299.78833300993, "ttft_ms": 38.87812502216548, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 61, "facts_retained": 7, "retention_rate": 0.11475409836065574, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.3076923076923077, "status": 0.0, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 690}, "ledger_by_hop": {"1": {"facts_total": 61, "facts_retained": 7, "retention_rate": 0.11475409836065574, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.3076923076923077, "status": 0.0, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 690}}}, "reference_key_id": "keys/synthetic/syn_03015_2000.json", "task_scores": {"correct": 1.0, "abstained": 1.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "e54cba3528ab78e09d5da03579b2a2035abb56934b1d0571c12acb32bf58b543", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "6b989ec13e67d63d79c9b9a0", "attempt_id": "531117f31237", "run_group_id": "E4_core:A5|closed|qwen3b-llama3b:h1", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:03:06.732887Z", "task_family": "synthetic", "document_id": "syn_03015_2000", "document_hash": "71a7284b9bc3f681d69d91d18b9e8d45a4dda486cd14f4145395786ac6a68a89", "question_id": "q14_unanswerable", "split": "test", "cluster_id": "syn_03015_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 1, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1922, "query_tokens": 424, "message_tokens": 328, "message_utf8_bytes": 754, "cache_tensor_bytes": null, "serialized_payload_bytes": 754, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 22787.491958006285, "hops": {"1": 22787.491958006285}, "receiver_prefill_ms": 300.26383290532976, "ttft_ms": 38.71024993713945, "decode_ms": 37.67654101829976}, "total_wall_ms": 23164.638499030843, "ttft_ms": 38.71024993713945, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 61, "facts_retained": 7, "retention_rate": 0.11475409836065574, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.3076923076923077, "status": 0.0, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 690}, "ledger_by_hop": {"1": {"facts_total": 61, "facts_retained": 7, "retention_rate": 0.11475409836065574, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.3076923076923077, "status": 0.0, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 690}}}, "reference_key_id": "keys/synthetic/syn_03015_2000.json", "task_scores": {"correct": 1.0, "abstained": 1.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "e54cba3528ab78e09d5da03579b2a2035abb56934b1d0571c12acb32bf58b543", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "4cd45d98884522c2d048781c", "attempt_id": "56cc7f3a1ca3", "run_group_id": "E4_core:A5|closed|qwen3b-llama3b:h2", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:03:18.554617Z", "task_family": "synthetic", "document_id": "syn_03015_2000", "document_hash": "71a7284b9bc3f681d69d91d18b9e8d45a4dda486cd14f4145395786ac6a68a89", "question_id": "q00_single_fact", "split": "test", "cluster_id": "syn_03015_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "tokenizer_hashes": {"receiver": "b9d8872cf3a99da99e1fc4b0f71232a451632dc49ca4c6771549b4efa5c1c763"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 2, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1741, "query_tokens": 392, "message_tokens": 288, "message_utf8_bytes": 811, "cache_tensor_bytes": null, "serialized_payload_bytes": 811, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 34213.04958302062, "hops": {"1": 22787.491958006285, "2": 11425.557625014335}, "receiver_prefill_ms": 359.44337502587587, "ttft_ms": 50.976915983483195, "decode_ms": 104.9944160040468}, "total_wall_ms": 34729.04558305163, "ttft_ms": 50.976915983483195, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 61, "facts_retained": 7, "retention_rate": 0.11475409836065574, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.3076923076923077, "status": 0.0, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 811}, "ledger_by_hop": {"1": {"facts_total": 61, "facts_retained": 7, "retention_rate": 0.11475409836065574, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.3076923076923077, "status": 0.0, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 690}, "2": {"facts_total": 61, "facts_retained": 7, "retention_rate": 0.11475409836065574, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.3076923076923077, "status": 0.0, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 811}}}, "reference_key_id": "keys/synthetic/syn_03015_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "4f973bd7c40fcf07e373e29e86c8315cdbd7790f931726a9494ca5cf77a0b82e", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "239755cf1d563e0894960d5b", "attempt_id": "930d68d0f6c7", "run_group_id": "E4_core:A5|closed|qwen3b-llama3b:h2", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:03:19.090013Z", "task_family": "synthetic", "document_id": "syn_03015_2000", "document_hash": "71a7284b9bc3f681d69d91d18b9e8d45a4dda486cd14f4145395786ac6a68a89", "question_id": "q01_single_fact", "split": "test", "cluster_id": "syn_03015_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "tokenizer_hashes": {"receiver": "b9d8872cf3a99da99e1fc4b0f71232a451632dc49ca4c6771549b4efa5c1c763"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 2, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1741, "query_tokens": 396, "message_tokens": 288, "message_utf8_bytes": 811, "cache_tensor_bytes": null, "serialized_payload_bytes": 811, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 34213.04958302062, "hops": {"1": 22787.491958006285, "2": 11425.557625014335}, "receiver_prefill_ms": 411.5952920401469, "ttft_ms": 23.810832994058728, "decode_ms": 38.43691700603813}, "total_wall_ms": 34687.56537500303, "ttft_ms": 23.810832994058728, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 61, "facts_retained": 7, "retention_rate": 0.11475409836065574, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.3076923076923077, "status": 0.0, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 811}, "ledger_by_hop": {"1": {"facts_total": 61, "facts_retained": 7, "retention_rate": 0.11475409836065574, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.3076923076923077, "status": 0.0, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 690}, "2": {"facts_total": 61, "facts_retained": 7, "retention_rate": 0.11475409836065574, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.3076923076923077, "status": 0.0, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 811}}}, "reference_key_id": "keys/synthetic/syn_03015_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "4f973bd7c40fcf07e373e29e86c8315cdbd7790f931726a9494ca5cf77a0b82e", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "1ab5f31c3f10cf3164f99f23", "attempt_id": "395f5c351df5", "run_group_id": "E4_core:A5|closed|qwen3b-llama3b:h2", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:03:19.584722Z", "task_family": "synthetic", "document_id": "syn_03015_2000", "document_hash": "71a7284b9bc3f681d69d91d18b9e8d45a4dda486cd14f4145395786ac6a68a89", "question_id": "q02_exact_copy", "split": "test", "cluster_id": "syn_03015_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "tokenizer_hashes": {"receiver": "b9d8872cf3a99da99e1fc4b0f71232a451632dc49ca4c6771549b4efa5c1c763"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 2, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1741, "query_tokens": 396, "message_tokens": 288, "message_utf8_bytes": 811, "cache_tensor_bytes": null, "serialized_payload_bytes": 811, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 34213.04958302062, "hops": {"1": 22787.491958006285, "2": 11425.557625014335}, "receiver_prefill_ms": 361.710709054023, "ttft_ms": 19.687958993017673, "decode_ms": 31.945333001203835}, "total_wall_ms": 34626.81812501978, "ttft_ms": 19.687958993017673, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 61, "facts_retained": 7, "retention_rate": 0.11475409836065574, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.3076923076923077, "status": 0.0, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 811}, "ledger_by_hop": {"1": {"facts_total": 61, "facts_retained": 7, "retention_rate": 0.11475409836065574, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.3076923076923077, "status": 0.0, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 690}, "2": {"facts_total": 61, "facts_retained": 7, "retention_rate": 0.11475409836065574, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.3076923076923077, "status": 0.0, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 811}}}, "reference_key_id": "keys/synthetic/syn_03015_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "4f973bd7c40fcf07e373e29e86c8315cdbd7790f931726a9494ca5cf77a0b82e", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "ba1670f6fa782330abf2d131", "attempt_id": "385597537c79", "run_group_id": "E4_core:A5|closed|qwen3b-llama3b:h2", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:03:20.016113Z", "task_family": "synthetic", "document_id": "syn_03015_2000", "document_hash": "71a7284b9bc3f681d69d91d18b9e8d45a4dda486cd14f4145395786ac6a68a89", "question_id": "q03_exact_copy", "split": "test", "cluster_id": "syn_03015_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "tokenizer_hashes": {"receiver": "b9d8872cf3a99da99e1fc4b0f71232a451632dc49ca4c6771549b4efa5c1c763"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 2, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1741, "query_tokens": 392, "message_tokens": 288, "message_utf8_bytes": 811, "cache_tensor_bytes": null, "serialized_payload_bytes": 811, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 34213.04958302062, "hops": {"1": 22787.491958006285, "2": 11425.557625014335}, "receiver_prefill_ms": 293.55500009842217, "ttft_ms": 38.04591705556959, "decode_ms": 443.9430000493303}, "total_wall_ms": 34989.11362502258, "ttft_ms": 38.04591705556959, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 6}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "Wilhelmina Dubrovnik", "evidence_artifact": {"ledger": {"facts_total": 61, "facts_retained": 7, "retention_rate": 0.11475409836065574, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.3076923076923077, "status": 0.0, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 811}, "ledger_by_hop": {"1": {"facts_total": 61, "facts_retained": 7, "retention_rate": 0.11475409836065574, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.3076923076923077, "status": 0.0, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 690}, "2": {"facts_total": 61, "facts_retained": 7, "retention_rate": 0.11475409836065574, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.3076923076923077, "status": 0.0, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 811}}}, "reference_key_id": "keys/synthetic/syn_03015_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "4f973bd7c40fcf07e373e29e86c8315cdbd7790f931726a9494ca5cf77a0b82e", "output_hash": "f42f05ebfc96f7258cbfbbb37cd65e114d3d15f869e68e6c95b71fc288c30a32"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "cccb3537cf68823bf226dc9b", "attempt_id": "586d2ee3ca3f", "run_group_id": "E4_core:A5|closed|qwen3b-llama3b:h2", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:03:20.810098Z", "task_family": "synthetic", "document_id": "syn_03015_2000", "document_hash": "71a7284b9bc3f681d69d91d18b9e8d45a4dda486cd14f4145395786ac6a68a89", "question_id": "q04_multi_hop", "split": "test", "cluster_id": "syn_03015_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "tokenizer_hashes": {"receiver": "b9d8872cf3a99da99e1fc4b0f71232a451632dc49ca4c6771549b4efa5c1c763"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 2, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1741, "query_tokens": 401, "message_tokens": 288, "message_utf8_bytes": 811, "cache_tensor_bytes": null, "serialized_payload_bytes": 811, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 34213.04958302062, "hops": {"1": 22787.491958006285, "2": 11425.557625014335}, "receiver_prefill_ms": 348.2957920059562, "ttft_ms": 19.390083965845406, "decode_ms": 155.04870901349932}, "total_wall_ms": 34736.22883297503, "ttft_ms": 19.390083965845406, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 5}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "8,421,725", "evidence_artifact": {"ledger": {"facts_total": 61, "facts_retained": 7, "retention_rate": 0.11475409836065574, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.3076923076923077, "status": 0.0, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 811}, "ledger_by_hop": {"1": {"facts_total": 61, "facts_retained": 7, "retention_rate": 0.11475409836065574, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.3076923076923077, "status": 0.0, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 690}, "2": {"facts_total": 61, "facts_retained": 7, "retention_rate": 0.11475409836065574, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.3076923076923077, "status": 0.0, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 811}}}, "reference_key_id": "keys/synthetic/syn_03015_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "4f973bd7c40fcf07e373e29e86c8315cdbd7790f931726a9494ca5cf77a0b82e", "output_hash": "6e31df6d4fb38417a5210ded39fc1e4c0e32f278ab9cba0da5b8e7585ddeb534"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "276a490864d06c6d419331c3", "attempt_id": "a42d83a5e97e", "run_group_id": "E4_core:A5|closed|qwen3b-llama3b:h2", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:03:21.351015Z", "task_family": "synthetic", "document_id": "syn_03015_2000", "document_hash": "71a7284b9bc3f681d69d91d18b9e8d45a4dda486cd14f4145395786ac6a68a89", "question_id": "q05_multi_hop", "split": "test", "cluster_id": "syn_03015_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "tokenizer_hashes": {"receiver": "b9d8872cf3a99da99e1fc4b0f71232a451632dc49ca4c6771549b4efa5c1c763"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 2, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1741, "query_tokens": 399, "message_tokens": 288, "message_utf8_bytes": 811, "cache_tensor_bytes": null, "serialized_payload_bytes": 811, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 34213.04958302062, "hops": {"1": 22787.491958006285, "2": 11425.557625014335}, "receiver_prefill_ms": 330.775290960446, "ttft_ms": 19.48016695678234, "decode_ms": 15.548084047622979}, "total_wall_ms": 34579.38266603742, "ttft_ms": 19.48016695678234, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 61, "facts_retained": 7, "retention_rate": 0.11475409836065574, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.3076923076923077, "status": 0.0, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 811}, "ledger_by_hop": {"1": {"facts_total": 61, "facts_retained": 7, "retention_rate": 0.11475409836065574, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.3076923076923077, "status": 0.0, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 690}, "2": {"facts_total": 61, "facts_retained": 7, "retention_rate": 0.11475409836065574, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.3076923076923077, "status": 0.0, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 811}}}, "reference_key_id": "keys/synthetic/syn_03015_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "4f973bd7c40fcf07e373e29e86c8315cdbd7790f931726a9494ca5cf77a0b82e", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "9e7c3e85aafd4a9f27ca3682", "attempt_id": "6912610d08c3", "run_group_id": "E4_core:A5|closed|qwen3b-llama3b:h2", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:03:21.734827Z", "task_family": "synthetic", "document_id": "syn_03015_2000", "document_hash": "71a7284b9bc3f681d69d91d18b9e8d45a4dda486cd14f4145395786ac6a68a89", "question_id": "q06_negation", "split": "test", "cluster_id": "syn_03015_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "tokenizer_hashes": {"receiver": "b9d8872cf3a99da99e1fc4b0f71232a451632dc49ca4c6771549b4efa5c1c763"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 2, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1741, "query_tokens": 393, "message_tokens": 288, "message_utf8_bytes": 811, "cache_tensor_bytes": null, "serialized_payload_bytes": 811, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 34213.04958302062, "hops": {"1": 22787.491958006285, "2": 11425.557625014335}, "receiver_prefill_ms": 321.5405000373721, "ttft_ms": 43.40829106513411, "decode_ms": 58.21549997199327}, "total_wall_ms": 34637.04808300827, "ttft_ms": 43.40829106513411, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 61, "facts_retained": 7, "retention_rate": 0.11475409836065574, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.3076923076923077, "status": 0.0, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 811}, "ledger_by_hop": {"1": {"facts_total": 61, "facts_retained": 7, "retention_rate": 0.11475409836065574, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.3076923076923077, "status": 0.0, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 690}, "2": {"facts_total": 61, "facts_retained": 7, "retention_rate": 0.11475409836065574, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.3076923076923077, "status": 0.0, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 811}}}, "reference_key_id": "keys/synthetic/syn_03015_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "4f973bd7c40fcf07e373e29e86c8315cdbd7790f931726a9494ca5cf77a0b82e", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "61088f0e796dca9d980634cf", "attempt_id": "8e39c6873bb4", "run_group_id": "E4_core:A5|closed|qwen3b-llama3b:h2", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:03:22.178441Z", "task_family": "synthetic", "document_id": "syn_03015_2000", "document_hash": "71a7284b9bc3f681d69d91d18b9e8d45a4dda486cd14f4145395786ac6a68a89", "question_id": "q07_negation", "split": "test", "cluster_id": "syn_03015_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "tokenizer_hashes": {"receiver": "b9d8872cf3a99da99e1fc4b0f71232a451632dc49ca4c6771549b4efa5c1c763"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 2, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1741, "query_tokens": 393, "message_tokens": 288, "message_utf8_bytes": 811, "cache_tensor_bytes": null, "serialized_payload_bytes": 811, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 34213.04958302062, "hops": {"1": 22787.491958006285, "2": 11425.557625014335}, "receiver_prefill_ms": 456.6399999894202, "ttft_ms": 33.96000002976507, "decode_ms": 30.166749958880246}, "total_wall_ms": 34734.30970800109, "ttft_ms": 33.96000002976507, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 61, "facts_retained": 7, "retention_rate": 0.11475409836065574, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.3076923076923077, "status": 0.0, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 811}, "ledger_by_hop": {"1": {"facts_total": 61, "facts_retained": 7, "retention_rate": 0.11475409836065574, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.3076923076923077, "status": 0.0, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 690}, "2": {"facts_total": 61, "facts_retained": 7, "retention_rate": 0.11475409836065574, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.3076923076923077, "status": 0.0, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 811}}}, "reference_key_id": "keys/synthetic/syn_03015_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "4f973bd7c40fcf07e373e29e86c8315cdbd7790f931726a9494ca5cf77a0b82e", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "e43059d4ab1671a9e082bdd3", "attempt_id": "1acbfa1c4716", "run_group_id": "E4_core:A5|closed|qwen3b-llama3b:h2", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:03:22.717602Z", "task_family": "synthetic", "document_id": "syn_03015_2000", "document_hash": "71a7284b9bc3f681d69d91d18b9e8d45a4dda486cd14f4145395786ac6a68a89", "question_id": "q08_revision", "split": "test", "cluster_id": "syn_03015_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "tokenizer_hashes": {"receiver": "b9d8872cf3a99da99e1fc4b0f71232a451632dc49ca4c6771549b4efa5c1c763"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 2, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1741, "query_tokens": 406, "message_tokens": 288, "message_utf8_bytes": 811, "cache_tensor_bytes": null, "serialized_payload_bytes": 811, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 34213.04958302062, "hops": {"1": 22787.491958006285, "2": 11425.557625014335}, "receiver_prefill_ms": 317.33462505508214, "ttft_ms": 38.76845899503678, "decode_ms": 22.36887498293072}, "total_wall_ms": 34592.031957930885, "ttft_ms": 38.76845899503678, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 61, "facts_retained": 7, "retention_rate": 0.11475409836065574, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.3076923076923077, "status": 0.0, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 811}, "ledger_by_hop": {"1": {"facts_total": 61, "facts_retained": 7, "retention_rate": 0.11475409836065574, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.3076923076923077, "status": 0.0, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 690}, "2": {"facts_total": 61, "facts_retained": 7, "retention_rate": 0.11475409836065574, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.3076923076923077, "status": 0.0, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 811}}}, "reference_key_id": "keys/synthetic/syn_03015_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "4f973bd7c40fcf07e373e29e86c8315cdbd7790f931726a9494ca5cf77a0b82e", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "1a086baf6e18b9a4eee97267", "attempt_id": "6d4bf5e2169c", "run_group_id": "E4_core:A5|closed|qwen3b-llama3b:h2", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:03:23.114843Z", "task_family": "synthetic", "document_id": "syn_03015_2000", "document_hash": "71a7284b9bc3f681d69d91d18b9e8d45a4dda486cd14f4145395786ac6a68a89", "question_id": "q09_revision", "split": "test", "cluster_id": "syn_03015_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "tokenizer_hashes": {"receiver": "b9d8872cf3a99da99e1fc4b0f71232a451632dc49ca4c6771549b4efa5c1c763"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 2, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1741, "query_tokens": 400, "message_tokens": 288, "message_utf8_bytes": 811, "cache_tensor_bytes": null, "serialized_payload_bytes": 811, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 34213.04958302062, "hops": {"1": 22787.491958006285, "2": 11425.557625014335}, "receiver_prefill_ms": 293.03125001024455, "ttft_ms": 52.19087493605912, "decode_ms": 281.6588329151273}, "total_wall_ms": 34840.54212504998, "ttft_ms": 52.19087493605912, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 5}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "7,028,925", "evidence_artifact": {"ledger": {"facts_total": 61, "facts_retained": 7, "retention_rate": 0.11475409836065574, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.3076923076923077, "status": 0.0, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 811}, "ledger_by_hop": {"1": {"facts_total": 61, "facts_retained": 7, "retention_rate": 0.11475409836065574, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.3076923076923077, "status": 0.0, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 690}, "2": {"facts_total": 61, "facts_retained": 7, "retention_rate": 0.11475409836065574, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.3076923076923077, "status": 0.0, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 811}}}, "reference_key_id": "keys/synthetic/syn_03015_2000.json", "task_scores": {"correct": 0.0, "abstained": 0.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "4f973bd7c40fcf07e373e29e86c8315cdbd7790f931726a9494ca5cf77a0b82e", "output_hash": "e83edb3f38a69e604be343e63e2e84442156fcb2bb30f36ba9716575d5ffd57a"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "59503cab7490cf17a5991f33", "attempt_id": "885a6f2615bc", "run_group_id": "E4_core:A5|closed|qwen3b-llama3b:h2", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:03:23.759594Z", "task_family": "synthetic", "document_id": "syn_03015_2000", "document_hash": "71a7284b9bc3f681d69d91d18b9e8d45a4dda486cd14f4145395786ac6a68a89", "question_id": "q10_temporal", "split": "test", "cluster_id": "syn_03015_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "tokenizer_hashes": {"receiver": "b9d8872cf3a99da99e1fc4b0f71232a451632dc49ca4c6771549b4efa5c1c763"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 2, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1741, "query_tokens": 409, "message_tokens": 288, "message_utf8_bytes": 811, "cache_tensor_bytes": null, "serialized_payload_bytes": 811, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 34213.04958302062, "hops": {"1": 22787.491958006285, "2": 11425.557625014335}, "receiver_prefill_ms": 455.2802500547841, "ttft_ms": 37.42616600356996, "decode_ms": 24.374332977458835}, "total_wall_ms": 34730.56345793884, "ttft_ms": 37.42616600356996, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 61, "facts_retained": 7, "retention_rate": 0.11475409836065574, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.3076923076923077, "status": 0.0, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 811}, "ledger_by_hop": {"1": {"facts_total": 61, "facts_retained": 7, "retention_rate": 0.11475409836065574, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.3076923076923077, "status": 0.0, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 690}, "2": {"facts_total": 61, "facts_retained": 7, "retention_rate": 0.11475409836065574, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.3076923076923077, "status": 0.0, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 811}}}, "reference_key_id": "keys/synthetic/syn_03015_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "4f973bd7c40fcf07e373e29e86c8315cdbd7790f931726a9494ca5cf77a0b82e", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "67ac899280affed06b923b6d", "attempt_id": "de1cfb1d1fd9", "run_group_id": "E4_core:A5|closed|qwen3b-llama3b:h2", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:03:24.294469Z", "task_family": "synthetic", "document_id": "syn_03015_2000", "document_hash": "71a7284b9bc3f681d69d91d18b9e8d45a4dda486cd14f4145395786ac6a68a89", "question_id": "q11_aggregation", "split": "test", "cluster_id": "syn_03015_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "tokenizer_hashes": {"receiver": "b9d8872cf3a99da99e1fc4b0f71232a451632dc49ca4c6771549b4efa5c1c763"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 2, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1741, "query_tokens": 403, "message_tokens": 288, "message_utf8_bytes": 811, "cache_tensor_bytes": null, "serialized_payload_bytes": 811, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 34213.04958302062, "hops": {"1": 22787.491958006285, "2": 11425.557625014335}, "receiver_prefill_ms": 353.06870797649026, "ttft_ms": 26.468042051419616, "decode_ms": 35.620792070403695}, "total_wall_ms": 34628.642500028946, "ttft_ms": 26.468042051419616, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 61, "facts_retained": 7, "retention_rate": 0.11475409836065574, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.3076923076923077, "status": 0.0, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 811}, "ledger_by_hop": {"1": {"facts_total": 61, "facts_retained": 7, "retention_rate": 0.11475409836065574, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.3076923076923077, "status": 0.0, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 690}, "2": {"facts_total": 61, "facts_retained": 7, "retention_rate": 0.11475409836065574, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.3076923076923077, "status": 0.0, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 811}}}, "reference_key_id": "keys/synthetic/syn_03015_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "4f973bd7c40fcf07e373e29e86c8315cdbd7790f931726a9494ca5cf77a0b82e", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "12257072790b86575ffbcb3c", "attempt_id": "c437868403d3", "run_group_id": "E4_core:A5|closed|qwen3b-llama3b:h2", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:03:24.726873Z", "task_family": "synthetic", "document_id": "syn_03015_2000", "document_hash": "71a7284b9bc3f681d69d91d18b9e8d45a4dda486cd14f4145395786ac6a68a89", "question_id": "q12_unit_conversion", "split": "test", "cluster_id": "syn_03015_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "tokenizer_hashes": {"receiver": "b9d8872cf3a99da99e1fc4b0f71232a451632dc49ca4c6771549b4efa5c1c763"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 2, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1741, "query_tokens": 400, "message_tokens": 288, "message_utf8_bytes": 811, "cache_tensor_bytes": null, "serialized_payload_bytes": 811, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 34213.04958302062, "hops": {"1": 22787.491958006285, "2": 11425.557625014335}, "receiver_prefill_ms": 315.1242920430377, "ttft_ms": 46.473124995827675, "decode_ms": 40.2259579859674}, "total_wall_ms": 34615.36262498703, "ttft_ms": 46.473124995827675, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 61, "facts_retained": 7, "retention_rate": 0.11475409836065574, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.3076923076923077, "status": 0.0, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 811}, "ledger_by_hop": {"1": {"facts_total": 61, "facts_retained": 7, "retention_rate": 0.11475409836065574, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.3076923076923077, "status": 0.0, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 690}, "2": {"facts_total": 61, "facts_retained": 7, "retention_rate": 0.11475409836065574, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.3076923076923077, "status": 0.0, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 811}}}, "reference_key_id": "keys/synthetic/syn_03015_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "4f973bd7c40fcf07e373e29e86c8315cdbd7790f931726a9494ca5cf77a0b82e", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "7e9b84e105d1360144cc36c5", "attempt_id": "4652f1f17ae9", "run_group_id": "E4_core:A5|closed|qwen3b-llama3b:h2", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:03:25.146095Z", "task_family": "synthetic", "document_id": "syn_03015_2000", "document_hash": "71a7284b9bc3f681d69d91d18b9e8d45a4dda486cd14f4145395786ac6a68a89", "question_id": "q13_unanswerable", "split": "test", "cluster_id": "syn_03015_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "tokenizer_hashes": {"receiver": "b9d8872cf3a99da99e1fc4b0f71232a451632dc49ca4c6771549b4efa5c1c763"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 2, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1741, "query_tokens": 403, "message_tokens": 288, "message_utf8_bytes": 811, "cache_tensor_bytes": null, "serialized_payload_bytes": 811, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 34213.04958302062, "hops": {"1": 22787.491958006285, "2": 11425.557625014335}, "receiver_prefill_ms": 495.51966693252325, "ttft_ms": 23.941374965943396, "decode_ms": 39.38354202546179}, "total_wall_ms": 34772.485083085485, "ttft_ms": 23.941374965943396, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 61, "facts_retained": 7, "retention_rate": 0.11475409836065574, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.3076923076923077, "status": 0.0, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 811}, "ledger_by_hop": {"1": {"facts_total": 61, "facts_retained": 7, "retention_rate": 0.11475409836065574, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.3076923076923077, "status": 0.0, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 690}, "2": {"facts_total": 61, "facts_retained": 7, "retention_rate": 0.11475409836065574, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.3076923076923077, "status": 0.0, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 811}}}, "reference_key_id": "keys/synthetic/syn_03015_2000.json", "task_scores": {"correct": 1.0, "abstained": 1.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "4f973bd7c40fcf07e373e29e86c8315cdbd7790f931726a9494ca5cf77a0b82e", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "48d9dfbd2f97531a435ccfe8", "attempt_id": "f95fe269339f", "run_group_id": "E4_core:A5|closed|qwen3b-llama3b:h2", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:03:25.722639Z", "task_family": "synthetic", "document_id": "syn_03015_2000", "document_hash": "71a7284b9bc3f681d69d91d18b9e8d45a4dda486cd14f4145395786ac6a68a89", "question_id": "q14_unanswerable", "split": "test", "cluster_id": "syn_03015_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "tokenizer_hashes": {"receiver": "b9d8872cf3a99da99e1fc4b0f71232a451632dc49ca4c6771549b4efa5c1c763"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 2, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1741, "query_tokens": 405, "message_tokens": 288, "message_utf8_bytes": 811, "cache_tensor_bytes": null, "serialized_payload_bytes": 811, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 34213.04958302062, "hops": {"1": 22787.491958006285, "2": 11425.557625014335}, "receiver_prefill_ms": 325.19337499979883, "ttft_ms": 38.280292064882815, "decode_ms": 31.63741691969335}, "total_wall_ms": 34608.7464580778, "ttft_ms": 38.280292064882815, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 61, "facts_retained": 7, "retention_rate": 0.11475409836065574, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.3076923076923077, "status": 0.0, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 811}, "ledger_by_hop": {"1": {"facts_total": 61, "facts_retained": 7, "retention_rate": 0.11475409836065574, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.3076923076923077, "status": 0.0, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 690}, "2": {"facts_total": 61, "facts_retained": 7, "retention_rate": 0.11475409836065574, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.3076923076923077, "status": 0.0, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 811}}}, "reference_key_id": "keys/synthetic/syn_03015_2000.json", "task_scores": {"correct": 1.0, "abstained": 1.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "4f973bd7c40fcf07e373e29e86c8315cdbd7790f931726a9494ca5cf77a0b82e", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "4eb37219a508bf1ff80dfc0f", "attempt_id": "bb0d7760b728", "run_group_id": "E4_core:A5|closed|qwen3b-llama3b:h4", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:03:52.021869Z", "task_family": "synthetic", "document_id": "syn_03015_2000", "document_hash": "71a7284b9bc3f681d69d91d18b9e8d45a4dda486cd14f4145395786ac6a68a89", "question_id": "q00_single_fact", "split": "test", "cluster_id": "syn_03015_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "tokenizer_hashes": {"receiver": "b9d8872cf3a99da99e1fc4b0f71232a451632dc49ca4c6771549b4efa5c1c763"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 4, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1741, "query_tokens": 392, "message_tokens": 288, "message_utf8_bytes": 811, "cache_tensor_bytes": null, "serialized_payload_bytes": 811, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 60096.57245792914, "hops": {"1": 22787.491958006285, "2": 11425.557625014335, "3": 14560.2815420134, "4": 11323.241332895122}, "receiver_prefill_ms": 339.4385409774259, "ttft_ms": 22.150499979034066, "decode_ms": 34.69612495973706}, "total_wall_ms": 60493.29145788215, "ttft_ms": 22.150499979034066, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 61, "facts_retained": 7, "retention_rate": 0.11475409836065574, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.3076923076923077, "status": 0.0, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 811}, "ledger_by_hop": {"1": {"facts_total": 61, "facts_retained": 7, "retention_rate": 0.11475409836065574, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.3076923076923077, "status": 0.0, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 690}, "2": {"facts_total": 61, "facts_retained": 7, "retention_rate": 0.11475409836065574, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.3076923076923077, "status": 0.0, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 811}, "3": {"facts_total": 61, "facts_retained": 7, "retention_rate": 0.11475409836065574, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.3076923076923077, "status": 0.0, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 811}, "4": {"facts_total": 61, "facts_retained": 7, "retention_rate": 0.11475409836065574, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.3076923076923077, "status": 0.0, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 811}}}, "reference_key_id": "keys/synthetic/syn_03015_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "4f973bd7c40fcf07e373e29e86c8315cdbd7790f931726a9494ca5cf77a0b82e", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "9384e8782ae961ad0bfa9cae", "attempt_id": "9c2ba4b80590", "run_group_id": "E4_core:A5|closed|qwen3b-llama3b:h4", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:03:52.436023Z", "task_family": "synthetic", "document_id": "syn_03015_2000", "document_hash": "71a7284b9bc3f681d69d91d18b9e8d45a4dda486cd14f4145395786ac6a68a89", "question_id": "q01_single_fact", "split": "test", "cluster_id": "syn_03015_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "tokenizer_hashes": {"receiver": "b9d8872cf3a99da99e1fc4b0f71232a451632dc49ca4c6771549b4efa5c1c763"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 4, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1741, "query_tokens": 396, "message_tokens": 288, "message_utf8_bytes": 811, "cache_tensor_bytes": null, "serialized_payload_bytes": 811, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 60096.57245792914, "hops": {"1": 22787.491958006285, "2": 11425.557625014335, "3": 14560.2815420134, "4": 11323.241332895122}, "receiver_prefill_ms": 309.25866693723947, "ttft_ms": 35.24116694461554, "decode_ms": 59.28154196590185}, "total_wall_ms": 60501.01195788011, "ttft_ms": 35.24116694461554, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 61, "facts_retained": 7, "retention_rate": 0.11475409836065574, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.3076923076923077, "status": 0.0, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 811}, "ledger_by_hop": {"1": {"facts_total": 61, "facts_retained": 7, "retention_rate": 0.11475409836065574, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.3076923076923077, "status": 0.0, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 690}, "2": {"facts_total": 61, "facts_retained": 7, "retention_rate": 0.11475409836065574, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.3076923076923077, "status": 0.0, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 811}, "3": {"facts_total": 61, "facts_retained": 7, "retention_rate": 0.11475409836065574, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.3076923076923077, "status": 0.0, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 811}, "4": {"facts_total": 61, "facts_retained": 7, "retention_rate": 0.11475409836065574, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.3076923076923077, "status": 0.0, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 811}}}, "reference_key_id": "keys/synthetic/syn_03015_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "4f973bd7c40fcf07e373e29e86c8315cdbd7790f931726a9494ca5cf77a0b82e", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "df16b36218c655cf47a85fcf", "attempt_id": "c32b8e08564f", "run_group_id": "E4_core:A5|closed|qwen3b-llama3b:h4", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:03:52.859195Z", "task_family": "synthetic", "document_id": "syn_03015_2000", "document_hash": "71a7284b9bc3f681d69d91d18b9e8d45a4dda486cd14f4145395786ac6a68a89", "question_id": "q02_exact_copy", "split": "test", "cluster_id": "syn_03015_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "tokenizer_hashes": {"receiver": "b9d8872cf3a99da99e1fc4b0f71232a451632dc49ca4c6771549b4efa5c1c763"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 4, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1741, "query_tokens": 396, "message_tokens": 288, "message_utf8_bytes": 811, "cache_tensor_bytes": null, "serialized_payload_bytes": 811, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 60096.57245792914, "hops": {"1": 22787.491958006285, "2": 11425.557625014335, "3": 14560.2815420134, "4": 11323.241332895122}, "receiver_prefill_ms": 315.4222499579191, "ttft_ms": 85.8754999935627, "decode_ms": 143.27995805069804}, "total_wall_ms": 60641.63024898153, "ttft_ms": 85.8754999935627, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 61, "facts_retained": 7, "retention_rate": 0.11475409836065574, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.3076923076923077, "status": 0.0, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 811}, "ledger_by_hop": {"1": {"facts_total": 61, "facts_retained": 7, "retention_rate": 0.11475409836065574, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.3076923076923077, "status": 0.0, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 690}, "2": {"facts_total": 61, "facts_retained": 7, "retention_rate": 0.11475409836065574, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.3076923076923077, "status": 0.0, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 811}, "3": {"facts_total": 61, "facts_retained": 7, "retention_rate": 0.11475409836065574, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.3076923076923077, "status": 0.0, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 811}, "4": {"facts_total": 61, "facts_retained": 7, "retention_rate": 0.11475409836065574, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.3076923076923077, "status": 0.0, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 811}}}, "reference_key_id": "keys/synthetic/syn_03015_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "4f973bd7c40fcf07e373e29e86c8315cdbd7790f931726a9494ca5cf77a0b82e", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "f9f500dd79a14888a244c6b7", "attempt_id": "3e07881cedd5", "run_group_id": "E4_core:A5|closed|qwen3b-llama3b:h4", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:03:53.421353Z", "task_family": "synthetic", "document_id": "syn_03015_2000", "document_hash": "71a7284b9bc3f681d69d91d18b9e8d45a4dda486cd14f4145395786ac6a68a89", "question_id": "q03_exact_copy", "split": "test", "cluster_id": "syn_03015_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "tokenizer_hashes": {"receiver": "b9d8872cf3a99da99e1fc4b0f71232a451632dc49ca4c6771549b4efa5c1c763"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 4, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1741, "query_tokens": 392, "message_tokens": 288, "message_utf8_bytes": 811, "cache_tensor_bytes": null, "serialized_payload_bytes": 811, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 60096.57245792914, "hops": {"1": 22787.491958006285, "2": 11425.557625014335, "3": 14560.2815420134, "4": 11323.241332895122}, "receiver_prefill_ms": 335.8255419880152, "ttft_ms": 20.339875016361475, "decode_ms": 187.10904102772474}, "total_wall_ms": 60640.460332972, "ttft_ms": 20.339875016361475, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 6}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "Wilhelmina Dubrovnik", "evidence_artifact": {"ledger": {"facts_total": 61, "facts_retained": 7, "retention_rate": 0.11475409836065574, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.3076923076923077, "status": 0.0, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 811}, "ledger_by_hop": {"1": {"facts_total": 61, "facts_retained": 7, "retention_rate": 0.11475409836065574, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.3076923076923077, "status": 0.0, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 690}, "2": {"facts_total": 61, "facts_retained": 7, "retention_rate": 0.11475409836065574, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.3076923076923077, "status": 0.0, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 811}, "3": {"facts_total": 61, "facts_retained": 7, "retention_rate": 0.11475409836065574, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.3076923076923077, "status": 0.0, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 811}, "4": {"facts_total": 61, "facts_retained": 7, "retention_rate": 0.11475409836065574, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.3076923076923077, "status": 0.0, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 811}}}, "reference_key_id": "keys/synthetic/syn_03015_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "4f973bd7c40fcf07e373e29e86c8315cdbd7790f931726a9494ca5cf77a0b82e", "output_hash": "f42f05ebfc96f7258cbfbbb37cd65e114d3d15f869e68e6c95b71fc288c30a32"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "3d6293be30a7c23910495c85", "attempt_id": "5b6ca636f72d", "run_group_id": "E4_core:A5|closed|qwen3b-llama3b:h4", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:03:53.982699Z", "task_family": "synthetic", "document_id": "syn_03015_2000", "document_hash": "71a7284b9bc3f681d69d91d18b9e8d45a4dda486cd14f4145395786ac6a68a89", "question_id": "q04_multi_hop", "split": "test", "cluster_id": "syn_03015_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "tokenizer_hashes": {"receiver": "b9d8872cf3a99da99e1fc4b0f71232a451632dc49ca4c6771549b4efa5c1c763"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 4, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1741, "query_tokens": 401, "message_tokens": 288, "message_utf8_bytes": 811, "cache_tensor_bytes": null, "serialized_payload_bytes": 811, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 60096.57245792914, "hops": {"1": 22787.491958006285, "2": 11425.557625014335, "3": 14560.2815420134, "4": 11323.241332895122}, "receiver_prefill_ms": 326.067709014751, "ttft_ms": 19.368916982784867, "decode_ms": 223.53637497872114}, "total_wall_ms": 60666.067332960665, "ttft_ms": 19.368916982784867, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 5}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "8,421,725", "evidence_artifact": {"ledger": {"facts_total": 61, "facts_retained": 7, "retention_rate": 0.11475409836065574, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.3076923076923077, "status": 0.0, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 811}, "ledger_by_hop": {"1": {"facts_total": 61, "facts_retained": 7, "retention_rate": 0.11475409836065574, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.3076923076923077, "status": 0.0, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 690}, "2": {"facts_total": 61, "facts_retained": 7, "retention_rate": 0.11475409836065574, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.3076923076923077, "status": 0.0, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 811}, "3": {"facts_total": 61, "facts_retained": 7, "retention_rate": 0.11475409836065574, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.3076923076923077, "status": 0.0, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 811}, "4": {"facts_total": 61, "facts_retained": 7, "retention_rate": 0.11475409836065574, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.3076923076923077, "status": 0.0, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 811}}}, "reference_key_id": "keys/synthetic/syn_03015_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "4f973bd7c40fcf07e373e29e86c8315cdbd7790f931726a9494ca5cf77a0b82e", "output_hash": "6e31df6d4fb38417a5210ded39fc1e4c0e32f278ab9cba0da5b8e7585ddeb534"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "fa2e0d1b81646bbc5b7730ec", "attempt_id": "976898288ac0", "run_group_id": "E4_core:A5|closed|qwen3b-llama3b:h4", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:03:54.570079Z", "task_family": "synthetic", "document_id": "syn_03015_2000", "document_hash": "71a7284b9bc3f681d69d91d18b9e8d45a4dda486cd14f4145395786ac6a68a89", "question_id": "q05_multi_hop", "split": "test", "cluster_id": "syn_03015_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "tokenizer_hashes": {"receiver": "b9d8872cf3a99da99e1fc4b0f71232a451632dc49ca4c6771549b4efa5c1c763"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 4, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1741, "query_tokens": 399, "message_tokens": 288, "message_utf8_bytes": 811, "cache_tensor_bytes": null, "serialized_payload_bytes": 811, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 60096.57245792914, "hops": {"1": 22787.491958006285, "2": 11425.557625014335, "3": 14560.2815420134, "4": 11323.241332895122}, "receiver_prefill_ms": 438.6909579625353, "ttft_ms": 30.52154101897031, "decode_ms": 39.82570790685713}, "total_wall_ms": 60606.06083297171, "ttft_ms": 30.52154101897031, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 61, "facts_retained": 7, "retention_rate": 0.11475409836065574, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.3076923076923077, "status": 0.0, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 811}, "ledger_by_hop": {"1": {"facts_total": 61, "facts_retained": 7, "retention_rate": 0.11475409836065574, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.3076923076923077, "status": 0.0, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 690}, "2": {"facts_total": 61, "facts_retained": 7, "retention_rate": 0.11475409836065574, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.3076923076923077, "status": 0.0, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 811}, "3": {"facts_total": 61, "facts_retained": 7, "retention_rate": 0.11475409836065574, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.3076923076923077, "status": 0.0, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 811}, "4": {"facts_total": 61, "facts_retained": 7, "retention_rate": 0.11475409836065574, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.3076923076923077, "status": 0.0, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 811}}}, "reference_key_id": "keys/synthetic/syn_03015_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "4f973bd7c40fcf07e373e29e86c8315cdbd7790f931726a9494ca5cf77a0b82e", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "e40829f1185a66a2dbc7f137", "attempt_id": "765d0845bb43", "run_group_id": "E4_core:A5|closed|qwen3b-llama3b:h4", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:03:55.097664Z", "task_family": "synthetic", "document_id": "syn_03015_2000", "document_hash": "71a7284b9bc3f681d69d91d18b9e8d45a4dda486cd14f4145395786ac6a68a89", "question_id": "q06_negation", "split": "test", "cluster_id": "syn_03015_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "tokenizer_hashes": {"receiver": "b9d8872cf3a99da99e1fc4b0f71232a451632dc49ca4c6771549b4efa5c1c763"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 4, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1741, "query_tokens": 393, "message_tokens": 288, "message_utf8_bytes": 811, "cache_tensor_bytes": null, "serialized_payload_bytes": 811, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 60096.57245792914, "hops": {"1": 22787.491958006285, "2": 11425.557625014335, "3": 14560.2815420134, "4": 11323.241332895122}, "receiver_prefill_ms": 344.05616600997746, "ttft_ms": 35.36454192362726, "decode_ms": 18.864457961171865}, "total_wall_ms": 60495.329540921375, "ttft_ms": 35.36454192362726, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 61, "facts_retained": 7, "retention_rate": 0.11475409836065574, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.3076923076923077, "status": 0.0, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 811}, "ledger_by_hop": {"1": {"facts_total": 61, "facts_retained": 7, "retention_rate": 0.11475409836065574, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.3076923076923077, "status": 0.0, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 690}, "2": {"facts_total": 61, "facts_retained": 7, "retention_rate": 0.11475409836065574, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.3076923076923077, "status": 0.0, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 811}, "3": {"facts_total": 61, "facts_retained": 7, "retention_rate": 0.11475409836065574, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.3076923076923077, "status": 0.0, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 811}, "4": {"facts_total": 61, "facts_retained": 7, "retention_rate": 0.11475409836065574, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.3076923076923077, "status": 0.0, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 811}}}, "reference_key_id": "keys/synthetic/syn_03015_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "4f973bd7c40fcf07e373e29e86c8315cdbd7790f931726a9494ca5cf77a0b82e", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "6c2172eb04c4f383d29e7f3a", "attempt_id": "d9479ea16eb2", "run_group_id": "E4_core:A5|closed|qwen3b-llama3b:h4", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:03:55.514365Z", "task_family": "synthetic", "document_id": "syn_03015_2000", "document_hash": "71a7284b9bc3f681d69d91d18b9e8d45a4dda486cd14f4145395786ac6a68a89", "question_id": "q07_negation", "split": "test", "cluster_id": "syn_03015_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "tokenizer_hashes": {"receiver": "b9d8872cf3a99da99e1fc4b0f71232a451632dc49ca4c6771549b4efa5c1c763"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 4, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1741, "query_tokens": 393, "message_tokens": 288, "message_utf8_bytes": 811, "cache_tensor_bytes": null, "serialized_payload_bytes": 811, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 60096.57245792914, "hops": {"1": 22787.491958006285, "2": 11425.557625014335, "3": 14560.2815420134, "4": 11323.241332895122}, "receiver_prefill_ms": 311.97516701649874, "ttft_ms": 32.206500065512955, "decode_ms": 18.33574997726828}, "total_wall_ms": 60459.48358299211, "ttft_ms": 32.206500065512955, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 61, "facts_retained": 7, "retention_rate": 0.11475409836065574, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.3076923076923077, "status": 0.0, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 811}, "ledger_by_hop": {"1": {"facts_total": 61, "facts_retained": 7, "retention_rate": 0.11475409836065574, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.3076923076923077, "status": 0.0, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 690}, "2": {"facts_total": 61, "facts_retained": 7, "retention_rate": 0.11475409836065574, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.3076923076923077, "status": 0.0, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 811}, "3": {"facts_total": 61, "facts_retained": 7, "retention_rate": 0.11475409836065574, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.3076923076923077, "status": 0.0, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 811}, "4": {"facts_total": 61, "facts_retained": 7, "retention_rate": 0.11475409836065574, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.3076923076923077, "status": 0.0, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 811}}}, "reference_key_id": "keys/synthetic/syn_03015_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "4f973bd7c40fcf07e373e29e86c8315cdbd7790f931726a9494ca5cf77a0b82e", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "56063cfa22f43d39573dc663", "attempt_id": "95f2400b1138", "run_group_id": "E4_core:A5|closed|qwen3b-llama3b:h4", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:03:55.893959Z", "task_family": "synthetic", "document_id": "syn_03015_2000", "document_hash": "71a7284b9bc3f681d69d91d18b9e8d45a4dda486cd14f4145395786ac6a68a89", "question_id": "q08_revision", "split": "test", "cluster_id": "syn_03015_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "tokenizer_hashes": {"receiver": "b9d8872cf3a99da99e1fc4b0f71232a451632dc49ca4c6771549b4efa5c1c763"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 4, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1741, "query_tokens": 406, "message_tokens": 288, "message_utf8_bytes": 811, "cache_tensor_bytes": null, "serialized_payload_bytes": 811, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 60096.57245792914, "hops": {"1": 22787.491958006285, "2": 11425.557625014335, "3": 14560.2815420134, "4": 11323.241332895122}, "receiver_prefill_ms": 310.7161660445854, "ttft_ms": 49.84562494792044, "decode_ms": 60.468874988146126}, "total_wall_ms": 60518.29154090956, "ttft_ms": 49.84562494792044, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 61, "facts_retained": 7, "retention_rate": 0.11475409836065574, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.3076923076923077, "status": 0.0, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 811}, "ledger_by_hop": {"1": {"facts_total": 61, "facts_retained": 7, "retention_rate": 0.11475409836065574, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.3076923076923077, "status": 0.0, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 690}, "2": {"facts_total": 61, "facts_retained": 7, "retention_rate": 0.11475409836065574, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.3076923076923077, "status": 0.0, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 811}, "3": {"facts_total": 61, "facts_retained": 7, "retention_rate": 0.11475409836065574, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.3076923076923077, "status": 0.0, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 811}, "4": {"facts_total": 61, "facts_retained": 7, "retention_rate": 0.11475409836065574, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.3076923076923077, "status": 0.0, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 811}}}, "reference_key_id": "keys/synthetic/syn_03015_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "4f973bd7c40fcf07e373e29e86c8315cdbd7790f931726a9494ca5cf77a0b82e", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "0354d09c12c30f06420fa9e5", "attempt_id": "5e6c7c756099", "run_group_id": "E4_core:A5|closed|qwen3b-llama3b:h4", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:03:56.334391Z", "task_family": "synthetic", "document_id": "syn_03015_2000", "document_hash": "71a7284b9bc3f681d69d91d18b9e8d45a4dda486cd14f4145395786ac6a68a89", "question_id": "q09_revision", "split": "test", "cluster_id": "syn_03015_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "tokenizer_hashes": {"receiver": "b9d8872cf3a99da99e1fc4b0f71232a451632dc49ca4c6771549b4efa5c1c763"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 4, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1741, "query_tokens": 400, "message_tokens": 288, "message_utf8_bytes": 811, "cache_tensor_bytes": null, "serialized_payload_bytes": 811, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 60096.57245792914, "hops": {"1": 22787.491958006285, "2": 11425.557625014335, "3": 14560.2815420134, "4": 11323.241332895122}, "receiver_prefill_ms": 458.6290419101715, "ttft_ms": 30.791042023338377, "decode_ms": 168.88958297204226}, "total_wall_ms": 60755.442624911666, "ttft_ms": 30.791042023338377, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 5}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "7,028,925", "evidence_artifact": {"ledger": {"facts_total": 61, "facts_retained": 7, "retention_rate": 0.11475409836065574, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.3076923076923077, "status": 0.0, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 811}, "ledger_by_hop": {"1": {"facts_total": 61, "facts_retained": 7, "retention_rate": 0.11475409836065574, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.3076923076923077, "status": 0.0, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 690}, "2": {"facts_total": 61, "facts_retained": 7, "retention_rate": 0.11475409836065574, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.3076923076923077, "status": 0.0, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 811}, "3": {"facts_total": 61, "facts_retained": 7, "retention_rate": 0.11475409836065574, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.3076923076923077, "status": 0.0, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 811}, "4": {"facts_total": 61, "facts_retained": 7, "retention_rate": 0.11475409836065574, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.3076923076923077, "status": 0.0, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 811}}}, "reference_key_id": "keys/synthetic/syn_03015_2000.json", "task_scores": {"correct": 0.0, "abstained": 0.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "4f973bd7c40fcf07e373e29e86c8315cdbd7790f931726a9494ca5cf77a0b82e", "output_hash": "e83edb3f38a69e604be343e63e2e84442156fcb2bb30f36ba9716575d5ffd57a"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "1db38b0708d3f5dac6ca2942", "attempt_id": "1a20513675a4", "run_group_id": "E4_core:A5|closed|qwen3b-llama3b:h4", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:03:57.011063Z", "task_family": "synthetic", "document_id": "syn_03015_2000", "document_hash": "71a7284b9bc3f681d69d91d18b9e8d45a4dda486cd14f4145395786ac6a68a89", "question_id": "q10_temporal", "split": "test", "cluster_id": "syn_03015_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "tokenizer_hashes": {"receiver": "b9d8872cf3a99da99e1fc4b0f71232a451632dc49ca4c6771549b4efa5c1c763"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 4, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1741, "query_tokens": 409, "message_tokens": 288, "message_utf8_bytes": 811, "cache_tensor_bytes": null, "serialized_payload_bytes": 811, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 60096.57245792914, "hops": {"1": 22787.491958006285, "2": 11425.557625014335, "3": 14560.2815420134, "4": 11323.241332895122}, "receiver_prefill_ms": 333.2694999407977, "ttft_ms": 34.95354193728417, "decode_ms": 22.025374928489327}, "total_wall_ms": 60487.248125020415, "ttft_ms": 34.95354193728417, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 61, "facts_retained": 7, "retention_rate": 0.11475409836065574, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.3076923076923077, "status": 0.0, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 811}, "ledger_by_hop": {"1": {"facts_total": 61, "facts_retained": 7, "retention_rate": 0.11475409836065574, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.3076923076923077, "status": 0.0, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 690}, "2": {"facts_total": 61, "facts_retained": 7, "retention_rate": 0.11475409836065574, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.3076923076923077, "status": 0.0, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 811}, "3": {"facts_total": 61, "facts_retained": 7, "retention_rate": 0.11475409836065574, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.3076923076923077, "status": 0.0, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 811}, "4": {"facts_total": 61, "facts_retained": 7, "retention_rate": 0.11475409836065574, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.3076923076923077, "status": 0.0, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 811}}}, "reference_key_id": "keys/synthetic/syn_03015_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "4f973bd7c40fcf07e373e29e86c8315cdbd7790f931726a9494ca5cf77a0b82e", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "7482c77ab57306dd55f3e1cb", "attempt_id": "c9a5b9b53d42", "run_group_id": "E4_core:A5|closed|qwen3b-llama3b:h4", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:03:57.419933Z", "task_family": "synthetic", "document_id": "syn_03015_2000", "document_hash": "71a7284b9bc3f681d69d91d18b9e8d45a4dda486cd14f4145395786ac6a68a89", "question_id": "q11_aggregation", "split": "test", "cluster_id": "syn_03015_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "tokenizer_hashes": {"receiver": "b9d8872cf3a99da99e1fc4b0f71232a451632dc49ca4c6771549b4efa5c1c763"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 4, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1741, "query_tokens": 403, "message_tokens": 288, "message_utf8_bytes": 811, "cache_tensor_bytes": null, "serialized_payload_bytes": 811, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 60096.57245792914, "hops": {"1": 22787.491958006285, "2": 11425.557625014335, "3": 14560.2815420134, "4": 11323.241332895122}, "receiver_prefill_ms": 295.4395830165595, "ttft_ms": 56.562499958090484, "decode_ms": 63.67795797996223}, "total_wall_ms": 60512.812582892366, "ttft_ms": 56.562499958090484, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 61, "facts_retained": 7, "retention_rate": 0.11475409836065574, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.3076923076923077, "status": 0.0, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 811}, "ledger_by_hop": {"1": {"facts_total": 61, "facts_retained": 7, "retention_rate": 0.11475409836065574, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.3076923076923077, "status": 0.0, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 690}, "2": {"facts_total": 61, "facts_retained": 7, "retention_rate": 0.11475409836065574, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.3076923076923077, "status": 0.0, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 811}, "3": {"facts_total": 61, "facts_retained": 7, "retention_rate": 0.11475409836065574, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.3076923076923077, "status": 0.0, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 811}, "4": {"facts_total": 61, "facts_retained": 7, "retention_rate": 0.11475409836065574, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.3076923076923077, "status": 0.0, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 811}}}, "reference_key_id": "keys/synthetic/syn_03015_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "4f973bd7c40fcf07e373e29e86c8315cdbd7790f931726a9494ca5cf77a0b82e", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "383a9cd69c072bc327292ca3", "attempt_id": "272342a47a74", "run_group_id": "E4_core:A5|closed|qwen3b-llama3b:h4", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:03:57.852914Z", "task_family": "synthetic", "document_id": "syn_03015_2000", "document_hash": "71a7284b9bc3f681d69d91d18b9e8d45a4dda486cd14f4145395786ac6a68a89", "question_id": "q12_unit_conversion", "split": "test", "cluster_id": "syn_03015_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "tokenizer_hashes": {"receiver": "b9d8872cf3a99da99e1fc4b0f71232a451632dc49ca4c6771549b4efa5c1c763"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 4, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1741, "query_tokens": 400, "message_tokens": 288, "message_utf8_bytes": 811, "cache_tensor_bytes": null, "serialized_payload_bytes": 811, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 60096.57245792914, "hops": {"1": 22787.491958006285, "2": 11425.557625014335, "3": 14560.2815420134, "4": 11323.241332895122}, "receiver_prefill_ms": 459.39820795319974, "ttft_ms": 30.278666992671788, "decode_ms": 40.322499931789935}, "total_wall_ms": 60627.15504085645, "ttft_ms": 30.278666992671788, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 61, "facts_retained": 7, "retention_rate": 0.11475409836065574, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.3076923076923077, "status": 0.0, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 811}, "ledger_by_hop": {"1": {"facts_total": 61, "facts_retained": 7, "retention_rate": 0.11475409836065574, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.3076923076923077, "status": 0.0, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 690}, "2": {"facts_total": 61, "facts_retained": 7, "retention_rate": 0.11475409836065574, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.3076923076923077, "status": 0.0, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 811}, "3": {"facts_total": 61, "facts_retained": 7, "retention_rate": 0.11475409836065574, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.3076923076923077, "status": 0.0, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 811}, "4": {"facts_total": 61, "facts_retained": 7, "retention_rate": 0.11475409836065574, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.3076923076923077, "status": 0.0, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 811}}}, "reference_key_id": "keys/synthetic/syn_03015_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "4f973bd7c40fcf07e373e29e86c8315cdbd7790f931726a9494ca5cf77a0b82e", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "12f41f46bc7df318525028dd", "attempt_id": "c5e2cbf365b7", "run_group_id": "E4_core:A5|closed|qwen3b-llama3b:h4", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:03:58.400951Z", "task_family": "synthetic", "document_id": "syn_03015_2000", "document_hash": "71a7284b9bc3f681d69d91d18b9e8d45a4dda486cd14f4145395786ac6a68a89", "question_id": "q13_unanswerable", "split": "test", "cluster_id": "syn_03015_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "tokenizer_hashes": {"receiver": "b9d8872cf3a99da99e1fc4b0f71232a451632dc49ca4c6771549b4efa5c1c763"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 4, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1741, "query_tokens": 403, "message_tokens": 288, "message_utf8_bytes": 811, "cache_tensor_bytes": null, "serialized_payload_bytes": 811, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 60096.57245792914, "hops": {"1": 22787.491958006285, "2": 11425.557625014335, "3": 14560.2815420134, "4": 11323.241332895122}, "receiver_prefill_ms": 332.68958306871355, "ttft_ms": 37.133665988221765, "decode_ms": 26.524416054598987}, "total_wall_ms": 60493.53429186158, "ttft_ms": 37.133665988221765, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 61, "facts_retained": 7, "retention_rate": 0.11475409836065574, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.3076923076923077, "status": 0.0, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 811}, "ledger_by_hop": {"1": {"facts_total": 61, "facts_retained": 7, "retention_rate": 0.11475409836065574, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.3076923076923077, "status": 0.0, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 690}, "2": {"facts_total": 61, "facts_retained": 7, "retention_rate": 0.11475409836065574, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.3076923076923077, "status": 0.0, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 811}, "3": {"facts_total": 61, "facts_retained": 7, "retention_rate": 0.11475409836065574, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.3076923076923077, "status": 0.0, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 811}, "4": {"facts_total": 61, "facts_retained": 7, "retention_rate": 0.11475409836065574, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.3076923076923077, "status": 0.0, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 811}}}, "reference_key_id": "keys/synthetic/syn_03015_2000.json", "task_scores": {"correct": 1.0, "abstained": 1.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "4f973bd7c40fcf07e373e29e86c8315cdbd7790f931726a9494ca5cf77a0b82e", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "d8fd996f4c85cd12fec26ba8", "attempt_id": "aa45a772b266", "run_group_id": "E4_core:A5|closed|qwen3b-llama3b:h4", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:03:58.816161Z", "task_family": "synthetic", "document_id": "syn_03015_2000", "document_hash": "71a7284b9bc3f681d69d91d18b9e8d45a4dda486cd14f4145395786ac6a68a89", "question_id": "q14_unanswerable", "split": "test", "cluster_id": "syn_03015_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "tokenizer_hashes": {"receiver": "b9d8872cf3a99da99e1fc4b0f71232a451632dc49ca4c6771549b4efa5c1c763"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 4, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1741, "query_tokens": 405, "message_tokens": 288, "message_utf8_bytes": 811, "cache_tensor_bytes": null, "serialized_payload_bytes": 811, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 60096.57245792914, "hops": {"1": 22787.491958006285, "2": 11425.557625014335, "3": 14560.2815420134, "4": 11323.241332895122}, "receiver_prefill_ms": 312.147207907401, "ttft_ms": 18.505832995288074, "decode_ms": 15.492290956899524}, "total_wall_ms": 60443.32683284301, "ttft_ms": 18.505832995288074, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 61, "facts_retained": 7, "retention_rate": 0.11475409836065574, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.3076923076923077, "status": 0.0, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 811}, "ledger_by_hop": {"1": {"facts_total": 61, "facts_retained": 7, "retention_rate": 0.11475409836065574, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.3076923076923077, "status": 0.0, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 690}, "2": {"facts_total": 61, "facts_retained": 7, "retention_rate": 0.11475409836065574, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.3076923076923077, "status": 0.0, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 811}, "3": {"facts_total": 61, "facts_retained": 7, "retention_rate": 0.11475409836065574, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.3076923076923077, "status": 0.0, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 811}, "4": {"facts_total": 61, "facts_retained": 7, "retention_rate": 0.11475409836065574, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.3076923076923077, "status": 0.0, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 811}}}, "reference_key_id": "keys/synthetic/syn_03015_2000.json", "task_scores": {"correct": 1.0, "abstained": 1.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "4f973bd7c40fcf07e373e29e86c8315cdbd7790f931726a9494ca5cf77a0b82e", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "f2e969448da0b9d511e79495", "attempt_id": "317844d2dea4", "run_group_id": "E4_core:A5|closed|qwen3b-llama3b:h8", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:04:51.002321Z", "task_family": "synthetic", "document_id": "syn_03015_2000", "document_hash": "71a7284b9bc3f681d69d91d18b9e8d45a4dda486cd14f4145395786ac6a68a89", "question_id": "q00_single_fact", "split": "test", "cluster_id": "syn_03015_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "tokenizer_hashes": {"receiver": "b9d8872cf3a99da99e1fc4b0f71232a451632dc49ca4c6771549b4efa5c1c763"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 8, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1741, "query_tokens": 392, "message_tokens": 288, "message_utf8_bytes": 811, "cache_tensor_bytes": null, "serialized_payload_bytes": 811, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 111914.81091594324, "hops": {"1": 22787.491958006285, "2": 11425.557625014335, "3": 14560.2815420134, "4": 11323.241332895122, "5": 14510.312750004232, "6": 11364.853915991262, "7": 14467.890792060643, "8": 11475.180999957956}, "receiver_prefill_ms": 353.5949169890955, "ttft_ms": 26.605874998494983, "decode_ms": 35.92920792289078}, "total_wall_ms": 112331.37416595127, "ttft_ms": 26.605874998494983, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 61, "facts_retained": 7, "retention_rate": 0.11475409836065574, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.3076923076923077, "status": 0.0, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 811}, "ledger_by_hop": {"1": {"facts_total": 61, "facts_retained": 7, "retention_rate": 0.11475409836065574, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.3076923076923077, "status": 0.0, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 690}, "2": {"facts_total": 61, "facts_retained": 7, "retention_rate": 0.11475409836065574, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.3076923076923077, "status": 0.0, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 811}, "3": {"facts_total": 61, "facts_retained": 7, "retention_rate": 0.11475409836065574, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.3076923076923077, "status": 0.0, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 811}, "4": {"facts_total": 61, "facts_retained": 7, "retention_rate": 0.11475409836065574, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.3076923076923077, "status": 0.0, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 811}, "5": {"facts_total": 61, "facts_retained": 7, "retention_rate": 0.11475409836065574, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.3076923076923077, "status": 0.0, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 811}, "6": {"facts_total": 61, "facts_retained": 7, "retention_rate": 0.11475409836065574, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.3076923076923077, "status": 0.0, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 811}, "7": {"facts_total": 61, "facts_retained": 7, "retention_rate": 0.11475409836065574, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.3076923076923077, "status": 0.0, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 811}, "8": {"facts_total": 61, "facts_retained": 7, "retention_rate": 0.11475409836065574, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.3076923076923077, "status": 0.0, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 811}}}, "reference_key_id": "keys/synthetic/syn_03015_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "4f973bd7c40fcf07e373e29e86c8315cdbd7790f931726a9494ca5cf77a0b82e", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "8e46d15c04c4a1e0269d3ce9", "attempt_id": "01fb7d88836c", "run_group_id": "E4_core:A5|closed|qwen3b-llama3b:h8", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:04:51.436540Z", "task_family": "synthetic", "document_id": "syn_03015_2000", "document_hash": "71a7284b9bc3f681d69d91d18b9e8d45a4dda486cd14f4145395786ac6a68a89", "question_id": "q01_single_fact", "split": "test", "cluster_id": "syn_03015_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "tokenizer_hashes": {"receiver": "b9d8872cf3a99da99e1fc4b0f71232a451632dc49ca4c6771549b4efa5c1c763"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 8, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1741, "query_tokens": 396, "message_tokens": 288, "message_utf8_bytes": 811, "cache_tensor_bytes": null, "serialized_payload_bytes": 811, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 111914.81091594324, "hops": {"1": 22787.491958006285, "2": 11425.557625014335, "3": 14560.2815420134, "4": 11323.241332895122, "5": 14510.312750004232, "6": 11364.853915991262, "7": 14467.890792060643, "8": 11475.180999957956}, "receiver_prefill_ms": 334.7237090347335, "ttft_ms": 61.0629579750821, "decode_ms": 39.91833399049938}, "total_wall_ms": 112351.12537501846, "ttft_ms": 61.0629579750821, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 61, "facts_retained": 7, "retention_rate": 0.11475409836065574, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.3076923076923077, "status": 0.0, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 811}, "ledger_by_hop": {"1": {"facts_total": 61, "facts_retained": 7, "retention_rate": 0.11475409836065574, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.3076923076923077, "status": 0.0, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 690}, "2": {"facts_total": 61, "facts_retained": 7, "retention_rate": 0.11475409836065574, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.3076923076923077, "status": 0.0, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 811}, "3": {"facts_total": 61, "facts_retained": 7, "retention_rate": 0.11475409836065574, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.3076923076923077, "status": 0.0, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 811}, "4": {"facts_total": 61, "facts_retained": 7, "retention_rate": 0.11475409836065574, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.3076923076923077, "status": 0.0, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 811}, "5": {"facts_total": 61, "facts_retained": 7, "retention_rate": 0.11475409836065574, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.3076923076923077, "status": 0.0, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 811}, "6": {"facts_total": 61, "facts_retained": 7, "retention_rate": 0.11475409836065574, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.3076923076923077, "status": 0.0, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 811}, "7": {"facts_total": 61, "facts_retained": 7, "retention_rate": 0.11475409836065574, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.3076923076923077, "status": 0.0, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 811}, "8": {"facts_total": 61, "facts_retained": 7, "retention_rate": 0.11475409836065574, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.3076923076923077, "status": 0.0, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 811}}}, "reference_key_id": "keys/synthetic/syn_03015_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "4f973bd7c40fcf07e373e29e86c8315cdbd7790f931726a9494ca5cf77a0b82e", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "95369a2a3af1cb386d31f936", "attempt_id": "b9464be788a2", "run_group_id": "E4_core:A5|closed|qwen3b-llama3b:h8", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:04:51.889661Z", "task_family": "synthetic", "document_id": "syn_03015_2000", "document_hash": "71a7284b9bc3f681d69d91d18b9e8d45a4dda486cd14f4145395786ac6a68a89", "question_id": "q02_exact_copy", "split": "test", "cluster_id": "syn_03015_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "tokenizer_hashes": {"receiver": "b9d8872cf3a99da99e1fc4b0f71232a451632dc49ca4c6771549b4efa5c1c763"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 8, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1741, "query_tokens": 396, "message_tokens": 288, "message_utf8_bytes": 811, "cache_tensor_bytes": null, "serialized_payload_bytes": 811, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 111914.81091594324, "hops": {"1": 22787.491958006285, "2": 11425.557625014335, "3": 14560.2815420134, "4": 11323.241332895122, "5": 14510.312750004232, "6": 11364.853915991262, "7": 14467.890792060643, "8": 11475.180999957956}, "receiver_prefill_ms": 437.2374579543248, "ttft_ms": 37.892374908551574, "decode_ms": 22.639458999037743}, "total_wall_ms": 112413.21479086764, "ttft_ms": 37.892374908551574, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 61, "facts_retained": 7, "retention_rate": 0.11475409836065574, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.3076923076923077, "status": 0.0, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 811}, "ledger_by_hop": {"1": {"facts_total": 61, "facts_retained": 7, "retention_rate": 0.11475409836065574, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.3076923076923077, "status": 0.0, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 690}, "2": {"facts_total": 61, "facts_retained": 7, "retention_rate": 0.11475409836065574, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.3076923076923077, "status": 0.0, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 811}, "3": {"facts_total": 61, "facts_retained": 7, "retention_rate": 0.11475409836065574, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.3076923076923077, "status": 0.0, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 811}, "4": {"facts_total": 61, "facts_retained": 7, "retention_rate": 0.11475409836065574, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.3076923076923077, "status": 0.0, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 811}, "5": {"facts_total": 61, "facts_retained": 7, "retention_rate": 0.11475409836065574, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.3076923076923077, "status": 0.0, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 811}, "6": {"facts_total": 61, "facts_retained": 7, "retention_rate": 0.11475409836065574, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.3076923076923077, "status": 0.0, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 811}, "7": {"facts_total": 61, "facts_retained": 7, "retention_rate": 0.11475409836065574, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.3076923076923077, "status": 0.0, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 811}, "8": {"facts_total": 61, "facts_retained": 7, "retention_rate": 0.11475409836065574, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.3076923076923077, "status": 0.0, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 811}}}, "reference_key_id": "keys/synthetic/syn_03015_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "4f973bd7c40fcf07e373e29e86c8315cdbd7790f931726a9494ca5cf77a0b82e", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "b68352ed3865bd61366fe3cf", "attempt_id": "8d17ddb32589", "run_group_id": "E4_core:A5|closed|qwen3b-llama3b:h8", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:04:52.405497Z", "task_family": "synthetic", "document_id": "syn_03015_2000", "document_hash": "71a7284b9bc3f681d69d91d18b9e8d45a4dda486cd14f4145395786ac6a68a89", "question_id": "q03_exact_copy", "split": "test", "cluster_id": "syn_03015_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "tokenizer_hashes": {"receiver": "b9d8872cf3a99da99e1fc4b0f71232a451632dc49ca4c6771549b4efa5c1c763"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 8, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1741, "query_tokens": 392, "message_tokens": 288, "message_utf8_bytes": 811, "cache_tensor_bytes": null, "serialized_payload_bytes": 811, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 111914.81091594324, "hops": {"1": 22787.491958006285, "2": 11425.557625014335, "3": 14560.2815420134, "4": 11323.241332895122, "5": 14510.312750004232, "6": 11364.853915991262, "7": 14467.890792060643, "8": 11475.180999957956}, "receiver_prefill_ms": 314.8652089294046, "ttft_ms": 33.96795899607241, "decode_ms": 183.20295796729624}, "total_wall_ms": 112447.27333297487, "ttft_ms": 33.96795899607241, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 6}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "Wilhelmina Dubrovnik", "evidence_artifact": {"ledger": {"facts_total": 61, "facts_retained": 7, "retention_rate": 0.11475409836065574, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.3076923076923077, "status": 0.0, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 811}, "ledger_by_hop": {"1": {"facts_total": 61, "facts_retained": 7, "retention_rate": 0.11475409836065574, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.3076923076923077, "status": 0.0, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 690}, "2": {"facts_total": 61, "facts_retained": 7, "retention_rate": 0.11475409836065574, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.3076923076923077, "status": 0.0, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 811}, "3": {"facts_total": 61, "facts_retained": 7, "retention_rate": 0.11475409836065574, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.3076923076923077, "status": 0.0, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 811}, "4": {"facts_total": 61, "facts_retained": 7, "retention_rate": 0.11475409836065574, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.3076923076923077, "status": 0.0, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 811}, "5": {"facts_total": 61, "facts_retained": 7, "retention_rate": 0.11475409836065574, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.3076923076923077, "status": 0.0, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 811}, "6": {"facts_total": 61, "facts_retained": 7, "retention_rate": 0.11475409836065574, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.3076923076923077, "status": 0.0, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 811}, "7": {"facts_total": 61, "facts_retained": 7, "retention_rate": 0.11475409836065574, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.3076923076923077, "status": 0.0, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 811}, "8": {"facts_total": 61, "facts_retained": 7, "retention_rate": 0.11475409836065574, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.3076923076923077, "status": 0.0, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 811}}}, "reference_key_id": "keys/synthetic/syn_03015_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "4f973bd7c40fcf07e373e29e86c8315cdbd7790f931726a9494ca5cf77a0b82e", "output_hash": "f42f05ebfc96f7258cbfbbb37cd65e114d3d15f869e68e6c95b71fc288c30a32"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "3b5230207a9b60ef0f35ad6f", "attempt_id": "e5ce4237c6c0", "run_group_id": "E4_core:A5|closed|qwen3b-llama3b:h8", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:04:52.954928Z", "task_family": "synthetic", "document_id": "syn_03015_2000", "document_hash": "71a7284b9bc3f681d69d91d18b9e8d45a4dda486cd14f4145395786ac6a68a89", "question_id": "q04_multi_hop", "split": "test", "cluster_id": "syn_03015_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "tokenizer_hashes": {"receiver": "b9d8872cf3a99da99e1fc4b0f71232a451632dc49ca4c6771549b4efa5c1c763"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 8, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1741, "query_tokens": 401, "message_tokens": 288, "message_utf8_bytes": 811, "cache_tensor_bytes": null, "serialized_payload_bytes": 811, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 111914.81091594324, "hops": {"1": 22787.491958006285, "2": 11425.557625014335, "3": 14560.2815420134, "4": 11323.241332895122, "5": 14510.312750004232, "6": 11364.853915991262, "7": 14467.890792060643, "8": 11475.180999957956}, "receiver_prefill_ms": 318.8440001104027, "ttft_ms": 57.039499981328845, "decode_ms": 386.4573750179261}, "total_wall_ms": 112677.6853329502, "ttft_ms": 57.039499981328845, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 5}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "8,421,725", "evidence_artifact": {"ledger": {"facts_total": 61, "facts_retained": 7, "retention_rate": 0.11475409836065574, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.3076923076923077, "status": 0.0, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 811}, "ledger_by_hop": {"1": {"facts_total": 61, "facts_retained": 7, "retention_rate": 0.11475409836065574, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.3076923076923077, "status": 0.0, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 690}, "2": {"facts_total": 61, "facts_retained": 7, "retention_rate": 0.11475409836065574, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.3076923076923077, "status": 0.0, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 811}, "3": {"facts_total": 61, "facts_retained": 7, "retention_rate": 0.11475409836065574, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.3076923076923077, "status": 0.0, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 811}, "4": {"facts_total": 61, "facts_retained": 7, "retention_rate": 0.11475409836065574, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.3076923076923077, "status": 0.0, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 811}, "5": {"facts_total": 61, "facts_retained": 7, "retention_rate": 0.11475409836065574, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.3076923076923077, "status": 0.0, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 811}, "6": {"facts_total": 61, "facts_retained": 7, "retention_rate": 0.11475409836065574, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.3076923076923077, "status": 0.0, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 811}, "7": {"facts_total": 61, "facts_retained": 7, "retention_rate": 0.11475409836065574, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.3076923076923077, "status": 0.0, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 811}, "8": {"facts_total": 61, "facts_retained": 7, "retention_rate": 0.11475409836065574, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.3076923076923077, "status": 0.0, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 811}}}, "reference_key_id": "keys/synthetic/syn_03015_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "4f973bd7c40fcf07e373e29e86c8315cdbd7790f931726a9494ca5cf77a0b82e", "output_hash": "6e31df6d4fb38417a5210ded39fc1e4c0e32f278ab9cba0da5b8e7585ddeb534"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "b0722fcbad269169a7b018e2", "attempt_id": "57dd3e286d67", "run_group_id": "E4_core:A5|closed|qwen3b-llama3b:h8", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:04:53.734882Z", "task_family": "synthetic", "document_id": "syn_03015_2000", "document_hash": "71a7284b9bc3f681d69d91d18b9e8d45a4dda486cd14f4145395786ac6a68a89", "question_id": "q05_multi_hop", "split": "test", "cluster_id": "syn_03015_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "tokenizer_hashes": {"receiver": "b9d8872cf3a99da99e1fc4b0f71232a451632dc49ca4c6771549b4efa5c1c763"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 8, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1741, "query_tokens": 399, "message_tokens": 288, "message_utf8_bytes": 811, "cache_tensor_bytes": null, "serialized_payload_bytes": 811, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 111914.81091594324, "hops": {"1": 22787.491958006285, "2": 11425.557625014335, "3": 14560.2815420134, "4": 11323.241332895122, "5": 14510.312750004232, "6": 11364.853915991262, "7": 14467.890792060643, "8": 11475.180999957956}, "receiver_prefill_ms": 326.53808291070163, "ttft_ms": 30.606457963585854, "decode_ms": 33.58504199422896}, "total_wall_ms": 112306.07662384864, "ttft_ms": 30.606457963585854, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 61, "facts_retained": 7, "retention_rate": 0.11475409836065574, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.3076923076923077, "status": 0.0, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 811}, "ledger_by_hop": {"1": {"facts_total": 61, "facts_retained": 7, "retention_rate": 0.11475409836065574, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.3076923076923077, "status": 0.0, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 690}, "2": {"facts_total": 61, "facts_retained": 7, "retention_rate": 0.11475409836065574, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.3076923076923077, "status": 0.0, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 811}, "3": {"facts_total": 61, "facts_retained": 7, "retention_rate": 0.11475409836065574, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.3076923076923077, "status": 0.0, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 811}, "4": {"facts_total": 61, "facts_retained": 7, "retention_rate": 0.11475409836065574, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.3076923076923077, "status": 0.0, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 811}, "5": {"facts_total": 61, "facts_retained": 7, "retention_rate": 0.11475409836065574, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.3076923076923077, "status": 0.0, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 811}, "6": {"facts_total": 61, "facts_retained": 7, "retention_rate": 0.11475409836065574, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.3076923076923077, "status": 0.0, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 811}, "7": {"facts_total": 61, "facts_retained": 7, "retention_rate": 0.11475409836065574, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.3076923076923077, "status": 0.0, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 811}, "8": {"facts_total": 61, "facts_retained": 7, "retention_rate": 0.11475409836065574, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.3076923076923077, "status": 0.0, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 811}}}, "reference_key_id": "keys/synthetic/syn_03015_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "4f973bd7c40fcf07e373e29e86c8315cdbd7790f931726a9494ca5cf77a0b82e", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "ac911e3c6242d63aec6d3b36", "attempt_id": "74e3a6c1a458", "run_group_id": "E4_core:A5|closed|qwen3b-llama3b:h8", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:04:54.143406Z", "task_family": "synthetic", "document_id": "syn_03015_2000", "document_hash": "71a7284b9bc3f681d69d91d18b9e8d45a4dda486cd14f4145395786ac6a68a89", "question_id": "q06_negation", "split": "test", "cluster_id": "syn_03015_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "tokenizer_hashes": {"receiver": "b9d8872cf3a99da99e1fc4b0f71232a451632dc49ca4c6771549b4efa5c1c763"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 8, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1741, "query_tokens": 393, "message_tokens": 288, "message_utf8_bytes": 811, "cache_tensor_bytes": null, "serialized_payload_bytes": 811, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 111914.81091594324, "hops": {"1": 22787.491958006285, "2": 11425.557625014335, "3": 14560.2815420134, "4": 11323.241332895122, "5": 14510.312750004232, "6": 11364.853915991262, "7": 14467.890792060643, "8": 11475.180999957956}, "receiver_prefill_ms": 333.0988750094548, "ttft_ms": 37.3265000525862, "decode_ms": 27.945791953243315}, "total_wall_ms": 112313.72866593301, "ttft_ms": 37.3265000525862, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 61, "facts_retained": 7, "retention_rate": 0.11475409836065574, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.3076923076923077, "status": 0.0, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 811}, "ledger_by_hop": {"1": {"facts_total": 61, "facts_retained": 7, "retention_rate": 0.11475409836065574, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.3076923076923077, "status": 0.0, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 690}, "2": {"facts_total": 61, "facts_retained": 7, "retention_rate": 0.11475409836065574, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.3076923076923077, "status": 0.0, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 811}, "3": {"facts_total": 61, "facts_retained": 7, "retention_rate": 0.11475409836065574, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.3076923076923077, "status": 0.0, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 811}, "4": {"facts_total": 61, "facts_retained": 7, "retention_rate": 0.11475409836065574, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.3076923076923077, "status": 0.0, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 811}, "5": {"facts_total": 61, "facts_retained": 7, "retention_rate": 0.11475409836065574, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.3076923076923077, "status": 0.0, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 811}, "6": {"facts_total": 61, "facts_retained": 7, "retention_rate": 0.11475409836065574, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.3076923076923077, "status": 0.0, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 811}, "7": {"facts_total": 61, "facts_retained": 7, "retention_rate": 0.11475409836065574, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.3076923076923077, "status": 0.0, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 811}, "8": {"facts_total": 61, "facts_retained": 7, "retention_rate": 0.11475409836065574, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.3076923076923077, "status": 0.0, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 811}}}, "reference_key_id": "keys/synthetic/syn_03015_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "4f973bd7c40fcf07e373e29e86c8315cdbd7790f931726a9494ca5cf77a0b82e", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "728a9728d35299d9f51a58b1", "attempt_id": "0bf5dafcd4aa", "run_group_id": "E4_core:A5|closed|qwen3b-llama3b:h8", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:04:54.560080Z", "task_family": "synthetic", "document_id": "syn_03015_2000", "document_hash": "71a7284b9bc3f681d69d91d18b9e8d45a4dda486cd14f4145395786ac6a68a89", "question_id": "q07_negation", "split": "test", "cluster_id": "syn_03015_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "tokenizer_hashes": {"receiver": "b9d8872cf3a99da99e1fc4b0f71232a451632dc49ca4c6771549b4efa5c1c763"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 8, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1741, "query_tokens": 393, "message_tokens": 288, "message_utf8_bytes": 811, "cache_tensor_bytes": null, "serialized_payload_bytes": 811, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 111914.81091594324, "hops": {"1": 22787.491958006285, "2": 11425.557625014335, "3": 14560.2815420134, "4": 11323.241332895122, "5": 14510.312750004232, "6": 11364.853915991262, "7": 14467.890792060643, "8": 11475.180999957956}, "receiver_prefill_ms": 295.3373750206083, "ttft_ms": 42.25966695230454, "decode_ms": 49.89045800175518}, "total_wall_ms": 112302.96524893492, "ttft_ms": 42.25966695230454, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 61, "facts_retained": 7, "retention_rate": 0.11475409836065574, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.3076923076923077, "status": 0.0, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 811}, "ledger_by_hop": {"1": {"facts_total": 61, "facts_retained": 7, "retention_rate": 0.11475409836065574, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.3076923076923077, "status": 0.0, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 690}, "2": {"facts_total": 61, "facts_retained": 7, "retention_rate": 0.11475409836065574, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.3076923076923077, "status": 0.0, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 811}, "3": {"facts_total": 61, "facts_retained": 7, "retention_rate": 0.11475409836065574, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.3076923076923077, "status": 0.0, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 811}, "4": {"facts_total": 61, "facts_retained": 7, "retention_rate": 0.11475409836065574, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.3076923076923077, "status": 0.0, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 811}, "5": {"facts_total": 61, "facts_retained": 7, "retention_rate": 0.11475409836065574, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.3076923076923077, "status": 0.0, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 811}, "6": {"facts_total": 61, "facts_retained": 7, "retention_rate": 0.11475409836065574, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.3076923076923077, "status": 0.0, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 811}, "7": {"facts_total": 61, "facts_retained": 7, "retention_rate": 0.11475409836065574, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.3076923076923077, "status": 0.0, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 811}, "8": {"facts_total": 61, "facts_retained": 7, "retention_rate": 0.11475409836065574, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.3076923076923077, "status": 0.0, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 811}}}, "reference_key_id": "keys/synthetic/syn_03015_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "4f973bd7c40fcf07e373e29e86c8315cdbd7790f931726a9494ca5cf77a0b82e", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "79bc3f010c265e3edc1d1e62", "attempt_id": "2f0196261aeb", "run_group_id": "E4_core:A5|closed|qwen3b-llama3b:h8", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:04:54.966158Z", "task_family": "synthetic", "document_id": "syn_03015_2000", "document_hash": "71a7284b9bc3f681d69d91d18b9e8d45a4dda486cd14f4145395786ac6a68a89", "question_id": "q08_revision", "split": "test", "cluster_id": "syn_03015_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "tokenizer_hashes": {"receiver": "b9d8872cf3a99da99e1fc4b0f71232a451632dc49ca4c6771549b4efa5c1c763"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 8, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1741, "query_tokens": 406, "message_tokens": 288, "message_utf8_bytes": 811, "cache_tensor_bytes": null, "serialized_payload_bytes": 811, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 111914.81091594324, "hops": {"1": 22787.491958006285, "2": 11425.557625014335, "3": 14560.2815420134, "4": 11323.241332895122, "5": 14510.312750004232, "6": 11364.853915991262, "7": 14467.890792060643, "8": 11475.180999957956}, "receiver_prefill_ms": 474.4279580190778, "ttft_ms": 35.866874968633056, "decode_ms": 23.245834046974778}, "total_wall_ms": 112448.83599888999, "ttft_ms": 35.866874968633056, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 61, "facts_retained": 7, "retention_rate": 0.11475409836065574, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.3076923076923077, "status": 0.0, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 811}, "ledger_by_hop": {"1": {"facts_total": 61, "facts_retained": 7, "retention_rate": 0.11475409836065574, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.3076923076923077, "status": 0.0, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 690}, "2": {"facts_total": 61, "facts_retained": 7, "retention_rate": 0.11475409836065574, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.3076923076923077, "status": 0.0, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 811}, "3": {"facts_total": 61, "facts_retained": 7, "retention_rate": 0.11475409836065574, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.3076923076923077, "status": 0.0, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 811}, "4": {"facts_total": 61, "facts_retained": 7, "retention_rate": 0.11475409836065574, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.3076923076923077, "status": 0.0, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 811}, "5": {"facts_total": 61, "facts_retained": 7, "retention_rate": 0.11475409836065574, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.3076923076923077, "status": 0.0, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 811}, "6": {"facts_total": 61, "facts_retained": 7, "retention_rate": 0.11475409836065574, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.3076923076923077, "status": 0.0, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 811}, "7": {"facts_total": 61, "facts_retained": 7, "retention_rate": 0.11475409836065574, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.3076923076923077, "status": 0.0, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 811}, "8": {"facts_total": 61, "facts_retained": 7, "retention_rate": 0.11475409836065574, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.3076923076923077, "status": 0.0, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 811}}}, "reference_key_id": "keys/synthetic/syn_03015_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "4f973bd7c40fcf07e373e29e86c8315cdbd7790f931726a9494ca5cf77a0b82e", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "deec06716fb193d2f483b407", "attempt_id": "03fbffd6c7a0", "run_group_id": "E4_core:A5|closed|qwen3b-llama3b:h8", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:04:55.516982Z", "task_family": "synthetic", "document_id": "syn_03015_2000", "document_hash": "71a7284b9bc3f681d69d91d18b9e8d45a4dda486cd14f4145395786ac6a68a89", "question_id": "q09_revision", "split": "test", "cluster_id": "syn_03015_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "tokenizer_hashes": {"receiver": "b9d8872cf3a99da99e1fc4b0f71232a451632dc49ca4c6771549b4efa5c1c763"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 8, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1741, "query_tokens": 400, "message_tokens": 288, "message_utf8_bytes": 811, "cache_tensor_bytes": null, "serialized_payload_bytes": 811, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 111914.81091594324, "hops": {"1": 22787.491958006285, "2": 11425.557625014335, "3": 14560.2815420134, "4": 11323.241332895122, "5": 14510.312750004232, "6": 11364.853915991262, "7": 14467.890792060643, "8": 11475.180999957956}, "receiver_prefill_ms": 339.5323340082541, "ttft_ms": 21.35837497189641, "decode_ms": 161.6004160605371}, "total_wall_ms": 112437.85899993964, "ttft_ms": 21.35837497189641, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 5}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "7,028,925", "evidence_artifact": {"ledger": {"facts_total": 61, "facts_retained": 7, "retention_rate": 0.11475409836065574, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.3076923076923077, "status": 0.0, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 811}, "ledger_by_hop": {"1": {"facts_total": 61, "facts_retained": 7, "retention_rate": 0.11475409836065574, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.3076923076923077, "status": 0.0, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 690}, "2": {"facts_total": 61, "facts_retained": 7, "retention_rate": 0.11475409836065574, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.3076923076923077, "status": 0.0, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 811}, "3": {"facts_total": 61, "facts_retained": 7, "retention_rate": 0.11475409836065574, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.3076923076923077, "status": 0.0, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 811}, "4": {"facts_total": 61, "facts_retained": 7, "retention_rate": 0.11475409836065574, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.3076923076923077, "status": 0.0, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 811}, "5": {"facts_total": 61, "facts_retained": 7, "retention_rate": 0.11475409836065574, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.3076923076923077, "status": 0.0, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 811}, "6": {"facts_total": 61, "facts_retained": 7, "retention_rate": 0.11475409836065574, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.3076923076923077, "status": 0.0, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 811}, "7": {"facts_total": 61, "facts_retained": 7, "retention_rate": 0.11475409836065574, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.3076923076923077, "status": 0.0, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 811}, "8": {"facts_total": 61, "facts_retained": 7, "retention_rate": 0.11475409836065574, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.3076923076923077, "status": 0.0, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 811}}}, "reference_key_id": "keys/synthetic/syn_03015_2000.json", "task_scores": {"correct": 0.0, "abstained": 0.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "4f973bd7c40fcf07e373e29e86c8315cdbd7790f931726a9494ca5cf77a0b82e", "output_hash": "e83edb3f38a69e604be343e63e2e84442156fcb2bb30f36ba9716575d5ffd57a"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "8644a1cad825525d9dfb13b7", "attempt_id": "0f919f95237c", "run_group_id": "E4_core:A5|closed|qwen3b-llama3b:h8", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:04:56.057234Z", "task_family": "synthetic", "document_id": "syn_03015_2000", "document_hash": "71a7284b9bc3f681d69d91d18b9e8d45a4dda486cd14f4145395786ac6a68a89", "question_id": "q10_temporal", "split": "test", "cluster_id": "syn_03015_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "tokenizer_hashes": {"receiver": "b9d8872cf3a99da99e1fc4b0f71232a451632dc49ca4c6771549b4efa5c1c763"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 8, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1741, "query_tokens": 409, "message_tokens": 288, "message_utf8_bytes": 811, "cache_tensor_bytes": null, "serialized_payload_bytes": 811, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 111914.81091594324, "hops": {"1": 22787.491958006285, "2": 11425.557625014335, "3": 14560.2815420134, "4": 11323.241332895122, "5": 14510.312750004232, "6": 11364.853915991262, "7": 14467.890792060643, "8": 11475.180999957956}, "receiver_prefill_ms": 311.04945798870176, "ttft_ms": 40.73020804207772, "decode_ms": 53.84716601110995}, "total_wall_ms": 112320.94333192799, "ttft_ms": 40.73020804207772, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 61, "facts_retained": 7, "retention_rate": 0.11475409836065574, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.3076923076923077, "status": 0.0, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 811}, "ledger_by_hop": {"1": {"facts_total": 61, "facts_retained": 7, "retention_rate": 0.11475409836065574, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.3076923076923077, "status": 0.0, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 690}, "2": {"facts_total": 61, "facts_retained": 7, "retention_rate": 0.11475409836065574, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.3076923076923077, "status": 0.0, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 811}, "3": {"facts_total": 61, "facts_retained": 7, "retention_rate": 0.11475409836065574, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.3076923076923077, "status": 0.0, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 811}, "4": {"facts_total": 61, "facts_retained": 7, "retention_rate": 0.11475409836065574, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.3076923076923077, "status": 0.0, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 811}, "5": {"facts_total": 61, "facts_retained": 7, "retention_rate": 0.11475409836065574, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.3076923076923077, "status": 0.0, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 811}, "6": {"facts_total": 61, "facts_retained": 7, "retention_rate": 0.11475409836065574, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.3076923076923077, "status": 0.0, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 811}, "7": {"facts_total": 61, "facts_retained": 7, "retention_rate": 0.11475409836065574, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.3076923076923077, "status": 0.0, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 811}, "8": {"facts_total": 61, "facts_retained": 7, "retention_rate": 0.11475409836065574, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.3076923076923077, "status": 0.0, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 811}}}, "reference_key_id": "keys/synthetic/syn_03015_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "4f973bd7c40fcf07e373e29e86c8315cdbd7790f931726a9494ca5cf77a0b82e", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "4b4e9a850ef54097856c156c", "attempt_id": "2ba09d5f46d5", "run_group_id": "E4_core:A5|closed|qwen3b-llama3b:h8", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:04:56.481425Z", "task_family": "synthetic", "document_id": "syn_03015_2000", "document_hash": "71a7284b9bc3f681d69d91d18b9e8d45a4dda486cd14f4145395786ac6a68a89", "question_id": "q11_aggregation", "split": "test", "cluster_id": "syn_03015_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "tokenizer_hashes": {"receiver": "b9d8872cf3a99da99e1fc4b0f71232a451632dc49ca4c6771549b4efa5c1c763"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 8, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1741, "query_tokens": 403, "message_tokens": 288, "message_utf8_bytes": 811, "cache_tensor_bytes": null, "serialized_payload_bytes": 811, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 111914.81091594324, "hops": {"1": 22787.491958006285, "2": 11425.557625014335, "3": 14560.2815420134, "4": 11323.241332895122, "5": 14510.312750004232, "6": 11364.853915991262, "7": 14467.890792060643, "8": 11475.180999957956}, "receiver_prefill_ms": 351.32783302105963, "ttft_ms": 129.74441691767424, "decode_ms": 38.79412496462464}, "total_wall_ms": 112435.17679092474, "ttft_ms": 129.74441691767424, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 61, "facts_retained": 7, "retention_rate": 0.11475409836065574, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.3076923076923077, "status": 0.0, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 811}, "ledger_by_hop": {"1": {"facts_total": 61, "facts_retained": 7, "retention_rate": 0.11475409836065574, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.3076923076923077, "status": 0.0, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 690}, "2": {"facts_total": 61, "facts_retained": 7, "retention_rate": 0.11475409836065574, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.3076923076923077, "status": 0.0, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 811}, "3": {"facts_total": 61, "facts_retained": 7, "retention_rate": 0.11475409836065574, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.3076923076923077, "status": 0.0, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 811}, "4": {"facts_total": 61, "facts_retained": 7, "retention_rate": 0.11475409836065574, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.3076923076923077, "status": 0.0, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 811}, "5": {"facts_total": 61, "facts_retained": 7, "retention_rate": 0.11475409836065574, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.3076923076923077, "status": 0.0, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 811}, "6": {"facts_total": 61, "facts_retained": 7, "retention_rate": 0.11475409836065574, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.3076923076923077, "status": 0.0, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 811}, "7": {"facts_total": 61, "facts_retained": 7, "retention_rate": 0.11475409836065574, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.3076923076923077, "status": 0.0, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 811}, "8": {"facts_total": 61, "facts_retained": 7, "retention_rate": 0.11475409836065574, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.3076923076923077, "status": 0.0, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 811}}}, "reference_key_id": "keys/synthetic/syn_03015_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "4f973bd7c40fcf07e373e29e86c8315cdbd7790f931726a9494ca5cf77a0b82e", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "c7becedba7274db39b3d9fac", "attempt_id": "07b992ce0f06", "run_group_id": "E4_core:A5|closed|qwen3b-llama3b:h8", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:04:57.019260Z", "task_family": "synthetic", "document_id": "syn_03015_2000", "document_hash": "71a7284b9bc3f681d69d91d18b9e8d45a4dda486cd14f4145395786ac6a68a89", "question_id": "q12_unit_conversion", "split": "test", "cluster_id": "syn_03015_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "tokenizer_hashes": {"receiver": "b9d8872cf3a99da99e1fc4b0f71232a451632dc49ca4c6771549b4efa5c1c763"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 8, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1741, "query_tokens": 400, "message_tokens": 288, "message_utf8_bytes": 811, "cache_tensor_bytes": null, "serialized_payload_bytes": 811, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 111914.81091594324, "hops": {"1": 22787.491958006285, "2": 11425.557625014335, "3": 14560.2815420134, "4": 11323.241332895122, "5": 14510.312750004232, "6": 11364.853915991262, "7": 14467.890792060643, "8": 11475.180999957956}, "receiver_prefill_ms": 354.3905420228839, "ttft_ms": 20.810249960049987, "decode_ms": 37.60083403903991}, "total_wall_ms": 112328.09008297045, "ttft_ms": 20.810249960049987, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 61, "facts_retained": 7, "retention_rate": 0.11475409836065574, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.3076923076923077, "status": 0.0, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 811}, "ledger_by_hop": {"1": {"facts_total": 61, "facts_retained": 7, "retention_rate": 0.11475409836065574, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.3076923076923077, "status": 0.0, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 690}, "2": {"facts_total": 61, "facts_retained": 7, "retention_rate": 0.11475409836065574, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.3076923076923077, "status": 0.0, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 811}, "3": {"facts_total": 61, "facts_retained": 7, "retention_rate": 0.11475409836065574, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.3076923076923077, "status": 0.0, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 811}, "4": {"facts_total": 61, "facts_retained": 7, "retention_rate": 0.11475409836065574, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.3076923076923077, "status": 0.0, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 811}, "5": {"facts_total": 61, "facts_retained": 7, "retention_rate": 0.11475409836065574, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.3076923076923077, "status": 0.0, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 811}, "6": {"facts_total": 61, "facts_retained": 7, "retention_rate": 0.11475409836065574, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.3076923076923077, "status": 0.0, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 811}, "7": {"facts_total": 61, "facts_retained": 7, "retention_rate": 0.11475409836065574, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.3076923076923077, "status": 0.0, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 811}, "8": {"facts_total": 61, "facts_retained": 7, "retention_rate": 0.11475409836065574, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.3076923076923077, "status": 0.0, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 811}}}, "reference_key_id": "keys/synthetic/syn_03015_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "4f973bd7c40fcf07e373e29e86c8315cdbd7790f931726a9494ca5cf77a0b82e", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "cf96f68fda95883f0ef911c8", "attempt_id": "b0bf1456773f", "run_group_id": "E4_core:A5|closed|qwen3b-llama3b:h8", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:04:57.450293Z", "task_family": "synthetic", "document_id": "syn_03015_2000", "document_hash": "71a7284b9bc3f681d69d91d18b9e8d45a4dda486cd14f4145395786ac6a68a89", "question_id": "q13_unanswerable", "split": "test", "cluster_id": "syn_03015_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "tokenizer_hashes": {"receiver": "b9d8872cf3a99da99e1fc4b0f71232a451632dc49ca4c6771549b4efa5c1c763"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 8, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1741, "query_tokens": 403, "message_tokens": 288, "message_utf8_bytes": 811, "cache_tensor_bytes": null, "serialized_payload_bytes": 811, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 111914.81091594324, "hops": {"1": 22787.491958006285, "2": 11425.557625014335, "3": 14560.2815420134, "4": 11323.241332895122, "5": 14510.312750004232, "6": 11364.853915991262, "7": 14467.890792060643, "8": 11475.180999957956}, "receiver_prefill_ms": 338.39670894667506, "ttft_ms": 31.30816703196615, "decode_ms": 20.238082972355187}, "total_wall_ms": 112305.16833299771, "ttft_ms": 31.30816703196615, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 61, "facts_retained": 7, "retention_rate": 0.11475409836065574, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.3076923076923077, "status": 0.0, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 811}, "ledger_by_hop": {"1": {"facts_total": 61, "facts_retained": 7, "retention_rate": 0.11475409836065574, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.3076923076923077, "status": 0.0, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 690}, "2": {"facts_total": 61, "facts_retained": 7, "retention_rate": 0.11475409836065574, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.3076923076923077, "status": 0.0, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 811}, "3": {"facts_total": 61, "facts_retained": 7, "retention_rate": 0.11475409836065574, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.3076923076923077, "status": 0.0, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 811}, "4": {"facts_total": 61, "facts_retained": 7, "retention_rate": 0.11475409836065574, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.3076923076923077, "status": 0.0, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 811}, "5": {"facts_total": 61, "facts_retained": 7, "retention_rate": 0.11475409836065574, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.3076923076923077, "status": 0.0, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 811}, "6": {"facts_total": 61, "facts_retained": 7, "retention_rate": 0.11475409836065574, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.3076923076923077, "status": 0.0, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 811}, "7": {"facts_total": 61, "facts_retained": 7, "retention_rate": 0.11475409836065574, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.3076923076923077, "status": 0.0, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 811}, "8": {"facts_total": 61, "facts_retained": 7, "retention_rate": 0.11475409836065574, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.3076923076923077, "status": 0.0, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 811}}}, "reference_key_id": "keys/synthetic/syn_03015_2000.json", "task_scores": {"correct": 1.0, "abstained": 1.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "4f973bd7c40fcf07e373e29e86c8315cdbd7790f931726a9494ca5cf77a0b82e", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "317a1472950b092bb1317168", "attempt_id": "d275b04caeea", "run_group_id": "E4_core:A5|closed|qwen3b-llama3b:h8", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:04:57.858417Z", "task_family": "synthetic", "document_id": "syn_03015_2000", "document_hash": "71a7284b9bc3f681d69d91d18b9e8d45a4dda486cd14f4145395786ac6a68a89", "question_id": "q14_unanswerable", "split": "test", "cluster_id": "syn_03015_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "tokenizer_hashes": {"receiver": "b9d8872cf3a99da99e1fc4b0f71232a451632dc49ca4c6771549b4efa5c1c763"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 8, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1741, "query_tokens": 405, "message_tokens": 288, "message_utf8_bytes": 811, "cache_tensor_bytes": null, "serialized_payload_bytes": 811, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 111914.81091594324, "hops": {"1": 22787.491958006285, "2": 11425.557625014335, "3": 14560.2815420134, "4": 11323.241332895122, "5": 14510.312750004232, "6": 11364.853915991262, "7": 14467.890792060643, "8": 11475.180999957956}, "receiver_prefill_ms": 304.52075006905943, "ttft_ms": 59.97804191429168, "decode_ms": 50.83116691093892}, "total_wall_ms": 112330.65587398596, "ttft_ms": 59.97804191429168, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 61, "facts_retained": 7, "retention_rate": 0.11475409836065574, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.3076923076923077, "status": 0.0, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 811}, "ledger_by_hop": {"1": {"facts_total": 61, "facts_retained": 7, "retention_rate": 0.11475409836065574, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.3076923076923077, "status": 0.0, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 690}, "2": {"facts_total": 61, "facts_retained": 7, "retention_rate": 0.11475409836065574, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.3076923076923077, "status": 0.0, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 811}, "3": {"facts_total": 61, "facts_retained": 7, "retention_rate": 0.11475409836065574, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.3076923076923077, "status": 0.0, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 811}, "4": {"facts_total": 61, "facts_retained": 7, "retention_rate": 0.11475409836065574, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.3076923076923077, "status": 0.0, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 811}, "5": {"facts_total": 61, "facts_retained": 7, "retention_rate": 0.11475409836065574, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.3076923076923077, "status": 0.0, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 811}, "6": {"facts_total": 61, "facts_retained": 7, "retention_rate": 0.11475409836065574, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.3076923076923077, "status": 0.0, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 811}, "7": {"facts_total": 61, "facts_retained": 7, "retention_rate": 0.11475409836065574, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.3076923076923077, "status": 0.0, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 811}, "8": {"facts_total": 61, "facts_retained": 7, "retention_rate": 0.11475409836065574, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.3076923076923077, "status": 0.0, "place": 0.0, "identifier": 0.0, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 811}}}, "reference_key_id": "keys/synthetic/syn_03015_2000.json", "task_scores": {"correct": 1.0, "abstained": 1.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "4f973bd7c40fcf07e373e29e86c8315cdbd7790f931726a9494ca5cf77a0b82e", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "924e3a3bc29749381cddad1d", "attempt_id": "1e35ec3ac812", "run_group_id": "E4_core:A4|closed|qwen3b:h1", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:05:21.214043Z", "task_family": "synthetic", "document_id": "syn_03018_2000", "document_hash": "d284eb07e27da742b9c2b67abccfdf72e4dd67c3259f44f3b41e34dd3e6fc1ff", "question_id": "q00_single_fact", "split": "test", "cluster_id": "syn_03018_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 1, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1982, "query_tokens": 595, "message_tokens": 512, "message_utf8_bytes": 1181, "cache_tensor_bytes": null, "serialized_payload_bytes": 1181, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 22836.384875001386, "hops": {"1": 22836.384875001386}, "receiver_prefill_ms": 367.6254170713946, "ttft_ms": 18.78579205367714, "decode_ms": 45.42887501884252}, "total_wall_ms": 23269.139166921377, "ttft_ms": 18.78579205367714, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 64, "facts_retained": 22, "retention_rate": 0.34375, "retention_by_kind": {"person": 0.5, "date": 0.5555555555555556, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1181}, "ledger_by_hop": {"1": {"facts_total": 64, "facts_retained": 22, "retention_rate": 0.34375, "retention_by_kind": {"person": 0.5, "date": 0.5555555555555556, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1181}}}, "reference_key_id": "keys/synthetic/syn_03018_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "499392c3597fdf0fd842c7907d9598f7d57a2cf7b9c4348cfba91a80adac26ec", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "ab2d8614de0ded2d8199b319", "attempt_id": "07ff21c42ddb", "run_group_id": "E4_core:A4|closed|qwen3b:h1", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:05:21.664370Z", "task_family": "synthetic", "document_id": "syn_03018_2000", "document_hash": "d284eb07e27da742b9c2b67abccfdf72e4dd67c3259f44f3b41e34dd3e6fc1ff", "question_id": "q01_single_fact", "split": "test", "cluster_id": "syn_03018_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 1, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1982, "query_tokens": 601, "message_tokens": 512, "message_utf8_bytes": 1181, "cache_tensor_bytes": null, "serialized_payload_bytes": 1181, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 22836.384875001386, "hops": {"1": 22836.384875001386}, "receiver_prefill_ms": 396.2372919777408, "ttft_ms": 36.908541922457516, "decode_ms": 573.8817499950528}, "total_wall_ms": 23844.12933397107, "ttft_ms": 36.908541922457516, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 12}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "Lot 376-E, Baie-Comeau", "evidence_artifact": {"ledger": {"facts_total": 64, "facts_retained": 22, "retention_rate": 0.34375, "retention_by_kind": {"person": 0.5, "date": 0.5555555555555556, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1181}, "ledger_by_hop": {"1": {"facts_total": 64, "facts_retained": 22, "retention_rate": 0.34375, "retention_by_kind": {"person": 0.5, "date": 0.5555555555555556, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1181}}}, "reference_key_id": "keys/synthetic/syn_03018_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 0.8571428571428571}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "499392c3597fdf0fd842c7907d9598f7d57a2cf7b9c4348cfba91a80adac26ec", "output_hash": "b58829a1440993c4af3307ba44494f021d726fd11e9337888b6b852a8b37351a"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "e5e0d5e57036df88c0c11690", "attempt_id": "dd868ccf6967", "run_group_id": "E4_core:A4|closed|qwen3b:h1", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:05:22.689055Z", "task_family": "synthetic", "document_id": "syn_03018_2000", "document_hash": "d284eb07e27da742b9c2b67abccfdf72e4dd67c3259f44f3b41e34dd3e6fc1ff", "question_id": "q02_exact_copy", "split": "test", "cluster_id": "syn_03018_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 1, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1982, "query_tokens": 600, "message_tokens": 512, "message_utf8_bytes": 1181, "cache_tensor_bytes": null, "serialized_payload_bytes": 1181, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 22836.384875001386, "hops": {"1": 22836.384875001386}, "receiver_prefill_ms": 408.0706670647487, "ttft_ms": 35.700792097486556, "decode_ms": 32.07066608592868}, "total_wall_ms": 23312.882750062272, "ttft_ms": 35.700792097486556, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 64, "facts_retained": 22, "retention_rate": 0.34375, "retention_by_kind": {"person": 0.5, "date": 0.5555555555555556, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1181}, "ledger_by_hop": {"1": {"facts_total": 64, "facts_retained": 22, "retention_rate": 0.34375, "retention_by_kind": {"person": 0.5, "date": 0.5555555555555556, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1181}}}, "reference_key_id": "keys/synthetic/syn_03018_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "499392c3597fdf0fd842c7907d9598f7d57a2cf7b9c4348cfba91a80adac26ec", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "a09d6fcb51e6cfa02523c416", "attempt_id": "a9461bff5521", "run_group_id": "E4_core:A4|closed|qwen3b:h1", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:05:23.183127Z", "task_family": "synthetic", "document_id": "syn_03018_2000", "document_hash": "d284eb07e27da742b9c2b67abccfdf72e4dd67c3259f44f3b41e34dd3e6fc1ff", "question_id": "q03_exact_copy", "split": "test", "cluster_id": "syn_03018_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 1, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1982, "query_tokens": 598, "message_tokens": 512, "message_utf8_bytes": 1181, "cache_tensor_bytes": null, "serialized_payload_bytes": 1181, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 22836.384875001386, "hops": {"1": 22836.384875001386}, "receiver_prefill_ms": 380.5222499649972, "ttft_ms": 54.16554200928658, "decode_ms": 474.62637501303107}, "total_wall_ms": 23746.367583051324, "ttft_ms": 54.16554200928658, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 9}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "Priyamvada Sørlie", "evidence_artifact": {"ledger": {"facts_total": 64, "facts_retained": 22, "retention_rate": 0.34375, "retention_by_kind": {"person": 0.5, "date": 0.5555555555555556, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1181}, "ledger_by_hop": {"1": {"facts_total": 64, "facts_retained": 22, "retention_rate": 0.34375, "retention_by_kind": {"person": 0.5, "date": 0.5555555555555556, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1181}}}, "reference_key_id": "keys/synthetic/syn_03018_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "499392c3597fdf0fd842c7907d9598f7d57a2cf7b9c4348cfba91a80adac26ec", "output_hash": "c5f554196ed23dd71ad995ec89f7142099d524dbb6534cc10d94a0c89d13e792"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "1e3aa0ed94f5ab637c48b20e", "attempt_id": "917a4fb61d45", "run_group_id": "E4_core:A4|closed|qwen3b:h1", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:05:24.111654Z", "task_family": "synthetic", "document_id": "syn_03018_2000", "document_hash": "d284eb07e27da742b9c2b67abccfdf72e4dd67c3259f44f3b41e34dd3e6fc1ff", "question_id": "q04_multi_hop", "split": "test", "cluster_id": "syn_03018_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 1, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1982, "query_tokens": 604, "message_tokens": 512, "message_utf8_bytes": 1181, "cache_tensor_bytes": null, "serialized_payload_bytes": 1181, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 22836.384875001386, "hops": {"1": 22836.384875001386}, "receiver_prefill_ms": 405.9368330053985, "ttft_ms": 34.62699998635799, "decode_ms": 32.60174999013543}, "total_wall_ms": 23310.333583969623, "ttft_ms": 34.62699998635799, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 64, "facts_retained": 22, "retention_rate": 0.34375, "retention_by_kind": {"person": 0.5, "date": 0.5555555555555556, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1181}, "ledger_by_hop": {"1": {"facts_total": 64, "facts_retained": 22, "retention_rate": 0.34375, "retention_by_kind": {"person": 0.5, "date": 0.5555555555555556, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1181}}}, "reference_key_id": "keys/synthetic/syn_03018_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "499392c3597fdf0fd842c7907d9598f7d57a2cf7b9c4348cfba91a80adac26ec", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "0dccb7bb356cf8c80fbd2119", "attempt_id": "b27b4d583027", "run_group_id": "E4_core:A4|closed|qwen3b:h1", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:05:24.603365Z", "task_family": "synthetic", "document_id": "syn_03018_2000", "document_hash": "d284eb07e27da742b9c2b67abccfdf72e4dd67c3259f44f3b41e34dd3e6fc1ff", "question_id": "q05_multi_hop", "split": "test", "cluster_id": "syn_03018_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 1, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1982, "query_tokens": 604, "message_tokens": 512, "message_utf8_bytes": 1181, "cache_tensor_bytes": null, "serialized_payload_bytes": 1181, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 22836.384875001386, "hops": {"1": 22836.384875001386}, "receiver_prefill_ms": 360.79304199665785, "ttft_ms": 49.51404198072851, "decode_ms": 615.7403340330347}, "total_wall_ms": 23863.223375054076, "ttft_ms": 49.51404198072851, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 10}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "HM-38K-5533", "evidence_artifact": {"ledger": {"facts_total": 64, "facts_retained": 22, "retention_rate": 0.34375, "retention_by_kind": {"person": 0.5, "date": 0.5555555555555556, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1181}, "ledger_by_hop": {"1": {"facts_total": 64, "facts_retained": 22, "retention_rate": 0.34375, "retention_by_kind": {"person": 0.5, "date": 0.5555555555555556, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1181}}}, "reference_key_id": "keys/synthetic/syn_03018_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "499392c3597fdf0fd842c7907d9598f7d57a2cf7b9c4348cfba91a80adac26ec", "output_hash": "18acd9807857243bbded3bceae55398316a3958b81d70364441bdb6748eaf878"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "4a33924ccc988939ed3984b5", "attempt_id": "d36dc110f446", "run_group_id": "E4_core:A4|closed|qwen3b:h1", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:05:25.647285Z", "task_family": "synthetic", "document_id": "syn_03018_2000", "document_hash": "d284eb07e27da742b9c2b67abccfdf72e4dd67c3259f44f3b41e34dd3e6fc1ff", "question_id": "q06_negation", "split": "test", "cluster_id": "syn_03018_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 1, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1982, "query_tokens": 596, "message_tokens": 512, "message_utf8_bytes": 1181, "cache_tensor_bytes": null, "serialized_payload_bytes": 1181, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 22836.384875001386, "hops": {"1": 22836.384875001386}, "receiver_prefill_ms": 395.58541600126773, "ttft_ms": 34.764416981488466, "decode_ms": 29.840542003512383}, "total_wall_ms": 23297.246958012693, "ttft_ms": 34.764416981488466, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 64, "facts_retained": 22, "retention_rate": 0.34375, "retention_by_kind": {"person": 0.5, "date": 0.5555555555555556, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1181}, "ledger_by_hop": {"1": {"facts_total": 64, "facts_retained": 22, "retention_rate": 0.34375, "retention_by_kind": {"person": 0.5, "date": 0.5555555555555556, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1181}}}, "reference_key_id": "keys/synthetic/syn_03018_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "499392c3597fdf0fd842c7907d9598f7d57a2cf7b9c4348cfba91a80adac26ec", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "1aa02ad7ba836764f709e06a", "attempt_id": "060894ba1186", "run_group_id": "E4_core:A4|closed|qwen3b:h1", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:05:26.125159Z", "task_family": "synthetic", "document_id": "syn_03018_2000", "document_hash": "d284eb07e27da742b9c2b67abccfdf72e4dd67c3259f44f3b41e34dd3e6fc1ff", "question_id": "q07_negation", "split": "test", "cluster_id": "syn_03018_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 1, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1982, "query_tokens": 595, "message_tokens": 512, "message_utf8_bytes": 1181, "cache_tensor_bytes": null, "serialized_payload_bytes": 1181, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 22836.384875001386, "hops": {"1": 22836.384875001386}, "receiver_prefill_ms": 380.65179204568267, "ttft_ms": 20.644415984861553, "decode_ms": 38.69212500285357}, "total_wall_ms": 23277.069707983173, "ttft_ms": 20.644415984861553, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "no", "evidence_artifact": {"ledger": {"facts_total": 64, "facts_retained": 22, "retention_rate": 0.34375, "retention_by_kind": {"person": 0.5, "date": 0.5555555555555556, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1181}, "ledger_by_hop": {"1": {"facts_total": 64, "facts_retained": 22, "retention_rate": 0.34375, "retention_by_kind": {"person": 0.5, "date": 0.5555555555555556, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1181}}}, "reference_key_id": "keys/synthetic/syn_03018_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "499392c3597fdf0fd842c7907d9598f7d57a2cf7b9c4348cfba91a80adac26ec", "output_hash": "9390298f3fb0c5b160498935d79cb139aef28e1c47358b4bbba61862b9c26e59"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "098a785f3d81bc4da590b894", "attempt_id": "8a208143a7c1", "run_group_id": "E4_core:A4|closed|qwen3b:h1", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:05:26.583601Z", "task_family": "synthetic", "document_id": "syn_03018_2000", "document_hash": "d284eb07e27da742b9c2b67abccfdf72e4dd67c3259f44f3b41e34dd3e6fc1ff", "question_id": "q08_revision", "split": "test", "cluster_id": "syn_03018_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 1, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1982, "query_tokens": 613, "message_tokens": 512, "message_utf8_bytes": 1181, "cache_tensor_bytes": null, "serialized_payload_bytes": 1181, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 22836.384875001386, "hops": {"1": 22836.384875001386}, "receiver_prefill_ms": 395.10524994693696, "ttft_ms": 47.29479202069342, "decode_ms": 360.74024997651577}, "total_wall_ms": 23640.227500000037, "ttft_ms": 47.29479202069342, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 6}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "304815", "evidence_artifact": {"ledger": {"facts_total": 64, "facts_retained": 22, "retention_rate": 0.34375, "retention_by_kind": {"person": 0.5, "date": 0.5555555555555556, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1181}, "ledger_by_hop": {"1": {"facts_total": 64, "facts_retained": 22, "retention_rate": 0.34375, "retention_by_kind": {"person": 0.5, "date": 0.5555555555555556, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1181}}}, "reference_key_id": "keys/synthetic/syn_03018_2000.json", "task_scores": {"correct": 0.0, "abstained": 0.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "499392c3597fdf0fd842c7907d9598f7d57a2cf7b9c4348cfba91a80adac26ec", "output_hash": "a480475d6855141c920f42d1bf3337ca7fd88bcf4087b0ed2e253ed68cca6926"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "82388a0342270f97b08c3ad0", "attempt_id": "a90f8f7822b7", "run_group_id": "E4_core:A4|closed|qwen3b:h1", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:05:27.405095Z", "task_family": "synthetic", "document_id": "syn_03018_2000", "document_hash": "d284eb07e27da742b9c2b67abccfdf72e4dd67c3259f44f3b41e34dd3e6fc1ff", "question_id": "q09_revision", "split": "test", "cluster_id": "syn_03018_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 1, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1982, "query_tokens": 606, "message_tokens": 512, "message_utf8_bytes": 1181, "cache_tensor_bytes": null, "serialized_payload_bytes": 1181, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 22836.384875001386, "hops": {"1": 22836.384875001386}, "receiver_prefill_ms": 405.8764170622453, "ttft_ms": 27.660832973197103, "decode_ms": 218.52304099593312}, "total_wall_ms": 23489.221125026233, "ttft_ms": 27.660832973197103, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 6}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "276340", "evidence_artifact": {"ledger": {"facts_total": 64, "facts_retained": 22, "retention_rate": 0.34375, "retention_by_kind": {"person": 0.5, "date": 0.5555555555555556, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1181}, "ledger_by_hop": {"1": {"facts_total": 64, "facts_retained": 22, "retention_rate": 0.34375, "retention_by_kind": {"person": 0.5, "date": 0.5555555555555556, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1181}}}, "reference_key_id": "keys/synthetic/syn_03018_2000.json", "task_scores": {"correct": 0.0, "abstained": 0.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "499392c3597fdf0fd842c7907d9598f7d57a2cf7b9c4348cfba91a80adac26ec", "output_hash": "bc7c3667cbe0ad74ec0a1b092d3bad7e6a02978fa67746e3adbf3174ed34b3a2"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "1a57629bc4a235dd7dec8f43", "attempt_id": "ea22f7ea7a95", "run_group_id": "E4_core:A4|closed|qwen3b:h1", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:05:28.075775Z", "task_family": "synthetic", "document_id": "syn_03018_2000", "document_hash": "d284eb07e27da742b9c2b67abccfdf72e4dd67c3259f44f3b41e34dd3e6fc1ff", "question_id": "q10_temporal", "split": "test", "cluster_id": "syn_03018_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 1, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1982, "query_tokens": 614, "message_tokens": 512, "message_utf8_bytes": 1181, "cache_tensor_bytes": null, "serialized_payload_bytes": 1181, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 22836.384875001386, "hops": {"1": 22836.384875001386}, "receiver_prefill_ms": 349.96183298062533, "ttft_ms": 57.49562499113381, "decode_ms": 409.8137919791043}, "total_wall_ms": 23654.58062500693, "ttft_ms": 57.49562499113381, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 6}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "Melymi-15", "evidence_artifact": {"ledger": {"facts_total": 64, "facts_retained": 22, "retention_rate": 0.34375, "retention_by_kind": {"person": 0.5, "date": 0.5555555555555556, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1181}, "ledger_by_hop": {"1": {"facts_total": 64, "facts_retained": 22, "retention_rate": 0.34375, "retention_by_kind": {"person": 0.5, "date": 0.5555555555555556, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1181}}}, "reference_key_id": "keys/synthetic/syn_03018_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "499392c3597fdf0fd842c7907d9598f7d57a2cf7b9c4348cfba91a80adac26ec", "output_hash": "fa08d1474b01d6dfbf3dea049be251aa3c3021c17b4818258064b156509af58b"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "cfe7479471a998b9e7192e4b", "attempt_id": "f4820ef1360d", "run_group_id": "E4_core:A4|closed|qwen3b:h1", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:05:28.912642Z", "task_family": "synthetic", "document_id": "syn_03018_2000", "document_hash": "d284eb07e27da742b9c2b67abccfdf72e4dd67c3259f44f3b41e34dd3e6fc1ff", "question_id": "q11_aggregation", "split": "test", "cluster_id": "syn_03018_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 1, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1982, "query_tokens": 606, "message_tokens": 512, "message_utf8_bytes": 1181, "cache_tensor_bytes": null, "serialized_payload_bytes": 1181, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 22836.384875001386, "hops": {"1": 22836.384875001386}, "receiver_prefill_ms": 396.0942089324817, "ttft_ms": 34.43633404094726, "decode_ms": 226.99937492143363}, "total_wall_ms": 23494.685790967196, "ttft_ms": 34.43633404094726, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 7}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "2414120", "evidence_artifact": {"ledger": {"facts_total": 64, "facts_retained": 22, "retention_rate": 0.34375, "retention_by_kind": {"person": 0.5, "date": 0.5555555555555556, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1181}, "ledger_by_hop": {"1": {"facts_total": 64, "facts_retained": 22, "retention_rate": 0.34375, "retention_by_kind": {"person": 0.5, "date": 0.5555555555555556, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1181}}}, "reference_key_id": "keys/synthetic/syn_03018_2000.json", "task_scores": {"correct": 0.0, "abstained": 0.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "499392c3597fdf0fd842c7907d9598f7d57a2cf7b9c4348cfba91a80adac26ec", "output_hash": "0a242f83095c1c1c7e4a3cd4bb88035f03abb655b6cf03cf7962945bb269a5e8"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "d95d1b1724f762387cbacb57", "attempt_id": "73fea84223e4", "run_group_id": "E4_core:A4|closed|qwen3b:h1", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:05:29.589420Z", "task_family": "synthetic", "document_id": "syn_03018_2000", "document_hash": "d284eb07e27da742b9c2b67abccfdf72e4dd67c3259f44f3b41e34dd3e6fc1ff", "question_id": "q12_unit_conversion", "split": "test", "cluster_id": "syn_03018_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 1, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1982, "query_tokens": 603, "message_tokens": 512, "message_utf8_bytes": 1181, "cache_tensor_bytes": null, "serialized_payload_bytes": 1181, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 22836.384875001386, "hops": {"1": 22836.384875001386}, "receiver_prefill_ms": 353.6238339729607, "ttft_ms": 48.00900002010167, "decode_ms": 54.37241599429399}, "total_wall_ms": 23293.234916985966, "ttft_ms": 48.00900002010167, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 64, "facts_retained": 22, "retention_rate": 0.34375, "retention_by_kind": {"person": 0.5, "date": 0.5555555555555556, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1181}, "ledger_by_hop": {"1": {"facts_total": 64, "facts_retained": 22, "retention_rate": 0.34375, "retention_by_kind": {"person": 0.5, "date": 0.5555555555555556, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1181}}}, "reference_key_id": "keys/synthetic/syn_03018_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "499392c3597fdf0fd842c7907d9598f7d57a2cf7b9c4348cfba91a80adac26ec", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "484f35718c45912bdcf70e9e", "attempt_id": "acc09f7dd8e2", "run_group_id": "E4_core:A4|closed|qwen3b:h1", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:05:30.063956Z", "task_family": "synthetic", "document_id": "syn_03018_2000", "document_hash": "d284eb07e27da742b9c2b67abccfdf72e4dd67c3259f44f3b41e34dd3e6fc1ff", "question_id": "q13_unanswerable", "split": "test", "cluster_id": "syn_03018_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 1, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1982, "query_tokens": 607, "message_tokens": 512, "message_utf8_bytes": 1181, "cache_tensor_bytes": null, "serialized_payload_bytes": 1181, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 22836.384875001386, "hops": {"1": 22836.384875001386}, "receiver_prefill_ms": 551.0705829365179, "ttft_ms": 37.835333961993456, "decode_ms": 24.179792031645775}, "total_wall_ms": 23450.15595899895, "ttft_ms": 37.835333961993456, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 64, "facts_retained": 22, "retention_rate": 0.34375, "retention_by_kind": {"person": 0.5, "date": 0.5555555555555556, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1181}, "ledger_by_hop": {"1": {"facts_total": 64, "facts_retained": 22, "retention_rate": 0.34375, "retention_by_kind": {"person": 0.5, "date": 0.5555555555555556, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1181}}}, "reference_key_id": "keys/synthetic/syn_03018_2000.json", "task_scores": {"correct": 1.0, "abstained": 1.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "499392c3597fdf0fd842c7907d9598f7d57a2cf7b9c4348cfba91a80adac26ec", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "7b304addc936a9df116f0660", "attempt_id": "7cc895255068", "run_group_id": "E4_core:A4|closed|qwen3b:h1", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:05:30.695550Z", "task_family": "synthetic", "document_id": "syn_03018_2000", "document_hash": "d284eb07e27da742b9c2b67abccfdf72e4dd67c3259f44f3b41e34dd3e6fc1ff", "question_id": "q14_unanswerable", "split": "test", "cluster_id": "syn_03018_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 1, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1982, "query_tokens": 608, "message_tokens": 512, "message_utf8_bytes": 1181, "cache_tensor_bytes": null, "serialized_payload_bytes": 1181, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 22836.384875001386, "hops": {"1": 22836.384875001386}, "receiver_prefill_ms": 400.50979203078896, "ttft_ms": 34.274542005732656, "decode_ms": 31.212375033646822}, "total_wall_ms": 23303.017833037302, "ttft_ms": 34.274542005732656, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 64, "facts_retained": 22, "retention_rate": 0.34375, "retention_by_kind": {"person": 0.5, "date": 0.5555555555555556, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1181}, "ledger_by_hop": {"1": {"facts_total": 64, "facts_retained": 22, "retention_rate": 0.34375, "retention_by_kind": {"person": 0.5, "date": 0.5555555555555556, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1181}}}, "reference_key_id": "keys/synthetic/syn_03018_2000.json", "task_scores": {"correct": 1.0, "abstained": 1.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "499392c3597fdf0fd842c7907d9598f7d57a2cf7b9c4348cfba91a80adac26ec", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "c64d204e3aeddab0b2adbe31", "attempt_id": "d3327dbec4c3", "run_group_id": "E4_core:A4|closed|qwen3b:h2", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:05:53.204297Z", "task_family": "synthetic", "document_id": "syn_03018_2000", "document_hash": "d284eb07e27da742b9c2b67abccfdf72e4dd67c3259f44f3b41e34dd3e6fc1ff", "question_id": "q00_single_fact", "split": "test", "cluster_id": "syn_03018_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 2, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1982, "query_tokens": 594, "message_tokens": 511, "message_utf8_bytes": 1176, "cache_tensor_bytes": null, "serialized_payload_bytes": 1176, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 44859.323624987155, "hops": {"1": 22836.384875001386, "2": 22022.93874998577}, "receiver_prefill_ms": 390.0061669992283, "ttft_ms": 51.46620899904519, "decode_ms": 39.66020804364234}, "total_wall_ms": 45341.2778750062, "ttft_ms": 51.46620899904519, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 64, "facts_retained": 22, "retention_rate": 0.34375, "retention_by_kind": {"person": 0.5, "date": 0.5555555555555556, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1176}, "ledger_by_hop": {"1": {"facts_total": 64, "facts_retained": 22, "retention_rate": 0.34375, "retention_by_kind": {"person": 0.5, "date": 0.5555555555555556, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1181}, "2": {"facts_total": 64, "facts_retained": 22, "retention_rate": 0.34375, "retention_by_kind": {"person": 0.5, "date": 0.5555555555555556, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1176}}}, "reference_key_id": "keys/synthetic/syn_03018_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "89438188d34426dbc3fdf77c29f9462c95075b81b820590bae58487694cd6183", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "eb037c3bb9f62e267d6f9233", "attempt_id": "4dad3b2c62b0", "run_group_id": "E4_core:A4|closed|qwen3b:h2", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:05:53.706247Z", "task_family": "synthetic", "document_id": "syn_03018_2000", "document_hash": "d284eb07e27da742b9c2b67abccfdf72e4dd67c3259f44f3b41e34dd3e6fc1ff", "question_id": "q01_single_fact", "split": "test", "cluster_id": "syn_03018_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 2, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1982, "query_tokens": 600, "message_tokens": 511, "message_utf8_bytes": 1176, "cache_tensor_bytes": null, "serialized_payload_bytes": 1176, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 44859.323624987155, "hops": {"1": 22836.384875001386, "2": 22022.93874998577}, "receiver_prefill_ms": 521.9700000016019, "ttft_ms": 36.713290959596634, "decode_ms": 387.6465830253437}, "total_wall_ms": 45806.41745892353, "ttft_ms": 36.713290959596634, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 12}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "Lot 376-E, Baie-Comeau", "evidence_artifact": {"ledger": {"facts_total": 64, "facts_retained": 22, "retention_rate": 0.34375, "retention_by_kind": {"person": 0.5, "date": 0.5555555555555556, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1176}, "ledger_by_hop": {"1": {"facts_total": 64, "facts_retained": 22, "retention_rate": 0.34375, "retention_by_kind": {"person": 0.5, "date": 0.5555555555555556, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1181}, "2": {"facts_total": 64, "facts_retained": 22, "retention_rate": 0.34375, "retention_by_kind": {"person": 0.5, "date": 0.5555555555555556, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1176}}}, "reference_key_id": "keys/synthetic/syn_03018_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 0.8571428571428571}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "89438188d34426dbc3fdf77c29f9462c95075b81b820590bae58487694cd6183", "output_hash": "b58829a1440993c4af3307ba44494f021d726fd11e9337888b6b852a8b37351a"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "d810130994fb821c35753300", "attempt_id": "4116c09fef45", "run_group_id": "E4_core:A4|closed|qwen3b:h2", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:05:54.670960Z", "task_family": "synthetic", "document_id": "syn_03018_2000", "document_hash": "d284eb07e27da742b9c2b67abccfdf72e4dd67c3259f44f3b41e34dd3e6fc1ff", "question_id": "q02_exact_copy", "split": "test", "cluster_id": "syn_03018_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 2, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1982, "query_tokens": 599, "message_tokens": 511, "message_utf8_bytes": 1176, "cache_tensor_bytes": null, "serialized_payload_bytes": 1176, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 44859.323624987155, "hops": {"1": 22836.384875001386, "2": 22022.93874998577}, "receiver_prefill_ms": 384.58233303390443, "ttft_ms": 42.94504097197205, "decode_ms": 71.78566697984934}, "total_wall_ms": 45359.3084999593, "ttft_ms": 42.94504097197205, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 64, "facts_retained": 22, "retention_rate": 0.34375, "retention_by_kind": {"person": 0.5, "date": 0.5555555555555556, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1176}, "ledger_by_hop": {"1": {"facts_total": 64, "facts_retained": 22, "retention_rate": 0.34375, "retention_by_kind": {"person": 0.5, "date": 0.5555555555555556, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1181}, "2": {"facts_total": 64, "facts_retained": 22, "retention_rate": 0.34375, "retention_by_kind": {"person": 0.5, "date": 0.5555555555555556, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1176}}}, "reference_key_id": "keys/synthetic/syn_03018_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "89438188d34426dbc3fdf77c29f9462c95075b81b820590bae58487694cd6183", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "0ad5c8eea7f91c1a88253c10", "attempt_id": "3fe6717d26e2", "run_group_id": "E4_core:A4|closed|qwen3b:h2", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:05:55.187936Z", "task_family": "synthetic", "document_id": "syn_03018_2000", "document_hash": "d284eb07e27da742b9c2b67abccfdf72e4dd67c3259f44f3b41e34dd3e6fc1ff", "question_id": "q03_exact_copy", "split": "test", "cluster_id": "syn_03018_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 2, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1982, "query_tokens": 597, "message_tokens": 511, "message_utf8_bytes": 1176, "cache_tensor_bytes": null, "serialized_payload_bytes": 1176, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 44859.323624987155, "hops": {"1": 22836.384875001386, "2": 22022.93874998577}, "receiver_prefill_ms": 523.088000016287, "ttft_ms": 36.57850006129593, "decode_ms": 297.030333080329}, "total_wall_ms": 45716.93145891186, "ttft_ms": 36.57850006129593, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 9}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "Priyamvada Sørlie", "evidence_artifact": {"ledger": {"facts_total": 64, "facts_retained": 22, "retention_rate": 0.34375, "retention_by_kind": {"person": 0.5, "date": 0.5555555555555556, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1176}, "ledger_by_hop": {"1": {"facts_total": 64, "facts_retained": 22, "retention_rate": 0.34375, "retention_by_kind": {"person": 0.5, "date": 0.5555555555555556, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1181}, "2": {"facts_total": 64, "facts_retained": 22, "retention_rate": 0.34375, "retention_by_kind": {"person": 0.5, "date": 0.5555555555555556, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1176}}}, "reference_key_id": "keys/synthetic/syn_03018_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "89438188d34426dbc3fdf77c29f9462c95075b81b820590bae58487694cd6183", "output_hash": "c5f554196ed23dd71ad995ec89f7142099d524dbb6534cc10d94a0c89d13e792"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "6dfbbf6c3d93eb7ba89c72ca", "attempt_id": "308d94195c02", "run_group_id": "E4_core:A4|closed|qwen3b:h2", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:05:56.063948Z", "task_family": "synthetic", "document_id": "syn_03018_2000", "document_hash": "d284eb07e27da742b9c2b67abccfdf72e4dd67c3259f44f3b41e34dd3e6fc1ff", "question_id": "q04_multi_hop", "split": "test", "cluster_id": "syn_03018_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 2, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1982, "query_tokens": 603, "message_tokens": 511, "message_utf8_bytes": 1176, "cache_tensor_bytes": null, "serialized_payload_bytes": 1176, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 44859.323624987155, "hops": {"1": 22836.384875001386, "2": 22022.93874998577}, "receiver_prefill_ms": 408.7845419999212, "ttft_ms": 31.216875067912042, "decode_ms": 20.334750064648688}, "total_wall_ms": 45320.28545800131, "ttft_ms": 31.216875067912042, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 64, "facts_retained": 22, "retention_rate": 0.34375, "retention_by_kind": {"person": 0.5, "date": 0.5555555555555556, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1176}, "ledger_by_hop": {"1": {"facts_total": 64, "facts_retained": 22, "retention_rate": 0.34375, "retention_by_kind": {"person": 0.5, "date": 0.5555555555555556, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1181}, "2": {"facts_total": 64, "facts_retained": 22, "retention_rate": 0.34375, "retention_by_kind": {"person": 0.5, "date": 0.5555555555555556, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1176}}}, "reference_key_id": "keys/synthetic/syn_03018_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "89438188d34426dbc3fdf77c29f9462c95075b81b820590bae58487694cd6183", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "ef0b997531115bc60feb1246", "attempt_id": "c2f0a3cc4cce", "run_group_id": "E4_core:A4|closed|qwen3b:h2", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:05:56.543162Z", "task_family": "synthetic", "document_id": "syn_03018_2000", "document_hash": "d284eb07e27da742b9c2b67abccfdf72e4dd67c3259f44f3b41e34dd3e6fc1ff", "question_id": "q05_multi_hop", "split": "test", "cluster_id": "syn_03018_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 2, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1982, "query_tokens": 603, "message_tokens": 511, "message_utf8_bytes": 1176, "cache_tensor_bytes": null, "serialized_payload_bytes": 1176, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 44859.323624987155, "hops": {"1": 22836.384875001386, "2": 22022.93874998577}, "receiver_prefill_ms": 367.39170795772225, "ttft_ms": 56.69104098342359, "decode_ms": 518.3175830170512}, "total_wall_ms": 45802.46699997224, "ttft_ms": 56.69104098342359, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 10}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "HM-38K-5533", "evidence_artifact": {"ledger": {"facts_total": 64, "facts_retained": 22, "retention_rate": 0.34375, "retention_by_kind": {"person": 0.5, "date": 0.5555555555555556, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1176}, "ledger_by_hop": {"1": {"facts_total": 64, "facts_retained": 22, "retention_rate": 0.34375, "retention_by_kind": {"person": 0.5, "date": 0.5555555555555556, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1181}, "2": {"facts_total": 64, "facts_retained": 22, "retention_rate": 0.34375, "retention_by_kind": {"person": 0.5, "date": 0.5555555555555556, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1176}}}, "reference_key_id": "keys/synthetic/syn_03018_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "89438188d34426dbc3fdf77c29f9462c95075b81b820590bae58487694cd6183", "output_hash": "18acd9807857243bbded3bceae55398316a3958b81d70364441bdb6748eaf878"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "3736c8720055cabe55dd46b4", "attempt_id": "6970cbc737eb", "run_group_id": "E4_core:A4|closed|qwen3b:h2", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:05:57.502251Z", "task_family": "synthetic", "document_id": "syn_03018_2000", "document_hash": "d284eb07e27da742b9c2b67abccfdf72e4dd67c3259f44f3b41e34dd3e6fc1ff", "question_id": "q06_negation", "split": "test", "cluster_id": "syn_03018_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 2, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1982, "query_tokens": 595, "message_tokens": 511, "message_utf8_bytes": 1176, "cache_tensor_bytes": null, "serialized_payload_bytes": 1176, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 44859.323624987155, "hops": {"1": 22836.384875001386, "2": 22022.93874998577}, "receiver_prefill_ms": 421.4527499862015, "ttft_ms": 35.16170894727111, "decode_ms": 29.222666984423995}, "total_wall_ms": 45345.7234170055, "ttft_ms": 35.16170894727111, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 64, "facts_retained": 22, "retention_rate": 0.34375, "retention_by_kind": {"person": 0.5, "date": 0.5555555555555556, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1176}, "ledger_by_hop": {"1": {"facts_total": 64, "facts_retained": 22, "retention_rate": 0.34375, "retention_by_kind": {"person": 0.5, "date": 0.5555555555555556, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1181}, "2": {"facts_total": 64, "facts_retained": 22, "retention_rate": 0.34375, "retention_by_kind": {"person": 0.5, "date": 0.5555555555555556, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1176}}}, "reference_key_id": "keys/synthetic/syn_03018_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "89438188d34426dbc3fdf77c29f9462c95075b81b820590bae58487694cd6183", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "9a0c854e905648d7e319e97d", "attempt_id": "144653a8c6b2", "run_group_id": "E4_core:A4|closed|qwen3b:h2", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:05:58.006397Z", "task_family": "synthetic", "document_id": "syn_03018_2000", "document_hash": "d284eb07e27da742b9c2b67abccfdf72e4dd67c3259f44f3b41e34dd3e6fc1ff", "question_id": "q07_negation", "split": "test", "cluster_id": "syn_03018_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 2, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1982, "query_tokens": 594, "message_tokens": 511, "message_utf8_bytes": 1176, "cache_tensor_bytes": null, "serialized_payload_bytes": 1176, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 44859.323624987155, "hops": {"1": 22836.384875001386, "2": 22022.93874998577}, "receiver_prefill_ms": 370.552874985151, "ttft_ms": 41.678500012494624, "decode_ms": 56.171709089539945}, "total_wall_ms": 45328.40837503318, "ttft_ms": 41.678500012494624, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "no", "evidence_artifact": {"ledger": {"facts_total": 64, "facts_retained": 22, "retention_rate": 0.34375, "retention_by_kind": {"person": 0.5, "date": 0.5555555555555556, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1176}, "ledger_by_hop": {"1": {"facts_total": 64, "facts_retained": 22, "retention_rate": 0.34375, "retention_by_kind": {"person": 0.5, "date": 0.5555555555555556, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1181}, "2": {"facts_total": 64, "facts_retained": 22, "retention_rate": 0.34375, "retention_by_kind": {"person": 0.5, "date": 0.5555555555555556, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1176}}}, "reference_key_id": "keys/synthetic/syn_03018_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "89438188d34426dbc3fdf77c29f9462c95075b81b820590bae58487694cd6183", "output_hash": "9390298f3fb0c5b160498935d79cb139aef28e1c47358b4bbba61862b9c26e59"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "33399e5badbac7079e583708", "attempt_id": "560d88cc3a00", "run_group_id": "E4_core:A4|closed|qwen3b:h2", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:05:58.492850Z", "task_family": "synthetic", "document_id": "syn_03018_2000", "document_hash": "d284eb07e27da742b9c2b67abccfdf72e4dd67c3259f44f3b41e34dd3e6fc1ff", "question_id": "q08_revision", "split": "test", "cluster_id": "syn_03018_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 2, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1982, "query_tokens": 612, "message_tokens": 511, "message_utf8_bytes": 1176, "cache_tensor_bytes": null, "serialized_payload_bytes": 1176, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 44859.323624987155, "hops": {"1": 22836.384875001386, "2": 22022.93874998577}, "receiver_prefill_ms": 542.3181669320911, "ttft_ms": 27.601583977229893, "decode_ms": 206.555167096667}, "total_wall_ms": 45636.566290981136, "ttft_ms": 27.601583977229893, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 6}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "304815", "evidence_artifact": {"ledger": {"facts_total": 64, "facts_retained": 22, "retention_rate": 0.34375, "retention_by_kind": {"person": 0.5, "date": 0.5555555555555556, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1176}, "ledger_by_hop": {"1": {"facts_total": 64, "facts_retained": 22, "retention_rate": 0.34375, "retention_by_kind": {"person": 0.5, "date": 0.5555555555555556, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1181}, "2": {"facts_total": 64, "facts_retained": 22, "retention_rate": 0.34375, "retention_by_kind": {"person": 0.5, "date": 0.5555555555555556, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1176}}}, "reference_key_id": "keys/synthetic/syn_03018_2000.json", "task_scores": {"correct": 0.0, "abstained": 0.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "89438188d34426dbc3fdf77c29f9462c95075b81b820590bae58487694cd6183", "output_hash": "a480475d6855141c920f42d1bf3337ca7fd88bcf4087b0ed2e253ed68cca6926"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "746f64306e70b48acf26b369", "attempt_id": "289dfee89ad9", "run_group_id": "E4_core:A4|closed|qwen3b:h2", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:05:59.288020Z", "task_family": "synthetic", "document_id": "syn_03018_2000", "document_hash": "d284eb07e27da742b9c2b67abccfdf72e4dd67c3259f44f3b41e34dd3e6fc1ff", "question_id": "q09_revision", "split": "test", "cluster_id": "syn_03018_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 2, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1982, "query_tokens": 605, "message_tokens": 511, "message_utf8_bytes": 1176, "cache_tensor_bytes": null, "serialized_payload_bytes": 1176, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 44859.323624987155, "hops": {"1": 22836.384875001386, "2": 22022.93874998577}, "receiver_prefill_ms": 408.8711249642074, "ttft_ms": 23.97041697986424, "decode_ms": 175.89075001887977}, "total_wall_ms": 45468.87791703921, "ttft_ms": 23.97041697986424, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 6}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "276340", "evidence_artifact": {"ledger": {"facts_total": 64, "facts_retained": 22, "retention_rate": 0.34375, "retention_by_kind": {"person": 0.5, "date": 0.5555555555555556, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1176}, "ledger_by_hop": {"1": {"facts_total": 64, "facts_retained": 22, "retention_rate": 0.34375, "retention_by_kind": {"person": 0.5, "date": 0.5555555555555556, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1181}, "2": {"facts_total": 64, "facts_retained": 22, "retention_rate": 0.34375, "retention_by_kind": {"person": 0.5, "date": 0.5555555555555556, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1176}}}, "reference_key_id": "keys/synthetic/syn_03018_2000.json", "task_scores": {"correct": 0.0, "abstained": 0.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "89438188d34426dbc3fdf77c29f9462c95075b81b820590bae58487694cd6183", "output_hash": "bc7c3667cbe0ad74ec0a1b092d3bad7e6a02978fa67746e3adbf3174ed34b3a2"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "316c0714658ca866f2859925", "attempt_id": "59bf9276ea86", "run_group_id": "E4_core:A4|closed|qwen3b:h2", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:05:59.915542Z", "task_family": "synthetic", "document_id": "syn_03018_2000", "document_hash": "d284eb07e27da742b9c2b67abccfdf72e4dd67c3259f44f3b41e34dd3e6fc1ff", "question_id": "q10_temporal", "split": "test", "cluster_id": "syn_03018_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 2, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1982, "query_tokens": 613, "message_tokens": 511, "message_utf8_bytes": 1176, "cache_tensor_bytes": null, "serialized_payload_bytes": 1176, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 44859.323624987155, "hops": {"1": 22836.384875001386, "2": 22022.93874998577}, "receiver_prefill_ms": 399.6879580663517, "ttft_ms": 38.64975005853921, "decode_ms": 370.82754098810256}, "total_wall_ms": 45669.2139999941, "ttft_ms": 38.64975005853921, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 6}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "Melymi-15", "evidence_artifact": {"ledger": {"facts_total": 64, "facts_retained": 22, "retention_rate": 0.34375, "retention_by_kind": {"person": 0.5, "date": 0.5555555555555556, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1176}, "ledger_by_hop": {"1": {"facts_total": 64, "facts_retained": 22, "retention_rate": 0.34375, "retention_by_kind": {"person": 0.5, "date": 0.5555555555555556, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1181}, "2": {"facts_total": 64, "facts_retained": 22, "retention_rate": 0.34375, "retention_by_kind": {"person": 0.5, "date": 0.5555555555555556, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1176}}}, "reference_key_id": "keys/synthetic/syn_03018_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "89438188d34426dbc3fdf77c29f9462c95075b81b820590bae58487694cd6183", "output_hash": "fa08d1474b01d6dfbf3dea049be251aa3c3021c17b4818258064b156509af58b"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "a30daed4c084286d4a1e99e5", "attempt_id": "a014e5b30a72", "run_group_id": "E4_core:A4|closed|qwen3b:h2", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:06:00.743064Z", "task_family": "synthetic", "document_id": "syn_03018_2000", "document_hash": "d284eb07e27da742b9c2b67abccfdf72e4dd67c3259f44f3b41e34dd3e6fc1ff", "question_id": "q11_aggregation", "split": "test", "cluster_id": "syn_03018_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 2, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1982, "query_tokens": 605, "message_tokens": 511, "message_utf8_bytes": 1176, "cache_tensor_bytes": null, "serialized_payload_bytes": 1176, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 44859.323624987155, "hops": {"1": 22836.384875001386, "2": 22022.93874998577}, "receiver_prefill_ms": 417.6570000126958, "ttft_ms": 35.68995790556073, "decode_ms": 213.54141703341156}, "total_wall_ms": 45526.88237500843, "ttft_ms": 35.68995790556073, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 7}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "2414100", "evidence_artifact": {"ledger": {"facts_total": 64, "facts_retained": 22, "retention_rate": 0.34375, "retention_by_kind": {"person": 0.5, "date": 0.5555555555555556, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1176}, "ledger_by_hop": {"1": {"facts_total": 64, "facts_retained": 22, "retention_rate": 0.34375, "retention_by_kind": {"person": 0.5, "date": 0.5555555555555556, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1181}, "2": {"facts_total": 64, "facts_retained": 22, "retention_rate": 0.34375, "retention_by_kind": {"person": 0.5, "date": 0.5555555555555556, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1176}}}, "reference_key_id": "keys/synthetic/syn_03018_2000.json", "task_scores": {"correct": 0.0, "abstained": 0.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "89438188d34426dbc3fdf77c29f9462c95075b81b820590bae58487694cd6183", "output_hash": "f8dcab5deb897eef941efd68c26d12f2731e44a034dd2933d74c1cd05e164f9b"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "15f182aaa2ad2427d2758915", "attempt_id": "41b130b86570", "run_group_id": "E4_core:A4|closed|qwen3b:h2", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:06:01.427859Z", "task_family": "synthetic", "document_id": "syn_03018_2000", "document_hash": "d284eb07e27da742b9c2b67abccfdf72e4dd67c3259f44f3b41e34dd3e6fc1ff", "question_id": "q12_unit_conversion", "split": "test", "cluster_id": "syn_03018_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 2, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1982, "query_tokens": 602, "message_tokens": 511, "message_utf8_bytes": 1176, "cache_tensor_bytes": null, "serialized_payload_bytes": 1176, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 44859.323624987155, "hops": {"1": 22836.384875001386, "2": 22022.93874998577}, "receiver_prefill_ms": 339.4380420213565, "ttft_ms": 51.79804190993309, "decode_ms": 46.453791903331876}, "total_wall_ms": 45297.80416702852, "ttft_ms": 51.79804190993309, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 64, "facts_retained": 22, "retention_rate": 0.34375, "retention_by_kind": {"person": 0.5, "date": 0.5555555555555556, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1176}, "ledger_by_hop": {"1": {"facts_total": 64, "facts_retained": 22, "retention_rate": 0.34375, "retention_by_kind": {"person": 0.5, "date": 0.5555555555555556, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1181}, "2": {"facts_total": 64, "facts_retained": 22, "retention_rate": 0.34375, "retention_by_kind": {"person": 0.5, "date": 0.5555555555555556, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1176}}}, "reference_key_id": "keys/synthetic/syn_03018_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "89438188d34426dbc3fdf77c29f9462c95075b81b820590bae58487694cd6183", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "97255b47071d8667833cee25", "attempt_id": "6a66282d87af", "run_group_id": "E4_core:A4|closed|qwen3b:h2", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:06:01.883397Z", "task_family": "synthetic", "document_id": "syn_03018_2000", "document_hash": "d284eb07e27da742b9c2b67abccfdf72e4dd67c3259f44f3b41e34dd3e6fc1ff", "question_id": "q13_unanswerable", "split": "test", "cluster_id": "syn_03018_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 2, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1982, "query_tokens": 606, "message_tokens": 511, "message_utf8_bytes": 1176, "cache_tensor_bytes": null, "serialized_payload_bytes": 1176, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 44859.323624987155, "hops": {"1": 22836.384875001386, "2": 22022.93874998577}, "receiver_prefill_ms": 553.697666968219, "ttft_ms": 25.619082967750728, "decode_ms": 42.09066601470113}, "total_wall_ms": 45481.65995802265, "ttft_ms": 25.619082967750728, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 64, "facts_retained": 22, "retention_rate": 0.34375, "retention_by_kind": {"person": 0.5, "date": 0.5555555555555556, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1176}, "ledger_by_hop": {"1": {"facts_total": 64, "facts_retained": 22, "retention_rate": 0.34375, "retention_by_kind": {"person": 0.5, "date": 0.5555555555555556, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1181}, "2": {"facts_total": 64, "facts_retained": 22, "retention_rate": 0.34375, "retention_by_kind": {"person": 0.5, "date": 0.5555555555555556, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1176}}}, "reference_key_id": "keys/synthetic/syn_03018_2000.json", "task_scores": {"correct": 1.0, "abstained": 1.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "89438188d34426dbc3fdf77c29f9462c95075b81b820590bae58487694cd6183", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "20e0946c0124443989c0057b", "attempt_id": "d4fc63145f06", "run_group_id": "E4_core:A4|closed|qwen3b:h2", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:06:02.523085Z", "task_family": "synthetic", "document_id": "syn_03018_2000", "document_hash": "d284eb07e27da742b9c2b67abccfdf72e4dd67c3259f44f3b41e34dd3e6fc1ff", "question_id": "q14_unanswerable", "split": "test", "cluster_id": "syn_03018_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 2, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1982, "query_tokens": 607, "message_tokens": 511, "message_utf8_bytes": 1176, "cache_tensor_bytes": null, "serialized_payload_bytes": 1176, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 44859.323624987155, "hops": {"1": 22836.384875001386, "2": 22022.93874998577}, "receiver_prefill_ms": 405.699499999173, "ttft_ms": 23.089208989404142, "decode_ms": 40.04424996674061}, "total_wall_ms": 45328.976792050526, "ttft_ms": 23.089208989404142, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 64, "facts_retained": 22, "retention_rate": 0.34375, "retention_by_kind": {"person": 0.5, "date": 0.5555555555555556, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1176}, "ledger_by_hop": {"1": {"facts_total": 64, "facts_retained": 22, "retention_rate": 0.34375, "retention_by_kind": {"person": 0.5, "date": 0.5555555555555556, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1181}, "2": {"facts_total": 64, "facts_retained": 22, "retention_rate": 0.34375, "retention_by_kind": {"person": 0.5, "date": 0.5555555555555556, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1176}}}, "reference_key_id": "keys/synthetic/syn_03018_2000.json", "task_scores": {"correct": 1.0, "abstained": 1.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "89438188d34426dbc3fdf77c29f9462c95075b81b820590bae58487694cd6183", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "a2b54bd4b30f69cfd683876c", "attempt_id": "7c882ef9d372", "run_group_id": "E4_core:A4|closed|qwen3b:h4", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:06:46.737996Z", "task_family": "synthetic", "document_id": "syn_03018_2000", "document_hash": "d284eb07e27da742b9c2b67abccfdf72e4dd67c3259f44f3b41e34dd3e6fc1ff", "question_id": "q00_single_fact", "split": "test", "cluster_id": "syn_03018_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 4, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1982, "query_tokens": 594, "message_tokens": 511, "message_utf8_bytes": 1176, "cache_tensor_bytes": null, "serialized_payload_bytes": 1176, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 88584.54354084097, "hops": {"1": 22836.384875001386, "2": 22022.93874998577, "3": 21884.76395793259, "4": 21840.455957921222}, "receiver_prefill_ms": 346.56025003641844, "ttft_ms": 51.19395803194493, "decode_ms": 57.95779102481902}, "total_wall_ms": 89040.93029082287, "ttft_ms": 51.19395803194493, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 64, "facts_retained": 22, "retention_rate": 0.34375, "retention_by_kind": {"person": 0.5, "date": 0.5555555555555556, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1176}, "ledger_by_hop": {"1": {"facts_total": 64, "facts_retained": 22, "retention_rate": 0.34375, "retention_by_kind": {"person": 0.5, "date": 0.5555555555555556, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1181}, "2": {"facts_total": 64, "facts_retained": 22, "retention_rate": 0.34375, "retention_by_kind": {"person": 0.5, "date": 0.5555555555555556, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1176}, "3": {"facts_total": 64, "facts_retained": 22, "retention_rate": 0.34375, "retention_by_kind": {"person": 0.5, "date": 0.5555555555555556, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1176}, "4": {"facts_total": 64, "facts_retained": 22, "retention_rate": 0.34375, "retention_by_kind": {"person": 0.5, "date": 0.5555555555555556, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1176}}}, "reference_key_id": "keys/synthetic/syn_03018_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "89438188d34426dbc3fdf77c29f9462c95075b81b820590bae58487694cd6183", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "65cc760250618786b712be1f", "attempt_id": "18c95b0e2273", "run_group_id": "E4_core:A4|closed|qwen3b:h4", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:06:47.212848Z", "task_family": "synthetic", "document_id": "syn_03018_2000", "document_hash": "d284eb07e27da742b9c2b67abccfdf72e4dd67c3259f44f3b41e34dd3e6fc1ff", "question_id": "q01_single_fact", "split": "test", "cluster_id": "syn_03018_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 4, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1982, "query_tokens": 600, "message_tokens": 511, "message_utf8_bytes": 1176, "cache_tensor_bytes": null, "serialized_payload_bytes": 1176, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 88584.54354084097, "hops": {"1": 22836.384875001386, "2": 22022.93874998577, "3": 21884.76395793259, "4": 21840.455957921222}, "receiver_prefill_ms": 515.5138750560582, "ttft_ms": 27.485208003781736, "decode_ms": 402.9367909533903}, "total_wall_ms": 89531.33854083717, "ttft_ms": 27.485208003781736, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 12}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "Lot 376-E, Baie-Comeau", "evidence_artifact": {"ledger": {"facts_total": 64, "facts_retained": 22, "retention_rate": 0.34375, "retention_by_kind": {"person": 0.5, "date": 0.5555555555555556, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1176}, "ledger_by_hop": {"1": {"facts_total": 64, "facts_retained": 22, "retention_rate": 0.34375, "retention_by_kind": {"person": 0.5, "date": 0.5555555555555556, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1181}, "2": {"facts_total": 64, "facts_retained": 22, "retention_rate": 0.34375, "retention_by_kind": {"person": 0.5, "date": 0.5555555555555556, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1176}, "3": {"facts_total": 64, "facts_retained": 22, "retention_rate": 0.34375, "retention_by_kind": {"person": 0.5, "date": 0.5555555555555556, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1176}, "4": {"facts_total": 64, "facts_retained": 22, "retention_rate": 0.34375, "retention_by_kind": {"person": 0.5, "date": 0.5555555555555556, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1176}}}, "reference_key_id": "keys/synthetic/syn_03018_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 0.8571428571428571}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "89438188d34426dbc3fdf77c29f9462c95075b81b820590bae58487694cd6183", "output_hash": "b58829a1440993c4af3307ba44494f021d726fd11e9337888b6b852a8b37351a"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "fc1baaad90d63e8701d30919", "attempt_id": "86bc0fa8550c", "run_group_id": "E4_core:A4|closed|qwen3b:h4", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:06:48.178329Z", "task_family": "synthetic", "document_id": "syn_03018_2000", "document_hash": "d284eb07e27da742b9c2b67abccfdf72e4dd67c3259f44f3b41e34dd3e6fc1ff", "question_id": "q02_exact_copy", "split": "test", "cluster_id": "syn_03018_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 4, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1982, "query_tokens": 599, "message_tokens": 511, "message_utf8_bytes": 1176, "cache_tensor_bytes": null, "serialized_payload_bytes": 1176, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 88584.54354084097, "hops": {"1": 22836.384875001386, "2": 22022.93874998577, "3": 21884.76395793259, "4": 21840.455957921222}, "receiver_prefill_ms": 371.2024170672521, "ttft_ms": 43.210458010435104, "decode_ms": 62.39266705233604}, "total_wall_ms": 89062.00995773543, "ttft_ms": 43.210458010435104, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 64, "facts_retained": 22, "retention_rate": 0.34375, "retention_by_kind": {"person": 0.5, "date": 0.5555555555555556, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1176}, "ledger_by_hop": {"1": {"facts_total": 64, "facts_retained": 22, "retention_rate": 0.34375, "retention_by_kind": {"person": 0.5, "date": 0.5555555555555556, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1181}, "2": {"facts_total": 64, "facts_retained": 22, "retention_rate": 0.34375, "retention_by_kind": {"person": 0.5, "date": 0.5555555555555556, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1176}, "3": {"facts_total": 64, "facts_retained": 22, "retention_rate": 0.34375, "retention_by_kind": {"person": 0.5, "date": 0.5555555555555556, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1176}, "4": {"facts_total": 64, "facts_retained": 22, "retention_rate": 0.34375, "retention_by_kind": {"person": 0.5, "date": 0.5555555555555556, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1176}}}, "reference_key_id": "keys/synthetic/syn_03018_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "89438188d34426dbc3fdf77c29f9462c95075b81b820590bae58487694cd6183", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "15f9bd4b044ad5e0dd1478a6", "attempt_id": "bdc533424e12", "run_group_id": "E4_core:A4|closed|qwen3b:h4", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:06:48.672777Z", "task_family": "synthetic", "document_id": "syn_03018_2000", "document_hash": "d284eb07e27da742b9c2b67abccfdf72e4dd67c3259f44f3b41e34dd3e6fc1ff", "question_id": "q03_exact_copy", "split": "test", "cluster_id": "syn_03018_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 4, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1982, "query_tokens": 597, "message_tokens": 511, "message_utf8_bytes": 1176, "cache_tensor_bytes": null, "serialized_payload_bytes": 1176, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 88584.54354084097, "hops": {"1": 22836.384875001386, "2": 22022.93874998577, "3": 21884.76395793259, "4": 21840.455957921222}, "receiver_prefill_ms": 562.540166079998, "ttft_ms": 37.88091603200883, "decode_ms": 299.55750005319715}, "total_wall_ms": 89485.50079076085, "ttft_ms": 37.88091603200883, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 9}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "Priyamvada Sørlie", "evidence_artifact": {"ledger": {"facts_total": 64, "facts_retained": 22, "retention_rate": 0.34375, "retention_by_kind": {"person": 0.5, "date": 0.5555555555555556, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1176}, "ledger_by_hop": {"1": {"facts_total": 64, "facts_retained": 22, "retention_rate": 0.34375, "retention_by_kind": {"person": 0.5, "date": 0.5555555555555556, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1181}, "2": {"facts_total": 64, "facts_retained": 22, "retention_rate": 0.34375, "retention_by_kind": {"person": 0.5, "date": 0.5555555555555556, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1176}, "3": {"facts_total": 64, "facts_retained": 22, "retention_rate": 0.34375, "retention_by_kind": {"person": 0.5, "date": 0.5555555555555556, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1176}, "4": {"facts_total": 64, "facts_retained": 22, "retention_rate": 0.34375, "retention_by_kind": {"person": 0.5, "date": 0.5555555555555556, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1176}}}, "reference_key_id": "keys/synthetic/syn_03018_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "89438188d34426dbc3fdf77c29f9462c95075b81b820590bae58487694cd6183", "output_hash": "c5f554196ed23dd71ad995ec89f7142099d524dbb6534cc10d94a0c89d13e792"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "9c26f1da7217f8486fda6ca1", "attempt_id": "adc1d5be821f", "run_group_id": "E4_core:A4|closed|qwen3b:h4", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:06:49.591151Z", "task_family": "synthetic", "document_id": "syn_03018_2000", "document_hash": "d284eb07e27da742b9c2b67abccfdf72e4dd67c3259f44f3b41e34dd3e6fc1ff", "question_id": "q04_multi_hop", "split": "test", "cluster_id": "syn_03018_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 4, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1982, "query_tokens": 603, "message_tokens": 511, "message_utf8_bytes": 1176, "cache_tensor_bytes": null, "serialized_payload_bytes": 1176, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 88584.54354084097, "hops": {"1": 22836.384875001386, "2": 22022.93874998577, "3": 21884.76395793259, "4": 21840.455957921222}, "receiver_prefill_ms": 404.5050840359181, "ttft_ms": 22.399667068384588, "decode_ms": 38.56958297546953}, "total_wall_ms": 89050.72758288588, "ttft_ms": 22.399667068384588, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 64, "facts_retained": 22, "retention_rate": 0.34375, "retention_by_kind": {"person": 0.5, "date": 0.5555555555555556, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1176}, "ledger_by_hop": {"1": {"facts_total": 64, "facts_retained": 22, "retention_rate": 0.34375, "retention_by_kind": {"person": 0.5, "date": 0.5555555555555556, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1181}, "2": {"facts_total": 64, "facts_retained": 22, "retention_rate": 0.34375, "retention_by_kind": {"person": 0.5, "date": 0.5555555555555556, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1176}, "3": {"facts_total": 64, "facts_retained": 22, "retention_rate": 0.34375, "retention_by_kind": {"person": 0.5, "date": 0.5555555555555556, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1176}, "4": {"facts_total": 64, "facts_retained": 22, "retention_rate": 0.34375, "retention_by_kind": {"person": 0.5, "date": 0.5555555555555556, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1176}}}, "reference_key_id": "keys/synthetic/syn_03018_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "89438188d34426dbc3fdf77c29f9462c95075b81b820590bae58487694cd6183", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "1dc9cca805fc57a8b77caaa1", "attempt_id": "5541e4bbb68d", "run_group_id": "E4_core:A4|closed|qwen3b:h4", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:06:50.074072Z", "task_family": "synthetic", "document_id": "syn_03018_2000", "document_hash": "d284eb07e27da742b9c2b67abccfdf72e4dd67c3259f44f3b41e34dd3e6fc1ff", "question_id": "q05_multi_hop", "split": "test", "cluster_id": "syn_03018_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 4, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1982, "query_tokens": 603, "message_tokens": 511, "message_utf8_bytes": 1176, "cache_tensor_bytes": null, "serialized_payload_bytes": 1176, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 88584.54354084097, "hops": {"1": 22836.384875001386, "2": 22022.93874998577, "3": 21884.76395793259, "4": 21840.455957921222}, "receiver_prefill_ms": 371.814375044778, "ttft_ms": 52.43583407718688, "decode_ms": 543.9182500122115}, "total_wall_ms": 89553.55924984906, "ttft_ms": 52.43583407718688, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 10}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "HM-38K-5533", "evidence_artifact": {"ledger": {"facts_total": 64, "facts_retained": 22, "retention_rate": 0.34375, "retention_by_kind": {"person": 0.5, "date": 0.5555555555555556, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1176}, "ledger_by_hop": {"1": {"facts_total": 64, "facts_retained": 22, "retention_rate": 0.34375, "retention_by_kind": {"person": 0.5, "date": 0.5555555555555556, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1181}, "2": {"facts_total": 64, "facts_retained": 22, "retention_rate": 0.34375, "retention_by_kind": {"person": 0.5, "date": 0.5555555555555556, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1176}, "3": {"facts_total": 64, "facts_retained": 22, "retention_rate": 0.34375, "retention_by_kind": {"person": 0.5, "date": 0.5555555555555556, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1176}, "4": {"facts_total": 64, "facts_retained": 22, "retention_rate": 0.34375, "retention_by_kind": {"person": 0.5, "date": 0.5555555555555556, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1176}}}, "reference_key_id": "keys/synthetic/syn_03018_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "89438188d34426dbc3fdf77c29f9462c95075b81b820590bae58487694cd6183", "output_hash": "18acd9807857243bbded3bceae55398316a3958b81d70364441bdb6748eaf878"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "8b2b1525ee3fe3dd6ce651ca", "attempt_id": "45402e0e0b55", "run_group_id": "E4_core:A4|closed|qwen3b:h4", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:06:51.061076Z", "task_family": "synthetic", "document_id": "syn_03018_2000", "document_hash": "d284eb07e27da742b9c2b67abccfdf72e4dd67c3259f44f3b41e34dd3e6fc1ff", "question_id": "q06_negation", "split": "test", "cluster_id": "syn_03018_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 4, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1982, "query_tokens": 595, "message_tokens": 511, "message_utf8_bytes": 1176, "cache_tensor_bytes": null, "serialized_payload_bytes": 1176, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 88584.54354084097, "hops": {"1": 22836.384875001386, "2": 22022.93874998577, "3": 21884.76395793259, "4": 21840.455957921222}, "receiver_prefill_ms": 419.89499994087964, "ttft_ms": 36.06329194735736, "decode_ms": 29.06779106706381}, "total_wall_ms": 89070.22562378552, "ttft_ms": 36.06329194735736, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 64, "facts_retained": 22, "retention_rate": 0.34375, "retention_by_kind": {"person": 0.5, "date": 0.5555555555555556, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1176}, "ledger_by_hop": {"1": {"facts_total": 64, "facts_retained": 22, "retention_rate": 0.34375, "retention_by_kind": {"person": 0.5, "date": 0.5555555555555556, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1181}, "2": {"facts_total": 64, "facts_retained": 22, "retention_rate": 0.34375, "retention_by_kind": {"person": 0.5, "date": 0.5555555555555556, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1176}, "3": {"facts_total": 64, "facts_retained": 22, "retention_rate": 0.34375, "retention_by_kind": {"person": 0.5, "date": 0.5555555555555556, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1176}, "4": {"facts_total": 64, "facts_retained": 22, "retention_rate": 0.34375, "retention_by_kind": {"person": 0.5, "date": 0.5555555555555556, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1176}}}, "reference_key_id": "keys/synthetic/syn_03018_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "89438188d34426dbc3fdf77c29f9462c95075b81b820590bae58487694cd6183", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "ba77668e5d841a475cd84b2f", "attempt_id": "4db4b2db40ac", "run_group_id": "E4_core:A4|closed|qwen3b:h4", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:06:51.564698Z", "task_family": "synthetic", "document_id": "syn_03018_2000", "document_hash": "d284eb07e27da742b9c2b67abccfdf72e4dd67c3259f44f3b41e34dd3e6fc1ff", "question_id": "q07_negation", "split": "test", "cluster_id": "syn_03018_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 4, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1982, "query_tokens": 594, "message_tokens": 511, "message_utf8_bytes": 1176, "cache_tensor_bytes": null, "serialized_payload_bytes": 1176, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 88584.54354084097, "hops": {"1": 22836.384875001386, "2": 22022.93874998577, "3": 21884.76395793259, "4": 21840.455957921222}, "receiver_prefill_ms": 356.22029192745686, "ttft_ms": 46.30854201968759, "decode_ms": 57.97870899550617}, "total_wall_ms": 89045.78874877188, "ttft_ms": 46.30854201968759, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "no", "evidence_artifact": {"ledger": {"facts_total": 64, "facts_retained": 22, "retention_rate": 0.34375, "retention_by_kind": {"person": 0.5, "date": 0.5555555555555556, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1176}, "ledger_by_hop": {"1": {"facts_total": 64, "facts_retained": 22, "retention_rate": 0.34375, "retention_by_kind": {"person": 0.5, "date": 0.5555555555555556, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1181}, "2": {"facts_total": 64, "facts_retained": 22, "retention_rate": 0.34375, "retention_by_kind": {"person": 0.5, "date": 0.5555555555555556, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1176}, "3": {"facts_total": 64, "facts_retained": 22, "retention_rate": 0.34375, "retention_by_kind": {"person": 0.5, "date": 0.5555555555555556, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1176}, "4": {"facts_total": 64, "facts_retained": 22, "retention_rate": 0.34375, "retention_by_kind": {"person": 0.5, "date": 0.5555555555555556, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1176}}}, "reference_key_id": "keys/synthetic/syn_03018_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "89438188d34426dbc3fdf77c29f9462c95075b81b820590bae58487694cd6183", "output_hash": "9390298f3fb0c5b160498935d79cb139aef28e1c47358b4bbba61862b9c26e59"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "bc882a5b3232fc3af7348794", "attempt_id": "d4c76b5e943e", "run_group_id": "E4_core:A4|closed|qwen3b:h4", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:06:52.043140Z", "task_family": "synthetic", "document_id": "syn_03018_2000", "document_hash": "d284eb07e27da742b9c2b67abccfdf72e4dd67c3259f44f3b41e34dd3e6fc1ff", "question_id": "q08_revision", "split": "test", "cluster_id": "syn_03018_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 4, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1982, "query_tokens": 612, "message_tokens": 511, "message_utf8_bytes": 1176, "cache_tensor_bytes": null, "serialized_payload_bytes": 1176, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 88584.54354084097, "hops": {"1": 22836.384875001386, "2": 22022.93874998577, "3": 21884.76395793259, "4": 21840.455957921222}, "receiver_prefill_ms": 555.3155828965828, "ttft_ms": 39.09787500742823, "decode_ms": 240.54625001735985}, "total_wall_ms": 89420.2646658523, "ttft_ms": 39.09787500742823, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 6}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "304815", "evidence_artifact": {"ledger": {"facts_total": 64, "facts_retained": 22, "retention_rate": 0.34375, "retention_by_kind": {"person": 0.5, "date": 0.5555555555555556, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1176}, "ledger_by_hop": {"1": {"facts_total": 64, "facts_retained": 22, "retention_rate": 0.34375, "retention_by_kind": {"person": 0.5, "date": 0.5555555555555556, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1181}, "2": {"facts_total": 64, "facts_retained": 22, "retention_rate": 0.34375, "retention_by_kind": {"person": 0.5, "date": 0.5555555555555556, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1176}, "3": {"facts_total": 64, "facts_retained": 22, "retention_rate": 0.34375, "retention_by_kind": {"person": 0.5, "date": 0.5555555555555556, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1176}, "4": {"facts_total": 64, "facts_retained": 22, "retention_rate": 0.34375, "retention_by_kind": {"person": 0.5, "date": 0.5555555555555556, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1176}}}, "reference_key_id": "keys/synthetic/syn_03018_2000.json", "task_scores": {"correct": 0.0, "abstained": 0.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "89438188d34426dbc3fdf77c29f9462c95075b81b820590bae58487694cd6183", "output_hash": "a480475d6855141c920f42d1bf3337ca7fd88bcf4087b0ed2e253ed68cca6926"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "491254a6834c11a4ce5e7a71", "attempt_id": "41a12bbdd756", "run_group_id": "E4_core:A4|closed|qwen3b:h4", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:06:52.897681Z", "task_family": "synthetic", "document_id": "syn_03018_2000", "document_hash": "d284eb07e27da742b9c2b67abccfdf72e4dd67c3259f44f3b41e34dd3e6fc1ff", "question_id": "q09_revision", "split": "test", "cluster_id": "syn_03018_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 4, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1982, "query_tokens": 605, "message_tokens": 511, "message_utf8_bytes": 1176, "cache_tensor_bytes": null, "serialized_payload_bytes": 1176, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 88584.54354084097, "hops": {"1": 22836.384875001386, "2": 22022.93874998577, "3": 21884.76395793259, "4": 21840.455957921222}, "receiver_prefill_ms": 400.26200003921986, "ttft_ms": 31.517833936959505, "decode_ms": 249.31500002276152}, "total_wall_ms": 89266.29466586746, "ttft_ms": 31.517833936959505, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 6}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "276340", "evidence_artifact": {"ledger": {"facts_total": 64, "facts_retained": 22, "retention_rate": 0.34375, "retention_by_kind": {"person": 0.5, "date": 0.5555555555555556, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1176}, "ledger_by_hop": {"1": {"facts_total": 64, "facts_retained": 22, "retention_rate": 0.34375, "retention_by_kind": {"person": 0.5, "date": 0.5555555555555556, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1181}, "2": {"facts_total": 64, "facts_retained": 22, "retention_rate": 0.34375, "retention_by_kind": {"person": 0.5, "date": 0.5555555555555556, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1176}, "3": {"facts_total": 64, "facts_retained": 22, "retention_rate": 0.34375, "retention_by_kind": {"person": 0.5, "date": 0.5555555555555556, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1176}, "4": {"facts_total": 64, "facts_retained": 22, "retention_rate": 0.34375, "retention_by_kind": {"person": 0.5, "date": 0.5555555555555556, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1176}}}, "reference_key_id": "keys/synthetic/syn_03018_2000.json", "task_scores": {"correct": 0.0, "abstained": 0.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "89438188d34426dbc3fdf77c29f9462c95075b81b820590bae58487694cd6183", "output_hash": "bc7c3667cbe0ad74ec0a1b092d3bad7e6a02978fa67746e3adbf3174ed34b3a2"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "6bdc5c33bebfc547353f0544", "attempt_id": "aff616350e2b", "run_group_id": "E4_core:A4|closed|qwen3b:h4", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:06:53.597133Z", "task_family": "synthetic", "document_id": "syn_03018_2000", "document_hash": "d284eb07e27da742b9c2b67abccfdf72e4dd67c3259f44f3b41e34dd3e6fc1ff", "question_id": "q10_temporal", "split": "test", "cluster_id": "syn_03018_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 4, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1982, "query_tokens": 613, "message_tokens": 511, "message_utf8_bytes": 1176, "cache_tensor_bytes": null, "serialized_payload_bytes": 1176, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 88584.54354084097, "hops": {"1": 22836.384875001386, "2": 22022.93874998577, "3": 21884.76395793259, "4": 21840.455957921222}, "receiver_prefill_ms": 498.8684169948101, "ttft_ms": 62.18537501990795, "decode_ms": 204.34466598089784}, "total_wall_ms": 89350.72187485639, "ttft_ms": 62.18537501990795, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 6}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "Melymi-15", "evidence_artifact": {"ledger": {"facts_total": 64, "facts_retained": 22, "retention_rate": 0.34375, "retention_by_kind": {"person": 0.5, "date": 0.5555555555555556, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1176}, "ledger_by_hop": {"1": {"facts_total": 64, "facts_retained": 22, "retention_rate": 0.34375, "retention_by_kind": {"person": 0.5, "date": 0.5555555555555556, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1181}, "2": {"facts_total": 64, "facts_retained": 22, "retention_rate": 0.34375, "retention_by_kind": {"person": 0.5, "date": 0.5555555555555556, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1176}, "3": {"facts_total": 64, "facts_retained": 22, "retention_rate": 0.34375, "retention_by_kind": {"person": 0.5, "date": 0.5555555555555556, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1176}, "4": {"facts_total": 64, "facts_retained": 22, "retention_rate": 0.34375, "retention_by_kind": {"person": 0.5, "date": 0.5555555555555556, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1176}}}, "reference_key_id": "keys/synthetic/syn_03018_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "89438188d34426dbc3fdf77c29f9462c95075b81b820590bae58487694cd6183", "output_hash": "fa08d1474b01d6dfbf3dea049be251aa3c3021c17b4818258064b156509af58b"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "6c8cdb7506ee6ca840d78660", "attempt_id": "0a804fe2da7d", "run_group_id": "E4_core:A4|closed|qwen3b:h4", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:06:54.382141Z", "task_family": "synthetic", "document_id": "syn_03018_2000", "document_hash": "d284eb07e27da742b9c2b67abccfdf72e4dd67c3259f44f3b41e34dd3e6fc1ff", "question_id": "q11_aggregation", "split": "test", "cluster_id": "syn_03018_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 4, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1982, "query_tokens": 605, "message_tokens": 511, "message_utf8_bytes": 1176, "cache_tensor_bytes": null, "serialized_payload_bytes": 1176, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 88584.54354084097, "hops": {"1": 22836.384875001386, "2": 22022.93874998577, "3": 21884.76395793259, "4": 21840.455957921222}, "receiver_prefill_ms": 419.5568748982623, "ttft_ms": 46.802875003777444, "decode_ms": 189.68483293429017}, "total_wall_ms": 89241.42924987245, "ttft_ms": 46.802875003777444, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 7}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "2410460", "evidence_artifact": {"ledger": {"facts_total": 64, "facts_retained": 22, "retention_rate": 0.34375, "retention_by_kind": {"person": 0.5, "date": 0.5555555555555556, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1176}, "ledger_by_hop": {"1": {"facts_total": 64, "facts_retained": 22, "retention_rate": 0.34375, "retention_by_kind": {"person": 0.5, "date": 0.5555555555555556, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1181}, "2": {"facts_total": 64, "facts_retained": 22, "retention_rate": 0.34375, "retention_by_kind": {"person": 0.5, "date": 0.5555555555555556, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1176}, "3": {"facts_total": 64, "facts_retained": 22, "retention_rate": 0.34375, "retention_by_kind": {"person": 0.5, "date": 0.5555555555555556, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1176}, "4": {"facts_total": 64, "facts_retained": 22, "retention_rate": 0.34375, "retention_by_kind": {"person": 0.5, "date": 0.5555555555555556, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1176}}}, "reference_key_id": "keys/synthetic/syn_03018_2000.json", "task_scores": {"correct": 0.0, "abstained": 0.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "89438188d34426dbc3fdf77c29f9462c95075b81b820590bae58487694cd6183", "output_hash": "63fe326637abd22b1b1e457840e07ea81e971b7c680f6bccdf09ebe8bd7c8b80"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "11f491e504ba614fd4ba0b62", "attempt_id": "1bf93aface86", "run_group_id": "E4_core:A4|closed|qwen3b:h4", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:06:55.057332Z", "task_family": "synthetic", "document_id": "syn_03018_2000", "document_hash": "d284eb07e27da742b9c2b67abccfdf72e4dd67c3259f44f3b41e34dd3e6fc1ff", "question_id": "q12_unit_conversion", "split": "test", "cluster_id": "syn_03018_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 4, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1982, "query_tokens": 602, "message_tokens": 511, "message_utf8_bytes": 1176, "cache_tensor_bytes": null, "serialized_payload_bytes": 1176, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 88584.54354084097, "hops": {"1": 22836.384875001386, "2": 22022.93874998577, "3": 21884.76395793259, "4": 21840.455957921222}, "receiver_prefill_ms": 417.7968750009313, "ttft_ms": 44.8345000622794, "decode_ms": 89.99895805027336}, "total_wall_ms": 89137.98466592561, "ttft_ms": 44.8345000622794, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 64, "facts_retained": 22, "retention_rate": 0.34375, "retention_by_kind": {"person": 0.5, "date": 0.5555555555555556, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1176}, "ledger_by_hop": {"1": {"facts_total": 64, "facts_retained": 22, "retention_rate": 0.34375, "retention_by_kind": {"person": 0.5, "date": 0.5555555555555556, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1181}, "2": {"facts_total": 64, "facts_retained": 22, "retention_rate": 0.34375, "retention_by_kind": {"person": 0.5, "date": 0.5555555555555556, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1176}, "3": {"facts_total": 64, "facts_retained": 22, "retention_rate": 0.34375, "retention_by_kind": {"person": 0.5, "date": 0.5555555555555556, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1176}, "4": {"facts_total": 64, "facts_retained": 22, "retention_rate": 0.34375, "retention_by_kind": {"person": 0.5, "date": 0.5555555555555556, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1176}}}, "reference_key_id": "keys/synthetic/syn_03018_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "89438188d34426dbc3fdf77c29f9462c95075b81b820590bae58487694cd6183", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "eb5ae4613374172b4f7c7410", "attempt_id": "bb25effaa986", "run_group_id": "E4_core:A4|closed|qwen3b:h4", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:06:55.629017Z", "task_family": "synthetic", "document_id": "syn_03018_2000", "document_hash": "d284eb07e27da742b9c2b67abccfdf72e4dd67c3259f44f3b41e34dd3e6fc1ff", "question_id": "q13_unanswerable", "split": "test", "cluster_id": "syn_03018_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 4, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1982, "query_tokens": 606, "message_tokens": 511, "message_utf8_bytes": 1176, "cache_tensor_bytes": null, "serialized_payload_bytes": 1176, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 88584.54354084097, "hops": {"1": 22836.384875001386, "2": 22022.93874998577, "3": 21884.76395793259, "4": 21840.455957921222}, "receiver_prefill_ms": 479.37266691587865, "ttft_ms": 39.454958983697, "decode_ms": 24.857333046384156}, "total_wall_ms": 89128.8884987589, "ttft_ms": 39.454958983697, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 64, "facts_retained": 22, "retention_rate": 0.34375, "retention_by_kind": {"person": 0.5, "date": 0.5555555555555556, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1176}, "ledger_by_hop": {"1": {"facts_total": 64, "facts_retained": 22, "retention_rate": 0.34375, "retention_by_kind": {"person": 0.5, "date": 0.5555555555555556, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1181}, "2": {"facts_total": 64, "facts_retained": 22, "retention_rate": 0.34375, "retention_by_kind": {"person": 0.5, "date": 0.5555555555555556, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1176}, "3": {"facts_total": 64, "facts_retained": 22, "retention_rate": 0.34375, "retention_by_kind": {"person": 0.5, "date": 0.5555555555555556, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1176}, "4": {"facts_total": 64, "facts_retained": 22, "retention_rate": 0.34375, "retention_by_kind": {"person": 0.5, "date": 0.5555555555555556, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1176}}}, "reference_key_id": "keys/synthetic/syn_03018_2000.json", "task_scores": {"correct": 1.0, "abstained": 1.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "89438188d34426dbc3fdf77c29f9462c95075b81b820590bae58487694cd6183", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "ce47da5abc875cbb3977dc7e", "attempt_id": "c81d9fdcc656", "run_group_id": "E4_core:A4|closed|qwen3b:h4", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:06:56.193052Z", "task_family": "synthetic", "document_id": "syn_03018_2000", "document_hash": "d284eb07e27da742b9c2b67abccfdf72e4dd67c3259f44f3b41e34dd3e6fc1ff", "question_id": "q14_unanswerable", "split": "test", "cluster_id": "syn_03018_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 4, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1982, "query_tokens": 607, "message_tokens": 511, "message_utf8_bytes": 1176, "cache_tensor_bytes": null, "serialized_payload_bytes": 1176, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 88584.54354084097, "hops": {"1": 22836.384875001386, "2": 22022.93874998577, "3": 21884.76395793259, "4": 21840.455957921222}, "receiver_prefill_ms": 393.2730419328436, "ttft_ms": 22.6130411028862, "decode_ms": 36.992958979681134}, "total_wall_ms": 89038.02904079203, "ttft_ms": 22.6130411028862, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 64, "facts_retained": 22, "retention_rate": 0.34375, "retention_by_kind": {"person": 0.5, "date": 0.5555555555555556, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1176}, "ledger_by_hop": {"1": {"facts_total": 64, "facts_retained": 22, "retention_rate": 0.34375, "retention_by_kind": {"person": 0.5, "date": 0.5555555555555556, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1181}, "2": {"facts_total": 64, "facts_retained": 22, "retention_rate": 0.34375, "retention_by_kind": {"person": 0.5, "date": 0.5555555555555556, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1176}, "3": {"facts_total": 64, "facts_retained": 22, "retention_rate": 0.34375, "retention_by_kind": {"person": 0.5, "date": 0.5555555555555556, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1176}, "4": {"facts_total": 64, "facts_retained": 22, "retention_rate": 0.34375, "retention_by_kind": {"person": 0.5, "date": 0.5555555555555556, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1176}}}, "reference_key_id": "keys/synthetic/syn_03018_2000.json", "task_scores": {"correct": 1.0, "abstained": 1.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "89438188d34426dbc3fdf77c29f9462c95075b81b820590bae58487694cd6183", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "e430db29c8a250544fe74512", "attempt_id": "344c125fb712", "run_group_id": "E4_core:A4|closed|qwen3b:h8", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:08:24.728821Z", "task_family": "synthetic", "document_id": "syn_03018_2000", "document_hash": "d284eb07e27da742b9c2b67abccfdf72e4dd67c3259f44f3b41e34dd3e6fc1ff", "question_id": "q00_single_fact", "split": "test", "cluster_id": "syn_03018_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 8, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1982, "query_tokens": 594, "message_tokens": 511, "message_utf8_bytes": 1176, "cache_tensor_bytes": null, "serialized_payload_bytes": 1176, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 176645.02341777552, "hops": {"1": 22836.384875001386, "2": 22022.93874998577, "3": 21884.76395793259, "4": 21840.455957921222, "5": 22013.46108398866, "6": 21984.005416976288, "7": 22096.49654198438, "8": 21966.516833985224}, "receiver_prefill_ms": 379.46245807688683, "ttft_ms": 25.453499983996153, "decode_ms": 36.27645899541676}, "total_wall_ms": 177086.83391776867, "ttft_ms": 25.453499983996153, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 64, "facts_retained": 22, "retention_rate": 0.34375, "retention_by_kind": {"person": 0.5, "date": 0.5555555555555556, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1176}, "ledger_by_hop": {"1": {"facts_total": 64, "facts_retained": 22, "retention_rate": 0.34375, "retention_by_kind": {"person": 0.5, "date": 0.5555555555555556, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1181}, "2": {"facts_total": 64, "facts_retained": 22, "retention_rate": 0.34375, "retention_by_kind": {"person": 0.5, "date": 0.5555555555555556, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1176}, "3": {"facts_total": 64, "facts_retained": 22, "retention_rate": 0.34375, "retention_by_kind": {"person": 0.5, "date": 0.5555555555555556, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1176}, "4": {"facts_total": 64, "facts_retained": 22, "retention_rate": 0.34375, "retention_by_kind": {"person": 0.5, "date": 0.5555555555555556, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1176}, "5": {"facts_total": 64, "facts_retained": 22, "retention_rate": 0.34375, "retention_by_kind": {"person": 0.5, "date": 0.5555555555555556, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1176}, "6": {"facts_total": 64, "facts_retained": 22, "retention_rate": 0.34375, "retention_by_kind": {"person": 0.5, "date": 0.5555555555555556, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1176}, "7": {"facts_total": 64, "facts_retained": 22, "retention_rate": 0.34375, "retention_by_kind": {"person": 0.5, "date": 0.5555555555555556, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1176}, "8": {"facts_total": 64, "facts_retained": 22, "retention_rate": 0.34375, "retention_by_kind": {"person": 0.5, "date": 0.5555555555555556, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1176}}}, "reference_key_id": "keys/synthetic/syn_03018_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "89438188d34426dbc3fdf77c29f9462c95075b81b820590bae58487694cd6183", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "43c2ff41109cca9f3d7cf376", "attempt_id": "a5aafea46567", "run_group_id": "E4_core:A4|closed|qwen3b:h8", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:08:25.189052Z", "task_family": "synthetic", "document_id": "syn_03018_2000", "document_hash": "d284eb07e27da742b9c2b67abccfdf72e4dd67c3259f44f3b41e34dd3e6fc1ff", "question_id": "q01_single_fact", "split": "test", "cluster_id": "syn_03018_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 8, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1982, "query_tokens": 600, "message_tokens": 511, "message_utf8_bytes": 1176, "cache_tensor_bytes": null, "serialized_payload_bytes": 1176, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 176645.02341777552, "hops": {"1": 22836.384875001386, "2": 22022.93874998577, "3": 21884.76395793259, "4": 21840.455957921222, "5": 22013.46108398866, "6": 21984.005416976288, "7": 22096.49654198438, "8": 21966.516833985224}, "receiver_prefill_ms": 354.47566700167954, "ttft_ms": 42.165499995462596, "decode_ms": 640.8868749858811}, "total_wall_ms": 177683.36537573487, "ttft_ms": 42.165499995462596, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 12}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "Lot 376-E, Baie-Comeau", "evidence_artifact": {"ledger": {"facts_total": 64, "facts_retained": 22, "retention_rate": 0.34375, "retention_by_kind": {"person": 0.5, "date": 0.5555555555555556, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1176}, "ledger_by_hop": {"1": {"facts_total": 64, "facts_retained": 22, "retention_rate": 0.34375, "retention_by_kind": {"person": 0.5, "date": 0.5555555555555556, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1181}, "2": {"facts_total": 64, "facts_retained": 22, "retention_rate": 0.34375, "retention_by_kind": {"person": 0.5, "date": 0.5555555555555556, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1176}, "3": {"facts_total": 64, "facts_retained": 22, "retention_rate": 0.34375, "retention_by_kind": {"person": 0.5, "date": 0.5555555555555556, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1176}, "4": {"facts_total": 64, "facts_retained": 22, "retention_rate": 0.34375, "retention_by_kind": {"person": 0.5, "date": 0.5555555555555556, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1176}, "5": {"facts_total": 64, "facts_retained": 22, "retention_rate": 0.34375, "retention_by_kind": {"person": 0.5, "date": 0.5555555555555556, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1176}, "6": {"facts_total": 64, "facts_retained": 22, "retention_rate": 0.34375, "retention_by_kind": {"person": 0.5, "date": 0.5555555555555556, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1176}, "7": {"facts_total": 64, "facts_retained": 22, "retention_rate": 0.34375, "retention_by_kind": {"person": 0.5, "date": 0.5555555555555556, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1176}, "8": {"facts_total": 64, "facts_retained": 22, "retention_rate": 0.34375, "retention_by_kind": {"person": 0.5, "date": 0.5555555555555556, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1176}}}, "reference_key_id": "keys/synthetic/syn_03018_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 0.8571428571428571}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "89438188d34426dbc3fdf77c29f9462c95075b81b820590bae58487694cd6183", "output_hash": "b58829a1440993c4af3307ba44494f021d726fd11e9337888b6b852a8b37351a"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "7c1c72c22a4c699910e47dce", "attempt_id": "3bc56f3ddc64", "run_group_id": "E4_core:A4|closed|qwen3b:h8", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:08:26.244101Z", "task_family": "synthetic", "document_id": "syn_03018_2000", "document_hash": "d284eb07e27da742b9c2b67abccfdf72e4dd67c3259f44f3b41e34dd3e6fc1ff", "question_id": "q02_exact_copy", "split": "test", "cluster_id": "syn_03018_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 8, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1982, "query_tokens": 599, "message_tokens": 511, "message_utf8_bytes": 1176, "cache_tensor_bytes": null, "serialized_payload_bytes": 1176, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 176645.02341777552, "hops": {"1": 22836.384875001386, "2": 22022.93874998577, "3": 21884.76395793259, "4": 21840.455957921222, "5": 22013.46108398866, "6": 21984.005416976288, "7": 22096.49654198438, "8": 21966.516833985224}, "receiver_prefill_ms": 401.7102080397308, "ttft_ms": 33.04104204289615, "decode_ms": 30.947416089475155}, "total_wall_ms": 177111.39491782524, "ttft_ms": 33.04104204289615, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 64, "facts_retained": 22, "retention_rate": 0.34375, "retention_by_kind": {"person": 0.5, "date": 0.5555555555555556, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1176}, "ledger_by_hop": {"1": {"facts_total": 64, "facts_retained": 22, "retention_rate": 0.34375, "retention_by_kind": {"person": 0.5, "date": 0.5555555555555556, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1181}, "2": {"facts_total": 64, "facts_retained": 22, "retention_rate": 0.34375, "retention_by_kind": {"person": 0.5, "date": 0.5555555555555556, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1176}, "3": {"facts_total": 64, "facts_retained": 22, "retention_rate": 0.34375, "retention_by_kind": {"person": 0.5, "date": 0.5555555555555556, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1176}, "4": {"facts_total": 64, "facts_retained": 22, "retention_rate": 0.34375, "retention_by_kind": {"person": 0.5, "date": 0.5555555555555556, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1176}, "5": {"facts_total": 64, "facts_retained": 22, "retention_rate": 0.34375, "retention_by_kind": {"person": 0.5, "date": 0.5555555555555556, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1176}, "6": {"facts_total": 64, "facts_retained": 22, "retention_rate": 0.34375, "retention_by_kind": {"person": 0.5, "date": 0.5555555555555556, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1176}, "7": {"facts_total": 64, "facts_retained": 22, "retention_rate": 0.34375, "retention_by_kind": {"person": 0.5, "date": 0.5555555555555556, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1176}, "8": {"facts_total": 64, "facts_retained": 22, "retention_rate": 0.34375, "retention_by_kind": {"person": 0.5, "date": 0.5555555555555556, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1176}}}, "reference_key_id": "keys/synthetic/syn_03018_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "89438188d34426dbc3fdf77c29f9462c95075b81b820590bae58487694cd6183", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "37d7f64132da2df96716314c", "attempt_id": "6627891c9706", "run_group_id": "E4_core:A4|closed|qwen3b:h8", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:08:26.728227Z", "task_family": "synthetic", "document_id": "syn_03018_2000", "document_hash": "d284eb07e27da742b9c2b67abccfdf72e4dd67c3259f44f3b41e34dd3e6fc1ff", "question_id": "q03_exact_copy", "split": "test", "cluster_id": "syn_03018_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 8, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1982, "query_tokens": 597, "message_tokens": 511, "message_utf8_bytes": 1176, "cache_tensor_bytes": null, "serialized_payload_bytes": 1176, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 176645.02341777552, "hops": {"1": 22836.384875001386, "2": 22022.93874998577, "3": 21884.76395793259, "4": 21840.455957921222, "5": 22013.46108398866, "6": 21984.005416976288, "7": 22096.49654198438, "8": 21966.516833985224}, "receiver_prefill_ms": 389.0474169747904, "ttft_ms": 52.92379204183817, "decode_ms": 556.8340409081429}, "total_wall_ms": 177644.5452927146, "ttft_ms": 52.92379204183817, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 9}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "Priyamvada Sørlie", "evidence_artifact": {"ledger": {"facts_total": 64, "facts_retained": 22, "retention_rate": 0.34375, "retention_by_kind": {"person": 0.5, "date": 0.5555555555555556, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1176}, "ledger_by_hop": {"1": {"facts_total": 64, "facts_retained": 22, "retention_rate": 0.34375, "retention_by_kind": {"person": 0.5, "date": 0.5555555555555556, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1181}, "2": {"facts_total": 64, "facts_retained": 22, "retention_rate": 0.34375, "retention_by_kind": {"person": 0.5, "date": 0.5555555555555556, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1176}, "3": {"facts_total": 64, "facts_retained": 22, "retention_rate": 0.34375, "retention_by_kind": {"person": 0.5, "date": 0.5555555555555556, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1176}, "4": {"facts_total": 64, "facts_retained": 22, "retention_rate": 0.34375, "retention_by_kind": {"person": 0.5, "date": 0.5555555555555556, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1176}, "5": {"facts_total": 64, "facts_retained": 22, "retention_rate": 0.34375, "retention_by_kind": {"person": 0.5, "date": 0.5555555555555556, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1176}, "6": {"facts_total": 64, "facts_retained": 22, "retention_rate": 0.34375, "retention_by_kind": {"person": 0.5, "date": 0.5555555555555556, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1176}, "7": {"facts_total": 64, "facts_retained": 22, "retention_rate": 0.34375, "retention_by_kind": {"person": 0.5, "date": 0.5555555555555556, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1176}, "8": {"facts_total": 64, "facts_retained": 22, "retention_rate": 0.34375, "retention_by_kind": {"person": 0.5, "date": 0.5555555555555556, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1176}}}, "reference_key_id": "keys/synthetic/syn_03018_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "89438188d34426dbc3fdf77c29f9462c95075b81b820590bae58487694cd6183", "output_hash": "c5f554196ed23dd71ad995ec89f7142099d524dbb6534cc10d94a0c89d13e792"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "97aab7d8aa55bff93ed7021a", "attempt_id": "9570593a5e8d", "run_group_id": "E4_core:A4|closed|qwen3b:h8", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:08:27.745528Z", "task_family": "synthetic", "document_id": "syn_03018_2000", "document_hash": "d284eb07e27da742b9c2b67abccfdf72e4dd67c3259f44f3b41e34dd3e6fc1ff", "question_id": "q04_multi_hop", "split": "test", "cluster_id": "syn_03018_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 8, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1982, "query_tokens": 603, "message_tokens": 511, "message_utf8_bytes": 1176, "cache_tensor_bytes": null, "serialized_payload_bytes": 1176, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 176645.02341777552, "hops": {"1": 22836.384875001386, "2": 22022.93874998577, "3": 21884.76395793259, "4": 21840.455957921222, "5": 22013.46108398866, "6": 21984.005416976288, "7": 22096.49654198438, "8": 21966.516833985224}, "receiver_prefill_ms": 415.1500000152737, "ttft_ms": 23.869708995334804, "decode_ms": 43.18512498866767}, "total_wall_ms": 177127.81879270915, "ttft_ms": 23.869708995334804, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 64, "facts_retained": 22, "retention_rate": 0.34375, "retention_by_kind": {"person": 0.5, "date": 0.5555555555555556, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1176}, "ledger_by_hop": {"1": {"facts_total": 64, "facts_retained": 22, "retention_rate": 0.34375, "retention_by_kind": {"person": 0.5, "date": 0.5555555555555556, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1181}, "2": {"facts_total": 64, "facts_retained": 22, "retention_rate": 0.34375, "retention_by_kind": {"person": 0.5, "date": 0.5555555555555556, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1176}, "3": {"facts_total": 64, "facts_retained": 22, "retention_rate": 0.34375, "retention_by_kind": {"person": 0.5, "date": 0.5555555555555556, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1176}, "4": {"facts_total": 64, "facts_retained": 22, "retention_rate": 0.34375, "retention_by_kind": {"person": 0.5, "date": 0.5555555555555556, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1176}, "5": {"facts_total": 64, "facts_retained": 22, "retention_rate": 0.34375, "retention_by_kind": {"person": 0.5, "date": 0.5555555555555556, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1176}, "6": {"facts_total": 64, "facts_retained": 22, "retention_rate": 0.34375, "retention_by_kind": {"person": 0.5, "date": 0.5555555555555556, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1176}, "7": {"facts_total": 64, "facts_retained": 22, "retention_rate": 0.34375, "retention_by_kind": {"person": 0.5, "date": 0.5555555555555556, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1176}, "8": {"facts_total": 64, "facts_retained": 22, "retention_rate": 0.34375, "retention_by_kind": {"person": 0.5, "date": 0.5555555555555556, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1176}}}, "reference_key_id": "keys/synthetic/syn_03018_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "89438188d34426dbc3fdf77c29f9462c95075b81b820590bae58487694cd6183", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "1739436f683b95f936b3de34", "attempt_id": "187c65b0dad9", "run_group_id": "E4_core:A4|closed|qwen3b:h8", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:08:28.246231Z", "task_family": "synthetic", "document_id": "syn_03018_2000", "document_hash": "d284eb07e27da742b9c2b67abccfdf72e4dd67c3259f44f3b41e34dd3e6fc1ff", "question_id": "q05_multi_hop", "split": "test", "cluster_id": "syn_03018_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 8, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1982, "query_tokens": 603, "message_tokens": 511, "message_utf8_bytes": 1176, "cache_tensor_bytes": null, "serialized_payload_bytes": 1176, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 176645.02341777552, "hops": {"1": 22836.384875001386, "2": 22022.93874998577, "3": 21884.76395793259, "4": 21840.455957921222, "5": 22013.46108398866, "6": 21984.005416976288, "7": 22096.49654198438, "8": 21966.516833985224}, "receiver_prefill_ms": 366.76441691815853, "ttft_ms": 42.800750001333654, "decode_ms": 647.5362499477342}, "total_wall_ms": 177702.91287673172, "ttft_ms": 42.800750001333654, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 10}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "HM-38K-5533", "evidence_artifact": {"ledger": {"facts_total": 64, "facts_retained": 22, "retention_rate": 0.34375, "retention_by_kind": {"person": 0.5, "date": 0.5555555555555556, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1176}, "ledger_by_hop": {"1": {"facts_total": 64, "facts_retained": 22, "retention_rate": 0.34375, "retention_by_kind": {"person": 0.5, "date": 0.5555555555555556, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1181}, "2": {"facts_total": 64, "facts_retained": 22, "retention_rate": 0.34375, "retention_by_kind": {"person": 0.5, "date": 0.5555555555555556, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1176}, "3": {"facts_total": 64, "facts_retained": 22, "retention_rate": 0.34375, "retention_by_kind": {"person": 0.5, "date": 0.5555555555555556, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1176}, "4": {"facts_total": 64, "facts_retained": 22, "retention_rate": 0.34375, "retention_by_kind": {"person": 0.5, "date": 0.5555555555555556, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1176}, "5": {"facts_total": 64, "facts_retained": 22, "retention_rate": 0.34375, "retention_by_kind": {"person": 0.5, "date": 0.5555555555555556, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1176}, "6": {"facts_total": 64, "facts_retained": 22, "retention_rate": 0.34375, "retention_by_kind": {"person": 0.5, "date": 0.5555555555555556, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1176}, "7": {"facts_total": 64, "facts_retained": 22, "retention_rate": 0.34375, "retention_by_kind": {"person": 0.5, "date": 0.5555555555555556, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1176}, "8": {"facts_total": 64, "facts_retained": 22, "retention_rate": 0.34375, "retention_by_kind": {"person": 0.5, "date": 0.5555555555555556, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1176}}}, "reference_key_id": "keys/synthetic/syn_03018_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "89438188d34426dbc3fdf77c29f9462c95075b81b820590bae58487694cd6183", "output_hash": "18acd9807857243bbded3bceae55398316a3958b81d70364441bdb6748eaf878"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "f171a06d01a334ab118cebb4", "attempt_id": "324787cf7508", "run_group_id": "E4_core:A4|closed|qwen3b:h8", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:08:29.320896Z", "task_family": "synthetic", "document_id": "syn_03018_2000", "document_hash": "d284eb07e27da742b9c2b67abccfdf72e4dd67c3259f44f3b41e34dd3e6fc1ff", "question_id": "q06_negation", "split": "test", "cluster_id": "syn_03018_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 8, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1982, "query_tokens": 595, "message_tokens": 511, "message_utf8_bytes": 1176, "cache_tensor_bytes": null, "serialized_payload_bytes": 1176, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 176645.02341777552, "hops": {"1": 22836.384875001386, "2": 22022.93874998577, "3": 21884.76395793259, "4": 21840.455957921222, "5": 22013.46108398866, "6": 21984.005416976288, "7": 22096.49654198438, "8": 21966.516833985224}, "receiver_prefill_ms": 410.9870829852298, "ttft_ms": 25.798625079914927, "decode_ms": 41.1711250199005}, "total_wall_ms": 177123.90704278369, "ttft_ms": 25.798625079914927, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 64, "facts_retained": 22, "retention_rate": 0.34375, "retention_by_kind": {"person": 0.5, "date": 0.5555555555555556, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1176}, "ledger_by_hop": {"1": {"facts_total": 64, "facts_retained": 22, "retention_rate": 0.34375, "retention_by_kind": {"person": 0.5, "date": 0.5555555555555556, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1181}, "2": {"facts_total": 64, "facts_retained": 22, "retention_rate": 0.34375, "retention_by_kind": {"person": 0.5, "date": 0.5555555555555556, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1176}, "3": {"facts_total": 64, "facts_retained": 22, "retention_rate": 0.34375, "retention_by_kind": {"person": 0.5, "date": 0.5555555555555556, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1176}, "4": {"facts_total": 64, "facts_retained": 22, "retention_rate": 0.34375, "retention_by_kind": {"person": 0.5, "date": 0.5555555555555556, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1176}, "5": {"facts_total": 64, "facts_retained": 22, "retention_rate": 0.34375, "retention_by_kind": {"person": 0.5, "date": 0.5555555555555556, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1176}, "6": {"facts_total": 64, "facts_retained": 22, "retention_rate": 0.34375, "retention_by_kind": {"person": 0.5, "date": 0.5555555555555556, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1176}, "7": {"facts_total": 64, "facts_retained": 22, "retention_rate": 0.34375, "retention_by_kind": {"person": 0.5, "date": 0.5555555555555556, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1176}, "8": {"facts_total": 64, "facts_retained": 22, "retention_rate": 0.34375, "retention_by_kind": {"person": 0.5, "date": 0.5555555555555556, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1176}}}, "reference_key_id": "keys/synthetic/syn_03018_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "89438188d34426dbc3fdf77c29f9462c95075b81b820590bae58487694cd6183", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "52bb32356bb28cd7f5ad3755", "attempt_id": "c12c20bdea81", "run_group_id": "E4_core:A4|closed|qwen3b:h8", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:08:29.818493Z", "task_family": "synthetic", "document_id": "syn_03018_2000", "document_hash": "d284eb07e27da742b9c2b67abccfdf72e4dd67c3259f44f3b41e34dd3e6fc1ff", "question_id": "q07_negation", "split": "test", "cluster_id": "syn_03018_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 8, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1982, "query_tokens": 594, "message_tokens": 511, "message_utf8_bytes": 1176, "cache_tensor_bytes": null, "serialized_payload_bytes": 1176, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 176645.02341777552, "hops": {"1": 22836.384875001386, "2": 22022.93874998577, "3": 21884.76395793259, "4": 21840.455957921222, "5": 22013.46108398866, "6": 21984.005416976288, "7": 22096.49654198438, "8": 21966.516833985224}, "receiver_prefill_ms": 369.17220905888826, "ttft_ms": 41.20454203803092, "decode_ms": 60.698291985318065}, "total_wall_ms": 177116.98925076053, "ttft_ms": 41.20454203803092, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "no", "evidence_artifact": {"ledger": {"facts_total": 64, "facts_retained": 22, "retention_rate": 0.34375, "retention_by_kind": {"person": 0.5, "date": 0.5555555555555556, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1176}, "ledger_by_hop": {"1": {"facts_total": 64, "facts_retained": 22, "retention_rate": 0.34375, "retention_by_kind": {"person": 0.5, "date": 0.5555555555555556, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1181}, "2": {"facts_total": 64, "facts_retained": 22, "retention_rate": 0.34375, "retention_by_kind": {"person": 0.5, "date": 0.5555555555555556, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1176}, "3": {"facts_total": 64, "facts_retained": 22, "retention_rate": 0.34375, "retention_by_kind": {"person": 0.5, "date": 0.5555555555555556, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1176}, "4": {"facts_total": 64, "facts_retained": 22, "retention_rate": 0.34375, "retention_by_kind": {"person": 0.5, "date": 0.5555555555555556, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1176}, "5": {"facts_total": 64, "facts_retained": 22, "retention_rate": 0.34375, "retention_by_kind": {"person": 0.5, "date": 0.5555555555555556, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1176}, "6": {"facts_total": 64, "facts_retained": 22, "retention_rate": 0.34375, "retention_by_kind": {"person": 0.5, "date": 0.5555555555555556, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1176}, "7": {"facts_total": 64, "facts_retained": 22, "retention_rate": 0.34375, "retention_by_kind": {"person": 0.5, "date": 0.5555555555555556, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1176}, "8": {"facts_total": 64, "facts_retained": 22, "retention_rate": 0.34375, "retention_by_kind": {"person": 0.5, "date": 0.5555555555555556, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1176}}}, "reference_key_id": "keys/synthetic/syn_03018_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "89438188d34426dbc3fdf77c29f9462c95075b81b820590bae58487694cd6183", "output_hash": "9390298f3fb0c5b160498935d79cb139aef28e1c47358b4bbba61862b9c26e59"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "7de9408a800cce9abe6c804b", "attempt_id": "8b79fb56a0a5", "run_group_id": "E4_core:A4|closed|qwen3b:h8", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:08:30.309661Z", "task_family": "synthetic", "document_id": "syn_03018_2000", "document_hash": "d284eb07e27da742b9c2b67abccfdf72e4dd67c3259f44f3b41e34dd3e6fc1ff", "question_id": "q08_revision", "split": "test", "cluster_id": "syn_03018_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 8, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1982, "query_tokens": 612, "message_tokens": 511, "message_utf8_bytes": 1176, "cache_tensor_bytes": null, "serialized_payload_bytes": 1176, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 176645.02341777552, "hops": {"1": 22836.384875001386, "2": 22022.93874998577, "3": 21884.76395793259, "4": 21840.455957921222, "5": 22013.46108398866, "6": 21984.005416976288, "7": 22096.49654198438, "8": 21966.516833985224}, "receiver_prefill_ms": 423.9280830370262, "ttft_ms": 139.12720896769315, "decode_ms": 187.16199998743832}, "total_wall_ms": 177396.0258347215, "ttft_ms": 139.12720896769315, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 6}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "304815", "evidence_artifact": {"ledger": {"facts_total": 64, "facts_retained": 22, "retention_rate": 0.34375, "retention_by_kind": {"person": 0.5, "date": 0.5555555555555556, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1176}, "ledger_by_hop": {"1": {"facts_total": 64, "facts_retained": 22, "retention_rate": 0.34375, "retention_by_kind": {"person": 0.5, "date": 0.5555555555555556, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1181}, "2": {"facts_total": 64, "facts_retained": 22, "retention_rate": 0.34375, "retention_by_kind": {"person": 0.5, "date": 0.5555555555555556, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1176}, "3": {"facts_total": 64, "facts_retained": 22, "retention_rate": 0.34375, "retention_by_kind": {"person": 0.5, "date": 0.5555555555555556, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1176}, "4": {"facts_total": 64, "facts_retained": 22, "retention_rate": 0.34375, "retention_by_kind": {"person": 0.5, "date": 0.5555555555555556, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1176}, "5": {"facts_total": 64, "facts_retained": 22, "retention_rate": 0.34375, "retention_by_kind": {"person": 0.5, "date": 0.5555555555555556, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1176}, "6": {"facts_total": 64, "facts_retained": 22, "retention_rate": 0.34375, "retention_by_kind": {"person": 0.5, "date": 0.5555555555555556, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1176}, "7": {"facts_total": 64, "facts_retained": 22, "retention_rate": 0.34375, "retention_by_kind": {"person": 0.5, "date": 0.5555555555555556, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1176}, "8": {"facts_total": 64, "facts_retained": 22, "retention_rate": 0.34375, "retention_by_kind": {"person": 0.5, "date": 0.5555555555555556, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1176}}}, "reference_key_id": "keys/synthetic/syn_03018_2000.json", "task_scores": {"correct": 0.0, "abstained": 0.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "89438188d34426dbc3fdf77c29f9462c95075b81b820590bae58487694cd6183", "output_hash": "a480475d6855141c920f42d1bf3337ca7fd88bcf4087b0ed2e253ed68cca6926"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "c58bdaf1740bb2cdae35093c", "attempt_id": "672b1072bded", "run_group_id": "E4_core:A4|closed|qwen3b:h8", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:08:31.077961Z", "task_family": "synthetic", "document_id": "syn_03018_2000", "document_hash": "d284eb07e27da742b9c2b67abccfdf72e4dd67c3259f44f3b41e34dd3e6fc1ff", "question_id": "q09_revision", "split": "test", "cluster_id": "syn_03018_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 8, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1982, "query_tokens": 605, "message_tokens": 511, "message_utf8_bytes": 1176, "cache_tensor_bytes": null, "serialized_payload_bytes": 1176, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 176645.02341777552, "hops": {"1": 22836.384875001386, "2": 22022.93874998577, "3": 21884.76395793259, "4": 21840.455957921222, "5": 22013.46108398866, "6": 21984.005416976288, "7": 22096.49654198438, "8": 21966.516833985224}, "receiver_prefill_ms": 398.1489579891786, "ttft_ms": 42.87891602143645, "decode_ms": 186.01433304138482}, "total_wall_ms": 177272.68812584225, "ttft_ms": 42.87891602143645, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 6}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "304815", "evidence_artifact": {"ledger": {"facts_total": 64, "facts_retained": 22, "retention_rate": 0.34375, "retention_by_kind": {"person": 0.5, "date": 0.5555555555555556, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1176}, "ledger_by_hop": {"1": {"facts_total": 64, "facts_retained": 22, "retention_rate": 0.34375, "retention_by_kind": {"person": 0.5, "date": 0.5555555555555556, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1181}, "2": {"facts_total": 64, "facts_retained": 22, "retention_rate": 0.34375, "retention_by_kind": {"person": 0.5, "date": 0.5555555555555556, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1176}, "3": {"facts_total": 64, "facts_retained": 22, "retention_rate": 0.34375, "retention_by_kind": {"person": 0.5, "date": 0.5555555555555556, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1176}, "4": {"facts_total": 64, "facts_retained": 22, "retention_rate": 0.34375, "retention_by_kind": {"person": 0.5, "date": 0.5555555555555556, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1176}, "5": {"facts_total": 64, "facts_retained": 22, "retention_rate": 0.34375, "retention_by_kind": {"person": 0.5, "date": 0.5555555555555556, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1176}, "6": {"facts_total": 64, "facts_retained": 22, "retention_rate": 0.34375, "retention_by_kind": {"person": 0.5, "date": 0.5555555555555556, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1176}, "7": {"facts_total": 64, "facts_retained": 22, "retention_rate": 0.34375, "retention_by_kind": {"person": 0.5, "date": 0.5555555555555556, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1176}, "8": {"facts_total": 64, "facts_retained": 22, "retention_rate": 0.34375, "retention_by_kind": {"person": 0.5, "date": 0.5555555555555556, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1176}}}, "reference_key_id": "keys/synthetic/syn_03018_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "89438188d34426dbc3fdf77c29f9462c95075b81b820590bae58487694cd6183", "output_hash": "a480475d6855141c920f42d1bf3337ca7fd88bcf4087b0ed2e253ed68cca6926"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "bda2a789d19386cc2e3b5d8d", "attempt_id": "3201aff37a13", "run_group_id": "E4_core:A4|closed|qwen3b:h8", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:08:31.723001Z", "task_family": "synthetic", "document_id": "syn_03018_2000", "document_hash": "d284eb07e27da742b9c2b67abccfdf72e4dd67c3259f44f3b41e34dd3e6fc1ff", "question_id": "q10_temporal", "split": "test", "cluster_id": "syn_03018_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 8, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1982, "query_tokens": 613, "message_tokens": 511, "message_utf8_bytes": 1176, "cache_tensor_bytes": null, "serialized_payload_bytes": 1176, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 176645.02341777552, "hops": {"1": 22836.384875001386, "2": 22022.93874998577, "3": 21884.76395793259, "4": 21840.455957921222, "5": 22013.46108398866, "6": 21984.005416976288, "7": 22096.49654198438, "8": 21966.516833985224}, "receiver_prefill_ms": 352.62091597542167, "ttft_ms": 66.45491695962846, "decode_ms": 391.6824589250609}, "total_wall_ms": 177456.4609597437, "ttft_ms": 66.45491695962846, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 6}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "Melymi-15", "evidence_artifact": {"ledger": {"facts_total": 64, "facts_retained": 22, "retention_rate": 0.34375, "retention_by_kind": {"person": 0.5, "date": 0.5555555555555556, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1176}, "ledger_by_hop": {"1": {"facts_total": 64, "facts_retained": 22, "retention_rate": 0.34375, "retention_by_kind": {"person": 0.5, "date": 0.5555555555555556, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1181}, "2": {"facts_total": 64, "facts_retained": 22, "retention_rate": 0.34375, "retention_by_kind": {"person": 0.5, "date": 0.5555555555555556, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1176}, "3": {"facts_total": 64, "facts_retained": 22, "retention_rate": 0.34375, "retention_by_kind": {"person": 0.5, "date": 0.5555555555555556, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1176}, "4": {"facts_total": 64, "facts_retained": 22, "retention_rate": 0.34375, "retention_by_kind": {"person": 0.5, "date": 0.5555555555555556, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1176}, "5": {"facts_total": 64, "facts_retained": 22, "retention_rate": 0.34375, "retention_by_kind": {"person": 0.5, "date": 0.5555555555555556, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1176}, "6": {"facts_total": 64, "facts_retained": 22, "retention_rate": 0.34375, "retention_by_kind": {"person": 0.5, "date": 0.5555555555555556, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1176}, "7": {"facts_total": 64, "facts_retained": 22, "retention_rate": 0.34375, "retention_by_kind": {"person": 0.5, "date": 0.5555555555555556, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1176}, "8": {"facts_total": 64, "facts_retained": 22, "retention_rate": 0.34375, "retention_by_kind": {"person": 0.5, "date": 0.5555555555555556, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1176}}}, "reference_key_id": "keys/synthetic/syn_03018_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "89438188d34426dbc3fdf77c29f9462c95075b81b820590bae58487694cd6183", "output_hash": "fa08d1474b01d6dfbf3dea049be251aa3c3021c17b4818258064b156509af58b"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "af0193ee9a4dad5898c88219", "attempt_id": "486d619d6b3b", "run_group_id": "E4_core:A4|closed|qwen3b:h8", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:08:32.552197Z", "task_family": "synthetic", "document_id": "syn_03018_2000", "document_hash": "d284eb07e27da742b9c2b67abccfdf72e4dd67c3259f44f3b41e34dd3e6fc1ff", "question_id": "q11_aggregation", "split": "test", "cluster_id": "syn_03018_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 8, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1982, "query_tokens": 605, "message_tokens": 511, "message_utf8_bytes": 1176, "cache_tensor_bytes": null, "serialized_payload_bytes": 1176, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 176645.02341777552, "hops": {"1": 22836.384875001386, "2": 22022.93874998577, "3": 21884.76395793259, "4": 21840.455957921222, "5": 22013.46108398866, "6": 21984.005416976288, "7": 22096.49654198438, "8": 21966.516833985224}, "receiver_prefill_ms": 405.4132499732077, "ttft_ms": 36.69562505092472, "decode_ms": 226.41474998090416}, "total_wall_ms": 177314.2910427414, "ttft_ms": 36.69562505092472, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 7}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "2410460", "evidence_artifact": {"ledger": {"facts_total": 64, "facts_retained": 22, "retention_rate": 0.34375, "retention_by_kind": {"person": 0.5, "date": 0.5555555555555556, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1176}, "ledger_by_hop": {"1": {"facts_total": 64, "facts_retained": 22, "retention_rate": 0.34375, "retention_by_kind": {"person": 0.5, "date": 0.5555555555555556, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1181}, "2": {"facts_total": 64, "facts_retained": 22, "retention_rate": 0.34375, "retention_by_kind": {"person": 0.5, "date": 0.5555555555555556, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1176}, "3": {"facts_total": 64, "facts_retained": 22, "retention_rate": 0.34375, "retention_by_kind": {"person": 0.5, "date": 0.5555555555555556, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1176}, "4": {"facts_total": 64, "facts_retained": 22, "retention_rate": 0.34375, "retention_by_kind": {"person": 0.5, "date": 0.5555555555555556, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1176}, "5": {"facts_total": 64, "facts_retained": 22, "retention_rate": 0.34375, "retention_by_kind": {"person": 0.5, "date": 0.5555555555555556, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1176}, "6": {"facts_total": 64, "facts_retained": 22, "retention_rate": 0.34375, "retention_by_kind": {"person": 0.5, "date": 0.5555555555555556, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1176}, "7": {"facts_total": 64, "facts_retained": 22, "retention_rate": 0.34375, "retention_by_kind": {"person": 0.5, "date": 0.5555555555555556, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1176}, "8": {"facts_total": 64, "facts_retained": 22, "retention_rate": 0.34375, "retention_by_kind": {"person": 0.5, "date": 0.5555555555555556, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1176}}}, "reference_key_id": "keys/synthetic/syn_03018_2000.json", "task_scores": {"correct": 0.0, "abstained": 0.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "89438188d34426dbc3fdf77c29f9462c95075b81b820590bae58487694cd6183", "output_hash": "63fe326637abd22b1b1e457840e07ea81e971b7c680f6bccdf09ebe8bd7c8b80"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "a2d1374eee4d4dfdb2503476", "attempt_id": "4e097e99fada", "run_group_id": "E4_core:A4|closed|qwen3b:h8", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:08:33.239371Z", "task_family": "synthetic", "document_id": "syn_03018_2000", "document_hash": "d284eb07e27da742b9c2b67abccfdf72e4dd67c3259f44f3b41e34dd3e6fc1ff", "question_id": "q12_unit_conversion", "split": "test", "cluster_id": "syn_03018_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 8, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1982, "query_tokens": 602, "message_tokens": 511, "message_utf8_bytes": 1176, "cache_tensor_bytes": null, "serialized_payload_bytes": 1176, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 176645.02341777552, "hops": {"1": 22836.384875001386, "2": 22022.93874998577, "3": 21884.76395793259, "4": 21840.455957921222, "5": 22013.46108398866, "6": 21984.005416976288, "7": 22096.49654198438, "8": 21966.516833985224}, "receiver_prefill_ms": 357.5739589286968, "ttft_ms": 42.982707964256406, "decode_ms": 55.423749960027635}, "total_wall_ms": 177101.72033472918, "ttft_ms": 42.982707964256406, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 64, "facts_retained": 22, "retention_rate": 0.34375, "retention_by_kind": {"person": 0.5, "date": 0.5555555555555556, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1176}, "ledger_by_hop": {"1": {"facts_total": 64, "facts_retained": 22, "retention_rate": 0.34375, "retention_by_kind": {"person": 0.5, "date": 0.5555555555555556, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1181}, "2": {"facts_total": 64, "facts_retained": 22, "retention_rate": 0.34375, "retention_by_kind": {"person": 0.5, "date": 0.5555555555555556, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1176}, "3": {"facts_total": 64, "facts_retained": 22, "retention_rate": 0.34375, "retention_by_kind": {"person": 0.5, "date": 0.5555555555555556, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1176}, "4": {"facts_total": 64, "facts_retained": 22, "retention_rate": 0.34375, "retention_by_kind": {"person": 0.5, "date": 0.5555555555555556, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1176}, "5": {"facts_total": 64, "facts_retained": 22, "retention_rate": 0.34375, "retention_by_kind": {"person": 0.5, "date": 0.5555555555555556, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1176}, "6": {"facts_total": 64, "facts_retained": 22, "retention_rate": 0.34375, "retention_by_kind": {"person": 0.5, "date": 0.5555555555555556, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1176}, "7": {"facts_total": 64, "facts_retained": 22, "retention_rate": 0.34375, "retention_by_kind": {"person": 0.5, "date": 0.5555555555555556, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1176}, "8": {"facts_total": 64, "facts_retained": 22, "retention_rate": 0.34375, "retention_by_kind": {"person": 0.5, "date": 0.5555555555555556, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1176}}}, "reference_key_id": "keys/synthetic/syn_03018_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "89438188d34426dbc3fdf77c29f9462c95075b81b820590bae58487694cd6183", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "26c8722f2c8d67c03ddc7d03", "attempt_id": "3bf8f404b1ce", "run_group_id": "E4_core:A4|closed|qwen3b:h8", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:08:33.713311Z", "task_family": "synthetic", "document_id": "syn_03018_2000", "document_hash": "d284eb07e27da742b9c2b67abccfdf72e4dd67c3259f44f3b41e34dd3e6fc1ff", "question_id": "q13_unanswerable", "split": "test", "cluster_id": "syn_03018_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 8, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1982, "query_tokens": 606, "message_tokens": 511, "message_utf8_bytes": 1176, "cache_tensor_bytes": null, "serialized_payload_bytes": 1176, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 176645.02341777552, "hops": {"1": 22836.384875001386, "2": 22022.93874998577, "3": 21884.76395793259, "4": 21840.455957921222, "5": 22013.46108398866, "6": 21984.005416976288, "7": 22096.49654198438, "8": 21966.516833985224}, "receiver_prefill_ms": 540.4582080664113, "ttft_ms": 22.890874999575317, "decode_ms": 41.249459027312696}, "total_wall_ms": 177250.53641782142, "ttft_ms": 22.890874999575317, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 64, "facts_retained": 22, "retention_rate": 0.34375, "retention_by_kind": {"person": 0.5, "date": 0.5555555555555556, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1176}, "ledger_by_hop": {"1": {"facts_total": 64, "facts_retained": 22, "retention_rate": 0.34375, "retention_by_kind": {"person": 0.5, "date": 0.5555555555555556, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1181}, "2": {"facts_total": 64, "facts_retained": 22, "retention_rate": 0.34375, "retention_by_kind": {"person": 0.5, "date": 0.5555555555555556, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1176}, "3": {"facts_total": 64, "facts_retained": 22, "retention_rate": 0.34375, "retention_by_kind": {"person": 0.5, "date": 0.5555555555555556, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1176}, "4": {"facts_total": 64, "facts_retained": 22, "retention_rate": 0.34375, "retention_by_kind": {"person": 0.5, "date": 0.5555555555555556, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1176}, "5": {"facts_total": 64, "facts_retained": 22, "retention_rate": 0.34375, "retention_by_kind": {"person": 0.5, "date": 0.5555555555555556, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1176}, "6": {"facts_total": 64, "facts_retained": 22, "retention_rate": 0.34375, "retention_by_kind": {"person": 0.5, "date": 0.5555555555555556, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1176}, "7": {"facts_total": 64, "facts_retained": 22, "retention_rate": 0.34375, "retention_by_kind": {"person": 0.5, "date": 0.5555555555555556, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1176}, "8": {"facts_total": 64, "facts_retained": 22, "retention_rate": 0.34375, "retention_by_kind": {"person": 0.5, "date": 0.5555555555555556, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1176}}}, "reference_key_id": "keys/synthetic/syn_03018_2000.json", "task_scores": {"correct": 1.0, "abstained": 1.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "89438188d34426dbc3fdf77c29f9462c95075b81b820590bae58487694cd6183", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "f2c54f4bddfdca134b95711f", "attempt_id": "d349c25a813b", "run_group_id": "E4_core:A4|closed|qwen3b:h8", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:08:34.336950Z", "task_family": "synthetic", "document_id": "syn_03018_2000", "document_hash": "d284eb07e27da742b9c2b67abccfdf72e4dd67c3259f44f3b41e34dd3e6fc1ff", "question_id": "q14_unanswerable", "split": "test", "cluster_id": "syn_03018_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 8, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1982, "query_tokens": 607, "message_tokens": 511, "message_utf8_bytes": 1176, "cache_tensor_bytes": null, "serialized_payload_bytes": 1176, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 176645.02341777552, "hops": {"1": 22836.384875001386, "2": 22022.93874998577, "3": 21884.76395793259, "4": 21840.455957921222, "5": 22013.46108398866, "6": 21984.005416976288, "7": 22096.49654198438, "8": 21966.516833985224}, "receiver_prefill_ms": 410.45966697856784, "ttft_ms": 43.45833300612867, "decode_ms": 25.14908299781382}, "total_wall_ms": 177124.75433479995, "ttft_ms": 43.45833300612867, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 64, "facts_retained": 22, "retention_rate": 0.34375, "retention_by_kind": {"person": 0.5, "date": 0.5555555555555556, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1176}, "ledger_by_hop": {"1": {"facts_total": 64, "facts_retained": 22, "retention_rate": 0.34375, "retention_by_kind": {"person": 0.5, "date": 0.5555555555555556, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1181}, "2": {"facts_total": 64, "facts_retained": 22, "retention_rate": 0.34375, "retention_by_kind": {"person": 0.5, "date": 0.5555555555555556, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1176}, "3": {"facts_total": 64, "facts_retained": 22, "retention_rate": 0.34375, "retention_by_kind": {"person": 0.5, "date": 0.5555555555555556, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1176}, "4": {"facts_total": 64, "facts_retained": 22, "retention_rate": 0.34375, "retention_by_kind": {"person": 0.5, "date": 0.5555555555555556, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1176}, "5": {"facts_total": 64, "facts_retained": 22, "retention_rate": 0.34375, "retention_by_kind": {"person": 0.5, "date": 0.5555555555555556, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1176}, "6": {"facts_total": 64, "facts_retained": 22, "retention_rate": 0.34375, "retention_by_kind": {"person": 0.5, "date": 0.5555555555555556, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1176}, "7": {"facts_total": 64, "facts_retained": 22, "retention_rate": 0.34375, "retention_by_kind": {"person": 0.5, "date": 0.5555555555555556, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1176}, "8": {"facts_total": 64, "facts_retained": 22, "retention_rate": 0.34375, "retention_by_kind": {"person": 0.5, "date": 0.5555555555555556, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1176}}}, "reference_key_id": "keys/synthetic/syn_03018_2000.json", "task_scores": {"correct": 1.0, "abstained": 1.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "89438188d34426dbc3fdf77c29f9462c95075b81b820590bae58487694cd6183", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "b35ff29b4ad9e4c749147935", "attempt_id": "3e412400af5f", "run_group_id": "E4_core:A5|closed|qwen3b:h1", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:08:57.636318Z", "task_family": "synthetic", "document_id": "syn_03018_2000", "document_hash": "d284eb07e27da742b9c2b67abccfdf72e4dd67c3259f44f3b41e34dd3e6fc1ff", "question_id": "q00_single_fact", "split": "test", "cluster_id": "syn_03018_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 1, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1982, "query_tokens": 484, "message_tokens": 401, "message_utf8_bytes": 940, "cache_tensor_bytes": null, "serialized_payload_bytes": 940, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 22800.499332952313, "hops": {"1": 22800.499332952313}, "receiver_prefill_ms": 460.31124994624406, "ttft_ms": 35.57279193773866, "decode_ms": 29.103583889082074}, "total_wall_ms": 23326.19033299852, "ttft_ms": 35.57279193773866, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 64, "facts_retained": 14, "retention_rate": 0.21875, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.2222222222222222, "amount": 0.15384615384615385, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 870}, "ledger_by_hop": {"1": {"facts_total": 64, "facts_retained": 14, "retention_rate": 0.21875, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.2222222222222222, "amount": 0.15384615384615385, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 870}}}, "reference_key_id": "keys/synthetic/syn_03018_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "b93ce255451151cce89ccd51cc89c5b4382f883b441b3818bc249f2f79c54d9e", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "4b61ddd4752e6237d870d2b5", "attempt_id": "a3071cb7587c", "run_group_id": "E4_core:A5|closed|qwen3b:h1", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:08:58.180592Z", "task_family": "synthetic", "document_id": "syn_03018_2000", "document_hash": "d284eb07e27da742b9c2b67abccfdf72e4dd67c3259f44f3b41e34dd3e6fc1ff", "question_id": "q01_single_fact", "split": "test", "cluster_id": "syn_03018_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 1, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1982, "query_tokens": 490, "message_tokens": 401, "message_utf8_bytes": 940, "cache_tensor_bytes": null, "serialized_payload_bytes": 940, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 22800.499332952313, "hops": {"1": 22800.499332952313}, "receiver_prefill_ms": 311.5231670672074, "ttft_ms": 31.878417008556426, "decode_ms": 39.48229202069342}, "total_wall_ms": 23183.964999974705, "ttft_ms": 31.878417008556426, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 64, "facts_retained": 14, "retention_rate": 0.21875, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.2222222222222222, "amount": 0.15384615384615385, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 870}, "ledger_by_hop": {"1": {"facts_total": 64, "facts_retained": 14, "retention_rate": 0.21875, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.2222222222222222, "amount": 0.15384615384615385, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 870}}}, "reference_key_id": "keys/synthetic/syn_03018_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "b93ce255451151cce89ccd51cc89c5b4382f883b441b3818bc249f2f79c54d9e", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "5183c66632b5c497e8da2a3c", "attempt_id": "541292a7fc2e", "run_group_id": "E4_core:A5|closed|qwen3b:h1", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:08:58.581769Z", "task_family": "synthetic", "document_id": "syn_03018_2000", "document_hash": "d284eb07e27da742b9c2b67abccfdf72e4dd67c3259f44f3b41e34dd3e6fc1ff", "question_id": "q02_exact_copy", "split": "test", "cluster_id": "syn_03018_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 1, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1982, "query_tokens": 489, "message_tokens": 401, "message_utf8_bytes": 940, "cache_tensor_bytes": null, "serialized_payload_bytes": 940, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 22800.499332952313, "hops": {"1": 22800.499332952313}, "receiver_prefill_ms": 318.3889579959214, "ttft_ms": 20.31237503979355, "decode_ms": 20.272583002224565}, "total_wall_ms": 23160.1190000074, "ttft_ms": 20.31237503979355, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 64, "facts_retained": 14, "retention_rate": 0.21875, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.2222222222222222, "amount": 0.15384615384615385, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 870}, "ledger_by_hop": {"1": {"facts_total": 64, "facts_retained": 14, "retention_rate": 0.21875, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.2222222222222222, "amount": 0.15384615384615385, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 870}}}, "reference_key_id": "keys/synthetic/syn_03018_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "b93ce255451151cce89ccd51cc89c5b4382f883b441b3818bc249f2f79c54d9e", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "8404b1836dd3294842ac6141", "attempt_id": "dee4d707f552", "run_group_id": "E4_core:A5|closed|qwen3b:h1", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:08:58.958806Z", "task_family": "synthetic", "document_id": "syn_03018_2000", "document_hash": "d284eb07e27da742b9c2b67abccfdf72e4dd67c3259f44f3b41e34dd3e6fc1ff", "question_id": "q03_exact_copy", "split": "test", "cluster_id": "syn_03018_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 1, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1982, "query_tokens": 487, "message_tokens": 401, "message_utf8_bytes": 940, "cache_tensor_bytes": null, "serialized_payload_bytes": 940, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 22800.499332952313, "hops": {"1": 22800.499332952313}, "receiver_prefill_ms": 351.4412089716643, "ttft_ms": 61.86950008850545, "decode_ms": 487.3210000805557}, "total_wall_ms": 23701.902249013074, "ttft_ms": 61.86950008850545, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 9}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "Priyamvada Sørlie", "evidence_artifact": {"ledger": {"facts_total": 64, "facts_retained": 14, "retention_rate": 0.21875, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.2222222222222222, "amount": 0.15384615384615385, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 870}, "ledger_by_hop": {"1": {"facts_total": 64, "facts_retained": 14, "retention_rate": 0.21875, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.2222222222222222, "amount": 0.15384615384615385, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 870}}}, "reference_key_id": "keys/synthetic/syn_03018_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "b93ce255451151cce89ccd51cc89c5b4382f883b441b3818bc249f2f79c54d9e", "output_hash": "c5f554196ed23dd71ad995ec89f7142099d524dbb6534cc10d94a0c89d13e792"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "eec40f27153cdd627b9b5111", "attempt_id": "9bc9cce7daec", "run_group_id": "E4_core:A5|closed|qwen3b:h1", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:08:59.881070Z", "task_family": "synthetic", "document_id": "syn_03018_2000", "document_hash": "d284eb07e27da742b9c2b67abccfdf72e4dd67c3259f44f3b41e34dd3e6fc1ff", "question_id": "q04_multi_hop", "split": "test", "cluster_id": "syn_03018_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 1, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1982, "query_tokens": 493, "message_tokens": 401, "message_utf8_bytes": 940, "cache_tensor_bytes": null, "serialized_payload_bytes": 940, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 22800.499332952313, "hops": {"1": 22800.499332952313}, "receiver_prefill_ms": 351.52599995490164, "ttft_ms": 24.29266704712063, "decode_ms": 38.44337502960116}, "total_wall_ms": 23215.488498914056, "ttft_ms": 24.29266704712063, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 64, "facts_retained": 14, "retention_rate": 0.21875, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.2222222222222222, "amount": 0.15384615384615385, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 870}, "ledger_by_hop": {"1": {"facts_total": 64, "facts_retained": 14, "retention_rate": 0.21875, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.2222222222222222, "amount": 0.15384615384615385, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 870}}}, "reference_key_id": "keys/synthetic/syn_03018_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "b93ce255451151cce89ccd51cc89c5b4382f883b441b3818bc249f2f79c54d9e", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "09c22a18732b95c785fdde52", "attempt_id": "fa86ebfc5037", "run_group_id": "E4_core:A5|closed|qwen3b:h1", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:09:00.314274Z", "task_family": "synthetic", "document_id": "syn_03018_2000", "document_hash": "d284eb07e27da742b9c2b67abccfdf72e4dd67c3259f44f3b41e34dd3e6fc1ff", "question_id": "q05_multi_hop", "split": "test", "cluster_id": "syn_03018_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 1, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1982, "query_tokens": 493, "message_tokens": 401, "message_utf8_bytes": 940, "cache_tensor_bytes": null, "serialized_payload_bytes": 940, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 22800.499332952313, "hops": {"1": 22800.499332952313}, "receiver_prefill_ms": 294.978209072724, "ttft_ms": 39.705540984869, "decode_ms": 55.12120795901865}, "total_wall_ms": 23191.15354202222, "ttft_ms": 39.705540984869, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 64, "facts_retained": 14, "retention_rate": 0.21875, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.2222222222222222, "amount": 0.15384615384615385, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 870}, "ledger_by_hop": {"1": {"facts_total": 64, "facts_retained": 14, "retention_rate": 0.21875, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.2222222222222222, "amount": 0.15384615384615385, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 870}}}, "reference_key_id": "keys/synthetic/syn_03018_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "b93ce255451151cce89ccd51cc89c5b4382f883b441b3818bc249f2f79c54d9e", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "d7cde2a0c2212f5f3bd369f4", "attempt_id": "a9a66d7809ff", "run_group_id": "E4_core:A5|closed|qwen3b:h1", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:09:00.723589Z", "task_family": "synthetic", "document_id": "syn_03018_2000", "document_hash": "d284eb07e27da742b9c2b67abccfdf72e4dd67c3259f44f3b41e34dd3e6fc1ff", "question_id": "q06_negation", "split": "test", "cluster_id": "syn_03018_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 1, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1982, "query_tokens": 485, "message_tokens": 401, "message_utf8_bytes": 940, "cache_tensor_bytes": null, "serialized_payload_bytes": 940, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 22800.499332952313, "hops": {"1": 22800.499332952313}, "receiver_prefill_ms": 395.734415971674, "ttft_ms": 131.4412079518661, "decode_ms": 41.8310419190675}, "total_wall_ms": 23370.26779097505, "ttft_ms": 131.4412079518661, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 64, "facts_retained": 14, "retention_rate": 0.21875, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.2222222222222222, "amount": 0.15384615384615385, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 870}, "ledger_by_hop": {"1": {"facts_total": 64, "facts_retained": 14, "retention_rate": 0.21875, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.2222222222222222, "amount": 0.15384615384615385, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 870}}}, "reference_key_id": "keys/synthetic/syn_03018_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "b93ce255451151cce89ccd51cc89c5b4382f883b441b3818bc249f2f79c54d9e", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "ada6e7d30fa523213a6a6844", "attempt_id": "8afcd83db97c", "run_group_id": "E4_core:A5|closed|qwen3b:h1", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:09:01.311426Z", "task_family": "synthetic", "document_id": "syn_03018_2000", "document_hash": "d284eb07e27da742b9c2b67abccfdf72e4dd67c3259f44f3b41e34dd3e6fc1ff", "question_id": "q07_negation", "split": "test", "cluster_id": "syn_03018_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 1, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1982, "query_tokens": 484, "message_tokens": 401, "message_utf8_bytes": 940, "cache_tensor_bytes": null, "serialized_payload_bytes": 940, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 22800.499332952313, "hops": {"1": 22800.499332952313}, "receiver_prefill_ms": 326.69887493830174, "ttft_ms": 39.28145801182836, "decode_ms": 24.9515000032261}, "total_wall_ms": 23192.043749033473, "ttft_ms": 39.28145801182836, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 64, "facts_retained": 14, "retention_rate": 0.21875, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.2222222222222222, "amount": 0.15384615384615385, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 870}, "ledger_by_hop": {"1": {"facts_total": 64, "facts_retained": 14, "retention_rate": 0.21875, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.2222222222222222, "amount": 0.15384615384615385, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 870}}}, "reference_key_id": "keys/synthetic/syn_03018_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "b93ce255451151cce89ccd51cc89c5b4382f883b441b3818bc249f2f79c54d9e", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "436e6153e9197c594aec7bea", "attempt_id": "c0a9c211d100", "run_group_id": "E4_core:A5|closed|qwen3b:h1", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:09:01.720705Z", "task_family": "synthetic", "document_id": "syn_03018_2000", "document_hash": "d284eb07e27da742b9c2b67abccfdf72e4dd67c3259f44f3b41e34dd3e6fc1ff", "question_id": "q08_revision", "split": "test", "cluster_id": "syn_03018_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 1, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1982, "query_tokens": 502, "message_tokens": 401, "message_utf8_bytes": 940, "cache_tensor_bytes": null, "serialized_payload_bytes": 940, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 22800.499332952313, "hops": {"1": 22800.499332952313}, "receiver_prefill_ms": 320.55962504819036, "ttft_ms": 37.032917025499046, "decode_ms": 171.41379206441343}, "total_wall_ms": 23330.28845791705, "ttft_ms": 37.032917025499046, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 6}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "272184", "evidence_artifact": {"ledger": {"facts_total": 64, "facts_retained": 14, "retention_rate": 0.21875, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.2222222222222222, "amount": 0.15384615384615385, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 870}, "ledger_by_hop": {"1": {"facts_total": 64, "facts_retained": 14, "retention_rate": 0.21875, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.2222222222222222, "amount": 0.15384615384615385, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 870}}}, "reference_key_id": "keys/synthetic/syn_03018_2000.json", "task_scores": {"correct": 0.0, "abstained": 0.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "b93ce255451151cce89ccd51cc89c5b4382f883b441b3818bc249f2f79c54d9e", "output_hash": "3aadaddf09c30b8bfe5dc7e2aa6984fcf0df4d216145cf925760dd5a946d6e05"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "631e9a621926de57395cd89f", "attempt_id": "e6b14b103324", "run_group_id": "E4_core:A5|closed|qwen3b:h1", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:09:02.268331Z", "task_family": "synthetic", "document_id": "syn_03018_2000", "document_hash": "d284eb07e27da742b9c2b67abccfdf72e4dd67c3259f44f3b41e34dd3e6fc1ff", "question_id": "q09_revision", "split": "test", "cluster_id": "syn_03018_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 1, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1982, "query_tokens": 495, "message_tokens": 401, "message_utf8_bytes": 940, "cache_tensor_bytes": null, "serialized_payload_bytes": 940, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 22800.499332952313, "hops": {"1": 22800.499332952313}, "receiver_prefill_ms": 334.1989580076188, "ttft_ms": 56.39816599432379, "decode_ms": 382.1712089702487}, "total_wall_ms": 23573.919832939282, "ttft_ms": 56.39816599432379, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 6}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "272184", "evidence_artifact": {"ledger": {"facts_total": 64, "facts_retained": 14, "retention_rate": 0.21875, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.2222222222222222, "amount": 0.15384615384615385, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 870}, "ledger_by_hop": {"1": {"facts_total": 64, "facts_retained": 14, "retention_rate": 0.21875, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.2222222222222222, "amount": 0.15384615384615385, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 870}}}, "reference_key_id": "keys/synthetic/syn_03018_2000.json", "task_scores": {"correct": 0.0, "abstained": 0.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "b93ce255451151cce89ccd51cc89c5b4382f883b441b3818bc249f2f79c54d9e", "output_hash": "3aadaddf09c30b8bfe5dc7e2aa6984fcf0df4d216145cf925760dd5a946d6e05"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "806091131a6a9225596fed6a", "attempt_id": "c57117fa25a3", "run_group_id": "E4_core:A5|closed|qwen3b:h1", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:09:03.059176Z", "task_family": "synthetic", "document_id": "syn_03018_2000", "document_hash": "d284eb07e27da742b9c2b67abccfdf72e4dd67c3259f44f3b41e34dd3e6fc1ff", "question_id": "q10_temporal", "split": "test", "cluster_id": "syn_03018_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 1, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1982, "query_tokens": 503, "message_tokens": 401, "message_utf8_bytes": 940, "cache_tensor_bytes": null, "serialized_payload_bytes": 940, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 22800.499332952313, "hops": {"1": 22800.499332952313}, "receiver_prefill_ms": 336.5895409369841, "ttft_ms": 39.228167035616934, "decode_ms": 32.40995795931667}, "total_wall_ms": 23209.57487495616, "ttft_ms": 39.228167035616934, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 64, "facts_retained": 14, "retention_rate": 0.21875, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.2222222222222222, "amount": 0.15384615384615385, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 870}, "ledger_by_hop": {"1": {"facts_total": 64, "facts_retained": 14, "retention_rate": 0.21875, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.2222222222222222, "amount": 0.15384615384615385, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 870}}}, "reference_key_id": "keys/synthetic/syn_03018_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "b93ce255451151cce89ccd51cc89c5b4382f883b441b3818bc249f2f79c54d9e", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "0b5a054b50adc4c2a74932f9", "attempt_id": "e11da4d001f1", "run_group_id": "E4_core:A5|closed|qwen3b:h1", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:09:03.486753Z", "task_family": "synthetic", "document_id": "syn_03018_2000", "document_hash": "d284eb07e27da742b9c2b67abccfdf72e4dd67c3259f44f3b41e34dd3e6fc1ff", "question_id": "q11_aggregation", "split": "test", "cluster_id": "syn_03018_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 1, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1982, "query_tokens": 495, "message_tokens": 401, "message_utf8_bytes": 940, "cache_tensor_bytes": null, "serialized_payload_bytes": 940, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 22800.499332952313, "hops": {"1": 22800.499332952313}, "receiver_prefill_ms": 316.4806250715628, "ttft_ms": 22.041582968086004, "decode_ms": 383.86679196264595}, "total_wall_ms": 23523.473541019484, "ttft_ms": 22.041582968086004, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 7}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "3189244", "evidence_artifact": {"ledger": {"facts_total": 64, "facts_retained": 14, "retention_rate": 0.21875, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.2222222222222222, "amount": 0.15384615384615385, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 870}, "ledger_by_hop": {"1": {"facts_total": 64, "facts_retained": 14, "retention_rate": 0.21875, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.2222222222222222, "amount": 0.15384615384615385, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 870}}}, "reference_key_id": "keys/synthetic/syn_03018_2000.json", "task_scores": {"correct": 0.0, "abstained": 0.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "b93ce255451151cce89ccd51cc89c5b4382f883b441b3818bc249f2f79c54d9e", "output_hash": "953508a9732586a80de7650cf183d3ca74cd590103fa92d25c9b3d80da907f61"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "eadf31aab6239d8a116ab5fa", "attempt_id": "7054d142f82c", "run_group_id": "E4_core:A5|closed|qwen3b:h1", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:09:04.228102Z", "task_family": "synthetic", "document_id": "syn_03018_2000", "document_hash": "d284eb07e27da742b9c2b67abccfdf72e4dd67c3259f44f3b41e34dd3e6fc1ff", "question_id": "q12_unit_conversion", "split": "test", "cluster_id": "syn_03018_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 1, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1982, "query_tokens": 492, "message_tokens": 401, "message_utf8_bytes": 940, "cache_tensor_bytes": null, "serialized_payload_bytes": 940, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 22800.499332952313, "hops": {"1": 22800.499332952313}, "receiver_prefill_ms": 439.4641669932753, "ttft_ms": 24.840707890689373, "decode_ms": 37.64387499541044}, "total_wall_ms": 23303.030374925584, "ttft_ms": 24.840707890689373, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 64, "facts_retained": 14, "retention_rate": 0.21875, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.2222222222222222, "amount": 0.15384615384615385, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 870}, "ledger_by_hop": {"1": {"facts_total": 64, "facts_retained": 14, "retention_rate": 0.21875, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.2222222222222222, "amount": 0.15384615384615385, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 870}}}, "reference_key_id": "keys/synthetic/syn_03018_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "b93ce255451151cce89ccd51cc89c5b4382f883b441b3818bc249f2f79c54d9e", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "29883bae251be51f732ac325", "attempt_id": "16866cbfb317", "run_group_id": "E4_core:A5|closed|qwen3b:h1", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:09:04.747431Z", "task_family": "synthetic", "document_id": "syn_03018_2000", "document_hash": "d284eb07e27da742b9c2b67abccfdf72e4dd67c3259f44f3b41e34dd3e6fc1ff", "question_id": "q13_unanswerable", "split": "test", "cluster_id": "syn_03018_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 1, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1982, "query_tokens": 496, "message_tokens": 401, "message_utf8_bytes": 940, "cache_tensor_bytes": null, "serialized_payload_bytes": 940, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 22800.499332952313, "hops": {"1": 22800.499332952313}, "receiver_prefill_ms": 324.5807499624789, "ttft_ms": 27.998582925647497, "decode_ms": 42.54237504210323}, "total_wall_ms": 23196.128332987428, "ttft_ms": 27.998582925647497, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 64, "facts_retained": 14, "retention_rate": 0.21875, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.2222222222222222, "amount": 0.15384615384615385, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 870}, "ledger_by_hop": {"1": {"facts_total": 64, "facts_retained": 14, "retention_rate": 0.21875, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.2222222222222222, "amount": 0.15384615384615385, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 870}}}, "reference_key_id": "keys/synthetic/syn_03018_2000.json", "task_scores": {"correct": 1.0, "abstained": 1.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "b93ce255451151cce89ccd51cc89c5b4382f883b441b3818bc249f2f79c54d9e", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "238e9ccab25f5e6f837c957a", "attempt_id": "5616b12365e9", "run_group_id": "E4_core:A5|closed|qwen3b:h1", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:09:05.160403Z", "task_family": "synthetic", "document_id": "syn_03018_2000", "document_hash": "d284eb07e27da742b9c2b67abccfdf72e4dd67c3259f44f3b41e34dd3e6fc1ff", "question_id": "q14_unanswerable", "split": "test", "cluster_id": "syn_03018_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 1, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1982, "query_tokens": 497, "message_tokens": 401, "message_utf8_bytes": 940, "cache_tensor_bytes": null, "serialized_payload_bytes": 940, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 22800.499332952313, "hops": {"1": 22800.499332952313}, "receiver_prefill_ms": 299.31041691452265, "ttft_ms": 40.43816600460559, "decode_ms": 59.3377910554409}, "total_wall_ms": 23200.146207935177, "ttft_ms": 40.43816600460559, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 64, "facts_retained": 14, "retention_rate": 0.21875, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.2222222222222222, "amount": 0.15384615384615385, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 870}, "ledger_by_hop": {"1": {"facts_total": 64, "facts_retained": 14, "retention_rate": 0.21875, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.2222222222222222, "amount": 0.15384615384615385, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 870}}}, "reference_key_id": "keys/synthetic/syn_03018_2000.json", "task_scores": {"correct": 1.0, "abstained": 1.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "b93ce255451151cce89ccd51cc89c5b4382f883b441b3818bc249f2f79c54d9e", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "e8a91a16b30be6f3b2c92cf8", "attempt_id": "624f71075d05", "run_group_id": "E4_core:A5|closed|qwen3b:h2", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:09:23.761300Z", "task_family": "synthetic", "document_id": "syn_03018_2000", "document_hash": "d284eb07e27da742b9c2b67abccfdf72e4dd67c3259f44f3b41e34dd3e6fc1ff", "question_id": "q00_single_fact", "split": "test", "cluster_id": "syn_03018_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 2, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1982, "query_tokens": 484, "message_tokens": 401, "message_utf8_bytes": 940, "cache_tensor_bytes": null, "serialized_payload_bytes": 940, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 40983.81449887529, "hops": {"1": 22800.499332952313, "2": 18183.315165922977}, "receiver_prefill_ms": 313.64524993114173, "ttft_ms": 32.06395800225437, "decode_ms": 32.57733304053545}, "total_wall_ms": 41362.777831964195, "ttft_ms": 32.06395800225437, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 64, "facts_retained": 14, "retention_rate": 0.21875, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.2222222222222222, "amount": 0.15384615384615385, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 870}, "ledger_by_hop": {"1": {"facts_total": 64, "facts_retained": 14, "retention_rate": 0.21875, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.2222222222222222, "amount": 0.15384615384615385, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 870}, "2": {"facts_total": 64, "facts_retained": 14, "retention_rate": 0.21875, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.2222222222222222, "amount": 0.15384615384615385, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 870}}}, "reference_key_id": "keys/synthetic/syn_03018_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "b93ce255451151cce89ccd51cc89c5b4382f883b441b3818bc249f2f79c54d9e", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "ba8f490d63d0e5ce6897bde9", "attempt_id": "ba300c0ddbfc", "run_group_id": "E4_core:A5|closed|qwen3b:h2", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:09:24.158004Z", "task_family": "synthetic", "document_id": "syn_03018_2000", "document_hash": "d284eb07e27da742b9c2b67abccfdf72e4dd67c3259f44f3b41e34dd3e6fc1ff", "question_id": "q01_single_fact", "split": "test", "cluster_id": "syn_03018_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 2, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1982, "query_tokens": 490, "message_tokens": 401, "message_utf8_bytes": 940, "cache_tensor_bytes": null, "serialized_payload_bytes": 940, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 40983.81449887529, "hops": {"1": 22800.499332952313, "2": 18183.315165922977}, "receiver_prefill_ms": 318.0516669526696, "ttft_ms": 32.90550003293902, "decode_ms": 29.614083003252745}, "total_wall_ms": 41364.88112388179, "ttft_ms": 32.90550003293902, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 64, "facts_retained": 14, "retention_rate": 0.21875, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.2222222222222222, "amount": 0.15384615384615385, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 870}, "ledger_by_hop": {"1": {"facts_total": 64, "facts_retained": 14, "retention_rate": 0.21875, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.2222222222222222, "amount": 0.15384615384615385, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 870}, "2": {"facts_total": 64, "facts_retained": 14, "retention_rate": 0.21875, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.2222222222222222, "amount": 0.15384615384615385, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 870}}}, "reference_key_id": "keys/synthetic/syn_03018_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "b93ce255451151cce89ccd51cc89c5b4382f883b441b3818bc249f2f79c54d9e", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "27ccce6c7c4132e1b805e502", "attempt_id": "54fc067af396", "run_group_id": "E4_core:A5|closed|qwen3b:h2", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:09:24.556062Z", "task_family": "synthetic", "document_id": "syn_03018_2000", "document_hash": "d284eb07e27da742b9c2b67abccfdf72e4dd67c3259f44f3b41e34dd3e6fc1ff", "question_id": "q02_exact_copy", "split": "test", "cluster_id": "syn_03018_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 2, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1982, "query_tokens": 489, "message_tokens": 401, "message_utf8_bytes": 940, "cache_tensor_bytes": null, "serialized_payload_bytes": 940, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 40983.81449887529, "hops": {"1": 22800.499332952313, "2": 18183.315165922977}, "receiver_prefill_ms": 289.6773749962449, "ttft_ms": 51.55812494922429, "decode_ms": 54.75112493149936}, "total_wall_ms": 41380.65103976987, "ttft_ms": 51.55812494922429, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 64, "facts_retained": 14, "retention_rate": 0.21875, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.2222222222222222, "amount": 0.15384615384615385, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 870}, "ledger_by_hop": {"1": {"facts_total": 64, "facts_retained": 14, "retention_rate": 0.21875, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.2222222222222222, "amount": 0.15384615384615385, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 870}, "2": {"facts_total": 64, "facts_retained": 14, "retention_rate": 0.21875, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.2222222222222222, "amount": 0.15384615384615385, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 870}}}, "reference_key_id": "keys/synthetic/syn_03018_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "b93ce255451151cce89ccd51cc89c5b4382f883b441b3818bc249f2f79c54d9e", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "b1f8e0817da3a4858c517768", "attempt_id": "bac36b5571a2", "run_group_id": "E4_core:A5|closed|qwen3b:h2", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:09:24.970416Z", "task_family": "synthetic", "document_id": "syn_03018_2000", "document_hash": "d284eb07e27da742b9c2b67abccfdf72e4dd67c3259f44f3b41e34dd3e6fc1ff", "question_id": "q03_exact_copy", "split": "test", "cluster_id": "syn_03018_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 2, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1982, "query_tokens": 487, "message_tokens": 401, "message_utf8_bytes": 940, "cache_tensor_bytes": null, "serialized_payload_bytes": 940, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 40983.81449887529, "hops": {"1": 22800.499332952313, "2": 18183.315165922977}, "receiver_prefill_ms": 358.0720419995487, "ttft_ms": 55.71008299011737, "decode_ms": 366.4042910095304}, "total_wall_ms": 41764.57870786544, "ttft_ms": 55.71008299011737, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 9}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "Priyamvada Sørlie", "evidence_artifact": {"ledger": {"facts_total": 64, "facts_retained": 14, "retention_rate": 0.21875, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.2222222222222222, "amount": 0.15384615384615385, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 870}, "ledger_by_hop": {"1": {"facts_total": 64, "facts_retained": 14, "retention_rate": 0.21875, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.2222222222222222, "amount": 0.15384615384615385, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 870}, "2": {"facts_total": 64, "facts_retained": 14, "retention_rate": 0.21875, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.2222222222222222, "amount": 0.15384615384615385, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 870}}}, "reference_key_id": "keys/synthetic/syn_03018_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "b93ce255451151cce89ccd51cc89c5b4382f883b441b3818bc249f2f79c54d9e", "output_hash": "c5f554196ed23dd71ad995ec89f7142099d524dbb6534cc10d94a0c89d13e792"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "d542c457edf2cfe20df69fe6", "attempt_id": "d175e44e2eae", "run_group_id": "E4_core:A5|closed|qwen3b:h2", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:09:25.771044Z", "task_family": "synthetic", "document_id": "syn_03018_2000", "document_hash": "d284eb07e27da742b9c2b67abccfdf72e4dd67c3259f44f3b41e34dd3e6fc1ff", "question_id": "q04_multi_hop", "split": "test", "cluster_id": "syn_03018_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 2, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1982, "query_tokens": 493, "message_tokens": 401, "message_utf8_bytes": 940, "cache_tensor_bytes": null, "serialized_payload_bytes": 940, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 40983.81449887529, "hops": {"1": 22800.499332952313, "2": 18183.315165922977}, "receiver_prefill_ms": 327.57766696158797, "ttft_ms": 31.797208008356392, "decode_ms": 40.84562498610467}, "total_wall_ms": 41384.54524893314, "ttft_ms": 31.797208008356392, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 64, "facts_retained": 14, "retention_rate": 0.21875, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.2222222222222222, "amount": 0.15384615384615385, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 870}, "ledger_by_hop": {"1": {"facts_total": 64, "facts_retained": 14, "retention_rate": 0.21875, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.2222222222222222, "amount": 0.15384615384615385, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 870}, "2": {"facts_total": 64, "facts_retained": 14, "retention_rate": 0.21875, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.2222222222222222, "amount": 0.15384615384615385, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 870}}}, "reference_key_id": "keys/synthetic/syn_03018_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "b93ce255451151cce89ccd51cc89c5b4382f883b441b3818bc249f2f79c54d9e", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "982895db51fc9071272ad077", "attempt_id": "423c0a693eae", "run_group_id": "E4_core:A5|closed|qwen3b:h2", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:09:26.189082Z", "task_family": "synthetic", "document_id": "syn_03018_2000", "document_hash": "d284eb07e27da742b9c2b67abccfdf72e4dd67c3259f44f3b41e34dd3e6fc1ff", "question_id": "q05_multi_hop", "split": "test", "cluster_id": "syn_03018_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 2, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1982, "query_tokens": 493, "message_tokens": 401, "message_utf8_bytes": 940, "cache_tensor_bytes": null, "serialized_payload_bytes": 940, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 40983.81449887529, "hops": {"1": 22800.499332952313, "2": 18183.315165922977}, "receiver_prefill_ms": 277.7492090826854, "ttft_ms": 56.792750023305416, "decode_ms": 63.168999971821904}, "total_wall_ms": 41382.29928992223, "ttft_ms": 56.792750023305416, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 64, "facts_retained": 14, "retention_rate": 0.21875, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.2222222222222222, "amount": 0.15384615384615385, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 870}, "ledger_by_hop": {"1": {"facts_total": 64, "facts_retained": 14, "retention_rate": 0.21875, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.2222222222222222, "amount": 0.15384615384615385, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 870}, "2": {"facts_total": 64, "facts_retained": 14, "retention_rate": 0.21875, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.2222222222222222, "amount": 0.15384615384615385, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 870}}}, "reference_key_id": "keys/synthetic/syn_03018_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "b93ce255451151cce89ccd51cc89c5b4382f883b441b3818bc249f2f79c54d9e", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "8606a226350bacfcf05916cc", "attempt_id": "2c76b8c91430", "run_group_id": "E4_core:A5|closed|qwen3b:h2", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:09:26.605558Z", "task_family": "synthetic", "document_id": "syn_03018_2000", "document_hash": "d284eb07e27da742b9c2b67abccfdf72e4dd67c3259f44f3b41e34dd3e6fc1ff", "question_id": "q06_negation", "split": "test", "cluster_id": "syn_03018_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 2, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1982, "query_tokens": 485, "message_tokens": 401, "message_utf8_bytes": 940, "cache_tensor_bytes": null, "serialized_payload_bytes": 940, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 40983.81449887529, "hops": {"1": 22800.499332952313, "2": 18183.315165922977}, "receiver_prefill_ms": 396.2065000087023, "ttft_ms": 124.7173750307411, "decode_ms": 41.12116701435298}, "total_wall_ms": 41546.57704080455, "ttft_ms": 124.7173750307411, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 64, "facts_retained": 14, "retention_rate": 0.21875, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.2222222222222222, "amount": 0.15384615384615385, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 870}, "ledger_by_hop": {"1": {"facts_total": 64, "facts_retained": 14, "retention_rate": 0.21875, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.2222222222222222, "amount": 0.15384615384615385, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 870}, "2": {"facts_total": 64, "facts_retained": 14, "retention_rate": 0.21875, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.2222222222222222, "amount": 0.15384615384615385, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 870}}}, "reference_key_id": "keys/synthetic/syn_03018_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "b93ce255451151cce89ccd51cc89c5b4382f883b441b3818bc249f2f79c54d9e", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "b654a660d66a9af4be5e518d", "attempt_id": "8de4e33bb08d", "run_group_id": "E4_core:A5|closed|qwen3b:h2", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:09:27.185791Z", "task_family": "synthetic", "document_id": "syn_03018_2000", "document_hash": "d284eb07e27da742b9c2b67abccfdf72e4dd67c3259f44f3b41e34dd3e6fc1ff", "question_id": "q07_negation", "split": "test", "cluster_id": "syn_03018_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 2, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1982, "query_tokens": 484, "message_tokens": 401, "message_utf8_bytes": 940, "cache_tensor_bytes": null, "serialized_payload_bytes": 940, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 40983.81449887529, "hops": {"1": 22800.499332952313, "2": 18183.315165922977}, "receiver_prefill_ms": 326.04145794175565, "ttft_ms": 44.069958035834134, "decode_ms": 26.84562501963228}, "total_wall_ms": 41381.33820681833, "ttft_ms": 44.069958035834134, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 64, "facts_retained": 14, "retention_rate": 0.21875, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.2222222222222222, "amount": 0.15384615384615385, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 870}, "ledger_by_hop": {"1": {"facts_total": 64, "facts_retained": 14, "retention_rate": 0.21875, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.2222222222222222, "amount": 0.15384615384615385, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 870}, "2": {"facts_total": 64, "facts_retained": 14, "retention_rate": 0.21875, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.2222222222222222, "amount": 0.15384615384615385, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 870}}}, "reference_key_id": "keys/synthetic/syn_03018_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "b93ce255451151cce89ccd51cc89c5b4382f883b441b3818bc249f2f79c54d9e", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "05bb4007fbbe81bd88d97a73", "attempt_id": "5f0f80ca5c90", "run_group_id": "E4_core:A5|closed|qwen3b:h2", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:09:27.600164Z", "task_family": "synthetic", "document_id": "syn_03018_2000", "document_hash": "d284eb07e27da742b9c2b67abccfdf72e4dd67c3259f44f3b41e34dd3e6fc1ff", "question_id": "q08_revision", "split": "test", "cluster_id": "syn_03018_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 2, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1982, "query_tokens": 502, "message_tokens": 401, "message_utf8_bytes": 940, "cache_tensor_bytes": null, "serialized_payload_bytes": 940, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 40983.81449887529, "hops": {"1": 22800.499332952313, "2": 18183.315165922977}, "receiver_prefill_ms": 306.7969169933349, "ttft_ms": 21.789041929878294, "decode_ms": 173.6887500155717}, "total_wall_ms": 41486.77912377752, "ttft_ms": 21.789041929878294, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 6}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "272184", "evidence_artifact": {"ledger": {"facts_total": 64, "facts_retained": 14, "retention_rate": 0.21875, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.2222222222222222, "amount": 0.15384615384615385, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 870}, "ledger_by_hop": {"1": {"facts_total": 64, "facts_retained": 14, "retention_rate": 0.21875, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.2222222222222222, "amount": 0.15384615384615385, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 870}, "2": {"facts_total": 64, "facts_retained": 14, "retention_rate": 0.21875, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.2222222222222222, "amount": 0.15384615384615385, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 870}}}, "reference_key_id": "keys/synthetic/syn_03018_2000.json", "task_scores": {"correct": 0.0, "abstained": 0.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "b93ce255451151cce89ccd51cc89c5b4382f883b441b3818bc249f2f79c54d9e", "output_hash": "3aadaddf09c30b8bfe5dc7e2aa6984fcf0df4d216145cf925760dd5a946d6e05"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "02065b10062fe0e05e0f0676", "attempt_id": "cdd764eb5e9b", "run_group_id": "E4_core:A5|closed|qwen3b:h2", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:09:28.120731Z", "task_family": "synthetic", "document_id": "syn_03018_2000", "document_hash": "d284eb07e27da742b9c2b67abccfdf72e4dd67c3259f44f3b41e34dd3e6fc1ff", "question_id": "q09_revision", "split": "test", "cluster_id": "syn_03018_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 2, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1982, "query_tokens": 495, "message_tokens": 401, "message_utf8_bytes": 940, "cache_tensor_bytes": null, "serialized_payload_bytes": 940, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 40983.81449887529, "hops": {"1": 22800.499332952313, "2": 18183.315165922977}, "receiver_prefill_ms": 300.1470830058679, "ttft_ms": 54.03954198118299, "decode_ms": 412.3252499848604}, "total_wall_ms": 41750.850123818964, "ttft_ms": 54.03954198118299, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 6}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "272184", "evidence_artifact": {"ledger": {"facts_total": 64, "facts_retained": 14, "retention_rate": 0.21875, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.2222222222222222, "amount": 0.15384615384615385, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 870}, "ledger_by_hop": {"1": {"facts_total": 64, "facts_retained": 14, "retention_rate": 0.21875, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.2222222222222222, "amount": 0.15384615384615385, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 870}, "2": {"facts_total": 64, "facts_retained": 14, "retention_rate": 0.21875, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.2222222222222222, "amount": 0.15384615384615385, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 870}}}, "reference_key_id": "keys/synthetic/syn_03018_2000.json", "task_scores": {"correct": 0.0, "abstained": 0.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "b93ce255451151cce89ccd51cc89c5b4382f883b441b3818bc249f2f79c54d9e", "output_hash": "3aadaddf09c30b8bfe5dc7e2aa6984fcf0df4d216145cf925760dd5a946d6e05"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "75514174cfca2887a39c42ed", "attempt_id": "fb7c84b82f0e", "run_group_id": "E4_core:A5|closed|qwen3b:h2", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:09:28.904792Z", "task_family": "synthetic", "document_id": "syn_03018_2000", "document_hash": "d284eb07e27da742b9c2b67abccfdf72e4dd67c3259f44f3b41e34dd3e6fc1ff", "question_id": "q10_temporal", "split": "test", "cluster_id": "syn_03018_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 2, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1982, "query_tokens": 503, "message_tokens": 401, "message_utf8_bytes": 940, "cache_tensor_bytes": null, "serialized_payload_bytes": 940, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 40983.81449887529, "hops": {"1": 22800.499332952313, "2": 18183.315165922977}, "receiver_prefill_ms": 325.05020895041525, "ttft_ms": 38.85020792949945, "decode_ms": 31.92075004335493}, "total_wall_ms": 41380.09270676412, "ttft_ms": 38.85020792949945, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 64, "facts_retained": 14, "retention_rate": 0.21875, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.2222222222222222, "amount": 0.15384615384615385, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 870}, "ledger_by_hop": {"1": {"facts_total": 64, "facts_retained": 14, "retention_rate": 0.21875, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.2222222222222222, "amount": 0.15384615384615385, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 870}, "2": {"facts_total": 64, "facts_retained": 14, "retention_rate": 0.21875, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.2222222222222222, "amount": 0.15384615384615385, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 870}}}, "reference_key_id": "keys/synthetic/syn_03018_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "b93ce255451151cce89ccd51cc89c5b4382f883b441b3818bc249f2f79c54d9e", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "7d4fbc7199f247d78bdd6b1e", "attempt_id": "070eab8ed60d", "run_group_id": "E4_core:A5|closed|qwen3b:h2", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:09:29.318486Z", "task_family": "synthetic", "document_id": "syn_03018_2000", "document_hash": "d284eb07e27da742b9c2b67abccfdf72e4dd67c3259f44f3b41e34dd3e6fc1ff", "question_id": "q11_aggregation", "split": "test", "cluster_id": "syn_03018_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 2, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1982, "query_tokens": 495, "message_tokens": 401, "message_utf8_bytes": 940, "cache_tensor_bytes": null, "serialized_payload_bytes": 940, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 40983.81449887529, "hops": {"1": 22800.499332952313, "2": 18183.315165922977}, "receiver_prefill_ms": 318.9265000401065, "ttft_ms": 40.788665995933115, "decode_ms": 333.5483750561252}, "total_wall_ms": 41677.684706868604, "ttft_ms": 40.788665995933115, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 7}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "3189244", "evidence_artifact": {"ledger": {"facts_total": 64, "facts_retained": 14, "retention_rate": 0.21875, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.2222222222222222, "amount": 0.15384615384615385, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 870}, "ledger_by_hop": {"1": {"facts_total": 64, "facts_retained": 14, "retention_rate": 0.21875, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.2222222222222222, "amount": 0.15384615384615385, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 870}, "2": {"facts_total": 64, "facts_retained": 14, "retention_rate": 0.21875, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.2222222222222222, "amount": 0.15384615384615385, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 870}}}, "reference_key_id": "keys/synthetic/syn_03018_2000.json", "task_scores": {"correct": 0.0, "abstained": 0.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "b93ce255451151cce89ccd51cc89c5b4382f883b441b3818bc249f2f79c54d9e", "output_hash": "953508a9732586a80de7650cf183d3ca74cd590103fa92d25c9b3d80da907f61"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "d6514b898c00159b9977687b", "attempt_id": "7291bbe16447", "run_group_id": "E4_core:A5|closed|qwen3b:h2", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:09:30.029052Z", "task_family": "synthetic", "document_id": "syn_03018_2000", "document_hash": "d284eb07e27da742b9c2b67abccfdf72e4dd67c3259f44f3b41e34dd3e6fc1ff", "question_id": "q12_unit_conversion", "split": "test", "cluster_id": "syn_03018_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 2, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1982, "query_tokens": 492, "message_tokens": 401, "message_utf8_bytes": 940, "cache_tensor_bytes": null, "serialized_payload_bytes": 940, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 40983.81449887529, "hops": {"1": 22800.499332952313, "2": 18183.315165922977}, "receiver_prefill_ms": 479.44883292075247, "ttft_ms": 40.893458062782884, "decode_ms": 28.305916930548847}, "total_wall_ms": 41533.14020787366, "ttft_ms": 40.893458062782884, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 64, "facts_retained": 14, "retention_rate": 0.21875, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.2222222222222222, "amount": 0.15384615384615385, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 870}, "ledger_by_hop": {"1": {"facts_total": 64, "facts_retained": 14, "retention_rate": 0.21875, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.2222222222222222, "amount": 0.15384615384615385, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 870}, "2": {"facts_total": 64, "facts_retained": 14, "retention_rate": 0.21875, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.2222222222222222, "amount": 0.15384615384615385, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 870}}}, "reference_key_id": "keys/synthetic/syn_03018_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "b93ce255451151cce89ccd51cc89c5b4382f883b441b3818bc249f2f79c54d9e", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "ac4e4c72f78df82984c559f6", "attempt_id": "5491fd57dd83", "run_group_id": "E4_core:A5|closed|qwen3b:h2", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:09:30.596334Z", "task_family": "synthetic", "document_id": "syn_03018_2000", "document_hash": "d284eb07e27da742b9c2b67abccfdf72e4dd67c3259f44f3b41e34dd3e6fc1ff", "question_id": "q13_unanswerable", "split": "test", "cluster_id": "syn_03018_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 2, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1982, "query_tokens": 496, "message_tokens": 401, "message_utf8_bytes": 940, "cache_tensor_bytes": null, "serialized_payload_bytes": 940, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 40983.81449887529, "hops": {"1": 22800.499332952313, "2": 18183.315165922977}, "receiver_prefill_ms": 318.82737507112324, "ttft_ms": 39.01904192753136, "decode_ms": 27.190541964955628}, "total_wall_ms": 41369.36195788439, "ttft_ms": 39.01904192753136, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 64, "facts_retained": 14, "retention_rate": 0.21875, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.2222222222222222, "amount": 0.15384615384615385, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 870}, "ledger_by_hop": {"1": {"facts_total": 64, "facts_retained": 14, "retention_rate": 0.21875, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.2222222222222222, "amount": 0.15384615384615385, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 870}, "2": {"facts_total": 64, "facts_retained": 14, "retention_rate": 0.21875, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.2222222222222222, "amount": 0.15384615384615385, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 870}}}, "reference_key_id": "keys/synthetic/syn_03018_2000.json", "task_scores": {"correct": 1.0, "abstained": 1.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "b93ce255451151cce89ccd51cc89c5b4382f883b441b3818bc249f2f79c54d9e", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "7bccc519dd0743f819a815fe", "attempt_id": "37850fad4395", "run_group_id": "E4_core:A5|closed|qwen3b:h2", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:09:30.998749Z", "task_family": "synthetic", "document_id": "syn_03018_2000", "document_hash": "d284eb07e27da742b9c2b67abccfdf72e4dd67c3259f44f3b41e34dd3e6fc1ff", "question_id": "q14_unanswerable", "split": "test", "cluster_id": "syn_03018_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 2, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1982, "query_tokens": 497, "message_tokens": 401, "message_utf8_bytes": 940, "cache_tensor_bytes": null, "serialized_payload_bytes": 940, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 40983.81449887529, "hops": {"1": 22800.499332952313, "2": 18183.315165922977}, "receiver_prefill_ms": 314.19362500309944, "ttft_ms": 20.642041927203536, "decode_ms": 20.32466698437929}, "total_wall_ms": 41339.49474885594, "ttft_ms": 20.642041927203536, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 64, "facts_retained": 14, "retention_rate": 0.21875, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.2222222222222222, "amount": 0.15384615384615385, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 870}, "ledger_by_hop": {"1": {"facts_total": 64, "facts_retained": 14, "retention_rate": 0.21875, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.2222222222222222, "amount": 0.15384615384615385, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 870}, "2": {"facts_total": 64, "facts_retained": 14, "retention_rate": 0.21875, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.2222222222222222, "amount": 0.15384615384615385, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 870}}}, "reference_key_id": "keys/synthetic/syn_03018_2000.json", "task_scores": {"correct": 1.0, "abstained": 1.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "b93ce255451151cce89ccd51cc89c5b4382f883b441b3818bc249f2f79c54d9e", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "f3c03f5b8ae38f299713a321", "attempt_id": "81516d436aa2", "run_group_id": "E4_core:A5|closed|qwen3b:h4", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:10:07.169473Z", "task_family": "synthetic", "document_id": "syn_03018_2000", "document_hash": "d284eb07e27da742b9c2b67abccfdf72e4dd67c3259f44f3b41e34dd3e6fc1ff", "question_id": "q00_single_fact", "split": "test", "cluster_id": "syn_03018_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 4, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1982, "query_tokens": 484, "message_tokens": 401, "message_utf8_bytes": 940, "cache_tensor_bytes": null, "serialized_payload_bytes": 940, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 76780.02479101997, "hops": {"1": 22800.499332952313, "2": 18183.315165922977, "3": 17950.87779208552, "4": 17845.332500059158}, "receiver_prefill_ms": 438.99479205720127, "ttft_ms": 37.26162505336106, "decode_ms": 25.184000027365983}, "total_wall_ms": 77282.11166604888, "ttft_ms": 37.26162505336106, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 64, "facts_retained": 14, "retention_rate": 0.21875, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.2222222222222222, "amount": 0.15384615384615385, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 870}, "ledger_by_hop": {"1": {"facts_total": 64, "facts_retained": 14, "retention_rate": 0.21875, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.2222222222222222, "amount": 0.15384615384615385, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 870}, "2": {"facts_total": 64, "facts_retained": 14, "retention_rate": 0.21875, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.2222222222222222, "amount": 0.15384615384615385, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 870}, "3": {"facts_total": 64, "facts_retained": 14, "retention_rate": 0.21875, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.2222222222222222, "amount": 0.15384615384615385, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 870}, "4": {"facts_total": 64, "facts_retained": 14, "retention_rate": 0.21875, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.2222222222222222, "amount": 0.15384615384615385, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 870}}}, "reference_key_id": "keys/synthetic/syn_03018_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "b93ce255451151cce89ccd51cc89c5b4382f883b441b3818bc249f2f79c54d9e", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "6e168563de97f3bf09e84a24", "attempt_id": "ce766d64ef12", "run_group_id": "E4_core:A5|closed|qwen3b:h4", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:10:07.691151Z", "task_family": "synthetic", "document_id": "syn_03018_2000", "document_hash": "d284eb07e27da742b9c2b67abccfdf72e4dd67c3259f44f3b41e34dd3e6fc1ff", "question_id": "q01_single_fact", "split": "test", "cluster_id": "syn_03018_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 4, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1982, "query_tokens": 490, "message_tokens": 401, "message_utf8_bytes": 940, "cache_tensor_bytes": null, "serialized_payload_bytes": 940, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 76780.02479101997, "hops": {"1": 22800.499332952313, "2": 18183.315165922977, "3": 17950.87779208552, "4": 17845.332500059158}, "receiver_prefill_ms": 317.71745800506324, "ttft_ms": 39.38099998049438, "decode_ms": 26.372459018602967}, "total_wall_ms": 77164.14416593034, "ttft_ms": 39.38099998049438, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 64, "facts_retained": 14, "retention_rate": 0.21875, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.2222222222222222, "amount": 0.15384615384615385, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 870}, "ledger_by_hop": {"1": {"facts_total": 64, "facts_retained": 14, "retention_rate": 0.21875, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.2222222222222222, "amount": 0.15384615384615385, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 870}, "2": {"facts_total": 64, "facts_retained": 14, "retention_rate": 0.21875, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.2222222222222222, "amount": 0.15384615384615385, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 870}, "3": {"facts_total": 64, "facts_retained": 14, "retention_rate": 0.21875, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.2222222222222222, "amount": 0.15384615384615385, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 870}, "4": {"facts_total": 64, "facts_retained": 14, "retention_rate": 0.21875, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.2222222222222222, "amount": 0.15384615384615385, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 870}}}, "reference_key_id": "keys/synthetic/syn_03018_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "b93ce255451151cce89ccd51cc89c5b4382f883b441b3818bc249f2f79c54d9e", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "dbc25b62fd0f5d8774793d0f", "attempt_id": "5741b13b6ba6", "run_group_id": "E4_core:A5|closed|qwen3b:h4", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:10:08.092847Z", "task_family": "synthetic", "document_id": "syn_03018_2000", "document_hash": "d284eb07e27da742b9c2b67abccfdf72e4dd67c3259f44f3b41e34dd3e6fc1ff", "question_id": "q02_exact_copy", "split": "test", "cluster_id": "syn_03018_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 4, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1982, "query_tokens": 489, "message_tokens": 401, "message_utf8_bytes": 940, "cache_tensor_bytes": null, "serialized_payload_bytes": 940, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 76780.02479101997, "hops": {"1": 22800.499332952313, "2": 18183.315165922977, "3": 17950.87779208552, "4": 17845.332500059158}, "receiver_prefill_ms": 319.9302910361439, "ttft_ms": 22.661875002086163, "decode_ms": 20.445250091142952}, "total_wall_ms": 77143.69583211374, "ttft_ms": 22.661875002086163, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 64, "facts_retained": 14, "retention_rate": 0.21875, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.2222222222222222, "amount": 0.15384615384615385, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 870}, "ledger_by_hop": {"1": {"facts_total": 64, "facts_retained": 14, "retention_rate": 0.21875, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.2222222222222222, "amount": 0.15384615384615385, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 870}, "2": {"facts_total": 64, "facts_retained": 14, "retention_rate": 0.21875, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.2222222222222222, "amount": 0.15384615384615385, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 870}, "3": {"facts_total": 64, "facts_retained": 14, "retention_rate": 0.21875, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.2222222222222222, "amount": 0.15384615384615385, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 870}, "4": {"facts_total": 64, "facts_retained": 14, "retention_rate": 0.21875, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.2222222222222222, "amount": 0.15384615384615385, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 870}}}, "reference_key_id": "keys/synthetic/syn_03018_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "b93ce255451151cce89ccd51cc89c5b4382f883b441b3818bc249f2f79c54d9e", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "dc7dc650ed4a6a7d03c67715", "attempt_id": "16f16b164ffd", "run_group_id": "E4_core:A5|closed|qwen3b:h4", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:10:08.474664Z", "task_family": "synthetic", "document_id": "syn_03018_2000", "document_hash": "d284eb07e27da742b9c2b67abccfdf72e4dd67c3259f44f3b41e34dd3e6fc1ff", "question_id": "q03_exact_copy", "split": "test", "cluster_id": "syn_03018_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 4, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1982, "query_tokens": 487, "message_tokens": 401, "message_utf8_bytes": 940, "cache_tensor_bytes": null, "serialized_payload_bytes": 940, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 76780.02479101997, "hops": {"1": 22800.499332952313, "2": 18183.315165922977, "3": 17950.87779208552, "4": 17845.332500059158}, "receiver_prefill_ms": 315.56254101451486, "ttft_ms": 48.276042100042105, "decode_ms": 478.6613329779357}, "total_wall_ms": 77623.15750005655, "ttft_ms": 48.276042100042105, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 9}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "Priyamvada Sørlie", "evidence_artifact": {"ledger": {"facts_total": 64, "facts_retained": 14, "retention_rate": 0.21875, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.2222222222222222, "amount": 0.15384615384615385, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 870}, "ledger_by_hop": {"1": {"facts_total": 64, "facts_retained": 14, "retention_rate": 0.21875, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.2222222222222222, "amount": 0.15384615384615385, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 870}, "2": {"facts_total": 64, "facts_retained": 14, "retention_rate": 0.21875, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.2222222222222222, "amount": 0.15384615384615385, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 870}, "3": {"facts_total": 64, "facts_retained": 14, "retention_rate": 0.21875, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.2222222222222222, "amount": 0.15384615384615385, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 870}, "4": {"facts_total": 64, "facts_retained": 14, "retention_rate": 0.21875, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.2222222222222222, "amount": 0.15384615384615385, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 870}}}, "reference_key_id": "keys/synthetic/syn_03018_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "b93ce255451151cce89ccd51cc89c5b4382f883b441b3818bc249f2f79c54d9e", "output_hash": "c5f554196ed23dd71ad995ec89f7142099d524dbb6534cc10d94a0c89d13e792"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "7261b4fb63588361c339f97d", "attempt_id": "7c975a947efd", "run_group_id": "E4_core:A5|closed|qwen3b:h4", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:10:09.336278Z", "task_family": "synthetic", "document_id": "syn_03018_2000", "document_hash": "d284eb07e27da742b9c2b67abccfdf72e4dd67c3259f44f3b41e34dd3e6fc1ff", "question_id": "q04_multi_hop", "split": "test", "cluster_id": "syn_03018_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 4, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1982, "query_tokens": 493, "message_tokens": 401, "message_utf8_bytes": 940, "cache_tensor_bytes": null, "serialized_payload_bytes": 940, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 76780.02479101997, "hops": {"1": 22800.499332952313, "2": 18183.315165922977, "3": 17950.87779208552, "4": 17845.332500059158}, "receiver_prefill_ms": 318.38370801415294, "ttft_ms": 32.79850003309548, "decode_ms": 29.060416971333325}, "total_wall_ms": 77160.75433196966, "ttft_ms": 32.79850003309548, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 64, "facts_retained": 14, "retention_rate": 0.21875, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.2222222222222222, "amount": 0.15384615384615385, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 870}, "ledger_by_hop": {"1": {"facts_total": 64, "facts_retained": 14, "retention_rate": 0.21875, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.2222222222222222, "amount": 0.15384615384615385, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 870}, "2": {"facts_total": 64, "facts_retained": 14, "retention_rate": 0.21875, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.2222222222222222, "amount": 0.15384615384615385, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 870}, "3": {"facts_total": 64, "facts_retained": 14, "retention_rate": 0.21875, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.2222222222222222, "amount": 0.15384615384615385, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 870}, "4": {"facts_total": 64, "facts_retained": 14, "retention_rate": 0.21875, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.2222222222222222, "amount": 0.15384615384615385, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 870}}}, "reference_key_id": "keys/synthetic/syn_03018_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "b93ce255451151cce89ccd51cc89c5b4382f883b441b3818bc249f2f79c54d9e", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "cd1417f11445847dbec93011", "attempt_id": "c989650e20a9", "run_group_id": "E4_core:A5|closed|qwen3b:h4", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:10:09.733929Z", "task_family": "synthetic", "document_id": "syn_03018_2000", "document_hash": "d284eb07e27da742b9c2b67abccfdf72e4dd67c3259f44f3b41e34dd3e6fc1ff", "question_id": "q05_multi_hop", "split": "test", "cluster_id": "syn_03018_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 4, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1982, "query_tokens": 493, "message_tokens": 401, "message_utf8_bytes": 940, "cache_tensor_bytes": null, "serialized_payload_bytes": 940, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 76780.02479101997, "hops": {"1": 22800.499332952313, "2": 18183.315165922977, "3": 17950.87779208552, "4": 17845.332500059158}, "receiver_prefill_ms": 328.1277499627322, "ttft_ms": 21.898709004744887, "decode_ms": 25.973041891120374}, "total_wall_ms": 77156.60320804454, "ttft_ms": 21.898709004744887, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 64, "facts_retained": 14, "retention_rate": 0.21875, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.2222222222222222, "amount": 0.15384615384615385, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 870}, "ledger_by_hop": {"1": {"facts_total": 64, "facts_retained": 14, "retention_rate": 0.21875, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.2222222222222222, "amount": 0.15384615384615385, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 870}, "2": {"facts_total": 64, "facts_retained": 14, "retention_rate": 0.21875, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.2222222222222222, "amount": 0.15384615384615385, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 870}, "3": {"facts_total": 64, "facts_retained": 14, "retention_rate": 0.21875, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.2222222222222222, "amount": 0.15384615384615385, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 870}, "4": {"facts_total": 64, "facts_retained": 14, "retention_rate": 0.21875, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.2222222222222222, "amount": 0.15384615384615385, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 870}}}, "reference_key_id": "keys/synthetic/syn_03018_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "b93ce255451151cce89ccd51cc89c5b4382f883b441b3818bc249f2f79c54d9e", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "a7433648c814b512463cab4d", "attempt_id": "7b47f2286a7e", "run_group_id": "E4_core:A5|closed|qwen3b:h4", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:10:10.128307Z", "task_family": "synthetic", "document_id": "syn_03018_2000", "document_hash": "d284eb07e27da742b9c2b67abccfdf72e4dd67c3259f44f3b41e34dd3e6fc1ff", "question_id": "q06_negation", "split": "test", "cluster_id": "syn_03018_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 4, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1982, "query_tokens": 485, "message_tokens": 401, "message_utf8_bytes": 940, "cache_tensor_bytes": null, "serialized_payload_bytes": 940, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 76780.02479101997, "hops": {"1": 22800.499332952313, "2": 18183.315165922977, "3": 17950.87779208552, "4": 17845.332500059158}, "receiver_prefill_ms": 305.4334579501301, "ttft_ms": 54.11949998233467, "decode_ms": 56.95687502156943}, "total_wall_ms": 77197.24412506912, "ttft_ms": 54.11949998233467, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 64, "facts_retained": 14, "retention_rate": 0.21875, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.2222222222222222, "amount": 0.15384615384615385, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 870}, "ledger_by_hop": {"1": {"facts_total": 64, "facts_retained": 14, "retention_rate": 0.21875, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.2222222222222222, "amount": 0.15384615384615385, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 870}, "2": {"facts_total": 64, "facts_retained": 14, "retention_rate": 0.21875, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.2222222222222222, "amount": 0.15384615384615385, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 870}, "3": {"facts_total": 64, "facts_retained": 14, "retention_rate": 0.21875, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.2222222222222222, "amount": 0.15384615384615385, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 870}, "4": {"facts_total": 64, "facts_retained": 14, "retention_rate": 0.21875, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.2222222222222222, "amount": 0.15384615384615385, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 870}}}, "reference_key_id": "keys/synthetic/syn_03018_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "b93ce255451151cce89ccd51cc89c5b4382f883b441b3818bc249f2f79c54d9e", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "0216223e65fbffd9d6d7a701", "attempt_id": "b2e9535c5a49", "run_group_id": "E4_core:A5|closed|qwen3b:h4", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:10:10.563182Z", "task_family": "synthetic", "document_id": "syn_03018_2000", "document_hash": "d284eb07e27da742b9c2b67abccfdf72e4dd67c3259f44f3b41e34dd3e6fc1ff", "question_id": "q07_negation", "split": "test", "cluster_id": "syn_03018_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 4, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1982, "query_tokens": 484, "message_tokens": 401, "message_utf8_bytes": 940, "cache_tensor_bytes": null, "serialized_payload_bytes": 940, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 76780.02479101997, "hops": {"1": 22800.499332952313, "2": 18183.315165922977, "3": 17950.87779208552, "4": 17845.332500059158}, "receiver_prefill_ms": 446.47249998524785, "ttft_ms": 37.48295898549259, "decode_ms": 42.73295903112739}, "total_wall_ms": 77307.39362398162, "ttft_ms": 37.48295898549259, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 64, "facts_retained": 14, "retention_rate": 0.21875, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.2222222222222222, "amount": 0.15384615384615385, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 870}, "ledger_by_hop": {"1": {"facts_total": 64, "facts_retained": 14, "retention_rate": 0.21875, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.2222222222222222, "amount": 0.15384615384615385, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 870}, "2": {"facts_total": 64, "facts_retained": 14, "retention_rate": 0.21875, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.2222222222222222, "amount": 0.15384615384615385, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 870}, "3": {"facts_total": 64, "facts_retained": 14, "retention_rate": 0.21875, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.2222222222222222, "amount": 0.15384615384615385, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 870}, "4": {"facts_total": 64, "facts_retained": 14, "retention_rate": 0.21875, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.2222222222222222, "amount": 0.15384615384615385, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 870}}}, "reference_key_id": "keys/synthetic/syn_03018_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "b93ce255451151cce89ccd51cc89c5b4382f883b441b3818bc249f2f79c54d9e", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "601f75d1e0d024fbb007813f", "attempt_id": "62f9dc401334", "run_group_id": "E4_core:A5|closed|qwen3b:h4", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:10:11.110314Z", "task_family": "synthetic", "document_id": "syn_03018_2000", "document_hash": "d284eb07e27da742b9c2b67abccfdf72e4dd67c3259f44f3b41e34dd3e6fc1ff", "question_id": "q08_revision", "split": "test", "cluster_id": "syn_03018_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 4, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1982, "query_tokens": 502, "message_tokens": 401, "message_utf8_bytes": 940, "cache_tensor_bytes": null, "serialized_payload_bytes": 940, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 76780.02479101997, "hops": {"1": 22800.499332952313, "2": 18183.315165922977, "3": 17950.87779208552, "4": 17845.332500059158}, "receiver_prefill_ms": 320.580166997388, "ttft_ms": 31.48241702001542, "decode_ms": 226.38391703367233}, "total_wall_ms": 77359.08266610932, "ttft_ms": 31.48241702001542, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 6}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "272184", "evidence_artifact": {"ledger": {"facts_total": 64, "facts_retained": 14, "retention_rate": 0.21875, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.2222222222222222, "amount": 0.15384615384615385, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 870}, "ledger_by_hop": {"1": {"facts_total": 64, "facts_retained": 14, "retention_rate": 0.21875, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.2222222222222222, "amount": 0.15384615384615385, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 870}, "2": {"facts_total": 64, "facts_retained": 14, "retention_rate": 0.21875, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.2222222222222222, "amount": 0.15384615384615385, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 870}, "3": {"facts_total": 64, "facts_retained": 14, "retention_rate": 0.21875, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.2222222222222222, "amount": 0.15384615384615385, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 870}, "4": {"facts_total": 64, "facts_retained": 14, "retention_rate": 0.21875, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.2222222222222222, "amount": 0.15384615384615385, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 870}}}, "reference_key_id": "keys/synthetic/syn_03018_2000.json", "task_scores": {"correct": 0.0, "abstained": 0.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "b93ce255451151cce89ccd51cc89c5b4382f883b441b3818bc249f2f79c54d9e", "output_hash": "3aadaddf09c30b8bfe5dc7e2aa6984fcf0df4d216145cf925760dd5a946d6e05"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "26f07a4376e85309bed44430", "attempt_id": "b76c2d0f6948", "run_group_id": "E4_core:A5|closed|qwen3b:h4", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:10:11.707071Z", "task_family": "synthetic", "document_id": "syn_03018_2000", "document_hash": "d284eb07e27da742b9c2b67abccfdf72e4dd67c3259f44f3b41e34dd3e6fc1ff", "question_id": "q09_revision", "split": "test", "cluster_id": "syn_03018_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 4, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1982, "query_tokens": 495, "message_tokens": 401, "message_utf8_bytes": 940, "cache_tensor_bytes": null, "serialized_payload_bytes": 940, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 76780.02479101997, "hops": {"1": 22800.499332952313, "2": 18183.315165922977, "3": 17950.87779208552, "4": 17845.332500059158}, "receiver_prefill_ms": 319.1392499720678, "ttft_ms": 50.28854194097221, "decode_ms": 404.46599991992116}, "total_wall_ms": 77554.71379100345, "ttft_ms": 50.28854194097221, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 6}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "272184", "evidence_artifact": {"ledger": {"facts_total": 64, "facts_retained": 14, "retention_rate": 0.21875, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.2222222222222222, "amount": 0.15384615384615385, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 870}, "ledger_by_hop": {"1": {"facts_total": 64, "facts_retained": 14, "retention_rate": 0.21875, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.2222222222222222, "amount": 0.15384615384615385, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 870}, "2": {"facts_total": 64, "facts_retained": 14, "retention_rate": 0.21875, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.2222222222222222, "amount": 0.15384615384615385, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 870}, "3": {"facts_total": 64, "facts_retained": 14, "retention_rate": 0.21875, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.2222222222222222, "amount": 0.15384615384615385, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 870}, "4": {"facts_total": 64, "facts_retained": 14, "retention_rate": 0.21875, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.2222222222222222, "amount": 0.15384615384615385, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 870}}}, "reference_key_id": "keys/synthetic/syn_03018_2000.json", "task_scores": {"correct": 0.0, "abstained": 0.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "b93ce255451151cce89ccd51cc89c5b4382f883b441b3818bc249f2f79c54d9e", "output_hash": "3aadaddf09c30b8bfe5dc7e2aa6984fcf0df4d216145cf925760dd5a946d6e05"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "a90d8904f3e4514c9c6104a0", "attempt_id": "3172f8e7680d", "run_group_id": "E4_core:A5|closed|qwen3b:h4", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:10:12.499249Z", "task_family": "synthetic", "document_id": "syn_03018_2000", "document_hash": "d284eb07e27da742b9c2b67abccfdf72e4dd67c3259f44f3b41e34dd3e6fc1ff", "question_id": "q10_temporal", "split": "test", "cluster_id": "syn_03018_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 4, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1982, "query_tokens": 503, "message_tokens": 401, "message_utf8_bytes": 940, "cache_tensor_bytes": null, "serialized_payload_bytes": 940, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 76780.02479101997, "hops": {"1": 22800.499332952313, "2": 18183.315165922977, "3": 17950.87779208552, "4": 17845.332500059158}, "receiver_prefill_ms": 317.5017921021208, "ttft_ms": 42.9642079398036, "decode_ms": 20.61650005634874}, "total_wall_ms": 77161.6784989601, "ttft_ms": 42.9642079398036, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 64, "facts_retained": 14, "retention_rate": 0.21875, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.2222222222222222, "amount": 0.15384615384615385, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 870}, "ledger_by_hop": {"1": {"facts_total": 64, "facts_retained": 14, "retention_rate": 0.21875, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.2222222222222222, "amount": 0.15384615384615385, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 870}, "2": {"facts_total": 64, "facts_retained": 14, "retention_rate": 0.21875, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.2222222222222222, "amount": 0.15384615384615385, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 870}, "3": {"facts_total": 64, "facts_retained": 14, "retention_rate": 0.21875, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.2222222222222222, "amount": 0.15384615384615385, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 870}, "4": {"facts_total": 64, "facts_retained": 14, "retention_rate": 0.21875, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.2222222222222222, "amount": 0.15384615384615385, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 870}}}, "reference_key_id": "keys/synthetic/syn_03018_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "b93ce255451151cce89ccd51cc89c5b4382f883b441b3818bc249f2f79c54d9e", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "c90250b38326fef36eac5bf4", "attempt_id": "c86b0de5c8ab", "run_group_id": "E4_core:A5|closed|qwen3b:h4", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:10:12.899497Z", "task_family": "synthetic", "document_id": "syn_03018_2000", "document_hash": "d284eb07e27da742b9c2b67abccfdf72e4dd67c3259f44f3b41e34dd3e6fc1ff", "question_id": "q11_aggregation", "split": "test", "cluster_id": "syn_03018_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 4, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1982, "query_tokens": 495, "message_tokens": 401, "message_utf8_bytes": 940, "cache_tensor_bytes": null, "serialized_payload_bytes": 940, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 76780.02479101997, "hops": {"1": 22800.499332952313, "2": 18183.315165922977, "3": 17950.87779208552, "4": 17845.332500059158}, "receiver_prefill_ms": 352.82604198437184, "ttft_ms": 22.867457941174507, "decode_ms": 286.0349171096459}, "total_wall_ms": 77442.3472910421, "ttft_ms": 22.867457941174507, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 7}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "3189245", "evidence_artifact": {"ledger": {"facts_total": 64, "facts_retained": 14, "retention_rate": 0.21875, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.2222222222222222, "amount": 0.15384615384615385, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 870}, "ledger_by_hop": {"1": {"facts_total": 64, "facts_retained": 14, "retention_rate": 0.21875, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.2222222222222222, "amount": 0.15384615384615385, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 870}, "2": {"facts_total": 64, "facts_retained": 14, "retention_rate": 0.21875, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.2222222222222222, "amount": 0.15384615384615385, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 870}, "3": {"facts_total": 64, "facts_retained": 14, "retention_rate": 0.21875, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.2222222222222222, "amount": 0.15384615384615385, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 870}, "4": {"facts_total": 64, "facts_retained": 14, "retention_rate": 0.21875, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.2222222222222222, "amount": 0.15384615384615385, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 870}}}, "reference_key_id": "keys/synthetic/syn_03018_2000.json", "task_scores": {"correct": 0.0, "abstained": 0.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "b93ce255451151cce89ccd51cc89c5b4382f883b441b3818bc249f2f79c54d9e", "output_hash": "82576d93a43637ab2b22d213df62e73fbc3c84f64f177d56d7fd57c0cee09155"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "eeb28356cbbe82b958e0028e", "attempt_id": "4904a4898a75", "run_group_id": "E4_core:A5|closed|qwen3b:h4", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:10:13.578933Z", "task_family": "synthetic", "document_id": "syn_03018_2000", "document_hash": "d284eb07e27da742b9c2b67abccfdf72e4dd67c3259f44f3b41e34dd3e6fc1ff", "question_id": "q12_unit_conversion", "split": "test", "cluster_id": "syn_03018_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 4, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1982, "query_tokens": 492, "message_tokens": 401, "message_utf8_bytes": 940, "cache_tensor_bytes": null, "serialized_payload_bytes": 940, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 76780.02479101997, "hops": {"1": 22800.499332952313, "2": 18183.315165922977, "3": 17950.87779208552, "4": 17845.332500059158}, "receiver_prefill_ms": 505.6297079427168, "ttft_ms": 34.20374996494502, "decode_ms": 31.82933304924518}, "total_wall_ms": 77352.24129108246, "ttft_ms": 34.20374996494502, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 64, "facts_retained": 14, "retention_rate": 0.21875, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.2222222222222222, "amount": 0.15384615384615385, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 870}, "ledger_by_hop": {"1": {"facts_total": 64, "facts_retained": 14, "retention_rate": 0.21875, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.2222222222222222, "amount": 0.15384615384615385, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 870}, "2": {"facts_total": 64, "facts_retained": 14, "retention_rate": 0.21875, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.2222222222222222, "amount": 0.15384615384615385, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 870}, "3": {"facts_total": 64, "facts_retained": 14, "retention_rate": 0.21875, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.2222222222222222, "amount": 0.15384615384615385, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 870}, "4": {"facts_total": 64, "facts_retained": 14, "retention_rate": 0.21875, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.2222222222222222, "amount": 0.15384615384615385, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 870}}}, "reference_key_id": "keys/synthetic/syn_03018_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "b93ce255451151cce89ccd51cc89c5b4382f883b441b3818bc249f2f79c54d9e", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "a2f64e583da416a3af4a931c", "attempt_id": "7acc4f732a75", "run_group_id": "E4_core:A5|closed|qwen3b:h4", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:10:14.169053Z", "task_family": "synthetic", "document_id": "syn_03018_2000", "document_hash": "d284eb07e27da742b9c2b67abccfdf72e4dd67c3259f44f3b41e34dd3e6fc1ff", "question_id": "q13_unanswerable", "split": "test", "cluster_id": "syn_03018_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 4, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1982, "query_tokens": 496, "message_tokens": 401, "message_utf8_bytes": 940, "cache_tensor_bytes": null, "serialized_payload_bytes": 940, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 76780.02479101997, "hops": {"1": 22800.499332952313, "2": 18183.315165922977, "3": 17950.87779208552, "4": 17845.332500059158}, "receiver_prefill_ms": 327.76941696647555, "ttft_ms": 41.45245801191777, "decode_ms": 21.275041974149644}, "total_wall_ms": 77171.12745798659, "ttft_ms": 41.45245801191777, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 64, "facts_retained": 14, "retention_rate": 0.21875, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.2222222222222222, "amount": 0.15384615384615385, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 870}, "ledger_by_hop": {"1": {"facts_total": 64, "facts_retained": 14, "retention_rate": 0.21875, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.2222222222222222, "amount": 0.15384615384615385, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 870}, "2": {"facts_total": 64, "facts_retained": 14, "retention_rate": 0.21875, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.2222222222222222, "amount": 0.15384615384615385, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 870}, "3": {"facts_total": 64, "facts_retained": 14, "retention_rate": 0.21875, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.2222222222222222, "amount": 0.15384615384615385, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 870}, "4": {"facts_total": 64, "facts_retained": 14, "retention_rate": 0.21875, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.2222222222222222, "amount": 0.15384615384615385, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 870}}}, "reference_key_id": "keys/synthetic/syn_03018_2000.json", "task_scores": {"correct": 1.0, "abstained": 1.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "b93ce255451151cce89ccd51cc89c5b4382f883b441b3818bc249f2f79c54d9e", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "3a3e736986b4ae15322ccea7", "attempt_id": "891e6e39272e", "run_group_id": "E4_core:A5|closed|qwen3b:h4", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:10:14.577422Z", "task_family": "synthetic", "document_id": "syn_03018_2000", "document_hash": "d284eb07e27da742b9c2b67abccfdf72e4dd67c3259f44f3b41e34dd3e6fc1ff", "question_id": "q14_unanswerable", "split": "test", "cluster_id": "syn_03018_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 4, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1982, "query_tokens": 497, "message_tokens": 401, "message_utf8_bytes": 940, "cache_tensor_bytes": null, "serialized_payload_bytes": 940, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 76780.02479101997, "hops": {"1": 22800.499332952313, "2": 18183.315165922977, "3": 17950.87779208552, "4": 17845.332500059158}, "receiver_prefill_ms": 328.781666001305, "ttft_ms": 38.84287504479289, "decode_ms": 19.28658306133002}, "total_wall_ms": 77167.48187399935, "ttft_ms": 38.84287504479289, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 64, "facts_retained": 14, "retention_rate": 0.21875, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.2222222222222222, "amount": 0.15384615384615385, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 870}, "ledger_by_hop": {"1": {"facts_total": 64, "facts_retained": 14, "retention_rate": 0.21875, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.2222222222222222, "amount": 0.15384615384615385, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 870}, "2": {"facts_total": 64, "facts_retained": 14, "retention_rate": 0.21875, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.2222222222222222, "amount": 0.15384615384615385, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 870}, "3": {"facts_total": 64, "facts_retained": 14, "retention_rate": 0.21875, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.2222222222222222, "amount": 0.15384615384615385, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 870}, "4": {"facts_total": 64, "facts_retained": 14, "retention_rate": 0.21875, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.2222222222222222, "amount": 0.15384615384615385, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 870}}}, "reference_key_id": "keys/synthetic/syn_03018_2000.json", "task_scores": {"correct": 1.0, "abstained": 1.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "b93ce255451151cce89ccd51cc89c5b4382f883b441b3818bc249f2f79c54d9e", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "121ee59bb4d4c1febfa3de8f", "attempt_id": "5bc3a9981356", "run_group_id": "E4_core:A5|closed|qwen3b:h8", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:11:26.050380Z", "task_family": "synthetic", "document_id": "syn_03018_2000", "document_hash": "d284eb07e27da742b9c2b67abccfdf72e4dd67c3259f44f3b41e34dd3e6fc1ff", "question_id": "q00_single_fact", "split": "test", "cluster_id": "syn_03018_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 8, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1982, "query_tokens": 484, "message_tokens": 401, "message_utf8_bytes": 940, "cache_tensor_bytes": null, "serialized_payload_bytes": 940, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 147844.79250002187, "hops": {"1": 22800.499332952313, "2": 18183.315165922977, "3": 17950.87779208552, "4": 17845.332500059158, "5": 17775.58920904994, "6": 17719.9630420655, "7": 17847.403332940303, "8": 17721.812124946155}, "receiver_prefill_ms": 317.7090829703957, "ttft_ms": 52.95616597868502, "decode_ms": 87.82350004184991}, "total_wall_ms": 148304.02679205872, "ttft_ms": 52.95616597868502, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 64, "facts_retained": 14, "retention_rate": 0.21875, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.2222222222222222, "amount": 0.15384615384615385, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 870}, "ledger_by_hop": {"1": {"facts_total": 64, "facts_retained": 14, "retention_rate": 0.21875, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.2222222222222222, "amount": 0.15384615384615385, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 870}, "2": {"facts_total": 64, "facts_retained": 14, "retention_rate": 0.21875, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.2222222222222222, "amount": 0.15384615384615385, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 870}, "3": {"facts_total": 64, "facts_retained": 14, "retention_rate": 0.21875, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.2222222222222222, "amount": 0.15384615384615385, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 870}, "4": {"facts_total": 64, "facts_retained": 14, "retention_rate": 0.21875, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.2222222222222222, "amount": 0.15384615384615385, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 870}, "5": {"facts_total": 64, "facts_retained": 14, "retention_rate": 0.21875, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.2222222222222222, "amount": 0.15384615384615385, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 870}, "6": {"facts_total": 64, "facts_retained": 14, "retention_rate": 0.21875, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.2222222222222222, "amount": 0.15384615384615385, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 870}, "7": {"facts_total": 64, "facts_retained": 14, "retention_rate": 0.21875, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.2222222222222222, "amount": 0.15384615384615385, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 870}, "8": {"facts_total": 64, "facts_retained": 14, "retention_rate": 0.21875, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.2222222222222222, "amount": 0.15384615384615385, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 870}}}, "reference_key_id": "keys/synthetic/syn_03018_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "b93ce255451151cce89ccd51cc89c5b4382f883b441b3818bc249f2f79c54d9e", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "a2458b73534b09211d6b3989", "attempt_id": "ed77e34f01d6", "run_group_id": "E4_core:A5|closed|qwen3b:h8", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:11:26.528719Z", "task_family": "synthetic", "document_id": "syn_03018_2000", "document_hash": "d284eb07e27da742b9c2b67abccfdf72e4dd67c3259f44f3b41e34dd3e6fc1ff", "question_id": "q01_single_fact", "split": "test", "cluster_id": "syn_03018_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 8, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1982, "query_tokens": 490, "message_tokens": 401, "message_utf8_bytes": 940, "cache_tensor_bytes": null, "serialized_payload_bytes": 940, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 147844.79250002187, "hops": {"1": 22800.499332952313, "2": 18183.315165922977, "3": 17950.87779208552, "4": 17845.332500059158, "5": 17775.58920904994, "6": 17719.9630420655, "7": 17847.403332940303, "8": 17721.812124946155}, "receiver_prefill_ms": 382.0837920065969, "ttft_ms": 38.07116695679724, "decode_ms": 24.18545796535909}, "total_wall_ms": 148289.61741703097, "ttft_ms": 38.07116695679724, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 64, "facts_retained": 14, "retention_rate": 0.21875, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.2222222222222222, "amount": 0.15384615384615385, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 870}, "ledger_by_hop": {"1": {"facts_total": 64, "facts_retained": 14, "retention_rate": 0.21875, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.2222222222222222, "amount": 0.15384615384615385, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 870}, "2": {"facts_total": 64, "facts_retained": 14, "retention_rate": 0.21875, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.2222222222222222, "amount": 0.15384615384615385, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 870}, "3": {"facts_total": 64, "facts_retained": 14, "retention_rate": 0.21875, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.2222222222222222, "amount": 0.15384615384615385, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 870}, "4": {"facts_total": 64, "facts_retained": 14, "retention_rate": 0.21875, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.2222222222222222, "amount": 0.15384615384615385, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 870}, "5": {"facts_total": 64, "facts_retained": 14, "retention_rate": 0.21875, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.2222222222222222, "amount": 0.15384615384615385, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 870}, "6": {"facts_total": 64, "facts_retained": 14, "retention_rate": 0.21875, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.2222222222222222, "amount": 0.15384615384615385, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 870}, "7": {"facts_total": 64, "facts_retained": 14, "retention_rate": 0.21875, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.2222222222222222, "amount": 0.15384615384615385, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 870}, "8": {"facts_total": 64, "facts_retained": 14, "retention_rate": 0.21875, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.2222222222222222, "amount": 0.15384615384615385, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 870}}}, "reference_key_id": "keys/synthetic/syn_03018_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "b93ce255451151cce89ccd51cc89c5b4382f883b441b3818bc249f2f79c54d9e", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "d0a34c1cce8c86747260f210", "attempt_id": "737ff56700d0", "run_group_id": "E4_core:A5|closed|qwen3b:h8", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:11:26.990102Z", "task_family": "synthetic", "document_id": "syn_03018_2000", "document_hash": "d284eb07e27da742b9c2b67abccfdf72e4dd67c3259f44f3b41e34dd3e6fc1ff", "question_id": "q02_exact_copy", "split": "test", "cluster_id": "syn_03018_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 8, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1982, "query_tokens": 489, "message_tokens": 401, "message_utf8_bytes": 940, "cache_tensor_bytes": null, "serialized_payload_bytes": 940, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 147844.79250002187, "hops": {"1": 22800.499332952313, "2": 18183.315165922977, "3": 17950.87779208552, "4": 17845.332500059158, "5": 17775.58920904994, "6": 17719.9630420655, "7": 17847.403332940303, "8": 17721.812124946155}, "receiver_prefill_ms": 339.9031669832766, "ttft_ms": 33.56720809824765, "decode_ms": 38.81637495942414}, "total_wall_ms": 148257.6449579792, "ttft_ms": 33.56720809824765, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 64, "facts_retained": 14, "retention_rate": 0.21875, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.2222222222222222, "amount": 0.15384615384615385, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 870}, "ledger_by_hop": {"1": {"facts_total": 64, "facts_retained": 14, "retention_rate": 0.21875, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.2222222222222222, "amount": 0.15384615384615385, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 870}, "2": {"facts_total": 64, "facts_retained": 14, "retention_rate": 0.21875, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.2222222222222222, "amount": 0.15384615384615385, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 870}, "3": {"facts_total": 64, "facts_retained": 14, "retention_rate": 0.21875, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.2222222222222222, "amount": 0.15384615384615385, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 870}, "4": {"facts_total": 64, "facts_retained": 14, "retention_rate": 0.21875, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.2222222222222222, "amount": 0.15384615384615385, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 870}, "5": {"facts_total": 64, "facts_retained": 14, "retention_rate": 0.21875, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.2222222222222222, "amount": 0.15384615384615385, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 870}, "6": {"facts_total": 64, "facts_retained": 14, "retention_rate": 0.21875, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.2222222222222222, "amount": 0.15384615384615385, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 870}, "7": {"facts_total": 64, "facts_retained": 14, "retention_rate": 0.21875, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.2222222222222222, "amount": 0.15384615384615385, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 870}, "8": {"facts_total": 64, "facts_retained": 14, "retention_rate": 0.21875, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.2222222222222222, "amount": 0.15384615384615385, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 870}}}, "reference_key_id": "keys/synthetic/syn_03018_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "b93ce255451151cce89ccd51cc89c5b4382f883b441b3818bc249f2f79c54d9e", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "abab109886462faffebe500b", "attempt_id": "bfd89cde3758", "run_group_id": "E4_core:A5|closed|qwen3b:h8", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:11:27.420541Z", "task_family": "synthetic", "document_id": "syn_03018_2000", "document_hash": "d284eb07e27da742b9c2b67abccfdf72e4dd67c3259f44f3b41e34dd3e6fc1ff", "question_id": "q03_exact_copy", "split": "test", "cluster_id": "syn_03018_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 8, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1982, "query_tokens": 487, "message_tokens": 401, "message_utf8_bytes": 940, "cache_tensor_bytes": null, "serialized_payload_bytes": 940, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 147844.79250002187, "hops": {"1": 22800.499332952313, "2": 18183.315165922977, "3": 17950.87779208552, "4": 17845.332500059158, "5": 17775.58920904994, "6": 17719.9630420655, "7": 17847.403332940303, "8": 17721.812124946155}, "receiver_prefill_ms": 278.4755000611767, "ttft_ms": 37.986209033988416, "decode_ms": 596.40637494158}, "total_wall_ms": 148758.2453750074, "ttft_ms": 37.986209033988416, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 9}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "Priyamvada Sørlie", "evidence_artifact": {"ledger": {"facts_total": 64, "facts_retained": 14, "retention_rate": 0.21875, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.2222222222222222, "amount": 0.15384615384615385, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 870}, "ledger_by_hop": {"1": {"facts_total": 64, "facts_retained": 14, "retention_rate": 0.21875, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.2222222222222222, "amount": 0.15384615384615385, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 870}, "2": {"facts_total": 64, "facts_retained": 14, "retention_rate": 0.21875, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.2222222222222222, "amount": 0.15384615384615385, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 870}, "3": {"facts_total": 64, "facts_retained": 14, "retention_rate": 0.21875, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.2222222222222222, "amount": 0.15384615384615385, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 870}, "4": {"facts_total": 64, "facts_retained": 14, "retention_rate": 0.21875, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.2222222222222222, "amount": 0.15384615384615385, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 870}, "5": {"facts_total": 64, "facts_retained": 14, "retention_rate": 0.21875, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.2222222222222222, "amount": 0.15384615384615385, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 870}, "6": {"facts_total": 64, "facts_retained": 14, "retention_rate": 0.21875, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.2222222222222222, "amount": 0.15384615384615385, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 870}, "7": {"facts_total": 64, "facts_retained": 14, "retention_rate": 0.21875, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.2222222222222222, "amount": 0.15384615384615385, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 870}, "8": {"facts_total": 64, "facts_retained": 14, "retention_rate": 0.21875, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.2222222222222222, "amount": 0.15384615384615385, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 870}}}, "reference_key_id": "keys/synthetic/syn_03018_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "b93ce255451151cce89ccd51cc89c5b4382f883b441b3818bc249f2f79c54d9e", "output_hash": "c5f554196ed23dd71ad995ec89f7142099d524dbb6534cc10d94a0c89d13e792"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "0c8f76d70153b8ddd1391cad", "attempt_id": "09d911dca919", "run_group_id": "E4_core:A5|closed|qwen3b:h8", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:11:28.351525Z", "task_family": "synthetic", "document_id": "syn_03018_2000", "document_hash": "d284eb07e27da742b9c2b67abccfdf72e4dd67c3259f44f3b41e34dd3e6fc1ff", "question_id": "q04_multi_hop", "split": "test", "cluster_id": "syn_03018_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 8, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1982, "query_tokens": 493, "message_tokens": 401, "message_utf8_bytes": 940, "cache_tensor_bytes": null, "serialized_payload_bytes": 940, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 147844.79250002187, "hops": {"1": 22800.499332952313, "2": 18183.315165922977, "3": 17950.87779208552, "4": 17845.332500059158, "5": 17775.58920904994, "6": 17719.9630420655, "7": 17847.403332940303, "8": 17721.812124946155}, "receiver_prefill_ms": 310.82529097329825, "ttft_ms": 40.20799999125302, "decode_ms": 34.78037496097386}, "total_wall_ms": 148231.15891695488, "ttft_ms": 40.20799999125302, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 64, "facts_retained": 14, "retention_rate": 0.21875, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.2222222222222222, "amount": 0.15384615384615385, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 870}, "ledger_by_hop": {"1": {"facts_total": 64, "facts_retained": 14, "retention_rate": 0.21875, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.2222222222222222, "amount": 0.15384615384615385, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 870}, "2": {"facts_total": 64, "facts_retained": 14, "retention_rate": 0.21875, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.2222222222222222, "amount": 0.15384615384615385, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 870}, "3": {"facts_total": 64, "facts_retained": 14, "retention_rate": 0.21875, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.2222222222222222, "amount": 0.15384615384615385, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 870}, "4": {"facts_total": 64, "facts_retained": 14, "retention_rate": 0.21875, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.2222222222222222, "amount": 0.15384615384615385, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 870}, "5": {"facts_total": 64, "facts_retained": 14, "retention_rate": 0.21875, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.2222222222222222, "amount": 0.15384615384615385, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 870}, "6": {"facts_total": 64, "facts_retained": 14, "retention_rate": 0.21875, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.2222222222222222, "amount": 0.15384615384615385, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 870}, "7": {"facts_total": 64, "facts_retained": 14, "retention_rate": 0.21875, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.2222222222222222, "amount": 0.15384615384615385, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 870}, "8": {"facts_total": 64, "facts_retained": 14, "retention_rate": 0.21875, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.2222222222222222, "amount": 0.15384615384615385, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 870}}}, "reference_key_id": "keys/synthetic/syn_03018_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "b93ce255451151cce89ccd51cc89c5b4382f883b441b3818bc249f2f79c54d9e", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "41a4dfaced23dbc46470123b", "attempt_id": "1f040752cda6", "run_group_id": "E4_core:A5|closed|qwen3b:h8", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:11:28.754891Z", "task_family": "synthetic", "document_id": "syn_03018_2000", "document_hash": "d284eb07e27da742b9c2b67abccfdf72e4dd67c3259f44f3b41e34dd3e6fc1ff", "question_id": "q05_multi_hop", "split": "test", "cluster_id": "syn_03018_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 8, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1982, "query_tokens": 493, "message_tokens": 401, "message_utf8_bytes": 940, "cache_tensor_bytes": null, "serialized_payload_bytes": 940, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 147844.79250002187, "hops": {"1": 22800.499332952313, "2": 18183.315165922977, "3": 17950.87779208552, "4": 17845.332500059158, "5": 17775.58920904994, "6": 17719.9630420655, "7": 17847.403332940303, "8": 17721.812124946155}, "receiver_prefill_ms": 338.8461669674143, "ttft_ms": 22.80649996828288, "decode_ms": 37.615290959365666}, "total_wall_ms": 148244.5553749567, "ttft_ms": 22.80649996828288, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 64, "facts_retained": 14, "retention_rate": 0.21875, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.2222222222222222, "amount": 0.15384615384615385, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 870}, "ledger_by_hop": {"1": {"facts_total": 64, "facts_retained": 14, "retention_rate": 0.21875, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.2222222222222222, "amount": 0.15384615384615385, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 870}, "2": {"facts_total": 64, "facts_retained": 14, "retention_rate": 0.21875, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.2222222222222222, "amount": 0.15384615384615385, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 870}, "3": {"facts_total": 64, "facts_retained": 14, "retention_rate": 0.21875, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.2222222222222222, "amount": 0.15384615384615385, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 870}, "4": {"facts_total": 64, "facts_retained": 14, "retention_rate": 0.21875, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.2222222222222222, "amount": 0.15384615384615385, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 870}, "5": {"facts_total": 64, "facts_retained": 14, "retention_rate": 0.21875, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.2222222222222222, "amount": 0.15384615384615385, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 870}, "6": {"facts_total": 64, "facts_retained": 14, "retention_rate": 0.21875, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.2222222222222222, "amount": 0.15384615384615385, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 870}, "7": {"facts_total": 64, "facts_retained": 14, "retention_rate": 0.21875, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.2222222222222222, "amount": 0.15384615384615385, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 870}, "8": {"facts_total": 64, "facts_retained": 14, "retention_rate": 0.21875, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.2222222222222222, "amount": 0.15384615384615385, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 870}}}, "reference_key_id": "keys/synthetic/syn_03018_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "b93ce255451151cce89ccd51cc89c5b4382f883b441b3818bc249f2f79c54d9e", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "ad33838efab08f723e4e2a3a", "attempt_id": "cc8cd9700627", "run_group_id": "E4_core:A5|closed|qwen3b:h8", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:11:29.171125Z", "task_family": "synthetic", "document_id": "syn_03018_2000", "document_hash": "d284eb07e27da742b9c2b67abccfdf72e4dd67c3259f44f3b41e34dd3e6fc1ff", "question_id": "q06_negation", "split": "test", "cluster_id": "syn_03018_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 8, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1982, "query_tokens": 485, "message_tokens": 401, "message_utf8_bytes": 940, "cache_tensor_bytes": null, "serialized_payload_bytes": 940, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 147844.79250002187, "hops": {"1": 22800.499332952313, "2": 18183.315165922977, "3": 17950.87779208552, "4": 17845.332500059158, "5": 17775.58920904994, "6": 17719.9630420655, "7": 17847.403332940303, "8": 17721.812124946155}, "receiver_prefill_ms": 305.37183303385973, "ttft_ms": 53.391499910503626, "decode_ms": 60.95179100520909}, "total_wall_ms": 148265.15133294743, "ttft_ms": 53.391499910503626, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 64, "facts_retained": 14, "retention_rate": 0.21875, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.2222222222222222, "amount": 0.15384615384615385, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 870}, "ledger_by_hop": {"1": {"facts_total": 64, "facts_retained": 14, "retention_rate": 0.21875, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.2222222222222222, "amount": 0.15384615384615385, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 870}, "2": {"facts_total": 64, "facts_retained": 14, "retention_rate": 0.21875, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.2222222222222222, "amount": 0.15384615384615385, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 870}, "3": {"facts_total": 64, "facts_retained": 14, "retention_rate": 0.21875, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.2222222222222222, "amount": 0.15384615384615385, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 870}, "4": {"facts_total": 64, "facts_retained": 14, "retention_rate": 0.21875, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.2222222222222222, "amount": 0.15384615384615385, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 870}, "5": {"facts_total": 64, "facts_retained": 14, "retention_rate": 0.21875, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.2222222222222222, "amount": 0.15384615384615385, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 870}, "6": {"facts_total": 64, "facts_retained": 14, "retention_rate": 0.21875, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.2222222222222222, "amount": 0.15384615384615385, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 870}, "7": {"facts_total": 64, "facts_retained": 14, "retention_rate": 0.21875, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.2222222222222222, "amount": 0.15384615384615385, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 870}, "8": {"facts_total": 64, "facts_retained": 14, "retention_rate": 0.21875, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.2222222222222222, "amount": 0.15384615384615385, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 870}}}, "reference_key_id": "keys/synthetic/syn_03018_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "b93ce255451151cce89ccd51cc89c5b4382f883b441b3818bc249f2f79c54d9e", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "c69145c0815adae81467ffc9", "attempt_id": "6ab9b0dbe6ff", "run_group_id": "E4_core:A5|closed|qwen3b:h8", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:11:29.608934Z", "task_family": "synthetic", "document_id": "syn_03018_2000", "document_hash": "d284eb07e27da742b9c2b67abccfdf72e4dd67c3259f44f3b41e34dd3e6fc1ff", "question_id": "q07_negation", "split": "test", "cluster_id": "syn_03018_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 8, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1982, "query_tokens": 484, "message_tokens": 401, "message_utf8_bytes": 940, "cache_tensor_bytes": null, "serialized_payload_bytes": 940, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 147844.79250002187, "hops": {"1": 22800.499332952313, "2": 18183.315165922977, "3": 17950.87779208552, "4": 17845.332500059158, "5": 17775.58920904994, "6": 17719.9630420655, "7": 17847.403332940303, "8": 17721.812124946155}, "receiver_prefill_ms": 479.0297499857843, "ttft_ms": 33.582541975192726, "decode_ms": 30.88866698089987}, "total_wall_ms": 148388.829459087, "ttft_ms": 33.582541975192726, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 64, "facts_retained": 14, "retention_rate": 0.21875, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.2222222222222222, "amount": 0.15384615384615385, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 870}, "ledger_by_hop": {"1": {"facts_total": 64, "facts_retained": 14, "retention_rate": 0.21875, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.2222222222222222, "amount": 0.15384615384615385, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 870}, "2": {"facts_total": 64, "facts_retained": 14, "retention_rate": 0.21875, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.2222222222222222, "amount": 0.15384615384615385, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 870}, "3": {"facts_total": 64, "facts_retained": 14, "retention_rate": 0.21875, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.2222222222222222, "amount": 0.15384615384615385, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 870}, "4": {"facts_total": 64, "facts_retained": 14, "retention_rate": 0.21875, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.2222222222222222, "amount": 0.15384615384615385, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 870}, "5": {"facts_total": 64, "facts_retained": 14, "retention_rate": 0.21875, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.2222222222222222, "amount": 0.15384615384615385, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 870}, "6": {"facts_total": 64, "facts_retained": 14, "retention_rate": 0.21875, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.2222222222222222, "amount": 0.15384615384615385, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 870}, "7": {"facts_total": 64, "facts_retained": 14, "retention_rate": 0.21875, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.2222222222222222, "amount": 0.15384615384615385, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 870}, "8": {"facts_total": 64, "facts_retained": 14, "retention_rate": 0.21875, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.2222222222222222, "amount": 0.15384615384615385, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 870}}}, "reference_key_id": "keys/synthetic/syn_03018_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "b93ce255451151cce89ccd51cc89c5b4382f883b441b3818bc249f2f79c54d9e", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "486cc485503898e66b5d0eff", "attempt_id": "fa3ab83bcdfb", "run_group_id": "E4_core:A5|closed|qwen3b:h8", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:11:30.171289Z", "task_family": "synthetic", "document_id": "syn_03018_2000", "document_hash": "d284eb07e27da742b9c2b67abccfdf72e4dd67c3259f44f3b41e34dd3e6fc1ff", "question_id": "q08_revision", "split": "test", "cluster_id": "syn_03018_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 8, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1982, "query_tokens": 502, "message_tokens": 401, "message_utf8_bytes": 940, "cache_tensor_bytes": null, "serialized_payload_bytes": 940, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 147844.79250002187, "hops": {"1": 22800.499332952313, "2": 18183.315165922977, "3": 17950.87779208552, "4": 17845.332500059158, "5": 17775.58920904994, "6": 17719.9630420655, "7": 17847.403332940303, "8": 17721.812124946155}, "receiver_prefill_ms": 322.0192500157282, "ttft_ms": 39.604166988283396, "decode_ms": 200.8325420320034}, "total_wall_ms": 148407.99108310603, "ttft_ms": 39.604166988283396, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 6}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "272184", "evidence_artifact": {"ledger": {"facts_total": 64, "facts_retained": 14, "retention_rate": 0.21875, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.2222222222222222, "amount": 0.15384615384615385, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 870}, "ledger_by_hop": {"1": {"facts_total": 64, "facts_retained": 14, "retention_rate": 0.21875, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.2222222222222222, "amount": 0.15384615384615385, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 870}, "2": {"facts_total": 64, "facts_retained": 14, "retention_rate": 0.21875, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.2222222222222222, "amount": 0.15384615384615385, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 870}, "3": {"facts_total": 64, "facts_retained": 14, "retention_rate": 0.21875, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.2222222222222222, "amount": 0.15384615384615385, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 870}, "4": {"facts_total": 64, "facts_retained": 14, "retention_rate": 0.21875, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.2222222222222222, "amount": 0.15384615384615385, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 870}, "5": {"facts_total": 64, "facts_retained": 14, "retention_rate": 0.21875, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.2222222222222222, "amount": 0.15384615384615385, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 870}, "6": {"facts_total": 64, "facts_retained": 14, "retention_rate": 0.21875, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.2222222222222222, "amount": 0.15384615384615385, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 870}, "7": {"facts_total": 64, "facts_retained": 14, "retention_rate": 0.21875, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.2222222222222222, "amount": 0.15384615384615385, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 870}, "8": {"facts_total": 64, "facts_retained": 14, "retention_rate": 0.21875, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.2222222222222222, "amount": 0.15384615384615385, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 870}}}, "reference_key_id": "keys/synthetic/syn_03018_2000.json", "task_scores": {"correct": 0.0, "abstained": 0.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "b93ce255451151cce89ccd51cc89c5b4382f883b441b3818bc249f2f79c54d9e", "output_hash": "3aadaddf09c30b8bfe5dc7e2aa6984fcf0df4d216145cf925760dd5a946d6e05"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "763836e337524401611d3a1c", "attempt_id": "5e6803a789fa", "run_group_id": "E4_core:A5|closed|qwen3b:h8", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:11:30.751913Z", "task_family": "synthetic", "document_id": "syn_03018_2000", "document_hash": "d284eb07e27da742b9c2b67abccfdf72e4dd67c3259f44f3b41e34dd3e6fc1ff", "question_id": "q09_revision", "split": "test", "cluster_id": "syn_03018_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 8, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1982, "query_tokens": 495, "message_tokens": 401, "message_utf8_bytes": 940, "cache_tensor_bytes": null, "serialized_payload_bytes": 940, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 147844.79250002187, "hops": {"1": 22800.499332952313, "2": 18183.315165922977, "3": 17950.87779208552, "4": 17845.332500059158, "5": 17775.58920904994, "6": 17719.9630420655, "7": 17847.403332940303, "8": 17721.812124946155}, "receiver_prefill_ms": 312.4349170830101, "ttft_ms": 42.03829204197973, "decode_ms": 447.0402500592172}, "total_wall_ms": 148646.96404198185, "ttft_ms": 42.03829204197973, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 6}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "272184", "evidence_artifact": {"ledger": {"facts_total": 64, "facts_retained": 14, "retention_rate": 0.21875, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.2222222222222222, "amount": 0.15384615384615385, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 870}, "ledger_by_hop": {"1": {"facts_total": 64, "facts_retained": 14, "retention_rate": 0.21875, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.2222222222222222, "amount": 0.15384615384615385, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 870}, "2": {"facts_total": 64, "facts_retained": 14, "retention_rate": 0.21875, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.2222222222222222, "amount": 0.15384615384615385, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 870}, "3": {"facts_total": 64, "facts_retained": 14, "retention_rate": 0.21875, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.2222222222222222, "amount": 0.15384615384615385, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 870}, "4": {"facts_total": 64, "facts_retained": 14, "retention_rate": 0.21875, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.2222222222222222, "amount": 0.15384615384615385, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 870}, "5": {"facts_total": 64, "facts_retained": 14, "retention_rate": 0.21875, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.2222222222222222, "amount": 0.15384615384615385, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 870}, "6": {"facts_total": 64, "facts_retained": 14, "retention_rate": 0.21875, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.2222222222222222, "amount": 0.15384615384615385, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 870}, "7": {"facts_total": 64, "facts_retained": 14, "retention_rate": 0.21875, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.2222222222222222, "amount": 0.15384615384615385, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 870}, "8": {"facts_total": 64, "facts_retained": 14, "retention_rate": 0.21875, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.2222222222222222, "amount": 0.15384615384615385, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 870}}}, "reference_key_id": "keys/synthetic/syn_03018_2000.json", "task_scores": {"correct": 0.0, "abstained": 0.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "b93ce255451151cce89ccd51cc89c5b4382f883b441b3818bc249f2f79c54d9e", "output_hash": "3aadaddf09c30b8bfe5dc7e2aa6984fcf0df4d216145cf925760dd5a946d6e05"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "0fd9183672f4f453ea06fc8e", "attempt_id": "b1e9f8aa2508", "run_group_id": "E4_core:A5|closed|qwen3b:h8", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:11:31.572089Z", "task_family": "synthetic", "document_id": "syn_03018_2000", "document_hash": "d284eb07e27da742b9c2b67abccfdf72e4dd67c3259f44f3b41e34dd3e6fc1ff", "question_id": "q10_temporal", "split": "test", "cluster_id": "syn_03018_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 8, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1982, "query_tokens": 503, "message_tokens": 401, "message_utf8_bytes": 940, "cache_tensor_bytes": null, "serialized_payload_bytes": 940, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 147844.79250002187, "hops": {"1": 22800.499332952313, "2": 18183.315165922977, "3": 17950.87779208552, "4": 17845.332500059158, "5": 17775.58920904994, "6": 17719.9630420655, "7": 17847.403332940303, "8": 17721.812124946155}, "receiver_prefill_ms": 328.3688339870423, "ttft_ms": 27.42395899258554, "decode_ms": 39.70916708931327}, "total_wall_ms": 148240.8621670911, "ttft_ms": 27.42395899258554, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 64, "facts_retained": 14, "retention_rate": 0.21875, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.2222222222222222, "amount": 0.15384615384615385, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 870}, "ledger_by_hop": {"1": {"facts_total": 64, "facts_retained": 14, "retention_rate": 0.21875, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.2222222222222222, "amount": 0.15384615384615385, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 870}, "2": {"facts_total": 64, "facts_retained": 14, "retention_rate": 0.21875, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.2222222222222222, "amount": 0.15384615384615385, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 870}, "3": {"facts_total": 64, "facts_retained": 14, "retention_rate": 0.21875, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.2222222222222222, "amount": 0.15384615384615385, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 870}, "4": {"facts_total": 64, "facts_retained": 14, "retention_rate": 0.21875, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.2222222222222222, "amount": 0.15384615384615385, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 870}, "5": {"facts_total": 64, "facts_retained": 14, "retention_rate": 0.21875, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.2222222222222222, "amount": 0.15384615384615385, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 870}, "6": {"facts_total": 64, "facts_retained": 14, "retention_rate": 0.21875, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.2222222222222222, "amount": 0.15384615384615385, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 870}, "7": {"facts_total": 64, "facts_retained": 14, "retention_rate": 0.21875, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.2222222222222222, "amount": 0.15384615384615385, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 870}, "8": {"facts_total": 64, "facts_retained": 14, "retention_rate": 0.21875, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.2222222222222222, "amount": 0.15384615384615385, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 870}}}, "reference_key_id": "keys/synthetic/syn_03018_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "b93ce255451151cce89ccd51cc89c5b4382f883b441b3818bc249f2f79c54d9e", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "18f7c5ce65c29c384f550c7d", "attempt_id": "8e0e422cc2d7", "run_group_id": "E4_core:A5|closed|qwen3b:h8", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:11:31.985599Z", "task_family": "synthetic", "document_id": "syn_03018_2000", "document_hash": "d284eb07e27da742b9c2b67abccfdf72e4dd67c3259f44f3b41e34dd3e6fc1ff", "question_id": "q11_aggregation", "split": "test", "cluster_id": "syn_03018_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 8, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1982, "query_tokens": 495, "message_tokens": 401, "message_utf8_bytes": 940, "cache_tensor_bytes": null, "serialized_payload_bytes": 940, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 147844.79250002187, "hops": {"1": 22800.499332952313, "2": 18183.315165922977, "3": 17950.87779208552, "4": 17845.332500059158, "5": 17775.58920904994, "6": 17719.9630420655, "7": 17847.403332940303, "8": 17721.812124946155}, "receiver_prefill_ms": 325.5466250702739, "ttft_ms": 37.71033394150436, "decode_ms": 226.74983297474682}, "total_wall_ms": 148435.48404204194, "ttft_ms": 37.71033394150436, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 7}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "3189245", "evidence_artifact": {"ledger": {"facts_total": 64, "facts_retained": 14, "retention_rate": 0.21875, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.2222222222222222, "amount": 0.15384615384615385, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 870}, "ledger_by_hop": {"1": {"facts_total": 64, "facts_retained": 14, "retention_rate": 0.21875, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.2222222222222222, "amount": 0.15384615384615385, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 870}, "2": {"facts_total": 64, "facts_retained": 14, "retention_rate": 0.21875, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.2222222222222222, "amount": 0.15384615384615385, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 870}, "3": {"facts_total": 64, "facts_retained": 14, "retention_rate": 0.21875, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.2222222222222222, "amount": 0.15384615384615385, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 870}, "4": {"facts_total": 64, "facts_retained": 14, "retention_rate": 0.21875, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.2222222222222222, "amount": 0.15384615384615385, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 870}, "5": {"facts_total": 64, "facts_retained": 14, "retention_rate": 0.21875, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.2222222222222222, "amount": 0.15384615384615385, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 870}, "6": {"facts_total": 64, "facts_retained": 14, "retention_rate": 0.21875, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.2222222222222222, "amount": 0.15384615384615385, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 870}, "7": {"facts_total": 64, "facts_retained": 14, "retention_rate": 0.21875, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.2222222222222222, "amount": 0.15384615384615385, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 870}, "8": {"facts_total": 64, "facts_retained": 14, "retention_rate": 0.21875, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.2222222222222222, "amount": 0.15384615384615385, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 870}}}, "reference_key_id": "keys/synthetic/syn_03018_2000.json", "task_scores": {"correct": 0.0, "abstained": 0.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "b93ce255451151cce89ccd51cc89c5b4382f883b441b3818bc249f2f79c54d9e", "output_hash": "82576d93a43637ab2b22d213df62e73fbc3c84f64f177d56d7fd57c0cee09155"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "bfdd7953343d48216e3c45be", "attempt_id": "ccef1bebec85", "run_group_id": "E4_core:A5|closed|qwen3b:h8", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:11:32.593743Z", "task_family": "synthetic", "document_id": "syn_03018_2000", "document_hash": "d284eb07e27da742b9c2b67abccfdf72e4dd67c3259f44f3b41e34dd3e6fc1ff", "question_id": "q12_unit_conversion", "split": "test", "cluster_id": "syn_03018_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 8, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1982, "query_tokens": 492, "message_tokens": 401, "message_utf8_bytes": 940, "cache_tensor_bytes": null, "serialized_payload_bytes": 940, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 147844.79250002187, "hops": {"1": 22800.499332952313, "2": 18183.315165922977, "3": 17950.87779208552, "4": 17845.332500059158, "5": 17775.58920904994, "6": 17719.9630420655, "7": 17847.403332940303, "8": 17721.812124946155}, "receiver_prefill_ms": 402.01079205144197, "ttft_ms": 65.97420899197459, "decode_ms": 102.35758393537253}, "total_wall_ms": 148415.94158299267, "ttft_ms": 65.97420899197459, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 64, "facts_retained": 14, "retention_rate": 0.21875, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.2222222222222222, "amount": 0.15384615384615385, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 870}, "ledger_by_hop": {"1": {"facts_total": 64, "facts_retained": 14, "retention_rate": 0.21875, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.2222222222222222, "amount": 0.15384615384615385, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 870}, "2": {"facts_total": 64, "facts_retained": 14, "retention_rate": 0.21875, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.2222222222222222, "amount": 0.15384615384615385, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 870}, "3": {"facts_total": 64, "facts_retained": 14, "retention_rate": 0.21875, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.2222222222222222, "amount": 0.15384615384615385, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 870}, "4": {"facts_total": 64, "facts_retained": 14, "retention_rate": 0.21875, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.2222222222222222, "amount": 0.15384615384615385, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 870}, "5": {"facts_total": 64, "facts_retained": 14, "retention_rate": 0.21875, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.2222222222222222, "amount": 0.15384615384615385, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 870}, "6": {"facts_total": 64, "facts_retained": 14, "retention_rate": 0.21875, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.2222222222222222, "amount": 0.15384615384615385, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 870}, "7": {"facts_total": 64, "facts_retained": 14, "retention_rate": 0.21875, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.2222222222222222, "amount": 0.15384615384615385, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 870}, "8": {"facts_total": 64, "facts_retained": 14, "retention_rate": 0.21875, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.2222222222222222, "amount": 0.15384615384615385, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 870}}}, "reference_key_id": "keys/synthetic/syn_03018_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "b93ce255451151cce89ccd51cc89c5b4382f883b441b3818bc249f2f79c54d9e", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "6d8eeaf5b875ba4b1479d8a7", "attempt_id": "1208c229003b", "run_group_id": "E4_core:A5|closed|qwen3b:h8", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:11:33.182477Z", "task_family": "synthetic", "document_id": "syn_03018_2000", "document_hash": "d284eb07e27da742b9c2b67abccfdf72e4dd67c3259f44f3b41e34dd3e6fc1ff", "question_id": "q13_unanswerable", "split": "test", "cluster_id": "syn_03018_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 8, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1982, "query_tokens": 496, "message_tokens": 401, "message_utf8_bytes": 940, "cache_tensor_bytes": null, "serialized_payload_bytes": 940, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 147844.79250002187, "hops": {"1": 22800.499332952313, "2": 18183.315165922977, "3": 17950.87779208552, "4": 17845.332500059158, "5": 17775.58920904994, "6": 17719.9630420655, "7": 17847.403332940303, "8": 17721.812124946155}, "receiver_prefill_ms": 344.4786670152098, "ttft_ms": 22.850875044241548, "decode_ms": 41.163875022903085}, "total_wall_ms": 148253.99295799434, "ttft_ms": 22.850875044241548, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 64, "facts_retained": 14, "retention_rate": 0.21875, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.2222222222222222, "amount": 0.15384615384615385, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 870}, "ledger_by_hop": {"1": {"facts_total": 64, "facts_retained": 14, "retention_rate": 0.21875, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.2222222222222222, "amount": 0.15384615384615385, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 870}, "2": {"facts_total": 64, "facts_retained": 14, "retention_rate": 0.21875, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.2222222222222222, "amount": 0.15384615384615385, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 870}, "3": {"facts_total": 64, "facts_retained": 14, "retention_rate": 0.21875, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.2222222222222222, "amount": 0.15384615384615385, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 870}, "4": {"facts_total": 64, "facts_retained": 14, "retention_rate": 0.21875, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.2222222222222222, "amount": 0.15384615384615385, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 870}, "5": {"facts_total": 64, "facts_retained": 14, "retention_rate": 0.21875, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.2222222222222222, "amount": 0.15384615384615385, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 870}, "6": {"facts_total": 64, "facts_retained": 14, "retention_rate": 0.21875, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.2222222222222222, "amount": 0.15384615384615385, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 870}, "7": {"facts_total": 64, "facts_retained": 14, "retention_rate": 0.21875, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.2222222222222222, "amount": 0.15384615384615385, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 870}, "8": {"facts_total": 64, "facts_retained": 14, "retention_rate": 0.21875, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.2222222222222222, "amount": 0.15384615384615385, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 870}}}, "reference_key_id": "keys/synthetic/syn_03018_2000.json", "task_scores": {"correct": 1.0, "abstained": 1.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "b93ce255451151cce89ccd51cc89c5b4382f883b441b3818bc249f2f79c54d9e", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "938d65b2c75ecb9dd754343f", "attempt_id": "b97fd126ce9a", "run_group_id": "E4_core:A5|closed|qwen3b:h8", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:11:33.608858Z", "task_family": "synthetic", "document_id": "syn_03018_2000", "document_hash": "d284eb07e27da742b9c2b67abccfdf72e4dd67c3259f44f3b41e34dd3e6fc1ff", "question_id": "q14_unanswerable", "split": "test", "cluster_id": "syn_03018_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 8, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1982, "query_tokens": 497, "message_tokens": 401, "message_utf8_bytes": 940, "cache_tensor_bytes": null, "serialized_payload_bytes": 940, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 147844.79250002187, "hops": {"1": 22800.499332952313, "2": 18183.315165922977, "3": 17950.87779208552, "4": 17845.332500059158, "5": 17775.58920904994, "6": 17719.9630420655, "7": 17847.403332940303, "8": 17721.812124946155}, "receiver_prefill_ms": 325.0279169296846, "ttft_ms": 25.843124953098595, "decode_ms": 38.782082963734865}, "total_wall_ms": 148234.9727920955, "ttft_ms": 25.843124953098595, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 64, "facts_retained": 14, "retention_rate": 0.21875, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.2222222222222222, "amount": 0.15384615384615385, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 870}, "ledger_by_hop": {"1": {"facts_total": 64, "facts_retained": 14, "retention_rate": 0.21875, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.2222222222222222, "amount": 0.15384615384615385, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 870}, "2": {"facts_total": 64, "facts_retained": 14, "retention_rate": 0.21875, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.2222222222222222, "amount": 0.15384615384615385, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 870}, "3": {"facts_total": 64, "facts_retained": 14, "retention_rate": 0.21875, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.2222222222222222, "amount": 0.15384615384615385, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 870}, "4": {"facts_total": 64, "facts_retained": 14, "retention_rate": 0.21875, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.2222222222222222, "amount": 0.15384615384615385, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 870}, "5": {"facts_total": 64, "facts_retained": 14, "retention_rate": 0.21875, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.2222222222222222, "amount": 0.15384615384615385, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 870}, "6": {"facts_total": 64, "facts_retained": 14, "retention_rate": 0.21875, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.2222222222222222, "amount": 0.15384615384615385, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 870}, "7": {"facts_total": 64, "facts_retained": 14, "retention_rate": 0.21875, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.2222222222222222, "amount": 0.15384615384615385, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 870}, "8": {"facts_total": 64, "facts_retained": 14, "retention_rate": 0.21875, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.2222222222222222, "amount": 0.15384615384615385, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 870}}}, "reference_key_id": "keys/synthetic/syn_03018_2000.json", "task_scores": {"correct": 1.0, "abstained": 1.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "b93ce255451151cce89ccd51cc89c5b4382f883b441b3818bc249f2f79c54d9e", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "fcf4b9760ceec60359295938", "attempt_id": "1a186c5ae436", "run_group_id": "E4_core:A4|repairable|qwen3b:h1", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:11:56.898414Z", "task_family": "synthetic", "document_id": "syn_03018_2000", "document_hash": "d284eb07e27da742b9c2b67abccfdf72e4dd67c3259f44f3b41e34dd3e6fc1ff", "question_id": "q00_single_fact", "split": "test", "cluster_id": "syn_03018_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 1, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1982, "query_tokens": 595, "message_tokens": 512, "message_utf8_bytes": 1181, "cache_tensor_bytes": null, "serialized_payload_bytes": 1181, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 22880.75100001879, "hops": {"1": 22880.75100001879}, "receiver_prefill_ms": 397.9618340963498, "ttft_ms": 43.49920793902129, "decode_ms": 25.301082991063595}, "total_wall_ms": 23348.17075007595, "ttft_ms": 43.49920793902129, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 64, "facts_retained": 22, "retention_rate": 0.34375, "retention_by_kind": {"person": 0.5, "date": 0.5555555555555556, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1181}, "ledger_by_hop": {"1": {"facts_total": 64, "facts_retained": 22, "retention_rate": 0.34375, "retention_by_kind": {"person": 0.5, "date": 0.5555555555555556, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1181}}}, "reference_key_id": "keys/synthetic/syn_03018_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "499392c3597fdf0fd842c7907d9598f7d57a2cf7b9c4348cfba91a80adac26ec", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "902fe2ee70c23012fca2c503", "attempt_id": "0b8716669c2b", "run_group_id": "E4_core:A4|repairable|qwen3b:h1", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:11:57.383462Z", "task_family": "synthetic", "document_id": "syn_03018_2000", "document_hash": "d284eb07e27da742b9c2b67abccfdf72e4dd67c3259f44f3b41e34dd3e6fc1ff", "question_id": "q01_single_fact", "split": "test", "cluster_id": "syn_03018_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 1, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1982, "query_tokens": 601, "message_tokens": 512, "message_utf8_bytes": 1181, "cache_tensor_bytes": null, "serialized_payload_bytes": 1181, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 22880.75100001879, "hops": {"1": 22880.75100001879}, "receiver_prefill_ms": 399.47366702836007, "ttft_ms": 32.21099998336285, "decode_ms": 561.5970829967409}, "total_wall_ms": 23874.724874971434, "ttft_ms": 32.21099998336285, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 12}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "Lot 376-E, Baie-Comeau", "evidence_artifact": {"ledger": {"facts_total": 64, "facts_retained": 22, "retention_rate": 0.34375, "retention_by_kind": {"person": 0.5, "date": 0.5555555555555556, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1181}, "ledger_by_hop": {"1": {"facts_total": 64, "facts_retained": 22, "retention_rate": 0.34375, "retention_by_kind": {"person": 0.5, "date": 0.5555555555555556, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1181}}}, "reference_key_id": "keys/synthetic/syn_03018_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 0.8571428571428571}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "499392c3597fdf0fd842c7907d9598f7d57a2cf7b9c4348cfba91a80adac26ec", "output_hash": "b58829a1440993c4af3307ba44494f021d726fd11e9337888b6b852a8b37351a"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "3378dd24dfd98a1051bf90e3", "attempt_id": "88c25012bbf8", "run_group_id": "E4_core:A4|repairable|qwen3b:h1", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:11:58.394122Z", "task_family": "synthetic", "document_id": "syn_03018_2000", "document_hash": "d284eb07e27da742b9c2b67abccfdf72e4dd67c3259f44f3b41e34dd3e6fc1ff", "question_id": "q02_exact_copy", "split": "test", "cluster_id": "syn_03018_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 1, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1982, "query_tokens": 600, "message_tokens": 512, "message_utf8_bytes": 1181, "cache_tensor_bytes": null, "serialized_payload_bytes": 1181, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 22880.75100001879, "hops": {"1": 22880.75100001879}, "receiver_prefill_ms": 472.8973750025034, "ttft_ms": 25.409082998521626, "decode_ms": 39.106959011405706}, "total_wall_ms": 23418.783292057924, "ttft_ms": 25.409082998521626, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 64, "facts_retained": 22, "retention_rate": 0.34375, "retention_by_kind": {"person": 0.5, "date": 0.5555555555555556, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1181}, "ledger_by_hop": {"1": {"facts_total": 64, "facts_retained": 22, "retention_rate": 0.34375, "retention_by_kind": {"person": 0.5, "date": 0.5555555555555556, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1181}}}, "reference_key_id": "keys/synthetic/syn_03018_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "499392c3597fdf0fd842c7907d9598f7d57a2cf7b9c4348cfba91a80adac26ec", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "3d6d85fcdf3d06543d4a9cb0", "attempt_id": "fe159c7d8d9d", "run_group_id": "E4_core:A4|repairable|qwen3b:h1", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:11:58.949567Z", "task_family": "synthetic", "document_id": "syn_03018_2000", "document_hash": "d284eb07e27da742b9c2b67abccfdf72e4dd67c3259f44f3b41e34dd3e6fc1ff", "question_id": "q03_exact_copy", "split": "test", "cluster_id": "syn_03018_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 1, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1982, "query_tokens": 598, "message_tokens": 512, "message_utf8_bytes": 1181, "cache_tensor_bytes": null, "serialized_payload_bytes": 1181, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 22880.75100001879, "hops": {"1": 22880.75100001879}, "receiver_prefill_ms": 411.4051670767367, "ttft_ms": 27.219625073485076, "decode_ms": 381.33991707582027}, "total_wall_ms": 23701.6739170067, "ttft_ms": 27.219625073485076, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 9}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "Priyamvada Sørlie", "evidence_artifact": {"ledger": {"facts_total": 64, "facts_retained": 22, "retention_rate": 0.34375, "retention_by_kind": {"person": 0.5, "date": 0.5555555555555556, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1181}, "ledger_by_hop": {"1": {"facts_total": 64, "facts_retained": 22, "retention_rate": 0.34375, "retention_by_kind": {"person": 0.5, "date": 0.5555555555555556, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1181}}}, "reference_key_id": "keys/synthetic/syn_03018_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "499392c3597fdf0fd842c7907d9598f7d57a2cf7b9c4348cfba91a80adac26ec", "output_hash": "c5f554196ed23dd71ad995ec89f7142099d524dbb6534cc10d94a0c89d13e792"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "e682bf028567cf38b24eade4", "attempt_id": "155c78a0b4ab", "run_group_id": "E4_core:A4|repairable|qwen3b:h1", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:11:59.788449Z", "task_family": "synthetic", "document_id": "syn_03018_2000", "document_hash": "d284eb07e27da742b9c2b67abccfdf72e4dd67c3259f44f3b41e34dd3e6fc1ff", "question_id": "q04_multi_hop", "split": "test", "cluster_id": "syn_03018_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 1, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1982, "query_tokens": 604, "message_tokens": 512, "message_utf8_bytes": 1181, "cache_tensor_bytes": null, "serialized_payload_bytes": 1181, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 22880.75100001879, "hops": {"1": 22880.75100001879}, "receiver_prefill_ms": 519.1848750691861, "ttft_ms": 36.041125073097646, "decode_ms": 26.578958029858768}, "total_wall_ms": 23463.25512509793, "ttft_ms": 36.041125073097646, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 64, "facts_retained": 22, "retention_rate": 0.34375, "retention_by_kind": {"person": 0.5, "date": 0.5555555555555556, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1181}, "ledger_by_hop": {"1": {"facts_total": 64, "facts_retained": 22, "retention_rate": 0.34375, "retention_by_kind": {"person": 0.5, "date": 0.5555555555555556, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1181}}}, "reference_key_id": "keys/synthetic/syn_03018_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "499392c3597fdf0fd842c7907d9598f7d57a2cf7b9c4348cfba91a80adac26ec", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "c3bd6e89beaa57efd37aed0d", "attempt_id": "d15000ec35fc", "run_group_id": "E4_core:A4|repairable|qwen3b:h1", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:12:00.387985Z", "task_family": "synthetic", "document_id": "syn_03018_2000", "document_hash": "d284eb07e27da742b9c2b67abccfdf72e4dd67c3259f44f3b41e34dd3e6fc1ff", "question_id": "q05_multi_hop", "split": "test", "cluster_id": "syn_03018_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 1, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1982, "query_tokens": 604, "message_tokens": 512, "message_utf8_bytes": 1181, "cache_tensor_bytes": null, "serialized_payload_bytes": 1181, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 22880.75100001879, "hops": {"1": 22880.75100001879}, "receiver_prefill_ms": 402.9275419889018, "ttft_ms": 42.70679200999439, "decode_ms": 326.09912496991456}, "total_wall_ms": 23653.401917079464, "ttft_ms": 42.70679200999439, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 10}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "HM-38K-5533", "evidence_artifact": {"ledger": {"facts_total": 64, "facts_retained": 22, "retention_rate": 0.34375, "retention_by_kind": {"person": 0.5, "date": 0.5555555555555556, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1181}, "ledger_by_hop": {"1": {"facts_total": 64, "facts_retained": 22, "retention_rate": 0.34375, "retention_by_kind": {"person": 0.5, "date": 0.5555555555555556, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1181}}}, "reference_key_id": "keys/synthetic/syn_03018_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "499392c3597fdf0fd842c7907d9598f7d57a2cf7b9c4348cfba91a80adac26ec", "output_hash": "18acd9807857243bbded3bceae55398316a3958b81d70364441bdb6748eaf878"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "aa1d76c1cc226f2888ec8bf2", "attempt_id": "300563a6e9dd", "run_group_id": "E4_core:A4|repairable|qwen3b:h1", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:12:01.177977Z", "task_family": "synthetic", "document_id": "syn_03018_2000", "document_hash": "d284eb07e27da742b9c2b67abccfdf72e4dd67c3259f44f3b41e34dd3e6fc1ff", "question_id": "q06_negation", "split": "test", "cluster_id": "syn_03018_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 1, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1982, "query_tokens": 596, "message_tokens": 512, "message_utf8_bytes": 1181, "cache_tensor_bytes": null, "serialized_payload_bytes": 1181, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 22880.75100001879, "hops": {"1": 22880.75100001879}, "receiver_prefill_ms": 369.70779206603765, "ttft_ms": 37.9831250756979, "decode_ms": 117.15029191691428}, "total_wall_ms": 23406.362624955364, "ttft_ms": 37.9831250756979, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 64, "facts_retained": 22, "retention_rate": 0.34375, "retention_by_kind": {"person": 0.5, "date": 0.5555555555555556, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1181}, "ledger_by_hop": {"1": {"facts_total": 64, "facts_retained": 22, "retention_rate": 0.34375, "retention_by_kind": {"person": 0.5, "date": 0.5555555555555556, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1181}}}, "reference_key_id": "keys/synthetic/syn_03018_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "499392c3597fdf0fd842c7907d9598f7d57a2cf7b9c4348cfba91a80adac26ec", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "319403d4362fbf818d2bdac9", "attempt_id": "0e473c3d26f2", "run_group_id": "E4_core:A4|repairable|qwen3b:h1", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:12:01.720759Z", "task_family": "synthetic", "document_id": "syn_03018_2000", "document_hash": "d284eb07e27da742b9c2b67abccfdf72e4dd67c3259f44f3b41e34dd3e6fc1ff", "question_id": "q07_negation", "split": "test", "cluster_id": "syn_03018_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 1, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1982, "query_tokens": 595, "message_tokens": 512, "message_utf8_bytes": 1181, "cache_tensor_bytes": null, "serialized_payload_bytes": 1181, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 22880.75100001879, "hops": {"1": 22880.75100001879}, "receiver_prefill_ms": 443.11383401509374, "ttft_ms": 26.3521249871701, "decode_ms": 36.35404200758785}, "total_wall_ms": 23387.136541074142, "ttft_ms": 26.3521249871701, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "no", "evidence_artifact": {"ledger": {"facts_total": 64, "facts_retained": 22, "retention_rate": 0.34375, "retention_by_kind": {"person": 0.5, "date": 0.5555555555555556, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1181}, "ledger_by_hop": {"1": {"facts_total": 64, "facts_retained": 22, "retention_rate": 0.34375, "retention_by_kind": {"person": 0.5, "date": 0.5555555555555556, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1181}}}, "reference_key_id": "keys/synthetic/syn_03018_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "499392c3597fdf0fd842c7907d9598f7d57a2cf7b9c4348cfba91a80adac26ec", "output_hash": "9390298f3fb0c5b160498935d79cb139aef28e1c47358b4bbba61862b9c26e59"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "9774a17acb664705c405a4ae", "attempt_id": "1eee59998482", "run_group_id": "E4_core:A4|repairable|qwen3b:h1", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:12:02.244525Z", "task_family": "synthetic", "document_id": "syn_03018_2000", "document_hash": "d284eb07e27da742b9c2b67abccfdf72e4dd67c3259f44f3b41e34dd3e6fc1ff", "question_id": "q08_revision", "split": "test", "cluster_id": "syn_03018_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 1, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1982, "query_tokens": 613, "message_tokens": 512, "message_utf8_bytes": 1181, "cache_tensor_bytes": null, "serialized_payload_bytes": 1181, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 22880.75100001879, "hops": {"1": 22880.75100001879}, "receiver_prefill_ms": 400.62687499448657, "ttft_ms": 29.11037509329617, "decode_ms": 167.38637501839548}, "total_wall_ms": 23478.899416979402, "ttft_ms": 29.11037509329617, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 6}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "304815", "evidence_artifact": {"ledger": {"facts_total": 64, "facts_retained": 22, "retention_rate": 0.34375, "retention_by_kind": {"person": 0.5, "date": 0.5555555555555556, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1181}, "ledger_by_hop": {"1": {"facts_total": 64, "facts_retained": 22, "retention_rate": 0.34375, "retention_by_kind": {"person": 0.5, "date": 0.5555555555555556, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1181}}}, "reference_key_id": "keys/synthetic/syn_03018_2000.json", "task_scores": {"correct": 0.0, "abstained": 0.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "499392c3597fdf0fd842c7907d9598f7d57a2cf7b9c4348cfba91a80adac26ec", "output_hash": "a480475d6855141c920f42d1bf3337ca7fd88bcf4087b0ed2e253ed68cca6926"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "e558f1517ee8e9ca3901fc0a", "attempt_id": "a8bd91222717", "run_group_id": "E4_core:A4|repairable|qwen3b:h1", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:12:02.860107Z", "task_family": "synthetic", "document_id": "syn_03018_2000", "document_hash": "d284eb07e27da742b9c2b67abccfdf72e4dd67c3259f44f3b41e34dd3e6fc1ff", "question_id": "q09_revision", "split": "test", "cluster_id": "syn_03018_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 1, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1982, "query_tokens": 606, "message_tokens": 512, "message_utf8_bytes": 1181, "cache_tensor_bytes": null, "serialized_payload_bytes": 1181, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 22880.75100001879, "hops": {"1": 22880.75100001879}, "receiver_prefill_ms": 428.6108750384301, "ttft_ms": 78.66312505211681, "decode_ms": 297.59412503335625}, "total_wall_ms": 23686.27787497826, "ttft_ms": 78.66312505211681, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 6}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "276340", "evidence_artifact": {"ledger": {"facts_total": 64, "facts_retained": 22, "retention_rate": 0.34375, "retention_by_kind": {"person": 0.5, "date": 0.5555555555555556, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1181}, "ledger_by_hop": {"1": {"facts_total": 64, "facts_retained": 22, "retention_rate": 0.34375, "retention_by_kind": {"person": 0.5, "date": 0.5555555555555556, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1181}}}, "reference_key_id": "keys/synthetic/syn_03018_2000.json", "task_scores": {"correct": 0.0, "abstained": 0.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "499392c3597fdf0fd842c7907d9598f7d57a2cf7b9c4348cfba91a80adac26ec", "output_hash": "bc7c3667cbe0ad74ec0a1b092d3bad7e6a02978fa67746e3adbf3174ed34b3a2"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "dc0364a55b38f5ef4b71be44", "attempt_id": "1293d5f79456", "run_group_id": "E4_core:A4|repairable|qwen3b:h1", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:12:03.683646Z", "task_family": "synthetic", "document_id": "syn_03018_2000", "document_hash": "d284eb07e27da742b9c2b67abccfdf72e4dd67c3259f44f3b41e34dd3e6fc1ff", "question_id": "q10_temporal", "split": "test", "cluster_id": "syn_03018_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 1, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1982, "query_tokens": 614, "message_tokens": 512, "message_utf8_bytes": 1181, "cache_tensor_bytes": null, "serialized_payload_bytes": 1181, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 22880.75100001879, "hops": {"1": 22880.75100001879}, "receiver_prefill_ms": 404.27429194096476, "ttft_ms": 41.301249992102385, "decode_ms": 188.88237501960248}, "total_wall_ms": 23515.889583039097, "ttft_ms": 41.301249992102385, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 6}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "Melymi-15", "evidence_artifact": {"ledger": {"facts_total": 64, "facts_retained": 22, "retention_rate": 0.34375, "retention_by_kind": {"person": 0.5, "date": 0.5555555555555556, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1181}, "ledger_by_hop": {"1": {"facts_total": 64, "facts_retained": 22, "retention_rate": 0.34375, "retention_by_kind": {"person": 0.5, "date": 0.5555555555555556, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1181}}}, "reference_key_id": "keys/synthetic/syn_03018_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "499392c3597fdf0fd842c7907d9598f7d57a2cf7b9c4348cfba91a80adac26ec", "output_hash": "fa08d1474b01d6dfbf3dea049be251aa3c3021c17b4818258064b156509af58b"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "d1f3ec898e309769141f50ee", "attempt_id": "8ee4e62ec461", "run_group_id": "E4_core:A4|repairable|qwen3b:h1", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:12:04.336003Z", "task_family": "synthetic", "document_id": "syn_03018_2000", "document_hash": "d284eb07e27da742b9c2b67abccfdf72e4dd67c3259f44f3b41e34dd3e6fc1ff", "question_id": "q11_aggregation", "split": "test", "cluster_id": "syn_03018_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 1, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1982, "query_tokens": 606, "message_tokens": 512, "message_utf8_bytes": 1181, "cache_tensor_bytes": null, "serialized_payload_bytes": 1181, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 22880.75100001879, "hops": {"1": 22880.75100001879}, "receiver_prefill_ms": 363.7227499857545, "ttft_ms": 40.477040922269225, "decode_ms": 467.1287910314277}, "total_wall_ms": 23752.7894169325, "ttft_ms": 40.477040922269225, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 7}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "2414120", "evidence_artifact": {"ledger": {"facts_total": 64, "facts_retained": 22, "retention_rate": 0.34375, "retention_by_kind": {"person": 0.5, "date": 0.5555555555555556, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1181}, "ledger_by_hop": {"1": {"facts_total": 64, "facts_retained": 22, "retention_rate": 0.34375, "retention_by_kind": {"person": 0.5, "date": 0.5555555555555556, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1181}}}, "reference_key_id": "keys/synthetic/syn_03018_2000.json", "task_scores": {"correct": 0.0, "abstained": 0.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "499392c3597fdf0fd842c7907d9598f7d57a2cf7b9c4348cfba91a80adac26ec", "output_hash": "0a242f83095c1c1c7e4a3cd4bb88035f03abb655b6cf03cf7962945bb269a5e8"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "30b05b6c5165d20cdb821d7c", "attempt_id": "6f3d1fa64621", "run_group_id": "E4_core:A4|repairable|qwen3b:h1", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:12:05.226111Z", "task_family": "synthetic", "document_id": "syn_03018_2000", "document_hash": "d284eb07e27da742b9c2b67abccfdf72e4dd67c3259f44f3b41e34dd3e6fc1ff", "question_id": "q12_unit_conversion", "split": "test", "cluster_id": "syn_03018_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 1, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1982, "query_tokens": 603, "message_tokens": 512, "message_utf8_bytes": 1181, "cache_tensor_bytes": null, "serialized_payload_bytes": 1181, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 22880.75100001879, "hops": {"1": 22880.75100001879}, "receiver_prefill_ms": 397.3505840403959, "ttft_ms": 26.323000085540116, "decode_ms": 38.21720799896866}, "total_wall_ms": 23343.22316700127, "ttft_ms": 26.323000085540116, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 64, "facts_retained": 22, "retention_rate": 0.34375, "retention_by_kind": {"person": 0.5, "date": 0.5555555555555556, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1181}, "ledger_by_hop": {"1": {"facts_total": 64, "facts_retained": 22, "retention_rate": 0.34375, "retention_by_kind": {"person": 0.5, "date": 0.5555555555555556, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1181}}}, "reference_key_id": "keys/synthetic/syn_03018_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "499392c3597fdf0fd842c7907d9598f7d57a2cf7b9c4348cfba91a80adac26ec", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "ca1a7f254521233faffa0629", "attempt_id": "e2e109e97eda", "run_group_id": "E4_core:A4|repairable|qwen3b:h1", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:12:05.705447Z", "task_family": "synthetic", "document_id": "syn_03018_2000", "document_hash": "d284eb07e27da742b9c2b67abccfdf72e4dd67c3259f44f3b41e34dd3e6fc1ff", "question_id": "q13_unanswerable", "split": "test", "cluster_id": "syn_03018_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 1, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1982, "query_tokens": 607, "message_tokens": 512, "message_utf8_bytes": 1181, "cache_tensor_bytes": null, "serialized_payload_bytes": 1181, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 22880.75100001879, "hops": {"1": 22880.75100001879}, "receiver_prefill_ms": 404.7868329798803, "ttft_ms": 24.982666946016252, "decode_ms": 19.60270805284381}, "total_wall_ms": 23330.695792101324, "ttft_ms": 24.982666946016252, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 64, "facts_retained": 22, "retention_rate": 0.34375, "retention_by_kind": {"person": 0.5, "date": 0.5555555555555556, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1181}, "ledger_by_hop": {"1": {"facts_total": 64, "facts_retained": 22, "retention_rate": 0.34375, "retention_by_kind": {"person": 0.5, "date": 0.5555555555555556, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1181}}}, "reference_key_id": "keys/synthetic/syn_03018_2000.json", "task_scores": {"correct": 1.0, "abstained": 1.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "499392c3597fdf0fd842c7907d9598f7d57a2cf7b9c4348cfba91a80adac26ec", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "64c4862952b6db7ab068e1c7", "attempt_id": "9a95bb26dbf1", "run_group_id": "E4_core:A4|repairable|qwen3b:h1", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:12:06.172463Z", "task_family": "synthetic", "document_id": "syn_03018_2000", "document_hash": "d284eb07e27da742b9c2b67abccfdf72e4dd67c3259f44f3b41e34dd3e6fc1ff", "question_id": "q14_unanswerable", "split": "test", "cluster_id": "syn_03018_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 1, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1982, "query_tokens": 608, "message_tokens": 512, "message_utf8_bytes": 1181, "cache_tensor_bytes": null, "serialized_payload_bytes": 1181, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 22880.75100001879, "hops": {"1": 22880.75100001879}, "receiver_prefill_ms": 406.81283397134393, "ttft_ms": 42.564457980915904, "decode_ms": 111.61920800805092}, "total_wall_ms": 23442.47800006997, "ttft_ms": 42.564457980915904, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 64, "facts_retained": 22, "retention_rate": 0.34375, "retention_by_kind": {"person": 0.5, "date": 0.5555555555555556, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1181}, "ledger_by_hop": {"1": {"facts_total": 64, "facts_retained": 22, "retention_rate": 0.34375, "retention_by_kind": {"person": 0.5, "date": 0.5555555555555556, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1181}}}, "reference_key_id": "keys/synthetic/syn_03018_2000.json", "task_scores": {"correct": 1.0, "abstained": 1.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "499392c3597fdf0fd842c7907d9598f7d57a2cf7b9c4348cfba91a80adac26ec", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "adc3477f9b9720a03767e796", "attempt_id": "1bf0ae3e0561", "run_group_id": "E4_core:A4|repairable|qwen3b:h2", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:12:25.912942Z", "task_family": "synthetic", "document_id": "syn_03018_2000", "document_hash": "d284eb07e27da742b9c2b67abccfdf72e4dd67c3259f44f3b41e34dd3e6fc1ff", "question_id": "q00_single_fact", "split": "test", "cluster_id": "syn_03018_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 2, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1982, "query_tokens": 578, "message_tokens": 495, "message_utf8_bytes": 1081, "cache_tensor_bytes": null, "serialized_payload_bytes": 1081, "side_information_bytes": 0, "fetched_bytes": 726, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 41957.59679109324, "hops": {"1": 22880.75100001879, "2": 19076.845791074447}, "receiver_prefill_ms": 226.7036250559613, "ttft_ms": 11.929291067644954, "decode_ms": 11.927374987863004}, "total_wall_ms": 42208.70708313305, "ttft_ms": 11.929291067644954, "input_usage": {"lookups": 4}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 64, "facts_retained": 22, "retention_rate": 0.34375, "retention_by_kind": {"person": 0.5, "date": 0.5555555555555556, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1081}, "ledger_by_hop": {"1": {"facts_total": 64, "facts_retained": 22, "retention_rate": 0.34375, "retention_by_kind": {"person": 0.5, "date": 0.5555555555555556, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1181}, "2": {"facts_total": 64, "facts_retained": 22, "retention_rate": 0.34375, "retention_by_kind": {"person": 0.5, "date": 0.5555555555555556, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1081}}}, "reference_key_id": "keys/synthetic/syn_03018_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "af090a9c821da67010b5031584205204273ce51579047d51345d5bf471ce81ec", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "c0889091751e8b7c62ebf74f", "attempt_id": "51847144fabd", "run_group_id": "E4_core:A4|repairable|qwen3b:h2", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:12:26.180316Z", "task_family": "synthetic", "document_id": "syn_03018_2000", "document_hash": "d284eb07e27da742b9c2b67abccfdf72e4dd67c3259f44f3b41e34dd3e6fc1ff", "question_id": "q01_single_fact", "split": "test", "cluster_id": "syn_03018_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 2, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1982, "query_tokens": 584, "message_tokens": 495, "message_utf8_bytes": 1081, "cache_tensor_bytes": null, "serialized_payload_bytes": 1081, "side_information_bytes": 0, "fetched_bytes": 726, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 41957.59679109324, "hops": {"1": 22880.75100001879, "2": 19076.845791074447}, "receiver_prefill_ms": 226.89920803532004, "ttft_ms": 11.941292090341449, "decode_ms": 142.81050011049956}, "total_wall_ms": 42339.752416010015, "ttft_ms": 11.941292090341449, "input_usage": {"lookups": 4}, "output_usage": {"output_tokens": 12}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "Lot 376-E, Baie-Comeau", "evidence_artifact": {"ledger": {"facts_total": 64, "facts_retained": 22, "retention_rate": 0.34375, "retention_by_kind": {"person": 0.5, "date": 0.5555555555555556, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1081}, "ledger_by_hop": {"1": {"facts_total": 64, "facts_retained": 22, "retention_rate": 0.34375, "retention_by_kind": {"person": 0.5, "date": 0.5555555555555556, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1181}, "2": {"facts_total": 64, "facts_retained": 22, "retention_rate": 0.34375, "retention_by_kind": {"person": 0.5, "date": 0.5555555555555556, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1081}}}, "reference_key_id": "keys/synthetic/syn_03018_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 0.8571428571428571}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "af090a9c821da67010b5031584205204273ce51579047d51345d5bf471ce81ec", "output_hash": "b58829a1440993c4af3307ba44494f021d726fd11e9337888b6b852a8b37351a"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "da04e883f2c0d956a4239fbf", "attempt_id": "eb097ad2262a", "run_group_id": "E4_core:A4|repairable|qwen3b:h2", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:12:26.577886Z", "task_family": "synthetic", "document_id": "syn_03018_2000", "document_hash": "d284eb07e27da742b9c2b67abccfdf72e4dd67c3259f44f3b41e34dd3e6fc1ff", "question_id": "q02_exact_copy", "split": "test", "cluster_id": "syn_03018_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 2, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1982, "query_tokens": 583, "message_tokens": 495, "message_utf8_bytes": 1081, "cache_tensor_bytes": null, "serialized_payload_bytes": 1081, "side_information_bytes": 0, "fetched_bytes": 726, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 41957.59679109324, "hops": {"1": 22880.75100001879, "2": 19076.845791074447}, "receiver_prefill_ms": 226.29070898983628, "ttft_ms": 12.075958075001836, "decode_ms": 12.217875104397535}, "total_wall_ms": 42208.78162514418, "ttft_ms": 12.075958075001836, "input_usage": {"lookups": 4}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 64, "facts_retained": 22, "retention_rate": 0.34375, "retention_by_kind": {"person": 0.5, "date": 0.5555555555555556, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1081}, "ledger_by_hop": {"1": {"facts_total": 64, "facts_retained": 22, "retention_rate": 0.34375, "retention_by_kind": {"person": 0.5, "date": 0.5555555555555556, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1181}, "2": {"facts_total": 64, "facts_retained": 22, "retention_rate": 0.34375, "retention_by_kind": {"person": 0.5, "date": 0.5555555555555556, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1081}}}, "reference_key_id": "keys/synthetic/syn_03018_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "af090a9c821da67010b5031584205204273ce51579047d51345d5bf471ce81ec", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "c5ee04c22b814a9da737a23f", "attempt_id": "6278519c46bf", "run_group_id": "E4_core:A4|repairable|qwen3b:h2", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:12:26.844760Z", "task_family": "synthetic", "document_id": "syn_03018_2000", "document_hash": "d284eb07e27da742b9c2b67abccfdf72e4dd67c3259f44f3b41e34dd3e6fc1ff", "question_id": "q03_exact_copy", "split": "test", "cluster_id": "syn_03018_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 2, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1982, "query_tokens": 581, "message_tokens": 495, "message_utf8_bytes": 1081, "cache_tensor_bytes": null, "serialized_payload_bytes": 1081, "side_information_bytes": 0, "fetched_bytes": 726, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 41957.59679109324, "hops": {"1": 22880.75100001879, "2": 19076.845791074447}, "receiver_prefill_ms": 226.42762504983693, "ttft_ms": 12.186916079372168, "decode_ms": 107.09191602654755}, "total_wall_ms": 42303.81891608704, "ttft_ms": 12.186916079372168, "input_usage": {"lookups": 4}, "output_usage": {"output_tokens": 9}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "Priyamvada Sørlie", "evidence_artifact": {"ledger": {"facts_total": 64, "facts_retained": 22, "retention_rate": 0.34375, "retention_by_kind": {"person": 0.5, "date": 0.5555555555555556, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1081}, "ledger_by_hop": {"1": {"facts_total": 64, "facts_retained": 22, "retention_rate": 0.34375, "retention_by_kind": {"person": 0.5, "date": 0.5555555555555556, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1181}, "2": {"facts_total": 64, "facts_retained": 22, "retention_rate": 0.34375, "retention_by_kind": {"person": 0.5, "date": 0.5555555555555556, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1081}}}, "reference_key_id": "keys/synthetic/syn_03018_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "af090a9c821da67010b5031584205204273ce51579047d51345d5bf471ce81ec", "output_hash": "c5f554196ed23dd71ad995ec89f7142099d524dbb6534cc10d94a0c89d13e792"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "e8f16eb418289e86fb935827", "attempt_id": "b6366482dcc3", "run_group_id": "E4_core:A4|repairable|qwen3b:h2", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:12:27.207406Z", "task_family": "synthetic", "document_id": "syn_03018_2000", "document_hash": "d284eb07e27da742b9c2b67abccfdf72e4dd67c3259f44f3b41e34dd3e6fc1ff", "question_id": "q04_multi_hop", "split": "test", "cluster_id": "syn_03018_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 2, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1982, "query_tokens": 587, "message_tokens": 495, "message_utf8_bytes": 1081, "cache_tensor_bytes": null, "serialized_payload_bytes": 1081, "side_information_bytes": 0, "fetched_bytes": 726, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 41957.59679109324, "hops": {"1": 22880.75100001879, "2": 19076.845791074447}, "receiver_prefill_ms": 226.8228749744594, "ttft_ms": 11.902958038263023, "decode_ms": 11.919458978809416}, "total_wall_ms": 42208.7246250594, "ttft_ms": 11.902958038263023, "input_usage": {"lookups": 4}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 64, "facts_retained": 22, "retention_rate": 0.34375, "retention_by_kind": {"person": 0.5, "date": 0.5555555555555556, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1081}, "ledger_by_hop": {"1": {"facts_total": 64, "facts_retained": 22, "retention_rate": 0.34375, "retention_by_kind": {"person": 0.5, "date": 0.5555555555555556, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1181}, "2": {"facts_total": 64, "facts_retained": 22, "retention_rate": 0.34375, "retention_by_kind": {"person": 0.5, "date": 0.5555555555555556, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1081}}}, "reference_key_id": "keys/synthetic/syn_03018_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "af090a9c821da67010b5031584205204273ce51579047d51345d5bf471ce81ec", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "d713cf59d62c83f6263824f4", "attempt_id": "4a31032306c1", "run_group_id": "E4_core:A4|repairable|qwen3b:h2", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:12:27.474182Z", "task_family": "synthetic", "document_id": "syn_03018_2000", "document_hash": "d284eb07e27da742b9c2b67abccfdf72e4dd67c3259f44f3b41e34dd3e6fc1ff", "question_id": "q05_multi_hop", "split": "test", "cluster_id": "syn_03018_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 2, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1982, "query_tokens": 587, "message_tokens": 495, "message_utf8_bytes": 1081, "cache_tensor_bytes": null, "serialized_payload_bytes": 1081, "side_information_bytes": 0, "fetched_bytes": 726, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 41957.59679109324, "hops": {"1": 22880.75100001879, "2": 19076.845791074447}, "receiver_prefill_ms": 226.64566698949784, "ttft_ms": 11.934167006984353, "decode_ms": 119.12450008094311}, "total_wall_ms": 42315.79204101581, "ttft_ms": 11.934167006984353, "input_usage": {"lookups": 4}, "output_usage": {"output_tokens": 10}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "HM-38K-5533", "evidence_artifact": {"ledger": {"facts_total": 64, "facts_retained": 22, "retention_rate": 0.34375, "retention_by_kind": {"person": 0.5, "date": 0.5555555555555556, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1081}, "ledger_by_hop": {"1": {"facts_total": 64, "facts_retained": 22, "retention_rate": 0.34375, "retention_by_kind": {"person": 0.5, "date": 0.5555555555555556, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1181}, "2": {"facts_total": 64, "facts_retained": 22, "retention_rate": 0.34375, "retention_by_kind": {"person": 0.5, "date": 0.5555555555555556, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1081}}}, "reference_key_id": "keys/synthetic/syn_03018_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "af090a9c821da67010b5031584205204273ce51579047d51345d5bf471ce81ec", "output_hash": "18acd9807857243bbded3bceae55398316a3958b81d70364441bdb6748eaf878"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "486a15c22f3b05033cf62920", "attempt_id": "1bbd8c34f91a", "run_group_id": "E4_core:A4|repairable|qwen3b:h2", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:12:27.848058Z", "task_family": "synthetic", "document_id": "syn_03018_2000", "document_hash": "d284eb07e27da742b9c2b67abccfdf72e4dd67c3259f44f3b41e34dd3e6fc1ff", "question_id": "q06_negation", "split": "test", "cluster_id": "syn_03018_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 2, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1982, "query_tokens": 579, "message_tokens": 495, "message_utf8_bytes": 1081, "cache_tensor_bytes": null, "serialized_payload_bytes": 1081, "side_information_bytes": 0, "fetched_bytes": 726, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 41957.59679109324, "hops": {"1": 22880.75100001879, "2": 19076.845791074447}, "receiver_prefill_ms": 226.47816699463874, "ttft_ms": 11.903417063876987, "decode_ms": 11.884624953381717}, "total_wall_ms": 42208.34962406661, "ttft_ms": 11.903417063876987, "input_usage": {"lookups": 4}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 64, "facts_retained": 22, "retention_rate": 0.34375, "retention_by_kind": {"person": 0.5, "date": 0.5555555555555556, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1081}, "ledger_by_hop": {"1": {"facts_total": 64, "facts_retained": 22, "retention_rate": 0.34375, "retention_by_kind": {"person": 0.5, "date": 0.5555555555555556, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1181}, "2": {"facts_total": 64, "facts_retained": 22, "retention_rate": 0.34375, "retention_by_kind": {"person": 0.5, "date": 0.5555555555555556, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1081}}}, "reference_key_id": "keys/synthetic/syn_03018_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "af090a9c821da67010b5031584205204273ce51579047d51345d5bf471ce81ec", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "5598c444e6a0ff970662cc61", "attempt_id": "c8f80042f0b4", "run_group_id": "E4_core:A4|repairable|qwen3b:h2", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:12:28.114024Z", "task_family": "synthetic", "document_id": "syn_03018_2000", "document_hash": "d284eb07e27da742b9c2b67abccfdf72e4dd67c3259f44f3b41e34dd3e6fc1ff", "question_id": "q07_negation", "split": "test", "cluster_id": "syn_03018_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 2, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1982, "query_tokens": 578, "message_tokens": 495, "message_utf8_bytes": 1081, "cache_tensor_bytes": null, "serialized_payload_bytes": 1081, "side_information_bytes": 0, "fetched_bytes": 726, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 41957.59679109324, "hops": {"1": 22880.75100001879, "2": 19076.845791074447}, "receiver_prefill_ms": 226.10858303960413, "ttft_ms": 11.850707931444049, "decode_ms": 11.920375050976872}, "total_wall_ms": 42207.95100007672, "ttft_ms": 11.850707931444049, "input_usage": {"lookups": 4}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 64, "facts_retained": 22, "retention_rate": 0.34375, "retention_by_kind": {"person": 0.5, "date": 0.5555555555555556, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1081}, "ledger_by_hop": {"1": {"facts_total": 64, "facts_retained": 22, "retention_rate": 0.34375, "retention_by_kind": {"person": 0.5, "date": 0.5555555555555556, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1181}, "2": {"facts_total": 64, "facts_retained": 22, "retention_rate": 0.34375, "retention_by_kind": {"person": 0.5, "date": 0.5555555555555556, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1081}}}, "reference_key_id": "keys/synthetic/syn_03018_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "af090a9c821da67010b5031584205204273ce51579047d51345d5bf471ce81ec", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "85ca05423c64b9df60a95968", "attempt_id": "e3b9ec47a83e", "run_group_id": "E4_core:A4|repairable|qwen3b:h2", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:12:28.379687Z", "task_family": "synthetic", "document_id": "syn_03018_2000", "document_hash": "d284eb07e27da742b9c2b67abccfdf72e4dd67c3259f44f3b41e34dd3e6fc1ff", "question_id": "q08_revision", "split": "test", "cluster_id": "syn_03018_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 2, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1982, "query_tokens": 596, "message_tokens": 495, "message_utf8_bytes": 1081, "cache_tensor_bytes": null, "serialized_payload_bytes": 1081, "side_information_bytes": 0, "fetched_bytes": 726, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 41957.59679109324, "hops": {"1": 22880.75100001879, "2": 19076.845791074447}, "receiver_prefill_ms": 227.0616249879822, "ttft_ms": 11.904333950951695, "decode_ms": 71.70412503182888}, "total_wall_ms": 42268.77958304249, "ttft_ms": 11.904333950951695, "input_usage": {"lookups": 4}, "output_usage": {"output_tokens": 6}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "304815", "evidence_artifact": {"ledger": {"facts_total": 64, "facts_retained": 22, "retention_rate": 0.34375, "retention_by_kind": {"person": 0.5, "date": 0.5555555555555556, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1081}, "ledger_by_hop": {"1": {"facts_total": 64, "facts_retained": 22, "retention_rate": 0.34375, "retention_by_kind": {"person": 0.5, "date": 0.5555555555555556, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1181}, "2": {"facts_total": 64, "facts_retained": 22, "retention_rate": 0.34375, "retention_by_kind": {"person": 0.5, "date": 0.5555555555555556, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1081}}}, "reference_key_id": "keys/synthetic/syn_03018_2000.json", "task_scores": {"correct": 0.0, "abstained": 0.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "af090a9c821da67010b5031584205204273ce51579047d51345d5bf471ce81ec", "output_hash": "a480475d6855141c920f42d1bf3337ca7fd88bcf4087b0ed2e253ed68cca6926"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "e7afbeefcdc5067daa0665ad", "attempt_id": "a5a6e514adba", "run_group_id": "E4_core:A4|repairable|qwen3b:h2", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:12:28.706399Z", "task_family": "synthetic", "document_id": "syn_03018_2000", "document_hash": "d284eb07e27da742b9c2b67abccfdf72e4dd67c3259f44f3b41e34dd3e6fc1ff", "question_id": "q09_revision", "split": "test", "cluster_id": "syn_03018_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 2, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1982, "query_tokens": 589, "message_tokens": 495, "message_utf8_bytes": 1081, "cache_tensor_bytes": null, "serialized_payload_bytes": 1081, "side_information_bytes": 0, "fetched_bytes": 726, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 41957.59679109324, "hops": {"1": 22880.75100001879, "2": 19076.845791074447}, "receiver_prefill_ms": 226.89287492539734, "ttft_ms": 12.089667026884854, "decode_ms": 71.35350001044571}, "total_wall_ms": 42268.43566610478, "ttft_ms": 12.089667026884854, "input_usage": {"lookups": 4}, "output_usage": {"output_tokens": 6}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "276340", "evidence_artifact": {"ledger": {"facts_total": 64, "facts_retained": 22, "retention_rate": 0.34375, "retention_by_kind": {"person": 0.5, "date": 0.5555555555555556, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1081}, "ledger_by_hop": {"1": {"facts_total": 64, "facts_retained": 22, "retention_rate": 0.34375, "retention_by_kind": {"person": 0.5, "date": 0.5555555555555556, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1181}, "2": {"facts_total": 64, "facts_retained": 22, "retention_rate": 0.34375, "retention_by_kind": {"person": 0.5, "date": 0.5555555555555556, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1081}}}, "reference_key_id": "keys/synthetic/syn_03018_2000.json", "task_scores": {"correct": 0.0, "abstained": 0.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "af090a9c821da67010b5031584205204273ce51579047d51345d5bf471ce81ec", "output_hash": "bc7c3667cbe0ad74ec0a1b092d3bad7e6a02978fa67746e3adbf3174ed34b3a2"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "9b3bab39215fb188aefb31fa", "attempt_id": "2fc1b7ba559e", "run_group_id": "E4_core:A4|repairable|qwen3b:h2", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:12:29.032479Z", "task_family": "synthetic", "document_id": "syn_03018_2000", "document_hash": "d284eb07e27da742b9c2b67abccfdf72e4dd67c3259f44f3b41e34dd3e6fc1ff", "question_id": "q10_temporal", "split": "test", "cluster_id": "syn_03018_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 2, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1982, "query_tokens": 597, "message_tokens": 495, "message_utf8_bytes": 1081, "cache_tensor_bytes": null, "serialized_payload_bytes": 1081, "side_information_bytes": 0, "fetched_bytes": 726, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 41957.59679109324, "hops": {"1": 22880.75100001879, "2": 19076.845791074447}, "receiver_prefill_ms": 227.0632500294596, "ttft_ms": 11.870209011249244, "decode_ms": 71.41162501648068}, "total_wall_ms": 42268.41962407343, "ttft_ms": 11.870209011249244, "input_usage": {"lookups": 4}, "output_usage": {"output_tokens": 6}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "Melymi-15", "evidence_artifact": {"ledger": {"facts_total": 64, "facts_retained": 22, "retention_rate": 0.34375, "retention_by_kind": {"person": 0.5, "date": 0.5555555555555556, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1081}, "ledger_by_hop": {"1": {"facts_total": 64, "facts_retained": 22, "retention_rate": 0.34375, "retention_by_kind": {"person": 0.5, "date": 0.5555555555555556, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1181}, "2": {"facts_total": 64, "facts_retained": 22, "retention_rate": 0.34375, "retention_by_kind": {"person": 0.5, "date": 0.5555555555555556, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1081}}}, "reference_key_id": "keys/synthetic/syn_03018_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "af090a9c821da67010b5031584205204273ce51579047d51345d5bf471ce81ec", "output_hash": "fa08d1474b01d6dfbf3dea049be251aa3c3021c17b4818258064b156509af58b"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "1ac0a42f7d4a7cfcdb3c7b45", "attempt_id": "9c8733975487", "run_group_id": "E4_core:A4|repairable|qwen3b:h2", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:12:29.358728Z", "task_family": "synthetic", "document_id": "syn_03018_2000", "document_hash": "d284eb07e27da742b9c2b67abccfdf72e4dd67c3259f44f3b41e34dd3e6fc1ff", "question_id": "q11_aggregation", "split": "test", "cluster_id": "syn_03018_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 2, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1982, "query_tokens": 589, "message_tokens": 495, "message_utf8_bytes": 1081, "cache_tensor_bytes": null, "serialized_payload_bytes": 1081, "side_information_bytes": 0, "fetched_bytes": 726, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 41957.59679109324, "hops": {"1": 22880.75100001879, "2": 19076.845791074447}, "receiver_prefill_ms": 226.78320901468396, "ttft_ms": 11.911791050806642, "decode_ms": 83.45945796463639}, "total_wall_ms": 42280.21733206697, "ttft_ms": 11.911791050806642, "input_usage": {"lookups": 4}, "output_usage": {"output_tokens": 7}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "2314480", "evidence_artifact": {"ledger": {"facts_total": 64, "facts_retained": 22, "retention_rate": 0.34375, "retention_by_kind": {"person": 0.5, "date": 0.5555555555555556, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1081}, "ledger_by_hop": {"1": {"facts_total": 64, "facts_retained": 22, "retention_rate": 0.34375, "retention_by_kind": {"person": 0.5, "date": 0.5555555555555556, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1181}, "2": {"facts_total": 64, "facts_retained": 22, "retention_rate": 0.34375, "retention_by_kind": {"person": 0.5, "date": 0.5555555555555556, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1081}}}, "reference_key_id": "keys/synthetic/syn_03018_2000.json", "task_scores": {"correct": 0.0, "abstained": 0.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "af090a9c821da67010b5031584205204273ce51579047d51345d5bf471ce81ec", "output_hash": "2dec5a6357460f713937a66e661c12ff90f39e7df214283ec28954b070efb06f"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "759706889be80c87df00eb27", "attempt_id": "8f41406f6b96", "run_group_id": "E4_core:A4|repairable|qwen3b:h2", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:12:29.696674Z", "task_family": "synthetic", "document_id": "syn_03018_2000", "document_hash": "d284eb07e27da742b9c2b67abccfdf72e4dd67c3259f44f3b41e34dd3e6fc1ff", "question_id": "q12_unit_conversion", "split": "test", "cluster_id": "syn_03018_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 2, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1982, "query_tokens": 586, "message_tokens": 495, "message_utf8_bytes": 1081, "cache_tensor_bytes": null, "serialized_payload_bytes": 1081, "side_information_bytes": 0, "fetched_bytes": 726, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 41957.59679109324, "hops": {"1": 22880.75100001879, "2": 19076.845791074447}, "receiver_prefill_ms": 226.5614999923855, "ttft_ms": 11.8940839311108, "decode_ms": 11.910000001080334}, "total_wall_ms": 42208.488166099414, "ttft_ms": 11.8940839311108, "input_usage": {"lookups": 4}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 64, "facts_retained": 22, "retention_rate": 0.34375, "retention_by_kind": {"person": 0.5, "date": 0.5555555555555556, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1081}, "ledger_by_hop": {"1": {"facts_total": 64, "facts_retained": 22, "retention_rate": 0.34375, "retention_by_kind": {"person": 0.5, "date": 0.5555555555555556, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1181}, "2": {"facts_total": 64, "facts_retained": 22, "retention_rate": 0.34375, "retention_by_kind": {"person": 0.5, "date": 0.5555555555555556, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1081}}}, "reference_key_id": "keys/synthetic/syn_03018_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "af090a9c821da67010b5031584205204273ce51579047d51345d5bf471ce81ec", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "83a410fee78062d888015ead", "attempt_id": "6d1b1a56cd41", "run_group_id": "E4_core:A4|repairable|qwen3b:h2", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:12:29.962818Z", "task_family": "synthetic", "document_id": "syn_03018_2000", "document_hash": "d284eb07e27da742b9c2b67abccfdf72e4dd67c3259f44f3b41e34dd3e6fc1ff", "question_id": "q13_unanswerable", "split": "test", "cluster_id": "syn_03018_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 2, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1982, "query_tokens": 590, "message_tokens": 495, "message_utf8_bytes": 1081, "cache_tensor_bytes": null, "serialized_payload_bytes": 1081, "side_information_bytes": 0, "fetched_bytes": 726, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 41957.59679109324, "hops": {"1": 22880.75100001879, "2": 19076.845791074447}, "receiver_prefill_ms": 226.99654195457697, "ttft_ms": 11.921582976356149, "decode_ms": 11.874249903485179}, "total_wall_ms": 42208.880666061305, "ttft_ms": 11.921582976356149, "input_usage": {"lookups": 4}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 64, "facts_retained": 22, "retention_rate": 0.34375, "retention_by_kind": {"person": 0.5, "date": 0.5555555555555556, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1081}, "ledger_by_hop": {"1": {"facts_total": 64, "facts_retained": 22, "retention_rate": 0.34375, "retention_by_kind": {"person": 0.5, "date": 0.5555555555555556, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1181}, "2": {"facts_total": 64, "facts_retained": 22, "retention_rate": 0.34375, "retention_by_kind": {"person": 0.5, "date": 0.5555555555555556, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1081}}}, "reference_key_id": "keys/synthetic/syn_03018_2000.json", "task_scores": {"correct": 1.0, "abstained": 1.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "af090a9c821da67010b5031584205204273ce51579047d51345d5bf471ce81ec", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "f7417eacd0b1aca21dfdb556", "attempt_id": "81e008d33431", "run_group_id": "E4_core:A4|repairable|qwen3b:h2", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:12:30.230259Z", "task_family": "synthetic", "document_id": "syn_03018_2000", "document_hash": "d284eb07e27da742b9c2b67abccfdf72e4dd67c3259f44f3b41e34dd3e6fc1ff", "question_id": "q14_unanswerable", "split": "test", "cluster_id": "syn_03018_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 2, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1982, "query_tokens": 591, "message_tokens": 495, "message_utf8_bytes": 1081, "cache_tensor_bytes": null, "serialized_payload_bytes": 1081, "side_information_bytes": 0, "fetched_bytes": 726, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 41957.59679109324, "hops": {"1": 22880.75100001879, "2": 19076.845791074447}, "receiver_prefill_ms": 226.86295805033296, "ttft_ms": 11.905915918760002, "decode_ms": 11.966541991569102}, "total_wall_ms": 42208.82866613101, "ttft_ms": 11.905915918760002, "input_usage": {"lookups": 4}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 64, "facts_retained": 22, "retention_rate": 0.34375, "retention_by_kind": {"person": 0.5, "date": 0.5555555555555556, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1081}, "ledger_by_hop": {"1": {"facts_total": 64, "facts_retained": 22, "retention_rate": 0.34375, "retention_by_kind": {"person": 0.5, "date": 0.5555555555555556, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1181}, "2": {"facts_total": 64, "facts_retained": 22, "retention_rate": 0.34375, "retention_by_kind": {"person": 0.5, "date": 0.5555555555555556, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1081}}}, "reference_key_id": "keys/synthetic/syn_03018_2000.json", "task_scores": {"correct": 1.0, "abstained": 1.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "af090a9c821da67010b5031584205204273ce51579047d51345d5bf471ce81ec", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "c691a2fbcb05f85f5a8d2447", "attempt_id": "a09e21f38876", "run_group_id": "E4_core:A4|repairable|qwen3b:h4", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:12:43.020781Z", "task_family": "synthetic", "document_id": "syn_03018_2000", "document_hash": "d284eb07e27da742b9c2b67abccfdf72e4dd67c3259f44f3b41e34dd3e6fc1ff", "question_id": "q00_single_fact", "split": "test", "cluster_id": "syn_03018_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 4, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1982, "query_tokens": 577, "message_tokens": 494, "message_utf8_bytes": 1077, "cache_tensor_bytes": null, "serialized_payload_bytes": 1077, "side_information_bytes": 0, "fetched_bytes": 2178, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 54478.767832159065, "hops": {"1": 22880.75100001879, "2": 19076.845791074447, "3": 6260.933833080344, "4": 6260.237207985483}, "receiver_prefill_ms": 208.8446670677513, "ttft_ms": 11.931665940210223, "decode_ms": 11.888333014212549}, "total_wall_ms": 54711.94158215076, "ttft_ms": 11.931665940210223, "input_usage": {"lookups": 12}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 64, "facts_retained": 22, "retention_rate": 0.34375, "retention_by_kind": {"person": 0.5, "date": 0.5555555555555556, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1077}, "ledger_by_hop": {"1": {"facts_total": 64, "facts_retained": 22, "retention_rate": 0.34375, "retention_by_kind": {"person": 0.5, "date": 0.5555555555555556, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1181}, "2": {"facts_total": 64, "facts_retained": 22, "retention_rate": 0.34375, "retention_by_kind": {"person": 0.5, "date": 0.5555555555555556, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1081}, "3": {"facts_total": 64, "facts_retained": 22, "retention_rate": 0.34375, "retention_by_kind": {"person": 0.5, "date": 0.5555555555555556, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1077}, "4": {"facts_total": 64, "facts_retained": 22, "retention_rate": 0.34375, "retention_by_kind": {"person": 0.5, "date": 0.5555555555555556, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1077}}}, "reference_key_id": "keys/synthetic/syn_03018_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "fb3497132b656390ec94098589be8cebe46dd0f8736439bf4730fca3b2cc4f4e", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "5a6ee2529cc21403845ca4b3", "attempt_id": "bda9dbaa1716", "run_group_id": "E4_core:A4|repairable|qwen3b:h4", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:12:43.269803Z", "task_family": "synthetic", "document_id": "syn_03018_2000", "document_hash": "d284eb07e27da742b9c2b67abccfdf72e4dd67c3259f44f3b41e34dd3e6fc1ff", "question_id": "q01_single_fact", "split": "test", "cluster_id": "syn_03018_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 4, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1982, "query_tokens": 583, "message_tokens": 494, "message_utf8_bytes": 1077, "cache_tensor_bytes": null, "serialized_payload_bytes": 1077, "side_information_bytes": 0, "fetched_bytes": 2178, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 54478.767832159065, "hops": {"1": 22880.75100001879, "2": 19076.845791074447, "3": 6260.933833080344, "4": 6260.237207985483}, "receiver_prefill_ms": 226.55616700649261, "ttft_ms": 11.993707972578704, "decode_ms": 143.13275006134063}, "total_wall_ms": 54860.947124194354, "ttft_ms": 11.993707972578704, "input_usage": {"lookups": 12}, "output_usage": {"output_tokens": 12}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "Lot 376-E, Baie-Comeau", "evidence_artifact": {"ledger": {"facts_total": 64, "facts_retained": 22, "retention_rate": 0.34375, "retention_by_kind": {"person": 0.5, "date": 0.5555555555555556, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1077}, "ledger_by_hop": {"1": {"facts_total": 64, "facts_retained": 22, "retention_rate": 0.34375, "retention_by_kind": {"person": 0.5, "date": 0.5555555555555556, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1181}, "2": {"facts_total": 64, "facts_retained": 22, "retention_rate": 0.34375, "retention_by_kind": {"person": 0.5, "date": 0.5555555555555556, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1081}, "3": {"facts_total": 64, "facts_retained": 22, "retention_rate": 0.34375, "retention_by_kind": {"person": 0.5, "date": 0.5555555555555556, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1077}, "4": {"facts_total": 64, "facts_retained": 22, "retention_rate": 0.34375, "retention_by_kind": {"person": 0.5, "date": 0.5555555555555556, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1077}}}, "reference_key_id": "keys/synthetic/syn_03018_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 0.8571428571428571}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "fb3497132b656390ec94098589be8cebe46dd0f8736439bf4730fca3b2cc4f4e", "output_hash": "b58829a1440993c4af3307ba44494f021d726fd11e9337888b6b852a8b37351a"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "e5a854dc76bdbfb5a4df698d", "attempt_id": "53eaeb849d2e", "run_group_id": "E4_core:A4|repairable|qwen3b:h4", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:12:43.667105Z", "task_family": "synthetic", "document_id": "syn_03018_2000", "document_hash": "d284eb07e27da742b9c2b67abccfdf72e4dd67c3259f44f3b41e34dd3e6fc1ff", "question_id": "q02_exact_copy", "split": "test", "cluster_id": "syn_03018_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 4, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1982, "query_tokens": 582, "message_tokens": 494, "message_utf8_bytes": 1077, "cache_tensor_bytes": null, "serialized_payload_bytes": 1077, "side_information_bytes": 0, "fetched_bytes": 2178, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 54478.767832159065, "hops": {"1": 22880.75100001879, "2": 19076.845791074447, "3": 6260.933833080344, "4": 6260.237207985483}, "receiver_prefill_ms": 226.43087501637638, "ttft_ms": 11.94604195188731, "decode_ms": 11.88850007019937}, "total_wall_ms": 54729.493748163804, "ttft_ms": 11.94604195188731, "input_usage": {"lookups": 12}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 64, "facts_retained": 22, "retention_rate": 0.34375, "retention_by_kind": {"person": 0.5, "date": 0.5555555555555556, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1077}, "ledger_by_hop": {"1": {"facts_total": 64, "facts_retained": 22, "retention_rate": 0.34375, "retention_by_kind": {"person": 0.5, "date": 0.5555555555555556, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1181}, "2": {"facts_total": 64, "facts_retained": 22, "retention_rate": 0.34375, "retention_by_kind": {"person": 0.5, "date": 0.5555555555555556, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1081}, "3": {"facts_total": 64, "facts_retained": 22, "retention_rate": 0.34375, "retention_by_kind": {"person": 0.5, "date": 0.5555555555555556, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1077}, "4": {"facts_total": 64, "facts_retained": 22, "retention_rate": 0.34375, "retention_by_kind": {"person": 0.5, "date": 0.5555555555555556, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1077}}}, "reference_key_id": "keys/synthetic/syn_03018_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "fb3497132b656390ec94098589be8cebe46dd0f8736439bf4730fca3b2cc4f4e", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "9fca8c98130a9bc0cf2016c1", "attempt_id": "081cf8561e7e", "run_group_id": "E4_core:A4|repairable|qwen3b:h4", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:12:43.933037Z", "task_family": "synthetic", "document_id": "syn_03018_2000", "document_hash": "d284eb07e27da742b9c2b67abccfdf72e4dd67c3259f44f3b41e34dd3e6fc1ff", "question_id": "q03_exact_copy", "split": "test", "cluster_id": "syn_03018_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 4, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1982, "query_tokens": 580, "message_tokens": 494, "message_utf8_bytes": 1077, "cache_tensor_bytes": null, "serialized_payload_bytes": 1077, "side_information_bytes": 0, "fetched_bytes": 2178, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 54478.767832159065, "hops": {"1": 22880.75100001879, "2": 19076.845791074447, "3": 6260.933833080344, "4": 6260.237207985483}, "receiver_prefill_ms": 226.49170795921236, "ttft_ms": 11.946166050620377, "decode_ms": 107.36045800149441}, "total_wall_ms": 54825.0311241718, "ttft_ms": 11.946166050620377, "input_usage": {"lookups": 12}, "output_usage": {"output_tokens": 9}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "Priyamvada Sørlie", "evidence_artifact": {"ledger": {"facts_total": 64, "facts_retained": 22, "retention_rate": 0.34375, "retention_by_kind": {"person": 0.5, "date": 0.5555555555555556, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1077}, "ledger_by_hop": {"1": {"facts_total": 64, "facts_retained": 22, "retention_rate": 0.34375, "retention_by_kind": {"person": 0.5, "date": 0.5555555555555556, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1181}, "2": {"facts_total": 64, "facts_retained": 22, "retention_rate": 0.34375, "retention_by_kind": {"person": 0.5, "date": 0.5555555555555556, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1081}, "3": {"facts_total": 64, "facts_retained": 22, "retention_rate": 0.34375, "retention_by_kind": {"person": 0.5, "date": 0.5555555555555556, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1077}, "4": {"facts_total": 64, "facts_retained": 22, "retention_rate": 0.34375, "retention_by_kind": {"person": 0.5, "date": 0.5555555555555556, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1077}}}, "reference_key_id": "keys/synthetic/syn_03018_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "fb3497132b656390ec94098589be8cebe46dd0f8736439bf4730fca3b2cc4f4e", "output_hash": "c5f554196ed23dd71ad995ec89f7142099d524dbb6534cc10d94a0c89d13e792"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "0e10ebf4ccba9f2f273fb58b", "attempt_id": "0d23a8f40b77", "run_group_id": "E4_core:A4|repairable|qwen3b:h4", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:12:44.294671Z", "task_family": "synthetic", "document_id": "syn_03018_2000", "document_hash": "d284eb07e27da742b9c2b67abccfdf72e4dd67c3259f44f3b41e34dd3e6fc1ff", "question_id": "q04_multi_hop", "split": "test", "cluster_id": "syn_03018_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 4, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1982, "query_tokens": 586, "message_tokens": 494, "message_utf8_bytes": 1077, "cache_tensor_bytes": null, "serialized_payload_bytes": 1077, "side_information_bytes": 0, "fetched_bytes": 2178, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 54478.767832159065, "hops": {"1": 22880.75100001879, "2": 19076.845791074447, "3": 6260.933833080344, "4": 6260.237207985483}, "receiver_prefill_ms": 226.514249923639, "ttft_ms": 11.947249993681908, "decode_ms": 11.969083920121193}, "total_wall_ms": 54729.64799916372, "ttft_ms": 11.947249993681908, "input_usage": {"lookups": 12}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 64, "facts_retained": 22, "retention_rate": 0.34375, "retention_by_kind": {"person": 0.5, "date": 0.5555555555555556, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1077}, "ledger_by_hop": {"1": {"facts_total": 64, "facts_retained": 22, "retention_rate": 0.34375, "retention_by_kind": {"person": 0.5, "date": 0.5555555555555556, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1181}, "2": {"facts_total": 64, "facts_retained": 22, "retention_rate": 0.34375, "retention_by_kind": {"person": 0.5, "date": 0.5555555555555556, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1081}, "3": {"facts_total": 64, "facts_retained": 22, "retention_rate": 0.34375, "retention_by_kind": {"person": 0.5, "date": 0.5555555555555556, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1077}, "4": {"facts_total": 64, "facts_retained": 22, "retention_rate": 0.34375, "retention_by_kind": {"person": 0.5, "date": 0.5555555555555556, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1077}}}, "reference_key_id": "keys/synthetic/syn_03018_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "fb3497132b656390ec94098589be8cebe46dd0f8736439bf4730fca3b2cc4f4e", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "f720baeb8f670bd0f122512d", "attempt_id": "62b8d755c890", "run_group_id": "E4_core:A4|repairable|qwen3b:h4", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:12:44.560619Z", "task_family": "synthetic", "document_id": "syn_03018_2000", "document_hash": "d284eb07e27da742b9c2b67abccfdf72e4dd67c3259f44f3b41e34dd3e6fc1ff", "question_id": "q05_multi_hop", "split": "test", "cluster_id": "syn_03018_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 4, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1982, "query_tokens": 586, "message_tokens": 494, "message_utf8_bytes": 1077, "cache_tensor_bytes": null, "serialized_payload_bytes": 1077, "side_information_bytes": 0, "fetched_bytes": 2178, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 54478.767832159065, "hops": {"1": 22880.75100001879, "2": 19076.845791074447, "3": 6260.933833080344, "4": 6260.237207985483}, "receiver_prefill_ms": 226.59970889799297, "ttft_ms": 11.911457986570895, "decode_ms": 118.9490839606151}, "total_wall_ms": 54836.667124181986, "ttft_ms": 11.911457986570895, "input_usage": {"lookups": 12}, "output_usage": {"output_tokens": 10}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "HM-38K-5533", "evidence_artifact": {"ledger": {"facts_total": 64, "facts_retained": 22, "retention_rate": 0.34375, "retention_by_kind": {"person": 0.5, "date": 0.5555555555555556, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1077}, "ledger_by_hop": {"1": {"facts_total": 64, "facts_retained": 22, "retention_rate": 0.34375, "retention_by_kind": {"person": 0.5, "date": 0.5555555555555556, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1181}, "2": {"facts_total": 64, "facts_retained": 22, "retention_rate": 0.34375, "retention_by_kind": {"person": 0.5, "date": 0.5555555555555556, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1081}, "3": {"facts_total": 64, "facts_retained": 22, "retention_rate": 0.34375, "retention_by_kind": {"person": 0.5, "date": 0.5555555555555556, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1077}, "4": {"facts_total": 64, "facts_retained": 22, "retention_rate": 0.34375, "retention_by_kind": {"person": 0.5, "date": 0.5555555555555556, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1077}}}, "reference_key_id": "keys/synthetic/syn_03018_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "fb3497132b656390ec94098589be8cebe46dd0f8736439bf4730fca3b2cc4f4e", "output_hash": "18acd9807857243bbded3bceae55398316a3958b81d70364441bdb6748eaf878"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "d99bcc5ed61f7e1a553ed56a", "attempt_id": "6ecf852b508c", "run_group_id": "E4_core:A4|repairable|qwen3b:h4", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:12:44.933592Z", "task_family": "synthetic", "document_id": "syn_03018_2000", "document_hash": "d284eb07e27da742b9c2b67abccfdf72e4dd67c3259f44f3b41e34dd3e6fc1ff", "question_id": "q06_negation", "split": "test", "cluster_id": "syn_03018_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 4, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1982, "query_tokens": 578, "message_tokens": 494, "message_utf8_bytes": 1077, "cache_tensor_bytes": null, "serialized_payload_bytes": 1077, "side_information_bytes": 0, "fetched_bytes": 2178, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 54478.767832159065, "hops": {"1": 22880.75100001879, "2": 19076.845791074447, "3": 6260.933833080344, "4": 6260.237207985483}, "receiver_prefill_ms": 226.02916706819087, "ttft_ms": 11.945915990509093, "decode_ms": 11.926249950192869}, "total_wall_ms": 54729.13633217104, "ttft_ms": 11.945915990509093, "input_usage": {"lookups": 12}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 64, "facts_retained": 22, "retention_rate": 0.34375, "retention_by_kind": {"person": 0.5, "date": 0.5555555555555556, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1077}, "ledger_by_hop": {"1": {"facts_total": 64, "facts_retained": 22, "retention_rate": 0.34375, "retention_by_kind": {"person": 0.5, "date": 0.5555555555555556, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1181}, "2": {"facts_total": 64, "facts_retained": 22, "retention_rate": 0.34375, "retention_by_kind": {"person": 0.5, "date": 0.5555555555555556, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1081}, "3": {"facts_total": 64, "facts_retained": 22, "retention_rate": 0.34375, "retention_by_kind": {"person": 0.5, "date": 0.5555555555555556, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1077}, "4": {"facts_total": 64, "facts_retained": 22, "retention_rate": 0.34375, "retention_by_kind": {"person": 0.5, "date": 0.5555555555555556, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1077}}}, "reference_key_id": "keys/synthetic/syn_03018_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "fb3497132b656390ec94098589be8cebe46dd0f8736439bf4730fca3b2cc4f4e", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "1f21a255077fd47921c73a02", "attempt_id": "1c5321537968", "run_group_id": "E4_core:A4|repairable|qwen3b:h4", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:12:45.199142Z", "task_family": "synthetic", "document_id": "syn_03018_2000", "document_hash": "d284eb07e27da742b9c2b67abccfdf72e4dd67c3259f44f3b41e34dd3e6fc1ff", "question_id": "q07_negation", "split": "test", "cluster_id": "syn_03018_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 4, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1982, "query_tokens": 577, "message_tokens": 494, "message_utf8_bytes": 1077, "cache_tensor_bytes": null, "serialized_payload_bytes": 1077, "side_information_bytes": 0, "fetched_bytes": 2178, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 54478.767832159065, "hops": {"1": 22880.75100001879, "2": 19076.845791074447, "3": 6260.933833080344, "4": 6260.237207985483}, "receiver_prefill_ms": 209.17541603557765, "ttft_ms": 11.900875018909574, "decode_ms": 11.938291951082647}, "total_wall_ms": 54712.23579009529, "ttft_ms": 11.900875018909574, "input_usage": {"lookups": 12}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 64, "facts_retained": 22, "retention_rate": 0.34375, "retention_by_kind": {"person": 0.5, "date": 0.5555555555555556, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1077}, "ledger_by_hop": {"1": {"facts_total": 64, "facts_retained": 22, "retention_rate": 0.34375, "retention_by_kind": {"person": 0.5, "date": 0.5555555555555556, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1181}, "2": {"facts_total": 64, "facts_retained": 22, "retention_rate": 0.34375, "retention_by_kind": {"person": 0.5, "date": 0.5555555555555556, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1081}, "3": {"facts_total": 64, "facts_retained": 22, "retention_rate": 0.34375, "retention_by_kind": {"person": 0.5, "date": 0.5555555555555556, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1077}, "4": {"facts_total": 64, "facts_retained": 22, "retention_rate": 0.34375, "retention_by_kind": {"person": 0.5, "date": 0.5555555555555556, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1077}}}, "reference_key_id": "keys/synthetic/syn_03018_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "fb3497132b656390ec94098589be8cebe46dd0f8736439bf4730fca3b2cc4f4e", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "b20a6a3847ea3affa4c64f99", "attempt_id": "44fe58985f44", "run_group_id": "E4_core:A4|repairable|qwen3b:h4", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:12:45.447517Z", "task_family": "synthetic", "document_id": "syn_03018_2000", "document_hash": "d284eb07e27da742b9c2b67abccfdf72e4dd67c3259f44f3b41e34dd3e6fc1ff", "question_id": "q08_revision", "split": "test", "cluster_id": "syn_03018_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 4, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1982, "query_tokens": 595, "message_tokens": 494, "message_utf8_bytes": 1077, "cache_tensor_bytes": null, "serialized_payload_bytes": 1077, "side_information_bytes": 0, "fetched_bytes": 2178, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 54478.767832159065, "hops": {"1": 22880.75100001879, "2": 19076.845791074447, "3": 6260.933833080344, "4": 6260.237207985483}, "receiver_prefill_ms": 226.9329170230776, "ttft_ms": 11.8787499377504, "decode_ms": 71.1424999171868}, "total_wall_ms": 54789.16420717724, "ttft_ms": 11.8787499377504, "input_usage": {"lookups": 12}, "output_usage": {"output_tokens": 6}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "304815", "evidence_artifact": {"ledger": {"facts_total": 64, "facts_retained": 22, "retention_rate": 0.34375, "retention_by_kind": {"person": 0.5, "date": 0.5555555555555556, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1077}, "ledger_by_hop": {"1": {"facts_total": 64, "facts_retained": 22, "retention_rate": 0.34375, "retention_by_kind": {"person": 0.5, "date": 0.5555555555555556, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1181}, "2": {"facts_total": 64, "facts_retained": 22, "retention_rate": 0.34375, "retention_by_kind": {"person": 0.5, "date": 0.5555555555555556, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1081}, "3": {"facts_total": 64, "facts_retained": 22, "retention_rate": 0.34375, "retention_by_kind": {"person": 0.5, "date": 0.5555555555555556, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1077}, "4": {"facts_total": 64, "facts_retained": 22, "retention_rate": 0.34375, "retention_by_kind": {"person": 0.5, "date": 0.5555555555555556, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1077}}}, "reference_key_id": "keys/synthetic/syn_03018_2000.json", "task_scores": {"correct": 0.0, "abstained": 0.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "fb3497132b656390ec94098589be8cebe46dd0f8736439bf4730fca3b2cc4f4e", "output_hash": "a480475d6855141c920f42d1bf3337ca7fd88bcf4087b0ed2e253ed68cca6926"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "77ae089902fccd567020404f", "attempt_id": "21597d943377", "run_group_id": "E4_core:A4|repairable|qwen3b:h4", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:12:45.773265Z", "task_family": "synthetic", "document_id": "syn_03018_2000", "document_hash": "d284eb07e27da742b9c2b67abccfdf72e4dd67c3259f44f3b41e34dd3e6fc1ff", "question_id": "q09_revision", "split": "test", "cluster_id": "syn_03018_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 4, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1982, "query_tokens": 588, "message_tokens": 494, "message_utf8_bytes": 1077, "cache_tensor_bytes": null, "serialized_payload_bytes": 1077, "side_information_bytes": 0, "fetched_bytes": 2178, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 54478.767832159065, "hops": {"1": 22880.75100001879, "2": 19076.845791074447, "3": 6260.933833080344, "4": 6260.237207985483}, "receiver_prefill_ms": 226.49766691029072, "ttft_ms": 12.20620796084404, "decode_ms": 106.95308400318027}, "total_wall_ms": 54824.89224825986, "ttft_ms": 12.20620796084404, "input_usage": {"lookups": 12}, "output_usage": {"output_tokens": 9}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "CAD 276,340", "evidence_artifact": {"ledger": {"facts_total": 64, "facts_retained": 22, "retention_rate": 0.34375, "retention_by_kind": {"person": 0.5, "date": 0.5555555555555556, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1077}, "ledger_by_hop": {"1": {"facts_total": 64, "facts_retained": 22, "retention_rate": 0.34375, "retention_by_kind": {"person": 0.5, "date": 0.5555555555555556, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1181}, "2": {"facts_total": 64, "facts_retained": 22, "retention_rate": 0.34375, "retention_by_kind": {"person": 0.5, "date": 0.5555555555555556, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1081}, "3": {"facts_total": 64, "facts_retained": 22, "retention_rate": 0.34375, "retention_by_kind": {"person": 0.5, "date": 0.5555555555555556, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1077}, "4": {"facts_total": 64, "facts_retained": 22, "retention_rate": 0.34375, "retention_by_kind": {"person": 0.5, "date": 0.5555555555555556, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1077}}}, "reference_key_id": "keys/synthetic/syn_03018_2000.json", "task_scores": {"correct": 0.0, "abstained": 0.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "fb3497132b656390ec94098589be8cebe46dd0f8736439bf4730fca3b2cc4f4e", "output_hash": "1f0e1d2e718e3493b4e945d50114918556754962d08324081d005dcca667be19"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "15868723e95140571f22a0ab", "attempt_id": "2316350fdc2c", "run_group_id": "E4_core:A4|repairable|qwen3b:h4", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:12:46.134827Z", "task_family": "synthetic", "document_id": "syn_03018_2000", "document_hash": "d284eb07e27da742b9c2b67abccfdf72e4dd67c3259f44f3b41e34dd3e6fc1ff", "question_id": "q10_temporal", "split": "test", "cluster_id": "syn_03018_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 4, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1982, "query_tokens": 596, "message_tokens": 494, "message_utf8_bytes": 1077, "cache_tensor_bytes": null, "serialized_payload_bytes": 1077, "side_information_bytes": 0, "fetched_bytes": 2178, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 54478.767832159065, "hops": {"1": 22880.75100001879, "2": 19076.845791074447, "3": 6260.933833080344, "4": 6260.237207985483}, "receiver_prefill_ms": 227.03658405225724, "ttft_ms": 11.93179190158844, "decode_ms": 71.69600005727261}, "total_wall_ms": 54789.902165182866, "ttft_ms": 11.93179190158844, "input_usage": {"lookups": 12}, "output_usage": {"output_tokens": 6}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "Melymi-15", "evidence_artifact": {"ledger": {"facts_total": 64, "facts_retained": 22, "retention_rate": 0.34375, "retention_by_kind": {"person": 0.5, "date": 0.5555555555555556, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1077}, "ledger_by_hop": {"1": {"facts_total": 64, "facts_retained": 22, "retention_rate": 0.34375, "retention_by_kind": {"person": 0.5, "date": 0.5555555555555556, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1181}, "2": {"facts_total": 64, "facts_retained": 22, "retention_rate": 0.34375, "retention_by_kind": {"person": 0.5, "date": 0.5555555555555556, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1081}, "3": {"facts_total": 64, "facts_retained": 22, "retention_rate": 0.34375, "retention_by_kind": {"person": 0.5, "date": 0.5555555555555556, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1077}, "4": {"facts_total": 64, "facts_retained": 22, "retention_rate": 0.34375, "retention_by_kind": {"person": 0.5, "date": 0.5555555555555556, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1077}}}, "reference_key_id": "keys/synthetic/syn_03018_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "fb3497132b656390ec94098589be8cebe46dd0f8736439bf4730fca3b2cc4f4e", "output_hash": "fa08d1474b01d6dfbf3dea049be251aa3c3021c17b4818258064b156509af58b"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "a75189aef0be7c1a8801b4fa", "attempt_id": "23b67c82050a", "run_group_id": "E4_core:A4|repairable|qwen3b:h4", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:12:46.461074Z", "task_family": "synthetic", "document_id": "syn_03018_2000", "document_hash": "d284eb07e27da742b9c2b67abccfdf72e4dd67c3259f44f3b41e34dd3e6fc1ff", "question_id": "q11_aggregation", "split": "test", "cluster_id": "syn_03018_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 4, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1982, "query_tokens": 588, "message_tokens": 494, "message_utf8_bytes": 1077, "cache_tensor_bytes": null, "serialized_payload_bytes": 1077, "side_information_bytes": 0, "fetched_bytes": 2178, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 54478.767832159065, "hops": {"1": 22880.75100001879, "2": 19076.845791074447, "3": 6260.933833080344, "4": 6260.237207985483}, "receiver_prefill_ms": 226.57950001303107, "ttft_ms": 11.892084032297134, "decode_ms": 107.11812507361174}, "total_wall_ms": 54824.799249181524, "ttft_ms": 11.892084032297134, "input_usage": {"lookups": 12}, "output_usage": {"output_tokens": 9}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "2,481,350", "evidence_artifact": {"ledger": {"facts_total": 64, "facts_retained": 22, "retention_rate": 0.34375, "retention_by_kind": {"person": 0.5, "date": 0.5555555555555556, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1077}, "ledger_by_hop": {"1": {"facts_total": 64, "facts_retained": 22, "retention_rate": 0.34375, "retention_by_kind": {"person": 0.5, "date": 0.5555555555555556, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1181}, "2": {"facts_total": 64, "facts_retained": 22, "retention_rate": 0.34375, "retention_by_kind": {"person": 0.5, "date": 0.5555555555555556, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1081}, "3": {"facts_total": 64, "facts_retained": 22, "retention_rate": 0.34375, "retention_by_kind": {"person": 0.5, "date": 0.5555555555555556, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1077}, "4": {"facts_total": 64, "facts_retained": 22, "retention_rate": 0.34375, "retention_by_kind": {"person": 0.5, "date": 0.5555555555555556, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1077}}}, "reference_key_id": "keys/synthetic/syn_03018_2000.json", "task_scores": {"correct": 0.0, "abstained": 0.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "fb3497132b656390ec94098589be8cebe46dd0f8736439bf4730fca3b2cc4f4e", "output_hash": "a59560fb797f568e2b2688b45d94559fccc0e769906b2badf279c95761025476"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "5733a23a92b6733270ff4988", "attempt_id": "c2cb279037a7", "run_group_id": "E4_core:A4|repairable|qwen3b:h4", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:12:46.822571Z", "task_family": "synthetic", "document_id": "syn_03018_2000", "document_hash": "d284eb07e27da742b9c2b67abccfdf72e4dd67c3259f44f3b41e34dd3e6fc1ff", "question_id": "q12_unit_conversion", "split": "test", "cluster_id": "syn_03018_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 4, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1982, "query_tokens": 585, "message_tokens": 494, "message_utf8_bytes": 1077, "cache_tensor_bytes": null, "serialized_payload_bytes": 1077, "side_information_bytes": 0, "fetched_bytes": 2178, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 54478.767832159065, "hops": {"1": 22880.75100001879, "2": 19076.845791074447, "3": 6260.933833080344, "4": 6260.237207985483}, "receiver_prefill_ms": 226.4375409577042, "ttft_ms": 11.876833974383771, "decode_ms": 11.865541920997202}, "total_wall_ms": 54729.39820715692, "ttft_ms": 11.876833974383771, "input_usage": {"lookups": 12}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 64, "facts_retained": 22, "retention_rate": 0.34375, "retention_by_kind": {"person": 0.5, "date": 0.5555555555555556, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1077}, "ledger_by_hop": {"1": {"facts_total": 64, "facts_retained": 22, "retention_rate": 0.34375, "retention_by_kind": {"person": 0.5, "date": 0.5555555555555556, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1181}, "2": {"facts_total": 64, "facts_retained": 22, "retention_rate": 0.34375, "retention_by_kind": {"person": 0.5, "date": 0.5555555555555556, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1081}, "3": {"facts_total": 64, "facts_retained": 22, "retention_rate": 0.34375, "retention_by_kind": {"person": 0.5, "date": 0.5555555555555556, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1077}, "4": {"facts_total": 64, "facts_retained": 22, "retention_rate": 0.34375, "retention_by_kind": {"person": 0.5, "date": 0.5555555555555556, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1077}}}, "reference_key_id": "keys/synthetic/syn_03018_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "fb3497132b656390ec94098589be8cebe46dd0f8736439bf4730fca3b2cc4f4e", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "0ebb702e2f06899b9aafd782", "attempt_id": "ba741d710499", "run_group_id": "E4_core:A4|repairable|qwen3b:h4", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:12:47.088051Z", "task_family": "synthetic", "document_id": "syn_03018_2000", "document_hash": "d284eb07e27da742b9c2b67abccfdf72e4dd67c3259f44f3b41e34dd3e6fc1ff", "question_id": "q13_unanswerable", "split": "test", "cluster_id": "syn_03018_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 4, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1982, "query_tokens": 589, "message_tokens": 494, "message_utf8_bytes": 1077, "cache_tensor_bytes": null, "serialized_payload_bytes": 1077, "side_information_bytes": 0, "fetched_bytes": 2178, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 54478.767832159065, "hops": {"1": 22880.75100001879, "2": 19076.845791074447, "3": 6260.933833080344, "4": 6260.237207985483}, "receiver_prefill_ms": 226.61558294203132, "ttft_ms": 11.879125027917325, "decode_ms": 11.939583928324282}, "total_wall_ms": 54729.64604012668, "ttft_ms": 11.879125027917325, "input_usage": {"lookups": 12}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 64, "facts_retained": 22, "retention_rate": 0.34375, "retention_by_kind": {"person": 0.5, "date": 0.5555555555555556, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1077}, "ledger_by_hop": {"1": {"facts_total": 64, "facts_retained": 22, "retention_rate": 0.34375, "retention_by_kind": {"person": 0.5, "date": 0.5555555555555556, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1181}, "2": {"facts_total": 64, "facts_retained": 22, "retention_rate": 0.34375, "retention_by_kind": {"person": 0.5, "date": 0.5555555555555556, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1081}, "3": {"facts_total": 64, "facts_retained": 22, "retention_rate": 0.34375, "retention_by_kind": {"person": 0.5, "date": 0.5555555555555556, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1077}, "4": {"facts_total": 64, "facts_retained": 22, "retention_rate": 0.34375, "retention_by_kind": {"person": 0.5, "date": 0.5555555555555556, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1077}}}, "reference_key_id": "keys/synthetic/syn_03018_2000.json", "task_scores": {"correct": 1.0, "abstained": 1.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "fb3497132b656390ec94098589be8cebe46dd0f8736439bf4730fca3b2cc4f4e", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "fca029a14ad55497bda5788a", "attempt_id": "2e8463116c7e", "run_group_id": "E4_core:A4|repairable|qwen3b:h4", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:12:47.354162Z", "task_family": "synthetic", "document_id": "syn_03018_2000", "document_hash": "d284eb07e27da742b9c2b67abccfdf72e4dd67c3259f44f3b41e34dd3e6fc1ff", "question_id": "q14_unanswerable", "split": "test", "cluster_id": "syn_03018_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 4, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1982, "query_tokens": 590, "message_tokens": 494, "message_utf8_bytes": 1077, "cache_tensor_bytes": null, "serialized_payload_bytes": 1077, "side_information_bytes": 0, "fetched_bytes": 2178, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 54478.767832159065, "hops": {"1": 22880.75100001879, "2": 19076.845791074447, "3": 6260.933833080344, "4": 6260.237207985483}, "receiver_prefill_ms": 226.64404206443578, "ttft_ms": 11.910707922652364, "decode_ms": 11.908625019714236}, "total_wall_ms": 54729.684124235064, "ttft_ms": 11.910707922652364, "input_usage": {"lookups": 12}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 64, "facts_retained": 22, "retention_rate": 0.34375, "retention_by_kind": {"person": 0.5, "date": 0.5555555555555556, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1077}, "ledger_by_hop": {"1": {"facts_total": 64, "facts_retained": 22, "retention_rate": 0.34375, "retention_by_kind": {"person": 0.5, "date": 0.5555555555555556, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1181}, "2": {"facts_total": 64, "facts_retained": 22, "retention_rate": 0.34375, "retention_by_kind": {"person": 0.5, "date": 0.5555555555555556, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1081}, "3": {"facts_total": 64, "facts_retained": 22, "retention_rate": 0.34375, "retention_by_kind": {"person": 0.5, "date": 0.5555555555555556, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1077}, "4": {"facts_total": 64, "facts_retained": 22, "retention_rate": 0.34375, "retention_by_kind": {"person": 0.5, "date": 0.5555555555555556, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1077}}}, "reference_key_id": "keys/synthetic/syn_03018_2000.json", "task_scores": {"correct": 1.0, "abstained": 1.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "fb3497132b656390ec94098589be8cebe46dd0f8736439bf4730fca3b2cc4f4e", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "d3e3fd17cbb7564c6a4c0447", "attempt_id": "2a0082d7bb8e", "run_group_id": "E4_core:A4|repairable|qwen3b:h8", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:13:12.717913Z", "task_family": "synthetic", "document_id": "syn_03018_2000", "document_hash": "d284eb07e27da742b9c2b67abccfdf72e4dd67c3259f44f3b41e34dd3e6fc1ff", "question_id": "q00_single_fact", "split": "test", "cluster_id": "syn_03018_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 8, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1982, "query_tokens": 577, "message_tokens": 494, "message_utf8_bytes": 1077, "cache_tensor_bytes": null, "serialized_payload_bytes": 1077, "side_information_bytes": 0, "fetched_bytes": 5082, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 79571.560457116, "hops": {"1": 22880.75100001879, "2": 19076.845791074447, "3": 6260.933833080344, "4": 6260.237207985483, "5": 6275.52599995397, "6": 6274.316166993231, "7": 6269.359500030987, "8": 6273.59095797874}, "receiver_prefill_ms": 209.41641693934798, "ttft_ms": 11.839249986223876, "decode_ms": 11.996665969491005}, "total_wall_ms": 79805.31874904409, "ttft_ms": 11.839249986223876, "input_usage": {"lookups": 28}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 64, "facts_retained": 22, "retention_rate": 0.34375, "retention_by_kind": {"person": 0.5, "date": 0.5555555555555556, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1077}, "ledger_by_hop": {"1": {"facts_total": 64, "facts_retained": 22, "retention_rate": 0.34375, "retention_by_kind": {"person": 0.5, "date": 0.5555555555555556, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1181}, "2": {"facts_total": 64, "facts_retained": 22, "retention_rate": 0.34375, "retention_by_kind": {"person": 0.5, "date": 0.5555555555555556, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1081}, "3": {"facts_total": 64, "facts_retained": 22, "retention_rate": 0.34375, "retention_by_kind": {"person": 0.5, "date": 0.5555555555555556, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1077}, "4": {"facts_total": 64, "facts_retained": 22, "retention_rate": 0.34375, "retention_by_kind": {"person": 0.5, "date": 0.5555555555555556, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1077}, "5": {"facts_total": 64, "facts_retained": 22, "retention_rate": 0.34375, "retention_by_kind": {"person": 0.5, "date": 0.5555555555555556, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1077}, "6": {"facts_total": 64, "facts_retained": 22, "retention_rate": 0.34375, "retention_by_kind": {"person": 0.5, "date": 0.5555555555555556, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1077}, "7": {"facts_total": 64, "facts_retained": 22, "retention_rate": 0.34375, "retention_by_kind": {"person": 0.5, "date": 0.5555555555555556, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1077}, "8": {"facts_total": 64, "facts_retained": 22, "retention_rate": 0.34375, "retention_by_kind": {"person": 0.5, "date": 0.5555555555555556, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1077}}}, "reference_key_id": "keys/synthetic/syn_03018_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "fb3497132b656390ec94098589be8cebe46dd0f8736439bf4730fca3b2cc4f4e", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "1f6bd0a48c0dc32c5f7a9d64", "attempt_id": "5df7f1d65105", "run_group_id": "E4_core:A4|repairable|qwen3b:h8", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:13:12.967377Z", "task_family": "synthetic", "document_id": "syn_03018_2000", "document_hash": "d284eb07e27da742b9c2b67abccfdf72e4dd67c3259f44f3b41e34dd3e6fc1ff", "question_id": "q01_single_fact", "split": "test", "cluster_id": "syn_03018_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 8, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1982, "query_tokens": 583, "message_tokens": 494, "message_utf8_bytes": 1077, "cache_tensor_bytes": null, "serialized_payload_bytes": 1077, "side_information_bytes": 0, "fetched_bytes": 5082, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 79571.560457116, "hops": {"1": 22880.75100001879, "2": 19076.845791074447, "3": 6260.933833080344, "4": 6260.237207985483, "5": 6275.52599995397, "6": 6274.316166993231, "7": 6269.359500030987, "8": 6273.59095797874}, "receiver_prefill_ms": 226.66575002949685, "ttft_ms": 11.906833038665354, "decode_ms": 143.40141601860523}, "total_wall_ms": 79954.11270717159, "ttft_ms": 11.906833038665354, "input_usage": {"lookups": 28}, "output_usage": {"output_tokens": 12}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "Lot 376-E, Baie-Comeau", "evidence_artifact": {"ledger": {"facts_total": 64, "facts_retained": 22, "retention_rate": 0.34375, "retention_by_kind": {"person": 0.5, "date": 0.5555555555555556, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1077}, "ledger_by_hop": {"1": {"facts_total": 64, "facts_retained": 22, "retention_rate": 0.34375, "retention_by_kind": {"person": 0.5, "date": 0.5555555555555556, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1181}, "2": {"facts_total": 64, "facts_retained": 22, "retention_rate": 0.34375, "retention_by_kind": {"person": 0.5, "date": 0.5555555555555556, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1081}, "3": {"facts_total": 64, "facts_retained": 22, "retention_rate": 0.34375, "retention_by_kind": {"person": 0.5, "date": 0.5555555555555556, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1077}, "4": {"facts_total": 64, "facts_retained": 22, "retention_rate": 0.34375, "retention_by_kind": {"person": 0.5, "date": 0.5555555555555556, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1077}, "5": {"facts_total": 64, "facts_retained": 22, "retention_rate": 0.34375, "retention_by_kind": {"person": 0.5, "date": 0.5555555555555556, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1077}, "6": {"facts_total": 64, "facts_retained": 22, "retention_rate": 0.34375, "retention_by_kind": {"person": 0.5, "date": 0.5555555555555556, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1077}, "7": {"facts_total": 64, "facts_retained": 22, "retention_rate": 0.34375, "retention_by_kind": {"person": 0.5, "date": 0.5555555555555556, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1077}, "8": {"facts_total": 64, "facts_retained": 22, "retention_rate": 0.34375, "retention_by_kind": {"person": 0.5, "date": 0.5555555555555556, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1077}}}, "reference_key_id": "keys/synthetic/syn_03018_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 0.8571428571428571}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "fb3497132b656390ec94098589be8cebe46dd0f8736439bf4730fca3b2cc4f4e", "output_hash": "b58829a1440993c4af3307ba44494f021d726fd11e9337888b6b852a8b37351a"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "cafc7ee26cb150cb37a98d62", "attempt_id": "1a0098d6569e", "run_group_id": "E4_core:A4|repairable|qwen3b:h8", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:13:13.365125Z", "task_family": "synthetic", "document_id": "syn_03018_2000", "document_hash": "d284eb07e27da742b9c2b67abccfdf72e4dd67c3259f44f3b41e34dd3e6fc1ff", "question_id": "q02_exact_copy", "split": "test", "cluster_id": "syn_03018_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 8, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1982, "query_tokens": 582, "message_tokens": 494, "message_utf8_bytes": 1077, "cache_tensor_bytes": null, "serialized_payload_bytes": 1077, "side_information_bytes": 0, "fetched_bytes": 5082, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 79571.560457116, "hops": {"1": 22880.75100001879, "2": 19076.845791074447, "3": 6260.933833080344, "4": 6260.237207985483, "5": 6275.52599995397, "6": 6274.316166993231, "7": 6269.359500030987, "8": 6273.59095797874}, "receiver_prefill_ms": 226.33020905777812, "ttft_ms": 11.960749980062246, "decode_ms": 11.899375007487833}, "total_wall_ms": 79822.23566516768, "ttft_ms": 11.960749980062246, "input_usage": {"lookups": 28}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 64, "facts_retained": 22, "retention_rate": 0.34375, "retention_by_kind": {"person": 0.5, "date": 0.5555555555555556, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1077}, "ledger_by_hop": {"1": {"facts_total": 64, "facts_retained": 22, "retention_rate": 0.34375, "retention_by_kind": {"person": 0.5, "date": 0.5555555555555556, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1181}, "2": {"facts_total": 64, "facts_retained": 22, "retention_rate": 0.34375, "retention_by_kind": {"person": 0.5, "date": 0.5555555555555556, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1081}, "3": {"facts_total": 64, "facts_retained": 22, "retention_rate": 0.34375, "retention_by_kind": {"person": 0.5, "date": 0.5555555555555556, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1077}, "4": {"facts_total": 64, "facts_retained": 22, "retention_rate": 0.34375, "retention_by_kind": {"person": 0.5, "date": 0.5555555555555556, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1077}, "5": {"facts_total": 64, "facts_retained": 22, "retention_rate": 0.34375, "retention_by_kind": {"person": 0.5, "date": 0.5555555555555556, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1077}, "6": {"facts_total": 64, "facts_retained": 22, "retention_rate": 0.34375, "retention_by_kind": {"person": 0.5, "date": 0.5555555555555556, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1077}, "7": {"facts_total": 64, "facts_retained": 22, "retention_rate": 0.34375, "retention_by_kind": {"person": 0.5, "date": 0.5555555555555556, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1077}, "8": {"facts_total": 64, "facts_retained": 22, "retention_rate": 0.34375, "retention_by_kind": {"person": 0.5, "date": 0.5555555555555556, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1077}}}, "reference_key_id": "keys/synthetic/syn_03018_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "fb3497132b656390ec94098589be8cebe46dd0f8736439bf4730fca3b2cc4f4e", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "4f29a12672257861620c78b4", "attempt_id": "66a9d34cf4c3", "run_group_id": "E4_core:A4|repairable|qwen3b:h8", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:13:13.631534Z", "task_family": "synthetic", "document_id": "syn_03018_2000", "document_hash": "d284eb07e27da742b9c2b67abccfdf72e4dd67c3259f44f3b41e34dd3e6fc1ff", "question_id": "q03_exact_copy", "split": "test", "cluster_id": "syn_03018_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 8, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1982, "query_tokens": 580, "message_tokens": 494, "message_utf8_bytes": 1077, "cache_tensor_bytes": null, "serialized_payload_bytes": 1077, "side_information_bytes": 0, "fetched_bytes": 5082, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 79571.560457116, "hops": {"1": 22880.75100001879, "2": 19076.845791074447, "3": 6260.933833080344, "4": 6260.237207985483, "5": 6275.52599995397, "6": 6274.316166993231, "7": 6269.359500030987, "8": 6273.59095797874}, "receiver_prefill_ms": 225.8349999319762, "ttft_ms": 11.982292053289711, "decode_ms": 106.7875410662964}, "total_wall_ms": 79916.64899815805, "ttft_ms": 11.982292053289711, "input_usage": {"lookups": 28}, "output_usage": {"output_tokens": 9}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "Priyamvada Sørlie", "evidence_artifact": {"ledger": {"facts_total": 64, "facts_retained": 22, "retention_rate": 0.34375, "retention_by_kind": {"person": 0.5, "date": 0.5555555555555556, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1077}, "ledger_by_hop": {"1": {"facts_total": 64, "facts_retained": 22, "retention_rate": 0.34375, "retention_by_kind": {"person": 0.5, "date": 0.5555555555555556, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1181}, "2": {"facts_total": 64, "facts_retained": 22, "retention_rate": 0.34375, "retention_by_kind": {"person": 0.5, "date": 0.5555555555555556, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1081}, "3": {"facts_total": 64, "facts_retained": 22, "retention_rate": 0.34375, "retention_by_kind": {"person": 0.5, "date": 0.5555555555555556, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1077}, "4": {"facts_total": 64, "facts_retained": 22, "retention_rate": 0.34375, "retention_by_kind": {"person": 0.5, "date": 0.5555555555555556, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1077}, "5": {"facts_total": 64, "facts_retained": 22, "retention_rate": 0.34375, "retention_by_kind": {"person": 0.5, "date": 0.5555555555555556, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1077}, "6": {"facts_total": 64, "facts_retained": 22, "retention_rate": 0.34375, "retention_by_kind": {"person": 0.5, "date": 0.5555555555555556, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1077}, "7": {"facts_total": 64, "facts_retained": 22, "retention_rate": 0.34375, "retention_by_kind": {"person": 0.5, "date": 0.5555555555555556, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1077}, "8": {"facts_total": 64, "facts_retained": 22, "retention_rate": 0.34375, "retention_by_kind": {"person": 0.5, "date": 0.5555555555555556, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1077}}}, "reference_key_id": "keys/synthetic/syn_03018_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "fb3497132b656390ec94098589be8cebe46dd0f8736439bf4730fca3b2cc4f4e", "output_hash": "c5f554196ed23dd71ad995ec89f7142099d524dbb6534cc10d94a0c89d13e792"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "757672f4d19b432202bfe0ff", "attempt_id": "826725ee6ab6", "run_group_id": "E4_core:A4|repairable|qwen3b:h8", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:13:13.991396Z", "task_family": "synthetic", "document_id": "syn_03018_2000", "document_hash": "d284eb07e27da742b9c2b67abccfdf72e4dd67c3259f44f3b41e34dd3e6fc1ff", "question_id": "q04_multi_hop", "split": "test", "cluster_id": "syn_03018_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 8, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1982, "query_tokens": 586, "message_tokens": 494, "message_utf8_bytes": 1077, "cache_tensor_bytes": null, "serialized_payload_bytes": 1077, "side_information_bytes": 0, "fetched_bytes": 5082, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 79571.560457116, "hops": {"1": 22880.75100001879, "2": 19076.845791074447, "3": 6260.933833080344, "4": 6260.237207985483, "5": 6275.52599995397, "6": 6274.316166993231, "7": 6269.359500030987, "8": 6273.59095797874}, "receiver_prefill_ms": 226.9362909719348, "ttft_ms": 11.92866696510464, "decode_ms": 12.017959030345082}, "total_wall_ms": 79822.92933214922, "ttft_ms": 11.92866696510464, "input_usage": {"lookups": 28}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 64, "facts_retained": 22, "retention_rate": 0.34375, "retention_by_kind": {"person": 0.5, "date": 0.5555555555555556, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1077}, "ledger_by_hop": {"1": {"facts_total": 64, "facts_retained": 22, "retention_rate": 0.34375, "retention_by_kind": {"person": 0.5, "date": 0.5555555555555556, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1181}, "2": {"facts_total": 64, "facts_retained": 22, "retention_rate": 0.34375, "retention_by_kind": {"person": 0.5, "date": 0.5555555555555556, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1081}, "3": {"facts_total": 64, "facts_retained": 22, "retention_rate": 0.34375, "retention_by_kind": {"person": 0.5, "date": 0.5555555555555556, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1077}, "4": {"facts_total": 64, "facts_retained": 22, "retention_rate": 0.34375, "retention_by_kind": {"person": 0.5, "date": 0.5555555555555556, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1077}, "5": {"facts_total": 64, "facts_retained": 22, "retention_rate": 0.34375, "retention_by_kind": {"person": 0.5, "date": 0.5555555555555556, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1077}, "6": {"facts_total": 64, "facts_retained": 22, "retention_rate": 0.34375, "retention_by_kind": {"person": 0.5, "date": 0.5555555555555556, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1077}, "7": {"facts_total": 64, "facts_retained": 22, "retention_rate": 0.34375, "retention_by_kind": {"person": 0.5, "date": 0.5555555555555556, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1077}, "8": {"facts_total": 64, "facts_retained": 22, "retention_rate": 0.34375, "retention_by_kind": {"person": 0.5, "date": 0.5555555555555556, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1077}}}, "reference_key_id": "keys/synthetic/syn_03018_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "fb3497132b656390ec94098589be8cebe46dd0f8736439bf4730fca3b2cc4f4e", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "5f5067d55e9d4b938800855a", "attempt_id": "28d5622baa08", "run_group_id": "E4_core:A4|repairable|qwen3b:h8", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:13:14.258143Z", "task_family": "synthetic", "document_id": "syn_03018_2000", "document_hash": "d284eb07e27da742b9c2b67abccfdf72e4dd67c3259f44f3b41e34dd3e6fc1ff", "question_id": "q05_multi_hop", "split": "test", "cluster_id": "syn_03018_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 8, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1982, "query_tokens": 586, "message_tokens": 494, "message_utf8_bytes": 1077, "cache_tensor_bytes": null, "serialized_payload_bytes": 1077, "side_information_bytes": 0, "fetched_bytes": 5082, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 79571.560457116, "hops": {"1": 22880.75100001879, "2": 19076.845791074447, "3": 6260.933833080344, "4": 6260.237207985483, "5": 6275.52599995397, "6": 6274.316166993231, "7": 6269.359500030987, "8": 6273.59095797874}, "receiver_prefill_ms": 226.71654098667204, "ttft_ms": 11.964999954216182, "decode_ms": 118.932000012137}, "total_wall_ms": 79929.62595715653, "ttft_ms": 11.964999954216182, "input_usage": {"lookups": 28}, "output_usage": {"output_tokens": 10}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "HM-38K-5533", "evidence_artifact": {"ledger": {"facts_total": 64, "facts_retained": 22, "retention_rate": 0.34375, "retention_by_kind": {"person": 0.5, "date": 0.5555555555555556, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1077}, "ledger_by_hop": {"1": {"facts_total": 64, "facts_retained": 22, "retention_rate": 0.34375, "retention_by_kind": {"person": 0.5, "date": 0.5555555555555556, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1181}, "2": {"facts_total": 64, "facts_retained": 22, "retention_rate": 0.34375, "retention_by_kind": {"person": 0.5, "date": 0.5555555555555556, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1081}, "3": {"facts_total": 64, "facts_retained": 22, "retention_rate": 0.34375, "retention_by_kind": {"person": 0.5, "date": 0.5555555555555556, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1077}, "4": {"facts_total": 64, "facts_retained": 22, "retention_rate": 0.34375, "retention_by_kind": {"person": 0.5, "date": 0.5555555555555556, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1077}, "5": {"facts_total": 64, "facts_retained": 22, "retention_rate": 0.34375, "retention_by_kind": {"person": 0.5, "date": 0.5555555555555556, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1077}, "6": {"facts_total": 64, "facts_retained": 22, "retention_rate": 0.34375, "retention_by_kind": {"person": 0.5, "date": 0.5555555555555556, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1077}, "7": {"facts_total": 64, "facts_retained": 22, "retention_rate": 0.34375, "retention_by_kind": {"person": 0.5, "date": 0.5555555555555556, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1077}, "8": {"facts_total": 64, "facts_retained": 22, "retention_rate": 0.34375, "retention_by_kind": {"person": 0.5, "date": 0.5555555555555556, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1077}}}, "reference_key_id": "keys/synthetic/syn_03018_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "fb3497132b656390ec94098589be8cebe46dd0f8736439bf4730fca3b2cc4f4e", "output_hash": "18acd9807857243bbded3bceae55398316a3958b81d70364441bdb6748eaf878"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "22dfdd73264cd2b051bc95f6", "attempt_id": "1990863d9495", "run_group_id": "E4_core:A4|repairable|qwen3b:h8", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:13:14.631397Z", "task_family": "synthetic", "document_id": "syn_03018_2000", "document_hash": "d284eb07e27da742b9c2b67abccfdf72e4dd67c3259f44f3b41e34dd3e6fc1ff", "question_id": "q06_negation", "split": "test", "cluster_id": "syn_03018_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 8, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1982, "query_tokens": 578, "message_tokens": 494, "message_utf8_bytes": 1077, "cache_tensor_bytes": null, "serialized_payload_bytes": 1077, "side_information_bytes": 0, "fetched_bytes": 5082, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 79571.560457116, "hops": {"1": 22880.75100001879, "2": 19076.845791074447, "3": 6260.933833080344, "4": 6260.237207985483, "5": 6275.52599995397, "6": 6274.316166993231, "7": 6269.359500030987, "8": 6273.59095797874}, "receiver_prefill_ms": 226.0989589849487, "ttft_ms": 11.972708976827562, "decode_ms": 11.934042093344033}, "total_wall_ms": 79822.01266509946, "ttft_ms": 11.972708976827562, "input_usage": {"lookups": 28}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 64, "facts_retained": 22, "retention_rate": 0.34375, "retention_by_kind": {"person": 0.5, "date": 0.5555555555555556, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1077}, "ledger_by_hop": {"1": {"facts_total": 64, "facts_retained": 22, "retention_rate": 0.34375, "retention_by_kind": {"person": 0.5, "date": 0.5555555555555556, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1181}, "2": {"facts_total": 64, "facts_retained": 22, "retention_rate": 0.34375, "retention_by_kind": {"person": 0.5, "date": 0.5555555555555556, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1081}, "3": {"facts_total": 64, "facts_retained": 22, "retention_rate": 0.34375, "retention_by_kind": {"person": 0.5, "date": 0.5555555555555556, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1077}, "4": {"facts_total": 64, "facts_retained": 22, "retention_rate": 0.34375, "retention_by_kind": {"person": 0.5, "date": 0.5555555555555556, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1077}, "5": {"facts_total": 64, "facts_retained": 22, "retention_rate": 0.34375, "retention_by_kind": {"person": 0.5, "date": 0.5555555555555556, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1077}, "6": {"facts_total": 64, "facts_retained": 22, "retention_rate": 0.34375, "retention_by_kind": {"person": 0.5, "date": 0.5555555555555556, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1077}, "7": {"facts_total": 64, "facts_retained": 22, "retention_rate": 0.34375, "retention_by_kind": {"person": 0.5, "date": 0.5555555555555556, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1077}, "8": {"facts_total": 64, "facts_retained": 22, "retention_rate": 0.34375, "retention_by_kind": {"person": 0.5, "date": 0.5555555555555556, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1077}}}, "reference_key_id": "keys/synthetic/syn_03018_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "fb3497132b656390ec94098589be8cebe46dd0f8736439bf4730fca3b2cc4f4e", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "a69b6c16174b5286c0d43ef9", "attempt_id": "f7004be08703", "run_group_id": "E4_core:A4|repairable|qwen3b:h8", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:13:14.896802Z", "task_family": "synthetic", "document_id": "syn_03018_2000", "document_hash": "d284eb07e27da742b9c2b67abccfdf72e4dd67c3259f44f3b41e34dd3e6fc1ff", "question_id": "q07_negation", "split": "test", "cluster_id": "syn_03018_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 8, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1982, "query_tokens": 577, "message_tokens": 494, "message_utf8_bytes": 1077, "cache_tensor_bytes": null, "serialized_payload_bytes": 1077, "side_information_bytes": 0, "fetched_bytes": 5082, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 79571.560457116, "hops": {"1": 22880.75100001879, "2": 19076.845791074447, "3": 6260.933833080344, "4": 6260.237207985483, "5": 6275.52599995397, "6": 6274.316166993231, "7": 6269.359500030987, "8": 6273.59095797874}, "receiver_prefill_ms": 209.01737501844764, "ttft_ms": 11.981290997937322, "decode_ms": 11.90829195547849}, "total_wall_ms": 79804.91091613658, "ttft_ms": 11.981290997937322, "input_usage": {"lookups": 28}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 64, "facts_retained": 22, "retention_rate": 0.34375, "retention_by_kind": {"person": 0.5, "date": 0.5555555555555556, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1077}, "ledger_by_hop": {"1": {"facts_total": 64, "facts_retained": 22, "retention_rate": 0.34375, "retention_by_kind": {"person": 0.5, "date": 0.5555555555555556, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1181}, "2": {"facts_total": 64, "facts_retained": 22, "retention_rate": 0.34375, "retention_by_kind": {"person": 0.5, "date": 0.5555555555555556, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1081}, "3": {"facts_total": 64, "facts_retained": 22, "retention_rate": 0.34375, "retention_by_kind": {"person": 0.5, "date": 0.5555555555555556, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1077}, "4": {"facts_total": 64, "facts_retained": 22, "retention_rate": 0.34375, "retention_by_kind": {"person": 0.5, "date": 0.5555555555555556, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1077}, "5": {"facts_total": 64, "facts_retained": 22, "retention_rate": 0.34375, "retention_by_kind": {"person": 0.5, "date": 0.5555555555555556, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1077}, "6": {"facts_total": 64, "facts_retained": 22, "retention_rate": 0.34375, "retention_by_kind": {"person": 0.5, "date": 0.5555555555555556, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1077}, "7": {"facts_total": 64, "facts_retained": 22, "retention_rate": 0.34375, "retention_by_kind": {"person": 0.5, "date": 0.5555555555555556, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1077}, "8": {"facts_total": 64, "facts_retained": 22, "retention_rate": 0.34375, "retention_by_kind": {"person": 0.5, "date": 0.5555555555555556, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1077}}}, "reference_key_id": "keys/synthetic/syn_03018_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "fb3497132b656390ec94098589be8cebe46dd0f8736439bf4730fca3b2cc4f4e", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "228a2e81c9819b729b1afd9b", "attempt_id": "7c64f49efd48", "run_group_id": "E4_core:A4|repairable|qwen3b:h8", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:13:15.144697Z", "task_family": "synthetic", "document_id": "syn_03018_2000", "document_hash": "d284eb07e27da742b9c2b67abccfdf72e4dd67c3259f44f3b41e34dd3e6fc1ff", "question_id": "q08_revision", "split": "test", "cluster_id": "syn_03018_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 8, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1982, "query_tokens": 595, "message_tokens": 494, "message_utf8_bytes": 1077, "cache_tensor_bytes": null, "serialized_payload_bytes": 1077, "side_information_bytes": 0, "fetched_bytes": 5082, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 79571.560457116, "hops": {"1": 22880.75100001879, "2": 19076.845791074447, "3": 6260.933833080344, "4": 6260.237207985483, "5": 6275.52599995397, "6": 6274.316166993231, "7": 6269.359500030987, "8": 6273.59095797874}, "receiver_prefill_ms": 226.85708303470165, "ttft_ms": 11.93816598970443, "decode_ms": 71.4245829731226}, "total_wall_ms": 79882.26508221123, "ttft_ms": 11.93816598970443, "input_usage": {"lookups": 28}, "output_usage": {"output_tokens": 6}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "304815", "evidence_artifact": {"ledger": {"facts_total": 64, "facts_retained": 22, "retention_rate": 0.34375, "retention_by_kind": {"person": 0.5, "date": 0.5555555555555556, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1077}, "ledger_by_hop": {"1": {"facts_total": 64, "facts_retained": 22, "retention_rate": 0.34375, "retention_by_kind": {"person": 0.5, "date": 0.5555555555555556, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1181}, "2": {"facts_total": 64, "facts_retained": 22, "retention_rate": 0.34375, "retention_by_kind": {"person": 0.5, "date": 0.5555555555555556, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1081}, "3": {"facts_total": 64, "facts_retained": 22, "retention_rate": 0.34375, "retention_by_kind": {"person": 0.5, "date": 0.5555555555555556, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1077}, "4": {"facts_total": 64, "facts_retained": 22, "retention_rate": 0.34375, "retention_by_kind": {"person": 0.5, "date": 0.5555555555555556, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1077}, "5": {"facts_total": 64, "facts_retained": 22, "retention_rate": 0.34375, "retention_by_kind": {"person": 0.5, "date": 0.5555555555555556, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1077}, "6": {"facts_total": 64, "facts_retained": 22, "retention_rate": 0.34375, "retention_by_kind": {"person": 0.5, "date": 0.5555555555555556, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1077}, "7": {"facts_total": 64, "facts_retained": 22, "retention_rate": 0.34375, "retention_by_kind": {"person": 0.5, "date": 0.5555555555555556, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1077}, "8": {"facts_total": 64, "facts_retained": 22, "retention_rate": 0.34375, "retention_by_kind": {"person": 0.5, "date": 0.5555555555555556, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1077}}}, "reference_key_id": "keys/synthetic/syn_03018_2000.json", "task_scores": {"correct": 0.0, "abstained": 0.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "fb3497132b656390ec94098589be8cebe46dd0f8736439bf4730fca3b2cc4f4e", "output_hash": "a480475d6855141c920f42d1bf3337ca7fd88bcf4087b0ed2e253ed68cca6926"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "bc298885df3aa92dca073732", "attempt_id": "7d2d06e095f5", "run_group_id": "E4_core:A4|repairable|qwen3b:h8", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:13:15.470466Z", "task_family": "synthetic", "document_id": "syn_03018_2000", "document_hash": "d284eb07e27da742b9c2b67abccfdf72e4dd67c3259f44f3b41e34dd3e6fc1ff", "question_id": "q09_revision", "split": "test", "cluster_id": "syn_03018_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 8, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1982, "query_tokens": 588, "message_tokens": 494, "message_utf8_bytes": 1077, "cache_tensor_bytes": null, "serialized_payload_bytes": 1077, "side_information_bytes": 0, "fetched_bytes": 5082, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 79571.560457116, "hops": {"1": 22880.75100001879, "2": 19076.845791074447, "3": 6260.933833080344, "4": 6260.237207985483, "5": 6275.52599995397, "6": 6274.316166993231, "7": 6269.359500030987, "8": 6273.59095797874}, "receiver_prefill_ms": 226.85800003819168, "ttft_ms": 11.973500018939376, "decode_ms": 83.18349998444319}, "total_wall_ms": 79894.03537404723, "ttft_ms": 11.973500018939376, "input_usage": {"lookups": 28}, "output_usage": {"output_tokens": 7}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "276,340", "evidence_artifact": {"ledger": {"facts_total": 64, "facts_retained": 22, "retention_rate": 0.34375, "retention_by_kind": {"person": 0.5, "date": 0.5555555555555556, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1077}, "ledger_by_hop": {"1": {"facts_total": 64, "facts_retained": 22, "retention_rate": 0.34375, "retention_by_kind": {"person": 0.5, "date": 0.5555555555555556, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1181}, "2": {"facts_total": 64, "facts_retained": 22, "retention_rate": 0.34375, "retention_by_kind": {"person": 0.5, "date": 0.5555555555555556, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1081}, "3": {"facts_total": 64, "facts_retained": 22, "retention_rate": 0.34375, "retention_by_kind": {"person": 0.5, "date": 0.5555555555555556, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1077}, "4": {"facts_total": 64, "facts_retained": 22, "retention_rate": 0.34375, "retention_by_kind": {"person": 0.5, "date": 0.5555555555555556, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1077}, "5": {"facts_total": 64, "facts_retained": 22, "retention_rate": 0.34375, "retention_by_kind": {"person": 0.5, "date": 0.5555555555555556, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1077}, "6": {"facts_total": 64, "facts_retained": 22, "retention_rate": 0.34375, "retention_by_kind": {"person": 0.5, "date": 0.5555555555555556, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1077}, "7": {"facts_total": 64, "facts_retained": 22, "retention_rate": 0.34375, "retention_by_kind": {"person": 0.5, "date": 0.5555555555555556, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1077}, "8": {"facts_total": 64, "facts_retained": 22, "retention_rate": 0.34375, "retention_by_kind": {"person": 0.5, "date": 0.5555555555555556, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1077}}}, "reference_key_id": "keys/synthetic/syn_03018_2000.json", "task_scores": {"correct": 0.0, "abstained": 0.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "fb3497132b656390ec94098589be8cebe46dd0f8736439bf4730fca3b2cc4f4e", "output_hash": "ef0dffffbd6c962fe44611aa2052fb7b78efaa1203b0a5de9937b8bd584d71c2"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "70c6e6a06ec1c0db8bf7abe8", "attempt_id": "5c58d0cc5b88", "run_group_id": "E4_core:A4|repairable|qwen3b:h8", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:13:15.808044Z", "task_family": "synthetic", "document_id": "syn_03018_2000", "document_hash": "d284eb07e27da742b9c2b67abccfdf72e4dd67c3259f44f3b41e34dd3e6fc1ff", "question_id": "q10_temporal", "split": "test", "cluster_id": "syn_03018_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 8, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1982, "query_tokens": 596, "message_tokens": 494, "message_utf8_bytes": 1077, "cache_tensor_bytes": null, "serialized_payload_bytes": 1077, "side_information_bytes": 0, "fetched_bytes": 5082, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 79571.560457116, "hops": {"1": 22880.75100001879, "2": 19076.845791074447, "3": 6260.933833080344, "4": 6260.237207985483, "5": 6275.52599995397, "6": 6274.316166993231, "7": 6269.359500030987, "8": 6273.59095797874}, "receiver_prefill_ms": 226.6878749942407, "ttft_ms": 11.98008400388062, "decode_ms": 71.63566700182855}, "total_wall_ms": 79882.32016505208, "ttft_ms": 11.98008400388062, "input_usage": {"lookups": 28}, "output_usage": {"output_tokens": 6}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "Melymi-15", "evidence_artifact": {"ledger": {"facts_total": 64, "facts_retained": 22, "retention_rate": 0.34375, "retention_by_kind": {"person": 0.5, "date": 0.5555555555555556, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1077}, "ledger_by_hop": {"1": {"facts_total": 64, "facts_retained": 22, "retention_rate": 0.34375, "retention_by_kind": {"person": 0.5, "date": 0.5555555555555556, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1181}, "2": {"facts_total": 64, "facts_retained": 22, "retention_rate": 0.34375, "retention_by_kind": {"person": 0.5, "date": 0.5555555555555556, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1081}, "3": {"facts_total": 64, "facts_retained": 22, "retention_rate": 0.34375, "retention_by_kind": {"person": 0.5, "date": 0.5555555555555556, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1077}, "4": {"facts_total": 64, "facts_retained": 22, "retention_rate": 0.34375, "retention_by_kind": {"person": 0.5, "date": 0.5555555555555556, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1077}, "5": {"facts_total": 64, "facts_retained": 22, "retention_rate": 0.34375, "retention_by_kind": {"person": 0.5, "date": 0.5555555555555556, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1077}, "6": {"facts_total": 64, "facts_retained": 22, "retention_rate": 0.34375, "retention_by_kind": {"person": 0.5, "date": 0.5555555555555556, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1077}, "7": {"facts_total": 64, "facts_retained": 22, "retention_rate": 0.34375, "retention_by_kind": {"person": 0.5, "date": 0.5555555555555556, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1077}, "8": {"facts_total": 64, "facts_retained": 22, "retention_rate": 0.34375, "retention_by_kind": {"person": 0.5, "date": 0.5555555555555556, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1077}}}, "reference_key_id": "keys/synthetic/syn_03018_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "fb3497132b656390ec94098589be8cebe46dd0f8736439bf4730fca3b2cc4f4e", "output_hash": "fa08d1474b01d6dfbf3dea049be251aa3c3021c17b4818258064b156509af58b"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "db3fbe1296a3d4262100a412", "attempt_id": "99fbe0b3abb3", "run_group_id": "E4_core:A4|repairable|qwen3b:h8", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:13:16.133709Z", "task_family": "synthetic", "document_id": "syn_03018_2000", "document_hash": "d284eb07e27da742b9c2b67abccfdf72e4dd67c3259f44f3b41e34dd3e6fc1ff", "question_id": "q11_aggregation", "split": "test", "cluster_id": "syn_03018_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 8, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1982, "query_tokens": 588, "message_tokens": 494, "message_utf8_bytes": 1077, "cache_tensor_bytes": null, "serialized_payload_bytes": 1077, "side_information_bytes": 0, "fetched_bytes": 5082, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 79571.560457116, "hops": {"1": 22880.75100001879, "2": 19076.845791074447, "3": 6260.933833080344, "4": 6260.237207985483, "5": 6275.52599995397, "6": 6274.316166993231, "7": 6269.359500030987, "8": 6273.59095797874}, "receiver_prefill_ms": 226.89175000414252, "ttft_ms": 11.910874978639185, "decode_ms": 83.34908401593566}, "total_wall_ms": 79894.16858216282, "ttft_ms": 11.910874978639185, "input_usage": {"lookups": 28}, "output_usage": {"output_tokens": 7}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "2414061", "evidence_artifact": {"ledger": {"facts_total": 64, "facts_retained": 22, "retention_rate": 0.34375, "retention_by_kind": {"person": 0.5, "date": 0.5555555555555556, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1077}, "ledger_by_hop": {"1": {"facts_total": 64, "facts_retained": 22, "retention_rate": 0.34375, "retention_by_kind": {"person": 0.5, "date": 0.5555555555555556, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1181}, "2": {"facts_total": 64, "facts_retained": 22, "retention_rate": 0.34375, "retention_by_kind": {"person": 0.5, "date": 0.5555555555555556, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1081}, "3": {"facts_total": 64, "facts_retained": 22, "retention_rate": 0.34375, "retention_by_kind": {"person": 0.5, "date": 0.5555555555555556, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1077}, "4": {"facts_total": 64, "facts_retained": 22, "retention_rate": 0.34375, "retention_by_kind": {"person": 0.5, "date": 0.5555555555555556, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1077}, "5": {"facts_total": 64, "facts_retained": 22, "retention_rate": 0.34375, "retention_by_kind": {"person": 0.5, "date": 0.5555555555555556, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1077}, "6": {"facts_total": 64, "facts_retained": 22, "retention_rate": 0.34375, "retention_by_kind": {"person": 0.5, "date": 0.5555555555555556, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1077}, "7": {"facts_total": 64, "facts_retained": 22, "retention_rate": 0.34375, "retention_by_kind": {"person": 0.5, "date": 0.5555555555555556, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1077}, "8": {"facts_total": 64, "facts_retained": 22, "retention_rate": 0.34375, "retention_by_kind": {"person": 0.5, "date": 0.5555555555555556, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1077}}}, "reference_key_id": "keys/synthetic/syn_03018_2000.json", "task_scores": {"correct": 0.0, "abstained": 0.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "fb3497132b656390ec94098589be8cebe46dd0f8736439bf4730fca3b2cc4f4e", "output_hash": "9a9e13b7936cee2163bc948619d2d5fd4768cdd23b133eb9231f47383fc12bf6"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "508dc7df9d75fd7186501c79", "attempt_id": "d207548a4d8c", "run_group_id": "E4_core:A4|repairable|qwen3b:h8", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:13:16.471180Z", "task_family": "synthetic", "document_id": "syn_03018_2000", "document_hash": "d284eb07e27da742b9c2b67abccfdf72e4dd67c3259f44f3b41e34dd3e6fc1ff", "question_id": "q12_unit_conversion", "split": "test", "cluster_id": "syn_03018_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 8, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1982, "query_tokens": 585, "message_tokens": 494, "message_utf8_bytes": 1077, "cache_tensor_bytes": null, "serialized_payload_bytes": 1077, "side_information_bytes": 0, "fetched_bytes": 5082, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 79571.560457116, "hops": {"1": 22880.75100001879, "2": 19076.845791074447, "3": 6260.933833080344, "4": 6260.237207985483, "5": 6275.52599995397, "6": 6274.316166993231, "7": 6269.359500030987, "8": 6273.59095797874}, "receiver_prefill_ms": 226.52500006370246, "ttft_ms": 11.969750048592687, "decode_ms": 11.947875027544796}, "total_wall_ms": 79822.45229114778, "ttft_ms": 11.969750048592687, "input_usage": {"lookups": 28}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 64, "facts_retained": 22, "retention_rate": 0.34375, "retention_by_kind": {"person": 0.5, "date": 0.5555555555555556, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1077}, "ledger_by_hop": {"1": {"facts_total": 64, "facts_retained": 22, "retention_rate": 0.34375, "retention_by_kind": {"person": 0.5, "date": 0.5555555555555556, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1181}, "2": {"facts_total": 64, "facts_retained": 22, "retention_rate": 0.34375, "retention_by_kind": {"person": 0.5, "date": 0.5555555555555556, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1081}, "3": {"facts_total": 64, "facts_retained": 22, "retention_rate": 0.34375, "retention_by_kind": {"person": 0.5, "date": 0.5555555555555556, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1077}, "4": {"facts_total": 64, "facts_retained": 22, "retention_rate": 0.34375, "retention_by_kind": {"person": 0.5, "date": 0.5555555555555556, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1077}, "5": {"facts_total": 64, "facts_retained": 22, "retention_rate": 0.34375, "retention_by_kind": {"person": 0.5, "date": 0.5555555555555556, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1077}, "6": {"facts_total": 64, "facts_retained": 22, "retention_rate": 0.34375, "retention_by_kind": {"person": 0.5, "date": 0.5555555555555556, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1077}, "7": {"facts_total": 64, "facts_retained": 22, "retention_rate": 0.34375, "retention_by_kind": {"person": 0.5, "date": 0.5555555555555556, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1077}, "8": {"facts_total": 64, "facts_retained": 22, "retention_rate": 0.34375, "retention_by_kind": {"person": 0.5, "date": 0.5555555555555556, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1077}}}, "reference_key_id": "keys/synthetic/syn_03018_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "fb3497132b656390ec94098589be8cebe46dd0f8736439bf4730fca3b2cc4f4e", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "d890693c1bfdb6f0944112c1", "attempt_id": "0052aa647103", "run_group_id": "E4_core:A4|repairable|qwen3b:h8", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:13:16.736955Z", "task_family": "synthetic", "document_id": "syn_03018_2000", "document_hash": "d284eb07e27da742b9c2b67abccfdf72e4dd67c3259f44f3b41e34dd3e6fc1ff", "question_id": "q13_unanswerable", "split": "test", "cluster_id": "syn_03018_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 8, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1982, "query_tokens": 589, "message_tokens": 494, "message_utf8_bytes": 1077, "cache_tensor_bytes": null, "serialized_payload_bytes": 1077, "side_information_bytes": 0, "fetched_bytes": 5082, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 79571.560457116, "hops": {"1": 22880.75100001879, "2": 19076.845791074447, "3": 6260.933833080344, "4": 6260.237207985483, "5": 6275.52599995397, "6": 6274.316166993231, "7": 6269.359500030987, "8": 6273.59095797874}, "receiver_prefill_ms": 227.1055830642581, "ttft_ms": 11.949167004786432, "decode_ms": 11.901834048330784}, "total_wall_ms": 79822.98283209093, "ttft_ms": 11.949167004786432, "input_usage": {"lookups": 28}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 64, "facts_retained": 22, "retention_rate": 0.34375, "retention_by_kind": {"person": 0.5, "date": 0.5555555555555556, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1077}, "ledger_by_hop": {"1": {"facts_total": 64, "facts_retained": 22, "retention_rate": 0.34375, "retention_by_kind": {"person": 0.5, "date": 0.5555555555555556, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1181}, "2": {"facts_total": 64, "facts_retained": 22, "retention_rate": 0.34375, "retention_by_kind": {"person": 0.5, "date": 0.5555555555555556, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1081}, "3": {"facts_total": 64, "facts_retained": 22, "retention_rate": 0.34375, "retention_by_kind": {"person": 0.5, "date": 0.5555555555555556, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1077}, "4": {"facts_total": 64, "facts_retained": 22, "retention_rate": 0.34375, "retention_by_kind": {"person": 0.5, "date": 0.5555555555555556, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1077}, "5": {"facts_total": 64, "facts_retained": 22, "retention_rate": 0.34375, "retention_by_kind": {"person": 0.5, "date": 0.5555555555555556, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1077}, "6": {"facts_total": 64, "facts_retained": 22, "retention_rate": 0.34375, "retention_by_kind": {"person": 0.5, "date": 0.5555555555555556, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1077}, "7": {"facts_total": 64, "facts_retained": 22, "retention_rate": 0.34375, "retention_by_kind": {"person": 0.5, "date": 0.5555555555555556, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1077}, "8": {"facts_total": 64, "facts_retained": 22, "retention_rate": 0.34375, "retention_by_kind": {"person": 0.5, "date": 0.5555555555555556, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1077}}}, "reference_key_id": "keys/synthetic/syn_03018_2000.json", "task_scores": {"correct": 1.0, "abstained": 1.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "fb3497132b656390ec94098589be8cebe46dd0f8736439bf4730fca3b2cc4f4e", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "5e992ab2523dc64e45326ee2", "attempt_id": "46b1b667aa00", "run_group_id": "E4_core:A4|repairable|qwen3b:h8", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:13:17.003272Z", "task_family": "synthetic", "document_id": "syn_03018_2000", "document_hash": "d284eb07e27da742b9c2b67abccfdf72e4dd67c3259f44f3b41e34dd3e6fc1ff", "question_id": "q14_unanswerable", "split": "test", "cluster_id": "syn_03018_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 8, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1982, "query_tokens": 590, "message_tokens": 494, "message_utf8_bytes": 1077, "cache_tensor_bytes": null, "serialized_payload_bytes": 1077, "side_information_bytes": 0, "fetched_bytes": 5082, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 79571.560457116, "hops": {"1": 22880.75100001879, "2": 19076.845791074447, "3": 6260.933833080344, "4": 6260.237207985483, "5": 6275.52599995397, "6": 6274.316166993231, "7": 6269.359500030987, "8": 6273.59095797874}, "receiver_prefill_ms": 227.24654199555516, "ttft_ms": 11.974625056609511, "decode_ms": 11.95666694547981}, "total_wall_ms": 79823.18724913057, "ttft_ms": 11.974625056609511, "input_usage": {"lookups": 28}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 64, "facts_retained": 22, "retention_rate": 0.34375, "retention_by_kind": {"person": 0.5, "date": 0.5555555555555556, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1077}, "ledger_by_hop": {"1": {"facts_total": 64, "facts_retained": 22, "retention_rate": 0.34375, "retention_by_kind": {"person": 0.5, "date": 0.5555555555555556, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1181}, "2": {"facts_total": 64, "facts_retained": 22, "retention_rate": 0.34375, "retention_by_kind": {"person": 0.5, "date": 0.5555555555555556, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1081}, "3": {"facts_total": 64, "facts_retained": 22, "retention_rate": 0.34375, "retention_by_kind": {"person": 0.5, "date": 0.5555555555555556, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1077}, "4": {"facts_total": 64, "facts_retained": 22, "retention_rate": 0.34375, "retention_by_kind": {"person": 0.5, "date": 0.5555555555555556, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1077}, "5": {"facts_total": 64, "facts_retained": 22, "retention_rate": 0.34375, "retention_by_kind": {"person": 0.5, "date": 0.5555555555555556, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1077}, "6": {"facts_total": 64, "facts_retained": 22, "retention_rate": 0.34375, "retention_by_kind": {"person": 0.5, "date": 0.5555555555555556, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1077}, "7": {"facts_total": 64, "facts_retained": 22, "retention_rate": 0.34375, "retention_by_kind": {"person": 0.5, "date": 0.5555555555555556, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1077}, "8": {"facts_total": 64, "facts_retained": 22, "retention_rate": 0.34375, "retention_by_kind": {"person": 0.5, "date": 0.5555555555555556, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1077}}}, "reference_key_id": "keys/synthetic/syn_03018_2000.json", "task_scores": {"correct": 1.0, "abstained": 1.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "fb3497132b656390ec94098589be8cebe46dd0f8736439bf4730fca3b2cc4f4e", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "49d7366a80d12f97ea08f7b3", "attempt_id": "b9256cce42cb", "run_group_id": "E4_core:A5|repairable|qwen3b:h1", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:13:24.194957Z", "task_family": "synthetic", "document_id": "syn_03018_2000", "document_hash": "d284eb07e27da742b9c2b67abccfdf72e4dd67c3259f44f3b41e34dd3e6fc1ff", "question_id": "q00_single_fact", "split": "test", "cluster_id": "syn_03018_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 1, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1982, "query_tokens": 484, "message_tokens": 401, "message_utf8_bytes": 940, "cache_tensor_bytes": null, "serialized_payload_bytes": 940, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 6923.268833081238, "hops": {"1": 6923.268833081238}, "receiver_prefill_ms": 188.24108398985118, "ttft_ms": 11.85308292042464, "decode_ms": 11.844582972116768}, "total_wall_ms": 7135.66499913577, "ttft_ms": 11.85308292042464, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 64, "facts_retained": 14, "retention_rate": 0.21875, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.2222222222222222, "amount": 0.15384615384615385, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 870}, "ledger_by_hop": {"1": {"facts_total": 64, "facts_retained": 14, "retention_rate": 0.21875, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.2222222222222222, "amount": 0.15384615384615385, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 870}}}, "reference_key_id": "keys/synthetic/syn_03018_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "b93ce255451151cce89ccd51cc89c5b4382f883b441b3818bc249f2f79c54d9e", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "c8346e8b9d7553e69fb56ddc", "attempt_id": "237e80df5aee", "run_group_id": "E4_core:A5|repairable|qwen3b:h1", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:13:24.423196Z", "task_family": "synthetic", "document_id": "syn_03018_2000", "document_hash": "d284eb07e27da742b9c2b67abccfdf72e4dd67c3259f44f3b41e34dd3e6fc1ff", "question_id": "q01_single_fact", "split": "test", "cluster_id": "syn_03018_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 1, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1982, "query_tokens": 490, "message_tokens": 401, "message_utf8_bytes": 940, "cache_tensor_bytes": null, "serialized_payload_bytes": 940, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 6923.268833081238, "hops": {"1": 6923.268833081238}, "receiver_prefill_ms": 188.60616604797542, "ttft_ms": 11.896583950147033, "decode_ms": 11.901583056896925}, "total_wall_ms": 7136.064458056353, "ttft_ms": 11.896583950147033, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 64, "facts_retained": 14, "retention_rate": 0.21875, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.2222222222222222, "amount": 0.15384615384615385, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 870}, "ledger_by_hop": {"1": {"facts_total": 64, "facts_retained": 14, "retention_rate": 0.21875, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.2222222222222222, "amount": 0.15384615384615385, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 870}}}, "reference_key_id": "keys/synthetic/syn_03018_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "b93ce255451151cce89ccd51cc89c5b4382f883b441b3818bc249f2f79c54d9e", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "ee6a6de6bd23d624bacd8ff0", "attempt_id": "275b9f74c4ed", "run_group_id": "E4_core:A5|repairable|qwen3b:h1", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:13:24.651069Z", "task_family": "synthetic", "document_id": "syn_03018_2000", "document_hash": "d284eb07e27da742b9c2b67abccfdf72e4dd67c3259f44f3b41e34dd3e6fc1ff", "question_id": "q02_exact_copy", "split": "test", "cluster_id": "syn_03018_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 1, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1982, "query_tokens": 489, "message_tokens": 401, "message_utf8_bytes": 940, "cache_tensor_bytes": null, "serialized_payload_bytes": 940, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 6923.268833081238, "hops": {"1": 6923.268833081238}, "receiver_prefill_ms": 188.58495797030628, "ttft_ms": 11.965416022576392, "decode_ms": 11.866291984915733}, "total_wall_ms": 7136.113583110273, "ttft_ms": 11.965416022576392, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 64, "facts_retained": 14, "retention_rate": 0.21875, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.2222222222222222, "amount": 0.15384615384615385, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 870}, "ledger_by_hop": {"1": {"facts_total": 64, "facts_retained": 14, "retention_rate": 0.21875, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.2222222222222222, "amount": 0.15384615384615385, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 870}}}, "reference_key_id": "keys/synthetic/syn_03018_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "b93ce255451151cce89ccd51cc89c5b4382f883b441b3818bc249f2f79c54d9e", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "f1f57ef706fe3e9b3bebeda6", "attempt_id": "c8129f648647", "run_group_id": "E4_core:A5|repairable|qwen3b:h1", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:13:24.879082Z", "task_family": "synthetic", "document_id": "syn_03018_2000", "document_hash": "d284eb07e27da742b9c2b67abccfdf72e4dd67c3259f44f3b41e34dd3e6fc1ff", "question_id": "q03_exact_copy", "split": "test", "cluster_id": "syn_03018_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 1, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1982, "query_tokens": 487, "message_tokens": 401, "message_utf8_bytes": 940, "cache_tensor_bytes": null, "serialized_payload_bytes": 940, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 6923.268833081238, "hops": {"1": 6923.268833081238}, "receiver_prefill_ms": 185.16691704280674, "ttft_ms": 11.931749992072582, "decode_ms": 106.64000001270324}, "total_wall_ms": 7227.409416111186, "ttft_ms": 11.931749992072582, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 9}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "Priyamvada Sørlie", "evidence_artifact": {"ledger": {"facts_total": 64, "facts_retained": 14, "retention_rate": 0.21875, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.2222222222222222, "amount": 0.15384615384615385, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 870}, "ledger_by_hop": {"1": {"facts_total": 64, "facts_retained": 14, "retention_rate": 0.21875, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.2222222222222222, "amount": 0.15384615384615385, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 870}}}, "reference_key_id": "keys/synthetic/syn_03018_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "b93ce255451151cce89ccd51cc89c5b4382f883b441b3818bc249f2f79c54d9e", "output_hash": "c5f554196ed23dd71ad995ec89f7142099d524dbb6534cc10d94a0c89d13e792"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "12c9981daa5fce7083faa799", "attempt_id": "2207d8b445e9", "run_group_id": "E4_core:A5|repairable|qwen3b:h1", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:13:25.197841Z", "task_family": "synthetic", "document_id": "syn_03018_2000", "document_hash": "d284eb07e27da742b9c2b67abccfdf72e4dd67c3259f44f3b41e34dd3e6fc1ff", "question_id": "q04_multi_hop", "split": "test", "cluster_id": "syn_03018_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 1, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1982, "query_tokens": 493, "message_tokens": 401, "message_utf8_bytes": 940, "cache_tensor_bytes": null, "serialized_payload_bytes": 940, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 6923.268833081238, "hops": {"1": 6923.268833081238}, "receiver_prefill_ms": 188.7896660482511, "ttft_ms": 11.872290982864797, "decode_ms": 11.85900007840246}, "total_wall_ms": 7136.196749168448, "ttft_ms": 11.872290982864797, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 64, "facts_retained": 14, "retention_rate": 0.21875, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.2222222222222222, "amount": 0.15384615384615385, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 870}, "ledger_by_hop": {"1": {"facts_total": 64, "facts_retained": 14, "retention_rate": 0.21875, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.2222222222222222, "amount": 0.15384615384615385, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 870}}}, "reference_key_id": "keys/synthetic/syn_03018_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "b93ce255451151cce89ccd51cc89c5b4382f883b441b3818bc249f2f79c54d9e", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "bb649e47485602450c731fae", "attempt_id": "2e493d3175dc", "run_group_id": "E4_core:A5|repairable|qwen3b:h1", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:13:25.425971Z", "task_family": "synthetic", "document_id": "syn_03018_2000", "document_hash": "d284eb07e27da742b9c2b67abccfdf72e4dd67c3259f44f3b41e34dd3e6fc1ff", "question_id": "q05_multi_hop", "split": "test", "cluster_id": "syn_03018_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 1, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1982, "query_tokens": 493, "message_tokens": 401, "message_utf8_bytes": 940, "cache_tensor_bytes": null, "serialized_payload_bytes": 940, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 6923.268833081238, "hops": {"1": 6923.268833081238}, "receiver_prefill_ms": 188.3699590107426, "ttft_ms": 11.907791951671243, "decode_ms": 11.840958031825721}, "total_wall_ms": 7135.774500085972, "ttft_ms": 11.907791951671243, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 64, "facts_retained": 14, "retention_rate": 0.21875, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.2222222222222222, "amount": 0.15384615384615385, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 870}, "ledger_by_hop": {"1": {"facts_total": 64, "facts_retained": 14, "retention_rate": 0.21875, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.2222222222222222, "amount": 0.15384615384615385, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 870}}}, "reference_key_id": "keys/synthetic/syn_03018_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "b93ce255451151cce89ccd51cc89c5b4382f883b441b3818bc249f2f79c54d9e", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "0cf1f20299a969e7012b006c", "attempt_id": "3eb7da7c1670", "run_group_id": "E4_core:A5|repairable|qwen3b:h1", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:13:25.653272Z", "task_family": "synthetic", "document_id": "syn_03018_2000", "document_hash": "d284eb07e27da742b9c2b67abccfdf72e4dd67c3259f44f3b41e34dd3e6fc1ff", "question_id": "q06_negation", "split": "test", "cluster_id": "syn_03018_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 1, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1982, "query_tokens": 485, "message_tokens": 401, "message_utf8_bytes": 940, "cache_tensor_bytes": null, "serialized_payload_bytes": 940, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 6923.268833081238, "hops": {"1": 6923.268833081238}, "receiver_prefill_ms": 188.43770900275558, "ttft_ms": 11.900209006853402, "decode_ms": 11.83570793364197}, "total_wall_ms": 7135.848417063244, "ttft_ms": 11.900209006853402, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 64, "facts_retained": 14, "retention_rate": 0.21875, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.2222222222222222, "amount": 0.15384615384615385, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 870}, "ledger_by_hop": {"1": {"facts_total": 64, "facts_retained": 14, "retention_rate": 0.21875, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.2222222222222222, "amount": 0.15384615384615385, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 870}}}, "reference_key_id": "keys/synthetic/syn_03018_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "b93ce255451151cce89ccd51cc89c5b4382f883b441b3818bc249f2f79c54d9e", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "9a56070e7f6af93f7cf0552a", "attempt_id": "655912e7f7fe", "run_group_id": "E4_core:A5|repairable|qwen3b:h1", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:13:25.880771Z", "task_family": "synthetic", "document_id": "syn_03018_2000", "document_hash": "d284eb07e27da742b9c2b67abccfdf72e4dd67c3259f44f3b41e34dd3e6fc1ff", "question_id": "q07_negation", "split": "test", "cluster_id": "syn_03018_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 1, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1982, "query_tokens": 484, "message_tokens": 401, "message_utf8_bytes": 940, "cache_tensor_bytes": null, "serialized_payload_bytes": 940, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 6923.268833081238, "hops": {"1": 6923.268833081238}, "receiver_prefill_ms": 188.38487495668232, "ttft_ms": 11.87725004274398, "decode_ms": 11.8559580296278}, "total_wall_ms": 7135.7745410641655, "ttft_ms": 11.87725004274398, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 64, "facts_retained": 14, "retention_rate": 0.21875, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.2222222222222222, "amount": 0.15384615384615385, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 870}, "ledger_by_hop": {"1": {"facts_total": 64, "facts_retained": 14, "retention_rate": 0.21875, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.2222222222222222, "amount": 0.15384615384615385, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 870}}}, "reference_key_id": "keys/synthetic/syn_03018_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "b93ce255451151cce89ccd51cc89c5b4382f883b441b3818bc249f2f79c54d9e", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "ce83fdfed5db2cc5a54a6906", "attempt_id": "479702e14722", "run_group_id": "E4_core:A5|repairable|qwen3b:h1", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:13:26.107873Z", "task_family": "synthetic", "document_id": "syn_03018_2000", "document_hash": "d284eb07e27da742b9c2b67abccfdf72e4dd67c3259f44f3b41e34dd3e6fc1ff", "question_id": "q08_revision", "split": "test", "cluster_id": "syn_03018_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 1, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1982, "query_tokens": 502, "message_tokens": 401, "message_utf8_bytes": 940, "cache_tensor_bytes": null, "serialized_payload_bytes": 940, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 6923.268833081238, "hops": {"1": 6923.268833081238}, "receiver_prefill_ms": 188.79174999892712, "ttft_ms": 11.91833394113928, "decode_ms": 71.18891703430563}, "total_wall_ms": 7195.565708098002, "ttft_ms": 11.91833394113928, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 6}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "272184", "evidence_artifact": {"ledger": {"facts_total": 64, "facts_retained": 14, "retention_rate": 0.21875, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.2222222222222222, "amount": 0.15384615384615385, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 870}, "ledger_by_hop": {"1": {"facts_total": 64, "facts_retained": 14, "retention_rate": 0.21875, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.2222222222222222, "amount": 0.15384615384615385, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 870}}}, "reference_key_id": "keys/synthetic/syn_03018_2000.json", "task_scores": {"correct": 0.0, "abstained": 0.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "b93ce255451151cce89ccd51cc89c5b4382f883b441b3818bc249f2f79c54d9e", "output_hash": "3aadaddf09c30b8bfe5dc7e2aa6984fcf0df4d216145cf925760dd5a946d6e05"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "39994851b96a5477b73a368a", "attempt_id": "ed3e7ca9f745", "run_group_id": "E4_core:A5|repairable|qwen3b:h1", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:13:26.394876Z", "task_family": "synthetic", "document_id": "syn_03018_2000", "document_hash": "d284eb07e27da742b9c2b67abccfdf72e4dd67c3259f44f3b41e34dd3e6fc1ff", "question_id": "q09_revision", "split": "test", "cluster_id": "syn_03018_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 1, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1982, "query_tokens": 495, "message_tokens": 401, "message_utf8_bytes": 940, "cache_tensor_bytes": null, "serialized_payload_bytes": 940, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 6923.268833081238, "hops": {"1": 6923.268833081238}, "receiver_prefill_ms": 188.93225002102554, "ttft_ms": 11.858125100843608, "decode_ms": 71.03370898403227}, "total_wall_ms": 7195.491291116923, "ttft_ms": 11.858125100843608, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 6}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "272184", "evidence_artifact": {"ledger": {"facts_total": 64, "facts_retained": 14, "retention_rate": 0.21875, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.2222222222222222, "amount": 0.15384615384615385, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 870}, "ledger_by_hop": {"1": {"facts_total": 64, "facts_retained": 14, "retention_rate": 0.21875, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.2222222222222222, "amount": 0.15384615384615385, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 870}}}, "reference_key_id": "keys/synthetic/syn_03018_2000.json", "task_scores": {"correct": 0.0, "abstained": 0.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "b93ce255451151cce89ccd51cc89c5b4382f883b441b3818bc249f2f79c54d9e", "output_hash": "3aadaddf09c30b8bfe5dc7e2aa6984fcf0df4d216145cf925760dd5a946d6e05"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "84e74658d60cc6b2b8a64ad2", "attempt_id": "a04494876fe3", "run_group_id": "E4_core:A5|repairable|qwen3b:h1", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:13:26.682359Z", "task_family": "synthetic", "document_id": "syn_03018_2000", "document_hash": "d284eb07e27da742b9c2b67abccfdf72e4dd67c3259f44f3b41e34dd3e6fc1ff", "question_id": "q10_temporal", "split": "test", "cluster_id": "syn_03018_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 1, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1982, "query_tokens": 503, "message_tokens": 401, "message_utf8_bytes": 940, "cache_tensor_bytes": null, "serialized_payload_bytes": 940, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 6923.268833081238, "hops": {"1": 6923.268833081238}, "receiver_prefill_ms": 188.6872500181198, "ttft_ms": 11.893333052285016, "decode_ms": 11.827792041003704}, "total_wall_ms": 7136.078874114901, "ttft_ms": 11.893333052285016, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 64, "facts_retained": 14, "retention_rate": 0.21875, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.2222222222222222, "amount": 0.15384615384615385, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 870}, "ledger_by_hop": {"1": {"facts_total": 64, "facts_retained": 14, "retention_rate": 0.21875, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.2222222222222222, "amount": 0.15384615384615385, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 870}}}, "reference_key_id": "keys/synthetic/syn_03018_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "b93ce255451151cce89ccd51cc89c5b4382f883b441b3818bc249f2f79c54d9e", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "28d0213a207cae64e63af1a9", "attempt_id": "66c1378ccc5f", "run_group_id": "E4_core:A5|repairable|qwen3b:h1", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:13:26.910023Z", "task_family": "synthetic", "document_id": "syn_03018_2000", "document_hash": "d284eb07e27da742b9c2b67abccfdf72e4dd67c3259f44f3b41e34dd3e6fc1ff", "question_id": "q11_aggregation", "split": "test", "cluster_id": "syn_03018_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 1, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1982, "query_tokens": 495, "message_tokens": 401, "message_utf8_bytes": 940, "cache_tensor_bytes": null, "serialized_payload_bytes": 940, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 6923.268833081238, "hops": {"1": 6923.268833081238}, "receiver_prefill_ms": 188.12725006137043, "ttft_ms": 11.95408299099654, "decode_ms": 82.83645904157311}, "total_wall_ms": 7206.567083136179, "ttft_ms": 11.95408299099654, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 7}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "3189244", "evidence_artifact": {"ledger": {"facts_total": 64, "facts_retained": 14, "retention_rate": 0.21875, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.2222222222222222, "amount": 0.15384615384615385, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 870}, "ledger_by_hop": {"1": {"facts_total": 64, "facts_retained": 14, "retention_rate": 0.21875, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.2222222222222222, "amount": 0.15384615384615385, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 870}}}, "reference_key_id": "keys/synthetic/syn_03018_2000.json", "task_scores": {"correct": 0.0, "abstained": 0.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "b93ce255451151cce89ccd51cc89c5b4382f883b441b3818bc249f2f79c54d9e", "output_hash": "953508a9732586a80de7650cf183d3ca74cd590103fa92d25c9b3d80da907f61"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "4b2af40c86d6a94ec0a009cd", "attempt_id": "e5caa2d8f810", "run_group_id": "E4_core:A5|repairable|qwen3b:h1", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:13:27.207815Z", "task_family": "synthetic", "document_id": "syn_03018_2000", "document_hash": "d284eb07e27da742b9c2b67abccfdf72e4dd67c3259f44f3b41e34dd3e6fc1ff", "question_id": "q12_unit_conversion", "split": "test", "cluster_id": "syn_03018_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 1, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1982, "query_tokens": 492, "message_tokens": 401, "message_utf8_bytes": 940, "cache_tensor_bytes": null, "serialized_payload_bytes": 940, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 6923.268833081238, "hops": {"1": 6923.268833081238}, "receiver_prefill_ms": 188.14566696528345, "ttft_ms": 11.934875044971704, "decode_ms": 11.901417048648}, "total_wall_ms": 7135.6450411258265, "ttft_ms": 11.934875044971704, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 64, "facts_retained": 14, "retention_rate": 0.21875, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.2222222222222222, "amount": 0.15384615384615385, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 870}, "ledger_by_hop": {"1": {"facts_total": 64, "facts_retained": 14, "retention_rate": 0.21875, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.2222222222222222, "amount": 0.15384615384615385, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 870}}}, "reference_key_id": "keys/synthetic/syn_03018_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "b93ce255451151cce89ccd51cc89c5b4382f883b441b3818bc249f2f79c54d9e", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "af1bdc8c47224f4516bd69a1", "attempt_id": "513531ce2616", "run_group_id": "E4_core:A5|repairable|qwen3b:h1", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:13:27.435437Z", "task_family": "synthetic", "document_id": "syn_03018_2000", "document_hash": "d284eb07e27da742b9c2b67abccfdf72e4dd67c3259f44f3b41e34dd3e6fc1ff", "question_id": "q13_unanswerable", "split": "test", "cluster_id": "syn_03018_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 1, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1982, "query_tokens": 496, "message_tokens": 401, "message_utf8_bytes": 940, "cache_tensor_bytes": null, "serialized_payload_bytes": 940, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 6923.268833081238, "hops": {"1": 6923.268833081238}, "receiver_prefill_ms": 188.4048340143636, "ttft_ms": 11.803708970546722, "decode_ms": 11.92416693083942}, "total_wall_ms": 7135.800749179907, "ttft_ms": 11.803708970546722, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 64, "facts_retained": 14, "retention_rate": 0.21875, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.2222222222222222, "amount": 0.15384615384615385, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 870}, "ledger_by_hop": {"1": {"facts_total": 64, "facts_retained": 14, "retention_rate": 0.21875, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.2222222222222222, "amount": 0.15384615384615385, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 870}}}, "reference_key_id": "keys/synthetic/syn_03018_2000.json", "task_scores": {"correct": 1.0, "abstained": 1.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "b93ce255451151cce89ccd51cc89c5b4382f883b441b3818bc249f2f79c54d9e", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "1657659cc1b8119276c7095e", "attempt_id": "dac96acbca60", "run_group_id": "E4_core:A5|repairable|qwen3b:h1", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:13:27.662791Z", "task_family": "synthetic", "document_id": "syn_03018_2000", "document_hash": "d284eb07e27da742b9c2b67abccfdf72e4dd67c3259f44f3b41e34dd3e6fc1ff", "question_id": "q14_unanswerable", "split": "test", "cluster_id": "syn_03018_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 1, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1982, "query_tokens": 497, "message_tokens": 401, "message_utf8_bytes": 940, "cache_tensor_bytes": null, "serialized_payload_bytes": 940, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 6923.268833081238, "hops": {"1": 6923.268833081238}, "receiver_prefill_ms": 188.83720808662474, "ttft_ms": 11.882207938469946, "decode_ms": 11.914792004972696}, "total_wall_ms": 7136.296291020699, "ttft_ms": 11.882207938469946, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 64, "facts_retained": 14, "retention_rate": 0.21875, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.2222222222222222, "amount": 0.15384615384615385, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 870}, "ledger_by_hop": {"1": {"facts_total": 64, "facts_retained": 14, "retention_rate": 0.21875, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.2222222222222222, "amount": 0.15384615384615385, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 870}}}, "reference_key_id": "keys/synthetic/syn_03018_2000.json", "task_scores": {"correct": 1.0, "abstained": 1.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "b93ce255451151cce89ccd51cc89c5b4382f883b441b3818bc249f2f79c54d9e", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "3b5d89dffbf3d63d84954b97", "attempt_id": "d84ef6f7b369", "run_group_id": "E4_core:A5|repairable|qwen3b:h2", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:13:31.547700Z", "task_family": "synthetic", "document_id": "syn_03018_2000", "document_hash": "d284eb07e27da742b9c2b67abccfdf72e4dd67c3259f44f3b41e34dd3e6fc1ff", "question_id": "q00_single_fact", "split": "test", "cluster_id": "syn_03018_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 2, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1982, "query_tokens": 355, "message_tokens": 273, "message_utf8_bytes": 657, "cache_tensor_bytes": null, "serialized_payload_bytes": 657, "side_information_bytes": 0, "fetched_bytes": 1091, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 10578.63804209046, "hops": {"1": 6923.268833081238, "2": 3655.3692090092227}, "receiver_prefill_ms": 141.80116599891335, "ttft_ms": 11.899332981556654, "decode_ms": 11.810790980234742}, "total_wall_ms": 10744.607667205855, "ttft_ms": 11.899332981556654, "input_usage": {"lookups": 2}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 64, "facts_retained": 11, "retention_rate": 0.171875, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1111111111111111, "amount": 0.07692307692307693, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 609}, "ledger_by_hop": {"1": {"facts_total": 64, "facts_retained": 14, "retention_rate": 0.21875, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.2222222222222222, "amount": 0.15384615384615385, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 870}, "2": {"facts_total": 64, "facts_retained": 11, "retention_rate": 0.171875, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1111111111111111, "amount": 0.07692307692307693, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 609}}}, "reference_key_id": "keys/synthetic/syn_03018_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "830183240b7f45e537ebfeb7fe898082030a951407aec8fa2f3cd9cc13fe9d49", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "455037cbddd2491669f24520", "attempt_id": "ae86bd07d181", "run_group_id": "E4_core:A5|repairable|qwen3b:h2", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:13:31.729229Z", "task_family": "synthetic", "document_id": "syn_03018_2000", "document_hash": "d284eb07e27da742b9c2b67abccfdf72e4dd67c3259f44f3b41e34dd3e6fc1ff", "question_id": "q01_single_fact", "split": "test", "cluster_id": "syn_03018_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 2, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1982, "query_tokens": 361, "message_tokens": 273, "message_utf8_bytes": 657, "cache_tensor_bytes": null, "serialized_payload_bytes": 657, "side_information_bytes": 0, "fetched_bytes": 1091, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 10578.63804209046, "hops": {"1": 6923.268833081238, "2": 3655.3692090092227}, "receiver_prefill_ms": 142.0962909469381, "ttft_ms": 12.09533296059817, "decode_ms": 12.054541031830013}, "total_wall_ms": 10745.368334115483, "ttft_ms": 12.09533296059817, "input_usage": {"lookups": 2}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 64, "facts_retained": 11, "retention_rate": 0.171875, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1111111111111111, "amount": 0.07692307692307693, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 609}, "ledger_by_hop": {"1": {"facts_total": 64, "facts_retained": 14, "retention_rate": 0.21875, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.2222222222222222, "amount": 0.15384615384615385, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 870}, "2": {"facts_total": 64, "facts_retained": 11, "retention_rate": 0.171875, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1111111111111111, "amount": 0.07692307692307693, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 609}}}, "reference_key_id": "keys/synthetic/syn_03018_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "830183240b7f45e537ebfeb7fe898082030a951407aec8fa2f3cd9cc13fe9d49", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "5dd5c882512fd06b3a3f9251", "attempt_id": "7c9d9ae5c941", "run_group_id": "E4_core:A5|repairable|qwen3b:h2", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:13:31.911195Z", "task_family": "synthetic", "document_id": "syn_03018_2000", "document_hash": "d284eb07e27da742b9c2b67abccfdf72e4dd67c3259f44f3b41e34dd3e6fc1ff", "question_id": "q02_exact_copy", "split": "test", "cluster_id": "syn_03018_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 2, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1982, "query_tokens": 360, "message_tokens": 273, "message_utf8_bytes": 657, "cache_tensor_bytes": null, "serialized_payload_bytes": 657, "side_information_bytes": 0, "fetched_bytes": 1091, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 10578.63804209046, "hops": {"1": 6923.268833081238, "2": 3655.3692090092227}, "receiver_prefill_ms": 141.89958304632455, "ttft_ms": 12.053541955538094, "decode_ms": 12.112082913517952}, "total_wall_ms": 10745.227750041522, "ttft_ms": 12.053541955538094, "input_usage": {"lookups": 2}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 64, "facts_retained": 11, "retention_rate": 0.171875, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1111111111111111, "amount": 0.07692307692307693, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 609}, "ledger_by_hop": {"1": {"facts_total": 64, "facts_retained": 14, "retention_rate": 0.21875, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.2222222222222222, "amount": 0.15384615384615385, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 870}, "2": {"facts_total": 64, "facts_retained": 11, "retention_rate": 0.171875, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1111111111111111, "amount": 0.07692307692307693, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 609}}}, "reference_key_id": "keys/synthetic/syn_03018_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "830183240b7f45e537ebfeb7fe898082030a951407aec8fa2f3cd9cc13fe9d49", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "595d1be6654add36512c0953", "attempt_id": "3639ca9daf29", "run_group_id": "E4_core:A5|repairable|qwen3b:h2", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:13:32.094480Z", "task_family": "synthetic", "document_id": "syn_03018_2000", "document_hash": "d284eb07e27da742b9c2b67abccfdf72e4dd67c3259f44f3b41e34dd3e6fc1ff", "question_id": "q03_exact_copy", "split": "test", "cluster_id": "syn_03018_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 2, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1982, "query_tokens": 358, "message_tokens": 273, "message_utf8_bytes": 657, "cache_tensor_bytes": null, "serialized_payload_bytes": 657, "side_information_bytes": 0, "fetched_bytes": 1091, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 10578.63804209046, "hops": {"1": 6923.268833081238, "2": 3655.3692090092227}, "receiver_prefill_ms": 142.24520896095783, "ttft_ms": 11.983374948613346, "decode_ms": 11.771750054322183}, "total_wall_ms": 10745.100417057984, "ttft_ms": 11.983374948613346, "input_usage": {"lookups": 2}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 64, "facts_retained": 11, "retention_rate": 0.171875, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1111111111111111, "amount": 0.07692307692307693, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 609}, "ledger_by_hop": {"1": {"facts_total": 64, "facts_retained": 14, "retention_rate": 0.21875, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.2222222222222222, "amount": 0.15384615384615385, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 870}, "2": {"facts_total": 64, "facts_retained": 11, "retention_rate": 0.171875, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1111111111111111, "amount": 0.07692307692307693, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 609}}}, "reference_key_id": "keys/synthetic/syn_03018_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "830183240b7f45e537ebfeb7fe898082030a951407aec8fa2f3cd9cc13fe9d49", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "02e5991d840bb28d7f3a1c45", "attempt_id": "08c2f3544cde", "run_group_id": "E4_core:A5|repairable|qwen3b:h2", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:13:32.276826Z", "task_family": "synthetic", "document_id": "syn_03018_2000", "document_hash": "d284eb07e27da742b9c2b67abccfdf72e4dd67c3259f44f3b41e34dd3e6fc1ff", "question_id": "q04_multi_hop", "split": "test", "cluster_id": "syn_03018_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 2, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1982, "query_tokens": 364, "message_tokens": 273, "message_utf8_bytes": 657, "cache_tensor_bytes": null, "serialized_payload_bytes": 657, "side_information_bytes": 0, "fetched_bytes": 1091, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 10578.63804209046, "hops": {"1": 6923.268833081238, "2": 3655.3692090092227}, "receiver_prefill_ms": 142.5310840131715, "ttft_ms": 11.799750034697354, "decode_ms": 11.865000007674098}, "total_wall_ms": 10745.194959104992, "ttft_ms": 11.799750034697354, "input_usage": {"lookups": 2}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 64, "facts_retained": 11, "retention_rate": 0.171875, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1111111111111111, "amount": 0.07692307692307693, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 609}, "ledger_by_hop": {"1": {"facts_total": 64, "facts_retained": 14, "retention_rate": 0.21875, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.2222222222222222, "amount": 0.15384615384615385, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 870}, "2": {"facts_total": 64, "facts_retained": 11, "retention_rate": 0.171875, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1111111111111111, "amount": 0.07692307692307693, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 609}}}, "reference_key_id": "keys/synthetic/syn_03018_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "830183240b7f45e537ebfeb7fe898082030a951407aec8fa2f3cd9cc13fe9d49", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "768d56194775d431f7ebc054", "attempt_id": "8f566207dfd5", "run_group_id": "E4_core:A5|repairable|qwen3b:h2", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:13:32.458682Z", "task_family": "synthetic", "document_id": "syn_03018_2000", "document_hash": "d284eb07e27da742b9c2b67abccfdf72e4dd67c3259f44f3b41e34dd3e6fc1ff", "question_id": "q05_multi_hop", "split": "test", "cluster_id": "syn_03018_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 2, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1982, "query_tokens": 364, "message_tokens": 273, "message_utf8_bytes": 657, "cache_tensor_bytes": null, "serialized_payload_bytes": 657, "side_information_bytes": 0, "fetched_bytes": 1091, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 10578.63804209046, "hops": {"1": 6923.268833081238, "2": 3655.3692090092227}, "receiver_prefill_ms": 142.67804101109505, "ttft_ms": 11.89591595903039, "decode_ms": 11.784582980908453}, "total_wall_ms": 10745.345167117193, "ttft_ms": 11.89591595903039, "input_usage": {"lookups": 2}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 64, "facts_retained": 11, "retention_rate": 0.171875, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1111111111111111, "amount": 0.07692307692307693, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 609}, "ledger_by_hop": {"1": {"facts_total": 64, "facts_retained": 14, "retention_rate": 0.21875, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.2222222222222222, "amount": 0.15384615384615385, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 870}, "2": {"facts_total": 64, "facts_retained": 11, "retention_rate": 0.171875, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1111111111111111, "amount": 0.07692307692307693, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 609}}}, "reference_key_id": "keys/synthetic/syn_03018_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "830183240b7f45e537ebfeb7fe898082030a951407aec8fa2f3cd9cc13fe9d49", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "a6c60c123157577be4578671", "attempt_id": "0cb2559f5c37", "run_group_id": "E4_core:A5|repairable|qwen3b:h2", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:13:32.640579Z", "task_family": "synthetic", "document_id": "syn_03018_2000", "document_hash": "d284eb07e27da742b9c2b67abccfdf72e4dd67c3259f44f3b41e34dd3e6fc1ff", "question_id": "q06_negation", "split": "test", "cluster_id": "syn_03018_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 2, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1982, "query_tokens": 356, "message_tokens": 273, "message_utf8_bytes": 657, "cache_tensor_bytes": null, "serialized_payload_bytes": 657, "side_information_bytes": 0, "fetched_bytes": 1091, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 10578.63804209046, "hops": {"1": 6923.268833081238, "2": 3655.3692090092227}, "receiver_prefill_ms": 142.09149999078363, "ttft_ms": 11.737708002328873, "decode_ms": 11.857416946440935}, "total_wall_ms": 10744.648792082444, "ttft_ms": 11.737708002328873, "input_usage": {"lookups": 2}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 64, "facts_retained": 11, "retention_rate": 0.171875, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1111111111111111, "amount": 0.07692307692307693, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 609}, "ledger_by_hop": {"1": {"facts_total": 64, "facts_retained": 14, "retention_rate": 0.21875, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.2222222222222222, "amount": 0.15384615384615385, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 870}, "2": {"facts_total": 64, "facts_retained": 11, "retention_rate": 0.171875, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1111111111111111, "amount": 0.07692307692307693, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 609}}}, "reference_key_id": "keys/synthetic/syn_03018_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "830183240b7f45e537ebfeb7fe898082030a951407aec8fa2f3cd9cc13fe9d49", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "7695bc63e6e029296c5d8d25", "attempt_id": "709aab5da649", "run_group_id": "E4_core:A5|repairable|qwen3b:h2", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:13:32.821660Z", "task_family": "synthetic", "document_id": "syn_03018_2000", "document_hash": "d284eb07e27da742b9c2b67abccfdf72e4dd67c3259f44f3b41e34dd3e6fc1ff", "question_id": "q07_negation", "split": "test", "cluster_id": "syn_03018_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 2, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1982, "query_tokens": 355, "message_tokens": 273, "message_utf8_bytes": 657, "cache_tensor_bytes": null, "serialized_payload_bytes": 657, "side_information_bytes": 0, "fetched_bytes": 1091, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 10578.63804209046, "hops": {"1": 6923.268833081238, "2": 3655.3692090092227}, "receiver_prefill_ms": 142.18770805746317, "ttft_ms": 11.834583012387156, "decode_ms": 12.019374989904463}, "total_wall_ms": 10744.997833156958, "ttft_ms": 11.834583012387156, "input_usage": {"lookups": 2}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 64, "facts_retained": 11, "retention_rate": 0.171875, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1111111111111111, "amount": 0.07692307692307693, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 609}, "ledger_by_hop": {"1": {"facts_total": 64, "facts_retained": 14, "retention_rate": 0.21875, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.2222222222222222, "amount": 0.15384615384615385, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 870}, "2": {"facts_total": 64, "facts_retained": 11, "retention_rate": 0.171875, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1111111111111111, "amount": 0.07692307692307693, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 609}}}, "reference_key_id": "keys/synthetic/syn_03018_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "830183240b7f45e537ebfeb7fe898082030a951407aec8fa2f3cd9cc13fe9d49", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "d4aef0c1f012a6f24242986e", "attempt_id": "a213801295b2", "run_group_id": "E4_core:A5|repairable|qwen3b:h2", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:13:33.002389Z", "task_family": "synthetic", "document_id": "syn_03018_2000", "document_hash": "d284eb07e27da742b9c2b67abccfdf72e4dd67c3259f44f3b41e34dd3e6fc1ff", "question_id": "q08_revision", "split": "test", "cluster_id": "syn_03018_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 2, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1982, "query_tokens": 373, "message_tokens": 273, "message_utf8_bytes": 657, "cache_tensor_bytes": null, "serialized_payload_bytes": 657, "side_information_bytes": 0, "fetched_bytes": 1091, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 10578.63804209046, "hops": {"1": 6923.268833081238, "2": 3655.3692090092227}, "receiver_prefill_ms": 143.14545807428658, "ttft_ms": 11.793458019383252, "decode_ms": 11.791665921919048}, "total_wall_ms": 10745.713501120917, "ttft_ms": 11.793458019383252, "input_usage": {"lookups": 2}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 64, "facts_retained": 11, "retention_rate": 0.171875, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1111111111111111, "amount": 0.07692307692307693, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 609}, "ledger_by_hop": {"1": {"facts_total": 64, "facts_retained": 14, "retention_rate": 0.21875, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.2222222222222222, "amount": 0.15384615384615385, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 870}, "2": {"facts_total": 64, "facts_retained": 11, "retention_rate": 0.171875, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1111111111111111, "amount": 0.07692307692307693, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 609}}}, "reference_key_id": "keys/synthetic/syn_03018_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "830183240b7f45e537ebfeb7fe898082030a951407aec8fa2f3cd9cc13fe9d49", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "11de929ec8635dc1276a58c8", "attempt_id": "c0f05ae2a67e", "run_group_id": "E4_core:A5|repairable|qwen3b:h2", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:13:33.184268Z", "task_family": "synthetic", "document_id": "syn_03018_2000", "document_hash": "d284eb07e27da742b9c2b67abccfdf72e4dd67c3259f44f3b41e34dd3e6fc1ff", "question_id": "q09_revision", "split": "test", "cluster_id": "syn_03018_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 2, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1982, "query_tokens": 366, "message_tokens": 273, "message_utf8_bytes": 657, "cache_tensor_bytes": null, "serialized_payload_bytes": 657, "side_information_bytes": 0, "fetched_bytes": 1091, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 10578.63804209046, "hops": {"1": 6923.268833081238, "2": 3655.3692090092227}, "receiver_prefill_ms": 142.65525003429502, "ttft_ms": 11.753124999813735, "decode_ms": 11.877333978191018}, "total_wall_ms": 10745.23695919197, "ttft_ms": 11.753124999813735, "input_usage": {"lookups": 2}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 64, "facts_retained": 11, "retention_rate": 0.171875, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1111111111111111, "amount": 0.07692307692307693, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 609}, "ledger_by_hop": {"1": {"facts_total": 64, "facts_retained": 14, "retention_rate": 0.21875, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.2222222222222222, "amount": 0.15384615384615385, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 870}, "2": {"facts_total": 64, "facts_retained": 11, "retention_rate": 0.171875, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1111111111111111, "amount": 0.07692307692307693, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 609}}}, "reference_key_id": "keys/synthetic/syn_03018_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "830183240b7f45e537ebfeb7fe898082030a951407aec8fa2f3cd9cc13fe9d49", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "ee920fc794cbc667a3527866", "attempt_id": "d1285affbda1", "run_group_id": "E4_core:A5|repairable|qwen3b:h2", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:13:33.366110Z", "task_family": "synthetic", "document_id": "syn_03018_2000", "document_hash": "d284eb07e27da742b9c2b67abccfdf72e4dd67c3259f44f3b41e34dd3e6fc1ff", "question_id": "q10_temporal", "split": "test", "cluster_id": "syn_03018_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 2, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1982, "query_tokens": 374, "message_tokens": 273, "message_utf8_bytes": 657, "cache_tensor_bytes": null, "serialized_payload_bytes": 657, "side_information_bytes": 0, "fetched_bytes": 1091, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 10578.63804209046, "hops": {"1": 6923.268833081238, "2": 3655.3692090092227}, "receiver_prefill_ms": 143.20079202298075, "ttft_ms": 11.778249987401068, "decode_ms": 11.817499995231628}, "total_wall_ms": 10745.763209182769, "ttft_ms": 11.778249987401068, "input_usage": {"lookups": 2}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 64, "facts_retained": 11, "retention_rate": 0.171875, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1111111111111111, "amount": 0.07692307692307693, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 609}, "ledger_by_hop": {"1": {"facts_total": 64, "facts_retained": 14, "retention_rate": 0.21875, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.2222222222222222, "amount": 0.15384615384615385, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 870}, "2": {"facts_total": 64, "facts_retained": 11, "retention_rate": 0.171875, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1111111111111111, "amount": 0.07692307692307693, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 609}}}, "reference_key_id": "keys/synthetic/syn_03018_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "830183240b7f45e537ebfeb7fe898082030a951407aec8fa2f3cd9cc13fe9d49", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "6b6b059ddeb9d5452fe27497", "attempt_id": "c6fb3cbc09ee", "run_group_id": "E4_core:A5|repairable|qwen3b:h2", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:13:33.547832Z", "task_family": "synthetic", "document_id": "syn_03018_2000", "document_hash": "d284eb07e27da742b9c2b67abccfdf72e4dd67c3259f44f3b41e34dd3e6fc1ff", "question_id": "q11_aggregation", "split": "test", "cluster_id": "syn_03018_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 2, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1982, "query_tokens": 366, "message_tokens": 273, "message_utf8_bytes": 657, "cache_tensor_bytes": null, "serialized_payload_bytes": 657, "side_information_bytes": 0, "fetched_bytes": 1091, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 10578.63804209046, "hops": {"1": 6923.268833081238, "2": 3655.3692090092227}, "receiver_prefill_ms": 142.56895799189806, "ttft_ms": 11.834333068691194, "decode_ms": 81.8743749987334}, "total_wall_ms": 10815.229292027652, "ttft_ms": 11.834333068691194, "input_usage": {"lookups": 2}, "output_usage": {"output_tokens": 7}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "2917061", "evidence_artifact": {"ledger": {"facts_total": 64, "facts_retained": 11, "retention_rate": 0.171875, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1111111111111111, "amount": 0.07692307692307693, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 609}, "ledger_by_hop": {"1": {"facts_total": 64, "facts_retained": 14, "retention_rate": 0.21875, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.2222222222222222, "amount": 0.15384615384615385, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 870}, "2": {"facts_total": 64, "facts_retained": 11, "retention_rate": 0.171875, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1111111111111111, "amount": 0.07692307692307693, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 609}}}, "reference_key_id": "keys/synthetic/syn_03018_2000.json", "task_scores": {"correct": 0.0, "abstained": 0.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "830183240b7f45e537ebfeb7fe898082030a951407aec8fa2f3cd9cc13fe9d49", "output_hash": "129c25eb43beebf0b464953aaabebfe914d11fa686f1dce105e62022958e1719"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "a93ba59a29afe1343c8c060c", "attempt_id": "62b33400aa8e", "run_group_id": "E4_core:A5|repairable|qwen3b:h2", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:13:33.799173Z", "task_family": "synthetic", "document_id": "syn_03018_2000", "document_hash": "d284eb07e27da742b9c2b67abccfdf72e4dd67c3259f44f3b41e34dd3e6fc1ff", "question_id": "q12_unit_conversion", "split": "test", "cluster_id": "syn_03018_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 2, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1982, "query_tokens": 363, "message_tokens": 273, "message_utf8_bytes": 657, "cache_tensor_bytes": null, "serialized_payload_bytes": 657, "side_information_bytes": 0, "fetched_bytes": 1091, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 10578.63804209046, "hops": {"1": 6923.268833081238, "2": 3655.3692090092227}, "receiver_prefill_ms": 142.50766695477068, "ttft_ms": 11.758708045817912, "decode_ms": 47.06854198593646}, "total_wall_ms": 10780.294001102448, "ttft_ms": 11.758708045817912, "input_usage": {"lookups": 2}, "output_usage": {"output_tokens": 4}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "6400", "evidence_artifact": {"ledger": {"facts_total": 64, "facts_retained": 11, "retention_rate": 0.171875, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1111111111111111, "amount": 0.07692307692307693, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 609}, "ledger_by_hop": {"1": {"facts_total": 64, "facts_retained": 14, "retention_rate": 0.21875, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.2222222222222222, "amount": 0.15384615384615385, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 870}, "2": {"facts_total": 64, "facts_retained": 11, "retention_rate": 0.171875, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1111111111111111, "amount": 0.07692307692307693, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 609}}}, "reference_key_id": "keys/synthetic/syn_03018_2000.json", "task_scores": {"correct": 0.0, "abstained": 0.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "830183240b7f45e537ebfeb7fe898082030a951407aec8fa2f3cd9cc13fe9d49", "output_hash": "a7dccef9ce1ae31c49a4cfc484f8411dd52b11c914c1fe3d781633642d1e1327"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "1f455e8b847e6bb997e1a475", "attempt_id": "6ba52abc15ad", "run_group_id": "E4_core:A5|repairable|qwen3b:h2", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:13:34.015456Z", "task_family": "synthetic", "document_id": "syn_03018_2000", "document_hash": "d284eb07e27da742b9c2b67abccfdf72e4dd67c3259f44f3b41e34dd3e6fc1ff", "question_id": "q13_unanswerable", "split": "test", "cluster_id": "syn_03018_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 2, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1982, "query_tokens": 367, "message_tokens": 273, "message_utf8_bytes": 657, "cache_tensor_bytes": null, "serialized_payload_bytes": 657, "side_information_bytes": 0, "fetched_bytes": 1091, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 10578.63804209046, "hops": {"1": 6923.268833081238, "2": 3655.3692090092227}, "receiver_prefill_ms": 142.52237498294562, "ttft_ms": 11.816791957244277, "decode_ms": 11.836874997243285}, "total_wall_ms": 10745.135501027107, "ttft_ms": 11.816791957244277, "input_usage": {"lookups": 2}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 64, "facts_retained": 11, "retention_rate": 0.171875, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1111111111111111, "amount": 0.07692307692307693, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 609}, "ledger_by_hop": {"1": {"facts_total": 64, "facts_retained": 14, "retention_rate": 0.21875, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.2222222222222222, "amount": 0.15384615384615385, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 870}, "2": {"facts_total": 64, "facts_retained": 11, "retention_rate": 0.171875, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1111111111111111, "amount": 0.07692307692307693, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 609}}}, "reference_key_id": "keys/synthetic/syn_03018_2000.json", "task_scores": {"correct": 1.0, "abstained": 1.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "830183240b7f45e537ebfeb7fe898082030a951407aec8fa2f3cd9cc13fe9d49", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "360ad30043ea270a40342f9e", "attempt_id": "f3fbc2101135", "run_group_id": "E4_core:A5|repairable|qwen3b:h2", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:13:34.196733Z", "task_family": "synthetic", "document_id": "syn_03018_2000", "document_hash": "d284eb07e27da742b9c2b67abccfdf72e4dd67c3259f44f3b41e34dd3e6fc1ff", "question_id": "q14_unanswerable", "split": "test", "cluster_id": "syn_03018_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 2, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1982, "query_tokens": 368, "message_tokens": 273, "message_utf8_bytes": 657, "cache_tensor_bytes": null, "serialized_payload_bytes": 657, "side_information_bytes": 0, "fetched_bytes": 1091, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 10578.63804209046, "hops": {"1": 6923.268833081238, "2": 3655.3692090092227}, "receiver_prefill_ms": 142.76483294088393, "ttft_ms": 11.776749975979328, "decode_ms": 11.84791699051857}, "total_wall_ms": 10745.344167109579, "ttft_ms": 11.776749975979328, "input_usage": {"lookups": 2}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 64, "facts_retained": 11, "retention_rate": 0.171875, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1111111111111111, "amount": 0.07692307692307693, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 609}, "ledger_by_hop": {"1": {"facts_total": 64, "facts_retained": 14, "retention_rate": 0.21875, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.2222222222222222, "amount": 0.15384615384615385, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 870}, "2": {"facts_total": 64, "facts_retained": 11, "retention_rate": 0.171875, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1111111111111111, "amount": 0.07692307692307693, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 609}}}, "reference_key_id": "keys/synthetic/syn_03018_2000.json", "task_scores": {"correct": 1.0, "abstained": 1.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "830183240b7f45e537ebfeb7fe898082030a951407aec8fa2f3cd9cc13fe9d49", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "12244dd98fb17e8115a91ced", "attempt_id": "40cbebda1478", "run_group_id": "E4_core:A5|repairable|qwen3b:h4", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:13:41.441712Z", "task_family": "synthetic", "document_id": "syn_03018_2000", "document_hash": "d284eb07e27da742b9c2b67abccfdf72e4dd67c3259f44f3b41e34dd3e6fc1ff", "question_id": "q00_single_fact", "split": "test", "cluster_id": "syn_03018_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 4, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1982, "query_tokens": 355, "message_tokens": 273, "message_utf8_bytes": 657, "cache_tensor_bytes": null, "serialized_payload_bytes": 657, "side_information_bytes": 0, "fetched_bytes": 2027, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 17640.172542189248, "hops": {"1": 6923.268833081238, "2": 3655.3692090092227, "3": 3530.2908330922946, "4": 3531.2436670064926}, "receiver_prefill_ms": 141.63216692395508, "ttft_ms": 11.743708048015833, "decode_ms": 11.817916063591838}, "total_wall_ms": 17805.734375142492, "ttft_ms": 11.743708048015833, "input_usage": {"lookups": 4}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 64, "facts_retained": 11, "retention_rate": 0.171875, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1111111111111111, "amount": 0.07692307692307693, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 609}, "ledger_by_hop": {"1": {"facts_total": 64, "facts_retained": 14, "retention_rate": 0.21875, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.2222222222222222, "amount": 0.15384615384615385, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 870}, "2": {"facts_total": 64, "facts_retained": 11, "retention_rate": 0.171875, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1111111111111111, "amount": 0.07692307692307693, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 609}, "3": {"facts_total": 64, "facts_retained": 11, "retention_rate": 0.171875, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1111111111111111, "amount": 0.07692307692307693, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 609}, "4": {"facts_total": 64, "facts_retained": 11, "retention_rate": 0.171875, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1111111111111111, "amount": 0.07692307692307693, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 609}}}, "reference_key_id": "keys/synthetic/syn_03018_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "830183240b7f45e537ebfeb7fe898082030a951407aec8fa2f3cd9cc13fe9d49", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "67c963a4ebc0d3d02b9ad537", "attempt_id": "0274ff008e00", "run_group_id": "E4_core:A5|repairable|qwen3b:h4", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:13:41.622946Z", "task_family": "synthetic", "document_id": "syn_03018_2000", "document_hash": "d284eb07e27da742b9c2b67abccfdf72e4dd67c3259f44f3b41e34dd3e6fc1ff", "question_id": "q01_single_fact", "split": "test", "cluster_id": "syn_03018_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 4, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1982, "query_tokens": 361, "message_tokens": 273, "message_utf8_bytes": 657, "cache_tensor_bytes": null, "serialized_payload_bytes": 657, "side_information_bytes": 0, "fetched_bytes": 2027, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 17640.172542189248, "hops": {"1": 6923.268833081238, "2": 3655.3692090092227, "3": 3530.2908330922946, "4": 3531.2436670064926}, "receiver_prefill_ms": 142.42545806337148, "ttft_ms": 11.860167025588453, "decode_ms": 11.812624987214804}, "total_wall_ms": 17806.6325832624, "ttft_ms": 11.860167025588453, "input_usage": {"lookups": 4}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 64, "facts_retained": 11, "retention_rate": 0.171875, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1111111111111111, "amount": 0.07692307692307693, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 609}, "ledger_by_hop": {"1": {"facts_total": 64, "facts_retained": 14, "retention_rate": 0.21875, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.2222222222222222, "amount": 0.15384615384615385, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 870}, "2": {"facts_total": 64, "facts_retained": 11, "retention_rate": 0.171875, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1111111111111111, "amount": 0.07692307692307693, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 609}, "3": {"facts_total": 64, "facts_retained": 11, "retention_rate": 0.171875, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1111111111111111, "amount": 0.07692307692307693, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 609}, "4": {"facts_total": 64, "facts_retained": 11, "retention_rate": 0.171875, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1111111111111111, "amount": 0.07692307692307693, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 609}}}, "reference_key_id": "keys/synthetic/syn_03018_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "830183240b7f45e537ebfeb7fe898082030a951407aec8fa2f3cd9cc13fe9d49", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "1be897492f347d1e065fe41d", "attempt_id": "d9c4f1adb9e7", "run_group_id": "E4_core:A5|repairable|qwen3b:h4", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:13:41.804709Z", "task_family": "synthetic", "document_id": "syn_03018_2000", "document_hash": "d284eb07e27da742b9c2b67abccfdf72e4dd67c3259f44f3b41e34dd3e6fc1ff", "question_id": "q02_exact_copy", "split": "test", "cluster_id": "syn_03018_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 4, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1982, "query_tokens": 360, "message_tokens": 273, "message_utf8_bytes": 657, "cache_tensor_bytes": null, "serialized_payload_bytes": 657, "side_information_bytes": 0, "fetched_bytes": 2027, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 17640.172542189248, "hops": {"1": 6923.268833081238, "2": 3655.3692090092227, "3": 3530.2908330922946, "4": 3531.2436670064926}, "receiver_prefill_ms": 142.13770802598447, "ttft_ms": 11.781750014051795, "decode_ms": 11.950542102567852}, "total_wall_ms": 17806.345667224377, "ttft_ms": 11.781750014051795, "input_usage": {"lookups": 4}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 64, "facts_retained": 11, "retention_rate": 0.171875, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1111111111111111, "amount": 0.07692307692307693, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 609}, "ledger_by_hop": {"1": {"facts_total": 64, "facts_retained": 14, "retention_rate": 0.21875, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.2222222222222222, "amount": 0.15384615384615385, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 870}, "2": {"facts_total": 64, "facts_retained": 11, "retention_rate": 0.171875, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1111111111111111, "amount": 0.07692307692307693, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 609}, "3": {"facts_total": 64, "facts_retained": 11, "retention_rate": 0.171875, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1111111111111111, "amount": 0.07692307692307693, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 609}, "4": {"facts_total": 64, "facts_retained": 11, "retention_rate": 0.171875, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1111111111111111, "amount": 0.07692307692307693, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 609}}}, "reference_key_id": "keys/synthetic/syn_03018_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "830183240b7f45e537ebfeb7fe898082030a951407aec8fa2f3cd9cc13fe9d49", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "3c0f23ce8842fe574f2a94ea", "attempt_id": "89e45af40008", "run_group_id": "E4_core:A5|repairable|qwen3b:h4", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:13:41.985568Z", "task_family": "synthetic", "document_id": "syn_03018_2000", "document_hash": "d284eb07e27da742b9c2b67abccfdf72e4dd67c3259f44f3b41e34dd3e6fc1ff", "question_id": "q03_exact_copy", "split": "test", "cluster_id": "syn_03018_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 4, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1982, "query_tokens": 358, "message_tokens": 273, "message_utf8_bytes": 657, "cache_tensor_bytes": null, "serialized_payload_bytes": 657, "side_information_bytes": 0, "fetched_bytes": 2027, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 17640.172542189248, "hops": {"1": 6923.268833081238, "2": 3655.3692090092227, "3": 3530.2908330922946, "4": 3531.2436670064926}, "receiver_prefill_ms": 142.4112500390038, "ttft_ms": 11.774083017371595, "decode_ms": 12.14437501039356}, "total_wall_ms": 17806.81570910383, "ttft_ms": 11.774083017371595, "input_usage": {"lookups": 4}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 64, "facts_retained": 11, "retention_rate": 0.171875, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1111111111111111, "amount": 0.07692307692307693, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 609}, "ledger_by_hop": {"1": {"facts_total": 64, "facts_retained": 14, "retention_rate": 0.21875, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.2222222222222222, "amount": 0.15384615384615385, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 870}, "2": {"facts_total": 64, "facts_retained": 11, "retention_rate": 0.171875, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1111111111111111, "amount": 0.07692307692307693, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 609}, "3": {"facts_total": 64, "facts_retained": 11, "retention_rate": 0.171875, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1111111111111111, "amount": 0.07692307692307693, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 609}, "4": {"facts_total": 64, "facts_retained": 11, "retention_rate": 0.171875, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1111111111111111, "amount": 0.07692307692307693, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 609}}}, "reference_key_id": "keys/synthetic/syn_03018_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "830183240b7f45e537ebfeb7fe898082030a951407aec8fa2f3cd9cc13fe9d49", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "2f417324a226e07838038a91", "attempt_id": "ed01d53d8ebd", "run_group_id": "E4_core:A5|repairable|qwen3b:h4", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:13:42.167112Z", "task_family": "synthetic", "document_id": "syn_03018_2000", "document_hash": "d284eb07e27da742b9c2b67abccfdf72e4dd67c3259f44f3b41e34dd3e6fc1ff", "question_id": "q04_multi_hop", "split": "test", "cluster_id": "syn_03018_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 4, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1982, "query_tokens": 364, "message_tokens": 273, "message_utf8_bytes": 657, "cache_tensor_bytes": null, "serialized_payload_bytes": 657, "side_information_bytes": 0, "fetched_bytes": 2027, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 17640.172542189248, "hops": {"1": 6923.268833081238, "2": 3655.3692090092227, "3": 3530.2908330922946, "4": 3531.2436670064926}, "receiver_prefill_ms": 142.73629209492356, "ttft_ms": 11.805832968093455, "decode_ms": 11.79333298932761}, "total_wall_ms": 17806.833750219084, "ttft_ms": 11.805832968093455, "input_usage": {"lookups": 4}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 64, "facts_retained": 11, "retention_rate": 0.171875, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1111111111111111, "amount": 0.07692307692307693, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 609}, "ledger_by_hop": {"1": {"facts_total": 64, "facts_retained": 14, "retention_rate": 0.21875, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.2222222222222222, "amount": 0.15384615384615385, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 870}, "2": {"facts_total": 64, "facts_retained": 11, "retention_rate": 0.171875, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1111111111111111, "amount": 0.07692307692307693, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 609}, "3": {"facts_total": 64, "facts_retained": 11, "retention_rate": 0.171875, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1111111111111111, "amount": 0.07692307692307693, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 609}, "4": {"facts_total": 64, "facts_retained": 11, "retention_rate": 0.171875, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1111111111111111, "amount": 0.07692307692307693, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 609}}}, "reference_key_id": "keys/synthetic/syn_03018_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "830183240b7f45e537ebfeb7fe898082030a951407aec8fa2f3cd9cc13fe9d49", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "b128cd27683bd95b48a8715a", "attempt_id": "c4243f56fd9d", "run_group_id": "E4_core:A5|repairable|qwen3b:h4", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:13:42.348761Z", "task_family": "synthetic", "document_id": "syn_03018_2000", "document_hash": "d284eb07e27da742b9c2b67abccfdf72e4dd67c3259f44f3b41e34dd3e6fc1ff", "question_id": "q05_multi_hop", "split": "test", "cluster_id": "syn_03018_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 4, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1982, "query_tokens": 364, "message_tokens": 273, "message_utf8_bytes": 657, "cache_tensor_bytes": null, "serialized_payload_bytes": 657, "side_information_bytes": 0, "fetched_bytes": 2027, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 17640.172542189248, "hops": {"1": 6923.268833081238, "2": 3655.3692090092227, "3": 3530.2908330922946, "4": 3531.2436670064926}, "receiver_prefill_ms": 142.2889579553157, "ttft_ms": 11.794666992500424, "decode_ms": 11.881208047270775}, "total_wall_ms": 17806.44554225728, "ttft_ms": 11.794666992500424, "input_usage": {"lookups": 4}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 64, "facts_retained": 11, "retention_rate": 0.171875, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1111111111111111, "amount": 0.07692307692307693, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 609}, "ledger_by_hop": {"1": {"facts_total": 64, "facts_retained": 14, "retention_rate": 0.21875, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.2222222222222222, "amount": 0.15384615384615385, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 870}, "2": {"facts_total": 64, "facts_retained": 11, "retention_rate": 0.171875, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1111111111111111, "amount": 0.07692307692307693, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 609}, "3": {"facts_total": 64, "facts_retained": 11, "retention_rate": 0.171875, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1111111111111111, "amount": 0.07692307692307693, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 609}, "4": {"facts_total": 64, "facts_retained": 11, "retention_rate": 0.171875, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1111111111111111, "amount": 0.07692307692307693, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 609}}}, "reference_key_id": "keys/synthetic/syn_03018_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "830183240b7f45e537ebfeb7fe898082030a951407aec8fa2f3cd9cc13fe9d49", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "0aae4b7f2e562985076a69a4", "attempt_id": "d885a0cfc85f", "run_group_id": "E4_core:A5|repairable|qwen3b:h4", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:13:42.529992Z", "task_family": "synthetic", "document_id": "syn_03018_2000", "document_hash": "d284eb07e27da742b9c2b67abccfdf72e4dd67c3259f44f3b41e34dd3e6fc1ff", "question_id": "q06_negation", "split": "test", "cluster_id": "syn_03018_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 4, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1982, "query_tokens": 356, "message_tokens": 273, "message_utf8_bytes": 657, "cache_tensor_bytes": null, "serialized_payload_bytes": 657, "side_information_bytes": 0, "fetched_bytes": 2027, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 17640.172542189248, "hops": {"1": 6923.268833081238, "2": 3655.3692090092227, "3": 3530.2908330922946, "4": 3531.2436670064926}, "receiver_prefill_ms": 142.23437500186265, "ttft_ms": 11.839333921670914, "decode_ms": 12.036875006742775}, "total_wall_ms": 17806.605626130477, "ttft_ms": 11.839333921670914, "input_usage": {"lookups": 4}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 64, "facts_retained": 11, "retention_rate": 0.171875, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1111111111111111, "amount": 0.07692307692307693, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 609}, "ledger_by_hop": {"1": {"facts_total": 64, "facts_retained": 14, "retention_rate": 0.21875, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.2222222222222222, "amount": 0.15384615384615385, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 870}, "2": {"facts_total": 64, "facts_retained": 11, "retention_rate": 0.171875, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1111111111111111, "amount": 0.07692307692307693, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 609}, "3": {"facts_total": 64, "facts_retained": 11, "retention_rate": 0.171875, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1111111111111111, "amount": 0.07692307692307693, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 609}, "4": {"facts_total": 64, "facts_retained": 11, "retention_rate": 0.171875, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1111111111111111, "amount": 0.07692307692307693, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 609}}}, "reference_key_id": "keys/synthetic/syn_03018_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "830183240b7f45e537ebfeb7fe898082030a951407aec8fa2f3cd9cc13fe9d49", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "145f13624ca0202772cbdd3e", "attempt_id": "1a2a1fa95b63", "run_group_id": "E4_core:A5|repairable|qwen3b:h4", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:13:42.710864Z", "task_family": "synthetic", "document_id": "syn_03018_2000", "document_hash": "d284eb07e27da742b9c2b67abccfdf72e4dd67c3259f44f3b41e34dd3e6fc1ff", "question_id": "q07_negation", "split": "test", "cluster_id": "syn_03018_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 4, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1982, "query_tokens": 355, "message_tokens": 273, "message_utf8_bytes": 657, "cache_tensor_bytes": null, "serialized_payload_bytes": 657, "side_information_bytes": 0, "fetched_bytes": 2027, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 17640.172542189248, "hops": {"1": 6923.268833081238, "2": 3655.3692090092227, "3": 3530.2908330922946, "4": 3531.2436670064926}, "receiver_prefill_ms": 141.64725004229695, "ttft_ms": 11.782416957430542, "decode_ms": 11.780875036492944}, "total_wall_ms": 17805.681209196337, "ttft_ms": 11.782416957430542, "input_usage": {"lookups": 4}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 64, "facts_retained": 11, "retention_rate": 0.171875, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1111111111111111, "amount": 0.07692307692307693, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 609}, "ledger_by_hop": {"1": {"facts_total": 64, "facts_retained": 14, "retention_rate": 0.21875, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.2222222222222222, "amount": 0.15384615384615385, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 870}, "2": {"facts_total": 64, "facts_retained": 11, "retention_rate": 0.171875, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1111111111111111, "amount": 0.07692307692307693, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 609}, "3": {"facts_total": 64, "facts_retained": 11, "retention_rate": 0.171875, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1111111111111111, "amount": 0.07692307692307693, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 609}, "4": {"facts_total": 64, "facts_retained": 11, "retention_rate": 0.171875, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1111111111111111, "amount": 0.07692307692307693, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 609}}}, "reference_key_id": "keys/synthetic/syn_03018_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "830183240b7f45e537ebfeb7fe898082030a951407aec8fa2f3cd9cc13fe9d49", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "f24326862eb37654b95300ef", "attempt_id": "49f274b19bac", "run_group_id": "E4_core:A5|repairable|qwen3b:h4", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:13:42.891023Z", "task_family": "synthetic", "document_id": "syn_03018_2000", "document_hash": "d284eb07e27da742b9c2b67abccfdf72e4dd67c3259f44f3b41e34dd3e6fc1ff", "question_id": "q08_revision", "split": "test", "cluster_id": "syn_03018_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 4, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1982, "query_tokens": 373, "message_tokens": 273, "message_utf8_bytes": 657, "cache_tensor_bytes": null, "serialized_payload_bytes": 657, "side_information_bytes": 0, "fetched_bytes": 2027, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 17640.172542189248, "hops": {"1": 6923.268833081238, "2": 3655.3692090092227, "3": 3530.2908330922946, "4": 3531.2436670064926}, "receiver_prefill_ms": 143.5209170449525, "ttft_ms": 11.878958088345826, "decode_ms": 11.81033393368125}, "total_wall_ms": 17807.7102501411, "ttft_ms": 11.878958088345826, "input_usage": {"lookups": 4}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 64, "facts_retained": 11, "retention_rate": 0.171875, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1111111111111111, "amount": 0.07692307692307693, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 609}, "ledger_by_hop": {"1": {"facts_total": 64, "facts_retained": 14, "retention_rate": 0.21875, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.2222222222222222, "amount": 0.15384615384615385, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 870}, "2": {"facts_total": 64, "facts_retained": 11, "retention_rate": 0.171875, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1111111111111111, "amount": 0.07692307692307693, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 609}, "3": {"facts_total": 64, "facts_retained": 11, "retention_rate": 0.171875, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1111111111111111, "amount": 0.07692307692307693, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 609}, "4": {"facts_total": 64, "facts_retained": 11, "retention_rate": 0.171875, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1111111111111111, "amount": 0.07692307692307693, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 609}}}, "reference_key_id": "keys/synthetic/syn_03018_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "830183240b7f45e537ebfeb7fe898082030a951407aec8fa2f3cd9cc13fe9d49", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "eae182da563ba2ed55db43c0", "attempt_id": "dd688117e252", "run_group_id": "E4_core:A5|repairable|qwen3b:h4", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:13:43.073260Z", "task_family": "synthetic", "document_id": "syn_03018_2000", "document_hash": "d284eb07e27da742b9c2b67abccfdf72e4dd67c3259f44f3b41e34dd3e6fc1ff", "question_id": "q09_revision", "split": "test", "cluster_id": "syn_03018_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 4, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1982, "query_tokens": 366, "message_tokens": 273, "message_utf8_bytes": 657, "cache_tensor_bytes": null, "serialized_payload_bytes": 657, "side_information_bytes": 0, "fetched_bytes": 2027, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 17640.172542189248, "hops": {"1": 6923.268833081238, "2": 3655.3692090092227, "3": 3530.2908330922946, "4": 3531.2436670064926}, "receiver_prefill_ms": 142.6349589601159, "ttft_ms": 11.783791007474065, "decode_ms": 11.788957985118032}, "total_wall_ms": 17806.680833222345, "ttft_ms": 11.783791007474065, "input_usage": {"lookups": 4}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 64, "facts_retained": 11, "retention_rate": 0.171875, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1111111111111111, "amount": 0.07692307692307693, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 609}, "ledger_by_hop": {"1": {"facts_total": 64, "facts_retained": 14, "retention_rate": 0.21875, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.2222222222222222, "amount": 0.15384615384615385, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 870}, "2": {"facts_total": 64, "facts_retained": 11, "retention_rate": 0.171875, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1111111111111111, "amount": 0.07692307692307693, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 609}, "3": {"facts_total": 64, "facts_retained": 11, "retention_rate": 0.171875, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1111111111111111, "amount": 0.07692307692307693, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 609}, "4": {"facts_total": 64, "facts_retained": 11, "retention_rate": 0.171875, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1111111111111111, "amount": 0.07692307692307693, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 609}}}, "reference_key_id": "keys/synthetic/syn_03018_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "830183240b7f45e537ebfeb7fe898082030a951407aec8fa2f3cd9cc13fe9d49", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "45dd1a146455b59c2877c03d", "attempt_id": "61c3bd753b1d", "run_group_id": "E4_core:A5|repairable|qwen3b:h4", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:13:43.253989Z", "task_family": "synthetic", "document_id": "syn_03018_2000", "document_hash": "d284eb07e27da742b9c2b67abccfdf72e4dd67c3259f44f3b41e34dd3e6fc1ff", "question_id": "q10_temporal", "split": "test", "cluster_id": "syn_03018_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 4, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1982, "query_tokens": 374, "message_tokens": 273, "message_utf8_bytes": 657, "cache_tensor_bytes": null, "serialized_payload_bytes": 657, "side_information_bytes": 0, "fetched_bytes": 2027, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 17640.172542189248, "hops": {"1": 6923.268833081238, "2": 3655.3692090092227, "3": 3530.2908330922946, "4": 3531.2436670064926}, "receiver_prefill_ms": 142.86141691263765, "ttft_ms": 11.821000021882355, "decode_ms": 11.823207954876125}, "total_wall_ms": 17806.99591722805, "ttft_ms": 11.821000021882355, "input_usage": {"lookups": 4}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 64, "facts_retained": 11, "retention_rate": 0.171875, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1111111111111111, "amount": 0.07692307692307693, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 609}, "ledger_by_hop": {"1": {"facts_total": 64, "facts_retained": 14, "retention_rate": 0.21875, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.2222222222222222, "amount": 0.15384615384615385, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 870}, "2": {"facts_total": 64, "facts_retained": 11, "retention_rate": 0.171875, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1111111111111111, "amount": 0.07692307692307693, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 609}, "3": {"facts_total": 64, "facts_retained": 11, "retention_rate": 0.171875, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1111111111111111, "amount": 0.07692307692307693, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 609}, "4": {"facts_total": 64, "facts_retained": 11, "retention_rate": 0.171875, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1111111111111111, "amount": 0.07692307692307693, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 609}}}, "reference_key_id": "keys/synthetic/syn_03018_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "830183240b7f45e537ebfeb7fe898082030a951407aec8fa2f3cd9cc13fe9d49", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "620954ff8cb43faef9058932", "attempt_id": "0d908e560d5f", "run_group_id": "E4_core:A5|repairable|qwen3b:h4", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:13:43.435376Z", "task_family": "synthetic", "document_id": "syn_03018_2000", "document_hash": "d284eb07e27da742b9c2b67abccfdf72e4dd67c3259f44f3b41e34dd3e6fc1ff", "question_id": "q11_aggregation", "split": "test", "cluster_id": "syn_03018_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 4, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1982, "query_tokens": 366, "message_tokens": 273, "message_utf8_bytes": 657, "cache_tensor_bytes": null, "serialized_payload_bytes": 657, "side_information_bytes": 0, "fetched_bytes": 2027, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 17640.172542189248, "hops": {"1": 6923.268833081238, "2": 3655.3692090092227, "3": 3530.2908330922946, "4": 3531.2436670064926}, "receiver_prefill_ms": 142.8623329848051, "ttft_ms": 11.771041085012257, "decode_ms": 82.20679208170623}, "total_wall_ms": 17877.31708411593, "ttft_ms": 11.771041085012257, "input_usage": {"lookups": 4}, "output_usage": {"output_tokens": 7}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "2917061", "evidence_artifact": {"ledger": {"facts_total": 64, "facts_retained": 11, "retention_rate": 0.171875, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1111111111111111, "amount": 0.07692307692307693, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 609}, "ledger_by_hop": {"1": {"facts_total": 64, "facts_retained": 14, "retention_rate": 0.21875, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.2222222222222222, "amount": 0.15384615384615385, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 870}, "2": {"facts_total": 64, "facts_retained": 11, "retention_rate": 0.171875, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1111111111111111, "amount": 0.07692307692307693, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 609}, "3": {"facts_total": 64, "facts_retained": 11, "retention_rate": 0.171875, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1111111111111111, "amount": 0.07692307692307693, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 609}, "4": {"facts_total": 64, "facts_retained": 11, "retention_rate": 0.171875, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1111111111111111, "amount": 0.07692307692307693, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 609}}}, "reference_key_id": "keys/synthetic/syn_03018_2000.json", "task_scores": {"correct": 0.0, "abstained": 0.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "830183240b7f45e537ebfeb7fe898082030a951407aec8fa2f3cd9cc13fe9d49", "output_hash": "129c25eb43beebf0b464953aaabebfe914d11fa686f1dce105e62022958e1719"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "c413c8849f1aa6310b4e664a", "attempt_id": "1973aa989c9a", "run_group_id": "E4_core:A5|repairable|qwen3b:h4", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:13:43.687717Z", "task_family": "synthetic", "document_id": "syn_03018_2000", "document_hash": "d284eb07e27da742b9c2b67abccfdf72e4dd67c3259f44f3b41e34dd3e6fc1ff", "question_id": "q12_unit_conversion", "split": "test", "cluster_id": "syn_03018_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 4, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1982, "query_tokens": 363, "message_tokens": 273, "message_utf8_bytes": 657, "cache_tensor_bytes": null, "serialized_payload_bytes": 657, "side_information_bytes": 0, "fetched_bytes": 2027, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 17640.172542189248, "hops": {"1": 6923.268833081238, "2": 3655.3692090092227, "3": 3530.2908330922946, "4": 3531.2436670064926}, "receiver_prefill_ms": 142.49300002120435, "ttft_ms": 11.763500049710274, "decode_ms": 47.24675009492785}, "total_wall_ms": 17841.99300117325, "ttft_ms": 11.763500049710274, "input_usage": {"lookups": 4}, "output_usage": {"output_tokens": 4}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "6400", "evidence_artifact": {"ledger": {"facts_total": 64, "facts_retained": 11, "retention_rate": 0.171875, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1111111111111111, "amount": 0.07692307692307693, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 609}, "ledger_by_hop": {"1": {"facts_total": 64, "facts_retained": 14, "retention_rate": 0.21875, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.2222222222222222, "amount": 0.15384615384615385, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 870}, "2": {"facts_total": 64, "facts_retained": 11, "retention_rate": 0.171875, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1111111111111111, "amount": 0.07692307692307693, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 609}, "3": {"facts_total": 64, "facts_retained": 11, "retention_rate": 0.171875, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1111111111111111, "amount": 0.07692307692307693, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 609}, "4": {"facts_total": 64, "facts_retained": 11, "retention_rate": 0.171875, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1111111111111111, "amount": 0.07692307692307693, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 609}}}, "reference_key_id": "keys/synthetic/syn_03018_2000.json", "task_scores": {"correct": 0.0, "abstained": 0.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "830183240b7f45e537ebfeb7fe898082030a951407aec8fa2f3cd9cc13fe9d49", "output_hash": "a7dccef9ce1ae31c49a4cfc484f8411dd52b11c914c1fe3d781633642d1e1327"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "74042d56dc59a00ea3f4bed6", "attempt_id": "96e332f4fcee", "run_group_id": "E4_core:A5|repairable|qwen3b:h4", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:13:43.904260Z", "task_family": "synthetic", "document_id": "syn_03018_2000", "document_hash": "d284eb07e27da742b9c2b67abccfdf72e4dd67c3259f44f3b41e34dd3e6fc1ff", "question_id": "q13_unanswerable", "split": "test", "cluster_id": "syn_03018_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 4, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1982, "query_tokens": 367, "message_tokens": 273, "message_utf8_bytes": 657, "cache_tensor_bytes": null, "serialized_payload_bytes": 657, "side_information_bytes": 0, "fetched_bytes": 2027, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 17640.172542189248, "hops": {"1": 6923.268833081238, "2": 3655.3692090092227, "3": 3530.2908330922946, "4": 3531.2436670064926}, "receiver_prefill_ms": 142.6909159636125, "ttft_ms": 11.818582890555263, "decode_ms": 11.862165993079543}, "total_wall_ms": 17806.85895914212, "ttft_ms": 11.818582890555263, "input_usage": {"lookups": 4}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 64, "facts_retained": 11, "retention_rate": 0.171875, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1111111111111111, "amount": 0.07692307692307693, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 609}, "ledger_by_hop": {"1": {"facts_total": 64, "facts_retained": 14, "retention_rate": 0.21875, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.2222222222222222, "amount": 0.15384615384615385, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 870}, "2": {"facts_total": 64, "facts_retained": 11, "retention_rate": 0.171875, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1111111111111111, "amount": 0.07692307692307693, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 609}, "3": {"facts_total": 64, "facts_retained": 11, "retention_rate": 0.171875, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1111111111111111, "amount": 0.07692307692307693, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 609}, "4": {"facts_total": 64, "facts_retained": 11, "retention_rate": 0.171875, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1111111111111111, "amount": 0.07692307692307693, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 609}}}, "reference_key_id": "keys/synthetic/syn_03018_2000.json", "task_scores": {"correct": 1.0, "abstained": 1.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "830183240b7f45e537ebfeb7fe898082030a951407aec8fa2f3cd9cc13fe9d49", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "3e96045d2cccd56993498306", "attempt_id": "b6ed21fcda44", "run_group_id": "E4_core:A5|repairable|qwen3b:h4", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:13:44.085754Z", "task_family": "synthetic", "document_id": "syn_03018_2000", "document_hash": "d284eb07e27da742b9c2b67abccfdf72e4dd67c3259f44f3b41e34dd3e6fc1ff", "question_id": "q14_unanswerable", "split": "test", "cluster_id": "syn_03018_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 4, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1982, "query_tokens": 368, "message_tokens": 273, "message_utf8_bytes": 657, "cache_tensor_bytes": null, "serialized_payload_bytes": 657, "side_information_bytes": 0, "fetched_bytes": 2027, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 17640.172542189248, "hops": {"1": 6923.268833081238, "2": 3655.3692090092227, "3": 3530.2908330922946, "4": 3531.2436670064926}, "receiver_prefill_ms": 142.79666601214558, "ttft_ms": 11.799917090684175, "decode_ms": 11.825458030216396}, "total_wall_ms": 17806.90412526019, "ttft_ms": 11.799917090684175, "input_usage": {"lookups": 4}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 64, "facts_retained": 11, "retention_rate": 0.171875, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1111111111111111, "amount": 0.07692307692307693, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 609}, "ledger_by_hop": {"1": {"facts_total": 64, "facts_retained": 14, "retention_rate": 0.21875, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.2222222222222222, "amount": 0.15384615384615385, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 870}, "2": {"facts_total": 64, "facts_retained": 11, "retention_rate": 0.171875, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1111111111111111, "amount": 0.07692307692307693, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 609}, "3": {"facts_total": 64, "facts_retained": 11, "retention_rate": 0.171875, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1111111111111111, "amount": 0.07692307692307693, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 609}, "4": {"facts_total": 64, "facts_retained": 11, "retention_rate": 0.171875, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1111111111111111, "amount": 0.07692307692307693, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 609}}}, "reference_key_id": "keys/synthetic/syn_03018_2000.json", "task_scores": {"correct": 1.0, "abstained": 1.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "830183240b7f45e537ebfeb7fe898082030a951407aec8fa2f3cd9cc13fe9d49", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "eb611142f0f56276caca3693", "attempt_id": "f279948b85f7", "run_group_id": "E4_core:A5|repairable|qwen3b:h8", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:13:58.367634Z", "task_family": "synthetic", "document_id": "syn_03018_2000", "document_hash": "d284eb07e27da742b9c2b67abccfdf72e4dd67c3259f44f3b41e34dd3e6fc1ff", "question_id": "q00_single_fact", "split": "test", "cluster_id": "syn_03018_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 8, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1982, "query_tokens": 355, "message_tokens": 273, "message_utf8_bytes": 657, "cache_tensor_bytes": null, "serialized_payload_bytes": 657, "side_information_bytes": 0, "fetched_bytes": 3899, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 31736.605958081782, "hops": {"1": 6923.268833081238, "2": 3655.3692090092227, "3": 3530.2908330922946, "4": 3531.2436670064926, "5": 3533.83945801761, "6": 3521.261749905534, "7": 3521.8319579726085, "8": 3519.5002499967813}, "receiver_prefill_ms": 142.30383303947747, "ttft_ms": 11.815540958195925, "decode_ms": 11.762707959860563}, "total_wall_ms": 31902.847542078234, "ttft_ms": 11.815540958195925, "input_usage": {"lookups": 8}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 64, "facts_retained": 11, "retention_rate": 0.171875, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1111111111111111, "amount": 0.07692307692307693, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 609}, "ledger_by_hop": {"1": {"facts_total": 64, "facts_retained": 14, "retention_rate": 0.21875, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.2222222222222222, "amount": 0.15384615384615385, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 870}, "2": {"facts_total": 64, "facts_retained": 11, "retention_rate": 0.171875, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1111111111111111, "amount": 0.07692307692307693, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 609}, "3": {"facts_total": 64, "facts_retained": 11, "retention_rate": 0.171875, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1111111111111111, "amount": 0.07692307692307693, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 609}, "4": {"facts_total": 64, "facts_retained": 11, "retention_rate": 0.171875, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1111111111111111, "amount": 0.07692307692307693, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 609}, "5": {"facts_total": 64, "facts_retained": 11, "retention_rate": 0.171875, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1111111111111111, "amount": 0.07692307692307693, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 609}, "6": {"facts_total": 64, "facts_retained": 11, "retention_rate": 0.171875, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1111111111111111, "amount": 0.07692307692307693, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 609}, "7": {"facts_total": 64, "facts_retained": 11, "retention_rate": 0.171875, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1111111111111111, "amount": 0.07692307692307693, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 609}, "8": {"facts_total": 64, "facts_retained": 11, "retention_rate": 0.171875, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1111111111111111, "amount": 0.07692307692307693, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 609}}}, "reference_key_id": "keys/synthetic/syn_03018_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "830183240b7f45e537ebfeb7fe898082030a951407aec8fa2f3cd9cc13fe9d49", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "9367bfd1dac487e318bd90d5", "attempt_id": "2748e88d332f", "run_group_id": "E4_core:A5|repairable|qwen3b:h8", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:13:58.549440Z", "task_family": "synthetic", "document_id": "syn_03018_2000", "document_hash": "d284eb07e27da742b9c2b67abccfdf72e4dd67c3259f44f3b41e34dd3e6fc1ff", "question_id": "q01_single_fact", "split": "test", "cluster_id": "syn_03018_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 8, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1982, "query_tokens": 361, "message_tokens": 273, "message_utf8_bytes": 657, "cache_tensor_bytes": null, "serialized_payload_bytes": 657, "side_information_bytes": 0, "fetched_bytes": 3899, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 31736.605958081782, "hops": {"1": 6923.268833081238, "2": 3655.3692090092227, "3": 3530.2908330922946, "4": 3531.2436670064926, "5": 3533.83945801761, "6": 3521.261749905534, "7": 3521.8319579726085, "8": 3519.5002499967813}, "receiver_prefill_ms": 142.3572499770671, "ttft_ms": 11.868042056448758, "decode_ms": 11.762625072151423}, "total_wall_ms": 31902.92379213497, "ttft_ms": 11.868042056448758, "input_usage": {"lookups": 8}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 64, "facts_retained": 11, "retention_rate": 0.171875, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1111111111111111, "amount": 0.07692307692307693, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 609}, "ledger_by_hop": {"1": {"facts_total": 64, "facts_retained": 14, "retention_rate": 0.21875, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.2222222222222222, "amount": 0.15384615384615385, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 870}, "2": {"facts_total": 64, "facts_retained": 11, "retention_rate": 0.171875, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1111111111111111, "amount": 0.07692307692307693, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 609}, "3": {"facts_total": 64, "facts_retained": 11, "retention_rate": 0.171875, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1111111111111111, "amount": 0.07692307692307693, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 609}, "4": {"facts_total": 64, "facts_retained": 11, "retention_rate": 0.171875, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1111111111111111, "amount": 0.07692307692307693, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 609}, "5": {"facts_total": 64, "facts_retained": 11, "retention_rate": 0.171875, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1111111111111111, "amount": 0.07692307692307693, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 609}, "6": {"facts_total": 64, "facts_retained": 11, "retention_rate": 0.171875, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1111111111111111, "amount": 0.07692307692307693, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 609}, "7": {"facts_total": 64, "facts_retained": 11, "retention_rate": 0.171875, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1111111111111111, "amount": 0.07692307692307693, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 609}, "8": {"facts_total": 64, "facts_retained": 11, "retention_rate": 0.171875, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1111111111111111, "amount": 0.07692307692307693, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 609}}}, "reference_key_id": "keys/synthetic/syn_03018_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "830183240b7f45e537ebfeb7fe898082030a951407aec8fa2f3cd9cc13fe9d49", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "2fd9e080e1b41ef2eb36d8b3", "attempt_id": "293d89d5b1c0", "run_group_id": "E4_core:A5|repairable|qwen3b:h8", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:13:58.730579Z", "task_family": "synthetic", "document_id": "syn_03018_2000", "document_hash": "d284eb07e27da742b9c2b67abccfdf72e4dd67c3259f44f3b41e34dd3e6fc1ff", "question_id": "q02_exact_copy", "split": "test", "cluster_id": "syn_03018_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 8, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1982, "query_tokens": 360, "message_tokens": 273, "message_utf8_bytes": 657, "cache_tensor_bytes": null, "serialized_payload_bytes": 657, "side_information_bytes": 0, "fetched_bytes": 3899, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 31736.605958081782, "hops": {"1": 6923.268833081238, "2": 3655.3692090092227, "3": 3530.2908330922946, "4": 3531.2436670064926, "5": 3533.83945801761, "6": 3521.261749905534, "7": 3521.8319579726085, "8": 3519.5002499967813}, "receiver_prefill_ms": 142.6954169292003, "ttft_ms": 11.841334053315222, "decode_ms": 11.755459010601044}, "total_wall_ms": 31903.2039581798, "ttft_ms": 11.841334053315222, "input_usage": {"lookups": 8}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 64, "facts_retained": 11, "retention_rate": 0.171875, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1111111111111111, "amount": 0.07692307692307693, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 609}, "ledger_by_hop": {"1": {"facts_total": 64, "facts_retained": 14, "retention_rate": 0.21875, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.2222222222222222, "amount": 0.15384615384615385, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 870}, "2": {"facts_total": 64, "facts_retained": 11, "retention_rate": 0.171875, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1111111111111111, "amount": 0.07692307692307693, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 609}, "3": {"facts_total": 64, "facts_retained": 11, "retention_rate": 0.171875, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1111111111111111, "amount": 0.07692307692307693, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 609}, "4": {"facts_total": 64, "facts_retained": 11, "retention_rate": 0.171875, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1111111111111111, "amount": 0.07692307692307693, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 609}, "5": {"facts_total": 64, "facts_retained": 11, "retention_rate": 0.171875, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1111111111111111, "amount": 0.07692307692307693, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 609}, "6": {"facts_total": 64, "facts_retained": 11, "retention_rate": 0.171875, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1111111111111111, "amount": 0.07692307692307693, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 609}, "7": {"facts_total": 64, "facts_retained": 11, "retention_rate": 0.171875, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1111111111111111, "amount": 0.07692307692307693, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 609}, "8": {"facts_total": 64, "facts_retained": 11, "retention_rate": 0.171875, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1111111111111111, "amount": 0.07692307692307693, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 609}}}, "reference_key_id": "keys/synthetic/syn_03018_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "830183240b7f45e537ebfeb7fe898082030a951407aec8fa2f3cd9cc13fe9d49", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "07eaeb1164641116a6ea8bb6", "attempt_id": "3209155461cf", "run_group_id": "E4_core:A5|repairable|qwen3b:h8", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:13:58.912059Z", "task_family": "synthetic", "document_id": "syn_03018_2000", "document_hash": "d284eb07e27da742b9c2b67abccfdf72e4dd67c3259f44f3b41e34dd3e6fc1ff", "question_id": "q03_exact_copy", "split": "test", "cluster_id": "syn_03018_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 8, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1982, "query_tokens": 358, "message_tokens": 273, "message_utf8_bytes": 657, "cache_tensor_bytes": null, "serialized_payload_bytes": 657, "side_information_bytes": 0, "fetched_bytes": 3899, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 31736.605958081782, "hops": {"1": 6923.268833081238, "2": 3655.3692090092227, "3": 3530.2908330922946, "4": 3531.2436670064926, "5": 3533.83945801761, "6": 3521.261749905534, "7": 3521.8319579726085, "8": 3519.5002499967813}, "receiver_prefill_ms": 142.29379105381668, "ttft_ms": 11.736041051335633, "decode_ms": 11.826999951153994}, "total_wall_ms": 31902.78416709043, "ttft_ms": 11.736041051335633, "input_usage": {"lookups": 8}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 64, "facts_retained": 11, "retention_rate": 0.171875, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1111111111111111, "amount": 0.07692307692307693, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 609}, "ledger_by_hop": {"1": {"facts_total": 64, "facts_retained": 14, "retention_rate": 0.21875, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.2222222222222222, "amount": 0.15384615384615385, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 870}, "2": {"facts_total": 64, "facts_retained": 11, "retention_rate": 0.171875, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1111111111111111, "amount": 0.07692307692307693, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 609}, "3": {"facts_total": 64, "facts_retained": 11, "retention_rate": 0.171875, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1111111111111111, "amount": 0.07692307692307693, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 609}, "4": {"facts_total": 64, "facts_retained": 11, "retention_rate": 0.171875, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1111111111111111, "amount": 0.07692307692307693, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 609}, "5": {"facts_total": 64, "facts_retained": 11, "retention_rate": 0.171875, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1111111111111111, "amount": 0.07692307692307693, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 609}, "6": {"facts_total": 64, "facts_retained": 11, "retention_rate": 0.171875, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1111111111111111, "amount": 0.07692307692307693, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 609}, "7": {"facts_total": 64, "facts_retained": 11, "retention_rate": 0.171875, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1111111111111111, "amount": 0.07692307692307693, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 609}, "8": {"facts_total": 64, "facts_retained": 11, "retention_rate": 0.171875, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1111111111111111, "amount": 0.07692307692307693, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 609}}}, "reference_key_id": "keys/synthetic/syn_03018_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "830183240b7f45e537ebfeb7fe898082030a951407aec8fa2f3cd9cc13fe9d49", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "e6012c98215261c82138b047", "attempt_id": "137681e8facd", "run_group_id": "E4_core:A5|repairable|qwen3b:h8", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:13:59.092907Z", "task_family": "synthetic", "document_id": "syn_03018_2000", "document_hash": "d284eb07e27da742b9c2b67abccfdf72e4dd67c3259f44f3b41e34dd3e6fc1ff", "question_id": "q04_multi_hop", "split": "test", "cluster_id": "syn_03018_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 8, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1982, "query_tokens": 364, "message_tokens": 273, "message_utf8_bytes": 657, "cache_tensor_bytes": null, "serialized_payload_bytes": 657, "side_information_bytes": 0, "fetched_bytes": 3899, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 31736.605958081782, "hops": {"1": 6923.268833081238, "2": 3655.3692090092227, "3": 3530.2908330922946, "4": 3531.2436670064926, "5": 3533.83945801761, "6": 3521.261749905534, "7": 3521.8319579726085, "8": 3519.5002499967813}, "receiver_prefill_ms": 142.69654091913253, "ttft_ms": 11.767250020056963, "decode_ms": 11.800625012256205}, "total_wall_ms": 31903.19800004363, "ttft_ms": 11.767250020056963, "input_usage": {"lookups": 8}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 64, "facts_retained": 11, "retention_rate": 0.171875, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1111111111111111, "amount": 0.07692307692307693, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 609}, "ledger_by_hop": {"1": {"facts_total": 64, "facts_retained": 14, "retention_rate": 0.21875, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.2222222222222222, "amount": 0.15384615384615385, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 870}, "2": {"facts_total": 64, "facts_retained": 11, "retention_rate": 0.171875, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1111111111111111, "amount": 0.07692307692307693, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 609}, "3": {"facts_total": 64, "facts_retained": 11, "retention_rate": 0.171875, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1111111111111111, "amount": 0.07692307692307693, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 609}, "4": {"facts_total": 64, "facts_retained": 11, "retention_rate": 0.171875, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1111111111111111, "amount": 0.07692307692307693, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 609}, "5": {"facts_total": 64, "facts_retained": 11, "retention_rate": 0.171875, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1111111111111111, "amount": 0.07692307692307693, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 609}, "6": {"facts_total": 64, "facts_retained": 11, "retention_rate": 0.171875, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1111111111111111, "amount": 0.07692307692307693, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 609}, "7": {"facts_total": 64, "facts_retained": 11, "retention_rate": 0.171875, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1111111111111111, "amount": 0.07692307692307693, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 609}, "8": {"facts_total": 64, "facts_retained": 11, "retention_rate": 0.171875, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1111111111111111, "amount": 0.07692307692307693, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 609}}}, "reference_key_id": "keys/synthetic/syn_03018_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "830183240b7f45e537ebfeb7fe898082030a951407aec8fa2f3cd9cc13fe9d49", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "c809c37cad7bc20a9ff0d7b2", "attempt_id": "08793990c828", "run_group_id": "E4_core:A5|repairable|qwen3b:h8", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:13:59.273889Z", "task_family": "synthetic", "document_id": "syn_03018_2000", "document_hash": "d284eb07e27da742b9c2b67abccfdf72e4dd67c3259f44f3b41e34dd3e6fc1ff", "question_id": "q05_multi_hop", "split": "test", "cluster_id": "syn_03018_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 8, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1982, "query_tokens": 364, "message_tokens": 273, "message_utf8_bytes": 657, "cache_tensor_bytes": null, "serialized_payload_bytes": 657, "side_information_bytes": 0, "fetched_bytes": 3899, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 31736.605958081782, "hops": {"1": 6923.268833081238, "2": 3655.3692090092227, "3": 3530.2908330922946, "4": 3531.2436670064926, "5": 3533.83945801761, "6": 3521.261749905534, "7": 3521.8319579726085, "8": 3519.5002499967813}, "receiver_prefill_ms": 141.93141588475555, "ttft_ms": 11.765584000386298, "decode_ms": 11.758457985706627}, "total_wall_ms": 31902.371583040804, "ttft_ms": 11.765584000386298, "input_usage": {"lookups": 8}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 64, "facts_retained": 11, "retention_rate": 0.171875, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1111111111111111, "amount": 0.07692307692307693, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 609}, "ledger_by_hop": {"1": {"facts_total": 64, "facts_retained": 14, "retention_rate": 0.21875, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.2222222222222222, "amount": 0.15384615384615385, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 870}, "2": {"facts_total": 64, "facts_retained": 11, "retention_rate": 0.171875, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1111111111111111, "amount": 0.07692307692307693, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 609}, "3": {"facts_total": 64, "facts_retained": 11, "retention_rate": 0.171875, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1111111111111111, "amount": 0.07692307692307693, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 609}, "4": {"facts_total": 64, "facts_retained": 11, "retention_rate": 0.171875, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1111111111111111, "amount": 0.07692307692307693, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 609}, "5": {"facts_total": 64, "facts_retained": 11, "retention_rate": 0.171875, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1111111111111111, "amount": 0.07692307692307693, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 609}, "6": {"facts_total": 64, "facts_retained": 11, "retention_rate": 0.171875, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1111111111111111, "amount": 0.07692307692307693, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 609}, "7": {"facts_total": 64, "facts_retained": 11, "retention_rate": 0.171875, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1111111111111111, "amount": 0.07692307692307693, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 609}, "8": {"facts_total": 64, "facts_retained": 11, "retention_rate": 0.171875, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1111111111111111, "amount": 0.07692307692307693, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 609}}}, "reference_key_id": "keys/synthetic/syn_03018_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "830183240b7f45e537ebfeb7fe898082030a951407aec8fa2f3cd9cc13fe9d49", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "d2717052b23a9183af847768", "attempt_id": "ba2f9c47aa92", "run_group_id": "E4_core:A5|repairable|qwen3b:h8", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:13:59.454228Z", "task_family": "synthetic", "document_id": "syn_03018_2000", "document_hash": "d284eb07e27da742b9c2b67abccfdf72e4dd67c3259f44f3b41e34dd3e6fc1ff", "question_id": "q06_negation", "split": "test", "cluster_id": "syn_03018_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 8, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1982, "query_tokens": 356, "message_tokens": 273, "message_utf8_bytes": 657, "cache_tensor_bytes": null, "serialized_payload_bytes": 657, "side_information_bytes": 0, "fetched_bytes": 3899, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 31736.605958081782, "hops": {"1": 6923.268833081238, "2": 3655.3692090092227, "3": 3530.2908330922946, "4": 3531.2436670064926, "5": 3533.83945801761, "6": 3521.261749905534, "7": 3521.8319579726085, "8": 3519.5002499967813}, "receiver_prefill_ms": 142.3501669196412, "ttft_ms": 11.757500004023314, "decode_ms": 11.75041706301272}, "total_wall_ms": 31902.785875136033, "ttft_ms": 11.757500004023314, "input_usage": {"lookups": 8}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 64, "facts_retained": 11, "retention_rate": 0.171875, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1111111111111111, "amount": 0.07692307692307693, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 609}, "ledger_by_hop": {"1": {"facts_total": 64, "facts_retained": 14, "retention_rate": 0.21875, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.2222222222222222, "amount": 0.15384615384615385, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 870}, "2": {"facts_total": 64, "facts_retained": 11, "retention_rate": 0.171875, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1111111111111111, "amount": 0.07692307692307693, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 609}, "3": {"facts_total": 64, "facts_retained": 11, "retention_rate": 0.171875, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1111111111111111, "amount": 0.07692307692307693, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 609}, "4": {"facts_total": 64, "facts_retained": 11, "retention_rate": 0.171875, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1111111111111111, "amount": 0.07692307692307693, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 609}, "5": {"facts_total": 64, "facts_retained": 11, "retention_rate": 0.171875, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1111111111111111, "amount": 0.07692307692307693, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 609}, "6": {"facts_total": 64, "facts_retained": 11, "retention_rate": 0.171875, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1111111111111111, "amount": 0.07692307692307693, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 609}, "7": {"facts_total": 64, "facts_retained": 11, "retention_rate": 0.171875, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1111111111111111, "amount": 0.07692307692307693, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 609}, "8": {"facts_total": 64, "facts_retained": 11, "retention_rate": 0.171875, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1111111111111111, "amount": 0.07692307692307693, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 609}}}, "reference_key_id": "keys/synthetic/syn_03018_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "830183240b7f45e537ebfeb7fe898082030a951407aec8fa2f3cd9cc13fe9d49", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "84570e2d967c3cfcf213e843", "attempt_id": "d364b0c20b9f", "run_group_id": "E4_core:A5|repairable|qwen3b:h8", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:13:59.634838Z", "task_family": "synthetic", "document_id": "syn_03018_2000", "document_hash": "d284eb07e27da742b9c2b67abccfdf72e4dd67c3259f44f3b41e34dd3e6fc1ff", "question_id": "q07_negation", "split": "test", "cluster_id": "syn_03018_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 8, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1982, "query_tokens": 355, "message_tokens": 273, "message_utf8_bytes": 657, "cache_tensor_bytes": null, "serialized_payload_bytes": 657, "side_information_bytes": 0, "fetched_bytes": 3899, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 31736.605958081782, "hops": {"1": 6923.268833081238, "2": 3655.3692090092227, "3": 3530.2908330922946, "4": 3531.2436670064926, "5": 3533.83945801761, "6": 3521.261749905534, "7": 3521.8319579726085, "8": 3519.5002499967813}, "receiver_prefill_ms": 142.26816594600677, "ttft_ms": 11.777333915233612, "decode_ms": 11.780166998505592}, "total_wall_ms": 31902.739083045162, "ttft_ms": 11.777333915233612, "input_usage": {"lookups": 8}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 64, "facts_retained": 11, "retention_rate": 0.171875, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1111111111111111, "amount": 0.07692307692307693, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 609}, "ledger_by_hop": {"1": {"facts_total": 64, "facts_retained": 14, "retention_rate": 0.21875, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.2222222222222222, "amount": 0.15384615384615385, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 870}, "2": {"facts_total": 64, "facts_retained": 11, "retention_rate": 0.171875, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1111111111111111, "amount": 0.07692307692307693, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 609}, "3": {"facts_total": 64, "facts_retained": 11, "retention_rate": 0.171875, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1111111111111111, "amount": 0.07692307692307693, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 609}, "4": {"facts_total": 64, "facts_retained": 11, "retention_rate": 0.171875, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1111111111111111, "amount": 0.07692307692307693, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 609}, "5": {"facts_total": 64, "facts_retained": 11, "retention_rate": 0.171875, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1111111111111111, "amount": 0.07692307692307693, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 609}, "6": {"facts_total": 64, "facts_retained": 11, "retention_rate": 0.171875, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1111111111111111, "amount": 0.07692307692307693, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 609}, "7": {"facts_total": 64, "facts_retained": 11, "retention_rate": 0.171875, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1111111111111111, "amount": 0.07692307692307693, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 609}, "8": {"facts_total": 64, "facts_retained": 11, "retention_rate": 0.171875, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1111111111111111, "amount": 0.07692307692307693, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 609}}}, "reference_key_id": "keys/synthetic/syn_03018_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "830183240b7f45e537ebfeb7fe898082030a951407aec8fa2f3cd9cc13fe9d49", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "07e471f0eefe71a82c950448", "attempt_id": "5a2ea52fe6e1", "run_group_id": "E4_core:A5|repairable|qwen3b:h8", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:13:59.815336Z", "task_family": "synthetic", "document_id": "syn_03018_2000", "document_hash": "d284eb07e27da742b9c2b67abccfdf72e4dd67c3259f44f3b41e34dd3e6fc1ff", "question_id": "q08_revision", "split": "test", "cluster_id": "syn_03018_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 8, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1982, "query_tokens": 373, "message_tokens": 273, "message_utf8_bytes": 657, "cache_tensor_bytes": null, "serialized_payload_bytes": 657, "side_information_bytes": 0, "fetched_bytes": 3899, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 31736.605958081782, "hops": {"1": 6923.268833081238, "2": 3655.3692090092227, "3": 3530.2908330922946, "4": 3531.2436670064926, "5": 3533.83945801761, "6": 3521.261749905534, "7": 3521.8319579726085, "8": 3519.5002499967813}, "receiver_prefill_ms": 142.80479203443974, "ttft_ms": 11.828084010630846, "decode_ms": 11.773915961384773}, "total_wall_ms": 31903.351916116662, "ttft_ms": 11.828084010630846, "input_usage": {"lookups": 8}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 64, "facts_retained": 11, "retention_rate": 0.171875, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1111111111111111, "amount": 0.07692307692307693, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 609}, "ledger_by_hop": {"1": {"facts_total": 64, "facts_retained": 14, "retention_rate": 0.21875, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.2222222222222222, "amount": 0.15384615384615385, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 870}, "2": {"facts_total": 64, "facts_retained": 11, "retention_rate": 0.171875, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1111111111111111, "amount": 0.07692307692307693, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 609}, "3": {"facts_total": 64, "facts_retained": 11, "retention_rate": 0.171875, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1111111111111111, "amount": 0.07692307692307693, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 609}, "4": {"facts_total": 64, "facts_retained": 11, "retention_rate": 0.171875, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1111111111111111, "amount": 0.07692307692307693, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 609}, "5": {"facts_total": 64, "facts_retained": 11, "retention_rate": 0.171875, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1111111111111111, "amount": 0.07692307692307693, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 609}, "6": {"facts_total": 64, "facts_retained": 11, "retention_rate": 0.171875, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1111111111111111, "amount": 0.07692307692307693, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 609}, "7": {"facts_total": 64, "facts_retained": 11, "retention_rate": 0.171875, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1111111111111111, "amount": 0.07692307692307693, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 609}, "8": {"facts_total": 64, "facts_retained": 11, "retention_rate": 0.171875, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1111111111111111, "amount": 0.07692307692307693, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 609}}}, "reference_key_id": "keys/synthetic/syn_03018_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "830183240b7f45e537ebfeb7fe898082030a951407aec8fa2f3cd9cc13fe9d49", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "c5a7f83551903562cf8d4281", "attempt_id": "f7c9980194c1", "run_group_id": "E4_core:A5|repairable|qwen3b:h8", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:13:59.996342Z", "task_family": "synthetic", "document_id": "syn_03018_2000", "document_hash": "d284eb07e27da742b9c2b67abccfdf72e4dd67c3259f44f3b41e34dd3e6fc1ff", "question_id": "q09_revision", "split": "test", "cluster_id": "syn_03018_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 8, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1982, "query_tokens": 366, "message_tokens": 273, "message_utf8_bytes": 657, "cache_tensor_bytes": null, "serialized_payload_bytes": 657, "side_information_bytes": 0, "fetched_bytes": 3899, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 31736.605958081782, "hops": {"1": 6923.268833081238, "2": 3655.3692090092227, "3": 3530.2908330922946, "4": 3531.2436670064926, "5": 3533.83945801761, "6": 3521.261749905534, "7": 3521.8319579726085, "8": 3519.5002499967813}, "receiver_prefill_ms": 142.6093749469146, "ttft_ms": 11.792000033892691, "decode_ms": 11.762792011722922}, "total_wall_ms": 31903.128541074693, "ttft_ms": 11.792000033892691, "input_usage": {"lookups": 8}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 64, "facts_retained": 11, "retention_rate": 0.171875, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1111111111111111, "amount": 0.07692307692307693, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 609}, "ledger_by_hop": {"1": {"facts_total": 64, "facts_retained": 14, "retention_rate": 0.21875, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.2222222222222222, "amount": 0.15384615384615385, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 870}, "2": {"facts_total": 64, "facts_retained": 11, "retention_rate": 0.171875, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1111111111111111, "amount": 0.07692307692307693, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 609}, "3": {"facts_total": 64, "facts_retained": 11, "retention_rate": 0.171875, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1111111111111111, "amount": 0.07692307692307693, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 609}, "4": {"facts_total": 64, "facts_retained": 11, "retention_rate": 0.171875, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1111111111111111, "amount": 0.07692307692307693, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 609}, "5": {"facts_total": 64, "facts_retained": 11, "retention_rate": 0.171875, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1111111111111111, "amount": 0.07692307692307693, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 609}, "6": {"facts_total": 64, "facts_retained": 11, "retention_rate": 0.171875, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1111111111111111, "amount": 0.07692307692307693, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 609}, "7": {"facts_total": 64, "facts_retained": 11, "retention_rate": 0.171875, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1111111111111111, "amount": 0.07692307692307693, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 609}, "8": {"facts_total": 64, "facts_retained": 11, "retention_rate": 0.171875, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1111111111111111, "amount": 0.07692307692307693, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 609}}}, "reference_key_id": "keys/synthetic/syn_03018_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "830183240b7f45e537ebfeb7fe898082030a951407aec8fa2f3cd9cc13fe9d49", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "4e5109ed84ac03dc800bcd64", "attempt_id": "3ece50c686d9", "run_group_id": "E4_core:A5|repairable|qwen3b:h8", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:14:00.177708Z", "task_family": "synthetic", "document_id": "syn_03018_2000", "document_hash": "d284eb07e27da742b9c2b67abccfdf72e4dd67c3259f44f3b41e34dd3e6fc1ff", "question_id": "q10_temporal", "split": "test", "cluster_id": "syn_03018_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 8, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1982, "query_tokens": 374, "message_tokens": 273, "message_utf8_bytes": 657, "cache_tensor_bytes": null, "serialized_payload_bytes": 657, "side_information_bytes": 0, "fetched_bytes": 3899, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 31736.605958081782, "hops": {"1": 6923.268833081238, "2": 3655.3692090092227, "3": 3530.2908330922946, "4": 3531.2436670064926, "5": 3533.83945801761, "6": 3521.261749905534, "7": 3521.8319579726085, "8": 3519.5002499967813}, "receiver_prefill_ms": 142.92066695634276, "ttft_ms": 11.774957994930446, "decode_ms": 11.769333039410412}, "total_wall_ms": 31903.39566709008, "ttft_ms": 11.774957994930446, "input_usage": {"lookups": 8}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 64, "facts_retained": 11, "retention_rate": 0.171875, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1111111111111111, "amount": 0.07692307692307693, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 609}, "ledger_by_hop": {"1": {"facts_total": 64, "facts_retained": 14, "retention_rate": 0.21875, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.2222222222222222, "amount": 0.15384615384615385, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 870}, "2": {"facts_total": 64, "facts_retained": 11, "retention_rate": 0.171875, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1111111111111111, "amount": 0.07692307692307693, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 609}, "3": {"facts_total": 64, "facts_retained": 11, "retention_rate": 0.171875, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1111111111111111, "amount": 0.07692307692307693, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 609}, "4": {"facts_total": 64, "facts_retained": 11, "retention_rate": 0.171875, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1111111111111111, "amount": 0.07692307692307693, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 609}, "5": {"facts_total": 64, "facts_retained": 11, "retention_rate": 0.171875, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1111111111111111, "amount": 0.07692307692307693, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 609}, "6": {"facts_total": 64, "facts_retained": 11, "retention_rate": 0.171875, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1111111111111111, "amount": 0.07692307692307693, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 609}, "7": {"facts_total": 64, "facts_retained": 11, "retention_rate": 0.171875, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1111111111111111, "amount": 0.07692307692307693, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 609}, "8": {"facts_total": 64, "facts_retained": 11, "retention_rate": 0.171875, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1111111111111111, "amount": 0.07692307692307693, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 609}}}, "reference_key_id": "keys/synthetic/syn_03018_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "830183240b7f45e537ebfeb7fe898082030a951407aec8fa2f3cd9cc13fe9d49", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "d6edc624743e1382cd3f7e4a", "attempt_id": "c58b5e2de92e", "run_group_id": "E4_core:A5|repairable|qwen3b:h8", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:14:00.359247Z", "task_family": "synthetic", "document_id": "syn_03018_2000", "document_hash": "d284eb07e27da742b9c2b67abccfdf72e4dd67c3259f44f3b41e34dd3e6fc1ff", "question_id": "q11_aggregation", "split": "test", "cluster_id": "syn_03018_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 8, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1982, "query_tokens": 366, "message_tokens": 273, "message_utf8_bytes": 657, "cache_tensor_bytes": null, "serialized_payload_bytes": 657, "side_information_bytes": 0, "fetched_bytes": 3899, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 31736.605958081782, "hops": {"1": 6923.268833081238, "2": 3655.3692090092227, "3": 3530.2908330922946, "4": 3531.2436670064926, "5": 3533.83945801761, "6": 3521.261749905534, "7": 3521.8319579726085, "8": 3519.5002499967813}, "receiver_prefill_ms": 142.41920900531113, "ttft_ms": 11.818083003163338, "decode_ms": 82.30450004339218}, "total_wall_ms": 31973.456083098426, "ttft_ms": 11.818083003163338, "input_usage": {"lookups": 8}, "output_usage": {"output_tokens": 7}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "2917061", "evidence_artifact": {"ledger": {"facts_total": 64, "facts_retained": 11, "retention_rate": 0.171875, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1111111111111111, "amount": 0.07692307692307693, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 609}, "ledger_by_hop": {"1": {"facts_total": 64, "facts_retained": 14, "retention_rate": 0.21875, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.2222222222222222, "amount": 0.15384615384615385, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 870}, "2": {"facts_total": 64, "facts_retained": 11, "retention_rate": 0.171875, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1111111111111111, "amount": 0.07692307692307693, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 609}, "3": {"facts_total": 64, "facts_retained": 11, "retention_rate": 0.171875, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1111111111111111, "amount": 0.07692307692307693, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 609}, "4": {"facts_total": 64, "facts_retained": 11, "retention_rate": 0.171875, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1111111111111111, "amount": 0.07692307692307693, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 609}, "5": {"facts_total": 64, "facts_retained": 11, "retention_rate": 0.171875, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1111111111111111, "amount": 0.07692307692307693, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 609}, "6": {"facts_total": 64, "facts_retained": 11, "retention_rate": 0.171875, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1111111111111111, "amount": 0.07692307692307693, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 609}, "7": {"facts_total": 64, "facts_retained": 11, "retention_rate": 0.171875, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1111111111111111, "amount": 0.07692307692307693, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 609}, "8": {"facts_total": 64, "facts_retained": 11, "retention_rate": 0.171875, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1111111111111111, "amount": 0.07692307692307693, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 609}}}, "reference_key_id": "keys/synthetic/syn_03018_2000.json", "task_scores": {"correct": 0.0, "abstained": 0.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "830183240b7f45e537ebfeb7fe898082030a951407aec8fa2f3cd9cc13fe9d49", "output_hash": "129c25eb43beebf0b464953aaabebfe914d11fa686f1dce105e62022958e1719"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "d27fd5daf9c17809005ba012", "attempt_id": "53afc507f522", "run_group_id": "E4_core:A5|repairable|qwen3b:h8", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:14:00.610695Z", "task_family": "synthetic", "document_id": "syn_03018_2000", "document_hash": "d284eb07e27da742b9c2b67abccfdf72e4dd67c3259f44f3b41e34dd3e6fc1ff", "question_id": "q12_unit_conversion", "split": "test", "cluster_id": "syn_03018_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 8, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1982, "query_tokens": 363, "message_tokens": 273, "message_utf8_bytes": 657, "cache_tensor_bytes": null, "serialized_payload_bytes": 657, "side_information_bytes": 0, "fetched_bytes": 3899, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 31736.605958081782, "hops": {"1": 6923.268833081238, "2": 3655.3692090092227, "3": 3530.2908330922946, "4": 3531.2436670064926, "5": 3533.83945801761, "6": 3521.261749905534, "7": 3521.8319579726085, "8": 3519.5002499967813}, "receiver_prefill_ms": 142.09987502545118, "ttft_ms": 11.78120810072869, "decode_ms": 46.96620802860707}, "total_wall_ms": 31937.772208126262, "ttft_ms": 11.78120810072869, "input_usage": {"lookups": 8}, "output_usage": {"output_tokens": 4}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "6400", "evidence_artifact": {"ledger": {"facts_total": 64, "facts_retained": 11, "retention_rate": 0.171875, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1111111111111111, "amount": 0.07692307692307693, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 609}, "ledger_by_hop": {"1": {"facts_total": 64, "facts_retained": 14, "retention_rate": 0.21875, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.2222222222222222, "amount": 0.15384615384615385, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 870}, "2": {"facts_total": 64, "facts_retained": 11, "retention_rate": 0.171875, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1111111111111111, "amount": 0.07692307692307693, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 609}, "3": {"facts_total": 64, "facts_retained": 11, "retention_rate": 0.171875, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1111111111111111, "amount": 0.07692307692307693, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 609}, "4": {"facts_total": 64, "facts_retained": 11, "retention_rate": 0.171875, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1111111111111111, "amount": 0.07692307692307693, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 609}, "5": {"facts_total": 64, "facts_retained": 11, "retention_rate": 0.171875, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1111111111111111, "amount": 0.07692307692307693, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 609}, "6": {"facts_total": 64, "facts_retained": 11, "retention_rate": 0.171875, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1111111111111111, "amount": 0.07692307692307693, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 609}, "7": {"facts_total": 64, "facts_retained": 11, "retention_rate": 0.171875, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1111111111111111, "amount": 0.07692307692307693, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 609}, "8": {"facts_total": 64, "facts_retained": 11, "retention_rate": 0.171875, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1111111111111111, "amount": 0.07692307692307693, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 609}}}, "reference_key_id": "keys/synthetic/syn_03018_2000.json", "task_scores": {"correct": 0.0, "abstained": 0.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "830183240b7f45e537ebfeb7fe898082030a951407aec8fa2f3cd9cc13fe9d49", "output_hash": "a7dccef9ce1ae31c49a4cfc484f8411dd52b11c914c1fe3d781633642d1e1327"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "1836f59a068c3788780e4d02", "attempt_id": "bab783abf4cb", "run_group_id": "E4_core:A5|repairable|qwen3b:h8", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:14:00.826170Z", "task_family": "synthetic", "document_id": "syn_03018_2000", "document_hash": "d284eb07e27da742b9c2b67abccfdf72e4dd67c3259f44f3b41e34dd3e6fc1ff", "question_id": "q13_unanswerable", "split": "test", "cluster_id": "syn_03018_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 8, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1982, "query_tokens": 367, "message_tokens": 273, "message_utf8_bytes": 657, "cache_tensor_bytes": null, "serialized_payload_bytes": 657, "side_information_bytes": 0, "fetched_bytes": 3899, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 31736.605958081782, "hops": {"1": 6923.268833081238, "2": 3655.3692090092227, "3": 3530.2908330922946, "4": 3531.2436670064926, "5": 3533.83945801761, "6": 3521.261749905534, "7": 3521.8319579726085, "8": 3519.5002499967813}, "receiver_prefill_ms": 142.47287495527416, "ttft_ms": 11.810875032097101, "decode_ms": 11.724666925147176}, "total_wall_ms": 31902.91479206644, "ttft_ms": 11.810875032097101, "input_usage": {"lookups": 8}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 64, "facts_retained": 11, "retention_rate": 0.171875, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1111111111111111, "amount": 0.07692307692307693, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 609}, "ledger_by_hop": {"1": {"facts_total": 64, "facts_retained": 14, "retention_rate": 0.21875, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.2222222222222222, "amount": 0.15384615384615385, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 870}, "2": {"facts_total": 64, "facts_retained": 11, "retention_rate": 0.171875, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1111111111111111, "amount": 0.07692307692307693, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 609}, "3": {"facts_total": 64, "facts_retained": 11, "retention_rate": 0.171875, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1111111111111111, "amount": 0.07692307692307693, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 609}, "4": {"facts_total": 64, "facts_retained": 11, "retention_rate": 0.171875, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1111111111111111, "amount": 0.07692307692307693, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 609}, "5": {"facts_total": 64, "facts_retained": 11, "retention_rate": 0.171875, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1111111111111111, "amount": 0.07692307692307693, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 609}, "6": {"facts_total": 64, "facts_retained": 11, "retention_rate": 0.171875, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1111111111111111, "amount": 0.07692307692307693, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 609}, "7": {"facts_total": 64, "facts_retained": 11, "retention_rate": 0.171875, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1111111111111111, "amount": 0.07692307692307693, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 609}, "8": {"facts_total": 64, "facts_retained": 11, "retention_rate": 0.171875, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1111111111111111, "amount": 0.07692307692307693, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 609}}}, "reference_key_id": "keys/synthetic/syn_03018_2000.json", "task_scores": {"correct": 1.0, "abstained": 1.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "830183240b7f45e537ebfeb7fe898082030a951407aec8fa2f3cd9cc13fe9d49", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "dea73f49b92aab0706b15908", "attempt_id": "5d16f119e8e4", "run_group_id": "E4_core:A5|repairable|qwen3b:h8", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:14:01.007671Z", "task_family": "synthetic", "document_id": "syn_03018_2000", "document_hash": "d284eb07e27da742b9c2b67abccfdf72e4dd67c3259f44f3b41e34dd3e6fc1ff", "question_id": "q14_unanswerable", "split": "test", "cluster_id": "syn_03018_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 8, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1982, "query_tokens": 368, "message_tokens": 273, "message_utf8_bytes": 657, "cache_tensor_bytes": null, "serialized_payload_bytes": 657, "side_information_bytes": 0, "fetched_bytes": 3899, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 31736.605958081782, "hops": {"1": 6923.268833081238, "2": 3655.3692090092227, "3": 3530.2908330922946, "4": 3531.2436670064926, "5": 3533.83945801761, "6": 3521.261749905534, "7": 3521.8319579726085, "8": 3519.5002499967813}, "receiver_prefill_ms": 142.0437089400366, "ttft_ms": 11.899166973307729, "decode_ms": 11.872124974615872}, "total_wall_ms": 31902.933040983044, "ttft_ms": 11.899166973307729, "input_usage": {"lookups": 8}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 64, "facts_retained": 11, "retention_rate": 0.171875, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1111111111111111, "amount": 0.07692307692307693, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 609}, "ledger_by_hop": {"1": {"facts_total": 64, "facts_retained": 14, "retention_rate": 0.21875, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.2222222222222222, "amount": 0.15384615384615385, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 870}, "2": {"facts_total": 64, "facts_retained": 11, "retention_rate": 0.171875, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1111111111111111, "amount": 0.07692307692307693, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 609}, "3": {"facts_total": 64, "facts_retained": 11, "retention_rate": 0.171875, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1111111111111111, "amount": 0.07692307692307693, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 609}, "4": {"facts_total": 64, "facts_retained": 11, "retention_rate": 0.171875, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1111111111111111, "amount": 0.07692307692307693, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 609}, "5": {"facts_total": 64, "facts_retained": 11, "retention_rate": 0.171875, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1111111111111111, "amount": 0.07692307692307693, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 609}, "6": {"facts_total": 64, "facts_retained": 11, "retention_rate": 0.171875, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1111111111111111, "amount": 0.07692307692307693, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 609}, "7": {"facts_total": 64, "facts_retained": 11, "retention_rate": 0.171875, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1111111111111111, "amount": 0.07692307692307693, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 609}, "8": {"facts_total": 64, "facts_retained": 11, "retention_rate": 0.171875, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1111111111111111, "amount": 0.07692307692307693, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 609}}}, "reference_key_id": "keys/synthetic/syn_03018_2000.json", "task_scores": {"correct": 1.0, "abstained": 1.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "830183240b7f45e537ebfeb7fe898082030a951407aec8fa2f3cd9cc13fe9d49", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "9cebb5471822871625ddbb0c", "attempt_id": "49079d175521", "run_group_id": "E4_core:A4|closed|qwen3b-llama3b:h1", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:14:08.068894Z", "task_family": "synthetic", "document_id": "syn_03018_2000", "document_hash": "d284eb07e27da742b9c2b67abccfdf72e4dd67c3259f44f3b41e34dd3e6fc1ff", "question_id": "q00_single_fact", "split": "test", "cluster_id": "syn_03018_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 1, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1982, "query_tokens": 595, "message_tokens": 512, "message_utf8_bytes": 1181, "cache_tensor_bytes": null, "serialized_payload_bytes": 1181, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 6878.585290978663, "hops": {"1": 6878.585290978663}, "receiver_prefill_ms": 227.10599994752556, "ttft_ms": 11.884500039741397, "decode_ms": 11.964875040575862}, "total_wall_ms": 7130.064790952019, "ttft_ms": 11.884500039741397, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 64, "facts_retained": 22, "retention_rate": 0.34375, "retention_by_kind": {"person": 0.5, "date": 0.5555555555555556, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1181}, "ledger_by_hop": {"1": {"facts_total": 64, "facts_retained": 22, "retention_rate": 0.34375, "retention_by_kind": {"person": 0.5, "date": 0.5555555555555556, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1181}}}, "reference_key_id": "keys/synthetic/syn_03018_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "499392c3597fdf0fd842c7907d9598f7d57a2cf7b9c4348cfba91a80adac26ec", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "ebd18f2d15b4603b85fa8658", "attempt_id": "719831221201", "run_group_id": "E4_core:A4|closed|qwen3b-llama3b:h1", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:14:08.336025Z", "task_family": "synthetic", "document_id": "syn_03018_2000", "document_hash": "d284eb07e27da742b9c2b67abccfdf72e4dd67c3259f44f3b41e34dd3e6fc1ff", "question_id": "q01_single_fact", "split": "test", "cluster_id": "syn_03018_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 1, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1982, "query_tokens": 601, "message_tokens": 512, "message_utf8_bytes": 1181, "cache_tensor_bytes": null, "serialized_payload_bytes": 1181, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 6878.585290978663, "hops": {"1": 6878.585290978663}, "receiver_prefill_ms": 227.2120409179479, "ttft_ms": 11.897666961885989, "decode_ms": 142.8483750205487}, "total_wall_ms": 7261.04412402492, "ttft_ms": 11.897666961885989, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 12}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "Lot 376-E, Baie-Comeau", "evidence_artifact": {"ledger": {"facts_total": 64, "facts_retained": 22, "retention_rate": 0.34375, "retention_by_kind": {"person": 0.5, "date": 0.5555555555555556, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1181}, "ledger_by_hop": {"1": {"facts_total": 64, "facts_retained": 22, "retention_rate": 0.34375, "retention_by_kind": {"person": 0.5, "date": 0.5555555555555556, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1181}}}, "reference_key_id": "keys/synthetic/syn_03018_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 0.8571428571428571}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "499392c3597fdf0fd842c7907d9598f7d57a2cf7b9c4348cfba91a80adac26ec", "output_hash": "b58829a1440993c4af3307ba44494f021d726fd11e9337888b6b852a8b37351a"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "970d7ac3fb6694a4081acfb4", "attempt_id": "8eca271e2095", "run_group_id": "E4_core:A4|closed|qwen3b-llama3b:h1", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:14:08.733155Z", "task_family": "synthetic", "document_id": "syn_03018_2000", "document_hash": "d284eb07e27da742b9c2b67abccfdf72e4dd67c3259f44f3b41e34dd3e6fc1ff", "question_id": "q02_exact_copy", "split": "test", "cluster_id": "syn_03018_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 1, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1982, "query_tokens": 600, "message_tokens": 512, "message_utf8_bytes": 1181, "cache_tensor_bytes": null, "serialized_payload_bytes": 1181, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 6878.585290978663, "hops": {"1": 6878.585290978663}, "receiver_prefill_ms": 227.1866250084713, "ttft_ms": 11.87649997882545, "decode_ms": 12.036541011184454}, "total_wall_ms": 7130.159249994904, "ttft_ms": 11.87649997882545, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 64, "facts_retained": 22, "retention_rate": 0.34375, "retention_by_kind": {"person": 0.5, "date": 0.5555555555555556, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1181}, "ledger_by_hop": {"1": {"facts_total": 64, "facts_retained": 22, "retention_rate": 0.34375, "retention_by_kind": {"person": 0.5, "date": 0.5555555555555556, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1181}}}, "reference_key_id": "keys/synthetic/syn_03018_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "499392c3597fdf0fd842c7907d9598f7d57a2cf7b9c4348cfba91a80adac26ec", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "e462d6b07ae4289f29712ca6", "attempt_id": "9486fec385f5", "run_group_id": "E4_core:A4|closed|qwen3b-llama3b:h1", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:14:08.999555Z", "task_family": "synthetic", "document_id": "syn_03018_2000", "document_hash": "d284eb07e27da742b9c2b67abccfdf72e4dd67c3259f44f3b41e34dd3e6fc1ff", "question_id": "q03_exact_copy", "split": "test", "cluster_id": "syn_03018_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 1, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1982, "query_tokens": 598, "message_tokens": 512, "message_utf8_bytes": 1181, "cache_tensor_bytes": null, "serialized_payload_bytes": 1181, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 6878.585290978663, "hops": {"1": 6878.585290978663}, "receiver_prefill_ms": 227.43145807180554, "ttft_ms": 11.880832957103848, "decode_ms": 107.05741599667817}, "total_wall_ms": 7225.438290974125, "ttft_ms": 11.880832957103848, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 9}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "Priyamvada Sørlie", "evidence_artifact": {"ledger": {"facts_total": 64, "facts_retained": 22, "retention_rate": 0.34375, "retention_by_kind": {"person": 0.5, "date": 0.5555555555555556, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1181}, "ledger_by_hop": {"1": {"facts_total": 64, "facts_retained": 22, "retention_rate": 0.34375, "retention_by_kind": {"person": 0.5, "date": 0.5555555555555556, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1181}}}, "reference_key_id": "keys/synthetic/syn_03018_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "499392c3597fdf0fd842c7907d9598f7d57a2cf7b9c4348cfba91a80adac26ec", "output_hash": "c5f554196ed23dd71ad995ec89f7142099d524dbb6534cc10d94a0c89d13e792"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "ea1dfed708a233864a44e10d", "attempt_id": "53365f72b9fd", "run_group_id": "E4_core:A4|closed|qwen3b-llama3b:h1", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:14:09.360904Z", "task_family": "synthetic", "document_id": "syn_03018_2000", "document_hash": "d284eb07e27da742b9c2b67abccfdf72e4dd67c3259f44f3b41e34dd3e6fc1ff", "question_id": "q04_multi_hop", "split": "test", "cluster_id": "syn_03018_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 1, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1982, "query_tokens": 604, "message_tokens": 512, "message_utf8_bytes": 1181, "cache_tensor_bytes": null, "serialized_payload_bytes": 1181, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 6878.585290978663, "hops": {"1": 6878.585290978663}, "receiver_prefill_ms": 227.15579194482416, "ttft_ms": 11.922750039957464, "decode_ms": 11.973584070801735}, "total_wall_ms": 7130.102291004732, "ttft_ms": 11.922750039957464, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 64, "facts_retained": 22, "retention_rate": 0.34375, "retention_by_kind": {"person": 0.5, "date": 0.5555555555555556, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1181}, "ledger_by_hop": {"1": {"facts_total": 64, "facts_retained": 22, "retention_rate": 0.34375, "retention_by_kind": {"person": 0.5, "date": 0.5555555555555556, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1181}}}, "reference_key_id": "keys/synthetic/syn_03018_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "499392c3597fdf0fd842c7907d9598f7d57a2cf7b9c4348cfba91a80adac26ec", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "3fa47848f00057250fbf7034", "attempt_id": "b093c80a4543", "run_group_id": "E4_core:A4|closed|qwen3b-llama3b:h1", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:14:09.626892Z", "task_family": "synthetic", "document_id": "syn_03018_2000", "document_hash": "d284eb07e27da742b9c2b67abccfdf72e4dd67c3259f44f3b41e34dd3e6fc1ff", "question_id": "q05_multi_hop", "split": "test", "cluster_id": "syn_03018_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 1, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1982, "query_tokens": 604, "message_tokens": 512, "message_utf8_bytes": 1181, "cache_tensor_bytes": null, "serialized_payload_bytes": 1181, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 6878.585290978663, "hops": {"1": 6878.585290978663}, "receiver_prefill_ms": 227.28470899164677, "ttft_ms": 11.926208040677011, "decode_ms": 118.76800004392862}, "total_wall_ms": 7237.039750092663, "ttft_ms": 11.926208040677011, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 10}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "HM-38K-5533", "evidence_artifact": {"ledger": {"facts_total": 64, "facts_retained": 22, "retention_rate": 0.34375, "retention_by_kind": {"person": 0.5, "date": 0.5555555555555556, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1181}, "ledger_by_hop": {"1": {"facts_total": 64, "facts_retained": 22, "retention_rate": 0.34375, "retention_by_kind": {"person": 0.5, "date": 0.5555555555555556, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1181}}}, "reference_key_id": "keys/synthetic/syn_03018_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "499392c3597fdf0fd842c7907d9598f7d57a2cf7b9c4348cfba91a80adac26ec", "output_hash": "18acd9807857243bbded3bceae55398316a3958b81d70364441bdb6748eaf878"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "8e09e1c06f8e8b18c1cb8290", "attempt_id": "475cac8144ff", "run_group_id": "E4_core:A4|closed|qwen3b-llama3b:h1", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:14:10.000394Z", "task_family": "synthetic", "document_id": "syn_03018_2000", "document_hash": "d284eb07e27da742b9c2b67abccfdf72e4dd67c3259f44f3b41e34dd3e6fc1ff", "question_id": "q06_negation", "split": "test", "cluster_id": "syn_03018_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 1, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1982, "query_tokens": 596, "message_tokens": 512, "message_utf8_bytes": 1181, "cache_tensor_bytes": null, "serialized_payload_bytes": 1181, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 6878.585290978663, "hops": {"1": 6878.585290978663}, "receiver_prefill_ms": 227.07691590767354, "ttft_ms": 11.91891694907099, "decode_ms": 12.13029189966619}, "total_wall_ms": 7130.188832990825, "ttft_ms": 11.91891694907099, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 64, "facts_retained": 22, "retention_rate": 0.34375, "retention_by_kind": {"person": 0.5, "date": 0.5555555555555556, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1181}, "ledger_by_hop": {"1": {"facts_total": 64, "facts_retained": 22, "retention_rate": 0.34375, "retention_by_kind": {"person": 0.5, "date": 0.5555555555555556, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1181}}}, "reference_key_id": "keys/synthetic/syn_03018_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "499392c3597fdf0fd842c7907d9598f7d57a2cf7b9c4348cfba91a80adac26ec", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "470cc53f14e6539897136211", "attempt_id": "f06d1b33fdce", "run_group_id": "E4_core:A4|closed|qwen3b-llama3b:h1", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:14:10.266871Z", "task_family": "synthetic", "document_id": "syn_03018_2000", "document_hash": "d284eb07e27da742b9c2b67abccfdf72e4dd67c3259f44f3b41e34dd3e6fc1ff", "question_id": "q07_negation", "split": "test", "cluster_id": "syn_03018_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 1, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1982, "query_tokens": 595, "message_tokens": 512, "message_utf8_bytes": 1181, "cache_tensor_bytes": null, "serialized_payload_bytes": 1181, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 6878.585290978663, "hops": {"1": 6878.585290978663}, "receiver_prefill_ms": 227.19424997922033, "ttft_ms": 11.87495794147253, "decode_ms": 12.092291959561408}, "total_wall_ms": 7130.221041035838, "ttft_ms": 11.87495794147253, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "no", "evidence_artifact": {"ledger": {"facts_total": 64, "facts_retained": 22, "retention_rate": 0.34375, "retention_by_kind": {"person": 0.5, "date": 0.5555555555555556, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1181}, "ledger_by_hop": {"1": {"facts_total": 64, "facts_retained": 22, "retention_rate": 0.34375, "retention_by_kind": {"person": 0.5, "date": 0.5555555555555556, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1181}}}, "reference_key_id": "keys/synthetic/syn_03018_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "499392c3597fdf0fd842c7907d9598f7d57a2cf7b9c4348cfba91a80adac26ec", "output_hash": "9390298f3fb0c5b160498935d79cb139aef28e1c47358b4bbba61862b9c26e59"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "0cbf2d01e6df95dca7663fa2", "attempt_id": "c4fd3d43b6d3", "run_group_id": "E4_core:A4|closed|qwen3b-llama3b:h1", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:14:10.533331Z", "task_family": "synthetic", "document_id": "syn_03018_2000", "document_hash": "d284eb07e27da742b9c2b67abccfdf72e4dd67c3259f44f3b41e34dd3e6fc1ff", "question_id": "q08_revision", "split": "test", "cluster_id": "syn_03018_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 1, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1982, "query_tokens": 613, "message_tokens": 512, "message_utf8_bytes": 1181, "cache_tensor_bytes": null, "serialized_payload_bytes": 1181, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 6878.585290978663, "hops": {"1": 6878.585290978663}, "receiver_prefill_ms": 229.33595802169293, "ttft_ms": 11.959917028434575, "decode_ms": 71.2857919279486}, "total_wall_ms": 7191.668874002062, "ttft_ms": 11.959917028434575, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 6}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "304815", "evidence_artifact": {"ledger": {"facts_total": 64, "facts_retained": 22, "retention_rate": 0.34375, "retention_by_kind": {"person": 0.5, "date": 0.5555555555555556, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1181}, "ledger_by_hop": {"1": {"facts_total": 64, "facts_retained": 22, "retention_rate": 0.34375, "retention_by_kind": {"person": 0.5, "date": 0.5555555555555556, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1181}}}, "reference_key_id": "keys/synthetic/syn_03018_2000.json", "task_scores": {"correct": 0.0, "abstained": 0.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "499392c3597fdf0fd842c7907d9598f7d57a2cf7b9c4348cfba91a80adac26ec", "output_hash": "a480475d6855141c920f42d1bf3337ca7fd88bcf4087b0ed2e253ed68cca6926"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "8c6501541b966201378fbc92", "attempt_id": "ee28277ab24c", "run_group_id": "E4_core:A4|closed|qwen3b-llama3b:h1", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:14:10.861542Z", "task_family": "synthetic", "document_id": "syn_03018_2000", "document_hash": "d284eb07e27da742b9c2b67abccfdf72e4dd67c3259f44f3b41e34dd3e6fc1ff", "question_id": "q09_revision", "split": "test", "cluster_id": "syn_03018_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 1, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1982, "query_tokens": 606, "message_tokens": 512, "message_utf8_bytes": 1181, "cache_tensor_bytes": null, "serialized_payload_bytes": 1181, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 6878.585290978663, "hops": {"1": 6878.585290978663}, "receiver_prefill_ms": 227.41229203529656, "ttft_ms": 11.927375104278326, "decode_ms": 71.23250002041459}, "total_wall_ms": 7189.655291032977, "ttft_ms": 11.927375104278326, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 6}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "276340", "evidence_artifact": {"ledger": {"facts_total": 64, "facts_retained": 22, "retention_rate": 0.34375, "retention_by_kind": {"person": 0.5, "date": 0.5555555555555556, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1181}, "ledger_by_hop": {"1": {"facts_total": 64, "facts_retained": 22, "retention_rate": 0.34375, "retention_by_kind": {"person": 0.5, "date": 0.5555555555555556, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1181}}}, "reference_key_id": "keys/synthetic/syn_03018_2000.json", "task_scores": {"correct": 0.0, "abstained": 0.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "499392c3597fdf0fd842c7907d9598f7d57a2cf7b9c4348cfba91a80adac26ec", "output_hash": "bc7c3667cbe0ad74ec0a1b092d3bad7e6a02978fa67746e3adbf3174ed34b3a2"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "8034ef00f8286a6135790952", "attempt_id": "1e32229a2994", "run_group_id": "E4_core:A4|closed|qwen3b-llama3b:h1", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:14:11.187257Z", "task_family": "synthetic", "document_id": "syn_03018_2000", "document_hash": "d284eb07e27da742b9c2b67abccfdf72e4dd67c3259f44f3b41e34dd3e6fc1ff", "question_id": "q10_temporal", "split": "test", "cluster_id": "syn_03018_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 1, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1982, "query_tokens": 614, "message_tokens": 512, "message_utf8_bytes": 1181, "cache_tensor_bytes": null, "serialized_payload_bytes": 1181, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 6878.585290978663, "hops": {"1": 6878.585290978663}, "receiver_prefill_ms": 229.51350000221282, "ttft_ms": 12.181458994746208, "decode_ms": 71.28554210066795}, "total_wall_ms": 7192.0656660804525, "ttft_ms": 12.181458994746208, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 6}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "Melymi-15", "evidence_artifact": {"ledger": {"facts_total": 64, "facts_retained": 22, "retention_rate": 0.34375, "retention_by_kind": {"person": 0.5, "date": 0.5555555555555556, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1181}, "ledger_by_hop": {"1": {"facts_total": 64, "facts_retained": 22, "retention_rate": 0.34375, "retention_by_kind": {"person": 0.5, "date": 0.5555555555555556, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1181}}}, "reference_key_id": "keys/synthetic/syn_03018_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "499392c3597fdf0fd842c7907d9598f7d57a2cf7b9c4348cfba91a80adac26ec", "output_hash": "fa08d1474b01d6dfbf3dea049be251aa3c3021c17b4818258064b156509af58b"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "a0961ec93ec071fe822660ac", "attempt_id": "c601651d843a", "run_group_id": "E4_core:A4|closed|qwen3b-llama3b:h1", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:14:11.515603Z", "task_family": "synthetic", "document_id": "syn_03018_2000", "document_hash": "d284eb07e27da742b9c2b67abccfdf72e4dd67c3259f44f3b41e34dd3e6fc1ff", "question_id": "q11_aggregation", "split": "test", "cluster_id": "syn_03018_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 1, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1982, "query_tokens": 606, "message_tokens": 512, "message_utf8_bytes": 1181, "cache_tensor_bytes": null, "serialized_payload_bytes": 1181, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 6878.585290978663, "hops": {"1": 6878.585290978663}, "receiver_prefill_ms": 226.76779108587652, "ttft_ms": 11.934167006984353, "decode_ms": 83.1693330546841}, "total_wall_ms": 7200.931748957373, "ttft_ms": 11.934167006984353, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 7}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "2414120", "evidence_artifact": {"ledger": {"facts_total": 64, "facts_retained": 22, "retention_rate": 0.34375, "retention_by_kind": {"person": 0.5, "date": 0.5555555555555556, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1181}, "ledger_by_hop": {"1": {"facts_total": 64, "facts_retained": 22, "retention_rate": 0.34375, "retention_by_kind": {"person": 0.5, "date": 0.5555555555555556, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1181}}}, "reference_key_id": "keys/synthetic/syn_03018_2000.json", "task_scores": {"correct": 0.0, "abstained": 0.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "499392c3597fdf0fd842c7907d9598f7d57a2cf7b9c4348cfba91a80adac26ec", "output_hash": "0a242f83095c1c1c7e4a3cd4bb88035f03abb655b6cf03cf7962945bb269a5e8"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "18b0d55e8190483c289108eb", "attempt_id": "3534be6b0a55", "run_group_id": "E4_core:A4|closed|qwen3b-llama3b:h1", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:14:11.852698Z", "task_family": "synthetic", "document_id": "syn_03018_2000", "document_hash": "d284eb07e27da742b9c2b67abccfdf72e4dd67c3259f44f3b41e34dd3e6fc1ff", "question_id": "q12_unit_conversion", "split": "test", "cluster_id": "syn_03018_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 1, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1982, "query_tokens": 603, "message_tokens": 512, "message_utf8_bytes": 1181, "cache_tensor_bytes": null, "serialized_payload_bytes": 1181, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 6878.585290978663, "hops": {"1": 6878.585290978663}, "receiver_prefill_ms": 226.7034159740433, "ttft_ms": 12.222625082358718, "decode_ms": 12.121666921302676}, "total_wall_ms": 7130.341624026187, "ttft_ms": 12.222625082358718, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 64, "facts_retained": 22, "retention_rate": 0.34375, "retention_by_kind": {"person": 0.5, "date": 0.5555555555555556, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1181}, "ledger_by_hop": {"1": {"facts_total": 64, "facts_retained": 22, "retention_rate": 0.34375, "retention_by_kind": {"person": 0.5, "date": 0.5555555555555556, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1181}}}, "reference_key_id": "keys/synthetic/syn_03018_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "499392c3597fdf0fd842c7907d9598f7d57a2cf7b9c4348cfba91a80adac26ec", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "a87d9efdb8331ed399544d7a", "attempt_id": "54e599ba63be", "run_group_id": "E4_core:A4|closed|qwen3b-llama3b:h1", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:14:12.119748Z", "task_family": "synthetic", "document_id": "syn_03018_2000", "document_hash": "d284eb07e27da742b9c2b67abccfdf72e4dd67c3259f44f3b41e34dd3e6fc1ff", "question_id": "q13_unanswerable", "split": "test", "cluster_id": "syn_03018_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 1, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1982, "query_tokens": 607, "message_tokens": 512, "message_utf8_bytes": 1181, "cache_tensor_bytes": null, "serialized_payload_bytes": 1181, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 6878.585290978663, "hops": {"1": 6878.585290978663}, "receiver_prefill_ms": 226.85174993239343, "ttft_ms": 12.336625019088387, "decode_ms": 12.23054202273488}, "total_wall_ms": 7130.673832958564, "ttft_ms": 12.336625019088387, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 64, "facts_retained": 22, "retention_rate": 0.34375, "retention_by_kind": {"person": 0.5, "date": 0.5555555555555556, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1181}, "ledger_by_hop": {"1": {"facts_total": 64, "facts_retained": 22, "retention_rate": 0.34375, "retention_by_kind": {"person": 0.5, "date": 0.5555555555555556, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1181}}}, "reference_key_id": "keys/synthetic/syn_03018_2000.json", "task_scores": {"correct": 1.0, "abstained": 1.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "499392c3597fdf0fd842c7907d9598f7d57a2cf7b9c4348cfba91a80adac26ec", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "e02586213b12e9b5743be385", "attempt_id": "78775e921b4c", "run_group_id": "E4_core:A4|closed|qwen3b-llama3b:h1", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:14:12.388210Z", "task_family": "synthetic", "document_id": "syn_03018_2000", "document_hash": "d284eb07e27da742b9c2b67abccfdf72e4dd67c3259f44f3b41e34dd3e6fc1ff", "question_id": "q14_unanswerable", "split": "test", "cluster_id": "syn_03018_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 1, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1982, "query_tokens": 608, "message_tokens": 512, "message_utf8_bytes": 1181, "cache_tensor_bytes": null, "serialized_payload_bytes": 1181, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 6878.585290978663, "hops": {"1": 6878.585290978663}, "receiver_prefill_ms": 227.3249999852851, "ttft_ms": 12.214915943332016, "decode_ms": 12.076959013938904}, "total_wall_ms": 7130.837582983077, "ttft_ms": 12.214915943332016, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 64, "facts_retained": 22, "retention_rate": 0.34375, "retention_by_kind": {"person": 0.5, "date": 0.5555555555555556, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1181}, "ledger_by_hop": {"1": {"facts_total": 64, "facts_retained": 22, "retention_rate": 0.34375, "retention_by_kind": {"person": 0.5, "date": 0.5555555555555556, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1181}}}, "reference_key_id": "keys/synthetic/syn_03018_2000.json", "task_scores": {"correct": 1.0, "abstained": 1.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "499392c3597fdf0fd842c7907d9598f7d57a2cf7b9c4348cfba91a80adac26ec", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "1e2494ac015c682bc3fe88e0", "attempt_id": "15ed5aacdf4e", "run_group_id": "E4_core:A4|closed|qwen3b-llama3b:h2", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:14:17.717022Z", "task_family": "synthetic", "document_id": "syn_03018_2000", "document_hash": "d284eb07e27da742b9c2b67abccfdf72e4dd67c3259f44f3b41e34dd3e6fc1ff", "question_id": "q00_single_fact", "split": "test", "cluster_id": "syn_03018_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "tokenizer_hashes": {"receiver": "b9d8872cf3a99da99e1fc4b0f71232a451632dc49ca4c6771549b4efa5c1c763"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 2, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1773, "query_tokens": 520, "message_tokens": 416, "message_utf8_bytes": 1085, "cache_tensor_bytes": null, "serialized_payload_bytes": 1085, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 11937.892624991946, "hops": {"1": 6878.585290978663, "2": 5059.307334013283}, "receiver_prefill_ms": 209.73079197574407, "ttft_ms": 11.672249995172024, "decode_ms": 11.577249970287085}, "total_wall_ms": 12171.442999970168, "ttft_ms": 11.672249995172024, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 64, "facts_retained": 22, "retention_rate": 0.34375, "retention_by_kind": {"person": 0.5, "date": 0.5555555555555556, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1085}, "ledger_by_hop": {"1": {"facts_total": 64, "facts_retained": 22, "retention_rate": 0.34375, "retention_by_kind": {"person": 0.5, "date": 0.5555555555555556, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1181}, "2": {"facts_total": 64, "facts_retained": 22, "retention_rate": 0.34375, "retention_by_kind": {"person": 0.5, "date": 0.5555555555555556, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1085}}}, "reference_key_id": "keys/synthetic/syn_03018_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "07068ac4be74eeb8732a0c40afecb5d8f0a37edd6a1196705e2f367d41b90792", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "57955b81e70a5f5d3c9cf4ae", "attempt_id": "1be8708926e4", "run_group_id": "E4_core:A4|closed|qwen3b-llama3b:h2", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:14:17.966234Z", "task_family": "synthetic", "document_id": "syn_03018_2000", "document_hash": "d284eb07e27da742b9c2b67abccfdf72e4dd67c3259f44f3b41e34dd3e6fc1ff", "question_id": "q01_single_fact", "split": "test", "cluster_id": "syn_03018_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "tokenizer_hashes": {"receiver": "b9d8872cf3a99da99e1fc4b0f71232a451632dc49ca4c6771549b4efa5c1c763"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 2, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1773, "query_tokens": 526, "message_tokens": 416, "message_utf8_bytes": 1085, "cache_tensor_bytes": null, "serialized_payload_bytes": 1085, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 11937.892624991946, "hops": {"1": 6878.585290978663, "2": 5059.307334013283}, "receiver_prefill_ms": 210.23587509989738, "ttft_ms": 11.595832998864353, "decode_ms": 126.70720892492682}, "total_wall_ms": 12286.897083045915, "ttft_ms": 11.595832998864353, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 11}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "Lot 376-E, Baie-Comeau.", "evidence_artifact": {"ledger": {"facts_total": 64, "facts_retained": 22, "retention_rate": 0.34375, "retention_by_kind": {"person": 0.5, "date": 0.5555555555555556, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1085}, "ledger_by_hop": {"1": {"facts_total": 64, "facts_retained": 22, "retention_rate": 0.34375, "retention_by_kind": {"person": 0.5, "date": 0.5555555555555556, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1181}, "2": {"facts_total": 64, "facts_retained": 22, "retention_rate": 0.34375, "retention_by_kind": {"person": 0.5, "date": 0.5555555555555556, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1085}}}, "reference_key_id": "keys/synthetic/syn_03018_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 0.8571428571428571}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "07068ac4be74eeb8732a0c40afecb5d8f0a37edd6a1196705e2f367d41b90792", "output_hash": "8884a56b4b1757cacee95b0917a46f02a41042e2887e8c0ed24b52ccb81335ba"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "b027369a62203c7f65292d3b", "attempt_id": "005cb2d7255a", "run_group_id": "E4_core:A4|closed|qwen3b-llama3b:h2", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:14:18.330680Z", "task_family": "synthetic", "document_id": "syn_03018_2000", "document_hash": "d284eb07e27da742b9c2b67abccfdf72e4dd67c3259f44f3b41e34dd3e6fc1ff", "question_id": "q02_exact_copy", "split": "test", "cluster_id": "syn_03018_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "tokenizer_hashes": {"receiver": "b9d8872cf3a99da99e1fc4b0f71232a451632dc49ca4c6771549b4efa5c1c763"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 2, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1773, "query_tokens": 525, "message_tokens": 416, "message_utf8_bytes": 1085, "cache_tensor_bytes": null, "serialized_payload_bytes": 1085, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 11937.892624991946, "hops": {"1": 6878.585290978663, "2": 5059.307334013283}, "receiver_prefill_ms": 210.29791689943522, "ttft_ms": 11.535624973475933, "decode_ms": 11.49912504479289}, "total_wall_ms": 12171.627875068225, "ttft_ms": 11.535624973475933, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 64, "facts_retained": 22, "retention_rate": 0.34375, "retention_by_kind": {"person": 0.5, "date": 0.5555555555555556, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1085}, "ledger_by_hop": {"1": {"facts_total": 64, "facts_retained": 22, "retention_rate": 0.34375, "retention_by_kind": {"person": 0.5, "date": 0.5555555555555556, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1181}, "2": {"facts_total": 64, "facts_retained": 22, "retention_rate": 0.34375, "retention_by_kind": {"person": 0.5, "date": 0.5555555555555556, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1085}}}, "reference_key_id": "keys/synthetic/syn_03018_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "07068ac4be74eeb8732a0c40afecb5d8f0a37edd6a1196705e2f367d41b90792", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "d7282007947b0ebe5d172cf1", "attempt_id": "b1f17757475e", "run_group_id": "E4_core:A4|closed|qwen3b-llama3b:h2", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:14:18.579087Z", "task_family": "synthetic", "document_id": "syn_03018_2000", "document_hash": "d284eb07e27da742b9c2b67abccfdf72e4dd67c3259f44f3b41e34dd3e6fc1ff", "question_id": "q03_exact_copy", "split": "test", "cluster_id": "syn_03018_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "tokenizer_hashes": {"receiver": "b9d8872cf3a99da99e1fc4b0f71232a451632dc49ca4c6771549b4efa5c1c763"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 2, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1773, "query_tokens": 522, "message_tokens": 416, "message_utf8_bytes": 1085, "cache_tensor_bytes": null, "serialized_payload_bytes": 1085, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 11937.892624991946, "hops": {"1": 6878.585290978663, "2": 5059.307334013283}, "receiver_prefill_ms": 209.9827079800889, "ttft_ms": 11.615250026807189, "decode_ms": 103.81341597530991}, "total_wall_ms": 12263.71312502306, "ttft_ms": 11.615250026807189, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 9}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "Priyamvada Sørlie", "evidence_artifact": {"ledger": {"facts_total": 64, "facts_retained": 22, "retention_rate": 0.34375, "retention_by_kind": {"person": 0.5, "date": 0.5555555555555556, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1085}, "ledger_by_hop": {"1": {"facts_total": 64, "facts_retained": 22, "retention_rate": 0.34375, "retention_by_kind": {"person": 0.5, "date": 0.5555555555555556, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1181}, "2": {"facts_total": 64, "facts_retained": 22, "retention_rate": 0.34375, "retention_by_kind": {"person": 0.5, "date": 0.5555555555555556, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1085}}}, "reference_key_id": "keys/synthetic/syn_03018_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "07068ac4be74eeb8732a0c40afecb5d8f0a37edd6a1196705e2f367d41b90792", "output_hash": "c5f554196ed23dd71ad995ec89f7142099d524dbb6534cc10d94a0c89d13e792"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "74cde5d55dbb3571df07108d", "attempt_id": "c15e2bb3fd81", "run_group_id": "E4_core:A4|closed|qwen3b-llama3b:h2", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:14:18.919210Z", "task_family": "synthetic", "document_id": "syn_03018_2000", "document_hash": "d284eb07e27da742b9c2b67abccfdf72e4dd67c3259f44f3b41e34dd3e6fc1ff", "question_id": "q04_multi_hop", "split": "test", "cluster_id": "syn_03018_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "tokenizer_hashes": {"receiver": "b9d8872cf3a99da99e1fc4b0f71232a451632dc49ca4c6771549b4efa5c1c763"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 2, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1773, "query_tokens": 530, "message_tokens": 416, "message_utf8_bytes": 1085, "cache_tensor_bytes": null, "serialized_payload_bytes": 1085, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 11937.892624991946, "hops": {"1": 6878.585290978663, "2": 5059.307334013283}, "receiver_prefill_ms": 210.23712505120784, "ttft_ms": 11.563915992155671, "decode_ms": 11.552250012755394}, "total_wall_ms": 12171.716499957256, "ttft_ms": 11.563915992155671, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 64, "facts_retained": 22, "retention_rate": 0.34375, "retention_by_kind": {"person": 0.5, "date": 0.5555555555555556, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1085}, "ledger_by_hop": {"1": {"facts_total": 64, "facts_retained": 22, "retention_rate": 0.34375, "retention_by_kind": {"person": 0.5, "date": 0.5555555555555556, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1181}, "2": {"facts_total": 64, "facts_retained": 22, "retention_rate": 0.34375, "retention_by_kind": {"person": 0.5, "date": 0.5555555555555556, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1085}}}, "reference_key_id": "keys/synthetic/syn_03018_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "07068ac4be74eeb8732a0c40afecb5d8f0a37edd6a1196705e2f367d41b90792", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "9db7526bfab39d81f43bc56e", "attempt_id": "94c3e6c69ba4", "run_group_id": "E4_core:A4|closed|qwen3b-llama3b:h2", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:14:19.167143Z", "task_family": "synthetic", "document_id": "syn_03018_2000", "document_hash": "d284eb07e27da742b9c2b67abccfdf72e4dd67c3259f44f3b41e34dd3e6fc1ff", "question_id": "q05_multi_hop", "split": "test", "cluster_id": "syn_03018_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "tokenizer_hashes": {"receiver": "b9d8872cf3a99da99e1fc4b0f71232a451632dc49ca4c6771549b4efa5c1c763"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 2, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1773, "query_tokens": 528, "message_tokens": 416, "message_utf8_bytes": 1085, "cache_tensor_bytes": null, "serialized_payload_bytes": 1085, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 11937.892624991946, "hops": {"1": 6878.585290978663, "2": 5059.307334013283}, "receiver_prefill_ms": 210.22883395198733, "ttft_ms": 11.602082988247275, "decode_ms": 80.74924990069121}, "total_wall_ms": 12240.897292038426, "ttft_ms": 11.602082988247275, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 7}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "HM-38K-5533", "evidence_artifact": {"ledger": {"facts_total": 64, "facts_retained": 22, "retention_rate": 0.34375, "retention_by_kind": {"person": 0.5, "date": 0.5555555555555556, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1085}, "ledger_by_hop": {"1": {"facts_total": 64, "facts_retained": 22, "retention_rate": 0.34375, "retention_by_kind": {"person": 0.5, "date": 0.5555555555555556, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1181}, "2": {"facts_total": 64, "facts_retained": 22, "retention_rate": 0.34375, "retention_by_kind": {"person": 0.5, "date": 0.5555555555555556, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1085}}}, "reference_key_id": "keys/synthetic/syn_03018_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "07068ac4be74eeb8732a0c40afecb5d8f0a37edd6a1196705e2f367d41b90792", "output_hash": "18acd9807857243bbded3bceae55398316a3958b81d70364441bdb6748eaf878"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "26ea12305deda1cff14b8a90", "attempt_id": "b2b161de9655", "run_group_id": "E4_core:A4|closed|qwen3b-llama3b:h2", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:14:19.484546Z", "task_family": "synthetic", "document_id": "syn_03018_2000", "document_hash": "d284eb07e27da742b9c2b67abccfdf72e4dd67c3259f44f3b41e34dd3e6fc1ff", "question_id": "q06_negation", "split": "test", "cluster_id": "syn_03018_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "tokenizer_hashes": {"receiver": "b9d8872cf3a99da99e1fc4b0f71232a451632dc49ca4c6771549b4efa5c1c763"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 2, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1773, "query_tokens": 521, "message_tokens": 416, "message_utf8_bytes": 1085, "cache_tensor_bytes": null, "serialized_payload_bytes": 1085, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 11937.892624991946, "hops": {"1": 6878.585290978663, "2": 5059.307334013283}, "receiver_prefill_ms": 209.8698749905452, "ttft_ms": 11.599791003391147, "decode_ms": 23.03295792080462}, "total_wall_ms": 12182.807874982245, "ttft_ms": 11.599791003391147, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 2}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "No.", "evidence_artifact": {"ledger": {"facts_total": 64, "facts_retained": 22, "retention_rate": 0.34375, "retention_by_kind": {"person": 0.5, "date": 0.5555555555555556, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1085}, "ledger_by_hop": {"1": {"facts_total": 64, "facts_retained": 22, "retention_rate": 0.34375, "retention_by_kind": {"person": 0.5, "date": 0.5555555555555556, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1181}, "2": {"facts_total": 64, "facts_retained": 22, "retention_rate": 0.34375, "retention_by_kind": {"person": 0.5, "date": 0.5555555555555556, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1085}}}, "reference_key_id": "keys/synthetic/syn_03018_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "07068ac4be74eeb8732a0c40afecb5d8f0a37edd6a1196705e2f367d41b90792", "output_hash": "38178a20b470cfd18299fb1593dd4ba706f1a83321dac051d01086efd8b7a96f"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "384989366509a83dfaf367aa", "attempt_id": "1adefe169d0e", "run_group_id": "E4_core:A4|closed|qwen3b-llama3b:h2", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:14:19.743784Z", "task_family": "synthetic", "document_id": "syn_03018_2000", "document_hash": "d284eb07e27da742b9c2b67abccfdf72e4dd67c3259f44f3b41e34dd3e6fc1ff", "question_id": "q07_negation", "split": "test", "cluster_id": "syn_03018_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "tokenizer_hashes": {"receiver": "b9d8872cf3a99da99e1fc4b0f71232a451632dc49ca4c6771549b4efa5c1c763"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 2, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1773, "query_tokens": 521, "message_tokens": 416, "message_utf8_bytes": 1085, "cache_tensor_bytes": null, "serialized_payload_bytes": 1085, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 11937.892624991946, "hops": {"1": 6878.585290978663, "2": 5059.307334013283}, "receiver_prefill_ms": 209.1656670672819, "ttft_ms": 11.551208095625043, "decode_ms": 23.038959014229476}, "total_wall_ms": 12182.06449993886, "ttft_ms": 11.551208095625043, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 2}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "No.", "evidence_artifact": {"ledger": {"facts_total": 64, "facts_retained": 22, "retention_rate": 0.34375, "retention_by_kind": {"person": 0.5, "date": 0.5555555555555556, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1085}, "ledger_by_hop": {"1": {"facts_total": 64, "facts_retained": 22, "retention_rate": 0.34375, "retention_by_kind": {"person": 0.5, "date": 0.5555555555555556, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1181}, "2": {"facts_total": 64, "facts_retained": 22, "retention_rate": 0.34375, "retention_by_kind": {"person": 0.5, "date": 0.5555555555555556, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1085}}}, "reference_key_id": "keys/synthetic/syn_03018_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "07068ac4be74eeb8732a0c40afecb5d8f0a37edd6a1196705e2f367d41b90792", "output_hash": "38178a20b470cfd18299fb1593dd4ba706f1a83321dac051d01086efd8b7a96f"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "71de9f943f128db5d9044db7", "attempt_id": "47795d75b93b", "run_group_id": "E4_core:A4|closed|qwen3b-llama3b:h2", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:14:20.002287Z", "task_family": "synthetic", "document_id": "syn_03018_2000", "document_hash": "d284eb07e27da742b9c2b67abccfdf72e4dd67c3259f44f3b41e34dd3e6fc1ff", "question_id": "q08_revision", "split": "test", "cluster_id": "syn_03018_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "tokenizer_hashes": {"receiver": "b9d8872cf3a99da99e1fc4b0f71232a451632dc49ca4c6771549b4efa5c1c763"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 2, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1773, "query_tokens": 537, "message_tokens": 416, "message_utf8_bytes": 1085, "cache_tensor_bytes": null, "serialized_payload_bytes": 1085, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 11937.892624991946, "hops": {"1": 6878.585290978663, "2": 5059.307334013283}, "receiver_prefill_ms": 210.6421251082793, "ttft_ms": 11.554124997928739, "decode_ms": 34.53833295498043}, "total_wall_ms": 12195.065208012238, "ttft_ms": 11.554124997928739, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 3}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "304,815", "evidence_artifact": {"ledger": {"facts_total": 64, "facts_retained": 22, "retention_rate": 0.34375, "retention_by_kind": {"person": 0.5, "date": 0.5555555555555556, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1085}, "ledger_by_hop": {"1": {"facts_total": 64, "facts_retained": 22, "retention_rate": 0.34375, "retention_by_kind": {"person": 0.5, "date": 0.5555555555555556, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1181}, "2": {"facts_total": 64, "facts_retained": 22, "retention_rate": 0.34375, "retention_by_kind": {"person": 0.5, "date": 0.5555555555555556, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1085}}}, "reference_key_id": "keys/synthetic/syn_03018_2000.json", "task_scores": {"correct": 0.0, "abstained": 0.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "07068ac4be74eeb8732a0c40afecb5d8f0a37edd6a1196705e2f367d41b90792", "output_hash": "763dbd4e77545c090115565b4c750695077f60f3d90c931f2917bbf660a928d3"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "abb389f1156592bda7819dc9", "attempt_id": "382a16cf9c3d", "run_group_id": "E4_core:A4|closed|qwen3b-llama3b:h2", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:14:20.274001Z", "task_family": "synthetic", "document_id": "syn_03018_2000", "document_hash": "d284eb07e27da742b9c2b67abccfdf72e4dd67c3259f44f3b41e34dd3e6fc1ff", "question_id": "q09_revision", "split": "test", "cluster_id": "syn_03018_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "tokenizer_hashes": {"receiver": "b9d8872cf3a99da99e1fc4b0f71232a451632dc49ca4c6771549b4efa5c1c763"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 2, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1773, "query_tokens": 530, "message_tokens": 416, "message_utf8_bytes": 1085, "cache_tensor_bytes": null, "serialized_payload_bytes": 1085, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 11937.892624991946, "hops": {"1": 6878.585290978663, "2": 5059.307334013283}, "receiver_prefill_ms": 209.57079192157835, "ttft_ms": 11.574124917387962, "decode_ms": 34.71762500703335}, "total_wall_ms": 12194.185708998702, "ttft_ms": 11.574124917387962, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 3}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "304,815", "evidence_artifact": {"ledger": {"facts_total": 64, "facts_retained": 22, "retention_rate": 0.34375, "retention_by_kind": {"person": 0.5, "date": 0.5555555555555556, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1085}, "ledger_by_hop": {"1": {"facts_total": 64, "facts_retained": 22, "retention_rate": 0.34375, "retention_by_kind": {"person": 0.5, "date": 0.5555555555555556, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1181}, "2": {"facts_total": 64, "facts_retained": 22, "retention_rate": 0.34375, "retention_by_kind": {"person": 0.5, "date": 0.5555555555555556, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1085}}}, "reference_key_id": "keys/synthetic/syn_03018_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "07068ac4be74eeb8732a0c40afecb5d8f0a37edd6a1196705e2f367d41b90792", "output_hash": "763dbd4e77545c090115565b4c750695077f60f3d90c931f2917bbf660a928d3"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "c36f32b220f113b50e6403a5", "attempt_id": "3f09407a932e", "run_group_id": "E4_core:A4|closed|qwen3b-llama3b:h2", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:14:20.544568Z", "task_family": "synthetic", "document_id": "syn_03018_2000", "document_hash": "d284eb07e27da742b9c2b67abccfdf72e4dd67c3259f44f3b41e34dd3e6fc1ff", "question_id": "q10_temporal", "split": "test", "cluster_id": "syn_03018_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "tokenizer_hashes": {"receiver": "b9d8872cf3a99da99e1fc4b0f71232a451632dc49ca4c6771549b4efa5c1c763"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 2, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1773, "query_tokens": 538, "message_tokens": 416, "message_utf8_bytes": 1085, "cache_tensor_bytes": null, "serialized_payload_bytes": 1085, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 11937.892624991946, "hops": {"1": 6878.585290978663, "2": 5059.307334013283}, "receiver_prefill_ms": 210.5684579582885, "ttft_ms": 11.626000050455332, "decode_ms": 11.60550001077354}, "total_wall_ms": 12172.120625036769, "ttft_ms": 11.626000050455332, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 64, "facts_retained": 22, "retention_rate": 0.34375, "retention_by_kind": {"person": 0.5, "date": 0.5555555555555556, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1085}, "ledger_by_hop": {"1": {"facts_total": 64, "facts_retained": 22, "retention_rate": 0.34375, "retention_by_kind": {"person": 0.5, "date": 0.5555555555555556, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1181}, "2": {"facts_total": 64, "facts_retained": 22, "retention_rate": 0.34375, "retention_by_kind": {"person": 0.5, "date": 0.5555555555555556, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1085}}}, "reference_key_id": "keys/synthetic/syn_03018_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "07068ac4be74eeb8732a0c40afecb5d8f0a37edd6a1196705e2f367d41b90792", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "444aae5da08a999817e5b7f3", "attempt_id": "39f562f09d2c", "run_group_id": "E4_core:A4|closed|qwen3b-llama3b:h2", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:14:20.793025Z", "task_family": "synthetic", "document_id": "syn_03018_2000", "document_hash": "d284eb07e27da742b9c2b67abccfdf72e4dd67c3259f44f3b41e34dd3e6fc1ff", "question_id": "q11_aggregation", "split": "test", "cluster_id": "syn_03018_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "tokenizer_hashes": {"receiver": "b9d8872cf3a99da99e1fc4b0f71232a451632dc49ca4c6771549b4efa5c1c763"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 2, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1773, "query_tokens": 532, "message_tokens": 416, "message_utf8_bytes": 1085, "cache_tensor_bytes": null, "serialized_payload_bytes": 1085, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 11937.892624991946, "hops": {"1": 6878.585290978663, "2": 5059.307334013283}, "receiver_prefill_ms": 210.53991606459022, "ttft_ms": 11.587082990445197, "decode_ms": 276.3855829834938}, "total_wall_ms": 12436.850750003941, "ttft_ms": 11.587082990445197, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 24}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "2,917,061 + 272,184 + 9,472,675 = 10,559,920", "evidence_artifact": {"ledger": {"facts_total": 64, "facts_retained": 22, "retention_rate": 0.34375, "retention_by_kind": {"person": 0.5, "date": 0.5555555555555556, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1085}, "ledger_by_hop": {"1": {"facts_total": 64, "facts_retained": 22, "retention_rate": 0.34375, "retention_by_kind": {"person": 0.5, "date": 0.5555555555555556, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1181}, "2": {"facts_total": 64, "facts_retained": 22, "retention_rate": 0.34375, "retention_by_kind": {"person": 0.5, "date": 0.5555555555555556, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1085}}}, "reference_key_id": "keys/synthetic/syn_03018_2000.json", "task_scores": {"correct": 0.0, "abstained": 0.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "07068ac4be74eeb8732a0c40afecb5d8f0a37edd6a1196705e2f367d41b90792", "output_hash": "96e748112886d948428329c3085eb4214b6634104561a490ddecd8deb1724d18"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "d55119c67c534a20ec99845e", "attempt_id": "7c81a96ab992", "run_group_id": "E4_core:A4|closed|qwen3b-llama3b:h2", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:14:21.306691Z", "task_family": "synthetic", "document_id": "syn_03018_2000", "document_hash": "d284eb07e27da742b9c2b67abccfdf72e4dd67c3259f44f3b41e34dd3e6fc1ff", "question_id": "q12_unit_conversion", "split": "test", "cluster_id": "syn_03018_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "tokenizer_hashes": {"receiver": "b9d8872cf3a99da99e1fc4b0f71232a451632dc49ca4c6771549b4efa5c1c763"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 2, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1773, "query_tokens": 528, "message_tokens": 416, "message_utf8_bytes": 1085, "cache_tensor_bytes": null, "serialized_payload_bytes": 1085, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 11937.892624991946, "hops": {"1": 6878.585290978663, "2": 5059.307334013283}, "receiver_prefill_ms": 209.68895789701492, "ttft_ms": 11.57012500334531, "decode_ms": 11.53745804913342}, "total_wall_ms": 12171.111540985294, "ttft_ms": 11.57012500334531, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 64, "facts_retained": 22, "retention_rate": 0.34375, "retention_by_kind": {"person": 0.5, "date": 0.5555555555555556, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1085}, "ledger_by_hop": {"1": {"facts_total": 64, "facts_retained": 22, "retention_rate": 0.34375, "retention_by_kind": {"person": 0.5, "date": 0.5555555555555556, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1181}, "2": {"facts_total": 64, "facts_retained": 22, "retention_rate": 0.34375, "retention_by_kind": {"person": 0.5, "date": 0.5555555555555556, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1085}}}, "reference_key_id": "keys/synthetic/syn_03018_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "07068ac4be74eeb8732a0c40afecb5d8f0a37edd6a1196705e2f367d41b90792", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "08609871bc34929fee6c8f94", "attempt_id": "8dd10299a998", "run_group_id": "E4_core:A4|closed|qwen3b-llama3b:h2", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:14:21.554576Z", "task_family": "synthetic", "document_id": "syn_03018_2000", "document_hash": "d284eb07e27da742b9c2b67abccfdf72e4dd67c3259f44f3b41e34dd3e6fc1ff", "question_id": "q13_unanswerable", "split": "test", "cluster_id": "syn_03018_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "tokenizer_hashes": {"receiver": "b9d8872cf3a99da99e1fc4b0f71232a451632dc49ca4c6771549b4efa5c1c763"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 2, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1773, "query_tokens": 533, "message_tokens": 416, "message_utf8_bytes": 1085, "cache_tensor_bytes": null, "serialized_payload_bytes": 1085, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 11937.892624991946, "hops": {"1": 6878.585290978663, "2": 5059.307334013283}, "receiver_prefill_ms": 210.44645889196545, "ttft_ms": 11.64487493224442, "decode_ms": 80.57695801835507}, "total_wall_ms": 12240.971165942028, "ttft_ms": 11.64487493224442, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 7}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "HM-38K-5533", "evidence_artifact": {"ledger": {"facts_total": 64, "facts_retained": 22, "retention_rate": 0.34375, "retention_by_kind": {"person": 0.5, "date": 0.5555555555555556, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1085}, "ledger_by_hop": {"1": {"facts_total": 64, "facts_retained": 22, "retention_rate": 0.34375, "retention_by_kind": {"person": 0.5, "date": 0.5555555555555556, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1181}, "2": {"facts_total": 64, "facts_retained": 22, "retention_rate": 0.34375, "retention_by_kind": {"person": 0.5, "date": 0.5555555555555556, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1085}}}, "reference_key_id": "keys/synthetic/syn_03018_2000.json", "task_scores": {"correct": 0.0, "abstained": 0.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "07068ac4be74eeb8732a0c40afecb5d8f0a37edd6a1196705e2f367d41b90792", "output_hash": "18acd9807857243bbded3bceae55398316a3958b81d70364441bdb6748eaf878"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "2d501ccf2338cb7f301cdc23", "attempt_id": "6c4db3cfc8dc", "run_group_id": "E4_core:A4|closed|qwen3b-llama3b:h2", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:14:21.871926Z", "task_family": "synthetic", "document_id": "syn_03018_2000", "document_hash": "d284eb07e27da742b9c2b67abccfdf72e4dd67c3259f44f3b41e34dd3e6fc1ff", "question_id": "q14_unanswerable", "split": "test", "cluster_id": "syn_03018_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "tokenizer_hashes": {"receiver": "b9d8872cf3a99da99e1fc4b0f71232a451632dc49ca4c6771549b4efa5c1c763"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 2, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1773, "query_tokens": 534, "message_tokens": 416, "message_utf8_bytes": 1085, "cache_tensor_bytes": null, "serialized_payload_bytes": 1085, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 11937.892624991946, "hops": {"1": 6878.585290978663, "2": 5059.307334013283}, "receiver_prefill_ms": 210.49125003628433, "ttft_ms": 11.556999990716577, "decode_ms": 11.596458032727242}, "total_wall_ms": 12171.986167086288, "ttft_ms": 11.556999990716577, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 64, "facts_retained": 22, "retention_rate": 0.34375, "retention_by_kind": {"person": 0.5, "date": 0.5555555555555556, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1085}, "ledger_by_hop": {"1": {"facts_total": 64, "facts_retained": 22, "retention_rate": 0.34375, "retention_by_kind": {"person": 0.5, "date": 0.5555555555555556, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1181}, "2": {"facts_total": 64, "facts_retained": 22, "retention_rate": 0.34375, "retention_by_kind": {"person": 0.5, "date": 0.5555555555555556, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1085}}}, "reference_key_id": "keys/synthetic/syn_03018_2000.json", "task_scores": {"correct": 1.0, "abstained": 1.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "07068ac4be74eeb8732a0c40afecb5d8f0a37edd6a1196705e2f367d41b90792", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "99a8ee87aabbc0e4be1a8c65", "attempt_id": "503d7bd24353", "run_group_id": "E4_core:A4|closed|qwen3b-llama3b:h4", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:14:33.321249Z", "task_family": "synthetic", "document_id": "syn_03018_2000", "document_hash": "d284eb07e27da742b9c2b67abccfdf72e4dd67c3259f44f3b41e34dd3e6fc1ff", "question_id": "q00_single_fact", "split": "test", "cluster_id": "syn_03018_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "tokenizer_hashes": {"receiver": "b9d8872cf3a99da99e1fc4b0f71232a451632dc49ca4c6771549b4efa5c1c763"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 4, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1773, "query_tokens": 513, "message_tokens": 409, "message_utf8_bytes": 1203, "cache_tensor_bytes": null, "serialized_payload_bytes": 1203, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 23135.95966703724, "hops": {"1": 6878.585290978663, "2": 5059.307334013283, "3": 6213.180000078864, "4": 4984.887041966431}, "receiver_prefill_ms": 204.8088749870658, "ttft_ms": 12.096709106117487, "decode_ms": 11.540499981492758}, "total_wall_ms": 23364.977917051874, "ttft_ms": 12.096709106117487, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 64, "facts_retained": 22, "retention_rate": 0.34375, "retention_by_kind": {"person": 0.5, "date": 0.5555555555555556, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1203}, "ledger_by_hop": {"1": {"facts_total": 64, "facts_retained": 22, "retention_rate": 0.34375, "retention_by_kind": {"person": 0.5, "date": 0.5555555555555556, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1181}, "2": {"facts_total": 64, "facts_retained": 22, "retention_rate": 0.34375, "retention_by_kind": {"person": 0.5, "date": 0.5555555555555556, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1085}, "3": {"facts_total": 64, "facts_retained": 22, "retention_rate": 0.34375, "retention_by_kind": {"person": 0.5, "date": 0.5555555555555556, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1085}, "4": {"facts_total": 64, "facts_retained": 22, "retention_rate": 0.34375, "retention_by_kind": {"person": 0.5, "date": 0.5555555555555556, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1203}}}, "reference_key_id": "keys/synthetic/syn_03018_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "cd70c40de1b204b635a25f1b065bcadd1379d53837e37154dc490cd58be455c6", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "c1bdcaa65fd1e8e7dc7e1c22", "attempt_id": "0b5c8374997b", "run_group_id": "E4_core:A4|closed|qwen3b-llama3b:h4", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:14:33.565507Z", "task_family": "synthetic", "document_id": "syn_03018_2000", "document_hash": "d284eb07e27da742b9c2b67abccfdf72e4dd67c3259f44f3b41e34dd3e6fc1ff", "question_id": "q01_single_fact", "split": "test", "cluster_id": "syn_03018_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "tokenizer_hashes": {"receiver": "b9d8872cf3a99da99e1fc4b0f71232a451632dc49ca4c6771549b4efa5c1c763"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 4, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1773, "query_tokens": 519, "message_tokens": 409, "message_utf8_bytes": 1203, "cache_tensor_bytes": null, "serialized_payload_bytes": 1203, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 23135.95966703724, "hops": {"1": 6878.585290978663, "2": 5059.307334013283, "3": 6213.180000078864, "4": 4984.887041966431}, "receiver_prefill_ms": 210.33149992581457, "ttft_ms": 11.544583016075194, "decode_ms": 125.98645896650851}, "total_wall_ms": 23484.399250126444, "ttft_ms": 11.544583016075194, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 11}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "Lot 376-E, Baie-Comeau.", "evidence_artifact": {"ledger": {"facts_total": 64, "facts_retained": 22, "retention_rate": 0.34375, "retention_by_kind": {"person": 0.5, "date": 0.5555555555555556, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1203}, "ledger_by_hop": {"1": {"facts_total": 64, "facts_retained": 22, "retention_rate": 0.34375, "retention_by_kind": {"person": 0.5, "date": 0.5555555555555556, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1181}, "2": {"facts_total": 64, "facts_retained": 22, "retention_rate": 0.34375, "retention_by_kind": {"person": 0.5, "date": 0.5555555555555556, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1085}, "3": {"facts_total": 64, "facts_retained": 22, "retention_rate": 0.34375, "retention_by_kind": {"person": 0.5, "date": 0.5555555555555556, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1085}, "4": {"facts_total": 64, "facts_retained": 22, "retention_rate": 0.34375, "retention_by_kind": {"person": 0.5, "date": 0.5555555555555556, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1203}}}, "reference_key_id": "keys/synthetic/syn_03018_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 0.8571428571428571}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "cd70c40de1b204b635a25f1b065bcadd1379d53837e37154dc490cd58be455c6", "output_hash": "8884a56b4b1757cacee95b0917a46f02a41042e2887e8c0ed24b52ccb81335ba"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "41ab2fac850f5e6caa5cce05", "attempt_id": "9cc7ebb9779e", "run_group_id": "E4_core:A4|closed|qwen3b-llama3b:h4", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:14:33.929450Z", "task_family": "synthetic", "document_id": "syn_03018_2000", "document_hash": "d284eb07e27da742b9c2b67abccfdf72e4dd67c3259f44f3b41e34dd3e6fc1ff", "question_id": "q02_exact_copy", "split": "test", "cluster_id": "syn_03018_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "tokenizer_hashes": {"receiver": "b9d8872cf3a99da99e1fc4b0f71232a451632dc49ca4c6771549b4efa5c1c763"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 4, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1773, "query_tokens": 518, "message_tokens": 409, "message_utf8_bytes": 1203, "cache_tensor_bytes": null, "serialized_payload_bytes": 1203, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 23135.95966703724, "hops": {"1": 6878.585290978663, "2": 5059.307334013283, "3": 6213.180000078864, "4": 4984.887041966431}, "receiver_prefill_ms": 210.02808399498463, "ttft_ms": 11.67124998755753, "decode_ms": 11.506500071845949}, "total_wall_ms": 23369.707750040106, "ttft_ms": 11.67124998755753, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 64, "facts_retained": 22, "retention_rate": 0.34375, "retention_by_kind": {"person": 0.5, "date": 0.5555555555555556, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1203}, "ledger_by_hop": {"1": {"facts_total": 64, "facts_retained": 22, "retention_rate": 0.34375, "retention_by_kind": {"person": 0.5, "date": 0.5555555555555556, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1181}, "2": {"facts_total": 64, "facts_retained": 22, "retention_rate": 0.34375, "retention_by_kind": {"person": 0.5, "date": 0.5555555555555556, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1085}, "3": {"facts_total": 64, "facts_retained": 22, "retention_rate": 0.34375, "retention_by_kind": {"person": 0.5, "date": 0.5555555555555556, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1085}, "4": {"facts_total": 64, "facts_retained": 22, "retention_rate": 0.34375, "retention_by_kind": {"person": 0.5, "date": 0.5555555555555556, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1203}}}, "reference_key_id": "keys/synthetic/syn_03018_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "cd70c40de1b204b635a25f1b065bcadd1379d53837e37154dc490cd58be455c6", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "4c3f909a5b0a7bfc735af6e0", "attempt_id": "d3e946b99314", "run_group_id": "E4_core:A4|closed|qwen3b-llama3b:h4", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:14:34.179464Z", "task_family": "synthetic", "document_id": "syn_03018_2000", "document_hash": "d284eb07e27da742b9c2b67abccfdf72e4dd67c3259f44f3b41e34dd3e6fc1ff", "question_id": "q03_exact_copy", "split": "test", "cluster_id": "syn_03018_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "tokenizer_hashes": {"receiver": "b9d8872cf3a99da99e1fc4b0f71232a451632dc49ca4c6771549b4efa5c1c763"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 4, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1773, "query_tokens": 515, "message_tokens": 409, "message_utf8_bytes": 1203, "cache_tensor_bytes": null, "serialized_payload_bytes": 1203, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 23135.95966703724, "hops": {"1": 6878.585290978663, "2": 5059.307334013283, "3": 6213.180000078864, "4": 4984.887041966431}, "receiver_prefill_ms": 209.58441705442965, "ttft_ms": 11.755125015042722, "decode_ms": 103.57808298431337}, "total_wall_ms": 23461.334708030336, "ttft_ms": 11.755125015042722, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 9}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "Priyamvada Sørlie", "evidence_artifact": {"ledger": {"facts_total": 64, "facts_retained": 22, "retention_rate": 0.34375, "retention_by_kind": {"person": 0.5, "date": 0.5555555555555556, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1203}, "ledger_by_hop": {"1": {"facts_total": 64, "facts_retained": 22, "retention_rate": 0.34375, "retention_by_kind": {"person": 0.5, "date": 0.5555555555555556, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1181}, "2": {"facts_total": 64, "facts_retained": 22, "retention_rate": 0.34375, "retention_by_kind": {"person": 0.5, "date": 0.5555555555555556, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1085}, "3": {"facts_total": 64, "facts_retained": 22, "retention_rate": 0.34375, "retention_by_kind": {"person": 0.5, "date": 0.5555555555555556, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1085}, "4": {"facts_total": 64, "facts_retained": 22, "retention_rate": 0.34375, "retention_by_kind": {"person": 0.5, "date": 0.5555555555555556, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1203}}}, "reference_key_id": "keys/synthetic/syn_03018_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "cd70c40de1b204b635a25f1b065bcadd1379d53837e37154dc490cd58be455c6", "output_hash": "c5f554196ed23dd71ad995ec89f7142099d524dbb6534cc10d94a0c89d13e792"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "c3aeaff308d3c4c22890fe42", "attempt_id": "0d197350a178", "run_group_id": "E4_core:A4|closed|qwen3b-llama3b:h4", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:14:34.520040Z", "task_family": "synthetic", "document_id": "syn_03018_2000", "document_hash": "d284eb07e27da742b9c2b67abccfdf72e4dd67c3259f44f3b41e34dd3e6fc1ff", "question_id": "q04_multi_hop", "split": "test", "cluster_id": "syn_03018_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "tokenizer_hashes": {"receiver": "b9d8872cf3a99da99e1fc4b0f71232a451632dc49ca4c6771549b4efa5c1c763"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 4, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1773, "query_tokens": 523, "message_tokens": 409, "message_utf8_bytes": 1203, "cache_tensor_bytes": null, "serialized_payload_bytes": 1203, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 23135.95966703724, "hops": {"1": 6878.585290978663, "2": 5059.307334013283, "3": 6213.180000078864, "4": 4984.887041966431}, "receiver_prefill_ms": 209.97358404565603, "ttft_ms": 11.570207891054451, "decode_ms": 11.501749977469444}, "total_wall_ms": 23369.45270805154, "ttft_ms": 11.570207891054451, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 64, "facts_retained": 22, "retention_rate": 0.34375, "retention_by_kind": {"person": 0.5, "date": 0.5555555555555556, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1203}, "ledger_by_hop": {"1": {"facts_total": 64, "facts_retained": 22, "retention_rate": 0.34375, "retention_by_kind": {"person": 0.5, "date": 0.5555555555555556, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1181}, "2": {"facts_total": 64, "facts_retained": 22, "retention_rate": 0.34375, "retention_by_kind": {"person": 0.5, "date": 0.5555555555555556, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1085}, "3": {"facts_total": 64, "facts_retained": 22, "retention_rate": 0.34375, "retention_by_kind": {"person": 0.5, "date": 0.5555555555555556, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1085}, "4": {"facts_total": 64, "facts_retained": 22, "retention_rate": 0.34375, "retention_by_kind": {"person": 0.5, "date": 0.5555555555555556, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1203}}}, "reference_key_id": "keys/synthetic/syn_03018_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "cd70c40de1b204b635a25f1b065bcadd1379d53837e37154dc490cd58be455c6", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "446e9fc2e004bb4ae99730f2", "attempt_id": "5977094caefb", "run_group_id": "E4_core:A4|closed|qwen3b-llama3b:h4", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:14:34.768166Z", "task_family": "synthetic", "document_id": "syn_03018_2000", "document_hash": "d284eb07e27da742b9c2b67abccfdf72e4dd67c3259f44f3b41e34dd3e6fc1ff", "question_id": "q05_multi_hop", "split": "test", "cluster_id": "syn_03018_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "tokenizer_hashes": {"receiver": "b9d8872cf3a99da99e1fc4b0f71232a451632dc49ca4c6771549b4efa5c1c763"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 4, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1773, "query_tokens": 521, "message_tokens": 409, "message_utf8_bytes": 1203, "cache_tensor_bytes": null, "serialized_payload_bytes": 1203, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 23135.95966703724, "hops": {"1": 6878.585290978663, "2": 5059.307334013283, "3": 6213.180000078864, "4": 4984.887041966431}, "receiver_prefill_ms": 209.91258299909532, "ttft_ms": 11.561832972802222, "decode_ms": 80.87033301126212}, "total_wall_ms": 23438.731667003594, "ttft_ms": 11.561832972802222, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 7}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "HM-38K-5533", "evidence_artifact": {"ledger": {"facts_total": 64, "facts_retained": 22, "retention_rate": 0.34375, "retention_by_kind": {"person": 0.5, "date": 0.5555555555555556, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1203}, "ledger_by_hop": {"1": {"facts_total": 64, "facts_retained": 22, "retention_rate": 0.34375, "retention_by_kind": {"person": 0.5, "date": 0.5555555555555556, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1181}, "2": {"facts_total": 64, "facts_retained": 22, "retention_rate": 0.34375, "retention_by_kind": {"person": 0.5, "date": 0.5555555555555556, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1085}, "3": {"facts_total": 64, "facts_retained": 22, "retention_rate": 0.34375, "retention_by_kind": {"person": 0.5, "date": 0.5555555555555556, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1085}, "4": {"facts_total": 64, "facts_retained": 22, "retention_rate": 0.34375, "retention_by_kind": {"person": 0.5, "date": 0.5555555555555556, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1203}}}, "reference_key_id": "keys/synthetic/syn_03018_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "cd70c40de1b204b635a25f1b065bcadd1379d53837e37154dc490cd58be455c6", "output_hash": "18acd9807857243bbded3bceae55398316a3958b81d70364441bdb6748eaf878"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "0fd259bc13df22fd3722c7a2", "attempt_id": "633ab6ef98ff", "run_group_id": "E4_core:A4|closed|qwen3b-llama3b:h4", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:14:35.085508Z", "task_family": "synthetic", "document_id": "syn_03018_2000", "document_hash": "d284eb07e27da742b9c2b67abccfdf72e4dd67c3259f44f3b41e34dd3e6fc1ff", "question_id": "q06_negation", "split": "test", "cluster_id": "syn_03018_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "tokenizer_hashes": {"receiver": "b9d8872cf3a99da99e1fc4b0f71232a451632dc49ca4c6771549b4efa5c1c763"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 4, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1773, "query_tokens": 514, "message_tokens": 409, "message_utf8_bytes": 1203, "cache_tensor_bytes": null, "serialized_payload_bytes": 1203, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 23135.95966703724, "hops": {"1": 6878.585290978663, "2": 5059.307334013283, "3": 6213.180000078864, "4": 4984.887041966431}, "receiver_prefill_ms": 209.41008301451802, "ttft_ms": 11.472541955299675, "decode_ms": 11.736208107322454}, "total_wall_ms": 23369.13041700609, "ttft_ms": 11.472541955299675, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 64, "facts_retained": 22, "retention_rate": 0.34375, "retention_by_kind": {"person": 0.5, "date": 0.5555555555555556, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1203}, "ledger_by_hop": {"1": {"facts_total": 64, "facts_retained": 22, "retention_rate": 0.34375, "retention_by_kind": {"person": 0.5, "date": 0.5555555555555556, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1181}, "2": {"facts_total": 64, "facts_retained": 22, "retention_rate": 0.34375, "retention_by_kind": {"person": 0.5, "date": 0.5555555555555556, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1085}, "3": {"facts_total": 64, "facts_retained": 22, "retention_rate": 0.34375, "retention_by_kind": {"person": 0.5, "date": 0.5555555555555556, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1085}, "4": {"facts_total": 64, "facts_retained": 22, "retention_rate": 0.34375, "retention_by_kind": {"person": 0.5, "date": 0.5555555555555556, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1203}}}, "reference_key_id": "keys/synthetic/syn_03018_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "cd70c40de1b204b635a25f1b065bcadd1379d53837e37154dc490cd58be455c6", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "4e3cd14f001e16afb8640140", "attempt_id": "601038fa9a84", "run_group_id": "E4_core:A4|closed|qwen3b-llama3b:h4", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:14:35.333673Z", "task_family": "synthetic", "document_id": "syn_03018_2000", "document_hash": "d284eb07e27da742b9c2b67abccfdf72e4dd67c3259f44f3b41e34dd3e6fc1ff", "question_id": "q07_negation", "split": "test", "cluster_id": "syn_03018_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "tokenizer_hashes": {"receiver": "b9d8872cf3a99da99e1fc4b0f71232a451632dc49ca4c6771549b4efa5c1c763"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 4, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1773, "query_tokens": 514, "message_tokens": 409, "message_utf8_bytes": 1203, "cache_tensor_bytes": null, "serialized_payload_bytes": 1203, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 23135.95966703724, "hops": {"1": 6878.585290978663, "2": 5059.307334013283, "3": 6213.180000078864, "4": 4984.887041966431}, "receiver_prefill_ms": 209.38062493223697, "ttft_ms": 11.713291984051466, "decode_ms": 23.321290966123343}, "total_wall_ms": 23380.93816698529, "ttft_ms": 11.713291984051466, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 2}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "No.", "evidence_artifact": {"ledger": {"facts_total": 64, "facts_retained": 22, "retention_rate": 0.34375, "retention_by_kind": {"person": 0.5, "date": 0.5555555555555556, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1203}, "ledger_by_hop": {"1": {"facts_total": 64, "facts_retained": 22, "retention_rate": 0.34375, "retention_by_kind": {"person": 0.5, "date": 0.5555555555555556, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1181}, "2": {"facts_total": 64, "facts_retained": 22, "retention_rate": 0.34375, "retention_by_kind": {"person": 0.5, "date": 0.5555555555555556, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1085}, "3": {"facts_total": 64, "facts_retained": 22, "retention_rate": 0.34375, "retention_by_kind": {"person": 0.5, "date": 0.5555555555555556, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1085}, "4": {"facts_total": 64, "facts_retained": 22, "retention_rate": 0.34375, "retention_by_kind": {"person": 0.5, "date": 0.5555555555555556, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1203}}}, "reference_key_id": "keys/synthetic/syn_03018_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "cd70c40de1b204b635a25f1b065bcadd1379d53837e37154dc490cd58be455c6", "output_hash": "38178a20b470cfd18299fb1593dd4ba706f1a83321dac051d01086efd8b7a96f"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "e05e9bcb86e91548010cca80", "attempt_id": "83b0c9ff4307", "run_group_id": "E4_core:A4|closed|qwen3b-llama3b:h4", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:14:35.594749Z", "task_family": "synthetic", "document_id": "syn_03018_2000", "document_hash": "d284eb07e27da742b9c2b67abccfdf72e4dd67c3259f44f3b41e34dd3e6fc1ff", "question_id": "q08_revision", "split": "test", "cluster_id": "syn_03018_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "tokenizer_hashes": {"receiver": "b9d8872cf3a99da99e1fc4b0f71232a451632dc49ca4c6771549b4efa5c1c763"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 4, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1773, "query_tokens": 530, "message_tokens": 409, "message_utf8_bytes": 1203, "cache_tensor_bytes": null, "serialized_payload_bytes": 1203, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 23135.95966703724, "hops": {"1": 6878.585290978663, "2": 5059.307334013283, "3": 6213.180000078864, "4": 4984.887041966431}, "receiver_prefill_ms": 210.20791691262275, "ttft_ms": 11.603874969296157, "decode_ms": 34.61199998855591}, "total_wall_ms": 23392.815207946114, "ttft_ms": 11.603874969296157, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 3}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "304,815", "evidence_artifact": {"ledger": {"facts_total": 64, "facts_retained": 22, "retention_rate": 0.34375, "retention_by_kind": {"person": 0.5, "date": 0.5555555555555556, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1203}, "ledger_by_hop": {"1": {"facts_total": 64, "facts_retained": 22, "retention_rate": 0.34375, "retention_by_kind": {"person": 0.5, "date": 0.5555555555555556, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1181}, "2": {"facts_total": 64, "facts_retained": 22, "retention_rate": 0.34375, "retention_by_kind": {"person": 0.5, "date": 0.5555555555555556, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1085}, "3": {"facts_total": 64, "facts_retained": 22, "retention_rate": 0.34375, "retention_by_kind": {"person": 0.5, "date": 0.5555555555555556, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1085}, "4": {"facts_total": 64, "facts_retained": 22, "retention_rate": 0.34375, "retention_by_kind": {"person": 0.5, "date": 0.5555555555555556, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1203}}}, "reference_key_id": "keys/synthetic/syn_03018_2000.json", "task_scores": {"correct": 0.0, "abstained": 0.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "cd70c40de1b204b635a25f1b065bcadd1379d53837e37154dc490cd58be455c6", "output_hash": "763dbd4e77545c090115565b4c750695077f60f3d90c931f2917bbf660a928d3"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "62d7d9975cab3c340f704c07", "attempt_id": "3233cd8a6acd", "run_group_id": "E4_core:A4|closed|qwen3b-llama3b:h4", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:14:35.867342Z", "task_family": "synthetic", "document_id": "syn_03018_2000", "document_hash": "d284eb07e27da742b9c2b67abccfdf72e4dd67c3259f44f3b41e34dd3e6fc1ff", "question_id": "q09_revision", "split": "test", "cluster_id": "syn_03018_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "tokenizer_hashes": {"receiver": "b9d8872cf3a99da99e1fc4b0f71232a451632dc49ca4c6771549b4efa5c1c763"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 4, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1773, "query_tokens": 523, "message_tokens": 409, "message_utf8_bytes": 1203, "cache_tensor_bytes": null, "serialized_payload_bytes": 1203, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 23135.95966703724, "hops": {"1": 6878.585290978663, "2": 5059.307334013283, "3": 6213.180000078864, "4": 4984.887041966431}, "receiver_prefill_ms": 210.29433398507535, "ttft_ms": 11.554625001735985, "decode_ms": 34.85437494236976}, "total_wall_ms": 23393.072625040077, "ttft_ms": 11.554625001735985, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 3}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "276,340", "evidence_artifact": {"ledger": {"facts_total": 64, "facts_retained": 22, "retention_rate": 0.34375, "retention_by_kind": {"person": 0.5, "date": 0.5555555555555556, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1203}, "ledger_by_hop": {"1": {"facts_total": 64, "facts_retained": 22, "retention_rate": 0.34375, "retention_by_kind": {"person": 0.5, "date": 0.5555555555555556, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1181}, "2": {"facts_total": 64, "facts_retained": 22, "retention_rate": 0.34375, "retention_by_kind": {"person": 0.5, "date": 0.5555555555555556, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1085}, "3": {"facts_total": 64, "facts_retained": 22, "retention_rate": 0.34375, "retention_by_kind": {"person": 0.5, "date": 0.5555555555555556, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1085}, "4": {"facts_total": 64, "facts_retained": 22, "retention_rate": 0.34375, "retention_by_kind": {"person": 0.5, "date": 0.5555555555555556, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1203}}}, "reference_key_id": "keys/synthetic/syn_03018_2000.json", "task_scores": {"correct": 0.0, "abstained": 0.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "cd70c40de1b204b635a25f1b065bcadd1379d53837e37154dc490cd58be455c6", "output_hash": "ef0dffffbd6c962fe44611aa2052fb7b78efaa1203b0a5de9937b8bd584d71c2"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "a421b027d252289f9e7e06a5", "attempt_id": "558c6ceec56a", "run_group_id": "E4_core:A4|closed|qwen3b-llama3b:h4", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:14:36.138711Z", "task_family": "synthetic", "document_id": "syn_03018_2000", "document_hash": "d284eb07e27da742b9c2b67abccfdf72e4dd67c3259f44f3b41e34dd3e6fc1ff", "question_id": "q10_temporal", "split": "test", "cluster_id": "syn_03018_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "tokenizer_hashes": {"receiver": "b9d8872cf3a99da99e1fc4b0f71232a451632dc49ca4c6771549b4efa5c1c763"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 4, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1773, "query_tokens": 531, "message_tokens": 409, "message_utf8_bytes": 1203, "cache_tensor_bytes": null, "serialized_payload_bytes": 1203, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 23135.95966703724, "hops": {"1": 6878.585290978663, "2": 5059.307334013283, "3": 6213.180000078864, "4": 4984.887041966431}, "receiver_prefill_ms": 210.27350006625056, "ttft_ms": 11.623417027294636, "decode_ms": 11.569708003662527}, "total_wall_ms": 23369.83483401127, "ttft_ms": 11.623417027294636, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 64, "facts_retained": 22, "retention_rate": 0.34375, "retention_by_kind": {"person": 0.5, "date": 0.5555555555555556, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1203}, "ledger_by_hop": {"1": {"facts_total": 64, "facts_retained": 22, "retention_rate": 0.34375, "retention_by_kind": {"person": 0.5, "date": 0.5555555555555556, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1181}, "2": {"facts_total": 64, "facts_retained": 22, "retention_rate": 0.34375, "retention_by_kind": {"person": 0.5, "date": 0.5555555555555556, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1085}, "3": {"facts_total": 64, "facts_retained": 22, "retention_rate": 0.34375, "retention_by_kind": {"person": 0.5, "date": 0.5555555555555556, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1085}, "4": {"facts_total": 64, "facts_retained": 22, "retention_rate": 0.34375, "retention_by_kind": {"person": 0.5, "date": 0.5555555555555556, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1203}}}, "reference_key_id": "keys/synthetic/syn_03018_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "cd70c40de1b204b635a25f1b065bcadd1379d53837e37154dc490cd58be455c6", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "69d11b7b8cc310d5af43d02a", "attempt_id": "bdb43791addd", "run_group_id": "E4_core:A4|closed|qwen3b-llama3b:h4", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:14:36.386891Z", "task_family": "synthetic", "document_id": "syn_03018_2000", "document_hash": "d284eb07e27da742b9c2b67abccfdf72e4dd67c3259f44f3b41e34dd3e6fc1ff", "question_id": "q11_aggregation", "split": "test", "cluster_id": "syn_03018_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "tokenizer_hashes": {"receiver": "b9d8872cf3a99da99e1fc4b0f71232a451632dc49ca4c6771549b4efa5c1c763"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 4, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1773, "query_tokens": 525, "message_tokens": 409, "message_utf8_bytes": 1203, "cache_tensor_bytes": null, "serialized_payload_bytes": 1203, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 23135.95966703724, "hops": {"1": 6878.585290978663, "2": 5059.307334013283, "3": 6213.180000078864, "4": 4984.887041966431}, "receiver_prefill_ms": 210.08395904209465, "ttft_ms": 11.56104204710573, "decode_ms": 276.4952910365537}, "total_wall_ms": 23634.526874986477, "ttft_ms": 11.56104204710573, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 24}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "2,917,061 + 272,184 + 9,472,675 = 10,659,920", "evidence_artifact": {"ledger": {"facts_total": 64, "facts_retained": 22, "retention_rate": 0.34375, "retention_by_kind": {"person": 0.5, "date": 0.5555555555555556, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1203}, "ledger_by_hop": {"1": {"facts_total": 64, "facts_retained": 22, "retention_rate": 0.34375, "retention_by_kind": {"person": 0.5, "date": 0.5555555555555556, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1181}, "2": {"facts_total": 64, "facts_retained": 22, "retention_rate": 0.34375, "retention_by_kind": {"person": 0.5, "date": 0.5555555555555556, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1085}, "3": {"facts_total": 64, "facts_retained": 22, "retention_rate": 0.34375, "retention_by_kind": {"person": 0.5, "date": 0.5555555555555556, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1085}, "4": {"facts_total": 64, "facts_retained": 22, "retention_rate": 0.34375, "retention_by_kind": {"person": 0.5, "date": 0.5555555555555556, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1203}}}, "reference_key_id": "keys/synthetic/syn_03018_2000.json", "task_scores": {"correct": 0.0, "abstained": 0.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "cd70c40de1b204b635a25f1b065bcadd1379d53837e37154dc490cd58be455c6", "output_hash": "8372b66d8642f24e584e75e6a98b660864a61369be60290c26d752e4e1673e71"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "53ff2ecb5e55279a633ac376", "attempt_id": "2e659d955202", "run_group_id": "E4_core:A4|closed|qwen3b-llama3b:h4", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:14:36.899883Z", "task_family": "synthetic", "document_id": "syn_03018_2000", "document_hash": "d284eb07e27da742b9c2b67abccfdf72e4dd67c3259f44f3b41e34dd3e6fc1ff", "question_id": "q12_unit_conversion", "split": "test", "cluster_id": "syn_03018_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "tokenizer_hashes": {"receiver": "b9d8872cf3a99da99e1fc4b0f71232a451632dc49ca4c6771549b4efa5c1c763"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 4, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1773, "query_tokens": 521, "message_tokens": 409, "message_utf8_bytes": 1203, "cache_tensor_bytes": null, "serialized_payload_bytes": 1203, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 23135.95966703724, "hops": {"1": 6878.585290978663, "2": 5059.307334013283, "3": 6213.180000078864, "4": 4984.887041966431}, "receiver_prefill_ms": 210.27958404738456, "ttft_ms": 11.59783301409334, "decode_ms": 11.525291949510574}, "total_wall_ms": 23369.79750113096, "ttft_ms": 11.59783301409334, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 64, "facts_retained": 22, "retention_rate": 0.34375, "retention_by_kind": {"person": 0.5, "date": 0.5555555555555556, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1203}, "ledger_by_hop": {"1": {"facts_total": 64, "facts_retained": 22, "retention_rate": 0.34375, "retention_by_kind": {"person": 0.5, "date": 0.5555555555555556, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1181}, "2": {"facts_total": 64, "facts_retained": 22, "retention_rate": 0.34375, "retention_by_kind": {"person": 0.5, "date": 0.5555555555555556, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1085}, "3": {"facts_total": 64, "facts_retained": 22, "retention_rate": 0.34375, "retention_by_kind": {"person": 0.5, "date": 0.5555555555555556, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1085}, "4": {"facts_total": 64, "facts_retained": 22, "retention_rate": 0.34375, "retention_by_kind": {"person": 0.5, "date": 0.5555555555555556, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1203}}}, "reference_key_id": "keys/synthetic/syn_03018_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "cd70c40de1b204b635a25f1b065bcadd1379d53837e37154dc490cd58be455c6", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "25bed1929e111b69681e907d", "attempt_id": "43ee49fb9da7", "run_group_id": "E4_core:A4|closed|qwen3b-llama3b:h4", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:14:37.148059Z", "task_family": "synthetic", "document_id": "syn_03018_2000", "document_hash": "d284eb07e27da742b9c2b67abccfdf72e4dd67c3259f44f3b41e34dd3e6fc1ff", "question_id": "q13_unanswerable", "split": "test", "cluster_id": "syn_03018_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "tokenizer_hashes": {"receiver": "b9d8872cf3a99da99e1fc4b0f71232a451632dc49ca4c6771549b4efa5c1c763"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 4, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1773, "query_tokens": 526, "message_tokens": 409, "message_utf8_bytes": 1203, "cache_tensor_bytes": null, "serialized_payload_bytes": 1203, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 23135.95966703724, "hops": {"1": 6878.585290978663, "2": 5059.307334013283, "3": 6213.180000078864, "4": 4984.887041966431}, "receiver_prefill_ms": 210.13891708571464, "ttft_ms": 11.507124989293516, "decode_ms": 80.62958298251033}, "total_wall_ms": 23438.678374979645, "ttft_ms": 11.507124989293516, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 7}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "HM-38K-5533", "evidence_artifact": {"ledger": {"facts_total": 64, "facts_retained": 22, "retention_rate": 0.34375, "retention_by_kind": {"person": 0.5, "date": 0.5555555555555556, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1203}, "ledger_by_hop": {"1": {"facts_total": 64, "facts_retained": 22, "retention_rate": 0.34375, "retention_by_kind": {"person": 0.5, "date": 0.5555555555555556, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1181}, "2": {"facts_total": 64, "facts_retained": 22, "retention_rate": 0.34375, "retention_by_kind": {"person": 0.5, "date": 0.5555555555555556, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1085}, "3": {"facts_total": 64, "facts_retained": 22, "retention_rate": 0.34375, "retention_by_kind": {"person": 0.5, "date": 0.5555555555555556, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1085}, "4": {"facts_total": 64, "facts_retained": 22, "retention_rate": 0.34375, "retention_by_kind": {"person": 0.5, "date": 0.5555555555555556, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1203}}}, "reference_key_id": "keys/synthetic/syn_03018_2000.json", "task_scores": {"correct": 0.0, "abstained": 0.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "cd70c40de1b204b635a25f1b065bcadd1379d53837e37154dc490cd58be455c6", "output_hash": "18acd9807857243bbded3bceae55398316a3958b81d70364441bdb6748eaf878"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "bf763110d3a8399465827ac0", "attempt_id": "9286a6de8139", "run_group_id": "E4_core:A4|closed|qwen3b-llama3b:h4", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:14:37.465342Z", "task_family": "synthetic", "document_id": "syn_03018_2000", "document_hash": "d284eb07e27da742b9c2b67abccfdf72e4dd67c3259f44f3b41e34dd3e6fc1ff", "question_id": "q14_unanswerable", "split": "test", "cluster_id": "syn_03018_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "tokenizer_hashes": {"receiver": "b9d8872cf3a99da99e1fc4b0f71232a451632dc49ca4c6771549b4efa5c1c763"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 4, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1773, "query_tokens": 527, "message_tokens": 409, "message_utf8_bytes": 1203, "cache_tensor_bytes": null, "serialized_payload_bytes": 1203, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 23135.95966703724, "hops": {"1": 6878.585290978663, "2": 5059.307334013283, "3": 6213.180000078864, "4": 4984.887041966431}, "receiver_prefill_ms": 210.1262080250308, "ttft_ms": 11.656166985630989, "decode_ms": 11.563999927602708}, "total_wall_ms": 23369.734959094785, "ttft_ms": 11.656166985630989, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 64, "facts_retained": 22, "retention_rate": 0.34375, "retention_by_kind": {"person": 0.5, "date": 0.5555555555555556, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1203}, "ledger_by_hop": {"1": {"facts_total": 64, "facts_retained": 22, "retention_rate": 0.34375, "retention_by_kind": {"person": 0.5, "date": 0.5555555555555556, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1181}, "2": {"facts_total": 64, "facts_retained": 22, "retention_rate": 0.34375, "retention_by_kind": {"person": 0.5, "date": 0.5555555555555556, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1085}, "3": {"facts_total": 64, "facts_retained": 22, "retention_rate": 0.34375, "retention_by_kind": {"person": 0.5, "date": 0.5555555555555556, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1085}, "4": {"facts_total": 64, "facts_retained": 22, "retention_rate": 0.34375, "retention_by_kind": {"person": 0.5, "date": 0.5555555555555556, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1203}}}, "reference_key_id": "keys/synthetic/syn_03018_2000.json", "task_scores": {"correct": 1.0, "abstained": 1.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "cd70c40de1b204b635a25f1b065bcadd1379d53837e37154dc490cd58be455c6", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "42105af50ab8892f528160cf", "attempt_id": "6a4dace9a982", "run_group_id": "E4_core:A4|closed|qwen3b-llama3b:h8", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:14:55.710405Z", "task_family": "synthetic", "document_id": "syn_03018_2000", "document_hash": "d284eb07e27da742b9c2b67abccfdf72e4dd67c3259f44f3b41e34dd3e6fc1ff", "question_id": "q00_single_fact", "split": "test", "cluster_id": "syn_03018_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "tokenizer_hashes": {"receiver": "b9d8872cf3a99da99e1fc4b0f71232a451632dc49ca4c6771549b4efa5c1c763"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 8, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1773, "query_tokens": 402, "message_tokens": 299, "message_utf8_bytes": 850, "cache_tensor_bytes": null, "serialized_payload_bytes": 850, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 41128.03645886015, "hops": {"1": 6878.585290978663, "2": 5059.307334013283, "3": 6213.180000078864, "4": 4984.887041966431, "5": 5136.557208956219, "6": 4697.909708949737, "7": 4502.605207962915, "8": 3655.0046659540385}, "receiver_prefill_ms": 185.54758303798735, "ttft_ms": 11.57779200002551, "decode_ms": 11.477124993689358}, "total_wall_ms": 41337.07479189616, "ttft_ms": 11.57779200002551, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 64, "facts_retained": 21, "retention_rate": 0.328125, "retention_by_kind": {"person": 0.5, "date": 0.2222222222222222, "amount": 0.0, "status": 0.5, "place": 0.3333333333333333, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 850}, "ledger_by_hop": {"1": {"facts_total": 64, "facts_retained": 22, "retention_rate": 0.34375, "retention_by_kind": {"person": 0.5, "date": 0.5555555555555556, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1181}, "2": {"facts_total": 64, "facts_retained": 22, "retention_rate": 0.34375, "retention_by_kind": {"person": 0.5, "date": 0.5555555555555556, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1085}, "3": {"facts_total": 64, "facts_retained": 22, "retention_rate": 0.34375, "retention_by_kind": {"person": 0.5, "date": 0.5555555555555556, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1085}, "4": {"facts_total": 64, "facts_retained": 22, "retention_rate": 0.34375, "retention_by_kind": {"person": 0.5, "date": 0.5555555555555556, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1203}, "5": {"facts_total": 64, "facts_retained": 22, "retention_rate": 0.34375, "retention_by_kind": {"person": 0.5, "date": 0.5555555555555556, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 926}, "6": {"facts_total": 64, "facts_retained": 24, "retention_rate": 0.375, "retention_by_kind": {"person": 0.5, "date": 0.5555555555555556, "amount": 0.0, "status": 0.5, "place": 0.3333333333333333, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1198}, "7": {"facts_total": 64, "facts_retained": 21, "retention_rate": 0.328125, "retention_by_kind": {"person": 0.5, "date": 0.2222222222222222, "amount": 0.0, "status": 0.5, "place": 0.3333333333333333, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 823}, "8": {"facts_total": 64, "facts_retained": 21, "retention_rate": 0.328125, "retention_by_kind": {"person": 0.5, "date": 0.2222222222222222, "amount": 0.0, "status": 0.5, "place": 0.3333333333333333, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 850}}}, "reference_key_id": "keys/synthetic/syn_03018_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "1b6718880d5561636cccbc918423b60943c018b7083fd37c82dd159683aaa309", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "afc50bb3f0a5d83d8de6847a", "attempt_id": "52b7e6ddb3ce", "run_group_id": "E4_core:A4|closed|qwen3b-llama3b:h8", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:14:55.934901Z", "task_family": "synthetic", "document_id": "syn_03018_2000", "document_hash": "d284eb07e27da742b9c2b67abccfdf72e4dd67c3259f44f3b41e34dd3e6fc1ff", "question_id": "q01_single_fact", "split": "test", "cluster_id": "syn_03018_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "tokenizer_hashes": {"receiver": "b9d8872cf3a99da99e1fc4b0f71232a451632dc49ca4c6771549b4efa5c1c763"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 8, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1773, "query_tokens": 408, "message_tokens": 299, "message_utf8_bytes": 850, "cache_tensor_bytes": null, "serialized_payload_bytes": 850, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 41128.03645886015, "hops": {"1": 6878.585290978663, "2": 5059.307334013283, "3": 6213.180000078864, "4": 4984.887041966431, "5": 5136.557208956219, "6": 4697.909708949737, "7": 4502.605207962915, "8": 3655.0046659540385}, "receiver_prefill_ms": 185.13512494973838, "ttft_ms": 11.562291998416185, "decode_ms": 147.99870899878442}, "total_wall_ms": 41473.11079187784, "ttft_ms": 11.562291998416185, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 13}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "Baie-Comeau, TA-34D-9894", "evidence_artifact": {"ledger": {"facts_total": 64, "facts_retained": 21, "retention_rate": 0.328125, "retention_by_kind": {"person": 0.5, "date": 0.2222222222222222, "amount": 0.0, "status": 0.5, "place": 0.3333333333333333, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 850}, "ledger_by_hop": {"1": {"facts_total": 64, "facts_retained": 22, "retention_rate": 0.34375, "retention_by_kind": {"person": 0.5, "date": 0.5555555555555556, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1181}, "2": {"facts_total": 64, "facts_retained": 22, "retention_rate": 0.34375, "retention_by_kind": {"person": 0.5, "date": 0.5555555555555556, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1085}, "3": {"facts_total": 64, "facts_retained": 22, "retention_rate": 0.34375, "retention_by_kind": {"person": 0.5, "date": 0.5555555555555556, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1085}, "4": {"facts_total": 64, "facts_retained": 22, "retention_rate": 0.34375, "retention_by_kind": {"person": 0.5, "date": 0.5555555555555556, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1203}, "5": {"facts_total": 64, "facts_retained": 22, "retention_rate": 0.34375, "retention_by_kind": {"person": 0.5, "date": 0.5555555555555556, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 926}, "6": {"facts_total": 64, "facts_retained": 24, "retention_rate": 0.375, "retention_by_kind": {"person": 0.5, "date": 0.5555555555555556, "amount": 0.0, "status": 0.5, "place": 0.3333333333333333, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1198}, "7": {"facts_total": 64, "facts_retained": 21, "retention_rate": 0.328125, "retention_by_kind": {"person": 0.5, "date": 0.2222222222222222, "amount": 0.0, "status": 0.5, "place": 0.3333333333333333, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 823}, "8": {"facts_total": 64, "facts_retained": 21, "retention_rate": 0.328125, "retention_by_kind": {"person": 0.5, "date": 0.2222222222222222, "amount": 0.0, "status": 0.5, "place": 0.3333333333333333, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 850}}}, "reference_key_id": "keys/synthetic/syn_03018_2000.json", "task_scores": {"correct": 0.0, "abstained": 0.0, "f1": 0.3333333333333333}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "1b6718880d5561636cccbc918423b60943c018b7083fd37c82dd159683aaa309", "output_hash": "d4c1648f22ebf96848f0476b30a87b341845734e8bbbbbe559bf1f401a5025ac"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "a45e9cb631f6999834118882", "attempt_id": "495425895ae1", "run_group_id": "E4_core:A4|closed|qwen3b-llama3b:h8", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:14:56.294787Z", "task_family": "synthetic", "document_id": "syn_03018_2000", "document_hash": "d284eb07e27da742b9c2b67abccfdf72e4dd67c3259f44f3b41e34dd3e6fc1ff", "question_id": "q02_exact_copy", "split": "test", "cluster_id": "syn_03018_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "tokenizer_hashes": {"receiver": "b9d8872cf3a99da99e1fc4b0f71232a451632dc49ca4c6771549b4efa5c1c763"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 8, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1773, "query_tokens": 407, "message_tokens": 299, "message_utf8_bytes": 850, "cache_tensor_bytes": null, "serialized_payload_bytes": 850, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 41128.03645886015, "hops": {"1": 6878.585290978663, "2": 5059.307334013283, "3": 6213.180000078864, "4": 4984.887041966431, "5": 5136.557208956219, "6": 4697.909708949737, "7": 4502.605207962915, "8": 3655.0046659540385}, "receiver_prefill_ms": 187.54629197064787, "ttft_ms": 11.554375058040023, "decode_ms": 11.51966699399054}, "total_wall_ms": 41339.02570896316, "ttft_ms": 11.554375058040023, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 64, "facts_retained": 21, "retention_rate": 0.328125, "retention_by_kind": {"person": 0.5, "date": 0.2222222222222222, "amount": 0.0, "status": 0.5, "place": 0.3333333333333333, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 850}, "ledger_by_hop": {"1": {"facts_total": 64, "facts_retained": 22, "retention_rate": 0.34375, "retention_by_kind": {"person": 0.5, "date": 0.5555555555555556, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1181}, "2": {"facts_total": 64, "facts_retained": 22, "retention_rate": 0.34375, "retention_by_kind": {"person": 0.5, "date": 0.5555555555555556, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1085}, "3": {"facts_total": 64, "facts_retained": 22, "retention_rate": 0.34375, "retention_by_kind": {"person": 0.5, "date": 0.5555555555555556, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1085}, "4": {"facts_total": 64, "facts_retained": 22, "retention_rate": 0.34375, "retention_by_kind": {"person": 0.5, "date": 0.5555555555555556, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1203}, "5": {"facts_total": 64, "facts_retained": 22, "retention_rate": 0.34375, "retention_by_kind": {"person": 0.5, "date": 0.5555555555555556, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 926}, "6": {"facts_total": 64, "facts_retained": 24, "retention_rate": 0.375, "retention_by_kind": {"person": 0.5, "date": 0.5555555555555556, "amount": 0.0, "status": 0.5, "place": 0.3333333333333333, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1198}, "7": {"facts_total": 64, "facts_retained": 21, "retention_rate": 0.328125, "retention_by_kind": {"person": 0.5, "date": 0.2222222222222222, "amount": 0.0, "status": 0.5, "place": 0.3333333333333333, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 823}, "8": {"facts_total": 64, "facts_retained": 21, "retention_rate": 0.328125, "retention_by_kind": {"person": 0.5, "date": 0.2222222222222222, "amount": 0.0, "status": 0.5, "place": 0.3333333333333333, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 850}}}, "reference_key_id": "keys/synthetic/syn_03018_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "1b6718880d5561636cccbc918423b60943c018b7083fd37c82dd159683aaa309", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "e6ea12e631b4633bc01dc269", "attempt_id": "5816bb48512e", "run_group_id": "E4_core:A4|closed|qwen3b-llama3b:h8", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:14:56.522861Z", "task_family": "synthetic", "document_id": "syn_03018_2000", "document_hash": "d284eb07e27da742b9c2b67abccfdf72e4dd67c3259f44f3b41e34dd3e6fc1ff", "question_id": "q03_exact_copy", "split": "test", "cluster_id": "syn_03018_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "tokenizer_hashes": {"receiver": "b9d8872cf3a99da99e1fc4b0f71232a451632dc49ca4c6771549b4efa5c1c763"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 8, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1773, "query_tokens": 404, "message_tokens": 299, "message_utf8_bytes": 850, "cache_tensor_bytes": null, "serialized_payload_bytes": 850, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 41128.03645886015, "hops": {"1": 6878.585290978663, "2": 5059.307334013283, "3": 6213.180000078864, "4": 4984.887041966431, "5": 5136.557208956219, "6": 4697.909708949737, "7": 4502.605207962915, "8": 3655.0046659540385}, "receiver_prefill_ms": 186.57979206182063, "ttft_ms": 11.551458039321005, "decode_ms": 103.2058340497315}, "total_wall_ms": 41429.726583883166, "ttft_ms": 11.551458039321005, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 9}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "Priyamvada Sørlie", "evidence_artifact": {"ledger": {"facts_total": 64, "facts_retained": 21, "retention_rate": 0.328125, "retention_by_kind": {"person": 0.5, "date": 0.2222222222222222, "amount": 0.0, "status": 0.5, "place": 0.3333333333333333, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 850}, "ledger_by_hop": {"1": {"facts_total": 64, "facts_retained": 22, "retention_rate": 0.34375, "retention_by_kind": {"person": 0.5, "date": 0.5555555555555556, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1181}, "2": {"facts_total": 64, "facts_retained": 22, "retention_rate": 0.34375, "retention_by_kind": {"person": 0.5, "date": 0.5555555555555556, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1085}, "3": {"facts_total": 64, "facts_retained": 22, "retention_rate": 0.34375, "retention_by_kind": {"person": 0.5, "date": 0.5555555555555556, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1085}, "4": {"facts_total": 64, "facts_retained": 22, "retention_rate": 0.34375, "retention_by_kind": {"person": 0.5, "date": 0.5555555555555556, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1203}, "5": {"facts_total": 64, "facts_retained": 22, "retention_rate": 0.34375, "retention_by_kind": {"person": 0.5, "date": 0.5555555555555556, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 926}, "6": {"facts_total": 64, "facts_retained": 24, "retention_rate": 0.375, "retention_by_kind": {"person": 0.5, "date": 0.5555555555555556, "amount": 0.0, "status": 0.5, "place": 0.3333333333333333, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1198}, "7": {"facts_total": 64, "facts_retained": 21, "retention_rate": 0.328125, "retention_by_kind": {"person": 0.5, "date": 0.2222222222222222, "amount": 0.0, "status": 0.5, "place": 0.3333333333333333, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 823}, "8": {"facts_total": 64, "facts_retained": 21, "retention_rate": 0.328125, "retention_by_kind": {"person": 0.5, "date": 0.2222222222222222, "amount": 0.0, "status": 0.5, "place": 0.3333333333333333, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 850}}}, "reference_key_id": "keys/synthetic/syn_03018_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "1b6718880d5561636cccbc918423b60943c018b7083fd37c82dd159683aaa309", "output_hash": "c5f554196ed23dd71ad995ec89f7142099d524dbb6534cc10d94a0c89d13e792"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "d14687cb5804fe8c5f9d95ef", "attempt_id": "a14c64ee92e7", "run_group_id": "E4_core:A4|closed|qwen3b-llama3b:h8", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:14:56.839547Z", "task_family": "synthetic", "document_id": "syn_03018_2000", "document_hash": "d284eb07e27da742b9c2b67abccfdf72e4dd67c3259f44f3b41e34dd3e6fc1ff", "question_id": "q04_multi_hop", "split": "test", "cluster_id": "syn_03018_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "tokenizer_hashes": {"receiver": "b9d8872cf3a99da99e1fc4b0f71232a451632dc49ca4c6771549b4efa5c1c763"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 8, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1773, "query_tokens": 412, "message_tokens": 299, "message_utf8_bytes": 850, "cache_tensor_bytes": null, "serialized_payload_bytes": 850, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 41128.03645886015, "hops": {"1": 6878.585290978663, "2": 5059.307334013283, "3": 6213.180000078864, "4": 4984.887041966431, "5": 5136.557208956219, "6": 4697.909708949737, "7": 4502.605207962915, "8": 3655.0046659540385}, "receiver_prefill_ms": 184.97916706837714, "ttft_ms": 11.586207896471024, "decode_ms": 11.550875031389296}, "total_wall_ms": 41336.54391695745, "ttft_ms": 11.586207896471024, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 64, "facts_retained": 21, "retention_rate": 0.328125, "retention_by_kind": {"person": 0.5, "date": 0.2222222222222222, "amount": 0.0, "status": 0.5, "place": 0.3333333333333333, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 850}, "ledger_by_hop": {"1": {"facts_total": 64, "facts_retained": 22, "retention_rate": 0.34375, "retention_by_kind": {"person": 0.5, "date": 0.5555555555555556, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1181}, "2": {"facts_total": 64, "facts_retained": 22, "retention_rate": 0.34375, "retention_by_kind": {"person": 0.5, "date": 0.5555555555555556, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1085}, "3": {"facts_total": 64, "facts_retained": 22, "retention_rate": 0.34375, "retention_by_kind": {"person": 0.5, "date": 0.5555555555555556, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1085}, "4": {"facts_total": 64, "facts_retained": 22, "retention_rate": 0.34375, "retention_by_kind": {"person": 0.5, "date": 0.5555555555555556, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1203}, "5": {"facts_total": 64, "facts_retained": 22, "retention_rate": 0.34375, "retention_by_kind": {"person": 0.5, "date": 0.5555555555555556, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 926}, "6": {"facts_total": 64, "facts_retained": 24, "retention_rate": 0.375, "retention_by_kind": {"person": 0.5, "date": 0.5555555555555556, "amount": 0.0, "status": 0.5, "place": 0.3333333333333333, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1198}, "7": {"facts_total": 64, "facts_retained": 21, "retention_rate": 0.328125, "retention_by_kind": {"person": 0.5, "date": 0.2222222222222222, "amount": 0.0, "status": 0.5, "place": 0.3333333333333333, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 823}, "8": {"facts_total": 64, "facts_retained": 21, "retention_rate": 0.328125, "retention_by_kind": {"person": 0.5, "date": 0.2222222222222222, "amount": 0.0, "status": 0.5, "place": 0.3333333333333333, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 850}}}, "reference_key_id": "keys/synthetic/syn_03018_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "1b6718880d5561636cccbc918423b60943c018b7083fd37c82dd159683aaa309", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "4bb5e4e148addcd41fad29fc", "attempt_id": "dd638b359297", "run_group_id": "E4_core:A4|closed|qwen3b-llama3b:h8", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:14:57.062349Z", "task_family": "synthetic", "document_id": "syn_03018_2000", "document_hash": "d284eb07e27da742b9c2b67abccfdf72e4dd67c3259f44f3b41e34dd3e6fc1ff", "question_id": "q05_multi_hop", "split": "test", "cluster_id": "syn_03018_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "tokenizer_hashes": {"receiver": "b9d8872cf3a99da99e1fc4b0f71232a451632dc49ca4c6771549b4efa5c1c763"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 8, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1773, "query_tokens": 410, "message_tokens": 299, "message_utf8_bytes": 850, "cache_tensor_bytes": null, "serialized_payload_bytes": 850, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 41128.03645886015, "hops": {"1": 6878.585290978663, "2": 5059.307334013283, "3": 6213.180000078864, "4": 4984.887041966431, "5": 5136.557208956219, "6": 4697.909708949737, "7": 4502.605207962915, "8": 3655.0046659540385}, "receiver_prefill_ms": 183.5532080149278, "ttft_ms": 11.523417080752552, "decode_ms": 103.67845906876028}, "total_wall_ms": 41427.14891687501, "ttft_ms": 11.523417080752552, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 9}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "Zéphyrine Qvistrand", "evidence_artifact": {"ledger": {"facts_total": 64, "facts_retained": 21, "retention_rate": 0.328125, "retention_by_kind": {"person": 0.5, "date": 0.2222222222222222, "amount": 0.0, "status": 0.5, "place": 0.3333333333333333, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 850}, "ledger_by_hop": {"1": {"facts_total": 64, "facts_retained": 22, "retention_rate": 0.34375, "retention_by_kind": {"person": 0.5, "date": 0.5555555555555556, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1181}, "2": {"facts_total": 64, "facts_retained": 22, "retention_rate": 0.34375, "retention_by_kind": {"person": 0.5, "date": 0.5555555555555556, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1085}, "3": {"facts_total": 64, "facts_retained": 22, "retention_rate": 0.34375, "retention_by_kind": {"person": 0.5, "date": 0.5555555555555556, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1085}, "4": {"facts_total": 64, "facts_retained": 22, "retention_rate": 0.34375, "retention_by_kind": {"person": 0.5, "date": 0.5555555555555556, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1203}, "5": {"facts_total": 64, "facts_retained": 22, "retention_rate": 0.34375, "retention_by_kind": {"person": 0.5, "date": 0.5555555555555556, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 926}, "6": {"facts_total": 64, "facts_retained": 24, "retention_rate": 0.375, "retention_by_kind": {"person": 0.5, "date": 0.5555555555555556, "amount": 0.0, "status": 0.5, "place": 0.3333333333333333, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1198}, "7": {"facts_total": 64, "facts_retained": 21, "retention_rate": 0.328125, "retention_by_kind": {"person": 0.5, "date": 0.2222222222222222, "amount": 0.0, "status": 0.5, "place": 0.3333333333333333, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 823}, "8": {"facts_total": 64, "facts_retained": 21, "retention_rate": 0.328125, "retention_by_kind": {"person": 0.5, "date": 0.2222222222222222, "amount": 0.0, "status": 0.5, "place": 0.3333333333333333, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 850}}}, "reference_key_id": "keys/synthetic/syn_03018_2000.json", "task_scores": {"correct": 0.0, "abstained": 0.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "1b6718880d5561636cccbc918423b60943c018b7083fd37c82dd159683aaa309", "output_hash": "97a6d4e2ba882d0f93375fd951719a1424d00737260e57feb372ff01ccd51f18"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "dafac3db26a6547812a6deb1", "attempt_id": "5086040806e0", "run_group_id": "E4_core:A4|closed|qwen3b-llama3b:h8", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:14:57.376286Z", "task_family": "synthetic", "document_id": "syn_03018_2000", "document_hash": "d284eb07e27da742b9c2b67abccfdf72e4dd67c3259f44f3b41e34dd3e6fc1ff", "question_id": "q06_negation", "split": "test", "cluster_id": "syn_03018_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "tokenizer_hashes": {"receiver": "b9d8872cf3a99da99e1fc4b0f71232a451632dc49ca4c6771549b4efa5c1c763"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 8, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1773, "query_tokens": 403, "message_tokens": 299, "message_utf8_bytes": 850, "cache_tensor_bytes": null, "serialized_payload_bytes": 850, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 41128.03645886015, "hops": {"1": 6878.585290978663, "2": 5059.307334013283, "3": 6213.180000078864, "4": 4984.887041966431, "5": 5136.557208956219, "6": 4697.909708949737, "7": 4502.605207962915, "8": 3655.0046659540385}, "receiver_prefill_ms": 185.8055000193417, "ttft_ms": 11.53387501835823, "decode_ms": 11.448708013631403}, "total_wall_ms": 41337.20570884179, "ttft_ms": 11.53387501835823, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 64, "facts_retained": 21, "retention_rate": 0.328125, "retention_by_kind": {"person": 0.5, "date": 0.2222222222222222, "amount": 0.0, "status": 0.5, "place": 0.3333333333333333, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 850}, "ledger_by_hop": {"1": {"facts_total": 64, "facts_retained": 22, "retention_rate": 0.34375, "retention_by_kind": {"person": 0.5, "date": 0.5555555555555556, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1181}, "2": {"facts_total": 64, "facts_retained": 22, "retention_rate": 0.34375, "retention_by_kind": {"person": 0.5, "date": 0.5555555555555556, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1085}, "3": {"facts_total": 64, "facts_retained": 22, "retention_rate": 0.34375, "retention_by_kind": {"person": 0.5, "date": 0.5555555555555556, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1085}, "4": {"facts_total": 64, "facts_retained": 22, "retention_rate": 0.34375, "retention_by_kind": {"person": 0.5, "date": 0.5555555555555556, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1203}, "5": {"facts_total": 64, "facts_retained": 22, "retention_rate": 0.34375, "retention_by_kind": {"person": 0.5, "date": 0.5555555555555556, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 926}, "6": {"facts_total": 64, "facts_retained": 24, "retention_rate": 0.375, "retention_by_kind": {"person": 0.5, "date": 0.5555555555555556, "amount": 0.0, "status": 0.5, "place": 0.3333333333333333, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1198}, "7": {"facts_total": 64, "facts_retained": 21, "retention_rate": 0.328125, "retention_by_kind": {"person": 0.5, "date": 0.2222222222222222, "amount": 0.0, "status": 0.5, "place": 0.3333333333333333, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 823}, "8": {"facts_total": 64, "facts_retained": 21, "retention_rate": 0.328125, "retention_by_kind": {"person": 0.5, "date": 0.2222222222222222, "amount": 0.0, "status": 0.5, "place": 0.3333333333333333, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 850}}}, "reference_key_id": "keys/synthetic/syn_03018_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "1b6718880d5561636cccbc918423b60943c018b7083fd37c82dd159683aaa309", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "a5769ccf7f32dde70faf3b58", "attempt_id": "40ba8b2fb855", "run_group_id": "E4_core:A4|closed|qwen3b-llama3b:h8", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:14:57.600189Z", "task_family": "synthetic", "document_id": "syn_03018_2000", "document_hash": "d284eb07e27da742b9c2b67abccfdf72e4dd67c3259f44f3b41e34dd3e6fc1ff", "question_id": "q07_negation", "split": "test", "cluster_id": "syn_03018_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "tokenizer_hashes": {"receiver": "b9d8872cf3a99da99e1fc4b0f71232a451632dc49ca4c6771549b4efa5c1c763"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 8, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1773, "query_tokens": 403, "message_tokens": 299, "message_utf8_bytes": 850, "cache_tensor_bytes": null, "serialized_payload_bytes": 850, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 41128.03645886015, "hops": {"1": 6878.585290978663, "2": 5059.307334013283, "3": 6213.180000078864, "4": 4984.887041966431, "5": 5136.557208956219, "6": 4697.909708949737, "7": 4502.605207962915, "8": 3655.0046659540385}, "receiver_prefill_ms": 185.50525000318885, "ttft_ms": 11.514957994222641, "decode_ms": 23.105249973013997}, "total_wall_ms": 41348.49375090562, "ttft_ms": 11.514957994222641, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 2}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "No.", "evidence_artifact": {"ledger": {"facts_total": 64, "facts_retained": 21, "retention_rate": 0.328125, "retention_by_kind": {"person": 0.5, "date": 0.2222222222222222, "amount": 0.0, "status": 0.5, "place": 0.3333333333333333, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 850}, "ledger_by_hop": {"1": {"facts_total": 64, "facts_retained": 22, "retention_rate": 0.34375, "retention_by_kind": {"person": 0.5, "date": 0.5555555555555556, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1181}, "2": {"facts_total": 64, "facts_retained": 22, "retention_rate": 0.34375, "retention_by_kind": {"person": 0.5, "date": 0.5555555555555556, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1085}, "3": {"facts_total": 64, "facts_retained": 22, "retention_rate": 0.34375, "retention_by_kind": {"person": 0.5, "date": 0.5555555555555556, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1085}, "4": {"facts_total": 64, "facts_retained": 22, "retention_rate": 0.34375, "retention_by_kind": {"person": 0.5, "date": 0.5555555555555556, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1203}, "5": {"facts_total": 64, "facts_retained": 22, "retention_rate": 0.34375, "retention_by_kind": {"person": 0.5, "date": 0.5555555555555556, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 926}, "6": {"facts_total": 64, "facts_retained": 24, "retention_rate": 0.375, "retention_by_kind": {"person": 0.5, "date": 0.5555555555555556, "amount": 0.0, "status": 0.5, "place": 0.3333333333333333, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1198}, "7": {"facts_total": 64, "facts_retained": 21, "retention_rate": 0.328125, "retention_by_kind": {"person": 0.5, "date": 0.2222222222222222, "amount": 0.0, "status": 0.5, "place": 0.3333333333333333, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 823}, "8": {"facts_total": 64, "facts_retained": 21, "retention_rate": 0.328125, "retention_by_kind": {"person": 0.5, "date": 0.2222222222222222, "amount": 0.0, "status": 0.5, "place": 0.3333333333333333, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 850}}}, "reference_key_id": "keys/synthetic/syn_03018_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "1b6718880d5561636cccbc918423b60943c018b7083fd37c82dd159683aaa309", "output_hash": "38178a20b470cfd18299fb1593dd4ba706f1a83321dac051d01086efd8b7a96f"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "b462f85675a3c0c131cf0489", "attempt_id": "8477fd63e264", "run_group_id": "E4_core:A4|closed|qwen3b-llama3b:h8", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:14:57.835194Z", "task_family": "synthetic", "document_id": "syn_03018_2000", "document_hash": "d284eb07e27da742b9c2b67abccfdf72e4dd67c3259f44f3b41e34dd3e6fc1ff", "question_id": "q08_revision", "split": "test", "cluster_id": "syn_03018_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "tokenizer_hashes": {"receiver": "b9d8872cf3a99da99e1fc4b0f71232a451632dc49ca4c6771549b4efa5c1c763"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 8, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1773, "query_tokens": 419, "message_tokens": 299, "message_utf8_bytes": 850, "cache_tensor_bytes": null, "serialized_payload_bytes": 850, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 41128.03645886015, "hops": {"1": 6878.585290978663, "2": 5059.307334013283, "3": 6213.180000078864, "4": 4984.887041966431, "5": 5136.557208956219, "6": 4697.909708949737, "7": 4502.605207962915, "8": 3655.0046659540385}, "receiver_prefill_ms": 183.27083403710276, "ttft_ms": 11.53666595928371, "decode_ms": 34.4559580553323}, "total_wall_ms": 41357.6728758635, "ttft_ms": 11.53666595928371, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 3}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "276,340", "evidence_artifact": {"ledger": {"facts_total": 64, "facts_retained": 21, "retention_rate": 0.328125, "retention_by_kind": {"person": 0.5, "date": 0.2222222222222222, "amount": 0.0, "status": 0.5, "place": 0.3333333333333333, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 850}, "ledger_by_hop": {"1": {"facts_total": 64, "facts_retained": 22, "retention_rate": 0.34375, "retention_by_kind": {"person": 0.5, "date": 0.5555555555555556, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1181}, "2": {"facts_total": 64, "facts_retained": 22, "retention_rate": 0.34375, "retention_by_kind": {"person": 0.5, "date": 0.5555555555555556, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1085}, "3": {"facts_total": 64, "facts_retained": 22, "retention_rate": 0.34375, "retention_by_kind": {"person": 0.5, "date": 0.5555555555555556, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1085}, "4": {"facts_total": 64, "facts_retained": 22, "retention_rate": 0.34375, "retention_by_kind": {"person": 0.5, "date": 0.5555555555555556, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1203}, "5": {"facts_total": 64, "facts_retained": 22, "retention_rate": 0.34375, "retention_by_kind": {"person": 0.5, "date": 0.5555555555555556, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 926}, "6": {"facts_total": 64, "facts_retained": 24, "retention_rate": 0.375, "retention_by_kind": {"person": 0.5, "date": 0.5555555555555556, "amount": 0.0, "status": 0.5, "place": 0.3333333333333333, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1198}, "7": {"facts_total": 64, "facts_retained": 21, "retention_rate": 0.328125, "retention_by_kind": {"person": 0.5, "date": 0.2222222222222222, "amount": 0.0, "status": 0.5, "place": 0.3333333333333333, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 823}, "8": {"facts_total": 64, "facts_retained": 21, "retention_rate": 0.328125, "retention_by_kind": {"person": 0.5, "date": 0.2222222222222222, "amount": 0.0, "status": 0.5, "place": 0.3333333333333333, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 850}}}, "reference_key_id": "keys/synthetic/syn_03018_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "1b6718880d5561636cccbc918423b60943c018b7083fd37c82dd159683aaa309", "output_hash": "ef0dffffbd6c962fe44611aa2052fb7b78efaa1203b0a5de9937b8bd584d71c2"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "a7ae311dd3b5830e90c9ba42", "attempt_id": "7e76b68685e0", "run_group_id": "E4_core:A4|closed|qwen3b-llama3b:h8", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:14:58.079366Z", "task_family": "synthetic", "document_id": "syn_03018_2000", "document_hash": "d284eb07e27da742b9c2b67abccfdf72e4dd67c3259f44f3b41e34dd3e6fc1ff", "question_id": "q09_revision", "split": "test", "cluster_id": "syn_03018_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "tokenizer_hashes": {"receiver": "b9d8872cf3a99da99e1fc4b0f71232a451632dc49ca4c6771549b4efa5c1c763"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 8, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1773, "query_tokens": 412, "message_tokens": 299, "message_utf8_bytes": 850, "cache_tensor_bytes": null, "serialized_payload_bytes": 850, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 41128.03645886015, "hops": {"1": 6878.585290978663, "2": 5059.307334013283, "3": 6213.180000078864, "4": 4984.887041966431, "5": 5136.557208956219, "6": 4697.909708949737, "7": 4502.605207962915, "8": 3655.0046659540385}, "receiver_prefill_ms": 184.90729096811265, "ttft_ms": 11.817832943052053, "decode_ms": 34.619375015608966}, "total_wall_ms": 41359.843917773105, "ttft_ms": 11.817832943052053, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 3}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "276,340", "evidence_artifact": {"ledger": {"facts_total": 64, "facts_retained": 21, "retention_rate": 0.328125, "retention_by_kind": {"person": 0.5, "date": 0.2222222222222222, "amount": 0.0, "status": 0.5, "place": 0.3333333333333333, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 850}, "ledger_by_hop": {"1": {"facts_total": 64, "facts_retained": 22, "retention_rate": 0.34375, "retention_by_kind": {"person": 0.5, "date": 0.5555555555555556, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1181}, "2": {"facts_total": 64, "facts_retained": 22, "retention_rate": 0.34375, "retention_by_kind": {"person": 0.5, "date": 0.5555555555555556, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1085}, "3": {"facts_total": 64, "facts_retained": 22, "retention_rate": 0.34375, "retention_by_kind": {"person": 0.5, "date": 0.5555555555555556, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1085}, "4": {"facts_total": 64, "facts_retained": 22, "retention_rate": 0.34375, "retention_by_kind": {"person": 0.5, "date": 0.5555555555555556, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1203}, "5": {"facts_total": 64, "facts_retained": 22, "retention_rate": 0.34375, "retention_by_kind": {"person": 0.5, "date": 0.5555555555555556, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 926}, "6": {"facts_total": 64, "facts_retained": 24, "retention_rate": 0.375, "retention_by_kind": {"person": 0.5, "date": 0.5555555555555556, "amount": 0.0, "status": 0.5, "place": 0.3333333333333333, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1198}, "7": {"facts_total": 64, "facts_retained": 21, "retention_rate": 0.328125, "retention_by_kind": {"person": 0.5, "date": 0.2222222222222222, "amount": 0.0, "status": 0.5, "place": 0.3333333333333333, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 823}, "8": {"facts_total": 64, "facts_retained": 21, "retention_rate": 0.328125, "retention_by_kind": {"person": 0.5, "date": 0.2222222222222222, "amount": 0.0, "status": 0.5, "place": 0.3333333333333333, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 850}}}, "reference_key_id": "keys/synthetic/syn_03018_2000.json", "task_scores": {"correct": 0.0, "abstained": 0.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "1b6718880d5561636cccbc918423b60943c018b7083fd37c82dd159683aaa309", "output_hash": "ef0dffffbd6c962fe44611aa2052fb7b78efaa1203b0a5de9937b8bd584d71c2"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "0c60d84be0b13e54e8b1de34", "attempt_id": "b2d8bf032787", "run_group_id": "E4_core:A4|closed|qwen3b-llama3b:h8", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:14:58.326343Z", "task_family": "synthetic", "document_id": "syn_03018_2000", "document_hash": "d284eb07e27da742b9c2b67abccfdf72e4dd67c3259f44f3b41e34dd3e6fc1ff", "question_id": "q10_temporal", "split": "test", "cluster_id": "syn_03018_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "tokenizer_hashes": {"receiver": "b9d8872cf3a99da99e1fc4b0f71232a451632dc49ca4c6771549b4efa5c1c763"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 8, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1773, "query_tokens": 420, "message_tokens": 299, "message_utf8_bytes": 850, "cache_tensor_bytes": null, "serialized_payload_bytes": 850, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 41128.03645886015, "hops": {"1": 6878.585290978663, "2": 5059.307334013283, "3": 6213.180000078864, "4": 4984.887041966431, "5": 5136.557208956219, "6": 4697.909708949737, "7": 4502.605207962915, "8": 3655.0046659540385}, "receiver_prefill_ms": 183.75370895955712, "ttft_ms": 11.610083049163222, "decode_ms": 11.532415985129774}, "total_wall_ms": 41335.303375846706, "ttft_ms": 11.610083049163222, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 64, "facts_retained": 21, "retention_rate": 0.328125, "retention_by_kind": {"person": 0.5, "date": 0.2222222222222222, "amount": 0.0, "status": 0.5, "place": 0.3333333333333333, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 850}, "ledger_by_hop": {"1": {"facts_total": 64, "facts_retained": 22, "retention_rate": 0.34375, "retention_by_kind": {"person": 0.5, "date": 0.5555555555555556, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1181}, "2": {"facts_total": 64, "facts_retained": 22, "retention_rate": 0.34375, "retention_by_kind": {"person": 0.5, "date": 0.5555555555555556, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1085}, "3": {"facts_total": 64, "facts_retained": 22, "retention_rate": 0.34375, "retention_by_kind": {"person": 0.5, "date": 0.5555555555555556, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1085}, "4": {"facts_total": 64, "facts_retained": 22, "retention_rate": 0.34375, "retention_by_kind": {"person": 0.5, "date": 0.5555555555555556, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1203}, "5": {"facts_total": 64, "facts_retained": 22, "retention_rate": 0.34375, "retention_by_kind": {"person": 0.5, "date": 0.5555555555555556, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 926}, "6": {"facts_total": 64, "facts_retained": 24, "retention_rate": 0.375, "retention_by_kind": {"person": 0.5, "date": 0.5555555555555556, "amount": 0.0, "status": 0.5, "place": 0.3333333333333333, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1198}, "7": {"facts_total": 64, "facts_retained": 21, "retention_rate": 0.328125, "retention_by_kind": {"person": 0.5, "date": 0.2222222222222222, "amount": 0.0, "status": 0.5, "place": 0.3333333333333333, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 823}, "8": {"facts_total": 64, "facts_retained": 21, "retention_rate": 0.328125, "retention_by_kind": {"person": 0.5, "date": 0.2222222222222222, "amount": 0.0, "status": 0.5, "place": 0.3333333333333333, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 850}}}, "reference_key_id": "keys/synthetic/syn_03018_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "1b6718880d5561636cccbc918423b60943c018b7083fd37c82dd159683aaa309", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "bd1c5cacb818e438dcb30919", "attempt_id": "f209d50142e4", "run_group_id": "E4_core:A4|closed|qwen3b-llama3b:h8", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:14:58.548745Z", "task_family": "synthetic", "document_id": "syn_03018_2000", "document_hash": "d284eb07e27da742b9c2b67abccfdf72e4dd67c3259f44f3b41e34dd3e6fc1ff", "question_id": "q11_aggregation", "split": "test", "cluster_id": "syn_03018_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "tokenizer_hashes": {"receiver": "b9d8872cf3a99da99e1fc4b0f71232a451632dc49ca4c6771549b4efa5c1c763"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 8, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1773, "query_tokens": 414, "message_tokens": 299, "message_utf8_bytes": 850, "cache_tensor_bytes": null, "serialized_payload_bytes": 850, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 41128.03645886015, "hops": {"1": 6878.585290978663, "2": 5059.307334013283, "3": 6213.180000078864, "4": 4984.887041966431, "5": 5136.557208956219, "6": 4697.909708949737, "7": 4502.605207962915, "8": 3655.0046659540385}, "receiver_prefill_ms": 184.81066601816565, "ttft_ms": 11.759291985072196, "decode_ms": 539.9706249590963}, "total_wall_ms": 41864.972333889455, "ttft_ms": 11.759291985072196, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 48}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "2,917,061 + 272,184 + 276,340 + 304,815 + 9,472,675 + 10,352,320 + 8,340,685 = 32,193,380", "evidence_artifact": {"ledger": {"facts_total": 64, "facts_retained": 21, "retention_rate": 0.328125, "retention_by_kind": {"person": 0.5, "date": 0.2222222222222222, "amount": 0.0, "status": 0.5, "place": 0.3333333333333333, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 850}, "ledger_by_hop": {"1": {"facts_total": 64, "facts_retained": 22, "retention_rate": 0.34375, "retention_by_kind": {"person": 0.5, "date": 0.5555555555555556, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1181}, "2": {"facts_total": 64, "facts_retained": 22, "retention_rate": 0.34375, "retention_by_kind": {"person": 0.5, "date": 0.5555555555555556, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1085}, "3": {"facts_total": 64, "facts_retained": 22, "retention_rate": 0.34375, "retention_by_kind": {"person": 0.5, "date": 0.5555555555555556, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1085}, "4": {"facts_total": 64, "facts_retained": 22, "retention_rate": 0.34375, "retention_by_kind": {"person": 0.5, "date": 0.5555555555555556, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1203}, "5": {"facts_total": 64, "facts_retained": 22, "retention_rate": 0.34375, "retention_by_kind": {"person": 0.5, "date": 0.5555555555555556, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 926}, "6": {"facts_total": 64, "facts_retained": 24, "retention_rate": 0.375, "retention_by_kind": {"person": 0.5, "date": 0.5555555555555556, "amount": 0.0, "status": 0.5, "place": 0.3333333333333333, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1198}, "7": {"facts_total": 64, "facts_retained": 21, "retention_rate": 0.328125, "retention_by_kind": {"person": 0.5, "date": 0.2222222222222222, "amount": 0.0, "status": 0.5, "place": 0.3333333333333333, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 823}, "8": {"facts_total": 64, "facts_retained": 21, "retention_rate": 0.328125, "retention_by_kind": {"person": 0.5, "date": 0.2222222222222222, "amount": 0.0, "status": 0.5, "place": 0.3333333333333333, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 850}}}, "reference_key_id": "keys/synthetic/syn_03018_2000.json", "task_scores": {"correct": 0.0, "abstained": 0.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "1b6718880d5561636cccbc918423b60943c018b7083fd37c82dd159683aaa309", "output_hash": "df9f9c81da80a26bbeec3f59a54e22b775bf48844ab6587599318c3c6247ef9e"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "0f4f2c9622788b00e397d94a", "attempt_id": "0637f5f15715", "run_group_id": "E4_core:A4|closed|qwen3b-llama3b:h8", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:14:59.300422Z", "task_family": "synthetic", "document_id": "syn_03018_2000", "document_hash": "d284eb07e27da742b9c2b67abccfdf72e4dd67c3259f44f3b41e34dd3e6fc1ff", "question_id": "q12_unit_conversion", "split": "test", "cluster_id": "syn_03018_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "tokenizer_hashes": {"receiver": "b9d8872cf3a99da99e1fc4b0f71232a451632dc49ca4c6771549b4efa5c1c763"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 8, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1773, "query_tokens": 410, "message_tokens": 299, "message_utf8_bytes": 850, "cache_tensor_bytes": null, "serialized_payload_bytes": 850, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 41128.03645886015, "hops": {"1": 6878.585290978663, "2": 5059.307334013283, "3": 6213.180000078864, "4": 4984.887041966431, "5": 5136.557208956219, "6": 4697.909708949737, "7": 4502.605207962915, "8": 3655.0046659540385}, "receiver_prefill_ms": 182.25987500045449, "ttft_ms": 11.606249958276749, "decode_ms": 11.51204202324152}, "total_wall_ms": 41333.78220885061, "ttft_ms": 11.606249958276749, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 64, "facts_retained": 21, "retention_rate": 0.328125, "retention_by_kind": {"person": 0.5, "date": 0.2222222222222222, "amount": 0.0, "status": 0.5, "place": 0.3333333333333333, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 850}, "ledger_by_hop": {"1": {"facts_total": 64, "facts_retained": 22, "retention_rate": 0.34375, "retention_by_kind": {"person": 0.5, "date": 0.5555555555555556, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1181}, "2": {"facts_total": 64, "facts_retained": 22, "retention_rate": 0.34375, "retention_by_kind": {"person": 0.5, "date": 0.5555555555555556, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1085}, "3": {"facts_total": 64, "facts_retained": 22, "retention_rate": 0.34375, "retention_by_kind": {"person": 0.5, "date": 0.5555555555555556, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1085}, "4": {"facts_total": 64, "facts_retained": 22, "retention_rate": 0.34375, "retention_by_kind": {"person": 0.5, "date": 0.5555555555555556, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1203}, "5": {"facts_total": 64, "facts_retained": 22, "retention_rate": 0.34375, "retention_by_kind": {"person": 0.5, "date": 0.5555555555555556, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 926}, "6": {"facts_total": 64, "facts_retained": 24, "retention_rate": 0.375, "retention_by_kind": {"person": 0.5, "date": 0.5555555555555556, "amount": 0.0, "status": 0.5, "place": 0.3333333333333333, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1198}, "7": {"facts_total": 64, "facts_retained": 21, "retention_rate": 0.328125, "retention_by_kind": {"person": 0.5, "date": 0.2222222222222222, "amount": 0.0, "status": 0.5, "place": 0.3333333333333333, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 823}, "8": {"facts_total": 64, "facts_retained": 21, "retention_rate": 0.328125, "retention_by_kind": {"person": 0.5, "date": 0.2222222222222222, "amount": 0.0, "status": 0.5, "place": 0.3333333333333333, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 850}}}, "reference_key_id": "keys/synthetic/syn_03018_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "1b6718880d5561636cccbc918423b60943c018b7083fd37c82dd159683aaa309", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "541045f9a70b4e6e660c7c88", "attempt_id": "482f86a5e36e", "run_group_id": "E4_core:A4|closed|qwen3b-llama3b:h8", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:14:59.520998Z", "task_family": "synthetic", "document_id": "syn_03018_2000", "document_hash": "d284eb07e27da742b9c2b67abccfdf72e4dd67c3259f44f3b41e34dd3e6fc1ff", "question_id": "q13_unanswerable", "split": "test", "cluster_id": "syn_03018_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "tokenizer_hashes": {"receiver": "b9d8872cf3a99da99e1fc4b0f71232a451632dc49ca4c6771549b4efa5c1c763"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 8, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1773, "query_tokens": 415, "message_tokens": 299, "message_utf8_bytes": 850, "cache_tensor_bytes": null, "serialized_payload_bytes": 850, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 41128.03645886015, "hops": {"1": 6878.585290978663, "2": 5059.307334013283, "3": 6213.180000078864, "4": 4984.887041966431, "5": 5136.557208956219, "6": 4697.909708949737, "7": 4502.605207962915, "8": 3655.0046659540385}, "receiver_prefill_ms": 184.33529196772724, "ttft_ms": 11.611916008405387, "decode_ms": 11.75508403684944}, "total_wall_ms": 41336.10725088511, "ttft_ms": 11.611916008405387, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 64, "facts_retained": 21, "retention_rate": 0.328125, "retention_by_kind": {"person": 0.5, "date": 0.2222222222222222, "amount": 0.0, "status": 0.5, "place": 0.3333333333333333, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 850}, "ledger_by_hop": {"1": {"facts_total": 64, "facts_retained": 22, "retention_rate": 0.34375, "retention_by_kind": {"person": 0.5, "date": 0.5555555555555556, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1181}, "2": {"facts_total": 64, "facts_retained": 22, "retention_rate": 0.34375, "retention_by_kind": {"person": 0.5, "date": 0.5555555555555556, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1085}, "3": {"facts_total": 64, "facts_retained": 22, "retention_rate": 0.34375, "retention_by_kind": {"person": 0.5, "date": 0.5555555555555556, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1085}, "4": {"facts_total": 64, "facts_retained": 22, "retention_rate": 0.34375, "retention_by_kind": {"person": 0.5, "date": 0.5555555555555556, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1203}, "5": {"facts_total": 64, "facts_retained": 22, "retention_rate": 0.34375, "retention_by_kind": {"person": 0.5, "date": 0.5555555555555556, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 926}, "6": {"facts_total": 64, "facts_retained": 24, "retention_rate": 0.375, "retention_by_kind": {"person": 0.5, "date": 0.5555555555555556, "amount": 0.0, "status": 0.5, "place": 0.3333333333333333, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1198}, "7": {"facts_total": 64, "facts_retained": 21, "retention_rate": 0.328125, "retention_by_kind": {"person": 0.5, "date": 0.2222222222222222, "amount": 0.0, "status": 0.5, "place": 0.3333333333333333, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 823}, "8": {"facts_total": 64, "facts_retained": 21, "retention_rate": 0.328125, "retention_by_kind": {"person": 0.5, "date": 0.2222222222222222, "amount": 0.0, "status": 0.5, "place": 0.3333333333333333, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 850}}}, "reference_key_id": "keys/synthetic/syn_03018_2000.json", "task_scores": {"correct": 1.0, "abstained": 1.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "1b6718880d5561636cccbc918423b60943c018b7083fd37c82dd159683aaa309", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "2b1f7a6b4e8cca6da4a25c38", "attempt_id": "ce0c6593d703", "run_group_id": "E4_core:A4|closed|qwen3b-llama3b:h8", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:14:59.743730Z", "task_family": "synthetic", "document_id": "syn_03018_2000", "document_hash": "d284eb07e27da742b9c2b67abccfdf72e4dd67c3259f44f3b41e34dd3e6fc1ff", "question_id": "q14_unanswerable", "split": "test", "cluster_id": "syn_03018_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "tokenizer_hashes": {"receiver": "b9d8872cf3a99da99e1fc4b0f71232a451632dc49ca4c6771549b4efa5c1c763"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 8, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1773, "query_tokens": 416, "message_tokens": 299, "message_utf8_bytes": 850, "cache_tensor_bytes": null, "serialized_payload_bytes": 850, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 41128.03645886015, "hops": {"1": 6878.585290978663, "2": 5059.307334013283, "3": 6213.180000078864, "4": 4984.887041966431, "5": 5136.557208956219, "6": 4697.909708949737, "7": 4502.605207962915, "8": 3655.0046659540385}, "receiver_prefill_ms": 185.90520904399455, "ttft_ms": 11.744167073629797, "decode_ms": 11.569540947675705}, "total_wall_ms": 41337.60004187934, "ttft_ms": 11.744167073629797, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 64, "facts_retained": 21, "retention_rate": 0.328125, "retention_by_kind": {"person": 0.5, "date": 0.2222222222222222, "amount": 0.0, "status": 0.5, "place": 0.3333333333333333, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 850}, "ledger_by_hop": {"1": {"facts_total": 64, "facts_retained": 22, "retention_rate": 0.34375, "retention_by_kind": {"person": 0.5, "date": 0.5555555555555556, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1181}, "2": {"facts_total": 64, "facts_retained": 22, "retention_rate": 0.34375, "retention_by_kind": {"person": 0.5, "date": 0.5555555555555556, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1085}, "3": {"facts_total": 64, "facts_retained": 22, "retention_rate": 0.34375, "retention_by_kind": {"person": 0.5, "date": 0.5555555555555556, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1085}, "4": {"facts_total": 64, "facts_retained": 22, "retention_rate": 0.34375, "retention_by_kind": {"person": 0.5, "date": 0.5555555555555556, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1203}, "5": {"facts_total": 64, "facts_retained": 22, "retention_rate": 0.34375, "retention_by_kind": {"person": 0.5, "date": 0.5555555555555556, "amount": 0.0, "status": 0.5, "place": 0.0, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 926}, "6": {"facts_total": 64, "facts_retained": 24, "retention_rate": 0.375, "retention_by_kind": {"person": 0.5, "date": 0.5555555555555556, "amount": 0.0, "status": 0.5, "place": 0.3333333333333333, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1198}, "7": {"facts_total": 64, "facts_retained": 21, "retention_rate": 0.328125, "retention_by_kind": {"person": 0.5, "date": 0.2222222222222222, "amount": 0.0, "status": 0.5, "place": 0.3333333333333333, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 823}, "8": {"facts_total": 64, "facts_retained": 21, "retention_rate": 0.328125, "retention_by_kind": {"person": 0.5, "date": 0.2222222222222222, "amount": 0.0, "status": 0.5, "place": 0.3333333333333333, "identifier": 0.4166666666666667, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 850}}}, "reference_key_id": "keys/synthetic/syn_03018_2000.json", "task_scores": {"correct": 1.0, "abstained": 1.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "1b6718880d5561636cccbc918423b60943c018b7083fd37c82dd159683aaa309", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "ef9d59d0d730f119dee3e537", "attempt_id": "0d7ad5d34b8a", "run_group_id": "E4_core:A5|closed|qwen3b-llama3b:h1", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:15:06.899078Z", "task_family": "synthetic", "document_id": "syn_03018_2000", "document_hash": "d284eb07e27da742b9c2b67abccfdf72e4dd67c3259f44f3b41e34dd3e6fc1ff", "question_id": "q00_single_fact", "split": "test", "cluster_id": "syn_03018_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 1, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1982, "query_tokens": 484, "message_tokens": 401, "message_utf8_bytes": 940, "cache_tensor_bytes": null, "serialized_payload_bytes": 940, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 6929.263000027277, "hops": {"1": 6929.263000027277}, "receiver_prefill_ms": 187.67795898020267, "ttft_ms": 11.746917036361992, "decode_ms": 11.756666004657745}, "total_wall_ms": 7140.908292029053, "ttft_ms": 11.746917036361992, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 64, "facts_retained": 14, "retention_rate": 0.21875, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.2222222222222222, "amount": 0.15384615384615385, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 870}, "ledger_by_hop": {"1": {"facts_total": 64, "facts_retained": 14, "retention_rate": 0.21875, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.2222222222222222, "amount": 0.15384615384615385, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 870}}}, "reference_key_id": "keys/synthetic/syn_03018_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "b93ce255451151cce89ccd51cc89c5b4382f883b441b3818bc249f2f79c54d9e", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "e011fd1d8572a231f09ce2d6", "attempt_id": "2164210ffdb2", "run_group_id": "E4_core:A5|closed|qwen3b-llama3b:h1", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:15:07.126567Z", "task_family": "synthetic", "document_id": "syn_03018_2000", "document_hash": "d284eb07e27da742b9c2b67abccfdf72e4dd67c3259f44f3b41e34dd3e6fc1ff", "question_id": "q01_single_fact", "split": "test", "cluster_id": "syn_03018_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 1, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1982, "query_tokens": 490, "message_tokens": 401, "message_utf8_bytes": 940, "cache_tensor_bytes": null, "serialized_payload_bytes": 940, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 6929.263000027277, "hops": {"1": 6929.263000027277}, "receiver_prefill_ms": 187.61270795948803, "ttft_ms": 11.777999927289784, "decode_ms": 11.730582918971777}, "total_wall_ms": 7140.8066670410335, "ttft_ms": 11.777999927289784, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 64, "facts_retained": 14, "retention_rate": 0.21875, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.2222222222222222, "amount": 0.15384615384615385, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 870}, "ledger_by_hop": {"1": {"facts_total": 64, "facts_retained": 14, "retention_rate": 0.21875, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.2222222222222222, "amount": 0.15384615384615385, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 870}}}, "reference_key_id": "keys/synthetic/syn_03018_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "b93ce255451151cce89ccd51cc89c5b4382f883b441b3818bc249f2f79c54d9e", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "39d0dc92c8838dcb2310fa7e", "attempt_id": "951ba3b0c71d", "run_group_id": "E4_core:A5|closed|qwen3b-llama3b:h1", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:15:07.353481Z", "task_family": "synthetic", "document_id": "syn_03018_2000", "document_hash": "d284eb07e27da742b9c2b67abccfdf72e4dd67c3259f44f3b41e34dd3e6fc1ff", "question_id": "q02_exact_copy", "split": "test", "cluster_id": "syn_03018_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 1, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1982, "query_tokens": 489, "message_tokens": 401, "message_utf8_bytes": 940, "cache_tensor_bytes": null, "serialized_payload_bytes": 940, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 6929.263000027277, "hops": {"1": 6929.263000027277}, "receiver_prefill_ms": 187.96954199206084, "ttft_ms": 11.712457984685898, "decode_ms": 11.775666964240372}, "total_wall_ms": 7141.115084057674, "ttft_ms": 11.712457984685898, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 64, "facts_retained": 14, "retention_rate": 0.21875, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.2222222222222222, "amount": 0.15384615384615385, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 870}, "ledger_by_hop": {"1": {"facts_total": 64, "facts_retained": 14, "retention_rate": 0.21875, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.2222222222222222, "amount": 0.15384615384615385, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 870}}}, "reference_key_id": "keys/synthetic/syn_03018_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "b93ce255451151cce89ccd51cc89c5b4382f883b441b3818bc249f2f79c54d9e", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "30563accebfd06be9129dc88", "attempt_id": "19a481380ba2", "run_group_id": "E4_core:A5|closed|qwen3b-llama3b:h1", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:15:07.580796Z", "task_family": "synthetic", "document_id": "syn_03018_2000", "document_hash": "d284eb07e27da742b9c2b67abccfdf72e4dd67c3259f44f3b41e34dd3e6fc1ff", "question_id": "q03_exact_copy", "split": "test", "cluster_id": "syn_03018_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 1, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1982, "query_tokens": 487, "message_tokens": 401, "message_utf8_bytes": 940, "cache_tensor_bytes": null, "serialized_payload_bytes": 940, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 6929.263000027277, "hops": {"1": 6929.263000027277}, "receiver_prefill_ms": 184.11799997556955, "ttft_ms": 11.852291994728148, "decode_ms": 105.4289999883622}, "total_wall_ms": 7231.058708042838, "ttft_ms": 11.852291994728148, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 9}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "Priyamvada Sørlie", "evidence_artifact": {"ledger": {"facts_total": 64, "facts_retained": 14, "retention_rate": 0.21875, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.2222222222222222, "amount": 0.15384615384615385, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 870}, "ledger_by_hop": {"1": {"facts_total": 64, "facts_retained": 14, "retention_rate": 0.21875, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.2222222222222222, "amount": 0.15384615384615385, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 870}}}, "reference_key_id": "keys/synthetic/syn_03018_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "b93ce255451151cce89ccd51cc89c5b4382f883b441b3818bc249f2f79c54d9e", "output_hash": "c5f554196ed23dd71ad995ec89f7142099d524dbb6534cc10d94a0c89d13e792"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "41502a237320af2b4e8452f4", "attempt_id": "c4e11a7bf16e", "run_group_id": "E4_core:A5|closed|qwen3b-llama3b:h1", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:15:07.897827Z", "task_family": "synthetic", "document_id": "syn_03018_2000", "document_hash": "d284eb07e27da742b9c2b67abccfdf72e4dd67c3259f44f3b41e34dd3e6fc1ff", "question_id": "q04_multi_hop", "split": "test", "cluster_id": "syn_03018_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 1, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1982, "query_tokens": 493, "message_tokens": 401, "message_utf8_bytes": 940, "cache_tensor_bytes": null, "serialized_payload_bytes": 940, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 6929.263000027277, "hops": {"1": 6929.263000027277}, "receiver_prefill_ms": 187.8489590017125, "ttft_ms": 11.719416943378747, "decode_ms": 11.778000043705106}, "total_wall_ms": 7140.995250083506, "ttft_ms": 11.719416943378747, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 64, "facts_retained": 14, "retention_rate": 0.21875, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.2222222222222222, "amount": 0.15384615384615385, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 870}, "ledger_by_hop": {"1": {"facts_total": 64, "facts_retained": 14, "retention_rate": 0.21875, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.2222222222222222, "amount": 0.15384615384615385, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 870}}}, "reference_key_id": "keys/synthetic/syn_03018_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "b93ce255451151cce89ccd51cc89c5b4382f883b441b3818bc249f2f79c54d9e", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "407ce17ec735e0f57421d1d2", "attempt_id": "d40fa8ba4d41", "run_group_id": "E4_core:A5|closed|qwen3b-llama3b:h1", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:15:08.124280Z", "task_family": "synthetic", "document_id": "syn_03018_2000", "document_hash": "d284eb07e27da742b9c2b67abccfdf72e4dd67c3259f44f3b41e34dd3e6fc1ff", "question_id": "q05_multi_hop", "split": "test", "cluster_id": "syn_03018_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 1, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1982, "query_tokens": 493, "message_tokens": 401, "message_utf8_bytes": 940, "cache_tensor_bytes": null, "serialized_payload_bytes": 940, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 6929.263000027277, "hops": {"1": 6929.263000027277}, "receiver_prefill_ms": 187.815707991831, "ttft_ms": 12.055499944835901, "decode_ms": 11.991166975349188}, "total_wall_ms": 7141.6594160255045, "ttft_ms": 12.055499944835901, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 64, "facts_retained": 14, "retention_rate": 0.21875, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.2222222222222222, "amount": 0.15384615384615385, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 870}, "ledger_by_hop": {"1": {"facts_total": 64, "facts_retained": 14, "retention_rate": 0.21875, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.2222222222222222, "amount": 0.15384615384615385, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 870}}}, "reference_key_id": "keys/synthetic/syn_03018_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "b93ce255451151cce89ccd51cc89c5b4382f883b441b3818bc249f2f79c54d9e", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "355452ad57790ae882eb8971", "attempt_id": "36c45934b36b", "run_group_id": "E4_core:A5|closed|qwen3b-llama3b:h1", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:15:08.351643Z", "task_family": "synthetic", "document_id": "syn_03018_2000", "document_hash": "d284eb07e27da742b9c2b67abccfdf72e4dd67c3259f44f3b41e34dd3e6fc1ff", "question_id": "q06_negation", "split": "test", "cluster_id": "syn_03018_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 1, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1982, "query_tokens": 485, "message_tokens": 401, "message_utf8_bytes": 940, "cache_tensor_bytes": null, "serialized_payload_bytes": 940, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 6929.263000027277, "hops": {"1": 6929.263000027277}, "receiver_prefill_ms": 188.0054579814896, "ttft_ms": 11.848166002891958, "decode_ms": 11.809833929874003}, "total_wall_ms": 7141.393958008848, "ttft_ms": 11.848166002891958, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 64, "facts_retained": 14, "retention_rate": 0.21875, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.2222222222222222, "amount": 0.15384615384615385, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 870}, "ledger_by_hop": {"1": {"facts_total": 64, "facts_retained": 14, "retention_rate": 0.21875, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.2222222222222222, "amount": 0.15384615384615385, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 870}}}, "reference_key_id": "keys/synthetic/syn_03018_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "b93ce255451151cce89ccd51cc89c5b4382f883b441b3818bc249f2f79c54d9e", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "bb8c98bb3041221ad850946e", "attempt_id": "409e9e025204", "run_group_id": "E4_core:A5|closed|qwen3b-llama3b:h1", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:15:08.579907Z", "task_family": "synthetic", "document_id": "syn_03018_2000", "document_hash": "d284eb07e27da742b9c2b67abccfdf72e4dd67c3259f44f3b41e34dd3e6fc1ff", "question_id": "q07_negation", "split": "test", "cluster_id": "syn_03018_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 1, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1982, "query_tokens": 484, "message_tokens": 401, "message_utf8_bytes": 940, "cache_tensor_bytes": null, "serialized_payload_bytes": 940, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 6929.263000027277, "hops": {"1": 6929.263000027277}, "receiver_prefill_ms": 187.9180419491604, "ttft_ms": 11.844624998047948, "decode_ms": 11.811083066277206}, "total_wall_ms": 7141.232416965067, "ttft_ms": 11.844624998047948, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 64, "facts_retained": 14, "retention_rate": 0.21875, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.2222222222222222, "amount": 0.15384615384615385, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 870}, "ledger_by_hop": {"1": {"facts_total": 64, "facts_retained": 14, "retention_rate": 0.21875, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.2222222222222222, "amount": 0.15384615384615385, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 870}}}, "reference_key_id": "keys/synthetic/syn_03018_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "b93ce255451151cce89ccd51cc89c5b4382f883b441b3818bc249f2f79c54d9e", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "17834dcc4bda22518a0e5089", "attempt_id": "b5afc5519827", "run_group_id": "E4_core:A5|closed|qwen3b-llama3b:h1", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:15:08.807185Z", "task_family": "synthetic", "document_id": "syn_03018_2000", "document_hash": "d284eb07e27da742b9c2b67abccfdf72e4dd67c3259f44f3b41e34dd3e6fc1ff", "question_id": "q08_revision", "split": "test", "cluster_id": "syn_03018_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 1, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1982, "query_tokens": 502, "message_tokens": 401, "message_utf8_bytes": 940, "cache_tensor_bytes": null, "serialized_payload_bytes": 940, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 6929.263000027277, "hops": {"1": 6929.263000027277}, "receiver_prefill_ms": 188.47216700669378, "ttft_ms": 12.001291965134442, "decode_ms": 70.52620907779783}, "total_wall_ms": 7200.6918750703335, "ttft_ms": 12.001291965134442, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 6}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "272184", "evidence_artifact": {"ledger": {"facts_total": 64, "facts_retained": 14, "retention_rate": 0.21875, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.2222222222222222, "amount": 0.15384615384615385, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 870}, "ledger_by_hop": {"1": {"facts_total": 64, "facts_retained": 14, "retention_rate": 0.21875, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.2222222222222222, "amount": 0.15384615384615385, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 870}}}, "reference_key_id": "keys/synthetic/syn_03018_2000.json", "task_scores": {"correct": 0.0, "abstained": 0.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "b93ce255451151cce89ccd51cc89c5b4382f883b441b3818bc249f2f79c54d9e", "output_hash": "3aadaddf09c30b8bfe5dc7e2aa6984fcf0df4d216145cf925760dd5a946d6e05"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "56cf7fa6d81e76d1f8515549", "attempt_id": "59ea90d556b5", "run_group_id": "E4_core:A5|closed|qwen3b-llama3b:h1", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:15:09.093657Z", "task_family": "synthetic", "document_id": "syn_03018_2000", "document_hash": "d284eb07e27da742b9c2b67abccfdf72e4dd67c3259f44f3b41e34dd3e6fc1ff", "question_id": "q09_revision", "split": "test", "cluster_id": "syn_03018_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 1, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1982, "query_tokens": 495, "message_tokens": 401, "message_utf8_bytes": 940, "cache_tensor_bytes": null, "serialized_payload_bytes": 940, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 6929.263000027277, "hops": {"1": 6929.263000027277}, "receiver_prefill_ms": 188.18745797034353, "ttft_ms": 11.758583947084844, "decode_ms": 70.63787500374019}, "total_wall_ms": 7200.272500049323, "ttft_ms": 11.758583947084844, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 6}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "272184", "evidence_artifact": {"ledger": {"facts_total": 64, "facts_retained": 14, "retention_rate": 0.21875, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.2222222222222222, "amount": 0.15384615384615385, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 870}, "ledger_by_hop": {"1": {"facts_total": 64, "facts_retained": 14, "retention_rate": 0.21875, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.2222222222222222, "amount": 0.15384615384615385, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 870}}}, "reference_key_id": "keys/synthetic/syn_03018_2000.json", "task_scores": {"correct": 0.0, "abstained": 0.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "b93ce255451151cce89ccd51cc89c5b4382f883b441b3818bc249f2f79c54d9e", "output_hash": "3aadaddf09c30b8bfe5dc7e2aa6984fcf0df4d216145cf925760dd5a946d6e05"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "a5da0b22e75559e1d6d6a036", "attempt_id": "6db2940ab5cf", "run_group_id": "E4_core:A5|closed|qwen3b-llama3b:h1", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:15:09.379670Z", "task_family": "synthetic", "document_id": "syn_03018_2000", "document_hash": "d284eb07e27da742b9c2b67abccfdf72e4dd67c3259f44f3b41e34dd3e6fc1ff", "question_id": "q10_temporal", "split": "test", "cluster_id": "syn_03018_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 1, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1982, "query_tokens": 503, "message_tokens": 401, "message_utf8_bytes": 940, "cache_tensor_bytes": null, "serialized_payload_bytes": 940, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 6929.263000027277, "hops": {"1": 6929.263000027277}, "receiver_prefill_ms": 188.4382499847561, "ttft_ms": 11.818040977232158, "decode_ms": 11.822791071608663}, "total_wall_ms": 7141.743667074479, "ttft_ms": 11.818040977232158, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 64, "facts_retained": 14, "retention_rate": 0.21875, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.2222222222222222, "amount": 0.15384615384615385, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 870}, "ledger_by_hop": {"1": {"facts_total": 64, "facts_retained": 14, "retention_rate": 0.21875, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.2222222222222222, "amount": 0.15384615384615385, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 870}}}, "reference_key_id": "keys/synthetic/syn_03018_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "b93ce255451151cce89ccd51cc89c5b4382f883b441b3818bc249f2f79c54d9e", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "af13f2f6fa630827a56442cd", "attempt_id": "1dfd3979474a", "run_group_id": "E4_core:A5|closed|qwen3b-llama3b:h1", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:15:09.607243Z", "task_family": "synthetic", "document_id": "syn_03018_2000", "document_hash": "d284eb07e27da742b9c2b67abccfdf72e4dd67c3259f44f3b41e34dd3e6fc1ff", "question_id": "q11_aggregation", "split": "test", "cluster_id": "syn_03018_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 1, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1982, "query_tokens": 495, "message_tokens": 401, "message_utf8_bytes": 940, "cache_tensor_bytes": null, "serialized_payload_bytes": 940, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 6929.263000027277, "hops": {"1": 6929.263000027277}, "receiver_prefill_ms": 188.26720805373043, "ttft_ms": 11.799500091001391, "decode_ms": 82.77991705108434}, "total_wall_ms": 7212.496457970701, "ttft_ms": 11.799500091001391, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 7}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "3189244", "evidence_artifact": {"ledger": {"facts_total": 64, "facts_retained": 14, "retention_rate": 0.21875, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.2222222222222222, "amount": 0.15384615384615385, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 870}, "ledger_by_hop": {"1": {"facts_total": 64, "facts_retained": 14, "retention_rate": 0.21875, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.2222222222222222, "amount": 0.15384615384615385, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 870}}}, "reference_key_id": "keys/synthetic/syn_03018_2000.json", "task_scores": {"correct": 0.0, "abstained": 0.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "b93ce255451151cce89ccd51cc89c5b4382f883b441b3818bc249f2f79c54d9e", "output_hash": "953508a9732586a80de7650cf183d3ca74cd590103fa92d25c9b3d80da907f61"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "df1db045ed7f031b60422ecf", "attempt_id": "aa50039336f9", "run_group_id": "E4_core:A5|closed|qwen3b-llama3b:h1", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:15:09.905123Z", "task_family": "synthetic", "document_id": "syn_03018_2000", "document_hash": "d284eb07e27da742b9c2b67abccfdf72e4dd67c3259f44f3b41e34dd3e6fc1ff", "question_id": "q12_unit_conversion", "split": "test", "cluster_id": "syn_03018_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 1, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1982, "query_tokens": 492, "message_tokens": 401, "message_utf8_bytes": 940, "cache_tensor_bytes": null, "serialized_payload_bytes": 940, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 6929.263000027277, "hops": {"1": 6929.263000027277}, "receiver_prefill_ms": 188.76395898405463, "ttft_ms": 11.85541704762727, "decode_ms": 11.912832967936993}, "total_wall_ms": 7142.205458017997, "ttft_ms": 11.85541704762727, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 64, "facts_retained": 14, "retention_rate": 0.21875, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.2222222222222222, "amount": 0.15384615384615385, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 870}, "ledger_by_hop": {"1": {"facts_total": 64, "facts_retained": 14, "retention_rate": 0.21875, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.2222222222222222, "amount": 0.15384615384615385, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 870}}}, "reference_key_id": "keys/synthetic/syn_03018_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "b93ce255451151cce89ccd51cc89c5b4382f883b441b3818bc249f2f79c54d9e", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "b537b10b65942896aa24088e", "attempt_id": "2f5fc4a3c407", "run_group_id": "E4_core:A5|closed|qwen3b-llama3b:h1", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:15:10.132631Z", "task_family": "synthetic", "document_id": "syn_03018_2000", "document_hash": "d284eb07e27da742b9c2b67abccfdf72e4dd67c3259f44f3b41e34dd3e6fc1ff", "question_id": "q13_unanswerable", "split": "test", "cluster_id": "syn_03018_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 1, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1982, "query_tokens": 496, "message_tokens": 401, "message_utf8_bytes": 940, "cache_tensor_bytes": null, "serialized_payload_bytes": 940, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 6929.263000027277, "hops": {"1": 6929.263000027277}, "receiver_prefill_ms": 188.65350005216897, "ttft_ms": 11.902958038263023, "decode_ms": 11.907750042155385}, "total_wall_ms": 7142.128125065938, "ttft_ms": 11.902958038263023, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 64, "facts_retained": 14, "retention_rate": 0.21875, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.2222222222222222, "amount": 0.15384615384615385, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 870}, "ledger_by_hop": {"1": {"facts_total": 64, "facts_retained": 14, "retention_rate": 0.21875, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.2222222222222222, "amount": 0.15384615384615385, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 870}}}, "reference_key_id": "keys/synthetic/syn_03018_2000.json", "task_scores": {"correct": 1.0, "abstained": 1.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "b93ce255451151cce89ccd51cc89c5b4382f883b441b3818bc249f2f79c54d9e", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "8ef8ba4b3c114fcfc15205b6", "attempt_id": "b4378d780c60", "run_group_id": "E4_core:A5|closed|qwen3b-llama3b:h1", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:15:10.360155Z", "task_family": "synthetic", "document_id": "syn_03018_2000", "document_hash": "d284eb07e27da742b9c2b67abccfdf72e4dd67c3259f44f3b41e34dd3e6fc1ff", "question_id": "q14_unanswerable", "split": "test", "cluster_id": "syn_03018_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 1, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1982, "query_tokens": 497, "message_tokens": 401, "message_utf8_bytes": 940, "cache_tensor_bytes": null, "serialized_payload_bytes": 940, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 6929.263000027277, "hops": {"1": 6929.263000027277}, "receiver_prefill_ms": 188.8059169286862, "ttft_ms": 11.928000021725893, "decode_ms": 11.919624987058342}, "total_wall_ms": 7142.308208043687, "ttft_ms": 11.928000021725893, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 64, "facts_retained": 14, "retention_rate": 0.21875, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.2222222222222222, "amount": 0.15384615384615385, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 870}, "ledger_by_hop": {"1": {"facts_total": 64, "facts_retained": 14, "retention_rate": 0.21875, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.2222222222222222, "amount": 0.15384615384615385, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 870}}}, "reference_key_id": "keys/synthetic/syn_03018_2000.json", "task_scores": {"correct": 1.0, "abstained": 1.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "b93ce255451151cce89ccd51cc89c5b4382f883b441b3818bc249f2f79c54d9e", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "b72a4a4a5ff60a327a090d84", "attempt_id": "be0b67c401ac", "run_group_id": "E4_core:A5|closed|qwen3b-llama3b:h2", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:15:15.030544Z", "task_family": "synthetic", "document_id": "syn_03018_2000", "document_hash": "d284eb07e27da742b9c2b67abccfdf72e4dd67c3259f44f3b41e34dd3e6fc1ff", "question_id": "q00_single_fact", "split": "test", "cluster_id": "syn_03018_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "tokenizer_hashes": {"receiver": "b9d8872cf3a99da99e1fc4b0f71232a451632dc49ca4c6771549b4efa5c1c763"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 2, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1773, "query_tokens": 459, "message_tokens": 356, "message_utf8_bytes": 969, "cache_tensor_bytes": null, "serialized_payload_bytes": 969, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 11370.702792075463, "hops": {"1": 6929.263000027277, "2": 4441.439792048186}, "receiver_prefill_ms": 205.04920894745737, "ttft_ms": 11.627916945144534, "decode_ms": 11.823083041235805}, "total_wall_ms": 11599.639792111702, "ttft_ms": 11.627916945144534, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 64, "facts_retained": 13, "retention_rate": 0.203125, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.2222222222222222, "amount": 0.15384615384615385, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 899}, "ledger_by_hop": {"1": {"facts_total": 64, "facts_retained": 14, "retention_rate": 0.21875, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.2222222222222222, "amount": 0.15384615384615385, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 870}, "2": {"facts_total": 64, "facts_retained": 13, "retention_rate": 0.203125, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.2222222222222222, "amount": 0.15384615384615385, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 899}}}, "reference_key_id": "keys/synthetic/syn_03018_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "6f2a234c96c0a3531585f9d596252011a84f9c2d971d967ceddbecbadbe2e8cf", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "faf17abf9cb164a7ced8133e", "attempt_id": "25f9bfadd051", "run_group_id": "E4_core:A5|closed|qwen3b-llama3b:h2", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:15:15.274872Z", "task_family": "synthetic", "document_id": "syn_03018_2000", "document_hash": "d284eb07e27da742b9c2b67abccfdf72e4dd67c3259f44f3b41e34dd3e6fc1ff", "question_id": "q01_single_fact", "split": "test", "cluster_id": "syn_03018_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "tokenizer_hashes": {"receiver": "b9d8872cf3a99da99e1fc4b0f71232a451632dc49ca4c6771549b4efa5c1c763"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 2, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1773, "query_tokens": 465, "message_tokens": 356, "message_utf8_bytes": 969, "cache_tensor_bytes": null, "serialized_payload_bytes": 969, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 11370.702792075463, "hops": {"1": 6929.263000027277, "2": 4441.439792048186}, "receiver_prefill_ms": 206.39225002378225, "ttft_ms": 11.61300006788224, "decode_ms": 11.594958021305501}, "total_wall_ms": 11600.70962505415, "ttft_ms": 11.61300006788224, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 64, "facts_retained": 13, "retention_rate": 0.203125, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.2222222222222222, "amount": 0.15384615384615385, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 899}, "ledger_by_hop": {"1": {"facts_total": 64, "facts_retained": 14, "retention_rate": 0.21875, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.2222222222222222, "amount": 0.15384615384615385, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 870}, "2": {"facts_total": 64, "facts_retained": 13, "retention_rate": 0.203125, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.2222222222222222, "amount": 0.15384615384615385, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 899}}}, "reference_key_id": "keys/synthetic/syn_03018_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "6f2a234c96c0a3531585f9d596252011a84f9c2d971d967ceddbecbadbe2e8cf", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "bc59e50e8f94e764e013d5ed", "attempt_id": "a382bb53e178", "run_group_id": "E4_core:A5|closed|qwen3b-llama3b:h2", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:15:15.519756Z", "task_family": "synthetic", "document_id": "syn_03018_2000", "document_hash": "d284eb07e27da742b9c2b67abccfdf72e4dd67c3259f44f3b41e34dd3e6fc1ff", "question_id": "q02_exact_copy", "split": "test", "cluster_id": "syn_03018_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "tokenizer_hashes": {"receiver": "b9d8872cf3a99da99e1fc4b0f71232a451632dc49ca4c6771549b4efa5c1c763"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 2, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1773, "query_tokens": 464, "message_tokens": 356, "message_utf8_bytes": 969, "cache_tensor_bytes": null, "serialized_payload_bytes": 969, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 11370.702792075463, "hops": {"1": 6929.263000027277, "2": 4441.439792048186}, "receiver_prefill_ms": 203.54579202830791, "ttft_ms": 11.82141702156514, "decode_ms": 11.59729203209281}, "total_wall_ms": 11598.17591717001, "ttft_ms": 11.82141702156514, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 64, "facts_retained": 13, "retention_rate": 0.203125, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.2222222222222222, "amount": 0.15384615384615385, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 899}, "ledger_by_hop": {"1": {"facts_total": 64, "facts_retained": 14, "retention_rate": 0.21875, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.2222222222222222, "amount": 0.15384615384615385, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 870}, "2": {"facts_total": 64, "facts_retained": 13, "retention_rate": 0.203125, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.2222222222222222, "amount": 0.15384615384615385, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 899}}}, "reference_key_id": "keys/synthetic/syn_03018_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "6f2a234c96c0a3531585f9d596252011a84f9c2d971d967ceddbecbadbe2e8cf", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "8424913933c0147478a69a27", "attempt_id": "43b472a8d856", "run_group_id": "E4_core:A5|closed|qwen3b-llama3b:h2", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:15:15.762093Z", "task_family": "synthetic", "document_id": "syn_03018_2000", "document_hash": "d284eb07e27da742b9c2b67abccfdf72e4dd67c3259f44f3b41e34dd3e6fc1ff", "question_id": "q03_exact_copy", "split": "test", "cluster_id": "syn_03018_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "tokenizer_hashes": {"receiver": "b9d8872cf3a99da99e1fc4b0f71232a451632dc49ca4c6771549b4efa5c1c763"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 2, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1773, "query_tokens": 461, "message_tokens": 356, "message_utf8_bytes": 969, "cache_tensor_bytes": null, "serialized_payload_bytes": 969, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 11370.702792075463, "hops": {"1": 6929.263000027277, "2": 4441.439792048186}, "receiver_prefill_ms": 204.07516707200557, "ttft_ms": 11.706209043040872, "decode_ms": 103.74600009527057}, "total_wall_ms": 11690.623959177174, "ttft_ms": 11.706209043040872, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 9}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "Priyamvada Sörlie", "evidence_artifact": {"ledger": {"facts_total": 64, "facts_retained": 13, "retention_rate": 0.203125, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.2222222222222222, "amount": 0.15384615384615385, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 899}, "ledger_by_hop": {"1": {"facts_total": 64, "facts_retained": 14, "retention_rate": 0.21875, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.2222222222222222, "amount": 0.15384615384615385, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 870}, "2": {"facts_total": 64, "facts_retained": 13, "retention_rate": 0.203125, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.2222222222222222, "amount": 0.15384615384615385, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 899}}}, "reference_key_id": "keys/synthetic/syn_03018_2000.json", "task_scores": {"correct": 0.0, "abstained": 0.0, "f1": 0.5}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "6f2a234c96c0a3531585f9d596252011a84f9c2d971d967ceddbecbadbe2e8cf", "output_hash": "0bc227b0b15021fbeca665b4eb3e111165a8d8518a5987c4a83aee81df5f53ee"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "8f903130f8027663b9bcb576", "attempt_id": "1d926a61db56", "run_group_id": "E4_core:A5|closed|qwen3b-llama3b:h2", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:15:16.097333Z", "task_family": "synthetic", "document_id": "syn_03018_2000", "document_hash": "d284eb07e27da742b9c2b67abccfdf72e4dd67c3259f44f3b41e34dd3e6fc1ff", "question_id": "q04_multi_hop", "split": "test", "cluster_id": "syn_03018_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "tokenizer_hashes": {"receiver": "b9d8872cf3a99da99e1fc4b0f71232a451632dc49ca4c6771549b4efa5c1c763"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 2, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1773, "query_tokens": 469, "message_tokens": 356, "message_utf8_bytes": 969, "cache_tensor_bytes": null, "serialized_payload_bytes": 969, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 11370.702792075463, "hops": {"1": 6929.263000027277, "2": 4441.439792048186}, "receiver_prefill_ms": 204.95858404319733, "ttft_ms": 11.655249982140958, "decode_ms": 11.463999981060624}, "total_wall_ms": 11599.195292103104, "ttft_ms": 11.655249982140958, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 64, "facts_retained": 13, "retention_rate": 0.203125, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.2222222222222222, "amount": 0.15384615384615385, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 899}, "ledger_by_hop": {"1": {"facts_total": 64, "facts_retained": 14, "retention_rate": 0.21875, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.2222222222222222, "amount": 0.15384615384615385, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 870}, "2": {"facts_total": 64, "facts_retained": 13, "retention_rate": 0.203125, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.2222222222222222, "amount": 0.15384615384615385, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 899}}}, "reference_key_id": "keys/synthetic/syn_03018_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "6f2a234c96c0a3531585f9d596252011a84f9c2d971d967ceddbecbadbe2e8cf", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "9890ba90e06b84ca7a1046d8", "attempt_id": "2488fcf22c58", "run_group_id": "E4_core:A5|closed|qwen3b-llama3b:h2", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:15:16.340231Z", "task_family": "synthetic", "document_id": "syn_03018_2000", "document_hash": "d284eb07e27da742b9c2b67abccfdf72e4dd67c3259f44f3b41e34dd3e6fc1ff", "question_id": "q05_multi_hop", "split": "test", "cluster_id": "syn_03018_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "tokenizer_hashes": {"receiver": "b9d8872cf3a99da99e1fc4b0f71232a451632dc49ca4c6771549b4efa5c1c763"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 2, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1773, "query_tokens": 467, "message_tokens": 356, "message_utf8_bytes": 969, "cache_tensor_bytes": null, "serialized_payload_bytes": 969, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 11370.702792075463, "hops": {"1": 6929.263000027277, "2": 4441.439792048186}, "receiver_prefill_ms": 205.1072500180453, "ttft_ms": 11.613084003329277, "decode_ms": 11.49624993558973}, "total_wall_ms": 11599.285167059861, "ttft_ms": 11.613084003329277, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 64, "facts_retained": 13, "retention_rate": 0.203125, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.2222222222222222, "amount": 0.15384615384615385, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 899}, "ledger_by_hop": {"1": {"facts_total": 64, "facts_retained": 14, "retention_rate": 0.21875, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.2222222222222222, "amount": 0.15384615384615385, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 870}, "2": {"facts_total": 64, "facts_retained": 13, "retention_rate": 0.203125, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.2222222222222222, "amount": 0.15384615384615385, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 899}}}, "reference_key_id": "keys/synthetic/syn_03018_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "6f2a234c96c0a3531585f9d596252011a84f9c2d971d967ceddbecbadbe2e8cf", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "c7ec6559cdb4bf5a9018b77b", "attempt_id": "01e23d89f171", "run_group_id": "E4_core:A5|closed|qwen3b-llama3b:h2", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:15:16.583179Z", "task_family": "synthetic", "document_id": "syn_03018_2000", "document_hash": "d284eb07e27da742b9c2b67abccfdf72e4dd67c3259f44f3b41e34dd3e6fc1ff", "question_id": "q06_negation", "split": "test", "cluster_id": "syn_03018_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "tokenizer_hashes": {"receiver": "b9d8872cf3a99da99e1fc4b0f71232a451632dc49ca4c6771549b4efa5c1c763"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 2, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1773, "query_tokens": 460, "message_tokens": 356, "message_utf8_bytes": 969, "cache_tensor_bytes": null, "serialized_payload_bytes": 969, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 11370.702792075463, "hops": {"1": 6929.263000027277, "2": 4441.439792048186}, "receiver_prefill_ms": 204.1886249789968, "ttft_ms": 11.603499995544553, "decode_ms": 11.499582906253636}, "total_wall_ms": 11598.36566715967, "ttft_ms": 11.603499995544553, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "No", "evidence_artifact": {"ledger": {"facts_total": 64, "facts_retained": 13, "retention_rate": 0.203125, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.2222222222222222, "amount": 0.15384615384615385, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 899}, "ledger_by_hop": {"1": {"facts_total": 64, "facts_retained": 14, "retention_rate": 0.21875, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.2222222222222222, "amount": 0.15384615384615385, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 870}, "2": {"facts_total": 64, "facts_retained": 13, "retention_rate": 0.203125, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.2222222222222222, "amount": 0.15384615384615385, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 899}}}, "reference_key_id": "keys/synthetic/syn_03018_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "6f2a234c96c0a3531585f9d596252011a84f9c2d971d967ceddbecbadbe2e8cf", "output_hash": "1ea442a134b2a184bd5d40104401f2a37fbc09ccf3f4bc9da161c6099be3691d"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "2bf6b4d3cded72f3975de332", "attempt_id": "85c9d5fe800a", "run_group_id": "E4_core:A5|closed|qwen3b-llama3b:h2", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:15:16.825275Z", "task_family": "synthetic", "document_id": "syn_03018_2000", "document_hash": "d284eb07e27da742b9c2b67abccfdf72e4dd67c3259f44f3b41e34dd3e6fc1ff", "question_id": "q07_negation", "split": "test", "cluster_id": "syn_03018_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "tokenizer_hashes": {"receiver": "b9d8872cf3a99da99e1fc4b0f71232a451632dc49ca4c6771549b4efa5c1c763"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 2, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1773, "query_tokens": 460, "message_tokens": 356, "message_utf8_bytes": 969, "cache_tensor_bytes": null, "serialized_payload_bytes": 969, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 11370.702792075463, "hops": {"1": 6929.263000027277, "2": 4441.439792048186}, "receiver_prefill_ms": 204.19241697527468, "ttft_ms": 11.739165987819433, "decode_ms": 11.534707969985902}, "total_wall_ms": 11598.541125073098, "ttft_ms": 11.739165987819433, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 64, "facts_retained": 13, "retention_rate": 0.203125, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.2222222222222222, "amount": 0.15384615384615385, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 899}, "ledger_by_hop": {"1": {"facts_total": 64, "facts_retained": 14, "retention_rate": 0.21875, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.2222222222222222, "amount": 0.15384615384615385, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 870}, "2": {"facts_total": 64, "facts_retained": 13, "retention_rate": 0.203125, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.2222222222222222, "amount": 0.15384615384615385, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 899}}}, "reference_key_id": "keys/synthetic/syn_03018_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "6f2a234c96c0a3531585f9d596252011a84f9c2d971d967ceddbecbadbe2e8cf", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "15a9aa30b2f31ca5188470df", "attempt_id": "cd1c247a060f", "run_group_id": "E4_core:A5|closed|qwen3b-llama3b:h2", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:15:17.067440Z", "task_family": "synthetic", "document_id": "syn_03018_2000", "document_hash": "d284eb07e27da742b9c2b67abccfdf72e4dd67c3259f44f3b41e34dd3e6fc1ff", "question_id": "q08_revision", "split": "test", "cluster_id": "syn_03018_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "tokenizer_hashes": {"receiver": "b9d8872cf3a99da99e1fc4b0f71232a451632dc49ca4c6771549b4efa5c1c763"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 2, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1773, "query_tokens": 476, "message_tokens": 356, "message_utf8_bytes": 969, "cache_tensor_bytes": null, "serialized_payload_bytes": 969, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 11370.702792075463, "hops": {"1": 6929.263000027277, "2": 4441.439792048186}, "receiver_prefill_ms": 207.92575005907565, "ttft_ms": 11.660875054076314, "decode_ms": 34.62995798327029}, "total_wall_ms": 11625.328709022142, "ttft_ms": 11.660875054076314, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 3}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "272,184", "evidence_artifact": {"ledger": {"facts_total": 64, "facts_retained": 13, "retention_rate": 0.203125, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.2222222222222222, "amount": 0.15384615384615385, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 899}, "ledger_by_hop": {"1": {"facts_total": 64, "facts_retained": 14, "retention_rate": 0.21875, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.2222222222222222, "amount": 0.15384615384615385, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 870}, "2": {"facts_total": 64, "facts_retained": 13, "retention_rate": 0.203125, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.2222222222222222, "amount": 0.15384615384615385, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 899}}}, "reference_key_id": "keys/synthetic/syn_03018_2000.json", "task_scores": {"correct": 0.0, "abstained": 0.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "6f2a234c96c0a3531585f9d596252011a84f9c2d971d967ceddbecbadbe2e8cf", "output_hash": "7ab8ea46f050f78e243a876f64862b441080883dc7365da1e4abc893edc12fb0"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "1c8060360c213268c980c984", "attempt_id": "47c379868677", "run_group_id": "E4_core:A5|closed|qwen3b-llama3b:h2", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:15:17.336270Z", "task_family": "synthetic", "document_id": "syn_03018_2000", "document_hash": "d284eb07e27da742b9c2b67abccfdf72e4dd67c3259f44f3b41e34dd3e6fc1ff", "question_id": "q09_revision", "split": "test", "cluster_id": "syn_03018_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "tokenizer_hashes": {"receiver": "b9d8872cf3a99da99e1fc4b0f71232a451632dc49ca4c6771549b4efa5c1c763"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 2, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1773, "query_tokens": 469, "message_tokens": 356, "message_utf8_bytes": 969, "cache_tensor_bytes": null, "serialized_payload_bytes": 969, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 11370.702792075463, "hops": {"1": 6929.263000027277, "2": 4441.439792048186}, "receiver_prefill_ms": 207.0920419646427, "ttft_ms": 11.598457931540906, "decode_ms": 34.72695802338421}, "total_wall_ms": 11624.514374998398, "ttft_ms": 11.598457931540906, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 3}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "272,184", "evidence_artifact": {"ledger": {"facts_total": 64, "facts_retained": 13, "retention_rate": 0.203125, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.2222222222222222, "amount": 0.15384615384615385, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 899}, "ledger_by_hop": {"1": {"facts_total": 64, "facts_retained": 14, "retention_rate": 0.21875, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.2222222222222222, "amount": 0.15384615384615385, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 870}, "2": {"facts_total": 64, "facts_retained": 13, "retention_rate": 0.203125, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.2222222222222222, "amount": 0.15384615384615385, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 899}}}, "reference_key_id": "keys/synthetic/syn_03018_2000.json", "task_scores": {"correct": 0.0, "abstained": 0.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "6f2a234c96c0a3531585f9d596252011a84f9c2d971d967ceddbecbadbe2e8cf", "output_hash": "7ab8ea46f050f78e243a876f64862b441080883dc7365da1e4abc893edc12fb0"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "78d2b1e311208729eeb63909", "attempt_id": "4647330c7ec9", "run_group_id": "E4_core:A5|closed|qwen3b-llama3b:h2", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:15:17.604463Z", "task_family": "synthetic", "document_id": "syn_03018_2000", "document_hash": "d284eb07e27da742b9c2b67abccfdf72e4dd67c3259f44f3b41e34dd3e6fc1ff", "question_id": "q10_temporal", "split": "test", "cluster_id": "syn_03018_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "tokenizer_hashes": {"receiver": "b9d8872cf3a99da99e1fc4b0f71232a451632dc49ca4c6771549b4efa5c1c763"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 2, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1773, "query_tokens": 477, "message_tokens": 356, "message_utf8_bytes": 969, "cache_tensor_bytes": null, "serialized_payload_bytes": 969, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 11370.702792075463, "hops": {"1": 6929.263000027277, "2": 4441.439792048186}, "receiver_prefill_ms": 207.33475009910762, "ttft_ms": 11.577790952287614, "decode_ms": 11.514584068208933}, "total_wall_ms": 11601.493584108539, "ttft_ms": 11.577790952287614, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 64, "facts_retained": 13, "retention_rate": 0.203125, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.2222222222222222, "amount": 0.15384615384615385, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 899}, "ledger_by_hop": {"1": {"facts_total": 64, "facts_retained": 14, "retention_rate": 0.21875, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.2222222222222222, "amount": 0.15384615384615385, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 870}, "2": {"facts_total": 64, "facts_retained": 13, "retention_rate": 0.203125, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.2222222222222222, "amount": 0.15384615384615385, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 899}}}, "reference_key_id": "keys/synthetic/syn_03018_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "6f2a234c96c0a3531585f9d596252011a84f9c2d971d967ceddbecbadbe2e8cf", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "9339de26638277071855792e", "attempt_id": "c936b4bd552b", "run_group_id": "E4_core:A5|closed|qwen3b-llama3b:h2", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:15:17.849722Z", "task_family": "synthetic", "document_id": "syn_03018_2000", "document_hash": "d284eb07e27da742b9c2b67abccfdf72e4dd67c3259f44f3b41e34dd3e6fc1ff", "question_id": "q11_aggregation", "split": "test", "cluster_id": "syn_03018_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "tokenizer_hashes": {"receiver": "b9d8872cf3a99da99e1fc4b0f71232a451632dc49ca4c6771549b4efa5c1c763"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 2, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1773, "query_tokens": 471, "message_tokens": 356, "message_utf8_bytes": 969, "cache_tensor_bytes": null, "serialized_payload_bytes": 969, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 11370.702792075463, "hops": {"1": 6929.263000027277, "2": 4441.439792048186}, "receiver_prefill_ms": 204.96133295819163, "ttft_ms": 11.493084020912647, "decode_ms": 57.65754100866616}, "total_wall_ms": 11645.205459091812, "ttft_ms": 11.493084020912647, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 5}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "2,189,245", "evidence_artifact": {"ledger": {"facts_total": 64, "facts_retained": 13, "retention_rate": 0.203125, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.2222222222222222, "amount": 0.15384615384615385, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 899}, "ledger_by_hop": {"1": {"facts_total": 64, "facts_retained": 14, "retention_rate": 0.21875, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.2222222222222222, "amount": 0.15384615384615385, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 870}, "2": {"facts_total": 64, "facts_retained": 13, "retention_rate": 0.203125, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.2222222222222222, "amount": 0.15384615384615385, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 899}}}, "reference_key_id": "keys/synthetic/syn_03018_2000.json", "task_scores": {"correct": 0.0, "abstained": 0.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "6f2a234c96c0a3531585f9d596252011a84f9c2d971d967ceddbecbadbe2e8cf", "output_hash": "6ebf362ebc451a5b2a0d5ab6fc41db9fea75444a45cc75aa55fa0a34bc04be37"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "a707226ac0cc68aef131ccb3", "attempt_id": "78a13613ddf9", "run_group_id": "E4_core:A5|closed|qwen3b-llama3b:h2", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:15:18.138838Z", "task_family": "synthetic", "document_id": "syn_03018_2000", "document_hash": "d284eb07e27da742b9c2b67abccfdf72e4dd67c3259f44f3b41e34dd3e6fc1ff", "question_id": "q12_unit_conversion", "split": "test", "cluster_id": "syn_03018_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "tokenizer_hashes": {"receiver": "b9d8872cf3a99da99e1fc4b0f71232a451632dc49ca4c6771549b4efa5c1c763"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 2, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1773, "query_tokens": 467, "message_tokens": 356, "message_utf8_bytes": 969, "cache_tensor_bytes": null, "serialized_payload_bytes": 969, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 11370.702792075463, "hops": {"1": 6929.263000027277, "2": 4441.439792048186}, "receiver_prefill_ms": 202.83091708552092, "ttft_ms": 11.598166078329086, "decode_ms": 11.628667009063065}, "total_wall_ms": 11597.123958053999, "ttft_ms": 11.598166078329086, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 64, "facts_retained": 13, "retention_rate": 0.203125, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.2222222222222222, "amount": 0.15384615384615385, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 899}, "ledger_by_hop": {"1": {"facts_total": 64, "facts_retained": 14, "retention_rate": 0.21875, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.2222222222222222, "amount": 0.15384615384615385, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 870}, "2": {"facts_total": 64, "facts_retained": 13, "retention_rate": 0.203125, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.2222222222222222, "amount": 0.15384615384615385, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 899}}}, "reference_key_id": "keys/synthetic/syn_03018_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "6f2a234c96c0a3531585f9d596252011a84f9c2d971d967ceddbecbadbe2e8cf", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "7f7719199a35680876cf4738", "attempt_id": "95dfbdcdd738", "run_group_id": "E4_core:A5|closed|qwen3b-llama3b:h2", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:15:18.379880Z", "task_family": "synthetic", "document_id": "syn_03018_2000", "document_hash": "d284eb07e27da742b9c2b67abccfdf72e4dd67c3259f44f3b41e34dd3e6fc1ff", "question_id": "q13_unanswerable", "split": "test", "cluster_id": "syn_03018_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "tokenizer_hashes": {"receiver": "b9d8872cf3a99da99e1fc4b0f71232a451632dc49ca4c6771549b4efa5c1c763"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 2, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1773, "query_tokens": 472, "message_tokens": 356, "message_utf8_bytes": 969, "cache_tensor_bytes": null, "serialized_payload_bytes": 969, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 11370.702792075463, "hops": {"1": 6929.263000027277, "2": 4441.439792048186}, "receiver_prefill_ms": 204.10470792558044, "ttft_ms": 11.566333007067442, "decode_ms": 11.596875032410026}, "total_wall_ms": 11598.341709119268, "ttft_ms": 11.566333007067442, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 64, "facts_retained": 13, "retention_rate": 0.203125, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.2222222222222222, "amount": 0.15384615384615385, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 899}, "ledger_by_hop": {"1": {"facts_total": 64, "facts_retained": 14, "retention_rate": 0.21875, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.2222222222222222, "amount": 0.15384615384615385, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 870}, "2": {"facts_total": 64, "facts_retained": 13, "retention_rate": 0.203125, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.2222222222222222, "amount": 0.15384615384615385, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 899}}}, "reference_key_id": "keys/synthetic/syn_03018_2000.json", "task_scores": {"correct": 1.0, "abstained": 1.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "6f2a234c96c0a3531585f9d596252011a84f9c2d971d967ceddbecbadbe2e8cf", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "07b3f3590a4dbd0201ba27e8", "attempt_id": "292d9e375f7e", "run_group_id": "E4_core:A5|closed|qwen3b-llama3b:h2", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:15:18.622198Z", "task_family": "synthetic", "document_id": "syn_03018_2000", "document_hash": "d284eb07e27da742b9c2b67abccfdf72e4dd67c3259f44f3b41e34dd3e6fc1ff", "question_id": "q14_unanswerable", "split": "test", "cluster_id": "syn_03018_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "tokenizer_hashes": {"receiver": "b9d8872cf3a99da99e1fc4b0f71232a451632dc49ca4c6771549b4efa5c1c763"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 2, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1773, "query_tokens": 473, "message_tokens": 356, "message_utf8_bytes": 969, "cache_tensor_bytes": null, "serialized_payload_bytes": 969, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 11370.702792075463, "hops": {"1": 6929.263000027277, "2": 4441.439792048186}, "receiver_prefill_ms": 205.04904200788587, "ttft_ms": 11.590666021220386, "decode_ms": 11.576791992411017}, "total_wall_ms": 11599.301626090892, "ttft_ms": 11.590666021220386, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 64, "facts_retained": 13, "retention_rate": 0.203125, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.2222222222222222, "amount": 0.15384615384615385, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 899}, "ledger_by_hop": {"1": {"facts_total": 64, "facts_retained": 14, "retention_rate": 0.21875, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.2222222222222222, "amount": 0.15384615384615385, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 870}, "2": {"facts_total": 64, "facts_retained": 13, "retention_rate": 0.203125, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.2222222222222222, "amount": 0.15384615384615385, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 899}}}, "reference_key_id": "keys/synthetic/syn_03018_2000.json", "task_scores": {"correct": 1.0, "abstained": 1.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "6f2a234c96c0a3531585f9d596252011a84f9c2d971d967ceddbecbadbe2e8cf", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "e84da1a25a478b0bad48a76e", "attempt_id": "fd3dd3fc77ad", "run_group_id": "E4_core:A5|closed|qwen3b-llama3b:h4", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:15:28.498760Z", "task_family": "synthetic", "document_id": "syn_03018_2000", "document_hash": "d284eb07e27da742b9c2b67abccfdf72e4dd67c3259f44f3b41e34dd3e6fc1ff", "question_id": "q00_single_fact", "split": "test", "cluster_id": "syn_03018_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "tokenizer_hashes": {"receiver": "b9d8872cf3a99da99e1fc4b0f71232a451632dc49ca4c6771549b4efa5c1c763"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 4, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1773, "query_tokens": 460, "message_tokens": 357, "message_utf8_bytes": 1012, "cache_tensor_bytes": null, "serialized_payload_bytes": 1012, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 21001.85354310088, "hops": {"1": 6929.263000027277, "2": 4441.439792048186, "3": 5179.9550839932635, "4": 4451.195667032152}, "receiver_prefill_ms": 204.40025010611862, "ttft_ms": 11.672458960674703, "decode_ms": 11.42370910383761}, "total_wall_ms": 21229.793251142837, "ttft_ms": 11.672458960674703, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 64, "facts_retained": 13, "retention_rate": 0.203125, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.2222222222222222, "amount": 0.15384615384615385, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1012}, "ledger_by_hop": {"1": {"facts_total": 64, "facts_retained": 14, "retention_rate": 0.21875, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.2222222222222222, "amount": 0.15384615384615385, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 870}, "2": {"facts_total": 64, "facts_retained": 13, "retention_rate": 0.203125, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.2222222222222222, "amount": 0.15384615384615385, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 899}, "3": {"facts_total": 64, "facts_retained": 13, "retention_rate": 0.203125, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.2222222222222222, "amount": 0.15384615384615385, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 899}, "4": {"facts_total": 64, "facts_retained": 13, "retention_rate": 0.203125, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.2222222222222222, "amount": 0.15384615384615385, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1012}}}, "reference_key_id": "keys/synthetic/syn_03018_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "549e967d2d3d6d2623cc1b9a89343ccef7eff9d43bd47b2d2a266fc5d9d74d04", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "279af06392bdf2e88301b53e", "attempt_id": "c8cedcdeab50", "run_group_id": "E4_core:A5|closed|qwen3b-llama3b:h4", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:15:28.742271Z", "task_family": "synthetic", "document_id": "syn_03018_2000", "document_hash": "d284eb07e27da742b9c2b67abccfdf72e4dd67c3259f44f3b41e34dd3e6fc1ff", "question_id": "q01_single_fact", "split": "test", "cluster_id": "syn_03018_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "tokenizer_hashes": {"receiver": "b9d8872cf3a99da99e1fc4b0f71232a451632dc49ca4c6771549b4efa5c1c763"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 4, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1773, "query_tokens": 466, "message_tokens": 357, "message_utf8_bytes": 1012, "cache_tensor_bytes": null, "serialized_payload_bytes": 1012, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 21001.85354310088, "hops": {"1": 6929.263000027277, "2": 4441.439792048186, "3": 5179.9550839932635, "4": 4451.195667032152}, "receiver_prefill_ms": 203.80770799238235, "ttft_ms": 11.639499920420349, "decode_ms": 11.539916973561049}, "total_wall_ms": 21229.232918121852, "ttft_ms": 11.639499920420349, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 64, "facts_retained": 13, "retention_rate": 0.203125, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.2222222222222222, "amount": 0.15384615384615385, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1012}, "ledger_by_hop": {"1": {"facts_total": 64, "facts_retained": 14, "retention_rate": 0.21875, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.2222222222222222, "amount": 0.15384615384615385, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 870}, "2": {"facts_total": 64, "facts_retained": 13, "retention_rate": 0.203125, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.2222222222222222, "amount": 0.15384615384615385, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 899}, "3": {"facts_total": 64, "facts_retained": 13, "retention_rate": 0.203125, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.2222222222222222, "amount": 0.15384615384615385, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 899}, "4": {"facts_total": 64, "facts_retained": 13, "retention_rate": 0.203125, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.2222222222222222, "amount": 0.15384615384615385, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1012}}}, "reference_key_id": "keys/synthetic/syn_03018_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "549e967d2d3d6d2623cc1b9a89343ccef7eff9d43bd47b2d2a266fc5d9d74d04", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "06a40593f25103d6a5c1cf16", "attempt_id": "15df6e929e64", "run_group_id": "E4_core:A5|closed|qwen3b-llama3b:h4", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:15:28.984500Z", "task_family": "synthetic", "document_id": "syn_03018_2000", "document_hash": "d284eb07e27da742b9c2b67abccfdf72e4dd67c3259f44f3b41e34dd3e6fc1ff", "question_id": "q02_exact_copy", "split": "test", "cluster_id": "syn_03018_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "tokenizer_hashes": {"receiver": "b9d8872cf3a99da99e1fc4b0f71232a451632dc49ca4c6771549b4efa5c1c763"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 4, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1773, "query_tokens": 465, "message_tokens": 357, "message_utf8_bytes": 1012, "cache_tensor_bytes": null, "serialized_payload_bytes": 1012, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 21001.85354310088, "hops": {"1": 6929.263000027277, "2": 4441.439792048186, "3": 5179.9550839932635, "4": 4451.195667032152}, "receiver_prefill_ms": 203.02958402317017, "ttft_ms": 11.599124991334975, "decode_ms": 11.552874930202961}, "total_wall_ms": 21228.400501073338, "ttft_ms": 11.599124991334975, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 64, "facts_retained": 13, "retention_rate": 0.203125, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.2222222222222222, "amount": 0.15384615384615385, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1012}, "ledger_by_hop": {"1": {"facts_total": 64, "facts_retained": 14, "retention_rate": 0.21875, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.2222222222222222, "amount": 0.15384615384615385, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 870}, "2": {"facts_total": 64, "facts_retained": 13, "retention_rate": 0.203125, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.2222222222222222, "amount": 0.15384615384615385, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 899}, "3": {"facts_total": 64, "facts_retained": 13, "retention_rate": 0.203125, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.2222222222222222, "amount": 0.15384615384615385, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 899}, "4": {"facts_total": 64, "facts_retained": 13, "retention_rate": 0.203125, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.2222222222222222, "amount": 0.15384615384615385, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1012}}}, "reference_key_id": "keys/synthetic/syn_03018_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "549e967d2d3d6d2623cc1b9a89343ccef7eff9d43bd47b2d2a266fc5d9d74d04", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "41ff6d374b6f9ee45bac040e", "attempt_id": "34b76078904c", "run_group_id": "E4_core:A5|closed|qwen3b-llama3b:h4", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:15:29.225362Z", "task_family": "synthetic", "document_id": "syn_03018_2000", "document_hash": "d284eb07e27da742b9c2b67abccfdf72e4dd67c3259f44f3b41e34dd3e6fc1ff", "question_id": "q03_exact_copy", "split": "test", "cluster_id": "syn_03018_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "tokenizer_hashes": {"receiver": "b9d8872cf3a99da99e1fc4b0f71232a451632dc49ca4c6771549b4efa5c1c763"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 4, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1773, "query_tokens": 462, "message_tokens": 357, "message_utf8_bytes": 1012, "cache_tensor_bytes": null, "serialized_payload_bytes": 1012, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 21001.85354310088, "hops": {"1": 6929.263000027277, "2": 4441.439792048186, "3": 5179.9550839932635, "4": 4451.195667032152}, "receiver_prefill_ms": 202.69408402964473, "ttft_ms": 11.55037502758205, "decode_ms": 103.94666704814881}, "total_wall_ms": 21320.440168026835, "ttft_ms": 11.55037502758205, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 9}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "Priyamvada Sörlie", "evidence_artifact": {"ledger": {"facts_total": 64, "facts_retained": 13, "retention_rate": 0.203125, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.2222222222222222, "amount": 0.15384615384615385, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1012}, "ledger_by_hop": {"1": {"facts_total": 64, "facts_retained": 14, "retention_rate": 0.21875, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.2222222222222222, "amount": 0.15384615384615385, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 870}, "2": {"facts_total": 64, "facts_retained": 13, "retention_rate": 0.203125, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.2222222222222222, "amount": 0.15384615384615385, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 899}, "3": {"facts_total": 64, "facts_retained": 13, "retention_rate": 0.203125, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.2222222222222222, "amount": 0.15384615384615385, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 899}, "4": {"facts_total": 64, "facts_retained": 13, "retention_rate": 0.203125, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.2222222222222222, "amount": 0.15384615384615385, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1012}}}, "reference_key_id": "keys/synthetic/syn_03018_2000.json", "task_scores": {"correct": 0.0, "abstained": 0.0, "f1": 0.5}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "549e967d2d3d6d2623cc1b9a89343ccef7eff9d43bd47b2d2a266fc5d9d74d04", "output_hash": "0bc227b0b15021fbeca665b4eb3e111165a8d8518a5987c4a83aee81df5f53ee"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "8641b20c72a36ee81224c053", "attempt_id": "9a3a8a0a8736", "run_group_id": "E4_core:A5|closed|qwen3b-llama3b:h4", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:15:29.559024Z", "task_family": "synthetic", "document_id": "syn_03018_2000", "document_hash": "d284eb07e27da742b9c2b67abccfdf72e4dd67c3259f44f3b41e34dd3e6fc1ff", "question_id": "q04_multi_hop", "split": "test", "cluster_id": "syn_03018_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "tokenizer_hashes": {"receiver": "b9d8872cf3a99da99e1fc4b0f71232a451632dc49ca4c6771549b4efa5c1c763"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 4, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1773, "query_tokens": 470, "message_tokens": 357, "message_utf8_bytes": 1012, "cache_tensor_bytes": null, "serialized_payload_bytes": 1012, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 21001.85354310088, "hops": {"1": 6929.263000027277, "2": 4441.439792048186, "3": 5179.9550839932635, "4": 4451.195667032152}, "receiver_prefill_ms": 203.31529201939702, "ttft_ms": 11.482749949209392, "decode_ms": 11.576582910493016}, "total_wall_ms": 21228.58537605498, "ttft_ms": 11.482749949209392, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 64, "facts_retained": 13, "retention_rate": 0.203125, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.2222222222222222, "amount": 0.15384615384615385, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1012}, "ledger_by_hop": {"1": {"facts_total": 64, "facts_retained": 14, "retention_rate": 0.21875, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.2222222222222222, "amount": 0.15384615384615385, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 870}, "2": {"facts_total": 64, "facts_retained": 13, "retention_rate": 0.203125, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.2222222222222222, "amount": 0.15384615384615385, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 899}, "3": {"facts_total": 64, "facts_retained": 13, "retention_rate": 0.203125, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.2222222222222222, "amount": 0.15384615384615385, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 899}, "4": {"facts_total": 64, "facts_retained": 13, "retention_rate": 0.203125, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.2222222222222222, "amount": 0.15384615384615385, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1012}}}, "reference_key_id": "keys/synthetic/syn_03018_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "549e967d2d3d6d2623cc1b9a89343ccef7eff9d43bd47b2d2a266fc5d9d74d04", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "2c88d4d1b6dd80c43dc0bc16", "attempt_id": "00fb96b36d2c", "run_group_id": "E4_core:A5|closed|qwen3b-llama3b:h4", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:15:29.799830Z", "task_family": "synthetic", "document_id": "syn_03018_2000", "document_hash": "d284eb07e27da742b9c2b67abccfdf72e4dd67c3259f44f3b41e34dd3e6fc1ff", "question_id": "q05_multi_hop", "split": "test", "cluster_id": "syn_03018_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "tokenizer_hashes": {"receiver": "b9d8872cf3a99da99e1fc4b0f71232a451632dc49ca4c6771549b4efa5c1c763"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 4, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1773, "query_tokens": 468, "message_tokens": 357, "message_utf8_bytes": 1012, "cache_tensor_bytes": null, "serialized_payload_bytes": 1012, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 21001.85354310088, "hops": {"1": 6929.263000027277, "2": 4441.439792048186, "3": 5179.9550839932635, "4": 4451.195667032152}, "receiver_prefill_ms": 203.9772500284016, "ttft_ms": 11.54962508007884, "decode_ms": 11.540916049852967}, "total_wall_ms": 21229.296793113463, "ttft_ms": 11.54962508007884, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 64, "facts_retained": 13, "retention_rate": 0.203125, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.2222222222222222, "amount": 0.15384615384615385, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1012}, "ledger_by_hop": {"1": {"facts_total": 64, "facts_retained": 14, "retention_rate": 0.21875, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.2222222222222222, "amount": 0.15384615384615385, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 870}, "2": {"facts_total": 64, "facts_retained": 13, "retention_rate": 0.203125, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.2222222222222222, "amount": 0.15384615384615385, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 899}, "3": {"facts_total": 64, "facts_retained": 13, "retention_rate": 0.203125, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.2222222222222222, "amount": 0.15384615384615385, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 899}, "4": {"facts_total": 64, "facts_retained": 13, "retention_rate": 0.203125, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.2222222222222222, "amount": 0.15384615384615385, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1012}}}, "reference_key_id": "keys/synthetic/syn_03018_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "549e967d2d3d6d2623cc1b9a89343ccef7eff9d43bd47b2d2a266fc5d9d74d04", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "09c6a72ab19cb9d1bd8c1aa9", "attempt_id": "3f81dad1ef58", "run_group_id": "E4_core:A5|closed|qwen3b-llama3b:h4", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:15:30.042679Z", "task_family": "synthetic", "document_id": "syn_03018_2000", "document_hash": "d284eb07e27da742b9c2b67abccfdf72e4dd67c3259f44f3b41e34dd3e6fc1ff", "question_id": "q06_negation", "split": "test", "cluster_id": "syn_03018_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "tokenizer_hashes": {"receiver": "b9d8872cf3a99da99e1fc4b0f71232a451632dc49ca4c6771549b4efa5c1c763"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 4, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1773, "query_tokens": 461, "message_tokens": 357, "message_utf8_bytes": 1012, "cache_tensor_bytes": null, "serialized_payload_bytes": 1012, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 21001.85354310088, "hops": {"1": 6929.263000027277, "2": 4441.439792048186, "3": 5179.9550839932635, "4": 4451.195667032152}, "receiver_prefill_ms": 205.1257910206914, "ttft_ms": 11.586499982513487, "decode_ms": 11.517875012941658}, "total_wall_ms": 21230.457252124324, "ttft_ms": 11.586499982513487, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "No", "evidence_artifact": {"ledger": {"facts_total": 64, "facts_retained": 13, "retention_rate": 0.203125, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.2222222222222222, "amount": 0.15384615384615385, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1012}, "ledger_by_hop": {"1": {"facts_total": 64, "facts_retained": 14, "retention_rate": 0.21875, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.2222222222222222, "amount": 0.15384615384615385, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 870}, "2": {"facts_total": 64, "facts_retained": 13, "retention_rate": 0.203125, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.2222222222222222, "amount": 0.15384615384615385, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 899}, "3": {"facts_total": 64, "facts_retained": 13, "retention_rate": 0.203125, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.2222222222222222, "amount": 0.15384615384615385, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 899}, "4": {"facts_total": 64, "facts_retained": 13, "retention_rate": 0.203125, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.2222222222222222, "amount": 0.15384615384615385, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1012}}}, "reference_key_id": "keys/synthetic/syn_03018_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "549e967d2d3d6d2623cc1b9a89343ccef7eff9d43bd47b2d2a266fc5d9d74d04", "output_hash": "1ea442a134b2a184bd5d40104401f2a37fbc09ccf3f4bc9da161c6099be3691d"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "9396c8287e7d3e8816f6bc54", "attempt_id": "ec24fb95b112", "run_group_id": "E4_core:A5|closed|qwen3b-llama3b:h4", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:15:30.285566Z", "task_family": "synthetic", "document_id": "syn_03018_2000", "document_hash": "d284eb07e27da742b9c2b67abccfdf72e4dd67c3259f44f3b41e34dd3e6fc1ff", "question_id": "q07_negation", "split": "test", "cluster_id": "syn_03018_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "tokenizer_hashes": {"receiver": "b9d8872cf3a99da99e1fc4b0f71232a451632dc49ca4c6771549b4efa5c1c763"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 4, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1773, "query_tokens": 461, "message_tokens": 357, "message_utf8_bytes": 1012, "cache_tensor_bytes": null, "serialized_payload_bytes": 1012, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 21001.85354310088, "hops": {"1": 6929.263000027277, "2": 4441.439792048186, "3": 5179.9550839932635, "4": 4451.195667032152}, "receiver_prefill_ms": 205.3372919326648, "ttft_ms": 11.583541985601187, "decode_ms": 11.616582982242107}, "total_wall_ms": 21230.748668080196, "ttft_ms": 11.583541985601187, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 64, "facts_retained": 13, "retention_rate": 0.203125, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.2222222222222222, "amount": 0.15384615384615385, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1012}, "ledger_by_hop": {"1": {"facts_total": 64, "facts_retained": 14, "retention_rate": 0.21875, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.2222222222222222, "amount": 0.15384615384615385, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 870}, "2": {"facts_total": 64, "facts_retained": 13, "retention_rate": 0.203125, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.2222222222222222, "amount": 0.15384615384615385, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 899}, "3": {"facts_total": 64, "facts_retained": 13, "retention_rate": 0.203125, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.2222222222222222, "amount": 0.15384615384615385, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 899}, "4": {"facts_total": 64, "facts_retained": 13, "retention_rate": 0.203125, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.2222222222222222, "amount": 0.15384615384615385, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1012}}}, "reference_key_id": "keys/synthetic/syn_03018_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "549e967d2d3d6d2623cc1b9a89343ccef7eff9d43bd47b2d2a266fc5d9d74d04", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "52d75d5cbe184142325d3910", "attempt_id": "07c51eef0fe1", "run_group_id": "E4_core:A5|closed|qwen3b-llama3b:h4", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:15:30.529062Z", "task_family": "synthetic", "document_id": "syn_03018_2000", "document_hash": "d284eb07e27da742b9c2b67abccfdf72e4dd67c3259f44f3b41e34dd3e6fc1ff", "question_id": "q08_revision", "split": "test", "cluster_id": "syn_03018_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "tokenizer_hashes": {"receiver": "b9d8872cf3a99da99e1fc4b0f71232a451632dc49ca4c6771549b4efa5c1c763"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 4, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1773, "query_tokens": 477, "message_tokens": 357, "message_utf8_bytes": 1012, "cache_tensor_bytes": null, "serialized_payload_bytes": 1012, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 21001.85354310088, "hops": {"1": 6929.263000027277, "2": 4441.439792048186, "3": 5179.9550839932635, "4": 4451.195667032152}, "receiver_prefill_ms": 208.63995794206858, "ttft_ms": 11.699834023602307, "decode_ms": 34.679041942581534}, "total_wall_ms": 21257.25962710567, "ttft_ms": 11.699834023602307, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 3}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "272,184", "evidence_artifact": {"ledger": {"facts_total": 64, "facts_retained": 13, "retention_rate": 0.203125, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.2222222222222222, "amount": 0.15384615384615385, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1012}, "ledger_by_hop": {"1": {"facts_total": 64, "facts_retained": 14, "retention_rate": 0.21875, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.2222222222222222, "amount": 0.15384615384615385, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 870}, "2": {"facts_total": 64, "facts_retained": 13, "retention_rate": 0.203125, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.2222222222222222, "amount": 0.15384615384615385, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 899}, "3": {"facts_total": 64, "facts_retained": 13, "retention_rate": 0.203125, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.2222222222222222, "amount": 0.15384615384615385, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 899}, "4": {"facts_total": 64, "facts_retained": 13, "retention_rate": 0.203125, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.2222222222222222, "amount": 0.15384615384615385, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1012}}}, "reference_key_id": "keys/synthetic/syn_03018_2000.json", "task_scores": {"correct": 0.0, "abstained": 0.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "549e967d2d3d6d2623cc1b9a89343ccef7eff9d43bd47b2d2a266fc5d9d74d04", "output_hash": "7ab8ea46f050f78e243a876f64862b441080883dc7365da1e4abc893edc12fb0"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "f73c6a023f3746b263487b99", "attempt_id": "b2588a4677cf", "run_group_id": "E4_core:A5|closed|qwen3b-llama3b:h4", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:15:30.798530Z", "task_family": "synthetic", "document_id": "syn_03018_2000", "document_hash": "d284eb07e27da742b9c2b67abccfdf72e4dd67c3259f44f3b41e34dd3e6fc1ff", "question_id": "q09_revision", "split": "test", "cluster_id": "syn_03018_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "tokenizer_hashes": {"receiver": "b9d8872cf3a99da99e1fc4b0f71232a451632dc49ca4c6771549b4efa5c1c763"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 4, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1773, "query_tokens": 470, "message_tokens": 357, "message_utf8_bytes": 1012, "cache_tensor_bytes": null, "serialized_payload_bytes": 1012, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 21001.85354310088, "hops": {"1": 6929.263000027277, "2": 4441.439792048186, "3": 5179.9550839932635, "4": 4451.195667032152}, "receiver_prefill_ms": 206.4403749536723, "ttft_ms": 11.59779098816216, "decode_ms": 34.92595802526921}, "total_wall_ms": 21255.183502100408, "ttft_ms": 11.59779098816216, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 3}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "272,184", "evidence_artifact": {"ledger": {"facts_total": 64, "facts_retained": 13, "retention_rate": 0.203125, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.2222222222222222, "amount": 0.15384615384615385, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1012}, "ledger_by_hop": {"1": {"facts_total": 64, "facts_retained": 14, "retention_rate": 0.21875, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.2222222222222222, "amount": 0.15384615384615385, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 870}, "2": {"facts_total": 64, "facts_retained": 13, "retention_rate": 0.203125, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.2222222222222222, "amount": 0.15384615384615385, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 899}, "3": {"facts_total": 64, "facts_retained": 13, "retention_rate": 0.203125, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.2222222222222222, "amount": 0.15384615384615385, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 899}, "4": {"facts_total": 64, "facts_retained": 13, "retention_rate": 0.203125, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.2222222222222222, "amount": 0.15384615384615385, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1012}}}, "reference_key_id": "keys/synthetic/syn_03018_2000.json", "task_scores": {"correct": 0.0, "abstained": 0.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "549e967d2d3d6d2623cc1b9a89343ccef7eff9d43bd47b2d2a266fc5d9d74d04", "output_hash": "7ab8ea46f050f78e243a876f64862b441080883dc7365da1e4abc893edc12fb0"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "21e5bf313cc1f46109de8895", "attempt_id": "b612866e6b02", "run_group_id": "E4_core:A5|closed|qwen3b-llama3b:h4", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:15:31.066448Z", "task_family": "synthetic", "document_id": "syn_03018_2000", "document_hash": "d284eb07e27da742b9c2b67abccfdf72e4dd67c3259f44f3b41e34dd3e6fc1ff", "question_id": "q10_temporal", "split": "test", "cluster_id": "syn_03018_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "tokenizer_hashes": {"receiver": "b9d8872cf3a99da99e1fc4b0f71232a451632dc49ca4c6771549b4efa5c1c763"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 4, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1773, "query_tokens": 478, "message_tokens": 357, "message_utf8_bytes": 1012, "cache_tensor_bytes": null, "serialized_payload_bytes": 1012, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 21001.85354310088, "hops": {"1": 6929.263000027277, "2": 4441.439792048186, "3": 5179.9550839932635, "4": 4451.195667032152}, "receiver_prefill_ms": 206.38529199641198, "ttft_ms": 11.906916974112391, "decode_ms": 11.552459094673395}, "total_wall_ms": 21232.063627103344, "ttft_ms": 11.906916974112391, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 64, "facts_retained": 13, "retention_rate": 0.203125, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.2222222222222222, "amount": 0.15384615384615385, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1012}, "ledger_by_hop": {"1": {"facts_total": 64, "facts_retained": 14, "retention_rate": 0.21875, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.2222222222222222, "amount": 0.15384615384615385, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 870}, "2": {"facts_total": 64, "facts_retained": 13, "retention_rate": 0.203125, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.2222222222222222, "amount": 0.15384615384615385, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 899}, "3": {"facts_total": 64, "facts_retained": 13, "retention_rate": 0.203125, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.2222222222222222, "amount": 0.15384615384615385, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 899}, "4": {"facts_total": 64, "facts_retained": 13, "retention_rate": 0.203125, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.2222222222222222, "amount": 0.15384615384615385, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1012}}}, "reference_key_id": "keys/synthetic/syn_03018_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "549e967d2d3d6d2623cc1b9a89343ccef7eff9d43bd47b2d2a266fc5d9d74d04", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "9ed03a1acd7753bae17f1684", "attempt_id": "f33c6fb568eb", "run_group_id": "E4_core:A5|closed|qwen3b-llama3b:h4", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:15:31.311272Z", "task_family": "synthetic", "document_id": "syn_03018_2000", "document_hash": "d284eb07e27da742b9c2b67abccfdf72e4dd67c3259f44f3b41e34dd3e6fc1ff", "question_id": "q11_aggregation", "split": "test", "cluster_id": "syn_03018_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "tokenizer_hashes": {"receiver": "b9d8872cf3a99da99e1fc4b0f71232a451632dc49ca4c6771549b4efa5c1c763"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 4, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1773, "query_tokens": 472, "message_tokens": 357, "message_utf8_bytes": 1012, "cache_tensor_bytes": null, "serialized_payload_bytes": 1012, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 21001.85354310088, "hops": {"1": 6929.263000027277, "2": 4441.439792048186, "3": 5179.9550839932635, "4": 4451.195667032152}, "receiver_prefill_ms": 204.59941599983722, "ttft_ms": 11.61766704171896, "decode_ms": 57.48212500475347}, "total_wall_ms": 21276.058126124553, "ttft_ms": 11.61766704171896, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 5}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "2,189,245", "evidence_artifact": {"ledger": {"facts_total": 64, "facts_retained": 13, "retention_rate": 0.203125, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.2222222222222222, "amount": 0.15384615384615385, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1012}, "ledger_by_hop": {"1": {"facts_total": 64, "facts_retained": 14, "retention_rate": 0.21875, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.2222222222222222, "amount": 0.15384615384615385, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 870}, "2": {"facts_total": 64, "facts_retained": 13, "retention_rate": 0.203125, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.2222222222222222, "amount": 0.15384615384615385, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 899}, "3": {"facts_total": 64, "facts_retained": 13, "retention_rate": 0.203125, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.2222222222222222, "amount": 0.15384615384615385, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 899}, "4": {"facts_total": 64, "facts_retained": 13, "retention_rate": 0.203125, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.2222222222222222, "amount": 0.15384615384615385, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1012}}}, "reference_key_id": "keys/synthetic/syn_03018_2000.json", "task_scores": {"correct": 0.0, "abstained": 0.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "549e967d2d3d6d2623cc1b9a89343ccef7eff9d43bd47b2d2a266fc5d9d74d04", "output_hash": "6ebf362ebc451a5b2a0d5ab6fc41db9fea75444a45cc75aa55fa0a34bc04be37"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "b7b2ba63039c798b3498570f", "attempt_id": "a55fa4a71778", "run_group_id": "E4_core:A5|closed|qwen3b-llama3b:h4", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:15:31.599908Z", "task_family": "synthetic", "document_id": "syn_03018_2000", "document_hash": "d284eb07e27da742b9c2b67abccfdf72e4dd67c3259f44f3b41e34dd3e6fc1ff", "question_id": "q12_unit_conversion", "split": "test", "cluster_id": "syn_03018_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "tokenizer_hashes": {"receiver": "b9d8872cf3a99da99e1fc4b0f71232a451632dc49ca4c6771549b4efa5c1c763"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 4, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1773, "query_tokens": 468, "message_tokens": 357, "message_utf8_bytes": 1012, "cache_tensor_bytes": null, "serialized_payload_bytes": 1012, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 21001.85354310088, "hops": {"1": 6929.263000027277, "2": 4441.439792048186, "3": 5179.9550839932635, "4": 4451.195667032152}, "receiver_prefill_ms": 206.066417042166, "ttft_ms": 11.779749998822808, "decode_ms": 11.574832955375314}, "total_wall_ms": 21231.797043117695, "ttft_ms": 11.779749998822808, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 64, "facts_retained": 13, "retention_rate": 0.203125, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.2222222222222222, "amount": 0.15384615384615385, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1012}, "ledger_by_hop": {"1": {"facts_total": 64, "facts_retained": 14, "retention_rate": 0.21875, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.2222222222222222, "amount": 0.15384615384615385, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 870}, "2": {"facts_total": 64, "facts_retained": 13, "retention_rate": 0.203125, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.2222222222222222, "amount": 0.15384615384615385, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 899}, "3": {"facts_total": 64, "facts_retained": 13, "retention_rate": 0.203125, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.2222222222222222, "amount": 0.15384615384615385, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 899}, "4": {"facts_total": 64, "facts_retained": 13, "retention_rate": 0.203125, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.2222222222222222, "amount": 0.15384615384615385, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1012}}}, "reference_key_id": "keys/synthetic/syn_03018_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "549e967d2d3d6d2623cc1b9a89343ccef7eff9d43bd47b2d2a266fc5d9d74d04", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "0f87dc927ad8052424bbf50a", "attempt_id": "70505dcd2ef1", "run_group_id": "E4_core:A5|closed|qwen3b-llama3b:h4", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:15:31.845458Z", "task_family": "synthetic", "document_id": "syn_03018_2000", "document_hash": "d284eb07e27da742b9c2b67abccfdf72e4dd67c3259f44f3b41e34dd3e6fc1ff", "question_id": "q13_unanswerable", "split": "test", "cluster_id": "syn_03018_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "tokenizer_hashes": {"receiver": "b9d8872cf3a99da99e1fc4b0f71232a451632dc49ca4c6771549b4efa5c1c763"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 4, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1773, "query_tokens": 473, "message_tokens": 357, "message_utf8_bytes": 1012, "cache_tensor_bytes": null, "serialized_payload_bytes": 1012, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 21001.85354310088, "hops": {"1": 6929.263000027277, "2": 4441.439792048186, "3": 5179.9550839932635, "4": 4451.195667032152}, "receiver_prefill_ms": 204.85804101917893, "ttft_ms": 11.64720905944705, "decode_ms": 11.716834036633372}, "total_wall_ms": 21230.458959005773, "ttft_ms": 11.64720905944705, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 64, "facts_retained": 13, "retention_rate": 0.203125, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.2222222222222222, "amount": 0.15384615384615385, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1012}, "ledger_by_hop": {"1": {"facts_total": 64, "facts_retained": 14, "retention_rate": 0.21875, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.2222222222222222, "amount": 0.15384615384615385, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 870}, "2": {"facts_total": 64, "facts_retained": 13, "retention_rate": 0.203125, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.2222222222222222, "amount": 0.15384615384615385, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 899}, "3": {"facts_total": 64, "facts_retained": 13, "retention_rate": 0.203125, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.2222222222222222, "amount": 0.15384615384615385, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 899}, "4": {"facts_total": 64, "facts_retained": 13, "retention_rate": 0.203125, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.2222222222222222, "amount": 0.15384615384615385, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1012}}}, "reference_key_id": "keys/synthetic/syn_03018_2000.json", "task_scores": {"correct": 1.0, "abstained": 1.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "549e967d2d3d6d2623cc1b9a89343ccef7eff9d43bd47b2d2a266fc5d9d74d04", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "48cf5d8c792009f1a7165262", "attempt_id": "07e16a8da70e", "run_group_id": "E4_core:A5|closed|qwen3b-llama3b:h4", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:15:32.089274Z", "task_family": "synthetic", "document_id": "syn_03018_2000", "document_hash": "d284eb07e27da742b9c2b67abccfdf72e4dd67c3259f44f3b41e34dd3e6fc1ff", "question_id": "q14_unanswerable", "split": "test", "cluster_id": "syn_03018_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "tokenizer_hashes": {"receiver": "b9d8872cf3a99da99e1fc4b0f71232a451632dc49ca4c6771549b4efa5c1c763"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 4, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1773, "query_tokens": 474, "message_tokens": 357, "message_utf8_bytes": 1012, "cache_tensor_bytes": null, "serialized_payload_bytes": 1012, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 21001.85354310088, "hops": {"1": 6929.263000027277, "2": 4441.439792048186, "3": 5179.9550839932635, "4": 4451.195667032152}, "receiver_prefill_ms": 205.25574998464435, "ttft_ms": 11.596917058341205, "decode_ms": 11.584667023271322}, "total_wall_ms": 21230.6735010352, "ttft_ms": 11.596917058341205, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 64, "facts_retained": 13, "retention_rate": 0.203125, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.2222222222222222, "amount": 0.15384615384615385, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1012}, "ledger_by_hop": {"1": {"facts_total": 64, "facts_retained": 14, "retention_rate": 0.21875, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.2222222222222222, "amount": 0.15384615384615385, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 870}, "2": {"facts_total": 64, "facts_retained": 13, "retention_rate": 0.203125, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.2222222222222222, "amount": 0.15384615384615385, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 899}, "3": {"facts_total": 64, "facts_retained": 13, "retention_rate": 0.203125, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.2222222222222222, "amount": 0.15384615384615385, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 899}, "4": {"facts_total": 64, "facts_retained": 13, "retention_rate": 0.203125, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.2222222222222222, "amount": 0.15384615384615385, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1012}}}, "reference_key_id": "keys/synthetic/syn_03018_2000.json", "task_scores": {"correct": 1.0, "abstained": 1.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "549e967d2d3d6d2623cc1b9a89343ccef7eff9d43bd47b2d2a266fc5d9d74d04", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "9ba0d073dc74415deedb7dff", "attempt_id": "ba00b7b5d324", "run_group_id": "E4_core:A5|closed|qwen3b-llama3b:h8", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:15:51.242114Z", "task_family": "synthetic", "document_id": "syn_03018_2000", "document_hash": "d284eb07e27da742b9c2b67abccfdf72e4dd67c3259f44f3b41e34dd3e6fc1ff", "question_id": "q00_single_fact", "split": "test", "cluster_id": "syn_03018_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "tokenizer_hashes": {"receiver": "b9d8872cf3a99da99e1fc4b0f71232a451632dc49ca4c6771549b4efa5c1c763"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 8, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1773, "query_tokens": 460, "message_tokens": 357, "message_utf8_bytes": 1012, "cache_tensor_bytes": null, "serialized_payload_bytes": 1012, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 39906.50916902814, "hops": {"1": 6929.263000027277, "2": 4441.439792048186, "3": 5179.9550839932635, "4": 4451.195667032152, "5": 5003.116499981843, "6": 4448.355708969757, "7": 4999.658874934539, "8": 4453.524542041123}, "receiver_prefill_ms": 201.5507910400629, "ttft_ms": 11.598624987527728, "decode_ms": 11.487292009405792}, "total_wall_ms": 40131.5859190654, "ttft_ms": 11.598624987527728, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 64, "facts_retained": 13, "retention_rate": 0.203125, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.2222222222222222, "amount": 0.15384615384615385, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1012}, "ledger_by_hop": {"1": {"facts_total": 64, "facts_retained": 14, "retention_rate": 0.21875, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.2222222222222222, "amount": 0.15384615384615385, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 870}, "2": {"facts_total": 64, "facts_retained": 13, "retention_rate": 0.203125, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.2222222222222222, "amount": 0.15384615384615385, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 899}, "3": {"facts_total": 64, "facts_retained": 13, "retention_rate": 0.203125, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.2222222222222222, "amount": 0.15384615384615385, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 899}, "4": {"facts_total": 64, "facts_retained": 13, "retention_rate": 0.203125, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.2222222222222222, "amount": 0.15384615384615385, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1012}, "5": {"facts_total": 64, "facts_retained": 13, "retention_rate": 0.203125, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.2222222222222222, "amount": 0.15384615384615385, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1012}, "6": {"facts_total": 64, "facts_retained": 13, "retention_rate": 0.203125, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.2222222222222222, "amount": 0.15384615384615385, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1012}, "7": {"facts_total": 64, "facts_retained": 13, "retention_rate": 0.203125, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.2222222222222222, "amount": 0.15384615384615385, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1012}, "8": {"facts_total": 64, "facts_retained": 13, "retention_rate": 0.203125, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.2222222222222222, "amount": 0.15384615384615385, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1012}}}, "reference_key_id": "keys/synthetic/syn_03018_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "549e967d2d3d6d2623cc1b9a89343ccef7eff9d43bd47b2d2a266fc5d9d74d04", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "e3b39001ab1f01c5eb6439bc", "attempt_id": "f1241caf1f3e", "run_group_id": "E4_core:A5|closed|qwen3b-llama3b:h8", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:15:51.483154Z", "task_family": "synthetic", "document_id": "syn_03018_2000", "document_hash": "d284eb07e27da742b9c2b67abccfdf72e4dd67c3259f44f3b41e34dd3e6fc1ff", "question_id": "q01_single_fact", "split": "test", "cluster_id": "syn_03018_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "tokenizer_hashes": {"receiver": "b9d8872cf3a99da99e1fc4b0f71232a451632dc49ca4c6771549b4efa5c1c763"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 8, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1773, "query_tokens": 466, "message_tokens": 357, "message_utf8_bytes": 1012, "cache_tensor_bytes": null, "serialized_payload_bytes": 1012, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 39906.50916902814, "hops": {"1": 6929.263000027277, "2": 4441.439792048186, "3": 5179.9550839932635, "4": 4451.195667032152, "5": 5003.116499981843, "6": 4448.355708969757, "7": 4999.658874934539, "8": 4453.524542041123}, "receiver_prefill_ms": 203.79433408379555, "ttft_ms": 11.580666992813349, "decode_ms": 11.570041999220848}, "total_wall_ms": 40133.848794037476, "ttft_ms": 11.580666992813349, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 64, "facts_retained": 13, "retention_rate": 0.203125, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.2222222222222222, "amount": 0.15384615384615385, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1012}, "ledger_by_hop": {"1": {"facts_total": 64, "facts_retained": 14, "retention_rate": 0.21875, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.2222222222222222, "amount": 0.15384615384615385, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 870}, "2": {"facts_total": 64, "facts_retained": 13, "retention_rate": 0.203125, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.2222222222222222, "amount": 0.15384615384615385, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 899}, "3": {"facts_total": 64, "facts_retained": 13, "retention_rate": 0.203125, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.2222222222222222, "amount": 0.15384615384615385, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 899}, "4": {"facts_total": 64, "facts_retained": 13, "retention_rate": 0.203125, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.2222222222222222, "amount": 0.15384615384615385, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1012}, "5": {"facts_total": 64, "facts_retained": 13, "retention_rate": 0.203125, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.2222222222222222, "amount": 0.15384615384615385, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1012}, "6": {"facts_total": 64, "facts_retained": 13, "retention_rate": 0.203125, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.2222222222222222, "amount": 0.15384615384615385, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1012}, "7": {"facts_total": 64, "facts_retained": 13, "retention_rate": 0.203125, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.2222222222222222, "amount": 0.15384615384615385, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1012}, "8": {"facts_total": 64, "facts_retained": 13, "retention_rate": 0.203125, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.2222222222222222, "amount": 0.15384615384615385, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1012}}}, "reference_key_id": "keys/synthetic/syn_03018_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "549e967d2d3d6d2623cc1b9a89343ccef7eff9d43bd47b2d2a266fc5d9d74d04", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "206842f1fe24c923e471216e", "attempt_id": "d02680eb5344", "run_group_id": "E4_core:A5|closed|qwen3b-llama3b:h8", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:15:51.725791Z", "task_family": "synthetic", "document_id": "syn_03018_2000", "document_hash": "d284eb07e27da742b9c2b67abccfdf72e4dd67c3259f44f3b41e34dd3e6fc1ff", "question_id": "q02_exact_copy", "split": "test", "cluster_id": "syn_03018_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "tokenizer_hashes": {"receiver": "b9d8872cf3a99da99e1fc4b0f71232a451632dc49ca4c6771549b4efa5c1c763"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 8, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1773, "query_tokens": 465, "message_tokens": 357, "message_utf8_bytes": 1012, "cache_tensor_bytes": null, "serialized_payload_bytes": 1012, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 39906.50916902814, "hops": {"1": 6929.263000027277, "2": 4441.439792048186, "3": 5179.9550839932635, "4": 4451.195667032152, "5": 5003.116499981843, "6": 4448.355708969757, "7": 4999.658874934539, "8": 4453.524542041123}, "receiver_prefill_ms": 205.93541592825204, "ttft_ms": 11.586125008761883, "decode_ms": 11.581624974496663}, "total_wall_ms": 40135.987127083354, "ttft_ms": 11.586125008761883, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 64, "facts_retained": 13, "retention_rate": 0.203125, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.2222222222222222, "amount": 0.15384615384615385, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1012}, "ledger_by_hop": {"1": {"facts_total": 64, "facts_retained": 14, "retention_rate": 0.21875, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.2222222222222222, "amount": 0.15384615384615385, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 870}, "2": {"facts_total": 64, "facts_retained": 13, "retention_rate": 0.203125, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.2222222222222222, "amount": 0.15384615384615385, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 899}, "3": {"facts_total": 64, "facts_retained": 13, "retention_rate": 0.203125, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.2222222222222222, "amount": 0.15384615384615385, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 899}, "4": {"facts_total": 64, "facts_retained": 13, "retention_rate": 0.203125, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.2222222222222222, "amount": 0.15384615384615385, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1012}, "5": {"facts_total": 64, "facts_retained": 13, "retention_rate": 0.203125, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.2222222222222222, "amount": 0.15384615384615385, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1012}, "6": {"facts_total": 64, "facts_retained": 13, "retention_rate": 0.203125, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.2222222222222222, "amount": 0.15384615384615385, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1012}, "7": {"facts_total": 64, "facts_retained": 13, "retention_rate": 0.203125, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.2222222222222222, "amount": 0.15384615384615385, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1012}, "8": {"facts_total": 64, "facts_retained": 13, "retention_rate": 0.203125, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.2222222222222222, "amount": 0.15384615384615385, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1012}}}, "reference_key_id": "keys/synthetic/syn_03018_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "549e967d2d3d6d2623cc1b9a89343ccef7eff9d43bd47b2d2a266fc5d9d74d04", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "f34c138338da40e639113efd", "attempt_id": "c471ae4256be", "run_group_id": "E4_core:A5|closed|qwen3b-llama3b:h8", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:15:51.969697Z", "task_family": "synthetic", "document_id": "syn_03018_2000", "document_hash": "d284eb07e27da742b9c2b67abccfdf72e4dd67c3259f44f3b41e34dd3e6fc1ff", "question_id": "q03_exact_copy", "split": "test", "cluster_id": "syn_03018_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "tokenizer_hashes": {"receiver": "b9d8872cf3a99da99e1fc4b0f71232a451632dc49ca4c6771549b4efa5c1c763"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 8, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1773, "query_tokens": 462, "message_tokens": 357, "message_utf8_bytes": 1012, "cache_tensor_bytes": null, "serialized_payload_bytes": 1012, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 39906.50916902814, "hops": {"1": 6929.263000027277, "2": 4441.439792048186, "3": 5179.9550839932635, "4": 4451.195667032152, "5": 5003.116499981843, "6": 4448.355708969757, "7": 4999.658874934539, "8": 4453.524542041123}, "receiver_prefill_ms": 201.30966696888208, "ttft_ms": 11.90291706006974, "decode_ms": 103.94812503363937}, "total_wall_ms": 40224.15137698408, "ttft_ms": 11.90291706006974, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 9}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "Priyamvada Sörlie", "evidence_artifact": {"ledger": {"facts_total": 64, "facts_retained": 13, "retention_rate": 0.203125, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.2222222222222222, "amount": 0.15384615384615385, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1012}, "ledger_by_hop": {"1": {"facts_total": 64, "facts_retained": 14, "retention_rate": 0.21875, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.2222222222222222, "amount": 0.15384615384615385, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 870}, "2": {"facts_total": 64, "facts_retained": 13, "retention_rate": 0.203125, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.2222222222222222, "amount": 0.15384615384615385, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 899}, "3": {"facts_total": 64, "facts_retained": 13, "retention_rate": 0.203125, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.2222222222222222, "amount": 0.15384615384615385, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 899}, "4": {"facts_total": 64, "facts_retained": 13, "retention_rate": 0.203125, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.2222222222222222, "amount": 0.15384615384615385, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1012}, "5": {"facts_total": 64, "facts_retained": 13, "retention_rate": 0.203125, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.2222222222222222, "amount": 0.15384615384615385, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1012}, "6": {"facts_total": 64, "facts_retained": 13, "retention_rate": 0.203125, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.2222222222222222, "amount": 0.15384615384615385, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1012}, "7": {"facts_total": 64, "facts_retained": 13, "retention_rate": 0.203125, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.2222222222222222, "amount": 0.15384615384615385, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1012}, "8": {"facts_total": 64, "facts_retained": 13, "retention_rate": 0.203125, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.2222222222222222, "amount": 0.15384615384615385, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1012}}}, "reference_key_id": "keys/synthetic/syn_03018_2000.json", "task_scores": {"correct": 0.0, "abstained": 0.0, "f1": 0.5}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "549e967d2d3d6d2623cc1b9a89343ccef7eff9d43bd47b2d2a266fc5d9d74d04", "output_hash": "0bc227b0b15021fbeca665b4eb3e111165a8d8518a5987c4a83aee81df5f53ee"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "ae223bac5e21ebb1ea6f95ee", "attempt_id": "56846200eb03", "run_group_id": "E4_core:A5|closed|qwen3b-llama3b:h8", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:15:52.302377Z", "task_family": "synthetic", "document_id": "syn_03018_2000", "document_hash": "d284eb07e27da742b9c2b67abccfdf72e4dd67c3259f44f3b41e34dd3e6fc1ff", "question_id": "q04_multi_hop", "split": "test", "cluster_id": "syn_03018_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "tokenizer_hashes": {"receiver": "b9d8872cf3a99da99e1fc4b0f71232a451632dc49ca4c6771549b4efa5c1c763"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 8, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1773, "query_tokens": 470, "message_tokens": 357, "message_utf8_bytes": 1012, "cache_tensor_bytes": null, "serialized_payload_bytes": 1012, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 39906.50916902814, "hops": {"1": 6929.263000027277, "2": 4441.439792048186, "3": 5179.9550839932635, "4": 4451.195667032152, "5": 5003.116499981843, "6": 4448.355708969757, "7": 4999.658874934539, "8": 4453.524542041123}, "receiver_prefill_ms": 205.1468340214342, "ttft_ms": 11.636042036116123, "decode_ms": 11.466542026028037}, "total_wall_ms": 40135.158461052924, "ttft_ms": 11.636042036116123, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 64, "facts_retained": 13, "retention_rate": 0.203125, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.2222222222222222, "amount": 0.15384615384615385, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1012}, "ledger_by_hop": {"1": {"facts_total": 64, "facts_retained": 14, "retention_rate": 0.21875, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.2222222222222222, "amount": 0.15384615384615385, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 870}, "2": {"facts_total": 64, "facts_retained": 13, "retention_rate": 0.203125, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.2222222222222222, "amount": 0.15384615384615385, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 899}, "3": {"facts_total": 64, "facts_retained": 13, "retention_rate": 0.203125, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.2222222222222222, "amount": 0.15384615384615385, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 899}, "4": {"facts_total": 64, "facts_retained": 13, "retention_rate": 0.203125, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.2222222222222222, "amount": 0.15384615384615385, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1012}, "5": {"facts_total": 64, "facts_retained": 13, "retention_rate": 0.203125, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.2222222222222222, "amount": 0.15384615384615385, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1012}, "6": {"facts_total": 64, "facts_retained": 13, "retention_rate": 0.203125, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.2222222222222222, "amount": 0.15384615384615385, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1012}, "7": {"facts_total": 64, "facts_retained": 13, "retention_rate": 0.203125, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.2222222222222222, "amount": 0.15384615384615385, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1012}, "8": {"facts_total": 64, "facts_retained": 13, "retention_rate": 0.203125, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.2222222222222222, "amount": 0.15384615384615385, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1012}}}, "reference_key_id": "keys/synthetic/syn_03018_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "549e967d2d3d6d2623cc1b9a89343ccef7eff9d43bd47b2d2a266fc5d9d74d04", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "eaa822db387622926a01943d", "attempt_id": "4b39127af540", "run_group_id": "E4_core:A5|closed|qwen3b-llama3b:h8", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:15:52.546023Z", "task_family": "synthetic", "document_id": "syn_03018_2000", "document_hash": "d284eb07e27da742b9c2b67abccfdf72e4dd67c3259f44f3b41e34dd3e6fc1ff", "question_id": "q05_multi_hop", "split": "test", "cluster_id": "syn_03018_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "tokenizer_hashes": {"receiver": "b9d8872cf3a99da99e1fc4b0f71232a451632dc49ca4c6771549b4efa5c1c763"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 8, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1773, "query_tokens": 468, "message_tokens": 357, "message_utf8_bytes": 1012, "cache_tensor_bytes": null, "serialized_payload_bytes": 1012, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 39906.50916902814, "hops": {"1": 6929.263000027277, "2": 4441.439792048186, "3": 5179.9550839932635, "4": 4451.195667032152, "5": 5003.116499981843, "6": 4448.355708969757, "7": 4999.658874934539, "8": 4453.524542041123}, "receiver_prefill_ms": 205.47741698101163, "ttft_ms": 11.822582921013236, "decode_ms": 11.609124951064587}, "total_wall_ms": 40135.98429400008, "ttft_ms": 11.822582921013236, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 64, "facts_retained": 13, "retention_rate": 0.203125, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.2222222222222222, "amount": 0.15384615384615385, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1012}, "ledger_by_hop": {"1": {"facts_total": 64, "facts_retained": 14, "retention_rate": 0.21875, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.2222222222222222, "amount": 0.15384615384615385, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 870}, "2": {"facts_total": 64, "facts_retained": 13, "retention_rate": 0.203125, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.2222222222222222, "amount": 0.15384615384615385, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 899}, "3": {"facts_total": 64, "facts_retained": 13, "retention_rate": 0.203125, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.2222222222222222, "amount": 0.15384615384615385, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 899}, "4": {"facts_total": 64, "facts_retained": 13, "retention_rate": 0.203125, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.2222222222222222, "amount": 0.15384615384615385, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1012}, "5": {"facts_total": 64, "facts_retained": 13, "retention_rate": 0.203125, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.2222222222222222, "amount": 0.15384615384615385, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1012}, "6": {"facts_total": 64, "facts_retained": 13, "retention_rate": 0.203125, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.2222222222222222, "amount": 0.15384615384615385, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1012}, "7": {"facts_total": 64, "facts_retained": 13, "retention_rate": 0.203125, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.2222222222222222, "amount": 0.15384615384615385, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1012}, "8": {"facts_total": 64, "facts_retained": 13, "retention_rate": 0.203125, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.2222222222222222, "amount": 0.15384615384615385, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1012}}}, "reference_key_id": "keys/synthetic/syn_03018_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "549e967d2d3d6d2623cc1b9a89343ccef7eff9d43bd47b2d2a266fc5d9d74d04", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "41e53b1f282180067f5559f6", "attempt_id": "dbb25d8dd43b", "run_group_id": "E4_core:A5|closed|qwen3b-llama3b:h8", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:15:52.790322Z", "task_family": "synthetic", "document_id": "syn_03018_2000", "document_hash": "d284eb07e27da742b9c2b67abccfdf72e4dd67c3259f44f3b41e34dd3e6fc1ff", "question_id": "q06_negation", "split": "test", "cluster_id": "syn_03018_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "tokenizer_hashes": {"receiver": "b9d8872cf3a99da99e1fc4b0f71232a451632dc49ca4c6771549b4efa5c1c763"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 8, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1773, "query_tokens": 461, "message_tokens": 357, "message_utf8_bytes": 1012, "cache_tensor_bytes": null, "serialized_payload_bytes": 1012, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 39906.50916902814, "hops": {"1": 6929.263000027277, "2": 4441.439792048186, "3": 5179.9550839932635, "4": 4451.195667032152, "5": 5003.116499981843, "6": 4448.355708969757, "7": 4999.658874934539, "8": 4453.524542041123}, "receiver_prefill_ms": 204.42754204850644, "ttft_ms": 11.834374978207052, "decode_ms": 11.744916904717684}, "total_wall_ms": 40135.096085025, "ttft_ms": 11.834374978207052, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "No", "evidence_artifact": {"ledger": {"facts_total": 64, "facts_retained": 13, "retention_rate": 0.203125, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.2222222222222222, "amount": 0.15384615384615385, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1012}, "ledger_by_hop": {"1": {"facts_total": 64, "facts_retained": 14, "retention_rate": 0.21875, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.2222222222222222, "amount": 0.15384615384615385, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 870}, "2": {"facts_total": 64, "facts_retained": 13, "retention_rate": 0.203125, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.2222222222222222, "amount": 0.15384615384615385, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 899}, "3": {"facts_total": 64, "facts_retained": 13, "retention_rate": 0.203125, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.2222222222222222, "amount": 0.15384615384615385, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 899}, "4": {"facts_total": 64, "facts_retained": 13, "retention_rate": 0.203125, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.2222222222222222, "amount": 0.15384615384615385, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1012}, "5": {"facts_total": 64, "facts_retained": 13, "retention_rate": 0.203125, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.2222222222222222, "amount": 0.15384615384615385, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1012}, "6": {"facts_total": 64, "facts_retained": 13, "retention_rate": 0.203125, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.2222222222222222, "amount": 0.15384615384615385, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1012}, "7": {"facts_total": 64, "facts_retained": 13, "retention_rate": 0.203125, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.2222222222222222, "amount": 0.15384615384615385, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1012}, "8": {"facts_total": 64, "facts_retained": 13, "retention_rate": 0.203125, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.2222222222222222, "amount": 0.15384615384615385, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1012}}}, "reference_key_id": "keys/synthetic/syn_03018_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "549e967d2d3d6d2623cc1b9a89343ccef7eff9d43bd47b2d2a266fc5d9d74d04", "output_hash": "1ea442a134b2a184bd5d40104401f2a37fbc09ccf3f4bc9da161c6099be3691d"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "29b844ba27b0b3d4f48db4d6", "attempt_id": "f2215ed5e313", "run_group_id": "E4_core:A5|closed|qwen3b-llama3b:h8", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:15:53.035792Z", "task_family": "synthetic", "document_id": "syn_03018_2000", "document_hash": "d284eb07e27da742b9c2b67abccfdf72e4dd67c3259f44f3b41e34dd3e6fc1ff", "question_id": "q07_negation", "split": "test", "cluster_id": "syn_03018_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "tokenizer_hashes": {"receiver": "b9d8872cf3a99da99e1fc4b0f71232a451632dc49ca4c6771549b4efa5c1c763"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 8, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1773, "query_tokens": 461, "message_tokens": 357, "message_utf8_bytes": 1012, "cache_tensor_bytes": null, "serialized_payload_bytes": 1012, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 39906.50916902814, "hops": {"1": 6929.263000027277, "2": 4441.439792048186, "3": 5179.9550839932635, "4": 4451.195667032152, "5": 5003.116499981843, "6": 4448.355708969757, "7": 4999.658874934539, "8": 4453.524542041123}, "receiver_prefill_ms": 204.5016250340268, "ttft_ms": 11.67029095813632, "decode_ms": 11.54808304272592}, "total_wall_ms": 40134.70412802417, "ttft_ms": 11.67029095813632, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "No", "evidence_artifact": {"ledger": {"facts_total": 64, "facts_retained": 13, "retention_rate": 0.203125, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.2222222222222222, "amount": 0.15384615384615385, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1012}, "ledger_by_hop": {"1": {"facts_total": 64, "facts_retained": 14, "retention_rate": 0.21875, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.2222222222222222, "amount": 0.15384615384615385, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 870}, "2": {"facts_total": 64, "facts_retained": 13, "retention_rate": 0.203125, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.2222222222222222, "amount": 0.15384615384615385, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 899}, "3": {"facts_total": 64, "facts_retained": 13, "retention_rate": 0.203125, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.2222222222222222, "amount": 0.15384615384615385, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 899}, "4": {"facts_total": 64, "facts_retained": 13, "retention_rate": 0.203125, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.2222222222222222, "amount": 0.15384615384615385, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1012}, "5": {"facts_total": 64, "facts_retained": 13, "retention_rate": 0.203125, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.2222222222222222, "amount": 0.15384615384615385, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1012}, "6": {"facts_total": 64, "facts_retained": 13, "retention_rate": 0.203125, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.2222222222222222, "amount": 0.15384615384615385, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1012}, "7": {"facts_total": 64, "facts_retained": 13, "retention_rate": 0.203125, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.2222222222222222, "amount": 0.15384615384615385, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1012}, "8": {"facts_total": 64, "facts_retained": 13, "retention_rate": 0.203125, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.2222222222222222, "amount": 0.15384615384615385, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1012}}}, "reference_key_id": "keys/synthetic/syn_03018_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "549e967d2d3d6d2623cc1b9a89343ccef7eff9d43bd47b2d2a266fc5d9d74d04", "output_hash": "1ea442a134b2a184bd5d40104401f2a37fbc09ccf3f4bc9da161c6099be3691d"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "cb6bd8d5534f6048de4631bb", "attempt_id": "083e83a1a8c3", "run_group_id": "E4_core:A5|closed|qwen3b-llama3b:h8", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:15:53.279068Z", "task_family": "synthetic", "document_id": "syn_03018_2000", "document_hash": "d284eb07e27da742b9c2b67abccfdf72e4dd67c3259f44f3b41e34dd3e6fc1ff", "question_id": "q08_revision", "split": "test", "cluster_id": "syn_03018_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "tokenizer_hashes": {"receiver": "b9d8872cf3a99da99e1fc4b0f71232a451632dc49ca4c6771549b4efa5c1c763"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 8, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1773, "query_tokens": 477, "message_tokens": 357, "message_utf8_bytes": 1012, "cache_tensor_bytes": null, "serialized_payload_bytes": 1012, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 39906.50916902814, "hops": {"1": 6929.263000027277, "2": 4441.439792048186, "3": 5179.9550839932635, "4": 4451.195667032152, "5": 5003.116499981843, "6": 4448.355708969757, "7": 4999.658874934539, "8": 4453.524542041123}, "receiver_prefill_ms": 208.3051250083372, "ttft_ms": 11.646791943348944, "decode_ms": 34.72245798911899}, "total_wall_ms": 40161.59912699368, "ttft_ms": 11.646791943348944, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 3}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "272,184", "evidence_artifact": {"ledger": {"facts_total": 64, "facts_retained": 13, "retention_rate": 0.203125, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.2222222222222222, "amount": 0.15384615384615385, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1012}, "ledger_by_hop": {"1": {"facts_total": 64, "facts_retained": 14, "retention_rate": 0.21875, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.2222222222222222, "amount": 0.15384615384615385, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 870}, "2": {"facts_total": 64, "facts_retained": 13, "retention_rate": 0.203125, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.2222222222222222, "amount": 0.15384615384615385, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 899}, "3": {"facts_total": 64, "facts_retained": 13, "retention_rate": 0.203125, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.2222222222222222, "amount": 0.15384615384615385, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 899}, "4": {"facts_total": 64, "facts_retained": 13, "retention_rate": 0.203125, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.2222222222222222, "amount": 0.15384615384615385, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1012}, "5": {"facts_total": 64, "facts_retained": 13, "retention_rate": 0.203125, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.2222222222222222, "amount": 0.15384615384615385, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1012}, "6": {"facts_total": 64, "facts_retained": 13, "retention_rate": 0.203125, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.2222222222222222, "amount": 0.15384615384615385, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1012}, "7": {"facts_total": 64, "facts_retained": 13, "retention_rate": 0.203125, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.2222222222222222, "amount": 0.15384615384615385, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1012}, "8": {"facts_total": 64, "facts_retained": 13, "retention_rate": 0.203125, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.2222222222222222, "amount": 0.15384615384615385, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1012}}}, "reference_key_id": "keys/synthetic/syn_03018_2000.json", "task_scores": {"correct": 0.0, "abstained": 0.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "549e967d2d3d6d2623cc1b9a89343ccef7eff9d43bd47b2d2a266fc5d9d74d04", "output_hash": "7ab8ea46f050f78e243a876f64862b441080883dc7365da1e4abc893edc12fb0"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "8f30d9123f90730e06e97fdd", "attempt_id": "a88c8070ad8f", "run_group_id": "E4_core:A5|closed|qwen3b-llama3b:h8", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:15:53.548929Z", "task_family": "synthetic", "document_id": "syn_03018_2000", "document_hash": "d284eb07e27da742b9c2b67abccfdf72e4dd67c3259f44f3b41e34dd3e6fc1ff", "question_id": "q09_revision", "split": "test", "cluster_id": "syn_03018_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "tokenizer_hashes": {"receiver": "b9d8872cf3a99da99e1fc4b0f71232a451632dc49ca4c6771549b4efa5c1c763"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 8, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1773, "query_tokens": 470, "message_tokens": 357, "message_utf8_bytes": 1012, "cache_tensor_bytes": null, "serialized_payload_bytes": 1012, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 39906.50916902814, "hops": {"1": 6929.263000027277, "2": 4441.439792048186, "3": 5179.9550839932635, "4": 4451.195667032152, "5": 5003.116499981843, "6": 4448.355708969757, "7": 4999.658874934539, "8": 4453.524542041123}, "receiver_prefill_ms": 206.4537089318037, "ttft_ms": 11.582499952055514, "decode_ms": 34.65300006791949}, "total_wall_ms": 40159.59129401017, "ttft_ms": 11.582499952055514, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 3}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "272,184", "evidence_artifact": {"ledger": {"facts_total": 64, "facts_retained": 13, "retention_rate": 0.203125, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.2222222222222222, "amount": 0.15384615384615385, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1012}, "ledger_by_hop": {"1": {"facts_total": 64, "facts_retained": 14, "retention_rate": 0.21875, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.2222222222222222, "amount": 0.15384615384615385, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 870}, "2": {"facts_total": 64, "facts_retained": 13, "retention_rate": 0.203125, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.2222222222222222, "amount": 0.15384615384615385, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 899}, "3": {"facts_total": 64, "facts_retained": 13, "retention_rate": 0.203125, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.2222222222222222, "amount": 0.15384615384615385, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 899}, "4": {"facts_total": 64, "facts_retained": 13, "retention_rate": 0.203125, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.2222222222222222, "amount": 0.15384615384615385, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1012}, "5": {"facts_total": 64, "facts_retained": 13, "retention_rate": 0.203125, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.2222222222222222, "amount": 0.15384615384615385, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1012}, "6": {"facts_total": 64, "facts_retained": 13, "retention_rate": 0.203125, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.2222222222222222, "amount": 0.15384615384615385, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1012}, "7": {"facts_total": 64, "facts_retained": 13, "retention_rate": 0.203125, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.2222222222222222, "amount": 0.15384615384615385, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1012}, "8": {"facts_total": 64, "facts_retained": 13, "retention_rate": 0.203125, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.2222222222222222, "amount": 0.15384615384615385, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1012}}}, "reference_key_id": "keys/synthetic/syn_03018_2000.json", "task_scores": {"correct": 0.0, "abstained": 0.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "549e967d2d3d6d2623cc1b9a89343ccef7eff9d43bd47b2d2a266fc5d9d74d04", "output_hash": "7ab8ea46f050f78e243a876f64862b441080883dc7365da1e4abc893edc12fb0"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "18ed563c6a445a240acdd9f0", "attempt_id": "eb206133e9d5", "run_group_id": "E4_core:A5|closed|qwen3b-llama3b:h8", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:15:53.816470Z", "task_family": "synthetic", "document_id": "syn_03018_2000", "document_hash": "d284eb07e27da742b9c2b67abccfdf72e4dd67c3259f44f3b41e34dd3e6fc1ff", "question_id": "q10_temporal", "split": "test", "cluster_id": "syn_03018_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "tokenizer_hashes": {"receiver": "b9d8872cf3a99da99e1fc4b0f71232a451632dc49ca4c6771549b4efa5c1c763"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 8, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1773, "query_tokens": 478, "message_tokens": 357, "message_utf8_bytes": 1012, "cache_tensor_bytes": null, "serialized_payload_bytes": 1012, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 39906.50916902814, "hops": {"1": 6929.263000027277, "2": 4441.439792048186, "3": 5179.9550839932635, "4": 4451.195667032152, "5": 5003.116499981843, "6": 4448.355708969757, "7": 4999.658874934539, "8": 4453.524542041123}, "receiver_prefill_ms": 206.67020801920444, "ttft_ms": 11.66870805900544, "decode_ms": 11.429999954998493}, "total_wall_ms": 40136.65537699126, "ttft_ms": 11.66870805900544, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 64, "facts_retained": 13, "retention_rate": 0.203125, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.2222222222222222, "amount": 0.15384615384615385, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1012}, "ledger_by_hop": {"1": {"facts_total": 64, "facts_retained": 14, "retention_rate": 0.21875, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.2222222222222222, "amount": 0.15384615384615385, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 870}, "2": {"facts_total": 64, "facts_retained": 13, "retention_rate": 0.203125, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.2222222222222222, "amount": 0.15384615384615385, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 899}, "3": {"facts_total": 64, "facts_retained": 13, "retention_rate": 0.203125, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.2222222222222222, "amount": 0.15384615384615385, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 899}, "4": {"facts_total": 64, "facts_retained": 13, "retention_rate": 0.203125, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.2222222222222222, "amount": 0.15384615384615385, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1012}, "5": {"facts_total": 64, "facts_retained": 13, "retention_rate": 0.203125, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.2222222222222222, "amount": 0.15384615384615385, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1012}, "6": {"facts_total": 64, "facts_retained": 13, "retention_rate": 0.203125, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.2222222222222222, "amount": 0.15384615384615385, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1012}, "7": {"facts_total": 64, "facts_retained": 13, "retention_rate": 0.203125, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.2222222222222222, "amount": 0.15384615384615385, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1012}, "8": {"facts_total": 64, "facts_retained": 13, "retention_rate": 0.203125, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.2222222222222222, "amount": 0.15384615384615385, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1012}}}, "reference_key_id": "keys/synthetic/syn_03018_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "549e967d2d3d6d2623cc1b9a89343ccef7eff9d43bd47b2d2a266fc5d9d74d04", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "ea39b2dab0d52a1300de4a36", "attempt_id": "7873510ccbe1", "run_group_id": "E4_core:A5|closed|qwen3b-llama3b:h8", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:15:54.061201Z", "task_family": "synthetic", "document_id": "syn_03018_2000", "document_hash": "d284eb07e27da742b9c2b67abccfdf72e4dd67c3259f44f3b41e34dd3e6fc1ff", "question_id": "q11_aggregation", "split": "test", "cluster_id": "syn_03018_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "tokenizer_hashes": {"receiver": "b9d8872cf3a99da99e1fc4b0f71232a451632dc49ca4c6771549b4efa5c1c763"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 8, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1773, "query_tokens": 472, "message_tokens": 357, "message_utf8_bytes": 1012, "cache_tensor_bytes": null, "serialized_payload_bytes": 1012, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 39906.50916902814, "hops": {"1": 6929.263000027277, "2": 4441.439792048186, "3": 5179.9550839932635, "4": 4451.195667032152, "5": 5003.116499981843, "6": 4448.355708969757, "7": 4999.658874934539, "8": 4453.524542041123}, "receiver_prefill_ms": 206.71325002331287, "ttft_ms": 11.643874924629927, "decode_ms": 57.57108307443559}, "total_wall_ms": 40182.859628112055, "ttft_ms": 11.643874924629927, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 5}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "2,189,245", "evidence_artifact": {"ledger": {"facts_total": 64, "facts_retained": 13, "retention_rate": 0.203125, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.2222222222222222, "amount": 0.15384615384615385, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1012}, "ledger_by_hop": {"1": {"facts_total": 64, "facts_retained": 14, "retention_rate": 0.21875, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.2222222222222222, "amount": 0.15384615384615385, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 870}, "2": {"facts_total": 64, "facts_retained": 13, "retention_rate": 0.203125, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.2222222222222222, "amount": 0.15384615384615385, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 899}, "3": {"facts_total": 64, "facts_retained": 13, "retention_rate": 0.203125, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.2222222222222222, "amount": 0.15384615384615385, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 899}, "4": {"facts_total": 64, "facts_retained": 13, "retention_rate": 0.203125, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.2222222222222222, "amount": 0.15384615384615385, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1012}, "5": {"facts_total": 64, "facts_retained": 13, "retention_rate": 0.203125, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.2222222222222222, "amount": 0.15384615384615385, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1012}, "6": {"facts_total": 64, "facts_retained": 13, "retention_rate": 0.203125, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.2222222222222222, "amount": 0.15384615384615385, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1012}, "7": {"facts_total": 64, "facts_retained": 13, "retention_rate": 0.203125, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.2222222222222222, "amount": 0.15384615384615385, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1012}, "8": {"facts_total": 64, "facts_retained": 13, "retention_rate": 0.203125, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.2222222222222222, "amount": 0.15384615384615385, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1012}}}, "reference_key_id": "keys/synthetic/syn_03018_2000.json", "task_scores": {"correct": 0.0, "abstained": 0.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "549e967d2d3d6d2623cc1b9a89343ccef7eff9d43bd47b2d2a266fc5d9d74d04", "output_hash": "6ebf362ebc451a5b2a0d5ab6fc41db9fea75444a45cc75aa55fa0a34bc04be37"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "3dc9d8ec9fcc03d2bdd7f3ed", "attempt_id": "d5ae336e1a9a", "run_group_id": "E4_core:A5|closed|qwen3b-llama3b:h8", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:15:54.352243Z", "task_family": "synthetic", "document_id": "syn_03018_2000", "document_hash": "d284eb07e27da742b9c2b67abccfdf72e4dd67c3259f44f3b41e34dd3e6fc1ff", "question_id": "q12_unit_conversion", "split": "test", "cluster_id": "syn_03018_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "tokenizer_hashes": {"receiver": "b9d8872cf3a99da99e1fc4b0f71232a451632dc49ca4c6771549b4efa5c1c763"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 8, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1773, "query_tokens": 468, "message_tokens": 357, "message_utf8_bytes": 1012, "cache_tensor_bytes": null, "serialized_payload_bytes": 1012, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 39906.50916902814, "hops": {"1": 6929.263000027277, "2": 4441.439792048186, "3": 5179.9550839932635, "4": 4451.195667032152, "5": 5003.116499981843, "6": 4448.355708969757, "7": 4999.658874934539, "8": 4453.524542041123}, "receiver_prefill_ms": 203.85058410465717, "ttft_ms": 11.6886249743402, "decode_ms": 11.592291994020343}, "total_wall_ms": 40134.03454399668, "ttft_ms": 11.6886249743402, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 64, "facts_retained": 13, "retention_rate": 0.203125, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.2222222222222222, "amount": 0.15384615384615385, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1012}, "ledger_by_hop": {"1": {"facts_total": 64, "facts_retained": 14, "retention_rate": 0.21875, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.2222222222222222, "amount": 0.15384615384615385, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 870}, "2": {"facts_total": 64, "facts_retained": 13, "retention_rate": 0.203125, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.2222222222222222, "amount": 0.15384615384615385, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 899}, "3": {"facts_total": 64, "facts_retained": 13, "retention_rate": 0.203125, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.2222222222222222, "amount": 0.15384615384615385, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 899}, "4": {"facts_total": 64, "facts_retained": 13, "retention_rate": 0.203125, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.2222222222222222, "amount": 0.15384615384615385, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1012}, "5": {"facts_total": 64, "facts_retained": 13, "retention_rate": 0.203125, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.2222222222222222, "amount": 0.15384615384615385, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1012}, "6": {"facts_total": 64, "facts_retained": 13, "retention_rate": 0.203125, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.2222222222222222, "amount": 0.15384615384615385, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1012}, "7": {"facts_total": 64, "facts_retained": 13, "retention_rate": 0.203125, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.2222222222222222, "amount": 0.15384615384615385, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1012}, "8": {"facts_total": 64, "facts_retained": 13, "retention_rate": 0.203125, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.2222222222222222, "amount": 0.15384615384615385, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1012}}}, "reference_key_id": "keys/synthetic/syn_03018_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "549e967d2d3d6d2623cc1b9a89343ccef7eff9d43bd47b2d2a266fc5d9d74d04", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "066159548d9a2e9bac152de0", "attempt_id": "765c9ef17c49", "run_group_id": "E4_core:A5|closed|qwen3b-llama3b:h8", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:15:54.594542Z", "task_family": "synthetic", "document_id": "syn_03018_2000", "document_hash": "d284eb07e27da742b9c2b67abccfdf72e4dd67c3259f44f3b41e34dd3e6fc1ff", "question_id": "q13_unanswerable", "split": "test", "cluster_id": "syn_03018_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "tokenizer_hashes": {"receiver": "b9d8872cf3a99da99e1fc4b0f71232a451632dc49ca4c6771549b4efa5c1c763"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 8, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1773, "query_tokens": 473, "message_tokens": 357, "message_utf8_bytes": 1012, "cache_tensor_bytes": null, "serialized_payload_bytes": 1012, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 39906.50916902814, "hops": {"1": 6929.263000027277, "2": 4441.439792048186, "3": 5179.9550839932635, "4": 4451.195667032152, "5": 5003.116499981843, "6": 4448.355708969757, "7": 4999.658874934539, "8": 4453.524542041123}, "receiver_prefill_ms": 203.87283305171877, "ttft_ms": 11.674707988277078, "decode_ms": 11.582250008359551}, "total_wall_ms": 40134.00783610996, "ttft_ms": 11.674707988277078, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 64, "facts_retained": 13, "retention_rate": 0.203125, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.2222222222222222, "amount": 0.15384615384615385, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1012}, "ledger_by_hop": {"1": {"facts_total": 64, "facts_retained": 14, "retention_rate": 0.21875, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.2222222222222222, "amount": 0.15384615384615385, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 870}, "2": {"facts_total": 64, "facts_retained": 13, "retention_rate": 0.203125, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.2222222222222222, "amount": 0.15384615384615385, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 899}, "3": {"facts_total": 64, "facts_retained": 13, "retention_rate": 0.203125, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.2222222222222222, "amount": 0.15384615384615385, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 899}, "4": {"facts_total": 64, "facts_retained": 13, "retention_rate": 0.203125, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.2222222222222222, "amount": 0.15384615384615385, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1012}, "5": {"facts_total": 64, "facts_retained": 13, "retention_rate": 0.203125, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.2222222222222222, "amount": 0.15384615384615385, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1012}, "6": {"facts_total": 64, "facts_retained": 13, "retention_rate": 0.203125, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.2222222222222222, "amount": 0.15384615384615385, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1012}, "7": {"facts_total": 64, "facts_retained": 13, "retention_rate": 0.203125, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.2222222222222222, "amount": 0.15384615384615385, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1012}, "8": {"facts_total": 64, "facts_retained": 13, "retention_rate": 0.203125, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.2222222222222222, "amount": 0.15384615384615385, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1012}}}, "reference_key_id": "keys/synthetic/syn_03018_2000.json", "task_scores": {"correct": 1.0, "abstained": 1.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "549e967d2d3d6d2623cc1b9a89343ccef7eff9d43bd47b2d2a266fc5d9d74d04", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "9f10750200ec745f23d817c3", "attempt_id": "f8ed4de254c0", "run_group_id": "E4_core:A5|closed|qwen3b-llama3b:h8", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:15:54.836867Z", "task_family": "synthetic", "document_id": "syn_03018_2000", "document_hash": "d284eb07e27da742b9c2b67abccfdf72e4dd67c3259f44f3b41e34dd3e6fc1ff", "question_id": "q14_unanswerable", "split": "test", "cluster_id": "syn_03018_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "tokenizer_hashes": {"receiver": "b9d8872cf3a99da99e1fc4b0f71232a451632dc49ca4c6771549b4efa5c1c763"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 8, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1773, "query_tokens": 474, "message_tokens": 357, "message_utf8_bytes": 1012, "cache_tensor_bytes": null, "serialized_payload_bytes": 1012, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 39906.50916902814, "hops": {"1": 6929.263000027277, "2": 4441.439792048186, "3": 5179.9550839932635, "4": 4451.195667032152, "5": 5003.116499981843, "6": 4448.355708969757, "7": 4999.658874934539, "8": 4453.524542041123}, "receiver_prefill_ms": 206.0652090003714, "ttft_ms": 11.582790990360081, "decode_ms": 11.608042055740952}, "total_wall_ms": 40136.132378014736, "ttft_ms": 11.582790990360081, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 64, "facts_retained": 13, "retention_rate": 0.203125, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.2222222222222222, "amount": 0.15384615384615385, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1012}, "ledger_by_hop": {"1": {"facts_total": 64, "facts_retained": 14, "retention_rate": 0.21875, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.2222222222222222, "amount": 0.15384615384615385, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 870}, "2": {"facts_total": 64, "facts_retained": 13, "retention_rate": 0.203125, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.2222222222222222, "amount": 0.15384615384615385, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 899}, "3": {"facts_total": 64, "facts_retained": 13, "retention_rate": 0.203125, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.2222222222222222, "amount": 0.15384615384615385, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 899}, "4": {"facts_total": 64, "facts_retained": 13, "retention_rate": 0.203125, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.2222222222222222, "amount": 0.15384615384615385, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1012}, "5": {"facts_total": 64, "facts_retained": 13, "retention_rate": 0.203125, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.2222222222222222, "amount": 0.15384615384615385, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1012}, "6": {"facts_total": 64, "facts_retained": 13, "retention_rate": 0.203125, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.2222222222222222, "amount": 0.15384615384615385, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1012}, "7": {"facts_total": 64, "facts_retained": 13, "retention_rate": 0.203125, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.2222222222222222, "amount": 0.15384615384615385, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1012}, "8": {"facts_total": 64, "facts_retained": 13, "retention_rate": 0.203125, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.2222222222222222, "amount": 0.15384615384615385, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1012}}}, "reference_key_id": "keys/synthetic/syn_03018_2000.json", "task_scores": {"correct": 1.0, "abstained": 1.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "549e967d2d3d6d2623cc1b9a89343ccef7eff9d43bd47b2d2a266fc5d9d74d04", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "6f2fa11b8debe5fa1828f954", "attempt_id": "c4f9db8162af", "run_group_id": "E4_core:A4|closed|qwen3b:h1", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:16:02.017079Z", "task_family": "synthetic", "document_id": "syn_03021_2000", "document_hash": "3ac205b464552356699a4d2b0a7bf62497ead36e5bca73982e11ae5ee3810731", "question_id": "q00_single_fact", "split": "test", "cluster_id": "syn_03021_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 1, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1914, "query_tokens": 592, "message_tokens": 512, "message_utf8_bytes": 1179, "cache_tensor_bytes": null, "serialized_payload_bytes": 1179, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 6856.280499952845, "hops": {"1": 6856.280499952845}, "receiver_prefill_ms": 226.9969170447439, "ttft_ms": 11.920208926312625, "decode_ms": 11.922957957722247}, "total_wall_ms": 7107.677374966443, "ttft_ms": 11.920208926312625, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 38, "retention_rate": 0.5846153846153846, "retention_by_kind": {"person": 0.8333333333333334, "date": 0.7, "amount": 0.38461538461538464, "status": 1.0, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1179}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 38, "retention_rate": 0.5846153846153846, "retention_by_kind": {"person": 0.8333333333333334, "date": 0.7, "amount": 0.38461538461538464, "status": 1.0, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1179}}}, "reference_key_id": "keys/synthetic/syn_03021_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "f6342dcd02ebd80508f967d8bfed98ca4bc1afeea750629dadac4b492752e270", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "7115831970b20d21f3a36f6e", "attempt_id": "6d2daca99cc7", "run_group_id": "E4_core:A4|closed|qwen3b:h1", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:16:02.283758Z", "task_family": "synthetic", "document_id": "syn_03021_2000", "document_hash": "3ac205b464552356699a4d2b0a7bf62497ead36e5bca73982e11ae5ee3810731", "question_id": "q01_single_fact", "split": "test", "cluster_id": "syn_03021_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 1, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1914, "query_tokens": 598, "message_tokens": 512, "message_utf8_bytes": 1179, "cache_tensor_bytes": null, "serialized_payload_bytes": 1179, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 6856.280499952845, "hops": {"1": 6856.280499952845}, "receiver_prefill_ms": 227.2167910123244, "ttft_ms": 11.98808301705867, "decode_ms": 130.8189999544993}, "total_wall_ms": 7226.821457967162, "ttft_ms": 11.98808301705867, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 11}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "Lot 536-R, Senneterre", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 38, "retention_rate": 0.5846153846153846, "retention_by_kind": {"person": 0.8333333333333334, "date": 0.7, "amount": 0.38461538461538464, "status": 1.0, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1179}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 38, "retention_rate": 0.5846153846153846, "retention_by_kind": {"person": 0.8333333333333334, "date": 0.7, "amount": 0.38461538461538464, "status": 1.0, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1179}}}, "reference_key_id": "keys/synthetic/syn_03021_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 0.8571428571428571}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "f6342dcd02ebd80508f967d8bfed98ca4bc1afeea750629dadac4b492752e270", "output_hash": "5e85af24aac0f1904d70a2fcee84f1b6f6912cc7450298af1f173002fd8e59df"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "4bea80245bb562832cd2515c", "attempt_id": "cfd570786c5e", "run_group_id": "E4_core:A4|closed|qwen3b:h1", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:16:02.669025Z", "task_family": "synthetic", "document_id": "syn_03021_2000", "document_hash": "3ac205b464552356699a4d2b0a7bf62497ead36e5bca73982e11ae5ee3810731", "question_id": "q02_exact_copy", "split": "test", "cluster_id": "syn_03021_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 1, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1914, "query_tokens": 598, "message_tokens": 512, "message_utf8_bytes": 1179, "cache_tensor_bytes": null, "serialized_payload_bytes": 1179, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 6856.280499952845, "hops": {"1": 6856.280499952845}, "receiver_prefill_ms": 227.31175005901605, "ttft_ms": 11.93924993276596, "decode_ms": 83.24283396359533}, "total_wall_ms": 7179.240999976173, "ttft_ms": 11.93924993276596, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 7}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "88.7751", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 38, "retention_rate": 0.5846153846153846, "retention_by_kind": {"person": 0.8333333333333334, "date": 0.7, "amount": 0.38461538461538464, "status": 1.0, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1179}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 38, "retention_rate": 0.5846153846153846, "retention_by_kind": {"person": 0.8333333333333334, "date": 0.7, "amount": 0.38461538461538464, "status": 1.0, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1179}}}, "reference_key_id": "keys/synthetic/syn_03021_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "f6342dcd02ebd80508f967d8bfed98ca4bc1afeea750629dadac4b492752e270", "output_hash": "1e01e62c69f9479b681e50e59e8565c6583f5cdb803446a05ed5cb13118acdf9"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "84af0678ed0fcb28b40020aa", "attempt_id": "681c947991eb", "run_group_id": "E4_core:A4|closed|qwen3b:h1", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:16:03.006877Z", "task_family": "synthetic", "document_id": "syn_03021_2000", "document_hash": "3ac205b464552356699a4d2b0a7bf62497ead36e5bca73982e11ae5ee3810731", "question_id": "q03_exact_copy", "split": "test", "cluster_id": "syn_03021_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 1, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1914, "query_tokens": 595, "message_tokens": 512, "message_utf8_bytes": 1179, "cache_tensor_bytes": null, "serialized_payload_bytes": 1179, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 6856.280499952845, "hops": {"1": 6856.280499952845}, "receiver_prefill_ms": 226.94133303593844, "ttft_ms": 11.952125001698732, "decode_ms": 47.772084013558924}, "total_wall_ms": 7143.445165944286, "ttft_ms": 11.952125001698732, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 4}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "Ingrid Mbeki", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 38, "retention_rate": 0.5846153846153846, "retention_by_kind": {"person": 0.8333333333333334, "date": 0.7, "amount": 0.38461538461538464, "status": 1.0, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1179}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 38, "retention_rate": 0.5846153846153846, "retention_by_kind": {"person": 0.8333333333333334, "date": 0.7, "amount": 0.38461538461538464, "status": 1.0, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1179}}}, "reference_key_id": "keys/synthetic/syn_03021_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "f6342dcd02ebd80508f967d8bfed98ca4bc1afeea750629dadac4b492752e270", "output_hash": "1dbd85090bb65fb3b72de8ca4ce32f2c234729ff4a239274e79fb574b800a79d"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "63e12edbb963cd852c99f786", "attempt_id": "b5d28ae187f7", "run_group_id": "E4_core:A4|closed|qwen3b:h1", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:16:03.308240Z", "task_family": "synthetic", "document_id": "syn_03021_2000", "document_hash": "3ac205b464552356699a4d2b0a7bf62497ead36e5bca73982e11ae5ee3810731", "question_id": "q04_multi_hop", "split": "test", "cluster_id": "syn_03021_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 1, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1914, "query_tokens": 602, "message_tokens": 512, "message_utf8_bytes": 1179, "cache_tensor_bytes": null, "serialized_payload_bytes": 1179, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 6856.280499952845, "hops": {"1": 6856.280499952845}, "receiver_prefill_ms": 227.50533290673047, "ttft_ms": 11.95412501692772, "decode_ms": 12.052333098836243}, "total_wall_ms": 7108.2649999298155, "ttft_ms": 11.95412501692772, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 38, "retention_rate": 0.5846153846153846, "retention_by_kind": {"person": 0.8333333333333334, "date": 0.7, "amount": 0.38461538461538464, "status": 1.0, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1179}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 38, "retention_rate": 0.5846153846153846, "retention_by_kind": {"person": 0.8333333333333334, "date": 0.7, "amount": 0.38461538461538464, "status": 1.0, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1179}}}, "reference_key_id": "keys/synthetic/syn_03021_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "f6342dcd02ebd80508f967d8bfed98ca4bc1afeea750629dadac4b492752e270", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "1bcf76b9c4dce1525fcd5309", "attempt_id": "558edb4d9bd1", "run_group_id": "E4_core:A4|closed|qwen3b:h1", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:16:03.574982Z", "task_family": "synthetic", "document_id": "syn_03021_2000", "document_hash": "3ac205b464552356699a4d2b0a7bf62497ead36e5bca73982e11ae5ee3810731", "question_id": "q05_multi_hop", "split": "test", "cluster_id": "syn_03021_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 1, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1914, "query_tokens": 603, "message_tokens": 512, "message_utf8_bytes": 1179, "cache_tensor_bytes": null, "serialized_payload_bytes": 1179, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 6856.280499952845, "hops": {"1": 6856.280499952845}, "receiver_prefill_ms": 227.0380420377478, "ttft_ms": 12.014499981887639, "decode_ms": 130.60462509747595}, "total_wall_ms": 7226.404333021492, "ttft_ms": 12.014499981887639, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 11}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "BQ-59Z-6753", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 38, "retention_rate": 0.5846153846153846, "retention_by_kind": {"person": 0.8333333333333334, "date": 0.7, "amount": 0.38461538461538464, "status": 1.0, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1179}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 38, "retention_rate": 0.5846153846153846, "retention_by_kind": {"person": 0.8333333333333334, "date": 0.7, "amount": 0.38461538461538464, "status": 1.0, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1179}}}, "reference_key_id": "keys/synthetic/syn_03021_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "f6342dcd02ebd80508f967d8bfed98ca4bc1afeea750629dadac4b492752e270", "output_hash": "db250f072c856fe7ddf4bbbd16e20c71b8bad1eef5e4480acd0dac30f8ffe9cd"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "0403642d96aaa614780dc9a0", "attempt_id": "1651d1f631c2", "run_group_id": "E4_core:A4|closed|qwen3b:h1", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:16:03.959226Z", "task_family": "synthetic", "document_id": "syn_03021_2000", "document_hash": "3ac205b464552356699a4d2b0a7bf62497ead36e5bca73982e11ae5ee3810731", "question_id": "q06_negation", "split": "test", "cluster_id": "syn_03021_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 1, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1914, "query_tokens": 593, "message_tokens": 512, "message_utf8_bytes": 1179, "cache_tensor_bytes": null, "serialized_payload_bytes": 1179, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 6856.280499952845, "hops": {"1": 6856.280499952845}, "receiver_prefill_ms": 227.23020799458027, "ttft_ms": 11.898207943886518, "decode_ms": 11.965375044383109}, "total_wall_ms": 7107.868833001703, "ttft_ms": 11.898207943886518, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "no", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 38, "retention_rate": 0.5846153846153846, "retention_by_kind": {"person": 0.8333333333333334, "date": 0.7, "amount": 0.38461538461538464, "status": 1.0, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1179}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 38, "retention_rate": 0.5846153846153846, "retention_by_kind": {"person": 0.8333333333333334, "date": 0.7, "amount": 0.38461538461538464, "status": 1.0, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1179}}}, "reference_key_id": "keys/synthetic/syn_03021_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "f6342dcd02ebd80508f967d8bfed98ca4bc1afeea750629dadac4b492752e270", "output_hash": "9390298f3fb0c5b160498935d79cb139aef28e1c47358b4bbba61862b9c26e59"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "12854c49f1cc2d64a802d53a", "attempt_id": "3c730b7902c7", "run_group_id": "E4_core:A4|closed|qwen3b:h1", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:16:04.225408Z", "task_family": "synthetic", "document_id": "syn_03021_2000", "document_hash": "3ac205b464552356699a4d2b0a7bf62497ead36e5bca73982e11ae5ee3810731", "question_id": "q07_negation", "split": "test", "cluster_id": "syn_03021_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 1, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1914, "query_tokens": 594, "message_tokens": 512, "message_utf8_bytes": 1179, "cache_tensor_bytes": null, "serialized_payload_bytes": 1179, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 6856.280499952845, "hops": {"1": 6856.280499952845}, "receiver_prefill_ms": 226.65870795026422, "ttft_ms": 11.926457984372973, "decode_ms": 11.911834008060396}, "total_wall_ms": 7107.234874856658, "ttft_ms": 11.926457984372973, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "yes", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 38, "retention_rate": 0.5846153846153846, "retention_by_kind": {"person": 0.8333333333333334, "date": 0.7, "amount": 0.38461538461538464, "status": 1.0, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1179}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 38, "retention_rate": 0.5846153846153846, "retention_by_kind": {"person": 0.8333333333333334, "date": 0.7, "amount": 0.38461538461538464, "status": 1.0, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1179}}}, "reference_key_id": "keys/synthetic/syn_03021_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "f6342dcd02ebd80508f967d8bfed98ca4bc1afeea750629dadac4b492752e270", "output_hash": "8a798890fe93817163b10b5f7bd2ca4d25d84c52739a645a889c173eee7d9d3d"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "ad266fa5b9b6e811cd5838ef", "attempt_id": "919221af8cce", "run_group_id": "E4_core:A4|closed|qwen3b:h1", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:16:04.490827Z", "task_family": "synthetic", "document_id": "syn_03021_2000", "document_hash": "3ac205b464552356699a4d2b0a7bf62497ead36e5bca73982e11ae5ee3810731", "question_id": "q08_revision", "split": "test", "cluster_id": "syn_03021_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 1, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1914, "query_tokens": 610, "message_tokens": 512, "message_utf8_bytes": 1179, "cache_tensor_bytes": null, "serialized_payload_bytes": 1179, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 6856.280499952845, "hops": {"1": 6856.280499952845}, "receiver_prefill_ms": 229.2604160029441, "ttft_ms": 11.948207975365222, "decode_ms": 11.89070800319314}, "total_wall_ms": 7109.883374883793, "ttft_ms": 11.948207975365222, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 38, "retention_rate": 0.5846153846153846, "retention_by_kind": {"person": 0.8333333333333334, "date": 0.7, "amount": 0.38461538461538464, "status": 1.0, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1179}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 38, "retention_rate": 0.5846153846153846, "retention_by_kind": {"person": 0.8333333333333334, "date": 0.7, "amount": 0.38461538461538464, "status": 1.0, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1179}}}, "reference_key_id": "keys/synthetic/syn_03021_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "f6342dcd02ebd80508f967d8bfed98ca4bc1afeea750629dadac4b492752e270", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "a46070e2414043c75270c0a7", "attempt_id": "07083e86658e", "run_group_id": "E4_core:A4|closed|qwen3b:h1", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:16:04.758886Z", "task_family": "synthetic", "document_id": "syn_03021_2000", "document_hash": "3ac205b464552356699a4d2b0a7bf62497ead36e5bca73982e11ae5ee3810731", "question_id": "q09_revision", "split": "test", "cluster_id": "syn_03021_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 1, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1914, "query_tokens": 601, "message_tokens": 512, "message_utf8_bytes": 1179, "cache_tensor_bytes": null, "serialized_payload_bytes": 1179, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 6856.280499952845, "hops": {"1": 6856.280499952845}, "receiver_prefill_ms": 227.30816598050296, "ttft_ms": 11.919124983251095, "decode_ms": 11.887500062584877}, "total_wall_ms": 7107.859125011601, "ttft_ms": 11.919124983251095, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 38, "retention_rate": 0.5846153846153846, "retention_by_kind": {"person": 0.8333333333333334, "date": 0.7, "amount": 0.38461538461538464, "status": 1.0, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1179}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 38, "retention_rate": 0.5846153846153846, "retention_by_kind": {"person": 0.8333333333333334, "date": 0.7, "amount": 0.38461538461538464, "status": 1.0, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1179}}}, "reference_key_id": "keys/synthetic/syn_03021_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "f6342dcd02ebd80508f967d8bfed98ca4bc1afeea750629dadac4b492752e270", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "7285ae262964cf104aaea068", "attempt_id": "52aab711ab24", "run_group_id": "E4_core:A4|closed|qwen3b:h1", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:16:05.025072Z", "task_family": "synthetic", "document_id": "syn_03021_2000", "document_hash": "3ac205b464552356699a4d2b0a7bf62497ead36e5bca73982e11ae5ee3810731", "question_id": "q10_temporal", "split": "test", "cluster_id": "syn_03021_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 1, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1914, "query_tokens": 609, "message_tokens": 512, "message_utf8_bytes": 1179, "cache_tensor_bytes": null, "serialized_payload_bytes": 1179, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 6856.280499952845, "hops": {"1": 6856.280499952845}, "receiver_prefill_ms": 227.06900001503527, "ttft_ms": 12.021416914649308, "decode_ms": 59.429083950817585}, "total_wall_ms": 7155.278082937002, "ttft_ms": 12.021416914649308, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 5}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "Kovar-9", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 38, "retention_rate": 0.5846153846153846, "retention_by_kind": {"person": 0.8333333333333334, "date": 0.7, "amount": 0.38461538461538464, "status": 1.0, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1179}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 38, "retention_rate": 0.5846153846153846, "retention_by_kind": {"person": 0.8333333333333334, "date": 0.7, "amount": 0.38461538461538464, "status": 1.0, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1179}}}, "reference_key_id": "keys/synthetic/syn_03021_2000.json", "task_scores": {"correct": 0.0, "abstained": 0.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "f6342dcd02ebd80508f967d8bfed98ca4bc1afeea750629dadac4b492752e270", "output_hash": "c969b5f7547945067f5b07e9f822d0043126326f979d950a8b1b045057fbb818"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "43dd85a530a24848d4ef866d", "attempt_id": "0c2184305b3e", "run_group_id": "E4_core:A4|closed|qwen3b:h1", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:16:05.338354Z", "task_family": "synthetic", "document_id": "syn_03021_2000", "document_hash": "3ac205b464552356699a4d2b0a7bf62497ead36e5bca73982e11ae5ee3810731", "question_id": "q11_aggregation", "split": "test", "cluster_id": "syn_03021_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 1, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1914, "query_tokens": 605, "message_tokens": 512, "message_utf8_bytes": 1179, "cache_tensor_bytes": null, "serialized_payload_bytes": 1179, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 6856.280499952845, "hops": {"1": 6856.280499952845}, "receiver_prefill_ms": 227.4395830463618, "ttft_ms": 11.9680839125067, "decode_ms": 83.41041707899421}, "total_wall_ms": 7179.584666970186, "ttft_ms": 11.9680839125067, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 7}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "1302481", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 38, "retention_rate": 0.5846153846153846, "retention_by_kind": {"person": 0.8333333333333334, "date": 0.7, "amount": 0.38461538461538464, "status": 1.0, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1179}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 38, "retention_rate": 0.5846153846153846, "retention_by_kind": {"person": 0.8333333333333334, "date": 0.7, "amount": 0.38461538461538464, "status": 1.0, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1179}}}, "reference_key_id": "keys/synthetic/syn_03021_2000.json", "task_scores": {"correct": 0.0, "abstained": 0.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "f6342dcd02ebd80508f967d8bfed98ca4bc1afeea750629dadac4b492752e270", "output_hash": "1c130ebf03e5051f031da47a49e8c91caae34e3c2144f451ec045a8e1c85549d"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "6e71a107ec1f597bbe504885", "attempt_id": "5b27d7a26464", "run_group_id": "E4_core:A4|closed|qwen3b:h1", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:16:05.676061Z", "task_family": "synthetic", "document_id": "syn_03021_2000", "document_hash": "3ac205b464552356699a4d2b0a7bf62497ead36e5bca73982e11ae5ee3810731", "question_id": "q12_unit_conversion", "split": "test", "cluster_id": "syn_03021_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 1, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1914, "query_tokens": 600, "message_tokens": 512, "message_utf8_bytes": 1179, "cache_tensor_bytes": null, "serialized_payload_bytes": 1179, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 6856.280499952845, "hops": {"1": 6856.280499952845}, "receiver_prefill_ms": 226.70395800378174, "ttft_ms": 11.91033294890076, "decode_ms": 11.966499965637922}, "total_wall_ms": 7107.375958934426, "ttft_ms": 11.91033294890076, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 38, "retention_rate": 0.5846153846153846, "retention_by_kind": {"person": 0.8333333333333334, "date": 0.7, "amount": 0.38461538461538464, "status": 1.0, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1179}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 38, "retention_rate": 0.5846153846153846, "retention_by_kind": {"person": 0.8333333333333334, "date": 0.7, "amount": 0.38461538461538464, "status": 1.0, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1179}}}, "reference_key_id": "keys/synthetic/syn_03021_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "f6342dcd02ebd80508f967d8bfed98ca4bc1afeea750629dadac4b492752e270", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "dc9f059a6ab44095fade86a7", "attempt_id": "94f1ad178251", "run_group_id": "E4_core:A4|closed|qwen3b:h1", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:16:05.941701Z", "task_family": "synthetic", "document_id": "syn_03021_2000", "document_hash": "3ac205b464552356699a4d2b0a7bf62497ead36e5bca73982e11ae5ee3810731", "question_id": "q13_unanswerable", "split": "test", "cluster_id": "syn_03021_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 1, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1914, "query_tokens": 605, "message_tokens": 512, "message_utf8_bytes": 1179, "cache_tensor_bytes": null, "serialized_payload_bytes": 1179, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 6856.280499952845, "hops": {"1": 6856.280499952845}, "receiver_prefill_ms": 227.0913339452818, "ttft_ms": 11.940250056795776, "decode_ms": 11.923625017516315}, "total_wall_ms": 7107.693374971859, "ttft_ms": 11.940250056795776, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 38, "retention_rate": 0.5846153846153846, "retention_by_kind": {"person": 0.8333333333333334, "date": 0.7, "amount": 0.38461538461538464, "status": 1.0, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1179}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 38, "retention_rate": 0.5846153846153846, "retention_by_kind": {"person": 0.8333333333333334, "date": 0.7, "amount": 0.38461538461538464, "status": 1.0, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1179}}}, "reference_key_id": "keys/synthetic/syn_03021_2000.json", "task_scores": {"correct": 1.0, "abstained": 1.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "f6342dcd02ebd80508f967d8bfed98ca4bc1afeea750629dadac4b492752e270", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "993937fb39ee048267304285", "attempt_id": "8d3b0c29a224", "run_group_id": "E4_core:A4|closed|qwen3b:h1", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:16:06.207492Z", "task_family": "synthetic", "document_id": "syn_03021_2000", "document_hash": "3ac205b464552356699a4d2b0a7bf62497ead36e5bca73982e11ae5ee3810731", "question_id": "q14_unanswerable", "split": "test", "cluster_id": "syn_03021_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 1, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1914, "query_tokens": 607, "message_tokens": 512, "message_utf8_bytes": 1179, "cache_tensor_bytes": null, "serialized_payload_bytes": 1179, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 6856.280499952845, "hops": {"1": 6856.280499952845}, "receiver_prefill_ms": 227.08895802497864, "ttft_ms": 11.945249978452921, "decode_ms": 11.939625022932887}, "total_wall_ms": 7107.757165911607, "ttft_ms": 11.945249978452921, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 38, "retention_rate": 0.5846153846153846, "retention_by_kind": {"person": 0.8333333333333334, "date": 0.7, "amount": 0.38461538461538464, "status": 1.0, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1179}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 38, "retention_rate": 0.5846153846153846, "retention_by_kind": {"person": 0.8333333333333334, "date": 0.7, "amount": 0.38461538461538464, "status": 1.0, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1179}}}, "reference_key_id": "keys/synthetic/syn_03021_2000.json", "task_scores": {"correct": 1.0, "abstained": 1.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "f6342dcd02ebd80508f967d8bfed98ca4bc1afeea750629dadac4b492752e270", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "55aeb289d759185c1873321e", "attempt_id": "ba990af0c5ef", "run_group_id": "E4_core:A4|closed|qwen3b:h2", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:16:12.112527Z", "task_family": "synthetic", "document_id": "syn_03021_2000", "document_hash": "3ac205b464552356699a4d2b0a7bf62497ead36e5bca73982e11ae5ee3810731", "question_id": "q00_single_fact", "split": "test", "cluster_id": "syn_03021_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 2, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1914, "query_tokens": 531, "message_tokens": 451, "message_utf8_bytes": 1078, "cache_tensor_bytes": null, "serialized_payload_bytes": 1078, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 12494.169082958251, "hops": {"1": 6856.280499952845, "2": 5637.888583005406}, "receiver_prefill_ms": 208.29166704788804, "ttft_ms": 11.86662504915148, "decode_ms": 11.971708037890494}, "total_wall_ms": 12726.778249954805, "ttft_ms": 11.86662504915148, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 38, "retention_rate": 0.5846153846153846, "retention_by_kind": {"person": 0.8333333333333334, "date": 0.7, "amount": 0.38461538461538464, "status": 1.0, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1078}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 38, "retention_rate": 0.5846153846153846, "retention_by_kind": {"person": 0.8333333333333334, "date": 0.7, "amount": 0.38461538461538464, "status": 1.0, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1179}, "2": {"facts_total": 65, "facts_retained": 38, "retention_rate": 0.5846153846153846, "retention_by_kind": {"person": 0.8333333333333334, "date": 0.7, "amount": 0.38461538461538464, "status": 1.0, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1078}}}, "reference_key_id": "keys/synthetic/syn_03021_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "c817fcf3104750482692c83aaa525b365e66a6edca57a9076fa77d315bd06a42", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "0841b8214cb67740f9635819", "attempt_id": "24e0814639ae", "run_group_id": "E4_core:A4|closed|qwen3b:h2", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:16:12.360487Z", "task_family": "synthetic", "document_id": "syn_03021_2000", "document_hash": "3ac205b464552356699a4d2b0a7bf62497ead36e5bca73982e11ae5ee3810731", "question_id": "q01_single_fact", "split": "test", "cluster_id": "syn_03021_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 2, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1914, "query_tokens": 537, "message_tokens": 451, "message_utf8_bytes": 1078, "cache_tensor_bytes": null, "serialized_payload_bytes": 1078, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 12494.169082958251, "hops": {"1": 6856.280499952845, "2": 5637.888583005406}, "receiver_prefill_ms": 208.25937495101243, "ttft_ms": 11.8968749884516, "decode_ms": 130.10574993677437}, "total_wall_ms": 12844.901375006884, "ttft_ms": 11.8968749884516, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 11}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "Lot 536-R, Senneterre", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 38, "retention_rate": 0.5846153846153846, "retention_by_kind": {"person": 0.8333333333333334, "date": 0.7, "amount": 0.38461538461538464, "status": 1.0, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1078}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 38, "retention_rate": 0.5846153846153846, "retention_by_kind": {"person": 0.8333333333333334, "date": 0.7, "amount": 0.38461538461538464, "status": 1.0, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1179}, "2": {"facts_total": 65, "facts_retained": 38, "retention_rate": 0.5846153846153846, "retention_by_kind": {"person": 0.8333333333333334, "date": 0.7, "amount": 0.38461538461538464, "status": 1.0, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1078}}}, "reference_key_id": "keys/synthetic/syn_03021_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 0.8571428571428571}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "c817fcf3104750482692c83aaa525b365e66a6edca57a9076fa77d315bd06a42", "output_hash": "5e85af24aac0f1904d70a2fcee84f1b6f6912cc7450298af1f173002fd8e59df"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "70fac0acd309f2ccb402635c", "attempt_id": "d9a4a6f4fbaf", "run_group_id": "E4_core:A4|closed|qwen3b:h2", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:16:12.725787Z", "task_family": "synthetic", "document_id": "syn_03021_2000", "document_hash": "3ac205b464552356699a4d2b0a7bf62497ead36e5bca73982e11ae5ee3810731", "question_id": "q02_exact_copy", "split": "test", "cluster_id": "syn_03021_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 2, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1914, "query_tokens": 537, "message_tokens": 451, "message_utf8_bytes": 1078, "cache_tensor_bytes": null, "serialized_payload_bytes": 1078, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 12494.169082958251, "hops": {"1": 6856.280499952845, "2": 5637.888583005406}, "receiver_prefill_ms": 207.3174169054255, "ttft_ms": 12.307040975429118, "decode_ms": 83.2267920486629}, "total_wall_ms": 12797.655750066042, "ttft_ms": 12.307040975429118, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 7}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "88.7751", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 38, "retention_rate": 0.5846153846153846, "retention_by_kind": {"person": 0.8333333333333334, "date": 0.7, "amount": 0.38461538461538464, "status": 1.0, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1078}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 38, "retention_rate": 0.5846153846153846, "retention_by_kind": {"person": 0.8333333333333334, "date": 0.7, "amount": 0.38461538461538464, "status": 1.0, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1179}, "2": {"facts_total": 65, "facts_retained": 38, "retention_rate": 0.5846153846153846, "retention_by_kind": {"person": 0.8333333333333334, "date": 0.7, "amount": 0.38461538461538464, "status": 1.0, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1078}}}, "reference_key_id": "keys/synthetic/syn_03021_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "c817fcf3104750482692c83aaa525b365e66a6edca57a9076fa77d315bd06a42", "output_hash": "1e01e62c69f9479b681e50e59e8565c6583f5cdb803446a05ed5cb13118acdf9"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "e4b1fa186f3eecbe413eed96", "attempt_id": "85a5b07f7c50", "run_group_id": "E4_core:A4|closed|qwen3b:h2", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:16:13.044422Z", "task_family": "synthetic", "document_id": "syn_03021_2000", "document_hash": "3ac205b464552356699a4d2b0a7bf62497ead36e5bca73982e11ae5ee3810731", "question_id": "q03_exact_copy", "split": "test", "cluster_id": "syn_03021_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 2, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1914, "query_tokens": 534, "message_tokens": 451, "message_utf8_bytes": 1078, "cache_tensor_bytes": null, "serialized_payload_bytes": 1078, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 12494.169082958251, "hops": {"1": 6856.280499952845, "2": 5637.888583005406}, "receiver_prefill_ms": 206.93858398590237, "ttft_ms": 12.103292043320835, "decode_ms": 47.89774992968887}, "total_wall_ms": 12761.691374937072, "ttft_ms": 12.103292043320835, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 4}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "Ingrid Mbeki", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 38, "retention_rate": 0.5846153846153846, "retention_by_kind": {"person": 0.8333333333333334, "date": 0.7, "amount": 0.38461538461538464, "status": 1.0, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1078}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 38, "retention_rate": 0.5846153846153846, "retention_by_kind": {"person": 0.8333333333333334, "date": 0.7, "amount": 0.38461538461538464, "status": 1.0, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1179}, "2": {"facts_total": 65, "facts_retained": 38, "retention_rate": 0.5846153846153846, "retention_by_kind": {"person": 0.8333333333333334, "date": 0.7, "amount": 0.38461538461538464, "status": 1.0, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1078}}}, "reference_key_id": "keys/synthetic/syn_03021_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "c817fcf3104750482692c83aaa525b365e66a6edca57a9076fa77d315bd06a42", "output_hash": "1dbd85090bb65fb3b72de8ca4ce32f2c234729ff4a239274e79fb574b800a79d"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "dfd4c4fc69f43092a4eaef3a", "attempt_id": "b91f2c005433", "run_group_id": "E4_core:A4|closed|qwen3b:h2", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:16:13.327870Z", "task_family": "synthetic", "document_id": "syn_03021_2000", "document_hash": "3ac205b464552356699a4d2b0a7bf62497ead36e5bca73982e11ae5ee3810731", "question_id": "q04_multi_hop", "split": "test", "cluster_id": "syn_03021_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 2, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1914, "query_tokens": 541, "message_tokens": 451, "message_utf8_bytes": 1078, "cache_tensor_bytes": null, "serialized_payload_bytes": 1078, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 12494.169082958251, "hops": {"1": 6856.280499952845, "2": 5637.888583005406}, "receiver_prefill_ms": 208.4630000172183, "ttft_ms": 11.983083095401525, "decode_ms": 11.85241702478379}, "total_wall_ms": 12726.951207965612, "ttft_ms": 11.983083095401525, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 38, "retention_rate": 0.5846153846153846, "retention_by_kind": {"person": 0.8333333333333334, "date": 0.7, "amount": 0.38461538461538464, "status": 1.0, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1078}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 38, "retention_rate": 0.5846153846153846, "retention_by_kind": {"person": 0.8333333333333334, "date": 0.7, "amount": 0.38461538461538464, "status": 1.0, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1179}, "2": {"facts_total": 65, "facts_retained": 38, "retention_rate": 0.5846153846153846, "retention_by_kind": {"person": 0.8333333333333334, "date": 0.7, "amount": 0.38461538461538464, "status": 1.0, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1078}}}, "reference_key_id": "keys/synthetic/syn_03021_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "c817fcf3104750482692c83aaa525b365e66a6edca57a9076fa77d315bd06a42", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "3020c6af46b1f7fca814c3c6", "attempt_id": "6d7d52bae356", "run_group_id": "E4_core:A4|closed|qwen3b:h2", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:16:13.576220Z", "task_family": "synthetic", "document_id": "syn_03021_2000", "document_hash": "3ac205b464552356699a4d2b0a7bf62497ead36e5bca73982e11ae5ee3810731", "question_id": "q05_multi_hop", "split": "test", "cluster_id": "syn_03021_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 2, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1914, "query_tokens": 542, "message_tokens": 451, "message_utf8_bytes": 1078, "cache_tensor_bytes": null, "serialized_payload_bytes": 1078, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 12494.169082958251, "hops": {"1": 6856.280499952845, "2": 5637.888583005406}, "receiver_prefill_ms": 208.38770805858076, "ttft_ms": 11.907624895684421, "decode_ms": 130.5852500954643}, "total_wall_ms": 12845.527249970473, "ttft_ms": 11.907624895684421, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 11}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "BQ-59Z-6753", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 38, "retention_rate": 0.5846153846153846, "retention_by_kind": {"person": 0.8333333333333334, "date": 0.7, "amount": 0.38461538461538464, "status": 1.0, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1078}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 38, "retention_rate": 0.5846153846153846, "retention_by_kind": {"person": 0.8333333333333334, "date": 0.7, "amount": 0.38461538461538464, "status": 1.0, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1179}, "2": {"facts_total": 65, "facts_retained": 38, "retention_rate": 0.5846153846153846, "retention_by_kind": {"person": 0.8333333333333334, "date": 0.7, "amount": 0.38461538461538464, "status": 1.0, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1078}}}, "reference_key_id": "keys/synthetic/syn_03021_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "c817fcf3104750482692c83aaa525b365e66a6edca57a9076fa77d315bd06a42", "output_hash": "db250f072c856fe7ddf4bbbd16e20c71b8bad1eef5e4480acd0dac30f8ffe9cd"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "c1981667253b5f4abb24dc7c", "attempt_id": "e9b99cecb57d", "run_group_id": "E4_core:A4|closed|qwen3b:h2", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:16:13.942404Z", "task_family": "synthetic", "document_id": "syn_03021_2000", "document_hash": "3ac205b464552356699a4d2b0a7bf62497ead36e5bca73982e11ae5ee3810731", "question_id": "q06_negation", "split": "test", "cluster_id": "syn_03021_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 2, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1914, "query_tokens": 532, "message_tokens": 451, "message_utf8_bytes": 1078, "cache_tensor_bytes": null, "serialized_payload_bytes": 1078, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 12494.169082958251, "hops": {"1": 6856.280499952845, "2": 5637.888583005406}, "receiver_prefill_ms": 207.95041602104902, "ttft_ms": 11.83516695164144, "decode_ms": 11.86908408999443}, "total_wall_ms": 12726.296123932116, "ttft_ms": 11.83516695164144, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "no", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 38, "retention_rate": 0.5846153846153846, "retention_by_kind": {"person": 0.8333333333333334, "date": 0.7, "amount": 0.38461538461538464, "status": 1.0, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1078}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 38, "retention_rate": 0.5846153846153846, "retention_by_kind": {"person": 0.8333333333333334, "date": 0.7, "amount": 0.38461538461538464, "status": 1.0, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1179}, "2": {"facts_total": 65, "facts_retained": 38, "retention_rate": 0.5846153846153846, "retention_by_kind": {"person": 0.8333333333333334, "date": 0.7, "amount": 0.38461538461538464, "status": 1.0, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1078}}}, "reference_key_id": "keys/synthetic/syn_03021_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "c817fcf3104750482692c83aaa525b365e66a6edca57a9076fa77d315bd06a42", "output_hash": "9390298f3fb0c5b160498935d79cb139aef28e1c47358b4bbba61862b9c26e59"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "f7cf92fb4444b7ed7e3afea2", "attempt_id": "bc876e2f07b6", "run_group_id": "E4_core:A4|closed|qwen3b:h2", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:16:14.189606Z", "task_family": "synthetic", "document_id": "syn_03021_2000", "document_hash": "3ac205b464552356699a4d2b0a7bf62497ead36e5bca73982e11ae5ee3810731", "question_id": "q07_negation", "split": "test", "cluster_id": "syn_03021_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 2, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1914, "query_tokens": 533, "message_tokens": 451, "message_utf8_bytes": 1078, "cache_tensor_bytes": null, "serialized_payload_bytes": 1078, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 12494.169082958251, "hops": {"1": 6856.280499952845, "2": 5637.888583005406}, "receiver_prefill_ms": 207.9217090504244, "ttft_ms": 12.097125058062375, "decode_ms": 11.866540997289121}, "total_wall_ms": 12726.486125029624, "ttft_ms": 12.097125058062375, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "yes", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 38, "retention_rate": 0.5846153846153846, "retention_by_kind": {"person": 0.8333333333333334, "date": 0.7, "amount": 0.38461538461538464, "status": 1.0, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1078}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 38, "retention_rate": 0.5846153846153846, "retention_by_kind": {"person": 0.8333333333333334, "date": 0.7, "amount": 0.38461538461538464, "status": 1.0, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1179}, "2": {"facts_total": 65, "facts_retained": 38, "retention_rate": 0.5846153846153846, "retention_by_kind": {"person": 0.8333333333333334, "date": 0.7, "amount": 0.38461538461538464, "status": 1.0, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1078}}}, "reference_key_id": "keys/synthetic/syn_03021_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "c817fcf3104750482692c83aaa525b365e66a6edca57a9076fa77d315bd06a42", "output_hash": "8a798890fe93817163b10b5f7bd2ca4d25d84c52739a645a889c173eee7d9d3d"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "6338f2ca221e993f0317d643", "attempt_id": "19d4d8397c00", "run_group_id": "E4_core:A4|closed|qwen3b:h2", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:16:14.436600Z", "task_family": "synthetic", "document_id": "syn_03021_2000", "document_hash": "3ac205b464552356699a4d2b0a7bf62497ead36e5bca73982e11ae5ee3810731", "question_id": "q08_revision", "split": "test", "cluster_id": "syn_03021_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 2, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1914, "query_tokens": 549, "message_tokens": 451, "message_utf8_bytes": 1078, "cache_tensor_bytes": null, "serialized_payload_bytes": 1078, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 12494.169082958251, "hops": {"1": 6856.280499952845, "2": 5637.888583005406}, "receiver_prefill_ms": 210.04316594917327, "ttft_ms": 11.878084042109549, "decode_ms": 11.866459040902555}, "total_wall_ms": 12728.417749982327, "ttft_ms": 11.878084042109549, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 38, "retention_rate": 0.5846153846153846, "retention_by_kind": {"person": 0.8333333333333334, "date": 0.7, "amount": 0.38461538461538464, "status": 1.0, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1078}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 38, "retention_rate": 0.5846153846153846, "retention_by_kind": {"person": 0.8333333333333334, "date": 0.7, "amount": 0.38461538461538464, "status": 1.0, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1179}, "2": {"facts_total": 65, "facts_retained": 38, "retention_rate": 0.5846153846153846, "retention_by_kind": {"person": 0.8333333333333334, "date": 0.7, "amount": 0.38461538461538464, "status": 1.0, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1078}}}, "reference_key_id": "keys/synthetic/syn_03021_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "c817fcf3104750482692c83aaa525b365e66a6edca57a9076fa77d315bd06a42", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "5a71fd4229ccee305604e2f2", "attempt_id": "feca3473e6fb", "run_group_id": "E4_core:A4|closed|qwen3b:h2", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:16:14.685165Z", "task_family": "synthetic", "document_id": "syn_03021_2000", "document_hash": "3ac205b464552356699a4d2b0a7bf62497ead36e5bca73982e11ae5ee3810731", "question_id": "q09_revision", "split": "test", "cluster_id": "syn_03021_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 2, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1914, "query_tokens": 540, "message_tokens": 451, "message_utf8_bytes": 1078, "cache_tensor_bytes": null, "serialized_payload_bytes": 1078, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 12494.169082958251, "hops": {"1": 6856.280499952845, "2": 5637.888583005406}, "receiver_prefill_ms": 208.4308749763295, "ttft_ms": 11.89295796211809, "decode_ms": 11.958708055317402}, "total_wall_ms": 12726.860332884826, "ttft_ms": 11.89295796211809, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 38, "retention_rate": 0.5846153846153846, "retention_by_kind": {"person": 0.8333333333333334, "date": 0.7, "amount": 0.38461538461538464, "status": 1.0, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1078}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 38, "retention_rate": 0.5846153846153846, "retention_by_kind": {"person": 0.8333333333333334, "date": 0.7, "amount": 0.38461538461538464, "status": 1.0, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1179}, "2": {"facts_total": 65, "facts_retained": 38, "retention_rate": 0.5846153846153846, "retention_by_kind": {"person": 0.8333333333333334, "date": 0.7, "amount": 0.38461538461538464, "status": 1.0, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1078}}}, "reference_key_id": "keys/synthetic/syn_03021_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "c817fcf3104750482692c83aaa525b365e66a6edca57a9076fa77d315bd06a42", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "6886f9d3d3a15ee89f9c50e6", "attempt_id": "e4cc70d2e677", "run_group_id": "E4_core:A4|closed|qwen3b:h2", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:16:14.932382Z", "task_family": "synthetic", "document_id": "syn_03021_2000", "document_hash": "3ac205b464552356699a4d2b0a7bf62497ead36e5bca73982e11ae5ee3810731", "question_id": "q10_temporal", "split": "test", "cluster_id": "syn_03021_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 2, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1914, "query_tokens": 548, "message_tokens": 451, "message_utf8_bytes": 1078, "cache_tensor_bytes": null, "serialized_payload_bytes": 1078, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 12494.169082958251, "hops": {"1": 6856.280499952845, "2": 5637.888583005406}, "receiver_prefill_ms": 210.2216249331832, "ttft_ms": 11.851957999169827, "decode_ms": 59.35849994421005}, "total_wall_ms": 12776.020916877314, "ttft_ms": 11.851957999169827, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 5}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "Kovar-9", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 38, "retention_rate": 0.5846153846153846, "retention_by_kind": {"person": 0.8333333333333334, "date": 0.7, "amount": 0.38461538461538464, "status": 1.0, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1078}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 38, "retention_rate": 0.5846153846153846, "retention_by_kind": {"person": 0.8333333333333334, "date": 0.7, "amount": 0.38461538461538464, "status": 1.0, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1179}, "2": {"facts_total": 65, "facts_retained": 38, "retention_rate": 0.5846153846153846, "retention_by_kind": {"person": 0.8333333333333334, "date": 0.7, "amount": 0.38461538461538464, "status": 1.0, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1078}}}, "reference_key_id": "keys/synthetic/syn_03021_2000.json", "task_scores": {"correct": 0.0, "abstained": 0.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "c817fcf3104750482692c83aaa525b365e66a6edca57a9076fa77d315bd06a42", "output_hash": "c969b5f7547945067f5b07e9f822d0043126326f979d950a8b1b045057fbb818"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "282a2d391045ed7c276455e6", "attempt_id": "b75a86b86757", "run_group_id": "E4_core:A4|closed|qwen3b:h2", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:16:15.228614Z", "task_family": "synthetic", "document_id": "syn_03021_2000", "document_hash": "3ac205b464552356699a4d2b0a7bf62497ead36e5bca73982e11ae5ee3810731", "question_id": "q11_aggregation", "split": "test", "cluster_id": "syn_03021_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 2, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1914, "query_tokens": 544, "message_tokens": 451, "message_utf8_bytes": 1078, "cache_tensor_bytes": null, "serialized_payload_bytes": 1078, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 12494.169082958251, "hops": {"1": 6856.280499952845, "2": 5637.888583005406}, "receiver_prefill_ms": 208.58450001105666, "ttft_ms": 11.888750013895333, "decode_ms": 83.24279205407947}, "total_wall_ms": 12798.34287497215, "ttft_ms": 11.888750013895333, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 7}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "1392497", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 38, "retention_rate": 0.5846153846153846, "retention_by_kind": {"person": 0.8333333333333334, "date": 0.7, "amount": 0.38461538461538464, "status": 1.0, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1078}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 38, "retention_rate": 0.5846153846153846, "retention_by_kind": {"person": 0.8333333333333334, "date": 0.7, "amount": 0.38461538461538464, "status": 1.0, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1179}, "2": {"facts_total": 65, "facts_retained": 38, "retention_rate": 0.5846153846153846, "retention_by_kind": {"person": 0.8333333333333334, "date": 0.7, "amount": 0.38461538461538464, "status": 1.0, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1078}}}, "reference_key_id": "keys/synthetic/syn_03021_2000.json", "task_scores": {"correct": 0.0, "abstained": 0.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "c817fcf3104750482692c83aaa525b365e66a6edca57a9076fa77d315bd06a42", "output_hash": "8eee62192f018b9d037d820d43ea01dddf7881d04a58f2b4bf1bd95cc43a4009"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "a5802f940c4e9d5b742693d6", "attempt_id": "332ca67aa172", "run_group_id": "E4_core:A4|closed|qwen3b:h2", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:16:15.547214Z", "task_family": "synthetic", "document_id": "syn_03021_2000", "document_hash": "3ac205b464552356699a4d2b0a7bf62497ead36e5bca73982e11ae5ee3810731", "question_id": "q12_unit_conversion", "split": "test", "cluster_id": "syn_03021_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 2, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1914, "query_tokens": 539, "message_tokens": 451, "message_utf8_bytes": 1078, "cache_tensor_bytes": null, "serialized_payload_bytes": 1078, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 12494.169082958251, "hops": {"1": 6856.280499952845, "2": 5637.888583005406}, "receiver_prefill_ms": 208.2550839986652, "ttft_ms": 11.881167069077492, "decode_ms": 11.858041980303824}, "total_wall_ms": 12726.592498947866, "ttft_ms": 11.881167069077492, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 38, "retention_rate": 0.5846153846153846, "retention_by_kind": {"person": 0.8333333333333334, "date": 0.7, "amount": 0.38461538461538464, "status": 1.0, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1078}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 38, "retention_rate": 0.5846153846153846, "retention_by_kind": {"person": 0.8333333333333334, "date": 0.7, "amount": 0.38461538461538464, "status": 1.0, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1179}, "2": {"facts_total": 65, "facts_retained": 38, "retention_rate": 0.5846153846153846, "retention_by_kind": {"person": 0.8333333333333334, "date": 0.7, "amount": 0.38461538461538464, "status": 1.0, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1078}}}, "reference_key_id": "keys/synthetic/syn_03021_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "c817fcf3104750482692c83aaa525b365e66a6edca57a9076fa77d315bd06a42", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "47505f266dd26746184f0550", "attempt_id": "27a1a2317273", "run_group_id": "E4_core:A4|closed|qwen3b:h2", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:16:15.794162Z", "task_family": "synthetic", "document_id": "syn_03021_2000", "document_hash": "3ac205b464552356699a4d2b0a7bf62497ead36e5bca73982e11ae5ee3810731", "question_id": "q13_unanswerable", "split": "test", "cluster_id": "syn_03021_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 2, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1914, "query_tokens": 544, "message_tokens": 451, "message_utf8_bytes": 1078, "cache_tensor_bytes": null, "serialized_payload_bytes": 1078, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 12494.169082958251, "hops": {"1": 6856.280499952845, "2": 5637.888583005406}, "receiver_prefill_ms": 208.54666701052338, "ttft_ms": 11.907874955795705, "decode_ms": 11.861459002830088}, "total_wall_ms": 12726.910749915987, "ttft_ms": 11.907874955795705, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 38, "retention_rate": 0.5846153846153846, "retention_by_kind": {"person": 0.8333333333333334, "date": 0.7, "amount": 0.38461538461538464, "status": 1.0, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1078}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 38, "retention_rate": 0.5846153846153846, "retention_by_kind": {"person": 0.8333333333333334, "date": 0.7, "amount": 0.38461538461538464, "status": 1.0, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1179}, "2": {"facts_total": 65, "facts_retained": 38, "retention_rate": 0.5846153846153846, "retention_by_kind": {"person": 0.8333333333333334, "date": 0.7, "amount": 0.38461538461538464, "status": 1.0, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1078}}}, "reference_key_id": "keys/synthetic/syn_03021_2000.json", "task_scores": {"correct": 1.0, "abstained": 1.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "c817fcf3104750482692c83aaa525b365e66a6edca57a9076fa77d315bd06a42", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "f4ad3db333572f3b27874399", "attempt_id": "b9fb07b1b86f", "run_group_id": "E4_core:A4|closed|qwen3b:h2", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:16:16.040948Z", "task_family": "synthetic", "document_id": "syn_03021_2000", "document_hash": "3ac205b464552356699a4d2b0a7bf62497ead36e5bca73982e11ae5ee3810731", "question_id": "q14_unanswerable", "split": "test", "cluster_id": "syn_03021_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 2, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1914, "query_tokens": 546, "message_tokens": 451, "message_utf8_bytes": 1078, "cache_tensor_bytes": null, "serialized_payload_bytes": 1078, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 12494.169082958251, "hops": {"1": 6856.280499952845, "2": 5637.888583005406}, "receiver_prefill_ms": 209.87904199864715, "ttft_ms": 11.911874986253679, "decode_ms": 11.905207997187972}, "total_wall_ms": 12728.29570795875, "ttft_ms": 11.911874986253679, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 38, "retention_rate": 0.5846153846153846, "retention_by_kind": {"person": 0.8333333333333334, "date": 0.7, "amount": 0.38461538461538464, "status": 1.0, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1078}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 38, "retention_rate": 0.5846153846153846, "retention_by_kind": {"person": 0.8333333333333334, "date": 0.7, "amount": 0.38461538461538464, "status": 1.0, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1179}, "2": {"facts_total": 65, "facts_retained": 38, "retention_rate": 0.5846153846153846, "retention_by_kind": {"person": 0.8333333333333334, "date": 0.7, "amount": 0.38461538461538464, "status": 1.0, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1078}}}, "reference_key_id": "keys/synthetic/syn_03021_2000.json", "task_scores": {"correct": 1.0, "abstained": 1.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "c817fcf3104750482692c83aaa525b365e66a6edca57a9076fa77d315bd06a42", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "c45ae542f900faeafb233914", "attempt_id": "848c89f75932", "run_group_id": "E4_core:A4|closed|qwen3b:h4", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:16:27.521209Z", "task_family": "synthetic", "document_id": "syn_03021_2000", "document_hash": "3ac205b464552356699a4d2b0a7bf62497ead36e5bca73982e11ae5ee3810731", "question_id": "q00_single_fact", "split": "test", "cluster_id": "syn_03021_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 4, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1914, "query_tokens": 531, "message_tokens": 451, "message_utf8_bytes": 1078, "cache_tensor_bytes": null, "serialized_payload_bytes": 1078, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 23723.33950106986, "hops": {"1": 6856.280499952845, "2": 5637.888583005406, "3": 5615.923584089614, "4": 5613.246834021993}, "receiver_prefill_ms": 207.71895803045481, "ttft_ms": 11.905624996870756, "decode_ms": 11.871958035044372}, "total_wall_ms": 23955.306043033488, "ttft_ms": 11.905624996870756, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 38, "retention_rate": 0.5846153846153846, "retention_by_kind": {"person": 0.8333333333333334, "date": 0.7, "amount": 0.38461538461538464, "status": 1.0, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1078}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 38, "retention_rate": 0.5846153846153846, "retention_by_kind": {"person": 0.8333333333333334, "date": 0.7, "amount": 0.38461538461538464, "status": 1.0, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1179}, "2": {"facts_total": 65, "facts_retained": 38, "retention_rate": 0.5846153846153846, "retention_by_kind": {"person": 0.8333333333333334, "date": 0.7, "amount": 0.38461538461538464, "status": 1.0, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1078}, "3": {"facts_total": 65, "facts_retained": 38, "retention_rate": 0.5846153846153846, "retention_by_kind": {"person": 0.8333333333333334, "date": 0.7, "amount": 0.38461538461538464, "status": 1.0, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1078}, "4": {"facts_total": 65, "facts_retained": 38, "retention_rate": 0.5846153846153846, "retention_by_kind": {"person": 0.8333333333333334, "date": 0.7, "amount": 0.38461538461538464, "status": 1.0, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1078}}}, "reference_key_id": "keys/synthetic/syn_03021_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "c817fcf3104750482692c83aaa525b365e66a6edca57a9076fa77d315bd06a42", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "3b50986d9c1eabbdc67ac697", "attempt_id": "4774bdac5070", "run_group_id": "E4_core:A4|closed|qwen3b:h4", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:16:27.768691Z", "task_family": "synthetic", "document_id": "syn_03021_2000", "document_hash": "3ac205b464552356699a4d2b0a7bf62497ead36e5bca73982e11ae5ee3810731", "question_id": "q01_single_fact", "split": "test", "cluster_id": "syn_03021_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 4, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1914, "query_tokens": 537, "message_tokens": 451, "message_utf8_bytes": 1078, "cache_tensor_bytes": null, "serialized_payload_bytes": 1078, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 23723.33950106986, "hops": {"1": 6856.280499952845, "2": 5637.888583005406, "3": 5615.923584089614, "4": 5613.246834021993}, "receiver_prefill_ms": 208.2517920061946, "ttft_ms": 11.945415986701846, "decode_ms": 130.2836670074612}, "total_wall_ms": 24074.26995912101, "ttft_ms": 11.945415986701846, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 11}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "Lot 536-R, Senneterre", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 38, "retention_rate": 0.5846153846153846, "retention_by_kind": {"person": 0.8333333333333334, "date": 0.7, "amount": 0.38461538461538464, "status": 1.0, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1078}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 38, "retention_rate": 0.5846153846153846, "retention_by_kind": {"person": 0.8333333333333334, "date": 0.7, "amount": 0.38461538461538464, "status": 1.0, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1179}, "2": {"facts_total": 65, "facts_retained": 38, "retention_rate": 0.5846153846153846, "retention_by_kind": {"person": 0.8333333333333334, "date": 0.7, "amount": 0.38461538461538464, "status": 1.0, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1078}, "3": {"facts_total": 65, "facts_retained": 38, "retention_rate": 0.5846153846153846, "retention_by_kind": {"person": 0.8333333333333334, "date": 0.7, "amount": 0.38461538461538464, "status": 1.0, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1078}, "4": {"facts_total": 65, "facts_retained": 38, "retention_rate": 0.5846153846153846, "retention_by_kind": {"person": 0.8333333333333334, "date": 0.7, "amount": 0.38461538461538464, "status": 1.0, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1078}}}, "reference_key_id": "keys/synthetic/syn_03021_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 0.8571428571428571}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "c817fcf3104750482692c83aaa525b365e66a6edca57a9076fa77d315bd06a42", "output_hash": "5e85af24aac0f1904d70a2fcee84f1b6f6912cc7450298af1f173002fd8e59df"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "6c519a61f78a29b044c79ecd", "attempt_id": "66ea42ce12f2", "run_group_id": "E4_core:A4|closed|qwen3b:h4", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:16:28.134371Z", "task_family": "synthetic", "document_id": "syn_03021_2000", "document_hash": "3ac205b464552356699a4d2b0a7bf62497ead36e5bca73982e11ae5ee3810731", "question_id": "q02_exact_copy", "split": "test", "cluster_id": "syn_03021_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 4, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1914, "query_tokens": 537, "message_tokens": 451, "message_utf8_bytes": 1078, "cache_tensor_bytes": null, "serialized_payload_bytes": 1078, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 23723.33950106986, "hops": {"1": 6856.280499952845, "2": 5637.888583005406, "3": 5615.923584089614, "4": 5613.246834021993}, "receiver_prefill_ms": 208.372832974419, "ttft_ms": 11.885083047673106, "decode_ms": 82.90870802011341}, "total_wall_ms": 24026.93921001628, "ttft_ms": 11.885083047673106, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 7}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "88.7751", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 38, "retention_rate": 0.5846153846153846, "retention_by_kind": {"person": 0.8333333333333334, "date": 0.7, "amount": 0.38461538461538464, "status": 1.0, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1078}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 38, "retention_rate": 0.5846153846153846, "retention_by_kind": {"person": 0.8333333333333334, "date": 0.7, "amount": 0.38461538461538464, "status": 1.0, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1179}, "2": {"facts_total": 65, "facts_retained": 38, "retention_rate": 0.5846153846153846, "retention_by_kind": {"person": 0.8333333333333334, "date": 0.7, "amount": 0.38461538461538464, "status": 1.0, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1078}, "3": {"facts_total": 65, "facts_retained": 38, "retention_rate": 0.5846153846153846, "retention_by_kind": {"person": 0.8333333333333334, "date": 0.7, "amount": 0.38461538461538464, "status": 1.0, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1078}, "4": {"facts_total": 65, "facts_retained": 38, "retention_rate": 0.5846153846153846, "retention_by_kind": {"person": 0.8333333333333334, "date": 0.7, "amount": 0.38461538461538464, "status": 1.0, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1078}}}, "reference_key_id": "keys/synthetic/syn_03021_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "c817fcf3104750482692c83aaa525b365e66a6edca57a9076fa77d315bd06a42", "output_hash": "1e01e62c69f9479b681e50e59e8565c6583f5cdb803446a05ed5cb13118acdf9"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "9896d1f3353cf1d7b927ea65", "attempt_id": "f91f061808e7", "run_group_id": "E4_core:A4|closed|qwen3b:h4", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:16:28.452909Z", "task_family": "synthetic", "document_id": "syn_03021_2000", "document_hash": "3ac205b464552356699a4d2b0a7bf62497ead36e5bca73982e11ae5ee3810731", "question_id": "q03_exact_copy", "split": "test", "cluster_id": "syn_03021_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 4, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1914, "query_tokens": 534, "message_tokens": 451, "message_utf8_bytes": 1078, "cache_tensor_bytes": null, "serialized_payload_bytes": 1078, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 23723.33950106986, "hops": {"1": 6856.280499952845, "2": 5637.888583005406, "3": 5615.923584089614, "4": 5613.246834021993}, "receiver_prefill_ms": 207.99854199867696, "ttft_ms": 11.85220805928111, "decode_ms": 47.604333027265966}, "total_wall_ms": 23991.20675108861, "ttft_ms": 11.85220805928111, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 4}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "Ingrid Mbeki", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 38, "retention_rate": 0.5846153846153846, "retention_by_kind": {"person": 0.8333333333333334, "date": 0.7, "amount": 0.38461538461538464, "status": 1.0, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1078}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 38, "retention_rate": 0.5846153846153846, "retention_by_kind": {"person": 0.8333333333333334, "date": 0.7, "amount": 0.38461538461538464, "status": 1.0, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1179}, "2": {"facts_total": 65, "facts_retained": 38, "retention_rate": 0.5846153846153846, "retention_by_kind": {"person": 0.8333333333333334, "date": 0.7, "amount": 0.38461538461538464, "status": 1.0, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1078}, "3": {"facts_total": 65, "facts_retained": 38, "retention_rate": 0.5846153846153846, "retention_by_kind": {"person": 0.8333333333333334, "date": 0.7, "amount": 0.38461538461538464, "status": 1.0, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1078}, "4": {"facts_total": 65, "facts_retained": 38, "retention_rate": 0.5846153846153846, "retention_by_kind": {"person": 0.8333333333333334, "date": 0.7, "amount": 0.38461538461538464, "status": 1.0, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1078}}}, "reference_key_id": "keys/synthetic/syn_03021_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "c817fcf3104750482692c83aaa525b365e66a6edca57a9076fa77d315bd06a42", "output_hash": "1dbd85090bb65fb3b72de8ca4ce32f2c234729ff4a239274e79fb574b800a79d"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "a673433cfd3f0ecc6a3917c2", "attempt_id": "598fa1124c27", "run_group_id": "E4_core:A4|closed|qwen3b:h4", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:16:28.735268Z", "task_family": "synthetic", "document_id": "syn_03021_2000", "document_hash": "3ac205b464552356699a4d2b0a7bf62497ead36e5bca73982e11ae5ee3810731", "question_id": "q04_multi_hop", "split": "test", "cluster_id": "syn_03021_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 4, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1914, "query_tokens": 541, "message_tokens": 451, "message_utf8_bytes": 1078, "cache_tensor_bytes": null, "serialized_payload_bytes": 1078, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 23723.33950106986, "hops": {"1": 6856.280499952845, "2": 5637.888583005406, "3": 5615.923584089614, "4": 5613.246834021993}, "receiver_prefill_ms": 208.59645796008408, "ttft_ms": 11.845584027469158, "decode_ms": 11.91420794930309}, "total_wall_ms": 23956.130584003404, "ttft_ms": 11.845584027469158, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 38, "retention_rate": 0.5846153846153846, "retention_by_kind": {"person": 0.8333333333333334, "date": 0.7, "amount": 0.38461538461538464, "status": 1.0, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1078}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 38, "retention_rate": 0.5846153846153846, "retention_by_kind": {"person": 0.8333333333333334, "date": 0.7, "amount": 0.38461538461538464, "status": 1.0, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1179}, "2": {"facts_total": 65, "facts_retained": 38, "retention_rate": 0.5846153846153846, "retention_by_kind": {"person": 0.8333333333333334, "date": 0.7, "amount": 0.38461538461538464, "status": 1.0, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1078}, "3": {"facts_total": 65, "facts_retained": 38, "retention_rate": 0.5846153846153846, "retention_by_kind": {"person": 0.8333333333333334, "date": 0.7, "amount": 0.38461538461538464, "status": 1.0, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1078}, "4": {"facts_total": 65, "facts_retained": 38, "retention_rate": 0.5846153846153846, "retention_by_kind": {"person": 0.8333333333333334, "date": 0.7, "amount": 0.38461538461538464, "status": 1.0, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1078}}}, "reference_key_id": "keys/synthetic/syn_03021_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "c817fcf3104750482692c83aaa525b365e66a6edca57a9076fa77d315bd06a42", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "d30ae8b9944b560a37ea9be5", "attempt_id": "0ef472be8842", "run_group_id": "E4_core:A4|closed|qwen3b:h4", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:16:28.982415Z", "task_family": "synthetic", "document_id": "syn_03021_2000", "document_hash": "3ac205b464552356699a4d2b0a7bf62497ead36e5bca73982e11ae5ee3810731", "question_id": "q05_multi_hop", "split": "test", "cluster_id": "syn_03021_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 4, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1914, "query_tokens": 542, "message_tokens": 451, "message_utf8_bytes": 1078, "cache_tensor_bytes": null, "serialized_payload_bytes": 1078, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 23723.33950106986, "hops": {"1": 6856.280499952845, "2": 5637.888583005406, "3": 5615.923584089614, "4": 5613.246834021993}, "receiver_prefill_ms": 208.33062496967614, "ttft_ms": 11.92233304027468, "decode_ms": 130.9022499481216}, "total_wall_ms": 24075.095584150404, "ttft_ms": 11.92233304027468, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 11}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "BQ-59Z-6753", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 38, "retention_rate": 0.5846153846153846, "retention_by_kind": {"person": 0.8333333333333334, "date": 0.7, "amount": 0.38461538461538464, "status": 1.0, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1078}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 38, "retention_rate": 0.5846153846153846, "retention_by_kind": {"person": 0.8333333333333334, "date": 0.7, "amount": 0.38461538461538464, "status": 1.0, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1179}, "2": {"facts_total": 65, "facts_retained": 38, "retention_rate": 0.5846153846153846, "retention_by_kind": {"person": 0.8333333333333334, "date": 0.7, "amount": 0.38461538461538464, "status": 1.0, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1078}, "3": {"facts_total": 65, "facts_retained": 38, "retention_rate": 0.5846153846153846, "retention_by_kind": {"person": 0.8333333333333334, "date": 0.7, "amount": 0.38461538461538464, "status": 1.0, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1078}, "4": {"facts_total": 65, "facts_retained": 38, "retention_rate": 0.5846153846153846, "retention_by_kind": {"person": 0.8333333333333334, "date": 0.7, "amount": 0.38461538461538464, "status": 1.0, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1078}}}, "reference_key_id": "keys/synthetic/syn_03021_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "c817fcf3104750482692c83aaa525b365e66a6edca57a9076fa77d315bd06a42", "output_hash": "db250f072c856fe7ddf4bbbd16e20c71b8bad1eef5e4480acd0dac30f8ffe9cd"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "77553aa33c4a6917fd682a65", "attempt_id": "0715ecae7adb", "run_group_id": "E4_core:A4|closed|qwen3b:h4", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:16:29.348654Z", "task_family": "synthetic", "document_id": "syn_03021_2000", "document_hash": "3ac205b464552356699a4d2b0a7bf62497ead36e5bca73982e11ae5ee3810731", "question_id": "q06_negation", "split": "test", "cluster_id": "syn_03021_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 4, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1914, "query_tokens": 532, "message_tokens": 451, "message_utf8_bytes": 1078, "cache_tensor_bytes": null, "serialized_payload_bytes": 1078, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 23723.33950106986, "hops": {"1": 6856.280499952845, "2": 5637.888583005406, "3": 5615.923584089614, "4": 5613.246834021993}, "receiver_prefill_ms": 207.52458297647536, "ttft_ms": 12.082790955901146, "decode_ms": 11.927583953365684}, "total_wall_ms": 23955.4495010525, "ttft_ms": 12.082790955901146, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "no", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 38, "retention_rate": 0.5846153846153846, "retention_by_kind": {"person": 0.8333333333333334, "date": 0.7, "amount": 0.38461538461538464, "status": 1.0, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1078}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 38, "retention_rate": 0.5846153846153846, "retention_by_kind": {"person": 0.8333333333333334, "date": 0.7, "amount": 0.38461538461538464, "status": 1.0, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1179}, "2": {"facts_total": 65, "facts_retained": 38, "retention_rate": 0.5846153846153846, "retention_by_kind": {"person": 0.8333333333333334, "date": 0.7, "amount": 0.38461538461538464, "status": 1.0, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1078}, "3": {"facts_total": 65, "facts_retained": 38, "retention_rate": 0.5846153846153846, "retention_by_kind": {"person": 0.8333333333333334, "date": 0.7, "amount": 0.38461538461538464, "status": 1.0, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1078}, "4": {"facts_total": 65, "facts_retained": 38, "retention_rate": 0.5846153846153846, "retention_by_kind": {"person": 0.8333333333333334, "date": 0.7, "amount": 0.38461538461538464, "status": 1.0, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1078}}}, "reference_key_id": "keys/synthetic/syn_03021_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "c817fcf3104750482692c83aaa525b365e66a6edca57a9076fa77d315bd06a42", "output_hash": "9390298f3fb0c5b160498935d79cb139aef28e1c47358b4bbba61862b9c26e59"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "b1f7b1377e722b4e5bbc503d", "attempt_id": "044e55dd9c76", "run_group_id": "E4_core:A4|closed|qwen3b:h4", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:16:29.597038Z", "task_family": "synthetic", "document_id": "syn_03021_2000", "document_hash": "3ac205b464552356699a4d2b0a7bf62497ead36e5bca73982e11ae5ee3810731", "question_id": "q07_negation", "split": "test", "cluster_id": "syn_03021_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 4, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1914, "query_tokens": 533, "message_tokens": 451, "message_utf8_bytes": 1078, "cache_tensor_bytes": null, "serialized_payload_bytes": 1078, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 23723.33950106986, "hops": {"1": 6856.280499952845, "2": 5637.888583005406, "3": 5615.923584089614, "4": 5613.246834021993}, "receiver_prefill_ms": 208.1073330482468, "ttft_ms": 12.144833104684949, "decode_ms": 11.877042008563876}, "total_wall_ms": 23955.89916803874, "ttft_ms": 12.144833104684949, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "yes", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 38, "retention_rate": 0.5846153846153846, "retention_by_kind": {"person": 0.8333333333333334, "date": 0.7, "amount": 0.38461538461538464, "status": 1.0, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1078}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 38, "retention_rate": 0.5846153846153846, "retention_by_kind": {"person": 0.8333333333333334, "date": 0.7, "amount": 0.38461538461538464, "status": 1.0, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1179}, "2": {"facts_total": 65, "facts_retained": 38, "retention_rate": 0.5846153846153846, "retention_by_kind": {"person": 0.8333333333333334, "date": 0.7, "amount": 0.38461538461538464, "status": 1.0, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1078}, "3": {"facts_total": 65, "facts_retained": 38, "retention_rate": 0.5846153846153846, "retention_by_kind": {"person": 0.8333333333333334, "date": 0.7, "amount": 0.38461538461538464, "status": 1.0, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1078}, "4": {"facts_total": 65, "facts_retained": 38, "retention_rate": 0.5846153846153846, "retention_by_kind": {"person": 0.8333333333333334, "date": 0.7, "amount": 0.38461538461538464, "status": 1.0, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1078}}}, "reference_key_id": "keys/synthetic/syn_03021_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "c817fcf3104750482692c83aaa525b365e66a6edca57a9076fa77d315bd06a42", "output_hash": "8a798890fe93817163b10b5f7bd2ca4d25d84c52739a645a889c173eee7d9d3d"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "eb5404bb25f3dd2e2d4be9a5", "attempt_id": "7cc4a8632a5f", "run_group_id": "E4_core:A4|closed|qwen3b:h4", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:16:29.844592Z", "task_family": "synthetic", "document_id": "syn_03021_2000", "document_hash": "3ac205b464552356699a4d2b0a7bf62497ead36e5bca73982e11ae5ee3810731", "question_id": "q08_revision", "split": "test", "cluster_id": "syn_03021_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 4, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1914, "query_tokens": 549, "message_tokens": 451, "message_utf8_bytes": 1078, "cache_tensor_bytes": null, "serialized_payload_bytes": 1078, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 23723.33950106986, "hops": {"1": 6856.280499952845, "2": 5637.888583005406, "3": 5615.923584089614, "4": 5613.246834021993}, "receiver_prefill_ms": 209.81887495145202, "ttft_ms": 11.928000021725893, "decode_ms": 11.84720906894654}, "total_wall_ms": 23957.41070911754, "ttft_ms": 11.928000021725893, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 38, "retention_rate": 0.5846153846153846, "retention_by_kind": {"person": 0.8333333333333334, "date": 0.7, "amount": 0.38461538461538464, "status": 1.0, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1078}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 38, "retention_rate": 0.5846153846153846, "retention_by_kind": {"person": 0.8333333333333334, "date": 0.7, "amount": 0.38461538461538464, "status": 1.0, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1179}, "2": {"facts_total": 65, "facts_retained": 38, "retention_rate": 0.5846153846153846, "retention_by_kind": {"person": 0.8333333333333334, "date": 0.7, "amount": 0.38461538461538464, "status": 1.0, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1078}, "3": {"facts_total": 65, "facts_retained": 38, "retention_rate": 0.5846153846153846, "retention_by_kind": {"person": 0.8333333333333334, "date": 0.7, "amount": 0.38461538461538464, "status": 1.0, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1078}, "4": {"facts_total": 65, "facts_retained": 38, "retention_rate": 0.5846153846153846, "retention_by_kind": {"person": 0.8333333333333334, "date": 0.7, "amount": 0.38461538461538464, "status": 1.0, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1078}}}, "reference_key_id": "keys/synthetic/syn_03021_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "c817fcf3104750482692c83aaa525b365e66a6edca57a9076fa77d315bd06a42", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "2b5851d831099e01acfc958f", "attempt_id": "58ca6262a0f9", "run_group_id": "E4_core:A4|closed|qwen3b:h4", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:16:30.093663Z", "task_family": "synthetic", "document_id": "syn_03021_2000", "document_hash": "3ac205b464552356699a4d2b0a7bf62497ead36e5bca73982e11ae5ee3810731", "question_id": "q09_revision", "split": "test", "cluster_id": "syn_03021_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 4, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1914, "query_tokens": 540, "message_tokens": 451, "message_utf8_bytes": 1078, "cache_tensor_bytes": null, "serialized_payload_bytes": 1078, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 23723.33950106986, "hops": {"1": 6856.280499952845, "2": 5637.888583005406, "3": 5615.923584089614, "4": 5613.246834021993}, "receiver_prefill_ms": 208.3978329319507, "ttft_ms": 11.909666936844587, "decode_ms": 11.908499989658594}, "total_wall_ms": 23955.988542060368, "ttft_ms": 11.909666936844587, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 38, "retention_rate": 0.5846153846153846, "retention_by_kind": {"person": 0.8333333333333334, "date": 0.7, "amount": 0.38461538461538464, "status": 1.0, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1078}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 38, "retention_rate": 0.5846153846153846, "retention_by_kind": {"person": 0.8333333333333334, "date": 0.7, "amount": 0.38461538461538464, "status": 1.0, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1179}, "2": {"facts_total": 65, "facts_retained": 38, "retention_rate": 0.5846153846153846, "retention_by_kind": {"person": 0.8333333333333334, "date": 0.7, "amount": 0.38461538461538464, "status": 1.0, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1078}, "3": {"facts_total": 65, "facts_retained": 38, "retention_rate": 0.5846153846153846, "retention_by_kind": {"person": 0.8333333333333334, "date": 0.7, "amount": 0.38461538461538464, "status": 1.0, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1078}, "4": {"facts_total": 65, "facts_retained": 38, "retention_rate": 0.5846153846153846, "retention_by_kind": {"person": 0.8333333333333334, "date": 0.7, "amount": 0.38461538461538464, "status": 1.0, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1078}}}, "reference_key_id": "keys/synthetic/syn_03021_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "c817fcf3104750482692c83aaa525b365e66a6edca57a9076fa77d315bd06a42", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "f020abc6f26c5b6af8c39fe3", "attempt_id": "e550a7fad138", "run_group_id": "E4_core:A4|closed|qwen3b:h4", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:16:30.340897Z", "task_family": "synthetic", "document_id": "syn_03021_2000", "document_hash": "3ac205b464552356699a4d2b0a7bf62497ead36e5bca73982e11ae5ee3810731", "question_id": "q10_temporal", "split": "test", "cluster_id": "syn_03021_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 4, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1914, "query_tokens": 548, "message_tokens": 451, "message_utf8_bytes": 1078, "cache_tensor_bytes": null, "serialized_payload_bytes": 1078, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 23723.33950106986, "hops": {"1": 6856.280499952845, "2": 5637.888583005406, "3": 5615.923584089614, "4": 5613.246834021993}, "receiver_prefill_ms": 209.40587494987994, "ttft_ms": 11.93516596686095, "decode_ms": 59.25687507260591}, "total_wall_ms": 24004.37254202552, "ttft_ms": 11.93516596686095, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 5}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "Kovar-9", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 38, "retention_rate": 0.5846153846153846, "retention_by_kind": {"person": 0.8333333333333334, "date": 0.7, "amount": 0.38461538461538464, "status": 1.0, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1078}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 38, "retention_rate": 0.5846153846153846, "retention_by_kind": {"person": 0.8333333333333334, "date": 0.7, "amount": 0.38461538461538464, "status": 1.0, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1179}, "2": {"facts_total": 65, "facts_retained": 38, "retention_rate": 0.5846153846153846, "retention_by_kind": {"person": 0.8333333333333334, "date": 0.7, "amount": 0.38461538461538464, "status": 1.0, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1078}, "3": {"facts_total": 65, "facts_retained": 38, "retention_rate": 0.5846153846153846, "retention_by_kind": {"person": 0.8333333333333334, "date": 0.7, "amount": 0.38461538461538464, "status": 1.0, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1078}, "4": {"facts_total": 65, "facts_retained": 38, "retention_rate": 0.5846153846153846, "retention_by_kind": {"person": 0.8333333333333334, "date": 0.7, "amount": 0.38461538461538464, "status": 1.0, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1078}}}, "reference_key_id": "keys/synthetic/syn_03021_2000.json", "task_scores": {"correct": 0.0, "abstained": 0.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "c817fcf3104750482692c83aaa525b365e66a6edca57a9076fa77d315bd06a42", "output_hash": "c969b5f7547945067f5b07e9f822d0043126326f979d950a8b1b045057fbb818"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "2f9ceb057ab120cf8d7f988f", "attempt_id": "4440fb5e918f", "run_group_id": "E4_core:A4|closed|qwen3b:h4", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:16:30.636327Z", "task_family": "synthetic", "document_id": "syn_03021_2000", "document_hash": "3ac205b464552356699a4d2b0a7bf62497ead36e5bca73982e11ae5ee3810731", "question_id": "q11_aggregation", "split": "test", "cluster_id": "syn_03021_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 4, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1914, "query_tokens": 544, "message_tokens": 451, "message_utf8_bytes": 1078, "cache_tensor_bytes": null, "serialized_payload_bytes": 1078, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 23723.33950106986, "hops": {"1": 6856.280499952845, "2": 5637.888583005406, "3": 5615.923584089614, "4": 5613.246834021993}, "receiver_prefill_ms": 208.48158397711813, "ttft_ms": 11.890792055055499, "decode_ms": 82.99558307044208}, "total_wall_ms": 24027.15254307259, "ttft_ms": 11.890792055055499, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 7}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "1392497", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 38, "retention_rate": 0.5846153846153846, "retention_by_kind": {"person": 0.8333333333333334, "date": 0.7, "amount": 0.38461538461538464, "status": 1.0, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1078}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 38, "retention_rate": 0.5846153846153846, "retention_by_kind": {"person": 0.8333333333333334, "date": 0.7, "amount": 0.38461538461538464, "status": 1.0, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1179}, "2": {"facts_total": 65, "facts_retained": 38, "retention_rate": 0.5846153846153846, "retention_by_kind": {"person": 0.8333333333333334, "date": 0.7, "amount": 0.38461538461538464, "status": 1.0, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1078}, "3": {"facts_total": 65, "facts_retained": 38, "retention_rate": 0.5846153846153846, "retention_by_kind": {"person": 0.8333333333333334, "date": 0.7, "amount": 0.38461538461538464, "status": 1.0, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1078}, "4": {"facts_total": 65, "facts_retained": 38, "retention_rate": 0.5846153846153846, "retention_by_kind": {"person": 0.8333333333333334, "date": 0.7, "amount": 0.38461538461538464, "status": 1.0, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1078}}}, "reference_key_id": "keys/synthetic/syn_03021_2000.json", "task_scores": {"correct": 0.0, "abstained": 0.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "c817fcf3104750482692c83aaa525b365e66a6edca57a9076fa77d315bd06a42", "output_hash": "8eee62192f018b9d037d820d43ea01dddf7881d04a58f2b4bf1bd95cc43a4009"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "9d2eaa18b66adeacc4b4cfc9", "attempt_id": "bfd13bfbf872", "run_group_id": "E4_core:A4|closed|qwen3b:h4", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:16:30.954667Z", "task_family": "synthetic", "document_id": "syn_03021_2000", "document_hash": "3ac205b464552356699a4d2b0a7bf62497ead36e5bca73982e11ae5ee3810731", "question_id": "q12_unit_conversion", "split": "test", "cluster_id": "syn_03021_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 4, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1914, "query_tokens": 539, "message_tokens": 451, "message_utf8_bytes": 1078, "cache_tensor_bytes": null, "serialized_payload_bytes": 1078, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 23723.33950106986, "hops": {"1": 6856.280499952845, "2": 5637.888583005406, "3": 5615.923584089614, "4": 5613.246834021993}, "receiver_prefill_ms": 208.29850004520267, "ttft_ms": 11.853625066578388, "decode_ms": 11.894791037775576}, "total_wall_ms": 23955.817168112844, "ttft_ms": 11.853625066578388, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 38, "retention_rate": 0.5846153846153846, "retention_by_kind": {"person": 0.8333333333333334, "date": 0.7, "amount": 0.38461538461538464, "status": 1.0, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1078}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 38, "retention_rate": 0.5846153846153846, "retention_by_kind": {"person": 0.8333333333333334, "date": 0.7, "amount": 0.38461538461538464, "status": 1.0, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1179}, "2": {"facts_total": 65, "facts_retained": 38, "retention_rate": 0.5846153846153846, "retention_by_kind": {"person": 0.8333333333333334, "date": 0.7, "amount": 0.38461538461538464, "status": 1.0, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1078}, "3": {"facts_total": 65, "facts_retained": 38, "retention_rate": 0.5846153846153846, "retention_by_kind": {"person": 0.8333333333333334, "date": 0.7, "amount": 0.38461538461538464, "status": 1.0, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1078}, "4": {"facts_total": 65, "facts_retained": 38, "retention_rate": 0.5846153846153846, "retention_by_kind": {"person": 0.8333333333333334, "date": 0.7, "amount": 0.38461538461538464, "status": 1.0, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1078}}}, "reference_key_id": "keys/synthetic/syn_03021_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "c817fcf3104750482692c83aaa525b365e66a6edca57a9076fa77d315bd06a42", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "3d0ffcc6dcceb70dd7b46e87", "attempt_id": "f0221e0cb1b6", "run_group_id": "E4_core:A4|closed|qwen3b:h4", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:16:31.201532Z", "task_family": "synthetic", "document_id": "syn_03021_2000", "document_hash": "3ac205b464552356699a4d2b0a7bf62497ead36e5bca73982e11ae5ee3810731", "question_id": "q13_unanswerable", "split": "test", "cluster_id": "syn_03021_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 4, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1914, "query_tokens": 544, "message_tokens": 451, "message_utf8_bytes": 1078, "cache_tensor_bytes": null, "serialized_payload_bytes": 1078, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 23723.33950106986, "hops": {"1": 6856.280499952845, "2": 5637.888583005406, "3": 5615.923584089614, "4": 5613.246834021993}, "receiver_prefill_ms": 208.82724993862212, "ttft_ms": 11.941834003664553, "decode_ms": 11.914625065401196}, "total_wall_ms": 23956.43720915541, "ttft_ms": 11.941834003664553, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 38, "retention_rate": 0.5846153846153846, "retention_by_kind": {"person": 0.8333333333333334, "date": 0.7, "amount": 0.38461538461538464, "status": 1.0, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1078}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 38, "retention_rate": 0.5846153846153846, "retention_by_kind": {"person": 0.8333333333333334, "date": 0.7, "amount": 0.38461538461538464, "status": 1.0, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1179}, "2": {"facts_total": 65, "facts_retained": 38, "retention_rate": 0.5846153846153846, "retention_by_kind": {"person": 0.8333333333333334, "date": 0.7, "amount": 0.38461538461538464, "status": 1.0, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1078}, "3": {"facts_total": 65, "facts_retained": 38, "retention_rate": 0.5846153846153846, "retention_by_kind": {"person": 0.8333333333333334, "date": 0.7, "amount": 0.38461538461538464, "status": 1.0, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1078}, "4": {"facts_total": 65, "facts_retained": 38, "retention_rate": 0.5846153846153846, "retention_by_kind": {"person": 0.8333333333333334, "date": 0.7, "amount": 0.38461538461538464, "status": 1.0, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1078}}}, "reference_key_id": "keys/synthetic/syn_03021_2000.json", "task_scores": {"correct": 1.0, "abstained": 1.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "c817fcf3104750482692c83aaa525b365e66a6edca57a9076fa77d315bd06a42", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "cd0854bc3a9bf804afc949a9", "attempt_id": "95118dfad954", "run_group_id": "E4_core:A4|closed|qwen3b:h4", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:16:31.448900Z", "task_family": "synthetic", "document_id": "syn_03021_2000", "document_hash": "3ac205b464552356699a4d2b0a7bf62497ead36e5bca73982e11ae5ee3810731", "question_id": "q14_unanswerable", "split": "test", "cluster_id": "syn_03021_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 4, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1914, "query_tokens": 546, "message_tokens": 451, "message_utf8_bytes": 1078, "cache_tensor_bytes": null, "serialized_payload_bytes": 1078, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 23723.33950106986, "hops": {"1": 6856.280499952845, "2": 5637.888583005406, "3": 5615.923584089614, "4": 5613.246834021993}, "receiver_prefill_ms": 209.55816691275686, "ttft_ms": 11.805249960161746, "decode_ms": 11.852832976728678}, "total_wall_ms": 23957.09595910739, "ttft_ms": 11.805249960161746, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 38, "retention_rate": 0.5846153846153846, "retention_by_kind": {"person": 0.8333333333333334, "date": 0.7, "amount": 0.38461538461538464, "status": 1.0, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1078}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 38, "retention_rate": 0.5846153846153846, "retention_by_kind": {"person": 0.8333333333333334, "date": 0.7, "amount": 0.38461538461538464, "status": 1.0, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1179}, "2": {"facts_total": 65, "facts_retained": 38, "retention_rate": 0.5846153846153846, "retention_by_kind": {"person": 0.8333333333333334, "date": 0.7, "amount": 0.38461538461538464, "status": 1.0, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1078}, "3": {"facts_total": 65, "facts_retained": 38, "retention_rate": 0.5846153846153846, "retention_by_kind": {"person": 0.8333333333333334, "date": 0.7, "amount": 0.38461538461538464, "status": 1.0, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1078}, "4": {"facts_total": 65, "facts_retained": 38, "retention_rate": 0.5846153846153846, "retention_by_kind": {"person": 0.8333333333333334, "date": 0.7, "amount": 0.38461538461538464, "status": 1.0, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1078}}}, "reference_key_id": "keys/synthetic/syn_03021_2000.json", "task_scores": {"correct": 1.0, "abstained": 1.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "c817fcf3104750482692c83aaa525b365e66a6edca57a9076fa77d315bd06a42", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "c5fd0297433f278797b1a72f", "attempt_id": "aefd5e7b17b6", "run_group_id": "E4_core:A4|closed|qwen3b:h8", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:16:54.178008Z", "task_family": "synthetic", "document_id": "syn_03021_2000", "document_hash": "3ac205b464552356699a4d2b0a7bf62497ead36e5bca73982e11ae5ee3810731", "question_id": "q00_single_fact", "split": "test", "cluster_id": "syn_03021_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 8, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1914, "query_tokens": 531, "message_tokens": 451, "message_utf8_bytes": 1078, "cache_tensor_bytes": null, "serialized_payload_bytes": 1078, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 46199.571542209014, "hops": {"1": 6856.280499952845, "2": 5637.888583005406, "3": 5615.923584089614, "4": 5613.246834021993, "5": 5621.278500068001, "6": 5617.707708035596, "7": 5616.204458056018, "8": 5621.041374979541}, "receiver_prefill_ms": 208.15920911263674, "ttft_ms": 11.85754092875868, "decode_ms": 11.932292021811008}, "total_wall_ms": 46431.99520919006, "ttft_ms": 11.85754092875868, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 38, "retention_rate": 0.5846153846153846, "retention_by_kind": {"person": 0.8333333333333334, "date": 0.7, "amount": 0.38461538461538464, "status": 1.0, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1078}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 38, "retention_rate": 0.5846153846153846, "retention_by_kind": {"person": 0.8333333333333334, "date": 0.7, "amount": 0.38461538461538464, "status": 1.0, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1179}, "2": {"facts_total": 65, "facts_retained": 38, "retention_rate": 0.5846153846153846, "retention_by_kind": {"person": 0.8333333333333334, "date": 0.7, "amount": 0.38461538461538464, "status": 1.0, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1078}, "3": {"facts_total": 65, "facts_retained": 38, "retention_rate": 0.5846153846153846, "retention_by_kind": {"person": 0.8333333333333334, "date": 0.7, "amount": 0.38461538461538464, "status": 1.0, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1078}, "4": {"facts_total": 65, "facts_retained": 38, "retention_rate": 0.5846153846153846, "retention_by_kind": {"person": 0.8333333333333334, "date": 0.7, "amount": 0.38461538461538464, "status": 1.0, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1078}, "5": {"facts_total": 65, "facts_retained": 38, "retention_rate": 0.5846153846153846, "retention_by_kind": {"person": 0.8333333333333334, "date": 0.7, "amount": 0.38461538461538464, "status": 1.0, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1078}, "6": {"facts_total": 65, "facts_retained": 38, "retention_rate": 0.5846153846153846, "retention_by_kind": {"person": 0.8333333333333334, "date": 0.7, "amount": 0.38461538461538464, "status": 1.0, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1078}, "7": {"facts_total": 65, "facts_retained": 38, "retention_rate": 0.5846153846153846, "retention_by_kind": {"person": 0.8333333333333334, "date": 0.7, "amount": 0.38461538461538464, "status": 1.0, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1078}, "8": {"facts_total": 65, "facts_retained": 38, "retention_rate": 0.5846153846153846, "retention_by_kind": {"person": 0.8333333333333334, "date": 0.7, "amount": 0.38461538461538464, "status": 1.0, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1078}}}, "reference_key_id": "keys/synthetic/syn_03021_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "c817fcf3104750482692c83aaa525b365e66a6edca57a9076fa77d315bd06a42", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "4e7fbaf1d6f5b3f542bb64d5", "attempt_id": "9590382aa076", "run_group_id": "E4_core:A4|closed|qwen3b:h8", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:16:54.425667Z", "task_family": "synthetic", "document_id": "syn_03021_2000", "document_hash": "3ac205b464552356699a4d2b0a7bf62497ead36e5bca73982e11ae5ee3810731", "question_id": "q01_single_fact", "split": "test", "cluster_id": "syn_03021_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 8, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1914, "query_tokens": 537, "message_tokens": 451, "message_utf8_bytes": 1078, "cache_tensor_bytes": null, "serialized_payload_bytes": 1078, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 46199.571542209014, "hops": {"1": 6856.280499952845, "2": 5637.888583005406, "3": 5615.923584089614, "4": 5613.246834021993, "5": 5621.278500068001, "6": 5617.707708035596, "7": 5616.204458056018, "8": 5621.041374979541}, "receiver_prefill_ms": 208.11033295467496, "ttft_ms": 11.999999987892807, "decode_ms": 130.72195800486952}, "total_wall_ms": 46550.856875255704, "ttft_ms": 11.999999987892807, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 11}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "Lot 536-R, Senneterre", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 38, "retention_rate": 0.5846153846153846, "retention_by_kind": {"person": 0.8333333333333334, "date": 0.7, "amount": 0.38461538461538464, "status": 1.0, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1078}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 38, "retention_rate": 0.5846153846153846, "retention_by_kind": {"person": 0.8333333333333334, "date": 0.7, "amount": 0.38461538461538464, "status": 1.0, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1179}, "2": {"facts_total": 65, "facts_retained": 38, "retention_rate": 0.5846153846153846, "retention_by_kind": {"person": 0.8333333333333334, "date": 0.7, "amount": 0.38461538461538464, "status": 1.0, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1078}, "3": {"facts_total": 65, "facts_retained": 38, "retention_rate": 0.5846153846153846, "retention_by_kind": {"person": 0.8333333333333334, "date": 0.7, "amount": 0.38461538461538464, "status": 1.0, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1078}, "4": {"facts_total": 65, "facts_retained": 38, "retention_rate": 0.5846153846153846, "retention_by_kind": {"person": 0.8333333333333334, "date": 0.7, "amount": 0.38461538461538464, "status": 1.0, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1078}, "5": {"facts_total": 65, "facts_retained": 38, "retention_rate": 0.5846153846153846, "retention_by_kind": {"person": 0.8333333333333334, "date": 0.7, "amount": 0.38461538461538464, "status": 1.0, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1078}, "6": {"facts_total": 65, "facts_retained": 38, "retention_rate": 0.5846153846153846, "retention_by_kind": {"person": 0.8333333333333334, "date": 0.7, "amount": 0.38461538461538464, "status": 1.0, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1078}, "7": {"facts_total": 65, "facts_retained": 38, "retention_rate": 0.5846153846153846, "retention_by_kind": {"person": 0.8333333333333334, "date": 0.7, "amount": 0.38461538461538464, "status": 1.0, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1078}, "8": {"facts_total": 65, "facts_retained": 38, "retention_rate": 0.5846153846153846, "retention_by_kind": {"person": 0.8333333333333334, "date": 0.7, "amount": 0.38461538461538464, "status": 1.0, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1078}}}, "reference_key_id": "keys/synthetic/syn_03021_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 0.8571428571428571}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "c817fcf3104750482692c83aaa525b365e66a6edca57a9076fa77d315bd06a42", "output_hash": "5e85af24aac0f1904d70a2fcee84f1b6f6912cc7450298af1f173002fd8e59df"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "41dabc51522514484ce9e3f1", "attempt_id": "08fb09606f98", "run_group_id": "E4_core:A4|closed|qwen3b:h8", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:16:54.791620Z", "task_family": "synthetic", "document_id": "syn_03021_2000", "document_hash": "3ac205b464552356699a4d2b0a7bf62497ead36e5bca73982e11ae5ee3810731", "question_id": "q02_exact_copy", "split": "test", "cluster_id": "syn_03021_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 8, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1914, "query_tokens": 537, "message_tokens": 451, "message_utf8_bytes": 1078, "cache_tensor_bytes": null, "serialized_payload_bytes": 1078, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 46199.571542209014, "hops": {"1": 6856.280499952845, "2": 5637.888583005406, "3": 5615.923584089614, "4": 5613.246834021993, "5": 5621.278500068001, "6": 5617.707708035596, "7": 5616.204458056018, "8": 5621.041374979541}, "receiver_prefill_ms": 208.0295829800889, "ttft_ms": 11.907916981726885, "decode_ms": 83.02745898254216}, "total_wall_ms": 46502.962167258374, "ttft_ms": 11.907916981726885, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 7}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "88.7751", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 38, "retention_rate": 0.5846153846153846, "retention_by_kind": {"person": 0.8333333333333334, "date": 0.7, "amount": 0.38461538461538464, "status": 1.0, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1078}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 38, "retention_rate": 0.5846153846153846, "retention_by_kind": {"person": 0.8333333333333334, "date": 0.7, "amount": 0.38461538461538464, "status": 1.0, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1179}, "2": {"facts_total": 65, "facts_retained": 38, "retention_rate": 0.5846153846153846, "retention_by_kind": {"person": 0.8333333333333334, "date": 0.7, "amount": 0.38461538461538464, "status": 1.0, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1078}, "3": {"facts_total": 65, "facts_retained": 38, "retention_rate": 0.5846153846153846, "retention_by_kind": {"person": 0.8333333333333334, "date": 0.7, "amount": 0.38461538461538464, "status": 1.0, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1078}, "4": {"facts_total": 65, "facts_retained": 38, "retention_rate": 0.5846153846153846, "retention_by_kind": {"person": 0.8333333333333334, "date": 0.7, "amount": 0.38461538461538464, "status": 1.0, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1078}, "5": {"facts_total": 65, "facts_retained": 38, "retention_rate": 0.5846153846153846, "retention_by_kind": {"person": 0.8333333333333334, "date": 0.7, "amount": 0.38461538461538464, "status": 1.0, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1078}, "6": {"facts_total": 65, "facts_retained": 38, "retention_rate": 0.5846153846153846, "retention_by_kind": {"person": 0.8333333333333334, "date": 0.7, "amount": 0.38461538461538464, "status": 1.0, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1078}, "7": {"facts_total": 65, "facts_retained": 38, "retention_rate": 0.5846153846153846, "retention_by_kind": {"person": 0.8333333333333334, "date": 0.7, "amount": 0.38461538461538464, "status": 1.0, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1078}, "8": {"facts_total": 65, "facts_retained": 38, "retention_rate": 0.5846153846153846, "retention_by_kind": {"person": 0.8333333333333334, "date": 0.7, "amount": 0.38461538461538464, "status": 1.0, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1078}}}, "reference_key_id": "keys/synthetic/syn_03021_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "c817fcf3104750482692c83aaa525b365e66a6edca57a9076fa77d315bd06a42", "output_hash": "1e01e62c69f9479b681e50e59e8565c6583f5cdb803446a05ed5cb13118acdf9"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "5b95b8c31e3e7afa919bc0ad", "attempt_id": "9a9b579b01b2", "run_group_id": "E4_core:A4|closed|qwen3b:h8", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:16:55.109318Z", "task_family": "synthetic", "document_id": "syn_03021_2000", "document_hash": "3ac205b464552356699a4d2b0a7bf62497ead36e5bca73982e11ae5ee3810731", "question_id": "q03_exact_copy", "split": "test", "cluster_id": "syn_03021_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 8, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1914, "query_tokens": 534, "message_tokens": 451, "message_utf8_bytes": 1078, "cache_tensor_bytes": null, "serialized_payload_bytes": 1078, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 46199.571542209014, "hops": {"1": 6856.280499952845, "2": 5637.888583005406, "3": 5615.923584089614, "4": 5613.246834021993, "5": 5621.278500068001, "6": 5617.707708035596, "7": 5616.204458056018, "8": 5621.041374979541}, "receiver_prefill_ms": 207.86695799324661, "ttft_ms": 11.864250060170889, "decode_ms": 47.48091695364565}, "total_wall_ms": 46467.20325015485, "ttft_ms": 11.864250060170889, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 4}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "Ingrid Mbeki", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 38, "retention_rate": 0.5846153846153846, "retention_by_kind": {"person": 0.8333333333333334, "date": 0.7, "amount": 0.38461538461538464, "status": 1.0, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1078}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 38, "retention_rate": 0.5846153846153846, "retention_by_kind": {"person": 0.8333333333333334, "date": 0.7, "amount": 0.38461538461538464, "status": 1.0, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1179}, "2": {"facts_total": 65, "facts_retained": 38, "retention_rate": 0.5846153846153846, "retention_by_kind": {"person": 0.8333333333333334, "date": 0.7, "amount": 0.38461538461538464, "status": 1.0, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1078}, "3": {"facts_total": 65, "facts_retained": 38, "retention_rate": 0.5846153846153846, "retention_by_kind": {"person": 0.8333333333333334, "date": 0.7, "amount": 0.38461538461538464, "status": 1.0, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1078}, "4": {"facts_total": 65, "facts_retained": 38, "retention_rate": 0.5846153846153846, "retention_by_kind": {"person": 0.8333333333333334, "date": 0.7, "amount": 0.38461538461538464, "status": 1.0, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1078}, "5": {"facts_total": 65, "facts_retained": 38, "retention_rate": 0.5846153846153846, "retention_by_kind": {"person": 0.8333333333333334, "date": 0.7, "amount": 0.38461538461538464, "status": 1.0, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1078}, "6": {"facts_total": 65, "facts_retained": 38, "retention_rate": 0.5846153846153846, "retention_by_kind": {"person": 0.8333333333333334, "date": 0.7, "amount": 0.38461538461538464, "status": 1.0, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1078}, "7": {"facts_total": 65, "facts_retained": 38, "retention_rate": 0.5846153846153846, "retention_by_kind": {"person": 0.8333333333333334, "date": 0.7, "amount": 0.38461538461538464, "status": 1.0, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1078}, "8": {"facts_total": 65, "facts_retained": 38, "retention_rate": 0.5846153846153846, "retention_by_kind": {"person": 0.8333333333333334, "date": 0.7, "amount": 0.38461538461538464, "status": 1.0, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1078}}}, "reference_key_id": "keys/synthetic/syn_03021_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "c817fcf3104750482692c83aaa525b365e66a6edca57a9076fa77d315bd06a42", "output_hash": "1dbd85090bb65fb3b72de8ca4ce32f2c234729ff4a239274e79fb574b800a79d"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "02bad975ca6e48cf6342d787", "attempt_id": "374f9ada1d2d", "run_group_id": "E4_core:A4|closed|qwen3b:h8", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:16:55.391623Z", "task_family": "synthetic", "document_id": "syn_03021_2000", "document_hash": "3ac205b464552356699a4d2b0a7bf62497ead36e5bca73982e11ae5ee3810731", "question_id": "q04_multi_hop", "split": "test", "cluster_id": "syn_03021_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 8, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1914, "query_tokens": 541, "message_tokens": 451, "message_utf8_bytes": 1078, "cache_tensor_bytes": null, "serialized_payload_bytes": 1078, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 46199.571542209014, "hops": {"1": 6856.280499952845, "2": 5637.888583005406, "3": 5615.923584089614, "4": 5613.246834021993, "5": 5621.278500068001, "6": 5617.707708035596, "7": 5616.204458056018, "8": 5621.041374979541}, "receiver_prefill_ms": 208.26212503015995, "ttft_ms": 12.061959016136825, "decode_ms": 11.867917026393116}, "total_wall_ms": 46432.177376234904, "ttft_ms": 12.061959016136825, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 38, "retention_rate": 0.5846153846153846, "retention_by_kind": {"person": 0.8333333333333334, "date": 0.7, "amount": 0.38461538461538464, "status": 1.0, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1078}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 38, "retention_rate": 0.5846153846153846, "retention_by_kind": {"person": 0.8333333333333334, "date": 0.7, "amount": 0.38461538461538464, "status": 1.0, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1179}, "2": {"facts_total": 65, "facts_retained": 38, "retention_rate": 0.5846153846153846, "retention_by_kind": {"person": 0.8333333333333334, "date": 0.7, "amount": 0.38461538461538464, "status": 1.0, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1078}, "3": {"facts_total": 65, "facts_retained": 38, "retention_rate": 0.5846153846153846, "retention_by_kind": {"person": 0.8333333333333334, "date": 0.7, "amount": 0.38461538461538464, "status": 1.0, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1078}, "4": {"facts_total": 65, "facts_retained": 38, "retention_rate": 0.5846153846153846, "retention_by_kind": {"person": 0.8333333333333334, "date": 0.7, "amount": 0.38461538461538464, "status": 1.0, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1078}, "5": {"facts_total": 65, "facts_retained": 38, "retention_rate": 0.5846153846153846, "retention_by_kind": {"person": 0.8333333333333334, "date": 0.7, "amount": 0.38461538461538464, "status": 1.0, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1078}, "6": {"facts_total": 65, "facts_retained": 38, "retention_rate": 0.5846153846153846, "retention_by_kind": {"person": 0.8333333333333334, "date": 0.7, "amount": 0.38461538461538464, "status": 1.0, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1078}, "7": {"facts_total": 65, "facts_retained": 38, "retention_rate": 0.5846153846153846, "retention_by_kind": {"person": 0.8333333333333334, "date": 0.7, "amount": 0.38461538461538464, "status": 1.0, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1078}, "8": {"facts_total": 65, "facts_retained": 38, "retention_rate": 0.5846153846153846, "retention_by_kind": {"person": 0.8333333333333334, "date": 0.7, "amount": 0.38461538461538464, "status": 1.0, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1078}}}, "reference_key_id": "keys/synthetic/syn_03021_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "c817fcf3104750482692c83aaa525b365e66a6edca57a9076fa77d315bd06a42", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "821fe18a60d2e24bad0d698a", "attempt_id": "7323712ae0ad", "run_group_id": "E4_core:A4|closed|qwen3b:h8", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:16:55.639005Z", "task_family": "synthetic", "document_id": "syn_03021_2000", "document_hash": "3ac205b464552356699a4d2b0a7bf62497ead36e5bca73982e11ae5ee3810731", "question_id": "q05_multi_hop", "split": "test", "cluster_id": "syn_03021_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 8, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1914, "query_tokens": 542, "message_tokens": 451, "message_utf8_bytes": 1078, "cache_tensor_bytes": null, "serialized_payload_bytes": 1078, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 46199.571542209014, "hops": {"1": 6856.280499952845, "2": 5637.888583005406, "3": 5615.923584089614, "4": 5613.246834021993, "5": 5621.278500068001, "6": 5617.707708035596, "7": 5616.204458056018, "8": 5621.041374979541}, "receiver_prefill_ms": 208.53637508116663, "ttft_ms": 12.160041951574385, "decode_ms": 130.03870809916407}, "total_wall_ms": 46550.75858416967, "ttft_ms": 12.160041951574385, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 11}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "BQ-59Z-6753", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 38, "retention_rate": 0.5846153846153846, "retention_by_kind": {"person": 0.8333333333333334, "date": 0.7, "amount": 0.38461538461538464, "status": 1.0, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1078}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 38, "retention_rate": 0.5846153846153846, "retention_by_kind": {"person": 0.8333333333333334, "date": 0.7, "amount": 0.38461538461538464, "status": 1.0, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1179}, "2": {"facts_total": 65, "facts_retained": 38, "retention_rate": 0.5846153846153846, "retention_by_kind": {"person": 0.8333333333333334, "date": 0.7, "amount": 0.38461538461538464, "status": 1.0, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1078}, "3": {"facts_total": 65, "facts_retained": 38, "retention_rate": 0.5846153846153846, "retention_by_kind": {"person": 0.8333333333333334, "date": 0.7, "amount": 0.38461538461538464, "status": 1.0, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1078}, "4": {"facts_total": 65, "facts_retained": 38, "retention_rate": 0.5846153846153846, "retention_by_kind": {"person": 0.8333333333333334, "date": 0.7, "amount": 0.38461538461538464, "status": 1.0, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1078}, "5": {"facts_total": 65, "facts_retained": 38, "retention_rate": 0.5846153846153846, "retention_by_kind": {"person": 0.8333333333333334, "date": 0.7, "amount": 0.38461538461538464, "status": 1.0, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1078}, "6": {"facts_total": 65, "facts_retained": 38, "retention_rate": 0.5846153846153846, "retention_by_kind": {"person": 0.8333333333333334, "date": 0.7, "amount": 0.38461538461538464, "status": 1.0, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1078}, "7": {"facts_total": 65, "facts_retained": 38, "retention_rate": 0.5846153846153846, "retention_by_kind": {"person": 0.8333333333333334, "date": 0.7, "amount": 0.38461538461538464, "status": 1.0, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1078}, "8": {"facts_total": 65, "facts_retained": 38, "retention_rate": 0.5846153846153846, "retention_by_kind": {"person": 0.8333333333333334, "date": 0.7, "amount": 0.38461538461538464, "status": 1.0, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1078}}}, "reference_key_id": "keys/synthetic/syn_03021_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "c817fcf3104750482692c83aaa525b365e66a6edca57a9076fa77d315bd06a42", "output_hash": "db250f072c856fe7ddf4bbbd16e20c71b8bad1eef5e4480acd0dac30f8ffe9cd"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "d9fc7171c52b1b46c7750590", "attempt_id": "0e2738668731", "run_group_id": "E4_core:A4|closed|qwen3b:h8", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:16:56.004929Z", "task_family": "synthetic", "document_id": "syn_03021_2000", "document_hash": "3ac205b464552356699a4d2b0a7bf62497ead36e5bca73982e11ae5ee3810731", "question_id": "q06_negation", "split": "test", "cluster_id": "syn_03021_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 8, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1914, "query_tokens": 532, "message_tokens": 451, "message_utf8_bytes": 1078, "cache_tensor_bytes": null, "serialized_payload_bytes": 1078, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 46199.571542209014, "hops": {"1": 6856.280499952845, "2": 5637.888583005406, "3": 5615.923584089614, "4": 5613.246834021993, "5": 5621.278500068001, "6": 5617.707708035596, "7": 5616.204458056018, "8": 5621.041374979541}, "receiver_prefill_ms": 208.19124998524785, "ttft_ms": 11.891125002875924, "decode_ms": 11.88912510406226}, "total_wall_ms": 46431.97183427401, "ttft_ms": 11.891125002875924, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "no", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 38, "retention_rate": 0.5846153846153846, "retention_by_kind": {"person": 0.8333333333333334, "date": 0.7, "amount": 0.38461538461538464, "status": 1.0, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1078}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 38, "retention_rate": 0.5846153846153846, "retention_by_kind": {"person": 0.8333333333333334, "date": 0.7, "amount": 0.38461538461538464, "status": 1.0, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1179}, "2": {"facts_total": 65, "facts_retained": 38, "retention_rate": 0.5846153846153846, "retention_by_kind": {"person": 0.8333333333333334, "date": 0.7, "amount": 0.38461538461538464, "status": 1.0, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1078}, "3": {"facts_total": 65, "facts_retained": 38, "retention_rate": 0.5846153846153846, "retention_by_kind": {"person": 0.8333333333333334, "date": 0.7, "amount": 0.38461538461538464, "status": 1.0, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1078}, "4": {"facts_total": 65, "facts_retained": 38, "retention_rate": 0.5846153846153846, "retention_by_kind": {"person": 0.8333333333333334, "date": 0.7, "amount": 0.38461538461538464, "status": 1.0, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1078}, "5": {"facts_total": 65, "facts_retained": 38, "retention_rate": 0.5846153846153846, "retention_by_kind": {"person": 0.8333333333333334, "date": 0.7, "amount": 0.38461538461538464, "status": 1.0, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1078}, "6": {"facts_total": 65, "facts_retained": 38, "retention_rate": 0.5846153846153846, "retention_by_kind": {"person": 0.8333333333333334, "date": 0.7, "amount": 0.38461538461538464, "status": 1.0, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1078}, "7": {"facts_total": 65, "facts_retained": 38, "retention_rate": 0.5846153846153846, "retention_by_kind": {"person": 0.8333333333333334, "date": 0.7, "amount": 0.38461538461538464, "status": 1.0, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1078}, "8": {"facts_total": 65, "facts_retained": 38, "retention_rate": 0.5846153846153846, "retention_by_kind": {"person": 0.8333333333333334, "date": 0.7, "amount": 0.38461538461538464, "status": 1.0, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1078}}}, "reference_key_id": "keys/synthetic/syn_03021_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "c817fcf3104750482692c83aaa525b365e66a6edca57a9076fa77d315bd06a42", "output_hash": "9390298f3fb0c5b160498935d79cb139aef28e1c47358b4bbba61862b9c26e59"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "4b5aa8497254d1421569b4eb", "attempt_id": "d88d116bcde8", "run_group_id": "E4_core:A4|closed|qwen3b:h8", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:16:56.252055Z", "task_family": "synthetic", "document_id": "syn_03021_2000", "document_hash": "3ac205b464552356699a4d2b0a7bf62497ead36e5bca73982e11ae5ee3810731", "question_id": "q07_negation", "split": "test", "cluster_id": "syn_03021_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 8, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1914, "query_tokens": 533, "message_tokens": 451, "message_utf8_bytes": 1078, "cache_tensor_bytes": null, "serialized_payload_bytes": 1078, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 46199.571542209014, "hops": {"1": 6856.280499952845, "2": 5637.888583005406, "3": 5615.923584089614, "4": 5613.246834021993, "5": 5621.278500068001, "6": 5617.707708035596, "7": 5616.204458056018, "8": 5621.041374979541}, "receiver_prefill_ms": 208.08266603853554, "ttft_ms": 11.856916942633688, "decode_ms": 11.943208053708076}, "total_wall_ms": 46431.88341718633, "ttft_ms": 11.856916942633688, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "yes", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 38, "retention_rate": 0.5846153846153846, "retention_by_kind": {"person": 0.8333333333333334, "date": 0.7, "amount": 0.38461538461538464, "status": 1.0, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1078}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 38, "retention_rate": 0.5846153846153846, "retention_by_kind": {"person": 0.8333333333333334, "date": 0.7, "amount": 0.38461538461538464, "status": 1.0, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1179}, "2": {"facts_total": 65, "facts_retained": 38, "retention_rate": 0.5846153846153846, "retention_by_kind": {"person": 0.8333333333333334, "date": 0.7, "amount": 0.38461538461538464, "status": 1.0, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1078}, "3": {"facts_total": 65, "facts_retained": 38, "retention_rate": 0.5846153846153846, "retention_by_kind": {"person": 0.8333333333333334, "date": 0.7, "amount": 0.38461538461538464, "status": 1.0, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1078}, "4": {"facts_total": 65, "facts_retained": 38, "retention_rate": 0.5846153846153846, "retention_by_kind": {"person": 0.8333333333333334, "date": 0.7, "amount": 0.38461538461538464, "status": 1.0, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1078}, "5": {"facts_total": 65, "facts_retained": 38, "retention_rate": 0.5846153846153846, "retention_by_kind": {"person": 0.8333333333333334, "date": 0.7, "amount": 0.38461538461538464, "status": 1.0, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1078}, "6": {"facts_total": 65, "facts_retained": 38, "retention_rate": 0.5846153846153846, "retention_by_kind": {"person": 0.8333333333333334, "date": 0.7, "amount": 0.38461538461538464, "status": 1.0, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1078}, "7": {"facts_total": 65, "facts_retained": 38, "retention_rate": 0.5846153846153846, "retention_by_kind": {"person": 0.8333333333333334, "date": 0.7, "amount": 0.38461538461538464, "status": 1.0, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1078}, "8": {"facts_total": 65, "facts_retained": 38, "retention_rate": 0.5846153846153846, "retention_by_kind": {"person": 0.8333333333333334, "date": 0.7, "amount": 0.38461538461538464, "status": 1.0, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1078}}}, "reference_key_id": "keys/synthetic/syn_03021_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "c817fcf3104750482692c83aaa525b365e66a6edca57a9076fa77d315bd06a42", "output_hash": "8a798890fe93817163b10b5f7bd2ca4d25d84c52739a645a889c173eee7d9d3d"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "564b579990d6b223456b4eaa", "attempt_id": "af18fe05ed3d", "run_group_id": "E4_core:A4|closed|qwen3b:h8", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:16:56.498743Z", "task_family": "synthetic", "document_id": "syn_03021_2000", "document_hash": "3ac205b464552356699a4d2b0a7bf62497ead36e5bca73982e11ae5ee3810731", "question_id": "q08_revision", "split": "test", "cluster_id": "syn_03021_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 8, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1914, "query_tokens": 549, "message_tokens": 451, "message_utf8_bytes": 1078, "cache_tensor_bytes": null, "serialized_payload_bytes": 1078, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 46199.571542209014, "hops": {"1": 6856.280499952845, "2": 5637.888583005406, "3": 5615.923584089614, "4": 5613.246834021993, "5": 5621.278500068001, "6": 5617.707708035596, "7": 5616.204458056018, "8": 5621.041374979541}, "receiver_prefill_ms": 210.3598329704255, "ttft_ms": 11.858084006235003, "decode_ms": 11.916041024960577}, "total_wall_ms": 46434.12825022824, "ttft_ms": 11.858084006235003, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 38, "retention_rate": 0.5846153846153846, "retention_by_kind": {"person": 0.8333333333333334, "date": 0.7, "amount": 0.38461538461538464, "status": 1.0, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1078}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 38, "retention_rate": 0.5846153846153846, "retention_by_kind": {"person": 0.8333333333333334, "date": 0.7, "amount": 0.38461538461538464, "status": 1.0, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1179}, "2": {"facts_total": 65, "facts_retained": 38, "retention_rate": 0.5846153846153846, "retention_by_kind": {"person": 0.8333333333333334, "date": 0.7, "amount": 0.38461538461538464, "status": 1.0, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1078}, "3": {"facts_total": 65, "facts_retained": 38, "retention_rate": 0.5846153846153846, "retention_by_kind": {"person": 0.8333333333333334, "date": 0.7, "amount": 0.38461538461538464, "status": 1.0, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1078}, "4": {"facts_total": 65, "facts_retained": 38, "retention_rate": 0.5846153846153846, "retention_by_kind": {"person": 0.8333333333333334, "date": 0.7, "amount": 0.38461538461538464, "status": 1.0, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1078}, "5": {"facts_total": 65, "facts_retained": 38, "retention_rate": 0.5846153846153846, "retention_by_kind": {"person": 0.8333333333333334, "date": 0.7, "amount": 0.38461538461538464, "status": 1.0, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1078}, "6": {"facts_total": 65, "facts_retained": 38, "retention_rate": 0.5846153846153846, "retention_by_kind": {"person": 0.8333333333333334, "date": 0.7, "amount": 0.38461538461538464, "status": 1.0, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1078}, "7": {"facts_total": 65, "facts_retained": 38, "retention_rate": 0.5846153846153846, "retention_by_kind": {"person": 0.8333333333333334, "date": 0.7, "amount": 0.38461538461538464, "status": 1.0, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1078}, "8": {"facts_total": 65, "facts_retained": 38, "retention_rate": 0.5846153846153846, "retention_by_kind": {"person": 0.8333333333333334, "date": 0.7, "amount": 0.38461538461538464, "status": 1.0, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1078}}}, "reference_key_id": "keys/synthetic/syn_03021_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "c817fcf3104750482692c83aaa525b365e66a6edca57a9076fa77d315bd06a42", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "034e5938aba6fd92da0571e6", "attempt_id": "a8b7fef1248d", "run_group_id": "E4_core:A4|closed|qwen3b:h8", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:16:56.748126Z", "task_family": "synthetic", "document_id": "syn_03021_2000", "document_hash": "3ac205b464552356699a4d2b0a7bf62497ead36e5bca73982e11ae5ee3810731", "question_id": "q09_revision", "split": "test", "cluster_id": "syn_03021_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 8, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1914, "query_tokens": 540, "message_tokens": 451, "message_utf8_bytes": 1078, "cache_tensor_bytes": null, "serialized_payload_bytes": 1078, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 46199.571542209014, "hops": {"1": 6856.280499952845, "2": 5637.888583005406, "3": 5615.923584089614, "4": 5613.246834021993, "5": 5621.278500068001, "6": 5617.707708035596, "7": 5616.204458056018, "8": 5621.041374979541}, "receiver_prefill_ms": 208.23116705287248, "ttft_ms": 11.911166948266327, "decode_ms": 11.85904210433364}, "total_wall_ms": 46432.02004220802, "ttft_ms": 11.911166948266327, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 38, "retention_rate": 0.5846153846153846, "retention_by_kind": {"person": 0.8333333333333334, "date": 0.7, "amount": 0.38461538461538464, "status": 1.0, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1078}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 38, "retention_rate": 0.5846153846153846, "retention_by_kind": {"person": 0.8333333333333334, "date": 0.7, "amount": 0.38461538461538464, "status": 1.0, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1179}, "2": {"facts_total": 65, "facts_retained": 38, "retention_rate": 0.5846153846153846, "retention_by_kind": {"person": 0.8333333333333334, "date": 0.7, "amount": 0.38461538461538464, "status": 1.0, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1078}, "3": {"facts_total": 65, "facts_retained": 38, "retention_rate": 0.5846153846153846, "retention_by_kind": {"person": 0.8333333333333334, "date": 0.7, "amount": 0.38461538461538464, "status": 1.0, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1078}, "4": {"facts_total": 65, "facts_retained": 38, "retention_rate": 0.5846153846153846, "retention_by_kind": {"person": 0.8333333333333334, "date": 0.7, "amount": 0.38461538461538464, "status": 1.0, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1078}, "5": {"facts_total": 65, "facts_retained": 38, "retention_rate": 0.5846153846153846, "retention_by_kind": {"person": 0.8333333333333334, "date": 0.7, "amount": 0.38461538461538464, "status": 1.0, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1078}, "6": {"facts_total": 65, "facts_retained": 38, "retention_rate": 0.5846153846153846, "retention_by_kind": {"person": 0.8333333333333334, "date": 0.7, "amount": 0.38461538461538464, "status": 1.0, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1078}, "7": {"facts_total": 65, "facts_retained": 38, "retention_rate": 0.5846153846153846, "retention_by_kind": {"person": 0.8333333333333334, "date": 0.7, "amount": 0.38461538461538464, "status": 1.0, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1078}, "8": {"facts_total": 65, "facts_retained": 38, "retention_rate": 0.5846153846153846, "retention_by_kind": {"person": 0.8333333333333334, "date": 0.7, "amount": 0.38461538461538464, "status": 1.0, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1078}}}, "reference_key_id": "keys/synthetic/syn_03021_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "c817fcf3104750482692c83aaa525b365e66a6edca57a9076fa77d315bd06a42", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "a4e0e54e0870a480c53bafc8", "attempt_id": "b1a1e0dfba79", "run_group_id": "E4_core:A4|closed|qwen3b:h8", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:16:56.995121Z", "task_family": "synthetic", "document_id": "syn_03021_2000", "document_hash": "3ac205b464552356699a4d2b0a7bf62497ead36e5bca73982e11ae5ee3810731", "question_id": "q10_temporal", "split": "test", "cluster_id": "syn_03021_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 8, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1914, "query_tokens": 548, "message_tokens": 451, "message_utf8_bytes": 1078, "cache_tensor_bytes": null, "serialized_payload_bytes": 1078, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 46199.571542209014, "hops": {"1": 6856.280499952845, "2": 5637.888583005406, "3": 5615.923584089614, "4": 5613.246834021993, "5": 5621.278500068001, "6": 5617.707708035596, "7": 5616.204458056018, "8": 5621.041374979541}, "receiver_prefill_ms": 210.09379206225276, "ttft_ms": 11.9454589439556, "decode_ms": 59.51929197181016}, "total_wall_ms": 46481.55795922503, "ttft_ms": 11.9454589439556, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 5}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "Kovar-9", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 38, "retention_rate": 0.5846153846153846, "retention_by_kind": {"person": 0.8333333333333334, "date": 0.7, "amount": 0.38461538461538464, "status": 1.0, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1078}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 38, "retention_rate": 0.5846153846153846, "retention_by_kind": {"person": 0.8333333333333334, "date": 0.7, "amount": 0.38461538461538464, "status": 1.0, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1179}, "2": {"facts_total": 65, "facts_retained": 38, "retention_rate": 0.5846153846153846, "retention_by_kind": {"person": 0.8333333333333334, "date": 0.7, "amount": 0.38461538461538464, "status": 1.0, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1078}, "3": {"facts_total": 65, "facts_retained": 38, "retention_rate": 0.5846153846153846, "retention_by_kind": {"person": 0.8333333333333334, "date": 0.7, "amount": 0.38461538461538464, "status": 1.0, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1078}, "4": {"facts_total": 65, "facts_retained": 38, "retention_rate": 0.5846153846153846, "retention_by_kind": {"person": 0.8333333333333334, "date": 0.7, "amount": 0.38461538461538464, "status": 1.0, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1078}, "5": {"facts_total": 65, "facts_retained": 38, "retention_rate": 0.5846153846153846, "retention_by_kind": {"person": 0.8333333333333334, "date": 0.7, "amount": 0.38461538461538464, "status": 1.0, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1078}, "6": {"facts_total": 65, "facts_retained": 38, "retention_rate": 0.5846153846153846, "retention_by_kind": {"person": 0.8333333333333334, "date": 0.7, "amount": 0.38461538461538464, "status": 1.0, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1078}, "7": {"facts_total": 65, "facts_retained": 38, "retention_rate": 0.5846153846153846, "retention_by_kind": {"person": 0.8333333333333334, "date": 0.7, "amount": 0.38461538461538464, "status": 1.0, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1078}, "8": {"facts_total": 65, "facts_retained": 38, "retention_rate": 0.5846153846153846, "retention_by_kind": {"person": 0.8333333333333334, "date": 0.7, "amount": 0.38461538461538464, "status": 1.0, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1078}}}, "reference_key_id": "keys/synthetic/syn_03021_2000.json", "task_scores": {"correct": 0.0, "abstained": 0.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "c817fcf3104750482692c83aaa525b365e66a6edca57a9076fa77d315bd06a42", "output_hash": "c969b5f7547945067f5b07e9f822d0043126326f979d950a8b1b045057fbb818"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "30e673a4fd8af7e1f020fb15", "attempt_id": "4843714a99dd", "run_group_id": "E4_core:A4|closed|qwen3b:h8", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:16:57.291527Z", "task_family": "synthetic", "document_id": "syn_03021_2000", "document_hash": "3ac205b464552356699a4d2b0a7bf62497ead36e5bca73982e11ae5ee3810731", "question_id": "q11_aggregation", "split": "test", "cluster_id": "syn_03021_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 8, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1914, "query_tokens": 544, "message_tokens": 451, "message_utf8_bytes": 1078, "cache_tensor_bytes": null, "serialized_payload_bytes": 1078, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 46199.571542209014, "hops": {"1": 6856.280499952845, "2": 5637.888583005406, "3": 5615.923584089614, "4": 5613.246834021993, "5": 5621.278500068001, "6": 5617.707708035596, "7": 5616.204458056018, "8": 5621.041374979541}, "receiver_prefill_ms": 208.53962504770607, "ttft_ms": 11.95858302526176, "decode_ms": 83.00124993547797}, "total_wall_ms": 46503.5193762742, "ttft_ms": 11.95858302526176, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 7}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "1402481", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 38, "retention_rate": 0.5846153846153846, "retention_by_kind": {"person": 0.8333333333333334, "date": 0.7, "amount": 0.38461538461538464, "status": 1.0, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1078}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 38, "retention_rate": 0.5846153846153846, "retention_by_kind": {"person": 0.8333333333333334, "date": 0.7, "amount": 0.38461538461538464, "status": 1.0, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1179}, "2": {"facts_total": 65, "facts_retained": 38, "retention_rate": 0.5846153846153846, "retention_by_kind": {"person": 0.8333333333333334, "date": 0.7, "amount": 0.38461538461538464, "status": 1.0, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1078}, "3": {"facts_total": 65, "facts_retained": 38, "retention_rate": 0.5846153846153846, "retention_by_kind": {"person": 0.8333333333333334, "date": 0.7, "amount": 0.38461538461538464, "status": 1.0, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1078}, "4": {"facts_total": 65, "facts_retained": 38, "retention_rate": 0.5846153846153846, "retention_by_kind": {"person": 0.8333333333333334, "date": 0.7, "amount": 0.38461538461538464, "status": 1.0, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1078}, "5": {"facts_total": 65, "facts_retained": 38, "retention_rate": 0.5846153846153846, "retention_by_kind": {"person": 0.8333333333333334, "date": 0.7, "amount": 0.38461538461538464, "status": 1.0, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1078}, "6": {"facts_total": 65, "facts_retained": 38, "retention_rate": 0.5846153846153846, "retention_by_kind": {"person": 0.8333333333333334, "date": 0.7, "amount": 0.38461538461538464, "status": 1.0, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1078}, "7": {"facts_total": 65, "facts_retained": 38, "retention_rate": 0.5846153846153846, "retention_by_kind": {"person": 0.8333333333333334, "date": 0.7, "amount": 0.38461538461538464, "status": 1.0, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1078}, "8": {"facts_total": 65, "facts_retained": 38, "retention_rate": 0.5846153846153846, "retention_by_kind": {"person": 0.8333333333333334, "date": 0.7, "amount": 0.38461538461538464, "status": 1.0, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1078}}}, "reference_key_id": "keys/synthetic/syn_03021_2000.json", "task_scores": {"correct": 0.0, "abstained": 0.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "c817fcf3104750482692c83aaa525b365e66a6edca57a9076fa77d315bd06a42", "output_hash": "d66bff6aad3f09bcc89b31dd53521f992c908ee990de72cb1b8a73da32f7c2b3"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "2edba60b7a66d55aed425778", "attempt_id": "49536c890363", "run_group_id": "E4_core:A4|closed|qwen3b:h8", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:16:57.609883Z", "task_family": "synthetic", "document_id": "syn_03021_2000", "document_hash": "3ac205b464552356699a4d2b0a7bf62497ead36e5bca73982e11ae5ee3810731", "question_id": "q12_unit_conversion", "split": "test", "cluster_id": "syn_03021_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 8, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1914, "query_tokens": 539, "message_tokens": 451, "message_utf8_bytes": 1078, "cache_tensor_bytes": null, "serialized_payload_bytes": 1078, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 46199.571542209014, "hops": {"1": 6856.280499952845, "2": 5637.888583005406, "3": 5615.923584089614, "4": 5613.246834021993, "5": 5621.278500068001, "6": 5617.707708035596, "7": 5616.204458056018, "8": 5621.041374979541}, "receiver_prefill_ms": 208.0205000238493, "ttft_ms": 11.923665995709598, "decode_ms": 11.861833045259118}, "total_wall_ms": 46431.813292205334, "ttft_ms": 11.923665995709598, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 38, "retention_rate": 0.5846153846153846, "retention_by_kind": {"person": 0.8333333333333334, "date": 0.7, "amount": 0.38461538461538464, "status": 1.0, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1078}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 38, "retention_rate": 0.5846153846153846, "retention_by_kind": {"person": 0.8333333333333334, "date": 0.7, "amount": 0.38461538461538464, "status": 1.0, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1179}, "2": {"facts_total": 65, "facts_retained": 38, "retention_rate": 0.5846153846153846, "retention_by_kind": {"person": 0.8333333333333334, "date": 0.7, "amount": 0.38461538461538464, "status": 1.0, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1078}, "3": {"facts_total": 65, "facts_retained": 38, "retention_rate": 0.5846153846153846, "retention_by_kind": {"person": 0.8333333333333334, "date": 0.7, "amount": 0.38461538461538464, "status": 1.0, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1078}, "4": {"facts_total": 65, "facts_retained": 38, "retention_rate": 0.5846153846153846, "retention_by_kind": {"person": 0.8333333333333334, "date": 0.7, "amount": 0.38461538461538464, "status": 1.0, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1078}, "5": {"facts_total": 65, "facts_retained": 38, "retention_rate": 0.5846153846153846, "retention_by_kind": {"person": 0.8333333333333334, "date": 0.7, "amount": 0.38461538461538464, "status": 1.0, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1078}, "6": {"facts_total": 65, "facts_retained": 38, "retention_rate": 0.5846153846153846, "retention_by_kind": {"person": 0.8333333333333334, "date": 0.7, "amount": 0.38461538461538464, "status": 1.0, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1078}, "7": {"facts_total": 65, "facts_retained": 38, "retention_rate": 0.5846153846153846, "retention_by_kind": {"person": 0.8333333333333334, "date": 0.7, "amount": 0.38461538461538464, "status": 1.0, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1078}, "8": {"facts_total": 65, "facts_retained": 38, "retention_rate": 0.5846153846153846, "retention_by_kind": {"person": 0.8333333333333334, "date": 0.7, "amount": 0.38461538461538464, "status": 1.0, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1078}}}, "reference_key_id": "keys/synthetic/syn_03021_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "c817fcf3104750482692c83aaa525b365e66a6edca57a9076fa77d315bd06a42", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "546009e99da1f69d470df875", "attempt_id": "b2bd244ed1fd", "run_group_id": "E4_core:A4|closed|qwen3b:h8", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:16:57.856635Z", "task_family": "synthetic", "document_id": "syn_03021_2000", "document_hash": "3ac205b464552356699a4d2b0a7bf62497ead36e5bca73982e11ae5ee3810731", "question_id": "q13_unanswerable", "split": "test", "cluster_id": "syn_03021_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 8, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1914, "query_tokens": 544, "message_tokens": 451, "message_utf8_bytes": 1078, "cache_tensor_bytes": null, "serialized_payload_bytes": 1078, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 46199.571542209014, "hops": {"1": 6856.280499952845, "2": 5637.888583005406, "3": 5615.923584089614, "4": 5613.246834021993, "5": 5621.278500068001, "6": 5617.707708035596, "7": 5616.204458056018, "8": 5621.041374979541}, "receiver_prefill_ms": 208.5955000948161, "ttft_ms": 11.932624969631433, "decode_ms": 11.910790926776826}, "total_wall_ms": 46432.44583415799, "ttft_ms": 11.932624969631433, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 38, "retention_rate": 0.5846153846153846, "retention_by_kind": {"person": 0.8333333333333334, "date": 0.7, "amount": 0.38461538461538464, "status": 1.0, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1078}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 38, "retention_rate": 0.5846153846153846, "retention_by_kind": {"person": 0.8333333333333334, "date": 0.7, "amount": 0.38461538461538464, "status": 1.0, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1179}, "2": {"facts_total": 65, "facts_retained": 38, "retention_rate": 0.5846153846153846, "retention_by_kind": {"person": 0.8333333333333334, "date": 0.7, "amount": 0.38461538461538464, "status": 1.0, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1078}, "3": {"facts_total": 65, "facts_retained": 38, "retention_rate": 0.5846153846153846, "retention_by_kind": {"person": 0.8333333333333334, "date": 0.7, "amount": 0.38461538461538464, "status": 1.0, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1078}, "4": {"facts_total": 65, "facts_retained": 38, "retention_rate": 0.5846153846153846, "retention_by_kind": {"person": 0.8333333333333334, "date": 0.7, "amount": 0.38461538461538464, "status": 1.0, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1078}, "5": {"facts_total": 65, "facts_retained": 38, "retention_rate": 0.5846153846153846, "retention_by_kind": {"person": 0.8333333333333334, "date": 0.7, "amount": 0.38461538461538464, "status": 1.0, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1078}, "6": {"facts_total": 65, "facts_retained": 38, "retention_rate": 0.5846153846153846, "retention_by_kind": {"person": 0.8333333333333334, "date": 0.7, "amount": 0.38461538461538464, "status": 1.0, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1078}, "7": {"facts_total": 65, "facts_retained": 38, "retention_rate": 0.5846153846153846, "retention_by_kind": {"person": 0.8333333333333334, "date": 0.7, "amount": 0.38461538461538464, "status": 1.0, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1078}, "8": {"facts_total": 65, "facts_retained": 38, "retention_rate": 0.5846153846153846, "retention_by_kind": {"person": 0.8333333333333334, "date": 0.7, "amount": 0.38461538461538464, "status": 1.0, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1078}}}, "reference_key_id": "keys/synthetic/syn_03021_2000.json", "task_scores": {"correct": 1.0, "abstained": 1.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "c817fcf3104750482692c83aaa525b365e66a6edca57a9076fa77d315bd06a42", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "ba80b4aa669bd64475e35286", "attempt_id": "4381dabdcb45", "run_group_id": "E4_core:A4|closed|qwen3b:h8", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:16:58.104136Z", "task_family": "synthetic", "document_id": "syn_03021_2000", "document_hash": "3ac205b464552356699a4d2b0a7bf62497ead36e5bca73982e11ae5ee3810731", "question_id": "q14_unanswerable", "split": "test", "cluster_id": "syn_03021_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 8, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1914, "query_tokens": 546, "message_tokens": 451, "message_utf8_bytes": 1078, "cache_tensor_bytes": null, "serialized_payload_bytes": 1078, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 46199.571542209014, "hops": {"1": 6856.280499952845, "2": 5637.888583005406, "3": 5615.923584089614, "4": 5613.246834021993, "5": 5621.278500068001, "6": 5617.707708035596, "7": 5616.204458056018, "8": 5621.041374979541}, "receiver_prefill_ms": 209.60583398118615, "ttft_ms": 11.916207964532077, "decode_ms": 11.901333928108215}, "total_wall_ms": 46433.49204224069, "ttft_ms": 11.916207964532077, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 38, "retention_rate": 0.5846153846153846, "retention_by_kind": {"person": 0.8333333333333334, "date": 0.7, "amount": 0.38461538461538464, "status": 1.0, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1078}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 38, "retention_rate": 0.5846153846153846, "retention_by_kind": {"person": 0.8333333333333334, "date": 0.7, "amount": 0.38461538461538464, "status": 1.0, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1179}, "2": {"facts_total": 65, "facts_retained": 38, "retention_rate": 0.5846153846153846, "retention_by_kind": {"person": 0.8333333333333334, "date": 0.7, "amount": 0.38461538461538464, "status": 1.0, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1078}, "3": {"facts_total": 65, "facts_retained": 38, "retention_rate": 0.5846153846153846, "retention_by_kind": {"person": 0.8333333333333334, "date": 0.7, "amount": 0.38461538461538464, "status": 1.0, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1078}, "4": {"facts_total": 65, "facts_retained": 38, "retention_rate": 0.5846153846153846, "retention_by_kind": {"person": 0.8333333333333334, "date": 0.7, "amount": 0.38461538461538464, "status": 1.0, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1078}, "5": {"facts_total": 65, "facts_retained": 38, "retention_rate": 0.5846153846153846, "retention_by_kind": {"person": 0.8333333333333334, "date": 0.7, "amount": 0.38461538461538464, "status": 1.0, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1078}, "6": {"facts_total": 65, "facts_retained": 38, "retention_rate": 0.5846153846153846, "retention_by_kind": {"person": 0.8333333333333334, "date": 0.7, "amount": 0.38461538461538464, "status": 1.0, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1078}, "7": {"facts_total": 65, "facts_retained": 38, "retention_rate": 0.5846153846153846, "retention_by_kind": {"person": 0.8333333333333334, "date": 0.7, "amount": 0.38461538461538464, "status": 1.0, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1078}, "8": {"facts_total": 65, "facts_retained": 38, "retention_rate": 0.5846153846153846, "retention_by_kind": {"person": 0.8333333333333334, "date": 0.7, "amount": 0.38461538461538464, "status": 1.0, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1078}}}, "reference_key_id": "keys/synthetic/syn_03021_2000.json", "task_scores": {"correct": 1.0, "abstained": 1.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "c817fcf3104750482692c83aaa525b365e66a6edca57a9076fa77d315bd06a42", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "ae7257d8f3b2bc4acef4fefe", "attempt_id": "ffd427b62e0f", "run_group_id": "E4_core:A5|closed|qwen3b:h1", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:17:05.279563Z", "task_family": "synthetic", "document_id": "syn_03021_2000", "document_hash": "3ac205b464552356699a4d2b0a7bf62497ead36e5bca73982e11ae5ee3810731", "question_id": "q00_single_fact", "split": "test", "cluster_id": "syn_03021_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 1, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1914, "query_tokens": 460, "message_tokens": 379, "message_utf8_bytes": 900, "cache_tensor_bytes": null, "serialized_payload_bytes": 900, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 6925.180334015749, "hops": {"1": 6925.180334015749}, "receiver_prefill_ms": 185.74941693805158, "ttft_ms": 11.932625086046755, "decode_ms": 11.890375055372715}, "total_wall_ms": 7135.222293087281, "ttft_ms": 11.932625086046755, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 19, "retention_rate": 0.2923076923076923, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.15384615384615385, "status": 1.0, "place": 0.0, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 836}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 19, "retention_rate": 0.2923076923076923, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.15384615384615385, "status": 1.0, "place": 0.0, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 836}}}, "reference_key_id": "keys/synthetic/syn_03021_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "2ae8e199245ad716748b3e420d7b2c8e50b7daeeebd965effa93b31557ba8942", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "5b1b1b76da59f52fea762c89", "attempt_id": "0855dcf667cc", "run_group_id": "E4_core:A5|closed|qwen3b:h1", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:17:05.504729Z", "task_family": "synthetic", "document_id": "syn_03021_2000", "document_hash": "3ac205b464552356699a4d2b0a7bf62497ead36e5bca73982e11ae5ee3810731", "question_id": "q01_single_fact", "split": "test", "cluster_id": "syn_03021_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 1, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1914, "query_tokens": 466, "message_tokens": 379, "message_utf8_bytes": 900, "cache_tensor_bytes": null, "serialized_payload_bytes": 900, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 6925.180334015749, "hops": {"1": 6925.180334015749}, "receiver_prefill_ms": 186.38054199982435, "ttft_ms": 12.015084037557244, "decode_ms": 11.870292015373707}, "total_wall_ms": 7135.912375058979, "ttft_ms": 12.015084037557244, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 19, "retention_rate": 0.2923076923076923, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.15384615384615385, "status": 1.0, "place": 0.0, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 836}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 19, "retention_rate": 0.2923076923076923, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.15384615384615385, "status": 1.0, "place": 0.0, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 836}}}, "reference_key_id": "keys/synthetic/syn_03021_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "2ae8e199245ad716748b3e420d7b2c8e50b7daeeebd965effa93b31557ba8942", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "a480562658a89d3ff96962af", "attempt_id": "669e07a19328", "run_group_id": "E4_core:A5|closed|qwen3b:h1", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:17:05.731053Z", "task_family": "synthetic", "document_id": "syn_03021_2000", "document_hash": "3ac205b464552356699a4d2b0a7bf62497ead36e5bca73982e11ae5ee3810731", "question_id": "q02_exact_copy", "split": "test", "cluster_id": "syn_03021_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 1, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1914, "query_tokens": 466, "message_tokens": 379, "message_utf8_bytes": 900, "cache_tensor_bytes": null, "serialized_payload_bytes": 900, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 6925.180334015749, "hops": {"1": 6925.180334015749}, "receiver_prefill_ms": 185.80212502274662, "ttft_ms": 11.93687494378537, "decode_ms": 11.84945902787149}, "total_wall_ms": 7135.175834060647, "ttft_ms": 11.93687494378537, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 19, "retention_rate": 0.2923076923076923, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.15384615384615385, "status": 1.0, "place": 0.0, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 836}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 19, "retention_rate": 0.2923076923076923, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.15384615384615385, "status": 1.0, "place": 0.0, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 836}}}, "reference_key_id": "keys/synthetic/syn_03021_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "2ae8e199245ad716748b3e420d7b2c8e50b7daeeebd965effa93b31557ba8942", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "cfc4f60d6494772527b53307", "attempt_id": "10ee17888aa5", "run_group_id": "E4_core:A5|closed|qwen3b:h1", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:17:05.956084Z", "task_family": "synthetic", "document_id": "syn_03021_2000", "document_hash": "3ac205b464552356699a4d2b0a7bf62497ead36e5bca73982e11ae5ee3810731", "question_id": "q03_exact_copy", "split": "test", "cluster_id": "syn_03021_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 1, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1914, "query_tokens": 463, "message_tokens": 379, "message_utf8_bytes": 900, "cache_tensor_bytes": null, "serialized_payload_bytes": 900, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 6925.180334015749, "hops": {"1": 6925.180334015749}, "receiver_prefill_ms": 185.72916695848107, "ttft_ms": 12.121542007662356, "decode_ms": 11.8968749884516}, "total_wall_ms": 7135.335543076508, "ttft_ms": 12.121542007662356, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 19, "retention_rate": 0.2923076923076923, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.15384615384615385, "status": 1.0, "place": 0.0, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 836}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 19, "retention_rate": 0.2923076923076923, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.15384615384615385, "status": 1.0, "place": 0.0, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 836}}}, "reference_key_id": "keys/synthetic/syn_03021_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "2ae8e199245ad716748b3e420d7b2c8e50b7daeeebd965effa93b31557ba8942", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "3cae994ed8018e0b3116169c", "attempt_id": "e477271412ee", "run_group_id": "E4_core:A5|closed|qwen3b:h1", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:17:06.180939Z", "task_family": "synthetic", "document_id": "syn_03021_2000", "document_hash": "3ac205b464552356699a4d2b0a7bf62497ead36e5bca73982e11ae5ee3810731", "question_id": "q04_multi_hop", "split": "test", "cluster_id": "syn_03021_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 1, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1914, "query_tokens": 470, "message_tokens": 379, "message_utf8_bytes": 900, "cache_tensor_bytes": null, "serialized_payload_bytes": 900, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 6925.180334015749, "hops": {"1": 6925.180334015749}, "receiver_prefill_ms": 185.81774993799627, "ttft_ms": 11.970542022027075, "decode_ms": 11.864624917507172}, "total_wall_ms": 7135.265375021845, "ttft_ms": 11.970542022027075, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 19, "retention_rate": 0.2923076923076923, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.15384615384615385, "status": 1.0, "place": 0.0, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 836}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 19, "retention_rate": 0.2923076923076923, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.15384615384615385, "status": 1.0, "place": 0.0, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 836}}}, "reference_key_id": "keys/synthetic/syn_03021_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "2ae8e199245ad716748b3e420d7b2c8e50b7daeeebd965effa93b31557ba8942", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "c6108ca8099c6640ec6301c4", "attempt_id": "7e64dde775c1", "run_group_id": "E4_core:A5|closed|qwen3b:h1", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:17:06.405328Z", "task_family": "synthetic", "document_id": "syn_03021_2000", "document_hash": "3ac205b464552356699a4d2b0a7bf62497ead36e5bca73982e11ae5ee3810731", "question_id": "q05_multi_hop", "split": "test", "cluster_id": "syn_03021_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 1, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1914, "query_tokens": 471, "message_tokens": 379, "message_utf8_bytes": 900, "cache_tensor_bytes": null, "serialized_payload_bytes": 900, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 6925.180334015749, "hops": {"1": 6925.180334015749}, "receiver_prefill_ms": 186.11291702836752, "ttft_ms": 11.900875018909574, "decode_ms": 118.42575005721301}, "total_wall_ms": 7242.054667090997, "ttft_ms": 11.900875018909574, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 10}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "MH-96F-2099", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 19, "retention_rate": 0.2923076923076923, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.15384615384615385, "status": 1.0, "place": 0.0, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 836}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 19, "retention_rate": 0.2923076923076923, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.15384615384615385, "status": 1.0, "place": 0.0, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 836}}}, "reference_key_id": "keys/synthetic/syn_03021_2000.json", "task_scores": {"correct": 0.0, "abstained": 0.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "2ae8e199245ad716748b3e420d7b2c8e50b7daeeebd965effa93b31557ba8942", "output_hash": "95c230d14a055d30e0e1d4355caf07808e1c968b1feab9e0607d8bd37b1b43dc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "fe7b55b4d45a22a11f6909ec", "attempt_id": "889a4c2a9b07", "run_group_id": "E4_core:A5|closed|qwen3b:h1", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:17:06.736779Z", "task_family": "synthetic", "document_id": "syn_03021_2000", "document_hash": "3ac205b464552356699a4d2b0a7bf62497ead36e5bca73982e11ae5ee3810731", "question_id": "q06_negation", "split": "test", "cluster_id": "syn_03021_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 1, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1914, "query_tokens": 461, "message_tokens": 379, "message_utf8_bytes": 900, "cache_tensor_bytes": null, "serialized_payload_bytes": 900, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 6925.180334015749, "hops": {"1": 6925.180334015749}, "receiver_prefill_ms": 185.79858297016472, "ttft_ms": 12.15145899914205, "decode_ms": 11.867084074765444}, "total_wall_ms": 7135.408833972178, "ttft_ms": 12.15145899914205, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "no", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 19, "retention_rate": 0.2923076923076923, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.15384615384615385, "status": 1.0, "place": 0.0, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 836}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 19, "retention_rate": 0.2923076923076923, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.15384615384615385, "status": 1.0, "place": 0.0, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 836}}}, "reference_key_id": "keys/synthetic/syn_03021_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "2ae8e199245ad716748b3e420d7b2c8e50b7daeeebd965effa93b31557ba8942", "output_hash": "9390298f3fb0c5b160498935d79cb139aef28e1c47358b4bbba61862b9c26e59"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "7b0b701d4880f5bc8b220c99", "attempt_id": "42687da5a2d4", "run_group_id": "E4_core:A5|closed|qwen3b:h1", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:17:06.961644Z", "task_family": "synthetic", "document_id": "syn_03021_2000", "document_hash": "3ac205b464552356699a4d2b0a7bf62497ead36e5bca73982e11ae5ee3810731", "question_id": "q07_negation", "split": "test", "cluster_id": "syn_03021_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 1, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1914, "query_tokens": 462, "message_tokens": 379, "message_utf8_bytes": 900, "cache_tensor_bytes": null, "serialized_payload_bytes": 900, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 6925.180334015749, "hops": {"1": 6925.180334015749}, "receiver_prefill_ms": 185.63987500965595, "ttft_ms": 11.912833899259567, "decode_ms": 11.90208294428885}, "total_wall_ms": 7135.017459047958, "ttft_ms": 11.912833899259567, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "no", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 19, "retention_rate": 0.2923076923076923, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.15384615384615385, "status": 1.0, "place": 0.0, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 836}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 19, "retention_rate": 0.2923076923076923, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.15384615384615385, "status": 1.0, "place": 0.0, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 836}}}, "reference_key_id": "keys/synthetic/syn_03021_2000.json", "task_scores": {"correct": 0.0, "abstained": 0.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "2ae8e199245ad716748b3e420d7b2c8e50b7daeeebd965effa93b31557ba8942", "output_hash": "9390298f3fb0c5b160498935d79cb139aef28e1c47358b4bbba61862b9c26e59"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "d1a8adcf2006739f449a8d0f", "attempt_id": "cb4b62c6b2cd", "run_group_id": "E4_core:A5|closed|qwen3b:h1", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:17:07.186161Z", "task_family": "synthetic", "document_id": "syn_03021_2000", "document_hash": "3ac205b464552356699a4d2b0a7bf62497ead36e5bca73982e11ae5ee3810731", "question_id": "q08_revision", "split": "test", "cluster_id": "syn_03021_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 1, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1914, "query_tokens": 478, "message_tokens": 379, "message_utf8_bytes": 900, "cache_tensor_bytes": null, "serialized_payload_bytes": 900, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 6925.180334015749, "hops": {"1": 6925.180334015749}, "receiver_prefill_ms": 186.61087495274842, "ttft_ms": 12.079833075404167, "decode_ms": 11.912542046047747}, "total_wall_ms": 7136.2303759669885, "ttft_ms": 12.079833075404167, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 19, "retention_rate": 0.2923076923076923, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.15384615384615385, "status": 1.0, "place": 0.0, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 836}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 19, "retention_rate": 0.2923076923076923, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.15384615384615385, "status": 1.0, "place": 0.0, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 836}}}, "reference_key_id": "keys/synthetic/syn_03021_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "2ae8e199245ad716748b3e420d7b2c8e50b7daeeebd965effa93b31557ba8942", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "2f2c042ef37fa519ed5596a4", "attempt_id": "3637701eb5db", "run_group_id": "E4_core:A5|closed|qwen3b:h1", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:17:07.411702Z", "task_family": "synthetic", "document_id": "syn_03021_2000", "document_hash": "3ac205b464552356699a4d2b0a7bf62497ead36e5bca73982e11ae5ee3810731", "question_id": "q09_revision", "split": "test", "cluster_id": "syn_03021_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 1, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1914, "query_tokens": 469, "message_tokens": 379, "message_utf8_bytes": 900, "cache_tensor_bytes": null, "serialized_payload_bytes": 900, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 6925.180334015749, "hops": {"1": 6925.180334015749}, "receiver_prefill_ms": 185.6529590440914, "ttft_ms": 11.886582942679524, "decode_ms": 82.92412501759827}, "total_wall_ms": 7206.051668035798, "ttft_ms": 11.886582942679524, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 7}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "5646720", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 19, "retention_rate": 0.2923076923076923, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.15384615384615385, "status": 1.0, "place": 0.0, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 836}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 19, "retention_rate": 0.2923076923076923, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.15384615384615385, "status": 1.0, "place": 0.0, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 836}}}, "reference_key_id": "keys/synthetic/syn_03021_2000.json", "task_scores": {"correct": 0.0, "abstained": 0.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "2ae8e199245ad716748b3e420d7b2c8e50b7daeeebd965effa93b31557ba8942", "output_hash": "e1d129e860627cf4623671ed186cb4ebbf5d63169d3d0dc14a62e1da2af61326"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "a10579345ff90ce0b51b095a", "attempt_id": "8e5a432518a5", "run_group_id": "E4_core:A5|closed|qwen3b:h1", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:17:07.707287Z", "task_family": "synthetic", "document_id": "syn_03021_2000", "document_hash": "3ac205b464552356699a4d2b0a7bf62497ead36e5bca73982e11ae5ee3810731", "question_id": "q10_temporal", "split": "test", "cluster_id": "syn_03021_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 1, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1914, "query_tokens": 477, "message_tokens": 379, "message_utf8_bytes": 900, "cache_tensor_bytes": null, "serialized_payload_bytes": 900, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 6925.180334015749, "hops": {"1": 6925.180334015749}, "receiver_prefill_ms": 186.01137504447252, "ttft_ms": 11.888124980032444, "decode_ms": 59.248958015814424}, "total_wall_ms": 7182.706166990101, "ttft_ms": 11.888124980032444, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 5}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "Kovar-9", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 19, "retention_rate": 0.2923076923076923, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.15384615384615385, "status": 1.0, "place": 0.0, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 836}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 19, "retention_rate": 0.2923076923076923, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.15384615384615385, "status": 1.0, "place": 0.0, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 836}}}, "reference_key_id": "keys/synthetic/syn_03021_2000.json", "task_scores": {"correct": 0.0, "abstained": 0.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "2ae8e199245ad716748b3e420d7b2c8e50b7daeeebd965effa93b31557ba8942", "output_hash": "c969b5f7547945067f5b07e9f822d0043126326f979d950a8b1b045057fbb818"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "49c2f1ac039c3e943de2f2bc", "attempt_id": "200b170d1e67", "run_group_id": "E4_core:A5|closed|qwen3b:h1", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:17:07.979554Z", "task_family": "synthetic", "document_id": "syn_03021_2000", "document_hash": "3ac205b464552356699a4d2b0a7bf62497ead36e5bca73982e11ae5ee3810731", "question_id": "q11_aggregation", "split": "test", "cluster_id": "syn_03021_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 1, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1914, "query_tokens": 473, "message_tokens": 379, "message_utf8_bytes": 900, "cache_tensor_bytes": null, "serialized_payload_bytes": 900, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 6925.180334015749, "hops": {"1": 6925.180334015749}, "receiver_prefill_ms": 185.95029204152524, "ttft_ms": 11.952874949201941, "decode_ms": 82.92195794638246}, "total_wall_ms": 7206.42341801431, "ttft_ms": 11.952874949201941, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 7}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "9484825", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 19, "retention_rate": 0.2923076923076923, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.15384615384615385, "status": 1.0, "place": 0.0, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 836}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 19, "retention_rate": 0.2923076923076923, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.15384615384615385, "status": 1.0, "place": 0.0, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 836}}}, "reference_key_id": "keys/synthetic/syn_03021_2000.json", "task_scores": {"correct": 0.0, "abstained": 0.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "2ae8e199245ad716748b3e420d7b2c8e50b7daeeebd965effa93b31557ba8942", "output_hash": "667b970507c4c6c7fb572757eb3b5eb04186ddb67a6ef680266b7a0924737993"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "27a0d757a7e29dc483a9839e", "attempt_id": "fd5182f9a59f", "run_group_id": "E4_core:A5|closed|qwen3b:h1", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:17:08.275184Z", "task_family": "synthetic", "document_id": "syn_03021_2000", "document_hash": "3ac205b464552356699a4d2b0a7bf62497ead36e5bca73982e11ae5ee3810731", "question_id": "q12_unit_conversion", "split": "test", "cluster_id": "syn_03021_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 1, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1914, "query_tokens": 468, "message_tokens": 379, "message_utf8_bytes": 900, "cache_tensor_bytes": null, "serialized_payload_bytes": 900, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 6925.180334015749, "hops": {"1": 6925.180334015749}, "receiver_prefill_ms": 185.07245904766023, "ttft_ms": 12.499416945502162, "decode_ms": 48.374790931120515}, "total_wall_ms": 7171.649417025037, "ttft_ms": 12.499416945502162, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 4}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "22.3", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 19, "retention_rate": 0.2923076923076923, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.15384615384615385, "status": 1.0, "place": 0.0, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 836}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 19, "retention_rate": 0.2923076923076923, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.15384615384615385, "status": 1.0, "place": 0.0, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 836}}}, "reference_key_id": "keys/synthetic/syn_03021_2000.json", "task_scores": {"correct": 0.0, "abstained": 0.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "2ae8e199245ad716748b3e420d7b2c8e50b7daeeebd965effa93b31557ba8942", "output_hash": "8afb0bb35e730fd32e7805d65081241bd21f43709df37165f4ec91c763f8a99d"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "72ca9413ae5d77d6e1f34eb2", "attempt_id": "6c5c85129b3a", "run_group_id": "E4_core:A5|closed|qwen3b:h1", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:17:08.536578Z", "task_family": "synthetic", "document_id": "syn_03021_2000", "document_hash": "3ac205b464552356699a4d2b0a7bf62497ead36e5bca73982e11ae5ee3810731", "question_id": "q13_unanswerable", "split": "test", "cluster_id": "syn_03021_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 1, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1914, "query_tokens": 473, "message_tokens": 379, "message_utf8_bytes": 900, "cache_tensor_bytes": null, "serialized_payload_bytes": 900, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 6925.180334015749, "hops": {"1": 6925.180334015749}, "receiver_prefill_ms": 186.29824998788536, "ttft_ms": 11.775708058848977, "decode_ms": 11.896333075128496}, "total_wall_ms": 7135.5698759434745, "ttft_ms": 11.775708058848977, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 19, "retention_rate": 0.2923076923076923, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.15384615384615385, "status": 1.0, "place": 0.0, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 836}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 19, "retention_rate": 0.2923076923076923, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.15384615384615385, "status": 1.0, "place": 0.0, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 836}}}, "reference_key_id": "keys/synthetic/syn_03021_2000.json", "task_scores": {"correct": 1.0, "abstained": 1.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "2ae8e199245ad716748b3e420d7b2c8e50b7daeeebd965effa93b31557ba8942", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "3ce81b4c08728288beee7033", "attempt_id": "9a524be5c1c1", "run_group_id": "E4_core:A5|closed|qwen3b:h1", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:17:08.762180Z", "task_family": "synthetic", "document_id": "syn_03021_2000", "document_hash": "3ac205b464552356699a4d2b0a7bf62497ead36e5bca73982e11ae5ee3810731", "question_id": "q14_unanswerable", "split": "test", "cluster_id": "syn_03021_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 1, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1914, "query_tokens": 475, "message_tokens": 379, "message_utf8_bytes": 900, "cache_tensor_bytes": null, "serialized_payload_bytes": 900, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 6925.180334015749, "hops": {"1": 6925.180334015749}, "receiver_prefill_ms": 186.34079198818654, "ttft_ms": 11.879625031724572, "decode_ms": 11.835790937766433}, "total_wall_ms": 7135.655333986506, "ttft_ms": 11.879625031724572, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 19, "retention_rate": 0.2923076923076923, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.15384615384615385, "status": 1.0, "place": 0.0, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 836}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 19, "retention_rate": 0.2923076923076923, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.15384615384615385, "status": 1.0, "place": 0.0, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 836}}}, "reference_key_id": "keys/synthetic/syn_03021_2000.json", "task_scores": {"correct": 1.0, "abstained": 1.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "2ae8e199245ad716748b3e420d7b2c8e50b7daeeebd965effa93b31557ba8942", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "3b61b5dba0499246cb30457d", "attempt_id": "dd3cb297941f", "run_group_id": "E4_core:A5|closed|qwen3b:h2", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:17:13.925481Z", "task_family": "synthetic", "document_id": "syn_03021_2000", "document_hash": "3ac205b464552356699a4d2b0a7bf62497ead36e5bca73982e11ae5ee3810731", "question_id": "q00_single_fact", "split": "test", "cluster_id": "syn_03021_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 2, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1914, "query_tokens": 460, "message_tokens": 379, "message_utf8_bytes": 900, "cache_tensor_bytes": null, "serialized_payload_bytes": 900, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 11861.411251011305, "hops": {"1": 6925.180334015749, "2": 4936.230916995555}, "receiver_prefill_ms": 185.55329099763185, "ttft_ms": 11.846916982904077, "decode_ms": 11.894957977347076}, "total_wall_ms": 12071.135417092592, "ttft_ms": 11.846916982904077, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 19, "retention_rate": 0.2923076923076923, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.15384615384615385, "status": 1.0, "place": 0.0, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 836}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 19, "retention_rate": 0.2923076923076923, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.15384615384615385, "status": 1.0, "place": 0.0, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 836}, "2": {"facts_total": 65, "facts_retained": 19, "retention_rate": 0.2923076923076923, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.15384615384615385, "status": 1.0, "place": 0.0, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 836}}}, "reference_key_id": "keys/synthetic/syn_03021_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "2ae8e199245ad716748b3e420d7b2c8e50b7daeeebd965effa93b31557ba8942", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "22fbf07a16461c29c6fc2551", "attempt_id": "2a6b4bbb8cf5", "run_group_id": "E4_core:A5|closed|qwen3b:h2", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:17:14.150563Z", "task_family": "synthetic", "document_id": "syn_03021_2000", "document_hash": "3ac205b464552356699a4d2b0a7bf62497ead36e5bca73982e11ae5ee3810731", "question_id": "q01_single_fact", "split": "test", "cluster_id": "syn_03021_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 2, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1914, "query_tokens": 466, "message_tokens": 379, "message_utf8_bytes": 900, "cache_tensor_bytes": null, "serialized_payload_bytes": 900, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 11861.411251011305, "hops": {"1": 6925.180334015749, "2": 4936.230916995555}, "receiver_prefill_ms": 185.66191603895277, "ttft_ms": 11.953208013437688, "decode_ms": 11.810999945737422}, "total_wall_ms": 12071.271167951636, "ttft_ms": 11.953208013437688, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 19, "retention_rate": 0.2923076923076923, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.15384615384615385, "status": 1.0, "place": 0.0, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 836}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 19, "retention_rate": 0.2923076923076923, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.15384615384615385, "status": 1.0, "place": 0.0, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 836}, "2": {"facts_total": 65, "facts_retained": 19, "retention_rate": 0.2923076923076923, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.15384615384615385, "status": 1.0, "place": 0.0, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 836}}}, "reference_key_id": "keys/synthetic/syn_03021_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "2ae8e199245ad716748b3e420d7b2c8e50b7daeeebd965effa93b31557ba8942", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "7b770824cf70736b0cf058dc", "attempt_id": "26e43239cab6", "run_group_id": "E4_core:A5|closed|qwen3b:h2", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:17:14.375324Z", "task_family": "synthetic", "document_id": "syn_03021_2000", "document_hash": "3ac205b464552356699a4d2b0a7bf62497ead36e5bca73982e11ae5ee3810731", "question_id": "q02_exact_copy", "split": "test", "cluster_id": "syn_03021_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 2, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1914, "query_tokens": 466, "message_tokens": 379, "message_utf8_bytes": 900, "cache_tensor_bytes": null, "serialized_payload_bytes": 900, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 11861.411251011305, "hops": {"1": 6925.180334015749, "2": 4936.230916995555}, "receiver_prefill_ms": 184.9977910751477, "ttft_ms": 11.88245799858123, "decode_ms": 11.951375054195523}, "total_wall_ms": 12070.611543022096, "ttft_ms": 11.88245799858123, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 19, "retention_rate": 0.2923076923076923, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.15384615384615385, "status": 1.0, "place": 0.0, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 836}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 19, "retention_rate": 0.2923076923076923, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.15384615384615385, "status": 1.0, "place": 0.0, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 836}, "2": {"facts_total": 65, "facts_retained": 19, "retention_rate": 0.2923076923076923, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.15384615384615385, "status": 1.0, "place": 0.0, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 836}}}, "reference_key_id": "keys/synthetic/syn_03021_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "2ae8e199245ad716748b3e420d7b2c8e50b7daeeebd965effa93b31557ba8942", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "7228469c085236a5d7cdbc53", "attempt_id": "1ed7c87c634c", "run_group_id": "E4_core:A5|closed|qwen3b:h2", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:17:14.599214Z", "task_family": "synthetic", "document_id": "syn_03021_2000", "document_hash": "3ac205b464552356699a4d2b0a7bf62497ead36e5bca73982e11ae5ee3810731", "question_id": "q03_exact_copy", "split": "test", "cluster_id": "syn_03021_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 2, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1914, "query_tokens": 463, "message_tokens": 379, "message_utf8_bytes": 900, "cache_tensor_bytes": null, "serialized_payload_bytes": 900, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 11861.411251011305, "hops": {"1": 6925.180334015749, "2": 4936.230916995555}, "receiver_prefill_ms": 185.88683393318206, "ttft_ms": 11.878125020302832, "decode_ms": 11.930083041079342}, "total_wall_ms": 12071.487543056719, "ttft_ms": 11.878125020302832, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 19, "retention_rate": 0.2923076923076923, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.15384615384615385, "status": 1.0, "place": 0.0, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 836}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 19, "retention_rate": 0.2923076923076923, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.15384615384615385, "status": 1.0, "place": 0.0, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 836}, "2": {"facts_total": 65, "facts_retained": 19, "retention_rate": 0.2923076923076923, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.15384615384615385, "status": 1.0, "place": 0.0, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 836}}}, "reference_key_id": "keys/synthetic/syn_03021_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "2ae8e199245ad716748b3e420d7b2c8e50b7daeeebd965effa93b31557ba8942", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "78e6f91e802f5422f07de9f9", "attempt_id": "5218beab345b", "run_group_id": "E4_core:A5|closed|qwen3b:h2", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:17:14.823563Z", "task_family": "synthetic", "document_id": "syn_03021_2000", "document_hash": "3ac205b464552356699a4d2b0a7bf62497ead36e5bca73982e11ae5ee3810731", "question_id": "q04_multi_hop", "split": "test", "cluster_id": "syn_03021_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 2, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1914, "query_tokens": 470, "message_tokens": 379, "message_utf8_bytes": 900, "cache_tensor_bytes": null, "serialized_payload_bytes": 900, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 11861.411251011305, "hops": {"1": 6925.180334015749, "2": 4936.230916995555}, "receiver_prefill_ms": 186.55491701792926, "ttft_ms": 11.879374971613288, "decode_ms": 11.874792049638927}, "total_wall_ms": 12072.130126063712, "ttft_ms": 11.879374971613288, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 19, "retention_rate": 0.2923076923076923, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.15384615384615385, "status": 1.0, "place": 0.0, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 836}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 19, "retention_rate": 0.2923076923076923, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.15384615384615385, "status": 1.0, "place": 0.0, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 836}, "2": {"facts_total": 65, "facts_retained": 19, "retention_rate": 0.2923076923076923, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.15384615384615385, "status": 1.0, "place": 0.0, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 836}}}, "reference_key_id": "keys/synthetic/syn_03021_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "2ae8e199245ad716748b3e420d7b2c8e50b7daeeebd965effa93b31557ba8942", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "12bc1b54d14bf090dc8bfc31", "attempt_id": "816c17885095", "run_group_id": "E4_core:A5|closed|qwen3b:h2", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:17:15.048493Z", "task_family": "synthetic", "document_id": "syn_03021_2000", "document_hash": "3ac205b464552356699a4d2b0a7bf62497ead36e5bca73982e11ae5ee3810731", "question_id": "q05_multi_hop", "split": "test", "cluster_id": "syn_03021_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 2, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1914, "query_tokens": 471, "message_tokens": 379, "message_utf8_bytes": 900, "cache_tensor_bytes": null, "serialized_payload_bytes": 900, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 11861.411251011305, "hops": {"1": 6925.180334015749, "2": 4936.230916995555}, "receiver_prefill_ms": 186.15012511145324, "ttft_ms": 11.892207898199558, "decode_ms": 11.85537502169609}, "total_wall_ms": 12071.711335098371, "ttft_ms": 11.892207898199558, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 19, "retention_rate": 0.2923076923076923, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.15384615384615385, "status": 1.0, "place": 0.0, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 836}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 19, "retention_rate": 0.2923076923076923, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.15384615384615385, "status": 1.0, "place": 0.0, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 836}, "2": {"facts_total": 65, "facts_retained": 19, "retention_rate": 0.2923076923076923, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.15384615384615385, "status": 1.0, "place": 0.0, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 836}}}, "reference_key_id": "keys/synthetic/syn_03021_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "2ae8e199245ad716748b3e420d7b2c8e50b7daeeebd965effa93b31557ba8942", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "d392ed93ed70cb1ea8037358", "attempt_id": "66ffea12902a", "run_group_id": "E4_core:A5|closed|qwen3b:h2", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:17:15.273435Z", "task_family": "synthetic", "document_id": "syn_03021_2000", "document_hash": "3ac205b464552356699a4d2b0a7bf62497ead36e5bca73982e11ae5ee3810731", "question_id": "q06_negation", "split": "test", "cluster_id": "syn_03021_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 2, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1914, "query_tokens": 461, "message_tokens": 379, "message_utf8_bytes": 900, "cache_tensor_bytes": null, "serialized_payload_bytes": 900, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 11861.411251011305, "hops": {"1": 6925.180334015749, "2": 4936.230916995555}, "receiver_prefill_ms": 185.0749159930274, "ttft_ms": 11.988957994617522, "decode_ms": 11.812624987214804}, "total_wall_ms": 12070.687875966541, "ttft_ms": 11.988957994617522, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "no", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 19, "retention_rate": 0.2923076923076923, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.15384615384615385, "status": 1.0, "place": 0.0, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 836}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 19, "retention_rate": 0.2923076923076923, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.15384615384615385, "status": 1.0, "place": 0.0, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 836}, "2": {"facts_total": 65, "facts_retained": 19, "retention_rate": 0.2923076923076923, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.15384615384615385, "status": 1.0, "place": 0.0, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 836}}}, "reference_key_id": "keys/synthetic/syn_03021_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "2ae8e199245ad716748b3e420d7b2c8e50b7daeeebd965effa93b31557ba8942", "output_hash": "9390298f3fb0c5b160498935d79cb139aef28e1c47358b4bbba61862b9c26e59"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "d04fe637e3c6d136095b49a7", "attempt_id": "6725e0e7fdb0", "run_group_id": "E4_core:A5|closed|qwen3b:h2", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:17:15.497281Z", "task_family": "synthetic", "document_id": "syn_03021_2000", "document_hash": "3ac205b464552356699a4d2b0a7bf62497ead36e5bca73982e11ae5ee3810731", "question_id": "q07_negation", "split": "test", "cluster_id": "syn_03021_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 2, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1914, "query_tokens": 462, "message_tokens": 379, "message_utf8_bytes": 900, "cache_tensor_bytes": null, "serialized_payload_bytes": 900, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 11861.411251011305, "hops": {"1": 6925.180334015749, "2": 4936.230916995555}, "receiver_prefill_ms": 185.89491699822247, "ttft_ms": 11.849834001623094, "decode_ms": 12.03279197216034}, "total_wall_ms": 12071.580250980332, "ttft_ms": 11.849834001623094, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "no", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 19, "retention_rate": 0.2923076923076923, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.15384615384615385, "status": 1.0, "place": 0.0, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 836}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 19, "retention_rate": 0.2923076923076923, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.15384615384615385, "status": 1.0, "place": 0.0, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 836}, "2": {"facts_total": 65, "facts_retained": 19, "retention_rate": 0.2923076923076923, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.15384615384615385, "status": 1.0, "place": 0.0, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 836}}}, "reference_key_id": "keys/synthetic/syn_03021_2000.json", "task_scores": {"correct": 0.0, "abstained": 0.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "2ae8e199245ad716748b3e420d7b2c8e50b7daeeebd965effa93b31557ba8942", "output_hash": "9390298f3fb0c5b160498935d79cb139aef28e1c47358b4bbba61862b9c26e59"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "5b7ea06de73f5adeb5cf68c5", "attempt_id": "464b88459adb", "run_group_id": "E4_core:A5|closed|qwen3b:h2", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:17:15.722059Z", "task_family": "synthetic", "document_id": "syn_03021_2000", "document_hash": "3ac205b464552356699a4d2b0a7bf62497ead36e5bca73982e11ae5ee3810731", "question_id": "q08_revision", "split": "test", "cluster_id": "syn_03021_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 2, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1914, "query_tokens": 478, "message_tokens": 379, "message_utf8_bytes": 900, "cache_tensor_bytes": null, "serialized_payload_bytes": 900, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 11861.411251011305, "hops": {"1": 6925.180334015749, "2": 4936.230916995555}, "receiver_prefill_ms": 186.34495802689344, "ttft_ms": 11.887083994224668, "decode_ms": 11.819291044957936}, "total_wall_ms": 12071.841417928226, "ttft_ms": 11.887083994224668, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 19, "retention_rate": 0.2923076923076923, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.15384615384615385, "status": 1.0, "place": 0.0, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 836}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 19, "retention_rate": 0.2923076923076923, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.15384615384615385, "status": 1.0, "place": 0.0, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 836}, "2": {"facts_total": 65, "facts_retained": 19, "retention_rate": 0.2923076923076923, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.15384615384615385, "status": 1.0, "place": 0.0, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 836}}}, "reference_key_id": "keys/synthetic/syn_03021_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "2ae8e199245ad716748b3e420d7b2c8e50b7daeeebd965effa93b31557ba8942", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "f782d9ff7e8018cc417d94a5", "attempt_id": "9143f2f28756", "run_group_id": "E4_core:A5|closed|qwen3b:h2", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:17:15.946778Z", "task_family": "synthetic", "document_id": "syn_03021_2000", "document_hash": "3ac205b464552356699a4d2b0a7bf62497ead36e5bca73982e11ae5ee3810731", "question_id": "q09_revision", "split": "test", "cluster_id": "syn_03021_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 2, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1914, "query_tokens": 469, "message_tokens": 379, "message_utf8_bytes": 900, "cache_tensor_bytes": null, "serialized_payload_bytes": 900, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 11861.411251011305, "hops": {"1": 6925.180334015749, "2": 4936.230916995555}, "receiver_prefill_ms": 185.57754205539823, "ttft_ms": 11.88474998343736, "decode_ms": 11.888041975907981}, "total_wall_ms": 12071.145292953588, "ttft_ms": 11.88474998343736, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 19, "retention_rate": 0.2923076923076923, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.15384615384615385, "status": 1.0, "place": 0.0, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 836}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 19, "retention_rate": 0.2923076923076923, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.15384615384615385, "status": 1.0, "place": 0.0, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 836}, "2": {"facts_total": 65, "facts_retained": 19, "retention_rate": 0.2923076923076923, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.15384615384615385, "status": 1.0, "place": 0.0, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 836}}}, "reference_key_id": "keys/synthetic/syn_03021_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "2ae8e199245ad716748b3e420d7b2c8e50b7daeeebd965effa93b31557ba8942", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "4815c4c9795727eda5b0286f", "attempt_id": "9d95a7a5c3bb", "run_group_id": "E4_core:A5|closed|qwen3b:h2", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:17:16.170934Z", "task_family": "synthetic", "document_id": "syn_03021_2000", "document_hash": "3ac205b464552356699a4d2b0a7bf62497ead36e5bca73982e11ae5ee3810731", "question_id": "q10_temporal", "split": "test", "cluster_id": "syn_03021_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 2, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1914, "query_tokens": 477, "message_tokens": 379, "message_utf8_bytes": 900, "cache_tensor_bytes": null, "serialized_payload_bytes": 900, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 11861.411251011305, "hops": {"1": 6925.180334015749, "2": 4936.230916995555}, "receiver_prefill_ms": 186.81025004480034, "ttft_ms": 11.913917027413845, "decode_ms": 59.32566698174924}, "total_wall_ms": 12119.859459111467, "ttft_ms": 11.913917027413845, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 5}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "Kovar-9", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 19, "retention_rate": 0.2923076923076923, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.15384615384615385, "status": 1.0, "place": 0.0, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 836}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 19, "retention_rate": 0.2923076923076923, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.15384615384615385, "status": 1.0, "place": 0.0, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 836}, "2": {"facts_total": 65, "facts_retained": 19, "retention_rate": 0.2923076923076923, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.15384615384615385, "status": 1.0, "place": 0.0, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 836}}}, "reference_key_id": "keys/synthetic/syn_03021_2000.json", "task_scores": {"correct": 0.0, "abstained": 0.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "2ae8e199245ad716748b3e420d7b2c8e50b7daeeebd965effa93b31557ba8942", "output_hash": "c969b5f7547945067f5b07e9f822d0043126326f979d950a8b1b045057fbb818"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "923a53439acca5a6be0c3377", "attempt_id": "34993c0109ee", "run_group_id": "E4_core:A5|closed|qwen3b:h2", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:17:16.444008Z", "task_family": "synthetic", "document_id": "syn_03021_2000", "document_hash": "3ac205b464552356699a4d2b0a7bf62497ead36e5bca73982e11ae5ee3810731", "question_id": "q11_aggregation", "split": "test", "cluster_id": "syn_03021_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 2, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1914, "query_tokens": 473, "message_tokens": 379, "message_utf8_bytes": 900, "cache_tensor_bytes": null, "serialized_payload_bytes": 900, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 11861.411251011305, "hops": {"1": 6925.180334015749, "2": 4936.230916995555}, "receiver_prefill_ms": 185.53416605573148, "ttft_ms": 11.885375017300248, "decode_ms": 71.10737496986985}, "total_wall_ms": 12130.342293065041, "ttft_ms": 11.885375017300248, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 6}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "948485", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 19, "retention_rate": 0.2923076923076923, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.15384615384615385, "status": 1.0, "place": 0.0, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 836}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 19, "retention_rate": 0.2923076923076923, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.15384615384615385, "status": 1.0, "place": 0.0, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 836}, "2": {"facts_total": 65, "facts_retained": 19, "retention_rate": 0.2923076923076923, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.15384615384615385, "status": 1.0, "place": 0.0, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 836}}}, "reference_key_id": "keys/synthetic/syn_03021_2000.json", "task_scores": {"correct": 0.0, "abstained": 0.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "2ae8e199245ad716748b3e420d7b2c8e50b7daeeebd965effa93b31557ba8942", "output_hash": "244f716d260fa5f07cbedd5d728983a25a4991af5624be73b358ad829974e9c4"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "7d5deff60354eb05c358d6fe", "attempt_id": "2ef1b9d7397f", "run_group_id": "E4_core:A5|closed|qwen3b:h2", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:17:16.727312Z", "task_family": "synthetic", "document_id": "syn_03021_2000", "document_hash": "3ac205b464552356699a4d2b0a7bf62497ead36e5bca73982e11ae5ee3810731", "question_id": "q12_unit_conversion", "split": "test", "cluster_id": "syn_03021_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 2, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1914, "query_tokens": 468, "message_tokens": 379, "message_utf8_bytes": 900, "cache_tensor_bytes": null, "serialized_payload_bytes": 900, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 11861.411251011305, "hops": {"1": 6925.180334015749, "2": 4936.230916995555}, "receiver_prefill_ms": 186.10749999061227, "ttft_ms": 11.895458097569644, "decode_ms": 47.342833946458995}, "total_wall_ms": 12107.148251030594, "ttft_ms": 11.895458097569644, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 4}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "22.3", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 19, "retention_rate": 0.2923076923076923, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.15384615384615385, "status": 1.0, "place": 0.0, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 836}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 19, "retention_rate": 0.2923076923076923, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.15384615384615385, "status": 1.0, "place": 0.0, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 836}, "2": {"facts_total": 65, "facts_retained": 19, "retention_rate": 0.2923076923076923, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.15384615384615385, "status": 1.0, "place": 0.0, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 836}}}, "reference_key_id": "keys/synthetic/syn_03021_2000.json", "task_scores": {"correct": 0.0, "abstained": 0.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "2ae8e199245ad716748b3e420d7b2c8e50b7daeeebd965effa93b31557ba8942", "output_hash": "8afb0bb35e730fd32e7805d65081241bd21f43709df37165f4ec91c763f8a99d"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "3f9ad0cfa4fa59893a8f4167", "attempt_id": "d6e4e50c1f17", "run_group_id": "E4_core:A5|closed|qwen3b:h2", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:17:16.987982Z", "task_family": "synthetic", "document_id": "syn_03021_2000", "document_hash": "3ac205b464552356699a4d2b0a7bf62497ead36e5bca73982e11ae5ee3810731", "question_id": "q13_unanswerable", "split": "test", "cluster_id": "syn_03021_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 2, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1914, "query_tokens": 473, "message_tokens": 379, "message_utf8_bytes": 900, "cache_tensor_bytes": null, "serialized_payload_bytes": 900, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 11861.411251011305, "hops": {"1": 6925.180334015749, "2": 4936.230916995555}, "receiver_prefill_ms": 186.39020796399564, "ttft_ms": 11.895709089003503, "decode_ms": 11.9151669787243}, "total_wall_ms": 12071.982958936132, "ttft_ms": 11.895709089003503, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 19, "retention_rate": 0.2923076923076923, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.15384615384615385, "status": 1.0, "place": 0.0, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 836}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 19, "retention_rate": 0.2923076923076923, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.15384615384615385, "status": 1.0, "place": 0.0, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 836}, "2": {"facts_total": 65, "facts_retained": 19, "retention_rate": 0.2923076923076923, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.15384615384615385, "status": 1.0, "place": 0.0, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 836}}}, "reference_key_id": "keys/synthetic/syn_03021_2000.json", "task_scores": {"correct": 1.0, "abstained": 1.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "2ae8e199245ad716748b3e420d7b2c8e50b7daeeebd965effa93b31557ba8942", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "b12fbc4830987bc406e7b478", "attempt_id": "3882d715257c", "run_group_id": "E4_core:A5|closed|qwen3b:h2", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:17:17.213008Z", "task_family": "synthetic", "document_id": "syn_03021_2000", "document_hash": "3ac205b464552356699a4d2b0a7bf62497ead36e5bca73982e11ae5ee3810731", "question_id": "q14_unanswerable", "split": "test", "cluster_id": "syn_03021_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 2, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1914, "query_tokens": 475, "message_tokens": 379, "message_utf8_bytes": 900, "cache_tensor_bytes": null, "serialized_payload_bytes": 900, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 11861.411251011305, "hops": {"1": 6925.180334015749, "2": 4936.230916995555}, "receiver_prefill_ms": 185.91550004202873, "ttft_ms": 11.890333029441535, "decode_ms": 11.884249979630113}, "total_wall_ms": 12071.47366704885, "ttft_ms": 11.890333029441535, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 19, "retention_rate": 0.2923076923076923, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.15384615384615385, "status": 1.0, "place": 0.0, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 836}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 19, "retention_rate": 0.2923076923076923, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.15384615384615385, "status": 1.0, "place": 0.0, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 836}, "2": {"facts_total": 65, "facts_retained": 19, "retention_rate": 0.2923076923076923, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.15384615384615385, "status": 1.0, "place": 0.0, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 836}}}, "reference_key_id": "keys/synthetic/syn_03021_2000.json", "task_scores": {"correct": 1.0, "abstained": 1.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "2ae8e199245ad716748b3e420d7b2c8e50b7daeeebd965effa93b31557ba8942", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "6188a92e08f0081dca6dee37", "attempt_id": "41447bf3d157", "run_group_id": "E4_core:A5|closed|qwen3b:h4", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:17:27.297102Z", "task_family": "synthetic", "document_id": "syn_03021_2000", "document_hash": "3ac205b464552356699a4d2b0a7bf62497ead36e5bca73982e11ae5ee3810731", "question_id": "q00_single_fact", "split": "test", "cluster_id": "syn_03021_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 4, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1914, "query_tokens": 460, "message_tokens": 379, "message_utf8_bytes": 900, "cache_tensor_bytes": null, "serialized_payload_bytes": 900, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 21718.831210047938, "hops": {"1": 6925.180334015749, "2": 4936.230916995555, "3": 4929.1688749799505, "4": 4928.251084056683}, "receiver_prefill_ms": 185.4541659122333, "ttft_ms": 12.101124972105026, "decode_ms": 11.930375010706484}, "total_wall_ms": 21928.757168119773, "ttft_ms": 12.101124972105026, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 19, "retention_rate": 0.2923076923076923, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.15384615384615385, "status": 1.0, "place": 0.0, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 836}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 19, "retention_rate": 0.2923076923076923, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.15384615384615385, "status": 1.0, "place": 0.0, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 836}, "2": {"facts_total": 65, "facts_retained": 19, "retention_rate": 0.2923076923076923, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.15384615384615385, "status": 1.0, "place": 0.0, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 836}, "3": {"facts_total": 65, "facts_retained": 19, "retention_rate": 0.2923076923076923, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.15384615384615385, "status": 1.0, "place": 0.0, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 836}, "4": {"facts_total": 65, "facts_retained": 19, "retention_rate": 0.2923076923076923, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.15384615384615385, "status": 1.0, "place": 0.0, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 836}}}, "reference_key_id": "keys/synthetic/syn_03021_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "2ae8e199245ad716748b3e420d7b2c8e50b7daeeebd965effa93b31557ba8942", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "b37a266cc8d9dfce80ad23ce", "attempt_id": "313c3d8b7482", "run_group_id": "E4_core:A5|closed|qwen3b:h4", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:17:27.522476Z", "task_family": "synthetic", "document_id": "syn_03021_2000", "document_hash": "3ac205b464552356699a4d2b0a7bf62497ead36e5bca73982e11ae5ee3810731", "question_id": "q01_single_fact", "split": "test", "cluster_id": "syn_03021_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 4, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1914, "query_tokens": 466, "message_tokens": 379, "message_utf8_bytes": 900, "cache_tensor_bytes": null, "serialized_payload_bytes": 900, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 21718.831210047938, "hops": {"1": 6925.180334015749, "2": 4936.230916995555, "3": 4929.1688749799505, "4": 4928.251084056683}, "receiver_prefill_ms": 185.8140000840649, "ttft_ms": 12.124125030823052, "decode_ms": 11.888958979398012}, "total_wall_ms": 21929.065418080427, "ttft_ms": 12.124125030823052, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 19, "retention_rate": 0.2923076923076923, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.15384615384615385, "status": 1.0, "place": 0.0, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 836}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 19, "retention_rate": 0.2923076923076923, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.15384615384615385, "status": 1.0, "place": 0.0, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 836}, "2": {"facts_total": 65, "facts_retained": 19, "retention_rate": 0.2923076923076923, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.15384615384615385, "status": 1.0, "place": 0.0, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 836}, "3": {"facts_total": 65, "facts_retained": 19, "retention_rate": 0.2923076923076923, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.15384615384615385, "status": 1.0, "place": 0.0, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 836}, "4": {"facts_total": 65, "facts_retained": 19, "retention_rate": 0.2923076923076923, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.15384615384615385, "status": 1.0, "place": 0.0, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 836}}}, "reference_key_id": "keys/synthetic/syn_03021_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "2ae8e199245ad716748b3e420d7b2c8e50b7daeeebd965effa93b31557ba8942", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "334c164a1bc5401a1d1d4a94", "attempt_id": "324ab047e1fa", "run_group_id": "E4_core:A5|closed|qwen3b:h4", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:17:27.747573Z", "task_family": "synthetic", "document_id": "syn_03021_2000", "document_hash": "3ac205b464552356699a4d2b0a7bf62497ead36e5bca73982e11ae5ee3810731", "question_id": "q02_exact_copy", "split": "test", "cluster_id": "syn_03021_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 4, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1914, "query_tokens": 466, "message_tokens": 379, "message_utf8_bytes": 900, "cache_tensor_bytes": null, "serialized_payload_bytes": 900, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 21718.831210047938, "hops": {"1": 6925.180334015749, "2": 4936.230916995555, "3": 4929.1688749799505, "4": 4928.251084056683}, "receiver_prefill_ms": 185.54466601926833, "ttft_ms": 11.909666936844587, "decode_ms": 11.862624902278185}, "total_wall_ms": 21928.5229180241, "ttft_ms": 11.909666936844587, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 19, "retention_rate": 0.2923076923076923, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.15384615384615385, "status": 1.0, "place": 0.0, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 836}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 19, "retention_rate": 0.2923076923076923, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.15384615384615385, "status": 1.0, "place": 0.0, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 836}, "2": {"facts_total": 65, "facts_retained": 19, "retention_rate": 0.2923076923076923, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.15384615384615385, "status": 1.0, "place": 0.0, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 836}, "3": {"facts_total": 65, "facts_retained": 19, "retention_rate": 0.2923076923076923, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.15384615384615385, "status": 1.0, "place": 0.0, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 836}, "4": {"facts_total": 65, "facts_retained": 19, "retention_rate": 0.2923076923076923, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.15384615384615385, "status": 1.0, "place": 0.0, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 836}}}, "reference_key_id": "keys/synthetic/syn_03021_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "2ae8e199245ad716748b3e420d7b2c8e50b7daeeebd965effa93b31557ba8942", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "2a24fe23550381fc6d12a493", "attempt_id": "14e482fc6afb", "run_group_id": "E4_core:A5|closed|qwen3b:h4", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:17:27.971741Z", "task_family": "synthetic", "document_id": "syn_03021_2000", "document_hash": "3ac205b464552356699a4d2b0a7bf62497ead36e5bca73982e11ae5ee3810731", "question_id": "q03_exact_copy", "split": "test", "cluster_id": "syn_03021_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 4, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1914, "query_tokens": 463, "message_tokens": 379, "message_utf8_bytes": 900, "cache_tensor_bytes": null, "serialized_payload_bytes": 900, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 21718.831210047938, "hops": {"1": 6925.180334015749, "2": 4936.230916995555, "3": 4929.1688749799505, "4": 4928.251084056683}, "receiver_prefill_ms": 185.74850005097687, "ttft_ms": 11.91266707610339, "decode_ms": 11.861374950967729}, "total_wall_ms": 21928.743835072964, "ttft_ms": 11.91266707610339, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 19, "retention_rate": 0.2923076923076923, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.15384615384615385, "status": 1.0, "place": 0.0, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 836}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 19, "retention_rate": 0.2923076923076923, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.15384615384615385, "status": 1.0, "place": 0.0, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 836}, "2": {"facts_total": 65, "facts_retained": 19, "retention_rate": 0.2923076923076923, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.15384615384615385, "status": 1.0, "place": 0.0, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 836}, "3": {"facts_total": 65, "facts_retained": 19, "retention_rate": 0.2923076923076923, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.15384615384615385, "status": 1.0, "place": 0.0, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 836}, "4": {"facts_total": 65, "facts_retained": 19, "retention_rate": 0.2923076923076923, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.15384615384615385, "status": 1.0, "place": 0.0, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 836}}}, "reference_key_id": "keys/synthetic/syn_03021_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "2ae8e199245ad716748b3e420d7b2c8e50b7daeeebd965effa93b31557ba8942", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "7ff9215247eee67040055cb4", "attempt_id": "1b763429023b", "run_group_id": "E4_core:A5|closed|qwen3b:h4", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:17:28.195920Z", "task_family": "synthetic", "document_id": "syn_03021_2000", "document_hash": "3ac205b464552356699a4d2b0a7bf62497ead36e5bca73982e11ae5ee3810731", "question_id": "q04_multi_hop", "split": "test", "cluster_id": "syn_03021_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 4, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1914, "query_tokens": 470, "message_tokens": 379, "message_utf8_bytes": 900, "cache_tensor_bytes": null, "serialized_payload_bytes": 900, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 21718.831210047938, "hops": {"1": 6925.180334015749, "2": 4936.230916995555, "3": 4929.1688749799505, "4": 4928.251084056683}, "receiver_prefill_ms": 185.8582500135526, "ttft_ms": 11.76045904867351, "decode_ms": 83.18525005597621}, "total_wall_ms": 22000.01679302659, "ttft_ms": 11.76045904867351, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 7}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "4862425", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 19, "retention_rate": 0.2923076923076923, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.15384615384615385, "status": 1.0, "place": 0.0, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 836}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 19, "retention_rate": 0.2923076923076923, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.15384615384615385, "status": 1.0, "place": 0.0, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 836}, "2": {"facts_total": 65, "facts_retained": 19, "retention_rate": 0.2923076923076923, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.15384615384615385, "status": 1.0, "place": 0.0, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 836}, "3": {"facts_total": 65, "facts_retained": 19, "retention_rate": 0.2923076923076923, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.15384615384615385, "status": 1.0, "place": 0.0, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 836}, "4": {"facts_total": 65, "facts_retained": 19, "retention_rate": 0.2923076923076923, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.15384615384615385, "status": 1.0, "place": 0.0, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 836}}}, "reference_key_id": "keys/synthetic/syn_03021_2000.json", "task_scores": {"correct": 0.0, "abstained": 0.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "2ae8e199245ad716748b3e420d7b2c8e50b7daeeebd965effa93b31557ba8942", "output_hash": "0424a5744857d229fe68c1068d2f0379845129a69ec2dc68e917df297c717d20"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "a9fb0641a28affd9d0e1ec1f", "attempt_id": "fd98b563ad32", "run_group_id": "E4_core:A5|closed|qwen3b:h4", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:17:28.491428Z", "task_family": "synthetic", "document_id": "syn_03021_2000", "document_hash": "3ac205b464552356699a4d2b0a7bf62497ead36e5bca73982e11ae5ee3810731", "question_id": "q05_multi_hop", "split": "test", "cluster_id": "syn_03021_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 4, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1914, "query_tokens": 471, "message_tokens": 379, "message_utf8_bytes": 900, "cache_tensor_bytes": null, "serialized_payload_bytes": 900, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 21718.831210047938, "hops": {"1": 6925.180334015749, "2": 4936.230916995555, "3": 4929.1688749799505, "4": 4928.251084056683}, "receiver_prefill_ms": 186.05199991725385, "ttft_ms": 11.880790931172669, "decode_ms": 11.9449159828946}, "total_wall_ms": 21929.09279302694, "ttft_ms": 11.880790931172669, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 19, "retention_rate": 0.2923076923076923, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.15384615384615385, "status": 1.0, "place": 0.0, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 836}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 19, "retention_rate": 0.2923076923076923, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.15384615384615385, "status": 1.0, "place": 0.0, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 836}, "2": {"facts_total": 65, "facts_retained": 19, "retention_rate": 0.2923076923076923, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.15384615384615385, "status": 1.0, "place": 0.0, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 836}, "3": {"facts_total": 65, "facts_retained": 19, "retention_rate": 0.2923076923076923, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.15384615384615385, "status": 1.0, "place": 0.0, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 836}, "4": {"facts_total": 65, "facts_retained": 19, "retention_rate": 0.2923076923076923, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.15384615384615385, "status": 1.0, "place": 0.0, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 836}}}, "reference_key_id": "keys/synthetic/syn_03021_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "2ae8e199245ad716748b3e420d7b2c8e50b7daeeebd965effa93b31557ba8942", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "bd53a0fb92bebbc2b1546988", "attempt_id": "b933262aaf69", "run_group_id": "E4_core:A5|closed|qwen3b:h4", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:17:28.715745Z", "task_family": "synthetic", "document_id": "syn_03021_2000", "document_hash": "3ac205b464552356699a4d2b0a7bf62497ead36e5bca73982e11ae5ee3810731", "question_id": "q06_negation", "split": "test", "cluster_id": "syn_03021_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 4, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1914, "query_tokens": 461, "message_tokens": 379, "message_utf8_bytes": 900, "cache_tensor_bytes": null, "serialized_payload_bytes": 900, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 21718.831210047938, "hops": {"1": 6925.180334015749, "2": 4936.230916995555, "3": 4929.1688749799505, "4": 4928.251084056683}, "receiver_prefill_ms": 185.33075007144362, "ttft_ms": 11.88700005877763, "decode_ms": 11.853583040647209}, "total_wall_ms": 21928.27733512968, "ttft_ms": 11.88700005877763, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "no", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 19, "retention_rate": 0.2923076923076923, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.15384615384615385, "status": 1.0, "place": 0.0, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 836}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 19, "retention_rate": 0.2923076923076923, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.15384615384615385, "status": 1.0, "place": 0.0, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 836}, "2": {"facts_total": 65, "facts_retained": 19, "retention_rate": 0.2923076923076923, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.15384615384615385, "status": 1.0, "place": 0.0, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 836}, "3": {"facts_total": 65, "facts_retained": 19, "retention_rate": 0.2923076923076923, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.15384615384615385, "status": 1.0, "place": 0.0, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 836}, "4": {"facts_total": 65, "facts_retained": 19, "retention_rate": 0.2923076923076923, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.15384615384615385, "status": 1.0, "place": 0.0, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 836}}}, "reference_key_id": "keys/synthetic/syn_03021_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "2ae8e199245ad716748b3e420d7b2c8e50b7daeeebd965effa93b31557ba8942", "output_hash": "9390298f3fb0c5b160498935d79cb139aef28e1c47358b4bbba61862b9c26e59"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "50d7cb9607e4aea486f16554", "attempt_id": "ad00f756bbb7", "run_group_id": "E4_core:A5|closed|qwen3b:h4", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:17:28.939639Z", "task_family": "synthetic", "document_id": "syn_03021_2000", "document_hash": "3ac205b464552356699a4d2b0a7bf62497ead36e5bca73982e11ae5ee3810731", "question_id": "q07_negation", "split": "test", "cluster_id": "syn_03021_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 4, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1914, "query_tokens": 462, "message_tokens": 379, "message_utf8_bytes": 900, "cache_tensor_bytes": null, "serialized_payload_bytes": 900, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 21718.831210047938, "hops": {"1": 6925.180334015749, "2": 4936.230916995555, "3": 4929.1688749799505, "4": 4928.251084056683}, "receiver_prefill_ms": 185.6141668977216, "ttft_ms": 12.056250008754432, "decode_ms": 11.868375004269183}, "total_wall_ms": 21928.752668085508, "ttft_ms": 12.056250008754432, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "no", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 19, "retention_rate": 0.2923076923076923, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.15384615384615385, "status": 1.0, "place": 0.0, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 836}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 19, "retention_rate": 0.2923076923076923, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.15384615384615385, "status": 1.0, "place": 0.0, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 836}, "2": {"facts_total": 65, "facts_retained": 19, "retention_rate": 0.2923076923076923, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.15384615384615385, "status": 1.0, "place": 0.0, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 836}, "3": {"facts_total": 65, "facts_retained": 19, "retention_rate": 0.2923076923076923, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.15384615384615385, "status": 1.0, "place": 0.0, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 836}, "4": {"facts_total": 65, "facts_retained": 19, "retention_rate": 0.2923076923076923, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.15384615384615385, "status": 1.0, "place": 0.0, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 836}}}, "reference_key_id": "keys/synthetic/syn_03021_2000.json", "task_scores": {"correct": 0.0, "abstained": 0.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "2ae8e199245ad716748b3e420d7b2c8e50b7daeeebd965effa93b31557ba8942", "output_hash": "9390298f3fb0c5b160498935d79cb139aef28e1c47358b4bbba61862b9c26e59"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "39ef09f1bc1b8779d26cd2f7", "attempt_id": "d363604cd250", "run_group_id": "E4_core:A5|closed|qwen3b:h4", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:17:29.163946Z", "task_family": "synthetic", "document_id": "syn_03021_2000", "document_hash": "3ac205b464552356699a4d2b0a7bf62497ead36e5bca73982e11ae5ee3810731", "question_id": "q08_revision", "split": "test", "cluster_id": "syn_03021_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 4, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1914, "query_tokens": 478, "message_tokens": 379, "message_utf8_bytes": 900, "cache_tensor_bytes": null, "serialized_payload_bytes": 900, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 21718.831210047938, "hops": {"1": 6925.180334015749, "2": 4936.230916995555, "3": 4929.1688749799505, "4": 4928.251084056683}, "receiver_prefill_ms": 186.0741659766063, "ttft_ms": 11.923792073503137, "decode_ms": 11.898875003680587}, "total_wall_ms": 21929.129543015733, "ttft_ms": 11.923792073503137, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 19, "retention_rate": 0.2923076923076923, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.15384615384615385, "status": 1.0, "place": 0.0, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 836}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 19, "retention_rate": 0.2923076923076923, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.15384615384615385, "status": 1.0, "place": 0.0, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 836}, "2": {"facts_total": 65, "facts_retained": 19, "retention_rate": 0.2923076923076923, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.15384615384615385, "status": 1.0, "place": 0.0, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 836}, "3": {"facts_total": 65, "facts_retained": 19, "retention_rate": 0.2923076923076923, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.15384615384615385, "status": 1.0, "place": 0.0, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 836}, "4": {"facts_total": 65, "facts_retained": 19, "retention_rate": 0.2923076923076923, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.15384615384615385, "status": 1.0, "place": 0.0, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 836}}}, "reference_key_id": "keys/synthetic/syn_03021_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "2ae8e199245ad716748b3e420d7b2c8e50b7daeeebd965effa93b31557ba8942", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "eede002bf2c062763323d238", "attempt_id": "d7f89023dd8a", "run_group_id": "E4_core:A5|closed|qwen3b:h4", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:17:29.388606Z", "task_family": "synthetic", "document_id": "syn_03021_2000", "document_hash": "3ac205b464552356699a4d2b0a7bf62497ead36e5bca73982e11ae5ee3810731", "question_id": "q09_revision", "split": "test", "cluster_id": "syn_03021_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 4, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1914, "query_tokens": 469, "message_tokens": 379, "message_utf8_bytes": 900, "cache_tensor_bytes": null, "serialized_payload_bytes": 900, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 21718.831210047938, "hops": {"1": 6925.180334015749, "2": 4936.230916995555, "3": 4929.1688749799505, "4": 4928.251084056683}, "receiver_prefill_ms": 185.64858299214393, "ttft_ms": 11.903666076250374, "decode_ms": 11.875040945596993}, "total_wall_ms": 21928.643668070436, "ttft_ms": 11.903666076250374, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 19, "retention_rate": 0.2923076923076923, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.15384615384615385, "status": 1.0, "place": 0.0, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 836}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 19, "retention_rate": 0.2923076923076923, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.15384615384615385, "status": 1.0, "place": 0.0, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 836}, "2": {"facts_total": 65, "facts_retained": 19, "retention_rate": 0.2923076923076923, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.15384615384615385, "status": 1.0, "place": 0.0, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 836}, "3": {"facts_total": 65, "facts_retained": 19, "retention_rate": 0.2923076923076923, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.15384615384615385, "status": 1.0, "place": 0.0, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 836}, "4": {"facts_total": 65, "facts_retained": 19, "retention_rate": 0.2923076923076923, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.15384615384615385, "status": 1.0, "place": 0.0, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 836}}}, "reference_key_id": "keys/synthetic/syn_03021_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "2ae8e199245ad716748b3e420d7b2c8e50b7daeeebd965effa93b31557ba8942", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "dc035f5702898dec1ec13e0f", "attempt_id": "4304156c7f74", "run_group_id": "E4_core:A5|closed|qwen3b:h4", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:17:29.612869Z", "task_family": "synthetic", "document_id": "syn_03021_2000", "document_hash": "3ac205b464552356699a4d2b0a7bf62497ead36e5bca73982e11ae5ee3810731", "question_id": "q10_temporal", "split": "test", "cluster_id": "syn_03021_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 4, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1914, "query_tokens": 477, "message_tokens": 379, "message_utf8_bytes": 900, "cache_tensor_bytes": null, "serialized_payload_bytes": 900, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 21718.831210047938, "hops": {"1": 6925.180334015749, "2": 4936.230916995555, "3": 4929.1688749799505, "4": 4928.251084056683}, "receiver_prefill_ms": 186.49454100523144, "ttft_ms": 11.791749973781407, "decode_ms": 59.556583990342915}, "total_wall_ms": 21977.058627060615, "ttft_ms": 11.791749973781407, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 5}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "Kovar-9", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 19, "retention_rate": 0.2923076923076923, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.15384615384615385, "status": 1.0, "place": 0.0, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 836}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 19, "retention_rate": 0.2923076923076923, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.15384615384615385, "status": 1.0, "place": 0.0, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 836}, "2": {"facts_total": 65, "facts_retained": 19, "retention_rate": 0.2923076923076923, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.15384615384615385, "status": 1.0, "place": 0.0, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 836}, "3": {"facts_total": 65, "facts_retained": 19, "retention_rate": 0.2923076923076923, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.15384615384615385, "status": 1.0, "place": 0.0, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 836}, "4": {"facts_total": 65, "facts_retained": 19, "retention_rate": 0.2923076923076923, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.15384615384615385, "status": 1.0, "place": 0.0, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 836}}}, "reference_key_id": "keys/synthetic/syn_03021_2000.json", "task_scores": {"correct": 0.0, "abstained": 0.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "2ae8e199245ad716748b3e420d7b2c8e50b7daeeebd965effa93b31557ba8942", "output_hash": "c969b5f7547945067f5b07e9f822d0043126326f979d950a8b1b045057fbb818"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "17a39860bce1f6b09bb5e166", "attempt_id": "6a987e520219", "run_group_id": "E4_core:A5|closed|qwen3b:h4", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:17:29.885376Z", "task_family": "synthetic", "document_id": "syn_03021_2000", "document_hash": "3ac205b464552356699a4d2b0a7bf62497ead36e5bca73982e11ae5ee3810731", "question_id": "q11_aggregation", "split": "test", "cluster_id": "syn_03021_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 4, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1914, "query_tokens": 473, "message_tokens": 379, "message_utf8_bytes": 900, "cache_tensor_bytes": null, "serialized_payload_bytes": 900, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 21718.831210047938, "hops": {"1": 6925.180334015749, "2": 4936.230916995555, "3": 4929.1688749799505, "4": 4928.251084056683}, "receiver_prefill_ms": 186.0176669433713, "ttft_ms": 11.870583053678274, "decode_ms": 71.49125006981194}, "total_wall_ms": 21988.602960132994, "ttft_ms": 11.870583053678274, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 6}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "948485", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 19, "retention_rate": 0.2923076923076923, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.15384615384615385, "status": 1.0, "place": 0.0, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 836}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 19, "retention_rate": 0.2923076923076923, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.15384615384615385, "status": 1.0, "place": 0.0, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 836}, "2": {"facts_total": 65, "facts_retained": 19, "retention_rate": 0.2923076923076923, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.15384615384615385, "status": 1.0, "place": 0.0, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 836}, "3": {"facts_total": 65, "facts_retained": 19, "retention_rate": 0.2923076923076923, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.15384615384615385, "status": 1.0, "place": 0.0, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 836}, "4": {"facts_total": 65, "facts_retained": 19, "retention_rate": 0.2923076923076923, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.15384615384615385, "status": 1.0, "place": 0.0, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 836}}}, "reference_key_id": "keys/synthetic/syn_03021_2000.json", "task_scores": {"correct": 0.0, "abstained": 0.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "2ae8e199245ad716748b3e420d7b2c8e50b7daeeebd965effa93b31557ba8942", "output_hash": "244f716d260fa5f07cbedd5d728983a25a4991af5624be73b358ad829974e9c4"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "b162f886218841a39399343c", "attempt_id": "e4e63600f593", "run_group_id": "E4_core:A5|closed|qwen3b:h4", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:17:30.169852Z", "task_family": "synthetic", "document_id": "syn_03021_2000", "document_hash": "3ac205b464552356699a4d2b0a7bf62497ead36e5bca73982e11ae5ee3810731", "question_id": "q12_unit_conversion", "split": "test", "cluster_id": "syn_03021_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 4, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1914, "query_tokens": 468, "message_tokens": 379, "message_utf8_bytes": 900, "cache_tensor_bytes": null, "serialized_payload_bytes": 900, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 21718.831210047938, "hops": {"1": 6925.180334015749, "2": 4936.230916995555, "3": 4929.1688749799505, "4": 4928.251084056683}, "receiver_prefill_ms": 185.62729097902775, "ttft_ms": 11.925874976441264, "decode_ms": 47.405832912772894}, "total_wall_ms": 21964.225627016276, "ttft_ms": 11.925874976441264, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 4}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "22.3", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 19, "retention_rate": 0.2923076923076923, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.15384615384615385, "status": 1.0, "place": 0.0, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 836}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 19, "retention_rate": 0.2923076923076923, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.15384615384615385, "status": 1.0, "place": 0.0, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 836}, "2": {"facts_total": 65, "facts_retained": 19, "retention_rate": 0.2923076923076923, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.15384615384615385, "status": 1.0, "place": 0.0, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 836}, "3": {"facts_total": 65, "facts_retained": 19, "retention_rate": 0.2923076923076923, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.15384615384615385, "status": 1.0, "place": 0.0, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 836}, "4": {"facts_total": 65, "facts_retained": 19, "retention_rate": 0.2923076923076923, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.15384615384615385, "status": 1.0, "place": 0.0, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 836}}}, "reference_key_id": "keys/synthetic/syn_03021_2000.json", "task_scores": {"correct": 0.0, "abstained": 0.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "2ae8e199245ad716748b3e420d7b2c8e50b7daeeebd965effa93b31557ba8942", "output_hash": "8afb0bb35e730fd32e7805d65081241bd21f43709df37165f4ec91c763f8a99d"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "cb082e8097f0ee67f129a690", "attempt_id": "754476efbb71", "run_group_id": "E4_core:A5|closed|qwen3b:h4", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:17:30.429713Z", "task_family": "synthetic", "document_id": "syn_03021_2000", "document_hash": "3ac205b464552356699a4d2b0a7bf62497ead36e5bca73982e11ae5ee3810731", "question_id": "q13_unanswerable", "split": "test", "cluster_id": "syn_03021_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 4, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1914, "query_tokens": 473, "message_tokens": 379, "message_utf8_bytes": 900, "cache_tensor_bytes": null, "serialized_payload_bytes": 900, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 21718.831210047938, "hops": {"1": 6925.180334015749, "2": 4936.230916995555, "3": 4929.1688749799505, "4": 4928.251084056683}, "receiver_prefill_ms": 186.26170803327113, "ttft_ms": 11.866291984915733, "decode_ms": 11.894959025084972}, "total_wall_ms": 21929.25254406873, "ttft_ms": 11.866291984915733, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 19, "retention_rate": 0.2923076923076923, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.15384615384615385, "status": 1.0, "place": 0.0, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 836}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 19, "retention_rate": 0.2923076923076923, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.15384615384615385, "status": 1.0, "place": 0.0, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 836}, "2": {"facts_total": 65, "facts_retained": 19, "retention_rate": 0.2923076923076923, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.15384615384615385, "status": 1.0, "place": 0.0, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 836}, "3": {"facts_total": 65, "facts_retained": 19, "retention_rate": 0.2923076923076923, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.15384615384615385, "status": 1.0, "place": 0.0, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 836}, "4": {"facts_total": 65, "facts_retained": 19, "retention_rate": 0.2923076923076923, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.15384615384615385, "status": 1.0, "place": 0.0, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 836}}}, "reference_key_id": "keys/synthetic/syn_03021_2000.json", "task_scores": {"correct": 1.0, "abstained": 1.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "2ae8e199245ad716748b3e420d7b2c8e50b7daeeebd965effa93b31557ba8942", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "89e13de313a27070f96961f5", "attempt_id": "1192f28acd67", "run_group_id": "E4_core:A5|closed|qwen3b:h4", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:17:30.655083Z", "task_family": "synthetic", "document_id": "syn_03021_2000", "document_hash": "3ac205b464552356699a4d2b0a7bf62497ead36e5bca73982e11ae5ee3810731", "question_id": "q14_unanswerable", "split": "test", "cluster_id": "syn_03021_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 4, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1914, "query_tokens": 475, "message_tokens": 379, "message_utf8_bytes": 900, "cache_tensor_bytes": null, "serialized_payload_bytes": 900, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 21718.831210047938, "hops": {"1": 6925.180334015749, "2": 4936.230916995555, "3": 4929.1688749799505, "4": 4928.251084056683}, "receiver_prefill_ms": 185.66583399660885, "ttft_ms": 11.722749914042652, "decode_ms": 11.803374974988401}, "total_wall_ms": 21928.473793086596, "ttft_ms": 11.722749914042652, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 19, "retention_rate": 0.2923076923076923, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.15384615384615385, "status": 1.0, "place": 0.0, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 836}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 19, "retention_rate": 0.2923076923076923, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.15384615384615385, "status": 1.0, "place": 0.0, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 836}, "2": {"facts_total": 65, "facts_retained": 19, "retention_rate": 0.2923076923076923, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.15384615384615385, "status": 1.0, "place": 0.0, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 836}, "3": {"facts_total": 65, "facts_retained": 19, "retention_rate": 0.2923076923076923, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.15384615384615385, "status": 1.0, "place": 0.0, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 836}, "4": {"facts_total": 65, "facts_retained": 19, "retention_rate": 0.2923076923076923, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.15384615384615385, "status": 1.0, "place": 0.0, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 836}}}, "reference_key_id": "keys/synthetic/syn_03021_2000.json", "task_scores": {"correct": 1.0, "abstained": 1.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "2ae8e199245ad716748b3e420d7b2c8e50b7daeeebd965effa93b31557ba8942", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "1f7aa4ac28db426e9462c704", "attempt_id": "509f0586ceb4", "run_group_id": "E4_core:A5|closed|qwen3b:h8", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:17:50.591976Z", "task_family": "synthetic", "document_id": "syn_03021_2000", "document_hash": "3ac205b464552356699a4d2b0a7bf62497ead36e5bca73982e11ae5ee3810731", "question_id": "q00_single_fact", "split": "test", "cluster_id": "syn_03021_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 8, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1914, "query_tokens": 460, "message_tokens": 379, "message_utf8_bytes": 900, "cache_tensor_bytes": null, "serialized_payload_bytes": 900, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 41427.24412702955, "hops": {"1": 6925.180334015749, "2": 4936.230916995555, "3": 4929.1688749799505, "4": 4928.251084056683, "5": 4930.099040968344, "6": 4925.894750049338, "7": 4927.353916922584, "8": 4925.06520904135}, "receiver_prefill_ms": 185.69120892789215, "ttft_ms": 11.868500034324825, "decode_ms": 11.89162500668317}, "total_wall_ms": 41637.12862704415, "ttft_ms": 11.868500034324825, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 19, "retention_rate": 0.2923076923076923, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.15384615384615385, "status": 1.0, "place": 0.0, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 836}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 19, "retention_rate": 0.2923076923076923, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.15384615384615385, "status": 1.0, "place": 0.0, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 836}, "2": {"facts_total": 65, "facts_retained": 19, "retention_rate": 0.2923076923076923, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.15384615384615385, "status": 1.0, "place": 0.0, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 836}, "3": {"facts_total": 65, "facts_retained": 19, "retention_rate": 0.2923076923076923, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.15384615384615385, "status": 1.0, "place": 0.0, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 836}, "4": {"facts_total": 65, "facts_retained": 19, "retention_rate": 0.2923076923076923, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.15384615384615385, "status": 1.0, "place": 0.0, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 836}, "5": {"facts_total": 65, "facts_retained": 19, "retention_rate": 0.2923076923076923, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.15384615384615385, "status": 1.0, "place": 0.0, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 836}, "6": {"facts_total": 65, "facts_retained": 19, "retention_rate": 0.2923076923076923, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.15384615384615385, "status": 1.0, "place": 0.0, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 836}, "7": {"facts_total": 65, "facts_retained": 19, "retention_rate": 0.2923076923076923, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.15384615384615385, "status": 1.0, "place": 0.0, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 836}, "8": {"facts_total": 65, "facts_retained": 19, "retention_rate": 0.2923076923076923, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.15384615384615385, "status": 1.0, "place": 0.0, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 836}}}, "reference_key_id": "keys/synthetic/syn_03021_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "2ae8e199245ad716748b3e420d7b2c8e50b7daeeebd965effa93b31557ba8942", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "30827b67cd75275a944ecfa3", "attempt_id": "4b956dd8efe7", "run_group_id": "E4_core:A5|closed|qwen3b:h8", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:17:50.817498Z", "task_family": "synthetic", "document_id": "syn_03021_2000", "document_hash": "3ac205b464552356699a4d2b0a7bf62497ead36e5bca73982e11ae5ee3810731", "question_id": "q01_single_fact", "split": "test", "cluster_id": "syn_03021_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 8, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1914, "query_tokens": 466, "message_tokens": 379, "message_utf8_bytes": 900, "cache_tensor_bytes": null, "serialized_payload_bytes": 900, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 41427.24412702955, "hops": {"1": 6925.180334015749, "2": 4936.230916995555, "3": 4929.1688749799505, "4": 4928.251084056683, "5": 4930.099040968344, "6": 4925.894750049338, "7": 4927.353916922584, "8": 4925.06520904135}, "receiver_prefill_ms": 185.96262496430427, "ttft_ms": 11.96454104501754, "decode_ms": 11.789167067036033}, "total_wall_ms": 41637.3655020725, "ttft_ms": 11.96454104501754, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 19, "retention_rate": 0.2923076923076923, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.15384615384615385, "status": 1.0, "place": 0.0, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 836}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 19, "retention_rate": 0.2923076923076923, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.15384615384615385, "status": 1.0, "place": 0.0, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 836}, "2": {"facts_total": 65, "facts_retained": 19, "retention_rate": 0.2923076923076923, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.15384615384615385, "status": 1.0, "place": 0.0, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 836}, "3": {"facts_total": 65, "facts_retained": 19, "retention_rate": 0.2923076923076923, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.15384615384615385, "status": 1.0, "place": 0.0, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 836}, "4": {"facts_total": 65, "facts_retained": 19, "retention_rate": 0.2923076923076923, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.15384615384615385, "status": 1.0, "place": 0.0, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 836}, "5": {"facts_total": 65, "facts_retained": 19, "retention_rate": 0.2923076923076923, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.15384615384615385, "status": 1.0, "place": 0.0, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 836}, "6": {"facts_total": 65, "facts_retained": 19, "retention_rate": 0.2923076923076923, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.15384615384615385, "status": 1.0, "place": 0.0, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 836}, "7": {"facts_total": 65, "facts_retained": 19, "retention_rate": 0.2923076923076923, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.15384615384615385, "status": 1.0, "place": 0.0, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 836}, "8": {"facts_total": 65, "facts_retained": 19, "retention_rate": 0.2923076923076923, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.15384615384615385, "status": 1.0, "place": 0.0, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 836}}}, "reference_key_id": "keys/synthetic/syn_03021_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "2ae8e199245ad716748b3e420d7b2c8e50b7daeeebd965effa93b31557ba8942", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "bcf0fac8f19bd29a06f64021", "attempt_id": "9dabebf42c6a", "run_group_id": "E4_core:A5|closed|qwen3b:h8", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:17:51.042525Z", "task_family": "synthetic", "document_id": "syn_03021_2000", "document_hash": "3ac205b464552356699a4d2b0a7bf62497ead36e5bca73982e11ae5ee3810731", "question_id": "q02_exact_copy", "split": "test", "cluster_id": "syn_03021_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 8, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1914, "query_tokens": 466, "message_tokens": 379, "message_utf8_bytes": 900, "cache_tensor_bytes": null, "serialized_payload_bytes": 900, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 41427.24412702955, "hops": {"1": 6925.180334015749, "2": 4936.230916995555, "3": 4929.1688749799505, "4": 4928.251084056683, "5": 4930.099040968344, "6": 4925.894750049338, "7": 4927.353916922584, "8": 4925.06520904135}, "receiver_prefill_ms": 184.82099997345358, "ttft_ms": 11.880457983352244, "decode_ms": 12.16229097917676}, "total_wall_ms": 41636.48425193969, "ttft_ms": 11.880457983352244, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 19, "retention_rate": 0.2923076923076923, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.15384615384615385, "status": 1.0, "place": 0.0, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 836}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 19, "retention_rate": 0.2923076923076923, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.15384615384615385, "status": 1.0, "place": 0.0, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 836}, "2": {"facts_total": 65, "facts_retained": 19, "retention_rate": 0.2923076923076923, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.15384615384615385, "status": 1.0, "place": 0.0, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 836}, "3": {"facts_total": 65, "facts_retained": 19, "retention_rate": 0.2923076923076923, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.15384615384615385, "status": 1.0, "place": 0.0, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 836}, "4": {"facts_total": 65, "facts_retained": 19, "retention_rate": 0.2923076923076923, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.15384615384615385, "status": 1.0, "place": 0.0, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 836}, "5": {"facts_total": 65, "facts_retained": 19, "retention_rate": 0.2923076923076923, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.15384615384615385, "status": 1.0, "place": 0.0, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 836}, "6": {"facts_total": 65, "facts_retained": 19, "retention_rate": 0.2923076923076923, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.15384615384615385, "status": 1.0, "place": 0.0, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 836}, "7": {"facts_total": 65, "facts_retained": 19, "retention_rate": 0.2923076923076923, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.15384615384615385, "status": 1.0, "place": 0.0, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 836}, "8": {"facts_total": 65, "facts_retained": 19, "retention_rate": 0.2923076923076923, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.15384615384615385, "status": 1.0, "place": 0.0, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 836}}}, "reference_key_id": "keys/synthetic/syn_03021_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "2ae8e199245ad716748b3e420d7b2c8e50b7daeeebd965effa93b31557ba8942", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "cd80899899222e560f14c4e9", "attempt_id": "78a54017cc1b", "run_group_id": "E4_core:A5|closed|qwen3b:h8", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:17:51.266420Z", "task_family": "synthetic", "document_id": "syn_03021_2000", "document_hash": "3ac205b464552356699a4d2b0a7bf62497ead36e5bca73982e11ae5ee3810731", "question_id": "q03_exact_copy", "split": "test", "cluster_id": "syn_03021_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 8, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1914, "query_tokens": 463, "message_tokens": 379, "message_utf8_bytes": 900, "cache_tensor_bytes": null, "serialized_payload_bytes": 900, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 41427.24412702955, "hops": {"1": 6925.180334015749, "2": 4936.230916995555, "3": 4929.1688749799505, "4": 4928.251084056683, "5": 4930.099040968344, "6": 4925.894750049338, "7": 4927.353916922584, "8": 4925.06520904135}, "receiver_prefill_ms": 185.6077500851825, "ttft_ms": 11.889124987646937, "decode_ms": 11.934250011108816}, "total_wall_ms": 41637.05754408147, "ttft_ms": 11.889124987646937, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 19, "retention_rate": 0.2923076923076923, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.15384615384615385, "status": 1.0, "place": 0.0, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 836}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 19, "retention_rate": 0.2923076923076923, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.15384615384615385, "status": 1.0, "place": 0.0, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 836}, "2": {"facts_total": 65, "facts_retained": 19, "retention_rate": 0.2923076923076923, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.15384615384615385, "status": 1.0, "place": 0.0, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 836}, "3": {"facts_total": 65, "facts_retained": 19, "retention_rate": 0.2923076923076923, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.15384615384615385, "status": 1.0, "place": 0.0, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 836}, "4": {"facts_total": 65, "facts_retained": 19, "retention_rate": 0.2923076923076923, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.15384615384615385, "status": 1.0, "place": 0.0, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 836}, "5": {"facts_total": 65, "facts_retained": 19, "retention_rate": 0.2923076923076923, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.15384615384615385, "status": 1.0, "place": 0.0, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 836}, "6": {"facts_total": 65, "facts_retained": 19, "retention_rate": 0.2923076923076923, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.15384615384615385, "status": 1.0, "place": 0.0, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 836}, "7": {"facts_total": 65, "facts_retained": 19, "retention_rate": 0.2923076923076923, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.15384615384615385, "status": 1.0, "place": 0.0, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 836}, "8": {"facts_total": 65, "facts_retained": 19, "retention_rate": 0.2923076923076923, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.15384615384615385, "status": 1.0, "place": 0.0, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 836}}}, "reference_key_id": "keys/synthetic/syn_03021_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "2ae8e199245ad716748b3e420d7b2c8e50b7daeeebd965effa93b31557ba8942", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "83ec9388a20a29c81d9609ee", "attempt_id": "332b65d6084c", "run_group_id": "E4_core:A5|closed|qwen3b:h8", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:17:51.491055Z", "task_family": "synthetic", "document_id": "syn_03021_2000", "document_hash": "3ac205b464552356699a4d2b0a7bf62497ead36e5bca73982e11ae5ee3810731", "question_id": "q04_multi_hop", "split": "test", "cluster_id": "syn_03021_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 8, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1914, "query_tokens": 470, "message_tokens": 379, "message_utf8_bytes": 900, "cache_tensor_bytes": null, "serialized_payload_bytes": 900, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 41427.24412702955, "hops": {"1": 6925.180334015749, "2": 4936.230916995555, "3": 4929.1688749799505, "4": 4928.251084056683, "5": 4930.099040968344, "6": 4925.894750049338, "7": 4927.353916922584, "8": 4925.06520904135}, "receiver_prefill_ms": 185.66837499383837, "ttft_ms": 11.86079194303602, "decode_ms": 11.88775000628084}, "total_wall_ms": 41637.0368769858, "ttft_ms": 11.86079194303602, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 19, "retention_rate": 0.2923076923076923, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.15384615384615385, "status": 1.0, "place": 0.0, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 836}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 19, "retention_rate": 0.2923076923076923, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.15384615384615385, "status": 1.0, "place": 0.0, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 836}, "2": {"facts_total": 65, "facts_retained": 19, "retention_rate": 0.2923076923076923, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.15384615384615385, "status": 1.0, "place": 0.0, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 836}, "3": {"facts_total": 65, "facts_retained": 19, "retention_rate": 0.2923076923076923, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.15384615384615385, "status": 1.0, "place": 0.0, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 836}, "4": {"facts_total": 65, "facts_retained": 19, "retention_rate": 0.2923076923076923, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.15384615384615385, "status": 1.0, "place": 0.0, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 836}, "5": {"facts_total": 65, "facts_retained": 19, "retention_rate": 0.2923076923076923, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.15384615384615385, "status": 1.0, "place": 0.0, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 836}, "6": {"facts_total": 65, "facts_retained": 19, "retention_rate": 0.2923076923076923, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.15384615384615385, "status": 1.0, "place": 0.0, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 836}, "7": {"facts_total": 65, "facts_retained": 19, "retention_rate": 0.2923076923076923, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.15384615384615385, "status": 1.0, "place": 0.0, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 836}, "8": {"facts_total": 65, "facts_retained": 19, "retention_rate": 0.2923076923076923, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.15384615384615385, "status": 1.0, "place": 0.0, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 836}}}, "reference_key_id": "keys/synthetic/syn_03021_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "2ae8e199245ad716748b3e420d7b2c8e50b7daeeebd965effa93b31557ba8942", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "30ffe3b485f5c222e3deccb5", "attempt_id": "d32427a19537", "run_group_id": "E4_core:A5|closed|qwen3b:h8", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:17:51.715250Z", "task_family": "synthetic", "document_id": "syn_03021_2000", "document_hash": "3ac205b464552356699a4d2b0a7bf62497ead36e5bca73982e11ae5ee3810731", "question_id": "q05_multi_hop", "split": "test", "cluster_id": "syn_03021_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 8, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1914, "query_tokens": 471, "message_tokens": 379, "message_utf8_bytes": 900, "cache_tensor_bytes": null, "serialized_payload_bytes": 900, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 41427.24412702955, "hops": {"1": 6925.180334015749, "2": 4936.230916995555, "3": 4929.1688749799505, "4": 4928.251084056683, "5": 4930.099040968344, "6": 4925.894750049338, "7": 4927.353916922584, "8": 4925.06520904135}, "receiver_prefill_ms": 186.16775004193187, "ttft_ms": 11.90533407498151, "decode_ms": 11.867707944475114}, "total_wall_ms": 41637.58175203111, "ttft_ms": 11.90533407498151, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 19, "retention_rate": 0.2923076923076923, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.15384615384615385, "status": 1.0, "place": 0.0, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 836}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 19, "retention_rate": 0.2923076923076923, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.15384615384615385, "status": 1.0, "place": 0.0, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 836}, "2": {"facts_total": 65, "facts_retained": 19, "retention_rate": 0.2923076923076923, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.15384615384615385, "status": 1.0, "place": 0.0, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 836}, "3": {"facts_total": 65, "facts_retained": 19, "retention_rate": 0.2923076923076923, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.15384615384615385, "status": 1.0, "place": 0.0, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 836}, "4": {"facts_total": 65, "facts_retained": 19, "retention_rate": 0.2923076923076923, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.15384615384615385, "status": 1.0, "place": 0.0, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 836}, "5": {"facts_total": 65, "facts_retained": 19, "retention_rate": 0.2923076923076923, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.15384615384615385, "status": 1.0, "place": 0.0, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 836}, "6": {"facts_total": 65, "facts_retained": 19, "retention_rate": 0.2923076923076923, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.15384615384615385, "status": 1.0, "place": 0.0, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 836}, "7": {"facts_total": 65, "facts_retained": 19, "retention_rate": 0.2923076923076923, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.15384615384615385, "status": 1.0, "place": 0.0, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 836}, "8": {"facts_total": 65, "facts_retained": 19, "retention_rate": 0.2923076923076923, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.15384615384615385, "status": 1.0, "place": 0.0, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 836}}}, "reference_key_id": "keys/synthetic/syn_03021_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "2ae8e199245ad716748b3e420d7b2c8e50b7daeeebd965effa93b31557ba8942", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "2f0b7943918e7a32b5613d88", "attempt_id": "5bcda0480d37", "run_group_id": "E4_core:A5|closed|qwen3b:h8", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:17:51.940219Z", "task_family": "synthetic", "document_id": "syn_03021_2000", "document_hash": "3ac205b464552356699a4d2b0a7bf62497ead36e5bca73982e11ae5ee3810731", "question_id": "q06_negation", "split": "test", "cluster_id": "syn_03021_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 8, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1914, "query_tokens": 461, "message_tokens": 379, "message_utf8_bytes": 900, "cache_tensor_bytes": null, "serialized_payload_bytes": 900, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 41427.24412702955, "hops": {"1": 6925.180334015749, "2": 4936.230916995555, "3": 4929.1688749799505, "4": 4928.251084056683, "5": 4930.099040968344, "6": 4925.894750049338, "7": 4927.353916922584, "8": 4925.06520904135}, "receiver_prefill_ms": 184.79249998927116, "ttft_ms": 11.864167056046426, "decode_ms": 11.8454999756068}, "total_wall_ms": 41636.138335918076, "ttft_ms": 11.864167056046426, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "no", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 19, "retention_rate": 0.2923076923076923, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.15384615384615385, "status": 1.0, "place": 0.0, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 836}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 19, "retention_rate": 0.2923076923076923, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.15384615384615385, "status": 1.0, "place": 0.0, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 836}, "2": {"facts_total": 65, "facts_retained": 19, "retention_rate": 0.2923076923076923, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.15384615384615385, "status": 1.0, "place": 0.0, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 836}, "3": {"facts_total": 65, "facts_retained": 19, "retention_rate": 0.2923076923076923, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.15384615384615385, "status": 1.0, "place": 0.0, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 836}, "4": {"facts_total": 65, "facts_retained": 19, "retention_rate": 0.2923076923076923, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.15384615384615385, "status": 1.0, "place": 0.0, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 836}, "5": {"facts_total": 65, "facts_retained": 19, "retention_rate": 0.2923076923076923, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.15384615384615385, "status": 1.0, "place": 0.0, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 836}, "6": {"facts_total": 65, "facts_retained": 19, "retention_rate": 0.2923076923076923, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.15384615384615385, "status": 1.0, "place": 0.0, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 836}, "7": {"facts_total": 65, "facts_retained": 19, "retention_rate": 0.2923076923076923, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.15384615384615385, "status": 1.0, "place": 0.0, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 836}, "8": {"facts_total": 65, "facts_retained": 19, "retention_rate": 0.2923076923076923, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.15384615384615385, "status": 1.0, "place": 0.0, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 836}}}, "reference_key_id": "keys/synthetic/syn_03021_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "2ae8e199245ad716748b3e420d7b2c8e50b7daeeebd965effa93b31557ba8942", "output_hash": "9390298f3fb0c5b160498935d79cb139aef28e1c47358b4bbba61862b9c26e59"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "05ffc38086f0787c6e85c95a", "attempt_id": "486fe2b6e0e0", "run_group_id": "E4_core:A5|closed|qwen3b:h8", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:17:52.163716Z", "task_family": "synthetic", "document_id": "syn_03021_2000", "document_hash": "3ac205b464552356699a4d2b0a7bf62497ead36e5bca73982e11ae5ee3810731", "question_id": "q07_negation", "split": "test", "cluster_id": "syn_03021_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 8, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1914, "query_tokens": 462, "message_tokens": 379, "message_utf8_bytes": 900, "cache_tensor_bytes": null, "serialized_payload_bytes": 900, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 41427.24412702955, "hops": {"1": 6925.180334015749, "2": 4936.230916995555, "3": 4929.1688749799505, "4": 4928.251084056683, "5": 4930.099040968344, "6": 4925.894750049338, "7": 4927.353916922584, "8": 4925.06520904135}, "receiver_prefill_ms": 185.69795798975974, "ttft_ms": 11.89908292144537, "decode_ms": 11.869458016008139}, "total_wall_ms": 41637.0997519698, "ttft_ms": 11.89908292144537, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "no", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 19, "retention_rate": 0.2923076923076923, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.15384615384615385, "status": 1.0, "place": 0.0, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 836}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 19, "retention_rate": 0.2923076923076923, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.15384615384615385, "status": 1.0, "place": 0.0, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 836}, "2": {"facts_total": 65, "facts_retained": 19, "retention_rate": 0.2923076923076923, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.15384615384615385, "status": 1.0, "place": 0.0, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 836}, "3": {"facts_total": 65, "facts_retained": 19, "retention_rate": 0.2923076923076923, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.15384615384615385, "status": 1.0, "place": 0.0, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 836}, "4": {"facts_total": 65, "facts_retained": 19, "retention_rate": 0.2923076923076923, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.15384615384615385, "status": 1.0, "place": 0.0, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 836}, "5": {"facts_total": 65, "facts_retained": 19, "retention_rate": 0.2923076923076923, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.15384615384615385, "status": 1.0, "place": 0.0, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 836}, "6": {"facts_total": 65, "facts_retained": 19, "retention_rate": 0.2923076923076923, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.15384615384615385, "status": 1.0, "place": 0.0, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 836}, "7": {"facts_total": 65, "facts_retained": 19, "retention_rate": 0.2923076923076923, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.15384615384615385, "status": 1.0, "place": 0.0, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 836}, "8": {"facts_total": 65, "facts_retained": 19, "retention_rate": 0.2923076923076923, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.15384615384615385, "status": 1.0, "place": 0.0, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 836}}}, "reference_key_id": "keys/synthetic/syn_03021_2000.json", "task_scores": {"correct": 0.0, "abstained": 0.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "2ae8e199245ad716748b3e420d7b2c8e50b7daeeebd965effa93b31557ba8942", "output_hash": "9390298f3fb0c5b160498935d79cb139aef28e1c47358b4bbba61862b9c26e59"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "5b795e637a0776a0efff044b", "attempt_id": "4d2df655a5d0", "run_group_id": "E4_core:A5|closed|qwen3b:h8", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:17:52.387894Z", "task_family": "synthetic", "document_id": "syn_03021_2000", "document_hash": "3ac205b464552356699a4d2b0a7bf62497ead36e5bca73982e11ae5ee3810731", "question_id": "q08_revision", "split": "test", "cluster_id": "syn_03021_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 8, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1914, "query_tokens": 478, "message_tokens": 379, "message_utf8_bytes": 900, "cache_tensor_bytes": null, "serialized_payload_bytes": 900, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 41427.24412702955, "hops": {"1": 6925.180334015749, "2": 4936.230916995555, "3": 4929.1688749799505, "4": 4928.251084056683, "5": 4930.099040968344, "6": 4925.894750049338, "7": 4927.353916922584, "8": 4925.06520904135}, "receiver_prefill_ms": 186.6634590551257, "ttft_ms": 11.852583033032715, "decode_ms": 11.909167049452662}, "total_wall_ms": 41638.05362698622, "ttft_ms": 11.852583033032715, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 19, "retention_rate": 0.2923076923076923, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.15384615384615385, "status": 1.0, "place": 0.0, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 836}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 19, "retention_rate": 0.2923076923076923, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.15384615384615385, "status": 1.0, "place": 0.0, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 836}, "2": {"facts_total": 65, "facts_retained": 19, "retention_rate": 0.2923076923076923, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.15384615384615385, "status": 1.0, "place": 0.0, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 836}, "3": {"facts_total": 65, "facts_retained": 19, "retention_rate": 0.2923076923076923, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.15384615384615385, "status": 1.0, "place": 0.0, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 836}, "4": {"facts_total": 65, "facts_retained": 19, "retention_rate": 0.2923076923076923, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.15384615384615385, "status": 1.0, "place": 0.0, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 836}, "5": {"facts_total": 65, "facts_retained": 19, "retention_rate": 0.2923076923076923, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.15384615384615385, "status": 1.0, "place": 0.0, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 836}, "6": {"facts_total": 65, "facts_retained": 19, "retention_rate": 0.2923076923076923, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.15384615384615385, "status": 1.0, "place": 0.0, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 836}, "7": {"facts_total": 65, "facts_retained": 19, "retention_rate": 0.2923076923076923, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.15384615384615385, "status": 1.0, "place": 0.0, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 836}, "8": {"facts_total": 65, "facts_retained": 19, "retention_rate": 0.2923076923076923, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.15384615384615385, "status": 1.0, "place": 0.0, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 836}}}, "reference_key_id": "keys/synthetic/syn_03021_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "2ae8e199245ad716748b3e420d7b2c8e50b7daeeebd965effa93b31557ba8942", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "5ea45deb13726ba60308bfec", "attempt_id": "5623b1124162", "run_group_id": "E4_core:A5|closed|qwen3b:h8", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:17:52.612954Z", "task_family": "synthetic", "document_id": "syn_03021_2000", "document_hash": "3ac205b464552356699a4d2b0a7bf62497ead36e5bca73982e11ae5ee3810731", "question_id": "q09_revision", "split": "test", "cluster_id": "syn_03021_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 8, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1914, "query_tokens": 469, "message_tokens": 379, "message_utf8_bytes": 900, "cache_tensor_bytes": null, "serialized_payload_bytes": 900, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 41427.24412702955, "hops": {"1": 6925.180334015749, "2": 4936.230916995555, "3": 4929.1688749799505, "4": 4928.251084056683, "5": 4930.099040968344, "6": 4925.894750049338, "7": 4927.353916922584, "8": 4925.06520904135}, "receiver_prefill_ms": 185.8509590383619, "ttft_ms": 11.948750005103648, "decode_ms": 11.846582987345755}, "total_wall_ms": 41637.2786690481, "ttft_ms": 11.948750005103648, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 19, "retention_rate": 0.2923076923076923, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.15384615384615385, "status": 1.0, "place": 0.0, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 836}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 19, "retention_rate": 0.2923076923076923, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.15384615384615385, "status": 1.0, "place": 0.0, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 836}, "2": {"facts_total": 65, "facts_retained": 19, "retention_rate": 0.2923076923076923, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.15384615384615385, "status": 1.0, "place": 0.0, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 836}, "3": {"facts_total": 65, "facts_retained": 19, "retention_rate": 0.2923076923076923, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.15384615384615385, "status": 1.0, "place": 0.0, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 836}, "4": {"facts_total": 65, "facts_retained": 19, "retention_rate": 0.2923076923076923, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.15384615384615385, "status": 1.0, "place": 0.0, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 836}, "5": {"facts_total": 65, "facts_retained": 19, "retention_rate": 0.2923076923076923, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.15384615384615385, "status": 1.0, "place": 0.0, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 836}, "6": {"facts_total": 65, "facts_retained": 19, "retention_rate": 0.2923076923076923, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.15384615384615385, "status": 1.0, "place": 0.0, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 836}, "7": {"facts_total": 65, "facts_retained": 19, "retention_rate": 0.2923076923076923, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.15384615384615385, "status": 1.0, "place": 0.0, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 836}, "8": {"facts_total": 65, "facts_retained": 19, "retention_rate": 0.2923076923076923, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.15384615384615385, "status": 1.0, "place": 0.0, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 836}}}, "reference_key_id": "keys/synthetic/syn_03021_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "2ae8e199245ad716748b3e420d7b2c8e50b7daeeebd965effa93b31557ba8942", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "c7b937c223f4ac6fb25e9cf4", "attempt_id": "16b457b0913a", "run_group_id": "E4_core:A5|closed|qwen3b:h8", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:17:52.837790Z", "task_family": "synthetic", "document_id": "syn_03021_2000", "document_hash": "3ac205b464552356699a4d2b0a7bf62497ead36e5bca73982e11ae5ee3810731", "question_id": "q10_temporal", "split": "test", "cluster_id": "syn_03021_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 8, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1914, "query_tokens": 477, "message_tokens": 379, "message_utf8_bytes": 900, "cache_tensor_bytes": null, "serialized_payload_bytes": 900, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 41427.24412702955, "hops": {"1": 6925.180334015749, "2": 4936.230916995555, "3": 4929.1688749799505, "4": 4928.251084056683, "5": 4930.099040968344, "6": 4925.894750049338, "7": 4927.353916922584, "8": 4925.06520904135}, "receiver_prefill_ms": 186.66258396115154, "ttft_ms": 11.848125024698675, "decode_ms": 59.24800003413111}, "total_wall_ms": 41685.39125204552, "ttft_ms": 11.848125024698675, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 5}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "Kovar-9", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 19, "retention_rate": 0.2923076923076923, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.15384615384615385, "status": 1.0, "place": 0.0, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 836}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 19, "retention_rate": 0.2923076923076923, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.15384615384615385, "status": 1.0, "place": 0.0, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 836}, "2": {"facts_total": 65, "facts_retained": 19, "retention_rate": 0.2923076923076923, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.15384615384615385, "status": 1.0, "place": 0.0, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 836}, "3": {"facts_total": 65, "facts_retained": 19, "retention_rate": 0.2923076923076923, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.15384615384615385, "status": 1.0, "place": 0.0, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 836}, "4": {"facts_total": 65, "facts_retained": 19, "retention_rate": 0.2923076923076923, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.15384615384615385, "status": 1.0, "place": 0.0, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 836}, "5": {"facts_total": 65, "facts_retained": 19, "retention_rate": 0.2923076923076923, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.15384615384615385, "status": 1.0, "place": 0.0, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 836}, "6": {"facts_total": 65, "facts_retained": 19, "retention_rate": 0.2923076923076923, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.15384615384615385, "status": 1.0, "place": 0.0, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 836}, "7": {"facts_total": 65, "facts_retained": 19, "retention_rate": 0.2923076923076923, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.15384615384615385, "status": 1.0, "place": 0.0, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 836}, "8": {"facts_total": 65, "facts_retained": 19, "retention_rate": 0.2923076923076923, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.15384615384615385, "status": 1.0, "place": 0.0, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 836}}}, "reference_key_id": "keys/synthetic/syn_03021_2000.json", "task_scores": {"correct": 0.0, "abstained": 0.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "2ae8e199245ad716748b3e420d7b2c8e50b7daeeebd965effa93b31557ba8942", "output_hash": "c969b5f7547945067f5b07e9f822d0043126326f979d950a8b1b045057fbb818"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "172399ef80b88f0477f16468", "attempt_id": "0e7ab39679a1", "run_group_id": "E4_core:A5|closed|qwen3b:h8", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:17:53.110680Z", "task_family": "synthetic", "document_id": "syn_03021_2000", "document_hash": "3ac205b464552356699a4d2b0a7bf62497ead36e5bca73982e11ae5ee3810731", "question_id": "q11_aggregation", "split": "test", "cluster_id": "syn_03021_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 8, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1914, "query_tokens": 473, "message_tokens": 379, "message_utf8_bytes": 900, "cache_tensor_bytes": null, "serialized_payload_bytes": 900, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 41427.24412702955, "hops": {"1": 6925.180334015749, "2": 4936.230916995555, "3": 4929.1688749799505, "4": 4928.251084056683, "5": 4930.099040968344, "6": 4925.894750049338, "7": 4927.353916922584, "8": 4925.06520904135}, "receiver_prefill_ms": 185.79341599252075, "ttft_ms": 11.93629193585366, "decode_ms": 71.10645901411772}, "total_wall_ms": 41696.49600202683, "ttft_ms": 11.93629193585366, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 6}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "972465", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 19, "retention_rate": 0.2923076923076923, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.15384615384615385, "status": 1.0, "place": 0.0, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 836}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 19, "retention_rate": 0.2923076923076923, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.15384615384615385, "status": 1.0, "place": 0.0, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 836}, "2": {"facts_total": 65, "facts_retained": 19, "retention_rate": 0.2923076923076923, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.15384615384615385, "status": 1.0, "place": 0.0, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 836}, "3": {"facts_total": 65, "facts_retained": 19, "retention_rate": 0.2923076923076923, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.15384615384615385, "status": 1.0, "place": 0.0, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 836}, "4": {"facts_total": 65, "facts_retained": 19, "retention_rate": 0.2923076923076923, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.15384615384615385, "status": 1.0, "place": 0.0, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 836}, "5": {"facts_total": 65, "facts_retained": 19, "retention_rate": 0.2923076923076923, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.15384615384615385, "status": 1.0, "place": 0.0, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 836}, "6": {"facts_total": 65, "facts_retained": 19, "retention_rate": 0.2923076923076923, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.15384615384615385, "status": 1.0, "place": 0.0, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 836}, "7": {"facts_total": 65, "facts_retained": 19, "retention_rate": 0.2923076923076923, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.15384615384615385, "status": 1.0, "place": 0.0, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 836}, "8": {"facts_total": 65, "facts_retained": 19, "retention_rate": 0.2923076923076923, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.15384615384615385, "status": 1.0, "place": 0.0, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 836}}}, "reference_key_id": "keys/synthetic/syn_03021_2000.json", "task_scores": {"correct": 0.0, "abstained": 0.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "2ae8e199245ad716748b3e420d7b2c8e50b7daeeebd965effa93b31557ba8942", "output_hash": "0a00e59631ae965bf9ab3d3d1f18cb59d92aea878baa86f032ccec5b0696dce0"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "c0cde1360a80dd776b398901", "attempt_id": "35867cd3df7c", "run_group_id": "E4_core:A5|closed|qwen3b:h8", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:17:53.394481Z", "task_family": "synthetic", "document_id": "syn_03021_2000", "document_hash": "3ac205b464552356699a4d2b0a7bf62497ead36e5bca73982e11ae5ee3810731", "question_id": "q12_unit_conversion", "split": "test", "cluster_id": "syn_03021_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 8, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1914, "query_tokens": 468, "message_tokens": 379, "message_utf8_bytes": 900, "cache_tensor_bytes": null, "serialized_payload_bytes": 900, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 41427.24412702955, "hops": {"1": 6925.180334015749, "2": 4936.230916995555, "3": 4929.1688749799505, "4": 4928.251084056683, "5": 4930.099040968344, "6": 4925.894750049338, "7": 4927.353916922584, "8": 4925.06520904135}, "receiver_prefill_ms": 185.3639589389786, "ttft_ms": 12.168999994173646, "decode_ms": 47.6904169190675}, "total_wall_ms": 41673.04825200699, "ttft_ms": 12.168999994173646, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 4}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "22.3", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 19, "retention_rate": 0.2923076923076923, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.15384615384615385, "status": 1.0, "place": 0.0, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 836}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 19, "retention_rate": 0.2923076923076923, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.15384615384615385, "status": 1.0, "place": 0.0, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 836}, "2": {"facts_total": 65, "facts_retained": 19, "retention_rate": 0.2923076923076923, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.15384615384615385, "status": 1.0, "place": 0.0, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 836}, "3": {"facts_total": 65, "facts_retained": 19, "retention_rate": 0.2923076923076923, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.15384615384615385, "status": 1.0, "place": 0.0, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 836}, "4": {"facts_total": 65, "facts_retained": 19, "retention_rate": 0.2923076923076923, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.15384615384615385, "status": 1.0, "place": 0.0, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 836}, "5": {"facts_total": 65, "facts_retained": 19, "retention_rate": 0.2923076923076923, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.15384615384615385, "status": 1.0, "place": 0.0, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 836}, "6": {"facts_total": 65, "facts_retained": 19, "retention_rate": 0.2923076923076923, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.15384615384615385, "status": 1.0, "place": 0.0, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 836}, "7": {"facts_total": 65, "facts_retained": 19, "retention_rate": 0.2923076923076923, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.15384615384615385, "status": 1.0, "place": 0.0, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 836}, "8": {"facts_total": 65, "facts_retained": 19, "retention_rate": 0.2923076923076923, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.15384615384615385, "status": 1.0, "place": 0.0, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 836}}}, "reference_key_id": "keys/synthetic/syn_03021_2000.json", "task_scores": {"correct": 0.0, "abstained": 0.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "2ae8e199245ad716748b3e420d7b2c8e50b7daeeebd965effa93b31557ba8942", "output_hash": "8afb0bb35e730fd32e7805d65081241bd21f43709df37165f4ec91c763f8a99d"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "c4b0e4f592543159cb5475f6", "attempt_id": "6c18494121ad", "run_group_id": "E4_core:A5|closed|qwen3b:h8", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:17:53.655585Z", "task_family": "synthetic", "document_id": "syn_03021_2000", "document_hash": "3ac205b464552356699a4d2b0a7bf62497ead36e5bca73982e11ae5ee3810731", "question_id": "q13_unanswerable", "split": "test", "cluster_id": "syn_03021_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 8, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1914, "query_tokens": 473, "message_tokens": 379, "message_utf8_bytes": 900, "cache_tensor_bytes": null, "serialized_payload_bytes": 900, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 41427.24412702955, "hops": {"1": 6925.180334015749, "2": 4936.230916995555, "3": 4929.1688749799505, "4": 4928.251084056683, "5": 4930.099040968344, "6": 4925.894750049338, "7": 4927.353916922584, "8": 4925.06520904135}, "receiver_prefill_ms": 186.04341603349894, "ttft_ms": 12.104833964258432, "decode_ms": 12.137249927036464}, "total_wall_ms": 41638.14587704837, "ttft_ms": 12.104833964258432, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 19, "retention_rate": 0.2923076923076923, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.15384615384615385, "status": 1.0, "place": 0.0, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 836}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 19, "retention_rate": 0.2923076923076923, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.15384615384615385, "status": 1.0, "place": 0.0, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 836}, "2": {"facts_total": 65, "facts_retained": 19, "retention_rate": 0.2923076923076923, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.15384615384615385, "status": 1.0, "place": 0.0, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 836}, "3": {"facts_total": 65, "facts_retained": 19, "retention_rate": 0.2923076923076923, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.15384615384615385, "status": 1.0, "place": 0.0, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 836}, "4": {"facts_total": 65, "facts_retained": 19, "retention_rate": 0.2923076923076923, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.15384615384615385, "status": 1.0, "place": 0.0, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 836}, "5": {"facts_total": 65, "facts_retained": 19, "retention_rate": 0.2923076923076923, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.15384615384615385, "status": 1.0, "place": 0.0, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 836}, "6": {"facts_total": 65, "facts_retained": 19, "retention_rate": 0.2923076923076923, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.15384615384615385, "status": 1.0, "place": 0.0, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 836}, "7": {"facts_total": 65, "facts_retained": 19, "retention_rate": 0.2923076923076923, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.15384615384615385, "status": 1.0, "place": 0.0, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 836}, "8": {"facts_total": 65, "facts_retained": 19, "retention_rate": 0.2923076923076923, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.15384615384615385, "status": 1.0, "place": 0.0, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 836}}}, "reference_key_id": "keys/synthetic/syn_03021_2000.json", "task_scores": {"correct": 1.0, "abstained": 1.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "2ae8e199245ad716748b3e420d7b2c8e50b7daeeebd965effa93b31557ba8942", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "185678dd0114cdb5fae5b0a9", "attempt_id": "267d568814ca", "run_group_id": "E4_core:A5|closed|qwen3b:h8", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:17:53.883722Z", "task_family": "synthetic", "document_id": "syn_03021_2000", "document_hash": "3ac205b464552356699a4d2b0a7bf62497ead36e5bca73982e11ae5ee3810731", "question_id": "q14_unanswerable", "split": "test", "cluster_id": "syn_03021_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 8, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1914, "query_tokens": 475, "message_tokens": 379, "message_utf8_bytes": 900, "cache_tensor_bytes": null, "serialized_payload_bytes": 900, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 41427.24412702955, "hops": {"1": 6925.180334015749, "2": 4936.230916995555, "3": 4929.1688749799505, "4": 4928.251084056683, "5": 4930.099040968344, "6": 4925.894750049338, "7": 4927.353916922584, "8": 4925.06520904135}, "receiver_prefill_ms": 186.3162920344621, "ttft_ms": 12.045166920870543, "decode_ms": 12.15504202991724}, "total_wall_ms": 41638.27246008441, "ttft_ms": 12.045166920870543, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 19, "retention_rate": 0.2923076923076923, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.15384615384615385, "status": 1.0, "place": 0.0, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 836}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 19, "retention_rate": 0.2923076923076923, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.15384615384615385, "status": 1.0, "place": 0.0, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 836}, "2": {"facts_total": 65, "facts_retained": 19, "retention_rate": 0.2923076923076923, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.15384615384615385, "status": 1.0, "place": 0.0, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 836}, "3": {"facts_total": 65, "facts_retained": 19, "retention_rate": 0.2923076923076923, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.15384615384615385, "status": 1.0, "place": 0.0, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 836}, "4": {"facts_total": 65, "facts_retained": 19, "retention_rate": 0.2923076923076923, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.15384615384615385, "status": 1.0, "place": 0.0, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 836}, "5": {"facts_total": 65, "facts_retained": 19, "retention_rate": 0.2923076923076923, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.15384615384615385, "status": 1.0, "place": 0.0, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 836}, "6": {"facts_total": 65, "facts_retained": 19, "retention_rate": 0.2923076923076923, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.15384615384615385, "status": 1.0, "place": 0.0, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 836}, "7": {"facts_total": 65, "facts_retained": 19, "retention_rate": 0.2923076923076923, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.15384615384615385, "status": 1.0, "place": 0.0, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 836}, "8": {"facts_total": 65, "facts_retained": 19, "retention_rate": 0.2923076923076923, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.15384615384615385, "status": 1.0, "place": 0.0, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 836}}}, "reference_key_id": "keys/synthetic/syn_03021_2000.json", "task_scores": {"correct": 1.0, "abstained": 1.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "2ae8e199245ad716748b3e420d7b2c8e50b7daeeebd965effa93b31557ba8942", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "af0e42d0b1084fed64200fde", "attempt_id": "989677fd86d5", "run_group_id": "E4_core:A4|repairable|qwen3b:h1", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:18:00.972787Z", "task_family": "synthetic", "document_id": "syn_03021_2000", "document_hash": "3ac205b464552356699a4d2b0a7bf62497ead36e5bca73982e11ae5ee3810731", "question_id": "q00_single_fact", "split": "test", "cluster_id": "syn_03021_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 1, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1914, "query_tokens": 592, "message_tokens": 512, "message_utf8_bytes": 1179, "cache_tensor_bytes": null, "serialized_payload_bytes": 1179, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 6860.111457994208, "hops": {"1": 6860.111457994208}, "receiver_prefill_ms": 226.402125088498, "ttft_ms": 12.21316703595221, "decode_ms": 11.887833010405302}, "total_wall_ms": 7111.151498975232, "ttft_ms": 12.21316703595221, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 38, "retention_rate": 0.5846153846153846, "retention_by_kind": {"person": 0.8333333333333334, "date": 0.7, "amount": 0.38461538461538464, "status": 1.0, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1179}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 38, "retention_rate": 0.5846153846153846, "retention_by_kind": {"person": 0.8333333333333334, "date": 0.7, "amount": 0.38461538461538464, "status": 1.0, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1179}}}, "reference_key_id": "keys/synthetic/syn_03021_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "f6342dcd02ebd80508f967d8bfed98ca4bc1afeea750629dadac4b492752e270", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "b2a09d16aff35ab7f9f40b15", "attempt_id": "db58afecd8ae", "run_group_id": "E4_core:A4|repairable|qwen3b:h1", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:18:01.239208Z", "task_family": "synthetic", "document_id": "syn_03021_2000", "document_hash": "3ac205b464552356699a4d2b0a7bf62497ead36e5bca73982e11ae5ee3810731", "question_id": "q01_single_fact", "split": "test", "cluster_id": "syn_03021_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 1, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1914, "query_tokens": 598, "message_tokens": 512, "message_utf8_bytes": 1179, "cache_tensor_bytes": null, "serialized_payload_bytes": 1179, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 6860.111457994208, "hops": {"1": 6860.111457994208}, "receiver_prefill_ms": 226.97008401155472, "ttft_ms": 12.136540957726538, "decode_ms": 130.98520797211677}, "total_wall_ms": 7230.709374998696, "ttft_ms": 12.136540957726538, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 11}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "Lot 536-R, Senneterre", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 38, "retention_rate": 0.5846153846153846, "retention_by_kind": {"person": 0.8333333333333334, "date": 0.7, "amount": 0.38461538461538464, "status": 1.0, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1179}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 38, "retention_rate": 0.5846153846153846, "retention_by_kind": {"person": 0.8333333333333334, "date": 0.7, "amount": 0.38461538461538464, "status": 1.0, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1179}}}, "reference_key_id": "keys/synthetic/syn_03021_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 0.8571428571428571}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "f6342dcd02ebd80508f967d8bfed98ca4bc1afeea750629dadac4b492752e270", "output_hash": "5e85af24aac0f1904d70a2fcee84f1b6f6912cc7450298af1f173002fd8e59df"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "da421a27d683d923570a3bc2", "attempt_id": "14e935e8b50a", "run_group_id": "E4_core:A4|repairable|qwen3b:h1", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:18:01.624983Z", "task_family": "synthetic", "document_id": "syn_03021_2000", "document_hash": "3ac205b464552356699a4d2b0a7bf62497ead36e5bca73982e11ae5ee3810731", "question_id": "q02_exact_copy", "split": "test", "cluster_id": "syn_03021_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 1, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1914, "query_tokens": 598, "message_tokens": 512, "message_utf8_bytes": 1179, "cache_tensor_bytes": null, "serialized_payload_bytes": 1179, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 6860.111457994208, "hops": {"1": 6860.111457994208}, "receiver_prefill_ms": 227.0039999857545, "ttft_ms": 11.921084020286798, "decode_ms": 83.41558300890028}, "total_wall_ms": 7183.151665958576, "ttft_ms": 11.921084020286798, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 7}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "88.7751", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 38, "retention_rate": 0.5846153846153846, "retention_by_kind": {"person": 0.8333333333333334, "date": 0.7, "amount": 0.38461538461538464, "status": 1.0, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1179}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 38, "retention_rate": 0.5846153846153846, "retention_by_kind": {"person": 0.8333333333333334, "date": 0.7, "amount": 0.38461538461538464, "status": 1.0, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1179}}}, "reference_key_id": "keys/synthetic/syn_03021_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "f6342dcd02ebd80508f967d8bfed98ca4bc1afeea750629dadac4b492752e270", "output_hash": "1e01e62c69f9479b681e50e59e8565c6583f5cdb803446a05ed5cb13118acdf9"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "87baee15405fb7a62bff50d2", "attempt_id": "7d40e380956a", "run_group_id": "E4_core:A4|repairable|qwen3b:h1", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:18:01.962819Z", "task_family": "synthetic", "document_id": "syn_03021_2000", "document_hash": "3ac205b464552356699a4d2b0a7bf62497ead36e5bca73982e11ae5ee3810731", "question_id": "q03_exact_copy", "split": "test", "cluster_id": "syn_03021_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 1, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1914, "query_tokens": 595, "message_tokens": 512, "message_utf8_bytes": 1179, "cache_tensor_bytes": null, "serialized_payload_bytes": 1179, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 6860.111457994208, "hops": {"1": 6860.111457994208}, "receiver_prefill_ms": 226.7918330617249, "ttft_ms": 12.025915901176631, "decode_ms": 47.71212500054389}, "total_wall_ms": 7147.156000020914, "ttft_ms": 12.025915901176631, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 4}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "Ingrid Mbeki", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 38, "retention_rate": 0.5846153846153846, "retention_by_kind": {"person": 0.8333333333333334, "date": 0.7, "amount": 0.38461538461538464, "status": 1.0, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1179}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 38, "retention_rate": 0.5846153846153846, "retention_by_kind": {"person": 0.8333333333333334, "date": 0.7, "amount": 0.38461538461538464, "status": 1.0, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1179}}}, "reference_key_id": "keys/synthetic/syn_03021_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "f6342dcd02ebd80508f967d8bfed98ca4bc1afeea750629dadac4b492752e270", "output_hash": "1dbd85090bb65fb3b72de8ca4ce32f2c234729ff4a239274e79fb574b800a79d"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "e23c03ec72a3a634e13ecf7e", "attempt_id": "f97a4fc05b84", "run_group_id": "E4_core:A4|repairable|qwen3b:h1", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:18:02.265293Z", "task_family": "synthetic", "document_id": "syn_03021_2000", "document_hash": "3ac205b464552356699a4d2b0a7bf62497ead36e5bca73982e11ae5ee3810731", "question_id": "q04_multi_hop", "split": "test", "cluster_id": "syn_03021_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 1, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1914, "query_tokens": 602, "message_tokens": 512, "message_utf8_bytes": 1179, "cache_tensor_bytes": null, "serialized_payload_bytes": 1179, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 6860.111457994208, "hops": {"1": 6860.111457994208}, "receiver_prefill_ms": 227.13754198048264, "ttft_ms": 12.27516692597419, "decode_ms": 12.184666004031897}, "total_wall_ms": 7112.408332992345, "ttft_ms": 12.27516692597419, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 38, "retention_rate": 0.5846153846153846, "retention_by_kind": {"person": 0.8333333333333334, "date": 0.7, "amount": 0.38461538461538464, "status": 1.0, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1179}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 38, "retention_rate": 0.5846153846153846, "retention_by_kind": {"person": 0.8333333333333334, "date": 0.7, "amount": 0.38461538461538464, "status": 1.0, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1179}}}, "reference_key_id": "keys/synthetic/syn_03021_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "f6342dcd02ebd80508f967d8bfed98ca4bc1afeea750629dadac4b492752e270", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "a55f5f9c37ddc26b42971966", "attempt_id": "44e010c85d45", "run_group_id": "E4_core:A4|repairable|qwen3b:h1", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:18:02.533199Z", "task_family": "synthetic", "document_id": "syn_03021_2000", "document_hash": "3ac205b464552356699a4d2b0a7bf62497ead36e5bca73982e11ae5ee3810731", "question_id": "q05_multi_hop", "split": "test", "cluster_id": "syn_03021_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 1, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1914, "query_tokens": 603, "message_tokens": 512, "message_utf8_bytes": 1179, "cache_tensor_bytes": null, "serialized_payload_bytes": 1179, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 6860.111457994208, "hops": {"1": 6860.111457994208}, "receiver_prefill_ms": 227.2484590066597, "ttft_ms": 12.140959035605192, "decode_ms": 131.11216691322625}, "total_wall_ms": 7231.239791959524, "ttft_ms": 12.140959035605192, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 11}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "BQ-59Z-6753", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 38, "retention_rate": 0.5846153846153846, "retention_by_kind": {"person": 0.8333333333333334, "date": 0.7, "amount": 0.38461538461538464, "status": 1.0, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1179}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 38, "retention_rate": 0.5846153846153846, "retention_by_kind": {"person": 0.8333333333333334, "date": 0.7, "amount": 0.38461538461538464, "status": 1.0, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1179}}}, "reference_key_id": "keys/synthetic/syn_03021_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "f6342dcd02ebd80508f967d8bfed98ca4bc1afeea750629dadac4b492752e270", "output_hash": "db250f072c856fe7ddf4bbbd16e20c71b8bad1eef5e4480acd0dac30f8ffe9cd"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "e0f9e0331ae77f1f3c4dcca6", "attempt_id": "16627c9a08a4", "run_group_id": "E4_core:A4|repairable|qwen3b:h1", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:18:02.921271Z", "task_family": "synthetic", "document_id": "syn_03021_2000", "document_hash": "3ac205b464552356699a4d2b0a7bf62497ead36e5bca73982e11ae5ee3810731", "question_id": "q06_negation", "split": "test", "cluster_id": "syn_03021_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 1, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1914, "query_tokens": 593, "message_tokens": 512, "message_utf8_bytes": 1179, "cache_tensor_bytes": null, "serialized_payload_bytes": 1179, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 6860.111457994208, "hops": {"1": 6860.111457994208}, "receiver_prefill_ms": 226.95445804856718, "ttft_ms": 11.989000020548701, "decode_ms": 11.9677500333637}, "total_wall_ms": 7111.529208021238, "ttft_ms": 11.989000020548701, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "no", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 38, "retention_rate": 0.5846153846153846, "retention_by_kind": {"person": 0.8333333333333334, "date": 0.7, "amount": 0.38461538461538464, "status": 1.0, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1179}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 38, "retention_rate": 0.5846153846153846, "retention_by_kind": {"person": 0.8333333333333334, "date": 0.7, "amount": 0.38461538461538464, "status": 1.0, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1179}}}, "reference_key_id": "keys/synthetic/syn_03021_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "f6342dcd02ebd80508f967d8bfed98ca4bc1afeea750629dadac4b492752e270", "output_hash": "9390298f3fb0c5b160498935d79cb139aef28e1c47358b4bbba61862b9c26e59"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "eaf351a567901b5d636b7901", "attempt_id": "7c7a9334f070", "run_group_id": "E4_core:A4|repairable|qwen3b:h1", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:18:03.188021Z", "task_family": "synthetic", "document_id": "syn_03021_2000", "document_hash": "3ac205b464552356699a4d2b0a7bf62497ead36e5bca73982e11ae5ee3810731", "question_id": "q07_negation", "split": "test", "cluster_id": "syn_03021_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 1, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1914, "query_tokens": 594, "message_tokens": 512, "message_utf8_bytes": 1179, "cache_tensor_bytes": null, "serialized_payload_bytes": 1179, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 6860.111457994208, "hops": {"1": 6860.111457994208}, "receiver_prefill_ms": 226.48525005206466, "ttft_ms": 12.032832950353622, "decode_ms": 11.912500020116568}, "total_wall_ms": 7111.016624025069, "ttft_ms": 12.032832950353622, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "yes", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 38, "retention_rate": 0.5846153846153846, "retention_by_kind": {"person": 0.8333333333333334, "date": 0.7, "amount": 0.38461538461538464, "status": 1.0, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1179}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 38, "retention_rate": 0.5846153846153846, "retention_by_kind": {"person": 0.8333333333333334, "date": 0.7, "amount": 0.38461538461538464, "status": 1.0, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1179}}}, "reference_key_id": "keys/synthetic/syn_03021_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "f6342dcd02ebd80508f967d8bfed98ca4bc1afeea750629dadac4b492752e270", "output_hash": "8a798890fe93817163b10b5f7bd2ca4d25d84c52739a645a889c173eee7d9d3d"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "451cfecc9918407050534893", "attempt_id": "74f1140d4484", "run_group_id": "E4_core:A4|repairable|qwen3b:h1", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:18:03.453727Z", "task_family": "synthetic", "document_id": "syn_03021_2000", "document_hash": "3ac205b464552356699a4d2b0a7bf62497ead36e5bca73982e11ae5ee3810731", "question_id": "q08_revision", "split": "test", "cluster_id": "syn_03021_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 1, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1914, "query_tokens": 610, "message_tokens": 512, "message_utf8_bytes": 1179, "cache_tensor_bytes": null, "serialized_payload_bytes": 1179, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 6860.111457994208, "hops": {"1": 6860.111457994208}, "receiver_prefill_ms": 229.28704204969108, "ttft_ms": 12.16325000859797, "decode_ms": 12.026750016957521}, "total_wall_ms": 7114.078290993348, "ttft_ms": 12.16325000859797, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 38, "retention_rate": 0.5846153846153846, "retention_by_kind": {"person": 0.8333333333333334, "date": 0.7, "amount": 0.38461538461538464, "status": 1.0, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1179}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 38, "retention_rate": 0.5846153846153846, "retention_by_kind": {"person": 0.8333333333333334, "date": 0.7, "amount": 0.38461538461538464, "status": 1.0, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1179}}}, "reference_key_id": "keys/synthetic/syn_03021_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "f6342dcd02ebd80508f967d8bfed98ca4bc1afeea750629dadac4b492752e270", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "378adc8a667fa8fddede73d1", "attempt_id": "45697d980867", "run_group_id": "E4_core:A4|repairable|qwen3b:h1", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:18:03.722608Z", "task_family": "synthetic", "document_id": "syn_03021_2000", "document_hash": "3ac205b464552356699a4d2b0a7bf62497ead36e5bca73982e11ae5ee3810731", "question_id": "q09_revision", "split": "test", "cluster_id": "syn_03021_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 1, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1914, "query_tokens": 601, "message_tokens": 512, "message_utf8_bytes": 1179, "cache_tensor_bytes": null, "serialized_payload_bytes": 1179, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 6860.111457994208, "hops": {"1": 6860.111457994208}, "receiver_prefill_ms": 226.62954207044095, "ttft_ms": 11.946208076551557, "decode_ms": 12.06508302129805}, "total_wall_ms": 7111.226874985732, "ttft_ms": 11.946208076551557, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 38, "retention_rate": 0.5846153846153846, "retention_by_kind": {"person": 0.8333333333333334, "date": 0.7, "amount": 0.38461538461538464, "status": 1.0, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1179}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 38, "retention_rate": 0.5846153846153846, "retention_by_kind": {"person": 0.8333333333333334, "date": 0.7, "amount": 0.38461538461538464, "status": 1.0, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1179}}}, "reference_key_id": "keys/synthetic/syn_03021_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "f6342dcd02ebd80508f967d8bfed98ca4bc1afeea750629dadac4b492752e270", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "bd5b8288eb43fbc74778ca88", "attempt_id": "fe63c1a2ee72", "run_group_id": "E4_core:A4|repairable|qwen3b:h1", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:18:03.988549Z", "task_family": "synthetic", "document_id": "syn_03021_2000", "document_hash": "3ac205b464552356699a4d2b0a7bf62497ead36e5bca73982e11ae5ee3810731", "question_id": "q10_temporal", "split": "test", "cluster_id": "syn_03021_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 1, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1914, "query_tokens": 609, "message_tokens": 512, "message_utf8_bytes": 1179, "cache_tensor_bytes": null, "serialized_payload_bytes": 1179, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 6860.111457994208, "hops": {"1": 6860.111457994208}, "receiver_prefill_ms": 227.37750003580004, "ttft_ms": 11.970457970164716, "decode_ms": 59.540791902691126}, "total_wall_ms": 7159.483207971789, "ttft_ms": 11.970457970164716, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 5}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "Kovar-9", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 38, "retention_rate": 0.5846153846153846, "retention_by_kind": {"person": 0.8333333333333334, "date": 0.7, "amount": 0.38461538461538464, "status": 1.0, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1179}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 38, "retention_rate": 0.5846153846153846, "retention_by_kind": {"person": 0.8333333333333334, "date": 0.7, "amount": 0.38461538461538464, "status": 1.0, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1179}}}, "reference_key_id": "keys/synthetic/syn_03021_2000.json", "task_scores": {"correct": 0.0, "abstained": 0.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "f6342dcd02ebd80508f967d8bfed98ca4bc1afeea750629dadac4b492752e270", "output_hash": "c969b5f7547945067f5b07e9f822d0043126326f979d950a8b1b045057fbb818"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "693d6ee093db0ab9f80e4ef8", "attempt_id": "3467a640ac8a", "run_group_id": "E4_core:A4|repairable|qwen3b:h1", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:18:04.302529Z", "task_family": "synthetic", "document_id": "syn_03021_2000", "document_hash": "3ac205b464552356699a4d2b0a7bf62497ead36e5bca73982e11ae5ee3810731", "question_id": "q11_aggregation", "split": "test", "cluster_id": "syn_03021_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 1, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1914, "query_tokens": 605, "message_tokens": 512, "message_utf8_bytes": 1179, "cache_tensor_bytes": null, "serialized_payload_bytes": 1179, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 6860.111457994208, "hops": {"1": 6860.111457994208}, "receiver_prefill_ms": 227.26708301343024, "ttft_ms": 11.9785419665277, "decode_ms": 83.4002080373466}, "total_wall_ms": 7183.243583072908, "ttft_ms": 11.9785419665277, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 7}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "1302481", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 38, "retention_rate": 0.5846153846153846, "retention_by_kind": {"person": 0.8333333333333334, "date": 0.7, "amount": 0.38461538461538464, "status": 1.0, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1179}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 38, "retention_rate": 0.5846153846153846, "retention_by_kind": {"person": 0.8333333333333334, "date": 0.7, "amount": 0.38461538461538464, "status": 1.0, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1179}}}, "reference_key_id": "keys/synthetic/syn_03021_2000.json", "task_scores": {"correct": 0.0, "abstained": 0.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "f6342dcd02ebd80508f967d8bfed98ca4bc1afeea750629dadac4b492752e270", "output_hash": "1c130ebf03e5051f031da47a49e8c91caae34e3c2144f451ec045a8e1c85549d"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "f1dec0bdc802d2531d269fdc", "attempt_id": "b6562f2ad885", "run_group_id": "E4_core:A4|repairable|qwen3b:h1", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:18:04.640930Z", "task_family": "synthetic", "document_id": "syn_03021_2000", "document_hash": "3ac205b464552356699a4d2b0a7bf62497ead36e5bca73982e11ae5ee3810731", "question_id": "q12_unit_conversion", "split": "test", "cluster_id": "syn_03021_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 1, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1914, "query_tokens": 600, "message_tokens": 512, "message_utf8_bytes": 1179, "cache_tensor_bytes": null, "serialized_payload_bytes": 1179, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 6860.111457994208, "hops": {"1": 6860.111457994208}, "receiver_prefill_ms": 226.34291695430875, "ttft_ms": 12.062499998137355, "decode_ms": 11.937125003896654}, "total_wall_ms": 7111.109748948365, "ttft_ms": 12.062499998137355, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 38, "retention_rate": 0.5846153846153846, "retention_by_kind": {"person": 0.8333333333333334, "date": 0.7, "amount": 0.38461538461538464, "status": 1.0, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1179}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 38, "retention_rate": 0.5846153846153846, "retention_by_kind": {"person": 0.8333333333333334, "date": 0.7, "amount": 0.38461538461538464, "status": 1.0, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1179}}}, "reference_key_id": "keys/synthetic/syn_03021_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "f6342dcd02ebd80508f967d8bfed98ca4bc1afeea750629dadac4b492752e270", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "225aa0bdffe8ab7c68a1bc2d", "attempt_id": "1284cd7b8c1e", "run_group_id": "E4_core:A4|repairable|qwen3b:h1", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:18:04.906734Z", "task_family": "synthetic", "document_id": "syn_03021_2000", "document_hash": "3ac205b464552356699a4d2b0a7bf62497ead36e5bca73982e11ae5ee3810731", "question_id": "q13_unanswerable", "split": "test", "cluster_id": "syn_03021_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 1, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1914, "query_tokens": 605, "message_tokens": 512, "message_utf8_bytes": 1179, "cache_tensor_bytes": null, "serialized_payload_bytes": 1179, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 6860.111457994208, "hops": {"1": 6860.111457994208}, "receiver_prefill_ms": 226.39533400069922, "ttft_ms": 12.255124980583787, "decode_ms": 11.995458044111729}, "total_wall_ms": 7111.378874978982, "ttft_ms": 12.255124980583787, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 38, "retention_rate": 0.5846153846153846, "retention_by_kind": {"person": 0.8333333333333334, "date": 0.7, "amount": 0.38461538461538464, "status": 1.0, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1179}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 38, "retention_rate": 0.5846153846153846, "retention_by_kind": {"person": 0.8333333333333334, "date": 0.7, "amount": 0.38461538461538464, "status": 1.0, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1179}}}, "reference_key_id": "keys/synthetic/syn_03021_2000.json", "task_scores": {"correct": 1.0, "abstained": 1.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "f6342dcd02ebd80508f967d8bfed98ca4bc1afeea750629dadac4b492752e270", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "c4f55ba6563dde2e253a68fa", "attempt_id": "74385a148ddf", "run_group_id": "E4_core:A4|repairable|qwen3b:h1", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:18:05.172487Z", "task_family": "synthetic", "document_id": "syn_03021_2000", "document_hash": "3ac205b464552356699a4d2b0a7bf62497ead36e5bca73982e11ae5ee3810731", "question_id": "q14_unanswerable", "split": "test", "cluster_id": "syn_03021_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 1, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1914, "query_tokens": 607, "message_tokens": 512, "message_utf8_bytes": 1179, "cache_tensor_bytes": null, "serialized_payload_bytes": 1179, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 6860.111457994208, "hops": {"1": 6860.111457994208}, "receiver_prefill_ms": 226.69262497220188, "ttft_ms": 12.230082997120917, "decode_ms": 12.032125028781593}, "total_wall_ms": 7111.5701659582555, "ttft_ms": 12.230082997120917, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 38, "retention_rate": 0.5846153846153846, "retention_by_kind": {"person": 0.8333333333333334, "date": 0.7, "amount": 0.38461538461538464, "status": 1.0, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1179}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 38, "retention_rate": 0.5846153846153846, "retention_by_kind": {"person": 0.8333333333333334, "date": 0.7, "amount": 0.38461538461538464, "status": 1.0, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1179}}}, "reference_key_id": "keys/synthetic/syn_03021_2000.json", "task_scores": {"correct": 1.0, "abstained": 1.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "f6342dcd02ebd80508f967d8bfed98ca4bc1afeea750629dadac4b492752e270", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "d62a89297f31d2f14e52f08a", "attempt_id": "ec06dd215e7b", "run_group_id": "E4_core:A4|repairable|qwen3b:h2", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:18:11.742463Z", "task_family": "synthetic", "document_id": "syn_03021_2000", "document_hash": "3ac205b464552356699a4d2b0a7bf62497ead36e5bca73982e11ae5ee3810731", "question_id": "q00_single_fact", "split": "test", "cluster_id": "syn_03021_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 2, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1914, "query_tokens": 577, "message_tokens": 497, "message_utf8_bytes": 1152, "cache_tensor_bytes": null, "serialized_payload_bytes": 1152, "side_information_bytes": 0, "fetched_bytes": 746, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 13161.659458070062, "hops": {"1": 6860.111457994208, "2": 6301.548000075854}, "receiver_prefill_ms": 209.32583301328123, "ttft_ms": 11.933750007301569, "decode_ms": 11.934750014916062}, "total_wall_ms": 13395.379541092552, "ttft_ms": 11.933750007301569, "input_usage": {"lookups": 4}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 36, "retention_rate": 0.5538461538461539, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.6, "amount": 0.38461538461538464, "status": 1.0, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1152}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 38, "retention_rate": 0.5846153846153846, "retention_by_kind": {"person": 0.8333333333333334, "date": 0.7, "amount": 0.38461538461538464, "status": 1.0, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1179}, "2": {"facts_total": 65, "facts_retained": 36, "retention_rate": 0.5538461538461539, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.6, "amount": 0.38461538461538464, "status": 1.0, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1152}}}, "reference_key_id": "keys/synthetic/syn_03021_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "29d19ef4cb16ed19e33980db8a0baa06391ee0314866e83c8ddf69b297e56071", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "7d7b6fdd2cccadcb17bc0a68", "attempt_id": "b476270061c4", "run_group_id": "E4_core:A4|repairable|qwen3b:h2", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:18:11.991795Z", "task_family": "synthetic", "document_id": "syn_03021_2000", "document_hash": "3ac205b464552356699a4d2b0a7bf62497ead36e5bca73982e11ae5ee3810731", "question_id": "q01_single_fact", "split": "test", "cluster_id": "syn_03021_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 2, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1914, "query_tokens": 583, "message_tokens": 497, "message_utf8_bytes": 1152, "cache_tensor_bytes": null, "serialized_payload_bytes": 1152, "side_information_bytes": 0, "fetched_bytes": 746, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 13161.659458070062, "hops": {"1": 6860.111457994208, "2": 6301.548000075854}, "receiver_prefill_ms": 227.2348329424858, "ttft_ms": 11.93300005979836, "decode_ms": 130.301000084728}, "total_wall_ms": 13531.631125020795, "ttft_ms": 11.93300005979836, "input_usage": {"lookups": 4}, "output_usage": {"output_tokens": 11}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "Lot 536-R, Senneterre", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 36, "retention_rate": 0.5538461538461539, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.6, "amount": 0.38461538461538464, "status": 1.0, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1152}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 38, "retention_rate": 0.5846153846153846, "retention_by_kind": {"person": 0.8333333333333334, "date": 0.7, "amount": 0.38461538461538464, "status": 1.0, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1179}, "2": {"facts_total": 65, "facts_retained": 36, "retention_rate": 0.5538461538461539, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.6, "amount": 0.38461538461538464, "status": 1.0, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1152}}}, "reference_key_id": "keys/synthetic/syn_03021_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 0.8571428571428571}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "29d19ef4cb16ed19e33980db8a0baa06391ee0314866e83c8ddf69b297e56071", "output_hash": "5e85af24aac0f1904d70a2fcee84f1b6f6912cc7450298af1f173002fd8e59df"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "e1f38e6b02f18b205ce94701", "attempt_id": "15f50fe3a956", "run_group_id": "E4_core:A4|repairable|qwen3b:h2", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:18:12.376929Z", "task_family": "synthetic", "document_id": "syn_03021_2000", "document_hash": "3ac205b464552356699a4d2b0a7bf62497ead36e5bca73982e11ae5ee3810731", "question_id": "q02_exact_copy", "split": "test", "cluster_id": "syn_03021_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 2, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1914, "query_tokens": 583, "message_tokens": 497, "message_utf8_bytes": 1152, "cache_tensor_bytes": null, "serialized_payload_bytes": 1152, "side_information_bytes": 0, "fetched_bytes": 746, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 13161.659458070062, "hops": {"1": 6860.111457994208, "2": 6301.548000075854}, "receiver_prefill_ms": 226.3869169401005, "ttft_ms": 11.981499963440001, "decode_ms": 83.10983295086771}, "total_wall_ms": 13483.626958099194, "ttft_ms": 11.981499963440001, "input_usage": {"lookups": 4}, "output_usage": {"output_tokens": 7}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "88.7751", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 36, "retention_rate": 0.5538461538461539, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.6, "amount": 0.38461538461538464, "status": 1.0, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1152}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 38, "retention_rate": 0.5846153846153846, "retention_by_kind": {"person": 0.8333333333333334, "date": 0.7, "amount": 0.38461538461538464, "status": 1.0, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1179}, "2": {"facts_total": 65, "facts_retained": 36, "retention_rate": 0.5538461538461539, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.6, "amount": 0.38461538461538464, "status": 1.0, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1152}}}, "reference_key_id": "keys/synthetic/syn_03021_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "29d19ef4cb16ed19e33980db8a0baa06391ee0314866e83c8ddf69b297e56071", "output_hash": "1e01e62c69f9479b681e50e59e8565c6583f5cdb803446a05ed5cb13118acdf9"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "cb03c0eeb92823b5de9ce689", "attempt_id": "e193f3d41d46", "run_group_id": "E4_core:A4|repairable|qwen3b:h2", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:18:12.713802Z", "task_family": "synthetic", "document_id": "syn_03021_2000", "document_hash": "3ac205b464552356699a4d2b0a7bf62497ead36e5bca73982e11ae5ee3810731", "question_id": "q03_exact_copy", "split": "test", "cluster_id": "syn_03021_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 2, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1914, "query_tokens": 580, "message_tokens": 497, "message_utf8_bytes": 1152, "cache_tensor_bytes": null, "serialized_payload_bytes": 1152, "side_information_bytes": 0, "fetched_bytes": 746, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 13161.659458070062, "hops": {"1": 6860.111457994208, "2": 6301.548000075854}, "receiver_prefill_ms": 226.5225830487907, "ttft_ms": 11.911417008377612, "decode_ms": 11.882040998898447}, "total_wall_ms": 13412.46295813471, "ttft_ms": 11.911417008377612, "input_usage": {"lookups": 4}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 36, "retention_rate": 0.5538461538461539, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.6, "amount": 0.38461538461538464, "status": 1.0, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1152}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 38, "retention_rate": 0.5846153846153846, "retention_by_kind": {"person": 0.8333333333333334, "date": 0.7, "amount": 0.38461538461538464, "status": 1.0, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1179}, "2": {"facts_total": 65, "facts_retained": 36, "retention_rate": 0.5538461538461539, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.6, "amount": 0.38461538461538464, "status": 1.0, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1152}}}, "reference_key_id": "keys/synthetic/syn_03021_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "29d19ef4cb16ed19e33980db8a0baa06391ee0314866e83c8ddf69b297e56071", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "d68ae32727b36042d2ee7c59", "attempt_id": "61f45f0a2116", "run_group_id": "E4_core:A4|repairable|qwen3b:h2", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:18:12.979220Z", "task_family": "synthetic", "document_id": "syn_03021_2000", "document_hash": "3ac205b464552356699a4d2b0a7bf62497ead36e5bca73982e11ae5ee3810731", "question_id": "q04_multi_hop", "split": "test", "cluster_id": "syn_03021_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 2, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1914, "query_tokens": 587, "message_tokens": 497, "message_utf8_bytes": 1152, "cache_tensor_bytes": null, "serialized_payload_bytes": 1152, "side_information_bytes": 0, "fetched_bytes": 746, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 13161.659458070062, "hops": {"1": 6860.111457994208, "2": 6301.548000075854}, "receiver_prefill_ms": 226.67000000365078, "ttft_ms": 12.13691697921604, "decode_ms": 11.901833000592887}, "total_wall_ms": 13412.861917051487, "ttft_ms": 12.13691697921604, "input_usage": {"lookups": 4}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 36, "retention_rate": 0.5538461538461539, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.6, "amount": 0.38461538461538464, "status": 1.0, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1152}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 38, "retention_rate": 0.5846153846153846, "retention_by_kind": {"person": 0.8333333333333334, "date": 0.7, "amount": 0.38461538461538464, "status": 1.0, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1179}, "2": {"facts_total": 65, "facts_retained": 36, "retention_rate": 0.5538461538461539, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.6, "amount": 0.38461538461538464, "status": 1.0, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1152}}}, "reference_key_id": "keys/synthetic/syn_03021_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "29d19ef4cb16ed19e33980db8a0baa06391ee0314866e83c8ddf69b297e56071", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "569e7bf5a4ec8f5c44cabea3", "attempt_id": "4d36945963de", "run_group_id": "E4_core:A4|repairable|qwen3b:h2", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:18:13.245674Z", "task_family": "synthetic", "document_id": "syn_03021_2000", "document_hash": "3ac205b464552356699a4d2b0a7bf62497ead36e5bca73982e11ae5ee3810731", "question_id": "q05_multi_hop", "split": "test", "cluster_id": "syn_03021_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 2, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1914, "query_tokens": 588, "message_tokens": 497, "message_utf8_bytes": 1152, "cache_tensor_bytes": null, "serialized_payload_bytes": 1152, "side_information_bytes": 0, "fetched_bytes": 746, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 13161.659458070062, "hops": {"1": 6860.111457994208, "2": 6301.548000075854}, "receiver_prefill_ms": 226.48012498393655, "ttft_ms": 11.998458998277783, "decode_ms": 131.08833401929587}, "total_wall_ms": 13531.844624085352, "ttft_ms": 11.998458998277783, "input_usage": {"lookups": 4}, "output_usage": {"output_tokens": 11}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "BQ-59Z-6753", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 36, "retention_rate": 0.5538461538461539, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.6, "amount": 0.38461538461538464, "status": 1.0, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1152}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 38, "retention_rate": 0.5846153846153846, "retention_by_kind": {"person": 0.8333333333333334, "date": 0.7, "amount": 0.38461538461538464, "status": 1.0, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1179}, "2": {"facts_total": 65, "facts_retained": 36, "retention_rate": 0.5538461538461539, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.6, "amount": 0.38461538461538464, "status": 1.0, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1152}}}, "reference_key_id": "keys/synthetic/syn_03021_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "29d19ef4cb16ed19e33980db8a0baa06391ee0314866e83c8ddf69b297e56071", "output_hash": "db250f072c856fe7ddf4bbbd16e20c71b8bad1eef5e4480acd0dac30f8ffe9cd"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "5439b7ea2f5d286c2f17ad45", "attempt_id": "7afd6978bee8", "run_group_id": "E4_core:A4|repairable|qwen3b:h2", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:18:13.631115Z", "task_family": "synthetic", "document_id": "syn_03021_2000", "document_hash": "3ac205b464552356699a4d2b0a7bf62497ead36e5bca73982e11ae5ee3810731", "question_id": "q06_negation", "split": "test", "cluster_id": "syn_03021_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 2, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1914, "query_tokens": 578, "message_tokens": 497, "message_utf8_bytes": 1152, "cache_tensor_bytes": null, "serialized_payload_bytes": 1152, "side_information_bytes": 0, "fetched_bytes": 746, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 13161.659458070062, "hops": {"1": 6860.111457994208, "2": 6301.548000075854}, "receiver_prefill_ms": 225.67512502428144, "ttft_ms": 12.151958071626723, "decode_ms": 11.959291994571686}, "total_wall_ms": 13412.100958172232, "ttft_ms": 12.151958071626723, "input_usage": {"lookups": 4}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "No", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 36, "retention_rate": 0.5538461538461539, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.6, "amount": 0.38461538461538464, "status": 1.0, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1152}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 38, "retention_rate": 0.5846153846153846, "retention_by_kind": {"person": 0.8333333333333334, "date": 0.7, "amount": 0.38461538461538464, "status": 1.0, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1179}, "2": {"facts_total": 65, "facts_retained": 36, "retention_rate": 0.5538461538461539, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.6, "amount": 0.38461538461538464, "status": 1.0, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1152}}}, "reference_key_id": "keys/synthetic/syn_03021_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "29d19ef4cb16ed19e33980db8a0baa06391ee0314866e83c8ddf69b297e56071", "output_hash": "1ea442a134b2a184bd5d40104401f2a37fbc09ccf3f4bc9da161c6099be3691d"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "4ae060020ba2a575f6a79448", "attempt_id": "06ba95a72aba", "run_group_id": "E4_core:A4|repairable|qwen3b:h2", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:18:13.896877Z", "task_family": "synthetic", "document_id": "syn_03021_2000", "document_hash": "3ac205b464552356699a4d2b0a7bf62497ead36e5bca73982e11ae5ee3810731", "question_id": "q07_negation", "split": "test", "cluster_id": "syn_03021_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 2, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1914, "query_tokens": 579, "message_tokens": 497, "message_utf8_bytes": 1152, "cache_tensor_bytes": null, "serialized_payload_bytes": 1152, "side_information_bytes": 0, "fetched_bytes": 746, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 13161.659458070062, "hops": {"1": 6860.111457994208, "2": 6301.548000075854}, "receiver_prefill_ms": 225.89400003198534, "ttft_ms": 12.093709083274007, "decode_ms": 11.991459061391652}, "total_wall_ms": 13412.250958150253, "ttft_ms": 12.093709083274007, "input_usage": {"lookups": 4}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "yes", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 36, "retention_rate": 0.5538461538461539, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.6, "amount": 0.38461538461538464, "status": 1.0, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1152}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 38, "retention_rate": 0.5846153846153846, "retention_by_kind": {"person": 0.8333333333333334, "date": 0.7, "amount": 0.38461538461538464, "status": 1.0, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1179}, "2": {"facts_total": 65, "facts_retained": 36, "retention_rate": 0.5538461538461539, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.6, "amount": 0.38461538461538464, "status": 1.0, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1152}}}, "reference_key_id": "keys/synthetic/syn_03021_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "29d19ef4cb16ed19e33980db8a0baa06391ee0314866e83c8ddf69b297e56071", "output_hash": "8a798890fe93817163b10b5f7bd2ca4d25d84c52739a645a889c173eee7d9d3d"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "555fcc2c0601b06b6ea40726", "attempt_id": "c21c08f7425e", "run_group_id": "E4_core:A4|repairable|qwen3b:h2", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:18:14.163548Z", "task_family": "synthetic", "document_id": "syn_03021_2000", "document_hash": "3ac205b464552356699a4d2b0a7bf62497ead36e5bca73982e11ae5ee3810731", "question_id": "q08_revision", "split": "test", "cluster_id": "syn_03021_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 2, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1914, "query_tokens": 595, "message_tokens": 497, "message_utf8_bytes": 1152, "cache_tensor_bytes": null, "serialized_payload_bytes": 1152, "side_information_bytes": 0, "fetched_bytes": 746, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 13161.659458070062, "hops": {"1": 6860.111457994208, "2": 6301.548000075854}, "receiver_prefill_ms": 226.92554094828665, "ttft_ms": 11.92904100753367, "decode_ms": 11.978833004832268}, "total_wall_ms": 13412.984083057381, "ttft_ms": 11.92904100753367, "input_usage": {"lookups": 4}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 36, "retention_rate": 0.5538461538461539, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.6, "amount": 0.38461538461538464, "status": 1.0, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1152}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 38, "retention_rate": 0.5846153846153846, "retention_by_kind": {"person": 0.8333333333333334, "date": 0.7, "amount": 0.38461538461538464, "status": 1.0, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1179}, "2": {"facts_total": 65, "facts_retained": 36, "retention_rate": 0.5538461538461539, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.6, "amount": 0.38461538461538464, "status": 1.0, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1152}}}, "reference_key_id": "keys/synthetic/syn_03021_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "29d19ef4cb16ed19e33980db8a0baa06391ee0314866e83c8ddf69b297e56071", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "ea6d7b1a40375117ca4a8985", "attempt_id": "dfd858c8d46d", "run_group_id": "E4_core:A4|repairable|qwen3b:h2", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:18:14.430204Z", "task_family": "synthetic", "document_id": "syn_03021_2000", "document_hash": "3ac205b464552356699a4d2b0a7bf62497ead36e5bca73982e11ae5ee3810731", "question_id": "q09_revision", "split": "test", "cluster_id": "syn_03021_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 2, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1914, "query_tokens": 586, "message_tokens": 497, "message_utf8_bytes": 1152, "cache_tensor_bytes": null, "serialized_payload_bytes": 1152, "side_information_bytes": 0, "fetched_bytes": 746, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 13161.659458070062, "hops": {"1": 6860.111457994208, "2": 6301.548000075854}, "receiver_prefill_ms": 226.34762490633875, "ttft_ms": 11.899958946742117, "decode_ms": 11.99108303990215}, "total_wall_ms": 13412.35208301805, "ttft_ms": 11.899958946742117, "input_usage": {"lookups": 4}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 36, "retention_rate": 0.5538461538461539, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.6, "amount": 0.38461538461538464, "status": 1.0, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1152}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 38, "retention_rate": 0.5846153846153846, "retention_by_kind": {"person": 0.8333333333333334, "date": 0.7, "amount": 0.38461538461538464, "status": 1.0, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1179}, "2": {"facts_total": 65, "facts_retained": 36, "retention_rate": 0.5538461538461539, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.6, "amount": 0.38461538461538464, "status": 1.0, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1152}}}, "reference_key_id": "keys/synthetic/syn_03021_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "29d19ef4cb16ed19e33980db8a0baa06391ee0314866e83c8ddf69b297e56071", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "777c33c64131b7f6665fa222", "attempt_id": "f9d9d7fa3a7f", "run_group_id": "E4_core:A4|repairable|qwen3b:h2", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:18:14.695586Z", "task_family": "synthetic", "document_id": "syn_03021_2000", "document_hash": "3ac205b464552356699a4d2b0a7bf62497ead36e5bca73982e11ae5ee3810731", "question_id": "q10_temporal", "split": "test", "cluster_id": "syn_03021_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 2, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1914, "query_tokens": 594, "message_tokens": 497, "message_utf8_bytes": 1152, "cache_tensor_bytes": null, "serialized_payload_bytes": 1152, "side_information_bytes": 0, "fetched_bytes": 746, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 13161.659458070062, "hops": {"1": 6860.111457994208, "2": 6301.548000075854}, "receiver_prefill_ms": 226.97137494105846, "ttft_ms": 11.974541936069727, "decode_ms": 59.44470898248255}, "total_wall_ms": 13460.535542108119, "ttft_ms": 11.974541936069727, "input_usage": {"lookups": 4}, "output_usage": {"output_tokens": 5}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "Kovar-9", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 36, "retention_rate": 0.5538461538461539, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.6, "amount": 0.38461538461538464, "status": 1.0, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1152}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 38, "retention_rate": 0.5846153846153846, "retention_by_kind": {"person": 0.8333333333333334, "date": 0.7, "amount": 0.38461538461538464, "status": 1.0, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1179}, "2": {"facts_total": 65, "facts_retained": 36, "retention_rate": 0.5538461538461539, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.6, "amount": 0.38461538461538464, "status": 1.0, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1152}}}, "reference_key_id": "keys/synthetic/syn_03021_2000.json", "task_scores": {"correct": 0.0, "abstained": 0.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "29d19ef4cb16ed19e33980db8a0baa06391ee0314866e83c8ddf69b297e56071", "output_hash": "c969b5f7547945067f5b07e9f822d0043126326f979d950a8b1b045057fbb818"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "9808688b5da67796aa80fe81", "attempt_id": "1b41db8d342e", "run_group_id": "E4_core:A4|repairable|qwen3b:h2", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:18:15.009317Z", "task_family": "synthetic", "document_id": "syn_03021_2000", "document_hash": "3ac205b464552356699a4d2b0a7bf62497ead36e5bca73982e11ae5ee3810731", "question_id": "q11_aggregation", "split": "test", "cluster_id": "syn_03021_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 2, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1914, "query_tokens": 590, "message_tokens": 497, "message_utf8_bytes": 1152, "cache_tensor_bytes": null, "serialized_payload_bytes": 1152, "side_information_bytes": 0, "fetched_bytes": 746, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 13161.659458070062, "hops": {"1": 6860.111457994208, "2": 6301.548000075854}, "receiver_prefill_ms": 226.5684159938246, "ttft_ms": 11.967374943196774, "decode_ms": 94.82549992389977}, "total_wall_ms": 13495.493583148345, "ttft_ms": 11.967374943196774, "input_usage": {"lookups": 4}, "output_usage": {"output_tokens": 8}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "13024817", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 36, "retention_rate": 0.5538461538461539, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.6, "amount": 0.38461538461538464, "status": 1.0, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1152}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 38, "retention_rate": 0.5846153846153846, "retention_by_kind": {"person": 0.8333333333333334, "date": 0.7, "amount": 0.38461538461538464, "status": 1.0, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1179}, "2": {"facts_total": 65, "facts_retained": 36, "retention_rate": 0.5538461538461539, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.6, "amount": 0.38461538461538464, "status": 1.0, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1152}}}, "reference_key_id": "keys/synthetic/syn_03021_2000.json", "task_scores": {"correct": 0.0, "abstained": 0.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "29d19ef4cb16ed19e33980db8a0baa06391ee0314866e83c8ddf69b297e56071", "output_hash": "672c00df3bfd31225e9d03325d167c25800a9007589a430995bacbf8ad8c6f21"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "40c70859f27d25f58330e92b", "attempt_id": "73c2aa411d76", "run_group_id": "E4_core:A4|repairable|qwen3b:h2", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:18:15.358163Z", "task_family": "synthetic", "document_id": "syn_03021_2000", "document_hash": "3ac205b464552356699a4d2b0a7bf62497ead36e5bca73982e11ae5ee3810731", "question_id": "q12_unit_conversion", "split": "test", "cluster_id": "syn_03021_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 2, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1914, "query_tokens": 585, "message_tokens": 497, "message_utf8_bytes": 1152, "cache_tensor_bytes": null, "serialized_payload_bytes": 1152, "side_information_bytes": 0, "fetched_bytes": 746, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 13161.659458070062, "hops": {"1": 6860.111457994208, "2": 6301.548000075854}, "receiver_prefill_ms": 226.84412507805973, "ttft_ms": 11.883749975822866, "decode_ms": 11.935291928239167}, "total_wall_ms": 13412.811083020642, "ttft_ms": 11.883749975822866, "input_usage": {"lookups": 4}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 36, "retention_rate": 0.5538461538461539, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.6, "amount": 0.38461538461538464, "status": 1.0, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1152}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 38, "retention_rate": 0.5846153846153846, "retention_by_kind": {"person": 0.8333333333333334, "date": 0.7, "amount": 0.38461538461538464, "status": 1.0, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1179}, "2": {"facts_total": 65, "facts_retained": 36, "retention_rate": 0.5538461538461539, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.6, "amount": 0.38461538461538464, "status": 1.0, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1152}}}, "reference_key_id": "keys/synthetic/syn_03021_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "29d19ef4cb16ed19e33980db8a0baa06391ee0314866e83c8ddf69b297e56071", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "f47aaff34d48a1545ac238ec", "attempt_id": "ec1a0206b938", "run_group_id": "E4_core:A4|repairable|qwen3b:h2", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:18:15.625521Z", "task_family": "synthetic", "document_id": "syn_03021_2000", "document_hash": "3ac205b464552356699a4d2b0a7bf62497ead36e5bca73982e11ae5ee3810731", "question_id": "q13_unanswerable", "split": "test", "cluster_id": "syn_03021_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 2, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1914, "query_tokens": 590, "message_tokens": 497, "message_utf8_bytes": 1152, "cache_tensor_bytes": null, "serialized_payload_bytes": 1152, "side_information_bytes": 0, "fetched_bytes": 746, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 13161.659458070062, "hops": {"1": 6860.111457994208, "2": 6301.548000075854}, "receiver_prefill_ms": 226.9573329249397, "ttft_ms": 11.8819170165807, "decode_ms": 11.953916982747614}, "total_wall_ms": 13412.913375068456, "ttft_ms": 11.8819170165807, "input_usage": {"lookups": 4}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 36, "retention_rate": 0.5538461538461539, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.6, "amount": 0.38461538461538464, "status": 1.0, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1152}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 38, "retention_rate": 0.5846153846153846, "retention_by_kind": {"person": 0.8333333333333334, "date": 0.7, "amount": 0.38461538461538464, "status": 1.0, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1179}, "2": {"facts_total": 65, "facts_retained": 36, "retention_rate": 0.5538461538461539, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.6, "amount": 0.38461538461538464, "status": 1.0, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1152}}}, "reference_key_id": "keys/synthetic/syn_03021_2000.json", "task_scores": {"correct": 1.0, "abstained": 1.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "29d19ef4cb16ed19e33980db8a0baa06391ee0314866e83c8ddf69b297e56071", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "16b41f5139a790c73426beec", "attempt_id": "e0c079376663", "run_group_id": "E4_core:A4|repairable|qwen3b:h2", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:18:15.891792Z", "task_family": "synthetic", "document_id": "syn_03021_2000", "document_hash": "3ac205b464552356699a4d2b0a7bf62497ead36e5bca73982e11ae5ee3810731", "question_id": "q14_unanswerable", "split": "test", "cluster_id": "syn_03021_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 2, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1914, "query_tokens": 592, "message_tokens": 497, "message_utf8_bytes": 1152, "cache_tensor_bytes": null, "serialized_payload_bytes": 1152, "side_information_bytes": 0, "fetched_bytes": 746, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 13161.659458070062, "hops": {"1": 6860.111457994208, "2": 6301.548000075854}, "receiver_prefill_ms": 226.9875839119777, "ttft_ms": 11.907708016224205, "decode_ms": 11.941958917304873}, "total_wall_ms": 13412.956041051075, "ttft_ms": 11.907708016224205, "input_usage": {"lookups": 4}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 36, "retention_rate": 0.5538461538461539, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.6, "amount": 0.38461538461538464, "status": 1.0, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1152}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 38, "retention_rate": 0.5846153846153846, "retention_by_kind": {"person": 0.8333333333333334, "date": 0.7, "amount": 0.38461538461538464, "status": 1.0, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1179}, "2": {"facts_total": 65, "facts_retained": 36, "retention_rate": 0.5538461538461539, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.6, "amount": 0.38461538461538464, "status": 1.0, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1152}}}, "reference_key_id": "keys/synthetic/syn_03021_2000.json", "task_scores": {"correct": 1.0, "abstained": 1.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "29d19ef4cb16ed19e33980db8a0baa06391ee0314866e83c8ddf69b297e56071", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "47dbca824955d1ae3187b5b2", "attempt_id": "4b33b63e5bab", "run_group_id": "E4_core:A4|repairable|qwen3b:h4", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:18:28.732598Z", "task_family": "synthetic", "document_id": "syn_03021_2000", "document_hash": "3ac205b464552356699a4d2b0a7bf62497ead36e5bca73982e11ae5ee3810731", "question_id": "q00_single_fact", "split": "test", "cluster_id": "syn_03021_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 4, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1914, "query_tokens": 577, "message_tokens": 497, "message_utf8_bytes": 1152, "cache_tensor_bytes": null, "serialized_payload_bytes": 1152, "side_information_bytes": 0, "fetched_bytes": 2238, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 25733.930833172053, "hops": {"1": 6860.111457994208, "2": 6301.548000075854, "3": 6284.723916091025, "4": 6287.547459010966}, "receiver_prefill_ms": 208.80850008688867, "ttft_ms": 11.889124987646937, "decode_ms": 11.852832976728678}, "total_wall_ms": 25966.977458214387, "ttft_ms": 11.889124987646937, "input_usage": {"lookups": 12}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 36, "retention_rate": 0.5538461538461539, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.6, "amount": 0.38461538461538464, "status": 1.0, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1152}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 38, "retention_rate": 0.5846153846153846, "retention_by_kind": {"person": 0.8333333333333334, "date": 0.7, "amount": 0.38461538461538464, "status": 1.0, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1179}, "2": {"facts_total": 65, "facts_retained": 36, "retention_rate": 0.5538461538461539, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.6, "amount": 0.38461538461538464, "status": 1.0, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1152}, "3": {"facts_total": 65, "facts_retained": 36, "retention_rate": 0.5538461538461539, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.6, "amount": 0.38461538461538464, "status": 1.0, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1152}, "4": {"facts_total": 65, "facts_retained": 36, "retention_rate": 0.5538461538461539, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.6, "amount": 0.38461538461538464, "status": 1.0, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1152}}}, "reference_key_id": "keys/synthetic/syn_03021_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "29d19ef4cb16ed19e33980db8a0baa06391ee0314866e83c8ddf69b297e56071", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "5fbbe72094418704b950ce2c", "attempt_id": "2d751a26a233", "run_group_id": "E4_core:A4|repairable|qwen3b:h4", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:18:28.980959Z", "task_family": "synthetic", "document_id": "syn_03021_2000", "document_hash": "3ac205b464552356699a4d2b0a7bf62497ead36e5bca73982e11ae5ee3810731", "question_id": "q01_single_fact", "split": "test", "cluster_id": "syn_03021_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 4, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1914, "query_tokens": 583, "message_tokens": 497, "message_utf8_bytes": 1152, "cache_tensor_bytes": null, "serialized_payload_bytes": 1152, "side_information_bytes": 0, "fetched_bytes": 2238, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 25733.930833172053, "hops": {"1": 6860.111457994208, "2": 6301.548000075854, "3": 6284.723916091025, "4": 6287.547459010966}, "receiver_prefill_ms": 226.34654201101512, "ttft_ms": 11.926042032428086, "decode_ms": 130.3043330553919}, "total_wall_ms": 26102.99900022801, "ttft_ms": 11.926042032428086, "input_usage": {"lookups": 12}, "output_usage": {"output_tokens": 11}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "Lot 536-R, Senneterre", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 36, "retention_rate": 0.5538461538461539, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.6, "amount": 0.38461538461538464, "status": 1.0, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1152}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 38, "retention_rate": 0.5846153846153846, "retention_by_kind": {"person": 0.8333333333333334, "date": 0.7, "amount": 0.38461538461538464, "status": 1.0, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1179}, "2": {"facts_total": 65, "facts_retained": 36, "retention_rate": 0.5538461538461539, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.6, "amount": 0.38461538461538464, "status": 1.0, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1152}, "3": {"facts_total": 65, "facts_retained": 36, "retention_rate": 0.5538461538461539, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.6, "amount": 0.38461538461538464, "status": 1.0, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1152}, "4": {"facts_total": 65, "facts_retained": 36, "retention_rate": 0.5538461538461539, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.6, "amount": 0.38461538461538464, "status": 1.0, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1152}}}, "reference_key_id": "keys/synthetic/syn_03021_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 0.8571428571428571}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "29d19ef4cb16ed19e33980db8a0baa06391ee0314866e83c8ddf69b297e56071", "output_hash": "5e85af24aac0f1904d70a2fcee84f1b6f6912cc7450298af1f173002fd8e59df"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "636b1db17d7394fa10e4db7b", "attempt_id": "676086342928", "run_group_id": "E4_core:A4|repairable|qwen3b:h4", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:18:29.365158Z", "task_family": "synthetic", "document_id": "syn_03021_2000", "document_hash": "3ac205b464552356699a4d2b0a7bf62497ead36e5bca73982e11ae5ee3810731", "question_id": "q02_exact_copy", "split": "test", "cluster_id": "syn_03021_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 4, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1914, "query_tokens": 583, "message_tokens": 497, "message_utf8_bytes": 1152, "cache_tensor_bytes": null, "serialized_payload_bytes": 1152, "side_information_bytes": 0, "fetched_bytes": 2238, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 25733.930833172053, "hops": {"1": 6860.111457994208, "2": 6301.548000075854, "3": 6284.723916091025, "4": 6287.547459010966}, "receiver_prefill_ms": 226.45491606090218, "ttft_ms": 12.106165988370776, "decode_ms": 82.95112499035895}, "total_wall_ms": 26055.92525016982, "ttft_ms": 12.106165988370776, "input_usage": {"lookups": 12}, "output_usage": {"output_tokens": 7}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "88.7751", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 36, "retention_rate": 0.5538461538461539, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.6, "amount": 0.38461538461538464, "status": 1.0, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1152}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 38, "retention_rate": 0.5846153846153846, "retention_by_kind": {"person": 0.8333333333333334, "date": 0.7, "amount": 0.38461538461538464, "status": 1.0, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1179}, "2": {"facts_total": 65, "facts_retained": 36, "retention_rate": 0.5538461538461539, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.6, "amount": 0.38461538461538464, "status": 1.0, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1152}, "3": {"facts_total": 65, "facts_retained": 36, "retention_rate": 0.5538461538461539, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.6, "amount": 0.38461538461538464, "status": 1.0, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1152}, "4": {"facts_total": 65, "facts_retained": 36, "retention_rate": 0.5538461538461539, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.6, "amount": 0.38461538461538464, "status": 1.0, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1152}}}, "reference_key_id": "keys/synthetic/syn_03021_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "29d19ef4cb16ed19e33980db8a0baa06391ee0314866e83c8ddf69b297e56071", "output_hash": "1e01e62c69f9479b681e50e59e8565c6583f5cdb803446a05ed5cb13118acdf9"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "9752381593d9c22229f9dad1", "attempt_id": "634285066f0d", "run_group_id": "E4_core:A4|repairable|qwen3b:h4", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:18:29.701866Z", "task_family": "synthetic", "document_id": "syn_03021_2000", "document_hash": "3ac205b464552356699a4d2b0a7bf62497ead36e5bca73982e11ae5ee3810731", "question_id": "q03_exact_copy", "split": "test", "cluster_id": "syn_03021_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 4, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1914, "query_tokens": 580, "message_tokens": 497, "message_utf8_bytes": 1152, "cache_tensor_bytes": null, "serialized_payload_bytes": 1152, "side_information_bytes": 0, "fetched_bytes": 2238, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 25733.930833172053, "hops": {"1": 6860.111457994208, "2": 6301.548000075854, "3": 6284.723916091025, "4": 6287.547459010966}, "receiver_prefill_ms": 226.81824991013855, "ttft_ms": 12.0340830180794, "decode_ms": 11.84329204261303}, "total_wall_ms": 25985.095583135262, "ttft_ms": 12.0340830180794, "input_usage": {"lookups": 12}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 36, "retention_rate": 0.5538461538461539, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.6, "amount": 0.38461538461538464, "status": 1.0, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1152}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 38, "retention_rate": 0.5846153846153846, "retention_by_kind": {"person": 0.8333333333333334, "date": 0.7, "amount": 0.38461538461538464, "status": 1.0, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1179}, "2": {"facts_total": 65, "facts_retained": 36, "retention_rate": 0.5538461538461539, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.6, "amount": 0.38461538461538464, "status": 1.0, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1152}, "3": {"facts_total": 65, "facts_retained": 36, "retention_rate": 0.5538461538461539, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.6, "amount": 0.38461538461538464, "status": 1.0, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1152}, "4": {"facts_total": 65, "facts_retained": 36, "retention_rate": 0.5538461538461539, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.6, "amount": 0.38461538461538464, "status": 1.0, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1152}}}, "reference_key_id": "keys/synthetic/syn_03021_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "29d19ef4cb16ed19e33980db8a0baa06391ee0314866e83c8ddf69b297e56071", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "0376c88c828bef134fbed9da", "attempt_id": "a2d2f553b9cf", "run_group_id": "E4_core:A4|repairable|qwen3b:h4", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:18:29.967992Z", "task_family": "synthetic", "document_id": "syn_03021_2000", "document_hash": "3ac205b464552356699a4d2b0a7bf62497ead36e5bca73982e11ae5ee3810731", "question_id": "q04_multi_hop", "split": "test", "cluster_id": "syn_03021_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 4, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1914, "query_tokens": 587, "message_tokens": 497, "message_utf8_bytes": 1152, "cache_tensor_bytes": null, "serialized_payload_bytes": 1152, "side_information_bytes": 0, "fetched_bytes": 2238, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 25733.930833172053, "hops": {"1": 6860.111457994208, "2": 6301.548000075854, "3": 6284.723916091025, "4": 6287.547459010966}, "receiver_prefill_ms": 226.58925002906471, "ttft_ms": 11.893625021912158, "decode_ms": 12.020291993394494}, "total_wall_ms": 25984.889708226547, "ttft_ms": 11.893625021912158, "input_usage": {"lookups": 12}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 36, "retention_rate": 0.5538461538461539, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.6, "amount": 0.38461538461538464, "status": 1.0, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1152}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 38, "retention_rate": 0.5846153846153846, "retention_by_kind": {"person": 0.8333333333333334, "date": 0.7, "amount": 0.38461538461538464, "status": 1.0, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1179}, "2": {"facts_total": 65, "facts_retained": 36, "retention_rate": 0.5538461538461539, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.6, "amount": 0.38461538461538464, "status": 1.0, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1152}, "3": {"facts_total": 65, "facts_retained": 36, "retention_rate": 0.5538461538461539, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.6, "amount": 0.38461538461538464, "status": 1.0, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1152}, "4": {"facts_total": 65, "facts_retained": 36, "retention_rate": 0.5538461538461539, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.6, "amount": 0.38461538461538464, "status": 1.0, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1152}}}, "reference_key_id": "keys/synthetic/syn_03021_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "29d19ef4cb16ed19e33980db8a0baa06391ee0314866e83c8ddf69b297e56071", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "f3a2ad21e7faa465c057bd52", "attempt_id": "56135811360c", "run_group_id": "E4_core:A4|repairable|qwen3b:h4", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:18:30.233621Z", "task_family": "synthetic", "document_id": "syn_03021_2000", "document_hash": "3ac205b464552356699a4d2b0a7bf62497ead36e5bca73982e11ae5ee3810731", "question_id": "q05_multi_hop", "split": "test", "cluster_id": "syn_03021_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 4, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1914, "query_tokens": 588, "message_tokens": 497, "message_utf8_bytes": 1152, "cache_tensor_bytes": null, "serialized_payload_bytes": 1152, "side_information_bytes": 0, "fetched_bytes": 2238, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 25733.930833172053, "hops": {"1": 6860.111457994208, "2": 6301.548000075854, "3": 6284.723916091025, "4": 6287.547459010966}, "receiver_prefill_ms": 226.74441698472947, "ttft_ms": 11.915334034711123, "decode_ms": 130.16837497707456}, "total_wall_ms": 26103.23541611433, "ttft_ms": 11.915334034711123, "input_usage": {"lookups": 12}, "output_usage": {"output_tokens": 11}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "BQ-59Z-6753", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 36, "retention_rate": 0.5538461538461539, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.6, "amount": 0.38461538461538464, "status": 1.0, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1152}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 38, "retention_rate": 0.5846153846153846, "retention_by_kind": {"person": 0.8333333333333334, "date": 0.7, "amount": 0.38461538461538464, "status": 1.0, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1179}, "2": {"facts_total": 65, "facts_retained": 36, "retention_rate": 0.5538461538461539, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.6, "amount": 0.38461538461538464, "status": 1.0, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1152}, "3": {"facts_total": 65, "facts_retained": 36, "retention_rate": 0.5538461538461539, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.6, "amount": 0.38461538461538464, "status": 1.0, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1152}, "4": {"facts_total": 65, "facts_retained": 36, "retention_rate": 0.5538461538461539, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.6, "amount": 0.38461538461538464, "status": 1.0, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1152}}}, "reference_key_id": "keys/synthetic/syn_03021_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "29d19ef4cb16ed19e33980db8a0baa06391ee0314866e83c8ddf69b297e56071", "output_hash": "db250f072c856fe7ddf4bbbd16e20c71b8bad1eef5e4480acd0dac30f8ffe9cd"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "a7e0415503bc1b18c154e71f", "attempt_id": "e2d867779f60", "run_group_id": "E4_core:A4|repairable|qwen3b:h4", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:18:30.617805Z", "task_family": "synthetic", "document_id": "syn_03021_2000", "document_hash": "3ac205b464552356699a4d2b0a7bf62497ead36e5bca73982e11ae5ee3810731", "question_id": "q06_negation", "split": "test", "cluster_id": "syn_03021_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 4, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1914, "query_tokens": 578, "message_tokens": 497, "message_utf8_bytes": 1152, "cache_tensor_bytes": null, "serialized_payload_bytes": 1152, "side_information_bytes": 0, "fetched_bytes": 2238, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 25733.930833172053, "hops": {"1": 6860.111457994208, "2": 6301.548000075854, "3": 6284.723916091025, "4": 6287.547459010966}, "receiver_prefill_ms": 226.10416694078594, "ttft_ms": 11.85033400543034, "decode_ms": 12.1004160027951}, "total_wall_ms": 25984.432583092712, "ttft_ms": 11.85033400543034, "input_usage": {"lookups": 12}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "No", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 36, "retention_rate": 0.5538461538461539, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.6, "amount": 0.38461538461538464, "status": 1.0, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1152}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 38, "retention_rate": 0.5846153846153846, "retention_by_kind": {"person": 0.8333333333333334, "date": 0.7, "amount": 0.38461538461538464, "status": 1.0, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1179}, "2": {"facts_total": 65, "facts_retained": 36, "retention_rate": 0.5538461538461539, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.6, "amount": 0.38461538461538464, "status": 1.0, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1152}, "3": {"facts_total": 65, "facts_retained": 36, "retention_rate": 0.5538461538461539, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.6, "amount": 0.38461538461538464, "status": 1.0, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1152}, "4": {"facts_total": 65, "facts_retained": 36, "retention_rate": 0.5538461538461539, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.6, "amount": 0.38461538461538464, "status": 1.0, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1152}}}, "reference_key_id": "keys/synthetic/syn_03021_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "29d19ef4cb16ed19e33980db8a0baa06391ee0314866e83c8ddf69b297e56071", "output_hash": "1ea442a134b2a184bd5d40104401f2a37fbc09ccf3f4bc9da161c6099be3691d"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "81cfa9d8f3ba1008aa6ebf45", "attempt_id": "2e2ab8e694c1", "run_group_id": "E4_core:A4|repairable|qwen3b:h4", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:18:30.883044Z", "task_family": "synthetic", "document_id": "syn_03021_2000", "document_hash": "3ac205b464552356699a4d2b0a7bf62497ead36e5bca73982e11ae5ee3810731", "question_id": "q07_negation", "split": "test", "cluster_id": "syn_03021_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 4, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1914, "query_tokens": 579, "message_tokens": 497, "message_utf8_bytes": 1152, "cache_tensor_bytes": null, "serialized_payload_bytes": 1152, "side_information_bytes": 0, "fetched_bytes": 2238, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 25733.930833172053, "hops": {"1": 6860.111457994208, "2": 6301.548000075854, "3": 6284.723916091025, "4": 6287.547459010966}, "receiver_prefill_ms": 226.06004192493856, "ttft_ms": 11.915459064766765, "decode_ms": 11.863666004501283}, "total_wall_ms": 25984.224875224754, "ttft_ms": 11.915459064766765, "input_usage": {"lookups": 12}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "yes", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 36, "retention_rate": 0.5538461538461539, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.6, "amount": 0.38461538461538464, "status": 1.0, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1152}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 38, "retention_rate": 0.5846153846153846, "retention_by_kind": {"person": 0.8333333333333334, "date": 0.7, "amount": 0.38461538461538464, "status": 1.0, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1179}, "2": {"facts_total": 65, "facts_retained": 36, "retention_rate": 0.5538461538461539, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.6, "amount": 0.38461538461538464, "status": 1.0, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1152}, "3": {"facts_total": 65, "facts_retained": 36, "retention_rate": 0.5538461538461539, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.6, "amount": 0.38461538461538464, "status": 1.0, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1152}, "4": {"facts_total": 65, "facts_retained": 36, "retention_rate": 0.5538461538461539, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.6, "amount": 0.38461538461538464, "status": 1.0, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1152}}}, "reference_key_id": "keys/synthetic/syn_03021_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "29d19ef4cb16ed19e33980db8a0baa06391ee0314866e83c8ddf69b297e56071", "output_hash": "8a798890fe93817163b10b5f7bd2ca4d25d84c52739a645a889c173eee7d9d3d"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "798d6eab719c01d6fa323a5d", "attempt_id": "cde957380797", "run_group_id": "E4_core:A4|repairable|qwen3b:h4", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:18:31.148078Z", "task_family": "synthetic", "document_id": "syn_03021_2000", "document_hash": "3ac205b464552356699a4d2b0a7bf62497ead36e5bca73982e11ae5ee3810731", "question_id": "q08_revision", "split": "test", "cluster_id": "syn_03021_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 4, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1914, "query_tokens": 595, "message_tokens": 497, "message_utf8_bytes": 1152, "cache_tensor_bytes": null, "serialized_payload_bytes": 1152, "side_information_bytes": 0, "fetched_bytes": 2238, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 25733.930833172053, "hops": {"1": 6860.111457994208, "2": 6301.548000075854, "3": 6284.723916091025, "4": 6287.547459010966}, "receiver_prefill_ms": 226.8334999680519, "ttft_ms": 11.853165924549103, "decode_ms": 11.917417054064572}, "total_wall_ms": 25985.012792167254, "ttft_ms": 11.853165924549103, "input_usage": {"lookups": 12}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 36, "retention_rate": 0.5538461538461539, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.6, "amount": 0.38461538461538464, "status": 1.0, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1152}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 38, "retention_rate": 0.5846153846153846, "retention_by_kind": {"person": 0.8333333333333334, "date": 0.7, "amount": 0.38461538461538464, "status": 1.0, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1179}, "2": {"facts_total": 65, "facts_retained": 36, "retention_rate": 0.5538461538461539, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.6, "amount": 0.38461538461538464, "status": 1.0, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1152}, "3": {"facts_total": 65, "facts_retained": 36, "retention_rate": 0.5538461538461539, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.6, "amount": 0.38461538461538464, "status": 1.0, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1152}, "4": {"facts_total": 65, "facts_retained": 36, "retention_rate": 0.5538461538461539, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.6, "amount": 0.38461538461538464, "status": 1.0, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1152}}}, "reference_key_id": "keys/synthetic/syn_03021_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "29d19ef4cb16ed19e33980db8a0baa06391ee0314866e83c8ddf69b297e56071", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "6f56a31eb6966de70cf5c503", "attempt_id": "c3315cc70b5f", "run_group_id": "E4_core:A4|repairable|qwen3b:h4", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:18:31.414421Z", "task_family": "synthetic", "document_id": "syn_03021_2000", "document_hash": "3ac205b464552356699a4d2b0a7bf62497ead36e5bca73982e11ae5ee3810731", "question_id": "q09_revision", "split": "test", "cluster_id": "syn_03021_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 4, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1914, "query_tokens": 586, "message_tokens": 497, "message_utf8_bytes": 1152, "cache_tensor_bytes": null, "serialized_payload_bytes": 1152, "side_information_bytes": 0, "fetched_bytes": 2238, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 25733.930833172053, "hops": {"1": 6860.111457994208, "2": 6301.548000075854, "3": 6284.723916091025, "4": 6287.547459010966}, "receiver_prefill_ms": 226.39208403415978, "ttft_ms": 11.858208919875324, "decode_ms": 11.979374918155372}, "total_wall_ms": 25984.62350014597, "ttft_ms": 11.858208919875324, "input_usage": {"lookups": 12}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 36, "retention_rate": 0.5538461538461539, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.6, "amount": 0.38461538461538464, "status": 1.0, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1152}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 38, "retention_rate": 0.5846153846153846, "retention_by_kind": {"person": 0.8333333333333334, "date": 0.7, "amount": 0.38461538461538464, "status": 1.0, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1179}, "2": {"facts_total": 65, "facts_retained": 36, "retention_rate": 0.5538461538461539, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.6, "amount": 0.38461538461538464, "status": 1.0, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1152}, "3": {"facts_total": 65, "facts_retained": 36, "retention_rate": 0.5538461538461539, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.6, "amount": 0.38461538461538464, "status": 1.0, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1152}, "4": {"facts_total": 65, "facts_retained": 36, "retention_rate": 0.5538461538461539, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.6, "amount": 0.38461538461538464, "status": 1.0, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1152}}}, "reference_key_id": "keys/synthetic/syn_03021_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "29d19ef4cb16ed19e33980db8a0baa06391ee0314866e83c8ddf69b297e56071", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "673baaca5ce68b663c4407c9", "attempt_id": "2d060ba1290a", "run_group_id": "E4_core:A4|repairable|qwen3b:h4", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:18:31.680363Z", "task_family": "synthetic", "document_id": "syn_03021_2000", "document_hash": "3ac205b464552356699a4d2b0a7bf62497ead36e5bca73982e11ae5ee3810731", "question_id": "q10_temporal", "split": "test", "cluster_id": "syn_03021_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 4, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1914, "query_tokens": 594, "message_tokens": 497, "message_utf8_bytes": 1152, "cache_tensor_bytes": null, "serialized_payload_bytes": 1152, "side_information_bytes": 0, "fetched_bytes": 2238, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 25733.930833172053, "hops": {"1": 6860.111457994208, "2": 6301.548000075854, "3": 6284.723916091025, "4": 6287.547459010966}, "receiver_prefill_ms": 226.7625000094995, "ttft_ms": 11.84979104436934, "decode_ms": 59.57145907450467}, "total_wall_ms": 26032.5781252468, "ttft_ms": 11.84979104436934, "input_usage": {"lookups": 12}, "output_usage": {"output_tokens": 5}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "Kovar-9", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 36, "retention_rate": 0.5538461538461539, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.6, "amount": 0.38461538461538464, "status": 1.0, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1152}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 38, "retention_rate": 0.5846153846153846, "retention_by_kind": {"person": 0.8333333333333334, "date": 0.7, "amount": 0.38461538461538464, "status": 1.0, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1179}, "2": {"facts_total": 65, "facts_retained": 36, "retention_rate": 0.5538461538461539, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.6, "amount": 0.38461538461538464, "status": 1.0, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1152}, "3": {"facts_total": 65, "facts_retained": 36, "retention_rate": 0.5538461538461539, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.6, "amount": 0.38461538461538464, "status": 1.0, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1152}, "4": {"facts_total": 65, "facts_retained": 36, "retention_rate": 0.5538461538461539, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.6, "amount": 0.38461538461538464, "status": 1.0, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1152}}}, "reference_key_id": "keys/synthetic/syn_03021_2000.json", "task_scores": {"correct": 0.0, "abstained": 0.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "29d19ef4cb16ed19e33980db8a0baa06391ee0314866e83c8ddf69b297e56071", "output_hash": "c969b5f7547945067f5b07e9f822d0043126326f979d950a8b1b045057fbb818"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "bbd43850f5370d90b0173997", "attempt_id": "e3d42e35d164", "run_group_id": "E4_core:A4|repairable|qwen3b:h4", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:18:31.993892Z", "task_family": "synthetic", "document_id": "syn_03021_2000", "document_hash": "3ac205b464552356699a4d2b0a7bf62497ead36e5bca73982e11ae5ee3810731", "question_id": "q11_aggregation", "split": "test", "cluster_id": "syn_03021_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 4, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1914, "query_tokens": 590, "message_tokens": 497, "message_utf8_bytes": 1152, "cache_tensor_bytes": null, "serialized_payload_bytes": 1152, "side_information_bytes": 0, "fetched_bytes": 2238, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 25733.930833172053, "hops": {"1": 6860.111457994208, "2": 6301.548000075854, "3": 6284.723916091025, "4": 6287.547459010966}, "receiver_prefill_ms": 226.64541599806398, "ttft_ms": 11.909915949217975, "decode_ms": 95.06816603243351}, "total_wall_ms": 26068.139667157084, "ttft_ms": 11.909915949217975, "input_usage": {"lookups": 12}, "output_usage": {"output_tokens": 8}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "13024817", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 36, "retention_rate": 0.5538461538461539, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.6, "amount": 0.38461538461538464, "status": 1.0, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1152}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 38, "retention_rate": 0.5846153846153846, "retention_by_kind": {"person": 0.8333333333333334, "date": 0.7, "amount": 0.38461538461538464, "status": 1.0, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1179}, "2": {"facts_total": 65, "facts_retained": 36, "retention_rate": 0.5538461538461539, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.6, "amount": 0.38461538461538464, "status": 1.0, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1152}, "3": {"facts_total": 65, "facts_retained": 36, "retention_rate": 0.5538461538461539, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.6, "amount": 0.38461538461538464, "status": 1.0, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1152}, "4": {"facts_total": 65, "facts_retained": 36, "retention_rate": 0.5538461538461539, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.6, "amount": 0.38461538461538464, "status": 1.0, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1152}}}, "reference_key_id": "keys/synthetic/syn_03021_2000.json", "task_scores": {"correct": 0.0, "abstained": 0.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "29d19ef4cb16ed19e33980db8a0baa06391ee0314866e83c8ddf69b297e56071", "output_hash": "672c00df3bfd31225e9d03325d167c25800a9007589a430995bacbf8ad8c6f21"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "ab4374f3eb03b45bd89473a7", "attempt_id": "701953fa6c1b", "run_group_id": "E4_core:A4|repairable|qwen3b:h4", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:18:32.342938Z", "task_family": "synthetic", "document_id": "syn_03021_2000", "document_hash": "3ac205b464552356699a4d2b0a7bf62497ead36e5bca73982e11ae5ee3810731", "question_id": "q12_unit_conversion", "split": "test", "cluster_id": "syn_03021_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 4, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1914, "query_tokens": 585, "message_tokens": 497, "message_utf8_bytes": 1152, "cache_tensor_bytes": null, "serialized_payload_bytes": 1152, "side_information_bytes": 0, "fetched_bytes": 2238, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 25733.930833172053, "hops": {"1": 6860.111457994208, "2": 6301.548000075854, "3": 6284.723916091025, "4": 6287.547459010966}, "receiver_prefill_ms": 225.72845791000873, "ttft_ms": 12.108584051020443, "decode_ms": 11.9452920043841}, "total_wall_ms": 25984.325875062495, "ttft_ms": 12.108584051020443, "input_usage": {"lookups": 12}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 36, "retention_rate": 0.5538461538461539, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.6, "amount": 0.38461538461538464, "status": 1.0, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1152}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 38, "retention_rate": 0.5846153846153846, "retention_by_kind": {"person": 0.8333333333333334, "date": 0.7, "amount": 0.38461538461538464, "status": 1.0, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1179}, "2": {"facts_total": 65, "facts_retained": 36, "retention_rate": 0.5538461538461539, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.6, "amount": 0.38461538461538464, "status": 1.0, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1152}, "3": {"facts_total": 65, "facts_retained": 36, "retention_rate": 0.5538461538461539, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.6, "amount": 0.38461538461538464, "status": 1.0, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1152}, "4": {"facts_total": 65, "facts_retained": 36, "retention_rate": 0.5538461538461539, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.6, "amount": 0.38461538461538464, "status": 1.0, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1152}}}, "reference_key_id": "keys/synthetic/syn_03021_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "29d19ef4cb16ed19e33980db8a0baa06391ee0314866e83c8ddf69b297e56071", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "028d9e49e3548ba4b7b15965", "attempt_id": "12e58e6b60d0", "run_group_id": "E4_core:A4|repairable|qwen3b:h4", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:18:32.608407Z", "task_family": "synthetic", "document_id": "syn_03021_2000", "document_hash": "3ac205b464552356699a4d2b0a7bf62497ead36e5bca73982e11ae5ee3810731", "question_id": "q13_unanswerable", "split": "test", "cluster_id": "syn_03021_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 4, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1914, "query_tokens": 590, "message_tokens": 497, "message_utf8_bytes": 1152, "cache_tensor_bytes": null, "serialized_payload_bytes": 1152, "side_information_bytes": 0, "fetched_bytes": 2238, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 25733.930833172053, "hops": {"1": 6860.111457994208, "2": 6301.548000075854, "3": 6284.723916091025, "4": 6287.547459010966}, "receiver_prefill_ms": 226.51729197241366, "ttft_ms": 11.971124913543463, "decode_ms": 12.001250055618584}, "total_wall_ms": 25984.878625255078, "ttft_ms": 11.971124913543463, "input_usage": {"lookups": 12}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 36, "retention_rate": 0.5538461538461539, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.6, "amount": 0.38461538461538464, "status": 1.0, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1152}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 38, "retention_rate": 0.5846153846153846, "retention_by_kind": {"person": 0.8333333333333334, "date": 0.7, "amount": 0.38461538461538464, "status": 1.0, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1179}, "2": {"facts_total": 65, "facts_retained": 36, "retention_rate": 0.5538461538461539, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.6, "amount": 0.38461538461538464, "status": 1.0, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1152}, "3": {"facts_total": 65, "facts_retained": 36, "retention_rate": 0.5538461538461539, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.6, "amount": 0.38461538461538464, "status": 1.0, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1152}, "4": {"facts_total": 65, "facts_retained": 36, "retention_rate": 0.5538461538461539, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.6, "amount": 0.38461538461538464, "status": 1.0, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1152}}}, "reference_key_id": "keys/synthetic/syn_03021_2000.json", "task_scores": {"correct": 1.0, "abstained": 1.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "29d19ef4cb16ed19e33980db8a0baa06391ee0314866e83c8ddf69b297e56071", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "bb7fcd436e6930e9dc5471f4", "attempt_id": "6c8911670c59", "run_group_id": "E4_core:A4|repairable|qwen3b:h4", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:18:32.874665Z", "task_family": "synthetic", "document_id": "syn_03021_2000", "document_hash": "3ac205b464552356699a4d2b0a7bf62497ead36e5bca73982e11ae5ee3810731", "question_id": "q14_unanswerable", "split": "test", "cluster_id": "syn_03021_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 4, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1914, "query_tokens": 592, "message_tokens": 497, "message_utf8_bytes": 1152, "cache_tensor_bytes": null, "serialized_payload_bytes": 1152, "side_information_bytes": 0, "fetched_bytes": 2238, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 25733.930833172053, "hops": {"1": 6860.111457994208, "2": 6301.548000075854, "3": 6284.723916091025, "4": 6287.547459010966}, "receiver_prefill_ms": 226.58804105594754, "ttft_ms": 11.91033294890076, "decode_ms": 11.964667006395757}, "total_wall_ms": 25984.893624205142, "ttft_ms": 11.91033294890076, "input_usage": {"lookups": 12}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 36, "retention_rate": 0.5538461538461539, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.6, "amount": 0.38461538461538464, "status": 1.0, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1152}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 38, "retention_rate": 0.5846153846153846, "retention_by_kind": {"person": 0.8333333333333334, "date": 0.7, "amount": 0.38461538461538464, "status": 1.0, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1179}, "2": {"facts_total": 65, "facts_retained": 36, "retention_rate": 0.5538461538461539, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.6, "amount": 0.38461538461538464, "status": 1.0, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1152}, "3": {"facts_total": 65, "facts_retained": 36, "retention_rate": 0.5538461538461539, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.6, "amount": 0.38461538461538464, "status": 1.0, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1152}, "4": {"facts_total": 65, "facts_retained": 36, "retention_rate": 0.5538461538461539, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.6, "amount": 0.38461538461538464, "status": 1.0, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1152}}}, "reference_key_id": "keys/synthetic/syn_03021_2000.json", "task_scores": {"correct": 1.0, "abstained": 1.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "29d19ef4cb16ed19e33980db8a0baa06391ee0314866e83c8ddf69b297e56071", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "4e3fe79b117646c8676c65d0", "attempt_id": "fb80622b766e", "run_group_id": "E4_core:A4|repairable|qwen3b:h8", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:18:58.272129Z", "task_family": "synthetic", "document_id": "syn_03021_2000", "document_hash": "3ac205b464552356699a4d2b0a7bf62497ead36e5bca73982e11ae5ee3810731", "question_id": "q00_single_fact", "split": "test", "cluster_id": "syn_03021_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 8, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1914, "query_tokens": 577, "message_tokens": 497, "message_utf8_bytes": 1152, "cache_tensor_bytes": null, "serialized_payload_bytes": 1152, "side_information_bytes": 0, "fetched_bytes": 5222, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 50860.583542031236, "hops": {"1": 6860.111457994208, "2": 6301.548000075854, "3": 6284.723916091025, "4": 6287.547459010966, "5": 6284.2733329162, "6": 6280.639917007647, "7": 6281.036333995871, "8": 6280.703124939464}, "receiver_prefill_ms": 208.69258395396173, "ttft_ms": 11.884875013493001, "decode_ms": 11.874084011651576}, "total_wall_ms": 51093.529292033054, "ttft_ms": 11.884875013493001, "input_usage": {"lookups": 28}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 36, "retention_rate": 0.5538461538461539, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.6, "amount": 0.38461538461538464, "status": 1.0, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1152}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 38, "retention_rate": 0.5846153846153846, "retention_by_kind": {"person": 0.8333333333333334, "date": 0.7, "amount": 0.38461538461538464, "status": 1.0, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1179}, "2": {"facts_total": 65, "facts_retained": 36, "retention_rate": 0.5538461538461539, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.6, "amount": 0.38461538461538464, "status": 1.0, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1152}, "3": {"facts_total": 65, "facts_retained": 36, "retention_rate": 0.5538461538461539, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.6, "amount": 0.38461538461538464, "status": 1.0, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1152}, "4": {"facts_total": 65, "facts_retained": 36, "retention_rate": 0.5538461538461539, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.6, "amount": 0.38461538461538464, "status": 1.0, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1152}, "5": {"facts_total": 65, "facts_retained": 36, "retention_rate": 0.5538461538461539, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.6, "amount": 0.38461538461538464, "status": 1.0, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1152}, "6": {"facts_total": 65, "facts_retained": 36, "retention_rate": 0.5538461538461539, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.6, "amount": 0.38461538461538464, "status": 1.0, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1152}, "7": {"facts_total": 65, "facts_retained": 36, "retention_rate": 0.5538461538461539, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.6, "amount": 0.38461538461538464, "status": 1.0, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1152}, "8": {"facts_total": 65, "facts_retained": 36, "retention_rate": 0.5538461538461539, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.6, "amount": 0.38461538461538464, "status": 1.0, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1152}}}, "reference_key_id": "keys/synthetic/syn_03021_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "29d19ef4cb16ed19e33980db8a0baa06391ee0314866e83c8ddf69b297e56071", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "b99d21b2d0bfc02fbbefdcb8", "attempt_id": "a04ee8808f84", "run_group_id": "E4_core:A4|repairable|qwen3b:h8", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:18:58.520777Z", "task_family": "synthetic", "document_id": "syn_03021_2000", "document_hash": "3ac205b464552356699a4d2b0a7bf62497ead36e5bca73982e11ae5ee3810731", "question_id": "q01_single_fact", "split": "test", "cluster_id": "syn_03021_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 8, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1914, "query_tokens": 583, "message_tokens": 497, "message_utf8_bytes": 1152, "cache_tensor_bytes": null, "serialized_payload_bytes": 1152, "side_information_bytes": 0, "fetched_bytes": 5222, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 50860.583542031236, "hops": {"1": 6860.111457994208, "2": 6301.548000075854, "3": 6284.723916091025, "4": 6287.547459010966, "5": 6284.2733329162, "6": 6280.639917007647, "7": 6281.036333995871, "8": 6280.703124939464}, "receiver_prefill_ms": 226.50037507992238, "ttft_ms": 11.87854097224772, "decode_ms": 130.59416692703962}, "total_wall_ms": 51230.06037506275, "ttft_ms": 11.87854097224772, "input_usage": {"lookups": 28}, "output_usage": {"output_tokens": 11}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "Lot 536-R, Senneterre", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 36, "retention_rate": 0.5538461538461539, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.6, "amount": 0.38461538461538464, "status": 1.0, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1152}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 38, "retention_rate": 0.5846153846153846, "retention_by_kind": {"person": 0.8333333333333334, "date": 0.7, "amount": 0.38461538461538464, "status": 1.0, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1179}, "2": {"facts_total": 65, "facts_retained": 36, "retention_rate": 0.5538461538461539, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.6, "amount": 0.38461538461538464, "status": 1.0, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1152}, "3": {"facts_total": 65, "facts_retained": 36, "retention_rate": 0.5538461538461539, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.6, "amount": 0.38461538461538464, "status": 1.0, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1152}, "4": {"facts_total": 65, "facts_retained": 36, "retention_rate": 0.5538461538461539, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.6, "amount": 0.38461538461538464, "status": 1.0, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1152}, "5": {"facts_total": 65, "facts_retained": 36, "retention_rate": 0.5538461538461539, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.6, "amount": 0.38461538461538464, "status": 1.0, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1152}, "6": {"facts_total": 65, "facts_retained": 36, "retention_rate": 0.5538461538461539, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.6, "amount": 0.38461538461538464, "status": 1.0, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1152}, "7": {"facts_total": 65, "facts_retained": 36, "retention_rate": 0.5538461538461539, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.6, "amount": 0.38461538461538464, "status": 1.0, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1152}, "8": {"facts_total": 65, "facts_retained": 36, "retention_rate": 0.5538461538461539, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.6, "amount": 0.38461538461538464, "status": 1.0, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1152}}}, "reference_key_id": "keys/synthetic/syn_03021_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 0.8571428571428571}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "29d19ef4cb16ed19e33980db8a0baa06391ee0314866e83c8ddf69b297e56071", "output_hash": "5e85af24aac0f1904d70a2fcee84f1b6f6912cc7450298af1f173002fd8e59df"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "fdb50f2b66efd8c03eeebc15", "attempt_id": "f04eb326e10a", "run_group_id": "E4_core:A4|repairable|qwen3b:h8", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:18:58.905448Z", "task_family": "synthetic", "document_id": "syn_03021_2000", "document_hash": "3ac205b464552356699a4d2b0a7bf62497ead36e5bca73982e11ae5ee3810731", "question_id": "q02_exact_copy", "split": "test", "cluster_id": "syn_03021_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 8, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1914, "query_tokens": 583, "message_tokens": 497, "message_utf8_bytes": 1152, "cache_tensor_bytes": null, "serialized_payload_bytes": 1152, "side_information_bytes": 0, "fetched_bytes": 5222, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 50860.583542031236, "hops": {"1": 6860.111457994208, "2": 6301.548000075854, "3": 6284.723916091025, "4": 6287.547459010966, "5": 6284.2733329162, "6": 6280.639917007647, "7": 6281.036333995871, "8": 6280.703124939464}, "receiver_prefill_ms": 226.56700003426522, "ttft_ms": 11.946541955694556, "decode_ms": 82.96749996952713}, "total_wall_ms": 51182.533667073585, "ttft_ms": 11.946541955694556, "input_usage": {"lookups": 28}, "output_usage": {"output_tokens": 7}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "88.7751", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 36, "retention_rate": 0.5538461538461539, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.6, "amount": 0.38461538461538464, "status": 1.0, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1152}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 38, "retention_rate": 0.5846153846153846, "retention_by_kind": {"person": 0.8333333333333334, "date": 0.7, "amount": 0.38461538461538464, "status": 1.0, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1179}, "2": {"facts_total": 65, "facts_retained": 36, "retention_rate": 0.5538461538461539, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.6, "amount": 0.38461538461538464, "status": 1.0, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1152}, "3": {"facts_total": 65, "facts_retained": 36, "retention_rate": 0.5538461538461539, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.6, "amount": 0.38461538461538464, "status": 1.0, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1152}, "4": {"facts_total": 65, "facts_retained": 36, "retention_rate": 0.5538461538461539, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.6, "amount": 0.38461538461538464, "status": 1.0, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1152}, "5": {"facts_total": 65, "facts_retained": 36, "retention_rate": 0.5538461538461539, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.6, "amount": 0.38461538461538464, "status": 1.0, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1152}, "6": {"facts_total": 65, "facts_retained": 36, "retention_rate": 0.5538461538461539, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.6, "amount": 0.38461538461538464, "status": 1.0, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1152}, "7": {"facts_total": 65, "facts_retained": 36, "retention_rate": 0.5538461538461539, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.6, "amount": 0.38461538461538464, "status": 1.0, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1152}, "8": {"facts_total": 65, "facts_retained": 36, "retention_rate": 0.5538461538461539, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.6, "amount": 0.38461538461538464, "status": 1.0, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1152}}}, "reference_key_id": "keys/synthetic/syn_03021_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "29d19ef4cb16ed19e33980db8a0baa06391ee0314866e83c8ddf69b297e56071", "output_hash": "1e01e62c69f9479b681e50e59e8565c6583f5cdb803446a05ed5cb13118acdf9"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "6d679d27a5f7afc5d3084456", "attempt_id": "4d333eabca8e", "run_group_id": "E4_core:A4|repairable|qwen3b:h8", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:18:59.242133Z", "task_family": "synthetic", "document_id": "syn_03021_2000", "document_hash": "3ac205b464552356699a4d2b0a7bf62497ead36e5bca73982e11ae5ee3810731", "question_id": "q03_exact_copy", "split": "test", "cluster_id": "syn_03021_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 8, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1914, "query_tokens": 580, "message_tokens": 497, "message_utf8_bytes": 1152, "cache_tensor_bytes": null, "serialized_payload_bytes": 1152, "side_information_bytes": 0, "fetched_bytes": 5222, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 50860.583542031236, "hops": {"1": 6860.111457994208, "2": 6301.548000075854, "3": 6284.723916091025, "4": 6287.547459010966, "5": 6284.2733329162, "6": 6280.639917007647, "7": 6281.036333995871, "8": 6280.703124939464}, "receiver_prefill_ms": 225.9958341019228, "ttft_ms": 11.976291076280177, "decode_ms": 11.83108298573643}, "total_wall_ms": 51110.87299999781, "ttft_ms": 11.976291076280177, "input_usage": {"lookups": 28}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 36, "retention_rate": 0.5538461538461539, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.6, "amount": 0.38461538461538464, "status": 1.0, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1152}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 38, "retention_rate": 0.5846153846153846, "retention_by_kind": {"person": 0.8333333333333334, "date": 0.7, "amount": 0.38461538461538464, "status": 1.0, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1179}, "2": {"facts_total": 65, "facts_retained": 36, "retention_rate": 0.5538461538461539, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.6, "amount": 0.38461538461538464, "status": 1.0, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1152}, "3": {"facts_total": 65, "facts_retained": 36, "retention_rate": 0.5538461538461539, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.6, "amount": 0.38461538461538464, "status": 1.0, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1152}, "4": {"facts_total": 65, "facts_retained": 36, "retention_rate": 0.5538461538461539, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.6, "amount": 0.38461538461538464, "status": 1.0, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1152}, "5": {"facts_total": 65, "facts_retained": 36, "retention_rate": 0.5538461538461539, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.6, "amount": 0.38461538461538464, "status": 1.0, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1152}, "6": {"facts_total": 65, "facts_retained": 36, "retention_rate": 0.5538461538461539, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.6, "amount": 0.38461538461538464, "status": 1.0, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1152}, "7": {"facts_total": 65, "facts_retained": 36, "retention_rate": 0.5538461538461539, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.6, "amount": 0.38461538461538464, "status": 1.0, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1152}, "8": {"facts_total": 65, "facts_retained": 36, "retention_rate": 0.5538461538461539, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.6, "amount": 0.38461538461538464, "status": 1.0, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1152}}}, "reference_key_id": "keys/synthetic/syn_03021_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "29d19ef4cb16ed19e33980db8a0baa06391ee0314866e83c8ddf69b297e56071", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "5c86d699770928bf84c3c791", "attempt_id": "ee4abff82877", "run_group_id": "E4_core:A4|repairable|qwen3b:h8", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:18:59.507138Z", "task_family": "synthetic", "document_id": "syn_03021_2000", "document_hash": "3ac205b464552356699a4d2b0a7bf62497ead36e5bca73982e11ae5ee3810731", "question_id": "q04_multi_hop", "split": "test", "cluster_id": "syn_03021_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 8, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1914, "query_tokens": 587, "message_tokens": 497, "message_utf8_bytes": 1152, "cache_tensor_bytes": null, "serialized_payload_bytes": 1152, "side_information_bytes": 0, "fetched_bytes": 5222, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 50860.583542031236, "hops": {"1": 6860.111457994208, "2": 6301.548000075854, "3": 6284.723916091025, "4": 6287.547459010966, "5": 6284.2733329162, "6": 6280.639917007647, "7": 6281.036333995871, "8": 6280.703124939464}, "receiver_prefill_ms": 226.3192079262808, "ttft_ms": 12.070041964761913, "decode_ms": 11.930875014513731}, "total_wall_ms": 51111.36558302678, "ttft_ms": 12.070041964761913, "input_usage": {"lookups": 28}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 36, "retention_rate": 0.5538461538461539, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.6, "amount": 0.38461538461538464, "status": 1.0, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1152}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 38, "retention_rate": 0.5846153846153846, "retention_by_kind": {"person": 0.8333333333333334, "date": 0.7, "amount": 0.38461538461538464, "status": 1.0, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1179}, "2": {"facts_total": 65, "facts_retained": 36, "retention_rate": 0.5538461538461539, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.6, "amount": 0.38461538461538464, "status": 1.0, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1152}, "3": {"facts_total": 65, "facts_retained": 36, "retention_rate": 0.5538461538461539, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.6, "amount": 0.38461538461538464, "status": 1.0, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1152}, "4": {"facts_total": 65, "facts_retained": 36, "retention_rate": 0.5538461538461539, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.6, "amount": 0.38461538461538464, "status": 1.0, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1152}, "5": {"facts_total": 65, "facts_retained": 36, "retention_rate": 0.5538461538461539, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.6, "amount": 0.38461538461538464, "status": 1.0, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1152}, "6": {"facts_total": 65, "facts_retained": 36, "retention_rate": 0.5538461538461539, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.6, "amount": 0.38461538461538464, "status": 1.0, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1152}, "7": {"facts_total": 65, "facts_retained": 36, "retention_rate": 0.5538461538461539, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.6, "amount": 0.38461538461538464, "status": 1.0, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1152}, "8": {"facts_total": 65, "facts_retained": 36, "retention_rate": 0.5538461538461539, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.6, "amount": 0.38461538461538464, "status": 1.0, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1152}}}, "reference_key_id": "keys/synthetic/syn_03021_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "29d19ef4cb16ed19e33980db8a0baa06391ee0314866e83c8ddf69b297e56071", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "cfae9e6ec10ad8e3ded42592", "attempt_id": "121fceb1df67", "run_group_id": "E4_core:A4|repairable|qwen3b:h8", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:18:59.772630Z", "task_family": "synthetic", "document_id": "syn_03021_2000", "document_hash": "3ac205b464552356699a4d2b0a7bf62497ead36e5bca73982e11ae5ee3810731", "question_id": "q05_multi_hop", "split": "test", "cluster_id": "syn_03021_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 8, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1914, "query_tokens": 588, "message_tokens": 497, "message_utf8_bytes": 1152, "cache_tensor_bytes": null, "serialized_payload_bytes": 1152, "side_information_bytes": 0, "fetched_bytes": 5222, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 50860.583542031236, "hops": {"1": 6860.111457994208, "2": 6301.548000075854, "3": 6284.723916091025, "4": 6287.547459010966, "5": 6284.2733329162, "6": 6280.639917007647, "7": 6281.036333995871, "8": 6280.703124939464}, "receiver_prefill_ms": 226.77674994338304, "ttft_ms": 12.181833968497813, "decode_ms": 131.2990830047056}, "total_wall_ms": 51231.40708403662, "ttft_ms": 12.181833968497813, "input_usage": {"lookups": 28}, "output_usage": {"output_tokens": 11}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "BQ-59Z-6753", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 36, "retention_rate": 0.5538461538461539, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.6, "amount": 0.38461538461538464, "status": 1.0, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1152}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 38, "retention_rate": 0.5846153846153846, "retention_by_kind": {"person": 0.8333333333333334, "date": 0.7, "amount": 0.38461538461538464, "status": 1.0, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1179}, "2": {"facts_total": 65, "facts_retained": 36, "retention_rate": 0.5538461538461539, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.6, "amount": 0.38461538461538464, "status": 1.0, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1152}, "3": {"facts_total": 65, "facts_retained": 36, "retention_rate": 0.5538461538461539, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.6, "amount": 0.38461538461538464, "status": 1.0, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1152}, "4": {"facts_total": 65, "facts_retained": 36, "retention_rate": 0.5538461538461539, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.6, "amount": 0.38461538461538464, "status": 1.0, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1152}, "5": {"facts_total": 65, "facts_retained": 36, "retention_rate": 0.5538461538461539, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.6, "amount": 0.38461538461538464, "status": 1.0, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1152}, "6": {"facts_total": 65, "facts_retained": 36, "retention_rate": 0.5538461538461539, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.6, "amount": 0.38461538461538464, "status": 1.0, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1152}, "7": {"facts_total": 65, "facts_retained": 36, "retention_rate": 0.5538461538461539, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.6, "amount": 0.38461538461538464, "status": 1.0, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1152}, "8": {"facts_total": 65, "facts_retained": 36, "retention_rate": 0.5538461538461539, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.6, "amount": 0.38461538461538464, "status": 1.0, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1152}}}, "reference_key_id": "keys/synthetic/syn_03021_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "29d19ef4cb16ed19e33980db8a0baa06391ee0314866e83c8ddf69b297e56071", "output_hash": "db250f072c856fe7ddf4bbbd16e20c71b8bad1eef5e4480acd0dac30f8ffe9cd"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "7996816ebde05369c712c59d", "attempt_id": "a68ad8d9efb1", "run_group_id": "E4_core:A4|repairable|qwen3b:h8", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:19:00.158309Z", "task_family": "synthetic", "document_id": "syn_03021_2000", "document_hash": "3ac205b464552356699a4d2b0a7bf62497ead36e5bca73982e11ae5ee3810731", "question_id": "q06_negation", "split": "test", "cluster_id": "syn_03021_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 8, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1914, "query_tokens": 578, "message_tokens": 497, "message_utf8_bytes": 1152, "cache_tensor_bytes": null, "serialized_payload_bytes": 1152, "side_information_bytes": 0, "fetched_bytes": 5222, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 50860.583542031236, "hops": {"1": 6860.111457994208, "2": 6301.548000075854, "3": 6284.723916091025, "4": 6287.547459010966, "5": 6284.2733329162, "6": 6280.639917007647, "7": 6281.036333995871, "8": 6280.703124939464}, "receiver_prefill_ms": 225.8157500764355, "ttft_ms": 11.92099996842444, "decode_ms": 11.865791981108487}, "total_wall_ms": 51110.675751115195, "ttft_ms": 11.92099996842444, "input_usage": {"lookups": 28}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "No", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 36, "retention_rate": 0.5538461538461539, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.6, "amount": 0.38461538461538464, "status": 1.0, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1152}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 38, "retention_rate": 0.5846153846153846, "retention_by_kind": {"person": 0.8333333333333334, "date": 0.7, "amount": 0.38461538461538464, "status": 1.0, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1179}, "2": {"facts_total": 65, "facts_retained": 36, "retention_rate": 0.5538461538461539, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.6, "amount": 0.38461538461538464, "status": 1.0, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1152}, "3": {"facts_total": 65, "facts_retained": 36, "retention_rate": 0.5538461538461539, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.6, "amount": 0.38461538461538464, "status": 1.0, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1152}, "4": {"facts_total": 65, "facts_retained": 36, "retention_rate": 0.5538461538461539, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.6, "amount": 0.38461538461538464, "status": 1.0, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1152}, "5": {"facts_total": 65, "facts_retained": 36, "retention_rate": 0.5538461538461539, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.6, "amount": 0.38461538461538464, "status": 1.0, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1152}, "6": {"facts_total": 65, "facts_retained": 36, "retention_rate": 0.5538461538461539, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.6, "amount": 0.38461538461538464, "status": 1.0, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1152}, "7": {"facts_total": 65, "facts_retained": 36, "retention_rate": 0.5538461538461539, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.6, "amount": 0.38461538461538464, "status": 1.0, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1152}, "8": {"facts_total": 65, "facts_retained": 36, "retention_rate": 0.5538461538461539, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.6, "amount": 0.38461538461538464, "status": 1.0, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1152}}}, "reference_key_id": "keys/synthetic/syn_03021_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "29d19ef4cb16ed19e33980db8a0baa06391ee0314866e83c8ddf69b297e56071", "output_hash": "1ea442a134b2a184bd5d40104401f2a37fbc09ccf3f4bc9da161c6099be3691d"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "49a0c8f91f796305ff856c8b", "attempt_id": "d34aa3289eb0", "run_group_id": "E4_core:A4|repairable|qwen3b:h8", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:19:00.423596Z", "task_family": "synthetic", "document_id": "syn_03021_2000", "document_hash": "3ac205b464552356699a4d2b0a7bf62497ead36e5bca73982e11ae5ee3810731", "question_id": "q07_negation", "split": "test", "cluster_id": "syn_03021_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 8, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1914, "query_tokens": 579, "message_tokens": 497, "message_utf8_bytes": 1152, "cache_tensor_bytes": null, "serialized_payload_bytes": 1152, "side_information_bytes": 0, "fetched_bytes": 5222, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 50860.583542031236, "hops": {"1": 6860.111457994208, "2": 6301.548000075854, "3": 6284.723916091025, "4": 6287.547459010966, "5": 6284.2733329162, "6": 6280.639917007647, "7": 6281.036333995871, "8": 6280.703124939464}, "receiver_prefill_ms": 226.30445798859, "ttft_ms": 12.106165988370776, "decode_ms": 11.934584006667137}, "total_wall_ms": 51111.38970812317, "ttft_ms": 12.106165988370776, "input_usage": {"lookups": 28}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "yes", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 36, "retention_rate": 0.5538461538461539, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.6, "amount": 0.38461538461538464, "status": 1.0, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1152}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 38, "retention_rate": 0.5846153846153846, "retention_by_kind": {"person": 0.8333333333333334, "date": 0.7, "amount": 0.38461538461538464, "status": 1.0, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1179}, "2": {"facts_total": 65, "facts_retained": 36, "retention_rate": 0.5538461538461539, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.6, "amount": 0.38461538461538464, "status": 1.0, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1152}, "3": {"facts_total": 65, "facts_retained": 36, "retention_rate": 0.5538461538461539, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.6, "amount": 0.38461538461538464, "status": 1.0, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1152}, "4": {"facts_total": 65, "facts_retained": 36, "retention_rate": 0.5538461538461539, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.6, "amount": 0.38461538461538464, "status": 1.0, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1152}, "5": {"facts_total": 65, "facts_retained": 36, "retention_rate": 0.5538461538461539, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.6, "amount": 0.38461538461538464, "status": 1.0, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1152}, "6": {"facts_total": 65, "facts_retained": 36, "retention_rate": 0.5538461538461539, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.6, "amount": 0.38461538461538464, "status": 1.0, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1152}, "7": {"facts_total": 65, "facts_retained": 36, "retention_rate": 0.5538461538461539, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.6, "amount": 0.38461538461538464, "status": 1.0, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1152}, "8": {"facts_total": 65, "facts_retained": 36, "retention_rate": 0.5538461538461539, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.6, "amount": 0.38461538461538464, "status": 1.0, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1152}}}, "reference_key_id": "keys/synthetic/syn_03021_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "29d19ef4cb16ed19e33980db8a0baa06391ee0314866e83c8ddf69b297e56071", "output_hash": "8a798890fe93817163b10b5f7bd2ca4d25d84c52739a645a889c173eee7d9d3d"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "8ea2c2bb519b47b7e2e55a3d", "attempt_id": "cfbf2567603c", "run_group_id": "E4_core:A4|repairable|qwen3b:h8", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:19:00.689159Z", "task_family": "synthetic", "document_id": "syn_03021_2000", "document_hash": "3ac205b464552356699a4d2b0a7bf62497ead36e5bca73982e11ae5ee3810731", "question_id": "q08_revision", "split": "test", "cluster_id": "syn_03021_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 8, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1914, "query_tokens": 595, "message_tokens": 497, "message_utf8_bytes": 1152, "cache_tensor_bytes": null, "serialized_payload_bytes": 1152, "side_information_bytes": 0, "fetched_bytes": 5222, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 50860.583542031236, "hops": {"1": 6860.111457994208, "2": 6301.548000075854, "3": 6284.723916091025, "4": 6287.547459010966, "5": 6284.2733329162, "6": 6280.639917007647, "7": 6281.036333995871, "8": 6280.703124939464}, "receiver_prefill_ms": 226.98754200246185, "ttft_ms": 11.919333948753774, "decode_ms": 11.920499964617193}, "total_wall_ms": 51111.89204198308, "ttft_ms": 11.919333948753774, "input_usage": {"lookups": 28}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 36, "retention_rate": 0.5538461538461539, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.6, "amount": 0.38461538461538464, "status": 1.0, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1152}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 38, "retention_rate": 0.5846153846153846, "retention_by_kind": {"person": 0.8333333333333334, "date": 0.7, "amount": 0.38461538461538464, "status": 1.0, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1179}, "2": {"facts_total": 65, "facts_retained": 36, "retention_rate": 0.5538461538461539, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.6, "amount": 0.38461538461538464, "status": 1.0, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1152}, "3": {"facts_total": 65, "facts_retained": 36, "retention_rate": 0.5538461538461539, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.6, "amount": 0.38461538461538464, "status": 1.0, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1152}, "4": {"facts_total": 65, "facts_retained": 36, "retention_rate": 0.5538461538461539, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.6, "amount": 0.38461538461538464, "status": 1.0, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1152}, "5": {"facts_total": 65, "facts_retained": 36, "retention_rate": 0.5538461538461539, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.6, "amount": 0.38461538461538464, "status": 1.0, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1152}, "6": {"facts_total": 65, "facts_retained": 36, "retention_rate": 0.5538461538461539, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.6, "amount": 0.38461538461538464, "status": 1.0, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1152}, "7": {"facts_total": 65, "facts_retained": 36, "retention_rate": 0.5538461538461539, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.6, "amount": 0.38461538461538464, "status": 1.0, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1152}, "8": {"facts_total": 65, "facts_retained": 36, "retention_rate": 0.5538461538461539, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.6, "amount": 0.38461538461538464, "status": 1.0, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1152}}}, "reference_key_id": "keys/synthetic/syn_03021_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "29d19ef4cb16ed19e33980db8a0baa06391ee0314866e83c8ddf69b297e56071", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "0167a7006f01719d616eb409", "attempt_id": "f61a511130c5", "run_group_id": "E4_core:A4|repairable|qwen3b:h8", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:19:00.955316Z", "task_family": "synthetic", "document_id": "syn_03021_2000", "document_hash": "3ac205b464552356699a4d2b0a7bf62497ead36e5bca73982e11ae5ee3810731", "question_id": "q09_revision", "split": "test", "cluster_id": "syn_03021_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 8, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1914, "query_tokens": 586, "message_tokens": 497, "message_utf8_bytes": 1152, "cache_tensor_bytes": null, "serialized_payload_bytes": 1152, "side_information_bytes": 0, "fetched_bytes": 5222, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 50860.583542031236, "hops": {"1": 6860.111457994208, "2": 6301.548000075854, "3": 6284.723916091025, "4": 6287.547459010966, "5": 6284.2733329162, "6": 6280.639917007647, "7": 6281.036333995871, "8": 6280.703124939464}, "receiver_prefill_ms": 226.63349995855242, "ttft_ms": 11.85654103755951, "decode_ms": 11.883166967891157}, "total_wall_ms": 51111.41916702036, "ttft_ms": 11.85654103755951, "input_usage": {"lookups": 28}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 36, "retention_rate": 0.5538461538461539, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.6, "amount": 0.38461538461538464, "status": 1.0, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1152}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 38, "retention_rate": 0.5846153846153846, "retention_by_kind": {"person": 0.8333333333333334, "date": 0.7, "amount": 0.38461538461538464, "status": 1.0, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1179}, "2": {"facts_total": 65, "facts_retained": 36, "retention_rate": 0.5538461538461539, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.6, "amount": 0.38461538461538464, "status": 1.0, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1152}, "3": {"facts_total": 65, "facts_retained": 36, "retention_rate": 0.5538461538461539, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.6, "amount": 0.38461538461538464, "status": 1.0, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1152}, "4": {"facts_total": 65, "facts_retained": 36, "retention_rate": 0.5538461538461539, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.6, "amount": 0.38461538461538464, "status": 1.0, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1152}, "5": {"facts_total": 65, "facts_retained": 36, "retention_rate": 0.5538461538461539, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.6, "amount": 0.38461538461538464, "status": 1.0, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1152}, "6": {"facts_total": 65, "facts_retained": 36, "retention_rate": 0.5538461538461539, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.6, "amount": 0.38461538461538464, "status": 1.0, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1152}, "7": {"facts_total": 65, "facts_retained": 36, "retention_rate": 0.5538461538461539, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.6, "amount": 0.38461538461538464, "status": 1.0, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1152}, "8": {"facts_total": 65, "facts_retained": 36, "retention_rate": 0.5538461538461539, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.6, "amount": 0.38461538461538464, "status": 1.0, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1152}}}, "reference_key_id": "keys/synthetic/syn_03021_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "29d19ef4cb16ed19e33980db8a0baa06391ee0314866e83c8ddf69b297e56071", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "da4946f419698264580b92a1", "attempt_id": "7f552d794ce9", "run_group_id": "E4_core:A4|repairable|qwen3b:h8", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:19:01.220838Z", "task_family": "synthetic", "document_id": "syn_03021_2000", "document_hash": "3ac205b464552356699a4d2b0a7bf62497ead36e5bca73982e11ae5ee3810731", "question_id": "q10_temporal", "split": "test", "cluster_id": "syn_03021_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 8, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1914, "query_tokens": 594, "message_tokens": 497, "message_utf8_bytes": 1152, "cache_tensor_bytes": null, "serialized_payload_bytes": 1152, "side_information_bytes": 0, "fetched_bytes": 5222, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 50860.583542031236, "hops": {"1": 6860.111457994208, "2": 6301.548000075854, "3": 6284.723916091025, "4": 6287.547459010966, "5": 6284.2733329162, "6": 6280.639917007647, "7": 6281.036333995871, "8": 6280.703124939464}, "receiver_prefill_ms": 227.0756249781698, "ttft_ms": 11.885665939189494, "decode_ms": 59.61816699709743}, "total_wall_ms": 51159.6483750036, "ttft_ms": 11.885665939189494, "input_usage": {"lookups": 28}, "output_usage": {"output_tokens": 5}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "Kovar-9", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 36, "retention_rate": 0.5538461538461539, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.6, "amount": 0.38461538461538464, "status": 1.0, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1152}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 38, "retention_rate": 0.5846153846153846, "retention_by_kind": {"person": 0.8333333333333334, "date": 0.7, "amount": 0.38461538461538464, "status": 1.0, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1179}, "2": {"facts_total": 65, "facts_retained": 36, "retention_rate": 0.5538461538461539, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.6, "amount": 0.38461538461538464, "status": 1.0, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1152}, "3": {"facts_total": 65, "facts_retained": 36, "retention_rate": 0.5538461538461539, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.6, "amount": 0.38461538461538464, "status": 1.0, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1152}, "4": {"facts_total": 65, "facts_retained": 36, "retention_rate": 0.5538461538461539, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.6, "amount": 0.38461538461538464, "status": 1.0, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1152}, "5": {"facts_total": 65, "facts_retained": 36, "retention_rate": 0.5538461538461539, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.6, "amount": 0.38461538461538464, "status": 1.0, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1152}, "6": {"facts_total": 65, "facts_retained": 36, "retention_rate": 0.5538461538461539, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.6, "amount": 0.38461538461538464, "status": 1.0, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1152}, "7": {"facts_total": 65, "facts_retained": 36, "retention_rate": 0.5538461538461539, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.6, "amount": 0.38461538461538464, "status": 1.0, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1152}, "8": {"facts_total": 65, "facts_retained": 36, "retention_rate": 0.5538461538461539, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.6, "amount": 0.38461538461538464, "status": 1.0, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1152}}}, "reference_key_id": "keys/synthetic/syn_03021_2000.json", "task_scores": {"correct": 0.0, "abstained": 0.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "29d19ef4cb16ed19e33980db8a0baa06391ee0314866e83c8ddf69b297e56071", "output_hash": "c969b5f7547945067f5b07e9f822d0043126326f979d950a8b1b045057fbb818"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "80ae66646370c00342102a8d", "attempt_id": "d18af4a3b85d", "run_group_id": "E4_core:A4|repairable|qwen3b:h8", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:19:01.534430Z", "task_family": "synthetic", "document_id": "syn_03021_2000", "document_hash": "3ac205b464552356699a4d2b0a7bf62497ead36e5bca73982e11ae5ee3810731", "question_id": "q11_aggregation", "split": "test", "cluster_id": "syn_03021_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 8, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1914, "query_tokens": 590, "message_tokens": 497, "message_utf8_bytes": 1152, "cache_tensor_bytes": null, "serialized_payload_bytes": 1152, "side_information_bytes": 0, "fetched_bytes": 5222, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 50860.583542031236, "hops": {"1": 6860.111457994208, "2": 6301.548000075854, "3": 6284.723916091025, "4": 6287.547459010966, "5": 6284.2733329162, "6": 6280.639917007647, "7": 6281.036333995871, "8": 6280.703124939464}, "receiver_prefill_ms": 226.76312504336238, "ttft_ms": 11.779709020629525, "decode_ms": 95.16625001560897}, "total_wall_ms": 51194.89991699811, "ttft_ms": 11.779709020629525, "input_usage": {"lookups": 28}, "output_usage": {"output_tokens": 8}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "13024817", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 36, "retention_rate": 0.5538461538461539, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.6, "amount": 0.38461538461538464, "status": 1.0, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1152}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 38, "retention_rate": 0.5846153846153846, "retention_by_kind": {"person": 0.8333333333333334, "date": 0.7, "amount": 0.38461538461538464, "status": 1.0, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1179}, "2": {"facts_total": 65, "facts_retained": 36, "retention_rate": 0.5538461538461539, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.6, "amount": 0.38461538461538464, "status": 1.0, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1152}, "3": {"facts_total": 65, "facts_retained": 36, "retention_rate": 0.5538461538461539, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.6, "amount": 0.38461538461538464, "status": 1.0, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1152}, "4": {"facts_total": 65, "facts_retained": 36, "retention_rate": 0.5538461538461539, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.6, "amount": 0.38461538461538464, "status": 1.0, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1152}, "5": {"facts_total": 65, "facts_retained": 36, "retention_rate": 0.5538461538461539, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.6, "amount": 0.38461538461538464, "status": 1.0, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1152}, "6": {"facts_total": 65, "facts_retained": 36, "retention_rate": 0.5538461538461539, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.6, "amount": 0.38461538461538464, "status": 1.0, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1152}, "7": {"facts_total": 65, "facts_retained": 36, "retention_rate": 0.5538461538461539, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.6, "amount": 0.38461538461538464, "status": 1.0, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1152}, "8": {"facts_total": 65, "facts_retained": 36, "retention_rate": 0.5538461538461539, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.6, "amount": 0.38461538461538464, "status": 1.0, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1152}}}, "reference_key_id": "keys/synthetic/syn_03021_2000.json", "task_scores": {"correct": 0.0, "abstained": 0.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "29d19ef4cb16ed19e33980db8a0baa06391ee0314866e83c8ddf69b297e56071", "output_hash": "672c00df3bfd31225e9d03325d167c25800a9007589a430995bacbf8ad8c6f21"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "1b9af61640909b3879536487", "attempt_id": "6ada368df423", "run_group_id": "E4_core:A4|repairable|qwen3b:h8", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:19:01.883642Z", "task_family": "synthetic", "document_id": "syn_03021_2000", "document_hash": "3ac205b464552356699a4d2b0a7bf62497ead36e5bca73982e11ae5ee3810731", "question_id": "q12_unit_conversion", "split": "test", "cluster_id": "syn_03021_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 8, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1914, "query_tokens": 585, "message_tokens": 497, "message_utf8_bytes": 1152, "cache_tensor_bytes": null, "serialized_payload_bytes": 1152, "side_information_bytes": 0, "fetched_bytes": 5222, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 50860.583542031236, "hops": {"1": 6860.111457994208, "2": 6301.548000075854, "3": 6284.723916091025, "4": 6287.547459010966, "5": 6284.2733329162, "6": 6280.639917007647, "7": 6281.036333995871, "8": 6280.703124939464}, "receiver_prefill_ms": 226.4832080109045, "ttft_ms": 11.929625063203275, "decode_ms": 11.864207917824388}, "total_wall_ms": 51111.36079195421, "ttft_ms": 11.929625063203275, "input_usage": {"lookups": 28}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 36, "retention_rate": 0.5538461538461539, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.6, "amount": 0.38461538461538464, "status": 1.0, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1152}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 38, "retention_rate": 0.5846153846153846, "retention_by_kind": {"person": 0.8333333333333334, "date": 0.7, "amount": 0.38461538461538464, "status": 1.0, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1179}, "2": {"facts_total": 65, "facts_retained": 36, "retention_rate": 0.5538461538461539, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.6, "amount": 0.38461538461538464, "status": 1.0, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1152}, "3": {"facts_total": 65, "facts_retained": 36, "retention_rate": 0.5538461538461539, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.6, "amount": 0.38461538461538464, "status": 1.0, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1152}, "4": {"facts_total": 65, "facts_retained": 36, "retention_rate": 0.5538461538461539, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.6, "amount": 0.38461538461538464, "status": 1.0, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1152}, "5": {"facts_total": 65, "facts_retained": 36, "retention_rate": 0.5538461538461539, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.6, "amount": 0.38461538461538464, "status": 1.0, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1152}, "6": {"facts_total": 65, "facts_retained": 36, "retention_rate": 0.5538461538461539, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.6, "amount": 0.38461538461538464, "status": 1.0, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1152}, "7": {"facts_total": 65, "facts_retained": 36, "retention_rate": 0.5538461538461539, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.6, "amount": 0.38461538461538464, "status": 1.0, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1152}, "8": {"facts_total": 65, "facts_retained": 36, "retention_rate": 0.5538461538461539, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.6, "amount": 0.38461538461538464, "status": 1.0, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1152}}}, "reference_key_id": "keys/synthetic/syn_03021_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "29d19ef4cb16ed19e33980db8a0baa06391ee0314866e83c8ddf69b297e56071", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "e15e5250ce71ddcc2658ebf3", "attempt_id": "4cfacac919cb", "run_group_id": "E4_core:A4|repairable|qwen3b:h8", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:19:02.150036Z", "task_family": "synthetic", "document_id": "syn_03021_2000", "document_hash": "3ac205b464552356699a4d2b0a7bf62497ead36e5bca73982e11ae5ee3810731", "question_id": "q13_unanswerable", "split": "test", "cluster_id": "syn_03021_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 8, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1914, "query_tokens": 590, "message_tokens": 497, "message_utf8_bytes": 1152, "cache_tensor_bytes": null, "serialized_payload_bytes": 1152, "side_information_bytes": 0, "fetched_bytes": 5222, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 50860.583542031236, "hops": {"1": 6860.111457994208, "2": 6301.548000075854, "3": 6284.723916091025, "4": 6287.547459010966, "5": 6284.2733329162, "6": 6280.639917007647, "7": 6281.036333995871, "8": 6280.703124939464}, "receiver_prefill_ms": 226.6715409932658, "ttft_ms": 11.828834074549377, "decode_ms": 11.788082891143858}, "total_wall_ms": 51111.34495900478, "ttft_ms": 11.828834074549377, "input_usage": {"lookups": 28}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 36, "retention_rate": 0.5538461538461539, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.6, "amount": 0.38461538461538464, "status": 1.0, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1152}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 38, "retention_rate": 0.5846153846153846, "retention_by_kind": {"person": 0.8333333333333334, "date": 0.7, "amount": 0.38461538461538464, "status": 1.0, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1179}, "2": {"facts_total": 65, "facts_retained": 36, "retention_rate": 0.5538461538461539, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.6, "amount": 0.38461538461538464, "status": 1.0, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1152}, "3": {"facts_total": 65, "facts_retained": 36, "retention_rate": 0.5538461538461539, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.6, "amount": 0.38461538461538464, "status": 1.0, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1152}, "4": {"facts_total": 65, "facts_retained": 36, "retention_rate": 0.5538461538461539, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.6, "amount": 0.38461538461538464, "status": 1.0, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1152}, "5": {"facts_total": 65, "facts_retained": 36, "retention_rate": 0.5538461538461539, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.6, "amount": 0.38461538461538464, "status": 1.0, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1152}, "6": {"facts_total": 65, "facts_retained": 36, "retention_rate": 0.5538461538461539, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.6, "amount": 0.38461538461538464, "status": 1.0, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1152}, "7": {"facts_total": 65, "facts_retained": 36, "retention_rate": 0.5538461538461539, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.6, "amount": 0.38461538461538464, "status": 1.0, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1152}, "8": {"facts_total": 65, "facts_retained": 36, "retention_rate": 0.5538461538461539, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.6, "amount": 0.38461538461538464, "status": 1.0, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1152}}}, "reference_key_id": "keys/synthetic/syn_03021_2000.json", "task_scores": {"correct": 1.0, "abstained": 1.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "29d19ef4cb16ed19e33980db8a0baa06391ee0314866e83c8ddf69b297e56071", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "fa40e97d71123f2aa82a925c", "attempt_id": "cb55cfe2065c", "run_group_id": "E4_core:A4|repairable|qwen3b:h8", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:19:02.415622Z", "task_family": "synthetic", "document_id": "syn_03021_2000", "document_hash": "3ac205b464552356699a4d2b0a7bf62497ead36e5bca73982e11ae5ee3810731", "question_id": "q14_unanswerable", "split": "test", "cluster_id": "syn_03021_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 8, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1914, "query_tokens": 592, "message_tokens": 497, "message_utf8_bytes": 1152, "cache_tensor_bytes": null, "serialized_payload_bytes": 1152, "side_information_bytes": 0, "fetched_bytes": 5222, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 50860.583542031236, "hops": {"1": 6860.111457994208, "2": 6301.548000075854, "3": 6284.723916091025, "4": 6287.547459010966, "5": 6284.2733329162, "6": 6280.639917007647, "7": 6281.036333995871, "8": 6280.703124939464}, "receiver_prefill_ms": 226.70029196888208, "ttft_ms": 12.28158304002136, "decode_ms": 12.00774998869747}, "total_wall_ms": 51112.22091712989, "ttft_ms": 12.28158304002136, "input_usage": {"lookups": 28}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 36, "retention_rate": 0.5538461538461539, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.6, "amount": 0.38461538461538464, "status": 1.0, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1152}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 38, "retention_rate": 0.5846153846153846, "retention_by_kind": {"person": 0.8333333333333334, "date": 0.7, "amount": 0.38461538461538464, "status": 1.0, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1179}, "2": {"facts_total": 65, "facts_retained": 36, "retention_rate": 0.5538461538461539, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.6, "amount": 0.38461538461538464, "status": 1.0, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1152}, "3": {"facts_total": 65, "facts_retained": 36, "retention_rate": 0.5538461538461539, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.6, "amount": 0.38461538461538464, "status": 1.0, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1152}, "4": {"facts_total": 65, "facts_retained": 36, "retention_rate": 0.5538461538461539, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.6, "amount": 0.38461538461538464, "status": 1.0, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1152}, "5": {"facts_total": 65, "facts_retained": 36, "retention_rate": 0.5538461538461539, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.6, "amount": 0.38461538461538464, "status": 1.0, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1152}, "6": {"facts_total": 65, "facts_retained": 36, "retention_rate": 0.5538461538461539, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.6, "amount": 0.38461538461538464, "status": 1.0, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1152}, "7": {"facts_total": 65, "facts_retained": 36, "retention_rate": 0.5538461538461539, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.6, "amount": 0.38461538461538464, "status": 1.0, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1152}, "8": {"facts_total": 65, "facts_retained": 36, "retention_rate": 0.5538461538461539, "retention_by_kind": {"person": 0.6666666666666666, "date": 0.6, "amount": 0.38461538461538464, "status": 1.0, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1152}}}, "reference_key_id": "keys/synthetic/syn_03021_2000.json", "task_scores": {"correct": 1.0, "abstained": 1.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "29d19ef4cb16ed19e33980db8a0baa06391ee0314866e83c8ddf69b297e56071", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "7002b67bf9eb2768afd17adb", "attempt_id": "fe8d1440d3d3", "run_group_id": "E4_core:A5|repairable|qwen3b:h1", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:19:09.542305Z", "task_family": "synthetic", "document_id": "syn_03021_2000", "document_hash": "3ac205b464552356699a4d2b0a7bf62497ead36e5bca73982e11ae5ee3810731", "question_id": "q00_single_fact", "split": "test", "cluster_id": "syn_03021_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 1, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1914, "query_tokens": 460, "message_tokens": 379, "message_utf8_bytes": 900, "cache_tensor_bytes": null, "serialized_payload_bytes": 900, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 6858.334750053473, "hops": {"1": 6858.334750053473}, "receiver_prefill_ms": 185.85645803250372, "ttft_ms": 11.812999960966408, "decode_ms": 11.776375002227724}, "total_wall_ms": 7068.241458036937, "ttft_ms": 11.812999960966408, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 19, "retention_rate": 0.2923076923076923, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.15384615384615385, "status": 1.0, "place": 0.0, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 836}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 19, "retention_rate": 0.2923076923076923, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.15384615384615385, "status": 1.0, "place": 0.0, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 836}}}, "reference_key_id": "keys/synthetic/syn_03021_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "2ae8e199245ad716748b3e420d7b2c8e50b7daeeebd965effa93b31557ba8942", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "e66adc333f2567da560f1834", "attempt_id": "c06f66d4b144", "run_group_id": "E4_core:A5|repairable|qwen3b:h1", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:19:09.768114Z", "task_family": "synthetic", "document_id": "syn_03021_2000", "document_hash": "3ac205b464552356699a4d2b0a7bf62497ead36e5bca73982e11ae5ee3810731", "question_id": "q01_single_fact", "split": "test", "cluster_id": "syn_03021_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 1, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1914, "query_tokens": 466, "message_tokens": 379, "message_utf8_bytes": 900, "cache_tensor_bytes": null, "serialized_payload_bytes": 900, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 6858.334750053473, "hops": {"1": 6858.334750053473}, "receiver_prefill_ms": 186.04087503626943, "ttft_ms": 11.82437501847744, "decode_ms": 11.819040984846652}, "total_wall_ms": 7068.433833075687, "ttft_ms": 11.82437501847744, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 19, "retention_rate": 0.2923076923076923, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.15384615384615385, "status": 1.0, "place": 0.0, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 836}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 19, "retention_rate": 0.2923076923076923, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.15384615384615385, "status": 1.0, "place": 0.0, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 836}}}, "reference_key_id": "keys/synthetic/syn_03021_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "2ae8e199245ad716748b3e420d7b2c8e50b7daeeebd965effa93b31557ba8942", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "056fe6d894d6b00f9a6e892f", "attempt_id": "1946bf0fb03a", "run_group_id": "E4_core:A5|repairable|qwen3b:h1", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:19:09.993313Z", "task_family": "synthetic", "document_id": "syn_03021_2000", "document_hash": "3ac205b464552356699a4d2b0a7bf62497ead36e5bca73982e11ae5ee3810731", "question_id": "q02_exact_copy", "split": "test", "cluster_id": "syn_03021_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 1, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1914, "query_tokens": 466, "message_tokens": 379, "message_utf8_bytes": 900, "cache_tensor_bytes": null, "serialized_payload_bytes": 900, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 6858.334750053473, "hops": {"1": 6858.334750053473}, "receiver_prefill_ms": 186.19808298535645, "ttft_ms": 11.798167019151151, "decode_ms": 12.033750070258975}, "total_wall_ms": 7068.761625094339, "ttft_ms": 11.798167019151151, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 19, "retention_rate": 0.2923076923076923, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.15384615384615385, "status": 1.0, "place": 0.0, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 836}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 19, "retention_rate": 0.2923076923076923, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.15384615384615385, "status": 1.0, "place": 0.0, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 836}}}, "reference_key_id": "keys/synthetic/syn_03021_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "2ae8e199245ad716748b3e420d7b2c8e50b7daeeebd965effa93b31557ba8942", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "3db0368256e4ba413236b871", "attempt_id": "15451b002a81", "run_group_id": "E4_core:A5|repairable|qwen3b:h1", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:19:10.218798Z", "task_family": "synthetic", "document_id": "syn_03021_2000", "document_hash": "3ac205b464552356699a4d2b0a7bf62497ead36e5bca73982e11ae5ee3810731", "question_id": "q03_exact_copy", "split": "test", "cluster_id": "syn_03021_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 1, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1914, "query_tokens": 463, "message_tokens": 379, "message_utf8_bytes": 900, "cache_tensor_bytes": null, "serialized_payload_bytes": 900, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 6858.334750053473, "hops": {"1": 6858.334750053473}, "receiver_prefill_ms": 186.48212496191263, "ttft_ms": 11.84791699051857, "decode_ms": 11.852917028591037}, "total_wall_ms": 7068.891209084541, "ttft_ms": 11.84791699051857, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 19, "retention_rate": 0.2923076923076923, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.15384615384615385, "status": 1.0, "place": 0.0, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 836}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 19, "retention_rate": 0.2923076923076923, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.15384615384615385, "status": 1.0, "place": 0.0, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 836}}}, "reference_key_id": "keys/synthetic/syn_03021_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "2ae8e199245ad716748b3e420d7b2c8e50b7daeeebd965effa93b31557ba8942", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "59aba285ef0deddc647f3aff", "attempt_id": "42307132b7d3", "run_group_id": "E4_core:A5|repairable|qwen3b:h1", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:19:10.444118Z", "task_family": "synthetic", "document_id": "syn_03021_2000", "document_hash": "3ac205b464552356699a4d2b0a7bf62497ead36e5bca73982e11ae5ee3810731", "question_id": "q04_multi_hop", "split": "test", "cluster_id": "syn_03021_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 1, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1914, "query_tokens": 470, "message_tokens": 379, "message_utf8_bytes": 900, "cache_tensor_bytes": null, "serialized_payload_bytes": 900, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 6858.334750053473, "hops": {"1": 6858.334750053473}, "receiver_prefill_ms": 186.20070908218622, "ttft_ms": 11.832916992716491, "decode_ms": 11.83708303142339}, "total_wall_ms": 7068.597042001784, "ttft_ms": 11.832916992716491, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 19, "retention_rate": 0.2923076923076923, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.15384615384615385, "status": 1.0, "place": 0.0, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 836}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 19, "retention_rate": 0.2923076923076923, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.15384615384615385, "status": 1.0, "place": 0.0, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 836}}}, "reference_key_id": "keys/synthetic/syn_03021_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "2ae8e199245ad716748b3e420d7b2c8e50b7daeeebd965effa93b31557ba8942", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "75f6d2adb964af62f2497c9f", "attempt_id": "daf9373e657b", "run_group_id": "E4_core:A5|repairable|qwen3b:h1", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:19:10.669199Z", "task_family": "synthetic", "document_id": "syn_03021_2000", "document_hash": "3ac205b464552356699a4d2b0a7bf62497ead36e5bca73982e11ae5ee3810731", "question_id": "q05_multi_hop", "split": "test", "cluster_id": "syn_03021_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 1, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1914, "query_tokens": 471, "message_tokens": 379, "message_utf8_bytes": 900, "cache_tensor_bytes": null, "serialized_payload_bytes": 900, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 6858.334750053473, "hops": {"1": 6858.334750053473}, "receiver_prefill_ms": 186.63504195865244, "ttft_ms": 11.767458054237068, "decode_ms": 117.8340419428423}, "total_wall_ms": 7174.971708096564, "ttft_ms": 11.767458054237068, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 10}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "MH-96F-2099", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 19, "retention_rate": 0.2923076923076923, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.15384615384615385, "status": 1.0, "place": 0.0, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 836}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 19, "retention_rate": 0.2923076923076923, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.15384615384615385, "status": 1.0, "place": 0.0, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 836}}}, "reference_key_id": "keys/synthetic/syn_03021_2000.json", "task_scores": {"correct": 0.0, "abstained": 0.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "2ae8e199245ad716748b3e420d7b2c8e50b7daeeebd965effa93b31557ba8942", "output_hash": "95c230d14a055d30e0e1d4355caf07808e1c968b1feab9e0607d8bd37b1b43dc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "3ff94bbdd1ac08e151c22d66", "attempt_id": "219a3b766dfe", "run_group_id": "E4_core:A5|repairable|qwen3b:h1", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:19:11.000924Z", "task_family": "synthetic", "document_id": "syn_03021_2000", "document_hash": "3ac205b464552356699a4d2b0a7bf62497ead36e5bca73982e11ae5ee3810731", "question_id": "q06_negation", "split": "test", "cluster_id": "syn_03021_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 1, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1914, "query_tokens": 461, "message_tokens": 379, "message_utf8_bytes": 900, "cache_tensor_bytes": null, "serialized_payload_bytes": 900, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 6858.334750053473, "hops": {"1": 6858.334750053473}, "receiver_prefill_ms": 186.40212493482977, "ttft_ms": 11.827416019514203, "decode_ms": 11.863624909892678}, "total_wall_ms": 7068.831958109513, "ttft_ms": 11.827416019514203, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "no", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 19, "retention_rate": 0.2923076923076923, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.15384615384615385, "status": 1.0, "place": 0.0, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 836}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 19, "retention_rate": 0.2923076923076923, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.15384615384615385, "status": 1.0, "place": 0.0, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 836}}}, "reference_key_id": "keys/synthetic/syn_03021_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "2ae8e199245ad716748b3e420d7b2c8e50b7daeeebd965effa93b31557ba8942", "output_hash": "9390298f3fb0c5b160498935d79cb139aef28e1c47358b4bbba61862b9c26e59"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "018ba3cc8c9a8cee768e2a80", "attempt_id": "b32d7dec6bc5", "run_group_id": "E4_core:A5|repairable|qwen3b:h1", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:19:11.226520Z", "task_family": "synthetic", "document_id": "syn_03021_2000", "document_hash": "3ac205b464552356699a4d2b0a7bf62497ead36e5bca73982e11ae5ee3810731", "question_id": "q07_negation", "split": "test", "cluster_id": "syn_03021_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 1, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1914, "query_tokens": 462, "message_tokens": 379, "message_utf8_bytes": 900, "cache_tensor_bytes": null, "serialized_payload_bytes": 900, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 6858.334750053473, "hops": {"1": 6858.334750053473}, "receiver_prefill_ms": 186.41974998172373, "ttft_ms": 11.842958047054708, "decode_ms": 11.901790974661708}, "total_wall_ms": 7068.881749990396, "ttft_ms": 11.842958047054708, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "no", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 19, "retention_rate": 0.2923076923076923, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.15384615384615385, "status": 1.0, "place": 0.0, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 836}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 19, "retention_rate": 0.2923076923076923, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.15384615384615385, "status": 1.0, "place": 0.0, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 836}}}, "reference_key_id": "keys/synthetic/syn_03021_2000.json", "task_scores": {"correct": 0.0, "abstained": 0.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "2ae8e199245ad716748b3e420d7b2c8e50b7daeeebd965effa93b31557ba8942", "output_hash": "9390298f3fb0c5b160498935d79cb139aef28e1c47358b4bbba61862b9c26e59"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "277d6954f787cd5d446f40a6", "attempt_id": "c6f0c2be24bd", "run_group_id": "E4_core:A5|repairable|qwen3b:h1", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:19:11.452007Z", "task_family": "synthetic", "document_id": "syn_03021_2000", "document_hash": "3ac205b464552356699a4d2b0a7bf62497ead36e5bca73982e11ae5ee3810731", "question_id": "q08_revision", "split": "test", "cluster_id": "syn_03021_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 1, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1914, "query_tokens": 478, "message_tokens": 379, "message_utf8_bytes": 900, "cache_tensor_bytes": null, "serialized_payload_bytes": 900, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 6858.334750053473, "hops": {"1": 6858.334750053473}, "receiver_prefill_ms": 186.70491699595004, "ttft_ms": 11.804500012658536, "decode_ms": 11.74224994610995}, "total_wall_ms": 7068.991417065263, "ttft_ms": 11.804500012658536, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 19, "retention_rate": 0.2923076923076923, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.15384615384615385, "status": 1.0, "place": 0.0, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 836}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 19, "retention_rate": 0.2923076923076923, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.15384615384615385, "status": 1.0, "place": 0.0, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 836}}}, "reference_key_id": "keys/synthetic/syn_03021_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "2ae8e199245ad716748b3e420d7b2c8e50b7daeeebd965effa93b31557ba8942", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "0b7f796434eadb85d1e52c8d", "attempt_id": "4d0a28b5c62d", "run_group_id": "E4_core:A5|repairable|qwen3b:h1", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:19:11.677314Z", "task_family": "synthetic", "document_id": "syn_03021_2000", "document_hash": "3ac205b464552356699a4d2b0a7bf62497ead36e5bca73982e11ae5ee3810731", "question_id": "q09_revision", "split": "test", "cluster_id": "syn_03021_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 1, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1914, "query_tokens": 469, "message_tokens": 379, "message_utf8_bytes": 900, "cache_tensor_bytes": null, "serialized_payload_bytes": 900, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 6858.334750053473, "hops": {"1": 6858.334750053473}, "receiver_prefill_ms": 185.86433306336403, "ttft_ms": 11.796500068157911, "decode_ms": 82.50654197763652}, "total_wall_ms": 7138.8948750682175, "ttft_ms": 11.796500068157911, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 7}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "5646720", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 19, "retention_rate": 0.2923076923076923, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.15384615384615385, "status": 1.0, "place": 0.0, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 836}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 19, "retention_rate": 0.2923076923076923, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.15384615384615385, "status": 1.0, "place": 0.0, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 836}}}, "reference_key_id": "keys/synthetic/syn_03021_2000.json", "task_scores": {"correct": 0.0, "abstained": 0.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "2ae8e199245ad716748b3e420d7b2c8e50b7daeeebd965effa93b31557ba8942", "output_hash": "e1d129e860627cf4623671ed186cb4ebbf5d63169d3d0dc14a62e1da2af61326"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "f2453af5a0ddc4ce71907a99", "attempt_id": "6955e20a5988", "run_group_id": "E4_core:A5|repairable|qwen3b:h1", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:19:11.972728Z", "task_family": "synthetic", "document_id": "syn_03021_2000", "document_hash": "3ac205b464552356699a4d2b0a7bf62497ead36e5bca73982e11ae5ee3810731", "question_id": "q10_temporal", "split": "test", "cluster_id": "syn_03021_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 1, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1914, "query_tokens": 477, "message_tokens": 379, "message_utf8_bytes": 900, "cache_tensor_bytes": null, "serialized_payload_bytes": 900, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 6858.334750053473, "hops": {"1": 6858.334750053473}, "receiver_prefill_ms": 186.30929209757596, "ttft_ms": 11.884625069797039, "decode_ms": 58.9661659905687}, "total_wall_ms": 7115.869125118479, "ttft_ms": 11.884625069797039, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 5}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "Kovar-9", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 19, "retention_rate": 0.2923076923076923, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.15384615384615385, "status": 1.0, "place": 0.0, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 836}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 19, "retention_rate": 0.2923076923076923, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.15384615384615385, "status": 1.0, "place": 0.0, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 836}}}, "reference_key_id": "keys/synthetic/syn_03021_2000.json", "task_scores": {"correct": 0.0, "abstained": 0.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "2ae8e199245ad716748b3e420d7b2c8e50b7daeeebd965effa93b31557ba8942", "output_hash": "c969b5f7547945067f5b07e9f822d0043126326f979d950a8b1b045057fbb818"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "f836d7809ff3f93d6c0488a1", "attempt_id": "d2acd35d4fdb", "run_group_id": "E4_core:A5|repairable|qwen3b:h1", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:19:12.244916Z", "task_family": "synthetic", "document_id": "syn_03021_2000", "document_hash": "3ac205b464552356699a4d2b0a7bf62497ead36e5bca73982e11ae5ee3810731", "question_id": "q11_aggregation", "split": "test", "cluster_id": "syn_03021_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 1, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1914, "query_tokens": 473, "message_tokens": 379, "message_utf8_bytes": 900, "cache_tensor_bytes": null, "serialized_payload_bytes": 900, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 6858.334750053473, "hops": {"1": 6858.334750053473}, "receiver_prefill_ms": 186.15483306348324, "ttft_ms": 11.910791974514723, "decode_ms": 82.62462494894862}, "total_wall_ms": 7139.431458082981, "ttft_ms": 11.910791974514723, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 7}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "9484825", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 19, "retention_rate": 0.2923076923076923, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.15384615384615385, "status": 1.0, "place": 0.0, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 836}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 19, "retention_rate": 0.2923076923076923, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.15384615384615385, "status": 1.0, "place": 0.0, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 836}}}, "reference_key_id": "keys/synthetic/syn_03021_2000.json", "task_scores": {"correct": 0.0, "abstained": 0.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "2ae8e199245ad716748b3e420d7b2c8e50b7daeeebd965effa93b31557ba8942", "output_hash": "667b970507c4c6c7fb572757eb3b5eb04186ddb67a6ef680266b7a0924737993"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "7b5d98d0561e060d6ec4e5f7", "attempt_id": "30aedc46933a", "run_group_id": "E4_core:A5|repairable|qwen3b:h1", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:19:12.541145Z", "task_family": "synthetic", "document_id": "syn_03021_2000", "document_hash": "3ac205b464552356699a4d2b0a7bf62497ead36e5bca73982e11ae5ee3810731", "question_id": "q12_unit_conversion", "split": "test", "cluster_id": "syn_03021_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 1, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1914, "query_tokens": 468, "message_tokens": 379, "message_utf8_bytes": 900, "cache_tensor_bytes": null, "serialized_payload_bytes": 900, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 6858.334750053473, "hops": {"1": 6858.334750053473}, "receiver_prefill_ms": 185.08358299732208, "ttft_ms": 11.85133308172226, "decode_ms": 47.42800001986325}, "total_wall_ms": 7103.088417090476, "ttft_ms": 11.85133308172226, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 4}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "22.3", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 19, "retention_rate": 0.2923076923076923, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.15384615384615385, "status": 1.0, "place": 0.0, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 836}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 19, "retention_rate": 0.2923076923076923, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.15384615384615385, "status": 1.0, "place": 0.0, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 836}}}, "reference_key_id": "keys/synthetic/syn_03021_2000.json", "task_scores": {"correct": 0.0, "abstained": 0.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "2ae8e199245ad716748b3e420d7b2c8e50b7daeeebd965effa93b31557ba8942", "output_hash": "8afb0bb35e730fd32e7805d65081241bd21f43709df37165f4ec91c763f8a99d"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "f88746fbc36c8068ed7d1ffa", "attempt_id": "48a2686128e3", "run_group_id": "E4_core:A5|repairable|qwen3b:h1", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:19:12.800718Z", "task_family": "synthetic", "document_id": "syn_03021_2000", "document_hash": "3ac205b464552356699a4d2b0a7bf62497ead36e5bca73982e11ae5ee3810731", "question_id": "q13_unanswerable", "split": "test", "cluster_id": "syn_03021_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 1, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1914, "query_tokens": 473, "message_tokens": 379, "message_utf8_bytes": 900, "cache_tensor_bytes": null, "serialized_payload_bytes": 900, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 6858.334750053473, "hops": {"1": 6858.334750053473}, "receiver_prefill_ms": 186.2459999974817, "ttft_ms": 11.850791983306408, "decode_ms": 11.878332938067615}, "total_wall_ms": 7068.681875127368, "ttft_ms": 11.850791983306408, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 19, "retention_rate": 0.2923076923076923, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.15384615384615385, "status": 1.0, "place": 0.0, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 836}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 19, "retention_rate": 0.2923076923076923, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.15384615384615385, "status": 1.0, "place": 0.0, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 836}}}, "reference_key_id": "keys/synthetic/syn_03021_2000.json", "task_scores": {"correct": 1.0, "abstained": 1.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "2ae8e199245ad716748b3e420d7b2c8e50b7daeeebd965effa93b31557ba8942", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "b5323b1dbad8832a4b42b95b", "attempt_id": "4d364c2e7123", "run_group_id": "E4_core:A5|repairable|qwen3b:h1", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:19:13.025886Z", "task_family": "synthetic", "document_id": "syn_03021_2000", "document_hash": "3ac205b464552356699a4d2b0a7bf62497ead36e5bca73982e11ae5ee3810731", "question_id": "q14_unanswerable", "split": "test", "cluster_id": "syn_03021_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 1, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1914, "query_tokens": 475, "message_tokens": 379, "message_utf8_bytes": 900, "cache_tensor_bytes": null, "serialized_payload_bytes": 900, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 6858.334750053473, "hops": {"1": 6858.334750053473}, "receiver_prefill_ms": 186.4105419954285, "ttft_ms": 11.826083064079285, "decode_ms": 11.851917020976543}, "total_wall_ms": 7068.8043751288205, "ttft_ms": 11.826083064079285, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 19, "retention_rate": 0.2923076923076923, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.15384615384615385, "status": 1.0, "place": 0.0, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 836}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 19, "retention_rate": 0.2923076923076923, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.15384615384615385, "status": 1.0, "place": 0.0, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 836}}}, "reference_key_id": "keys/synthetic/syn_03021_2000.json", "task_scores": {"correct": 1.0, "abstained": 1.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "2ae8e199245ad716748b3e420d7b2c8e50b7daeeebd965effa93b31557ba8942", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "4e6cbc70328e25c8357cf35b", "attempt_id": "256aa5c2e015", "run_group_id": "E4_core:A5|repairable|qwen3b:h2", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:19:18.228976Z", "task_family": "synthetic", "document_id": "syn_03021_2000", "document_hash": "3ac205b464552356699a4d2b0a7bf62497ead36e5bca73982e11ae5ee3810731", "question_id": "q00_single_fact", "split": "test", "cluster_id": "syn_03021_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 2, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1914, "query_tokens": 469, "message_tokens": 389, "message_utf8_bytes": 931, "cache_tensor_bytes": null, "serialized_payload_bytes": 931, "side_information_bytes": 0, "fetched_bytes": 497, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 11834.982667001896, "hops": {"1": 6858.334750053473, "2": 4976.647916948423}, "receiver_prefill_ms": 186.03099999018013, "ttft_ms": 11.82962500024587, "decode_ms": 11.953708017244935}, "total_wall_ms": 12045.24783405941, "ttft_ms": 11.82962500024587, "input_usage": {"lookups": 1}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 19, "retention_rate": 0.2923076923076923, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.15384615384615385, "status": 1.0, "place": 0.0, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 867}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 19, "retention_rate": 0.2923076923076923, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.15384615384615385, "status": 1.0, "place": 0.0, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 836}, "2": {"facts_total": 65, "facts_retained": 19, "retention_rate": 0.2923076923076923, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.15384615384615385, "status": 1.0, "place": 0.0, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 867}}}, "reference_key_id": "keys/synthetic/syn_03021_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "92d15207bf13f35ba83fc20f1408c2fcf07fd4b0ed40fc18f51a5580586b6137", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "7d2c180a1c7770a34353ebea", "attempt_id": "1b45fcde11d4", "run_group_id": "E4_core:A5|repairable|qwen3b:h2", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:19:18.455315Z", "task_family": "synthetic", "document_id": "syn_03021_2000", "document_hash": "3ac205b464552356699a4d2b0a7bf62497ead36e5bca73982e11ae5ee3810731", "question_id": "q01_single_fact", "split": "test", "cluster_id": "syn_03021_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 2, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1914, "query_tokens": 475, "message_tokens": 389, "message_utf8_bytes": 931, "cache_tensor_bytes": null, "serialized_payload_bytes": 931, "side_information_bytes": 0, "fetched_bytes": 497, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 11834.982667001896, "hops": {"1": 6858.334750053473, "2": 4976.647916948423}, "receiver_prefill_ms": 185.98220800049603, "ttft_ms": 11.92508393432945, "decode_ms": 11.82125008199364}, "total_wall_ms": 12045.16641702503, "ttft_ms": 11.92508393432945, "input_usage": {"lookups": 1}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 19, "retention_rate": 0.2923076923076923, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.15384615384615385, "status": 1.0, "place": 0.0, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 867}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 19, "retention_rate": 0.2923076923076923, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.15384615384615385, "status": 1.0, "place": 0.0, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 836}, "2": {"facts_total": 65, "facts_retained": 19, "retention_rate": 0.2923076923076923, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.15384615384615385, "status": 1.0, "place": 0.0, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 867}}}, "reference_key_id": "keys/synthetic/syn_03021_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "92d15207bf13f35ba83fc20f1408c2fcf07fd4b0ed40fc18f51a5580586b6137", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "4065a27e12390281dc1a284d", "attempt_id": "8d1669b73dd2", "run_group_id": "E4_core:A5|repairable|qwen3b:h2", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:19:18.680748Z", "task_family": "synthetic", "document_id": "syn_03021_2000", "document_hash": "3ac205b464552356699a4d2b0a7bf62497ead36e5bca73982e11ae5ee3810731", "question_id": "q02_exact_copy", "split": "test", "cluster_id": "syn_03021_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 2, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1914, "query_tokens": 475, "message_tokens": 389, "message_utf8_bytes": 931, "cache_tensor_bytes": null, "serialized_payload_bytes": 931, "side_information_bytes": 0, "fetched_bytes": 497, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 11834.982667001896, "hops": {"1": 6858.334750053473, "2": 4976.647916948423}, "receiver_prefill_ms": 186.37154099997133, "ttft_ms": 11.898875003680587, "decode_ms": 11.90529193263501}, "total_wall_ms": 12045.558709069155, "ttft_ms": 11.898875003680587, "input_usage": {"lookups": 1}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 19, "retention_rate": 0.2923076923076923, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.15384615384615385, "status": 1.0, "place": 0.0, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 867}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 19, "retention_rate": 0.2923076923076923, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.15384615384615385, "status": 1.0, "place": 0.0, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 836}, "2": {"facts_total": 65, "facts_retained": 19, "retention_rate": 0.2923076923076923, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.15384615384615385, "status": 1.0, "place": 0.0, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 867}}}, "reference_key_id": "keys/synthetic/syn_03021_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "92d15207bf13f35ba83fc20f1408c2fcf07fd4b0ed40fc18f51a5580586b6137", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "c1aaf0deef0986db8d193e60", "attempt_id": "3cd5be4201c1", "run_group_id": "E4_core:A5|repairable|qwen3b:h2", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:19:18.906734Z", "task_family": "synthetic", "document_id": "syn_03021_2000", "document_hash": "3ac205b464552356699a4d2b0a7bf62497ead36e5bca73982e11ae5ee3810731", "question_id": "q03_exact_copy", "split": "test", "cluster_id": "syn_03021_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 2, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1914, "query_tokens": 472, "message_tokens": 389, "message_utf8_bytes": 931, "cache_tensor_bytes": null, "serialized_payload_bytes": 931, "side_information_bytes": 0, "fetched_bytes": 497, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 11834.982667001896, "hops": {"1": 6858.334750053473, "2": 4976.647916948423}, "receiver_prefill_ms": 186.2717499025166, "ttft_ms": 11.856916011311114, "decode_ms": 11.921083088964224}, "total_wall_ms": 12045.455584069714, "ttft_ms": 11.856916011311114, "input_usage": {"lookups": 1}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 19, "retention_rate": 0.2923076923076923, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.15384615384615385, "status": 1.0, "place": 0.0, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 867}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 19, "retention_rate": 0.2923076923076923, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.15384615384615385, "status": 1.0, "place": 0.0, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 836}, "2": {"facts_total": 65, "facts_retained": 19, "retention_rate": 0.2923076923076923, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.15384615384615385, "status": 1.0, "place": 0.0, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 867}}}, "reference_key_id": "keys/synthetic/syn_03021_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "92d15207bf13f35ba83fc20f1408c2fcf07fd4b0ed40fc18f51a5580586b6137", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "252e4c859c4a0d299f7534e9", "attempt_id": "6deedad5e522", "run_group_id": "E4_core:A5|repairable|qwen3b:h2", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:19:19.131823Z", "task_family": "synthetic", "document_id": "syn_03021_2000", "document_hash": "3ac205b464552356699a4d2b0a7bf62497ead36e5bca73982e11ae5ee3810731", "question_id": "q04_multi_hop", "split": "test", "cluster_id": "syn_03021_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 2, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1914, "query_tokens": 479, "message_tokens": 389, "message_utf8_bytes": 931, "cache_tensor_bytes": null, "serialized_payload_bytes": 931, "side_information_bytes": 0, "fetched_bytes": 497, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 11834.982667001896, "hops": {"1": 6858.334750053473, "2": 4976.647916948423}, "receiver_prefill_ms": 186.42104207538068, "ttft_ms": 11.926540988497436, "decode_ms": 11.951540946029127}, "total_wall_ms": 12045.738041982986, "ttft_ms": 11.926540988497436, "input_usage": {"lookups": 1}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 19, "retention_rate": 0.2923076923076923, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.15384615384615385, "status": 1.0, "place": 0.0, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 867}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 19, "retention_rate": 0.2923076923076923, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.15384615384615385, "status": 1.0, "place": 0.0, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 836}, "2": {"facts_total": 65, "facts_retained": 19, "retention_rate": 0.2923076923076923, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.15384615384615385, "status": 1.0, "place": 0.0, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 867}}}, "reference_key_id": "keys/synthetic/syn_03021_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "92d15207bf13f35ba83fc20f1408c2fcf07fd4b0ed40fc18f51a5580586b6137", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "a21420be37a7f74a720ff260", "attempt_id": "f5dca43f25c8", "run_group_id": "E4_core:A5|repairable|qwen3b:h2", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:19:19.357510Z", "task_family": "synthetic", "document_id": "syn_03021_2000", "document_hash": "3ac205b464552356699a4d2b0a7bf62497ead36e5bca73982e11ae5ee3810731", "question_id": "q05_multi_hop", "split": "test", "cluster_id": "syn_03021_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 2, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1914, "query_tokens": 480, "message_tokens": 389, "message_utf8_bytes": 931, "cache_tensor_bytes": null, "serialized_payload_bytes": 931, "side_information_bytes": 0, "fetched_bytes": 497, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 11834.982667001896, "hops": {"1": 6858.334750053473, "2": 4976.647916948423}, "receiver_prefill_ms": 186.9509998941794, "ttft_ms": 11.850583949126303, "decode_ms": 11.91483298316598}, "total_wall_ms": 12046.148416935466, "ttft_ms": 11.850583949126303, "input_usage": {"lookups": 1}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 19, "retention_rate": 0.2923076923076923, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.15384615384615385, "status": 1.0, "place": 0.0, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 867}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 19, "retention_rate": 0.2923076923076923, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.15384615384615385, "status": 1.0, "place": 0.0, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 836}, "2": {"facts_total": 65, "facts_retained": 19, "retention_rate": 0.2923076923076923, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.15384615384615385, "status": 1.0, "place": 0.0, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 867}}}, "reference_key_id": "keys/synthetic/syn_03021_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "92d15207bf13f35ba83fc20f1408c2fcf07fd4b0ed40fc18f51a5580586b6137", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "94c80fa0c3585f8f9a6a4371", "attempt_id": "06c04ad86cff", "run_group_id": "E4_core:A5|repairable|qwen3b:h2", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:19:19.583359Z", "task_family": "synthetic", "document_id": "syn_03021_2000", "document_hash": "3ac205b464552356699a4d2b0a7bf62497ead36e5bca73982e11ae5ee3810731", "question_id": "q06_negation", "split": "test", "cluster_id": "syn_03021_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 2, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1914, "query_tokens": 470, "message_tokens": 389, "message_utf8_bytes": 931, "cache_tensor_bytes": null, "serialized_payload_bytes": 931, "side_information_bytes": 0, "fetched_bytes": 497, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 11834.982667001896, "hops": {"1": 6858.334750053473, "2": 4976.647916948423}, "receiver_prefill_ms": 185.8234580140561, "ttft_ms": 11.889416025951505, "decode_ms": 11.819042032584548}, "total_wall_ms": 12044.914458063431, "ttft_ms": 11.889416025951505, "input_usage": {"lookups": 1}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "no", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 19, "retention_rate": 0.2923076923076923, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.15384615384615385, "status": 1.0, "place": 0.0, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 867}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 19, "retention_rate": 0.2923076923076923, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.15384615384615385, "status": 1.0, "place": 0.0, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 836}, "2": {"facts_total": 65, "facts_retained": 19, "retention_rate": 0.2923076923076923, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.15384615384615385, "status": 1.0, "place": 0.0, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 867}}}, "reference_key_id": "keys/synthetic/syn_03021_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "92d15207bf13f35ba83fc20f1408c2fcf07fd4b0ed40fc18f51a5580586b6137", "output_hash": "9390298f3fb0c5b160498935d79cb139aef28e1c47358b4bbba61862b9c26e59"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "13920be4e8b3077047f4bcb1", "attempt_id": "ca2e4b7fb25e", "run_group_id": "E4_core:A5|repairable|qwen3b:h2", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:19:19.808278Z", "task_family": "synthetic", "document_id": "syn_03021_2000", "document_hash": "3ac205b464552356699a4d2b0a7bf62497ead36e5bca73982e11ae5ee3810731", "question_id": "q07_negation", "split": "test", "cluster_id": "syn_03021_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 2, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1914, "query_tokens": 471, "message_tokens": 389, "message_utf8_bytes": 931, "cache_tensor_bytes": null, "serialized_payload_bytes": 931, "side_information_bytes": 0, "fetched_bytes": 497, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 11834.982667001896, "hops": {"1": 6858.334750053473, "2": 4976.647916948423}, "receiver_prefill_ms": 185.40650000795722, "ttft_ms": 11.837625061161816, "decode_ms": 12.097875005565584}, "total_wall_ms": 12044.712999952026, "ttft_ms": 11.837625061161816, "input_usage": {"lookups": 1}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "no", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 19, "retention_rate": 0.2923076923076923, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.15384615384615385, "status": 1.0, "place": 0.0, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 867}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 19, "retention_rate": 0.2923076923076923, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.15384615384615385, "status": 1.0, "place": 0.0, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 836}, "2": {"facts_total": 65, "facts_retained": 19, "retention_rate": 0.2923076923076923, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.15384615384615385, "status": 1.0, "place": 0.0, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 867}}}, "reference_key_id": "keys/synthetic/syn_03021_2000.json", "task_scores": {"correct": 0.0, "abstained": 0.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "92d15207bf13f35ba83fc20f1408c2fcf07fd4b0ed40fc18f51a5580586b6137", "output_hash": "9390298f3fb0c5b160498935d79cb139aef28e1c47358b4bbba61862b9c26e59"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "f1e1bde8d469b9511959f7dd", "attempt_id": "3b5ca0e31235", "run_group_id": "E4_core:A5|repairable|qwen3b:h2", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:19:20.032809Z", "task_family": "synthetic", "document_id": "syn_03021_2000", "document_hash": "3ac205b464552356699a4d2b0a7bf62497ead36e5bca73982e11ae5ee3810731", "question_id": "q08_revision", "split": "test", "cluster_id": "syn_03021_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 2, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1914, "query_tokens": 487, "message_tokens": 389, "message_utf8_bytes": 931, "cache_tensor_bytes": null, "serialized_payload_bytes": 931, "side_information_bytes": 0, "fetched_bytes": 497, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 11834.982667001896, "hops": {"1": 6858.334750053473, "2": 4976.647916948423}, "receiver_prefill_ms": 187.7643340267241, "ttft_ms": 12.007332989014685, "decode_ms": 11.868707952089608}, "total_wall_ms": 12047.183291986585, "ttft_ms": 12.007332989014685, "input_usage": {"lookups": 1}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 19, "retention_rate": 0.2923076923076923, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.15384615384615385, "status": 1.0, "place": 0.0, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 867}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 19, "retention_rate": 0.2923076923076923, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.15384615384615385, "status": 1.0, "place": 0.0, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 836}, "2": {"facts_total": 65, "facts_retained": 19, "retention_rate": 0.2923076923076923, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.15384615384615385, "status": 1.0, "place": 0.0, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 867}}}, "reference_key_id": "keys/synthetic/syn_03021_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "92d15207bf13f35ba83fc20f1408c2fcf07fd4b0ed40fc18f51a5580586b6137", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "626fc8db67d1025bfd518da4", "attempt_id": "185429d9d04f", "run_group_id": "E4_core:A5|repairable|qwen3b:h2", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:19:20.259865Z", "task_family": "synthetic", "document_id": "syn_03021_2000", "document_hash": "3ac205b464552356699a4d2b0a7bf62497ead36e5bca73982e11ae5ee3810731", "question_id": "q09_revision", "split": "test", "cluster_id": "syn_03021_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 2, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1914, "query_tokens": 478, "message_tokens": 389, "message_utf8_bytes": 931, "cache_tensor_bytes": null, "serialized_payload_bytes": 931, "side_information_bytes": 0, "fetched_bytes": 497, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 11834.982667001896, "hops": {"1": 6858.334750053473, "2": 4976.647916948423}, "receiver_prefill_ms": 186.68425001669675, "ttft_ms": 12.029709061607718, "decode_ms": 11.898832977749407}, "total_wall_ms": 12046.112125040963, "ttft_ms": 12.029709061607718, "input_usage": {"lookups": 1}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 19, "retention_rate": 0.2923076923076923, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.15384615384615385, "status": 1.0, "place": 0.0, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 867}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 19, "retention_rate": 0.2923076923076923, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.15384615384615385, "status": 1.0, "place": 0.0, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 836}, "2": {"facts_total": 65, "facts_retained": 19, "retention_rate": 0.2923076923076923, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.15384615384615385, "status": 1.0, "place": 0.0, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 867}}}, "reference_key_id": "keys/synthetic/syn_03021_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "92d15207bf13f35ba83fc20f1408c2fcf07fd4b0ed40fc18f51a5580586b6137", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "fd4d061eb9a4a8b88d58adca", "attempt_id": "4381706f570f", "run_group_id": "E4_core:A5|repairable|qwen3b:h2", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:19:20.487599Z", "task_family": "synthetic", "document_id": "syn_03021_2000", "document_hash": "3ac205b464552356699a4d2b0a7bf62497ead36e5bca73982e11ae5ee3810731", "question_id": "q10_temporal", "split": "test", "cluster_id": "syn_03021_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 2, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1914, "query_tokens": 486, "message_tokens": 389, "message_utf8_bytes": 931, "cache_tensor_bytes": null, "serialized_payload_bytes": 931, "side_information_bytes": 0, "fetched_bytes": 497, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 11834.982667001896, "hops": {"1": 6858.334750053473, "2": 4976.647916948423}, "receiver_prefill_ms": 188.31787491217256, "ttft_ms": 11.744957999326289, "decode_ms": 59.13312500342727}, "total_wall_ms": 12094.612958026119, "ttft_ms": 11.744957999326289, "input_usage": {"lookups": 1}, "output_usage": {"output_tokens": 5}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "Kovar-9", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 19, "retention_rate": 0.2923076923076923, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.15384615384615385, "status": 1.0, "place": 0.0, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 867}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 19, "retention_rate": 0.2923076923076923, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.15384615384615385, "status": 1.0, "place": 0.0, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 836}, "2": {"facts_total": 65, "facts_retained": 19, "retention_rate": 0.2923076923076923, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.15384615384615385, "status": 1.0, "place": 0.0, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 867}}}, "reference_key_id": "keys/synthetic/syn_03021_2000.json", "task_scores": {"correct": 0.0, "abstained": 0.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "92d15207bf13f35ba83fc20f1408c2fcf07fd4b0ed40fc18f51a5580586b6137", "output_hash": "c969b5f7547945067f5b07e9f822d0043126326f979d950a8b1b045057fbb818"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "bcbac3d81b7e97b6b137539c", "attempt_id": "aba4e720a3a9", "run_group_id": "E4_core:A5|repairable|qwen3b:h2", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:19:20.762724Z", "task_family": "synthetic", "document_id": "syn_03021_2000", "document_hash": "3ac205b464552356699a4d2b0a7bf62497ead36e5bca73982e11ae5ee3810731", "question_id": "q11_aggregation", "split": "test", "cluster_id": "syn_03021_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 2, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1914, "query_tokens": 482, "message_tokens": 389, "message_utf8_bytes": 931, "cache_tensor_bytes": null, "serialized_payload_bytes": 931, "side_information_bytes": 0, "fetched_bytes": 497, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 11834.982667001896, "hops": {"1": 6858.334750053473, "2": 4976.647916948423}, "receiver_prefill_ms": 184.85312501434237, "ttft_ms": 11.861666920594871, "decode_ms": 71.2572910124436}, "total_wall_ms": 12103.36525097955, "ttft_ms": 11.861666920594871, "input_usage": {"lookups": 1}, "output_usage": {"output_tokens": 6}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "948485", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 19, "retention_rate": 0.2923076923076923, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.15384615384615385, "status": 1.0, "place": 0.0, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 867}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 19, "retention_rate": 0.2923076923076923, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.15384615384615385, "status": 1.0, "place": 0.0, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 836}, "2": {"facts_total": 65, "facts_retained": 19, "retention_rate": 0.2923076923076923, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.15384615384615385, "status": 1.0, "place": 0.0, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 867}}}, "reference_key_id": "keys/synthetic/syn_03021_2000.json", "task_scores": {"correct": 0.0, "abstained": 0.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "92d15207bf13f35ba83fc20f1408c2fcf07fd4b0ed40fc18f51a5580586b6137", "output_hash": "244f716d260fa5f07cbedd5d728983a25a4991af5624be73b358ad829974e9c4"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "f3de17234aeea84dd5b95577", "attempt_id": "c15b8c38d8ec", "run_group_id": "E4_core:A5|repairable|qwen3b:h2", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:19:21.046208Z", "task_family": "synthetic", "document_id": "syn_03021_2000", "document_hash": "3ac205b464552356699a4d2b0a7bf62497ead36e5bca73982e11ae5ee3810731", "question_id": "q12_unit_conversion", "split": "test", "cluster_id": "syn_03021_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 2, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1914, "query_tokens": 477, "message_tokens": 389, "message_utf8_bytes": 931, "cache_tensor_bytes": null, "serialized_payload_bytes": 931, "side_information_bytes": 0, "fetched_bytes": 497, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 11834.982667001896, "hops": {"1": 6858.334750053473, "2": 4976.647916948423}, "receiver_prefill_ms": 186.66945898439735, "ttft_ms": 11.865791049785912, "decode_ms": 47.31558298226446}, "total_wall_ms": 12081.220626016147, "ttft_ms": 11.865791049785912, "input_usage": {"lookups": 1}, "output_usage": {"output_tokens": 4}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "22.3", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 19, "retention_rate": 0.2923076923076923, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.15384615384615385, "status": 1.0, "place": 0.0, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 867}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 19, "retention_rate": 0.2923076923076923, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.15384615384615385, "status": 1.0, "place": 0.0, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 836}, "2": {"facts_total": 65, "facts_retained": 19, "retention_rate": 0.2923076923076923, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.15384615384615385, "status": 1.0, "place": 0.0, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 867}}}, "reference_key_id": "keys/synthetic/syn_03021_2000.json", "task_scores": {"correct": 0.0, "abstained": 0.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "92d15207bf13f35ba83fc20f1408c2fcf07fd4b0ed40fc18f51a5580586b6137", "output_hash": "8afb0bb35e730fd32e7805d65081241bd21f43709df37165f4ec91c763f8a99d"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "a81ec4ec9bd3a601e58289fe", "attempt_id": "40dbfa1c2926", "run_group_id": "E4_core:A5|repairable|qwen3b:h2", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:19:21.307238Z", "task_family": "synthetic", "document_id": "syn_03021_2000", "document_hash": "3ac205b464552356699a4d2b0a7bf62497ead36e5bca73982e11ae5ee3810731", "question_id": "q13_unanswerable", "split": "test", "cluster_id": "syn_03021_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 2, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1914, "query_tokens": 482, "message_tokens": 389, "message_utf8_bytes": 931, "cache_tensor_bytes": null, "serialized_payload_bytes": 931, "side_information_bytes": 0, "fetched_bytes": 497, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 11834.982667001896, "hops": {"1": 6858.334750053473, "2": 4976.647916948423}, "receiver_prefill_ms": 183.93512489274144, "ttft_ms": 11.840457911603153, "decode_ms": 11.890291003510356}, "total_wall_ms": 12043.01712603774, "ttft_ms": 11.840457911603153, "input_usage": {"lookups": 1}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 19, "retention_rate": 0.2923076923076923, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.15384615384615385, "status": 1.0, "place": 0.0, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 867}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 19, "retention_rate": 0.2923076923076923, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.15384615384615385, "status": 1.0, "place": 0.0, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 836}, "2": {"facts_total": 65, "facts_retained": 19, "retention_rate": 0.2923076923076923, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.15384615384615385, "status": 1.0, "place": 0.0, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 867}}}, "reference_key_id": "keys/synthetic/syn_03021_2000.json", "task_scores": {"correct": 1.0, "abstained": 1.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "92d15207bf13f35ba83fc20f1408c2fcf07fd4b0ed40fc18f51a5580586b6137", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "b069866f2a9d4a0469f8926e", "attempt_id": "d95444748c58", "run_group_id": "E4_core:A5|repairable|qwen3b:h2", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:19:21.529885Z", "task_family": "synthetic", "document_id": "syn_03021_2000", "document_hash": "3ac205b464552356699a4d2b0a7bf62497ead36e5bca73982e11ae5ee3810731", "question_id": "q14_unanswerable", "split": "test", "cluster_id": "syn_03021_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 2, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1914, "query_tokens": 484, "message_tokens": 389, "message_utf8_bytes": 931, "cache_tensor_bytes": null, "serialized_payload_bytes": 931, "side_information_bytes": 0, "fetched_bytes": 497, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 11834.982667001896, "hops": {"1": 6858.334750053473, "2": 4976.647916948423}, "receiver_prefill_ms": 184.3267499934882, "ttft_ms": 12.121375068090856, "decode_ms": 11.842624982818961}, "total_wall_ms": 12043.780250940472, "ttft_ms": 12.121375068090856, "input_usage": {"lookups": 1}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 19, "retention_rate": 0.2923076923076923, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.15384615384615385, "status": 1.0, "place": 0.0, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 867}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 19, "retention_rate": 0.2923076923076923, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.15384615384615385, "status": 1.0, "place": 0.0, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 836}, "2": {"facts_total": 65, "facts_retained": 19, "retention_rate": 0.2923076923076923, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.15384615384615385, "status": 1.0, "place": 0.0, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 867}}}, "reference_key_id": "keys/synthetic/syn_03021_2000.json", "task_scores": {"correct": 1.0, "abstained": 1.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "92d15207bf13f35ba83fc20f1408c2fcf07fd4b0ed40fc18f51a5580586b6137", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "dfa2aa0a6797ae01ac3bb479", "attempt_id": "2e5c39521ba7", "run_group_id": "E4_core:A5|repairable|qwen3b:h4", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:19:31.702812Z", "task_family": "synthetic", "document_id": "syn_03021_2000", "document_hash": "3ac205b464552356699a4d2b0a7bf62497ead36e5bca73982e11ae5ee3810731", "question_id": "q00_single_fact", "split": "test", "cluster_id": "syn_03021_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 4, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1914, "query_tokens": 469, "message_tokens": 389, "message_utf8_bytes": 931, "cache_tensor_bytes": null, "serialized_payload_bytes": 931, "side_information_bytes": 0, "fetched_bytes": 1491, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 21781.92445903551, "hops": {"1": 6858.334750053473, "2": 4976.647916948423, "3": 4970.926292007789, "4": 4976.015500025824}, "receiver_prefill_ms": 185.75866695027798, "ttft_ms": 11.88416697550565, "decode_ms": 11.851958930492401}, "total_wall_ms": 21991.85891705565, "ttft_ms": 11.88416697550565, "input_usage": {"lookups": 3}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 19, "retention_rate": 0.2923076923076923, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.15384615384615385, "status": 1.0, "place": 0.0, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 867}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 19, "retention_rate": 0.2923076923076923, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.15384615384615385, "status": 1.0, "place": 0.0, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 836}, "2": {"facts_total": 65, "facts_retained": 19, "retention_rate": 0.2923076923076923, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.15384615384615385, "status": 1.0, "place": 0.0, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 867}, "3": {"facts_total": 65, "facts_retained": 19, "retention_rate": 0.2923076923076923, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.15384615384615385, "status": 1.0, "place": 0.0, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 867}, "4": {"facts_total": 65, "facts_retained": 19, "retention_rate": 0.2923076923076923, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.15384615384615385, "status": 1.0, "place": 0.0, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 867}}}, "reference_key_id": "keys/synthetic/syn_03021_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "92d15207bf13f35ba83fc20f1408c2fcf07fd4b0ed40fc18f51a5580586b6137", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "15a75188a5ffb34386213b74", "attempt_id": "73a93a1758d3", "run_group_id": "E4_core:A5|repairable|qwen3b:h4", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:19:31.928435Z", "task_family": "synthetic", "document_id": "syn_03021_2000", "document_hash": "3ac205b464552356699a4d2b0a7bf62497ead36e5bca73982e11ae5ee3810731", "question_id": "q01_single_fact", "split": "test", "cluster_id": "syn_03021_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 4, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1914, "query_tokens": 475, "message_tokens": 389, "message_utf8_bytes": 931, "cache_tensor_bytes": null, "serialized_payload_bytes": 931, "side_information_bytes": 0, "fetched_bytes": 1491, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 21781.92445903551, "hops": {"1": 6858.334750053473, "2": 4976.647916948423, "3": 4970.926292007789, "4": 4976.015500025824}, "receiver_prefill_ms": 185.9312909655273, "ttft_ms": 11.89075002912432, "decode_ms": 11.830749921500683}, "total_wall_ms": 21992.001833976246, "ttft_ms": 11.89075002912432, "input_usage": {"lookups": 3}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 19, "retention_rate": 0.2923076923076923, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.15384615384615385, "status": 1.0, "place": 0.0, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 867}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 19, "retention_rate": 0.2923076923076923, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.15384615384615385, "status": 1.0, "place": 0.0, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 836}, "2": {"facts_total": 65, "facts_retained": 19, "retention_rate": 0.2923076923076923, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.15384615384615385, "status": 1.0, "place": 0.0, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 867}, "3": {"facts_total": 65, "facts_retained": 19, "retention_rate": 0.2923076923076923, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.15384615384615385, "status": 1.0, "place": 0.0, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 867}, "4": {"facts_total": 65, "facts_retained": 19, "retention_rate": 0.2923076923076923, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.15384615384615385, "status": 1.0, "place": 0.0, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 867}}}, "reference_key_id": "keys/synthetic/syn_03021_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "92d15207bf13f35ba83fc20f1408c2fcf07fd4b0ed40fc18f51a5580586b6137", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "eba8144f5bd7dde0d3a2d64c", "attempt_id": "a941fdee95e0", "run_group_id": "E4_core:A5|repairable|qwen3b:h4", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:19:32.153994Z", "task_family": "synthetic", "document_id": "syn_03021_2000", "document_hash": "3ac205b464552356699a4d2b0a7bf62497ead36e5bca73982e11ae5ee3810731", "question_id": "q02_exact_copy", "split": "test", "cluster_id": "syn_03021_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 4, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1914, "query_tokens": 475, "message_tokens": 389, "message_utf8_bytes": 931, "cache_tensor_bytes": null, "serialized_payload_bytes": 931, "side_information_bytes": 0, "fetched_bytes": 1491, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 21781.92445903551, "hops": {"1": 6858.334750053473, "2": 4976.647916948423, "3": 4970.926292007789, "4": 4976.015500025824}, "receiver_prefill_ms": 186.0080419573933, "ttft_ms": 11.859874939545989, "decode_ms": 11.83879200834781}, "total_wall_ms": 21992.018583929166, "ttft_ms": 11.859874939545989, "input_usage": {"lookups": 3}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 19, "retention_rate": 0.2923076923076923, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.15384615384615385, "status": 1.0, "place": 0.0, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 867}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 19, "retention_rate": 0.2923076923076923, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.15384615384615385, "status": 1.0, "place": 0.0, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 836}, "2": {"facts_total": 65, "facts_retained": 19, "retention_rate": 0.2923076923076923, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.15384615384615385, "status": 1.0, "place": 0.0, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 867}, "3": {"facts_total": 65, "facts_retained": 19, "retention_rate": 0.2923076923076923, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.15384615384615385, "status": 1.0, "place": 0.0, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 867}, "4": {"facts_total": 65, "facts_retained": 19, "retention_rate": 0.2923076923076923, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.15384615384615385, "status": 1.0, "place": 0.0, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 867}}}, "reference_key_id": "keys/synthetic/syn_03021_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "92d15207bf13f35ba83fc20f1408c2fcf07fd4b0ed40fc18f51a5580586b6137", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "f724d68fa8bcbe7d6e1577d2", "attempt_id": "4f84aeec5f94", "run_group_id": "E4_core:A5|repairable|qwen3b:h4", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:19:32.378976Z", "task_family": "synthetic", "document_id": "syn_03021_2000", "document_hash": "3ac205b464552356699a4d2b0a7bf62497ead36e5bca73982e11ae5ee3810731", "question_id": "q03_exact_copy", "split": "test", "cluster_id": "syn_03021_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 4, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1914, "query_tokens": 472, "message_tokens": 389, "message_utf8_bytes": 931, "cache_tensor_bytes": null, "serialized_payload_bytes": 931, "side_information_bytes": 0, "fetched_bytes": 1491, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 21781.92445903551, "hops": {"1": 6858.334750053473, "2": 4976.647916948423, "3": 4970.926292007789, "4": 4976.015500025824}, "receiver_prefill_ms": 185.7952909776941, "ttft_ms": 11.80870900861919, "decode_ms": 11.859916034154594}, "total_wall_ms": 21991.88129301183, "ttft_ms": 11.80870900861919, "input_usage": {"lookups": 3}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 19, "retention_rate": 0.2923076923076923, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.15384615384615385, "status": 1.0, "place": 0.0, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 867}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 19, "retention_rate": 0.2923076923076923, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.15384615384615385, "status": 1.0, "place": 0.0, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 836}, "2": {"facts_total": 65, "facts_retained": 19, "retention_rate": 0.2923076923076923, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.15384615384615385, "status": 1.0, "place": 0.0, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 867}, "3": {"facts_total": 65, "facts_retained": 19, "retention_rate": 0.2923076923076923, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.15384615384615385, "status": 1.0, "place": 0.0, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 867}, "4": {"facts_total": 65, "facts_retained": 19, "retention_rate": 0.2923076923076923, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.15384615384615385, "status": 1.0, "place": 0.0, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 867}}}, "reference_key_id": "keys/synthetic/syn_03021_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "92d15207bf13f35ba83fc20f1408c2fcf07fd4b0ed40fc18f51a5580586b6137", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "36a85823539f6a10202e88a7", "attempt_id": "127c2829dc22", "run_group_id": "E4_core:A5|repairable|qwen3b:h4", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:19:32.603630Z", "task_family": "synthetic", "document_id": "syn_03021_2000", "document_hash": "3ac205b464552356699a4d2b0a7bf62497ead36e5bca73982e11ae5ee3810731", "question_id": "q04_multi_hop", "split": "test", "cluster_id": "syn_03021_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 4, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1914, "query_tokens": 479, "message_tokens": 389, "message_utf8_bytes": 931, "cache_tensor_bytes": null, "serialized_payload_bytes": 931, "side_information_bytes": 0, "fetched_bytes": 1491, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 21781.92445903551, "hops": {"1": 6858.334750053473, "2": 4976.647916948423, "3": 4970.926292007789, "4": 4976.015500025824}, "receiver_prefill_ms": 186.8701659841463, "ttft_ms": 12.091000098735094, "decode_ms": 12.0086669921875}, "total_wall_ms": 21993.4917089995, "ttft_ms": 12.091000098735094, "input_usage": {"lookups": 3}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 19, "retention_rate": 0.2923076923076923, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.15384615384615385, "status": 1.0, "place": 0.0, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 867}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 19, "retention_rate": 0.2923076923076923, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.15384615384615385, "status": 1.0, "place": 0.0, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 836}, "2": {"facts_total": 65, "facts_retained": 19, "retention_rate": 0.2923076923076923, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.15384615384615385, "status": 1.0, "place": 0.0, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 867}, "3": {"facts_total": 65, "facts_retained": 19, "retention_rate": 0.2923076923076923, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.15384615384615385, "status": 1.0, "place": 0.0, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 867}, "4": {"facts_total": 65, "facts_retained": 19, "retention_rate": 0.2923076923076923, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.15384615384615385, "status": 1.0, "place": 0.0, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 867}}}, "reference_key_id": "keys/synthetic/syn_03021_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "92d15207bf13f35ba83fc20f1408c2fcf07fd4b0ed40fc18f51a5580586b6137", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "37830f8f492bf43790a4f5cd", "attempt_id": "b5793ea260aa", "run_group_id": "E4_core:A5|repairable|qwen3b:h4", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:19:32.830677Z", "task_family": "synthetic", "document_id": "syn_03021_2000", "document_hash": "3ac205b464552356699a4d2b0a7bf62497ead36e5bca73982e11ae5ee3810731", "question_id": "q05_multi_hop", "split": "test", "cluster_id": "syn_03021_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 4, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1914, "query_tokens": 480, "message_tokens": 389, "message_utf8_bytes": 931, "cache_tensor_bytes": null, "serialized_payload_bytes": 931, "side_information_bytes": 0, "fetched_bytes": 1491, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 21781.92445903551, "hops": {"1": 6858.334750053473, "2": 4976.647916948423, "3": 4970.926292007789, "4": 4976.015500025824}, "receiver_prefill_ms": 186.83437502477318, "ttft_ms": 11.896374984644353, "decode_ms": 118.63562499638647}, "total_wall_ms": 22099.718958954327, "ttft_ms": 11.896374984644353, "input_usage": {"lookups": 3}, "output_usage": {"output_tokens": 10}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "MH-96F-2099", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 19, "retention_rate": 0.2923076923076923, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.15384615384615385, "status": 1.0, "place": 0.0, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 867}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 19, "retention_rate": 0.2923076923076923, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.15384615384615385, "status": 1.0, "place": 0.0, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 836}, "2": {"facts_total": 65, "facts_retained": 19, "retention_rate": 0.2923076923076923, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.15384615384615385, "status": 1.0, "place": 0.0, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 867}, "3": {"facts_total": 65, "facts_retained": 19, "retention_rate": 0.2923076923076923, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.15384615384615385, "status": 1.0, "place": 0.0, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 867}, "4": {"facts_total": 65, "facts_retained": 19, "retention_rate": 0.2923076923076923, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.15384615384615385, "status": 1.0, "place": 0.0, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 867}}}, "reference_key_id": "keys/synthetic/syn_03021_2000.json", "task_scores": {"correct": 0.0, "abstained": 0.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "92d15207bf13f35ba83fc20f1408c2fcf07fd4b0ed40fc18f51a5580586b6137", "output_hash": "95c230d14a055d30e0e1d4355caf07808e1c968b1feab9e0607d8bd37b1b43dc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "1b4a1c5c1a5cb96858488e5c", "attempt_id": "083ea0124ec7", "run_group_id": "E4_core:A5|repairable|qwen3b:h4", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:19:33.163667Z", "task_family": "synthetic", "document_id": "syn_03021_2000", "document_hash": "3ac205b464552356699a4d2b0a7bf62497ead36e5bca73982e11ae5ee3810731", "question_id": "q06_negation", "split": "test", "cluster_id": "syn_03021_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 4, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1914, "query_tokens": 470, "message_tokens": 389, "message_utf8_bytes": 931, "cache_tensor_bytes": null, "serialized_payload_bytes": 931, "side_information_bytes": 0, "fetched_bytes": 1491, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 21781.92445903551, "hops": {"1": 6858.334750053473, "2": 4976.647916948423, "3": 4970.926292007789, "4": 4976.015500025824}, "receiver_prefill_ms": 186.04454200249165, "ttft_ms": 12.004374992102385, "decode_ms": 12.01070798560977}, "total_wall_ms": 21992.376458947547, "ttft_ms": 12.004374992102385, "input_usage": {"lookups": 3}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "no", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 19, "retention_rate": 0.2923076923076923, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.15384615384615385, "status": 1.0, "place": 0.0, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 867}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 19, "retention_rate": 0.2923076923076923, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.15384615384615385, "status": 1.0, "place": 0.0, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 836}, "2": {"facts_total": 65, "facts_retained": 19, "retention_rate": 0.2923076923076923, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.15384615384615385, "status": 1.0, "place": 0.0, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 867}, "3": {"facts_total": 65, "facts_retained": 19, "retention_rate": 0.2923076923076923, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.15384615384615385, "status": 1.0, "place": 0.0, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 867}, "4": {"facts_total": 65, "facts_retained": 19, "retention_rate": 0.2923076923076923, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.15384615384615385, "status": 1.0, "place": 0.0, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 867}}}, "reference_key_id": "keys/synthetic/syn_03021_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "92d15207bf13f35ba83fc20f1408c2fcf07fd4b0ed40fc18f51a5580586b6137", "output_hash": "9390298f3fb0c5b160498935d79cb139aef28e1c47358b4bbba61862b9c26e59"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "382141b8893e0a3ed2316441", "attempt_id": "050db41f693c", "run_group_id": "E4_core:A5|repairable|qwen3b:h4", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:19:33.388935Z", "task_family": "synthetic", "document_id": "syn_03021_2000", "document_hash": "3ac205b464552356699a4d2b0a7bf62497ead36e5bca73982e11ae5ee3810731", "question_id": "q07_negation", "split": "test", "cluster_id": "syn_03021_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 4, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1914, "query_tokens": 471, "message_tokens": 389, "message_utf8_bytes": 931, "cache_tensor_bytes": null, "serialized_payload_bytes": 931, "side_information_bytes": 0, "fetched_bytes": 1491, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 21781.92445903551, "hops": {"1": 6858.334750053473, "2": 4976.647916948423, "3": 4970.926292007789, "4": 4976.015500025824}, "receiver_prefill_ms": 186.2477499525994, "ttft_ms": 11.954624904319644, "decode_ms": 11.891416972503066}, "total_wall_ms": 21992.477542953566, "ttft_ms": 11.954624904319644, "input_usage": {"lookups": 3}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "no", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 19, "retention_rate": 0.2923076923076923, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.15384615384615385, "status": 1.0, "place": 0.0, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 867}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 19, "retention_rate": 0.2923076923076923, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.15384615384615385, "status": 1.0, "place": 0.0, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 836}, "2": {"facts_total": 65, "facts_retained": 19, "retention_rate": 0.2923076923076923, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.15384615384615385, "status": 1.0, "place": 0.0, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 867}, "3": {"facts_total": 65, "facts_retained": 19, "retention_rate": 0.2923076923076923, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.15384615384615385, "status": 1.0, "place": 0.0, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 867}, "4": {"facts_total": 65, "facts_retained": 19, "retention_rate": 0.2923076923076923, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.15384615384615385, "status": 1.0, "place": 0.0, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 867}}}, "reference_key_id": "keys/synthetic/syn_03021_2000.json", "task_scores": {"correct": 0.0, "abstained": 0.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "92d15207bf13f35ba83fc20f1408c2fcf07fd4b0ed40fc18f51a5580586b6137", "output_hash": "9390298f3fb0c5b160498935d79cb139aef28e1c47358b4bbba61862b9c26e59"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "29696173109d11f819ab2338", "attempt_id": "b43101214d10", "run_group_id": "E4_core:A5|repairable|qwen3b:h4", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:19:33.614253Z", "task_family": "synthetic", "document_id": "syn_03021_2000", "document_hash": "3ac205b464552356699a4d2b0a7bf62497ead36e5bca73982e11ae5ee3810731", "question_id": "q08_revision", "split": "test", "cluster_id": "syn_03021_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 4, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1914, "query_tokens": 487, "message_tokens": 389, "message_utf8_bytes": 931, "cache_tensor_bytes": null, "serialized_payload_bytes": 931, "side_information_bytes": 0, "fetched_bytes": 1491, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 21781.92445903551, "hops": {"1": 6858.334750053473, "2": 4976.647916948423, "3": 4970.926292007789, "4": 4976.015500025824}, "receiver_prefill_ms": 188.15254094079137, "ttft_ms": 11.985791963525116, "decode_ms": 11.900500045157969}, "total_wall_ms": 21994.395583984442, "ttft_ms": 11.985791963525116, "input_usage": {"lookups": 3}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 19, "retention_rate": 0.2923076923076923, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.15384615384615385, "status": 1.0, "place": 0.0, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 867}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 19, "retention_rate": 0.2923076923076923, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.15384615384615385, "status": 1.0, "place": 0.0, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 836}, "2": {"facts_total": 65, "facts_retained": 19, "retention_rate": 0.2923076923076923, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.15384615384615385, "status": 1.0, "place": 0.0, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 867}, "3": {"facts_total": 65, "facts_retained": 19, "retention_rate": 0.2923076923076923, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.15384615384615385, "status": 1.0, "place": 0.0, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 867}, "4": {"facts_total": 65, "facts_retained": 19, "retention_rate": 0.2923076923076923, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.15384615384615385, "status": 1.0, "place": 0.0, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 867}}}, "reference_key_id": "keys/synthetic/syn_03021_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "92d15207bf13f35ba83fc20f1408c2fcf07fd4b0ed40fc18f51a5580586b6137", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "5ad9569a760a8b4a55e2f797", "attempt_id": "66a285a6515c", "run_group_id": "E4_core:A5|repairable|qwen3b:h4", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:19:33.842033Z", "task_family": "synthetic", "document_id": "syn_03021_2000", "document_hash": "3ac205b464552356699a4d2b0a7bf62497ead36e5bca73982e11ae5ee3810731", "question_id": "q09_revision", "split": "test", "cluster_id": "syn_03021_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 4, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1914, "query_tokens": 478, "message_tokens": 389, "message_utf8_bytes": 931, "cache_tensor_bytes": null, "serialized_payload_bytes": 931, "side_information_bytes": 0, "fetched_bytes": 1491, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 21781.92445903551, "hops": {"1": 6858.334750053473, "2": 4976.647916948423, "3": 4970.926292007789, "4": 4976.015500025824}, "receiver_prefill_ms": 186.2962500890717, "ttft_ms": 11.930083041079342, "decode_ms": 82.44408294558525}, "total_wall_ms": 22063.01591708325, "ttft_ms": 11.930083041079342, "input_usage": {"lookups": 3}, "output_usage": {"output_tokens": 7}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "5646720", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 19, "retention_rate": 0.2923076923076923, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.15384615384615385, "status": 1.0, "place": 0.0, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 867}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 19, "retention_rate": 0.2923076923076923, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.15384615384615385, "status": 1.0, "place": 0.0, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 836}, "2": {"facts_total": 65, "facts_retained": 19, "retention_rate": 0.2923076923076923, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.15384615384615385, "status": 1.0, "place": 0.0, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 867}, "3": {"facts_total": 65, "facts_retained": 19, "retention_rate": 0.2923076923076923, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.15384615384615385, "status": 1.0, "place": 0.0, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 867}, "4": {"facts_total": 65, "facts_retained": 19, "retention_rate": 0.2923076923076923, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.15384615384615385, "status": 1.0, "place": 0.0, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 867}}}, "reference_key_id": "keys/synthetic/syn_03021_2000.json", "task_scores": {"correct": 0.0, "abstained": 0.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "92d15207bf13f35ba83fc20f1408c2fcf07fd4b0ed40fc18f51a5580586b6137", "output_hash": "e1d129e860627cf4623671ed186cb4ebbf5d63169d3d0dc14a62e1da2af61326"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "028a9e0044c5d46d286f6fed", "attempt_id": "60c5a1ca23d5", "run_group_id": "E4_core:A5|repairable|qwen3b:h4", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:19:34.138017Z", "task_family": "synthetic", "document_id": "syn_03021_2000", "document_hash": "3ac205b464552356699a4d2b0a7bf62497ead36e5bca73982e11ae5ee3810731", "question_id": "q10_temporal", "split": "test", "cluster_id": "syn_03021_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 4, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1914, "query_tokens": 486, "message_tokens": 389, "message_utf8_bytes": 931, "cache_tensor_bytes": null, "serialized_payload_bytes": 931, "side_information_bytes": 0, "fetched_bytes": 1491, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 21781.92445903551, "hops": {"1": 6858.334750053473, "2": 4976.647916948423, "3": 4970.926292007789, "4": 4976.015500025824}, "receiver_prefill_ms": 187.87783302832395, "ttft_ms": 12.163541978225112, "decode_ms": 59.53324993606657}, "total_wall_ms": 22042.043125955388, "ttft_ms": 12.163541978225112, "input_usage": {"lookups": 3}, "output_usage": {"output_tokens": 5}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "Kovar-9", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 19, "retention_rate": 0.2923076923076923, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.15384615384615385, "status": 1.0, "place": 0.0, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 867}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 19, "retention_rate": 0.2923076923076923, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.15384615384615385, "status": 1.0, "place": 0.0, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 836}, "2": {"facts_total": 65, "facts_retained": 19, "retention_rate": 0.2923076923076923, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.15384615384615385, "status": 1.0, "place": 0.0, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 867}, "3": {"facts_total": 65, "facts_retained": 19, "retention_rate": 0.2923076923076923, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.15384615384615385, "status": 1.0, "place": 0.0, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 867}, "4": {"facts_total": 65, "facts_retained": 19, "retention_rate": 0.2923076923076923, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.15384615384615385, "status": 1.0, "place": 0.0, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 867}}}, "reference_key_id": "keys/synthetic/syn_03021_2000.json", "task_scores": {"correct": 0.0, "abstained": 0.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "92d15207bf13f35ba83fc20f1408c2fcf07fd4b0ed40fc18f51a5580586b6137", "output_hash": "c969b5f7547945067f5b07e9f822d0043126326f979d950a8b1b045057fbb818"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "527e2d10941436d9843e61a1", "attempt_id": "c592b95a8f96", "run_group_id": "E4_core:A5|repairable|qwen3b:h4", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:19:34.413111Z", "task_family": "synthetic", "document_id": "syn_03021_2000", "document_hash": "3ac205b464552356699a4d2b0a7bf62497ead36e5bca73982e11ae5ee3810731", "question_id": "q11_aggregation", "split": "test", "cluster_id": "syn_03021_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 4, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1914, "query_tokens": 482, "message_tokens": 389, "message_utf8_bytes": 931, "cache_tensor_bytes": null, "serialized_payload_bytes": 931, "side_information_bytes": 0, "fetched_bytes": 1491, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 21781.92445903551, "hops": {"1": 6858.334750053473, "2": 4976.647916948423, "3": 4970.926292007789, "4": 4976.015500025824}, "receiver_prefill_ms": 183.64629102870822, "ttft_ms": 12.180499965324998, "decode_ms": 71.41625008080155}, "total_wall_ms": 22049.708334030584, "ttft_ms": 12.180499965324998, "input_usage": {"lookups": 3}, "output_usage": {"output_tokens": 6}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "948485", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 19, "retention_rate": 0.2923076923076923, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.15384615384615385, "status": 1.0, "place": 0.0, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 867}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 19, "retention_rate": 0.2923076923076923, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.15384615384615385, "status": 1.0, "place": 0.0, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 836}, "2": {"facts_total": 65, "facts_retained": 19, "retention_rate": 0.2923076923076923, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.15384615384615385, "status": 1.0, "place": 0.0, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 867}, "3": {"facts_total": 65, "facts_retained": 19, "retention_rate": 0.2923076923076923, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.15384615384615385, "status": 1.0, "place": 0.0, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 867}, "4": {"facts_total": 65, "facts_retained": 19, "retention_rate": 0.2923076923076923, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.15384615384615385, "status": 1.0, "place": 0.0, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 867}}}, "reference_key_id": "keys/synthetic/syn_03021_2000.json", "task_scores": {"correct": 0.0, "abstained": 0.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "92d15207bf13f35ba83fc20f1408c2fcf07fd4b0ed40fc18f51a5580586b6137", "output_hash": "244f716d260fa5f07cbedd5d728983a25a4991af5624be73b358ad829974e9c4"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "aeae98ccabe2505393002aa1", "attempt_id": "766c03efe5d0", "run_group_id": "E4_core:A5|repairable|qwen3b:h4", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:19:34.696891Z", "task_family": "synthetic", "document_id": "syn_03021_2000", "document_hash": "3ac205b464552356699a4d2b0a7bf62497ead36e5bca73982e11ae5ee3810731", "question_id": "q12_unit_conversion", "split": "test", "cluster_id": "syn_03021_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 4, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1914, "query_tokens": 477, "message_tokens": 389, "message_utf8_bytes": 931, "cache_tensor_bytes": null, "serialized_payload_bytes": 931, "side_information_bytes": 0, "fetched_bytes": 1491, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 21781.92445903551, "hops": {"1": 6858.334750053473, "2": 4976.647916948423, "3": 4970.926292007789, "4": 4976.015500025824}, "receiver_prefill_ms": 186.47487496491522, "ttft_ms": 11.89762493595481, "decode_ms": 47.2524999640882}, "total_wall_ms": 22027.974458993413, "ttft_ms": 11.89762493595481, "input_usage": {"lookups": 3}, "output_usage": {"output_tokens": 4}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "22.3", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 19, "retention_rate": 0.2923076923076923, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.15384615384615385, "status": 1.0, "place": 0.0, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 867}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 19, "retention_rate": 0.2923076923076923, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.15384615384615385, "status": 1.0, "place": 0.0, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 836}, "2": {"facts_total": 65, "facts_retained": 19, "retention_rate": 0.2923076923076923, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.15384615384615385, "status": 1.0, "place": 0.0, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 867}, "3": {"facts_total": 65, "facts_retained": 19, "retention_rate": 0.2923076923076923, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.15384615384615385, "status": 1.0, "place": 0.0, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 867}, "4": {"facts_total": 65, "facts_retained": 19, "retention_rate": 0.2923076923076923, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.15384615384615385, "status": 1.0, "place": 0.0, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 867}}}, "reference_key_id": "keys/synthetic/syn_03021_2000.json", "task_scores": {"correct": 0.0, "abstained": 0.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "92d15207bf13f35ba83fc20f1408c2fcf07fd4b0ed40fc18f51a5580586b6137", "output_hash": "8afb0bb35e730fd32e7805d65081241bd21f43709df37165f4ec91c763f8a99d"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "5c41f0753cb706d74e42b6b3", "attempt_id": "c4ac4a286d9f", "run_group_id": "E4_core:A5|repairable|qwen3b:h4", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:19:34.958329Z", "task_family": "synthetic", "document_id": "syn_03021_2000", "document_hash": "3ac205b464552356699a4d2b0a7bf62497ead36e5bca73982e11ae5ee3810731", "question_id": "q13_unanswerable", "split": "test", "cluster_id": "syn_03021_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 4, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1914, "query_tokens": 482, "message_tokens": 389, "message_utf8_bytes": 931, "cache_tensor_bytes": null, "serialized_payload_bytes": 931, "side_information_bytes": 0, "fetched_bytes": 1491, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 21781.92445903551, "hops": {"1": 6858.334750053473, "2": 4976.647916948423, "3": 4970.926292007789, "4": 4976.015500025824}, "receiver_prefill_ms": 184.58412494510412, "ttft_ms": 11.8815409950912, "decode_ms": 11.83712505735457}, "total_wall_ms": 21990.61270907987, "ttft_ms": 11.8815409950912, "input_usage": {"lookups": 3}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 19, "retention_rate": 0.2923076923076923, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.15384615384615385, "status": 1.0, "place": 0.0, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 867}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 19, "retention_rate": 0.2923076923076923, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.15384615384615385, "status": 1.0, "place": 0.0, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 836}, "2": {"facts_total": 65, "facts_retained": 19, "retention_rate": 0.2923076923076923, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.15384615384615385, "status": 1.0, "place": 0.0, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 867}, "3": {"facts_total": 65, "facts_retained": 19, "retention_rate": 0.2923076923076923, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.15384615384615385, "status": 1.0, "place": 0.0, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 867}, "4": {"facts_total": 65, "facts_retained": 19, "retention_rate": 0.2923076923076923, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.15384615384615385, "status": 1.0, "place": 0.0, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 867}}}, "reference_key_id": "keys/synthetic/syn_03021_2000.json", "task_scores": {"correct": 1.0, "abstained": 1.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "92d15207bf13f35ba83fc20f1408c2fcf07fd4b0ed40fc18f51a5580586b6137", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "8d38122e717d899535c9eb40", "attempt_id": "38249278f441", "run_group_id": "E4_core:A5|repairable|qwen3b:h4", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:19:35.181575Z", "task_family": "synthetic", "document_id": "syn_03021_2000", "document_hash": "3ac205b464552356699a4d2b0a7bf62497ead36e5bca73982e11ae5ee3810731", "question_id": "q14_unanswerable", "split": "test", "cluster_id": "syn_03021_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 4, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1914, "query_tokens": 484, "message_tokens": 389, "message_utf8_bytes": 931, "cache_tensor_bytes": null, "serialized_payload_bytes": 931, "side_information_bytes": 0, "fetched_bytes": 1491, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 21781.92445903551, "hops": {"1": 6858.334750053473, "2": 4976.647916948423, "3": 4970.926292007789, "4": 4976.015500025824}, "receiver_prefill_ms": 184.33633400127292, "ttft_ms": 12.082625064067543, "decode_ms": 11.769582983106375}, "total_wall_ms": 21990.528916940093, "ttft_ms": 12.082625064067543, "input_usage": {"lookups": 3}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 19, "retention_rate": 0.2923076923076923, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.15384615384615385, "status": 1.0, "place": 0.0, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 867}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 19, "retention_rate": 0.2923076923076923, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.15384615384615385, "status": 1.0, "place": 0.0, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 836}, "2": {"facts_total": 65, "facts_retained": 19, "retention_rate": 0.2923076923076923, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.15384615384615385, "status": 1.0, "place": 0.0, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 867}, "3": {"facts_total": 65, "facts_retained": 19, "retention_rate": 0.2923076923076923, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.15384615384615385, "status": 1.0, "place": 0.0, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 867}, "4": {"facts_total": 65, "facts_retained": 19, "retention_rate": 0.2923076923076923, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.15384615384615385, "status": 1.0, "place": 0.0, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 867}}}, "reference_key_id": "keys/synthetic/syn_03021_2000.json", "task_scores": {"correct": 1.0, "abstained": 1.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "92d15207bf13f35ba83fc20f1408c2fcf07fd4b0ed40fc18f51a5580586b6137", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "0a5e8c30bfca2b101baa1e68", "attempt_id": "b78523874eab", "run_group_id": "E4_core:A5|repairable|qwen3b:h8", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:20:28.022760Z", "task_family": "synthetic", "document_id": "syn_03021_2000", "document_hash": "3ac205b464552356699a4d2b0a7bf62497ead36e5bca73982e11ae5ee3810731", "question_id": "q00_single_fact", "split": "test", "cluster_id": "syn_03021_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 8, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1914, "query_tokens": 469, "message_tokens": 389, "message_utf8_bytes": 931, "cache_tensor_bytes": null, "serialized_payload_bytes": 931, "side_information_bytes": 0, "fetched_bytes": 3479, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 74394.34491703287, "hops": {"1": 6858.334750053473, "2": 4976.647916948423, "3": 4970.926292007789, "4": 4976.015500025824, "5": 4970.625541987829, "6": 7763.661833014339, "7": 19990.075874957256, "8": 19888.057208037935}, "receiver_prefill_ms": 340.0229170219973, "ttft_ms": 45.39320897310972, "decode_ms": 45.83379102405161}, "total_wall_ms": 74826.11566700507, "ttft_ms": 45.39320897310972, "input_usage": {"lookups": 7}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 19, "retention_rate": 0.2923076923076923, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.15384615384615385, "status": 1.0, "place": 0.0, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 867}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 19, "retention_rate": 0.2923076923076923, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.15384615384615385, "status": 1.0, "place": 0.0, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 836}, "2": {"facts_total": 65, "facts_retained": 19, "retention_rate": 0.2923076923076923, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.15384615384615385, "status": 1.0, "place": 0.0, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 867}, "3": {"facts_total": 65, "facts_retained": 19, "retention_rate": 0.2923076923076923, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.15384615384615385, "status": 1.0, "place": 0.0, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 867}, "4": {"facts_total": 65, "facts_retained": 19, "retention_rate": 0.2923076923076923, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.15384615384615385, "status": 1.0, "place": 0.0, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 867}, "5": {"facts_total": 65, "facts_retained": 19, "retention_rate": 0.2923076923076923, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.15384615384615385, "status": 1.0, "place": 0.0, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 867}, "6": {"facts_total": 65, "facts_retained": 19, "retention_rate": 0.2923076923076923, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.15384615384615385, "status": 1.0, "place": 0.0, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 867}, "7": {"facts_total": 65, "facts_retained": 19, "retention_rate": 0.2923076923076923, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.15384615384615385, "status": 1.0, "place": 0.0, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 867}, "8": {"facts_total": 65, "facts_retained": 19, "retention_rate": 0.2923076923076923, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.15384615384615385, "status": 1.0, "place": 0.0, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 867}}}, "reference_key_id": "keys/synthetic/syn_03021_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "92d15207bf13f35ba83fc20f1408c2fcf07fd4b0ed40fc18f51a5580586b6137", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "9a683658fa9392dd17ce6440", "attempt_id": "eb547cf2883a", "run_group_id": "E4_core:A5|repairable|qwen3b:h8", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:20:28.472013Z", "task_family": "synthetic", "document_id": "syn_03021_2000", "document_hash": "3ac205b464552356699a4d2b0a7bf62497ead36e5bca73982e11ae5ee3810731", "question_id": "q01_single_fact", "split": "test", "cluster_id": "syn_03021_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 8, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1914, "query_tokens": 475, "message_tokens": 389, "message_utf8_bytes": 931, "cache_tensor_bytes": null, "serialized_payload_bytes": 931, "side_information_bytes": 0, "fetched_bytes": 3479, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 74394.34491703287, "hops": {"1": 6858.334750053473, "2": 4976.647916948423, "3": 4970.926292007789, "4": 4976.015500025824, "5": 4970.625541987829, "6": 7763.661833014339, "7": 19990.075874957256, "8": 19888.057208037935}, "receiver_prefill_ms": 352.55520802456886, "ttft_ms": 40.868874988518655, "decode_ms": 49.819333013147116}, "total_wall_ms": 74838.15391699318, "ttft_ms": 40.868874988518655, "input_usage": {"lookups": 7}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 19, "retention_rate": 0.2923076923076923, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.15384615384615385, "status": 1.0, "place": 0.0, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 867}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 19, "retention_rate": 0.2923076923076923, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.15384615384615385, "status": 1.0, "place": 0.0, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 836}, "2": {"facts_total": 65, "facts_retained": 19, "retention_rate": 0.2923076923076923, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.15384615384615385, "status": 1.0, "place": 0.0, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 867}, "3": {"facts_total": 65, "facts_retained": 19, "retention_rate": 0.2923076923076923, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.15384615384615385, "status": 1.0, "place": 0.0, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 867}, "4": {"facts_total": 65, "facts_retained": 19, "retention_rate": 0.2923076923076923, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.15384615384615385, "status": 1.0, "place": 0.0, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 867}, "5": {"facts_total": 65, "facts_retained": 19, "retention_rate": 0.2923076923076923, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.15384615384615385, "status": 1.0, "place": 0.0, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 867}, "6": {"facts_total": 65, "facts_retained": 19, "retention_rate": 0.2923076923076923, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.15384615384615385, "status": 1.0, "place": 0.0, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 867}, "7": {"facts_total": 65, "facts_retained": 19, "retention_rate": 0.2923076923076923, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.15384615384615385, "status": 1.0, "place": 0.0, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 867}, "8": {"facts_total": 65, "facts_retained": 19, "retention_rate": 0.2923076923076923, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.15384615384615385, "status": 1.0, "place": 0.0, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 867}}}, "reference_key_id": "keys/synthetic/syn_03021_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "92d15207bf13f35ba83fc20f1408c2fcf07fd4b0ed40fc18f51a5580586b6137", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "8fb2dac235b6f6ce7584c34b", "attempt_id": "d2685afb6944", "run_group_id": "E4_core:A5|repairable|qwen3b:h8", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:20:28.932361Z", "task_family": "synthetic", "document_id": "syn_03021_2000", "document_hash": "3ac205b464552356699a4d2b0a7bf62497ead36e5bca73982e11ae5ee3810731", "question_id": "q02_exact_copy", "split": "test", "cluster_id": "syn_03021_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 8, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1914, "query_tokens": 475, "message_tokens": 389, "message_utf8_bytes": 931, "cache_tensor_bytes": null, "serialized_payload_bytes": 931, "side_information_bytes": 0, "fetched_bytes": 3479, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 74394.34491703287, "hops": {"1": 6858.334750053473, "2": 4976.647916948423, "3": 4970.926292007789, "4": 4976.015500025824, "5": 4970.625541987829, "6": 7763.661833014339, "7": 19990.075874957256, "8": 19888.057208037935}, "receiver_prefill_ms": 349.36150000430644, "ttft_ms": 45.4847919754684, "decode_ms": 45.87041598279029}, "total_wall_ms": 74835.56275104638, "ttft_ms": 45.4847919754684, "input_usage": {"lookups": 7}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 19, "retention_rate": 0.2923076923076923, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.15384615384615385, "status": 1.0, "place": 0.0, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 867}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 19, "retention_rate": 0.2923076923076923, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.15384615384615385, "status": 1.0, "place": 0.0, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 836}, "2": {"facts_total": 65, "facts_retained": 19, "retention_rate": 0.2923076923076923, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.15384615384615385, "status": 1.0, "place": 0.0, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 867}, "3": {"facts_total": 65, "facts_retained": 19, "retention_rate": 0.2923076923076923, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.15384615384615385, "status": 1.0, "place": 0.0, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 867}, "4": {"facts_total": 65, "facts_retained": 19, "retention_rate": 0.2923076923076923, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.15384615384615385, "status": 1.0, "place": 0.0, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 867}, "5": {"facts_total": 65, "facts_retained": 19, "retention_rate": 0.2923076923076923, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.15384615384615385, "status": 1.0, "place": 0.0, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 867}, "6": {"facts_total": 65, "facts_retained": 19, "retention_rate": 0.2923076923076923, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.15384615384615385, "status": 1.0, "place": 0.0, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 867}, "7": {"facts_total": 65, "facts_retained": 19, "retention_rate": 0.2923076923076923, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.15384615384615385, "status": 1.0, "place": 0.0, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 867}, "8": {"facts_total": 65, "facts_retained": 19, "retention_rate": 0.2923076923076923, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.15384615384615385, "status": 1.0, "place": 0.0, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 867}}}, "reference_key_id": "keys/synthetic/syn_03021_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "92d15207bf13f35ba83fc20f1408c2fcf07fd4b0ed40fc18f51a5580586b6137", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "a9f593f2ff39fcf69b942f42", "attempt_id": "6d7e7ec41a6c", "run_group_id": "E4_core:A5|repairable|qwen3b:h8", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:20:29.390259Z", "task_family": "synthetic", "document_id": "syn_03021_2000", "document_hash": "3ac205b464552356699a4d2b0a7bf62497ead36e5bca73982e11ae5ee3810731", "question_id": "q03_exact_copy", "split": "test", "cluster_id": "syn_03021_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 8, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1914, "query_tokens": 472, "message_tokens": 389, "message_utf8_bytes": 931, "cache_tensor_bytes": null, "serialized_payload_bytes": 931, "side_information_bytes": 0, "fetched_bytes": 3479, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 74394.34491703287, "hops": {"1": 6858.334750053473, "2": 4976.647916948423, "3": 4970.926292007789, "4": 4976.015500025824, "5": 4970.625541987829, "6": 7763.661833014339, "7": 19990.075874957256, "8": 19888.057208037935}, "receiver_prefill_ms": 347.4786669248715, "ttft_ms": 43.20362501312047, "decode_ms": 43.96199993789196}, "total_wall_ms": 74829.48395900894, "ttft_ms": 43.20362501312047, "input_usage": {"lookups": 7}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 19, "retention_rate": 0.2923076923076923, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.15384615384615385, "status": 1.0, "place": 0.0, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 867}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 19, "retention_rate": 0.2923076923076923, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.15384615384615385, "status": 1.0, "place": 0.0, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 836}, "2": {"facts_total": 65, "facts_retained": 19, "retention_rate": 0.2923076923076923, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.15384615384615385, "status": 1.0, "place": 0.0, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 867}, "3": {"facts_total": 65, "facts_retained": 19, "retention_rate": 0.2923076923076923, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.15384615384615385, "status": 1.0, "place": 0.0, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 867}, "4": {"facts_total": 65, "facts_retained": 19, "retention_rate": 0.2923076923076923, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.15384615384615385, "status": 1.0, "place": 0.0, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 867}, "5": {"facts_total": 65, "facts_retained": 19, "retention_rate": 0.2923076923076923, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.15384615384615385, "status": 1.0, "place": 0.0, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 867}, "6": {"facts_total": 65, "facts_retained": 19, "retention_rate": 0.2923076923076923, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.15384615384615385, "status": 1.0, "place": 0.0, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 867}, "7": {"facts_total": 65, "facts_retained": 19, "retention_rate": 0.2923076923076923, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.15384615384615385, "status": 1.0, "place": 0.0, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 867}, "8": {"facts_total": 65, "facts_retained": 19, "retention_rate": 0.2923076923076923, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.15384615384615385, "status": 1.0, "place": 0.0, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 867}}}, "reference_key_id": "keys/synthetic/syn_03021_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "92d15207bf13f35ba83fc20f1408c2fcf07fd4b0ed40fc18f51a5580586b6137", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "ad0f49cc3de245521249fc25", "attempt_id": "21cfa4f1a7f0", "run_group_id": "E4_core:A5|repairable|qwen3b:h8", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:20:29.841743Z", "task_family": "synthetic", "document_id": "syn_03021_2000", "document_hash": "3ac205b464552356699a4d2b0a7bf62497ead36e5bca73982e11ae5ee3810731", "question_id": "q04_multi_hop", "split": "test", "cluster_id": "syn_03021_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 8, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1914, "query_tokens": 479, "message_tokens": 389, "message_utf8_bytes": 931, "cache_tensor_bytes": null, "serialized_payload_bytes": 931, "side_information_bytes": 0, "fetched_bytes": 3479, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 74394.34491703287, "hops": {"1": 6858.334750053473, "2": 4976.647916948423, "3": 4970.926292007789, "4": 4976.015500025824, "5": 4970.625541987829, "6": 7763.661833014339, "7": 19990.075874957256, "8": 19888.057208037935}, "receiver_prefill_ms": 352.6987920049578, "ttft_ms": 43.187833041884005, "decode_ms": 42.309709009714425}, "total_wall_ms": 74833.044292056, "ttft_ms": 43.187833041884005, "input_usage": {"lookups": 7}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 19, "retention_rate": 0.2923076923076923, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.15384615384615385, "status": 1.0, "place": 0.0, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 867}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 19, "retention_rate": 0.2923076923076923, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.15384615384615385, "status": 1.0, "place": 0.0, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 836}, "2": {"facts_total": 65, "facts_retained": 19, "retention_rate": 0.2923076923076923, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.15384615384615385, "status": 1.0, "place": 0.0, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 867}, "3": {"facts_total": 65, "facts_retained": 19, "retention_rate": 0.2923076923076923, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.15384615384615385, "status": 1.0, "place": 0.0, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 867}, "4": {"facts_total": 65, "facts_retained": 19, "retention_rate": 0.2923076923076923, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.15384615384615385, "status": 1.0, "place": 0.0, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 867}, "5": {"facts_total": 65, "facts_retained": 19, "retention_rate": 0.2923076923076923, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.15384615384615385, "status": 1.0, "place": 0.0, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 867}, "6": {"facts_total": 65, "facts_retained": 19, "retention_rate": 0.2923076923076923, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.15384615384615385, "status": 1.0, "place": 0.0, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 867}, "7": {"facts_total": 65, "facts_retained": 19, "retention_rate": 0.2923076923076923, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.15384615384615385, "status": 1.0, "place": 0.0, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 867}, "8": {"facts_total": 65, "facts_retained": 19, "retention_rate": 0.2923076923076923, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.15384615384615385, "status": 1.0, "place": 0.0, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 867}}}, "reference_key_id": "keys/synthetic/syn_03021_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "92d15207bf13f35ba83fc20f1408c2fcf07fd4b0ed40fc18f51a5580586b6137", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "3502e86bf97404a55715cc00", "attempt_id": "e0d344a6d861", "run_group_id": "E4_core:A5|repairable|qwen3b:h8", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:20:30.297083Z", "task_family": "synthetic", "document_id": "syn_03021_2000", "document_hash": "3ac205b464552356699a4d2b0a7bf62497ead36e5bca73982e11ae5ee3810731", "question_id": "q05_multi_hop", "split": "test", "cluster_id": "syn_03021_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 8, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1914, "query_tokens": 480, "message_tokens": 389, "message_utf8_bytes": 931, "cache_tensor_bytes": null, "serialized_payload_bytes": 931, "side_information_bytes": 0, "fetched_bytes": 3479, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 74394.34491703287, "hops": {"1": 6858.334750053473, "2": 4976.647916948423, "3": 4970.926292007789, "4": 4976.015500025824, "5": 4970.625541987829, "6": 7763.661833014339, "7": 19990.075874957256, "8": 19888.057208037935}, "receiver_prefill_ms": 345.83670797292143, "ttft_ms": 42.64020803384483, "decode_ms": 44.390999944880605}, "total_wall_ms": 74827.73033308331, "ttft_ms": 42.64020803384483, "input_usage": {"lookups": 7}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 19, "retention_rate": 0.2923076923076923, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.15384615384615385, "status": 1.0, "place": 0.0, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 867}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 19, "retention_rate": 0.2923076923076923, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.15384615384615385, "status": 1.0, "place": 0.0, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 836}, "2": {"facts_total": 65, "facts_retained": 19, "retention_rate": 0.2923076923076923, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.15384615384615385, "status": 1.0, "place": 0.0, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 867}, "3": {"facts_total": 65, "facts_retained": 19, "retention_rate": 0.2923076923076923, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.15384615384615385, "status": 1.0, "place": 0.0, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 867}, "4": {"facts_total": 65, "facts_retained": 19, "retention_rate": 0.2923076923076923, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.15384615384615385, "status": 1.0, "place": 0.0, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 867}, "5": {"facts_total": 65, "facts_retained": 19, "retention_rate": 0.2923076923076923, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.15384615384615385, "status": 1.0, "place": 0.0, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 867}, "6": {"facts_total": 65, "facts_retained": 19, "retention_rate": 0.2923076923076923, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.15384615384615385, "status": 1.0, "place": 0.0, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 867}, "7": {"facts_total": 65, "facts_retained": 19, "retention_rate": 0.2923076923076923, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.15384615384615385, "status": 1.0, "place": 0.0, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 867}, "8": {"facts_total": 65, "facts_retained": 19, "retention_rate": 0.2923076923076923, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.15384615384615385, "status": 1.0, "place": 0.0, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 867}}}, "reference_key_id": "keys/synthetic/syn_03021_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "92d15207bf13f35ba83fc20f1408c2fcf07fd4b0ed40fc18f51a5580586b6137", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "8ac43d333871d75b68cd3c0e", "attempt_id": "9fa8def0c378", "run_group_id": "E4_core:A5|repairable|qwen3b:h8", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:20:30.747369Z", "task_family": "synthetic", "document_id": "syn_03021_2000", "document_hash": "3ac205b464552356699a4d2b0a7bf62497ead36e5bca73982e11ae5ee3810731", "question_id": "q06_negation", "split": "test", "cluster_id": "syn_03021_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 8, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1914, "query_tokens": 470, "message_tokens": 389, "message_utf8_bytes": 931, "cache_tensor_bytes": null, "serialized_payload_bytes": 931, "side_information_bytes": 0, "fetched_bytes": 3479, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 74394.34491703287, "hops": {"1": 6858.334750053473, "2": 4976.647916948423, "3": 4970.926292007789, "4": 4976.015500025824, "5": 4970.625541987829, "6": 7763.661833014339, "7": 19990.075874957256, "8": 19888.057208037935}, "receiver_prefill_ms": 352.1522500086576, "ttft_ms": 42.748916894197464, "decode_ms": 43.836083030328155}, "total_wall_ms": 74833.56412500143, "ttft_ms": 42.748916894197464, "input_usage": {"lookups": 7}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "no", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 19, "retention_rate": 0.2923076923076923, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.15384615384615385, "status": 1.0, "place": 0.0, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 867}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 19, "retention_rate": 0.2923076923076923, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.15384615384615385, "status": 1.0, "place": 0.0, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 836}, "2": {"facts_total": 65, "facts_retained": 19, "retention_rate": 0.2923076923076923, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.15384615384615385, "status": 1.0, "place": 0.0, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 867}, "3": {"facts_total": 65, "facts_retained": 19, "retention_rate": 0.2923076923076923, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.15384615384615385, "status": 1.0, "place": 0.0, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 867}, "4": {"facts_total": 65, "facts_retained": 19, "retention_rate": 0.2923076923076923, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.15384615384615385, "status": 1.0, "place": 0.0, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 867}, "5": {"facts_total": 65, "facts_retained": 19, "retention_rate": 0.2923076923076923, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.15384615384615385, "status": 1.0, "place": 0.0, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 867}, "6": {"facts_total": 65, "facts_retained": 19, "retention_rate": 0.2923076923076923, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.15384615384615385, "status": 1.0, "place": 0.0, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 867}, "7": {"facts_total": 65, "facts_retained": 19, "retention_rate": 0.2923076923076923, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.15384615384615385, "status": 1.0, "place": 0.0, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 867}, "8": {"facts_total": 65, "facts_retained": 19, "retention_rate": 0.2923076923076923, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.15384615384615385, "status": 1.0, "place": 0.0, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 867}}}, "reference_key_id": "keys/synthetic/syn_03021_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "92d15207bf13f35ba83fc20f1408c2fcf07fd4b0ed40fc18f51a5580586b6137", "output_hash": "9390298f3fb0c5b160498935d79cb139aef28e1c47358b4bbba61862b9c26e59"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "0bd0d9236e7e6858679ebca3", "attempt_id": "df60b3a28f0c", "run_group_id": "E4_core:A5|repairable|qwen3b:h8", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:20:31.203193Z", "task_family": "synthetic", "document_id": "syn_03021_2000", "document_hash": "3ac205b464552356699a4d2b0a7bf62497ead36e5bca73982e11ae5ee3810731", "question_id": "q07_negation", "split": "test", "cluster_id": "syn_03021_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 8, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1914, "query_tokens": 471, "message_tokens": 389, "message_utf8_bytes": 931, "cache_tensor_bytes": null, "serialized_payload_bytes": 931, "side_information_bytes": 0, "fetched_bytes": 3479, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 74394.34491703287, "hops": {"1": 6858.334750053473, "2": 4976.647916948423, "3": 4970.926292007789, "4": 4976.015500025824, "5": 4970.625541987829, "6": 7763.661833014339, "7": 19990.075874957256, "8": 19888.057208037935}, "receiver_prefill_ms": 348.0954170227051, "ttft_ms": 43.39912498835474, "decode_ms": 45.071667060256004}, "total_wall_ms": 74831.42375107855, "ttft_ms": 43.39912498835474, "input_usage": {"lookups": 7}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "no", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 19, "retention_rate": 0.2923076923076923, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.15384615384615385, "status": 1.0, "place": 0.0, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 867}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 19, "retention_rate": 0.2923076923076923, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.15384615384615385, "status": 1.0, "place": 0.0, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 836}, "2": {"facts_total": 65, "facts_retained": 19, "retention_rate": 0.2923076923076923, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.15384615384615385, "status": 1.0, "place": 0.0, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 867}, "3": {"facts_total": 65, "facts_retained": 19, "retention_rate": 0.2923076923076923, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.15384615384615385, "status": 1.0, "place": 0.0, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 867}, "4": {"facts_total": 65, "facts_retained": 19, "retention_rate": 0.2923076923076923, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.15384615384615385, "status": 1.0, "place": 0.0, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 867}, "5": {"facts_total": 65, "facts_retained": 19, "retention_rate": 0.2923076923076923, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.15384615384615385, "status": 1.0, "place": 0.0, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 867}, "6": {"facts_total": 65, "facts_retained": 19, "retention_rate": 0.2923076923076923, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.15384615384615385, "status": 1.0, "place": 0.0, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 867}, "7": {"facts_total": 65, "facts_retained": 19, "retention_rate": 0.2923076923076923, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.15384615384615385, "status": 1.0, "place": 0.0, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 867}, "8": {"facts_total": 65, "facts_retained": 19, "retention_rate": 0.2923076923076923, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.15384615384615385, "status": 1.0, "place": 0.0, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 867}}}, "reference_key_id": "keys/synthetic/syn_03021_2000.json", "task_scores": {"correct": 0.0, "abstained": 0.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "92d15207bf13f35ba83fc20f1408c2fcf07fd4b0ed40fc18f51a5580586b6137", "output_hash": "9390298f3fb0c5b160498935d79cb139aef28e1c47358b4bbba61862b9c26e59"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "f90358cc02c8683295d680bb", "attempt_id": "78ca28800c77", "run_group_id": "E4_core:A5|repairable|qwen3b:h8", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:20:31.656557Z", "task_family": "synthetic", "document_id": "syn_03021_2000", "document_hash": "3ac205b464552356699a4d2b0a7bf62497ead36e5bca73982e11ae5ee3810731", "question_id": "q08_revision", "split": "test", "cluster_id": "syn_03021_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 8, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1914, "query_tokens": 487, "message_tokens": 389, "message_utf8_bytes": 931, "cache_tensor_bytes": null, "serialized_payload_bytes": 931, "side_information_bytes": 0, "fetched_bytes": 3479, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 74394.34491703287, "hops": {"1": 6858.334750053473, "2": 4976.647916948423, "3": 4970.926292007789, "4": 4976.015500025824, "5": 4970.625541987829, "6": 7763.661833014339, "7": 19990.075874957256, "8": 19888.057208037935}, "receiver_prefill_ms": 343.14599994104356, "ttft_ms": 43.34687499795109, "decode_ms": 44.94879103731364}, "total_wall_ms": 74826.35062502231, "ttft_ms": 43.34687499795109, "input_usage": {"lookups": 7}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 19, "retention_rate": 0.2923076923076923, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.15384615384615385, "status": 1.0, "place": 0.0, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 867}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 19, "retention_rate": 0.2923076923076923, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.15384615384615385, "status": 1.0, "place": 0.0, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 836}, "2": {"facts_total": 65, "facts_retained": 19, "retention_rate": 0.2923076923076923, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.15384615384615385, "status": 1.0, "place": 0.0, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 867}, "3": {"facts_total": 65, "facts_retained": 19, "retention_rate": 0.2923076923076923, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.15384615384615385, "status": 1.0, "place": 0.0, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 867}, "4": {"facts_total": 65, "facts_retained": 19, "retention_rate": 0.2923076923076923, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.15384615384615385, "status": 1.0, "place": 0.0, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 867}, "5": {"facts_total": 65, "facts_retained": 19, "retention_rate": 0.2923076923076923, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.15384615384615385, "status": 1.0, "place": 0.0, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 867}, "6": {"facts_total": 65, "facts_retained": 19, "retention_rate": 0.2923076923076923, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.15384615384615385, "status": 1.0, "place": 0.0, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 867}, "7": {"facts_total": 65, "facts_retained": 19, "retention_rate": 0.2923076923076923, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.15384615384615385, "status": 1.0, "place": 0.0, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 867}, "8": {"facts_total": 65, "facts_retained": 19, "retention_rate": 0.2923076923076923, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.15384615384615385, "status": 1.0, "place": 0.0, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 867}}}, "reference_key_id": "keys/synthetic/syn_03021_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "92d15207bf13f35ba83fc20f1408c2fcf07fd4b0ed40fc18f51a5580586b6137", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "cc9ce330ccf4853dfb3dcb38", "attempt_id": "f825c6f6578c", "run_group_id": "E4_core:A5|repairable|qwen3b:h8", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:20:32.105447Z", "task_family": "synthetic", "document_id": "syn_03021_2000", "document_hash": "3ac205b464552356699a4d2b0a7bf62497ead36e5bca73982e11ae5ee3810731", "question_id": "q09_revision", "split": "test", "cluster_id": "syn_03021_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 8, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1914, "query_tokens": 478, "message_tokens": 389, "message_utf8_bytes": 931, "cache_tensor_bytes": null, "serialized_payload_bytes": 931, "side_information_bytes": 0, "fetched_bytes": 3479, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 74394.34491703287, "hops": {"1": 6858.334750053473, "2": 4976.647916948423, "3": 4970.926292007789, "4": 4976.015500025824, "5": 4970.625541987829, "6": 7763.661833014339, "7": 19990.075874957256, "8": 19888.057208037935}, "receiver_prefill_ms": 345.6582920625806, "ttft_ms": 47.286625020205975, "decode_ms": 47.979500028304756}, "total_wall_ms": 74835.76629206073, "ttft_ms": 47.286625020205975, "input_usage": {"lookups": 7}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 19, "retention_rate": 0.2923076923076923, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.15384615384615385, "status": 1.0, "place": 0.0, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 867}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 19, "retention_rate": 0.2923076923076923, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.15384615384615385, "status": 1.0, "place": 0.0, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 836}, "2": {"facts_total": 65, "facts_retained": 19, "retention_rate": 0.2923076923076923, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.15384615384615385, "status": 1.0, "place": 0.0, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 867}, "3": {"facts_total": 65, "facts_retained": 19, "retention_rate": 0.2923076923076923, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.15384615384615385, "status": 1.0, "place": 0.0, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 867}, "4": {"facts_total": 65, "facts_retained": 19, "retention_rate": 0.2923076923076923, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.15384615384615385, "status": 1.0, "place": 0.0, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 867}, "5": {"facts_total": 65, "facts_retained": 19, "retention_rate": 0.2923076923076923, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.15384615384615385, "status": 1.0, "place": 0.0, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 867}, "6": {"facts_total": 65, "facts_retained": 19, "retention_rate": 0.2923076923076923, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.15384615384615385, "status": 1.0, "place": 0.0, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 867}, "7": {"facts_total": 65, "facts_retained": 19, "retention_rate": 0.2923076923076923, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.15384615384615385, "status": 1.0, "place": 0.0, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 867}, "8": {"facts_total": 65, "facts_retained": 19, "retention_rate": 0.2923076923076923, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.15384615384615385, "status": 1.0, "place": 0.0, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 867}}}, "reference_key_id": "keys/synthetic/syn_03021_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "92d15207bf13f35ba83fc20f1408c2fcf07fd4b0ed40fc18f51a5580586b6137", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "a41acb50667e7f864294bac0", "attempt_id": "c11804a977a9", "run_group_id": "E4_core:A5|repairable|qwen3b:h8", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:20:32.563433Z", "task_family": "synthetic", "document_id": "syn_03021_2000", "document_hash": "3ac205b464552356699a4d2b0a7bf62497ead36e5bca73982e11ae5ee3810731", "question_id": "q10_temporal", "split": "test", "cluster_id": "syn_03021_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 8, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1914, "query_tokens": 486, "message_tokens": 389, "message_utf8_bytes": 931, "cache_tensor_bytes": null, "serialized_payload_bytes": 931, "side_information_bytes": 0, "fetched_bytes": 3479, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 74394.34491703287, "hops": {"1": 6858.334750053473, "2": 4976.647916948423, "3": 4970.926292007789, "4": 4976.015500025824, "5": 4970.625541987829, "6": 7763.661833014339, "7": 19990.075874957256, "8": 19888.057208037935}, "receiver_prefill_ms": 349.50512496288866, "ttft_ms": 43.788625043816864, "decode_ms": 230.5075420299545}, "total_wall_ms": 75018.74208403751, "ttft_ms": 43.788625043816864, "input_usage": {"lookups": 7}, "output_usage": {"output_tokens": 5}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "Kovar-9", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 19, "retention_rate": 0.2923076923076923, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.15384615384615385, "status": 1.0, "place": 0.0, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 867}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 19, "retention_rate": 0.2923076923076923, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.15384615384615385, "status": 1.0, "place": 0.0, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 836}, "2": {"facts_total": 65, "facts_retained": 19, "retention_rate": 0.2923076923076923, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.15384615384615385, "status": 1.0, "place": 0.0, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 867}, "3": {"facts_total": 65, "facts_retained": 19, "retention_rate": 0.2923076923076923, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.15384615384615385, "status": 1.0, "place": 0.0, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 867}, "4": {"facts_total": 65, "facts_retained": 19, "retention_rate": 0.2923076923076923, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.15384615384615385, "status": 1.0, "place": 0.0, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 867}, "5": {"facts_total": 65, "facts_retained": 19, "retention_rate": 0.2923076923076923, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.15384615384615385, "status": 1.0, "place": 0.0, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 867}, "6": {"facts_total": 65, "facts_retained": 19, "retention_rate": 0.2923076923076923, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.15384615384615385, "status": 1.0, "place": 0.0, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 867}, "7": {"facts_total": 65, "facts_retained": 19, "retention_rate": 0.2923076923076923, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.15384615384615385, "status": 1.0, "place": 0.0, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 867}, "8": {"facts_total": 65, "facts_retained": 19, "retention_rate": 0.2923076923076923, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.15384615384615385, "status": 1.0, "place": 0.0, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 867}}}, "reference_key_id": "keys/synthetic/syn_03021_2000.json", "task_scores": {"correct": 0.0, "abstained": 0.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "92d15207bf13f35ba83fc20f1408c2fcf07fd4b0ed40fc18f51a5580586b6137", "output_hash": "c969b5f7547945067f5b07e9f822d0043126326f979d950a8b1b045057fbb818"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "f41941a5d8bcebe43141d429", "attempt_id": "4628dc9ce16e", "run_group_id": "E4_core:A5|repairable|qwen3b:h8", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:20:33.204580Z", "task_family": "synthetic", "document_id": "syn_03021_2000", "document_hash": "3ac205b464552356699a4d2b0a7bf62497ead36e5bca73982e11ae5ee3810731", "question_id": "q11_aggregation", "split": "test", "cluster_id": "syn_03021_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 8, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1914, "query_tokens": 482, "message_tokens": 389, "message_utf8_bytes": 931, "cache_tensor_bytes": null, "serialized_payload_bytes": 931, "side_information_bytes": 0, "fetched_bytes": 3479, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 74394.34491703287, "hops": {"1": 6858.334750053473, "2": 4976.647916948423, "3": 4970.926292007789, "4": 4976.015500025824, "5": 4970.625541987829, "6": 7763.661833014339, "7": 19990.075874957256, "8": 19888.057208037935}, "receiver_prefill_ms": 322.29170901700854, "ttft_ms": 41.97966691572219, "decode_ms": 336.1370420316234}, "total_wall_ms": 75095.3966670204, "ttft_ms": 41.97966691572219, "input_usage": {"lookups": 7}, "output_usage": {"output_tokens": 7}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "9484825", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 19, "retention_rate": 0.2923076923076923, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.15384615384615385, "status": 1.0, "place": 0.0, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 867}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 19, "retention_rate": 0.2923076923076923, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.15384615384615385, "status": 1.0, "place": 0.0, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 836}, "2": {"facts_total": 65, "facts_retained": 19, "retention_rate": 0.2923076923076923, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.15384615384615385, "status": 1.0, "place": 0.0, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 867}, "3": {"facts_total": 65, "facts_retained": 19, "retention_rate": 0.2923076923076923, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.15384615384615385, "status": 1.0, "place": 0.0, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 867}, "4": {"facts_total": 65, "facts_retained": 19, "retention_rate": 0.2923076923076923, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.15384615384615385, "status": 1.0, "place": 0.0, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 867}, "5": {"facts_total": 65, "facts_retained": 19, "retention_rate": 0.2923076923076923, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.15384615384615385, "status": 1.0, "place": 0.0, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 867}, "6": {"facts_total": 65, "facts_retained": 19, "retention_rate": 0.2923076923076923, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.15384615384615385, "status": 1.0, "place": 0.0, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 867}, "7": {"facts_total": 65, "facts_retained": 19, "retention_rate": 0.2923076923076923, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.15384615384615385, "status": 1.0, "place": 0.0, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 867}, "8": {"facts_total": 65, "facts_retained": 19, "retention_rate": 0.2923076923076923, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.15384615384615385, "status": 1.0, "place": 0.0, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 867}}}, "reference_key_id": "keys/synthetic/syn_03021_2000.json", "task_scores": {"correct": 0.0, "abstained": 0.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "92d15207bf13f35ba83fc20f1408c2fcf07fd4b0ed40fc18f51a5580586b6137", "output_hash": "667b970507c4c6c7fb572757eb3b5eb04186ddb67a6ef680266b7a0924737993"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "7e420a79e238512cb1897a09", "attempt_id": "4dffb167c771", "run_group_id": "E4_core:A5|repairable|qwen3b:h8", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:20:33.922899Z", "task_family": "synthetic", "document_id": "syn_03021_2000", "document_hash": "3ac205b464552356699a4d2b0a7bf62497ead36e5bca73982e11ae5ee3810731", "question_id": "q12_unit_conversion", "split": "test", "cluster_id": "syn_03021_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 8, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1914, "query_tokens": 477, "message_tokens": 389, "message_utf8_bytes": 931, "cache_tensor_bytes": null, "serialized_payload_bytes": 931, "side_information_bytes": 0, "fetched_bytes": 3479, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 74394.34491703287, "hops": {"1": 6858.334750053473, "2": 4976.647916948423, "3": 4970.926292007789, "4": 4976.015500025824, "5": 4970.625541987829, "6": 7763.661833014339, "7": 19990.075874957256, "8": 19888.057208037935}, "receiver_prefill_ms": 334.5014579826966, "ttft_ms": 43.67445793468505, "decode_ms": 177.05166607629508}, "total_wall_ms": 74950.17541700508, "ttft_ms": 43.67445793468505, "input_usage": {"lookups": 7}, "output_usage": {"output_tokens": 4}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "22.3", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 19, "retention_rate": 0.2923076923076923, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.15384615384615385, "status": 1.0, "place": 0.0, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 867}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 19, "retention_rate": 0.2923076923076923, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.15384615384615385, "status": 1.0, "place": 0.0, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 836}, "2": {"facts_total": 65, "facts_retained": 19, "retention_rate": 0.2923076923076923, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.15384615384615385, "status": 1.0, "place": 0.0, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 867}, "3": {"facts_total": 65, "facts_retained": 19, "retention_rate": 0.2923076923076923, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.15384615384615385, "status": 1.0, "place": 0.0, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 867}, "4": {"facts_total": 65, "facts_retained": 19, "retention_rate": 0.2923076923076923, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.15384615384615385, "status": 1.0, "place": 0.0, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 867}, "5": {"facts_total": 65, "facts_retained": 19, "retention_rate": 0.2923076923076923, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.15384615384615385, "status": 1.0, "place": 0.0, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 867}, "6": {"facts_total": 65, "facts_retained": 19, "retention_rate": 0.2923076923076923, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.15384615384615385, "status": 1.0, "place": 0.0, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 867}, "7": {"facts_total": 65, "facts_retained": 19, "retention_rate": 0.2923076923076923, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.15384615384615385, "status": 1.0, "place": 0.0, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 867}, "8": {"facts_total": 65, "facts_retained": 19, "retention_rate": 0.2923076923076923, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.15384615384615385, "status": 1.0, "place": 0.0, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 867}}}, "reference_key_id": "keys/synthetic/syn_03021_2000.json", "task_scores": {"correct": 0.0, "abstained": 0.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "92d15207bf13f35ba83fc20f1408c2fcf07fd4b0ed40fc18f51a5580586b6137", "output_hash": "8afb0bb35e730fd32e7805d65081241bd21f43709df37165f4ec91c763f8a99d"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "14202cc2f32a4176dd106b01", "attempt_id": "b035e5911950", "run_group_id": "E4_core:A5|repairable|qwen3b:h8", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:20:34.496099Z", "task_family": "synthetic", "document_id": "syn_03021_2000", "document_hash": "3ac205b464552356699a4d2b0a7bf62497ead36e5bca73982e11ae5ee3810731", "question_id": "q13_unanswerable", "split": "test", "cluster_id": "syn_03021_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 8, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1914, "query_tokens": 482, "message_tokens": 389, "message_utf8_bytes": 931, "cache_tensor_bytes": null, "serialized_payload_bytes": 931, "side_information_bytes": 0, "fetched_bytes": 3479, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 74394.34491703287, "hops": {"1": 6858.334750053473, "2": 4976.647916948423, "3": 4970.926292007789, "4": 4976.015500025824, "5": 4970.625541987829, "6": 7763.661833014339, "7": 19990.075874957256, "8": 19888.057208037935}, "receiver_prefill_ms": 287.6303750090301, "ttft_ms": 50.758292083628476, "decode_ms": 45.91450002044439}, "total_wall_ms": 74779.32237612549, "ttft_ms": 50.758292083628476, "input_usage": {"lookups": 7}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 19, "retention_rate": 0.2923076923076923, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.15384615384615385, "status": 1.0, "place": 0.0, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 867}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 19, "retention_rate": 0.2923076923076923, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.15384615384615385, "status": 1.0, "place": 0.0, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 836}, "2": {"facts_total": 65, "facts_retained": 19, "retention_rate": 0.2923076923076923, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.15384615384615385, "status": 1.0, "place": 0.0, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 867}, "3": {"facts_total": 65, "facts_retained": 19, "retention_rate": 0.2923076923076923, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.15384615384615385, "status": 1.0, "place": 0.0, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 867}, "4": {"facts_total": 65, "facts_retained": 19, "retention_rate": 0.2923076923076923, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.15384615384615385, "status": 1.0, "place": 0.0, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 867}, "5": {"facts_total": 65, "facts_retained": 19, "retention_rate": 0.2923076923076923, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.15384615384615385, "status": 1.0, "place": 0.0, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 867}, "6": {"facts_total": 65, "facts_retained": 19, "retention_rate": 0.2923076923076923, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.15384615384615385, "status": 1.0, "place": 0.0, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 867}, "7": {"facts_total": 65, "facts_retained": 19, "retention_rate": 0.2923076923076923, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.15384615384615385, "status": 1.0, "place": 0.0, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 867}, "8": {"facts_total": 65, "facts_retained": 19, "retention_rate": 0.2923076923076923, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.15384615384615385, "status": 1.0, "place": 0.0, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 867}}}, "reference_key_id": "keys/synthetic/syn_03021_2000.json", "task_scores": {"correct": 1.0, "abstained": 1.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "92d15207bf13f35ba83fc20f1408c2fcf07fd4b0ed40fc18f51a5580586b6137", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "fd33d63090bdafcef3f49ccf", "attempt_id": "eed2993c9598", "run_group_id": "E4_core:A5|repairable|qwen3b:h8", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:20:34.898102Z", "task_family": "synthetic", "document_id": "syn_03021_2000", "document_hash": "3ac205b464552356699a4d2b0a7bf62497ead36e5bca73982e11ae5ee3810731", "question_id": "q14_unanswerable", "split": "test", "cluster_id": "syn_03021_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 8, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1914, "query_tokens": 484, "message_tokens": 389, "message_utf8_bytes": 931, "cache_tensor_bytes": null, "serialized_payload_bytes": 931, "side_information_bytes": 0, "fetched_bytes": 3479, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 74394.34491703287, "hops": {"1": 6858.334750053473, "2": 4976.647916948423, "3": 4970.926292007789, "4": 4976.015500025824, "5": 4970.625541987829, "6": 7763.661833014339, "7": 19990.075874957256, "8": 19888.057208037935}, "receiver_prefill_ms": 312.707083998248, "ttft_ms": 66.90520804841071, "decode_ms": 30.21133306901902}, "total_wall_ms": 74804.78116706945, "ttft_ms": 66.90520804841071, "input_usage": {"lookups": 7}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 19, "retention_rate": 0.2923076923076923, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.15384615384615385, "status": 1.0, "place": 0.0, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 867}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 19, "retention_rate": 0.2923076923076923, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.15384615384615385, "status": 1.0, "place": 0.0, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 836}, "2": {"facts_total": 65, "facts_retained": 19, "retention_rate": 0.2923076923076923, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.15384615384615385, "status": 1.0, "place": 0.0, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 867}, "3": {"facts_total": 65, "facts_retained": 19, "retention_rate": 0.2923076923076923, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.15384615384615385, "status": 1.0, "place": 0.0, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 867}, "4": {"facts_total": 65, "facts_retained": 19, "retention_rate": 0.2923076923076923, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.15384615384615385, "status": 1.0, "place": 0.0, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 867}, "5": {"facts_total": 65, "facts_retained": 19, "retention_rate": 0.2923076923076923, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.15384615384615385, "status": 1.0, "place": 0.0, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 867}, "6": {"facts_total": 65, "facts_retained": 19, "retention_rate": 0.2923076923076923, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.15384615384615385, "status": 1.0, "place": 0.0, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 867}, "7": {"facts_total": 65, "facts_retained": 19, "retention_rate": 0.2923076923076923, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.15384615384615385, "status": 1.0, "place": 0.0, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 867}, "8": {"facts_total": 65, "facts_retained": 19, "retention_rate": 0.2923076923076923, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.15384615384615385, "status": 1.0, "place": 0.0, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 867}}}, "reference_key_id": "keys/synthetic/syn_03021_2000.json", "task_scores": {"correct": 1.0, "abstained": 1.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "92d15207bf13f35ba83fc20f1408c2fcf07fd4b0ed40fc18f51a5580586b6137", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "3f32aff76e950cc1f431ac0b", "attempt_id": "18e8a15f2269", "run_group_id": "E4_core:A4|closed|qwen3b-llama3b:h1", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:20:57.689777Z", "task_family": "synthetic", "document_id": "syn_03021_2000", "document_hash": "3ac205b464552356699a4d2b0a7bf62497ead36e5bca73982e11ae5ee3810731", "question_id": "q00_single_fact", "split": "test", "cluster_id": "syn_03021_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 1, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1914, "query_tokens": 592, "message_tokens": 512, "message_utf8_bytes": 1179, "cache_tensor_bytes": null, "serialized_payload_bytes": 1179, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 22361.685457988642, "hops": {"1": 22361.685457988642}, "receiver_prefill_ms": 227.15991700533777, "ttft_ms": 11.977292015217245, "decode_ms": 12.12245796341449}, "total_wall_ms": 22613.57029201463, "ttft_ms": 11.977292015217245, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 38, "retention_rate": 0.5846153846153846, "retention_by_kind": {"person": 0.8333333333333334, "date": 0.7, "amount": 0.38461538461538464, "status": 1.0, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1179}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 38, "retention_rate": 0.5846153846153846, "retention_by_kind": {"person": 0.8333333333333334, "date": 0.7, "amount": 0.38461538461538464, "status": 1.0, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1179}}}, "reference_key_id": "keys/synthetic/syn_03021_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "f6342dcd02ebd80508f967d8bfed98ca4bc1afeea750629dadac4b492752e270", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "4ccce6a133688b4134eca730", "attempt_id": "69d1ebce5c5d", "run_group_id": "E4_core:A4|closed|qwen3b-llama3b:h1", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:20:57.958203Z", "task_family": "synthetic", "document_id": "syn_03021_2000", "document_hash": "3ac205b464552356699a4d2b0a7bf62497ead36e5bca73982e11ae5ee3810731", "question_id": "q01_single_fact", "split": "test", "cluster_id": "syn_03021_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 1, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1914, "query_tokens": 598, "message_tokens": 512, "message_utf8_bytes": 1179, "cache_tensor_bytes": null, "serialized_payload_bytes": 1179, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 22361.685457988642, "hops": {"1": 22361.685457988642}, "receiver_prefill_ms": 227.10354207083583, "ttft_ms": 11.963792028836906, "decode_ms": 130.18362503498793}, "total_wall_ms": 22731.602748972364, "ttft_ms": 11.963792028836906, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 11}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "Lot 536-R, Senneterre", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 38, "retention_rate": 0.5846153846153846, "retention_by_kind": {"person": 0.8333333333333334, "date": 0.7, "amount": 0.38461538461538464, "status": 1.0, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1179}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 38, "retention_rate": 0.5846153846153846, "retention_by_kind": {"person": 0.8333333333333334, "date": 0.7, "amount": 0.38461538461538464, "status": 1.0, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1179}}}, "reference_key_id": "keys/synthetic/syn_03021_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 0.8571428571428571}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "f6342dcd02ebd80508f967d8bfed98ca4bc1afeea750629dadac4b492752e270", "output_hash": "5e85af24aac0f1904d70a2fcee84f1b6f6912cc7450298af1f173002fd8e59df"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "69fd52a6e9e971b5b0f1a5be", "attempt_id": "c3cd2d356187", "run_group_id": "E4_core:A4|closed|qwen3b-llama3b:h1", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:20:58.344277Z", "task_family": "synthetic", "document_id": "syn_03021_2000", "document_hash": "3ac205b464552356699a4d2b0a7bf62497ead36e5bca73982e11ae5ee3810731", "question_id": "q02_exact_copy", "split": "test", "cluster_id": "syn_03021_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 1, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1914, "query_tokens": 598, "message_tokens": 512, "message_utf8_bytes": 1179, "cache_tensor_bytes": null, "serialized_payload_bytes": 1179, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 22361.685457988642, "hops": {"1": 22361.685457988642}, "receiver_prefill_ms": 226.86120797879994, "ttft_ms": 12.056584004312754, "decode_ms": 83.35712493862957}, "total_wall_ms": 22684.603373985738, "ttft_ms": 12.056584004312754, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 7}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "88.7751", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 38, "retention_rate": 0.5846153846153846, "retention_by_kind": {"person": 0.8333333333333334, "date": 0.7, "amount": 0.38461538461538464, "status": 1.0, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1179}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 38, "retention_rate": 0.5846153846153846, "retention_by_kind": {"person": 0.8333333333333334, "date": 0.7, "amount": 0.38461538461538464, "status": 1.0, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1179}}}, "reference_key_id": "keys/synthetic/syn_03021_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "f6342dcd02ebd80508f967d8bfed98ca4bc1afeea750629dadac4b492752e270", "output_hash": "1e01e62c69f9479b681e50e59e8565c6583f5cdb803446a05ed5cb13118acdf9"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "086a7f165c3fbc3ff0a9a94d", "attempt_id": "8fe5df06bc82", "run_group_id": "E4_core:A4|closed|qwen3b-llama3b:h1", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:20:58.683917Z", "task_family": "synthetic", "document_id": "syn_03021_2000", "document_hash": "3ac205b464552356699a4d2b0a7bf62497ead36e5bca73982e11ae5ee3810731", "question_id": "q03_exact_copy", "split": "test", "cluster_id": "syn_03021_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 1, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1914, "query_tokens": 595, "message_tokens": 512, "message_utf8_bytes": 1179, "cache_tensor_bytes": null, "serialized_payload_bytes": 1179, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 22361.685457988642, "hops": {"1": 22361.685457988642}, "receiver_prefill_ms": 226.997458958067, "ttft_ms": 11.92891609389335, "decode_ms": 47.57245793007314}, "total_wall_ms": 22648.729874985293, "ttft_ms": 11.92891609389335, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 4}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "Ingrid Mbeki", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 38, "retention_rate": 0.5846153846153846, "retention_by_kind": {"person": 0.8333333333333334, "date": 0.7, "amount": 0.38461538461538464, "status": 1.0, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1179}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 38, "retention_rate": 0.5846153846153846, "retention_by_kind": {"person": 0.8333333333333334, "date": 0.7, "amount": 0.38461538461538464, "status": 1.0, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1179}}}, "reference_key_id": "keys/synthetic/syn_03021_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "f6342dcd02ebd80508f967d8bfed98ca4bc1afeea750629dadac4b492752e270", "output_hash": "1dbd85090bb65fb3b72de8ca4ce32f2c234729ff4a239274e79fb574b800a79d"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "3b07f98b72002e5a2beb5675", "attempt_id": "480b6b5c4b68", "run_group_id": "E4_core:A4|closed|qwen3b-llama3b:h1", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:20:58.987077Z", "task_family": "synthetic", "document_id": "syn_03021_2000", "document_hash": "3ac205b464552356699a4d2b0a7bf62497ead36e5bca73982e11ae5ee3810731", "question_id": "q04_multi_hop", "split": "test", "cluster_id": "syn_03021_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 1, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1914, "query_tokens": 602, "message_tokens": 512, "message_utf8_bytes": 1179, "cache_tensor_bytes": null, "serialized_payload_bytes": 1179, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 22361.685457988642, "hops": {"1": 22361.685457988642}, "receiver_prefill_ms": 227.1020410116762, "ttft_ms": 11.852875002659857, "decode_ms": 11.78449997678399}, "total_wall_ms": 22612.94329096563, "ttft_ms": 11.852875002659857, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 38, "retention_rate": 0.5846153846153846, "retention_by_kind": {"person": 0.8333333333333334, "date": 0.7, "amount": 0.38461538461538464, "status": 1.0, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1179}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 38, "retention_rate": 0.5846153846153846, "retention_by_kind": {"person": 0.8333333333333334, "date": 0.7, "amount": 0.38461538461538464, "status": 1.0, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1179}}}, "reference_key_id": "keys/synthetic/syn_03021_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "f6342dcd02ebd80508f967d8bfed98ca4bc1afeea750629dadac4b492752e270", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "a70ed564ccfaa4f69e25a5db", "attempt_id": "180361820b0d", "run_group_id": "E4_core:A4|closed|qwen3b-llama3b:h1", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:20:59.253998Z", "task_family": "synthetic", "document_id": "syn_03021_2000", "document_hash": "3ac205b464552356699a4d2b0a7bf62497ead36e5bca73982e11ae5ee3810731", "question_id": "q05_multi_hop", "split": "test", "cluster_id": "syn_03021_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 1, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1914, "query_tokens": 603, "message_tokens": 512, "message_utf8_bytes": 1179, "cache_tensor_bytes": null, "serialized_payload_bytes": 1179, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 22361.685457988642, "hops": {"1": 22361.685457988642}, "receiver_prefill_ms": 227.24337503314018, "ttft_ms": 12.091082986444235, "decode_ms": 130.82462491001934}, "total_wall_ms": 22732.50016605016, "ttft_ms": 12.091082986444235, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 11}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "BQ-59Z-6753", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 38, "retention_rate": 0.5846153846153846, "retention_by_kind": {"person": 0.8333333333333334, "date": 0.7, "amount": 0.38461538461538464, "status": 1.0, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1179}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 38, "retention_rate": 0.5846153846153846, "retention_by_kind": {"person": 0.8333333333333334, "date": 0.7, "amount": 0.38461538461538464, "status": 1.0, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1179}}}, "reference_key_id": "keys/synthetic/syn_03021_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "f6342dcd02ebd80508f967d8bfed98ca4bc1afeea750629dadac4b492752e270", "output_hash": "db250f072c856fe7ddf4bbbd16e20c71b8bad1eef5e4480acd0dac30f8ffe9cd"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "55ab9256a03c14bb7dd4931b", "attempt_id": "ef477ebc6624", "run_group_id": "E4_core:A4|closed|qwen3b-llama3b:h1", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:20:59.640788Z", "task_family": "synthetic", "document_id": "syn_03021_2000", "document_hash": "3ac205b464552356699a4d2b0a7bf62497ead36e5bca73982e11ae5ee3810731", "question_id": "q06_negation", "split": "test", "cluster_id": "syn_03021_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 1, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1914, "query_tokens": 593, "message_tokens": 512, "message_utf8_bytes": 1179, "cache_tensor_bytes": null, "serialized_payload_bytes": 1179, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 22361.685457988642, "hops": {"1": 22361.685457988642}, "receiver_prefill_ms": 226.3879170641303, "ttft_ms": 12.00712495483458, "decode_ms": 11.904167011380196}, "total_wall_ms": 22612.498832982965, "ttft_ms": 12.00712495483458, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "no", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 38, "retention_rate": 0.5846153846153846, "retention_by_kind": {"person": 0.8333333333333334, "date": 0.7, "amount": 0.38461538461538464, "status": 1.0, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1179}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 38, "retention_rate": 0.5846153846153846, "retention_by_kind": {"person": 0.8333333333333334, "date": 0.7, "amount": 0.38461538461538464, "status": 1.0, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1179}}}, "reference_key_id": "keys/synthetic/syn_03021_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "f6342dcd02ebd80508f967d8bfed98ca4bc1afeea750629dadac4b492752e270", "output_hash": "9390298f3fb0c5b160498935d79cb139aef28e1c47358b4bbba61862b9c26e59"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "7abf703b61fcd40e55e50017", "attempt_id": "8c998977055d", "run_group_id": "E4_core:A4|closed|qwen3b-llama3b:h1", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:20:59.907437Z", "task_family": "synthetic", "document_id": "syn_03021_2000", "document_hash": "3ac205b464552356699a4d2b0a7bf62497ead36e5bca73982e11ae5ee3810731", "question_id": "q07_negation", "split": "test", "cluster_id": "syn_03021_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 1, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1914, "query_tokens": 594, "message_tokens": 512, "message_utf8_bytes": 1179, "cache_tensor_bytes": null, "serialized_payload_bytes": 1179, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 22361.685457988642, "hops": {"1": 22361.685457988642}, "receiver_prefill_ms": 226.28329205326736, "ttft_ms": 12.032249942421913, "decode_ms": 11.911874986253679}, "total_wall_ms": 22612.483248929493, "ttft_ms": 12.032249942421913, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "yes", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 38, "retention_rate": 0.5846153846153846, "retention_by_kind": {"person": 0.8333333333333334, "date": 0.7, "amount": 0.38461538461538464, "status": 1.0, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1179}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 38, "retention_rate": 0.5846153846153846, "retention_by_kind": {"person": 0.8333333333333334, "date": 0.7, "amount": 0.38461538461538464, "status": 1.0, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1179}}}, "reference_key_id": "keys/synthetic/syn_03021_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "f6342dcd02ebd80508f967d8bfed98ca4bc1afeea750629dadac4b492752e270", "output_hash": "8a798890fe93817163b10b5f7bd2ca4d25d84c52739a645a889c173eee7d9d3d"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "d65af73e32722271da5a697f", "attempt_id": "69e2265cd084", "run_group_id": "E4_core:A4|closed|qwen3b-llama3b:h1", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:21:00.173848Z", "task_family": "synthetic", "document_id": "syn_03021_2000", "document_hash": "3ac205b464552356699a4d2b0a7bf62497ead36e5bca73982e11ae5ee3810731", "question_id": "q08_revision", "split": "test", "cluster_id": "syn_03021_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 1, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1914, "query_tokens": 610, "message_tokens": 512, "message_utf8_bytes": 1179, "cache_tensor_bytes": null, "serialized_payload_bytes": 1179, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 22361.685457988642, "hops": {"1": 22361.685457988642}, "receiver_prefill_ms": 228.54887507855892, "ttft_ms": 12.094540987163782, "decode_ms": 11.97579107247293}, "total_wall_ms": 22614.94124995079, "ttft_ms": 12.094540987163782, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 38, "retention_rate": 0.5846153846153846, "retention_by_kind": {"person": 0.8333333333333334, "date": 0.7, "amount": 0.38461538461538464, "status": 1.0, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1179}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 38, "retention_rate": 0.5846153846153846, "retention_by_kind": {"person": 0.8333333333333334, "date": 0.7, "amount": 0.38461538461538464, "status": 1.0, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1179}}}, "reference_key_id": "keys/synthetic/syn_03021_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "f6342dcd02ebd80508f967d8bfed98ca4bc1afeea750629dadac4b492752e270", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "1b77b378869a751190db2006", "attempt_id": "dc6e82d8f82f", "run_group_id": "E4_core:A4|closed|qwen3b-llama3b:h1", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:21:00.442933Z", "task_family": "synthetic", "document_id": "syn_03021_2000", "document_hash": "3ac205b464552356699a4d2b0a7bf62497ead36e5bca73982e11ae5ee3810731", "question_id": "q09_revision", "split": "test", "cluster_id": "syn_03021_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 1, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1914, "query_tokens": 601, "message_tokens": 512, "message_utf8_bytes": 1179, "cache_tensor_bytes": null, "serialized_payload_bytes": 1179, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 22361.685457988642, "hops": {"1": 22361.685457988642}, "receiver_prefill_ms": 226.77829093299806, "ttft_ms": 11.983916047029197, "decode_ms": 11.876125005073845}, "total_wall_ms": 22612.849042052403, "ttft_ms": 11.983916047029197, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 38, "retention_rate": 0.5846153846153846, "retention_by_kind": {"person": 0.8333333333333334, "date": 0.7, "amount": 0.38461538461538464, "status": 1.0, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1179}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 38, "retention_rate": 0.5846153846153846, "retention_by_kind": {"person": 0.8333333333333334, "date": 0.7, "amount": 0.38461538461538464, "status": 1.0, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1179}}}, "reference_key_id": "keys/synthetic/syn_03021_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "f6342dcd02ebd80508f967d8bfed98ca4bc1afeea750629dadac4b492752e270", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "e648daf57fc4b7499c862983", "attempt_id": "8933bb3596ed", "run_group_id": "E4_core:A4|closed|qwen3b-llama3b:h1", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:21:00.709933Z", "task_family": "synthetic", "document_id": "syn_03021_2000", "document_hash": "3ac205b464552356699a4d2b0a7bf62497ead36e5bca73982e11ae5ee3810731", "question_id": "q10_temporal", "split": "test", "cluster_id": "syn_03021_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 1, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1914, "query_tokens": 609, "message_tokens": 512, "message_utf8_bytes": 1179, "cache_tensor_bytes": null, "serialized_payload_bytes": 1179, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 22361.685457988642, "hops": {"1": 22361.685457988642}, "receiver_prefill_ms": 226.77929198835045, "ttft_ms": 11.996500077657402, "decode_ms": 59.55533299129456}, "total_wall_ms": 22660.552707966417, "ttft_ms": 11.996500077657402, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 5}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "Kovar-9", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 38, "retention_rate": 0.5846153846153846, "retention_by_kind": {"person": 0.8333333333333334, "date": 0.7, "amount": 0.38461538461538464, "status": 1.0, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1179}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 38, "retention_rate": 0.5846153846153846, "retention_by_kind": {"person": 0.8333333333333334, "date": 0.7, "amount": 0.38461538461538464, "status": 1.0, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1179}}}, "reference_key_id": "keys/synthetic/syn_03021_2000.json", "task_scores": {"correct": 0.0, "abstained": 0.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "f6342dcd02ebd80508f967d8bfed98ca4bc1afeea750629dadac4b492752e270", "output_hash": "c969b5f7547945067f5b07e9f822d0043126326f979d950a8b1b045057fbb818"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "36038c95f552b868a77e602c", "attempt_id": "ed6f2fe9ce1e", "run_group_id": "E4_core:A4|closed|qwen3b-llama3b:h1", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:21:01.024375Z", "task_family": "synthetic", "document_id": "syn_03021_2000", "document_hash": "3ac205b464552356699a4d2b0a7bf62497ead36e5bca73982e11ae5ee3810731", "question_id": "q11_aggregation", "split": "test", "cluster_id": "syn_03021_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 1, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1914, "query_tokens": 605, "message_tokens": 512, "message_utf8_bytes": 1179, "cache_tensor_bytes": null, "serialized_payload_bytes": 1179, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 22361.685457988642, "hops": {"1": 22361.685457988642}, "receiver_prefill_ms": 226.84737492818385, "ttft_ms": 11.92341698333621, "decode_ms": 83.45675002783537}, "total_wall_ms": 22684.485999983735, "ttft_ms": 11.92341698333621, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 7}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "1302481", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 38, "retention_rate": 0.5846153846153846, "retention_by_kind": {"person": 0.8333333333333334, "date": 0.7, "amount": 0.38461538461538464, "status": 1.0, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1179}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 38, "retention_rate": 0.5846153846153846, "retention_by_kind": {"person": 0.8333333333333334, "date": 0.7, "amount": 0.38461538461538464, "status": 1.0, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1179}}}, "reference_key_id": "keys/synthetic/syn_03021_2000.json", "task_scores": {"correct": 0.0, "abstained": 0.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "f6342dcd02ebd80508f967d8bfed98ca4bc1afeea750629dadac4b492752e270", "output_hash": "1c130ebf03e5051f031da47a49e8c91caae34e3c2144f451ec045a8e1c85549d"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "492c0d6f11ac682eb4512d80", "attempt_id": "1fafdc280733", "run_group_id": "E4_core:A4|closed|qwen3b-llama3b:h1", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:21:01.362678Z", "task_family": "synthetic", "document_id": "syn_03021_2000", "document_hash": "3ac205b464552356699a4d2b0a7bf62497ead36e5bca73982e11ae5ee3810731", "question_id": "q12_unit_conversion", "split": "test", "cluster_id": "syn_03021_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 1, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1914, "query_tokens": 600, "message_tokens": 512, "message_utf8_bytes": 1179, "cache_tensor_bytes": null, "serialized_payload_bytes": 1179, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 22361.685457988642, "hops": {"1": 22361.685457988642}, "receiver_prefill_ms": 227.03187505248934, "ttft_ms": 11.922708945348859, "decode_ms": 11.946249986067414}, "total_wall_ms": 22613.093541003764, "ttft_ms": 11.922708945348859, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 38, "retention_rate": 0.5846153846153846, "retention_by_kind": {"person": 0.8333333333333334, "date": 0.7, "amount": 0.38461538461538464, "status": 1.0, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1179}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 38, "retention_rate": 0.5846153846153846, "retention_by_kind": {"person": 0.8333333333333334, "date": 0.7, "amount": 0.38461538461538464, "status": 1.0, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1179}}}, "reference_key_id": "keys/synthetic/syn_03021_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "f6342dcd02ebd80508f967d8bfed98ca4bc1afeea750629dadac4b492752e270", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "2df4dd884149fe98087de143", "attempt_id": "2afa5408fb88", "run_group_id": "E4_core:A4|closed|qwen3b-llama3b:h1", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:21:01.630041Z", "task_family": "synthetic", "document_id": "syn_03021_2000", "document_hash": "3ac205b464552356699a4d2b0a7bf62497ead36e5bca73982e11ae5ee3810731", "question_id": "q13_unanswerable", "split": "test", "cluster_id": "syn_03021_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 1, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1914, "query_tokens": 605, "message_tokens": 512, "message_utf8_bytes": 1179, "cache_tensor_bytes": null, "serialized_payload_bytes": 1179, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 22361.685457988642, "hops": {"1": 22361.685457988642}, "receiver_prefill_ms": 227.51516697462648, "ttft_ms": 12.134875054471195, "decode_ms": 11.938791954889894}, "total_wall_ms": 22613.76391700469, "ttft_ms": 12.134875054471195, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 38, "retention_rate": 0.5846153846153846, "retention_by_kind": {"person": 0.8333333333333334, "date": 0.7, "amount": 0.38461538461538464, "status": 1.0, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1179}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 38, "retention_rate": 0.5846153846153846, "retention_by_kind": {"person": 0.8333333333333334, "date": 0.7, "amount": 0.38461538461538464, "status": 1.0, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1179}}}, "reference_key_id": "keys/synthetic/syn_03021_2000.json", "task_scores": {"correct": 1.0, "abstained": 1.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "f6342dcd02ebd80508f967d8bfed98ca4bc1afeea750629dadac4b492752e270", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "c686aa3b751a76688e4f0425", "attempt_id": "2f78f3dc67fc", "run_group_id": "E4_core:A4|closed|qwen3b-llama3b:h1", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:21:01.897387Z", "task_family": "synthetic", "document_id": "syn_03021_2000", "document_hash": "3ac205b464552356699a4d2b0a7bf62497ead36e5bca73982e11ae5ee3810731", "question_id": "q14_unanswerable", "split": "test", "cluster_id": "syn_03021_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 1, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1914, "query_tokens": 607, "message_tokens": 512, "message_utf8_bytes": 1179, "cache_tensor_bytes": null, "serialized_payload_bytes": 1179, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 22361.685457988642, "hops": {"1": 22361.685457988642}, "receiver_prefill_ms": 227.29337494820356, "ttft_ms": 12.00641703326255, "decode_ms": 11.979499948211014}, "total_wall_ms": 22613.481333013624, "ttft_ms": 12.00641703326255, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 38, "retention_rate": 0.5846153846153846, "retention_by_kind": {"person": 0.8333333333333334, "date": 0.7, "amount": 0.38461538461538464, "status": 1.0, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1179}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 38, "retention_rate": 0.5846153846153846, "retention_by_kind": {"person": 0.8333333333333334, "date": 0.7, "amount": 0.38461538461538464, "status": 1.0, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1179}}}, "reference_key_id": "keys/synthetic/syn_03021_2000.json", "task_scores": {"correct": 1.0, "abstained": 1.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "f6342dcd02ebd80508f967d8bfed98ca4bc1afeea750629dadac4b492752e270", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "f862c2ccc6a03ef548071aab", "attempt_id": "045bce2d6652", "run_group_id": "E4_core:A4|closed|qwen3b-llama3b:h2", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:21:07.294655Z", "task_family": "synthetic", "document_id": "syn_03021_2000", "document_hash": "3ac205b464552356699a4d2b0a7bf62497ead36e5bca73982e11ae5ee3810731", "question_id": "q00_single_fact", "split": "test", "cluster_id": "syn_03021_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "tokenizer_hashes": {"receiver": "b9d8872cf3a99da99e1fc4b0f71232a451632dc49ca4c6771549b4efa5c1c763"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 2, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1725, "query_tokens": 523, "message_tokens": 421, "message_utf8_bytes": 1181, "cache_tensor_bytes": null, "serialized_payload_bytes": 1181, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 27490.246208035387, "hops": {"1": 22361.685457988642, "2": 5128.560750046745}, "receiver_prefill_ms": 210.22116695530713, "ttft_ms": 11.660957941785455, "decode_ms": 11.498666950501502}, "total_wall_ms": 27724.15650007315, "ttft_ms": 11.660957941785455, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 38, "retention_rate": 0.5846153846153846, "retention_by_kind": {"person": 0.8333333333333334, "date": 0.7, "amount": 0.38461538461538464, "status": 1.0, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1181}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 38, "retention_rate": 0.5846153846153846, "retention_by_kind": {"person": 0.8333333333333334, "date": 0.7, "amount": 0.38461538461538464, "status": 1.0, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1179}, "2": {"facts_total": 65, "facts_retained": 38, "retention_rate": 0.5846153846153846, "retention_by_kind": {"person": 0.8333333333333334, "date": 0.7, "amount": 0.38461538461538464, "status": 1.0, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1181}}}, "reference_key_id": "keys/synthetic/syn_03021_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "8669fc537fe26ddfe1b74689f872379bdcad840ad09064ca93b4d5535b301a41", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "db12f48a9bbcdbd11e580978", "attempt_id": "4d033e5560f0", "run_group_id": "E4_core:A4|closed|qwen3b-llama3b:h2", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:21:07.544880Z", "task_family": "synthetic", "document_id": "syn_03021_2000", "document_hash": "3ac205b464552356699a4d2b0a7bf62497ead36e5bca73982e11ae5ee3810731", "question_id": "q01_single_fact", "split": "test", "cluster_id": "syn_03021_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "tokenizer_hashes": {"receiver": "b9d8872cf3a99da99e1fc4b0f71232a451632dc49ca4c6771549b4efa5c1c763"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 2, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1725, "query_tokens": 529, "message_tokens": 421, "message_utf8_bytes": 1181, "cache_tensor_bytes": null, "serialized_payload_bytes": 1181, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 27490.246208035387, "hops": {"1": 22361.685457988642, "2": 5128.560750046745}, "receiver_prefill_ms": 209.99633392784745, "ttft_ms": 11.670165928080678, "decode_ms": 115.16645806841552}, "total_wall_ms": 27827.570000081323, "ttft_ms": 11.670165928080678, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 10}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "Lot 536-R, Senneterre.", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 38, "retention_rate": 0.5846153846153846, "retention_by_kind": {"person": 0.8333333333333334, "date": 0.7, "amount": 0.38461538461538464, "status": 1.0, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1181}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 38, "retention_rate": 0.5846153846153846, "retention_by_kind": {"person": 0.8333333333333334, "date": 0.7, "amount": 0.38461538461538464, "status": 1.0, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1179}, "2": {"facts_total": 65, "facts_retained": 38, "retention_rate": 0.5846153846153846, "retention_by_kind": {"person": 0.8333333333333334, "date": 0.7, "amount": 0.38461538461538464, "status": 1.0, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1181}}}, "reference_key_id": "keys/synthetic/syn_03021_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 0.8571428571428571}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "8669fc537fe26ddfe1b74689f872379bdcad840ad09064ca93b4d5535b301a41", "output_hash": "d34fb8d4553d115f1c915ac9be75f68ef887916efddf5ed339f7cf4236ce8fa8"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "20467eed2b5a943e1a5f0fa7", "attempt_id": "4d1c73b5afe8", "run_group_id": "E4_core:A4|closed|qwen3b-llama3b:h2", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:21:07.898088Z", "task_family": "synthetic", "document_id": "syn_03021_2000", "document_hash": "3ac205b464552356699a4d2b0a7bf62497ead36e5bca73982e11ae5ee3810731", "question_id": "q02_exact_copy", "split": "test", "cluster_id": "syn_03021_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "tokenizer_hashes": {"receiver": "b9d8872cf3a99da99e1fc4b0f71232a451632dc49ca4c6771549b4efa5c1c763"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 2, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1725, "query_tokens": 529, "message_tokens": 421, "message_utf8_bytes": 1181, "cache_tensor_bytes": null, "serialized_payload_bytes": 1181, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 27490.246208035387, "hops": {"1": 22361.685457988642, "2": 5128.560750046745}, "receiver_prefill_ms": 210.57508408557624, "ttft_ms": 11.598790995776653, "decode_ms": 45.95241602510214}, "total_wall_ms": 27758.832333027385, "ttft_ms": 11.598790995776653, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 4}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "88.7751", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 38, "retention_rate": 0.5846153846153846, "retention_by_kind": {"person": 0.8333333333333334, "date": 0.7, "amount": 0.38461538461538464, "status": 1.0, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1181}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 38, "retention_rate": 0.5846153846153846, "retention_by_kind": {"person": 0.8333333333333334, "date": 0.7, "amount": 0.38461538461538464, "status": 1.0, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1179}, "2": {"facts_total": 65, "facts_retained": 38, "retention_rate": 0.5846153846153846, "retention_by_kind": {"person": 0.8333333333333334, "date": 0.7, "amount": 0.38461538461538464, "status": 1.0, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1181}}}, "reference_key_id": "keys/synthetic/syn_03021_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "8669fc537fe26ddfe1b74689f872379bdcad840ad09064ca93b4d5535b301a41", "output_hash": "1e01e62c69f9479b681e50e59e8565c6583f5cdb803446a05ed5cb13118acdf9"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "763d60481be4b5a97d68cdf6", "attempt_id": "9929b23e05b4", "run_group_id": "E4_core:A4|closed|qwen3b-llama3b:h2", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:21:08.182280Z", "task_family": "synthetic", "document_id": "syn_03021_2000", "document_hash": "3ac205b464552356699a4d2b0a7bf62497ead36e5bca73982e11ae5ee3810731", "question_id": "q03_exact_copy", "split": "test", "cluster_id": "syn_03021_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "tokenizer_hashes": {"receiver": "b9d8872cf3a99da99e1fc4b0f71232a451632dc49ca4c6771549b4efa5c1c763"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 2, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1725, "query_tokens": 526, "message_tokens": 421, "message_utf8_bytes": 1181, "cache_tensor_bytes": null, "serialized_payload_bytes": 1181, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 27490.246208035387, "hops": {"1": 22361.685457988642, "2": 5128.560750046745}, "receiver_prefill_ms": 210.09791700635105, "ttft_ms": 11.771790916100144, "decode_ms": 57.52425000537187}, "total_wall_ms": 27770.269833039492, "ttft_ms": 11.771790916100144, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 5}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "Ingrid Mbeki.", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 38, "retention_rate": 0.5846153846153846, "retention_by_kind": {"person": 0.8333333333333334, "date": 0.7, "amount": 0.38461538461538464, "status": 1.0, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1181}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 38, "retention_rate": 0.5846153846153846, "retention_by_kind": {"person": 0.8333333333333334, "date": 0.7, "amount": 0.38461538461538464, "status": 1.0, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1179}, "2": {"facts_total": 65, "facts_retained": 38, "retention_rate": 0.5846153846153846, "retention_by_kind": {"person": 0.8333333333333334, "date": 0.7, "amount": 0.38461538461538464, "status": 1.0, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1181}}}, "reference_key_id": "keys/synthetic/syn_03021_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "8669fc537fe26ddfe1b74689f872379bdcad840ad09064ca93b4d5535b301a41", "output_hash": "4b9af6bc1085ca0df8d10d93274bfa44fbaf913b3380b2525ac5097c55375881"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "03a50238aa5cbe37b7e41a6b", "attempt_id": "0db3937a6c23", "run_group_id": "E4_core:A4|closed|qwen3b-llama3b:h2", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:21:08.477645Z", "task_family": "synthetic", "document_id": "syn_03021_2000", "document_hash": "3ac205b464552356699a4d2b0a7bf62497ead36e5bca73982e11ae5ee3810731", "question_id": "q04_multi_hop", "split": "test", "cluster_id": "syn_03021_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "tokenizer_hashes": {"receiver": "b9d8872cf3a99da99e1fc4b0f71232a451632dc49ca4c6771549b4efa5c1c763"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 2, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1725, "query_tokens": 533, "message_tokens": 421, "message_utf8_bytes": 1181, "cache_tensor_bytes": null, "serialized_payload_bytes": 1181, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 27490.246208035387, "hops": {"1": 22361.685457988642, "2": 5128.560750046745}, "receiver_prefill_ms": 210.60975000727922, "ttft_ms": 11.655749985948205, "decode_ms": 11.569624999538064}, "total_wall_ms": 27724.703874089755, "ttft_ms": 11.655749985948205, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 38, "retention_rate": 0.5846153846153846, "retention_by_kind": {"person": 0.8333333333333334, "date": 0.7, "amount": 0.38461538461538464, "status": 1.0, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1181}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 38, "retention_rate": 0.5846153846153846, "retention_by_kind": {"person": 0.8333333333333334, "date": 0.7, "amount": 0.38461538461538464, "status": 1.0, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1179}, "2": {"facts_total": 65, "facts_retained": 38, "retention_rate": 0.5846153846153846, "retention_by_kind": {"person": 0.8333333333333334, "date": 0.7, "amount": 0.38461538461538464, "status": 1.0, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1181}}}, "reference_key_id": "keys/synthetic/syn_03021_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "8669fc537fe26ddfe1b74689f872379bdcad840ad09064ca93b4d5535b301a41", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "43c03f1b00667b18510f48ab", "attempt_id": "eb438b0eca2a", "run_group_id": "E4_core:A4|closed|qwen3b-llama3b:h2", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:21:08.728978Z", "task_family": "synthetic", "document_id": "syn_03021_2000", "document_hash": "3ac205b464552356699a4d2b0a7bf62497ead36e5bca73982e11ae5ee3810731", "question_id": "q05_multi_hop", "split": "test", "cluster_id": "syn_03021_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "tokenizer_hashes": {"receiver": "b9d8872cf3a99da99e1fc4b0f71232a451632dc49ca4c6771549b4efa5c1c763"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 2, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1725, "query_tokens": 532, "message_tokens": 421, "message_utf8_bytes": 1181, "cache_tensor_bytes": null, "serialized_payload_bytes": 1181, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 27490.246208035387, "hops": {"1": 22361.685457988642, "2": 5128.560750046745}, "receiver_prefill_ms": 210.614874958992, "ttft_ms": 11.602125014178455, "decode_ms": 92.22799993585795}, "total_wall_ms": 27805.317124002613, "ttft_ms": 11.602125014178455, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 8}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "BQ-59Z-6753", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 38, "retention_rate": 0.5846153846153846, "retention_by_kind": {"person": 0.8333333333333334, "date": 0.7, "amount": 0.38461538461538464, "status": 1.0, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1181}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 38, "retention_rate": 0.5846153846153846, "retention_by_kind": {"person": 0.8333333333333334, "date": 0.7, "amount": 0.38461538461538464, "status": 1.0, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1179}, "2": {"facts_total": 65, "facts_retained": 38, "retention_rate": 0.5846153846153846, "retention_by_kind": {"person": 0.8333333333333334, "date": 0.7, "amount": 0.38461538461538464, "status": 1.0, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1181}}}, "reference_key_id": "keys/synthetic/syn_03021_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "8669fc537fe26ddfe1b74689f872379bdcad840ad09064ca93b4d5535b301a41", "output_hash": "db250f072c856fe7ddf4bbbd16e20c71b8bad1eef5e4480acd0dac30f8ffe9cd"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "c675939ad6211efbb6373c30", "attempt_id": "92017bd50999", "run_group_id": "E4_core:A4|closed|qwen3b-llama3b:h2", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:21:09.060414Z", "task_family": "synthetic", "document_id": "syn_03021_2000", "document_hash": "3ac205b464552356699a4d2b0a7bf62497ead36e5bca73982e11ae5ee3810731", "question_id": "q06_negation", "split": "test", "cluster_id": "syn_03021_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "tokenizer_hashes": {"receiver": "b9d8872cf3a99da99e1fc4b0f71232a451632dc49ca4c6771549b4efa5c1c763"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 2, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1725, "query_tokens": 524, "message_tokens": 421, "message_utf8_bytes": 1181, "cache_tensor_bytes": null, "serialized_payload_bytes": 1181, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 27490.246208035387, "hops": {"1": 22361.685457988642, "2": 5128.560750046745}, "receiver_prefill_ms": 209.64662497863173, "ttft_ms": 11.932790977880359, "decode_ms": 23.21420900989324}, "total_wall_ms": 27735.703124082647, "ttft_ms": 11.932790977880359, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 2}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "No.", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 38, "retention_rate": 0.5846153846153846, "retention_by_kind": {"person": 0.8333333333333334, "date": 0.7, "amount": 0.38461538461538464, "status": 1.0, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1181}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 38, "retention_rate": 0.5846153846153846, "retention_by_kind": {"person": 0.8333333333333334, "date": 0.7, "amount": 0.38461538461538464, "status": 1.0, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1179}, "2": {"facts_total": 65, "facts_retained": 38, "retention_rate": 0.5846153846153846, "retention_by_kind": {"person": 0.8333333333333334, "date": 0.7, "amount": 0.38461538461538464, "status": 1.0, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1181}}}, "reference_key_id": "keys/synthetic/syn_03021_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "8669fc537fe26ddfe1b74689f872379bdcad840ad09064ca93b4d5535b301a41", "output_hash": "38178a20b470cfd18299fb1593dd4ba706f1a83321dac051d01086efd8b7a96f"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "83fbc27fe5935e587845e2bb", "attempt_id": "72dfab763b07", "run_group_id": "E4_core:A4|closed|qwen3b-llama3b:h2", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:21:09.321282Z", "task_family": "synthetic", "document_id": "syn_03021_2000", "document_hash": "3ac205b464552356699a4d2b0a7bf62497ead36e5bca73982e11ae5ee3810731", "question_id": "q07_negation", "split": "test", "cluster_id": "syn_03021_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "tokenizer_hashes": {"receiver": "b9d8872cf3a99da99e1fc4b0f71232a451632dc49ca4c6771549b4efa5c1c763"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 2, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1725, "query_tokens": 525, "message_tokens": 421, "message_utf8_bytes": 1181, "cache_tensor_bytes": null, "serialized_payload_bytes": 1181, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 27490.246208035387, "hops": {"1": 22361.685457988642, "2": 5128.560750046745}, "receiver_prefill_ms": 210.05875000264496, "ttft_ms": 11.809790972620249, "decode_ms": 23.053209064528346}, "total_wall_ms": 27735.625999979675, "ttft_ms": 11.809790972620249, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 2}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "Yes.", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 38, "retention_rate": 0.5846153846153846, "retention_by_kind": {"person": 0.8333333333333334, "date": 0.7, "amount": 0.38461538461538464, "status": 1.0, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1181}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 38, "retention_rate": 0.5846153846153846, "retention_by_kind": {"person": 0.8333333333333334, "date": 0.7, "amount": 0.38461538461538464, "status": 1.0, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1179}, "2": {"facts_total": 65, "facts_retained": 38, "retention_rate": 0.5846153846153846, "retention_by_kind": {"person": 0.8333333333333334, "date": 0.7, "amount": 0.38461538461538464, "status": 1.0, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1181}}}, "reference_key_id": "keys/synthetic/syn_03021_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "8669fc537fe26ddfe1b74689f872379bdcad840ad09064ca93b4d5535b301a41", "output_hash": "5f9a2b795615ba6a3d5455fd5624d773fbca5bcd16249c421fd37411dc9837da"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "ad46b8a8e7e6aac75a20b736", "attempt_id": "67cbc96658cc", "run_group_id": "E4_core:A4|closed|qwen3b-llama3b:h2", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:21:09.583398Z", "task_family": "synthetic", "document_id": "syn_03021_2000", "document_hash": "3ac205b464552356699a4d2b0a7bf62497ead36e5bca73982e11ae5ee3810731", "question_id": "q08_revision", "split": "test", "cluster_id": "syn_03021_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "tokenizer_hashes": {"receiver": "b9d8872cf3a99da99e1fc4b0f71232a451632dc49ca4c6771549b4efa5c1c763"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 2, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1725, "query_tokens": 541, "message_tokens": 421, "message_utf8_bytes": 1181, "cache_tensor_bytes": null, "serialized_payload_bytes": 1181, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 27490.246208035387, "hops": {"1": 22361.685457988642, "2": 5128.560750046745}, "receiver_prefill_ms": 210.89633298106492, "ttft_ms": 11.642707977443933, "decode_ms": 11.571875074878335}, "total_wall_ms": 27724.89658300765, "ttft_ms": 11.642707977443933, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 38, "retention_rate": 0.5846153846153846, "retention_by_kind": {"person": 0.8333333333333334, "date": 0.7, "amount": 0.38461538461538464, "status": 1.0, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1181}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 38, "retention_rate": 0.5846153846153846, "retention_by_kind": {"person": 0.8333333333333334, "date": 0.7, "amount": 0.38461538461538464, "status": 1.0, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1179}, "2": {"facts_total": 65, "facts_retained": 38, "retention_rate": 0.5846153846153846, "retention_by_kind": {"person": 0.8333333333333334, "date": 0.7, "amount": 0.38461538461538464, "status": 1.0, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1181}}}, "reference_key_id": "keys/synthetic/syn_03021_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "8669fc537fe26ddfe1b74689f872379bdcad840ad09064ca93b4d5535b301a41", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "c05955bfb683065921a22812", "attempt_id": "1e21de775118", "run_group_id": "E4_core:A4|closed|qwen3b-llama3b:h2", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:21:09.833444Z", "task_family": "synthetic", "document_id": "syn_03021_2000", "document_hash": "3ac205b464552356699a4d2b0a7bf62497ead36e5bca73982e11ae5ee3810731", "question_id": "q09_revision", "split": "test", "cluster_id": "syn_03021_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "tokenizer_hashes": {"receiver": "b9d8872cf3a99da99e1fc4b0f71232a451632dc49ca4c6771549b4efa5c1c763"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 2, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1725, "query_tokens": 532, "message_tokens": 421, "message_utf8_bytes": 1181, "cache_tensor_bytes": null, "serialized_payload_bytes": 1181, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 27490.246208035387, "hops": {"1": 22361.685457988642, "2": 5128.560750046745}, "receiver_prefill_ms": 210.49687499180436, "ttft_ms": 11.573707917705178, "decode_ms": 11.591500020585954}, "total_wall_ms": 27724.35154102277, "ttft_ms": 11.573707917705178, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 38, "retention_rate": 0.5846153846153846, "retention_by_kind": {"person": 0.8333333333333334, "date": 0.7, "amount": 0.38461538461538464, "status": 1.0, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1181}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 38, "retention_rate": 0.5846153846153846, "retention_by_kind": {"person": 0.8333333333333334, "date": 0.7, "amount": 0.38461538461538464, "status": 1.0, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1179}, "2": {"facts_total": 65, "facts_retained": 38, "retention_rate": 0.5846153846153846, "retention_by_kind": {"person": 0.8333333333333334, "date": 0.7, "amount": 0.38461538461538464, "status": 1.0, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1181}}}, "reference_key_id": "keys/synthetic/syn_03021_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "8669fc537fe26ddfe1b74689f872379bdcad840ad09064ca93b4d5535b301a41", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "9b4347ff7944f073058090bf", "attempt_id": "be92416beb33", "run_group_id": "E4_core:A4|closed|qwen3b-llama3b:h2", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:21:10.083222Z", "task_family": "synthetic", "document_id": "syn_03021_2000", "document_hash": "3ac205b464552356699a4d2b0a7bf62497ead36e5bca73982e11ae5ee3810731", "question_id": "q10_temporal", "split": "test", "cluster_id": "syn_03021_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "tokenizer_hashes": {"receiver": "b9d8872cf3a99da99e1fc4b0f71232a451632dc49ca4c6771549b4efa5c1c763"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 2, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1725, "query_tokens": 540, "message_tokens": 421, "message_utf8_bytes": 1181, "cache_tensor_bytes": null, "serialized_payload_bytes": 1181, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 27490.246208035387, "hops": {"1": 22361.685457988642, "2": 5128.560750046745}, "receiver_prefill_ms": 210.98874998278916, "ttft_ms": 11.540542007423937, "decode_ms": 11.573166935704648}, "total_wall_ms": 27724.788957973942, "ttft_ms": 11.540542007423937, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 38, "retention_rate": 0.5846153846153846, "retention_by_kind": {"person": 0.8333333333333334, "date": 0.7, "amount": 0.38461538461538464, "status": 1.0, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1181}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 38, "retention_rate": 0.5846153846153846, "retention_by_kind": {"person": 0.8333333333333334, "date": 0.7, "amount": 0.38461538461538464, "status": 1.0, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1179}, "2": {"facts_total": 65, "facts_retained": 38, "retention_rate": 0.5846153846153846, "retention_by_kind": {"person": 0.8333333333333334, "date": 0.7, "amount": 0.38461538461538464, "status": 1.0, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1181}}}, "reference_key_id": "keys/synthetic/syn_03021_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "8669fc537fe26ddfe1b74689f872379bdcad840ad09064ca93b4d5535b301a41", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "c3923e64928b533ab815197f", "attempt_id": "974a25d8a3b2", "run_group_id": "E4_core:A4|closed|qwen3b-llama3b:h2", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:21:10.332740Z", "task_family": "synthetic", "document_id": "syn_03021_2000", "document_hash": "3ac205b464552356699a4d2b0a7bf62497ead36e5bca73982e11ae5ee3810731", "question_id": "q11_aggregation", "split": "test", "cluster_id": "syn_03021_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "tokenizer_hashes": {"receiver": "b9d8872cf3a99da99e1fc4b0f71232a451632dc49ca4c6771549b4efa5c1c763"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 2, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1725, "query_tokens": 536, "message_tokens": 421, "message_utf8_bytes": 1181, "cache_tensor_bytes": null, "serialized_payload_bytes": 1181, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 27490.246208035387, "hops": {"1": 22361.685457988642, "2": 5128.560750046745}, "receiver_prefill_ms": 210.7438329840079, "ttft_ms": 11.577583965845406, "decode_ms": 219.00204208213836}, "total_wall_ms": 27932.060915976763, "ttft_ms": 11.577583965845406, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 19}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "4,862,425 + 5,646,720 = 10,509,145", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 38, "retention_rate": 0.5846153846153846, "retention_by_kind": {"person": 0.8333333333333334, "date": 0.7, "amount": 0.38461538461538464, "status": 1.0, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1181}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 38, "retention_rate": 0.5846153846153846, "retention_by_kind": {"person": 0.8333333333333334, "date": 0.7, "amount": 0.38461538461538464, "status": 1.0, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1179}, "2": {"facts_total": 65, "facts_retained": 38, "retention_rate": 0.5846153846153846, "retention_by_kind": {"person": 0.8333333333333334, "date": 0.7, "amount": 0.38461538461538464, "status": 1.0, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1181}}}, "reference_key_id": "keys/synthetic/syn_03021_2000.json", "task_scores": {"correct": 0.0, "abstained": 0.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "8669fc537fe26ddfe1b74689f872379bdcad840ad09064ca93b4d5535b301a41", "output_hash": "038d9ed23c4639e6ae998b25ae021dada1e0ab0526113d0a0e4d8884ea4b4e6f"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "b8340747be15e7671266d69d", "attempt_id": "74b2ac82de5f", "run_group_id": "E4_core:A4|closed|qwen3b-llama3b:h2", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:21:10.789438Z", "task_family": "synthetic", "document_id": "syn_03021_2000", "document_hash": "3ac205b464552356699a4d2b0a7bf62497ead36e5bca73982e11ae5ee3810731", "question_id": "q12_unit_conversion", "split": "test", "cluster_id": "syn_03021_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "tokenizer_hashes": {"receiver": "b9d8872cf3a99da99e1fc4b0f71232a451632dc49ca4c6771549b4efa5c1c763"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 2, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1725, "query_tokens": 531, "message_tokens": 421, "message_utf8_bytes": 1181, "cache_tensor_bytes": null, "serialized_payload_bytes": 1181, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 27490.246208035387, "hops": {"1": 22361.685457988642, "2": 5128.560750046745}, "receiver_prefill_ms": 210.57391597423702, "ttft_ms": 11.641624965704978, "decode_ms": 11.491750017739832}, "total_wall_ms": 27724.412208073772, "ttft_ms": 11.641624965704978, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 38, "retention_rate": 0.5846153846153846, "retention_by_kind": {"person": 0.8333333333333334, "date": 0.7, "amount": 0.38461538461538464, "status": 1.0, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1181}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 38, "retention_rate": 0.5846153846153846, "retention_by_kind": {"person": 0.8333333333333334, "date": 0.7, "amount": 0.38461538461538464, "status": 1.0, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1179}, "2": {"facts_total": 65, "facts_retained": 38, "retention_rate": 0.5846153846153846, "retention_by_kind": {"person": 0.8333333333333334, "date": 0.7, "amount": 0.38461538461538464, "status": 1.0, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1181}}}, "reference_key_id": "keys/synthetic/syn_03021_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "8669fc537fe26ddfe1b74689f872379bdcad840ad09064ca93b4d5535b301a41", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "54a7b3e6aea0779e89e63106", "attempt_id": "6291ce6a3a2a", "run_group_id": "E4_core:A4|closed|qwen3b-llama3b:h2", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:21:11.038833Z", "task_family": "synthetic", "document_id": "syn_03021_2000", "document_hash": "3ac205b464552356699a4d2b0a7bf62497ead36e5bca73982e11ae5ee3810731", "question_id": "q13_unanswerable", "split": "test", "cluster_id": "syn_03021_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "tokenizer_hashes": {"receiver": "b9d8872cf3a99da99e1fc4b0f71232a451632dc49ca4c6771549b4efa5c1c763"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 2, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1725, "query_tokens": 535, "message_tokens": 421, "message_utf8_bytes": 1181, "cache_tensor_bytes": null, "serialized_payload_bytes": 1181, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 27490.246208035387, "hops": {"1": 22361.685457988642, "2": 5128.560750046745}, "receiver_prefill_ms": 210.2422920288518, "ttft_ms": 11.586208012886345, "decode_ms": 11.50383404456079}, "total_wall_ms": 27724.0187501302, "ttft_ms": 11.586208012886345, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 38, "retention_rate": 0.5846153846153846, "retention_by_kind": {"person": 0.8333333333333334, "date": 0.7, "amount": 0.38461538461538464, "status": 1.0, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1181}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 38, "retention_rate": 0.5846153846153846, "retention_by_kind": {"person": 0.8333333333333334, "date": 0.7, "amount": 0.38461538461538464, "status": 1.0, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1179}, "2": {"facts_total": 65, "facts_retained": 38, "retention_rate": 0.5846153846153846, "retention_by_kind": {"person": 0.8333333333333334, "date": 0.7, "amount": 0.38461538461538464, "status": 1.0, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1181}}}, "reference_key_id": "keys/synthetic/syn_03021_2000.json", "task_scores": {"correct": 1.0, "abstained": 1.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "8669fc537fe26ddfe1b74689f872379bdcad840ad09064ca93b4d5535b301a41", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "50a80ab68c959295f0b3f3d5", "attempt_id": "15b0346e5d8c", "run_group_id": "E4_core:A4|closed|qwen3b-llama3b:h2", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:21:11.287613Z", "task_family": "synthetic", "document_id": "syn_03021_2000", "document_hash": "3ac205b464552356699a4d2b0a7bf62497ead36e5bca73982e11ae5ee3810731", "question_id": "q14_unanswerable", "split": "test", "cluster_id": "syn_03021_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "tokenizer_hashes": {"receiver": "b9d8872cf3a99da99e1fc4b0f71232a451632dc49ca4c6771549b4efa5c1c763"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 2, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1725, "query_tokens": 538, "message_tokens": 421, "message_utf8_bytes": 1181, "cache_tensor_bytes": null, "serialized_payload_bytes": 1181, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 27490.246208035387, "hops": {"1": 22361.685457988642, "2": 5128.560750046745}, "receiver_prefill_ms": 210.84425004664809, "ttft_ms": 11.568416957743466, "decode_ms": 11.542583000846207}, "total_wall_ms": 27724.645416019484, "ttft_ms": 11.568416957743466, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 38, "retention_rate": 0.5846153846153846, "retention_by_kind": {"person": 0.8333333333333334, "date": 0.7, "amount": 0.38461538461538464, "status": 1.0, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1181}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 38, "retention_rate": 0.5846153846153846, "retention_by_kind": {"person": 0.8333333333333334, "date": 0.7, "amount": 0.38461538461538464, "status": 1.0, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1179}, "2": {"facts_total": 65, "facts_retained": 38, "retention_rate": 0.5846153846153846, "retention_by_kind": {"person": 0.8333333333333334, "date": 0.7, "amount": 0.38461538461538464, "status": 1.0, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1181}}}, "reference_key_id": "keys/synthetic/syn_03021_2000.json", "task_scores": {"correct": 1.0, "abstained": 1.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "8669fc537fe26ddfe1b74689f872379bdcad840ad09064ca93b4d5535b301a41", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "8f8e838994a5174ead46b7d5", "attempt_id": "8cf5d1f0b1e4", "run_group_id": "E4_core:A4|closed|qwen3b-llama3b:h4", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:21:21.918304Z", "task_family": "synthetic", "document_id": "syn_03021_2000", "document_hash": "3ac205b464552356699a4d2b0a7bf62497ead36e5bca73982e11ae5ee3810731", "question_id": "q00_single_fact", "split": "test", "cluster_id": "syn_03021_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "tokenizer_hashes": {"receiver": "b9d8872cf3a99da99e1fc4b0f71232a451632dc49ca4c6771549b4efa5c1c763"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 4, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1725, "query_tokens": 487, "message_tokens": 385, "message_utf8_bytes": 1111, "cache_tensor_bytes": null, "serialized_payload_bytes": 1111, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 37869.45949902292, "hops": {"1": 22361.685457988642, "2": 5128.560750046745, "3": 5686.298708082177, "4": 4692.91458290536}, "receiver_prefill_ms": 207.54650002345443, "ttft_ms": 11.57233293633908, "decode_ms": 11.76929206121713}, "total_wall_ms": 38100.84308194928, "ttft_ms": 11.57233293633908, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 38, "retention_rate": 0.5846153846153846, "retention_by_kind": {"person": 0.8333333333333334, "date": 0.7, "amount": 0.38461538461538464, "status": 1.0, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1111}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 38, "retention_rate": 0.5846153846153846, "retention_by_kind": {"person": 0.8333333333333334, "date": 0.7, "amount": 0.38461538461538464, "status": 1.0, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1179}, "2": {"facts_total": 65, "facts_retained": 38, "retention_rate": 0.5846153846153846, "retention_by_kind": {"person": 0.8333333333333334, "date": 0.7, "amount": 0.38461538461538464, "status": 1.0, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1181}, "3": {"facts_total": 65, "facts_retained": 38, "retention_rate": 0.5846153846153846, "retention_by_kind": {"person": 0.8333333333333334, "date": 0.7, "amount": 0.38461538461538464, "status": 1.0, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1095}, "4": {"facts_total": 65, "facts_retained": 38, "retention_rate": 0.5846153846153846, "retention_by_kind": {"person": 0.8333333333333334, "date": 0.7, "amount": 0.38461538461538464, "status": 1.0, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1111}}}, "reference_key_id": "keys/synthetic/syn_03021_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "1fb6b0e63657986ace0df6d3c1195f462c1359afc69ed5a7dad3ca3220aa6437", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "2db0d38245b2405e048a54e0", "attempt_id": "70b698c53a4f", "run_group_id": "E4_core:A4|closed|qwen3b-llama3b:h4", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:21:22.166273Z", "task_family": "synthetic", "document_id": "syn_03021_2000", "document_hash": "3ac205b464552356699a4d2b0a7bf62497ead36e5bca73982e11ae5ee3810731", "question_id": "q01_single_fact", "split": "test", "cluster_id": "syn_03021_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "tokenizer_hashes": {"receiver": "b9d8872cf3a99da99e1fc4b0f71232a451632dc49ca4c6771549b4efa5c1c763"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 4, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1725, "query_tokens": 493, "message_tokens": 385, "message_utf8_bytes": 1111, "cache_tensor_bytes": null, "serialized_payload_bytes": 1111, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 37869.45949902292, "hops": {"1": 22361.685457988642, "2": 5128.560750046745, "3": 5686.298708082177, "4": 4692.91458290536}, "receiver_prefill_ms": 209.6134580206126, "ttft_ms": 11.551541974768043, "decode_ms": 115.2444159379229}, "total_wall_ms": 38206.339915981516, "ttft_ms": 11.551541974768043, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 10}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "Lot 536-R, Senneterre.", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 38, "retention_rate": 0.5846153846153846, "retention_by_kind": {"person": 0.8333333333333334, "date": 0.7, "amount": 0.38461538461538464, "status": 1.0, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1111}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 38, "retention_rate": 0.5846153846153846, "retention_by_kind": {"person": 0.8333333333333334, "date": 0.7, "amount": 0.38461538461538464, "status": 1.0, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1179}, "2": {"facts_total": 65, "facts_retained": 38, "retention_rate": 0.5846153846153846, "retention_by_kind": {"person": 0.8333333333333334, "date": 0.7, "amount": 0.38461538461538464, "status": 1.0, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1181}, "3": {"facts_total": 65, "facts_retained": 38, "retention_rate": 0.5846153846153846, "retention_by_kind": {"person": 0.8333333333333334, "date": 0.7, "amount": 0.38461538461538464, "status": 1.0, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1095}, "4": {"facts_total": 65, "facts_retained": 38, "retention_rate": 0.5846153846153846, "retention_by_kind": {"person": 0.8333333333333334, "date": 0.7, "amount": 0.38461538461538464, "status": 1.0, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1111}}}, "reference_key_id": "keys/synthetic/syn_03021_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 0.8571428571428571}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "1fb6b0e63657986ace0df6d3c1195f462c1359afc69ed5a7dad3ca3220aa6437", "output_hash": "d34fb8d4553d115f1c915ac9be75f68ef887916efddf5ed339f7cf4236ce8fa8"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "c274063c0b31713f29296ab0", "attempt_id": "ff432b7730f3", "run_group_id": "E4_core:A4|closed|qwen3b-llama3b:h4", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:21:22.518638Z", "task_family": "synthetic", "document_id": "syn_03021_2000", "document_hash": "3ac205b464552356699a4d2b0a7bf62497ead36e5bca73982e11ae5ee3810731", "question_id": "q02_exact_copy", "split": "test", "cluster_id": "syn_03021_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "tokenizer_hashes": {"receiver": "b9d8872cf3a99da99e1fc4b0f71232a451632dc49ca4c6771549b4efa5c1c763"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 4, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1725, "query_tokens": 493, "message_tokens": 385, "message_utf8_bytes": 1111, "cache_tensor_bytes": null, "serialized_payload_bytes": 1111, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 37869.45949902292, "hops": {"1": 22361.685457988642, "2": 5128.560750046745, "3": 5686.298708082177, "4": 4692.91458290536}, "receiver_prefill_ms": 206.88049995806068, "ttft_ms": 11.547166970558465, "decode_ms": 45.98920908756554}, "total_wall_ms": 38134.30274894927, "ttft_ms": 11.547166970558465, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 4}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "88.7751", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 38, "retention_rate": 0.5846153846153846, "retention_by_kind": {"person": 0.8333333333333334, "date": 0.7, "amount": 0.38461538461538464, "status": 1.0, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1111}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 38, "retention_rate": 0.5846153846153846, "retention_by_kind": {"person": 0.8333333333333334, "date": 0.7, "amount": 0.38461538461538464, "status": 1.0, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1179}, "2": {"facts_total": 65, "facts_retained": 38, "retention_rate": 0.5846153846153846, "retention_by_kind": {"person": 0.8333333333333334, "date": 0.7, "amount": 0.38461538461538464, "status": 1.0, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1181}, "3": {"facts_total": 65, "facts_retained": 38, "retention_rate": 0.5846153846153846, "retention_by_kind": {"person": 0.8333333333333334, "date": 0.7, "amount": 0.38461538461538464, "status": 1.0, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1095}, "4": {"facts_total": 65, "facts_retained": 38, "retention_rate": 0.5846153846153846, "retention_by_kind": {"person": 0.8333333333333334, "date": 0.7, "amount": 0.38461538461538464, "status": 1.0, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1111}}}, "reference_key_id": "keys/synthetic/syn_03021_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "1fb6b0e63657986ace0df6d3c1195f462c1359afc69ed5a7dad3ca3220aa6437", "output_hash": "1e01e62c69f9479b681e50e59e8565c6583f5cdb803446a05ed5cb13118acdf9"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "08abe6ac6fa2a1af4a65fef3", "attempt_id": "5d5f05aef9e0", "run_group_id": "E4_core:A4|closed|qwen3b-llama3b:h4", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:21:22.798662Z", "task_family": "synthetic", "document_id": "syn_03021_2000", "document_hash": "3ac205b464552356699a4d2b0a7bf62497ead36e5bca73982e11ae5ee3810731", "question_id": "q03_exact_copy", "split": "test", "cluster_id": "syn_03021_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "tokenizer_hashes": {"receiver": "b9d8872cf3a99da99e1fc4b0f71232a451632dc49ca4c6771549b4efa5c1c763"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 4, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1725, "query_tokens": 490, "message_tokens": 385, "message_utf8_bytes": 1111, "cache_tensor_bytes": null, "serialized_payload_bytes": 1111, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 37869.45949902292, "hops": {"1": 22361.685457988642, "2": 5128.560750046745, "3": 5686.298708082177, "4": 4692.91458290536}, "receiver_prefill_ms": 205.76804096344858, "ttft_ms": 11.562332976609468, "decode_ms": 45.92741699889302}, "total_wall_ms": 38133.16487404518, "ttft_ms": 11.562332976609468, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 4}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "Ingrid Mbeki", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 38, "retention_rate": 0.5846153846153846, "retention_by_kind": {"person": 0.8333333333333334, "date": 0.7, "amount": 0.38461538461538464, "status": 1.0, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1111}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 38, "retention_rate": 0.5846153846153846, "retention_by_kind": {"person": 0.8333333333333334, "date": 0.7, "amount": 0.38461538461538464, "status": 1.0, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1179}, "2": {"facts_total": 65, "facts_retained": 38, "retention_rate": 0.5846153846153846, "retention_by_kind": {"person": 0.8333333333333334, "date": 0.7, "amount": 0.38461538461538464, "status": 1.0, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1181}, "3": {"facts_total": 65, "facts_retained": 38, "retention_rate": 0.5846153846153846, "retention_by_kind": {"person": 0.8333333333333334, "date": 0.7, "amount": 0.38461538461538464, "status": 1.0, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1095}, "4": {"facts_total": 65, "facts_retained": 38, "retention_rate": 0.5846153846153846, "retention_by_kind": {"person": 0.8333333333333334, "date": 0.7, "amount": 0.38461538461538464, "status": 1.0, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1111}}}, "reference_key_id": "keys/synthetic/syn_03021_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "1fb6b0e63657986ace0df6d3c1195f462c1359afc69ed5a7dad3ca3220aa6437", "output_hash": "1dbd85090bb65fb3b72de8ca4ce32f2c234729ff4a239274e79fb574b800a79d"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "79e4060b1f6f6b4884b63681", "attempt_id": "172508a55104", "run_group_id": "E4_core:A4|closed|qwen3b-llama3b:h4", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:21:23.077990Z", "task_family": "synthetic", "document_id": "syn_03021_2000", "document_hash": "3ac205b464552356699a4d2b0a7bf62497ead36e5bca73982e11ae5ee3810731", "question_id": "q04_multi_hop", "split": "test", "cluster_id": "syn_03021_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "tokenizer_hashes": {"receiver": "b9d8872cf3a99da99e1fc4b0f71232a451632dc49ca4c6771549b4efa5c1c763"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 4, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1725, "query_tokens": 497, "message_tokens": 385, "message_utf8_bytes": 1111, "cache_tensor_bytes": null, "serialized_payload_bytes": 1111, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 37869.45949902292, "hops": {"1": 22361.685457988642, "2": 5128.560750046745, "3": 5686.298708082177, "4": 4692.91458290536}, "receiver_prefill_ms": 208.45854096114635, "ttft_ms": 11.590208043344319, "decode_ms": 11.560541926883161}, "total_wall_ms": 38101.5318329446, "ttft_ms": 11.590208043344319, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 38, "retention_rate": 0.5846153846153846, "retention_by_kind": {"person": 0.8333333333333334, "date": 0.7, "amount": 0.38461538461538464, "status": 1.0, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1111}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 38, "retention_rate": 0.5846153846153846, "retention_by_kind": {"person": 0.8333333333333334, "date": 0.7, "amount": 0.38461538461538464, "status": 1.0, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1179}, "2": {"facts_total": 65, "facts_retained": 38, "retention_rate": 0.5846153846153846, "retention_by_kind": {"person": 0.8333333333333334, "date": 0.7, "amount": 0.38461538461538464, "status": 1.0, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1181}, "3": {"facts_total": 65, "facts_retained": 38, "retention_rate": 0.5846153846153846, "retention_by_kind": {"person": 0.8333333333333334, "date": 0.7, "amount": 0.38461538461538464, "status": 1.0, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1095}, "4": {"facts_total": 65, "facts_retained": 38, "retention_rate": 0.5846153846153846, "retention_by_kind": {"person": 0.8333333333333334, "date": 0.7, "amount": 0.38461538461538464, "status": 1.0, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1111}}}, "reference_key_id": "keys/synthetic/syn_03021_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "1fb6b0e63657986ace0df6d3c1195f462c1359afc69ed5a7dad3ca3220aa6437", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "a562deb1d153141db63bc3ae", "attempt_id": "c81fbe85b986", "run_group_id": "E4_core:A4|closed|qwen3b-llama3b:h4", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:21:23.325117Z", "task_family": "synthetic", "document_id": "syn_03021_2000", "document_hash": "3ac205b464552356699a4d2b0a7bf62497ead36e5bca73982e11ae5ee3810731", "question_id": "q05_multi_hop", "split": "test", "cluster_id": "syn_03021_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "tokenizer_hashes": {"receiver": "b9d8872cf3a99da99e1fc4b0f71232a451632dc49ca4c6771549b4efa5c1c763"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 4, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1725, "query_tokens": 496, "message_tokens": 385, "message_utf8_bytes": 1111, "cache_tensor_bytes": null, "serialized_payload_bytes": 1111, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 37869.45949902292, "hops": {"1": 22361.685457988642, "2": 5128.560750046745, "3": 5686.298708082177, "4": 4692.91458290536}, "receiver_prefill_ms": 207.76049990672618, "ttft_ms": 11.609584093093872, "decode_ms": 91.86570800375193}, "total_wall_ms": 38181.12691491842, "ttft_ms": 11.609584093093872, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 8}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "BQ-59Z-6753", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 38, "retention_rate": 0.5846153846153846, "retention_by_kind": {"person": 0.8333333333333334, "date": 0.7, "amount": 0.38461538461538464, "status": 1.0, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1111}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 38, "retention_rate": 0.5846153846153846, "retention_by_kind": {"person": 0.8333333333333334, "date": 0.7, "amount": 0.38461538461538464, "status": 1.0, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1179}, "2": {"facts_total": 65, "facts_retained": 38, "retention_rate": 0.5846153846153846, "retention_by_kind": {"person": 0.8333333333333334, "date": 0.7, "amount": 0.38461538461538464, "status": 1.0, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1181}, "3": {"facts_total": 65, "facts_retained": 38, "retention_rate": 0.5846153846153846, "retention_by_kind": {"person": 0.8333333333333334, "date": 0.7, "amount": 0.38461538461538464, "status": 1.0, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1095}, "4": {"facts_total": 65, "facts_retained": 38, "retention_rate": 0.5846153846153846, "retention_by_kind": {"person": 0.8333333333333334, "date": 0.7, "amount": 0.38461538461538464, "status": 1.0, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1111}}}, "reference_key_id": "keys/synthetic/syn_03021_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "1fb6b0e63657986ace0df6d3c1195f462c1359afc69ed5a7dad3ca3220aa6437", "output_hash": "db250f072c856fe7ddf4bbbd16e20c71b8bad1eef5e4480acd0dac30f8ffe9cd"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "3fbff6745fe8870007f76375", "attempt_id": "a444b81acc61", "run_group_id": "E4_core:A4|closed|qwen3b-llama3b:h4", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:21:23.652075Z", "task_family": "synthetic", "document_id": "syn_03021_2000", "document_hash": "3ac205b464552356699a4d2b0a7bf62497ead36e5bca73982e11ae5ee3810731", "question_id": "q06_negation", "split": "test", "cluster_id": "syn_03021_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "tokenizer_hashes": {"receiver": "b9d8872cf3a99da99e1fc4b0f71232a451632dc49ca4c6771549b4efa5c1c763"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 4, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1725, "query_tokens": 488, "message_tokens": 385, "message_utf8_bytes": 1111, "cache_tensor_bytes": null, "serialized_payload_bytes": 1111, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 37869.45949902292, "hops": {"1": 22361.685457988642, "2": 5128.560750046745, "3": 5686.298708082177, "4": 4692.91458290536}, "receiver_prefill_ms": 205.86283307056874, "ttft_ms": 11.583667015656829, "decode_ms": 23.060834035277367}, "total_wall_ms": 38110.413582995534, "ttft_ms": 11.583667015656829, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 2}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "No.", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 38, "retention_rate": 0.5846153846153846, "retention_by_kind": {"person": 0.8333333333333334, "date": 0.7, "amount": 0.38461538461538464, "status": 1.0, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1111}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 38, "retention_rate": 0.5846153846153846, "retention_by_kind": {"person": 0.8333333333333334, "date": 0.7, "amount": 0.38461538461538464, "status": 1.0, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1179}, "2": {"facts_total": 65, "facts_retained": 38, "retention_rate": 0.5846153846153846, "retention_by_kind": {"person": 0.8333333333333334, "date": 0.7, "amount": 0.38461538461538464, "status": 1.0, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1181}, "3": {"facts_total": 65, "facts_retained": 38, "retention_rate": 0.5846153846153846, "retention_by_kind": {"person": 0.8333333333333334, "date": 0.7, "amount": 0.38461538461538464, "status": 1.0, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1095}, "4": {"facts_total": 65, "facts_retained": 38, "retention_rate": 0.5846153846153846, "retention_by_kind": {"person": 0.8333333333333334, "date": 0.7, "amount": 0.38461538461538464, "status": 1.0, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1111}}}, "reference_key_id": "keys/synthetic/syn_03021_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "1fb6b0e63657986ace0df6d3c1195f462c1359afc69ed5a7dad3ca3220aa6437", "output_hash": "38178a20b470cfd18299fb1593dd4ba706f1a83321dac051d01086efd8b7a96f"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "4e3cec7c9b0ec4868d6642da", "attempt_id": "2f59ef09a633", "run_group_id": "E4_core:A4|closed|qwen3b-llama3b:h4", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:21:23.908264Z", "task_family": "synthetic", "document_id": "syn_03021_2000", "document_hash": "3ac205b464552356699a4d2b0a7bf62497ead36e5bca73982e11ae5ee3810731", "question_id": "q07_negation", "split": "test", "cluster_id": "syn_03021_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "tokenizer_hashes": {"receiver": "b9d8872cf3a99da99e1fc4b0f71232a451632dc49ca4c6771549b4efa5c1c763"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 4, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1725, "query_tokens": 489, "message_tokens": 385, "message_utf8_bytes": 1111, "cache_tensor_bytes": null, "serialized_payload_bytes": 1111, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 37869.45949902292, "hops": {"1": 22361.685457988642, "2": 5128.560750046745, "3": 5686.298708082177, "4": 4692.91458290536}, "receiver_prefill_ms": 204.29266605060548, "ttft_ms": 11.598250013776124, "decode_ms": 23.074583965353668}, "total_wall_ms": 38108.83570695296, "ttft_ms": 11.598250013776124, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 2}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "Yes.", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 38, "retention_rate": 0.5846153846153846, "retention_by_kind": {"person": 0.8333333333333334, "date": 0.7, "amount": 0.38461538461538464, "status": 1.0, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1111}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 38, "retention_rate": 0.5846153846153846, "retention_by_kind": {"person": 0.8333333333333334, "date": 0.7, "amount": 0.38461538461538464, "status": 1.0, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1179}, "2": {"facts_total": 65, "facts_retained": 38, "retention_rate": 0.5846153846153846, "retention_by_kind": {"person": 0.8333333333333334, "date": 0.7, "amount": 0.38461538461538464, "status": 1.0, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1181}, "3": {"facts_total": 65, "facts_retained": 38, "retention_rate": 0.5846153846153846, "retention_by_kind": {"person": 0.8333333333333334, "date": 0.7, "amount": 0.38461538461538464, "status": 1.0, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1095}, "4": {"facts_total": 65, "facts_retained": 38, "retention_rate": 0.5846153846153846, "retention_by_kind": {"person": 0.8333333333333334, "date": 0.7, "amount": 0.38461538461538464, "status": 1.0, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1111}}}, "reference_key_id": "keys/synthetic/syn_03021_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "1fb6b0e63657986ace0df6d3c1195f462c1359afc69ed5a7dad3ca3220aa6437", "output_hash": "5f9a2b795615ba6a3d5455fd5624d773fbca5bcd16249c421fd37411dc9837da"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "3d866a4a0d91dc939570e711", "attempt_id": "186e9400ec55", "run_group_id": "E4_core:A4|closed|qwen3b-llama3b:h4", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:21:24.162872Z", "task_family": "synthetic", "document_id": "syn_03021_2000", "document_hash": "3ac205b464552356699a4d2b0a7bf62497ead36e5bca73982e11ae5ee3810731", "question_id": "q08_revision", "split": "test", "cluster_id": "syn_03021_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "tokenizer_hashes": {"receiver": "b9d8872cf3a99da99e1fc4b0f71232a451632dc49ca4c6771549b4efa5c1c763"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 4, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1725, "query_tokens": 505, "message_tokens": 385, "message_utf8_bytes": 1111, "cache_tensor_bytes": null, "serialized_payload_bytes": 1111, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 37869.45949902292, "hops": {"1": 22361.685457988642, "2": 5128.560750046745, "3": 5686.298708082177, "4": 4692.91458290536}, "receiver_prefill_ms": 207.75370893534273, "ttft_ms": 11.625792016275227, "decode_ms": 11.582374921999872}, "total_wall_ms": 38100.87904101238, "ttft_ms": 11.625792016275227, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 38, "retention_rate": 0.5846153846153846, "retention_by_kind": {"person": 0.8333333333333334, "date": 0.7, "amount": 0.38461538461538464, "status": 1.0, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1111}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 38, "retention_rate": 0.5846153846153846, "retention_by_kind": {"person": 0.8333333333333334, "date": 0.7, "amount": 0.38461538461538464, "status": 1.0, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1179}, "2": {"facts_total": 65, "facts_retained": 38, "retention_rate": 0.5846153846153846, "retention_by_kind": {"person": 0.8333333333333334, "date": 0.7, "amount": 0.38461538461538464, "status": 1.0, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1181}, "3": {"facts_total": 65, "facts_retained": 38, "retention_rate": 0.5846153846153846, "retention_by_kind": {"person": 0.8333333333333334, "date": 0.7, "amount": 0.38461538461538464, "status": 1.0, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1095}, "4": {"facts_total": 65, "facts_retained": 38, "retention_rate": 0.5846153846153846, "retention_by_kind": {"person": 0.8333333333333334, "date": 0.7, "amount": 0.38461538461538464, "status": 1.0, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1111}}}, "reference_key_id": "keys/synthetic/syn_03021_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "1fb6b0e63657986ace0df6d3c1195f462c1359afc69ed5a7dad3ca3220aa6437", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "cc7b085d691e3ff4e3a0e760", "attempt_id": "757bc48ff199", "run_group_id": "E4_core:A4|closed|qwen3b-llama3b:h4", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:21:24.409225Z", "task_family": "synthetic", "document_id": "syn_03021_2000", "document_hash": "3ac205b464552356699a4d2b0a7bf62497ead36e5bca73982e11ae5ee3810731", "question_id": "q09_revision", "split": "test", "cluster_id": "syn_03021_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "tokenizer_hashes": {"receiver": "b9d8872cf3a99da99e1fc4b0f71232a451632dc49ca4c6771549b4efa5c1c763"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 4, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1725, "query_tokens": 496, "message_tokens": 385, "message_utf8_bytes": 1111, "cache_tensor_bytes": null, "serialized_payload_bytes": 1111, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 37869.45949902292, "hops": {"1": 22361.685457988642, "2": 5128.560750046745, "3": 5686.298708082177, "4": 4692.91458290536}, "receiver_prefill_ms": 203.9944170974195, "ttft_ms": 11.524083907715976, "decode_ms": 11.50795805733651}, "total_wall_ms": 38096.91137401387, "ttft_ms": 11.524083907715976, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 38, "retention_rate": 0.5846153846153846, "retention_by_kind": {"person": 0.8333333333333334, "date": 0.7, "amount": 0.38461538461538464, "status": 1.0, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1111}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 38, "retention_rate": 0.5846153846153846, "retention_by_kind": {"person": 0.8333333333333334, "date": 0.7, "amount": 0.38461538461538464, "status": 1.0, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1179}, "2": {"facts_total": 65, "facts_retained": 38, "retention_rate": 0.5846153846153846, "retention_by_kind": {"person": 0.8333333333333334, "date": 0.7, "amount": 0.38461538461538464, "status": 1.0, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1181}, "3": {"facts_total": 65, "facts_retained": 38, "retention_rate": 0.5846153846153846, "retention_by_kind": {"person": 0.8333333333333334, "date": 0.7, "amount": 0.38461538461538464, "status": 1.0, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1095}, "4": {"facts_total": 65, "facts_retained": 38, "retention_rate": 0.5846153846153846, "retention_by_kind": {"person": 0.8333333333333334, "date": 0.7, "amount": 0.38461538461538464, "status": 1.0, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1111}}}, "reference_key_id": "keys/synthetic/syn_03021_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "1fb6b0e63657986ace0df6d3c1195f462c1359afc69ed5a7dad3ca3220aa6437", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "1b2c19d6fc08f28efbe8c682", "attempt_id": "f4f7ab28d7a2", "run_group_id": "E4_core:A4|closed|qwen3b-llama3b:h4", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:21:24.652131Z", "task_family": "synthetic", "document_id": "syn_03021_2000", "document_hash": "3ac205b464552356699a4d2b0a7bf62497ead36e5bca73982e11ae5ee3810731", "question_id": "q10_temporal", "split": "test", "cluster_id": "syn_03021_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "tokenizer_hashes": {"receiver": "b9d8872cf3a99da99e1fc4b0f71232a451632dc49ca4c6771549b4efa5c1c763"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 4, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1725, "query_tokens": 504, "message_tokens": 385, "message_utf8_bytes": 1111, "cache_tensor_bytes": null, "serialized_payload_bytes": 1111, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 37869.45949902292, "hops": {"1": 22361.685457988642, "2": 5128.560750046745, "3": 5686.298708082177, "4": 4692.91458290536}, "receiver_prefill_ms": 206.81166602298617, "ttft_ms": 11.575083015486598, "decode_ms": 11.601999984122813}, "total_wall_ms": 38099.86699896399, "ttft_ms": 11.575083015486598, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 38, "retention_rate": 0.5846153846153846, "retention_by_kind": {"person": 0.8333333333333334, "date": 0.7, "amount": 0.38461538461538464, "status": 1.0, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1111}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 38, "retention_rate": 0.5846153846153846, "retention_by_kind": {"person": 0.8333333333333334, "date": 0.7, "amount": 0.38461538461538464, "status": 1.0, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1179}, "2": {"facts_total": 65, "facts_retained": 38, "retention_rate": 0.5846153846153846, "retention_by_kind": {"person": 0.8333333333333334, "date": 0.7, "amount": 0.38461538461538464, "status": 1.0, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1181}, "3": {"facts_total": 65, "facts_retained": 38, "retention_rate": 0.5846153846153846, "retention_by_kind": {"person": 0.8333333333333334, "date": 0.7, "amount": 0.38461538461538464, "status": 1.0, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1095}, "4": {"facts_total": 65, "facts_retained": 38, "retention_rate": 0.5846153846153846, "retention_by_kind": {"person": 0.8333333333333334, "date": 0.7, "amount": 0.38461538461538464, "status": 1.0, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1111}}}, "reference_key_id": "keys/synthetic/syn_03021_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "1fb6b0e63657986ace0df6d3c1195f462c1359afc69ed5a7dad3ca3220aa6437", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "3a1f23850d4f95f7b6b6870c", "attempt_id": "99402b636d75", "run_group_id": "E4_core:A4|closed|qwen3b-llama3b:h4", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:21:24.897630Z", "task_family": "synthetic", "document_id": "syn_03021_2000", "document_hash": "3ac205b464552356699a4d2b0a7bf62497ead36e5bca73982e11ae5ee3810731", "question_id": "q11_aggregation", "split": "test", "cluster_id": "syn_03021_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "tokenizer_hashes": {"receiver": "b9d8872cf3a99da99e1fc4b0f71232a451632dc49ca4c6771549b4efa5c1c763"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 4, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1725, "query_tokens": 500, "message_tokens": 385, "message_utf8_bytes": 1111, "cache_tensor_bytes": null, "serialized_payload_bytes": 1111, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 37869.45949902292, "hops": {"1": 22361.685457988642, "2": 5128.560750046745, "3": 5686.298708082177, "4": 4692.91458290536}, "receiver_prefill_ms": 206.85012498870492, "ttft_ms": 11.604624916799366, "decode_ms": 219.0847919555381}, "total_wall_ms": 38307.52749904059, "ttft_ms": 11.604624916799366, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 19}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "4,862,425 + 5,646,720 = 10,509,145", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 38, "retention_rate": 0.5846153846153846, "retention_by_kind": {"person": 0.8333333333333334, "date": 0.7, "amount": 0.38461538461538464, "status": 1.0, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1111}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 38, "retention_rate": 0.5846153846153846, "retention_by_kind": {"person": 0.8333333333333334, "date": 0.7, "amount": 0.38461538461538464, "status": 1.0, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1179}, "2": {"facts_total": 65, "facts_retained": 38, "retention_rate": 0.5846153846153846, "retention_by_kind": {"person": 0.8333333333333334, "date": 0.7, "amount": 0.38461538461538464, "status": 1.0, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1181}, "3": {"facts_total": 65, "facts_retained": 38, "retention_rate": 0.5846153846153846, "retention_by_kind": {"person": 0.8333333333333334, "date": 0.7, "amount": 0.38461538461538464, "status": 1.0, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1095}, "4": {"facts_total": 65, "facts_retained": 38, "retention_rate": 0.5846153846153846, "retention_by_kind": {"person": 0.8333333333333334, "date": 0.7, "amount": 0.38461538461538464, "status": 1.0, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1111}}}, "reference_key_id": "keys/synthetic/syn_03021_2000.json", "task_scores": {"correct": 0.0, "abstained": 0.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "1fb6b0e63657986ace0df6d3c1195f462c1359afc69ed5a7dad3ca3220aa6437", "output_hash": "038d9ed23c4639e6ae998b25ae021dada1e0ab0526113d0a0e4d8884ea4b4e6f"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "87302980abc81826c43804e4", "attempt_id": "e37a2124be1c", "run_group_id": "E4_core:A4|closed|qwen3b-llama3b:h4", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:21:25.350792Z", "task_family": "synthetic", "document_id": "syn_03021_2000", "document_hash": "3ac205b464552356699a4d2b0a7bf62497ead36e5bca73982e11ae5ee3810731", "question_id": "q12_unit_conversion", "split": "test", "cluster_id": "syn_03021_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "tokenizer_hashes": {"receiver": "b9d8872cf3a99da99e1fc4b0f71232a451632dc49ca4c6771549b4efa5c1c763"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 4, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1725, "query_tokens": 495, "message_tokens": 385, "message_utf8_bytes": 1111, "cache_tensor_bytes": null, "serialized_payload_bytes": 1111, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 37869.45949902292, "hops": {"1": 22361.685457988642, "2": 5128.560750046745, "3": 5686.298708082177, "4": 4692.91458290536}, "receiver_prefill_ms": 207.85816700663418, "ttft_ms": 11.542499996721745, "decode_ms": 11.541875079274178}, "total_wall_ms": 38100.84054002073, "ttft_ms": 11.542499996721745, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 38, "retention_rate": 0.5846153846153846, "retention_by_kind": {"person": 0.8333333333333334, "date": 0.7, "amount": 0.38461538461538464, "status": 1.0, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1111}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 38, "retention_rate": 0.5846153846153846, "retention_by_kind": {"person": 0.8333333333333334, "date": 0.7, "amount": 0.38461538461538464, "status": 1.0, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1179}, "2": {"facts_total": 65, "facts_retained": 38, "retention_rate": 0.5846153846153846, "retention_by_kind": {"person": 0.8333333333333334, "date": 0.7, "amount": 0.38461538461538464, "status": 1.0, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1181}, "3": {"facts_total": 65, "facts_retained": 38, "retention_rate": 0.5846153846153846, "retention_by_kind": {"person": 0.8333333333333334, "date": 0.7, "amount": 0.38461538461538464, "status": 1.0, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1095}, "4": {"facts_total": 65, "facts_retained": 38, "retention_rate": 0.5846153846153846, "retention_by_kind": {"person": 0.8333333333333334, "date": 0.7, "amount": 0.38461538461538464, "status": 1.0, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1111}}}, "reference_key_id": "keys/synthetic/syn_03021_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "1fb6b0e63657986ace0df6d3c1195f462c1359afc69ed5a7dad3ca3220aa6437", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "9d9fda703c1bf5aac43ed159", "attempt_id": "dda852f30875", "run_group_id": "E4_core:A4|closed|qwen3b-llama3b:h4", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:21:25.597921Z", "task_family": "synthetic", "document_id": "syn_03021_2000", "document_hash": "3ac205b464552356699a4d2b0a7bf62497ead36e5bca73982e11ae5ee3810731", "question_id": "q13_unanswerable", "split": "test", "cluster_id": "syn_03021_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "tokenizer_hashes": {"receiver": "b9d8872cf3a99da99e1fc4b0f71232a451632dc49ca4c6771549b4efa5c1c763"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 4, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1725, "query_tokens": 499, "message_tokens": 385, "message_utf8_bytes": 1111, "cache_tensor_bytes": null, "serialized_payload_bytes": 1111, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 37869.45949902292, "hops": {"1": 22361.685457988642, "2": 5128.560750046745, "3": 5686.298708082177, "4": 4692.91458290536}, "receiver_prefill_ms": 206.54475002083927, "ttft_ms": 11.62362506147474, "decode_ms": 11.549124959856272}, "total_wall_ms": 38099.6261240216, "ttft_ms": 11.62362506147474, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 38, "retention_rate": 0.5846153846153846, "retention_by_kind": {"person": 0.8333333333333334, "date": 0.7, "amount": 0.38461538461538464, "status": 1.0, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1111}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 38, "retention_rate": 0.5846153846153846, "retention_by_kind": {"person": 0.8333333333333334, "date": 0.7, "amount": 0.38461538461538464, "status": 1.0, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1179}, "2": {"facts_total": 65, "facts_retained": 38, "retention_rate": 0.5846153846153846, "retention_by_kind": {"person": 0.8333333333333334, "date": 0.7, "amount": 0.38461538461538464, "status": 1.0, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1181}, "3": {"facts_total": 65, "facts_retained": 38, "retention_rate": 0.5846153846153846, "retention_by_kind": {"person": 0.8333333333333334, "date": 0.7, "amount": 0.38461538461538464, "status": 1.0, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1095}, "4": {"facts_total": 65, "facts_retained": 38, "retention_rate": 0.5846153846153846, "retention_by_kind": {"person": 0.8333333333333334, "date": 0.7, "amount": 0.38461538461538464, "status": 1.0, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1111}}}, "reference_key_id": "keys/synthetic/syn_03021_2000.json", "task_scores": {"correct": 1.0, "abstained": 1.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "1fb6b0e63657986ace0df6d3c1195f462c1359afc69ed5a7dad3ca3220aa6437", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "2098f6b968e839a34bf81661", "attempt_id": "5f675f70215c", "run_group_id": "E4_core:A4|closed|qwen3b-llama3b:h4", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:21:25.843454Z", "task_family": "synthetic", "document_id": "syn_03021_2000", "document_hash": "3ac205b464552356699a4d2b0a7bf62497ead36e5bca73982e11ae5ee3810731", "question_id": "q14_unanswerable", "split": "test", "cluster_id": "syn_03021_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "tokenizer_hashes": {"receiver": "b9d8872cf3a99da99e1fc4b0f71232a451632dc49ca4c6771549b4efa5c1c763"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 4, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1725, "query_tokens": 502, "message_tokens": 385, "message_utf8_bytes": 1111, "cache_tensor_bytes": null, "serialized_payload_bytes": 1111, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 37869.45949902292, "hops": {"1": 22361.685457988642, "2": 5128.560750046745, "3": 5686.298708082177, "4": 4692.91458290536}, "receiver_prefill_ms": 207.72512489929795, "ttft_ms": 11.560374987311661, "decode_ms": 11.518125073052943}, "total_wall_ms": 38100.68291600328, "ttft_ms": 11.560374987311661, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 38, "retention_rate": 0.5846153846153846, "retention_by_kind": {"person": 0.8333333333333334, "date": 0.7, "amount": 0.38461538461538464, "status": 1.0, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1111}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 38, "retention_rate": 0.5846153846153846, "retention_by_kind": {"person": 0.8333333333333334, "date": 0.7, "amount": 0.38461538461538464, "status": 1.0, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1179}, "2": {"facts_total": 65, "facts_retained": 38, "retention_rate": 0.5846153846153846, "retention_by_kind": {"person": 0.8333333333333334, "date": 0.7, "amount": 0.38461538461538464, "status": 1.0, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1181}, "3": {"facts_total": 65, "facts_retained": 38, "retention_rate": 0.5846153846153846, "retention_by_kind": {"person": 0.8333333333333334, "date": 0.7, "amount": 0.38461538461538464, "status": 1.0, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1095}, "4": {"facts_total": 65, "facts_retained": 38, "retention_rate": 0.5846153846153846, "retention_by_kind": {"person": 0.8333333333333334, "date": 0.7, "amount": 0.38461538461538464, "status": 1.0, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1111}}}, "reference_key_id": "keys/synthetic/syn_03021_2000.json", "task_scores": {"correct": 1.0, "abstained": 1.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "1fb6b0e63657986ace0df6d3c1195f462c1359afc69ed5a7dad3ca3220aa6437", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "47b0c435f5590667052b70cc", "attempt_id": "9f605633dd38", "run_group_id": "E4_core:A4|closed|qwen3b-llama3b:h8", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:21:47.377868Z", "task_family": "synthetic", "document_id": "syn_03021_2000", "document_hash": "3ac205b464552356699a4d2b0a7bf62497ead36e5bca73982e11ae5ee3810731", "question_id": "q00_single_fact", "split": "test", "cluster_id": "syn_03021_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "tokenizer_hashes": {"receiver": "b9d8872cf3a99da99e1fc4b0f71232a451632dc49ca4c6771549b4efa5c1c763"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 8, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1725, "query_tokens": 496, "message_tokens": 394, "message_utf8_bytes": 1153, "cache_tensor_bytes": null, "serialized_payload_bytes": 1153, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 59152.118457946926, "hops": {"1": 22361.685457988642, "2": 5128.560750046745, "3": 5686.298708082177, "4": 4692.91458290536, "5": 5842.431625002064, "6": 4801.743416930549, "7": 5841.1449999548495, "8": 4797.338917036541}, "receiver_prefill_ms": 205.95662505365908, "ttft_ms": 11.59316604025662, "decode_ms": 11.459583067335188}, "total_wall_ms": 59381.6127498867, "ttft_ms": 11.59316604025662, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 38, "retention_rate": 0.5846153846153846, "retention_by_kind": {"person": 0.8333333333333334, "date": 0.7, "amount": 0.38461538461538464, "status": 1.0, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1153}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 38, "retention_rate": 0.5846153846153846, "retention_by_kind": {"person": 0.8333333333333334, "date": 0.7, "amount": 0.38461538461538464, "status": 1.0, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1179}, "2": {"facts_total": 65, "facts_retained": 38, "retention_rate": 0.5846153846153846, "retention_by_kind": {"person": 0.8333333333333334, "date": 0.7, "amount": 0.38461538461538464, "status": 1.0, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1181}, "3": {"facts_total": 65, "facts_retained": 38, "retention_rate": 0.5846153846153846, "retention_by_kind": {"person": 0.8333333333333334, "date": 0.7, "amount": 0.38461538461538464, "status": 1.0, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1095}, "4": {"facts_total": 65, "facts_retained": 38, "retention_rate": 0.5846153846153846, "retention_by_kind": {"person": 0.8333333333333334, "date": 0.7, "amount": 0.38461538461538464, "status": 1.0, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1111}, "5": {"facts_total": 65, "facts_retained": 38, "retention_rate": 0.5846153846153846, "retention_by_kind": {"person": 0.8333333333333334, "date": 0.7, "amount": 0.38461538461538464, "status": 1.0, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1111}, "6": {"facts_total": 65, "facts_retained": 38, "retention_rate": 0.5846153846153846, "retention_by_kind": {"person": 0.8333333333333334, "date": 0.7, "amount": 0.38461538461538464, "status": 1.0, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1153}, "7": {"facts_total": 65, "facts_retained": 38, "retention_rate": 0.5846153846153846, "retention_by_kind": {"person": 0.8333333333333334, "date": 0.7, "amount": 0.38461538461538464, "status": 1.0, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1111}, "8": {"facts_total": 65, "facts_retained": 38, "retention_rate": 0.5846153846153846, "retention_by_kind": {"person": 0.8333333333333334, "date": 0.7, "amount": 0.38461538461538464, "status": 1.0, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1153}}}, "reference_key_id": "keys/synthetic/syn_03021_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "1d0e469668b88ffe7676989aa63ff9a7d27eb7c55484b06c14ba4e2a98a13623", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "99a3d543aa37f01f328ea23a", "attempt_id": "9ab8015c9dc0", "run_group_id": "E4_core:A4|closed|qwen3b-llama3b:h8", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:21:47.623528Z", "task_family": "synthetic", "document_id": "syn_03021_2000", "document_hash": "3ac205b464552356699a4d2b0a7bf62497ead36e5bca73982e11ae5ee3810731", "question_id": "q01_single_fact", "split": "test", "cluster_id": "syn_03021_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "tokenizer_hashes": {"receiver": "b9d8872cf3a99da99e1fc4b0f71232a451632dc49ca4c6771549b4efa5c1c763"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 8, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1725, "query_tokens": 502, "message_tokens": 394, "message_utf8_bytes": 1153, "cache_tensor_bytes": null, "serialized_payload_bytes": 1153, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 59152.118457946926, "hops": {"1": 22361.685457988642, "2": 5128.560750046745, "3": 5686.298708082177, "4": 4692.91458290536, "5": 5842.431625002064, "6": 4801.743416930549, "7": 5841.1449999548495, "8": 4797.338917036541}, "receiver_prefill_ms": 206.6960830707103, "ttft_ms": 11.563332984223962, "decode_ms": 114.8346250411123}, "total_wall_ms": 59485.684790997766, "ttft_ms": 11.563332984223962, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 10}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "Lot 536-R, Senneterre.", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 38, "retention_rate": 0.5846153846153846, "retention_by_kind": {"person": 0.8333333333333334, "date": 0.7, "amount": 0.38461538461538464, "status": 1.0, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1153}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 38, "retention_rate": 0.5846153846153846, "retention_by_kind": {"person": 0.8333333333333334, "date": 0.7, "amount": 0.38461538461538464, "status": 1.0, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1179}, "2": {"facts_total": 65, "facts_retained": 38, "retention_rate": 0.5846153846153846, "retention_by_kind": {"person": 0.8333333333333334, "date": 0.7, "amount": 0.38461538461538464, "status": 1.0, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1181}, "3": {"facts_total": 65, "facts_retained": 38, "retention_rate": 0.5846153846153846, "retention_by_kind": {"person": 0.8333333333333334, "date": 0.7, "amount": 0.38461538461538464, "status": 1.0, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1095}, "4": {"facts_total": 65, "facts_retained": 38, "retention_rate": 0.5846153846153846, "retention_by_kind": {"person": 0.8333333333333334, "date": 0.7, "amount": 0.38461538461538464, "status": 1.0, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1111}, "5": {"facts_total": 65, "facts_retained": 38, "retention_rate": 0.5846153846153846, "retention_by_kind": {"person": 0.8333333333333334, "date": 0.7, "amount": 0.38461538461538464, "status": 1.0, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1111}, "6": {"facts_total": 65, "facts_retained": 38, "retention_rate": 0.5846153846153846, "retention_by_kind": {"person": 0.8333333333333334, "date": 0.7, "amount": 0.38461538461538464, "status": 1.0, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1153}, "7": {"facts_total": 65, "facts_retained": 38, "retention_rate": 0.5846153846153846, "retention_by_kind": {"person": 0.8333333333333334, "date": 0.7, "amount": 0.38461538461538464, "status": 1.0, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1111}, "8": {"facts_total": 65, "facts_retained": 38, "retention_rate": 0.5846153846153846, "retention_by_kind": {"person": 0.8333333333333334, "date": 0.7, "amount": 0.38461538461538464, "status": 1.0, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1153}}}, "reference_key_id": "keys/synthetic/syn_03021_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 0.8571428571428571}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "1d0e469668b88ffe7676989aa63ff9a7d27eb7c55484b06c14ba4e2a98a13623", "output_hash": "d34fb8d4553d115f1c915ac9be75f68ef887916efddf5ed339f7cf4236ce8fa8"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "89a65bba6e1b336011f8aec9", "attempt_id": "20a4b51c9156", "run_group_id": "E4_core:A4|closed|qwen3b-llama3b:h8", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:21:47.972924Z", "task_family": "synthetic", "document_id": "syn_03021_2000", "document_hash": "3ac205b464552356699a4d2b0a7bf62497ead36e5bca73982e11ae5ee3810731", "question_id": "q02_exact_copy", "split": "test", "cluster_id": "syn_03021_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "tokenizer_hashes": {"receiver": "b9d8872cf3a99da99e1fc4b0f71232a451632dc49ca4c6771549b4efa5c1c763"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 8, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1725, "query_tokens": 502, "message_tokens": 394, "message_utf8_bytes": 1153, "cache_tensor_bytes": null, "serialized_payload_bytes": 1153, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 59152.118457946926, "hops": {"1": 22361.685457988642, "2": 5128.560750046745, "3": 5686.298708082177, "4": 4692.91458290536, "5": 5842.431625002064, "6": 4801.743416930549, "7": 5841.1449999548495, "8": 4797.338917036541}, "receiver_prefill_ms": 206.66645804885775, "ttft_ms": 11.547207948751748, "decode_ms": 46.02350003551692}, "total_wall_ms": 59416.78358300123, "ttft_ms": 11.547207948751748, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 4}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "88.7751", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 38, "retention_rate": 0.5846153846153846, "retention_by_kind": {"person": 0.8333333333333334, "date": 0.7, "amount": 0.38461538461538464, "status": 1.0, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1153}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 38, "retention_rate": 0.5846153846153846, "retention_by_kind": {"person": 0.8333333333333334, "date": 0.7, "amount": 0.38461538461538464, "status": 1.0, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1179}, "2": {"facts_total": 65, "facts_retained": 38, "retention_rate": 0.5846153846153846, "retention_by_kind": {"person": 0.8333333333333334, "date": 0.7, "amount": 0.38461538461538464, "status": 1.0, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1181}, "3": {"facts_total": 65, "facts_retained": 38, "retention_rate": 0.5846153846153846, "retention_by_kind": {"person": 0.8333333333333334, "date": 0.7, "amount": 0.38461538461538464, "status": 1.0, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1095}, "4": {"facts_total": 65, "facts_retained": 38, "retention_rate": 0.5846153846153846, "retention_by_kind": {"person": 0.8333333333333334, "date": 0.7, "amount": 0.38461538461538464, "status": 1.0, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1111}, "5": {"facts_total": 65, "facts_retained": 38, "retention_rate": 0.5846153846153846, "retention_by_kind": {"person": 0.8333333333333334, "date": 0.7, "amount": 0.38461538461538464, "status": 1.0, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1111}, "6": {"facts_total": 65, "facts_retained": 38, "retention_rate": 0.5846153846153846, "retention_by_kind": {"person": 0.8333333333333334, "date": 0.7, "amount": 0.38461538461538464, "status": 1.0, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1153}, "7": {"facts_total": 65, "facts_retained": 38, "retention_rate": 0.5846153846153846, "retention_by_kind": {"person": 0.8333333333333334, "date": 0.7, "amount": 0.38461538461538464, "status": 1.0, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1111}, "8": {"facts_total": 65, "facts_retained": 38, "retention_rate": 0.5846153846153846, "retention_by_kind": {"person": 0.8333333333333334, "date": 0.7, "amount": 0.38461538461538464, "status": 1.0, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1153}}}, "reference_key_id": "keys/synthetic/syn_03021_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "1d0e469668b88ffe7676989aa63ff9a7d27eb7c55484b06c14ba4e2a98a13623", "output_hash": "1e01e62c69f9479b681e50e59e8565c6583f5cdb803446a05ed5cb13118acdf9"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "b00612f22399e784b78945b2", "attempt_id": "01cbb4d53e4c", "run_group_id": "E4_core:A4|closed|qwen3b-llama3b:h8", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:21:48.253248Z", "task_family": "synthetic", "document_id": "syn_03021_2000", "document_hash": "3ac205b464552356699a4d2b0a7bf62497ead36e5bca73982e11ae5ee3810731", "question_id": "q03_exact_copy", "split": "test", "cluster_id": "syn_03021_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "tokenizer_hashes": {"receiver": "b9d8872cf3a99da99e1fc4b0f71232a451632dc49ca4c6771549b4efa5c1c763"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 8, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1725, "query_tokens": 499, "message_tokens": 394, "message_utf8_bytes": 1153, "cache_tensor_bytes": null, "serialized_payload_bytes": 1153, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 59152.118457946926, "hops": {"1": 22361.685457988642, "2": 5128.560750046745, "3": 5686.298708082177, "4": 4692.91458290536, "5": 5842.431625002064, "6": 4801.743416930549, "7": 5841.1449999548495, "8": 4797.338917036541}, "receiver_prefill_ms": 210.11308289598674, "ttft_ms": 11.517458013258874, "decode_ms": 45.92083301395178}, "total_wall_ms": 59420.10579095222, "ttft_ms": 11.517458013258874, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 4}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "Ingrid Mbeki", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 38, "retention_rate": 0.5846153846153846, "retention_by_kind": {"person": 0.8333333333333334, "date": 0.7, "amount": 0.38461538461538464, "status": 1.0, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1153}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 38, "retention_rate": 0.5846153846153846, "retention_by_kind": {"person": 0.8333333333333334, "date": 0.7, "amount": 0.38461538461538464, "status": 1.0, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1179}, "2": {"facts_total": 65, "facts_retained": 38, "retention_rate": 0.5846153846153846, "retention_by_kind": {"person": 0.8333333333333334, "date": 0.7, "amount": 0.38461538461538464, "status": 1.0, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1181}, "3": {"facts_total": 65, "facts_retained": 38, "retention_rate": 0.5846153846153846, "retention_by_kind": {"person": 0.8333333333333334, "date": 0.7, "amount": 0.38461538461538464, "status": 1.0, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1095}, "4": {"facts_total": 65, "facts_retained": 38, "retention_rate": 0.5846153846153846, "retention_by_kind": {"person": 0.8333333333333334, "date": 0.7, "amount": 0.38461538461538464, "status": 1.0, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1111}, "5": {"facts_total": 65, "facts_retained": 38, "retention_rate": 0.5846153846153846, "retention_by_kind": {"person": 0.8333333333333334, "date": 0.7, "amount": 0.38461538461538464, "status": 1.0, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1111}, "6": {"facts_total": 65, "facts_retained": 38, "retention_rate": 0.5846153846153846, "retention_by_kind": {"person": 0.8333333333333334, "date": 0.7, "amount": 0.38461538461538464, "status": 1.0, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1153}, "7": {"facts_total": 65, "facts_retained": 38, "retention_rate": 0.5846153846153846, "retention_by_kind": {"person": 0.8333333333333334, "date": 0.7, "amount": 0.38461538461538464, "status": 1.0, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1111}, "8": {"facts_total": 65, "facts_retained": 38, "retention_rate": 0.5846153846153846, "retention_by_kind": {"person": 0.8333333333333334, "date": 0.7, "amount": 0.38461538461538464, "status": 1.0, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1153}}}, "reference_key_id": "keys/synthetic/syn_03021_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "1d0e469668b88ffe7676989aa63ff9a7d27eb7c55484b06c14ba4e2a98a13623", "output_hash": "1dbd85090bb65fb3b72de8ca4ce32f2c234729ff4a239274e79fb574b800a79d"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "0b66a30ba67b1da9ef5ac9a7", "attempt_id": "b74ae4e23191", "run_group_id": "E4_core:A4|closed|qwen3b-llama3b:h8", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:21:48.536596Z", "task_family": "synthetic", "document_id": "syn_03021_2000", "document_hash": "3ac205b464552356699a4d2b0a7bf62497ead36e5bca73982e11ae5ee3810731", "question_id": "q04_multi_hop", "split": "test", "cluster_id": "syn_03021_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "tokenizer_hashes": {"receiver": "b9d8872cf3a99da99e1fc4b0f71232a451632dc49ca4c6771549b4efa5c1c763"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 8, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1725, "query_tokens": 506, "message_tokens": 394, "message_utf8_bytes": 1153, "cache_tensor_bytes": null, "serialized_payload_bytes": 1153, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 59152.118457946926, "hops": {"1": 22361.685457988642, "2": 5128.560750046745, "3": 5686.298708082177, "4": 4692.91458290536, "5": 5842.431625002064, "6": 4801.743416930549, "7": 5841.1449999548495, "8": 4797.338917036541}, "receiver_prefill_ms": 208.13258294947445, "ttft_ms": 11.538416962139308, "decode_ms": 11.483917012810707}, "total_wall_ms": 59383.72379099019, "ttft_ms": 11.538416962139308, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 38, "retention_rate": 0.5846153846153846, "retention_by_kind": {"person": 0.8333333333333334, "date": 0.7, "amount": 0.38461538461538464, "status": 1.0, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1153}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 38, "retention_rate": 0.5846153846153846, "retention_by_kind": {"person": 0.8333333333333334, "date": 0.7, "amount": 0.38461538461538464, "status": 1.0, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1179}, "2": {"facts_total": 65, "facts_retained": 38, "retention_rate": 0.5846153846153846, "retention_by_kind": {"person": 0.8333333333333334, "date": 0.7, "amount": 0.38461538461538464, "status": 1.0, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1181}, "3": {"facts_total": 65, "facts_retained": 38, "retention_rate": 0.5846153846153846, "retention_by_kind": {"person": 0.8333333333333334, "date": 0.7, "amount": 0.38461538461538464, "status": 1.0, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1095}, "4": {"facts_total": 65, "facts_retained": 38, "retention_rate": 0.5846153846153846, "retention_by_kind": {"person": 0.8333333333333334, "date": 0.7, "amount": 0.38461538461538464, "status": 1.0, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1111}, "5": {"facts_total": 65, "facts_retained": 38, "retention_rate": 0.5846153846153846, "retention_by_kind": {"person": 0.8333333333333334, "date": 0.7, "amount": 0.38461538461538464, "status": 1.0, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1111}, "6": {"facts_total": 65, "facts_retained": 38, "retention_rate": 0.5846153846153846, "retention_by_kind": {"person": 0.8333333333333334, "date": 0.7, "amount": 0.38461538461538464, "status": 1.0, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1153}, "7": {"facts_total": 65, "facts_retained": 38, "retention_rate": 0.5846153846153846, "retention_by_kind": {"person": 0.8333333333333334, "date": 0.7, "amount": 0.38461538461538464, "status": 1.0, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1111}, "8": {"facts_total": 65, "facts_retained": 38, "retention_rate": 0.5846153846153846, "retention_by_kind": {"person": 0.8333333333333334, "date": 0.7, "amount": 0.38461538461538464, "status": 1.0, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1153}}}, "reference_key_id": "keys/synthetic/syn_03021_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "1d0e469668b88ffe7676989aa63ff9a7d27eb7c55484b06c14ba4e2a98a13623", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "d81a69863438829b555f8f8f", "attempt_id": "f13f602f2686", "run_group_id": "E4_core:A4|closed|qwen3b-llama3b:h8", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:21:48.783277Z", "task_family": "synthetic", "document_id": "syn_03021_2000", "document_hash": "3ac205b464552356699a4d2b0a7bf62497ead36e5bca73982e11ae5ee3810731", "question_id": "q05_multi_hop", "split": "test", "cluster_id": "syn_03021_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "tokenizer_hashes": {"receiver": "b9d8872cf3a99da99e1fc4b0f71232a451632dc49ca4c6771549b4efa5c1c763"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 8, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1725, "query_tokens": 505, "message_tokens": 394, "message_utf8_bytes": 1153, "cache_tensor_bytes": null, "serialized_payload_bytes": 1153, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 59152.118457946926, "hops": {"1": 22361.685457988642, "2": 5128.560750046745, "3": 5686.298708082177, "4": 4692.91458290536, "5": 5842.431625002064, "6": 4801.743416930549, "7": 5841.1449999548495, "8": 4797.338917036541}, "receiver_prefill_ms": 206.34704106487334, "ttft_ms": 11.641249991953373, "decode_ms": 92.6754999672994}, "total_wall_ms": 59463.234957889654, "ttft_ms": 11.641249991953373, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 8}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "BQ-59Z-6753", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 38, "retention_rate": 0.5846153846153846, "retention_by_kind": {"person": 0.8333333333333334, "date": 0.7, "amount": 0.38461538461538464, "status": 1.0, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1153}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 38, "retention_rate": 0.5846153846153846, "retention_by_kind": {"person": 0.8333333333333334, "date": 0.7, "amount": 0.38461538461538464, "status": 1.0, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1179}, "2": {"facts_total": 65, "facts_retained": 38, "retention_rate": 0.5846153846153846, "retention_by_kind": {"person": 0.8333333333333334, "date": 0.7, "amount": 0.38461538461538464, "status": 1.0, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1181}, "3": {"facts_total": 65, "facts_retained": 38, "retention_rate": 0.5846153846153846, "retention_by_kind": {"person": 0.8333333333333334, "date": 0.7, "amount": 0.38461538461538464, "status": 1.0, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1095}, "4": {"facts_total": 65, "facts_retained": 38, "retention_rate": 0.5846153846153846, "retention_by_kind": {"person": 0.8333333333333334, "date": 0.7, "amount": 0.38461538461538464, "status": 1.0, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1111}, "5": {"facts_total": 65, "facts_retained": 38, "retention_rate": 0.5846153846153846, "retention_by_kind": {"person": 0.8333333333333334, "date": 0.7, "amount": 0.38461538461538464, "status": 1.0, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1111}, "6": {"facts_total": 65, "facts_retained": 38, "retention_rate": 0.5846153846153846, "retention_by_kind": {"person": 0.8333333333333334, "date": 0.7, "amount": 0.38461538461538464, "status": 1.0, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1153}, "7": {"facts_total": 65, "facts_retained": 38, "retention_rate": 0.5846153846153846, "retention_by_kind": {"person": 0.8333333333333334, "date": 0.7, "amount": 0.38461538461538464, "status": 1.0, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1111}, "8": {"facts_total": 65, "facts_retained": 38, "retention_rate": 0.5846153846153846, "retention_by_kind": {"person": 0.8333333333333334, "date": 0.7, "amount": 0.38461538461538464, "status": 1.0, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1153}}}, "reference_key_id": "keys/synthetic/syn_03021_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "1d0e469668b88ffe7676989aa63ff9a7d27eb7c55484b06c14ba4e2a98a13623", "output_hash": "db250f072c856fe7ddf4bbbd16e20c71b8bad1eef5e4480acd0dac30f8ffe9cd"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "bce7b4e9487dea641eff9f6b", "attempt_id": "70b01a767e43", "run_group_id": "E4_core:A4|closed|qwen3b-llama3b:h8", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:21:49.109760Z", "task_family": "synthetic", "document_id": "syn_03021_2000", "document_hash": "3ac205b464552356699a4d2b0a7bf62497ead36e5bca73982e11ae5ee3810731", "question_id": "q06_negation", "split": "test", "cluster_id": "syn_03021_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "tokenizer_hashes": {"receiver": "b9d8872cf3a99da99e1fc4b0f71232a451632dc49ca4c6771549b4efa5c1c763"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 8, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1725, "query_tokens": 497, "message_tokens": 394, "message_utf8_bytes": 1153, "cache_tensor_bytes": null, "serialized_payload_bytes": 1153, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 59152.118457946926, "hops": {"1": 22361.685457988642, "2": 5128.560750046745, "3": 5686.298708082177, "4": 4692.91458290536, "5": 5842.431625002064, "6": 4801.743416930549, "7": 5841.1449999548495, "8": 4797.338917036541}, "receiver_prefill_ms": 207.62108301278204, "ttft_ms": 11.487792013213038, "decode_ms": 22.975333034992218}, "total_wall_ms": 59394.65216593817, "ttft_ms": 11.487792013213038, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 2}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "No.", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 38, "retention_rate": 0.5846153846153846, "retention_by_kind": {"person": 0.8333333333333334, "date": 0.7, "amount": 0.38461538461538464, "status": 1.0, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1153}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 38, "retention_rate": 0.5846153846153846, "retention_by_kind": {"person": 0.8333333333333334, "date": 0.7, "amount": 0.38461538461538464, "status": 1.0, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1179}, "2": {"facts_total": 65, "facts_retained": 38, "retention_rate": 0.5846153846153846, "retention_by_kind": {"person": 0.8333333333333334, "date": 0.7, "amount": 0.38461538461538464, "status": 1.0, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1181}, "3": {"facts_total": 65, "facts_retained": 38, "retention_rate": 0.5846153846153846, "retention_by_kind": {"person": 0.8333333333333334, "date": 0.7, "amount": 0.38461538461538464, "status": 1.0, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1095}, "4": {"facts_total": 65, "facts_retained": 38, "retention_rate": 0.5846153846153846, "retention_by_kind": {"person": 0.8333333333333334, "date": 0.7, "amount": 0.38461538461538464, "status": 1.0, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1111}, "5": {"facts_total": 65, "facts_retained": 38, "retention_rate": 0.5846153846153846, "retention_by_kind": {"person": 0.8333333333333334, "date": 0.7, "amount": 0.38461538461538464, "status": 1.0, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1111}, "6": {"facts_total": 65, "facts_retained": 38, "retention_rate": 0.5846153846153846, "retention_by_kind": {"person": 0.8333333333333334, "date": 0.7, "amount": 0.38461538461538464, "status": 1.0, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1153}, "7": {"facts_total": 65, "facts_retained": 38, "retention_rate": 0.5846153846153846, "retention_by_kind": {"person": 0.8333333333333334, "date": 0.7, "amount": 0.38461538461538464, "status": 1.0, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1111}, "8": {"facts_total": 65, "facts_retained": 38, "retention_rate": 0.5846153846153846, "retention_by_kind": {"person": 0.8333333333333334, "date": 0.7, "amount": 0.38461538461538464, "status": 1.0, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1153}}}, "reference_key_id": "keys/synthetic/syn_03021_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "1d0e469668b88ffe7676989aa63ff9a7d27eb7c55484b06c14ba4e2a98a13623", "output_hash": "38178a20b470cfd18299fb1593dd4ba706f1a83321dac051d01086efd8b7a96f"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "11f6ab043bd0d6445d4a22fc", "attempt_id": "e1d62530d976", "run_group_id": "E4_core:A4|closed|qwen3b-llama3b:h8", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:21:49.367570Z", "task_family": "synthetic", "document_id": "syn_03021_2000", "document_hash": "3ac205b464552356699a4d2b0a7bf62497ead36e5bca73982e11ae5ee3810731", "question_id": "q07_negation", "split": "test", "cluster_id": "syn_03021_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "tokenizer_hashes": {"receiver": "b9d8872cf3a99da99e1fc4b0f71232a451632dc49ca4c6771549b4efa5c1c763"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 8, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1725, "query_tokens": 498, "message_tokens": 394, "message_utf8_bytes": 1153, "cache_tensor_bytes": null, "serialized_payload_bytes": 1153, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 59152.118457946926, "hops": {"1": 22361.685457988642, "2": 5128.560750046745, "3": 5686.298708082177, "4": 4692.91458290536, "5": 5842.431625002064, "6": 4801.743416930549, "7": 5841.1449999548495, "8": 4797.338917036541}, "receiver_prefill_ms": 206.44120790529996, "ttft_ms": 11.569041991606355, "decode_ms": 23.052542004734278}, "total_wall_ms": 59393.60666600987, "ttft_ms": 11.569041991606355, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 2}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "Yes.", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 38, "retention_rate": 0.5846153846153846, "retention_by_kind": {"person": 0.8333333333333334, "date": 0.7, "amount": 0.38461538461538464, "status": 1.0, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1153}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 38, "retention_rate": 0.5846153846153846, "retention_by_kind": {"person": 0.8333333333333334, "date": 0.7, "amount": 0.38461538461538464, "status": 1.0, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1179}, "2": {"facts_total": 65, "facts_retained": 38, "retention_rate": 0.5846153846153846, "retention_by_kind": {"person": 0.8333333333333334, "date": 0.7, "amount": 0.38461538461538464, "status": 1.0, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1181}, "3": {"facts_total": 65, "facts_retained": 38, "retention_rate": 0.5846153846153846, "retention_by_kind": {"person": 0.8333333333333334, "date": 0.7, "amount": 0.38461538461538464, "status": 1.0, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1095}, "4": {"facts_total": 65, "facts_retained": 38, "retention_rate": 0.5846153846153846, "retention_by_kind": {"person": 0.8333333333333334, "date": 0.7, "amount": 0.38461538461538464, "status": 1.0, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1111}, "5": {"facts_total": 65, "facts_retained": 38, "retention_rate": 0.5846153846153846, "retention_by_kind": {"person": 0.8333333333333334, "date": 0.7, "amount": 0.38461538461538464, "status": 1.0, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1111}, "6": {"facts_total": 65, "facts_retained": 38, "retention_rate": 0.5846153846153846, "retention_by_kind": {"person": 0.8333333333333334, "date": 0.7, "amount": 0.38461538461538464, "status": 1.0, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1153}, "7": {"facts_total": 65, "facts_retained": 38, "retention_rate": 0.5846153846153846, "retention_by_kind": {"person": 0.8333333333333334, "date": 0.7, "amount": 0.38461538461538464, "status": 1.0, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1111}, "8": {"facts_total": 65, "facts_retained": 38, "retention_rate": 0.5846153846153846, "retention_by_kind": {"person": 0.8333333333333334, "date": 0.7, "amount": 0.38461538461538464, "status": 1.0, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1153}}}, "reference_key_id": "keys/synthetic/syn_03021_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "1d0e469668b88ffe7676989aa63ff9a7d27eb7c55484b06c14ba4e2a98a13623", "output_hash": "5f9a2b795615ba6a3d5455fd5624d773fbca5bcd16249c421fd37411dc9837da"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "4e69b55d29e1ef6ce3f95370", "attempt_id": "91ed5581f304", "run_group_id": "E4_core:A4|closed|qwen3b-llama3b:h8", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:21:49.624050Z", "task_family": "synthetic", "document_id": "syn_03021_2000", "document_hash": "3ac205b464552356699a4d2b0a7bf62497ead36e5bca73982e11ae5ee3810731", "question_id": "q08_revision", "split": "test", "cluster_id": "syn_03021_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "tokenizer_hashes": {"receiver": "b9d8872cf3a99da99e1fc4b0f71232a451632dc49ca4c6771549b4efa5c1c763"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 8, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1725, "query_tokens": 514, "message_tokens": 394, "message_utf8_bytes": 1153, "cache_tensor_bytes": null, "serialized_payload_bytes": 1153, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 59152.118457946926, "hops": {"1": 22361.685457988642, "2": 5128.560750046745, "3": 5686.298708082177, "4": 4692.91458290536, "5": 5842.431625002064, "6": 4801.743416930549, "7": 5841.1449999548495, "8": 4797.338917036541}, "receiver_prefill_ms": 209.82333295978606, "ttft_ms": 11.583083076402545, "decode_ms": 11.585792060941458}, "total_wall_ms": 59385.55579201784, "ttft_ms": 11.583083076402545, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 38, "retention_rate": 0.5846153846153846, "retention_by_kind": {"person": 0.8333333333333334, "date": 0.7, "amount": 0.38461538461538464, "status": 1.0, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1153}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 38, "retention_rate": 0.5846153846153846, "retention_by_kind": {"person": 0.8333333333333334, "date": 0.7, "amount": 0.38461538461538464, "status": 1.0, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1179}, "2": {"facts_total": 65, "facts_retained": 38, "retention_rate": 0.5846153846153846, "retention_by_kind": {"person": 0.8333333333333334, "date": 0.7, "amount": 0.38461538461538464, "status": 1.0, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1181}, "3": {"facts_total": 65, "facts_retained": 38, "retention_rate": 0.5846153846153846, "retention_by_kind": {"person": 0.8333333333333334, "date": 0.7, "amount": 0.38461538461538464, "status": 1.0, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1095}, "4": {"facts_total": 65, "facts_retained": 38, "retention_rate": 0.5846153846153846, "retention_by_kind": {"person": 0.8333333333333334, "date": 0.7, "amount": 0.38461538461538464, "status": 1.0, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1111}, "5": {"facts_total": 65, "facts_retained": 38, "retention_rate": 0.5846153846153846, "retention_by_kind": {"person": 0.8333333333333334, "date": 0.7, "amount": 0.38461538461538464, "status": 1.0, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1111}, "6": {"facts_total": 65, "facts_retained": 38, "retention_rate": 0.5846153846153846, "retention_by_kind": {"person": 0.8333333333333334, "date": 0.7, "amount": 0.38461538461538464, "status": 1.0, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1153}, "7": {"facts_total": 65, "facts_retained": 38, "retention_rate": 0.5846153846153846, "retention_by_kind": {"person": 0.8333333333333334, "date": 0.7, "amount": 0.38461538461538464, "status": 1.0, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1111}, "8": {"facts_total": 65, "facts_retained": 38, "retention_rate": 0.5846153846153846, "retention_by_kind": {"person": 0.8333333333333334, "date": 0.7, "amount": 0.38461538461538464, "status": 1.0, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1153}}}, "reference_key_id": "keys/synthetic/syn_03021_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "1d0e469668b88ffe7676989aa63ff9a7d27eb7c55484b06c14ba4e2a98a13623", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "db2e2ca503c3437617be4518", "attempt_id": "2e08cbf2b593", "run_group_id": "E4_core:A4|closed|qwen3b-llama3b:h8", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:21:49.872062Z", "task_family": "synthetic", "document_id": "syn_03021_2000", "document_hash": "3ac205b464552356699a4d2b0a7bf62497ead36e5bca73982e11ae5ee3810731", "question_id": "q09_revision", "split": "test", "cluster_id": "syn_03021_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "tokenizer_hashes": {"receiver": "b9d8872cf3a99da99e1fc4b0f71232a451632dc49ca4c6771549b4efa5c1c763"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 8, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1725, "query_tokens": 505, "message_tokens": 394, "message_utf8_bytes": 1153, "cache_tensor_bytes": null, "serialized_payload_bytes": 1153, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 59152.118457946926, "hops": {"1": 22361.685457988642, "2": 5128.560750046745, "3": 5686.298708082177, "4": 4692.91458290536, "5": 5842.431625002064, "6": 4801.743416930549, "7": 5841.1449999548495, "8": 4797.338917036541}, "receiver_prefill_ms": 207.5721659930423, "ttft_ms": 11.556250043213367, "decode_ms": 11.70158397872001}, "total_wall_ms": 59383.36149987299, "ttft_ms": 11.556250043213367, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 38, "retention_rate": 0.5846153846153846, "retention_by_kind": {"person": 0.8333333333333334, "date": 0.7, "amount": 0.38461538461538464, "status": 1.0, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1153}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 38, "retention_rate": 0.5846153846153846, "retention_by_kind": {"person": 0.8333333333333334, "date": 0.7, "amount": 0.38461538461538464, "status": 1.0, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1179}, "2": {"facts_total": 65, "facts_retained": 38, "retention_rate": 0.5846153846153846, "retention_by_kind": {"person": 0.8333333333333334, "date": 0.7, "amount": 0.38461538461538464, "status": 1.0, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1181}, "3": {"facts_total": 65, "facts_retained": 38, "retention_rate": 0.5846153846153846, "retention_by_kind": {"person": 0.8333333333333334, "date": 0.7, "amount": 0.38461538461538464, "status": 1.0, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1095}, "4": {"facts_total": 65, "facts_retained": 38, "retention_rate": 0.5846153846153846, "retention_by_kind": {"person": 0.8333333333333334, "date": 0.7, "amount": 0.38461538461538464, "status": 1.0, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1111}, "5": {"facts_total": 65, "facts_retained": 38, "retention_rate": 0.5846153846153846, "retention_by_kind": {"person": 0.8333333333333334, "date": 0.7, "amount": 0.38461538461538464, "status": 1.0, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1111}, "6": {"facts_total": 65, "facts_retained": 38, "retention_rate": 0.5846153846153846, "retention_by_kind": {"person": 0.8333333333333334, "date": 0.7, "amount": 0.38461538461538464, "status": 1.0, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1153}, "7": {"facts_total": 65, "facts_retained": 38, "retention_rate": 0.5846153846153846, "retention_by_kind": {"person": 0.8333333333333334, "date": 0.7, "amount": 0.38461538461538464, "status": 1.0, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1111}, "8": {"facts_total": 65, "facts_retained": 38, "retention_rate": 0.5846153846153846, "retention_by_kind": {"person": 0.8333333333333334, "date": 0.7, "amount": 0.38461538461538464, "status": 1.0, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1153}}}, "reference_key_id": "keys/synthetic/syn_03021_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "1d0e469668b88ffe7676989aa63ff9a7d27eb7c55484b06c14ba4e2a98a13623", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "554d52fa0c367eb1ec9d15aa", "attempt_id": "5324c66f387a", "run_group_id": "E4_core:A4|closed|qwen3b-llama3b:h8", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:21:50.118455Z", "task_family": "synthetic", "document_id": "syn_03021_2000", "document_hash": "3ac205b464552356699a4d2b0a7bf62497ead36e5bca73982e11ae5ee3810731", "question_id": "q10_temporal", "split": "test", "cluster_id": "syn_03021_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "tokenizer_hashes": {"receiver": "b9d8872cf3a99da99e1fc4b0f71232a451632dc49ca4c6771549b4efa5c1c763"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 8, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1725, "query_tokens": 513, "message_tokens": 394, "message_utf8_bytes": 1153, "cache_tensor_bytes": null, "serialized_payload_bytes": 1153, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 59152.118457946926, "hops": {"1": 22361.685457988642, "2": 5128.560750046745, "3": 5686.298708082177, "4": 4692.91458290536, "5": 5842.431625002064, "6": 4801.743416930549, "7": 5841.1449999548495, "8": 4797.338917036541}, "receiver_prefill_ms": 201.75858295988292, "ttft_ms": 12.069458025507629, "decode_ms": 11.520624975673854}, "total_wall_ms": 59377.917541889474, "ttft_ms": 12.069458025507629, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 38, "retention_rate": 0.5846153846153846, "retention_by_kind": {"person": 0.8333333333333334, "date": 0.7, "amount": 0.38461538461538464, "status": 1.0, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1153}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 38, "retention_rate": 0.5846153846153846, "retention_by_kind": {"person": 0.8333333333333334, "date": 0.7, "amount": 0.38461538461538464, "status": 1.0, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1179}, "2": {"facts_total": 65, "facts_retained": 38, "retention_rate": 0.5846153846153846, "retention_by_kind": {"person": 0.8333333333333334, "date": 0.7, "amount": 0.38461538461538464, "status": 1.0, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1181}, "3": {"facts_total": 65, "facts_retained": 38, "retention_rate": 0.5846153846153846, "retention_by_kind": {"person": 0.8333333333333334, "date": 0.7, "amount": 0.38461538461538464, "status": 1.0, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1095}, "4": {"facts_total": 65, "facts_retained": 38, "retention_rate": 0.5846153846153846, "retention_by_kind": {"person": 0.8333333333333334, "date": 0.7, "amount": 0.38461538461538464, "status": 1.0, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1111}, "5": {"facts_total": 65, "facts_retained": 38, "retention_rate": 0.5846153846153846, "retention_by_kind": {"person": 0.8333333333333334, "date": 0.7, "amount": 0.38461538461538464, "status": 1.0, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1111}, "6": {"facts_total": 65, "facts_retained": 38, "retention_rate": 0.5846153846153846, "retention_by_kind": {"person": 0.8333333333333334, "date": 0.7, "amount": 0.38461538461538464, "status": 1.0, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1153}, "7": {"facts_total": 65, "facts_retained": 38, "retention_rate": 0.5846153846153846, "retention_by_kind": {"person": 0.8333333333333334, "date": 0.7, "amount": 0.38461538461538464, "status": 1.0, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1111}, "8": {"facts_total": 65, "facts_retained": 38, "retention_rate": 0.5846153846153846, "retention_by_kind": {"person": 0.8333333333333334, "date": 0.7, "amount": 0.38461538461538464, "status": 1.0, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1153}}}, "reference_key_id": "keys/synthetic/syn_03021_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "1d0e469668b88ffe7676989aa63ff9a7d27eb7c55484b06c14ba4e2a98a13623", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "316c03a73ff4f029fa238073", "attempt_id": "ada79664ec3a", "run_group_id": "E4_core:A4|closed|qwen3b-llama3b:h8", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:21:50.359256Z", "task_family": "synthetic", "document_id": "syn_03021_2000", "document_hash": "3ac205b464552356699a4d2b0a7bf62497ead36e5bca73982e11ae5ee3810731", "question_id": "q11_aggregation", "split": "test", "cluster_id": "syn_03021_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "tokenizer_hashes": {"receiver": "b9d8872cf3a99da99e1fc4b0f71232a451632dc49ca4c6771549b4efa5c1c763"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 8, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1725, "query_tokens": 509, "message_tokens": 394, "message_utf8_bytes": 1153, "cache_tensor_bytes": null, "serialized_payload_bytes": 1153, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 59152.118457946926, "hops": {"1": 22361.685457988642, "2": 5128.560750046745, "3": 5686.298708082177, "4": 4692.91458290536, "5": 5842.431625002064, "6": 4801.743416930549, "7": 5841.1449999548495, "8": 4797.338917036541}, "receiver_prefill_ms": 209.5855000661686, "ttft_ms": 11.568957939743996, "decode_ms": 57.92950000613928}, "total_wall_ms": 59431.65408296045, "ttft_ms": 11.568957939743996, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 5}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "6,462,111", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 38, "retention_rate": 0.5846153846153846, "retention_by_kind": {"person": 0.8333333333333334, "date": 0.7, "amount": 0.38461538461538464, "status": 1.0, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1153}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 38, "retention_rate": 0.5846153846153846, "retention_by_kind": {"person": 0.8333333333333334, "date": 0.7, "amount": 0.38461538461538464, "status": 1.0, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1179}, "2": {"facts_total": 65, "facts_retained": 38, "retention_rate": 0.5846153846153846, "retention_by_kind": {"person": 0.8333333333333334, "date": 0.7, "amount": 0.38461538461538464, "status": 1.0, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1181}, "3": {"facts_total": 65, "facts_retained": 38, "retention_rate": 0.5846153846153846, "retention_by_kind": {"person": 0.8333333333333334, "date": 0.7, "amount": 0.38461538461538464, "status": 1.0, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1095}, "4": {"facts_total": 65, "facts_retained": 38, "retention_rate": 0.5846153846153846, "retention_by_kind": {"person": 0.8333333333333334, "date": 0.7, "amount": 0.38461538461538464, "status": 1.0, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1111}, "5": {"facts_total": 65, "facts_retained": 38, "retention_rate": 0.5846153846153846, "retention_by_kind": {"person": 0.8333333333333334, "date": 0.7, "amount": 0.38461538461538464, "status": 1.0, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1111}, "6": {"facts_total": 65, "facts_retained": 38, "retention_rate": 0.5846153846153846, "retention_by_kind": {"person": 0.8333333333333334, "date": 0.7, "amount": 0.38461538461538464, "status": 1.0, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1153}, "7": {"facts_total": 65, "facts_retained": 38, "retention_rate": 0.5846153846153846, "retention_by_kind": {"person": 0.8333333333333334, "date": 0.7, "amount": 0.38461538461538464, "status": 1.0, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1111}, "8": {"facts_total": 65, "facts_retained": 38, "retention_rate": 0.5846153846153846, "retention_by_kind": {"person": 0.8333333333333334, "date": 0.7, "amount": 0.38461538461538464, "status": 1.0, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1153}}}, "reference_key_id": "keys/synthetic/syn_03021_2000.json", "task_scores": {"correct": 0.0, "abstained": 0.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "1d0e469668b88ffe7676989aa63ff9a7d27eb7c55484b06c14ba4e2a98a13623", "output_hash": "1e8c5e2011479311a255cce5a1f78bfee345f6bb870c300bc4a818f9676ae0f7"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "daec267862fd7be55147e162", "attempt_id": "4eb8785ee3ae", "run_group_id": "E4_core:A4|closed|qwen3b-llama3b:h8", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:21:50.653864Z", "task_family": "synthetic", "document_id": "syn_03021_2000", "document_hash": "3ac205b464552356699a4d2b0a7bf62497ead36e5bca73982e11ae5ee3810731", "question_id": "q12_unit_conversion", "split": "test", "cluster_id": "syn_03021_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "tokenizer_hashes": {"receiver": "b9d8872cf3a99da99e1fc4b0f71232a451632dc49ca4c6771549b4efa5c1c763"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 8, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1725, "query_tokens": 504, "message_tokens": 394, "message_utf8_bytes": 1153, "cache_tensor_bytes": null, "serialized_payload_bytes": 1153, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 59152.118457946926, "hops": {"1": 22361.685457988642, "2": 5128.560750046745, "3": 5686.298708082177, "4": 4692.91458290536, "5": 5842.431625002064, "6": 4801.743416930549, "7": 5841.1449999548495, "8": 4797.338917036541}, "receiver_prefill_ms": 207.30849995743483, "ttft_ms": 11.553292046301067, "decode_ms": 11.53462496586144}, "total_wall_ms": 59382.938416907564, "ttft_ms": 11.553292046301067, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 38, "retention_rate": 0.5846153846153846, "retention_by_kind": {"person": 0.8333333333333334, "date": 0.7, "amount": 0.38461538461538464, "status": 1.0, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1153}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 38, "retention_rate": 0.5846153846153846, "retention_by_kind": {"person": 0.8333333333333334, "date": 0.7, "amount": 0.38461538461538464, "status": 1.0, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1179}, "2": {"facts_total": 65, "facts_retained": 38, "retention_rate": 0.5846153846153846, "retention_by_kind": {"person": 0.8333333333333334, "date": 0.7, "amount": 0.38461538461538464, "status": 1.0, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1181}, "3": {"facts_total": 65, "facts_retained": 38, "retention_rate": 0.5846153846153846, "retention_by_kind": {"person": 0.8333333333333334, "date": 0.7, "amount": 0.38461538461538464, "status": 1.0, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1095}, "4": {"facts_total": 65, "facts_retained": 38, "retention_rate": 0.5846153846153846, "retention_by_kind": {"person": 0.8333333333333334, "date": 0.7, "amount": 0.38461538461538464, "status": 1.0, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1111}, "5": {"facts_total": 65, "facts_retained": 38, "retention_rate": 0.5846153846153846, "retention_by_kind": {"person": 0.8333333333333334, "date": 0.7, "amount": 0.38461538461538464, "status": 1.0, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1111}, "6": {"facts_total": 65, "facts_retained": 38, "retention_rate": 0.5846153846153846, "retention_by_kind": {"person": 0.8333333333333334, "date": 0.7, "amount": 0.38461538461538464, "status": 1.0, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1153}, "7": {"facts_total": 65, "facts_retained": 38, "retention_rate": 0.5846153846153846, "retention_by_kind": {"person": 0.8333333333333334, "date": 0.7, "amount": 0.38461538461538464, "status": 1.0, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1111}, "8": {"facts_total": 65, "facts_retained": 38, "retention_rate": 0.5846153846153846, "retention_by_kind": {"person": 0.8333333333333334, "date": 0.7, "amount": 0.38461538461538464, "status": 1.0, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1153}}}, "reference_key_id": "keys/synthetic/syn_03021_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "1d0e469668b88ffe7676989aa63ff9a7d27eb7c55484b06c14ba4e2a98a13623", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "16ee8ff78db3d832aa9646c6", "attempt_id": "f12a737cf5bc", "run_group_id": "E4_core:A4|closed|qwen3b-llama3b:h8", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:21:50.899998Z", "task_family": "synthetic", "document_id": "syn_03021_2000", "document_hash": "3ac205b464552356699a4d2b0a7bf62497ead36e5bca73982e11ae5ee3810731", "question_id": "q13_unanswerable", "split": "test", "cluster_id": "syn_03021_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "tokenizer_hashes": {"receiver": "b9d8872cf3a99da99e1fc4b0f71232a451632dc49ca4c6771549b4efa5c1c763"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 8, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1725, "query_tokens": 508, "message_tokens": 394, "message_utf8_bytes": 1153, "cache_tensor_bytes": null, "serialized_payload_bytes": 1153, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 59152.118457946926, "hops": {"1": 22361.685457988642, "2": 5128.560750046745, "3": 5686.298708082177, "4": 4692.91458290536, "5": 5842.431625002064, "6": 4801.743416930549, "7": 5841.1449999548495, "8": 4797.338917036541}, "receiver_prefill_ms": 208.131332998164, "ttft_ms": 11.484291986562312, "decode_ms": 11.48550002835691}, "total_wall_ms": 59383.632665965706, "ttft_ms": 11.484291986562312, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 38, "retention_rate": 0.5846153846153846, "retention_by_kind": {"person": 0.8333333333333334, "date": 0.7, "amount": 0.38461538461538464, "status": 1.0, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1153}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 38, "retention_rate": 0.5846153846153846, "retention_by_kind": {"person": 0.8333333333333334, "date": 0.7, "amount": 0.38461538461538464, "status": 1.0, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1179}, "2": {"facts_total": 65, "facts_retained": 38, "retention_rate": 0.5846153846153846, "retention_by_kind": {"person": 0.8333333333333334, "date": 0.7, "amount": 0.38461538461538464, "status": 1.0, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1181}, "3": {"facts_total": 65, "facts_retained": 38, "retention_rate": 0.5846153846153846, "retention_by_kind": {"person": 0.8333333333333334, "date": 0.7, "amount": 0.38461538461538464, "status": 1.0, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1095}, "4": {"facts_total": 65, "facts_retained": 38, "retention_rate": 0.5846153846153846, "retention_by_kind": {"person": 0.8333333333333334, "date": 0.7, "amount": 0.38461538461538464, "status": 1.0, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1111}, "5": {"facts_total": 65, "facts_retained": 38, "retention_rate": 0.5846153846153846, "retention_by_kind": {"person": 0.8333333333333334, "date": 0.7, "amount": 0.38461538461538464, "status": 1.0, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1111}, "6": {"facts_total": 65, "facts_retained": 38, "retention_rate": 0.5846153846153846, "retention_by_kind": {"person": 0.8333333333333334, "date": 0.7, "amount": 0.38461538461538464, "status": 1.0, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1153}, "7": {"facts_total": 65, "facts_retained": 38, "retention_rate": 0.5846153846153846, "retention_by_kind": {"person": 0.8333333333333334, "date": 0.7, "amount": 0.38461538461538464, "status": 1.0, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1111}, "8": {"facts_total": 65, "facts_retained": 38, "retention_rate": 0.5846153846153846, "retention_by_kind": {"person": 0.8333333333333334, "date": 0.7, "amount": 0.38461538461538464, "status": 1.0, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1153}}}, "reference_key_id": "keys/synthetic/syn_03021_2000.json", "task_scores": {"correct": 1.0, "abstained": 1.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "1d0e469668b88ffe7676989aa63ff9a7d27eb7c55484b06c14ba4e2a98a13623", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "403a324c3a443931c24c3464", "attempt_id": "667c30b3b48f", "run_group_id": "E4_core:A4|closed|qwen3b-llama3b:h8", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:21:51.146374Z", "task_family": "synthetic", "document_id": "syn_03021_2000", "document_hash": "3ac205b464552356699a4d2b0a7bf62497ead36e5bca73982e11ae5ee3810731", "question_id": "q14_unanswerable", "split": "test", "cluster_id": "syn_03021_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "tokenizer_hashes": {"receiver": "b9d8872cf3a99da99e1fc4b0f71232a451632dc49ca4c6771549b4efa5c1c763"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 8, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1725, "query_tokens": 511, "message_tokens": 394, "message_utf8_bytes": 1153, "cache_tensor_bytes": null, "serialized_payload_bytes": 1153, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 59152.118457946926, "hops": {"1": 22361.685457988642, "2": 5128.560750046745, "3": 5686.298708082177, "4": 4692.91458290536, "5": 5842.431625002064, "6": 4801.743416930549, "7": 5841.1449999548495, "8": 4797.338917036541}, "receiver_prefill_ms": 209.0514999581501, "ttft_ms": 11.6252499865368, "decode_ms": 11.530208983458579}, "total_wall_ms": 59384.7572079394, "ttft_ms": 11.6252499865368, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 38, "retention_rate": 0.5846153846153846, "retention_by_kind": {"person": 0.8333333333333334, "date": 0.7, "amount": 0.38461538461538464, "status": 1.0, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1153}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 38, "retention_rate": 0.5846153846153846, "retention_by_kind": {"person": 0.8333333333333334, "date": 0.7, "amount": 0.38461538461538464, "status": 1.0, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1179}, "2": {"facts_total": 65, "facts_retained": 38, "retention_rate": 0.5846153846153846, "retention_by_kind": {"person": 0.8333333333333334, "date": 0.7, "amount": 0.38461538461538464, "status": 1.0, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1181}, "3": {"facts_total": 65, "facts_retained": 38, "retention_rate": 0.5846153846153846, "retention_by_kind": {"person": 0.8333333333333334, "date": 0.7, "amount": 0.38461538461538464, "status": 1.0, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1095}, "4": {"facts_total": 65, "facts_retained": 38, "retention_rate": 0.5846153846153846, "retention_by_kind": {"person": 0.8333333333333334, "date": 0.7, "amount": 0.38461538461538464, "status": 1.0, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1111}, "5": {"facts_total": 65, "facts_retained": 38, "retention_rate": 0.5846153846153846, "retention_by_kind": {"person": 0.8333333333333334, "date": 0.7, "amount": 0.38461538461538464, "status": 1.0, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1111}, "6": {"facts_total": 65, "facts_retained": 38, "retention_rate": 0.5846153846153846, "retention_by_kind": {"person": 0.8333333333333334, "date": 0.7, "amount": 0.38461538461538464, "status": 1.0, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1153}, "7": {"facts_total": 65, "facts_retained": 38, "retention_rate": 0.5846153846153846, "retention_by_kind": {"person": 0.8333333333333334, "date": 0.7, "amount": 0.38461538461538464, "status": 1.0, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1111}, "8": {"facts_total": 65, "facts_retained": 38, "retention_rate": 0.5846153846153846, "retention_by_kind": {"person": 0.8333333333333334, "date": 0.7, "amount": 0.38461538461538464, "status": 1.0, "place": 0.0, "identifier": 0.5, "quantity": 0.5}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1153}}}, "reference_key_id": "keys/synthetic/syn_03021_2000.json", "task_scores": {"correct": 1.0, "abstained": 1.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "1d0e469668b88ffe7676989aa63ff9a7d27eb7c55484b06c14ba4e2a98a13623", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "d94bd2b391c9476d7942e1a9", "attempt_id": "d24b59948acb", "run_group_id": "E4_core:A5|closed|qwen3b-llama3b:h1", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:21:58.292749Z", "task_family": "synthetic", "document_id": "syn_03021_2000", "document_hash": "3ac205b464552356699a4d2b0a7bf62497ead36e5bca73982e11ae5ee3810731", "question_id": "q00_single_fact", "split": "test", "cluster_id": "syn_03021_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 1, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1914, "query_tokens": 460, "message_tokens": 379, "message_utf8_bytes": 900, "cache_tensor_bytes": null, "serialized_payload_bytes": 900, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 6896.219208021648, "hops": {"1": 6896.219208021648}, "receiver_prefill_ms": 185.3049589553848, "ttft_ms": 11.899915989488363, "decode_ms": 11.902792030014098}, "total_wall_ms": 7105.795498937368, "ttft_ms": 11.899915989488363, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 19, "retention_rate": 0.2923076923076923, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.15384615384615385, "status": 1.0, "place": 0.0, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 836}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 19, "retention_rate": 0.2923076923076923, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.15384615384615385, "status": 1.0, "place": 0.0, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 836}}}, "reference_key_id": "keys/synthetic/syn_03021_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "2ae8e199245ad716748b3e420d7b2c8e50b7daeeebd965effa93b31557ba8942", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "c627238105d4610f85562eb6", "attempt_id": "a5f5dcd96290", "run_group_id": "E4_core:A5|closed|qwen3b-llama3b:h1", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:21:58.519084Z", "task_family": "synthetic", "document_id": "syn_03021_2000", "document_hash": "3ac205b464552356699a4d2b0a7bf62497ead36e5bca73982e11ae5ee3810731", "question_id": "q01_single_fact", "split": "test", "cluster_id": "syn_03021_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 1, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1914, "query_tokens": 466, "message_tokens": 379, "message_utf8_bytes": 900, "cache_tensor_bytes": null, "serialized_payload_bytes": 900, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 6896.219208021648, "hops": {"1": 6896.219208021648}, "receiver_prefill_ms": 185.94200001098216, "ttft_ms": 11.91279198974371, "decode_ms": 11.917916941456497}, "total_wall_ms": 7106.413958012126, "ttft_ms": 11.91279198974371, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 19, "retention_rate": 0.2923076923076923, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.15384615384615385, "status": 1.0, "place": 0.0, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 836}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 19, "retention_rate": 0.2923076923076923, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.15384615384615385, "status": 1.0, "place": 0.0, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 836}}}, "reference_key_id": "keys/synthetic/syn_03021_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "2ae8e199245ad716748b3e420d7b2c8e50b7daeeebd965effa93b31557ba8942", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "ad23cd890228829d4da8d7a1", "attempt_id": "68906bce5b5d", "run_group_id": "E4_core:A5|closed|qwen3b-llama3b:h1", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:21:58.744947Z", "task_family": "synthetic", "document_id": "syn_03021_2000", "document_hash": "3ac205b464552356699a4d2b0a7bf62497ead36e5bca73982e11ae5ee3810731", "question_id": "q02_exact_copy", "split": "test", "cluster_id": "syn_03021_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 1, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1914, "query_tokens": 466, "message_tokens": 379, "message_utf8_bytes": 900, "cache_tensor_bytes": null, "serialized_payload_bytes": 900, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 6896.219208021648, "hops": {"1": 6896.219208021648}, "receiver_prefill_ms": 185.74570794589818, "ttft_ms": 11.916583054699004, "decode_ms": 11.870707967318594}, "total_wall_ms": 7106.15970811341, "ttft_ms": 11.916583054699004, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 19, "retention_rate": 0.2923076923076923, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.15384615384615385, "status": 1.0, "place": 0.0, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 836}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 19, "retention_rate": 0.2923076923076923, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.15384615384615385, "status": 1.0, "place": 0.0, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 836}}}, "reference_key_id": "keys/synthetic/syn_03021_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "2ae8e199245ad716748b3e420d7b2c8e50b7daeeebd965effa93b31557ba8942", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "298ded8dadbaead1a4fe8379", "attempt_id": "e90500f4de7f", "run_group_id": "E4_core:A5|closed|qwen3b-llama3b:h1", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:21:58.970603Z", "task_family": "synthetic", "document_id": "syn_03021_2000", "document_hash": "3ac205b464552356699a4d2b0a7bf62497ead36e5bca73982e11ae5ee3810731", "question_id": "q03_exact_copy", "split": "test", "cluster_id": "syn_03021_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 1, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1914, "query_tokens": 463, "message_tokens": 379, "message_utf8_bytes": 900, "cache_tensor_bytes": null, "serialized_payload_bytes": 900, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 6896.219208021648, "hops": {"1": 6896.219208021648}, "receiver_prefill_ms": 185.68716698791832, "ttft_ms": 11.910875095054507, "decode_ms": 11.900958023034036}, "total_wall_ms": 7106.149208033457, "ttft_ms": 11.910875095054507, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 19, "retention_rate": 0.2923076923076923, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.15384615384615385, "status": 1.0, "place": 0.0, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 836}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 19, "retention_rate": 0.2923076923076923, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.15384615384615385, "status": 1.0, "place": 0.0, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 836}}}, "reference_key_id": "keys/synthetic/syn_03021_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "2ae8e199245ad716748b3e420d7b2c8e50b7daeeebd965effa93b31557ba8942", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "fdb95e87f6532bec0e8ea726", "attempt_id": "161443b5114e", "run_group_id": "E4_core:A5|closed|qwen3b-llama3b:h1", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:21:59.196006Z", "task_family": "synthetic", "document_id": "syn_03021_2000", "document_hash": "3ac205b464552356699a4d2b0a7bf62497ead36e5bca73982e11ae5ee3810731", "question_id": "q04_multi_hop", "split": "test", "cluster_id": "syn_03021_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 1, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1914, "query_tokens": 470, "message_tokens": 379, "message_utf8_bytes": 900, "cache_tensor_bytes": null, "serialized_payload_bytes": 900, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 6896.219208021648, "hops": {"1": 6896.219208021648}, "receiver_prefill_ms": 186.10466609243304, "ttft_ms": 11.861750041134655, "decode_ms": 11.738874949514866}, "total_wall_ms": 7106.345208012499, "ttft_ms": 11.861750041134655, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 19, "retention_rate": 0.2923076923076923, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.15384615384615385, "status": 1.0, "place": 0.0, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 836}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 19, "retention_rate": 0.2923076923076923, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.15384615384615385, "status": 1.0, "place": 0.0, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 836}}}, "reference_key_id": "keys/synthetic/syn_03021_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "2ae8e199245ad716748b3e420d7b2c8e50b7daeeebd965effa93b31557ba8942", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "4691649577f490c260a5e72b", "attempt_id": "590d592a7558", "run_group_id": "E4_core:A5|closed|qwen3b-llama3b:h1", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:21:59.421712Z", "task_family": "synthetic", "document_id": "syn_03021_2000", "document_hash": "3ac205b464552356699a4d2b0a7bf62497ead36e5bca73982e11ae5ee3810731", "question_id": "q05_multi_hop", "split": "test", "cluster_id": "syn_03021_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 1, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1914, "query_tokens": 471, "message_tokens": 379, "message_utf8_bytes": 900, "cache_tensor_bytes": null, "serialized_payload_bytes": 900, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 6896.219208021648, "hops": {"1": 6896.219208021648}, "receiver_prefill_ms": 186.07658299151808, "ttft_ms": 11.910666013136506, "decode_ms": 118.41008299961686}, "total_wall_ms": 7213.05958298035, "ttft_ms": 11.910666013136506, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 10}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "MH-96F-2099", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 19, "retention_rate": 0.2923076923076923, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.15384615384615385, "status": 1.0, "place": 0.0, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 836}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 19, "retention_rate": 0.2923076923076923, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.15384615384615385, "status": 1.0, "place": 0.0, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 836}}}, "reference_key_id": "keys/synthetic/syn_03021_2000.json", "task_scores": {"correct": 0.0, "abstained": 0.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "2ae8e199245ad716748b3e420d7b2c8e50b7daeeebd965effa93b31557ba8942", "output_hash": "95c230d14a055d30e0e1d4355caf07808e1c968b1feab9e0607d8bd37b1b43dc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "21efce31b54db81fd2083aa9", "attempt_id": "9d1eaba4b5f2", "run_group_id": "E4_core:A5|closed|qwen3b-llama3b:h1", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:21:59.755399Z", "task_family": "synthetic", "document_id": "syn_03021_2000", "document_hash": "3ac205b464552356699a4d2b0a7bf62497ead36e5bca73982e11ae5ee3810731", "question_id": "q06_negation", "split": "test", "cluster_id": "syn_03021_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 1, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1914, "query_tokens": 461, "message_tokens": 379, "message_utf8_bytes": 900, "cache_tensor_bytes": null, "serialized_payload_bytes": 900, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 6896.219208021648, "hops": {"1": 6896.219208021648}, "receiver_prefill_ms": 185.44695805758238, "ttft_ms": 11.878374963998795, "decode_ms": 11.877959012053907}, "total_wall_ms": 7105.824374011718, "ttft_ms": 11.878374963998795, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "no", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 19, "retention_rate": 0.2923076923076923, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.15384615384615385, "status": 1.0, "place": 0.0, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 836}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 19, "retention_rate": 0.2923076923076923, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.15384615384615385, "status": 1.0, "place": 0.0, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 836}}}, "reference_key_id": "keys/synthetic/syn_03021_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "2ae8e199245ad716748b3e420d7b2c8e50b7daeeebd965effa93b31557ba8942", "output_hash": "9390298f3fb0c5b160498935d79cb139aef28e1c47358b4bbba61862b9c26e59"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "70651103e2d6240d7d54aa63", "attempt_id": "beaf88b29f38", "run_group_id": "E4_core:A5|closed|qwen3b-llama3b:h1", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:21:59.980738Z", "task_family": "synthetic", "document_id": "syn_03021_2000", "document_hash": "3ac205b464552356699a4d2b0a7bf62497ead36e5bca73982e11ae5ee3810731", "question_id": "q07_negation", "split": "test", "cluster_id": "syn_03021_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 1, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1914, "query_tokens": 462, "message_tokens": 379, "message_utf8_bytes": 900, "cache_tensor_bytes": null, "serialized_payload_bytes": 900, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 6896.219208021648, "hops": {"1": 6896.219208021648}, "receiver_prefill_ms": 185.67370797973126, "ttft_ms": 11.92712492775172, "decode_ms": 11.881958926096559}, "total_wall_ms": 7106.162457959726, "ttft_ms": 11.92712492775172, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "no", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 19, "retention_rate": 0.2923076923076923, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.15384615384615385, "status": 1.0, "place": 0.0, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 836}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 19, "retention_rate": 0.2923076923076923, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.15384615384615385, "status": 1.0, "place": 0.0, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 836}}}, "reference_key_id": "keys/synthetic/syn_03021_2000.json", "task_scores": {"correct": 0.0, "abstained": 0.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "2ae8e199245ad716748b3e420d7b2c8e50b7daeeebd965effa93b31557ba8942", "output_hash": "9390298f3fb0c5b160498935d79cb139aef28e1c47358b4bbba61862b9c26e59"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "6bac3956a13836af56a96c9f", "attempt_id": "ec25bf611281", "run_group_id": "E4_core:A5|closed|qwen3b-llama3b:h1", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:22:00.205976Z", "task_family": "synthetic", "document_id": "syn_03021_2000", "document_hash": "3ac205b464552356699a4d2b0a7bf62497ead36e5bca73982e11ae5ee3810731", "question_id": "q08_revision", "split": "test", "cluster_id": "syn_03021_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 1, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1914, "query_tokens": 478, "message_tokens": 379, "message_utf8_bytes": 900, "cache_tensor_bytes": null, "serialized_payload_bytes": 900, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 6896.219208021648, "hops": {"1": 6896.219208021648}, "receiver_prefill_ms": 186.0346250468865, "ttft_ms": 11.902834055945277, "decode_ms": 11.930583976209164}, "total_wall_ms": 7106.514499988407, "ttft_ms": 11.902834055945277, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 19, "retention_rate": 0.2923076923076923, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.15384615384615385, "status": 1.0, "place": 0.0, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 836}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 19, "retention_rate": 0.2923076923076923, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.15384615384615385, "status": 1.0, "place": 0.0, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 836}}}, "reference_key_id": "keys/synthetic/syn_03021_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "2ae8e199245ad716748b3e420d7b2c8e50b7daeeebd965effa93b31557ba8942", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "ec08c98c3933b9421a97c0dd", "attempt_id": "3688ec5e64f1", "run_group_id": "E4_core:A5|closed|qwen3b-llama3b:h1", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:22:00.432038Z", "task_family": "synthetic", "document_id": "syn_03021_2000", "document_hash": "3ac205b464552356699a4d2b0a7bf62497ead36e5bca73982e11ae5ee3810731", "question_id": "q09_revision", "split": "test", "cluster_id": "syn_03021_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 1, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1914, "query_tokens": 469, "message_tokens": 379, "message_utf8_bytes": 900, "cache_tensor_bytes": null, "serialized_payload_bytes": 900, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 6896.219208021648, "hops": {"1": 6896.219208021648}, "receiver_prefill_ms": 185.93170901294798, "ttft_ms": 11.897167074494064, "decode_ms": 82.97449990641326}, "total_wall_ms": 7177.434250130318, "ttft_ms": 11.897167074494064, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 7}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "5646720", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 19, "retention_rate": 0.2923076923076923, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.15384615384615385, "status": 1.0, "place": 0.0, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 836}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 19, "retention_rate": 0.2923076923076923, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.15384615384615385, "status": 1.0, "place": 0.0, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 836}}}, "reference_key_id": "keys/synthetic/syn_03021_2000.json", "task_scores": {"correct": 0.0, "abstained": 0.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "2ae8e199245ad716748b3e420d7b2c8e50b7daeeebd965effa93b31557ba8942", "output_hash": "e1d129e860627cf4623671ed186cb4ebbf5d63169d3d0dc14a62e1da2af61326"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "99ead5cb42605d885af16045", "attempt_id": "6006dfe167ac", "run_group_id": "E4_core:A5|closed|qwen3b-llama3b:h1", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:22:00.728630Z", "task_family": "synthetic", "document_id": "syn_03021_2000", "document_hash": "3ac205b464552356699a4d2b0a7bf62497ead36e5bca73982e11ae5ee3810731", "question_id": "q10_temporal", "split": "test", "cluster_id": "syn_03021_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 1, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1914, "query_tokens": 477, "message_tokens": 379, "message_utf8_bytes": 900, "cache_tensor_bytes": null, "serialized_payload_bytes": 900, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 6896.219208021648, "hops": {"1": 6896.219208021648}, "receiver_prefill_ms": 186.8581249145791, "ttft_ms": 12.082583038136363, "decode_ms": 59.241166920401156}, "total_wall_ms": 7154.818832990713, "ttft_ms": 12.082583038136363, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 5}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "Kovar-9", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 19, "retention_rate": 0.2923076923076923, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.15384615384615385, "status": 1.0, "place": 0.0, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 836}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 19, "retention_rate": 0.2923076923076923, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.15384615384615385, "status": 1.0, "place": 0.0, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 836}}}, "reference_key_id": "keys/synthetic/syn_03021_2000.json", "task_scores": {"correct": 0.0, "abstained": 0.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "2ae8e199245ad716748b3e420d7b2c8e50b7daeeebd965effa93b31557ba8942", "output_hash": "c969b5f7547945067f5b07e9f822d0043126326f979d950a8b1b045057fbb818"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "5e2d348c636307add8dd93fa", "attempt_id": "95599e5221b8", "run_group_id": "E4_core:A5|closed|qwen3b-llama3b:h1", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:22:01.002997Z", "task_family": "synthetic", "document_id": "syn_03021_2000", "document_hash": "3ac205b464552356699a4d2b0a7bf62497ead36e5bca73982e11ae5ee3810731", "question_id": "q11_aggregation", "split": "test", "cluster_id": "syn_03021_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 1, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1914, "query_tokens": 473, "message_tokens": 379, "message_utf8_bytes": 900, "cache_tensor_bytes": null, "serialized_payload_bytes": 900, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 6896.219208021648, "hops": {"1": 6896.219208021648}, "receiver_prefill_ms": 186.658582999371, "ttft_ms": 11.848250054754317, "decode_ms": 83.23183306492865}, "total_wall_ms": 7178.38500009384, "ttft_ms": 11.848250054754317, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 7}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "9484825", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 19, "retention_rate": 0.2923076923076923, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.15384615384615385, "status": 1.0, "place": 0.0, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 836}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 19, "retention_rate": 0.2923076923076923, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.15384615384615385, "status": 1.0, "place": 0.0, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 836}}}, "reference_key_id": "keys/synthetic/syn_03021_2000.json", "task_scores": {"correct": 0.0, "abstained": 0.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "2ae8e199245ad716748b3e420d7b2c8e50b7daeeebd965effa93b31557ba8942", "output_hash": "667b970507c4c6c7fb572757eb3b5eb04186ddb67a6ef680266b7a0924737993"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "4cc7f5f99bcd42d25d3b32c8", "attempt_id": "db0b1823d1af", "run_group_id": "E4_core:A5|closed|qwen3b-llama3b:h1", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:22:01.300606Z", "task_family": "synthetic", "document_id": "syn_03021_2000", "document_hash": "3ac205b464552356699a4d2b0a7bf62497ead36e5bca73982e11ae5ee3810731", "question_id": "q12_unit_conversion", "split": "test", "cluster_id": "syn_03021_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 1, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1914, "query_tokens": 468, "message_tokens": 379, "message_utf8_bytes": 900, "cache_tensor_bytes": null, "serialized_payload_bytes": 900, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 6896.219208021648, "hops": {"1": 6896.219208021648}, "receiver_prefill_ms": 185.568124987185, "ttft_ms": 11.901542078703642, "decode_ms": 47.460832982324064}, "total_wall_ms": 7141.546333092265, "ttft_ms": 11.901542078703642, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 4}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "22.3", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 19, "retention_rate": 0.2923076923076923, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.15384615384615385, "status": 1.0, "place": 0.0, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 836}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 19, "retention_rate": 0.2923076923076923, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.15384615384615385, "status": 1.0, "place": 0.0, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 836}}}, "reference_key_id": "keys/synthetic/syn_03021_2000.json", "task_scores": {"correct": 0.0, "abstained": 0.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "2ae8e199245ad716748b3e420d7b2c8e50b7daeeebd965effa93b31557ba8942", "output_hash": "8afb0bb35e730fd32e7805d65081241bd21f43709df37165f4ec91c763f8a99d"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "bd0dda3d16fc4715cfe7cd86", "attempt_id": "d57050504468", "run_group_id": "E4_core:A5|closed|qwen3b-llama3b:h1", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:22:01.561261Z", "task_family": "synthetic", "document_id": "syn_03021_2000", "document_hash": "3ac205b464552356699a4d2b0a7bf62497ead36e5bca73982e11ae5ee3810731", "question_id": "q13_unanswerable", "split": "test", "cluster_id": "syn_03021_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 1, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1914, "query_tokens": 473, "message_tokens": 379, "message_utf8_bytes": 900, "cache_tensor_bytes": null, "serialized_payload_bytes": 900, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 6896.219208021648, "hops": {"1": 6896.219208021648}, "receiver_prefill_ms": 185.87141600437462, "ttft_ms": 11.87141693662852, "decode_ms": 11.888000066392124}, "total_wall_ms": 7106.263792025857, "ttft_ms": 11.87141693662852, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 19, "retention_rate": 0.2923076923076923, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.15384615384615385, "status": 1.0, "place": 0.0, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 836}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 19, "retention_rate": 0.2923076923076923, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.15384615384615385, "status": 1.0, "place": 0.0, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 836}}}, "reference_key_id": "keys/synthetic/syn_03021_2000.json", "task_scores": {"correct": 1.0, "abstained": 1.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "2ae8e199245ad716748b3e420d7b2c8e50b7daeeebd965effa93b31557ba8942", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "b6fb0cc24c8692d1e1421e0c", "attempt_id": "2263ec42fefc", "run_group_id": "E4_core:A5|closed|qwen3b-llama3b:h1", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:22:01.786559Z", "task_family": "synthetic", "document_id": "syn_03021_2000", "document_hash": "3ac205b464552356699a4d2b0a7bf62497ead36e5bca73982e11ae5ee3810731", "question_id": "q14_unanswerable", "split": "test", "cluster_id": "syn_03021_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 1, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1914, "query_tokens": 475, "message_tokens": 379, "message_utf8_bytes": 900, "cache_tensor_bytes": null, "serialized_payload_bytes": 900, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 6896.219208021648, "hops": {"1": 6896.219208021648}, "receiver_prefill_ms": 185.947582940571, "ttft_ms": 12.029374949634075, "decode_ms": 11.90191600471735}, "total_wall_ms": 7106.496624997817, "ttft_ms": 12.029374949634075, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 19, "retention_rate": 0.2923076923076923, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.15384615384615385, "status": 1.0, "place": 0.0, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 836}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 19, "retention_rate": 0.2923076923076923, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.15384615384615385, "status": 1.0, "place": 0.0, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 836}}}, "reference_key_id": "keys/synthetic/syn_03021_2000.json", "task_scores": {"correct": 1.0, "abstained": 1.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "2ae8e199245ad716748b3e420d7b2c8e50b7daeeebd965effa93b31557ba8942", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "7ee5cdf80b64d488a3b00110", "attempt_id": "0eb57dbca981", "run_group_id": "E4_core:A5|closed|qwen3b-llama3b:h2", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:22:06.307676Z", "task_family": "synthetic", "document_id": "syn_03021_2000", "document_hash": "3ac205b464552356699a4d2b0a7bf62497ead36e5bca73982e11ae5ee3810731", "question_id": "q00_single_fact", "split": "test", "cluster_id": "syn_03021_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "tokenizer_hashes": {"receiver": "b9d8872cf3a99da99e1fc4b0f71232a451632dc49ca4c6771549b4efa5c1c763"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 2, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1725, "query_tokens": 452, "message_tokens": 350, "message_utf8_bytes": 985, "cache_tensor_bytes": null, "serialized_payload_bytes": 985, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 11190.593583043665, "hops": {"1": 6896.219208021648, "2": 4294.3743750220165}, "receiver_prefill_ms": 200.07650007028133, "ttft_ms": 11.643542093224823, "decode_ms": 11.542499996721745}, "total_wall_ms": 11414.357333094813, "ttft_ms": 11.643542093224823, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 19, "retention_rate": 0.2923076923076923, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.15384615384615385, "status": 1.0, "place": 0.0, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 985}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 19, "retention_rate": 0.2923076923076923, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.15384615384615385, "status": 1.0, "place": 0.0, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 836}, "2": {"facts_total": 65, "facts_retained": 19, "retention_rate": 0.2923076923076923, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.15384615384615385, "status": 1.0, "place": 0.0, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 985}}}, "reference_key_id": "keys/synthetic/syn_03021_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "dc8b132a30e8989cf3c15b7b1a161c9730440bcba369b175bdd47033bb2266fc", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "4519603f860f348a0a6c6b2d", "attempt_id": "ebd3a9b74a86", "run_group_id": "E4_core:A5|closed|qwen3b-llama3b:h2", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:22:06.547993Z", "task_family": "synthetic", "document_id": "syn_03021_2000", "document_hash": "3ac205b464552356699a4d2b0a7bf62497ead36e5bca73982e11ae5ee3810731", "question_id": "q01_single_fact", "split": "test", "cluster_id": "syn_03021_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "tokenizer_hashes": {"receiver": "b9d8872cf3a99da99e1fc4b0f71232a451632dc49ca4c6771549b4efa5c1c763"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 2, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1725, "query_tokens": 458, "message_tokens": 350, "message_utf8_bytes": 985, "cache_tensor_bytes": null, "serialized_payload_bytes": 985, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 11190.593583043665, "hops": {"1": 6896.219208021648, "2": 4294.3743750220165}, "receiver_prefill_ms": 202.44499994441867, "ttft_ms": 11.635957984253764, "decode_ms": 11.510084033943713}, "total_wall_ms": 11416.62091598846, "ttft_ms": 11.635957984253764, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 19, "retention_rate": 0.2923076923076923, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.15384615384615385, "status": 1.0, "place": 0.0, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 985}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 19, "retention_rate": 0.2923076923076923, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.15384615384615385, "status": 1.0, "place": 0.0, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 836}, "2": {"facts_total": 65, "facts_retained": 19, "retention_rate": 0.2923076923076923, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.15384615384615385, "status": 1.0, "place": 0.0, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 985}}}, "reference_key_id": "keys/synthetic/syn_03021_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "dc8b132a30e8989cf3c15b7b1a161c9730440bcba369b175bdd47033bb2266fc", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "0210c299a673a58a8657a332", "attempt_id": "f57bb37548a9", "run_group_id": "E4_core:A5|closed|qwen3b-llama3b:h2", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:22:06.790192Z", "task_family": "synthetic", "document_id": "syn_03021_2000", "document_hash": "3ac205b464552356699a4d2b0a7bf62497ead36e5bca73982e11ae5ee3810731", "question_id": "q02_exact_copy", "split": "test", "cluster_id": "syn_03021_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "tokenizer_hashes": {"receiver": "b9d8872cf3a99da99e1fc4b0f71232a451632dc49ca4c6771549b4efa5c1c763"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 2, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1725, "query_tokens": 458, "message_tokens": 350, "message_utf8_bytes": 985, "cache_tensor_bytes": null, "serialized_payload_bytes": 985, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 11190.593583043665, "hops": {"1": 6896.219208021648, "2": 4294.3743750220165}, "receiver_prefill_ms": 202.78220798354596, "ttft_ms": 11.736541986465454, "decode_ms": 11.570625007152557}, "total_wall_ms": 11417.214958113618, "ttft_ms": 11.736541986465454, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 19, "retention_rate": 0.2923076923076923, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.15384615384615385, "status": 1.0, "place": 0.0, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 985}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 19, "retention_rate": 0.2923076923076923, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.15384615384615385, "status": 1.0, "place": 0.0, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 836}, "2": {"facts_total": 65, "facts_retained": 19, "retention_rate": 0.2923076923076923, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.15384615384615385, "status": 1.0, "place": 0.0, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 985}}}, "reference_key_id": "keys/synthetic/syn_03021_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "dc8b132a30e8989cf3c15b7b1a161c9730440bcba369b175bdd47033bb2266fc", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "c81f41666a0553f5a57da6eb", "attempt_id": "d81a7cc6762e", "run_group_id": "E4_core:A5|closed|qwen3b-llama3b:h2", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:22:07.032580Z", "task_family": "synthetic", "document_id": "syn_03021_2000", "document_hash": "3ac205b464552356699a4d2b0a7bf62497ead36e5bca73982e11ae5ee3810731", "question_id": "q03_exact_copy", "split": "test", "cluster_id": "syn_03021_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "tokenizer_hashes": {"receiver": "b9d8872cf3a99da99e1fc4b0f71232a451632dc49ca4c6771549b4efa5c1c763"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 2, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1725, "query_tokens": 455, "message_tokens": 350, "message_utf8_bytes": 985, "cache_tensor_bytes": null, "serialized_payload_bytes": 985, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 11190.593583043665, "hops": {"1": 6896.219208021648, "2": 4294.3743750220165}, "receiver_prefill_ms": 201.61716593429446, "ttft_ms": 11.642958037555218, "decode_ms": 11.505665956065059}, "total_wall_ms": 11415.777583024465, "ttft_ms": 11.642958037555218, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 19, "retention_rate": 0.2923076923076923, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.15384615384615385, "status": 1.0, "place": 0.0, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 985}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 19, "retention_rate": 0.2923076923076923, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.15384615384615385, "status": 1.0, "place": 0.0, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 836}, "2": {"facts_total": 65, "facts_retained": 19, "retention_rate": 0.2923076923076923, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.15384615384615385, "status": 1.0, "place": 0.0, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 985}}}, "reference_key_id": "keys/synthetic/syn_03021_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "dc8b132a30e8989cf3c15b7b1a161c9730440bcba369b175bdd47033bb2266fc", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "ff5dbe3458154d43b364f052", "attempt_id": "b993906eee4b", "run_group_id": "E4_core:A5|closed|qwen3b-llama3b:h2", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:22:07.273243Z", "task_family": "synthetic", "document_id": "syn_03021_2000", "document_hash": "3ac205b464552356699a4d2b0a7bf62497ead36e5bca73982e11ae5ee3810731", "question_id": "q04_multi_hop", "split": "test", "cluster_id": "syn_03021_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "tokenizer_hashes": {"receiver": "b9d8872cf3a99da99e1fc4b0f71232a451632dc49ca4c6771549b4efa5c1c763"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 2, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1725, "query_tokens": 462, "message_tokens": 350, "message_utf8_bytes": 985, "cache_tensor_bytes": null, "serialized_payload_bytes": 985, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 11190.593583043665, "hops": {"1": 6896.219208021648, "2": 4294.3743750220165}, "receiver_prefill_ms": 203.95933301188052, "ttft_ms": 11.543333996087313, "decode_ms": 11.669458006508648}, "total_wall_ms": 11418.174999998882, "ttft_ms": 11.543333996087313, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 19, "retention_rate": 0.2923076923076923, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.15384615384615385, "status": 1.0, "place": 0.0, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 985}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 19, "retention_rate": 0.2923076923076923, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.15384615384615385, "status": 1.0, "place": 0.0, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 836}, "2": {"facts_total": 65, "facts_retained": 19, "retention_rate": 0.2923076923076923, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.15384615384615385, "status": 1.0, "place": 0.0, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 985}}}, "reference_key_id": "keys/synthetic/syn_03021_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "dc8b132a30e8989cf3c15b7b1a161c9730440bcba369b175bdd47033bb2266fc", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "28281e6eec70eee2481a6751", "attempt_id": "5d4bd0a99718", "run_group_id": "E4_core:A5|closed|qwen3b-llama3b:h2", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:22:07.516176Z", "task_family": "synthetic", "document_id": "syn_03021_2000", "document_hash": "3ac205b464552356699a4d2b0a7bf62497ead36e5bca73982e11ae5ee3810731", "question_id": "q05_multi_hop", "split": "test", "cluster_id": "syn_03021_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "tokenizer_hashes": {"receiver": "b9d8872cf3a99da99e1fc4b0f71232a451632dc49ca4c6771549b4efa5c1c763"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 2, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1725, "query_tokens": 461, "message_tokens": 350, "message_utf8_bytes": 985, "cache_tensor_bytes": null, "serialized_payload_bytes": 985, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 11190.593583043665, "hops": {"1": 6896.219208021648, "2": 4294.3743750220165}, "receiver_prefill_ms": 203.4829999320209, "ttft_ms": 11.53916702605784, "decode_ms": 11.539792059920728}, "total_wall_ms": 11417.54320799373, "ttft_ms": 11.53916702605784, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 19, "retention_rate": 0.2923076923076923, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.15384615384615385, "status": 1.0, "place": 0.0, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 985}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 19, "retention_rate": 0.2923076923076923, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.15384615384615385, "status": 1.0, "place": 0.0, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 836}, "2": {"facts_total": 65, "facts_retained": 19, "retention_rate": 0.2923076923076923, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.15384615384615385, "status": 1.0, "place": 0.0, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 985}}}, "reference_key_id": "keys/synthetic/syn_03021_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "dc8b132a30e8989cf3c15b7b1a161c9730440bcba369b175bdd47033bb2266fc", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "cc418b36d9accce7b295a3e3", "attempt_id": "c706385040be", "run_group_id": "E4_core:A5|closed|qwen3b-llama3b:h2", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:22:07.758247Z", "task_family": "synthetic", "document_id": "syn_03021_2000", "document_hash": "3ac205b464552356699a4d2b0a7bf62497ead36e5bca73982e11ae5ee3810731", "question_id": "q06_negation", "split": "test", "cluster_id": "syn_03021_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "tokenizer_hashes": {"receiver": "b9d8872cf3a99da99e1fc4b0f71232a451632dc49ca4c6771549b4efa5c1c763"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 2, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1725, "query_tokens": 453, "message_tokens": 350, "message_utf8_bytes": 985, "cache_tensor_bytes": null, "serialized_payload_bytes": 985, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 11190.593583043665, "hops": {"1": 6896.219208021648, "2": 4294.3743750220165}, "receiver_prefill_ms": 202.41237501613796, "ttft_ms": 11.804334004409611, "decode_ms": 23.27070792671293}, "total_wall_ms": 11428.596249083057, "ttft_ms": 11.804334004409611, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 2}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "No.", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 19, "retention_rate": 0.2923076923076923, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.15384615384615385, "status": 1.0, "place": 0.0, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 985}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 19, "retention_rate": 0.2923076923076923, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.15384615384615385, "status": 1.0, "place": 0.0, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 836}, "2": {"facts_total": 65, "facts_retained": 19, "retention_rate": 0.2923076923076923, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.15384615384615385, "status": 1.0, "place": 0.0, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 985}}}, "reference_key_id": "keys/synthetic/syn_03021_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "dc8b132a30e8989cf3c15b7b1a161c9730440bcba369b175bdd47033bb2266fc", "output_hash": "38178a20b470cfd18299fb1593dd4ba706f1a83321dac051d01086efd8b7a96f"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "03dab83f7c9088cb70f1f04f", "attempt_id": "8074c2d95dc9", "run_group_id": "E4_core:A5|closed|qwen3b-llama3b:h2", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:22:08.011637Z", "task_family": "synthetic", "document_id": "syn_03021_2000", "document_hash": "3ac205b464552356699a4d2b0a7bf62497ead36e5bca73982e11ae5ee3810731", "question_id": "q07_negation", "split": "test", "cluster_id": "syn_03021_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "tokenizer_hashes": {"receiver": "b9d8872cf3a99da99e1fc4b0f71232a451632dc49ca4c6771549b4efa5c1c763"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 2, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1725, "query_tokens": 454, "message_tokens": 350, "message_utf8_bytes": 985, "cache_tensor_bytes": null, "serialized_payload_bytes": 985, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 11190.593583043665, "hops": {"1": 6896.219208021648, "2": 4294.3743750220165}, "receiver_prefill_ms": 199.69187502283603, "ttft_ms": 11.559708043932915, "decode_ms": 11.51358406059444}, "total_wall_ms": 11413.757000002079, "ttft_ms": 11.559708043932915, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "No", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 19, "retention_rate": 0.2923076923076923, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.15384615384615385, "status": 1.0, "place": 0.0, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 985}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 19, "retention_rate": 0.2923076923076923, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.15384615384615385, "status": 1.0, "place": 0.0, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 836}, "2": {"facts_total": 65, "facts_retained": 19, "retention_rate": 0.2923076923076923, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.15384615384615385, "status": 1.0, "place": 0.0, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 985}}}, "reference_key_id": "keys/synthetic/syn_03021_2000.json", "task_scores": {"correct": 0.0, "abstained": 0.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "dc8b132a30e8989cf3c15b7b1a161c9730440bcba369b175bdd47033bb2266fc", "output_hash": "1ea442a134b2a184bd5d40104401f2a37fbc09ccf3f4bc9da161c6099be3691d"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "fbd28c25e407a34fa567bdb9", "attempt_id": "a6f8fc675d36", "run_group_id": "E4_core:A5|closed|qwen3b-llama3b:h2", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:22:08.250525Z", "task_family": "synthetic", "document_id": "syn_03021_2000", "document_hash": "3ac205b464552356699a4d2b0a7bf62497ead36e5bca73982e11ae5ee3810731", "question_id": "q08_revision", "split": "test", "cluster_id": "syn_03021_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "tokenizer_hashes": {"receiver": "b9d8872cf3a99da99e1fc4b0f71232a451632dc49ca4c6771549b4efa5c1c763"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 2, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1725, "query_tokens": 470, "message_tokens": 350, "message_utf8_bytes": 985, "cache_tensor_bytes": null, "serialized_payload_bytes": 985, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 11190.593583043665, "hops": {"1": 6896.219208021648, "2": 4294.3743750220165}, "receiver_prefill_ms": 204.81916691642255, "ttft_ms": 11.57983299344778, "decode_ms": 11.55637507326901}, "total_wall_ms": 11418.969124089926, "ttft_ms": 11.57983299344778, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 19, "retention_rate": 0.2923076923076923, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.15384615384615385, "status": 1.0, "place": 0.0, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 985}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 19, "retention_rate": 0.2923076923076923, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.15384615384615385, "status": 1.0, "place": 0.0, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 836}, "2": {"facts_total": 65, "facts_retained": 19, "retention_rate": 0.2923076923076923, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.15384615384615385, "status": 1.0, "place": 0.0, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 985}}}, "reference_key_id": "keys/synthetic/syn_03021_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "dc8b132a30e8989cf3c15b7b1a161c9730440bcba369b175bdd47033bb2266fc", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "3367c76ca1ea311009197e69", "attempt_id": "4a9a55e26ed9", "run_group_id": "E4_core:A5|closed|qwen3b-llama3b:h2", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:22:08.494368Z", "task_family": "synthetic", "document_id": "syn_03021_2000", "document_hash": "3ac205b464552356699a4d2b0a7bf62497ead36e5bca73982e11ae5ee3810731", "question_id": "q09_revision", "split": "test", "cluster_id": "syn_03021_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "tokenizer_hashes": {"receiver": "b9d8872cf3a99da99e1fc4b0f71232a451632dc49ca4c6771549b4efa5c1c763"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 2, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1725, "query_tokens": 461, "message_tokens": 350, "message_utf8_bytes": 985, "cache_tensor_bytes": null, "serialized_payload_bytes": 985, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 11190.593583043665, "hops": {"1": 6896.219208021648, "2": 4294.3743750220165}, "receiver_prefill_ms": 203.2025830121711, "ttft_ms": 11.617250042036176, "decode_ms": 11.478040949441493}, "total_wall_ms": 11417.28666704148, "ttft_ms": 11.617250042036176, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 19, "retention_rate": 0.2923076923076923, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.15384615384615385, "status": 1.0, "place": 0.0, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 985}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 19, "retention_rate": 0.2923076923076923, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.15384615384615385, "status": 1.0, "place": 0.0, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 836}, "2": {"facts_total": 65, "facts_retained": 19, "retention_rate": 0.2923076923076923, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.15384615384615385, "status": 1.0, "place": 0.0, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 985}}}, "reference_key_id": "keys/synthetic/syn_03021_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "dc8b132a30e8989cf3c15b7b1a161c9730440bcba369b175bdd47033bb2266fc", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "d07793136fa65ff0a9ef5424", "attempt_id": "3dd7673edfd8", "run_group_id": "E4_core:A5|closed|qwen3b-llama3b:h2", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:22:08.736237Z", "task_family": "synthetic", "document_id": "syn_03021_2000", "document_hash": "3ac205b464552356699a4d2b0a7bf62497ead36e5bca73982e11ae5ee3810731", "question_id": "q10_temporal", "split": "test", "cluster_id": "syn_03021_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "tokenizer_hashes": {"receiver": "b9d8872cf3a99da99e1fc4b0f71232a451632dc49ca4c6771549b4efa5c1c763"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 2, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1725, "query_tokens": 469, "message_tokens": 350, "message_utf8_bytes": 985, "cache_tensor_bytes": null, "serialized_payload_bytes": 985, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 11190.593583043665, "hops": {"1": 6896.219208021648, "2": 4294.3743750220165}, "receiver_prefill_ms": 203.61879095435143, "ttft_ms": 11.561291990801692, "decode_ms": 11.525708017870784}, "total_wall_ms": 11417.696125106886, "ttft_ms": 11.561291990801692, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 19, "retention_rate": 0.2923076923076923, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.15384615384615385, "status": 1.0, "place": 0.0, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 985}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 19, "retention_rate": 0.2923076923076923, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.15384615384615385, "status": 1.0, "place": 0.0, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 836}, "2": {"facts_total": 65, "facts_retained": 19, "retention_rate": 0.2923076923076923, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.15384615384615385, "status": 1.0, "place": 0.0, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 985}}}, "reference_key_id": "keys/synthetic/syn_03021_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "dc8b132a30e8989cf3c15b7b1a161c9730440bcba369b175bdd47033bb2266fc", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "310b3a162b944435e740c295", "attempt_id": "b2e905b5590e", "run_group_id": "E4_core:A5|closed|qwen3b-llama3b:h2", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:22:08.978325Z", "task_family": "synthetic", "document_id": "syn_03021_2000", "document_hash": "3ac205b464552356699a4d2b0a7bf62497ead36e5bca73982e11ae5ee3810731", "question_id": "q11_aggregation", "split": "test", "cluster_id": "syn_03021_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "tokenizer_hashes": {"receiver": "b9d8872cf3a99da99e1fc4b0f71232a451632dc49ca4c6771549b4efa5c1c763"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 2, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1725, "query_tokens": 465, "message_tokens": 350, "message_utf8_bytes": 985, "cache_tensor_bytes": null, "serialized_payload_bytes": 985, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 11190.593583043665, "hops": {"1": 6896.219208021648, "2": 4294.3743750220165}, "receiver_prefill_ms": 202.004500082694, "ttft_ms": 11.60787499975413, "decode_ms": 57.55241704173386}, "total_wall_ms": 11462.183666066267, "ttft_ms": 11.60787499975413, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 5}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "5,646,720", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 19, "retention_rate": 0.2923076923076923, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.15384615384615385, "status": 1.0, "place": 0.0, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 985}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 19, "retention_rate": 0.2923076923076923, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.15384615384615385, "status": 1.0, "place": 0.0, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 836}, "2": {"facts_total": 65, "facts_retained": 19, "retention_rate": 0.2923076923076923, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.15384615384615385, "status": 1.0, "place": 0.0, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 985}}}, "reference_key_id": "keys/synthetic/syn_03021_2000.json", "task_scores": {"correct": 0.0, "abstained": 0.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "dc8b132a30e8989cf3c15b7b1a161c9730440bcba369b175bdd47033bb2266fc", "output_hash": "2c0bb14ad8ba61e6d23fb3dea5205f2f02c43c95a0cf8e7f93c240ddea626ad4"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "cc7d9218e620071898a4957b", "attempt_id": "4131ea95c074", "run_group_id": "E4_core:A5|closed|qwen3b-llama3b:h2", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:22:09.264754Z", "task_family": "synthetic", "document_id": "syn_03021_2000", "document_hash": "3ac205b464552356699a4d2b0a7bf62497ead36e5bca73982e11ae5ee3810731", "question_id": "q12_unit_conversion", "split": "test", "cluster_id": "syn_03021_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "tokenizer_hashes": {"receiver": "b9d8872cf3a99da99e1fc4b0f71232a451632dc49ca4c6771549b4efa5c1c763"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 2, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1725, "query_tokens": 460, "message_tokens": 350, "message_utf8_bytes": 985, "cache_tensor_bytes": null, "serialized_payload_bytes": 985, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 11190.593583043665, "hops": {"1": 6896.219208021648, "2": 4294.3743750220165}, "receiver_prefill_ms": 202.61016709264368, "ttft_ms": 11.743166018277407, "decode_ms": 11.507582967169583}, "total_wall_ms": 11416.990875033662, "ttft_ms": 11.743166018277407, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 19, "retention_rate": 0.2923076923076923, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.15384615384615385, "status": 1.0, "place": 0.0, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 985}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 19, "retention_rate": 0.2923076923076923, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.15384615384615385, "status": 1.0, "place": 0.0, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 836}, "2": {"facts_total": 65, "facts_retained": 19, "retention_rate": 0.2923076923076923, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.15384615384615385, "status": 1.0, "place": 0.0, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 985}}}, "reference_key_id": "keys/synthetic/syn_03021_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "dc8b132a30e8989cf3c15b7b1a161c9730440bcba369b175bdd47033bb2266fc", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "e762b48fc37a7271532bdabc", "attempt_id": "01e20a9e4f12", "run_group_id": "E4_core:A5|closed|qwen3b-llama3b:h2", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:22:09.506764Z", "task_family": "synthetic", "document_id": "syn_03021_2000", "document_hash": "3ac205b464552356699a4d2b0a7bf62497ead36e5bca73982e11ae5ee3810731", "question_id": "q13_unanswerable", "split": "test", "cluster_id": "syn_03021_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "tokenizer_hashes": {"receiver": "b9d8872cf3a99da99e1fc4b0f71232a451632dc49ca4c6771549b4efa5c1c763"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 2, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1725, "query_tokens": 464, "message_tokens": 350, "message_utf8_bytes": 985, "cache_tensor_bytes": null, "serialized_payload_bytes": 985, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 11190.593583043665, "hops": {"1": 6896.219208021648, "2": 4294.3743750220165}, "receiver_prefill_ms": 203.15854204818606, "ttft_ms": 11.63029100280255, "decode_ms": 11.580833001062274}, "total_wall_ms": 11417.369041126221, "ttft_ms": 11.63029100280255, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 19, "retention_rate": 0.2923076923076923, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.15384615384615385, "status": 1.0, "place": 0.0, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 985}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 19, "retention_rate": 0.2923076923076923, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.15384615384615385, "status": 1.0, "place": 0.0, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 836}, "2": {"facts_total": 65, "facts_retained": 19, "retention_rate": 0.2923076923076923, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.15384615384615385, "status": 1.0, "place": 0.0, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 985}}}, "reference_key_id": "keys/synthetic/syn_03021_2000.json", "task_scores": {"correct": 1.0, "abstained": 1.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "dc8b132a30e8989cf3c15b7b1a161c9730440bcba369b175bdd47033bb2266fc", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "d258f30414555b6c349fd536", "attempt_id": "f3e75c686786", "run_group_id": "E4_core:A5|closed|qwen3b-llama3b:h2", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:22:09.750179Z", "task_family": "synthetic", "document_id": "syn_03021_2000", "document_hash": "3ac205b464552356699a4d2b0a7bf62497ead36e5bca73982e11ae5ee3810731", "question_id": "q14_unanswerable", "split": "test", "cluster_id": "syn_03021_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "tokenizer_hashes": {"receiver": "b9d8872cf3a99da99e1fc4b0f71232a451632dc49ca4c6771549b4efa5c1c763"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 2, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1725, "query_tokens": 467, "message_tokens": 350, "message_utf8_bytes": 985, "cache_tensor_bytes": null, "serialized_payload_bytes": 985, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 11190.593583043665, "hops": {"1": 6896.219208021648, "2": 4294.3743750220165}, "receiver_prefill_ms": 206.60366595257074, "ttft_ms": 11.58149994444102, "decode_ms": 11.544415960088372}, "total_wall_ms": 11420.725042000413, "ttft_ms": 11.58149994444102, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 19, "retention_rate": 0.2923076923076923, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.15384615384615385, "status": 1.0, "place": 0.0, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 985}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 19, "retention_rate": 0.2923076923076923, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.15384615384615385, "status": 1.0, "place": 0.0, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 836}, "2": {"facts_total": 65, "facts_retained": 19, "retention_rate": 0.2923076923076923, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.15384615384615385, "status": 1.0, "place": 0.0, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 985}}}, "reference_key_id": "keys/synthetic/syn_03021_2000.json", "task_scores": {"correct": 1.0, "abstained": 1.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "dc8b132a30e8989cf3c15b7b1a161c9730440bcba369b175bdd47033bb2266fc", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "8f51afd6a3544ce47ce942b1", "attempt_id": "b3f44e92663a", "run_group_id": "E4_core:A5|closed|qwen3b-llama3b:h4", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:22:19.234183Z", "task_family": "synthetic", "document_id": "syn_03021_2000", "document_hash": "3ac205b464552356699a4d2b0a7bf62497ead36e5bca73982e11ae5ee3810731", "question_id": "q00_single_fact", "split": "test", "cluster_id": "syn_03021_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "tokenizer_hashes": {"receiver": "b9d8872cf3a99da99e1fc4b0f71232a451632dc49ca4c6771549b4efa5c1c763"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 4, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1725, "query_tokens": 452, "message_tokens": 350, "message_utf8_bytes": 985, "cache_tensor_bytes": null, "serialized_payload_bytes": 985, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 20426.742333103903, "hops": {"1": 6896.219208021648, "2": 4294.3743750220165, "3": 4930.94241595827, "4": 4305.2063341019675}, "receiver_prefill_ms": 200.9144159965217, "ttft_ms": 11.578792007640004, "decode_ms": 11.575500015169382}, "total_wall_ms": 20651.235541095957, "ttft_ms": 11.578792007640004, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 19, "retention_rate": 0.2923076923076923, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.15384615384615385, "status": 1.0, "place": 0.0, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 985}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 19, "retention_rate": 0.2923076923076923, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.15384615384615385, "status": 1.0, "place": 0.0, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 836}, "2": {"facts_total": 65, "facts_retained": 19, "retention_rate": 0.2923076923076923, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.15384615384615385, "status": 1.0, "place": 0.0, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 985}, "3": {"facts_total": 65, "facts_retained": 19, "retention_rate": 0.2923076923076923, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.15384615384615385, "status": 1.0, "place": 0.0, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 985}, "4": {"facts_total": 65, "facts_retained": 19, "retention_rate": 0.2923076923076923, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.15384615384615385, "status": 1.0, "place": 0.0, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 985}}}, "reference_key_id": "keys/synthetic/syn_03021_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "dc8b132a30e8989cf3c15b7b1a161c9730440bcba369b175bdd47033bb2266fc", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "e4298900a03447465092fed8", "attempt_id": "1d53ea775d50", "run_group_id": "E4_core:A5|closed|qwen3b-llama3b:h4", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:22:19.475048Z", "task_family": "synthetic", "document_id": "syn_03021_2000", "document_hash": "3ac205b464552356699a4d2b0a7bf62497ead36e5bca73982e11ae5ee3810731", "question_id": "q01_single_fact", "split": "test", "cluster_id": "syn_03021_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "tokenizer_hashes": {"receiver": "b9d8872cf3a99da99e1fc4b0f71232a451632dc49ca4c6771549b4efa5c1c763"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 4, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1725, "query_tokens": 458, "message_tokens": 350, "message_utf8_bytes": 985, "cache_tensor_bytes": null, "serialized_payload_bytes": 985, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 20426.742333103903, "hops": {"1": 6896.219208021648, "2": 4294.3743750220165, "3": 4930.94241595827, "4": 4305.2063341019675}, "receiver_prefill_ms": 201.46641694009304, "ttft_ms": 11.554333032108843, "decode_ms": 11.557165998965502}, "total_wall_ms": 20651.710916077718, "ttft_ms": 11.554333032108843, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 19, "retention_rate": 0.2923076923076923, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.15384615384615385, "status": 1.0, "place": 0.0, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 985}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 19, "retention_rate": 0.2923076923076923, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.15384615384615385, "status": 1.0, "place": 0.0, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 836}, "2": {"facts_total": 65, "facts_retained": 19, "retention_rate": 0.2923076923076923, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.15384615384615385, "status": 1.0, "place": 0.0, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 985}, "3": {"facts_total": 65, "facts_retained": 19, "retention_rate": 0.2923076923076923, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.15384615384615385, "status": 1.0, "place": 0.0, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 985}, "4": {"facts_total": 65, "facts_retained": 19, "retention_rate": 0.2923076923076923, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.15384615384615385, "status": 1.0, "place": 0.0, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 985}}}, "reference_key_id": "keys/synthetic/syn_03021_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "dc8b132a30e8989cf3c15b7b1a161c9730440bcba369b175bdd47033bb2266fc", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "c37b86cc101392b4e8285445", "attempt_id": "a99efe5d33dd", "run_group_id": "E4_core:A5|closed|qwen3b-llama3b:h4", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:22:19.715598Z", "task_family": "synthetic", "document_id": "syn_03021_2000", "document_hash": "3ac205b464552356699a4d2b0a7bf62497ead36e5bca73982e11ae5ee3810731", "question_id": "q02_exact_copy", "split": "test", "cluster_id": "syn_03021_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "tokenizer_hashes": {"receiver": "b9d8872cf3a99da99e1fc4b0f71232a451632dc49ca4c6771549b4efa5c1c763"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 4, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1725, "query_tokens": 458, "message_tokens": 350, "message_utf8_bytes": 985, "cache_tensor_bytes": null, "serialized_payload_bytes": 985, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 20426.742333103903, "hops": {"1": 6896.219208021648, "2": 4294.3743750220165, "3": 4930.94241595827, "4": 4305.2063341019675}, "receiver_prefill_ms": 201.89791696611792, "ttft_ms": 11.592500028200448, "decode_ms": 11.51408301666379}, "total_wall_ms": 20652.13854110334, "ttft_ms": 11.592500028200448, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 19, "retention_rate": 0.2923076923076923, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.15384615384615385, "status": 1.0, "place": 0.0, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 985}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 19, "retention_rate": 0.2923076923076923, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.15384615384615385, "status": 1.0, "place": 0.0, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 836}, "2": {"facts_total": 65, "facts_retained": 19, "retention_rate": 0.2923076923076923, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.15384615384615385, "status": 1.0, "place": 0.0, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 985}, "3": {"facts_total": 65, "facts_retained": 19, "retention_rate": 0.2923076923076923, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.15384615384615385, "status": 1.0, "place": 0.0, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 985}, "4": {"facts_total": 65, "facts_retained": 19, "retention_rate": 0.2923076923076923, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.15384615384615385, "status": 1.0, "place": 0.0, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 985}}}, "reference_key_id": "keys/synthetic/syn_03021_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "dc8b132a30e8989cf3c15b7b1a161c9730440bcba369b175bdd47033bb2266fc", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "7c6a520be8ad0b768553c595", "attempt_id": "99eef91f6b57", "run_group_id": "E4_core:A5|closed|qwen3b-llama3b:h4", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:22:19.956900Z", "task_family": "synthetic", "document_id": "syn_03021_2000", "document_hash": "3ac205b464552356699a4d2b0a7bf62497ead36e5bca73982e11ae5ee3810731", "question_id": "q03_exact_copy", "split": "test", "cluster_id": "syn_03021_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "tokenizer_hashes": {"receiver": "b9d8872cf3a99da99e1fc4b0f71232a451632dc49ca4c6771549b4efa5c1c763"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 4, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1725, "query_tokens": 455, "message_tokens": 350, "message_utf8_bytes": 985, "cache_tensor_bytes": null, "serialized_payload_bytes": 985, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 20426.742333103903, "hops": {"1": 6896.219208021648, "2": 4294.3743750220165, "3": 4930.94241595827, "4": 4305.2063341019675}, "receiver_prefill_ms": 204.12816607858986, "ttft_ms": 11.556084034964442, "decode_ms": 11.61312498152256}, "total_wall_ms": 20654.435958131216, "ttft_ms": 11.556084034964442, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 19, "retention_rate": 0.2923076923076923, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.15384615384615385, "status": 1.0, "place": 0.0, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 985}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 19, "retention_rate": 0.2923076923076923, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.15384615384615385, "status": 1.0, "place": 0.0, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 836}, "2": {"facts_total": 65, "facts_retained": 19, "retention_rate": 0.2923076923076923, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.15384615384615385, "status": 1.0, "place": 0.0, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 985}, "3": {"facts_total": 65, "facts_retained": 19, "retention_rate": 0.2923076923076923, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.15384615384615385, "status": 1.0, "place": 0.0, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 985}, "4": {"facts_total": 65, "facts_retained": 19, "retention_rate": 0.2923076923076923, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.15384615384615385, "status": 1.0, "place": 0.0, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 985}}}, "reference_key_id": "keys/synthetic/syn_03021_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "dc8b132a30e8989cf3c15b7b1a161c9730440bcba369b175bdd47033bb2266fc", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "efe37abcbe89ef07627af361", "attempt_id": "e072f435244f", "run_group_id": "E4_core:A5|closed|qwen3b-llama3b:h4", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:22:20.199770Z", "task_family": "synthetic", "document_id": "syn_03021_2000", "document_hash": "3ac205b464552356699a4d2b0a7bf62497ead36e5bca73982e11ae5ee3810731", "question_id": "q04_multi_hop", "split": "test", "cluster_id": "syn_03021_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "tokenizer_hashes": {"receiver": "b9d8872cf3a99da99e1fc4b0f71232a451632dc49ca4c6771549b4efa5c1c763"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 4, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1725, "query_tokens": 462, "message_tokens": 350, "message_utf8_bytes": 985, "cache_tensor_bytes": null, "serialized_payload_bytes": 985, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 20426.742333103903, "hops": {"1": 6896.219208021648, "2": 4294.3743750220165, "3": 4930.94241595827, "4": 4305.2063341019675}, "receiver_prefill_ms": 202.47225009370595, "ttft_ms": 11.574000003747642, "decode_ms": 11.491082957945764}, "total_wall_ms": 20652.672000112943, "ttft_ms": 11.574000003747642, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 19, "retention_rate": 0.2923076923076923, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.15384615384615385, "status": 1.0, "place": 0.0, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 985}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 19, "retention_rate": 0.2923076923076923, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.15384615384615385, "status": 1.0, "place": 0.0, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 836}, "2": {"facts_total": 65, "facts_retained": 19, "retention_rate": 0.2923076923076923, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.15384615384615385, "status": 1.0, "place": 0.0, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 985}, "3": {"facts_total": 65, "facts_retained": 19, "retention_rate": 0.2923076923076923, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.15384615384615385, "status": 1.0, "place": 0.0, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 985}, "4": {"facts_total": 65, "facts_retained": 19, "retention_rate": 0.2923076923076923, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.15384615384615385, "status": 1.0, "place": 0.0, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 985}}}, "reference_key_id": "keys/synthetic/syn_03021_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "dc8b132a30e8989cf3c15b7b1a161c9730440bcba369b175bdd47033bb2266fc", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "107b07f4145dc353204941b7", "attempt_id": "d7e68286f56b", "run_group_id": "E4_core:A5|closed|qwen3b-llama3b:h4", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:22:20.440472Z", "task_family": "synthetic", "document_id": "syn_03021_2000", "document_hash": "3ac205b464552356699a4d2b0a7bf62497ead36e5bca73982e11ae5ee3810731", "question_id": "q05_multi_hop", "split": "test", "cluster_id": "syn_03021_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "tokenizer_hashes": {"receiver": "b9d8872cf3a99da99e1fc4b0f71232a451632dc49ca4c6771549b4efa5c1c763"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 4, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1725, "query_tokens": 461, "message_tokens": 350, "message_utf8_bytes": 985, "cache_tensor_bytes": null, "serialized_payload_bytes": 985, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 20426.742333103903, "hops": {"1": 6896.219208021648, "2": 4294.3743750220165, "3": 4930.94241595827, "4": 4305.2063341019675}, "receiver_prefill_ms": 201.78941695485264, "ttft_ms": 11.508582974784076, "decode_ms": 11.591374990530312}, "total_wall_ms": 20652.00604114216, "ttft_ms": 11.508582974784076, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 19, "retention_rate": 0.2923076923076923, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.15384615384615385, "status": 1.0, "place": 0.0, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 985}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 19, "retention_rate": 0.2923076923076923, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.15384615384615385, "status": 1.0, "place": 0.0, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 836}, "2": {"facts_total": 65, "facts_retained": 19, "retention_rate": 0.2923076923076923, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.15384615384615385, "status": 1.0, "place": 0.0, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 985}, "3": {"facts_total": 65, "facts_retained": 19, "retention_rate": 0.2923076923076923, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.15384615384615385, "status": 1.0, "place": 0.0, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 985}, "4": {"facts_total": 65, "facts_retained": 19, "retention_rate": 0.2923076923076923, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.15384615384615385, "status": 1.0, "place": 0.0, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 985}}}, "reference_key_id": "keys/synthetic/syn_03021_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "dc8b132a30e8989cf3c15b7b1a161c9730440bcba369b175bdd47033bb2266fc", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "4e18f2a58fbce166f190266a", "attempt_id": "a66e4550b950", "run_group_id": "E4_core:A5|closed|qwen3b-llama3b:h4", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:22:20.681374Z", "task_family": "synthetic", "document_id": "syn_03021_2000", "document_hash": "3ac205b464552356699a4d2b0a7bf62497ead36e5bca73982e11ae5ee3810731", "question_id": "q06_negation", "split": "test", "cluster_id": "syn_03021_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "tokenizer_hashes": {"receiver": "b9d8872cf3a99da99e1fc4b0f71232a451632dc49ca4c6771549b4efa5c1c763"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 4, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1725, "query_tokens": 453, "message_tokens": 350, "message_utf8_bytes": 985, "cache_tensor_bytes": null, "serialized_payload_bytes": 985, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 20426.742333103903, "hops": {"1": 6896.219208021648, "2": 4294.3743750220165, "3": 4930.94241595827, "4": 4305.2063341019675}, "receiver_prefill_ms": 201.94341707974672, "ttft_ms": 11.617708951234818, "decode_ms": 22.87674997933209}, "total_wall_ms": 20663.58983318787, "ttft_ms": 11.617708951234818, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 2}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "No.", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 19, "retention_rate": 0.2923076923076923, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.15384615384615385, "status": 1.0, "place": 0.0, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 985}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 19, "retention_rate": 0.2923076923076923, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.15384615384615385, "status": 1.0, "place": 0.0, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 836}, "2": {"facts_total": 65, "facts_retained": 19, "retention_rate": 0.2923076923076923, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.15384615384615385, "status": 1.0, "place": 0.0, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 985}, "3": {"facts_total": 65, "facts_retained": 19, "retention_rate": 0.2923076923076923, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.15384615384615385, "status": 1.0, "place": 0.0, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 985}, "4": {"facts_total": 65, "facts_retained": 19, "retention_rate": 0.2923076923076923, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.15384615384615385, "status": 1.0, "place": 0.0, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 985}}}, "reference_key_id": "keys/synthetic/syn_03021_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "dc8b132a30e8989cf3c15b7b1a161c9730440bcba369b175bdd47033bb2266fc", "output_hash": "38178a20b470cfd18299fb1593dd4ba706f1a83321dac051d01086efd8b7a96f"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "f2ce6f482e9f27ffae024381", "attempt_id": "dca34cf1804f", "run_group_id": "E4_core:A5|closed|qwen3b-llama3b:h4", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:22:20.932602Z", "task_family": "synthetic", "document_id": "syn_03021_2000", "document_hash": "3ac205b464552356699a4d2b0a7bf62497ead36e5bca73982e11ae5ee3810731", "question_id": "q07_negation", "split": "test", "cluster_id": "syn_03021_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "tokenizer_hashes": {"receiver": "b9d8872cf3a99da99e1fc4b0f71232a451632dc49ca4c6771549b4efa5c1c763"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 4, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1725, "query_tokens": 454, "message_tokens": 350, "message_utf8_bytes": 985, "cache_tensor_bytes": null, "serialized_payload_bytes": 985, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 20426.742333103903, "hops": {"1": 6896.219208021648, "2": 4294.3743750220165, "3": 4930.94241595827, "4": 4305.2063341019675}, "receiver_prefill_ms": 200.86958399042487, "ttft_ms": 11.52995799202472, "decode_ms": 11.586250038817525}, "total_wall_ms": 20651.094458065927, "ttft_ms": 11.52995799202472, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "No", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 19, "retention_rate": 0.2923076923076923, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.15384615384615385, "status": 1.0, "place": 0.0, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 985}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 19, "retention_rate": 0.2923076923076923, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.15384615384615385, "status": 1.0, "place": 0.0, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 836}, "2": {"facts_total": 65, "facts_retained": 19, "retention_rate": 0.2923076923076923, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.15384615384615385, "status": 1.0, "place": 0.0, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 985}, "3": {"facts_total": 65, "facts_retained": 19, "retention_rate": 0.2923076923076923, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.15384615384615385, "status": 1.0, "place": 0.0, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 985}, "4": {"facts_total": 65, "facts_retained": 19, "retention_rate": 0.2923076923076923, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.15384615384615385, "status": 1.0, "place": 0.0, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 985}}}, "reference_key_id": "keys/synthetic/syn_03021_2000.json", "task_scores": {"correct": 0.0, "abstained": 0.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "dc8b132a30e8989cf3c15b7b1a161c9730440bcba369b175bdd47033bb2266fc", "output_hash": "1ea442a134b2a184bd5d40104401f2a37fbc09ccf3f4bc9da161c6099be3691d"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "f4011ae97631851c3ee7e766", "attempt_id": "d466ff971558", "run_group_id": "E4_core:A5|closed|qwen3b-llama3b:h4", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:22:21.172480Z", "task_family": "synthetic", "document_id": "syn_03021_2000", "document_hash": "3ac205b464552356699a4d2b0a7bf62497ead36e5bca73982e11ae5ee3810731", "question_id": "q08_revision", "split": "test", "cluster_id": "syn_03021_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "tokenizer_hashes": {"receiver": "b9d8872cf3a99da99e1fc4b0f71232a451632dc49ca4c6771549b4efa5c1c763"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 4, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1725, "query_tokens": 470, "message_tokens": 350, "message_utf8_bytes": 985, "cache_tensor_bytes": null, "serialized_payload_bytes": 985, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 20426.742333103903, "hops": {"1": 6896.219208021648, "2": 4294.3743750220165, "3": 4930.94241595827, "4": 4305.2063341019675}, "receiver_prefill_ms": 204.5193329686299, "ttft_ms": 11.535957921296358, "decode_ms": 11.52787497267127}, "total_wall_ms": 20654.727374087088, "ttft_ms": 11.535957921296358, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 19, "retention_rate": 0.2923076923076923, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.15384615384615385, "status": 1.0, "place": 0.0, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 985}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 19, "retention_rate": 0.2923076923076923, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.15384615384615385, "status": 1.0, "place": 0.0, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 836}, "2": {"facts_total": 65, "facts_retained": 19, "retention_rate": 0.2923076923076923, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.15384615384615385, "status": 1.0, "place": 0.0, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 985}, "3": {"facts_total": 65, "facts_retained": 19, "retention_rate": 0.2923076923076923, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.15384615384615385, "status": 1.0, "place": 0.0, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 985}, "4": {"facts_total": 65, "facts_retained": 19, "retention_rate": 0.2923076923076923, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.15384615384615385, "status": 1.0, "place": 0.0, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 985}}}, "reference_key_id": "keys/synthetic/syn_03021_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "dc8b132a30e8989cf3c15b7b1a161c9730440bcba369b175bdd47033bb2266fc", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "ae4a65c8cac8fe92ce597634", "attempt_id": "ac2cc66452c4", "run_group_id": "E4_core:A5|closed|qwen3b-llama3b:h4", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:22:21.415617Z", "task_family": "synthetic", "document_id": "syn_03021_2000", "document_hash": "3ac205b464552356699a4d2b0a7bf62497ead36e5bca73982e11ae5ee3810731", "question_id": "q09_revision", "split": "test", "cluster_id": "syn_03021_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "tokenizer_hashes": {"receiver": "b9d8872cf3a99da99e1fc4b0f71232a451632dc49ca4c6771549b4efa5c1c763"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 4, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1725, "query_tokens": 461, "message_tokens": 350, "message_utf8_bytes": 985, "cache_tensor_bytes": null, "serialized_payload_bytes": 985, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 20426.742333103903, "hops": {"1": 6896.219208021648, "2": 4294.3743750220165, "3": 4930.94241595827, "4": 4305.2063341019675}, "receiver_prefill_ms": 203.30787496641278, "ttft_ms": 11.5481250686571, "decode_ms": 11.503709014505148}, "total_wall_ms": 20653.487000148743, "ttft_ms": 11.5481250686571, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 19, "retention_rate": 0.2923076923076923, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.15384615384615385, "status": 1.0, "place": 0.0, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 985}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 19, "retention_rate": 0.2923076923076923, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.15384615384615385, "status": 1.0, "place": 0.0, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 836}, "2": {"facts_total": 65, "facts_retained": 19, "retention_rate": 0.2923076923076923, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.15384615384615385, "status": 1.0, "place": 0.0, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 985}, "3": {"facts_total": 65, "facts_retained": 19, "retention_rate": 0.2923076923076923, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.15384615384615385, "status": 1.0, "place": 0.0, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 985}, "4": {"facts_total": 65, "facts_retained": 19, "retention_rate": 0.2923076923076923, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.15384615384615385, "status": 1.0, "place": 0.0, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 985}}}, "reference_key_id": "keys/synthetic/syn_03021_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "dc8b132a30e8989cf3c15b7b1a161c9730440bcba369b175bdd47033bb2266fc", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "b171b72fc315c9e512c9c1ab", "attempt_id": "df8d91e4625f", "run_group_id": "E4_core:A5|closed|qwen3b-llama3b:h4", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:22:21.657764Z", "task_family": "synthetic", "document_id": "syn_03021_2000", "document_hash": "3ac205b464552356699a4d2b0a7bf62497ead36e5bca73982e11ae5ee3810731", "question_id": "q10_temporal", "split": "test", "cluster_id": "syn_03021_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "tokenizer_hashes": {"receiver": "b9d8872cf3a99da99e1fc4b0f71232a451632dc49ca4c6771549b4efa5c1c763"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 4, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1725, "query_tokens": 469, "message_tokens": 350, "message_utf8_bytes": 985, "cache_tensor_bytes": null, "serialized_payload_bytes": 985, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 20426.742333103903, "hops": {"1": 6896.219208021648, "2": 4294.3743750220165, "3": 4930.94241595827, "4": 4305.2063341019675}, "receiver_prefill_ms": 208.12520803883672, "ttft_ms": 11.61370798945427, "decode_ms": 11.469958000816405}, "total_wall_ms": 20658.339500194415, "ttft_ms": 11.61370798945427, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 19, "retention_rate": 0.2923076923076923, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.15384615384615385, "status": 1.0, "place": 0.0, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 985}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 19, "retention_rate": 0.2923076923076923, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.15384615384615385, "status": 1.0, "place": 0.0, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 836}, "2": {"facts_total": 65, "facts_retained": 19, "retention_rate": 0.2923076923076923, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.15384615384615385, "status": 1.0, "place": 0.0, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 985}, "3": {"facts_total": 65, "facts_retained": 19, "retention_rate": 0.2923076923076923, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.15384615384615385, "status": 1.0, "place": 0.0, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 985}, "4": {"facts_total": 65, "facts_retained": 19, "retention_rate": 0.2923076923076923, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.15384615384615385, "status": 1.0, "place": 0.0, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 985}}}, "reference_key_id": "keys/synthetic/syn_03021_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "dc8b132a30e8989cf3c15b7b1a161c9730440bcba369b175bdd47033bb2266fc", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "acc849b3d08b2c5057915d7c", "attempt_id": "43f3de5bba98", "run_group_id": "E4_core:A5|closed|qwen3b-llama3b:h4", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:22:21.904469Z", "task_family": "synthetic", "document_id": "syn_03021_2000", "document_hash": "3ac205b464552356699a4d2b0a7bf62497ead36e5bca73982e11ae5ee3810731", "question_id": "q11_aggregation", "split": "test", "cluster_id": "syn_03021_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "tokenizer_hashes": {"receiver": "b9d8872cf3a99da99e1fc4b0f71232a451632dc49ca4c6771549b4efa5c1c763"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 4, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1725, "query_tokens": 465, "message_tokens": 350, "message_utf8_bytes": 985, "cache_tensor_bytes": null, "serialized_payload_bytes": 985, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 20426.742333103903, "hops": {"1": 6896.219208021648, "2": 4294.3743750220165, "3": 4930.94241595827, "4": 4305.2063341019675}, "receiver_prefill_ms": 204.23579204361886, "ttft_ms": 11.56979100778699, "decode_ms": 57.78304091654718}, "total_wall_ms": 20700.720541062765, "ttft_ms": 11.56979100778699, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 5}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "5,646,720", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 19, "retention_rate": 0.2923076923076923, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.15384615384615385, "status": 1.0, "place": 0.0, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 985}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 19, "retention_rate": 0.2923076923076923, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.15384615384615385, "status": 1.0, "place": 0.0, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 836}, "2": {"facts_total": 65, "facts_retained": 19, "retention_rate": 0.2923076923076923, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.15384615384615385, "status": 1.0, "place": 0.0, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 985}, "3": {"facts_total": 65, "facts_retained": 19, "retention_rate": 0.2923076923076923, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.15384615384615385, "status": 1.0, "place": 0.0, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 985}, "4": {"facts_total": 65, "facts_retained": 19, "retention_rate": 0.2923076923076923, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.15384615384615385, "status": 1.0, "place": 0.0, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 985}}}, "reference_key_id": "keys/synthetic/syn_03021_2000.json", "task_scores": {"correct": 0.0, "abstained": 0.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "dc8b132a30e8989cf3c15b7b1a161c9730440bcba369b175bdd47033bb2266fc", "output_hash": "2c0bb14ad8ba61e6d23fb3dea5205f2f02c43c95a0cf8e7f93c240ddea626ad4"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "80b55c21aa769e11f06232e2", "attempt_id": "806784074db2", "run_group_id": "E4_core:A5|closed|qwen3b-llama3b:h4", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:22:22.193386Z", "task_family": "synthetic", "document_id": "syn_03021_2000", "document_hash": "3ac205b464552356699a4d2b0a7bf62497ead36e5bca73982e11ae5ee3810731", "question_id": "q12_unit_conversion", "split": "test", "cluster_id": "syn_03021_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "tokenizer_hashes": {"receiver": "b9d8872cf3a99da99e1fc4b0f71232a451632dc49ca4c6771549b4efa5c1c763"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 4, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1725, "query_tokens": 460, "message_tokens": 350, "message_utf8_bytes": 985, "cache_tensor_bytes": null, "serialized_payload_bytes": 985, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 20426.742333103903, "hops": {"1": 6896.219208021648, "2": 4294.3743750220165, "3": 4930.94241595827, "4": 4305.2063341019675}, "receiver_prefill_ms": 203.2241669949144, "ttft_ms": 11.535875033587217, "decode_ms": 11.553250020369887}, "total_wall_ms": 20653.444375027902, "ttft_ms": 11.535875033587217, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 19, "retention_rate": 0.2923076923076923, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.15384615384615385, "status": 1.0, "place": 0.0, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 985}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 19, "retention_rate": 0.2923076923076923, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.15384615384615385, "status": 1.0, "place": 0.0, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 836}, "2": {"facts_total": 65, "facts_retained": 19, "retention_rate": 0.2923076923076923, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.15384615384615385, "status": 1.0, "place": 0.0, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 985}, "3": {"facts_total": 65, "facts_retained": 19, "retention_rate": 0.2923076923076923, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.15384615384615385, "status": 1.0, "place": 0.0, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 985}, "4": {"facts_total": 65, "facts_retained": 19, "retention_rate": 0.2923076923076923, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.15384615384615385, "status": 1.0, "place": 0.0, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 985}}}, "reference_key_id": "keys/synthetic/syn_03021_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "dc8b132a30e8989cf3c15b7b1a161c9730440bcba369b175bdd47033bb2266fc", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "2045fa722f973b166629ff76", "attempt_id": "5194ab2ee3c4", "run_group_id": "E4_core:A5|closed|qwen3b-llama3b:h4", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:22:22.435091Z", "task_family": "synthetic", "document_id": "syn_03021_2000", "document_hash": "3ac205b464552356699a4d2b0a7bf62497ead36e5bca73982e11ae5ee3810731", "question_id": "q13_unanswerable", "split": "test", "cluster_id": "syn_03021_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "tokenizer_hashes": {"receiver": "b9d8872cf3a99da99e1fc4b0f71232a451632dc49ca4c6771549b4efa5c1c763"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 4, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1725, "query_tokens": 464, "message_tokens": 350, "message_utf8_bytes": 985, "cache_tensor_bytes": null, "serialized_payload_bytes": 985, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 20426.742333103903, "hops": {"1": 6896.219208021648, "2": 4294.3743750220165, "3": 4930.94241595827, "4": 4305.2063341019675}, "receiver_prefill_ms": 204.76945908740163, "ttft_ms": 11.539041996002197, "decode_ms": 11.492499965243042}, "total_wall_ms": 20654.90045817569, "ttft_ms": 11.539041996002197, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 19, "retention_rate": 0.2923076923076923, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.15384615384615385, "status": 1.0, "place": 0.0, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 985}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 19, "retention_rate": 0.2923076923076923, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.15384615384615385, "status": 1.0, "place": 0.0, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 836}, "2": {"facts_total": 65, "facts_retained": 19, "retention_rate": 0.2923076923076923, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.15384615384615385, "status": 1.0, "place": 0.0, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 985}, "3": {"facts_total": 65, "facts_retained": 19, "retention_rate": 0.2923076923076923, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.15384615384615385, "status": 1.0, "place": 0.0, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 985}, "4": {"facts_total": 65, "facts_retained": 19, "retention_rate": 0.2923076923076923, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.15384615384615385, "status": 1.0, "place": 0.0, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 985}}}, "reference_key_id": "keys/synthetic/syn_03021_2000.json", "task_scores": {"correct": 1.0, "abstained": 1.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "dc8b132a30e8989cf3c15b7b1a161c9730440bcba369b175bdd47033bb2266fc", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "277cd0ddb1dc0228579c6f0b", "attempt_id": "5e894ac62eb3", "run_group_id": "E4_core:A5|closed|qwen3b-llama3b:h4", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:22:22.678642Z", "task_family": "synthetic", "document_id": "syn_03021_2000", "document_hash": "3ac205b464552356699a4d2b0a7bf62497ead36e5bca73982e11ae5ee3810731", "question_id": "q14_unanswerable", "split": "test", "cluster_id": "syn_03021_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "tokenizer_hashes": {"receiver": "b9d8872cf3a99da99e1fc4b0f71232a451632dc49ca4c6771549b4efa5c1c763"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 4, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1725, "query_tokens": 467, "message_tokens": 350, "message_utf8_bytes": 985, "cache_tensor_bytes": null, "serialized_payload_bytes": 985, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 20426.742333103903, "hops": {"1": 6896.219208021648, "2": 4294.3743750220165, "3": 4930.94241595827, "4": 4305.2063341019675}, "receiver_prefill_ms": 204.35237500350922, "ttft_ms": 11.583790997974575, "decode_ms": 11.511042015627027}, "total_wall_ms": 20654.59216607269, "ttft_ms": 11.583790997974575, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 19, "retention_rate": 0.2923076923076923, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.15384615384615385, "status": 1.0, "place": 0.0, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 985}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 19, "retention_rate": 0.2923076923076923, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.15384615384615385, "status": 1.0, "place": 0.0, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 836}, "2": {"facts_total": 65, "facts_retained": 19, "retention_rate": 0.2923076923076923, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.15384615384615385, "status": 1.0, "place": 0.0, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 985}, "3": {"facts_total": 65, "facts_retained": 19, "retention_rate": 0.2923076923076923, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.15384615384615385, "status": 1.0, "place": 0.0, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 985}, "4": {"facts_total": 65, "facts_retained": 19, "retention_rate": 0.2923076923076923, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.15384615384615385, "status": 1.0, "place": 0.0, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 985}}}, "reference_key_id": "keys/synthetic/syn_03021_2000.json", "task_scores": {"correct": 1.0, "abstained": 1.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "dc8b132a30e8989cf3c15b7b1a161c9730440bcba369b175bdd47033bb2266fc", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "8684cca1bde15a7bf65cb552", "attempt_id": "66022a20a14e", "run_group_id": "E4_core:A5|closed|qwen3b-llama3b:h8", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:22:41.408286Z", "task_family": "synthetic", "document_id": "syn_03021_2000", "document_hash": "3ac205b464552356699a4d2b0a7bf62497ead36e5bca73982e11ae5ee3810731", "question_id": "q00_single_fact", "split": "test", "cluster_id": "syn_03021_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "tokenizer_hashes": {"receiver": "b9d8872cf3a99da99e1fc4b0f71232a451632dc49ca4c6771549b4efa5c1c763"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 8, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1725, "query_tokens": 452, "message_tokens": 350, "message_utf8_bytes": 985, "cache_tensor_bytes": null, "serialized_payload_bytes": 985, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 38908.40608323924, "hops": {"1": 6896.219208021648, "2": 4294.3743750220165, "3": 4930.94241595827, "4": 4305.2063341019675, "5": 4931.834749993868, "6": 4301.946583087556, "7": 4939.746499992907, "8": 4308.1359170610085}, "receiver_prefill_ms": 200.94579097349197, "ttft_ms": 11.417291942052543, "decode_ms": 11.446291930042207}, "total_wall_ms": 39132.63620832004, "ttft_ms": 11.417291942052543, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 19, "retention_rate": 0.2923076923076923, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.15384615384615385, "status": 1.0, "place": 0.0, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 985}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 19, "retention_rate": 0.2923076923076923, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.15384615384615385, "status": 1.0, "place": 0.0, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 836}, "2": {"facts_total": 65, "facts_retained": 19, "retention_rate": 0.2923076923076923, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.15384615384615385, "status": 1.0, "place": 0.0, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 985}, "3": {"facts_total": 65, "facts_retained": 19, "retention_rate": 0.2923076923076923, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.15384615384615385, "status": 1.0, "place": 0.0, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 985}, "4": {"facts_total": 65, "facts_retained": 19, "retention_rate": 0.2923076923076923, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.15384615384615385, "status": 1.0, "place": 0.0, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 985}, "5": {"facts_total": 65, "facts_retained": 19, "retention_rate": 0.2923076923076923, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.15384615384615385, "status": 1.0, "place": 0.0, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 985}, "6": {"facts_total": 65, "facts_retained": 19, "retention_rate": 0.2923076923076923, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.15384615384615385, "status": 1.0, "place": 0.0, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 985}, "7": {"facts_total": 65, "facts_retained": 19, "retention_rate": 0.2923076923076923, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.15384615384615385, "status": 1.0, "place": 0.0, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 985}, "8": {"facts_total": 65, "facts_retained": 19, "retention_rate": 0.2923076923076923, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.15384615384615385, "status": 1.0, "place": 0.0, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 985}}}, "reference_key_id": "keys/synthetic/syn_03021_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "dc8b132a30e8989cf3c15b7b1a161c9730440bcba369b175bdd47033bb2266fc", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "5168a70642e8a5690249bd9b", "attempt_id": "f0c41f519d89", "run_group_id": "E4_core:A5|closed|qwen3b-llama3b:h8", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:22:41.648717Z", "task_family": "synthetic", "document_id": "syn_03021_2000", "document_hash": "3ac205b464552356699a4d2b0a7bf62497ead36e5bca73982e11ae5ee3810731", "question_id": "q01_single_fact", "split": "test", "cluster_id": "syn_03021_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "tokenizer_hashes": {"receiver": "b9d8872cf3a99da99e1fc4b0f71232a451632dc49ca4c6771549b4efa5c1c763"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 8, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1725, "query_tokens": 458, "message_tokens": 350, "message_utf8_bytes": 985, "cache_tensor_bytes": null, "serialized_payload_bytes": 985, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 38908.40608323924, "hops": {"1": 6896.219208021648, "2": 4294.3743750220165, "3": 4930.94241595827, "4": 4305.2063341019675, "5": 4931.834749993868, "6": 4301.946583087556, "7": 4939.746499992907, "8": 4308.1359170610085}, "receiver_prefill_ms": 200.74433297850192, "ttft_ms": 11.597416014410555, "decode_ms": 11.473416001535952}, "total_wall_ms": 39132.653500302695, "ttft_ms": 11.597416014410555, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 19, "retention_rate": 0.2923076923076923, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.15384615384615385, "status": 1.0, "place": 0.0, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 985}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 19, "retention_rate": 0.2923076923076923, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.15384615384615385, "status": 1.0, "place": 0.0, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 836}, "2": {"facts_total": 65, "facts_retained": 19, "retention_rate": 0.2923076923076923, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.15384615384615385, "status": 1.0, "place": 0.0, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 985}, "3": {"facts_total": 65, "facts_retained": 19, "retention_rate": 0.2923076923076923, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.15384615384615385, "status": 1.0, "place": 0.0, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 985}, "4": {"facts_total": 65, "facts_retained": 19, "retention_rate": 0.2923076923076923, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.15384615384615385, "status": 1.0, "place": 0.0, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 985}, "5": {"facts_total": 65, "facts_retained": 19, "retention_rate": 0.2923076923076923, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.15384615384615385, "status": 1.0, "place": 0.0, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 985}, "6": {"facts_total": 65, "facts_retained": 19, "retention_rate": 0.2923076923076923, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.15384615384615385, "status": 1.0, "place": 0.0, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 985}, "7": {"facts_total": 65, "facts_retained": 19, "retention_rate": 0.2923076923076923, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.15384615384615385, "status": 1.0, "place": 0.0, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 985}, "8": {"facts_total": 65, "facts_retained": 19, "retention_rate": 0.2923076923076923, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.15384615384615385, "status": 1.0, "place": 0.0, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 985}}}, "reference_key_id": "keys/synthetic/syn_03021_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "dc8b132a30e8989cf3c15b7b1a161c9730440bcba369b175bdd47033bb2266fc", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "b0c25fc019d195294f129824", "attempt_id": "8b740728e3ce", "run_group_id": "E4_core:A5|closed|qwen3b-llama3b:h8", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:22:41.888559Z", "task_family": "synthetic", "document_id": "syn_03021_2000", "document_hash": "3ac205b464552356699a4d2b0a7bf62497ead36e5bca73982e11ae5ee3810731", "question_id": "q02_exact_copy", "split": "test", "cluster_id": "syn_03021_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "tokenizer_hashes": {"receiver": "b9d8872cf3a99da99e1fc4b0f71232a451632dc49ca4c6771549b4efa5c1c763"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 8, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1725, "query_tokens": 458, "message_tokens": 350, "message_utf8_bytes": 985, "cache_tensor_bytes": null, "serialized_payload_bytes": 985, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 38908.40608323924, "hops": {"1": 6896.219208021648, "2": 4294.3743750220165, "3": 4930.94241595827, "4": 4305.2063341019675, "5": 4931.834749993868, "6": 4301.946583087556, "7": 4939.746499992907, "8": 4308.1359170610085}, "receiver_prefill_ms": 202.5256670312956, "ttft_ms": 11.530166957527399, "decode_ms": 11.525416048243642}, "total_wall_ms": 39134.37458325643, "ttft_ms": 11.530166957527399, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 19, "retention_rate": 0.2923076923076923, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.15384615384615385, "status": 1.0, "place": 0.0, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 985}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 19, "retention_rate": 0.2923076923076923, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.15384615384615385, "status": 1.0, "place": 0.0, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 836}, "2": {"facts_total": 65, "facts_retained": 19, "retention_rate": 0.2923076923076923, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.15384615384615385, "status": 1.0, "place": 0.0, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 985}, "3": {"facts_total": 65, "facts_retained": 19, "retention_rate": 0.2923076923076923, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.15384615384615385, "status": 1.0, "place": 0.0, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 985}, "4": {"facts_total": 65, "facts_retained": 19, "retention_rate": 0.2923076923076923, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.15384615384615385, "status": 1.0, "place": 0.0, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 985}, "5": {"facts_total": 65, "facts_retained": 19, "retention_rate": 0.2923076923076923, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.15384615384615385, "status": 1.0, "place": 0.0, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 985}, "6": {"facts_total": 65, "facts_retained": 19, "retention_rate": 0.2923076923076923, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.15384615384615385, "status": 1.0, "place": 0.0, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 985}, "7": {"facts_total": 65, "facts_retained": 19, "retention_rate": 0.2923076923076923, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.15384615384615385, "status": 1.0, "place": 0.0, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 985}, "8": {"facts_total": 65, "facts_retained": 19, "retention_rate": 0.2923076923076923, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.15384615384615385, "status": 1.0, "place": 0.0, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 985}}}, "reference_key_id": "keys/synthetic/syn_03021_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "dc8b132a30e8989cf3c15b7b1a161c9730440bcba369b175bdd47033bb2266fc", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "0753c06f450b0678c6344cba", "attempt_id": "c09165b2ab24", "run_group_id": "E4_core:A5|closed|qwen3b-llama3b:h8", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:22:42.129945Z", "task_family": "synthetic", "document_id": "syn_03021_2000", "document_hash": "3ac205b464552356699a4d2b0a7bf62497ead36e5bca73982e11ae5ee3810731", "question_id": "q03_exact_copy", "split": "test", "cluster_id": "syn_03021_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "tokenizer_hashes": {"receiver": "b9d8872cf3a99da99e1fc4b0f71232a451632dc49ca4c6771549b4efa5c1c763"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 8, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1725, "query_tokens": 455, "message_tokens": 350, "message_utf8_bytes": 985, "cache_tensor_bytes": null, "serialized_payload_bytes": 985, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 38908.40608323924, "hops": {"1": 6896.219208021648, "2": 4294.3743750220165, "3": 4930.94241595827, "4": 4305.2063341019675, "5": 4931.834749993868, "6": 4301.946583087556, "7": 4939.746499992907, "8": 4308.1359170610085}, "receiver_prefill_ms": 203.65537505131215, "ttft_ms": 11.6252499865368, "decode_ms": 11.536624981090426}, "total_wall_ms": 39135.623958311044, "ttft_ms": 11.6252499865368, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 19, "retention_rate": 0.2923076923076923, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.15384615384615385, "status": 1.0, "place": 0.0, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 985}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 19, "retention_rate": 0.2923076923076923, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.15384615384615385, "status": 1.0, "place": 0.0, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 836}, "2": {"facts_total": 65, "facts_retained": 19, "retention_rate": 0.2923076923076923, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.15384615384615385, "status": 1.0, "place": 0.0, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 985}, "3": {"facts_total": 65, "facts_retained": 19, "retention_rate": 0.2923076923076923, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.15384615384615385, "status": 1.0, "place": 0.0, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 985}, "4": {"facts_total": 65, "facts_retained": 19, "retention_rate": 0.2923076923076923, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.15384615384615385, "status": 1.0, "place": 0.0, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 985}, "5": {"facts_total": 65, "facts_retained": 19, "retention_rate": 0.2923076923076923, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.15384615384615385, "status": 1.0, "place": 0.0, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 985}, "6": {"facts_total": 65, "facts_retained": 19, "retention_rate": 0.2923076923076923, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.15384615384615385, "status": 1.0, "place": 0.0, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 985}, "7": {"facts_total": 65, "facts_retained": 19, "retention_rate": 0.2923076923076923, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.15384615384615385, "status": 1.0, "place": 0.0, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 985}, "8": {"facts_total": 65, "facts_retained": 19, "retention_rate": 0.2923076923076923, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.15384615384615385, "status": 1.0, "place": 0.0, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 985}}}, "reference_key_id": "keys/synthetic/syn_03021_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "dc8b132a30e8989cf3c15b7b1a161c9730440bcba369b175bdd47033bb2266fc", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "7658de8168a54706ea455ea9", "attempt_id": "ed79af2a252d", "run_group_id": "E4_core:A5|closed|qwen3b-llama3b:h8", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:22:42.372543Z", "task_family": "synthetic", "document_id": "syn_03021_2000", "document_hash": "3ac205b464552356699a4d2b0a7bf62497ead36e5bca73982e11ae5ee3810731", "question_id": "q04_multi_hop", "split": "test", "cluster_id": "syn_03021_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "tokenizer_hashes": {"receiver": "b9d8872cf3a99da99e1fc4b0f71232a451632dc49ca4c6771549b4efa5c1c763"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 8, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1725, "query_tokens": 462, "message_tokens": 350, "message_utf8_bytes": 985, "cache_tensor_bytes": null, "serialized_payload_bytes": 985, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 38908.40608323924, "hops": {"1": 6896.219208021648, "2": 4294.3743750220165, "3": 4930.94241595827, "4": 4305.2063341019675, "5": 4931.834749993868, "6": 4301.946583087556, "7": 4939.746499992907, "8": 4308.1359170610085}, "receiver_prefill_ms": 200.44012495782226, "ttft_ms": 11.534958030097187, "decode_ms": 11.554750031791627}, "total_wall_ms": 39132.342500262894, "ttft_ms": 11.534958030097187, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 19, "retention_rate": 0.2923076923076923, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.15384615384615385, "status": 1.0, "place": 0.0, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 985}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 19, "retention_rate": 0.2923076923076923, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.15384615384615385, "status": 1.0, "place": 0.0, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 836}, "2": {"facts_total": 65, "facts_retained": 19, "retention_rate": 0.2923076923076923, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.15384615384615385, "status": 1.0, "place": 0.0, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 985}, "3": {"facts_total": 65, "facts_retained": 19, "retention_rate": 0.2923076923076923, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.15384615384615385, "status": 1.0, "place": 0.0, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 985}, "4": {"facts_total": 65, "facts_retained": 19, "retention_rate": 0.2923076923076923, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.15384615384615385, "status": 1.0, "place": 0.0, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 985}, "5": {"facts_total": 65, "facts_retained": 19, "retention_rate": 0.2923076923076923, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.15384615384615385, "status": 1.0, "place": 0.0, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 985}, "6": {"facts_total": 65, "facts_retained": 19, "retention_rate": 0.2923076923076923, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.15384615384615385, "status": 1.0, "place": 0.0, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 985}, "7": {"facts_total": 65, "facts_retained": 19, "retention_rate": 0.2923076923076923, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.15384615384615385, "status": 1.0, "place": 0.0, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 985}, "8": {"facts_total": 65, "facts_retained": 19, "retention_rate": 0.2923076923076923, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.15384615384615385, "status": 1.0, "place": 0.0, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 985}}}, "reference_key_id": "keys/synthetic/syn_03021_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "dc8b132a30e8989cf3c15b7b1a161c9730440bcba369b175bdd47033bb2266fc", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "86558a65a115b2723d2ca5d1", "attempt_id": "584f714fa3cd", "run_group_id": "E4_core:A5|closed|qwen3b-llama3b:h8", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:22:42.611746Z", "task_family": "synthetic", "document_id": "syn_03021_2000", "document_hash": "3ac205b464552356699a4d2b0a7bf62497ead36e5bca73982e11ae5ee3810731", "question_id": "q05_multi_hop", "split": "test", "cluster_id": "syn_03021_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "tokenizer_hashes": {"receiver": "b9d8872cf3a99da99e1fc4b0f71232a451632dc49ca4c6771549b4efa5c1c763"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 8, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1725, "query_tokens": 461, "message_tokens": 350, "message_utf8_bytes": 985, "cache_tensor_bytes": null, "serialized_payload_bytes": 985, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 38908.40608323924, "hops": {"1": 6896.219208021648, "2": 4294.3743750220165, "3": 4930.94241595827, "4": 4305.2063341019675, "5": 4931.834749993868, "6": 4301.946583087556, "7": 4939.746499992907, "8": 4308.1359170610085}, "receiver_prefill_ms": 203.5017079906538, "ttft_ms": 11.600124998949468, "decode_ms": 11.545417015440762}, "total_wall_ms": 39135.43791731354, "ttft_ms": 11.600124998949468, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 19, "retention_rate": 0.2923076923076923, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.15384615384615385, "status": 1.0, "place": 0.0, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 985}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 19, "retention_rate": 0.2923076923076923, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.15384615384615385, "status": 1.0, "place": 0.0, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 836}, "2": {"facts_total": 65, "facts_retained": 19, "retention_rate": 0.2923076923076923, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.15384615384615385, "status": 1.0, "place": 0.0, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 985}, "3": {"facts_total": 65, "facts_retained": 19, "retention_rate": 0.2923076923076923, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.15384615384615385, "status": 1.0, "place": 0.0, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 985}, "4": {"facts_total": 65, "facts_retained": 19, "retention_rate": 0.2923076923076923, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.15384615384615385, "status": 1.0, "place": 0.0, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 985}, "5": {"facts_total": 65, "facts_retained": 19, "retention_rate": 0.2923076923076923, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.15384615384615385, "status": 1.0, "place": 0.0, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 985}, "6": {"facts_total": 65, "facts_retained": 19, "retention_rate": 0.2923076923076923, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.15384615384615385, "status": 1.0, "place": 0.0, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 985}, "7": {"facts_total": 65, "facts_retained": 19, "retention_rate": 0.2923076923076923, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.15384615384615385, "status": 1.0, "place": 0.0, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 985}, "8": {"facts_total": 65, "facts_retained": 19, "retention_rate": 0.2923076923076923, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.15384615384615385, "status": 1.0, "place": 0.0, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 985}}}, "reference_key_id": "keys/synthetic/syn_03021_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "dc8b132a30e8989cf3c15b7b1a161c9730440bcba369b175bdd47033bb2266fc", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "4705add754afca10dd943642", "attempt_id": "109062cabb54", "run_group_id": "E4_core:A5|closed|qwen3b-llama3b:h8", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:22:42.854345Z", "task_family": "synthetic", "document_id": "syn_03021_2000", "document_hash": "3ac205b464552356699a4d2b0a7bf62497ead36e5bca73982e11ae5ee3810731", "question_id": "q06_negation", "split": "test", "cluster_id": "syn_03021_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "tokenizer_hashes": {"receiver": "b9d8872cf3a99da99e1fc4b0f71232a451632dc49ca4c6771549b4efa5c1c763"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 8, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1725, "query_tokens": 453, "message_tokens": 350, "message_utf8_bytes": 985, "cache_tensor_bytes": null, "serialized_payload_bytes": 985, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 38908.40608323924, "hops": {"1": 6896.219208021648, "2": 4294.3743750220165, "3": 4930.94241595827, "4": 4305.2063341019675, "5": 4931.834749993868, "6": 4301.946583087556, "7": 4939.746499992907, "8": 4308.1359170610085}, "receiver_prefill_ms": 202.36729201860726, "ttft_ms": 11.570916045457125, "decode_ms": 23.007625015452504}, "total_wall_ms": 39145.730374148116, "ttft_ms": 11.570916045457125, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 2}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "No.", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 19, "retention_rate": 0.2923076923076923, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.15384615384615385, "status": 1.0, "place": 0.0, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 985}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 19, "retention_rate": 0.2923076923076923, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.15384615384615385, "status": 1.0, "place": 0.0, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 836}, "2": {"facts_total": 65, "facts_retained": 19, "retention_rate": 0.2923076923076923, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.15384615384615385, "status": 1.0, "place": 0.0, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 985}, "3": {"facts_total": 65, "facts_retained": 19, "retention_rate": 0.2923076923076923, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.15384615384615385, "status": 1.0, "place": 0.0, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 985}, "4": {"facts_total": 65, "facts_retained": 19, "retention_rate": 0.2923076923076923, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.15384615384615385, "status": 1.0, "place": 0.0, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 985}, "5": {"facts_total": 65, "facts_retained": 19, "retention_rate": 0.2923076923076923, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.15384615384615385, "status": 1.0, "place": 0.0, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 985}, "6": {"facts_total": 65, "facts_retained": 19, "retention_rate": 0.2923076923076923, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.15384615384615385, "status": 1.0, "place": 0.0, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 985}, "7": {"facts_total": 65, "facts_retained": 19, "retention_rate": 0.2923076923076923, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.15384615384615385, "status": 1.0, "place": 0.0, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 985}, "8": {"facts_total": 65, "facts_retained": 19, "retention_rate": 0.2923076923076923, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.15384615384615385, "status": 1.0, "place": 0.0, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 985}}}, "reference_key_id": "keys/synthetic/syn_03021_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "dc8b132a30e8989cf3c15b7b1a161c9730440bcba369b175bdd47033bb2266fc", "output_hash": "38178a20b470cfd18299fb1593dd4ba706f1a83321dac051d01086efd8b7a96f"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "29acfc567719fb94547bab5c", "attempt_id": "4e95b5ef1d66", "run_group_id": "E4_core:A5|closed|qwen3b-llama3b:h8", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:22:43.106889Z", "task_family": "synthetic", "document_id": "syn_03021_2000", "document_hash": "3ac205b464552356699a4d2b0a7bf62497ead36e5bca73982e11ae5ee3810731", "question_id": "q07_negation", "split": "test", "cluster_id": "syn_03021_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "tokenizer_hashes": {"receiver": "b9d8872cf3a99da99e1fc4b0f71232a451632dc49ca4c6771549b4efa5c1c763"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 8, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1725, "query_tokens": 454, "message_tokens": 350, "message_utf8_bytes": 985, "cache_tensor_bytes": null, "serialized_payload_bytes": 985, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 38908.40608323924, "hops": {"1": 6896.219208021648, "2": 4294.3743750220165, "3": 4930.94241595827, "4": 4305.2063341019675, "5": 4931.834749993868, "6": 4301.946583087556, "7": 4939.746499992907, "8": 4308.1359170610085}, "receiver_prefill_ms": 202.103832969442, "ttft_ms": 11.636374983936548, "decode_ms": 11.507041985169053}, "total_wall_ms": 39134.05295822304, "ttft_ms": 11.636374983936548, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "No", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 19, "retention_rate": 0.2923076923076923, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.15384615384615385, "status": 1.0, "place": 0.0, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 985}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 19, "retention_rate": 0.2923076923076923, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.15384615384615385, "status": 1.0, "place": 0.0, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 836}, "2": {"facts_total": 65, "facts_retained": 19, "retention_rate": 0.2923076923076923, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.15384615384615385, "status": 1.0, "place": 0.0, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 985}, "3": {"facts_total": 65, "facts_retained": 19, "retention_rate": 0.2923076923076923, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.15384615384615385, "status": 1.0, "place": 0.0, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 985}, "4": {"facts_total": 65, "facts_retained": 19, "retention_rate": 0.2923076923076923, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.15384615384615385, "status": 1.0, "place": 0.0, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 985}, "5": {"facts_total": 65, "facts_retained": 19, "retention_rate": 0.2923076923076923, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.15384615384615385, "status": 1.0, "place": 0.0, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 985}, "6": {"facts_total": 65, "facts_retained": 19, "retention_rate": 0.2923076923076923, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.15384615384615385, "status": 1.0, "place": 0.0, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 985}, "7": {"facts_total": 65, "facts_retained": 19, "retention_rate": 0.2923076923076923, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.15384615384615385, "status": 1.0, "place": 0.0, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 985}, "8": {"facts_total": 65, "facts_retained": 19, "retention_rate": 0.2923076923076923, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.15384615384615385, "status": 1.0, "place": 0.0, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 985}}}, "reference_key_id": "keys/synthetic/syn_03021_2000.json", "task_scores": {"correct": 0.0, "abstained": 0.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "dc8b132a30e8989cf3c15b7b1a161c9730440bcba369b175bdd47033bb2266fc", "output_hash": "1ea442a134b2a184bd5d40104401f2a37fbc09ccf3f4bc9da161c6099be3691d"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "203b89cda76a88117927acfd", "attempt_id": "5d3ae8fda989", "run_group_id": "E4_core:A5|closed|qwen3b-llama3b:h8", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:22:43.347861Z", "task_family": "synthetic", "document_id": "syn_03021_2000", "document_hash": "3ac205b464552356699a4d2b0a7bf62497ead36e5bca73982e11ae5ee3810731", "question_id": "q08_revision", "split": "test", "cluster_id": "syn_03021_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "tokenizer_hashes": {"receiver": "b9d8872cf3a99da99e1fc4b0f71232a451632dc49ca4c6771549b4efa5c1c763"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 8, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1725, "query_tokens": 470, "message_tokens": 350, "message_utf8_bytes": 985, "cache_tensor_bytes": null, "serialized_payload_bytes": 985, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 38908.40608323924, "hops": {"1": 6896.219208021648, "2": 4294.3743750220165, "3": 4930.94241595827, "4": 4305.2063341019675, "5": 4931.834749993868, "6": 4301.946583087556, "7": 4939.746499992907, "8": 4308.1359170610085}, "receiver_prefill_ms": 202.15658308006823, "ttft_ms": 11.588749941438437, "decode_ms": 11.586457956582308}, "total_wall_ms": 39134.16225020774, "ttft_ms": 11.588749941438437, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 19, "retention_rate": 0.2923076923076923, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.15384615384615385, "status": 1.0, "place": 0.0, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 985}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 19, "retention_rate": 0.2923076923076923, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.15384615384615385, "status": 1.0, "place": 0.0, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 836}, "2": {"facts_total": 65, "facts_retained": 19, "retention_rate": 0.2923076923076923, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.15384615384615385, "status": 1.0, "place": 0.0, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 985}, "3": {"facts_total": 65, "facts_retained": 19, "retention_rate": 0.2923076923076923, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.15384615384615385, "status": 1.0, "place": 0.0, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 985}, "4": {"facts_total": 65, "facts_retained": 19, "retention_rate": 0.2923076923076923, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.15384615384615385, "status": 1.0, "place": 0.0, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 985}, "5": {"facts_total": 65, "facts_retained": 19, "retention_rate": 0.2923076923076923, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.15384615384615385, "status": 1.0, "place": 0.0, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 985}, "6": {"facts_total": 65, "facts_retained": 19, "retention_rate": 0.2923076923076923, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.15384615384615385, "status": 1.0, "place": 0.0, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 985}, "7": {"facts_total": 65, "facts_retained": 19, "retention_rate": 0.2923076923076923, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.15384615384615385, "status": 1.0, "place": 0.0, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 985}, "8": {"facts_total": 65, "facts_retained": 19, "retention_rate": 0.2923076923076923, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.15384615384615385, "status": 1.0, "place": 0.0, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 985}}}, "reference_key_id": "keys/synthetic/syn_03021_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "dc8b132a30e8989cf3c15b7b1a161c9730440bcba369b175bdd47033bb2266fc", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "cb0112a54b2ab91530061881", "attempt_id": "b4625ba61717", "run_group_id": "E4_core:A5|closed|qwen3b-llama3b:h8", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:22:43.588943Z", "task_family": "synthetic", "document_id": "syn_03021_2000", "document_hash": "3ac205b464552356699a4d2b0a7bf62497ead36e5bca73982e11ae5ee3810731", "question_id": "q09_revision", "split": "test", "cluster_id": "syn_03021_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "tokenizer_hashes": {"receiver": "b9d8872cf3a99da99e1fc4b0f71232a451632dc49ca4c6771549b4efa5c1c763"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 8, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1725, "query_tokens": 461, "message_tokens": 350, "message_utf8_bytes": 985, "cache_tensor_bytes": null, "serialized_payload_bytes": 985, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 38908.40608323924, "hops": {"1": 6896.219208021648, "2": 4294.3743750220165, "3": 4930.94241595827, "4": 4305.2063341019675, "5": 4931.834749993868, "6": 4301.946583087556, "7": 4939.746499992907, "8": 4308.1359170610085}, "receiver_prefill_ms": 201.65483397431672, "ttft_ms": 11.532332981005311, "decode_ms": 11.554541997611523}, "total_wall_ms": 39133.52791615762, "ttft_ms": 11.532332981005311, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 19, "retention_rate": 0.2923076923076923, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.15384615384615385, "status": 1.0, "place": 0.0, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 985}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 19, "retention_rate": 0.2923076923076923, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.15384615384615385, "status": 1.0, "place": 0.0, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 836}, "2": {"facts_total": 65, "facts_retained": 19, "retention_rate": 0.2923076923076923, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.15384615384615385, "status": 1.0, "place": 0.0, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 985}, "3": {"facts_total": 65, "facts_retained": 19, "retention_rate": 0.2923076923076923, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.15384615384615385, "status": 1.0, "place": 0.0, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 985}, "4": {"facts_total": 65, "facts_retained": 19, "retention_rate": 0.2923076923076923, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.15384615384615385, "status": 1.0, "place": 0.0, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 985}, "5": {"facts_total": 65, "facts_retained": 19, "retention_rate": 0.2923076923076923, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.15384615384615385, "status": 1.0, "place": 0.0, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 985}, "6": {"facts_total": 65, "facts_retained": 19, "retention_rate": 0.2923076923076923, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.15384615384615385, "status": 1.0, "place": 0.0, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 985}, "7": {"facts_total": 65, "facts_retained": 19, "retention_rate": 0.2923076923076923, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.15384615384615385, "status": 1.0, "place": 0.0, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 985}, "8": {"facts_total": 65, "facts_retained": 19, "retention_rate": 0.2923076923076923, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.15384615384615385, "status": 1.0, "place": 0.0, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 985}}}, "reference_key_id": "keys/synthetic/syn_03021_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "dc8b132a30e8989cf3c15b7b1a161c9730440bcba369b175bdd47033bb2266fc", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "f0ec1d9f0601ce9b0d52e097", "attempt_id": "f119ec23efa4", "run_group_id": "E4_core:A5|closed|qwen3b-llama3b:h8", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:22:43.829172Z", "task_family": "synthetic", "document_id": "syn_03021_2000", "document_hash": "3ac205b464552356699a4d2b0a7bf62497ead36e5bca73982e11ae5ee3810731", "question_id": "q10_temporal", "split": "test", "cluster_id": "syn_03021_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "tokenizer_hashes": {"receiver": "b9d8872cf3a99da99e1fc4b0f71232a451632dc49ca4c6771549b4efa5c1c763"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 8, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1725, "query_tokens": 469, "message_tokens": 350, "message_utf8_bytes": 985, "cache_tensor_bytes": null, "serialized_payload_bytes": 985, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 38908.40608323924, "hops": {"1": 6896.219208021648, "2": 4294.3743750220165, "3": 4930.94241595827, "4": 4305.2063341019675, "5": 4931.834749993868, "6": 4301.946583087556, "7": 4939.746499992907, "8": 4308.1359170610085}, "receiver_prefill_ms": 207.1392919169739, "ttft_ms": 11.553416028618813, "decode_ms": 11.523749912157655}, "total_wall_ms": 39139.00450034998, "ttft_ms": 11.553416028618813, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 19, "retention_rate": 0.2923076923076923, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.15384615384615385, "status": 1.0, "place": 0.0, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 985}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 19, "retention_rate": 0.2923076923076923, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.15384615384615385, "status": 1.0, "place": 0.0, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 836}, "2": {"facts_total": 65, "facts_retained": 19, "retention_rate": 0.2923076923076923, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.15384615384615385, "status": 1.0, "place": 0.0, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 985}, "3": {"facts_total": 65, "facts_retained": 19, "retention_rate": 0.2923076923076923, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.15384615384615385, "status": 1.0, "place": 0.0, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 985}, "4": {"facts_total": 65, "facts_retained": 19, "retention_rate": 0.2923076923076923, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.15384615384615385, "status": 1.0, "place": 0.0, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 985}, "5": {"facts_total": 65, "facts_retained": 19, "retention_rate": 0.2923076923076923, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.15384615384615385, "status": 1.0, "place": 0.0, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 985}, "6": {"facts_total": 65, "facts_retained": 19, "retention_rate": 0.2923076923076923, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.15384615384615385, "status": 1.0, "place": 0.0, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 985}, "7": {"facts_total": 65, "facts_retained": 19, "retention_rate": 0.2923076923076923, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.15384615384615385, "status": 1.0, "place": 0.0, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 985}, "8": {"facts_total": 65, "facts_retained": 19, "retention_rate": 0.2923076923076923, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.15384615384615385, "status": 1.0, "place": 0.0, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 985}}}, "reference_key_id": "keys/synthetic/syn_03021_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "dc8b132a30e8989cf3c15b7b1a161c9730440bcba369b175bdd47033bb2266fc", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "12056658164e80e5b64a4408", "attempt_id": "6f8c86e6c2bc", "run_group_id": "E4_core:A5|closed|qwen3b-llama3b:h8", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:22:44.074741Z", "task_family": "synthetic", "document_id": "syn_03021_2000", "document_hash": "3ac205b464552356699a4d2b0a7bf62497ead36e5bca73982e11ae5ee3810731", "question_id": "q11_aggregation", "split": "test", "cluster_id": "syn_03021_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "tokenizer_hashes": {"receiver": "b9d8872cf3a99da99e1fc4b0f71232a451632dc49ca4c6771549b4efa5c1c763"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 8, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1725, "query_tokens": 465, "message_tokens": 350, "message_utf8_bytes": 985, "cache_tensor_bytes": null, "serialized_payload_bytes": 985, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 38908.40608323924, "hops": {"1": 6896.219208021648, "2": 4294.3743750220165, "3": 4930.94241595827, "4": 4305.2063341019675, "5": 4931.834749993868, "6": 4301.946583087556, "7": 4939.746499992907, "8": 4308.1359170610085}, "receiver_prefill_ms": 203.10245803557336, "ttft_ms": 11.572084040381014, "decode_ms": 57.69333301577717}, "total_wall_ms": 39181.191000156105, "ttft_ms": 11.572084040381014, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 5}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "5,646,720", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 19, "retention_rate": 0.2923076923076923, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.15384615384615385, "status": 1.0, "place": 0.0, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 985}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 19, "retention_rate": 0.2923076923076923, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.15384615384615385, "status": 1.0, "place": 0.0, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 836}, "2": {"facts_total": 65, "facts_retained": 19, "retention_rate": 0.2923076923076923, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.15384615384615385, "status": 1.0, "place": 0.0, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 985}, "3": {"facts_total": 65, "facts_retained": 19, "retention_rate": 0.2923076923076923, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.15384615384615385, "status": 1.0, "place": 0.0, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 985}, "4": {"facts_total": 65, "facts_retained": 19, "retention_rate": 0.2923076923076923, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.15384615384615385, "status": 1.0, "place": 0.0, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 985}, "5": {"facts_total": 65, "facts_retained": 19, "retention_rate": 0.2923076923076923, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.15384615384615385, "status": 1.0, "place": 0.0, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 985}, "6": {"facts_total": 65, "facts_retained": 19, "retention_rate": 0.2923076923076923, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.15384615384615385, "status": 1.0, "place": 0.0, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 985}, "7": {"facts_total": 65, "facts_retained": 19, "retention_rate": 0.2923076923076923, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.15384615384615385, "status": 1.0, "place": 0.0, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 985}, "8": {"facts_total": 65, "facts_retained": 19, "retention_rate": 0.2923076923076923, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.15384615384615385, "status": 1.0, "place": 0.0, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 985}}}, "reference_key_id": "keys/synthetic/syn_03021_2000.json", "task_scores": {"correct": 0.0, "abstained": 0.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "dc8b132a30e8989cf3c15b7b1a161c9730440bcba369b175bdd47033bb2266fc", "output_hash": "2c0bb14ad8ba61e6d23fb3dea5205f2f02c43c95a0cf8e7f93c240ddea626ad4"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "4c5ddf64c06c4bf305520ffd", "attempt_id": "985c363a0213", "run_group_id": "E4_core:A5|closed|qwen3b-llama3b:h8", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:22:44.362681Z", "task_family": "synthetic", "document_id": "syn_03021_2000", "document_hash": "3ac205b464552356699a4d2b0a7bf62497ead36e5bca73982e11ae5ee3810731", "question_id": "q12_unit_conversion", "split": "test", "cluster_id": "syn_03021_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "tokenizer_hashes": {"receiver": "b9d8872cf3a99da99e1fc4b0f71232a451632dc49ca4c6771549b4efa5c1c763"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 8, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1725, "query_tokens": 460, "message_tokens": 350, "message_utf8_bytes": 985, "cache_tensor_bytes": null, "serialized_payload_bytes": 985, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 38908.40608323924, "hops": {"1": 6896.219208021648, "2": 4294.3743750220165, "3": 4930.94241595827, "4": 4305.2063341019675, "5": 4931.834749993868, "6": 4301.946583087556, "7": 4939.746499992907, "8": 4308.1359170610085}, "receiver_prefill_ms": 201.77362498361617, "ttft_ms": 11.604042025282979, "decode_ms": 11.506457929499447}, "total_wall_ms": 39133.6852501845, "ttft_ms": 11.604042025282979, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 19, "retention_rate": 0.2923076923076923, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.15384615384615385, "status": 1.0, "place": 0.0, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 985}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 19, "retention_rate": 0.2923076923076923, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.15384615384615385, "status": 1.0, "place": 0.0, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 836}, "2": {"facts_total": 65, "facts_retained": 19, "retention_rate": 0.2923076923076923, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.15384615384615385, "status": 1.0, "place": 0.0, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 985}, "3": {"facts_total": 65, "facts_retained": 19, "retention_rate": 0.2923076923076923, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.15384615384615385, "status": 1.0, "place": 0.0, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 985}, "4": {"facts_total": 65, "facts_retained": 19, "retention_rate": 0.2923076923076923, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.15384615384615385, "status": 1.0, "place": 0.0, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 985}, "5": {"facts_total": 65, "facts_retained": 19, "retention_rate": 0.2923076923076923, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.15384615384615385, "status": 1.0, "place": 0.0, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 985}, "6": {"facts_total": 65, "facts_retained": 19, "retention_rate": 0.2923076923076923, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.15384615384615385, "status": 1.0, "place": 0.0, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 985}, "7": {"facts_total": 65, "facts_retained": 19, "retention_rate": 0.2923076923076923, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.15384615384615385, "status": 1.0, "place": 0.0, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 985}, "8": {"facts_total": 65, "facts_retained": 19, "retention_rate": 0.2923076923076923, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.15384615384615385, "status": 1.0, "place": 0.0, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 985}}}, "reference_key_id": "keys/synthetic/syn_03021_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "dc8b132a30e8989cf3c15b7b1a161c9730440bcba369b175bdd47033bb2266fc", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "293149f634f67396053180ce", "attempt_id": "dcd61e6735b2", "run_group_id": "E4_core:A5|closed|qwen3b-llama3b:h8", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:22:44.603148Z", "task_family": "synthetic", "document_id": "syn_03021_2000", "document_hash": "3ac205b464552356699a4d2b0a7bf62497ead36e5bca73982e11ae5ee3810731", "question_id": "q13_unanswerable", "split": "test", "cluster_id": "syn_03021_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "tokenizer_hashes": {"receiver": "b9d8872cf3a99da99e1fc4b0f71232a451632dc49ca4c6771549b4efa5c1c763"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 8, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1725, "query_tokens": 464, "message_tokens": 350, "message_utf8_bytes": 985, "cache_tensor_bytes": null, "serialized_payload_bytes": 985, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 38908.40608323924, "hops": {"1": 6896.219208021648, "2": 4294.3743750220165, "3": 4930.94241595827, "4": 4305.2063341019675, "5": 4931.834749993868, "6": 4301.946583087556, "7": 4939.746499992907, "8": 4308.1359170610085}, "receiver_prefill_ms": 204.5530000468716, "ttft_ms": 11.566541972570121, "decode_ms": 11.507750023156404}, "total_wall_ms": 39136.406833305955, "ttft_ms": 11.566541972570121, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 19, "retention_rate": 0.2923076923076923, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.15384615384615385, "status": 1.0, "place": 0.0, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 985}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 19, "retention_rate": 0.2923076923076923, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.15384615384615385, "status": 1.0, "place": 0.0, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 836}, "2": {"facts_total": 65, "facts_retained": 19, "retention_rate": 0.2923076923076923, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.15384615384615385, "status": 1.0, "place": 0.0, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 985}, "3": {"facts_total": 65, "facts_retained": 19, "retention_rate": 0.2923076923076923, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.15384615384615385, "status": 1.0, "place": 0.0, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 985}, "4": {"facts_total": 65, "facts_retained": 19, "retention_rate": 0.2923076923076923, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.15384615384615385, "status": 1.0, "place": 0.0, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 985}, "5": {"facts_total": 65, "facts_retained": 19, "retention_rate": 0.2923076923076923, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.15384615384615385, "status": 1.0, "place": 0.0, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 985}, "6": {"facts_total": 65, "facts_retained": 19, "retention_rate": 0.2923076923076923, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.15384615384615385, "status": 1.0, "place": 0.0, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 985}, "7": {"facts_total": 65, "facts_retained": 19, "retention_rate": 0.2923076923076923, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.15384615384615385, "status": 1.0, "place": 0.0, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 985}, "8": {"facts_total": 65, "facts_retained": 19, "retention_rate": 0.2923076923076923, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.15384615384615385, "status": 1.0, "place": 0.0, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 985}}}, "reference_key_id": "keys/synthetic/syn_03021_2000.json", "task_scores": {"correct": 1.0, "abstained": 1.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "dc8b132a30e8989cf3c15b7b1a161c9730440bcba369b175bdd47033bb2266fc", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "da6f4356196da5ddf1f1bc9f", "attempt_id": "c93620885371", "run_group_id": "E4_core:A5|closed|qwen3b-llama3b:h8", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:22:44.846193Z", "task_family": "synthetic", "document_id": "syn_03021_2000", "document_hash": "3ac205b464552356699a4d2b0a7bf62497ead36e5bca73982e11ae5ee3810731", "question_id": "q14_unanswerable", "split": "test", "cluster_id": "syn_03021_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "tokenizer_hashes": {"receiver": "b9d8872cf3a99da99e1fc4b0f71232a451632dc49ca4c6771549b4efa5c1c763"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 8, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1725, "query_tokens": 467, "message_tokens": 350, "message_utf8_bytes": 985, "cache_tensor_bytes": null, "serialized_payload_bytes": 985, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 38908.40608323924, "hops": {"1": 6896.219208021648, "2": 4294.3743750220165, "3": 4930.94241595827, "4": 4305.2063341019675, "5": 4931.834749993868, "6": 4301.946583087556, "7": 4939.746499992907, "8": 4308.1359170610085}, "receiver_prefill_ms": 205.05233306903392, "ttft_ms": 11.624832986854017, "decode_ms": 11.526959016919136}, "total_wall_ms": 39136.99137524236, "ttft_ms": 11.624832986854017, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 65, "facts_retained": 19, "retention_rate": 0.2923076923076923, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.15384615384615385, "status": 1.0, "place": 0.0, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 985}, "ledger_by_hop": {"1": {"facts_total": 65, "facts_retained": 19, "retention_rate": 0.2923076923076923, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.15384615384615385, "status": 1.0, "place": 0.0, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 836}, "2": {"facts_total": 65, "facts_retained": 19, "retention_rate": 0.2923076923076923, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.15384615384615385, "status": 1.0, "place": 0.0, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 985}, "3": {"facts_total": 65, "facts_retained": 19, "retention_rate": 0.2923076923076923, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.15384615384615385, "status": 1.0, "place": 0.0, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 985}, "4": {"facts_total": 65, "facts_retained": 19, "retention_rate": 0.2923076923076923, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.15384615384615385, "status": 1.0, "place": 0.0, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 985}, "5": {"facts_total": 65, "facts_retained": 19, "retention_rate": 0.2923076923076923, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.15384615384615385, "status": 1.0, "place": 0.0, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 985}, "6": {"facts_total": 65, "facts_retained": 19, "retention_rate": 0.2923076923076923, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.15384615384615385, "status": 1.0, "place": 0.0, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 985}, "7": {"facts_total": 65, "facts_retained": 19, "retention_rate": 0.2923076923076923, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.15384615384615385, "status": 1.0, "place": 0.0, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 985}, "8": {"facts_total": 65, "facts_retained": 19, "retention_rate": 0.2923076923076923, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.1, "amount": 0.15384615384615385, "status": 1.0, "place": 0.0, "identifier": 0.16666666666666666, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 985}}}, "reference_key_id": "keys/synthetic/syn_03021_2000.json", "task_scores": {"correct": 1.0, "abstained": 1.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "dc8b132a30e8989cf3c15b7b1a161c9730440bcba369b175bdd47033bb2266fc", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "7f94f53a18e209d47bfb8ddd", "attempt_id": "ec035f1a57c0", "run_group_id": "E4_core:A4|closed|qwen3b:h1", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:22:52.137546Z", "task_family": "synthetic", "document_id": "syn_03024_2000", "document_hash": "ea10cd11ad08f01c50b0712bc481668857c9d2ddc8e6423968f2a31eeb97ac55", "question_id": "q00_single_fact", "split": "test", "cluster_id": "syn_03024_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 1, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 2068, "query_tokens": 595, "message_tokens": 512, "message_utf8_bytes": 1508, "cache_tensor_bytes": null, "serialized_payload_bytes": 1508, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 6965.8001670613885, "hops": {"1": 6965.8001670613885}, "receiver_prefill_ms": 227.33754105865955, "ttft_ms": 11.96925004478544, "decode_ms": 11.930791893973947}, "total_wall_ms": 7217.62954199221, "ttft_ms": 11.96925004478544, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 66, "facts_retained": 21, "retention_rate": 0.3181818181818182, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.35294117647058826, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1508}, "ledger_by_hop": {"1": {"facts_total": 66, "facts_retained": 21, "retention_rate": 0.3181818181818182, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.35294117647058826, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1508}}}, "reference_key_id": "keys/synthetic/syn_03024_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "34fcc1e23e5a866b5b7d7e222f0380ba771c4005d3732e9747f026c211840d43", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "6e4edc15fcdb423774811fe5", "attempt_id": "9b2dbf8d61df", "run_group_id": "E4_core:A4|closed|qwen3b:h1", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:22:52.405480Z", "task_family": "synthetic", "document_id": "syn_03024_2000", "document_hash": "ea10cd11ad08f01c50b0712bc481668857c9d2ddc8e6423968f2a31eeb97ac55", "question_id": "q01_single_fact", "split": "test", "cluster_id": "syn_03024_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 1, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 2068, "query_tokens": 600, "message_tokens": 512, "message_utf8_bytes": 1508, "cache_tensor_bytes": null, "serialized_payload_bytes": 1508, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 6965.8001670613885, "hops": {"1": 6965.8001670613885}, "receiver_prefill_ms": 227.46345796622336, "ttft_ms": 11.999250040389597, "decode_ms": 11.93166698794812}, "total_wall_ms": 7217.751334072091, "ttft_ms": 11.999250040389597, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 66, "facts_retained": 21, "retention_rate": 0.3181818181818182, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.35294117647058826, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1508}, "ledger_by_hop": {"1": {"facts_total": 66, "facts_retained": 21, "retention_rate": 0.3181818181818182, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.35294117647058826, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1508}}}, "reference_key_id": "keys/synthetic/syn_03024_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "34fcc1e23e5a866b5b7d7e222f0380ba771c4005d3732e9747f026c211840d43", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "5dfa922561234f365efc4e06", "attempt_id": "b35ff64d0baa", "run_group_id": "E4_core:A4|closed|qwen3b:h1", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:22:52.673131Z", "task_family": "synthetic", "document_id": "syn_03024_2000", "document_hash": "ea10cd11ad08f01c50b0712bc481668857c9d2ddc8e6423968f2a31eeb97ac55", "question_id": "q02_exact_copy", "split": "test", "cluster_id": "syn_03024_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 1, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 2068, "query_tokens": 600, "message_tokens": 512, "message_utf8_bytes": 1508, "cache_tensor_bytes": null, "serialized_payload_bytes": 1508, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 6965.8001670613885, "hops": {"1": 6965.8001670613885}, "receiver_prefill_ms": 226.3631250243634, "ttft_ms": 11.959125055000186, "decode_ms": 83.38604203891009}, "total_wall_ms": 7288.033208111301, "ttft_ms": 11.959125055000186, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 7}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "30.0224", "evidence_artifact": {"ledger": {"facts_total": 66, "facts_retained": 21, "retention_rate": 0.3181818181818182, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.35294117647058826, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1508}, "ledger_by_hop": {"1": {"facts_total": 66, "facts_retained": 21, "retention_rate": 0.3181818181818182, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.35294117647058826, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1508}}}, "reference_key_id": "keys/synthetic/syn_03024_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "34fcc1e23e5a866b5b7d7e222f0380ba771c4005d3732e9747f026c211840d43", "output_hash": "e28a78bc4deb3cec8beab4bf5b87297d7cc55a3e6f839b43fe6408735d464923"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "b5efa056aa8a9ab05e5a1ea1", "attempt_id": "8b686fd86132", "run_group_id": "E4_core:A4|closed|qwen3b:h1", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:22:53.010857Z", "task_family": "synthetic", "document_id": "syn_03024_2000", "document_hash": "ea10cd11ad08f01c50b0712bc481668857c9d2ddc8e6423968f2a31eeb97ac55", "question_id": "q03_exact_copy", "split": "test", "cluster_id": "syn_03024_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 1, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 2068, "query_tokens": 595, "message_tokens": 512, "message_utf8_bytes": 1508, "cache_tensor_bytes": null, "serialized_payload_bytes": 1508, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 6965.8001670613885, "hops": {"1": 6965.8001670613885}, "receiver_prefill_ms": 226.75329202320427, "ttft_ms": 11.945249978452921, "decode_ms": 12.137541081756353}, "total_wall_ms": 7217.132042045705, "ttft_ms": 11.945249978452921, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 66, "facts_retained": 21, "retention_rate": 0.3181818181818182, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.35294117647058826, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1508}, "ledger_by_hop": {"1": {"facts_total": 66, "facts_retained": 21, "retention_rate": 0.3181818181818182, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.35294117647058826, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1508}}}, "reference_key_id": "keys/synthetic/syn_03024_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "34fcc1e23e5a866b5b7d7e222f0380ba771c4005d3732e9747f026c211840d43", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "5c837f57c51575c785d18044", "attempt_id": "0585fa7f8963", "run_group_id": "E4_core:A4|closed|qwen3b:h1", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:22:53.277268Z", "task_family": "synthetic", "document_id": "syn_03024_2000", "document_hash": "ea10cd11ad08f01c50b0712bc481668857c9d2ddc8e6423968f2a31eeb97ac55", "question_id": "q04_multi_hop", "split": "test", "cluster_id": "syn_03024_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 1, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 2068, "query_tokens": 601, "message_tokens": 512, "message_utf8_bytes": 1508, "cache_tensor_bytes": null, "serialized_payload_bytes": 1508, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 6965.8001670613885, "hops": {"1": 6965.8001670613885}, "receiver_prefill_ms": 227.10254206322134, "ttft_ms": 11.983082978986204, "decode_ms": 107.37483308184892}, "total_wall_ms": 7312.79041711241, "ttft_ms": 11.983082978986204, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 9}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "6,009,480", "evidence_artifact": {"ledger": {"facts_total": 66, "facts_retained": 21, "retention_rate": 0.3181818181818182, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.35294117647058826, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1508}, "ledger_by_hop": {"1": {"facts_total": 66, "facts_retained": 21, "retention_rate": 0.3181818181818182, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.35294117647058826, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1508}}}, "reference_key_id": "keys/synthetic/syn_03024_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "34fcc1e23e5a866b5b7d7e222f0380ba771c4005d3732e9747f026c211840d43", "output_hash": "30b20d9aae4d7b0587edeefdc2a9a159f73faf6e2570f1d408a415fe92cbf45e"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "df42cc1d9e3606b312e48498", "attempt_id": "ef899a5d9889", "run_group_id": "E4_core:A4|closed|qwen3b:h1", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:22:53.639445Z", "task_family": "synthetic", "document_id": "syn_03024_2000", "document_hash": "ea10cd11ad08f01c50b0712bc481668857c9d2ddc8e6423968f2a31eeb97ac55", "question_id": "q05_multi_hop", "split": "test", "cluster_id": "syn_03024_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 1, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 2068, "query_tokens": 605, "message_tokens": 512, "message_utf8_bytes": 1508, "cache_tensor_bytes": null, "serialized_payload_bytes": 1508, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 6965.8001670613885, "hops": {"1": 6965.8001670613885}, "receiver_prefill_ms": 227.3269579745829, "ttft_ms": 12.147915898822248, "decode_ms": 12.042958987876773}, "total_wall_ms": 7217.956667067483, "ttft_ms": 12.147915898822248, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 66, "facts_retained": 21, "retention_rate": 0.3181818181818182, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.35294117647058826, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1508}, "ledger_by_hop": {"1": {"facts_total": 66, "facts_retained": 21, "retention_rate": 0.3181818181818182, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.35294117647058826, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1508}}}, "reference_key_id": "keys/synthetic/syn_03024_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "34fcc1e23e5a866b5b7d7e222f0380ba771c4005d3732e9747f026c211840d43", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "71c20d3d7f450645cb2b88f2", "attempt_id": "4c0a3b666a66", "run_group_id": "E4_core:A4|closed|qwen3b:h1", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:22:53.906924Z", "task_family": "synthetic", "document_id": "syn_03024_2000", "document_hash": "ea10cd11ad08f01c50b0712bc481668857c9d2ddc8e6423968f2a31eeb97ac55", "question_id": "q06_negation", "split": "test", "cluster_id": "syn_03024_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 1, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 2068, "query_tokens": 594, "message_tokens": 512, "message_utf8_bytes": 1508, "cache_tensor_bytes": null, "serialized_payload_bytes": 1508, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 6965.8001670613885, "hops": {"1": 6965.8001670613885}, "receiver_prefill_ms": 226.65075003169477, "ttft_ms": 11.850790935568511, "decode_ms": 11.824292014352977}, "total_wall_ms": 7216.691459063441, "ttft_ms": 11.850790935568511, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "No", "evidence_artifact": {"ledger": {"facts_total": 66, "facts_retained": 21, "retention_rate": 0.3181818181818182, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.35294117647058826, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1508}, "ledger_by_hop": {"1": {"facts_total": 66, "facts_retained": 21, "retention_rate": 0.3181818181818182, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.35294117647058826, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1508}}}, "reference_key_id": "keys/synthetic/syn_03024_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "34fcc1e23e5a866b5b7d7e222f0380ba771c4005d3732e9747f026c211840d43", "output_hash": "1ea442a134b2a184bd5d40104401f2a37fbc09ccf3f4bc9da161c6099be3691d"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "20691cdd16bcd0ee2b7376ce", "attempt_id": "53cd8bbb4df2", "run_group_id": "E4_core:A4|closed|qwen3b:h1", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:22:54.173242Z", "task_family": "synthetic", "document_id": "syn_03024_2000", "document_hash": "ea10cd11ad08f01c50b0712bc481668857c9d2ddc8e6423968f2a31eeb97ac55", "question_id": "q07_negation", "split": "test", "cluster_id": "syn_03024_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 1, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 2068, "query_tokens": 596, "message_tokens": 512, "message_utf8_bytes": 1508, "cache_tensor_bytes": null, "serialized_payload_bytes": 1508, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 6965.8001670613885, "hops": {"1": 6965.8001670613885}, "receiver_prefill_ms": 226.6510840272531, "ttft_ms": 12.024041963741183, "decode_ms": 11.98533398564905}, "total_wall_ms": 7216.995583148673, "ttft_ms": 12.024041963741183, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "no", "evidence_artifact": {"ledger": {"facts_total": 66, "facts_retained": 21, "retention_rate": 0.3181818181818182, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.35294117647058826, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1508}, "ledger_by_hop": {"1": {"facts_total": 66, "facts_retained": 21, "retention_rate": 0.3181818181818182, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.35294117647058826, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1508}}}, "reference_key_id": "keys/synthetic/syn_03024_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "34fcc1e23e5a866b5b7d7e222f0380ba771c4005d3732e9747f026c211840d43", "output_hash": "9390298f3fb0c5b160498935d79cb139aef28e1c47358b4bbba61862b9c26e59"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "e325942b42e596df339f4ce2", "attempt_id": "b84a2ed3fc50", "run_group_id": "E4_core:A4|closed|qwen3b:h1", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:22:54.440186Z", "task_family": "synthetic", "document_id": "syn_03024_2000", "document_hash": "ea10cd11ad08f01c50b0712bc481668857c9d2ddc8e6423968f2a31eeb97ac55", "question_id": "q08_revision", "split": "test", "cluster_id": "syn_03024_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 1, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 2068, "query_tokens": 612, "message_tokens": 512, "message_utf8_bytes": 1508, "cache_tensor_bytes": null, "serialized_payload_bytes": 1508, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 6965.8001670613885, "hops": {"1": 6965.8001670613885}, "receiver_prefill_ms": 229.6456249896437, "ttft_ms": 12.049874989315867, "decode_ms": 11.993665946647525}, "total_wall_ms": 7220.0570000568405, "ttft_ms": 12.049874989315867, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 66, "facts_retained": 21, "retention_rate": 0.3181818181818182, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.35294117647058826, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1508}, "ledger_by_hop": {"1": {"facts_total": 66, "facts_retained": 21, "retention_rate": 0.3181818181818182, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.35294117647058826, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1508}}}, "reference_key_id": "keys/synthetic/syn_03024_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "34fcc1e23e5a866b5b7d7e222f0380ba771c4005d3732e9747f026c211840d43", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "8992ac2d14527f84647b37e4", "attempt_id": "eba93941dcbd", "run_group_id": "E4_core:A4|closed|qwen3b:h1", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:22:54.709948Z", "task_family": "synthetic", "document_id": "syn_03024_2000", "document_hash": "ea10cd11ad08f01c50b0712bc481668857c9d2ddc8e6423968f2a31eeb97ac55", "question_id": "q09_revision", "split": "test", "cluster_id": "syn_03024_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 1, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 2068, "query_tokens": 603, "message_tokens": 512, "message_utf8_bytes": 1508, "cache_tensor_bytes": null, "serialized_payload_bytes": 1508, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 6965.8001670613885, "hops": {"1": 6965.8001670613885}, "receiver_prefill_ms": 227.03824995551258, "ttft_ms": 12.024249997921288, "decode_ms": 11.960541014559567}, "total_wall_ms": 7217.3483750084415, "ttft_ms": 12.024249997921288, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 66, "facts_retained": 21, "retention_rate": 0.3181818181818182, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.35294117647058826, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1508}, "ledger_by_hop": {"1": {"facts_total": 66, "facts_retained": 21, "retention_rate": 0.3181818181818182, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.35294117647058826, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1508}}}, "reference_key_id": "keys/synthetic/syn_03024_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "34fcc1e23e5a866b5b7d7e222f0380ba771c4005d3732e9747f026c211840d43", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "714a3b5afc6e23c38f513703", "attempt_id": "dc9527a73532", "run_group_id": "E4_core:A4|closed|qwen3b:h1", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:22:54.976827Z", "task_family": "synthetic", "document_id": "syn_03024_2000", "document_hash": "ea10cd11ad08f01c50b0712bc481668857c9d2ddc8e6423968f2a31eeb97ac55", "question_id": "q10_temporal", "split": "test", "cluster_id": "syn_03024_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 1, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 2068, "query_tokens": 613, "message_tokens": 512, "message_utf8_bytes": 1508, "cache_tensor_bytes": null, "serialized_payload_bytes": 1508, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 6965.8001670613885, "hops": {"1": 6965.8001670613885}, "receiver_prefill_ms": 229.2202499229461, "ttft_ms": 12.016541906632483, "decode_ms": 11.87279203440994}, "total_wall_ms": 7219.4512920686975, "ttft_ms": 12.016541906632483, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 66, "facts_retained": 21, "retention_rate": 0.3181818181818182, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.35294117647058826, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1508}, "ledger_by_hop": {"1": {"facts_total": 66, "facts_retained": 21, "retention_rate": 0.3181818181818182, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.35294117647058826, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1508}}}, "reference_key_id": "keys/synthetic/syn_03024_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "34fcc1e23e5a866b5b7d7e222f0380ba771c4005d3732e9747f026c211840d43", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "d242cfb6a8997d4430ef185c", "attempt_id": "2b7e40d65534", "run_group_id": "E4_core:A4|closed|qwen3b:h1", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:22:55.246057Z", "task_family": "synthetic", "document_id": "syn_03024_2000", "document_hash": "ea10cd11ad08f01c50b0712bc481668857c9d2ddc8e6423968f2a31eeb97ac55", "question_id": "q11_aggregation", "split": "test", "cluster_id": "syn_03024_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 1, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 2068, "query_tokens": 606, "message_tokens": 512, "message_utf8_bytes": 1508, "cache_tensor_bytes": null, "serialized_payload_bytes": 1508, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 6965.8001670613885, "hops": {"1": 6965.8001670613885}, "receiver_prefill_ms": 227.04345802776515, "ttft_ms": 12.091790908016264, "decode_ms": 83.68970802985132}, "total_wall_ms": 7289.284917060286, "ttft_ms": 12.091790908016264, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 7}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "2707654", "evidence_artifact": {"ledger": {"facts_total": 66, "facts_retained": 21, "retention_rate": 0.3181818181818182, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.35294117647058826, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1508}, "ledger_by_hop": {"1": {"facts_total": 66, "facts_retained": 21, "retention_rate": 0.3181818181818182, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.35294117647058826, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1508}}}, "reference_key_id": "keys/synthetic/syn_03024_2000.json", "task_scores": {"correct": 0.0, "abstained": 0.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "34fcc1e23e5a866b5b7d7e222f0380ba771c4005d3732e9747f026c211840d43", "output_hash": "7799e805760fa6568574360f61d4c7cc9e162842191ff434dfed71076922855e"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "1464ad9d7ff02bdb87b9ffa8", "attempt_id": "3c39669a8a37", "run_group_id": "E4_core:A4|closed|qwen3b:h1", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:22:55.584340Z", "task_family": "synthetic", "document_id": "syn_03024_2000", "document_hash": "ea10cd11ad08f01c50b0712bc481668857c9d2ddc8e6423968f2a31eeb97ac55", "question_id": "q12_unit_conversion", "split": "test", "cluster_id": "syn_03024_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 1, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 2068, "query_tokens": 603, "message_tokens": 512, "message_utf8_bytes": 1508, "cache_tensor_bytes": null, "serialized_payload_bytes": 1508, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 6965.8001670613885, "hops": {"1": 6965.8001670613885}, "receiver_prefill_ms": 226.7504578921944, "ttft_ms": 12.062042020261288, "decode_ms": 11.881207930855453}, "total_wall_ms": 7217.175125959329, "ttft_ms": 12.062042020261288, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 66, "facts_retained": 21, "retention_rate": 0.3181818181818182, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.35294117647058826, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1508}, "ledger_by_hop": {"1": {"facts_total": 66, "facts_retained": 21, "retention_rate": 0.3181818181818182, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.35294117647058826, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1508}}}, "reference_key_id": "keys/synthetic/syn_03024_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "34fcc1e23e5a866b5b7d7e222f0380ba771c4005d3732e9747f026c211840d43", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "9ab0b72e3db8ca45d469b799", "attempt_id": "a55451fc0127", "run_group_id": "E4_core:A4|closed|qwen3b:h1", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:22:55.851616Z", "task_family": "synthetic", "document_id": "syn_03024_2000", "document_hash": "ea10cd11ad08f01c50b0712bc481668857c9d2ddc8e6423968f2a31eeb97ac55", "question_id": "q13_unanswerable", "split": "test", "cluster_id": "syn_03024_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 1, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 2068, "query_tokens": 607, "message_tokens": 512, "message_utf8_bytes": 1508, "cache_tensor_bytes": null, "serialized_payload_bytes": 1508, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 6965.8001670613885, "hops": {"1": 6965.8001670613885}, "receiver_prefill_ms": 227.45554195716977, "ttft_ms": 12.18508300371468, "decode_ms": 12.01075001154095}, "total_wall_ms": 7218.157959054224, "ttft_ms": 12.18508300371468, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 66, "facts_retained": 21, "retention_rate": 0.3181818181818182, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.35294117647058826, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1508}, "ledger_by_hop": {"1": {"facts_total": 66, "facts_retained": 21, "retention_rate": 0.3181818181818182, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.35294117647058826, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1508}}}, "reference_key_id": "keys/synthetic/syn_03024_2000.json", "task_scores": {"correct": 1.0, "abstained": 1.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "34fcc1e23e5a866b5b7d7e222f0380ba771c4005d3732e9747f026c211840d43", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "b56cb03e9a57414caae162ea", "attempt_id": "48c7a46ea91f", "run_group_id": "E4_core:A4|closed|qwen3b:h1", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:22:56.120825Z", "task_family": "synthetic", "document_id": "syn_03024_2000", "document_hash": "ea10cd11ad08f01c50b0712bc481668857c9d2ddc8e6423968f2a31eeb97ac55", "question_id": "q14_unanswerable", "split": "test", "cluster_id": "syn_03024_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 1, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 2068, "query_tokens": 608, "message_tokens": 512, "message_utf8_bytes": 1508, "cache_tensor_bytes": null, "serialized_payload_bytes": 1508, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 6965.8001670613885, "hops": {"1": 6965.8001670613885}, "receiver_prefill_ms": 227.2582909790799, "ttft_ms": 11.99937507044524, "decode_ms": 12.003959040157497}, "total_wall_ms": 7217.628707992844, "ttft_ms": 11.99937507044524, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 66, "facts_retained": 21, "retention_rate": 0.3181818181818182, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.35294117647058826, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1508}, "ledger_by_hop": {"1": {"facts_total": 66, "facts_retained": 21, "retention_rate": 0.3181818181818182, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.35294117647058826, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1508}}}, "reference_key_id": "keys/synthetic/syn_03024_2000.json", "task_scores": {"correct": 1.0, "abstained": 1.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "34fcc1e23e5a866b5b7d7e222f0380ba771c4005d3732e9747f026c211840d43", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "753f744d5590b2274579aaaa", "attempt_id": "3a6577be9302", "run_group_id": "E4_core:A4|closed|qwen3b:h2", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:23:02.774149Z", "task_family": "synthetic", "document_id": "syn_03024_2000", "document_hash": "ea10cd11ad08f01c50b0712bc481668857c9d2ddc8e6423968f2a31eeb97ac55", "question_id": "q00_single_fact", "split": "test", "cluster_id": "syn_03024_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 2, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 2068, "query_tokens": 595, "message_tokens": 512, "message_utf8_bytes": 1508, "cache_tensor_bytes": null, "serialized_payload_bytes": 1508, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 13349.597501102835, "hops": {"1": 6965.8001670613885, "2": 6383.797334041446}, "receiver_prefill_ms": 226.68308403808624, "ttft_ms": 11.846457957290113, "decode_ms": 11.902624974027276}, "total_wall_ms": 13600.601210142486, "ttft_ms": 11.846457957290113, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 66, "facts_retained": 21, "retention_rate": 0.3181818181818182, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.35294117647058826, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1508}, "ledger_by_hop": {"1": {"facts_total": 66, "facts_retained": 21, "retention_rate": 0.3181818181818182, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.35294117647058826, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1508}, "2": {"facts_total": 66, "facts_retained": 21, "retention_rate": 0.3181818181818182, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.35294117647058826, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1508}}}, "reference_key_id": "keys/synthetic/syn_03024_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "34fcc1e23e5a866b5b7d7e222f0380ba771c4005d3732e9747f026c211840d43", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "e39732132c6f0d74391922ec", "attempt_id": "20e5b09ce7ad", "run_group_id": "E4_core:A4|closed|qwen3b:h2", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:23:03.041491Z", "task_family": "synthetic", "document_id": "syn_03024_2000", "document_hash": "ea10cd11ad08f01c50b0712bc481668857c9d2ddc8e6423968f2a31eeb97ac55", "question_id": "q01_single_fact", "split": "test", "cluster_id": "syn_03024_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 2, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 2068, "query_tokens": 600, "message_tokens": 512, "message_utf8_bytes": 1508, "cache_tensor_bytes": null, "serialized_payload_bytes": 1508, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 13349.597501102835, "hops": {"1": 6965.8001670613885, "2": 6383.797334041446}, "receiver_prefill_ms": 226.7764579737559, "ttft_ms": 11.951458058319986, "decode_ms": 11.919416952878237}, "total_wall_ms": 13600.782084162347, "ttft_ms": 11.951458058319986, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 66, "facts_retained": 21, "retention_rate": 0.3181818181818182, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.35294117647058826, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1508}, "ledger_by_hop": {"1": {"facts_total": 66, "facts_retained": 21, "retention_rate": 0.3181818181818182, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.35294117647058826, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1508}, "2": {"facts_total": 66, "facts_retained": 21, "retention_rate": 0.3181818181818182, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.35294117647058826, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1508}}}, "reference_key_id": "keys/synthetic/syn_03024_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "34fcc1e23e5a866b5b7d7e222f0380ba771c4005d3732e9747f026c211840d43", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "baeb2d37a0177444fe208edd", "attempt_id": "b5dd2a0849b9", "run_group_id": "E4_core:A4|closed|qwen3b:h2", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:23:03.308153Z", "task_family": "synthetic", "document_id": "syn_03024_2000", "document_hash": "ea10cd11ad08f01c50b0712bc481668857c9d2ddc8e6423968f2a31eeb97ac55", "question_id": "q02_exact_copy", "split": "test", "cluster_id": "syn_03024_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 2, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 2068, "query_tokens": 600, "message_tokens": 512, "message_utf8_bytes": 1508, "cache_tensor_bytes": null, "serialized_payload_bytes": 1508, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 13349.597501102835, "hops": {"1": 6965.8001670613885, "2": 6383.797334041446}, "receiver_prefill_ms": 226.385583053343, "ttft_ms": 12.16383301652968, "decode_ms": 84.16995801962912}, "total_wall_ms": 13673.025210155174, "ttft_ms": 12.16383301652968, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 7}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "30.0224", "evidence_artifact": {"ledger": {"facts_total": 66, "facts_retained": 21, "retention_rate": 0.3181818181818182, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.35294117647058826, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1508}, "ledger_by_hop": {"1": {"facts_total": 66, "facts_retained": 21, "retention_rate": 0.3181818181818182, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.35294117647058826, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1508}, "2": {"facts_total": 66, "facts_retained": 21, "retention_rate": 0.3181818181818182, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.35294117647058826, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1508}}}, "reference_key_id": "keys/synthetic/syn_03024_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "34fcc1e23e5a866b5b7d7e222f0380ba771c4005d3732e9747f026c211840d43", "output_hash": "e28a78bc4deb3cec8beab4bf5b87297d7cc55a3e6f839b43fe6408735d464923"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "5af9030e32b929ada6ae4d65", "attempt_id": "e81ec9708515", "run_group_id": "E4_core:A4|closed|qwen3b:h2", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:23:03.646825Z", "task_family": "synthetic", "document_id": "syn_03024_2000", "document_hash": "ea10cd11ad08f01c50b0712bc481668857c9d2ddc8e6423968f2a31eeb97ac55", "question_id": "q03_exact_copy", "split": "test", "cluster_id": "syn_03024_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 2, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 2068, "query_tokens": 595, "message_tokens": 512, "message_utf8_bytes": 1508, "cache_tensor_bytes": null, "serialized_payload_bytes": 1508, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 13349.597501102835, "hops": {"1": 6965.8001670613885, "2": 6383.797334041446}, "receiver_prefill_ms": 226.614834042266, "ttft_ms": 12.10945809725672, "decode_ms": 12.029916979372501}, "total_wall_ms": 13601.05179308448, "ttft_ms": 12.10945809725672, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 66, "facts_retained": 21, "retention_rate": 0.3181818181818182, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.35294117647058826, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1508}, "ledger_by_hop": {"1": {"facts_total": 66, "facts_retained": 21, "retention_rate": 0.3181818181818182, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.35294117647058826, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1508}, "2": {"facts_total": 66, "facts_retained": 21, "retention_rate": 0.3181818181818182, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.35294117647058826, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1508}}}, "reference_key_id": "keys/synthetic/syn_03024_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "34fcc1e23e5a866b5b7d7e222f0380ba771c4005d3732e9747f026c211840d43", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "2b0289cc401be1a010b7970e", "attempt_id": "205c4aa8f5bd", "run_group_id": "E4_core:A4|closed|qwen3b:h2", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:23:03.914060Z", "task_family": "synthetic", "document_id": "syn_03024_2000", "document_hash": "ea10cd11ad08f01c50b0712bc481668857c9d2ddc8e6423968f2a31eeb97ac55", "question_id": "q04_multi_hop", "split": "test", "cluster_id": "syn_03024_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 2, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 2068, "query_tokens": 601, "message_tokens": 512, "message_utf8_bytes": 1508, "cache_tensor_bytes": null, "serialized_payload_bytes": 1508, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 13349.597501102835, "hops": {"1": 6965.8001670613885, "2": 6383.797334041446}, "receiver_prefill_ms": 226.63891699630767, "ttft_ms": 11.959582916460931, "decode_ms": 107.48100001364946}, "total_wall_ms": 13696.246209088713, "ttft_ms": 11.959582916460931, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 9}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "6,009,480", "evidence_artifact": {"ledger": {"facts_total": 66, "facts_retained": 21, "retention_rate": 0.3181818181818182, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.35294117647058826, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1508}, "ledger_by_hop": {"1": {"facts_total": 66, "facts_retained": 21, "retention_rate": 0.3181818181818182, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.35294117647058826, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1508}, "2": {"facts_total": 66, "facts_retained": 21, "retention_rate": 0.3181818181818182, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.35294117647058826, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1508}}}, "reference_key_id": "keys/synthetic/syn_03024_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "34fcc1e23e5a866b5b7d7e222f0380ba771c4005d3732e9747f026c211840d43", "output_hash": "30b20d9aae4d7b0587edeefdc2a9a159f73faf6e2570f1d408a415fe92cbf45e"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "49d09c2a99ad0f6c510fe16c", "attempt_id": "333621cb6a37", "run_group_id": "E4_core:A4|closed|qwen3b:h2", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:23:04.277096Z", "task_family": "synthetic", "document_id": "syn_03024_2000", "document_hash": "ea10cd11ad08f01c50b0712bc481668857c9d2ddc8e6423968f2a31eeb97ac55", "question_id": "q05_multi_hop", "split": "test", "cluster_id": "syn_03024_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 2, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 2068, "query_tokens": 605, "message_tokens": 512, "message_utf8_bytes": 1508, "cache_tensor_bytes": null, "serialized_payload_bytes": 1508, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 13349.597501102835, "hops": {"1": 6965.8001670613885, "2": 6383.797334041446}, "receiver_prefill_ms": 226.90795792732388, "ttft_ms": 11.956249945797026, "decode_ms": 12.165375053882599}, "total_wall_ms": 13601.181835052557, "ttft_ms": 11.956249945797026, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 66, "facts_retained": 21, "retention_rate": 0.3181818181818182, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.35294117647058826, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1508}, "ledger_by_hop": {"1": {"facts_total": 66, "facts_retained": 21, "retention_rate": 0.3181818181818182, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.35294117647058826, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1508}, "2": {"facts_total": 66, "facts_retained": 21, "retention_rate": 0.3181818181818182, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.35294117647058826, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1508}}}, "reference_key_id": "keys/synthetic/syn_03024_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "34fcc1e23e5a866b5b7d7e222f0380ba771c4005d3732e9747f026c211840d43", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "e771788796b151bfd01affc6", "attempt_id": "adc35ad88ada", "run_group_id": "E4_core:A4|closed|qwen3b:h2", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:23:04.543903Z", "task_family": "synthetic", "document_id": "syn_03024_2000", "document_hash": "ea10cd11ad08f01c50b0712bc481668857c9d2ddc8e6423968f2a31eeb97ac55", "question_id": "q06_negation", "split": "test", "cluster_id": "syn_03024_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 2, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 2068, "query_tokens": 594, "message_tokens": 512, "message_utf8_bytes": 1508, "cache_tensor_bytes": null, "serialized_payload_bytes": 1508, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 13349.597501102835, "hops": {"1": 6965.8001670613885, "2": 6383.797334041446}, "receiver_prefill_ms": 226.6862919786945, "ttft_ms": 11.993833002634346, "decode_ms": 11.997084016911685}, "total_wall_ms": 13600.79825099092, "ttft_ms": 11.993833002634346, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "No", "evidence_artifact": {"ledger": {"facts_total": 66, "facts_retained": 21, "retention_rate": 0.3181818181818182, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.35294117647058826, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1508}, "ledger_by_hop": {"1": {"facts_total": 66, "facts_retained": 21, "retention_rate": 0.3181818181818182, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.35294117647058826, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1508}, "2": {"facts_total": 66, "facts_retained": 21, "retention_rate": 0.3181818181818182, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.35294117647058826, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1508}}}, "reference_key_id": "keys/synthetic/syn_03024_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "34fcc1e23e5a866b5b7d7e222f0380ba771c4005d3732e9747f026c211840d43", "output_hash": "1ea442a134b2a184bd5d40104401f2a37fbc09ccf3f4bc9da161c6099be3691d"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "a73dc4b558ae698227c1b34c", "attempt_id": "4fb299208db8", "run_group_id": "E4_core:A4|closed|qwen3b:h2", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:23:04.810395Z", "task_family": "synthetic", "document_id": "syn_03024_2000", "document_hash": "ea10cd11ad08f01c50b0712bc481668857c9d2ddc8e6423968f2a31eeb97ac55", "question_id": "q07_negation", "split": "test", "cluster_id": "syn_03024_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 2, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 2068, "query_tokens": 596, "message_tokens": 512, "message_utf8_bytes": 1508, "cache_tensor_bytes": null, "serialized_payload_bytes": 1508, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 13349.597501102835, "hops": {"1": 6965.8001670613885, "2": 6383.797334041446}, "receiver_prefill_ms": 226.766207953915, "ttft_ms": 11.945624952204525, "decode_ms": 11.92041696049273}, "total_wall_ms": 13600.741876056418, "ttft_ms": 11.945624952204525, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "no", "evidence_artifact": {"ledger": {"facts_total": 66, "facts_retained": 21, "retention_rate": 0.3181818181818182, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.35294117647058826, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1508}, "ledger_by_hop": {"1": {"facts_total": 66, "facts_retained": 21, "retention_rate": 0.3181818181818182, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.35294117647058826, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1508}, "2": {"facts_total": 66, "facts_retained": 21, "retention_rate": 0.3181818181818182, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.35294117647058826, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1508}}}, "reference_key_id": "keys/synthetic/syn_03024_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "34fcc1e23e5a866b5b7d7e222f0380ba771c4005d3732e9747f026c211840d43", "output_hash": "9390298f3fb0c5b160498935d79cb139aef28e1c47358b4bbba61862b9c26e59"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "6779ed992be9c3bd7594f65c", "attempt_id": "be021fa9a18a", "run_group_id": "E4_core:A4|closed|qwen3b:h2", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:23:05.076781Z", "task_family": "synthetic", "document_id": "syn_03024_2000", "document_hash": "ea10cd11ad08f01c50b0712bc481668857c9d2ddc8e6423968f2a31eeb97ac55", "question_id": "q08_revision", "split": "test", "cluster_id": "syn_03024_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 2, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 2068, "query_tokens": 612, "message_tokens": 512, "message_utf8_bytes": 1508, "cache_tensor_bytes": null, "serialized_payload_bytes": 1508, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 13349.597501102835, "hops": {"1": 6965.8001670613885, "2": 6383.797334041446}, "receiver_prefill_ms": 229.57179101649672, "ttft_ms": 12.232957989908755, "decode_ms": 11.908375076018274}, "total_wall_ms": 13603.986417059787, "ttft_ms": 12.232957989908755, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 66, "facts_retained": 21, "retention_rate": 0.3181818181818182, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.35294117647058826, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1508}, "ledger_by_hop": {"1": {"facts_total": 66, "facts_retained": 21, "retention_rate": 0.3181818181818182, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.35294117647058826, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1508}, "2": {"facts_total": 66, "facts_retained": 21, "retention_rate": 0.3181818181818182, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.35294117647058826, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1508}}}, "reference_key_id": "keys/synthetic/syn_03024_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "34fcc1e23e5a866b5b7d7e222f0380ba771c4005d3732e9747f026c211840d43", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "9df0f4564d4b81ae7ac37146", "attempt_id": "54eec9ad4c02", "run_group_id": "E4_core:A4|closed|qwen3b:h2", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:23:05.346551Z", "task_family": "synthetic", "document_id": "syn_03024_2000", "document_hash": "ea10cd11ad08f01c50b0712bc481668857c9d2ddc8e6423968f2a31eeb97ac55", "question_id": "q09_revision", "split": "test", "cluster_id": "syn_03024_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 2, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 2068, "query_tokens": 603, "message_tokens": 512, "message_utf8_bytes": 1508, "cache_tensor_bytes": null, "serialized_payload_bytes": 1508, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 13349.597501102835, "hops": {"1": 6965.8001670613885, "2": 6383.797334041446}, "receiver_prefill_ms": 227.05583297647536, "ttft_ms": 12.183125014416873, "decode_ms": 11.90149993635714}, "total_wall_ms": 13601.28316807095, "ttft_ms": 12.183125014416873, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 66, "facts_retained": 21, "retention_rate": 0.3181818181818182, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.35294117647058826, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1508}, "ledger_by_hop": {"1": {"facts_total": 66, "facts_retained": 21, "retention_rate": 0.3181818181818182, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.35294117647058826, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1508}, "2": {"facts_total": 66, "facts_retained": 21, "retention_rate": 0.3181818181818182, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.35294117647058826, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1508}}}, "reference_key_id": "keys/synthetic/syn_03024_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "34fcc1e23e5a866b5b7d7e222f0380ba771c4005d3732e9747f026c211840d43", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "75d767fc490e28cee390633b", "attempt_id": "7ea57903743a", "run_group_id": "E4_core:A4|closed|qwen3b:h2", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:23:05.613935Z", "task_family": "synthetic", "document_id": "syn_03024_2000", "document_hash": "ea10cd11ad08f01c50b0712bc481668857c9d2ddc8e6423968f2a31eeb97ac55", "question_id": "q10_temporal", "split": "test", "cluster_id": "syn_03024_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 2, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 2068, "query_tokens": 613, "message_tokens": 512, "message_utf8_bytes": 1508, "cache_tensor_bytes": null, "serialized_payload_bytes": 1508, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 13349.597501102835, "hops": {"1": 6965.8001670613885, "2": 6383.797334041446}, "receiver_prefill_ms": 229.49620801955462, "ttft_ms": 11.895584058947861, "decode_ms": 11.921458994038403}, "total_wall_ms": 13603.455210104585, "ttft_ms": 11.895584058947861, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 66, "facts_retained": 21, "retention_rate": 0.3181818181818182, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.35294117647058826, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1508}, "ledger_by_hop": {"1": {"facts_total": 66, "facts_retained": 21, "retention_rate": 0.3181818181818182, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.35294117647058826, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1508}, "2": {"facts_total": 66, "facts_retained": 21, "retention_rate": 0.3181818181818182, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.35294117647058826, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1508}}}, "reference_key_id": "keys/synthetic/syn_03024_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "34fcc1e23e5a866b5b7d7e222f0380ba771c4005d3732e9747f026c211840d43", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "ba97a7bffcd10007b5c6a840", "attempt_id": "e914bfbd5cd2", "run_group_id": "E4_core:A4|closed|qwen3b:h2", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:23:05.883243Z", "task_family": "synthetic", "document_id": "syn_03024_2000", "document_hash": "ea10cd11ad08f01c50b0712bc481668857c9d2ddc8e6423968f2a31eeb97ac55", "question_id": "q11_aggregation", "split": "test", "cluster_id": "syn_03024_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 2, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 2068, "query_tokens": 606, "message_tokens": 512, "message_utf8_bytes": 1508, "cache_tensor_bytes": null, "serialized_payload_bytes": 1508, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 13349.597501102835, "hops": {"1": 6965.8001670613885, "2": 6383.797334041446}, "receiver_prefill_ms": 227.29645809158683, "ttft_ms": 11.892999988049269, "decode_ms": 83.18300009705126}, "total_wall_ms": 13672.52783512231, "ttft_ms": 11.892999988049269, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 7}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "2707654", "evidence_artifact": {"ledger": {"facts_total": 66, "facts_retained": 21, "retention_rate": 0.3181818181818182, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.35294117647058826, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1508}, "ledger_by_hop": {"1": {"facts_total": 66, "facts_retained": 21, "retention_rate": 0.3181818181818182, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.35294117647058826, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1508}, "2": {"facts_total": 66, "facts_retained": 21, "retention_rate": 0.3181818181818182, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.35294117647058826, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1508}}}, "reference_key_id": "keys/synthetic/syn_03024_2000.json", "task_scores": {"correct": 0.0, "abstained": 0.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "34fcc1e23e5a866b5b7d7e222f0380ba771c4005d3732e9747f026c211840d43", "output_hash": "7799e805760fa6568574360f61d4c7cc9e162842191ff434dfed71076922855e"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "02bd6a199cfb3648b21e2001", "attempt_id": "2cd7243371e7", "run_group_id": "E4_core:A4|closed|qwen3b:h2", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:23:06.221612Z", "task_family": "synthetic", "document_id": "syn_03024_2000", "document_hash": "ea10cd11ad08f01c50b0712bc481668857c9d2ddc8e6423968f2a31eeb97ac55", "question_id": "q12_unit_conversion", "split": "test", "cluster_id": "syn_03024_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 2, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 2068, "query_tokens": 603, "message_tokens": 512, "message_utf8_bytes": 1508, "cache_tensor_bytes": null, "serialized_payload_bytes": 1508, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 13349.597501102835, "hops": {"1": 6965.8001670613885, "2": 6383.797334041446}, "receiver_prefill_ms": 227.43054199963808, "ttft_ms": 11.801416985690594, "decode_ms": 11.833208962343633}, "total_wall_ms": 13601.18412610609, "ttft_ms": 11.801416985690594, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 66, "facts_retained": 21, "retention_rate": 0.3181818181818182, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.35294117647058826, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1508}, "ledger_by_hop": {"1": {"facts_total": 66, "facts_retained": 21, "retention_rate": 0.3181818181818182, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.35294117647058826, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1508}, "2": {"facts_total": 66, "facts_retained": 21, "retention_rate": 0.3181818181818182, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.35294117647058826, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1508}}}, "reference_key_id": "keys/synthetic/syn_03024_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "34fcc1e23e5a866b5b7d7e222f0380ba771c4005d3732e9747f026c211840d43", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "0e3ca81812ba9a2213a67243", "attempt_id": "f02994db75e3", "run_group_id": "E4_core:A4|closed|qwen3b:h2", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:23:06.488573Z", "task_family": "synthetic", "document_id": "syn_03024_2000", "document_hash": "ea10cd11ad08f01c50b0712bc481668857c9d2ddc8e6423968f2a31eeb97ac55", "question_id": "q13_unanswerable", "split": "test", "cluster_id": "syn_03024_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 2, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 2068, "query_tokens": 607, "message_tokens": 512, "message_utf8_bytes": 1508, "cache_tensor_bytes": null, "serialized_payload_bytes": 1508, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 13349.597501102835, "hops": {"1": 6965.8001670613885, "2": 6383.797334041446}, "receiver_prefill_ms": 227.21154196187854, "ttft_ms": 11.821249965578318, "decode_ms": 11.791165918111801}, "total_wall_ms": 13600.940084084868, "ttft_ms": 11.821249965578318, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 66, "facts_retained": 21, "retention_rate": 0.3181818181818182, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.35294117647058826, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1508}, "ledger_by_hop": {"1": {"facts_total": 66, "facts_retained": 21, "retention_rate": 0.3181818181818182, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.35294117647058826, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1508}, "2": {"facts_total": 66, "facts_retained": 21, "retention_rate": 0.3181818181818182, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.35294117647058826, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1508}}}, "reference_key_id": "keys/synthetic/syn_03024_2000.json", "task_scores": {"correct": 1.0, "abstained": 1.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "34fcc1e23e5a866b5b7d7e222f0380ba771c4005d3732e9747f026c211840d43", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "77a55ff61a5cdb2f879b907c", "attempt_id": "7ff7f01f492d", "run_group_id": "E4_core:A4|closed|qwen3b:h2", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:23:06.754885Z", "task_family": "synthetic", "document_id": "syn_03024_2000", "document_hash": "ea10cd11ad08f01c50b0712bc481668857c9d2ddc8e6423968f2a31eeb97ac55", "question_id": "q14_unanswerable", "split": "test", "cluster_id": "syn_03024_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 2, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 2068, "query_tokens": 608, "message_tokens": 512, "message_utf8_bytes": 1508, "cache_tensor_bytes": null, "serialized_payload_bytes": 1508, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 13349.597501102835, "hops": {"1": 6965.8001670613885, "2": 6383.797334041446}, "receiver_prefill_ms": 227.14358288794756, "ttft_ms": 11.809874908067286, "decode_ms": 11.813458055257797}, "total_wall_ms": 13600.895918090828, "ttft_ms": 11.809874908067286, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 66, "facts_retained": 21, "retention_rate": 0.3181818181818182, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.35294117647058826, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1508}, "ledger_by_hop": {"1": {"facts_total": 66, "facts_retained": 21, "retention_rate": 0.3181818181818182, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.35294117647058826, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1508}, "2": {"facts_total": 66, "facts_retained": 21, "retention_rate": 0.3181818181818182, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.35294117647058826, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1508}}}, "reference_key_id": "keys/synthetic/syn_03024_2000.json", "task_scores": {"correct": 1.0, "abstained": 1.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "34fcc1e23e5a866b5b7d7e222f0380ba771c4005d3732e9747f026c211840d43", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "726fbb3883c8b239a09d00f6", "attempt_id": "8bd825633ac5", "run_group_id": "E4_core:A4|closed|qwen3b:h4", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:23:19.667794Z", "task_family": "synthetic", "document_id": "syn_03024_2000", "document_hash": "ea10cd11ad08f01c50b0712bc481668857c9d2ddc8e6423968f2a31eeb97ac55", "question_id": "q00_single_fact", "split": "test", "cluster_id": "syn_03024_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 4, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 2068, "query_tokens": 595, "message_tokens": 512, "message_utf8_bytes": 1508, "cache_tensor_bytes": null, "serialized_payload_bytes": 1508, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 25992.931168177165, "hops": {"1": 6965.8001670613885, "2": 6383.797334041446, "3": 6320.953875081614, "4": 6322.3797919927165}, "receiver_prefill_ms": 226.7713330220431, "ttft_ms": 11.888499953784049, "decode_ms": 11.843792046420276}, "total_wall_ms": 26244.002543156967, "ttft_ms": 11.888499953784049, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 66, "facts_retained": 21, "retention_rate": 0.3181818181818182, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.35294117647058826, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1508}, "ledger_by_hop": {"1": {"facts_total": 66, "facts_retained": 21, "retention_rate": 0.3181818181818182, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.35294117647058826, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1508}, "2": {"facts_total": 66, "facts_retained": 21, "retention_rate": 0.3181818181818182, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.35294117647058826, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1508}, "3": {"facts_total": 66, "facts_retained": 21, "retention_rate": 0.3181818181818182, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.35294117647058826, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1508}, "4": {"facts_total": 66, "facts_retained": 21, "retention_rate": 0.3181818181818182, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.35294117647058826, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1508}}}, "reference_key_id": "keys/synthetic/syn_03024_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "34fcc1e23e5a866b5b7d7e222f0380ba771c4005d3732e9747f026c211840d43", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "408729447d65b602d16fda14", "attempt_id": "a416a32a0131", "run_group_id": "E4_core:A4|closed|qwen3b:h4", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:23:19.935054Z", "task_family": "synthetic", "document_id": "syn_03024_2000", "document_hash": "ea10cd11ad08f01c50b0712bc481668857c9d2ddc8e6423968f2a31eeb97ac55", "question_id": "q01_single_fact", "split": "test", "cluster_id": "syn_03024_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 4, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 2068, "query_tokens": 600, "message_tokens": 512, "message_utf8_bytes": 1508, "cache_tensor_bytes": null, "serialized_payload_bytes": 1508, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 25992.931168177165, "hops": {"1": 6965.8001670613885, "2": 6383.797334041446, "3": 6320.953875081614, "4": 6322.3797919927165}, "receiver_prefill_ms": 227.02529199887067, "ttft_ms": 11.941000004298985, "decode_ms": 11.769249918870628}, "total_wall_ms": 26244.22429315746, "ttft_ms": 11.941000004298985, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 66, "facts_retained": 21, "retention_rate": 0.3181818181818182, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.35294117647058826, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1508}, "ledger_by_hop": {"1": {"facts_total": 66, "facts_retained": 21, "retention_rate": 0.3181818181818182, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.35294117647058826, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1508}, "2": {"facts_total": 66, "facts_retained": 21, "retention_rate": 0.3181818181818182, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.35294117647058826, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1508}, "3": {"facts_total": 66, "facts_retained": 21, "retention_rate": 0.3181818181818182, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.35294117647058826, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1508}, "4": {"facts_total": 66, "facts_retained": 21, "retention_rate": 0.3181818181818182, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.35294117647058826, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1508}}}, "reference_key_id": "keys/synthetic/syn_03024_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "34fcc1e23e5a866b5b7d7e222f0380ba771c4005d3732e9747f026c211840d43", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "8118c69625bce55bc1b63621", "attempt_id": "5e81113ab588", "run_group_id": "E4_core:A4|closed|qwen3b:h4", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:23:20.202408Z", "task_family": "synthetic", "document_id": "syn_03024_2000", "document_hash": "ea10cd11ad08f01c50b0712bc481668857c9d2ddc8e6423968f2a31eeb97ac55", "question_id": "q02_exact_copy", "split": "test", "cluster_id": "syn_03024_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 4, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 2068, "query_tokens": 600, "message_tokens": 512, "message_utf8_bytes": 1508, "cache_tensor_bytes": null, "serialized_payload_bytes": 1508, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 25992.931168177165, "hops": {"1": 6965.8001670613885, "2": 6383.797334041446, "3": 6320.953875081614, "4": 6322.3797919927165}, "receiver_prefill_ms": 227.03762503806502, "ttft_ms": 11.819500010460615, "decode_ms": 82.41174998693168}, "total_wall_ms": 26314.70904313028, "ttft_ms": 11.819500010460615, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 7}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "30.0224", "evidence_artifact": {"ledger": {"facts_total": 66, "facts_retained": 21, "retention_rate": 0.3181818181818182, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.35294117647058826, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1508}, "ledger_by_hop": {"1": {"facts_total": 66, "facts_retained": 21, "retention_rate": 0.3181818181818182, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.35294117647058826, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1508}, "2": {"facts_total": 66, "facts_retained": 21, "retention_rate": 0.3181818181818182, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.35294117647058826, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1508}, "3": {"facts_total": 66, "facts_retained": 21, "retention_rate": 0.3181818181818182, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.35294117647058826, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1508}, "4": {"facts_total": 66, "facts_retained": 21, "retention_rate": 0.3181818181818182, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.35294117647058826, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1508}}}, "reference_key_id": "keys/synthetic/syn_03024_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "34fcc1e23e5a866b5b7d7e222f0380ba771c4005d3732e9747f026c211840d43", "output_hash": "e28a78bc4deb3cec8beab4bf5b87297d7cc55a3e6f839b43fe6408735d464923"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "b41c2cd136f522f97551146f", "attempt_id": "53ab612bb802", "run_group_id": "E4_core:A4|closed|qwen3b:h4", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:23:20.539873Z", "task_family": "synthetic", "document_id": "syn_03024_2000", "document_hash": "ea10cd11ad08f01c50b0712bc481668857c9d2ddc8e6423968f2a31eeb97ac55", "question_id": "q03_exact_copy", "split": "test", "cluster_id": "syn_03024_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 4, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 2068, "query_tokens": 595, "message_tokens": 512, "message_utf8_bytes": 1508, "cache_tensor_bytes": null, "serialized_payload_bytes": 1508, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 25992.931168177165, "hops": {"1": 6965.8001670613885, "2": 6383.797334041446, "3": 6320.953875081614, "4": 6322.3797919927165}, "receiver_prefill_ms": 226.2377500301227, "ttft_ms": 11.864291969686747, "decode_ms": 11.791291064582765}, "total_wall_ms": 26243.310668156482, "ttft_ms": 11.864291969686747, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 66, "facts_retained": 21, "retention_rate": 0.3181818181818182, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.35294117647058826, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1508}, "ledger_by_hop": {"1": {"facts_total": 66, "facts_retained": 21, "retention_rate": 0.3181818181818182, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.35294117647058826, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1508}, "2": {"facts_total": 66, "facts_retained": 21, "retention_rate": 0.3181818181818182, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.35294117647058826, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1508}, "3": {"facts_total": 66, "facts_retained": 21, "retention_rate": 0.3181818181818182, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.35294117647058826, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1508}, "4": {"facts_total": 66, "facts_retained": 21, "retention_rate": 0.3181818181818182, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.35294117647058826, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1508}}}, "reference_key_id": "keys/synthetic/syn_03024_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "34fcc1e23e5a866b5b7d7e222f0380ba771c4005d3732e9747f026c211840d43", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "9264f3eb2fa8e2bc881a3644", "attempt_id": "4cc7500fc678", "run_group_id": "E4_core:A4|closed|qwen3b:h4", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:23:20.805585Z", "task_family": "synthetic", "document_id": "syn_03024_2000", "document_hash": "ea10cd11ad08f01c50b0712bc481668857c9d2ddc8e6423968f2a31eeb97ac55", "question_id": "q04_multi_hop", "split": "test", "cluster_id": "syn_03024_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 4, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 2068, "query_tokens": 601, "message_tokens": 512, "message_utf8_bytes": 1508, "cache_tensor_bytes": null, "serialized_payload_bytes": 1508, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 25992.931168177165, "hops": {"1": 6965.8001670613885, "2": 6383.797334041446, "3": 6320.953875081614, "4": 6322.3797919927165}, "receiver_prefill_ms": 226.80370800662786, "ttft_ms": 11.763916932977736, "decode_ms": 105.7549580000341}, "total_wall_ms": 26337.785501149483, "ttft_ms": 11.763916932977736, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 9}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "6,009,480", "evidence_artifact": {"ledger": {"facts_total": 66, "facts_retained": 21, "retention_rate": 0.3181818181818182, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.35294117647058826, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1508}, "ledger_by_hop": {"1": {"facts_total": 66, "facts_retained": 21, "retention_rate": 0.3181818181818182, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.35294117647058826, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1508}, "2": {"facts_total": 66, "facts_retained": 21, "retention_rate": 0.3181818181818182, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.35294117647058826, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1508}, "3": {"facts_total": 66, "facts_retained": 21, "retention_rate": 0.3181818181818182, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.35294117647058826, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1508}, "4": {"facts_total": 66, "facts_retained": 21, "retention_rate": 0.3181818181818182, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.35294117647058826, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1508}}}, "reference_key_id": "keys/synthetic/syn_03024_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "34fcc1e23e5a866b5b7d7e222f0380ba771c4005d3732e9747f026c211840d43", "output_hash": "30b20d9aae4d7b0587edeefdc2a9a159f73faf6e2570f1d408a415fe92cbf45e"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "379fdd74771035e2eb6e28a3", "attempt_id": "bfabffb1648e", "run_group_id": "E4_core:A4|closed|qwen3b:h4", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:23:21.165503Z", "task_family": "synthetic", "document_id": "syn_03024_2000", "document_hash": "ea10cd11ad08f01c50b0712bc481668857c9d2ddc8e6423968f2a31eeb97ac55", "question_id": "q05_multi_hop", "split": "test", "cluster_id": "syn_03024_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 4, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 2068, "query_tokens": 605, "message_tokens": 512, "message_utf8_bytes": 1508, "cache_tensor_bytes": null, "serialized_payload_bytes": 1508, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 25992.931168177165, "hops": {"1": 6965.8001670613885, "2": 6383.797334041446, "3": 6320.953875081614, "4": 6322.3797919927165}, "receiver_prefill_ms": 226.89308295957744, "ttft_ms": 11.791458004154265, "decode_ms": 11.826207977719605}, "total_wall_ms": 26243.955460144207, "ttft_ms": 11.791458004154265, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 66, "facts_retained": 21, "retention_rate": 0.3181818181818182, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.35294117647058826, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1508}, "ledger_by_hop": {"1": {"facts_total": 66, "facts_retained": 21, "retention_rate": 0.3181818181818182, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.35294117647058826, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1508}, "2": {"facts_total": 66, "facts_retained": 21, "retention_rate": 0.3181818181818182, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.35294117647058826, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1508}, "3": {"facts_total": 66, "facts_retained": 21, "retention_rate": 0.3181818181818182, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.35294117647058826, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1508}, "4": {"facts_total": 66, "facts_retained": 21, "retention_rate": 0.3181818181818182, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.35294117647058826, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1508}}}, "reference_key_id": "keys/synthetic/syn_03024_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "34fcc1e23e5a866b5b7d7e222f0380ba771c4005d3732e9747f026c211840d43", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "a1c0cc02f0feb55977a59249", "attempt_id": "4b5e754b980d", "run_group_id": "E4_core:A4|closed|qwen3b:h4", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:23:21.431760Z", "task_family": "synthetic", "document_id": "syn_03024_2000", "document_hash": "ea10cd11ad08f01c50b0712bc481668857c9d2ddc8e6423968f2a31eeb97ac55", "question_id": "q06_negation", "split": "test", "cluster_id": "syn_03024_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 4, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 2068, "query_tokens": 594, "message_tokens": 512, "message_utf8_bytes": 1508, "cache_tensor_bytes": null, "serialized_payload_bytes": 1508, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 25992.931168177165, "hops": {"1": 6965.8001670613885, "2": 6383.797334041446, "3": 6320.953875081614, "4": 6322.3797919927165}, "receiver_prefill_ms": 226.97441698983312, "ttft_ms": 11.828124988824129, "decode_ms": 11.864541913382709}, "total_wall_ms": 26244.10158523824, "ttft_ms": 11.828124988824129, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "No", "evidence_artifact": {"ledger": {"facts_total": 66, "facts_retained": 21, "retention_rate": 0.3181818181818182, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.35294117647058826, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1508}, "ledger_by_hop": {"1": {"facts_total": 66, "facts_retained": 21, "retention_rate": 0.3181818181818182, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.35294117647058826, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1508}, "2": {"facts_total": 66, "facts_retained": 21, "retention_rate": 0.3181818181818182, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.35294117647058826, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1508}, "3": {"facts_total": 66, "facts_retained": 21, "retention_rate": 0.3181818181818182, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.35294117647058826, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1508}, "4": {"facts_total": 66, "facts_retained": 21, "retention_rate": 0.3181818181818182, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.35294117647058826, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1508}}}, "reference_key_id": "keys/synthetic/syn_03024_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "34fcc1e23e5a866b5b7d7e222f0380ba771c4005d3732e9747f026c211840d43", "output_hash": "1ea442a134b2a184bd5d40104401f2a37fbc09ccf3f4bc9da161c6099be3691d"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "cfd7f566c9336b798f169eef", "attempt_id": "0d87558657c2", "run_group_id": "E4_core:A4|closed|qwen3b:h4", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:23:21.698172Z", "task_family": "synthetic", "document_id": "syn_03024_2000", "document_hash": "ea10cd11ad08f01c50b0712bc481668857c9d2ddc8e6423968f2a31eeb97ac55", "question_id": "q07_negation", "split": "test", "cluster_id": "syn_03024_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 4, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 2068, "query_tokens": 596, "message_tokens": 512, "message_utf8_bytes": 1508, "cache_tensor_bytes": null, "serialized_payload_bytes": 1508, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 25992.931168177165, "hops": {"1": 6965.8001670613885, "2": 6383.797334041446, "3": 6320.953875081614, "4": 6322.3797919927165}, "receiver_prefill_ms": 227.1051249699667, "ttft_ms": 11.855500051751733, "decode_ms": 11.904416023753583}, "total_wall_ms": 26244.29912713822, "ttft_ms": 11.855500051751733, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "no", "evidence_artifact": {"ledger": {"facts_total": 66, "facts_retained": 21, "retention_rate": 0.3181818181818182, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.35294117647058826, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1508}, "ledger_by_hop": {"1": {"facts_total": 66, "facts_retained": 21, "retention_rate": 0.3181818181818182, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.35294117647058826, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1508}, "2": {"facts_total": 66, "facts_retained": 21, "retention_rate": 0.3181818181818182, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.35294117647058826, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1508}, "3": {"facts_total": 66, "facts_retained": 21, "retention_rate": 0.3181818181818182, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.35294117647058826, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1508}, "4": {"facts_total": 66, "facts_retained": 21, "retention_rate": 0.3181818181818182, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.35294117647058826, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1508}}}, "reference_key_id": "keys/synthetic/syn_03024_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "34fcc1e23e5a866b5b7d7e222f0380ba771c4005d3732e9747f026c211840d43", "output_hash": "9390298f3fb0c5b160498935d79cb139aef28e1c47358b4bbba61862b9c26e59"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "a21f1bbe758ee8cef8f172c3", "attempt_id": "2cbd47ca92e9", "run_group_id": "E4_core:A4|closed|qwen3b:h4", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:23:21.965020Z", "task_family": "synthetic", "document_id": "syn_03024_2000", "document_hash": "ea10cd11ad08f01c50b0712bc481668857c9d2ddc8e6423968f2a31eeb97ac55", "question_id": "q08_revision", "split": "test", "cluster_id": "syn_03024_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 4, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 2068, "query_tokens": 612, "message_tokens": 512, "message_utf8_bytes": 1508, "cache_tensor_bytes": null, "serialized_payload_bytes": 1508, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 25992.931168177165, "hops": {"1": 6965.8001670613885, "2": 6383.797334041446, "3": 6320.953875081614, "4": 6322.3797919927165}, "receiver_prefill_ms": 229.09416700713336, "ttft_ms": 11.718250112608075, "decode_ms": 11.877542012371123}, "total_wall_ms": 26246.146918158047, "ttft_ms": 11.718250112608075, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 66, "facts_retained": 21, "retention_rate": 0.3181818181818182, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.35294117647058826, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1508}, "ledger_by_hop": {"1": {"facts_total": 66, "facts_retained": 21, "retention_rate": 0.3181818181818182, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.35294117647058826, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1508}, "2": {"facts_total": 66, "facts_retained": 21, "retention_rate": 0.3181818181818182, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.35294117647058826, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1508}, "3": {"facts_total": 66, "facts_retained": 21, "retention_rate": 0.3181818181818182, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.35294117647058826, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1508}, "4": {"facts_total": 66, "facts_retained": 21, "retention_rate": 0.3181818181818182, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.35294117647058826, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1508}}}, "reference_key_id": "keys/synthetic/syn_03024_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "34fcc1e23e5a866b5b7d7e222f0380ba771c4005d3732e9747f026c211840d43", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "58d9a604baa62d519acba4ee", "attempt_id": "de7fc439cc86", "run_group_id": "E4_core:A4|closed|qwen3b:h4", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:23:22.233270Z", "task_family": "synthetic", "document_id": "syn_03024_2000", "document_hash": "ea10cd11ad08f01c50b0712bc481668857c9d2ddc8e6423968f2a31eeb97ac55", "question_id": "q09_revision", "split": "test", "cluster_id": "syn_03024_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 4, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 2068, "query_tokens": 603, "message_tokens": 512, "message_utf8_bytes": 1508, "cache_tensor_bytes": null, "serialized_payload_bytes": 1508, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 25992.931168177165, "hops": {"1": 6965.8001670613885, "2": 6383.797334041446, "3": 6320.953875081614, "4": 6322.3797919927165}, "receiver_prefill_ms": 226.9183750031516, "ttft_ms": 11.839916929602623, "decode_ms": 11.85870892368257}, "total_wall_ms": 26244.057793170214, "ttft_ms": 11.839916929602623, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 66, "facts_retained": 21, "retention_rate": 0.3181818181818182, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.35294117647058826, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1508}, "ledger_by_hop": {"1": {"facts_total": 66, "facts_retained": 21, "retention_rate": 0.3181818181818182, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.35294117647058826, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1508}, "2": {"facts_total": 66, "facts_retained": 21, "retention_rate": 0.3181818181818182, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.35294117647058826, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1508}, "3": {"facts_total": 66, "facts_retained": 21, "retention_rate": 0.3181818181818182, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.35294117647058826, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1508}, "4": {"facts_total": 66, "facts_retained": 21, "retention_rate": 0.3181818181818182, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.35294117647058826, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1508}}}, "reference_key_id": "keys/synthetic/syn_03024_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "34fcc1e23e5a866b5b7d7e222f0380ba771c4005d3732e9747f026c211840d43", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "73c3858f29e819400209778f", "attempt_id": "a32525a59de5", "run_group_id": "E4_core:A4|closed|qwen3b:h4", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:23:22.500143Z", "task_family": "synthetic", "document_id": "syn_03024_2000", "document_hash": "ea10cd11ad08f01c50b0712bc481668857c9d2ddc8e6423968f2a31eeb97ac55", "question_id": "q10_temporal", "split": "test", "cluster_id": "syn_03024_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 4, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 2068, "query_tokens": 613, "message_tokens": 512, "message_utf8_bytes": 1508, "cache_tensor_bytes": null, "serialized_payload_bytes": 1508, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 25992.931168177165, "hops": {"1": 6965.8001670613885, "2": 6383.797334041446, "3": 6320.953875081614, "4": 6322.3797919927165}, "receiver_prefill_ms": 229.04245904646814, "ttft_ms": 11.789832962676883, "decode_ms": 11.730375001206994}, "total_wall_ms": 26246.00646016188, "ttft_ms": 11.789832962676883, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 66, "facts_retained": 21, "retention_rate": 0.3181818181818182, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.35294117647058826, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1508}, "ledger_by_hop": {"1": {"facts_total": 66, "facts_retained": 21, "retention_rate": 0.3181818181818182, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.35294117647058826, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1508}, "2": {"facts_total": 66, "facts_retained": 21, "retention_rate": 0.3181818181818182, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.35294117647058826, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1508}, "3": {"facts_total": 66, "facts_retained": 21, "retention_rate": 0.3181818181818182, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.35294117647058826, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1508}, "4": {"facts_total": 66, "facts_retained": 21, "retention_rate": 0.3181818181818182, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.35294117647058826, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1508}}}, "reference_key_id": "keys/synthetic/syn_03024_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "34fcc1e23e5a866b5b7d7e222f0380ba771c4005d3732e9747f026c211840d43", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "58d2d7acf81624a6f0c6dda3", "attempt_id": "c98af0ed5f82", "run_group_id": "E4_core:A4|closed|qwen3b:h4", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:23:22.768782Z", "task_family": "synthetic", "document_id": "syn_03024_2000", "document_hash": "ea10cd11ad08f01c50b0712bc481668857c9d2ddc8e6423968f2a31eeb97ac55", "question_id": "q11_aggregation", "split": "test", "cluster_id": "syn_03024_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 4, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 2068, "query_tokens": 606, "message_tokens": 512, "message_utf8_bytes": 1508, "cache_tensor_bytes": null, "serialized_payload_bytes": 1508, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 25992.931168177165, "hops": {"1": 6965.8001670613885, "2": 6383.797334041446, "3": 6320.953875081614, "4": 6322.3797919927165}, "receiver_prefill_ms": 227.17649990227073, "ttft_ms": 12.067832984030247, "decode_ms": 82.63779198750854}, "total_wall_ms": 26315.369501244277, "ttft_ms": 12.067832984030247, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 7}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "2707654", "evidence_artifact": {"ledger": {"facts_total": 66, "facts_retained": 21, "retention_rate": 0.3181818181818182, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.35294117647058826, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1508}, "ledger_by_hop": {"1": {"facts_total": 66, "facts_retained": 21, "retention_rate": 0.3181818181818182, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.35294117647058826, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1508}, "2": {"facts_total": 66, "facts_retained": 21, "retention_rate": 0.3181818181818182, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.35294117647058826, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1508}, "3": {"facts_total": 66, "facts_retained": 21, "retention_rate": 0.3181818181818182, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.35294117647058826, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1508}, "4": {"facts_total": 66, "facts_retained": 21, "retention_rate": 0.3181818181818182, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.35294117647058826, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1508}}}, "reference_key_id": "keys/synthetic/syn_03024_2000.json", "task_scores": {"correct": 0.0, "abstained": 0.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "34fcc1e23e5a866b5b7d7e222f0380ba771c4005d3732e9747f026c211840d43", "output_hash": "7799e805760fa6568574360f61d4c7cc9e162842191ff434dfed71076922855e"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "724f02747b38012dd4315efb", "attempt_id": "ab1a4f595e42", "run_group_id": "E4_core:A4|closed|qwen3b:h4", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:23:23.107341Z", "task_family": "synthetic", "document_id": "syn_03024_2000", "document_hash": "ea10cd11ad08f01c50b0712bc481668857c9d2ddc8e6423968f2a31eeb97ac55", "question_id": "q12_unit_conversion", "split": "test", "cluster_id": "syn_03024_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 4, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 2068, "query_tokens": 603, "message_tokens": 512, "message_utf8_bytes": 1508, "cache_tensor_bytes": null, "serialized_payload_bytes": 1508, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 25992.931168177165, "hops": {"1": 6965.8001670613885, "2": 6383.797334041446, "3": 6320.953875081614, "4": 6322.3797919927165}, "receiver_prefill_ms": 226.6135000390932, "ttft_ms": 11.80862495675683, "decode_ms": 11.871666996739805}, "total_wall_ms": 26243.756584241055, "ttft_ms": 11.80862495675683, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 66, "facts_retained": 21, "retention_rate": 0.3181818181818182, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.35294117647058826, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1508}, "ledger_by_hop": {"1": {"facts_total": 66, "facts_retained": 21, "retention_rate": 0.3181818181818182, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.35294117647058826, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1508}, "2": {"facts_total": 66, "facts_retained": 21, "retention_rate": 0.3181818181818182, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.35294117647058826, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1508}, "3": {"facts_total": 66, "facts_retained": 21, "retention_rate": 0.3181818181818182, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.35294117647058826, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1508}, "4": {"facts_total": 66, "facts_retained": 21, "retention_rate": 0.3181818181818182, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.35294117647058826, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1508}}}, "reference_key_id": "keys/synthetic/syn_03024_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "34fcc1e23e5a866b5b7d7e222f0380ba771c4005d3732e9747f026c211840d43", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "21be50c1530487d70d7108d6", "attempt_id": "e8a6251fe433", "run_group_id": "E4_core:A4|closed|qwen3b:h4", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:23:23.373315Z", "task_family": "synthetic", "document_id": "syn_03024_2000", "document_hash": "ea10cd11ad08f01c50b0712bc481668857c9d2ddc8e6423968f2a31eeb97ac55", "question_id": "q13_unanswerable", "split": "test", "cluster_id": "syn_03024_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 4, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 2068, "query_tokens": 607, "message_tokens": 512, "message_utf8_bytes": 1508, "cache_tensor_bytes": null, "serialized_payload_bytes": 1508, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 25992.931168177165, "hops": {"1": 6965.8001670613885, "2": 6383.797334041446, "3": 6320.953875081614, "4": 6322.3797919927165}, "receiver_prefill_ms": 226.42879199702293, "ttft_ms": 11.861625011079013, "decode_ms": 11.85912499204278}, "total_wall_ms": 26243.610210251063, "ttft_ms": 11.861625011079013, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 66, "facts_retained": 21, "retention_rate": 0.3181818181818182, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.35294117647058826, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1508}, "ledger_by_hop": {"1": {"facts_total": 66, "facts_retained": 21, "retention_rate": 0.3181818181818182, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.35294117647058826, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1508}, "2": {"facts_total": 66, "facts_retained": 21, "retention_rate": 0.3181818181818182, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.35294117647058826, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1508}, "3": {"facts_total": 66, "facts_retained": 21, "retention_rate": 0.3181818181818182, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.35294117647058826, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1508}, "4": {"facts_total": 66, "facts_retained": 21, "retention_rate": 0.3181818181818182, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.35294117647058826, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1508}}}, "reference_key_id": "keys/synthetic/syn_03024_2000.json", "task_scores": {"correct": 1.0, "abstained": 1.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "34fcc1e23e5a866b5b7d7e222f0380ba771c4005d3732e9747f026c211840d43", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "ef5155954acc1c0016fa85be", "attempt_id": "c49f4b2a1761", "run_group_id": "E4_core:A4|closed|qwen3b:h4", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:23:23.639231Z", "task_family": "synthetic", "document_id": "syn_03024_2000", "document_hash": "ea10cd11ad08f01c50b0712bc481668857c9d2ddc8e6423968f2a31eeb97ac55", "question_id": "q14_unanswerable", "split": "test", "cluster_id": "syn_03024_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 4, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 2068, "query_tokens": 608, "message_tokens": 512, "message_utf8_bytes": 1508, "cache_tensor_bytes": null, "serialized_payload_bytes": 1508, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 25992.931168177165, "hops": {"1": 6965.8001670613885, "2": 6383.797334041446, "3": 6320.953875081614, "4": 6322.3797919927165}, "receiver_prefill_ms": 227.08595800213516, "ttft_ms": 12.17612496111542, "decode_ms": 11.82437501847744}, "total_wall_ms": 26244.54262619838, "ttft_ms": 12.17612496111542, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 66, "facts_retained": 21, "retention_rate": 0.3181818181818182, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.35294117647058826, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1508}, "ledger_by_hop": {"1": {"facts_total": 66, "facts_retained": 21, "retention_rate": 0.3181818181818182, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.35294117647058826, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1508}, "2": {"facts_total": 66, "facts_retained": 21, "retention_rate": 0.3181818181818182, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.35294117647058826, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1508}, "3": {"facts_total": 66, "facts_retained": 21, "retention_rate": 0.3181818181818182, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.35294117647058826, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1508}, "4": {"facts_total": 66, "facts_retained": 21, "retention_rate": 0.3181818181818182, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.35294117647058826, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1508}}}, "reference_key_id": "keys/synthetic/syn_03024_2000.json", "task_scores": {"correct": 1.0, "abstained": 1.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "34fcc1e23e5a866b5b7d7e222f0380ba771c4005d3732e9747f026c211840d43", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "751370cd81892dd40de17903", "attempt_id": "769757d4d58d", "run_group_id": "E4_core:A4|closed|qwen3b:h8", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:23:49.213249Z", "task_family": "synthetic", "document_id": "syn_03024_2000", "document_hash": "ea10cd11ad08f01c50b0712bc481668857c9d2ddc8e6423968f2a31eeb97ac55", "question_id": "q00_single_fact", "split": "test", "cluster_id": "syn_03024_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 8, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 2068, "query_tokens": 595, "message_tokens": 512, "message_utf8_bytes": 1508, "cache_tensor_bytes": null, "serialized_payload_bytes": 1508, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 51294.487459235825, "hops": {"1": 6965.8001670613885, "2": 6383.797334041446, "3": 6320.953875081614, "4": 6322.3797919927165, "5": 6315.4535830253735, "6": 6335.9819169854745, "7": 6312.74179101456, "8": 6337.379000033252}, "receiver_prefill_ms": 226.88433295115829, "ttft_ms": 12.047790922224522, "decode_ms": 11.889833025634289}, "total_wall_ms": 51545.84158421494, "ttft_ms": 12.047790922224522, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 66, "facts_retained": 21, "retention_rate": 0.3181818181818182, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.35294117647058826, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1508}, "ledger_by_hop": {"1": {"facts_total": 66, "facts_retained": 21, "retention_rate": 0.3181818181818182, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.35294117647058826, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1508}, "2": {"facts_total": 66, "facts_retained": 21, "retention_rate": 0.3181818181818182, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.35294117647058826, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1508}, "3": {"facts_total": 66, "facts_retained": 21, "retention_rate": 0.3181818181818182, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.35294117647058826, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1508}, "4": {"facts_total": 66, "facts_retained": 21, "retention_rate": 0.3181818181818182, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.35294117647058826, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1508}, "5": {"facts_total": 66, "facts_retained": 21, "retention_rate": 0.3181818181818182, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.35294117647058826, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1508}, "6": {"facts_total": 66, "facts_retained": 21, "retention_rate": 0.3181818181818182, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.35294117647058826, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1508}, "7": {"facts_total": 66, "facts_retained": 21, "retention_rate": 0.3181818181818182, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.35294117647058826, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1508}, "8": {"facts_total": 66, "facts_retained": 21, "retention_rate": 0.3181818181818182, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.35294117647058826, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1508}}}, "reference_key_id": "keys/synthetic/syn_03024_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "34fcc1e23e5a866b5b7d7e222f0380ba771c4005d3732e9747f026c211840d43", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "880b3a3219cb96b64f4ccaf4", "attempt_id": "b087eaeacc35", "run_group_id": "E4_core:A4|closed|qwen3b:h8", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:23:49.480557Z", "task_family": "synthetic", "document_id": "syn_03024_2000", "document_hash": "ea10cd11ad08f01c50b0712bc481668857c9d2ddc8e6423968f2a31eeb97ac55", "question_id": "q01_single_fact", "split": "test", "cluster_id": "syn_03024_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 8, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 2068, "query_tokens": 600, "message_tokens": 512, "message_utf8_bytes": 1508, "cache_tensor_bytes": null, "serialized_payload_bytes": 1508, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 51294.487459235825, "hops": {"1": 6965.8001670613885, "2": 6383.797334041446, "3": 6320.953875081614, "4": 6322.3797919927165, "5": 6315.4535830253735, "6": 6335.9819169854745, "7": 6312.74179101456, "8": 6337.379000033252}, "receiver_prefill_ms": 227.0825420273468, "ttft_ms": 11.990125058218837, "decode_ms": 11.91887492313981}, "total_wall_ms": 51545.9824182326, "ttft_ms": 11.990125058218837, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 66, "facts_retained": 21, "retention_rate": 0.3181818181818182, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.35294117647058826, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1508}, "ledger_by_hop": {"1": {"facts_total": 66, "facts_retained": 21, "retention_rate": 0.3181818181818182, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.35294117647058826, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1508}, "2": {"facts_total": 66, "facts_retained": 21, "retention_rate": 0.3181818181818182, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.35294117647058826, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1508}, "3": {"facts_total": 66, "facts_retained": 21, "retention_rate": 0.3181818181818182, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.35294117647058826, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1508}, "4": {"facts_total": 66, "facts_retained": 21, "retention_rate": 0.3181818181818182, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.35294117647058826, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1508}, "5": {"facts_total": 66, "facts_retained": 21, "retention_rate": 0.3181818181818182, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.35294117647058826, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1508}, "6": {"facts_total": 66, "facts_retained": 21, "retention_rate": 0.3181818181818182, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.35294117647058826, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1508}, "7": {"facts_total": 66, "facts_retained": 21, "retention_rate": 0.3181818181818182, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.35294117647058826, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1508}, "8": {"facts_total": 66, "facts_retained": 21, "retention_rate": 0.3181818181818182, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.35294117647058826, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1508}}}, "reference_key_id": "keys/synthetic/syn_03024_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "34fcc1e23e5a866b5b7d7e222f0380ba771c4005d3732e9747f026c211840d43", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "e1dd3029f476bf30c5bc5707", "attempt_id": "bc96537f758d", "run_group_id": "E4_core:A4|closed|qwen3b:h8", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:23:49.747123Z", "task_family": "synthetic", "document_id": "syn_03024_2000", "document_hash": "ea10cd11ad08f01c50b0712bc481668857c9d2ddc8e6423968f2a31eeb97ac55", "question_id": "q02_exact_copy", "split": "test", "cluster_id": "syn_03024_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 8, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 2068, "query_tokens": 600, "message_tokens": 512, "message_utf8_bytes": 1508, "cache_tensor_bytes": null, "serialized_payload_bytes": 1508, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 51294.487459235825, "hops": {"1": 6965.8001670613885, "2": 6383.797334041446, "3": 6320.953875081614, "4": 6322.3797919927165, "5": 6315.4535830253735, "6": 6335.9819169854745, "7": 6312.74179101456, "8": 6337.379000033252}, "receiver_prefill_ms": 227.089999942109, "ttft_ms": 12.071042088791728, "decode_ms": 82.94683299027383}, "total_wall_ms": 51617.09904228337, "ttft_ms": 12.071042088791728, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 7}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "30.0224", "evidence_artifact": {"ledger": {"facts_total": 66, "facts_retained": 21, "retention_rate": 0.3181818181818182, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.35294117647058826, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1508}, "ledger_by_hop": {"1": {"facts_total": 66, "facts_retained": 21, "retention_rate": 0.3181818181818182, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.35294117647058826, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1508}, "2": {"facts_total": 66, "facts_retained": 21, "retention_rate": 0.3181818181818182, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.35294117647058826, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1508}, "3": {"facts_total": 66, "facts_retained": 21, "retention_rate": 0.3181818181818182, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.35294117647058826, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1508}, "4": {"facts_total": 66, "facts_retained": 21, "retention_rate": 0.3181818181818182, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.35294117647058826, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1508}, "5": {"facts_total": 66, "facts_retained": 21, "retention_rate": 0.3181818181818182, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.35294117647058826, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1508}, "6": {"facts_total": 66, "facts_retained": 21, "retention_rate": 0.3181818181818182, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.35294117647058826, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1508}, "7": {"facts_total": 66, "facts_retained": 21, "retention_rate": 0.3181818181818182, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.35294117647058826, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1508}, "8": {"facts_total": 66, "facts_retained": 21, "retention_rate": 0.3181818181818182, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.35294117647058826, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1508}}}, "reference_key_id": "keys/synthetic/syn_03024_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "34fcc1e23e5a866b5b7d7e222f0380ba771c4005d3732e9747f026c211840d43", "output_hash": "e28a78bc4deb3cec8beab4bf5b87297d7cc55a3e6f839b43fe6408735d464923"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "565e07b4e0655f8bacabac69", "attempt_id": "1b59b9220139", "run_group_id": "E4_core:A4|closed|qwen3b:h8", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:23:50.085031Z", "task_family": "synthetic", "document_id": "syn_03024_2000", "document_hash": "ea10cd11ad08f01c50b0712bc481668857c9d2ddc8e6423968f2a31eeb97ac55", "question_id": "q03_exact_copy", "split": "test", "cluster_id": "syn_03024_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 8, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 2068, "query_tokens": 595, "message_tokens": 512, "message_utf8_bytes": 1508, "cache_tensor_bytes": null, "serialized_payload_bytes": 1508, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 51294.487459235825, "hops": {"1": 6965.8001670613885, "2": 6383.797334041446, "3": 6320.953875081614, "4": 6322.3797919927165, "5": 6315.4535830253735, "6": 6335.9819169854745, "7": 6312.74179101456, "8": 6337.379000033252}, "receiver_prefill_ms": 227.21041610930115, "ttft_ms": 11.838041944429278, "decode_ms": 11.842875042930245}, "total_wall_ms": 51545.88004213292, "ttft_ms": 11.838041944429278, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 66, "facts_retained": 21, "retention_rate": 0.3181818181818182, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.35294117647058826, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1508}, "ledger_by_hop": {"1": {"facts_total": 66, "facts_retained": 21, "retention_rate": 0.3181818181818182, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.35294117647058826, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1508}, "2": {"facts_total": 66, "facts_retained": 21, "retention_rate": 0.3181818181818182, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.35294117647058826, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1508}, "3": {"facts_total": 66, "facts_retained": 21, "retention_rate": 0.3181818181818182, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.35294117647058826, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1508}, "4": {"facts_total": 66, "facts_retained": 21, "retention_rate": 0.3181818181818182, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.35294117647058826, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1508}, "5": {"facts_total": 66, "facts_retained": 21, "retention_rate": 0.3181818181818182, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.35294117647058826, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1508}, "6": {"facts_total": 66, "facts_retained": 21, "retention_rate": 0.3181818181818182, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.35294117647058826, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1508}, "7": {"facts_total": 66, "facts_retained": 21, "retention_rate": 0.3181818181818182, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.35294117647058826, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1508}, "8": {"facts_total": 66, "facts_retained": 21, "retention_rate": 0.3181818181818182, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.35294117647058826, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1508}}}, "reference_key_id": "keys/synthetic/syn_03024_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "34fcc1e23e5a866b5b7d7e222f0380ba771c4005d3732e9747f026c211840d43", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "b6ce8280ab8b1f23b48b50e5", "attempt_id": "25daa4a443a8", "run_group_id": "E4_core:A4|closed|qwen3b:h8", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:23:50.351570Z", "task_family": "synthetic", "document_id": "syn_03024_2000", "document_hash": "ea10cd11ad08f01c50b0712bc481668857c9d2ddc8e6423968f2a31eeb97ac55", "question_id": "q04_multi_hop", "split": "test", "cluster_id": "syn_03024_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 8, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 2068, "query_tokens": 601, "message_tokens": 512, "message_utf8_bytes": 1508, "cache_tensor_bytes": null, "serialized_payload_bytes": 1508, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 51294.487459235825, "hops": {"1": 6965.8001670613885, "2": 6383.797334041446, "3": 6320.953875081614, "4": 6322.3797919927165, "5": 6315.4535830253735, "6": 6335.9819169854745, "7": 6312.74179101456, "8": 6337.379000033252}, "receiver_prefill_ms": 227.19112504273653, "ttft_ms": 11.835957993753254, "decode_ms": 106.63254198152572}, "total_wall_ms": 51640.638959244825, "ttft_ms": 11.835957993753254, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 9}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "6,009,480", "evidence_artifact": {"ledger": {"facts_total": 66, "facts_retained": 21, "retention_rate": 0.3181818181818182, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.35294117647058826, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1508}, "ledger_by_hop": {"1": {"facts_total": 66, "facts_retained": 21, "retention_rate": 0.3181818181818182, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.35294117647058826, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1508}, "2": {"facts_total": 66, "facts_retained": 21, "retention_rate": 0.3181818181818182, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.35294117647058826, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1508}, "3": {"facts_total": 66, "facts_retained": 21, "retention_rate": 0.3181818181818182, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.35294117647058826, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1508}, "4": {"facts_total": 66, "facts_retained": 21, "retention_rate": 0.3181818181818182, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.35294117647058826, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1508}, "5": {"facts_total": 66, "facts_retained": 21, "retention_rate": 0.3181818181818182, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.35294117647058826, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1508}, "6": {"facts_total": 66, "facts_retained": 21, "retention_rate": 0.3181818181818182, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.35294117647058826, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1508}, "7": {"facts_total": 66, "facts_retained": 21, "retention_rate": 0.3181818181818182, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.35294117647058826, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1508}, "8": {"facts_total": 66, "facts_retained": 21, "retention_rate": 0.3181818181818182, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.35294117647058826, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1508}}}, "reference_key_id": "keys/synthetic/syn_03024_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "34fcc1e23e5a866b5b7d7e222f0380ba771c4005d3732e9747f026c211840d43", "output_hash": "30b20d9aae4d7b0587edeefdc2a9a159f73faf6e2570f1d408a415fe92cbf45e"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "8b453767f8031f595750575f", "attempt_id": "a2ea420b2a16", "run_group_id": "E4_core:A4|closed|qwen3b:h8", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:23:50.712977Z", "task_family": "synthetic", "document_id": "syn_03024_2000", "document_hash": "ea10cd11ad08f01c50b0712bc481668857c9d2ddc8e6423968f2a31eeb97ac55", "question_id": "q05_multi_hop", "split": "test", "cluster_id": "syn_03024_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 8, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 2068, "query_tokens": 605, "message_tokens": 512, "message_utf8_bytes": 1508, "cache_tensor_bytes": null, "serialized_payload_bytes": 1508, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 51294.487459235825, "hops": {"1": 6965.8001670613885, "2": 6383.797334041446, "3": 6320.953875081614, "4": 6322.3797919927165, "5": 6315.4535830253735, "6": 6335.9819169854745, "7": 6312.74179101456, "8": 6337.379000033252}, "receiver_prefill_ms": 227.34225005842745, "ttft_ms": 11.902292026206851, "decode_ms": 11.866874992847443}, "total_wall_ms": 51546.0947091924, "ttft_ms": 11.902292026206851, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 66, "facts_retained": 21, "retention_rate": 0.3181818181818182, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.35294117647058826, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1508}, "ledger_by_hop": {"1": {"facts_total": 66, "facts_retained": 21, "retention_rate": 0.3181818181818182, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.35294117647058826, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1508}, "2": {"facts_total": 66, "facts_retained": 21, "retention_rate": 0.3181818181818182, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.35294117647058826, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1508}, "3": {"facts_total": 66, "facts_retained": 21, "retention_rate": 0.3181818181818182, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.35294117647058826, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1508}, "4": {"facts_total": 66, "facts_retained": 21, "retention_rate": 0.3181818181818182, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.35294117647058826, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1508}, "5": {"facts_total": 66, "facts_retained": 21, "retention_rate": 0.3181818181818182, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.35294117647058826, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1508}, "6": {"facts_total": 66, "facts_retained": 21, "retention_rate": 0.3181818181818182, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.35294117647058826, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1508}, "7": {"facts_total": 66, "facts_retained": 21, "retention_rate": 0.3181818181818182, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.35294117647058826, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1508}, "8": {"facts_total": 66, "facts_retained": 21, "retention_rate": 0.3181818181818182, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.35294117647058826, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1508}}}, "reference_key_id": "keys/synthetic/syn_03024_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "34fcc1e23e5a866b5b7d7e222f0380ba771c4005d3732e9747f026c211840d43", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "113e37ab97f0e52acb231f53", "attempt_id": "3d0b15461043", "run_group_id": "E4_core:A4|closed|qwen3b:h8", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:23:50.979840Z", "task_family": "synthetic", "document_id": "syn_03024_2000", "document_hash": "ea10cd11ad08f01c50b0712bc481668857c9d2ddc8e6423968f2a31eeb97ac55", "question_id": "q06_negation", "split": "test", "cluster_id": "syn_03024_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 8, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 2068, "query_tokens": 594, "message_tokens": 512, "message_utf8_bytes": 1508, "cache_tensor_bytes": null, "serialized_payload_bytes": 1508, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 51294.487459235825, "hops": {"1": 6965.8001670613885, "2": 6383.797334041446, "3": 6320.953875081614, "4": 6322.3797919927165, "5": 6315.4535830253735, "6": 6335.9819169854745, "7": 6312.74179101456, "8": 6337.379000033252}, "receiver_prefill_ms": 226.8207500455901, "ttft_ms": 11.867458000779152, "decode_ms": 11.909457971341908}, "total_wall_ms": 51545.57116725482, "ttft_ms": 11.867458000779152, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "No", "evidence_artifact": {"ledger": {"facts_total": 66, "facts_retained": 21, "retention_rate": 0.3181818181818182, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.35294117647058826, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1508}, "ledger_by_hop": {"1": {"facts_total": 66, "facts_retained": 21, "retention_rate": 0.3181818181818182, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.35294117647058826, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1508}, "2": {"facts_total": 66, "facts_retained": 21, "retention_rate": 0.3181818181818182, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.35294117647058826, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1508}, "3": {"facts_total": 66, "facts_retained": 21, "retention_rate": 0.3181818181818182, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.35294117647058826, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1508}, "4": {"facts_total": 66, "facts_retained": 21, "retention_rate": 0.3181818181818182, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.35294117647058826, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1508}, "5": {"facts_total": 66, "facts_retained": 21, "retention_rate": 0.3181818181818182, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.35294117647058826, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1508}, "6": {"facts_total": 66, "facts_retained": 21, "retention_rate": 0.3181818181818182, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.35294117647058826, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1508}, "7": {"facts_total": 66, "facts_retained": 21, "retention_rate": 0.3181818181818182, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.35294117647058826, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1508}, "8": {"facts_total": 66, "facts_retained": 21, "retention_rate": 0.3181818181818182, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.35294117647058826, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1508}}}, "reference_key_id": "keys/synthetic/syn_03024_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "34fcc1e23e5a866b5b7d7e222f0380ba771c4005d3732e9747f026c211840d43", "output_hash": "1ea442a134b2a184bd5d40104401f2a37fbc09ccf3f4bc9da161c6099be3691d"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "da5ac261e1d1012fb619c1c6", "attempt_id": "4a72e4c1f38c", "run_group_id": "E4_core:A4|closed|qwen3b:h8", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:23:51.245966Z", "task_family": "synthetic", "document_id": "syn_03024_2000", "document_hash": "ea10cd11ad08f01c50b0712bc481668857c9d2ddc8e6423968f2a31eeb97ac55", "question_id": "q07_negation", "split": "test", "cluster_id": "syn_03024_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 8, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 2068, "query_tokens": 596, "message_tokens": 512, "message_utf8_bytes": 1508, "cache_tensor_bytes": null, "serialized_payload_bytes": 1508, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 51294.487459235825, "hops": {"1": 6965.8001670613885, "2": 6383.797334041446, "3": 6320.953875081614, "4": 6322.3797919927165, "5": 6315.4535830253735, "6": 6335.9819169854745, "7": 6312.74179101456, "8": 6337.379000033252}, "receiver_prefill_ms": 226.81791603099555, "ttft_ms": 12.087582959793508, "decode_ms": 11.90095895435661}, "total_wall_ms": 51545.81116826739, "ttft_ms": 12.087582959793508, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "no", "evidence_artifact": {"ledger": {"facts_total": 66, "facts_retained": 21, "retention_rate": 0.3181818181818182, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.35294117647058826, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1508}, "ledger_by_hop": {"1": {"facts_total": 66, "facts_retained": 21, "retention_rate": 0.3181818181818182, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.35294117647058826, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1508}, "2": {"facts_total": 66, "facts_retained": 21, "retention_rate": 0.3181818181818182, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.35294117647058826, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1508}, "3": {"facts_total": 66, "facts_retained": 21, "retention_rate": 0.3181818181818182, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.35294117647058826, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1508}, "4": {"facts_total": 66, "facts_retained": 21, "retention_rate": 0.3181818181818182, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.35294117647058826, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1508}, "5": {"facts_total": 66, "facts_retained": 21, "retention_rate": 0.3181818181818182, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.35294117647058826, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1508}, "6": {"facts_total": 66, "facts_retained": 21, "retention_rate": 0.3181818181818182, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.35294117647058826, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1508}, "7": {"facts_total": 66, "facts_retained": 21, "retention_rate": 0.3181818181818182, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.35294117647058826, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1508}, "8": {"facts_total": 66, "facts_retained": 21, "retention_rate": 0.3181818181818182, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.35294117647058826, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1508}}}, "reference_key_id": "keys/synthetic/syn_03024_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "34fcc1e23e5a866b5b7d7e222f0380ba771c4005d3732e9747f026c211840d43", "output_hash": "9390298f3fb0c5b160498935d79cb139aef28e1c47358b4bbba61862b9c26e59"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "203e68c27d9d23918f1d07cd", "attempt_id": "8c48eb1ca7ec", "run_group_id": "E4_core:A4|closed|qwen3b:h8", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:23:51.512118Z", "task_family": "synthetic", "document_id": "syn_03024_2000", "document_hash": "ea10cd11ad08f01c50b0712bc481668857c9d2ddc8e6423968f2a31eeb97ac55", "question_id": "q08_revision", "split": "test", "cluster_id": "syn_03024_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 8, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 2068, "query_tokens": 612, "message_tokens": 512, "message_utf8_bytes": 1508, "cache_tensor_bytes": null, "serialized_payload_bytes": 1508, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 51294.487459235825, "hops": {"1": 6965.8001670613885, "2": 6383.797334041446, "3": 6320.953875081614, "4": 6322.3797919927165, "5": 6315.4535830253735, "6": 6335.9819169854745, "7": 6312.74179101456, "8": 6337.379000033252}, "receiver_prefill_ms": 228.855749941431, "ttft_ms": 12.015124899335206, "decode_ms": 11.869041016325355}, "total_wall_ms": 51547.71729221102, "ttft_ms": 12.015124899335206, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 66, "facts_retained": 21, "retention_rate": 0.3181818181818182, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.35294117647058826, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1508}, "ledger_by_hop": {"1": {"facts_total": 66, "facts_retained": 21, "retention_rate": 0.3181818181818182, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.35294117647058826, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1508}, "2": {"facts_total": 66, "facts_retained": 21, "retention_rate": 0.3181818181818182, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.35294117647058826, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1508}, "3": {"facts_total": 66, "facts_retained": 21, "retention_rate": 0.3181818181818182, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.35294117647058826, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1508}, "4": {"facts_total": 66, "facts_retained": 21, "retention_rate": 0.3181818181818182, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.35294117647058826, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1508}, "5": {"facts_total": 66, "facts_retained": 21, "retention_rate": 0.3181818181818182, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.35294117647058826, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1508}, "6": {"facts_total": 66, "facts_retained": 21, "retention_rate": 0.3181818181818182, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.35294117647058826, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1508}, "7": {"facts_total": 66, "facts_retained": 21, "retention_rate": 0.3181818181818182, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.35294117647058826, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1508}, "8": {"facts_total": 66, "facts_retained": 21, "retention_rate": 0.3181818181818182, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.35294117647058826, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1508}}}, "reference_key_id": "keys/synthetic/syn_03024_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "34fcc1e23e5a866b5b7d7e222f0380ba771c4005d3732e9747f026c211840d43", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "12e68f1843b22f3e1c63ca8f", "attempt_id": "fcc416ee8e86", "run_group_id": "E4_core:A4|closed|qwen3b:h8", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:23:51.780479Z", "task_family": "synthetic", "document_id": "syn_03024_2000", "document_hash": "ea10cd11ad08f01c50b0712bc481668857c9d2ddc8e6423968f2a31eeb97ac55", "question_id": "q09_revision", "split": "test", "cluster_id": "syn_03024_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 8, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 2068, "query_tokens": 603, "message_tokens": 512, "message_utf8_bytes": 1508, "cache_tensor_bytes": null, "serialized_payload_bytes": 1508, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 51294.487459235825, "hops": {"1": 6965.8001670613885, "2": 6383.797334041446, "3": 6320.953875081614, "4": 6322.3797919927165, "5": 6315.4535830253735, "6": 6335.9819169854745, "7": 6312.74179101456, "8": 6337.379000033252}, "receiver_prefill_ms": 227.07966703455895, "ttft_ms": 12.022125069051981, "decode_ms": 11.914458009414375}, "total_wall_ms": 51546.01620929316, "ttft_ms": 12.022125069051981, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 66, "facts_retained": 21, "retention_rate": 0.3181818181818182, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.35294117647058826, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1508}, "ledger_by_hop": {"1": {"facts_total": 66, "facts_retained": 21, "retention_rate": 0.3181818181818182, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.35294117647058826, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1508}, "2": {"facts_total": 66, "facts_retained": 21, "retention_rate": 0.3181818181818182, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.35294117647058826, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1508}, "3": {"facts_total": 66, "facts_retained": 21, "retention_rate": 0.3181818181818182, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.35294117647058826, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1508}, "4": {"facts_total": 66, "facts_retained": 21, "retention_rate": 0.3181818181818182, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.35294117647058826, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1508}, "5": {"facts_total": 66, "facts_retained": 21, "retention_rate": 0.3181818181818182, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.35294117647058826, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1508}, "6": {"facts_total": 66, "facts_retained": 21, "retention_rate": 0.3181818181818182, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.35294117647058826, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1508}, "7": {"facts_total": 66, "facts_retained": 21, "retention_rate": 0.3181818181818182, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.35294117647058826, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1508}, "8": {"facts_total": 66, "facts_retained": 21, "retention_rate": 0.3181818181818182, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.35294117647058826, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1508}}}, "reference_key_id": "keys/synthetic/syn_03024_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "34fcc1e23e5a866b5b7d7e222f0380ba771c4005d3732e9747f026c211840d43", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "7ec61b8c3502b3dd603f5b34", "attempt_id": "c2c2c824e6f2", "run_group_id": "E4_core:A4|closed|qwen3b:h8", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:23:52.047154Z", "task_family": "synthetic", "document_id": "syn_03024_2000", "document_hash": "ea10cd11ad08f01c50b0712bc481668857c9d2ddc8e6423968f2a31eeb97ac55", "question_id": "q10_temporal", "split": "test", "cluster_id": "syn_03024_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 8, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 2068, "query_tokens": 613, "message_tokens": 512, "message_utf8_bytes": 1508, "cache_tensor_bytes": null, "serialized_payload_bytes": 1508, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 51294.487459235825, "hops": {"1": 6965.8001670613885, "2": 6383.797334041446, "3": 6320.953875081614, "4": 6322.3797919927165, "5": 6315.4535830253735, "6": 6335.9819169854745, "7": 6312.74179101456, "8": 6337.379000033252}, "receiver_prefill_ms": 229.16283400263637, "ttft_ms": 11.858124984428287, "decode_ms": 11.946166981942952}, "total_wall_ms": 51547.945667174645, "ttft_ms": 11.858124984428287, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 66, "facts_retained": 21, "retention_rate": 0.3181818181818182, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.35294117647058826, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1508}, "ledger_by_hop": {"1": {"facts_total": 66, "facts_retained": 21, "retention_rate": 0.3181818181818182, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.35294117647058826, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1508}, "2": {"facts_total": 66, "facts_retained": 21, "retention_rate": 0.3181818181818182, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.35294117647058826, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1508}, "3": {"facts_total": 66, "facts_retained": 21, "retention_rate": 0.3181818181818182, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.35294117647058826, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1508}, "4": {"facts_total": 66, "facts_retained": 21, "retention_rate": 0.3181818181818182, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.35294117647058826, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1508}, "5": {"facts_total": 66, "facts_retained": 21, "retention_rate": 0.3181818181818182, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.35294117647058826, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1508}, "6": {"facts_total": 66, "facts_retained": 21, "retention_rate": 0.3181818181818182, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.35294117647058826, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1508}, "7": {"facts_total": 66, "facts_retained": 21, "retention_rate": 0.3181818181818182, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.35294117647058826, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1508}, "8": {"facts_total": 66, "facts_retained": 21, "retention_rate": 0.3181818181818182, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.35294117647058826, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1508}}}, "reference_key_id": "keys/synthetic/syn_03024_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "34fcc1e23e5a866b5b7d7e222f0380ba771c4005d3732e9747f026c211840d43", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "1d0d129ea06720cfa36d3a6f", "attempt_id": "68c3d99dd648", "run_group_id": "E4_core:A4|closed|qwen3b:h8", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:23:52.315792Z", "task_family": "synthetic", "document_id": "syn_03024_2000", "document_hash": "ea10cd11ad08f01c50b0712bc481668857c9d2ddc8e6423968f2a31eeb97ac55", "question_id": "q11_aggregation", "split": "test", "cluster_id": "syn_03024_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 8, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 2068, "query_tokens": 606, "message_tokens": 512, "message_utf8_bytes": 1508, "cache_tensor_bytes": null, "serialized_payload_bytes": 1508, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 51294.487459235825, "hops": {"1": 6965.8001670613885, "2": 6383.797334041446, "3": 6320.953875081614, "4": 6322.3797919927165, "5": 6315.4535830253735, "6": 6335.9819169854745, "7": 6312.74179101456, "8": 6337.379000033252}, "receiver_prefill_ms": 227.40304097533226, "ttft_ms": 11.915542068891227, "decode_ms": 82.97120907809585}, "total_wall_ms": 51617.27400019299, "ttft_ms": 11.915542068891227, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 7}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "2707654", "evidence_artifact": {"ledger": {"facts_total": 66, "facts_retained": 21, "retention_rate": 0.3181818181818182, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.35294117647058826, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1508}, "ledger_by_hop": {"1": {"facts_total": 66, "facts_retained": 21, "retention_rate": 0.3181818181818182, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.35294117647058826, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1508}, "2": {"facts_total": 66, "facts_retained": 21, "retention_rate": 0.3181818181818182, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.35294117647058826, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1508}, "3": {"facts_total": 66, "facts_retained": 21, "retention_rate": 0.3181818181818182, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.35294117647058826, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1508}, "4": {"facts_total": 66, "facts_retained": 21, "retention_rate": 0.3181818181818182, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.35294117647058826, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1508}, "5": {"facts_total": 66, "facts_retained": 21, "retention_rate": 0.3181818181818182, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.35294117647058826, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1508}, "6": {"facts_total": 66, "facts_retained": 21, "retention_rate": 0.3181818181818182, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.35294117647058826, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1508}, "7": {"facts_total": 66, "facts_retained": 21, "retention_rate": 0.3181818181818182, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.35294117647058826, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1508}, "8": {"facts_total": 66, "facts_retained": 21, "retention_rate": 0.3181818181818182, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.35294117647058826, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1508}}}, "reference_key_id": "keys/synthetic/syn_03024_2000.json", "task_scores": {"correct": 0.0, "abstained": 0.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "34fcc1e23e5a866b5b7d7e222f0380ba771c4005d3732e9747f026c211840d43", "output_hash": "7799e805760fa6568574360f61d4c7cc9e162842191ff434dfed71076922855e"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "02c02a185e7fd154972915ca", "attempt_id": "f4d5caed64a8", "run_group_id": "E4_core:A4|closed|qwen3b:h8", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:23:52.653727Z", "task_family": "synthetic", "document_id": "syn_03024_2000", "document_hash": "ea10cd11ad08f01c50b0712bc481668857c9d2ddc8e6423968f2a31eeb97ac55", "question_id": "q12_unit_conversion", "split": "test", "cluster_id": "syn_03024_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 8, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 2068, "query_tokens": 603, "message_tokens": 512, "message_utf8_bytes": 1508, "cache_tensor_bytes": null, "serialized_payload_bytes": 1508, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 51294.487459235825, "hops": {"1": 6965.8001670613885, "2": 6383.797334041446, "3": 6320.953875081614, "4": 6322.3797919927165, "5": 6315.4535830253735, "6": 6335.9819169854745, "7": 6312.74179101456, "8": 6337.379000033252}, "receiver_prefill_ms": 227.34025004319847, "ttft_ms": 11.922917095944285, "decode_ms": 11.88620796892792}, "total_wall_ms": 51546.11158417538, "ttft_ms": 11.922917095944285, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 66, "facts_retained": 21, "retention_rate": 0.3181818181818182, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.35294117647058826, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1508}, "ledger_by_hop": {"1": {"facts_total": 66, "facts_retained": 21, "retention_rate": 0.3181818181818182, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.35294117647058826, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1508}, "2": {"facts_total": 66, "facts_retained": 21, "retention_rate": 0.3181818181818182, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.35294117647058826, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1508}, "3": {"facts_total": 66, "facts_retained": 21, "retention_rate": 0.3181818181818182, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.35294117647058826, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1508}, "4": {"facts_total": 66, "facts_retained": 21, "retention_rate": 0.3181818181818182, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.35294117647058826, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1508}, "5": {"facts_total": 66, "facts_retained": 21, "retention_rate": 0.3181818181818182, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.35294117647058826, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1508}, "6": {"facts_total": 66, "facts_retained": 21, "retention_rate": 0.3181818181818182, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.35294117647058826, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1508}, "7": {"facts_total": 66, "facts_retained": 21, "retention_rate": 0.3181818181818182, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.35294117647058826, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1508}, "8": {"facts_total": 66, "facts_retained": 21, "retention_rate": 0.3181818181818182, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.35294117647058826, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1508}}}, "reference_key_id": "keys/synthetic/syn_03024_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "34fcc1e23e5a866b5b7d7e222f0380ba771c4005d3732e9747f026c211840d43", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "7372cffe659dae5fcf305de9", "attempt_id": "39d376023187", "run_group_id": "E4_core:A4|closed|qwen3b:h8", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:23:52.920359Z", "task_family": "synthetic", "document_id": "syn_03024_2000", "document_hash": "ea10cd11ad08f01c50b0712bc481668857c9d2ddc8e6423968f2a31eeb97ac55", "question_id": "q13_unanswerable", "split": "test", "cluster_id": "syn_03024_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 8, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 2068, "query_tokens": 607, "message_tokens": 512, "message_utf8_bytes": 1508, "cache_tensor_bytes": null, "serialized_payload_bytes": 1508, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 51294.487459235825, "hops": {"1": 6965.8001670613885, "2": 6383.797334041446, "3": 6320.953875081614, "4": 6322.3797919927165, "5": 6315.4535830253735, "6": 6335.9819169854745, "7": 6312.74179101456, "8": 6337.379000033252}, "receiver_prefill_ms": 227.38029202446342, "ttft_ms": 11.933458037674427, "decode_ms": 11.882791994139552}, "total_wall_ms": 51546.17308429442, "ttft_ms": 11.933458037674427, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 66, "facts_retained": 21, "retention_rate": 0.3181818181818182, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.35294117647058826, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1508}, "ledger_by_hop": {"1": {"facts_total": 66, "facts_retained": 21, "retention_rate": 0.3181818181818182, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.35294117647058826, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1508}, "2": {"facts_total": 66, "facts_retained": 21, "retention_rate": 0.3181818181818182, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.35294117647058826, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1508}, "3": {"facts_total": 66, "facts_retained": 21, "retention_rate": 0.3181818181818182, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.35294117647058826, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1508}, "4": {"facts_total": 66, "facts_retained": 21, "retention_rate": 0.3181818181818182, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.35294117647058826, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1508}, "5": {"facts_total": 66, "facts_retained": 21, "retention_rate": 0.3181818181818182, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.35294117647058826, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1508}, "6": {"facts_total": 66, "facts_retained": 21, "retention_rate": 0.3181818181818182, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.35294117647058826, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1508}, "7": {"facts_total": 66, "facts_retained": 21, "retention_rate": 0.3181818181818182, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.35294117647058826, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1508}, "8": {"facts_total": 66, "facts_retained": 21, "retention_rate": 0.3181818181818182, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.35294117647058826, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1508}}}, "reference_key_id": "keys/synthetic/syn_03024_2000.json", "task_scores": {"correct": 1.0, "abstained": 1.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "34fcc1e23e5a866b5b7d7e222f0380ba771c4005d3732e9747f026c211840d43", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "aaa3fa059e7f87cdad36afbb", "attempt_id": "d98f39344b57", "run_group_id": "E4_core:A4|closed|qwen3b:h8", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:23:53.186754Z", "task_family": "synthetic", "document_id": "syn_03024_2000", "document_hash": "ea10cd11ad08f01c50b0712bc481668857c9d2ddc8e6423968f2a31eeb97ac55", "question_id": "q14_unanswerable", "split": "test", "cluster_id": "syn_03024_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 8, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 2068, "query_tokens": 608, "message_tokens": 512, "message_utf8_bytes": 1508, "cache_tensor_bytes": null, "serialized_payload_bytes": 1508, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 51294.487459235825, "hops": {"1": 6965.8001670613885, "2": 6383.797334041446, "3": 6320.953875081614, "4": 6322.3797919927165, "5": 6315.4535830253735, "6": 6335.9819169854745, "7": 6312.74179101456, "8": 6337.379000033252}, "receiver_prefill_ms": 227.1588749717921, "ttft_ms": 11.959374998696148, "decode_ms": 11.879750061780214}, "total_wall_ms": 51545.975125278346, "ttft_ms": 11.959374998696148, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 66, "facts_retained": 21, "retention_rate": 0.3181818181818182, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.35294117647058826, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1508}, "ledger_by_hop": {"1": {"facts_total": 66, "facts_retained": 21, "retention_rate": 0.3181818181818182, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.35294117647058826, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1508}, "2": {"facts_total": 66, "facts_retained": 21, "retention_rate": 0.3181818181818182, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.35294117647058826, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1508}, "3": {"facts_total": 66, "facts_retained": 21, "retention_rate": 0.3181818181818182, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.35294117647058826, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1508}, "4": {"facts_total": 66, "facts_retained": 21, "retention_rate": 0.3181818181818182, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.35294117647058826, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1508}, "5": {"facts_total": 66, "facts_retained": 21, "retention_rate": 0.3181818181818182, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.35294117647058826, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1508}, "6": {"facts_total": 66, "facts_retained": 21, "retention_rate": 0.3181818181818182, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.35294117647058826, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1508}, "7": {"facts_total": 66, "facts_retained": 21, "retention_rate": 0.3181818181818182, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.35294117647058826, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1508}, "8": {"facts_total": 66, "facts_retained": 21, "retention_rate": 0.3181818181818182, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.35294117647058826, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1508}}}, "reference_key_id": "keys/synthetic/syn_03024_2000.json", "task_scores": {"correct": 1.0, "abstained": 1.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "34fcc1e23e5a866b5b7d7e222f0380ba771c4005d3732e9747f026c211840d43", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "5c0a46794d470a83dc5eaf06", "attempt_id": "f6a53a4ae398", "run_group_id": "E4_core:A5|closed|qwen3b:h1", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:24:00.391363Z", "task_family": "synthetic", "document_id": "syn_03024_2000", "document_hash": "ea10cd11ad08f01c50b0712bc481668857c9d2ddc8e6423968f2a31eeb97ac55", "question_id": "q00_single_fact", "split": "test", "cluster_id": "syn_03024_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 1, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 2068, "query_tokens": 440, "message_tokens": 357, "message_utf8_bytes": 818, "cache_tensor_bytes": null, "serialized_payload_bytes": 818, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 6936.468709027395, "hops": {"1": 6936.468709027395}, "receiver_prefill_ms": 164.26933300681412, "ttft_ms": 11.80416694842279, "decode_ms": 11.790709104388952}, "total_wall_ms": 7124.787209089845, "ttft_ms": 11.80416694842279, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 66, "facts_retained": 6, "retention_rate": 0.09090909090909091, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.17647058823529413, "status": 0.0, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 754}, "ledger_by_hop": {"1": {"facts_total": 66, "facts_retained": 6, "retention_rate": 0.09090909090909091, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.17647058823529413, "status": 0.0, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 754}}}, "reference_key_id": "keys/synthetic/syn_03024_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "3c07df4f32c3dffe23905714a167378f1096d7812f8c55ff7cf92b920f04befa", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "9a054945779b95b752f6a5b5", "attempt_id": "b9fefcd8a26d", "run_group_id": "E4_core:A5|closed|qwen3b:h1", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:24:00.595688Z", "task_family": "synthetic", "document_id": "syn_03024_2000", "document_hash": "ea10cd11ad08f01c50b0712bc481668857c9d2ddc8e6423968f2a31eeb97ac55", "question_id": "q01_single_fact", "split": "test", "cluster_id": "syn_03024_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 1, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 2068, "query_tokens": 445, "message_tokens": 357, "message_utf8_bytes": 818, "cache_tensor_bytes": null, "serialized_payload_bytes": 818, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 6936.468709027395, "hops": {"1": 6936.468709027395}, "receiver_prefill_ms": 164.79070798959583, "ttft_ms": 12.047125026583672, "decode_ms": 130.07595902308822}, "total_wall_ms": 7243.934668018483, "ttft_ms": 12.047125026583672, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 11}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "Lot 568-G in Mont-Laurier", "evidence_artifact": {"ledger": {"facts_total": 66, "facts_retained": 6, "retention_rate": 0.09090909090909091, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.17647058823529413, "status": 0.0, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 754}, "ledger_by_hop": {"1": {"facts_total": 66, "facts_retained": 6, "retention_rate": 0.09090909090909091, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.17647058823529413, "status": 0.0, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 754}}}, "reference_key_id": "keys/synthetic/syn_03024_2000.json", "task_scores": {"correct": 0.0, "abstained": 0.0, "f1": 0.5}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "3c07df4f32c3dffe23905714a167378f1096d7812f8c55ff7cf92b920f04befa", "output_hash": "98e007faa5a10742ad74f2778a64e3ea0d6b72bec1c0392200d475073d7bf082"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "6c1dfbb4e681b18651889980", "attempt_id": "a001a2bb4d9c", "run_group_id": "E4_core:A5|closed|qwen3b:h1", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:24:00.918833Z", "task_family": "synthetic", "document_id": "syn_03024_2000", "document_hash": "ea10cd11ad08f01c50b0712bc481668857c9d2ddc8e6423968f2a31eeb97ac55", "question_id": "q02_exact_copy", "split": "test", "cluster_id": "syn_03024_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 1, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 2068, "query_tokens": 445, "message_tokens": 357, "message_utf8_bytes": 818, "cache_tensor_bytes": null, "serialized_payload_bytes": 818, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 6936.468709027395, "hops": {"1": 6936.468709027395}, "receiver_prefill_ms": 165.1084590703249, "ttft_ms": 11.83262502308935, "decode_ms": 11.728542041964829}, "total_wall_ms": 7125.536459032446, "ttft_ms": 11.83262502308935, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 66, "facts_retained": 6, "retention_rate": 0.09090909090909091, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.17647058823529413, "status": 0.0, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 754}, "ledger_by_hop": {"1": {"facts_total": 66, "facts_retained": 6, "retention_rate": 0.09090909090909091, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.17647058823529413, "status": 0.0, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 754}}}, "reference_key_id": "keys/synthetic/syn_03024_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "3c07df4f32c3dffe23905714a167378f1096d7812f8c55ff7cf92b920f04befa", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "002b83b453a65db7e45424d8", "attempt_id": "1dddba50ec46", "run_group_id": "E4_core:A5|closed|qwen3b:h1", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:24:01.124226Z", "task_family": "synthetic", "document_id": "syn_03024_2000", "document_hash": "ea10cd11ad08f01c50b0712bc481668857c9d2ddc8e6423968f2a31eeb97ac55", "question_id": "q03_exact_copy", "split": "test", "cluster_id": "syn_03024_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 1, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 2068, "query_tokens": 440, "message_tokens": 357, "message_utf8_bytes": 818, "cache_tensor_bytes": null, "serialized_payload_bytes": 818, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 6936.468709027395, "hops": {"1": 6936.468709027395}, "receiver_prefill_ms": 164.0583339612931, "ttft_ms": 12.081082910299301, "decode_ms": 11.792624951340258}, "total_wall_ms": 7124.9332500156015, "ttft_ms": 12.081082910299301, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 66, "facts_retained": 6, "retention_rate": 0.09090909090909091, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.17647058823529413, "status": 0.0, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 754}, "ledger_by_hop": {"1": {"facts_total": 66, "facts_retained": 6, "retention_rate": 0.09090909090909091, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.17647058823529413, "status": 0.0, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 754}}}, "reference_key_id": "keys/synthetic/syn_03024_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "3c07df4f32c3dffe23905714a167378f1096d7812f8c55ff7cf92b920f04befa", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "63e2b46825b1675b0b6f8fb1", "attempt_id": "f04a9c2fefc3", "run_group_id": "E4_core:A5|closed|qwen3b:h1", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:24:01.328199Z", "task_family": "synthetic", "document_id": "syn_03024_2000", "document_hash": "ea10cd11ad08f01c50b0712bc481668857c9d2ddc8e6423968f2a31eeb97ac55", "question_id": "q04_multi_hop", "split": "test", "cluster_id": "syn_03024_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 1, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 2068, "query_tokens": 446, "message_tokens": 357, "message_utf8_bytes": 818, "cache_tensor_bytes": null, "serialized_payload_bytes": 818, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 6936.468709027395, "hops": {"1": 6936.468709027395}, "receiver_prefill_ms": 164.67762493994087, "ttft_ms": 11.896417010575533, "decode_ms": 11.777332983911037}, "total_wall_ms": 7125.347375986166, "ttft_ms": 11.896417010575533, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 66, "facts_retained": 6, "retention_rate": 0.09090909090909091, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.17647058823529413, "status": 0.0, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 754}, "ledger_by_hop": {"1": {"facts_total": 66, "facts_retained": 6, "retention_rate": 0.09090909090909091, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.17647058823529413, "status": 0.0, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 754}}}, "reference_key_id": "keys/synthetic/syn_03024_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "3c07df4f32c3dffe23905714a167378f1096d7812f8c55ff7cf92b920f04befa", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "eab56dedd883c8843d5ff1cb", "attempt_id": "903634cd4f59", "run_group_id": "E4_core:A5|closed|qwen3b:h1", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:24:01.533843Z", "task_family": "synthetic", "document_id": "syn_03024_2000", "document_hash": "ea10cd11ad08f01c50b0712bc481668857c9d2ddc8e6423968f2a31eeb97ac55", "question_id": "q05_multi_hop", "split": "test", "cluster_id": "syn_03024_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 1, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 2068, "query_tokens": 450, "message_tokens": 357, "message_utf8_bytes": 818, "cache_tensor_bytes": null, "serialized_payload_bytes": 818, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 6936.468709027395, "hops": {"1": 6936.468709027395}, "receiver_prefill_ms": 184.89924992900342, "ttft_ms": 12.040583998896182, "decode_ms": 11.881624930538237}, "total_wall_ms": 7145.81962605007, "ttft_ms": 12.040583998896182, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 66, "facts_retained": 6, "retention_rate": 0.09090909090909091, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.17647058823529413, "status": 0.0, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 754}, "ledger_by_hop": {"1": {"facts_total": 66, "facts_retained": 6, "retention_rate": 0.09090909090909091, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.17647058823529413, "status": 0.0, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 754}}}, "reference_key_id": "keys/synthetic/syn_03024_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "3c07df4f32c3dffe23905714a167378f1096d7812f8c55ff7cf92b920f04befa", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "b8da8ea509495646817ef35e", "attempt_id": "e9cddc7c6db5", "run_group_id": "E4_core:A5|closed|qwen3b:h1", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:24:01.759234Z", "task_family": "synthetic", "document_id": "syn_03024_2000", "document_hash": "ea10cd11ad08f01c50b0712bc481668857c9d2ddc8e6423968f2a31eeb97ac55", "question_id": "q06_negation", "split": "test", "cluster_id": "syn_03024_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 1, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 2068, "query_tokens": 439, "message_tokens": 357, "message_utf8_bytes": 818, "cache_tensor_bytes": null, "serialized_payload_bytes": 818, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 6936.468709027395, "hops": {"1": 6936.468709027395}, "receiver_prefill_ms": 164.23270793166012, "ttft_ms": 11.796916020102799, "decode_ms": 11.788291041739285}, "total_wall_ms": 7124.669751035981, "ttft_ms": 11.796916020102799, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "no", "evidence_artifact": {"ledger": {"facts_total": 66, "facts_retained": 6, "retention_rate": 0.09090909090909091, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.17647058823529413, "status": 0.0, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 754}, "ledger_by_hop": {"1": {"facts_total": 66, "facts_retained": 6, "retention_rate": 0.09090909090909091, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.17647058823529413, "status": 0.0, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 754}}}, "reference_key_id": "keys/synthetic/syn_03024_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "3c07df4f32c3dffe23905714a167378f1096d7812f8c55ff7cf92b920f04befa", "output_hash": "9390298f3fb0c5b160498935d79cb139aef28e1c47358b4bbba61862b9c26e59"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "bedfd838642aeacb43c7693e", "attempt_id": "16774b407c62", "run_group_id": "E4_core:A5|closed|qwen3b:h1", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:24:01.963156Z", "task_family": "synthetic", "document_id": "syn_03024_2000", "document_hash": "ea10cd11ad08f01c50b0712bc481668857c9d2ddc8e6423968f2a31eeb97ac55", "question_id": "q07_negation", "split": "test", "cluster_id": "syn_03024_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 1, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 2068, "query_tokens": 441, "message_tokens": 357, "message_utf8_bytes": 818, "cache_tensor_bytes": null, "serialized_payload_bytes": 818, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 6936.468709027395, "hops": {"1": 6936.468709027395}, "receiver_prefill_ms": 164.2329579917714, "ttft_ms": 11.775291990488768, "decode_ms": 11.805000016465783}, "total_wall_ms": 7124.653875012882, "ttft_ms": 11.775291990488768, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "no", "evidence_artifact": {"ledger": {"facts_total": 66, "facts_retained": 6, "retention_rate": 0.09090909090909091, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.17647058823529413, "status": 0.0, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 754}, "ledger_by_hop": {"1": {"facts_total": 66, "facts_retained": 6, "retention_rate": 0.09090909090909091, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.17647058823529413, "status": 0.0, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 754}}}, "reference_key_id": "keys/synthetic/syn_03024_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "3c07df4f32c3dffe23905714a167378f1096d7812f8c55ff7cf92b920f04befa", "output_hash": "9390298f3fb0c5b160498935d79cb139aef28e1c47358b4bbba61862b9c26e59"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "0155b74fcff3d4ac35ff5f40", "attempt_id": "b6ea30e798b4", "run_group_id": "E4_core:A5|closed|qwen3b:h1", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:24:02.166460Z", "task_family": "synthetic", "document_id": "syn_03024_2000", "document_hash": "ea10cd11ad08f01c50b0712bc481668857c9d2ddc8e6423968f2a31eeb97ac55", "question_id": "q08_revision", "split": "test", "cluster_id": "syn_03024_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 1, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 2068, "query_tokens": 457, "message_tokens": 357, "message_utf8_bytes": 818, "cache_tensor_bytes": null, "serialized_payload_bytes": 818, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 6936.468709027395, "hops": {"1": 6936.468709027395}, "receiver_prefill_ms": 185.86870795115829, "ttft_ms": 11.836208985187113, "decode_ms": 11.81591593194753}, "total_wall_ms": 7146.448876010254, "ttft_ms": 11.836208985187113, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 66, "facts_retained": 6, "retention_rate": 0.09090909090909091, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.17647058823529413, "status": 0.0, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 754}, "ledger_by_hop": {"1": {"facts_total": 66, "facts_retained": 6, "retention_rate": 0.09090909090909091, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.17647058823529413, "status": 0.0, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 754}}}, "reference_key_id": "keys/synthetic/syn_03024_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "3c07df4f32c3dffe23905714a167378f1096d7812f8c55ff7cf92b920f04befa", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "ee38a8c6ca620154c4030cb8", "attempt_id": "d0c81646384e", "run_group_id": "E4_core:A5|closed|qwen3b:h1", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:24:02.392062Z", "task_family": "synthetic", "document_id": "syn_03024_2000", "document_hash": "ea10cd11ad08f01c50b0712bc481668857c9d2ddc8e6423968f2a31eeb97ac55", "question_id": "q09_revision", "split": "test", "cluster_id": "syn_03024_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 1, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 2068, "query_tokens": 448, "message_tokens": 357, "message_utf8_bytes": 818, "cache_tensor_bytes": null, "serialized_payload_bytes": 818, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 6936.468709027395, "hops": {"1": 6936.468709027395}, "receiver_prefill_ms": 165.6753340503201, "ttft_ms": 11.78512501064688, "decode_ms": 82.25158299319446}, "total_wall_ms": 7196.570542058907, "ttft_ms": 11.78512501064688, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 7}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "9134435", "evidence_artifact": {"ledger": {"facts_total": 66, "facts_retained": 6, "retention_rate": 0.09090909090909091, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.17647058823529413, "status": 0.0, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 754}, "ledger_by_hop": {"1": {"facts_total": 66, "facts_retained": 6, "retention_rate": 0.09090909090909091, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.17647058823529413, "status": 0.0, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 754}}}, "reference_key_id": "keys/synthetic/syn_03024_2000.json", "task_scores": {"correct": 0.0, "abstained": 0.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "3c07df4f32c3dffe23905714a167378f1096d7812f8c55ff7cf92b920f04befa", "output_hash": "f7384892cb4f9a38f6115f91da5fec3f0d7e24e47877737b0f34e2b4a6af7d5a"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "c2ac17ce3dc8a07f36ec94ec", "attempt_id": "9e15ad02f4e2", "run_group_id": "E4_core:A5|closed|qwen3b:h1", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:24:02.667654Z", "task_family": "synthetic", "document_id": "syn_03024_2000", "document_hash": "ea10cd11ad08f01c50b0712bc481668857c9d2ddc8e6423968f2a31eeb97ac55", "question_id": "q10_temporal", "split": "test", "cluster_id": "syn_03024_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 1, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 2068, "query_tokens": 458, "message_tokens": 357, "message_utf8_bytes": 818, "cache_tensor_bytes": null, "serialized_payload_bytes": 818, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 6936.468709027395, "hops": {"1": 6936.468709027395}, "receiver_prefill_ms": 185.01112493686378, "ttft_ms": 11.85000000987202, "decode_ms": 11.756583000533283}, "total_wall_ms": 7145.507792942226, "ttft_ms": 11.85000000987202, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 66, "facts_retained": 6, "retention_rate": 0.09090909090909091, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.17647058823529413, "status": 0.0, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 754}, "ledger_by_hop": {"1": {"facts_total": 66, "facts_retained": 6, "retention_rate": 0.09090909090909091, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.17647058823529413, "status": 0.0, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 754}}}, "reference_key_id": "keys/synthetic/syn_03024_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "3c07df4f32c3dffe23905714a167378f1096d7812f8c55ff7cf92b920f04befa", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "ae74c942b2725919916b8221", "attempt_id": "786b6dceeadf", "run_group_id": "E4_core:A5|closed|qwen3b:h1", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:24:02.892070Z", "task_family": "synthetic", "document_id": "syn_03024_2000", "document_hash": "ea10cd11ad08f01c50b0712bc481668857c9d2ddc8e6423968f2a31eeb97ac55", "question_id": "q11_aggregation", "split": "test", "cluster_id": "syn_03024_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 1, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 2068, "query_tokens": 451, "message_tokens": 357, "message_utf8_bytes": 818, "cache_tensor_bytes": null, "serialized_payload_bytes": 818, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 6936.468709027395, "hops": {"1": 6936.468709027395}, "receiver_prefill_ms": 184.76804194506258, "ttft_ms": 11.788417003117502, "decode_ms": 82.40608300548047}, "total_wall_ms": 7215.8218760741875, "ttft_ms": 11.788417003117502, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 7}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "8066734", "evidence_artifact": {"ledger": {"facts_total": 66, "facts_retained": 6, "retention_rate": 0.09090909090909091, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.17647058823529413, "status": 0.0, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 754}, "ledger_by_hop": {"1": {"facts_total": 66, "facts_retained": 6, "retention_rate": 0.09090909090909091, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.17647058823529413, "status": 0.0, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 754}}}, "reference_key_id": "keys/synthetic/syn_03024_2000.json", "task_scores": {"correct": 0.0, "abstained": 0.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "3c07df4f32c3dffe23905714a167378f1096d7812f8c55ff7cf92b920f04befa", "output_hash": "167c75fdd6a41fca64228797ab8f3a1a3f994475246df0bffb2e060d5d2e4ff0"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "6327d0f270950a8d968009a3", "attempt_id": "7f3d287266c7", "run_group_id": "E4_core:A5|closed|qwen3b:h1", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:24:03.186225Z", "task_family": "synthetic", "document_id": "syn_03024_2000", "document_hash": "ea10cd11ad08f01c50b0712bc481668857c9d2ddc8e6423968f2a31eeb97ac55", "question_id": "q12_unit_conversion", "split": "test", "cluster_id": "syn_03024_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 1, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 2068, "query_tokens": 448, "message_tokens": 357, "message_utf8_bytes": 818, "cache_tensor_bytes": null, "serialized_payload_bytes": 818, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 6936.468709027395, "hops": {"1": 6936.468709027395}, "receiver_prefill_ms": 165.31150008086115, "ttft_ms": 11.79787504952401, "decode_ms": 11.764207971282303}, "total_wall_ms": 7125.712918001227, "ttft_ms": 11.79787504952401, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 66, "facts_retained": 6, "retention_rate": 0.09090909090909091, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.17647058823529413, "status": 0.0, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 754}, "ledger_by_hop": {"1": {"facts_total": 66, "facts_retained": 6, "retention_rate": 0.09090909090909091, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.17647058823529413, "status": 0.0, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 754}}}, "reference_key_id": "keys/synthetic/syn_03024_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "3c07df4f32c3dffe23905714a167378f1096d7812f8c55ff7cf92b920f04befa", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "387acff2e427afe098414fcd", "attempt_id": "153b3034f20c", "run_group_id": "E4_core:A5|closed|qwen3b:h1", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:24:03.390365Z", "task_family": "synthetic", "document_id": "syn_03024_2000", "document_hash": "ea10cd11ad08f01c50b0712bc481668857c9d2ddc8e6423968f2a31eeb97ac55", "question_id": "q13_unanswerable", "split": "test", "cluster_id": "syn_03024_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 1, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 2068, "query_tokens": 452, "message_tokens": 357, "message_utf8_bytes": 818, "cache_tensor_bytes": null, "serialized_payload_bytes": 818, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 6936.468709027395, "hops": {"1": 6936.468709027395}, "receiver_prefill_ms": 185.54616696201265, "ttft_ms": 11.814708006568253, "decode_ms": 11.842208099551499}, "total_wall_ms": 7146.036084042862, "ttft_ms": 11.814708006568253, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 66, "facts_retained": 6, "retention_rate": 0.09090909090909091, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.17647058823529413, "status": 0.0, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 754}, "ledger_by_hop": {"1": {"facts_total": 66, "facts_retained": 6, "retention_rate": 0.09090909090909091, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.17647058823529413, "status": 0.0, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 754}}}, "reference_key_id": "keys/synthetic/syn_03024_2000.json", "task_scores": {"correct": 1.0, "abstained": 1.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "3c07df4f32c3dffe23905714a167378f1096d7812f8c55ff7cf92b920f04befa", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "2e89e7502d1437bc71263336", "attempt_id": "f89836bf53e4", "run_group_id": "E4_core:A5|closed|qwen3b:h1", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:24:03.614845Z", "task_family": "synthetic", "document_id": "syn_03024_2000", "document_hash": "ea10cd11ad08f01c50b0712bc481668857c9d2ddc8e6423968f2a31eeb97ac55", "question_id": "q14_unanswerable", "split": "test", "cluster_id": "syn_03024_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 1, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 2068, "query_tokens": 453, "message_tokens": 357, "message_utf8_bytes": 818, "cache_tensor_bytes": null, "serialized_payload_bytes": 818, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 6936.468709027395, "hops": {"1": 6936.468709027395}, "receiver_prefill_ms": 185.07358303759247, "ttft_ms": 11.805834015831351, "decode_ms": 11.80404203478247}, "total_wall_ms": 7145.520791993476, "ttft_ms": 11.805834015831351, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 66, "facts_retained": 6, "retention_rate": 0.09090909090909091, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.17647058823529413, "status": 0.0, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 754}, "ledger_by_hop": {"1": {"facts_total": 66, "facts_retained": 6, "retention_rate": 0.09090909090909091, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.17647058823529413, "status": 0.0, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 754}}}, "reference_key_id": "keys/synthetic/syn_03024_2000.json", "task_scores": {"correct": 1.0, "abstained": 1.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "3c07df4f32c3dffe23905714a167378f1096d7812f8c55ff7cf92b920f04befa", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "fadf6b80a3a17a3f7a9ffb6b", "attempt_id": "e2bf4ebd5cee", "run_group_id": "E4_core:A5|closed|qwen3b:h2", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:24:08.387120Z", "task_family": "synthetic", "document_id": "syn_03024_2000", "document_hash": "ea10cd11ad08f01c50b0712bc481668857c9d2ddc8e6423968f2a31eeb97ac55", "question_id": "q00_single_fact", "split": "test", "cluster_id": "syn_03024_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 2, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 2068, "query_tokens": 445, "message_tokens": 363, "message_utf8_bytes": 834, "cache_tensor_bytes": null, "serialized_payload_bytes": 834, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 11483.192084007896, "hops": {"1": 6936.468709027395, "2": 4546.723374980502}, "receiver_prefill_ms": 165.30312492977828, "ttft_ms": 11.850125039927661, "decode_ms": 117.14154202491045}, "total_wall_ms": 11777.991416980512, "ttft_ms": 11.850125039927661, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 10}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "BU-55G-3254", "evidence_artifact": {"ledger": {"facts_total": 66, "facts_retained": 8, "retention_rate": 0.12121212121212122, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.17647058823529413, "status": 0.16666666666666666, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 770}, "ledger_by_hop": {"1": {"facts_total": 66, "facts_retained": 6, "retention_rate": 0.09090909090909091, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.17647058823529413, "status": 0.0, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 754}, "2": {"facts_total": 66, "facts_retained": 8, "retention_rate": 0.12121212121212122, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.17647058823529413, "status": 0.16666666666666666, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 770}}}, "reference_key_id": "keys/synthetic/syn_03024_2000.json", "task_scores": {"correct": 0.0, "abstained": 0.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "b7386bd5ac9c9b04a9f81a350d93a018a74bcc7cc6eb41441f5a45c9ea28d0ba", "output_hash": "98227e32c2121a7856e6d2b32b74598ce6f4183c2ec3eec7ce4b9b0b3f66b98b"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "e1feda63120a084bc52c6f3c", "attempt_id": "5d5d2a002589", "run_group_id": "E4_core:A5|closed|qwen3b:h2", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:24:08.698202Z", "task_family": "synthetic", "document_id": "syn_03024_2000", "document_hash": "ea10cd11ad08f01c50b0712bc481668857c9d2ddc8e6423968f2a31eeb97ac55", "question_id": "q01_single_fact", "split": "test", "cluster_id": "syn_03024_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 2, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 2068, "query_tokens": 450, "message_tokens": 363, "message_utf8_bytes": 834, "cache_tensor_bytes": null, "serialized_payload_bytes": 834, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 11483.192084007896, "hops": {"1": 6936.468709027395, "2": 4546.723374980502}, "receiver_prefill_ms": 184.7597499145195, "ttft_ms": 12.01541698537767, "decode_ms": 129.88954095635563}, "total_wall_ms": 11810.371709056199, "ttft_ms": 12.01541698537767, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 11}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "Lot 568-G in Mont-Laurier", "evidence_artifact": {"ledger": {"facts_total": 66, "facts_retained": 8, "retention_rate": 0.12121212121212122, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.17647058823529413, "status": 0.16666666666666666, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 770}, "ledger_by_hop": {"1": {"facts_total": 66, "facts_retained": 6, "retention_rate": 0.09090909090909091, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.17647058823529413, "status": 0.0, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 754}, "2": {"facts_total": 66, "facts_retained": 8, "retention_rate": 0.12121212121212122, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.17647058823529413, "status": 0.16666666666666666, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 770}}}, "reference_key_id": "keys/synthetic/syn_03024_2000.json", "task_scores": {"correct": 0.0, "abstained": 0.0, "f1": 0.5}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "b7386bd5ac9c9b04a9f81a350d93a018a74bcc7cc6eb41441f5a45c9ea28d0ba", "output_hash": "98e007faa5a10742ad74f2778a64e3ea0d6b72bec1c0392200d475073d7bf082"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "66066b99de29d4f4d57fa7c0", "attempt_id": "9636d61fa06e", "run_group_id": "E4_core:A5|closed|qwen3b:h2", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:24:09.041975Z", "task_family": "synthetic", "document_id": "syn_03024_2000", "document_hash": "ea10cd11ad08f01c50b0712bc481668857c9d2ddc8e6423968f2a31eeb97ac55", "question_id": "q02_exact_copy", "split": "test", "cluster_id": "syn_03024_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 2, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 2068, "query_tokens": 450, "message_tokens": 363, "message_utf8_bytes": 834, "cache_tensor_bytes": null, "serialized_payload_bytes": 834, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 11483.192084007896, "hops": {"1": 6936.468709027395, "2": 4546.723374980502}, "receiver_prefill_ms": 184.02974994387478, "ttft_ms": 11.894749943166971, "decode_ms": 11.993292020633817}, "total_wall_ms": 11691.599541925825, "ttft_ms": 11.894749943166971, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 66, "facts_retained": 8, "retention_rate": 0.12121212121212122, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.17647058823529413, "status": 0.16666666666666666, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 770}, "ledger_by_hop": {"1": {"facts_total": 66, "facts_retained": 6, "retention_rate": 0.09090909090909091, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.17647058823529413, "status": 0.0, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 754}, "2": {"facts_total": 66, "facts_retained": 8, "retention_rate": 0.12121212121212122, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.17647058823529413, "status": 0.16666666666666666, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 770}}}, "reference_key_id": "keys/synthetic/syn_03024_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "b7386bd5ac9c9b04a9f81a350d93a018a74bcc7cc6eb41441f5a45c9ea28d0ba", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "7e2ebc2683ad6e5785804f22", "attempt_id": "3589f293cda4", "run_group_id": "E4_core:A5|closed|qwen3b:h2", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:24:09.266158Z", "task_family": "synthetic", "document_id": "syn_03024_2000", "document_hash": "ea10cd11ad08f01c50b0712bc481668857c9d2ddc8e6423968f2a31eeb97ac55", "question_id": "q03_exact_copy", "split": "test", "cluster_id": "syn_03024_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 2, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 2068, "query_tokens": 445, "message_tokens": 363, "message_utf8_bytes": 834, "cache_tensor_bytes": null, "serialized_payload_bytes": 834, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 11483.192084007896, "hops": {"1": 6936.468709027395, "2": 4546.723374980502}, "receiver_prefill_ms": 164.52316695358604, "ttft_ms": 11.815292062237859, "decode_ms": 11.850083945319057}, "total_wall_ms": 11671.759251039475, "ttft_ms": 11.815292062237859, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 66, "facts_retained": 8, "retention_rate": 0.12121212121212122, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.17647058823529413, "status": 0.16666666666666666, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 770}, "ledger_by_hop": {"1": {"facts_total": 66, "facts_retained": 6, "retention_rate": 0.09090909090909091, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.17647058823529413, "status": 0.0, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 754}, "2": {"facts_total": 66, "facts_retained": 8, "retention_rate": 0.12121212121212122, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.17647058823529413, "status": 0.16666666666666666, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 770}}}, "reference_key_id": "keys/synthetic/syn_03024_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "b7386bd5ac9c9b04a9f81a350d93a018a74bcc7cc6eb41441f5a45c9ea28d0ba", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "50db5edd817167b057d9d502", "attempt_id": "92b62d67fc8f", "run_group_id": "E4_core:A5|closed|qwen3b:h2", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:24:09.470107Z", "task_family": "synthetic", "document_id": "syn_03024_2000", "document_hash": "ea10cd11ad08f01c50b0712bc481668857c9d2ddc8e6423968f2a31eeb97ac55", "question_id": "q04_multi_hop", "split": "test", "cluster_id": "syn_03024_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 2, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 2068, "query_tokens": 451, "message_tokens": 363, "message_utf8_bytes": 834, "cache_tensor_bytes": null, "serialized_payload_bytes": 834, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 11483.192084007896, "hops": {"1": 6936.468709027395, "2": 4546.723374980502}, "receiver_prefill_ms": 184.84904197975993, "ttft_ms": 11.632499983534217, "decode_ms": 11.661750031635165}, "total_wall_ms": 11691.708042053506, "ttft_ms": 11.632499983534217, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 66, "facts_retained": 8, "retention_rate": 0.12121212121212122, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.17647058823529413, "status": 0.16666666666666666, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 770}, "ledger_by_hop": {"1": {"facts_total": 66, "facts_retained": 6, "retention_rate": 0.09090909090909091, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.17647058823529413, "status": 0.0, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 754}, "2": {"facts_total": 66, "facts_retained": 8, "retention_rate": 0.12121212121212122, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.17647058823529413, "status": 0.16666666666666666, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 770}}}, "reference_key_id": "keys/synthetic/syn_03024_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "b7386bd5ac9c9b04a9f81a350d93a018a74bcc7cc6eb41441f5a45c9ea28d0ba", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "71337a6adc8a5870360ed22b", "attempt_id": "24fef414b72a", "run_group_id": "E4_core:A5|closed|qwen3b:h2", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:24:09.693550Z", "task_family": "synthetic", "document_id": "syn_03024_2000", "document_hash": "ea10cd11ad08f01c50b0712bc481668857c9d2ddc8e6423968f2a31eeb97ac55", "question_id": "q05_multi_hop", "split": "test", "cluster_id": "syn_03024_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 2, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 2068, "query_tokens": 455, "message_tokens": 363, "message_utf8_bytes": 834, "cache_tensor_bytes": null, "serialized_payload_bytes": 834, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 11483.192084007896, "hops": {"1": 6936.468709027395, "2": 4546.723374980502}, "receiver_prefill_ms": 185.38833293132484, "ttft_ms": 12.018500012345612, "decode_ms": 11.815333971753716}, "total_wall_ms": 11692.906376090832, "ttft_ms": 12.018500012345612, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 66, "facts_retained": 8, "retention_rate": 0.12121212121212122, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.17647058823529413, "status": 0.16666666666666666, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 770}, "ledger_by_hop": {"1": {"facts_total": 66, "facts_retained": 6, "retention_rate": 0.09090909090909091, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.17647058823529413, "status": 0.0, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 754}, "2": {"facts_total": 66, "facts_retained": 8, "retention_rate": 0.12121212121212122, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.17647058823529413, "status": 0.16666666666666666, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 770}}}, "reference_key_id": "keys/synthetic/syn_03024_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "b7386bd5ac9c9b04a9f81a350d93a018a74bcc7cc6eb41441f5a45c9ea28d0ba", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "799e8a4d60036fcaf14ef86f", "attempt_id": "8057b02d18cf", "run_group_id": "E4_core:A5|closed|qwen3b:h2", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:24:09.918867Z", "task_family": "synthetic", "document_id": "syn_03024_2000", "document_hash": "ea10cd11ad08f01c50b0712bc481668857c9d2ddc8e6423968f2a31eeb97ac55", "question_id": "q06_negation", "split": "test", "cluster_id": "syn_03024_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 2, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 2068, "query_tokens": 444, "message_tokens": 363, "message_utf8_bytes": 834, "cache_tensor_bytes": null, "serialized_payload_bytes": 834, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 11483.192084007896, "hops": {"1": 6936.468709027395, "2": 4546.723374980502}, "receiver_prefill_ms": 164.20979204121977, "ttft_ms": 11.923499987460673, "decode_ms": 11.82508294004947}, "total_wall_ms": 11671.63574998267, "ttft_ms": 11.923499987460673, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "no", "evidence_artifact": {"ledger": {"facts_total": 66, "facts_retained": 8, "retention_rate": 0.12121212121212122, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.17647058823529413, "status": 0.16666666666666666, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 770}, "ledger_by_hop": {"1": {"facts_total": 66, "facts_retained": 6, "retention_rate": 0.09090909090909091, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.17647058823529413, "status": 0.0, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 754}, "2": {"facts_total": 66, "facts_retained": 8, "retention_rate": 0.12121212121212122, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.17647058823529413, "status": 0.16666666666666666, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 770}}}, "reference_key_id": "keys/synthetic/syn_03024_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "b7386bd5ac9c9b04a9f81a350d93a018a74bcc7cc6eb41441f5a45c9ea28d0ba", "output_hash": "9390298f3fb0c5b160498935d79cb139aef28e1c47358b4bbba61862b9c26e59"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "abef668c0b7a6ccb6652cae4", "attempt_id": "446a113001eb", "run_group_id": "E4_core:A5|closed|qwen3b:h2", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:24:10.123429Z", "task_family": "synthetic", "document_id": "syn_03024_2000", "document_hash": "ea10cd11ad08f01c50b0712bc481668857c9d2ddc8e6423968f2a31eeb97ac55", "question_id": "q07_negation", "split": "test", "cluster_id": "syn_03024_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 2, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 2068, "query_tokens": 446, "message_tokens": 363, "message_utf8_bytes": 834, "cache_tensor_bytes": null, "serialized_payload_bytes": 834, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 11483.192084007896, "hops": {"1": 6936.468709027395, "2": 4546.723374980502}, "receiver_prefill_ms": 164.78812508285046, "ttft_ms": 11.933541973121464, "decode_ms": 11.839375016279519}, "total_wall_ms": 11672.191416961141, "ttft_ms": 11.933541973121464, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "no", "evidence_artifact": {"ledger": {"facts_total": 66, "facts_retained": 8, "retention_rate": 0.12121212121212122, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.17647058823529413, "status": 0.16666666666666666, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 770}, "ledger_by_hop": {"1": {"facts_total": 66, "facts_retained": 6, "retention_rate": 0.09090909090909091, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.17647058823529413, "status": 0.0, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 754}, "2": {"facts_total": 66, "facts_retained": 8, "retention_rate": 0.12121212121212122, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.17647058823529413, "status": 0.16666666666666666, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 770}}}, "reference_key_id": "keys/synthetic/syn_03024_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "b7386bd5ac9c9b04a9f81a350d93a018a74bcc7cc6eb41441f5a45c9ea28d0ba", "output_hash": "9390298f3fb0c5b160498935d79cb139aef28e1c47358b4bbba61862b9c26e59"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "2e1246988ad4c1c037916b9e", "attempt_id": "990405aa52e1", "run_group_id": "E4_core:A5|closed|qwen3b:h2", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:24:10.328215Z", "task_family": "synthetic", "document_id": "syn_03024_2000", "document_hash": "ea10cd11ad08f01c50b0712bc481668857c9d2ddc8e6423968f2a31eeb97ac55", "question_id": "q08_revision", "split": "test", "cluster_id": "syn_03024_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 2, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 2068, "query_tokens": 462, "message_tokens": 363, "message_utf8_bytes": 834, "cache_tensor_bytes": null, "serialized_payload_bytes": 834, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 11483.192084007896, "hops": {"1": 6936.468709027395, "2": 4546.723374980502}, "receiver_prefill_ms": 185.66583306528628, "ttft_ms": 11.888416949659586, "decode_ms": 11.815374949947}, "total_wall_ms": 11693.00425099209, "ttft_ms": 11.888416949659586, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 66, "facts_retained": 8, "retention_rate": 0.12121212121212122, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.17647058823529413, "status": 0.16666666666666666, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 770}, "ledger_by_hop": {"1": {"facts_total": 66, "facts_retained": 6, "retention_rate": 0.09090909090909091, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.17647058823529413, "status": 0.0, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 754}, "2": {"facts_total": 66, "facts_retained": 8, "retention_rate": 0.12121212121212122, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.17647058823529413, "status": 0.16666666666666666, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 770}}}, "reference_key_id": "keys/synthetic/syn_03024_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "b7386bd5ac9c9b04a9f81a350d93a018a74bcc7cc6eb41441f5a45c9ea28d0ba", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "7a0d4a27493151e512f10672", "attempt_id": "0bb8379b6927", "run_group_id": "E4_core:A5|closed|qwen3b:h2", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:24:10.553594Z", "task_family": "synthetic", "document_id": "syn_03024_2000", "document_hash": "ea10cd11ad08f01c50b0712bc481668857c9d2ddc8e6423968f2a31eeb97ac55", "question_id": "q09_revision", "split": "test", "cluster_id": "syn_03024_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 2, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 2068, "query_tokens": 453, "message_tokens": 363, "message_utf8_bytes": 834, "cache_tensor_bytes": null, "serialized_payload_bytes": 834, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 11483.192084007896, "hops": {"1": 6936.468709027395, "2": 4546.723374980502}, "receiver_prefill_ms": 184.838708024472, "ttft_ms": 11.858458048664033, "decode_ms": 11.804417008534074}, "total_wall_ms": 11692.200417048298, "ttft_ms": 11.858458048664033, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 66, "facts_retained": 8, "retention_rate": 0.12121212121212122, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.17647058823529413, "status": 0.16666666666666666, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 770}, "ledger_by_hop": {"1": {"facts_total": 66, "facts_retained": 6, "retention_rate": 0.09090909090909091, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.17647058823529413, "status": 0.0, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 754}, "2": {"facts_total": 66, "facts_retained": 8, "retention_rate": 0.12121212121212122, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.17647058823529413, "status": 0.16666666666666666, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 770}}}, "reference_key_id": "keys/synthetic/syn_03024_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "b7386bd5ac9c9b04a9f81a350d93a018a74bcc7cc6eb41441f5a45c9ea28d0ba", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "10079aa7ca414ab52eafb69e", "attempt_id": "e2ed84f5328f", "run_group_id": "E4_core:A5|closed|qwen3b:h2", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:24:10.777591Z", "task_family": "synthetic", "document_id": "syn_03024_2000", "document_hash": "ea10cd11ad08f01c50b0712bc481668857c9d2ddc8e6423968f2a31eeb97ac55", "question_id": "q10_temporal", "split": "test", "cluster_id": "syn_03024_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 2, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 2068, "query_tokens": 463, "message_tokens": 363, "message_utf8_bytes": 834, "cache_tensor_bytes": null, "serialized_payload_bytes": 834, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 11483.192084007896, "hops": {"1": 6936.468709027395, "2": 4546.723374980502}, "receiver_prefill_ms": 185.35545794293284, "ttft_ms": 11.964083067141473, "decode_ms": 11.907332926057279}, "total_wall_ms": 11692.946959054098, "ttft_ms": 11.964083067141473, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 66, "facts_retained": 8, "retention_rate": 0.12121212121212122, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.17647058823529413, "status": 0.16666666666666666, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 770}, "ledger_by_hop": {"1": {"facts_total": 66, "facts_retained": 6, "retention_rate": 0.09090909090909091, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.17647058823529413, "status": 0.0, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 754}, "2": {"facts_total": 66, "facts_retained": 8, "retention_rate": 0.12121212121212122, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.17647058823529413, "status": 0.16666666666666666, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 770}}}, "reference_key_id": "keys/synthetic/syn_03024_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "b7386bd5ac9c9b04a9f81a350d93a018a74bcc7cc6eb41441f5a45c9ea28d0ba", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "f1751334c61a6544fcf83837", "attempt_id": "570bba19f253", "run_group_id": "E4_core:A5|closed|qwen3b:h2", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:24:11.003711Z", "task_family": "synthetic", "document_id": "syn_03024_2000", "document_hash": "ea10cd11ad08f01c50b0712bc481668857c9d2ddc8e6423968f2a31eeb97ac55", "question_id": "q11_aggregation", "split": "test", "cluster_id": "syn_03024_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 2, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 2068, "query_tokens": 456, "message_tokens": 363, "message_utf8_bytes": 834, "cache_tensor_bytes": null, "serialized_payload_bytes": 834, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 11483.192084007896, "hops": {"1": 6936.468709027395, "2": 4546.723374980502}, "receiver_prefill_ms": 185.12491695582867, "ttft_ms": 11.866500019095838, "decode_ms": 82.7463329769671}, "total_wall_ms": 11763.31404200755, "ttft_ms": 11.866500019095838, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 7}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "8066734", "evidence_artifact": {"ledger": {"facts_total": 66, "facts_retained": 8, "retention_rate": 0.12121212121212122, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.17647058823529413, "status": 0.16666666666666666, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 770}, "ledger_by_hop": {"1": {"facts_total": 66, "facts_retained": 6, "retention_rate": 0.09090909090909091, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.17647058823529413, "status": 0.0, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 754}, "2": {"facts_total": 66, "facts_retained": 8, "retention_rate": 0.12121212121212122, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.17647058823529413, "status": 0.16666666666666666, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 770}}}, "reference_key_id": "keys/synthetic/syn_03024_2000.json", "task_scores": {"correct": 0.0, "abstained": 0.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "b7386bd5ac9c9b04a9f81a350d93a018a74bcc7cc6eb41441f5a45c9ea28d0ba", "output_hash": "167c75fdd6a41fca64228797ab8f3a1a3f994475246df0bffb2e060d5d2e4ff0"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "e89b2dcd24531c8826a8bab5", "attempt_id": "5beaba253e10", "run_group_id": "E4_core:A5|closed|qwen3b:h2", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:24:11.299335Z", "task_family": "synthetic", "document_id": "syn_03024_2000", "document_hash": "ea10cd11ad08f01c50b0712bc481668857c9d2ddc8e6423968f2a31eeb97ac55", "question_id": "q12_unit_conversion", "split": "test", "cluster_id": "syn_03024_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 2, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 2068, "query_tokens": 453, "message_tokens": 363, "message_utf8_bytes": 834, "cache_tensor_bytes": null, "serialized_payload_bytes": 834, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 11483.192084007896, "hops": {"1": 6936.468709027395, "2": 4546.723374980502}, "receiver_prefill_ms": 185.42724999133497, "ttft_ms": 11.843333020806313, "decode_ms": 11.862916988320649}, "total_wall_ms": 11692.689917981625, "ttft_ms": 11.843333020806313, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 66, "facts_retained": 8, "retention_rate": 0.12121212121212122, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.17647058823529413, "status": 0.16666666666666666, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 770}, "ledger_by_hop": {"1": {"facts_total": 66, "facts_retained": 6, "retention_rate": 0.09090909090909091, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.17647058823529413, "status": 0.0, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 754}, "2": {"facts_total": 66, "facts_retained": 8, "retention_rate": 0.12121212121212122, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.17647058823529413, "status": 0.16666666666666666, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 770}}}, "reference_key_id": "keys/synthetic/syn_03024_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "b7386bd5ac9c9b04a9f81a350d93a018a74bcc7cc6eb41441f5a45c9ea28d0ba", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "fc6513c10527df06b064bcb0", "attempt_id": "818234321c94", "run_group_id": "E4_core:A5|closed|qwen3b:h2", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:24:11.523472Z", "task_family": "synthetic", "document_id": "syn_03024_2000", "document_hash": "ea10cd11ad08f01c50b0712bc481668857c9d2ddc8e6423968f2a31eeb97ac55", "question_id": "q13_unanswerable", "split": "test", "cluster_id": "syn_03024_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 2, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 2068, "query_tokens": 457, "message_tokens": 363, "message_utf8_bytes": 834, "cache_tensor_bytes": null, "serialized_payload_bytes": 834, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 11483.192084007896, "hops": {"1": 6936.468709027395, "2": 4546.723374980502}, "receiver_prefill_ms": 185.0474999519065, "ttft_ms": 11.861416045576334, "decode_ms": 11.87649997882545}, "total_wall_ms": 11692.399959079921, "ttft_ms": 11.861416045576334, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 66, "facts_retained": 8, "retention_rate": 0.12121212121212122, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.17647058823529413, "status": 0.16666666666666666, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 770}, "ledger_by_hop": {"1": {"facts_total": 66, "facts_retained": 6, "retention_rate": 0.09090909090909091, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.17647058823529413, "status": 0.0, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 754}, "2": {"facts_total": 66, "facts_retained": 8, "retention_rate": 0.12121212121212122, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.17647058823529413, "status": 0.16666666666666666, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 770}}}, "reference_key_id": "keys/synthetic/syn_03024_2000.json", "task_scores": {"correct": 1.0, "abstained": 1.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "b7386bd5ac9c9b04a9f81a350d93a018a74bcc7cc6eb41441f5a45c9ea28d0ba", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "c1330581de4bcfb078be8cd3", "attempt_id": "4143f91773ab", "run_group_id": "E4_core:A5|closed|qwen3b:h2", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:24:11.748188Z", "task_family": "synthetic", "document_id": "syn_03024_2000", "document_hash": "ea10cd11ad08f01c50b0712bc481668857c9d2ddc8e6423968f2a31eeb97ac55", "question_id": "q14_unanswerable", "split": "test", "cluster_id": "syn_03024_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 2, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 2068, "query_tokens": 458, "message_tokens": 363, "message_utf8_bytes": 834, "cache_tensor_bytes": null, "serialized_payload_bytes": 834, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 11483.192084007896, "hops": {"1": 6936.468709027395, "2": 4546.723374980502}, "receiver_prefill_ms": 185.00295898411423, "ttft_ms": 11.922916048206389, "decode_ms": 11.834541917778552}, "total_wall_ms": 11692.35262600705, "ttft_ms": 11.922916048206389, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 66, "facts_retained": 8, "retention_rate": 0.12121212121212122, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.17647058823529413, "status": 0.16666666666666666, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 770}, "ledger_by_hop": {"1": {"facts_total": 66, "facts_retained": 6, "retention_rate": 0.09090909090909091, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.17647058823529413, "status": 0.0, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 754}, "2": {"facts_total": 66, "facts_retained": 8, "retention_rate": 0.12121212121212122, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.17647058823529413, "status": 0.16666666666666666, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 770}}}, "reference_key_id": "keys/synthetic/syn_03024_2000.json", "task_scores": {"correct": 1.0, "abstained": 1.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "b7386bd5ac9c9b04a9f81a350d93a018a74bcc7cc6eb41441f5a45c9ea28d0ba", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "16860c18e815a31fc1d6290c", "attempt_id": "af19488cc824", "run_group_id": "E4_core:A5|closed|qwen3b:h4", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:24:21.095631Z", "task_family": "synthetic", "document_id": "syn_03024_2000", "document_hash": "ea10cd11ad08f01c50b0712bc481668857c9d2ddc8e6423968f2a31eeb97ac55", "question_id": "q00_single_fact", "split": "test", "cluster_id": "syn_03024_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 4, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 2068, "query_tokens": 445, "message_tokens": 363, "message_utf8_bytes": 834, "cache_tensor_bytes": null, "serialized_payload_bytes": 834, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 20603.92000002321, "hops": {"1": 6936.468709027395, "2": 4546.723374980502, "3": 4562.344291014597, "4": 4558.383625000715}, "receiver_prefill_ms": 164.44495902396739, "ttft_ms": 11.853790958411992, "decode_ms": 12.021875008940697}, "total_wall_ms": 20792.664124979638, "ttft_ms": 11.853790958411992, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 66, "facts_retained": 8, "retention_rate": 0.12121212121212122, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.17647058823529413, "status": 0.16666666666666666, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 770}, "ledger_by_hop": {"1": {"facts_total": 66, "facts_retained": 6, "retention_rate": 0.09090909090909091, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.17647058823529413, "status": 0.0, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 754}, "2": {"facts_total": 66, "facts_retained": 8, "retention_rate": 0.12121212121212122, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.17647058823529413, "status": 0.16666666666666666, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 770}, "3": {"facts_total": 66, "facts_retained": 8, "retention_rate": 0.12121212121212122, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.17647058823529413, "status": 0.16666666666666666, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 770}, "4": {"facts_total": 66, "facts_retained": 8, "retention_rate": 0.12121212121212122, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.17647058823529413, "status": 0.16666666666666666, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 770}}}, "reference_key_id": "keys/synthetic/syn_03024_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "b7386bd5ac9c9b04a9f81a350d93a018a74bcc7cc6eb41441f5a45c9ea28d0ba", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "e1b0632f3877eef1cdeffd83", "attempt_id": "d18f86d9645f", "run_group_id": "E4_core:A5|closed|qwen3b:h4", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:24:21.300363Z", "task_family": "synthetic", "document_id": "syn_03024_2000", "document_hash": "ea10cd11ad08f01c50b0712bc481668857c9d2ddc8e6423968f2a31eeb97ac55", "question_id": "q01_single_fact", "split": "test", "cluster_id": "syn_03024_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 4, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 2068, "query_tokens": 450, "message_tokens": 363, "message_utf8_bytes": 834, "cache_tensor_bytes": null, "serialized_payload_bytes": 834, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 20603.92000002321, "hops": {"1": 6936.468709027395, "2": 4546.723374980502, "3": 4562.344291014597, "4": 4558.383625000715}, "receiver_prefill_ms": 184.59491699468344, "ttft_ms": 11.866250075399876, "decode_ms": 129.9334999639541}, "total_wall_ms": 20930.726041086018, "ttft_ms": 11.866250075399876, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 11}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "Lot 568-G in Mont-Laurier", "evidence_artifact": {"ledger": {"facts_total": 66, "facts_retained": 8, "retention_rate": 0.12121212121212122, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.17647058823529413, "status": 0.16666666666666666, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 770}, "ledger_by_hop": {"1": {"facts_total": 66, "facts_retained": 6, "retention_rate": 0.09090909090909091, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.17647058823529413, "status": 0.0, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 754}, "2": {"facts_total": 66, "facts_retained": 8, "retention_rate": 0.12121212121212122, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.17647058823529413, "status": 0.16666666666666666, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 770}, "3": {"facts_total": 66, "facts_retained": 8, "retention_rate": 0.12121212121212122, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.17647058823529413, "status": 0.16666666666666666, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 770}, "4": {"facts_total": 66, "facts_retained": 8, "retention_rate": 0.12121212121212122, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.17647058823529413, "status": 0.16666666666666666, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 770}}}, "reference_key_id": "keys/synthetic/syn_03024_2000.json", "task_scores": {"correct": 0.0, "abstained": 0.0, "f1": 0.5}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "b7386bd5ac9c9b04a9f81a350d93a018a74bcc7cc6eb41441f5a45c9ea28d0ba", "output_hash": "98e007faa5a10742ad74f2778a64e3ea0d6b72bec1c0392200d475073d7bf082"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "ad77b3a14eb57439dd764b47", "attempt_id": "2ec5efef1d5e", "run_group_id": "E4_core:A5|closed|qwen3b:h4", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:24:21.642320Z", "task_family": "synthetic", "document_id": "syn_03024_2000", "document_hash": "ea10cd11ad08f01c50b0712bc481668857c9d2ddc8e6423968f2a31eeb97ac55", "question_id": "q02_exact_copy", "split": "test", "cluster_id": "syn_03024_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 4, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 2068, "query_tokens": 450, "message_tokens": 363, "message_utf8_bytes": 834, "cache_tensor_bytes": null, "serialized_payload_bytes": 834, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 20603.92000002321, "hops": {"1": 6936.468709027395, "2": 4546.723374980502, "3": 4562.344291014597, "4": 4558.383625000715}, "receiver_prefill_ms": 183.82549996022135, "ttft_ms": 11.845917091704905, "decode_ms": 11.891374946571887}, "total_wall_ms": 20811.870167031884, "ttft_ms": 11.845917091704905, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 66, "facts_retained": 8, "retention_rate": 0.12121212121212122, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.17647058823529413, "status": 0.16666666666666666, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 770}, "ledger_by_hop": {"1": {"facts_total": 66, "facts_retained": 6, "retention_rate": 0.09090909090909091, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.17647058823529413, "status": 0.0, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 754}, "2": {"facts_total": 66, "facts_retained": 8, "retention_rate": 0.12121212121212122, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.17647058823529413, "status": 0.16666666666666666, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 770}, "3": {"facts_total": 66, "facts_retained": 8, "retention_rate": 0.12121212121212122, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.17647058823529413, "status": 0.16666666666666666, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 770}, "4": {"facts_total": 66, "facts_retained": 8, "retention_rate": 0.12121212121212122, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.17647058823529413, "status": 0.16666666666666666, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 770}}}, "reference_key_id": "keys/synthetic/syn_03024_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "b7386bd5ac9c9b04a9f81a350d93a018a74bcc7cc6eb41441f5a45c9ea28d0ba", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "50c979e2fbd85e95a0c564d8", "attempt_id": "e23e623ad151", "run_group_id": "E4_core:A5|closed|qwen3b:h4", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:24:21.865499Z", "task_family": "synthetic", "document_id": "syn_03024_2000", "document_hash": "ea10cd11ad08f01c50b0712bc481668857c9d2ddc8e6423968f2a31eeb97ac55", "question_id": "q03_exact_copy", "split": "test", "cluster_id": "syn_03024_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 4, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 2068, "query_tokens": 445, "message_tokens": 363, "message_utf8_bytes": 834, "cache_tensor_bytes": null, "serialized_payload_bytes": 834, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 20603.92000002321, "hops": {"1": 6936.468709027395, "2": 4546.723374980502, "3": 4562.344291014597, "4": 4558.383625000715}, "receiver_prefill_ms": 164.1758339246735, "ttft_ms": 11.8742500199005, "decode_ms": 11.805207934230566}, "total_wall_ms": 20792.14008396957, "ttft_ms": 11.8742500199005, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 66, "facts_retained": 8, "retention_rate": 0.12121212121212122, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.17647058823529413, "status": 0.16666666666666666, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 770}, "ledger_by_hop": {"1": {"facts_total": 66, "facts_retained": 6, "retention_rate": 0.09090909090909091, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.17647058823529413, "status": 0.0, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 754}, "2": {"facts_total": 66, "facts_retained": 8, "retention_rate": 0.12121212121212122, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.17647058823529413, "status": 0.16666666666666666, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 770}, "3": {"facts_total": 66, "facts_retained": 8, "retention_rate": 0.12121212121212122, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.17647058823529413, "status": 0.16666666666666666, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 770}, "4": {"facts_total": 66, "facts_retained": 8, "retention_rate": 0.12121212121212122, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.17647058823529413, "status": 0.16666666666666666, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 770}}}, "reference_key_id": "keys/synthetic/syn_03024_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "b7386bd5ac9c9b04a9f81a350d93a018a74bcc7cc6eb41441f5a45c9ea28d0ba", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "92f883c1e23f82b06a6a65f2", "attempt_id": "d3df79261999", "run_group_id": "E4_core:A5|closed|qwen3b:h4", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:24:22.068800Z", "task_family": "synthetic", "document_id": "syn_03024_2000", "document_hash": "ea10cd11ad08f01c50b0712bc481668857c9d2ddc8e6423968f2a31eeb97ac55", "question_id": "q04_multi_hop", "split": "test", "cluster_id": "syn_03024_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 4, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 2068, "query_tokens": 451, "message_tokens": 363, "message_utf8_bytes": 834, "cache_tensor_bytes": null, "serialized_payload_bytes": 834, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 20603.92000002321, "hops": {"1": 6936.468709027395, "2": 4546.723374980502, "3": 4562.344291014597, "4": 4558.383625000715}, "receiver_prefill_ms": 185.03374990541488, "ttft_ms": 11.841833009384573, "decode_ms": 11.808292008936405}, "total_wall_ms": 20812.969458987936, "ttft_ms": 11.841833009384573, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 66, "facts_retained": 8, "retention_rate": 0.12121212121212122, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.17647058823529413, "status": 0.16666666666666666, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 770}, "ledger_by_hop": {"1": {"facts_total": 66, "facts_retained": 6, "retention_rate": 0.09090909090909091, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.17647058823529413, "status": 0.0, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 754}, "2": {"facts_total": 66, "facts_retained": 8, "retention_rate": 0.12121212121212122, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.17647058823529413, "status": 0.16666666666666666, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 770}, "3": {"facts_total": 66, "facts_retained": 8, "retention_rate": 0.12121212121212122, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.17647058823529413, "status": 0.16666666666666666, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 770}, "4": {"facts_total": 66, "facts_retained": 8, "retention_rate": 0.12121212121212122, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.17647058823529413, "status": 0.16666666666666666, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 770}}}, "reference_key_id": "keys/synthetic/syn_03024_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "b7386bd5ac9c9b04a9f81a350d93a018a74bcc7cc6eb41441f5a45c9ea28d0ba", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "349cca834979b7ec2d89e18e", "attempt_id": "7ec2854dbc72", "run_group_id": "E4_core:A5|closed|qwen3b:h4", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:24:22.292678Z", "task_family": "synthetic", "document_id": "syn_03024_2000", "document_hash": "ea10cd11ad08f01c50b0712bc481668857c9d2ddc8e6423968f2a31eeb97ac55", "question_id": "q05_multi_hop", "split": "test", "cluster_id": "syn_03024_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 4, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 2068, "query_tokens": 455, "message_tokens": 363, "message_utf8_bytes": 834, "cache_tensor_bytes": null, "serialized_payload_bytes": 834, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 20603.92000002321, "hops": {"1": 6936.468709027395, "2": 4546.723374980502, "3": 4562.344291014597, "4": 4558.383625000715}, "receiver_prefill_ms": 185.4181250091642, "ttft_ms": 11.826375033706427, "decode_ms": 11.85033400543034}, "total_wall_ms": 20813.36970801931, "ttft_ms": 11.826375033706427, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 66, "facts_retained": 8, "retention_rate": 0.12121212121212122, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.17647058823529413, "status": 0.16666666666666666, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 770}, "ledger_by_hop": {"1": {"facts_total": 66, "facts_retained": 6, "retention_rate": 0.09090909090909091, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.17647058823529413, "status": 0.0, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 754}, "2": {"facts_total": 66, "facts_retained": 8, "retention_rate": 0.12121212121212122, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.17647058823529413, "status": 0.16666666666666666, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 770}, "3": {"facts_total": 66, "facts_retained": 8, "retention_rate": 0.12121212121212122, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.17647058823529413, "status": 0.16666666666666666, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 770}, "4": {"facts_total": 66, "facts_retained": 8, "retention_rate": 0.12121212121212122, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.17647058823529413, "status": 0.16666666666666666, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 770}}}, "reference_key_id": "keys/synthetic/syn_03024_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "b7386bd5ac9c9b04a9f81a350d93a018a74bcc7cc6eb41441f5a45c9ea28d0ba", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "a447592486df7a43f7858ee0", "attempt_id": "4ad05350b006", "run_group_id": "E4_core:A5|closed|qwen3b:h4", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:24:22.516739Z", "task_family": "synthetic", "document_id": "syn_03024_2000", "document_hash": "ea10cd11ad08f01c50b0712bc481668857c9d2ddc8e6423968f2a31eeb97ac55", "question_id": "q06_negation", "split": "test", "cluster_id": "syn_03024_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 4, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 2068, "query_tokens": 444, "message_tokens": 363, "message_utf8_bytes": 834, "cache_tensor_bytes": null, "serialized_payload_bytes": 834, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 20603.92000002321, "hops": {"1": 6936.468709027395, "2": 4546.723374980502, "3": 4562.344291014597, "4": 4558.383625000715}, "receiver_prefill_ms": 164.16779207065701, "ttft_ms": 11.946332990191877, "decode_ms": 11.776415980421007}, "total_wall_ms": 20792.149875080213, "ttft_ms": 11.946332990191877, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "no", "evidence_artifact": {"ledger": {"facts_total": 66, "facts_retained": 8, "retention_rate": 0.12121212121212122, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.17647058823529413, "status": 0.16666666666666666, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 770}, "ledger_by_hop": {"1": {"facts_total": 66, "facts_retained": 6, "retention_rate": 0.09090909090909091, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.17647058823529413, "status": 0.0, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 754}, "2": {"facts_total": 66, "facts_retained": 8, "retention_rate": 0.12121212121212122, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.17647058823529413, "status": 0.16666666666666666, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 770}, "3": {"facts_total": 66, "facts_retained": 8, "retention_rate": 0.12121212121212122, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.17647058823529413, "status": 0.16666666666666666, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 770}, "4": {"facts_total": 66, "facts_retained": 8, "retention_rate": 0.12121212121212122, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.17647058823529413, "status": 0.16666666666666666, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 770}}}, "reference_key_id": "keys/synthetic/syn_03024_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "b7386bd5ac9c9b04a9f81a350d93a018a74bcc7cc6eb41441f5a45c9ea28d0ba", "output_hash": "9390298f3fb0c5b160498935d79cb139aef28e1c47358b4bbba61862b9c26e59"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "90eccaf77c9a6a72aa1e095d", "attempt_id": "ffb7e7135bf0", "run_group_id": "E4_core:A5|closed|qwen3b:h4", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:24:22.719505Z", "task_family": "synthetic", "document_id": "syn_03024_2000", "document_hash": "ea10cd11ad08f01c50b0712bc481668857c9d2ddc8e6423968f2a31eeb97ac55", "question_id": "q07_negation", "split": "test", "cluster_id": "syn_03024_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 4, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 2068, "query_tokens": 446, "message_tokens": 363, "message_utf8_bytes": 834, "cache_tensor_bytes": null, "serialized_payload_bytes": 834, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 20603.92000002321, "hops": {"1": 6936.468709027395, "2": 4546.723374980502, "3": 4562.344291014597, "4": 4558.383625000715}, "receiver_prefill_ms": 164.6102089434862, "ttft_ms": 11.85841707047075, "decode_ms": 11.880625039339066}, "total_wall_ms": 20792.62541711796, "ttft_ms": 11.85841707047075, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "no", "evidence_artifact": {"ledger": {"facts_total": 66, "facts_retained": 8, "retention_rate": 0.12121212121212122, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.17647058823529413, "status": 0.16666666666666666, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 770}, "ledger_by_hop": {"1": {"facts_total": 66, "facts_retained": 6, "retention_rate": 0.09090909090909091, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.17647058823529413, "status": 0.0, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 754}, "2": {"facts_total": 66, "facts_retained": 8, "retention_rate": 0.12121212121212122, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.17647058823529413, "status": 0.16666666666666666, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 770}, "3": {"facts_total": 66, "facts_retained": 8, "retention_rate": 0.12121212121212122, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.17647058823529413, "status": 0.16666666666666666, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 770}, "4": {"facts_total": 66, "facts_retained": 8, "retention_rate": 0.12121212121212122, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.17647058823529413, "status": 0.16666666666666666, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 770}}}, "reference_key_id": "keys/synthetic/syn_03024_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "b7386bd5ac9c9b04a9f81a350d93a018a74bcc7cc6eb41441f5a45c9ea28d0ba", "output_hash": "9390298f3fb0c5b160498935d79cb139aef28e1c47358b4bbba61862b9c26e59"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "936324d65d8c0a69e0195c8d", "attempt_id": "b0e05e044c09", "run_group_id": "E4_core:A5|closed|qwen3b:h4", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:24:22.923247Z", "task_family": "synthetic", "document_id": "syn_03024_2000", "document_hash": "ea10cd11ad08f01c50b0712bc481668857c9d2ddc8e6423968f2a31eeb97ac55", "question_id": "q08_revision", "split": "test", "cluster_id": "syn_03024_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 4, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 2068, "query_tokens": 462, "message_tokens": 363, "message_utf8_bytes": 834, "cache_tensor_bytes": null, "serialized_payload_bytes": 834, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 20603.92000002321, "hops": {"1": 6936.468709027395, "2": 4546.723374980502, "3": 4562.344291014597, "4": 4558.383625000715}, "receiver_prefill_ms": 185.19008299335837, "ttft_ms": 11.860250029712915, "decode_ms": 11.795790982432663}, "total_wall_ms": 20813.140416983515, "ttft_ms": 11.860250029712915, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 66, "facts_retained": 8, "retention_rate": 0.12121212121212122, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.17647058823529413, "status": 0.16666666666666666, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 770}, "ledger_by_hop": {"1": {"facts_total": 66, "facts_retained": 6, "retention_rate": 0.09090909090909091, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.17647058823529413, "status": 0.0, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 754}, "2": {"facts_total": 66, "facts_retained": 8, "retention_rate": 0.12121212121212122, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.17647058823529413, "status": 0.16666666666666666, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 770}, "3": {"facts_total": 66, "facts_retained": 8, "retention_rate": 0.12121212121212122, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.17647058823529413, "status": 0.16666666666666666, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 770}, "4": {"facts_total": 66, "facts_retained": 8, "retention_rate": 0.12121212121212122, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.17647058823529413, "status": 0.16666666666666666, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 770}}}, "reference_key_id": "keys/synthetic/syn_03024_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "b7386bd5ac9c9b04a9f81a350d93a018a74bcc7cc6eb41441f5a45c9ea28d0ba", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "73ad64ba128ba84d04f6c580", "attempt_id": "03eaa230538e", "run_group_id": "E4_core:A5|closed|qwen3b:h4", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:24:23.147509Z", "task_family": "synthetic", "document_id": "syn_03024_2000", "document_hash": "ea10cd11ad08f01c50b0712bc481668857c9d2ddc8e6423968f2a31eeb97ac55", "question_id": "q09_revision", "split": "test", "cluster_id": "syn_03024_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 4, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 2068, "query_tokens": 453, "message_tokens": 363, "message_utf8_bytes": 834, "cache_tensor_bytes": null, "serialized_payload_bytes": 834, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 20603.92000002321, "hops": {"1": 6936.468709027395, "2": 4546.723374980502, "3": 4562.344291014597, "4": 4558.383625000715}, "receiver_prefill_ms": 184.54370903782547, "ttft_ms": 11.871750000864267, "decode_ms": 11.827250011265278}, "total_wall_ms": 20812.533666961826, "ttft_ms": 11.871750000864267, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 66, "facts_retained": 8, "retention_rate": 0.12121212121212122, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.17647058823529413, "status": 0.16666666666666666, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 770}, "ledger_by_hop": {"1": {"facts_total": 66, "facts_retained": 6, "retention_rate": 0.09090909090909091, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.17647058823529413, "status": 0.0, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 754}, "2": {"facts_total": 66, "facts_retained": 8, "retention_rate": 0.12121212121212122, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.17647058823529413, "status": 0.16666666666666666, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 770}, "3": {"facts_total": 66, "facts_retained": 8, "retention_rate": 0.12121212121212122, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.17647058823529413, "status": 0.16666666666666666, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 770}, "4": {"facts_total": 66, "facts_retained": 8, "retention_rate": 0.12121212121212122, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.17647058823529413, "status": 0.16666666666666666, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 770}}}, "reference_key_id": "keys/synthetic/syn_03024_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "b7386bd5ac9c9b04a9f81a350d93a018a74bcc7cc6eb41441f5a45c9ea28d0ba", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "22dacd8f2646ee15f7da5c41", "attempt_id": "3fb733cecad0", "run_group_id": "E4_core:A5|closed|qwen3b:h4", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:24:23.371012Z", "task_family": "synthetic", "document_id": "syn_03024_2000", "document_hash": "ea10cd11ad08f01c50b0712bc481668857c9d2ddc8e6423968f2a31eeb97ac55", "question_id": "q10_temporal", "split": "test", "cluster_id": "syn_03024_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 4, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 2068, "query_tokens": 463, "message_tokens": 363, "message_utf8_bytes": 834, "cache_tensor_bytes": null, "serialized_payload_bytes": 834, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 20603.92000002321, "hops": {"1": 6936.468709027395, "2": 4546.723374980502, "3": 4562.344291014597, "4": 4558.383625000715}, "receiver_prefill_ms": 185.72470801882446, "ttft_ms": 11.868500034324825, "decode_ms": 11.852708994410932}, "total_wall_ms": 20813.73854202684, "ttft_ms": 11.868500034324825, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 66, "facts_retained": 8, "retention_rate": 0.12121212121212122, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.17647058823529413, "status": 0.16666666666666666, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 770}, "ledger_by_hop": {"1": {"facts_total": 66, "facts_retained": 6, "retention_rate": 0.09090909090909091, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.17647058823529413, "status": 0.0, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 754}, "2": {"facts_total": 66, "facts_retained": 8, "retention_rate": 0.12121212121212122, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.17647058823529413, "status": 0.16666666666666666, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 770}, "3": {"facts_total": 66, "facts_retained": 8, "retention_rate": 0.12121212121212122, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.17647058823529413, "status": 0.16666666666666666, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 770}, "4": {"facts_total": 66, "facts_retained": 8, "retention_rate": 0.12121212121212122, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.17647058823529413, "status": 0.16666666666666666, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 770}}}, "reference_key_id": "keys/synthetic/syn_03024_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "b7386bd5ac9c9b04a9f81a350d93a018a74bcc7cc6eb41441f5a45c9ea28d0ba", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "7746f175dc12b1ba6e7f3a25", "attempt_id": "2d4c8067449c", "run_group_id": "E4_core:A5|closed|qwen3b:h4", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:24:23.595542Z", "task_family": "synthetic", "document_id": "syn_03024_2000", "document_hash": "ea10cd11ad08f01c50b0712bc481668857c9d2ddc8e6423968f2a31eeb97ac55", "question_id": "q11_aggregation", "split": "test", "cluster_id": "syn_03024_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 4, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 2068, "query_tokens": 456, "message_tokens": 363, "message_utf8_bytes": 834, "cache_tensor_bytes": null, "serialized_payload_bytes": 834, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 20603.92000002321, "hops": {"1": 6936.468709027395, "2": 4546.723374980502, "3": 4562.344291014597, "4": 4558.383625000715}, "receiver_prefill_ms": 185.0237080361694, "ttft_ms": 11.849625036120415, "decode_ms": 82.70766702480614}, "total_wall_ms": 20883.87183402665, "ttft_ms": 11.849625036120415, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 7}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "8066734", "evidence_artifact": {"ledger": {"facts_total": 66, "facts_retained": 8, "retention_rate": 0.12121212121212122, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.17647058823529413, "status": 0.16666666666666666, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 770}, "ledger_by_hop": {"1": {"facts_total": 66, "facts_retained": 6, "retention_rate": 0.09090909090909091, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.17647058823529413, "status": 0.0, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 754}, "2": {"facts_total": 66, "facts_retained": 8, "retention_rate": 0.12121212121212122, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.17647058823529413, "status": 0.16666666666666666, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 770}, "3": {"facts_total": 66, "facts_retained": 8, "retention_rate": 0.12121212121212122, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.17647058823529413, "status": 0.16666666666666666, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 770}, "4": {"facts_total": 66, "facts_retained": 8, "retention_rate": 0.12121212121212122, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.17647058823529413, "status": 0.16666666666666666, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 770}}}, "reference_key_id": "keys/synthetic/syn_03024_2000.json", "task_scores": {"correct": 0.0, "abstained": 0.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "b7386bd5ac9c9b04a9f81a350d93a018a74bcc7cc6eb41441f5a45c9ea28d0ba", "output_hash": "167c75fdd6a41fca64228797ab8f3a1a3f994475246df0bffb2e060d5d2e4ff0"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "913c805305f049c259216d9e", "attempt_id": "87891aab8478", "run_group_id": "E4_core:A5|closed|qwen3b:h4", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:24:23.890402Z", "task_family": "synthetic", "document_id": "syn_03024_2000", "document_hash": "ea10cd11ad08f01c50b0712bc481668857c9d2ddc8e6423968f2a31eeb97ac55", "question_id": "q12_unit_conversion", "split": "test", "cluster_id": "syn_03024_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 4, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 2068, "query_tokens": 453, "message_tokens": 363, "message_utf8_bytes": 834, "cache_tensor_bytes": null, "serialized_payload_bytes": 834, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 20603.92000002321, "hops": {"1": 6936.468709027395, "2": 4546.723374980502, "3": 4562.344291014597, "4": 4558.383625000715}, "receiver_prefill_ms": 185.41999999433756, "ttft_ms": 11.828709044493735, "decode_ms": 11.834709090180695}, "total_wall_ms": 20813.37879097555, "ttft_ms": 11.828709044493735, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 66, "facts_retained": 8, "retention_rate": 0.12121212121212122, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.17647058823529413, "status": 0.16666666666666666, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 770}, "ledger_by_hop": {"1": {"facts_total": 66, "facts_retained": 6, "retention_rate": 0.09090909090909091, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.17647058823529413, "status": 0.0, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 754}, "2": {"facts_total": 66, "facts_retained": 8, "retention_rate": 0.12121212121212122, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.17647058823529413, "status": 0.16666666666666666, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 770}, "3": {"facts_total": 66, "facts_retained": 8, "retention_rate": 0.12121212121212122, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.17647058823529413, "status": 0.16666666666666666, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 770}, "4": {"facts_total": 66, "facts_retained": 8, "retention_rate": 0.12121212121212122, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.17647058823529413, "status": 0.16666666666666666, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 770}}}, "reference_key_id": "keys/synthetic/syn_03024_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "b7386bd5ac9c9b04a9f81a350d93a018a74bcc7cc6eb41441f5a45c9ea28d0ba", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "d478b15190aca223b51f5c76", "attempt_id": "3e5fe5dde7f6", "run_group_id": "E4_core:A5|closed|qwen3b:h4", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:24:24.115247Z", "task_family": "synthetic", "document_id": "syn_03024_2000", "document_hash": "ea10cd11ad08f01c50b0712bc481668857c9d2ddc8e6423968f2a31eeb97ac55", "question_id": "q13_unanswerable", "split": "test", "cluster_id": "syn_03024_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 4, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 2068, "query_tokens": 457, "message_tokens": 363, "message_utf8_bytes": 834, "cache_tensor_bytes": null, "serialized_payload_bytes": 834, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 20603.92000002321, "hops": {"1": 6936.468709027395, "2": 4546.723374980502, "3": 4562.344291014597, "4": 4558.383625000715}, "receiver_prefill_ms": 184.9063750123605, "ttft_ms": 11.823124950751662, "decode_ms": 11.84625003952533}, "total_wall_ms": 20812.855791067705, "ttft_ms": 11.823124950751662, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 66, "facts_retained": 8, "retention_rate": 0.12121212121212122, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.17647058823529413, "status": 0.16666666666666666, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 770}, "ledger_by_hop": {"1": {"facts_total": 66, "facts_retained": 6, "retention_rate": 0.09090909090909091, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.17647058823529413, "status": 0.0, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 754}, "2": {"facts_total": 66, "facts_retained": 8, "retention_rate": 0.12121212121212122, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.17647058823529413, "status": 0.16666666666666666, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 770}, "3": {"facts_total": 66, "facts_retained": 8, "retention_rate": 0.12121212121212122, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.17647058823529413, "status": 0.16666666666666666, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 770}, "4": {"facts_total": 66, "facts_retained": 8, "retention_rate": 0.12121212121212122, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.17647058823529413, "status": 0.16666666666666666, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 770}}}, "reference_key_id": "keys/synthetic/syn_03024_2000.json", "task_scores": {"correct": 1.0, "abstained": 1.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "b7386bd5ac9c9b04a9f81a350d93a018a74bcc7cc6eb41441f5a45c9ea28d0ba", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "cf124bb63534ad2b693d63c7", "attempt_id": "2b5c55ab6000", "run_group_id": "E4_core:A5|closed|qwen3b:h4", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:24:24.338817Z", "task_family": "synthetic", "document_id": "syn_03024_2000", "document_hash": "ea10cd11ad08f01c50b0712bc481668857c9d2ddc8e6423968f2a31eeb97ac55", "question_id": "q14_unanswerable", "split": "test", "cluster_id": "syn_03024_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 4, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 2068, "query_tokens": 458, "message_tokens": 363, "message_utf8_bytes": 834, "cache_tensor_bytes": null, "serialized_payload_bytes": 834, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 20603.92000002321, "hops": {"1": 6936.468709027395, "2": 4546.723374980502, "3": 4562.344291014597, "4": 4558.383625000715}, "receiver_prefill_ms": 185.54220802616328, "ttft_ms": 11.914875009097159, "decode_ms": 11.851749964989722}, "total_wall_ms": 20813.61258297693, "ttft_ms": 11.914875009097159, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 66, "facts_retained": 8, "retention_rate": 0.12121212121212122, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.17647058823529413, "status": 0.16666666666666666, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 770}, "ledger_by_hop": {"1": {"facts_total": 66, "facts_retained": 6, "retention_rate": 0.09090909090909091, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.17647058823529413, "status": 0.0, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 754}, "2": {"facts_total": 66, "facts_retained": 8, "retention_rate": 0.12121212121212122, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.17647058823529413, "status": 0.16666666666666666, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 770}, "3": {"facts_total": 66, "facts_retained": 8, "retention_rate": 0.12121212121212122, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.17647058823529413, "status": 0.16666666666666666, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 770}, "4": {"facts_total": 66, "facts_retained": 8, "retention_rate": 0.12121212121212122, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.17647058823529413, "status": 0.16666666666666666, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 770}}}, "reference_key_id": "keys/synthetic/syn_03024_2000.json", "task_scores": {"correct": 1.0, "abstained": 1.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "b7386bd5ac9c9b04a9f81a350d93a018a74bcc7cc6eb41441f5a45c9ea28d0ba", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "27407fe5ddd3bded10c90e65", "attempt_id": "29fc9dd12987", "run_group_id": "E4_core:A5|closed|qwen3b:h8", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:24:42.808196Z", "task_family": "synthetic", "document_id": "syn_03024_2000", "document_hash": "ea10cd11ad08f01c50b0712bc481668857c9d2ddc8e6423968f2a31eeb97ac55", "question_id": "q00_single_fact", "split": "test", "cluster_id": "syn_03024_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 8, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 2068, "query_tokens": 445, "message_tokens": 363, "message_utf8_bytes": 834, "cache_tensor_bytes": null, "serialized_payload_bytes": 834, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 38844.63216701988, "hops": {"1": 6936.468709027395, "2": 4546.723374980502, "3": 4562.344291014597, "4": 4558.383625000715, "5": 4559.562999987975, "6": 4558.797375066206, "7": 4563.401291961782, "8": 4558.95049998071}, "receiver_prefill_ms": 164.1359580680728, "ttft_ms": 11.880041914992034, "decode_ms": 11.849083006381989}, "total_wall_ms": 39032.90354204364, "ttft_ms": 11.880041914992034, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 66, "facts_retained": 8, "retention_rate": 0.12121212121212122, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.17647058823529413, "status": 0.16666666666666666, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 770}, "ledger_by_hop": {"1": {"facts_total": 66, "facts_retained": 6, "retention_rate": 0.09090909090909091, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.17647058823529413, "status": 0.0, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 754}, "2": {"facts_total": 66, "facts_retained": 8, "retention_rate": 0.12121212121212122, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.17647058823529413, "status": 0.16666666666666666, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 770}, "3": {"facts_total": 66, "facts_retained": 8, "retention_rate": 0.12121212121212122, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.17647058823529413, "status": 0.16666666666666666, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 770}, "4": {"facts_total": 66, "facts_retained": 8, "retention_rate": 0.12121212121212122, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.17647058823529413, "status": 0.16666666666666666, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 770}, "5": {"facts_total": 66, "facts_retained": 8, "retention_rate": 0.12121212121212122, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.17647058823529413, "status": 0.16666666666666666, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 770}, "6": {"facts_total": 66, "facts_retained": 8, "retention_rate": 0.12121212121212122, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.17647058823529413, "status": 0.16666666666666666, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 770}, "7": {"facts_total": 66, "facts_retained": 8, "retention_rate": 0.12121212121212122, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.17647058823529413, "status": 0.16666666666666666, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 770}, "8": {"facts_total": 66, "facts_retained": 8, "retention_rate": 0.12121212121212122, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.17647058823529413, "status": 0.16666666666666666, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 770}}}, "reference_key_id": "keys/synthetic/syn_03024_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "b7386bd5ac9c9b04a9f81a350d93a018a74bcc7cc6eb41441f5a45c9ea28d0ba", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "55d70990fc57abdde8c1c8d5", "attempt_id": "ae05d511537a", "run_group_id": "E4_core:A5|closed|qwen3b:h8", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:24:43.012743Z", "task_family": "synthetic", "document_id": "syn_03024_2000", "document_hash": "ea10cd11ad08f01c50b0712bc481668857c9d2ddc8e6423968f2a31eeb97ac55", "question_id": "q01_single_fact", "split": "test", "cluster_id": "syn_03024_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 8, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 2068, "query_tokens": 450, "message_tokens": 363, "message_utf8_bytes": 834, "cache_tensor_bytes": null, "serialized_payload_bytes": 834, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 38844.63216701988, "hops": {"1": 6936.468709027395, "2": 4546.723374980502, "3": 4562.344291014597, "4": 4558.383625000715, "5": 4559.562999987975, "6": 4558.797375066206, "7": 4563.401291961782, "8": 4558.95049998071}, "receiver_prefill_ms": 184.66995900962502, "ttft_ms": 11.843707994557917, "decode_ms": 129.85899997875094}, "total_wall_ms": 39171.41804203857, "ttft_ms": 11.843707994557917, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 11}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "Lot 568-G in Mont-Laurier", "evidence_artifact": {"ledger": {"facts_total": 66, "facts_retained": 8, "retention_rate": 0.12121212121212122, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.17647058823529413, "status": 0.16666666666666666, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 770}, "ledger_by_hop": {"1": {"facts_total": 66, "facts_retained": 6, "retention_rate": 0.09090909090909091, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.17647058823529413, "status": 0.0, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 754}, "2": {"facts_total": 66, "facts_retained": 8, "retention_rate": 0.12121212121212122, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.17647058823529413, "status": 0.16666666666666666, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 770}, "3": {"facts_total": 66, "facts_retained": 8, "retention_rate": 0.12121212121212122, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.17647058823529413, "status": 0.16666666666666666, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 770}, "4": {"facts_total": 66, "facts_retained": 8, "retention_rate": 0.12121212121212122, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.17647058823529413, "status": 0.16666666666666666, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 770}, "5": {"facts_total": 66, "facts_retained": 8, "retention_rate": 0.12121212121212122, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.17647058823529413, "status": 0.16666666666666666, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 770}, "6": {"facts_total": 66, "facts_retained": 8, "retention_rate": 0.12121212121212122, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.17647058823529413, "status": 0.16666666666666666, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 770}, "7": {"facts_total": 66, "facts_retained": 8, "retention_rate": 0.12121212121212122, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.17647058823529413, "status": 0.16666666666666666, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 770}, "8": {"facts_total": 66, "facts_retained": 8, "retention_rate": 0.12121212121212122, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.17647058823529413, "status": 0.16666666666666666, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 770}}}, "reference_key_id": "keys/synthetic/syn_03024_2000.json", "task_scores": {"correct": 0.0, "abstained": 0.0, "f1": 0.5}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "b7386bd5ac9c9b04a9f81a350d93a018a74bcc7cc6eb41441f5a45c9ea28d0ba", "output_hash": "98e007faa5a10742ad74f2778a64e3ea0d6b72bec1c0392200d475073d7bf082"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "befb512ec2e1301efd8e1569", "attempt_id": "55d8bcd5417a", "run_group_id": "E4_core:A5|closed|qwen3b:h8", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:24:43.355125Z", "task_family": "synthetic", "document_id": "syn_03024_2000", "document_hash": "ea10cd11ad08f01c50b0712bc481668857c9d2ddc8e6423968f2a31eeb97ac55", "question_id": "q02_exact_copy", "split": "test", "cluster_id": "syn_03024_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 8, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 2068, "query_tokens": 450, "message_tokens": 363, "message_utf8_bytes": 834, "cache_tensor_bytes": null, "serialized_payload_bytes": 834, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 38844.63216701988, "hops": {"1": 6936.468709027395, "2": 4546.723374980502, "3": 4562.344291014597, "4": 4558.383625000715, "5": 4559.562999987975, "6": 4558.797375066206, "7": 4563.401291961782, "8": 4558.95049998071}, "receiver_prefill_ms": 184.86441601999104, "ttft_ms": 11.810792027972639, "decode_ms": 11.849874979816377}, "total_wall_ms": 39053.5123340087, "ttft_ms": 11.810792027972639, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 66, "facts_retained": 8, "retention_rate": 0.12121212121212122, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.17647058823529413, "status": 0.16666666666666666, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 770}, "ledger_by_hop": {"1": {"facts_total": 66, "facts_retained": 6, "retention_rate": 0.09090909090909091, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.17647058823529413, "status": 0.0, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 754}, "2": {"facts_total": 66, "facts_retained": 8, "retention_rate": 0.12121212121212122, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.17647058823529413, "status": 0.16666666666666666, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 770}, "3": {"facts_total": 66, "facts_retained": 8, "retention_rate": 0.12121212121212122, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.17647058823529413, "status": 0.16666666666666666, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 770}, "4": {"facts_total": 66, "facts_retained": 8, "retention_rate": 0.12121212121212122, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.17647058823529413, "status": 0.16666666666666666, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 770}, "5": {"facts_total": 66, "facts_retained": 8, "retention_rate": 0.12121212121212122, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.17647058823529413, "status": 0.16666666666666666, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 770}, "6": {"facts_total": 66, "facts_retained": 8, "retention_rate": 0.12121212121212122, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.17647058823529413, "status": 0.16666666666666666, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 770}, "7": {"facts_total": 66, "facts_retained": 8, "retention_rate": 0.12121212121212122, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.17647058823529413, "status": 0.16666666666666666, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 770}, "8": {"facts_total": 66, "facts_retained": 8, "retention_rate": 0.12121212121212122, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.17647058823529413, "status": 0.16666666666666666, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 770}}}, "reference_key_id": "keys/synthetic/syn_03024_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "b7386bd5ac9c9b04a9f81a350d93a018a74bcc7cc6eb41441f5a45c9ea28d0ba", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "221f2a58197af486576ea3fa", "attempt_id": "7cd0aaa92bbc", "run_group_id": "E4_core:A5|closed|qwen3b:h8", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:24:43.579415Z", "task_family": "synthetic", "document_id": "syn_03024_2000", "document_hash": "ea10cd11ad08f01c50b0712bc481668857c9d2ddc8e6423968f2a31eeb97ac55", "question_id": "q03_exact_copy", "split": "test", "cluster_id": "syn_03024_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 8, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 2068, "query_tokens": 445, "message_tokens": 363, "message_utf8_bytes": 834, "cache_tensor_bytes": null, "serialized_payload_bytes": 834, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 38844.63216701988, "hops": {"1": 6936.468709027395, "2": 4546.723374980502, "3": 4562.344291014597, "4": 4558.383625000715, "5": 4559.562999987975, "6": 4558.797375066206, "7": 4563.401291961782, "8": 4558.95049998071}, "receiver_prefill_ms": 164.83379201963544, "ttft_ms": 11.806166963651776, "decode_ms": 11.875291005708277}, "total_wall_ms": 39033.50550006144, "ttft_ms": 11.806166963651776, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 66, "facts_retained": 8, "retention_rate": 0.12121212121212122, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.17647058823529413, "status": 0.16666666666666666, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 770}, "ledger_by_hop": {"1": {"facts_total": 66, "facts_retained": 6, "retention_rate": 0.09090909090909091, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.17647058823529413, "status": 0.0, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 754}, "2": {"facts_total": 66, "facts_retained": 8, "retention_rate": 0.12121212121212122, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.17647058823529413, "status": 0.16666666666666666, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 770}, "3": {"facts_total": 66, "facts_retained": 8, "retention_rate": 0.12121212121212122, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.17647058823529413, "status": 0.16666666666666666, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 770}, "4": {"facts_total": 66, "facts_retained": 8, "retention_rate": 0.12121212121212122, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.17647058823529413, "status": 0.16666666666666666, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 770}, "5": {"facts_total": 66, "facts_retained": 8, "retention_rate": 0.12121212121212122, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.17647058823529413, "status": 0.16666666666666666, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 770}, "6": {"facts_total": 66, "facts_retained": 8, "retention_rate": 0.12121212121212122, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.17647058823529413, "status": 0.16666666666666666, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 770}, "7": {"facts_total": 66, "facts_retained": 8, "retention_rate": 0.12121212121212122, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.17647058823529413, "status": 0.16666666666666666, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 770}, "8": {"facts_total": 66, "facts_retained": 8, "retention_rate": 0.12121212121212122, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.17647058823529413, "status": 0.16666666666666666, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 770}}}, "reference_key_id": "keys/synthetic/syn_03024_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "b7386bd5ac9c9b04a9f81a350d93a018a74bcc7cc6eb41441f5a45c9ea28d0ba", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "4f91af91cf769eed164604b9", "attempt_id": "754de6a6eaf6", "run_group_id": "E4_core:A5|closed|qwen3b:h8", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:24:43.783284Z", "task_family": "synthetic", "document_id": "syn_03024_2000", "document_hash": "ea10cd11ad08f01c50b0712bc481668857c9d2ddc8e6423968f2a31eeb97ac55", "question_id": "q04_multi_hop", "split": "test", "cluster_id": "syn_03024_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 8, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 2068, "query_tokens": 451, "message_tokens": 363, "message_utf8_bytes": 834, "cache_tensor_bytes": null, "serialized_payload_bytes": 834, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 38844.63216701988, "hops": {"1": 6936.468709027395, "2": 4546.723374980502, "3": 4562.344291014597, "4": 4558.383625000715, "5": 4559.562999987975, "6": 4558.797375066206, "7": 4563.401291961782, "8": 4558.95049998071}, "receiver_prefill_ms": 185.05345797166228, "ttft_ms": 11.875708936713636, "decode_ms": 12.00695801526308}, "total_wall_ms": 39053.94362506922, "ttft_ms": 11.875708936713636, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 66, "facts_retained": 8, "retention_rate": 0.12121212121212122, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.17647058823529413, "status": 0.16666666666666666, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 770}, "ledger_by_hop": {"1": {"facts_total": 66, "facts_retained": 6, "retention_rate": 0.09090909090909091, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.17647058823529413, "status": 0.0, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 754}, "2": {"facts_total": 66, "facts_retained": 8, "retention_rate": 0.12121212121212122, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.17647058823529413, "status": 0.16666666666666666, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 770}, "3": {"facts_total": 66, "facts_retained": 8, "retention_rate": 0.12121212121212122, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.17647058823529413, "status": 0.16666666666666666, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 770}, "4": {"facts_total": 66, "facts_retained": 8, "retention_rate": 0.12121212121212122, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.17647058823529413, "status": 0.16666666666666666, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 770}, "5": {"facts_total": 66, "facts_retained": 8, "retention_rate": 0.12121212121212122, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.17647058823529413, "status": 0.16666666666666666, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 770}, "6": {"facts_total": 66, "facts_retained": 8, "retention_rate": 0.12121212121212122, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.17647058823529413, "status": 0.16666666666666666, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 770}, "7": {"facts_total": 66, "facts_retained": 8, "retention_rate": 0.12121212121212122, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.17647058823529413, "status": 0.16666666666666666, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 770}, "8": {"facts_total": 66, "facts_retained": 8, "retention_rate": 0.12121212121212122, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.17647058823529413, "status": 0.16666666666666666, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 770}}}, "reference_key_id": "keys/synthetic/syn_03024_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "b7386bd5ac9c9b04a9f81a350d93a018a74bcc7cc6eb41441f5a45c9ea28d0ba", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "73f47c6d5c89beeb639b5467", "attempt_id": "3471a8240b39", "run_group_id": "E4_core:A5|closed|qwen3b:h8", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:24:44.007696Z", "task_family": "synthetic", "document_id": "syn_03024_2000", "document_hash": "ea10cd11ad08f01c50b0712bc481668857c9d2ddc8e6423968f2a31eeb97ac55", "question_id": "q05_multi_hop", "split": "test", "cluster_id": "syn_03024_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 8, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 2068, "query_tokens": 455, "message_tokens": 363, "message_utf8_bytes": 834, "cache_tensor_bytes": null, "serialized_payload_bytes": 834, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 38844.63216701988, "hops": {"1": 6936.468709027395, "2": 4546.723374980502, "3": 4562.344291014597, "4": 4558.383625000715, "5": 4559.562999987975, "6": 4558.797375066206, "7": 4563.401291961782, "8": 4558.95049998071}, "receiver_prefill_ms": 185.17374992370605, "ttft_ms": 11.88095798715949, "decode_ms": 11.854749987833202}, "total_wall_ms": 39053.93200099934, "ttft_ms": 11.88095798715949, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 66, "facts_retained": 8, "retention_rate": 0.12121212121212122, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.17647058823529413, "status": 0.16666666666666666, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 770}, "ledger_by_hop": {"1": {"facts_total": 66, "facts_retained": 6, "retention_rate": 0.09090909090909091, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.17647058823529413, "status": 0.0, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 754}, "2": {"facts_total": 66, "facts_retained": 8, "retention_rate": 0.12121212121212122, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.17647058823529413, "status": 0.16666666666666666, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 770}, "3": {"facts_total": 66, "facts_retained": 8, "retention_rate": 0.12121212121212122, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.17647058823529413, "status": 0.16666666666666666, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 770}, "4": {"facts_total": 66, "facts_retained": 8, "retention_rate": 0.12121212121212122, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.17647058823529413, "status": 0.16666666666666666, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 770}, "5": {"facts_total": 66, "facts_retained": 8, "retention_rate": 0.12121212121212122, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.17647058823529413, "status": 0.16666666666666666, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 770}, "6": {"facts_total": 66, "facts_retained": 8, "retention_rate": 0.12121212121212122, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.17647058823529413, "status": 0.16666666666666666, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 770}, "7": {"facts_total": 66, "facts_retained": 8, "retention_rate": 0.12121212121212122, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.17647058823529413, "status": 0.16666666666666666, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 770}, "8": {"facts_total": 66, "facts_retained": 8, "retention_rate": 0.12121212121212122, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.17647058823529413, "status": 0.16666666666666666, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 770}}}, "reference_key_id": "keys/synthetic/syn_03024_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "b7386bd5ac9c9b04a9f81a350d93a018a74bcc7cc6eb41441f5a45c9ea28d0ba", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "df07df6dea0cf6065addfa76", "attempt_id": "733b36c45770", "run_group_id": "E4_core:A5|closed|qwen3b:h8", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:24:44.231889Z", "task_family": "synthetic", "document_id": "syn_03024_2000", "document_hash": "ea10cd11ad08f01c50b0712bc481668857c9d2ddc8e6423968f2a31eeb97ac55", "question_id": "q06_negation", "split": "test", "cluster_id": "syn_03024_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 8, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 2068, "query_tokens": 444, "message_tokens": 363, "message_utf8_bytes": 834, "cache_tensor_bytes": null, "serialized_payload_bytes": 834, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 38844.63216701988, "hops": {"1": 6936.468709027395, "2": 4546.723374980502, "3": 4562.344291014597, "4": 4558.383625000715, "5": 4559.562999987975, "6": 4558.797375066206, "7": 4563.401291961782, "8": 4558.95049998071}, "receiver_prefill_ms": 164.25391589291394, "ttft_ms": 11.809166986495256, "decode_ms": 11.838875012472272}, "total_wall_ms": 39032.8872500686, "ttft_ms": 11.809166986495256, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "no", "evidence_artifact": {"ledger": {"facts_total": 66, "facts_retained": 8, "retention_rate": 0.12121212121212122, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.17647058823529413, "status": 0.16666666666666666, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 770}, "ledger_by_hop": {"1": {"facts_total": 66, "facts_retained": 6, "retention_rate": 0.09090909090909091, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.17647058823529413, "status": 0.0, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 754}, "2": {"facts_total": 66, "facts_retained": 8, "retention_rate": 0.12121212121212122, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.17647058823529413, "status": 0.16666666666666666, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 770}, "3": {"facts_total": 66, "facts_retained": 8, "retention_rate": 0.12121212121212122, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.17647058823529413, "status": 0.16666666666666666, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 770}, "4": {"facts_total": 66, "facts_retained": 8, "retention_rate": 0.12121212121212122, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.17647058823529413, "status": 0.16666666666666666, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 770}, "5": {"facts_total": 66, "facts_retained": 8, "retention_rate": 0.12121212121212122, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.17647058823529413, "status": 0.16666666666666666, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 770}, "6": {"facts_total": 66, "facts_retained": 8, "retention_rate": 0.12121212121212122, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.17647058823529413, "status": 0.16666666666666666, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 770}, "7": {"facts_total": 66, "facts_retained": 8, "retention_rate": 0.12121212121212122, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.17647058823529413, "status": 0.16666666666666666, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 770}, "8": {"facts_total": 66, "facts_retained": 8, "retention_rate": 0.12121212121212122, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.17647058823529413, "status": 0.16666666666666666, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 770}}}, "reference_key_id": "keys/synthetic/syn_03024_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "b7386bd5ac9c9b04a9f81a350d93a018a74bcc7cc6eb41441f5a45c9ea28d0ba", "output_hash": "9390298f3fb0c5b160498935d79cb139aef28e1c47358b4bbba61862b9c26e59"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "ab9e8dba7be7b85d63382125", "attempt_id": "ff1491509e5e", "run_group_id": "E4_core:A5|closed|qwen3b:h8", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:24:44.435308Z", "task_family": "synthetic", "document_id": "syn_03024_2000", "document_hash": "ea10cd11ad08f01c50b0712bc481668857c9d2ddc8e6423968f2a31eeb97ac55", "question_id": "q07_negation", "split": "test", "cluster_id": "syn_03024_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 8, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 2068, "query_tokens": 446, "message_tokens": 363, "message_utf8_bytes": 834, "cache_tensor_bytes": null, "serialized_payload_bytes": 834, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 38844.63216701988, "hops": {"1": 6936.468709027395, "2": 4546.723374980502, "3": 4562.344291014597, "4": 4558.383625000715, "5": 4559.562999987975, "6": 4558.797375066206, "7": 4563.401291961782, "8": 4558.95049998071}, "receiver_prefill_ms": 165.02370801754296, "ttft_ms": 11.89695904031396, "decode_ms": 11.877209064550698}, "total_wall_ms": 39033.79162505735, "ttft_ms": 11.89695904031396, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "no", "evidence_artifact": {"ledger": {"facts_total": 66, "facts_retained": 8, "retention_rate": 0.12121212121212122, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.17647058823529413, "status": 0.16666666666666666, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 770}, "ledger_by_hop": {"1": {"facts_total": 66, "facts_retained": 6, "retention_rate": 0.09090909090909091, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.17647058823529413, "status": 0.0, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 754}, "2": {"facts_total": 66, "facts_retained": 8, "retention_rate": 0.12121212121212122, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.17647058823529413, "status": 0.16666666666666666, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 770}, "3": {"facts_total": 66, "facts_retained": 8, "retention_rate": 0.12121212121212122, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.17647058823529413, "status": 0.16666666666666666, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 770}, "4": {"facts_total": 66, "facts_retained": 8, "retention_rate": 0.12121212121212122, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.17647058823529413, "status": 0.16666666666666666, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 770}, "5": {"facts_total": 66, "facts_retained": 8, "retention_rate": 0.12121212121212122, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.17647058823529413, "status": 0.16666666666666666, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 770}, "6": {"facts_total": 66, "facts_retained": 8, "retention_rate": 0.12121212121212122, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.17647058823529413, "status": 0.16666666666666666, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 770}, "7": {"facts_total": 66, "facts_retained": 8, "retention_rate": 0.12121212121212122, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.17647058823529413, "status": 0.16666666666666666, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 770}, "8": {"facts_total": 66, "facts_retained": 8, "retention_rate": 0.12121212121212122, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.17647058823529413, "status": 0.16666666666666666, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 770}}}, "reference_key_id": "keys/synthetic/syn_03024_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "b7386bd5ac9c9b04a9f81a350d93a018a74bcc7cc6eb41441f5a45c9ea28d0ba", "output_hash": "9390298f3fb0c5b160498935d79cb139aef28e1c47358b4bbba61862b9c26e59"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "52fa2b2c3cb580da7ec91eef", "attempt_id": "92f1081e2251", "run_group_id": "E4_core:A5|closed|qwen3b:h8", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:24:44.639515Z", "task_family": "synthetic", "document_id": "syn_03024_2000", "document_hash": "ea10cd11ad08f01c50b0712bc481668857c9d2ddc8e6423968f2a31eeb97ac55", "question_id": "q08_revision", "split": "test", "cluster_id": "syn_03024_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 8, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 2068, "query_tokens": 462, "message_tokens": 363, "message_utf8_bytes": 834, "cache_tensor_bytes": null, "serialized_payload_bytes": 834, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 38844.63216701988, "hops": {"1": 6936.468709027395, "2": 4546.723374980502, "3": 4562.344291014597, "4": 4558.383625000715, "5": 4559.562999987975, "6": 4558.797375066206, "7": 4563.401291961782, "8": 4558.95049998071}, "receiver_prefill_ms": 185.40212500374764, "ttft_ms": 11.866374989040196, "decode_ms": 11.911208042874932}, "total_wall_ms": 39054.18004200328, "ttft_ms": 11.866374989040196, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 66, "facts_retained": 8, "retention_rate": 0.12121212121212122, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.17647058823529413, "status": 0.16666666666666666, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 770}, "ledger_by_hop": {"1": {"facts_total": 66, "facts_retained": 6, "retention_rate": 0.09090909090909091, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.17647058823529413, "status": 0.0, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 754}, "2": {"facts_total": 66, "facts_retained": 8, "retention_rate": 0.12121212121212122, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.17647058823529413, "status": 0.16666666666666666, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 770}, "3": {"facts_total": 66, "facts_retained": 8, "retention_rate": 0.12121212121212122, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.17647058823529413, "status": 0.16666666666666666, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 770}, "4": {"facts_total": 66, "facts_retained": 8, "retention_rate": 0.12121212121212122, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.17647058823529413, "status": 0.16666666666666666, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 770}, "5": {"facts_total": 66, "facts_retained": 8, "retention_rate": 0.12121212121212122, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.17647058823529413, "status": 0.16666666666666666, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 770}, "6": {"facts_total": 66, "facts_retained": 8, "retention_rate": 0.12121212121212122, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.17647058823529413, "status": 0.16666666666666666, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 770}, "7": {"facts_total": 66, "facts_retained": 8, "retention_rate": 0.12121212121212122, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.17647058823529413, "status": 0.16666666666666666, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 770}, "8": {"facts_total": 66, "facts_retained": 8, "retention_rate": 0.12121212121212122, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.17647058823529413, "status": 0.16666666666666666, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 770}}}, "reference_key_id": "keys/synthetic/syn_03024_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "b7386bd5ac9c9b04a9f81a350d93a018a74bcc7cc6eb41441f5a45c9ea28d0ba", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "70862c8c72ebe95f2f2bb9b5", "attempt_id": "6e7c77d22f4b", "run_group_id": "E4_core:A5|closed|qwen3b:h8", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:24:44.864096Z", "task_family": "synthetic", "document_id": "syn_03024_2000", "document_hash": "ea10cd11ad08f01c50b0712bc481668857c9d2ddc8e6423968f2a31eeb97ac55", "question_id": "q09_revision", "split": "test", "cluster_id": "syn_03024_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 8, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 2068, "query_tokens": 453, "message_tokens": 363, "message_utf8_bytes": 834, "cache_tensor_bytes": null, "serialized_payload_bytes": 834, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 38844.63216701988, "hops": {"1": 6936.468709027395, "2": 4546.723374980502, "3": 4562.344291014597, "4": 4558.383625000715, "5": 4559.562999987975, "6": 4558.797375066206, "7": 4563.401291961782, "8": 4558.95049998071}, "receiver_prefill_ms": 185.03620906267315, "ttft_ms": 11.9964579353109, "decode_ms": 12.023207964375615}, "total_wall_ms": 39054.06704195775, "ttft_ms": 11.9964579353109, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 66, "facts_retained": 8, "retention_rate": 0.12121212121212122, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.17647058823529413, "status": 0.16666666666666666, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 770}, "ledger_by_hop": {"1": {"facts_total": 66, "facts_retained": 6, "retention_rate": 0.09090909090909091, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.17647058823529413, "status": 0.0, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 754}, "2": {"facts_total": 66, "facts_retained": 8, "retention_rate": 0.12121212121212122, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.17647058823529413, "status": 0.16666666666666666, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 770}, "3": {"facts_total": 66, "facts_retained": 8, "retention_rate": 0.12121212121212122, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.17647058823529413, "status": 0.16666666666666666, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 770}, "4": {"facts_total": 66, "facts_retained": 8, "retention_rate": 0.12121212121212122, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.17647058823529413, "status": 0.16666666666666666, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 770}, "5": {"facts_total": 66, "facts_retained": 8, "retention_rate": 0.12121212121212122, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.17647058823529413, "status": 0.16666666666666666, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 770}, "6": {"facts_total": 66, "facts_retained": 8, "retention_rate": 0.12121212121212122, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.17647058823529413, "status": 0.16666666666666666, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 770}, "7": {"facts_total": 66, "facts_retained": 8, "retention_rate": 0.12121212121212122, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.17647058823529413, "status": 0.16666666666666666, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 770}, "8": {"facts_total": 66, "facts_retained": 8, "retention_rate": 0.12121212121212122, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.17647058823529413, "status": 0.16666666666666666, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 770}}}, "reference_key_id": "keys/synthetic/syn_03024_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "b7386bd5ac9c9b04a9f81a350d93a018a74bcc7cc6eb41441f5a45c9ea28d0ba", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "2dc0a964239c683b3f179055", "attempt_id": "c4b0d24a0d0b", "run_group_id": "E4_core:A5|closed|qwen3b:h8", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:24:45.088682Z", "task_family": "synthetic", "document_id": "syn_03024_2000", "document_hash": "ea10cd11ad08f01c50b0712bc481668857c9d2ddc8e6423968f2a31eeb97ac55", "question_id": "q10_temporal", "split": "test", "cluster_id": "syn_03024_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 8, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 2068, "query_tokens": 463, "message_tokens": 363, "message_utf8_bytes": 834, "cache_tensor_bytes": null, "serialized_payload_bytes": 834, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 38844.63216701988, "hops": {"1": 6936.468709027395, "2": 4546.723374980502, "3": 4562.344291014597, "4": 4558.383625000715, "5": 4559.562999987975, "6": 4558.797375066206, "7": 4563.401291961782, "8": 4558.95049998071}, "receiver_prefill_ms": 185.9790829475969, "ttft_ms": 12.004542048089206, "decode_ms": 11.836834019050002}, "total_wall_ms": 39054.81691705063, "ttft_ms": 12.004542048089206, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 66, "facts_retained": 8, "retention_rate": 0.12121212121212122, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.17647058823529413, "status": 0.16666666666666666, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 770}, "ledger_by_hop": {"1": {"facts_total": 66, "facts_retained": 6, "retention_rate": 0.09090909090909091, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.17647058823529413, "status": 0.0, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 754}, "2": {"facts_total": 66, "facts_retained": 8, "retention_rate": 0.12121212121212122, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.17647058823529413, "status": 0.16666666666666666, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 770}, "3": {"facts_total": 66, "facts_retained": 8, "retention_rate": 0.12121212121212122, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.17647058823529413, "status": 0.16666666666666666, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 770}, "4": {"facts_total": 66, "facts_retained": 8, "retention_rate": 0.12121212121212122, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.17647058823529413, "status": 0.16666666666666666, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 770}, "5": {"facts_total": 66, "facts_retained": 8, "retention_rate": 0.12121212121212122, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.17647058823529413, "status": 0.16666666666666666, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 770}, "6": {"facts_total": 66, "facts_retained": 8, "retention_rate": 0.12121212121212122, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.17647058823529413, "status": 0.16666666666666666, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 770}, "7": {"facts_total": 66, "facts_retained": 8, "retention_rate": 0.12121212121212122, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.17647058823529413, "status": 0.16666666666666666, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 770}, "8": {"facts_total": 66, "facts_retained": 8, "retention_rate": 0.12121212121212122, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.17647058823529413, "status": 0.16666666666666666, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 770}}}, "reference_key_id": "keys/synthetic/syn_03024_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "b7386bd5ac9c9b04a9f81a350d93a018a74bcc7cc6eb41441f5a45c9ea28d0ba", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "3caaad1c53a1ddb32a816a58", "attempt_id": "3a00a9188b29", "run_group_id": "E4_core:A5|closed|qwen3b:h8", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:24:45.313751Z", "task_family": "synthetic", "document_id": "syn_03024_2000", "document_hash": "ea10cd11ad08f01c50b0712bc481668857c9d2ddc8e6423968f2a31eeb97ac55", "question_id": "q11_aggregation", "split": "test", "cluster_id": "syn_03024_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 8, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 2068, "query_tokens": 456, "message_tokens": 363, "message_utf8_bytes": 834, "cache_tensor_bytes": null, "serialized_payload_bytes": 834, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 38844.63216701988, "hops": {"1": 6936.468709027395, "2": 4546.723374980502, "3": 4562.344291014597, "4": 4558.383625000715, "5": 4559.562999987975, "6": 4558.797375066206, "7": 4563.401291961782, "8": 4558.95049998071}, "receiver_prefill_ms": 185.0465410389006, "ttft_ms": 11.896957992576063, "decode_ms": 82.89233292452991}, "total_wall_ms": 39124.85325103626, "ttft_ms": 11.896957992576063, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 7}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "8066734", "evidence_artifact": {"ledger": {"facts_total": 66, "facts_retained": 8, "retention_rate": 0.12121212121212122, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.17647058823529413, "status": 0.16666666666666666, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 770}, "ledger_by_hop": {"1": {"facts_total": 66, "facts_retained": 6, "retention_rate": 0.09090909090909091, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.17647058823529413, "status": 0.0, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 754}, "2": {"facts_total": 66, "facts_retained": 8, "retention_rate": 0.12121212121212122, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.17647058823529413, "status": 0.16666666666666666, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 770}, "3": {"facts_total": 66, "facts_retained": 8, "retention_rate": 0.12121212121212122, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.17647058823529413, "status": 0.16666666666666666, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 770}, "4": {"facts_total": 66, "facts_retained": 8, "retention_rate": 0.12121212121212122, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.17647058823529413, "status": 0.16666666666666666, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 770}, "5": {"facts_total": 66, "facts_retained": 8, "retention_rate": 0.12121212121212122, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.17647058823529413, "status": 0.16666666666666666, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 770}, "6": {"facts_total": 66, "facts_retained": 8, "retention_rate": 0.12121212121212122, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.17647058823529413, "status": 0.16666666666666666, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 770}, "7": {"facts_total": 66, "facts_retained": 8, "retention_rate": 0.12121212121212122, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.17647058823529413, "status": 0.16666666666666666, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 770}, "8": {"facts_total": 66, "facts_retained": 8, "retention_rate": 0.12121212121212122, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.17647058823529413, "status": 0.16666666666666666, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 770}}}, "reference_key_id": "keys/synthetic/syn_03024_2000.json", "task_scores": {"correct": 0.0, "abstained": 0.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "b7386bd5ac9c9b04a9f81a350d93a018a74bcc7cc6eb41441f5a45c9ea28d0ba", "output_hash": "167c75fdd6a41fca64228797ab8f3a1a3f994475246df0bffb2e060d5d2e4ff0"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "79fb8fd228a9c85c726bd42f", "attempt_id": "e7ec78e988b7", "run_group_id": "E4_core:A5|closed|qwen3b:h8", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:24:45.609438Z", "task_family": "synthetic", "document_id": "syn_03024_2000", "document_hash": "ea10cd11ad08f01c50b0712bc481668857c9d2ddc8e6423968f2a31eeb97ac55", "question_id": "q12_unit_conversion", "split": "test", "cluster_id": "syn_03024_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 8, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 2068, "query_tokens": 453, "message_tokens": 363, "message_utf8_bytes": 834, "cache_tensor_bytes": null, "serialized_payload_bytes": 834, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 38844.63216701988, "hops": {"1": 6936.468709027395, "2": 4546.723374980502, "3": 4562.344291014597, "4": 4558.383625000715, "5": 4559.562999987975, "6": 4558.797375066206, "7": 4563.401291961782, "8": 4558.95049998071}, "receiver_prefill_ms": 184.29700005799532, "ttft_ms": 11.732916929759085, "decode_ms": 11.949750012718141}, "total_wall_ms": 39053.147459053434, "ttft_ms": 11.732916929759085, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 66, "facts_retained": 8, "retention_rate": 0.12121212121212122, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.17647058823529413, "status": 0.16666666666666666, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 770}, "ledger_by_hop": {"1": {"facts_total": 66, "facts_retained": 6, "retention_rate": 0.09090909090909091, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.17647058823529413, "status": 0.0, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 754}, "2": {"facts_total": 66, "facts_retained": 8, "retention_rate": 0.12121212121212122, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.17647058823529413, "status": 0.16666666666666666, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 770}, "3": {"facts_total": 66, "facts_retained": 8, "retention_rate": 0.12121212121212122, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.17647058823529413, "status": 0.16666666666666666, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 770}, "4": {"facts_total": 66, "facts_retained": 8, "retention_rate": 0.12121212121212122, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.17647058823529413, "status": 0.16666666666666666, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 770}, "5": {"facts_total": 66, "facts_retained": 8, "retention_rate": 0.12121212121212122, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.17647058823529413, "status": 0.16666666666666666, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 770}, "6": {"facts_total": 66, "facts_retained": 8, "retention_rate": 0.12121212121212122, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.17647058823529413, "status": 0.16666666666666666, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 770}, "7": {"facts_total": 66, "facts_retained": 8, "retention_rate": 0.12121212121212122, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.17647058823529413, "status": 0.16666666666666666, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 770}, "8": {"facts_total": 66, "facts_retained": 8, "retention_rate": 0.12121212121212122, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.17647058823529413, "status": 0.16666666666666666, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 770}}}, "reference_key_id": "keys/synthetic/syn_03024_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "b7386bd5ac9c9b04a9f81a350d93a018a74bcc7cc6eb41441f5a45c9ea28d0ba", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "c97d85eed8d58bde93d7c368", "attempt_id": "24ad152e9ec5", "run_group_id": "E4_core:A5|closed|qwen3b:h8", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:24:45.833559Z", "task_family": "synthetic", "document_id": "syn_03024_2000", "document_hash": "ea10cd11ad08f01c50b0712bc481668857c9d2ddc8e6423968f2a31eeb97ac55", "question_id": "q13_unanswerable", "split": "test", "cluster_id": "syn_03024_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 8, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 2068, "query_tokens": 457, "message_tokens": 363, "message_utf8_bytes": 834, "cache_tensor_bytes": null, "serialized_payload_bytes": 834, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 38844.63216701988, "hops": {"1": 6936.468709027395, "2": 4546.723374980502, "3": 4562.344291014597, "4": 4558.383625000715, "5": 4559.562999987975, "6": 4558.797375066206, "7": 4563.401291961782, "8": 4558.95049998071}, "receiver_prefill_ms": 185.73754199314862, "ttft_ms": 11.87799999024719, "decode_ms": 11.943457997404039}, "total_wall_ms": 39054.58787502721, "ttft_ms": 11.87799999024719, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 66, "facts_retained": 8, "retention_rate": 0.12121212121212122, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.17647058823529413, "status": 0.16666666666666666, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 770}, "ledger_by_hop": {"1": {"facts_total": 66, "facts_retained": 6, "retention_rate": 0.09090909090909091, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.17647058823529413, "status": 0.0, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 754}, "2": {"facts_total": 66, "facts_retained": 8, "retention_rate": 0.12121212121212122, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.17647058823529413, "status": 0.16666666666666666, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 770}, "3": {"facts_total": 66, "facts_retained": 8, "retention_rate": 0.12121212121212122, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.17647058823529413, "status": 0.16666666666666666, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 770}, "4": {"facts_total": 66, "facts_retained": 8, "retention_rate": 0.12121212121212122, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.17647058823529413, "status": 0.16666666666666666, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 770}, "5": {"facts_total": 66, "facts_retained": 8, "retention_rate": 0.12121212121212122, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.17647058823529413, "status": 0.16666666666666666, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 770}, "6": {"facts_total": 66, "facts_retained": 8, "retention_rate": 0.12121212121212122, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.17647058823529413, "status": 0.16666666666666666, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 770}, "7": {"facts_total": 66, "facts_retained": 8, "retention_rate": 0.12121212121212122, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.17647058823529413, "status": 0.16666666666666666, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 770}, "8": {"facts_total": 66, "facts_retained": 8, "retention_rate": 0.12121212121212122, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.17647058823529413, "status": 0.16666666666666666, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 770}}}, "reference_key_id": "keys/synthetic/syn_03024_2000.json", "task_scores": {"correct": 1.0, "abstained": 1.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "b7386bd5ac9c9b04a9f81a350d93a018a74bcc7cc6eb41441f5a45c9ea28d0ba", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "6b0507a68f2d9d20b2279c54", "attempt_id": "38e783ff4ab5", "run_group_id": "E4_core:A5|closed|qwen3b:h8", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:24:46.059629Z", "task_family": "synthetic", "document_id": "syn_03024_2000", "document_hash": "ea10cd11ad08f01c50b0712bc481668857c9d2ddc8e6423968f2a31eeb97ac55", "question_id": "q14_unanswerable", "split": "test", "cluster_id": "syn_03024_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 8, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 2068, "query_tokens": 458, "message_tokens": 363, "message_utf8_bytes": 834, "cache_tensor_bytes": null, "serialized_payload_bytes": 834, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 38844.63216701988, "hops": {"1": 6936.468709027395, "2": 4546.723374980502, "3": 4562.344291014597, "4": 4558.383625000715, "5": 4559.562999987975, "6": 4558.797375066206, "7": 4563.401291961782, "8": 4558.95049998071}, "receiver_prefill_ms": 185.37674995604903, "ttft_ms": 11.859874939545989, "decode_ms": 11.881249956786633}, "total_wall_ms": 39054.13695902098, "ttft_ms": 11.859874939545989, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 66, "facts_retained": 8, "retention_rate": 0.12121212121212122, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.17647058823529413, "status": 0.16666666666666666, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 770}, "ledger_by_hop": {"1": {"facts_total": 66, "facts_retained": 6, "retention_rate": 0.09090909090909091, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.17647058823529413, "status": 0.0, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 754}, "2": {"facts_total": 66, "facts_retained": 8, "retention_rate": 0.12121212121212122, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.17647058823529413, "status": 0.16666666666666666, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 770}, "3": {"facts_total": 66, "facts_retained": 8, "retention_rate": 0.12121212121212122, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.17647058823529413, "status": 0.16666666666666666, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 770}, "4": {"facts_total": 66, "facts_retained": 8, "retention_rate": 0.12121212121212122, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.17647058823529413, "status": 0.16666666666666666, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 770}, "5": {"facts_total": 66, "facts_retained": 8, "retention_rate": 0.12121212121212122, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.17647058823529413, "status": 0.16666666666666666, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 770}, "6": {"facts_total": 66, "facts_retained": 8, "retention_rate": 0.12121212121212122, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.17647058823529413, "status": 0.16666666666666666, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 770}, "7": {"facts_total": 66, "facts_retained": 8, "retention_rate": 0.12121212121212122, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.17647058823529413, "status": 0.16666666666666666, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 770}, "8": {"facts_total": 66, "facts_retained": 8, "retention_rate": 0.12121212121212122, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.17647058823529413, "status": 0.16666666666666666, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 770}}}, "reference_key_id": "keys/synthetic/syn_03024_2000.json", "task_scores": {"correct": 1.0, "abstained": 1.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "b7386bd5ac9c9b04a9f81a350d93a018a74bcc7cc6eb41441f5a45c9ea28d0ba", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "ecc468121b857927f149d66b", "attempt_id": "7b3835396b72", "run_group_id": "E4_core:A4|repairable|qwen3b:h1", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:24:53.231843Z", "task_family": "synthetic", "document_id": "syn_03024_2000", "document_hash": "ea10cd11ad08f01c50b0712bc481668857c9d2ddc8e6423968f2a31eeb97ac55", "question_id": "q00_single_fact", "split": "test", "cluster_id": "syn_03024_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 1, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 2068, "query_tokens": 595, "message_tokens": 512, "message_utf8_bytes": 1508, "cache_tensor_bytes": null, "serialized_payload_bytes": 1508, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 6945.187625009567, "hops": {"1": 6945.187625009567}, "receiver_prefill_ms": 227.01366699766368, "ttft_ms": 11.890541994944215, "decode_ms": 12.150709051638842}, "total_wall_ms": 7196.774625102989, "ttft_ms": 11.890541994944215, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 66, "facts_retained": 21, "retention_rate": 0.3181818181818182, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.35294117647058826, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1508}, "ledger_by_hop": {"1": {"facts_total": 66, "facts_retained": 21, "retention_rate": 0.3181818181818182, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.35294117647058826, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1508}}}, "reference_key_id": "keys/synthetic/syn_03024_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "34fcc1e23e5a866b5b7d7e222f0380ba771c4005d3732e9747f026c211840d43", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "c9c81ead6074b873905fcace", "attempt_id": "36ba97098c36", "run_group_id": "E4_core:A4|repairable|qwen3b:h1", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:24:53.499518Z", "task_family": "synthetic", "document_id": "syn_03024_2000", "document_hash": "ea10cd11ad08f01c50b0712bc481668857c9d2ddc8e6423968f2a31eeb97ac55", "question_id": "q01_single_fact", "split": "test", "cluster_id": "syn_03024_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 1, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 2068, "query_tokens": 600, "message_tokens": 512, "message_utf8_bytes": 1508, "cache_tensor_bytes": null, "serialized_payload_bytes": 1508, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 6945.187625009567, "hops": {"1": 6945.187625009567}, "receiver_prefill_ms": 227.2832499584183, "ttft_ms": 11.905667022801936, "decode_ms": 12.08170794416219}, "total_wall_ms": 7196.945791016333, "ttft_ms": 11.905667022801936, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 66, "facts_retained": 21, "retention_rate": 0.3181818181818182, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.35294117647058826, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1508}, "ledger_by_hop": {"1": {"facts_total": 66, "facts_retained": 21, "retention_rate": 0.3181818181818182, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.35294117647058826, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1508}}}, "reference_key_id": "keys/synthetic/syn_03024_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "34fcc1e23e5a866b5b7d7e222f0380ba771c4005d3732e9747f026c211840d43", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "0aa9738e1cb9411ae4f12633", "attempt_id": "adf434ace87d", "run_group_id": "E4_core:A4|repairable|qwen3b:h1", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:24:53.766828Z", "task_family": "synthetic", "document_id": "syn_03024_2000", "document_hash": "ea10cd11ad08f01c50b0712bc481668857c9d2ddc8e6423968f2a31eeb97ac55", "question_id": "q02_exact_copy", "split": "test", "cluster_id": "syn_03024_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 1, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 2068, "query_tokens": 600, "message_tokens": 512, "message_utf8_bytes": 1508, "cache_tensor_bytes": null, "serialized_payload_bytes": 1508, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 6945.187625009567, "hops": {"1": 6945.187625009567}, "receiver_prefill_ms": 227.04312507994473, "ttft_ms": 11.928042047657073, "decode_ms": 83.2612500526011}, "total_wall_ms": 7267.958375043236, "ttft_ms": 11.928042047657073, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 7}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "30.0224", "evidence_artifact": {"ledger": {"facts_total": 66, "facts_retained": 21, "retention_rate": 0.3181818181818182, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.35294117647058826, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1508}, "ledger_by_hop": {"1": {"facts_total": 66, "facts_retained": 21, "retention_rate": 0.3181818181818182, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.35294117647058826, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1508}}}, "reference_key_id": "keys/synthetic/syn_03024_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "34fcc1e23e5a866b5b7d7e222f0380ba771c4005d3732e9747f026c211840d43", "output_hash": "e28a78bc4deb3cec8beab4bf5b87297d7cc55a3e6f839b43fe6408735d464923"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "b819f4cf4d70789765f56d45", "attempt_id": "72973140ba2c", "run_group_id": "E4_core:A4|repairable|qwen3b:h1", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:24:54.104944Z", "task_family": "synthetic", "document_id": "syn_03024_2000", "document_hash": "ea10cd11ad08f01c50b0712bc481668857c9d2ddc8e6423968f2a31eeb97ac55", "question_id": "q03_exact_copy", "split": "test", "cluster_id": "syn_03024_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 1, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 2068, "query_tokens": 595, "message_tokens": 512, "message_utf8_bytes": 1508, "cache_tensor_bytes": null, "serialized_payload_bytes": 1508, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 6945.187625009567, "hops": {"1": 6945.187625009567}, "receiver_prefill_ms": 226.95950011257082, "ttft_ms": 12.000125017948449, "decode_ms": 11.916665942408144}, "total_wall_ms": 7196.552124922164, "ttft_ms": 12.000125017948449, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 66, "facts_retained": 21, "retention_rate": 0.3181818181818182, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.35294117647058826, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1508}, "ledger_by_hop": {"1": {"facts_total": 66, "facts_retained": 21, "retention_rate": 0.3181818181818182, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.35294117647058826, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1508}}}, "reference_key_id": "keys/synthetic/syn_03024_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "34fcc1e23e5a866b5b7d7e222f0380ba771c4005d3732e9747f026c211840d43", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "9aa49802748ba50ac7ef879f", "attempt_id": "ea79a43d161d", "run_group_id": "E4_core:A4|repairable|qwen3b:h1", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:24:54.371379Z", "task_family": "synthetic", "document_id": "syn_03024_2000", "document_hash": "ea10cd11ad08f01c50b0712bc481668857c9d2ddc8e6423968f2a31eeb97ac55", "question_id": "q04_multi_hop", "split": "test", "cluster_id": "syn_03024_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 1, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 2068, "query_tokens": 601, "message_tokens": 512, "message_utf8_bytes": 1508, "cache_tensor_bytes": null, "serialized_payload_bytes": 1508, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 6945.187625009567, "hops": {"1": 6945.187625009567}, "receiver_prefill_ms": 227.11995895951986, "ttft_ms": 11.928292107768357, "decode_ms": 107.15870803687721}, "total_wall_ms": 7291.916416957974, "ttft_ms": 11.928292107768357, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 9}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "6,009,480", "evidence_artifact": {"ledger": {"facts_total": 66, "facts_retained": 21, "retention_rate": 0.3181818181818182, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.35294117647058826, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1508}, "ledger_by_hop": {"1": {"facts_total": 66, "facts_retained": 21, "retention_rate": 0.3181818181818182, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.35294117647058826, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1508}}}, "reference_key_id": "keys/synthetic/syn_03024_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "34fcc1e23e5a866b5b7d7e222f0380ba771c4005d3732e9747f026c211840d43", "output_hash": "30b20d9aae4d7b0587edeefdc2a9a159f73faf6e2570f1d408a415fe92cbf45e"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "b4e0a18d8c2966c1ad66e40e", "attempt_id": "b38a9e749d7a", "run_group_id": "E4_core:A4|repairable|qwen3b:h1", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:24:54.733269Z", "task_family": "synthetic", "document_id": "syn_03024_2000", "document_hash": "ea10cd11ad08f01c50b0712bc481668857c9d2ddc8e6423968f2a31eeb97ac55", "question_id": "q05_multi_hop", "split": "test", "cluster_id": "syn_03024_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 1, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 2068, "query_tokens": 605, "message_tokens": 512, "message_utf8_bytes": 1508, "cache_tensor_bytes": null, "serialized_payload_bytes": 1508, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 6945.187625009567, "hops": {"1": 6945.187625009567}, "receiver_prefill_ms": 227.19416697509587, "ttft_ms": 11.907875072211027, "decode_ms": 11.93504105322063}, "total_wall_ms": 7196.724709006958, "ttft_ms": 11.907875072211027, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 66, "facts_retained": 21, "retention_rate": 0.3181818181818182, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.35294117647058826, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1508}, "ledger_by_hop": {"1": {"facts_total": 66, "facts_retained": 21, "retention_rate": 0.3181818181818182, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.35294117647058826, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1508}}}, "reference_key_id": "keys/synthetic/syn_03024_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "34fcc1e23e5a866b5b7d7e222f0380ba771c4005d3732e9747f026c211840d43", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "a00451a766a29b7098803d89", "attempt_id": "3727adecc409", "run_group_id": "E4_core:A4|repairable|qwen3b:h1", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:24:55.000408Z", "task_family": "synthetic", "document_id": "syn_03024_2000", "document_hash": "ea10cd11ad08f01c50b0712bc481668857c9d2ddc8e6423968f2a31eeb97ac55", "question_id": "q06_negation", "split": "test", "cluster_id": "syn_03024_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 1, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 2068, "query_tokens": 594, "message_tokens": 512, "message_utf8_bytes": 1508, "cache_tensor_bytes": null, "serialized_payload_bytes": 1508, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 6945.187625009567, "hops": {"1": 6945.187625009567}, "receiver_prefill_ms": 226.95954202208668, "ttft_ms": 12.039124965667725, "decode_ms": 11.90358295571059}, "total_wall_ms": 7196.595625020564, "ttft_ms": 12.039124965667725, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "No", "evidence_artifact": {"ledger": {"facts_total": 66, "facts_retained": 21, "retention_rate": 0.3181818181818182, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.35294117647058826, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1508}, "ledger_by_hop": {"1": {"facts_total": 66, "facts_retained": 21, "retention_rate": 0.3181818181818182, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.35294117647058826, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1508}}}, "reference_key_id": "keys/synthetic/syn_03024_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "34fcc1e23e5a866b5b7d7e222f0380ba771c4005d3732e9747f026c211840d43", "output_hash": "1ea442a134b2a184bd5d40104401f2a37fbc09ccf3f4bc9da161c6099be3691d"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "f2baa60c53536a3dd4482fbe", "attempt_id": "5121887d28f2", "run_group_id": "E4_core:A4|repairable|qwen3b:h1", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:24:55.267121Z", "task_family": "synthetic", "document_id": "syn_03024_2000", "document_hash": "ea10cd11ad08f01c50b0712bc481668857c9d2ddc8e6423968f2a31eeb97ac55", "question_id": "q07_negation", "split": "test", "cluster_id": "syn_03024_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 1, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 2068, "query_tokens": 596, "message_tokens": 512, "message_utf8_bytes": 1508, "cache_tensor_bytes": null, "serialized_payload_bytes": 1508, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 6945.187625009567, "hops": {"1": 6945.187625009567}, "receiver_prefill_ms": 227.23612503614277, "ttft_ms": 11.964207980781794, "decode_ms": 11.92641700617969}, "total_wall_ms": 7196.817999938503, "ttft_ms": 11.964207980781794, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "no", "evidence_artifact": {"ledger": {"facts_total": 66, "facts_retained": 21, "retention_rate": 0.3181818181818182, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.35294117647058826, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1508}, "ledger_by_hop": {"1": {"facts_total": 66, "facts_retained": 21, "retention_rate": 0.3181818181818182, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.35294117647058826, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1508}}}, "reference_key_id": "keys/synthetic/syn_03024_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "34fcc1e23e5a866b5b7d7e222f0380ba771c4005d3732e9747f026c211840d43", "output_hash": "9390298f3fb0c5b160498935d79cb139aef28e1c47358b4bbba61862b9c26e59"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "3e2159de9daf206c7825ef1d", "attempt_id": "6c9a6172f2ee", "run_group_id": "E4_core:A4|repairable|qwen3b:h1", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:24:55.533946Z", "task_family": "synthetic", "document_id": "syn_03024_2000", "document_hash": "ea10cd11ad08f01c50b0712bc481668857c9d2ddc8e6423968f2a31eeb97ac55", "question_id": "q08_revision", "split": "test", "cluster_id": "syn_03024_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 1, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 2068, "query_tokens": 612, "message_tokens": 512, "message_utf8_bytes": 1508, "cache_tensor_bytes": null, "serialized_payload_bytes": 1508, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 6945.187625009567, "hops": {"1": 6945.187625009567}, "receiver_prefill_ms": 229.46875006891787, "ttft_ms": 11.937875067815185, "decode_ms": 11.968082981184125}, "total_wall_ms": 7199.072916992009, "ttft_ms": 11.937875067815185, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 66, "facts_retained": 21, "retention_rate": 0.3181818181818182, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.35294117647058826, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1508}, "ledger_by_hop": {"1": {"facts_total": 66, "facts_retained": 21, "retention_rate": 0.3181818181818182, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.35294117647058826, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1508}}}, "reference_key_id": "keys/synthetic/syn_03024_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "34fcc1e23e5a866b5b7d7e222f0380ba771c4005d3732e9747f026c211840d43", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "1c2d6f113f606cf9fbcb401b", "attempt_id": "3b8cc8974e49", "run_group_id": "E4_core:A4|repairable|qwen3b:h1", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:24:55.803170Z", "task_family": "synthetic", "document_id": "syn_03024_2000", "document_hash": "ea10cd11ad08f01c50b0712bc481668857c9d2ddc8e6423968f2a31eeb97ac55", "question_id": "q09_revision", "split": "test", "cluster_id": "syn_03024_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 1, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 2068, "query_tokens": 603, "message_tokens": 512, "message_utf8_bytes": 1508, "cache_tensor_bytes": null, "serialized_payload_bytes": 1508, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 6945.187625009567, "hops": {"1": 6945.187625009567}, "receiver_prefill_ms": 227.52887499518692, "ttft_ms": 11.987083009444177, "decode_ms": 11.924542021006346}, "total_wall_ms": 7197.124334052205, "ttft_ms": 11.987083009444177, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 66, "facts_retained": 21, "retention_rate": 0.3181818181818182, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.35294117647058826, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1508}, "ledger_by_hop": {"1": {"facts_total": 66, "facts_retained": 21, "retention_rate": 0.3181818181818182, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.35294117647058826, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1508}}}, "reference_key_id": "keys/synthetic/syn_03024_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "34fcc1e23e5a866b5b7d7e222f0380ba771c4005d3732e9747f026c211840d43", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "84e974276dc7ae1c7f8ac55b", "attempt_id": "0210cc9c3b1b", "run_group_id": "E4_core:A4|repairable|qwen3b:h1", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:24:56.070518Z", "task_family": "synthetic", "document_id": "syn_03024_2000", "document_hash": "ea10cd11ad08f01c50b0712bc481668857c9d2ddc8e6423968f2a31eeb97ac55", "question_id": "q10_temporal", "split": "test", "cluster_id": "syn_03024_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 1, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 2068, "query_tokens": 613, "message_tokens": 512, "message_utf8_bytes": 1508, "cache_tensor_bytes": null, "serialized_payload_bytes": 1508, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 6945.187625009567, "hops": {"1": 6945.187625009567}, "receiver_prefill_ms": 229.18304102495313, "ttft_ms": 12.186208041384816, "decode_ms": 12.122917105443776}, "total_wall_ms": 7199.391000089236, "ttft_ms": 12.186208041384816, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 66, "facts_retained": 21, "retention_rate": 0.3181818181818182, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.35294117647058826, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1508}, "ledger_by_hop": {"1": {"facts_total": 66, "facts_retained": 21, "retention_rate": 0.3181818181818182, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.35294117647058826, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1508}}}, "reference_key_id": "keys/synthetic/syn_03024_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "34fcc1e23e5a866b5b7d7e222f0380ba771c4005d3732e9747f026c211840d43", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "661a3cebc51120ec584eec23", "attempt_id": "9874aa566701", "run_group_id": "E4_core:A4|repairable|qwen3b:h1", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:24:56.340077Z", "task_family": "synthetic", "document_id": "syn_03024_2000", "document_hash": "ea10cd11ad08f01c50b0712bc481668857c9d2ddc8e6423968f2a31eeb97ac55", "question_id": "q11_aggregation", "split": "test", "cluster_id": "syn_03024_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 1, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 2068, "query_tokens": 606, "message_tokens": 512, "message_utf8_bytes": 1508, "cache_tensor_bytes": null, "serialized_payload_bytes": 1508, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 6945.187625009567, "hops": {"1": 6945.187625009567}, "receiver_prefill_ms": 226.54004103969783, "ttft_ms": 12.3287090100348, "decode_ms": 84.38154100440443}, "total_wall_ms": 7269.193916930817, "ttft_ms": 12.3287090100348, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 7}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "2707654", "evidence_artifact": {"ledger": {"facts_total": 66, "facts_retained": 21, "retention_rate": 0.3181818181818182, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.35294117647058826, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1508}, "ledger_by_hop": {"1": {"facts_total": 66, "facts_retained": 21, "retention_rate": 0.3181818181818182, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.35294117647058826, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1508}}}, "reference_key_id": "keys/synthetic/syn_03024_2000.json", "task_scores": {"correct": 0.0, "abstained": 0.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "34fcc1e23e5a866b5b7d7e222f0380ba771c4005d3732e9747f026c211840d43", "output_hash": "7799e805760fa6568574360f61d4c7cc9e162842191ff434dfed71076922855e"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "40b645465198bd7914f472e1", "attempt_id": "800fc667bc14", "run_group_id": "E4_core:A4|repairable|qwen3b:h1", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:24:56.680556Z", "task_family": "synthetic", "document_id": "syn_03024_2000", "document_hash": "ea10cd11ad08f01c50b0712bc481668857c9d2ddc8e6423968f2a31eeb97ac55", "question_id": "q12_unit_conversion", "split": "test", "cluster_id": "syn_03024_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 1, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 2068, "query_tokens": 603, "message_tokens": 512, "message_utf8_bytes": 1508, "cache_tensor_bytes": null, "serialized_payload_bytes": 1508, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 6945.187625009567, "hops": {"1": 6945.187625009567}, "receiver_prefill_ms": 227.1463330835104, "ttft_ms": 12.037250096909702, "decode_ms": 11.95787510368973}, "total_wall_ms": 7196.986290975474, "ttft_ms": 12.037250096909702, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 66, "facts_retained": 21, "retention_rate": 0.3181818181818182, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.35294117647058826, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1508}, "ledger_by_hop": {"1": {"facts_total": 66, "facts_retained": 21, "retention_rate": 0.3181818181818182, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.35294117647058826, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1508}}}, "reference_key_id": "keys/synthetic/syn_03024_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "34fcc1e23e5a866b5b7d7e222f0380ba771c4005d3732e9747f026c211840d43", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "4182d2221720c6ea031a231d", "attempt_id": "e07e62f59299", "run_group_id": "E4_core:A4|repairable|qwen3b:h1", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:24:56.948105Z", "task_family": "synthetic", "document_id": "syn_03024_2000", "document_hash": "ea10cd11ad08f01c50b0712bc481668857c9d2ddc8e6423968f2a31eeb97ac55", "question_id": "q13_unanswerable", "split": "test", "cluster_id": "syn_03024_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 1, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 2068, "query_tokens": 607, "message_tokens": 512, "message_utf8_bytes": 1508, "cache_tensor_bytes": null, "serialized_payload_bytes": 1508, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 6945.187625009567, "hops": {"1": 6945.187625009567}, "receiver_prefill_ms": 227.4645830038935, "ttft_ms": 12.003249954432249, "decode_ms": 11.94899994879961}, "total_wall_ms": 7197.1197499660775, "ttft_ms": 12.003249954432249, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 66, "facts_retained": 21, "retention_rate": 0.3181818181818182, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.35294117647058826, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1508}, "ledger_by_hop": {"1": {"facts_total": 66, "facts_retained": 21, "retention_rate": 0.3181818181818182, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.35294117647058826, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1508}}}, "reference_key_id": "keys/synthetic/syn_03024_2000.json", "task_scores": {"correct": 1.0, "abstained": 1.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "34fcc1e23e5a866b5b7d7e222f0380ba771c4005d3732e9747f026c211840d43", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "c7f5eccb5b99d80bd36304cc", "attempt_id": "26ab68fe59b4", "run_group_id": "E4_core:A4|repairable|qwen3b:h1", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:24:57.215664Z", "task_family": "synthetic", "document_id": "syn_03024_2000", "document_hash": "ea10cd11ad08f01c50b0712bc481668857c9d2ddc8e6423968f2a31eeb97ac55", "question_id": "q14_unanswerable", "split": "test", "cluster_id": "syn_03024_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 1, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 2068, "query_tokens": 608, "message_tokens": 512, "message_utf8_bytes": 1508, "cache_tensor_bytes": null, "serialized_payload_bytes": 1508, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 6945.187625009567, "hops": {"1": 6945.187625009567}, "receiver_prefill_ms": 227.46837500017136, "ttft_ms": 11.944042053073645, "decode_ms": 11.833416996523738}, "total_wall_ms": 7196.95229199715, "ttft_ms": 11.944042053073645, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 66, "facts_retained": 21, "retention_rate": 0.3181818181818182, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.35294117647058826, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1508}, "ledger_by_hop": {"1": {"facts_total": 66, "facts_retained": 21, "retention_rate": 0.3181818181818182, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.35294117647058826, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1508}}}, "reference_key_id": "keys/synthetic/syn_03024_2000.json", "task_scores": {"correct": 1.0, "abstained": 1.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "34fcc1e23e5a866b5b7d7e222f0380ba771c4005d3732e9747f026c211840d43", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "ddd0a2ed0dcb48c768ecfe32", "attempt_id": "4d72fc89187d", "run_group_id": "E4_core:A4|repairable|qwen3b:h2", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:25:04.013453Z", "task_family": "synthetic", "document_id": "syn_03024_2000", "document_hash": "ea10cd11ad08f01c50b0712bc481668857c9d2ddc8e6423968f2a31eeb97ac55", "question_id": "q00_single_fact", "split": "test", "cluster_id": "syn_03024_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 2, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 2068, "query_tokens": 595, "message_tokens": 512, "message_utf8_bytes": 1498, "cache_tensor_bytes": null, "serialized_payload_bytes": 1498, "side_information_bytes": 0, "fetched_bytes": 1169, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 13474.19674997218, "hops": {"1": 6945.187625009567, "2": 6529.009124962613}, "receiver_prefill_ms": 226.34804202243686, "ttft_ms": 12.367833056487143, "decode_ms": 12.113792006857693}, "total_wall_ms": 13725.749083911069, "ttft_ms": 12.367833056487143, "input_usage": {"lookups": 2}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 66, "facts_retained": 21, "retention_rate": 0.3181818181818182, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.35294117647058826, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 1, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1498}, "ledger_by_hop": {"1": {"facts_total": 66, "facts_retained": 21, "retention_rate": 0.3181818181818182, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.35294117647058826, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1508}, "2": {"facts_total": 66, "facts_retained": 21, "retention_rate": 0.3181818181818182, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.35294117647058826, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 1, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1498}}}, "reference_key_id": "keys/synthetic/syn_03024_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "dbc18f4d960624946c884bb20c9368119b30b62a2cad122ffe8ba81f0d70bc39", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "9b17bc6001af4428d4595a97", "attempt_id": "1aafde050649", "run_group_id": "E4_core:A4|repairable|qwen3b:h2", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:25:04.281307Z", "task_family": "synthetic", "document_id": "syn_03024_2000", "document_hash": "ea10cd11ad08f01c50b0712bc481668857c9d2ddc8e6423968f2a31eeb97ac55", "question_id": "q01_single_fact", "split": "test", "cluster_id": "syn_03024_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 2, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 2068, "query_tokens": 600, "message_tokens": 512, "message_utf8_bytes": 1498, "cache_tensor_bytes": null, "serialized_payload_bytes": 1498, "side_information_bytes": 0, "fetched_bytes": 1169, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 13474.19674997218, "hops": {"1": 6945.187625009567, "2": 6529.009124962613}, "receiver_prefill_ms": 226.9915830111131, "ttft_ms": 11.986832949332893, "decode_ms": 11.93108398001641}, "total_wall_ms": 13725.823416025378, "ttft_ms": 11.986832949332893, "input_usage": {"lookups": 2}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 66, "facts_retained": 21, "retention_rate": 0.3181818181818182, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.35294117647058826, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 1, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1498}, "ledger_by_hop": {"1": {"facts_total": 66, "facts_retained": 21, "retention_rate": 0.3181818181818182, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.35294117647058826, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1508}, "2": {"facts_total": 66, "facts_retained": 21, "retention_rate": 0.3181818181818182, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.35294117647058826, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 1, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1498}}}, "reference_key_id": "keys/synthetic/syn_03024_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "dbc18f4d960624946c884bb20c9368119b30b62a2cad122ffe8ba81f0d70bc39", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "ca1ccfe4498967618aac5e4c", "attempt_id": "65c3b66a4269", "run_group_id": "E4_core:A4|repairable|qwen3b:h2", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:25:04.550550Z", "task_family": "synthetic", "document_id": "syn_03024_2000", "document_hash": "ea10cd11ad08f01c50b0712bc481668857c9d2ddc8e6423968f2a31eeb97ac55", "question_id": "q02_exact_copy", "split": "test", "cluster_id": "syn_03024_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 2, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 2068, "query_tokens": 600, "message_tokens": 512, "message_utf8_bytes": 1498, "cache_tensor_bytes": null, "serialized_payload_bytes": 1498, "side_information_bytes": 0, "fetched_bytes": 1169, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 13474.19674997218, "hops": {"1": 6945.187625009567, "2": 6529.009124962613}, "receiver_prefill_ms": 227.24337503314018, "ttft_ms": 12.113417033106089, "decode_ms": 83.23362492956221}, "total_wall_ms": 13797.474874998443, "ttft_ms": 12.113417033106089, "input_usage": {"lookups": 2}, "output_usage": {"output_tokens": 7}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "30.0224", "evidence_artifact": {"ledger": {"facts_total": 66, "facts_retained": 21, "retention_rate": 0.3181818181818182, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.35294117647058826, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 1, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1498}, "ledger_by_hop": {"1": {"facts_total": 66, "facts_retained": 21, "retention_rate": 0.3181818181818182, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.35294117647058826, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1508}, "2": {"facts_total": 66, "facts_retained": 21, "retention_rate": 0.3181818181818182, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.35294117647058826, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 1, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1498}}}, "reference_key_id": "keys/synthetic/syn_03024_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "dbc18f4d960624946c884bb20c9368119b30b62a2cad122ffe8ba81f0d70bc39", "output_hash": "e28a78bc4deb3cec8beab4bf5b87297d7cc55a3e6f839b43fe6408735d464923"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "1495cd4007f1a99adeea5c5c", "attempt_id": "bff779179295", "run_group_id": "E4_core:A4|repairable|qwen3b:h2", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:25:04.890664Z", "task_family": "synthetic", "document_id": "syn_03024_2000", "document_hash": "ea10cd11ad08f01c50b0712bc481668857c9d2ddc8e6423968f2a31eeb97ac55", "question_id": "q03_exact_copy", "split": "test", "cluster_id": "syn_03024_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 2, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 2068, "query_tokens": 595, "message_tokens": 512, "message_utf8_bytes": 1498, "cache_tensor_bytes": null, "serialized_payload_bytes": 1498, "side_information_bytes": 0, "fetched_bytes": 1169, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 13474.19674997218, "hops": {"1": 6945.187625009567, "2": 6529.009124962613}, "receiver_prefill_ms": 226.6717499587685, "ttft_ms": 12.021500035189092, "decode_ms": 11.90883293747902}, "total_wall_ms": 13725.447041913867, "ttft_ms": 12.021500035189092, "input_usage": {"lookups": 2}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 66, "facts_retained": 21, "retention_rate": 0.3181818181818182, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.35294117647058826, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 1, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1498}, "ledger_by_hop": {"1": {"facts_total": 66, "facts_retained": 21, "retention_rate": 0.3181818181818182, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.35294117647058826, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1508}, "2": {"facts_total": 66, "facts_retained": 21, "retention_rate": 0.3181818181818182, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.35294117647058826, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 1, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1498}}}, "reference_key_id": "keys/synthetic/syn_03024_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "dbc18f4d960624946c884bb20c9368119b30b62a2cad122ffe8ba81f0d70bc39", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "82923a9b27c5c4f3b4d0441f", "attempt_id": "ddcb031505aa", "run_group_id": "E4_core:A4|repairable|qwen3b:h2", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:25:05.158105Z", "task_family": "synthetic", "document_id": "syn_03024_2000", "document_hash": "ea10cd11ad08f01c50b0712bc481668857c9d2ddc8e6423968f2a31eeb97ac55", "question_id": "q04_multi_hop", "split": "test", "cluster_id": "syn_03024_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 2, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 2068, "query_tokens": 601, "message_tokens": 512, "message_utf8_bytes": 1498, "cache_tensor_bytes": null, "serialized_payload_bytes": 1498, "side_information_bytes": 0, "fetched_bytes": 1169, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 13474.19674997218, "hops": {"1": 6945.187625009567, "2": 6529.009124962613}, "receiver_prefill_ms": 227.34483401291072, "ttft_ms": 11.980792041867971, "decode_ms": 106.91137495450675}, "total_wall_ms": 13821.044415934011, "ttft_ms": 11.980792041867971, "input_usage": {"lookups": 2}, "output_usage": {"output_tokens": 9}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "6,009,480", "evidence_artifact": {"ledger": {"facts_total": 66, "facts_retained": 21, "retention_rate": 0.3181818181818182, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.35294117647058826, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 1, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1498}, "ledger_by_hop": {"1": {"facts_total": 66, "facts_retained": 21, "retention_rate": 0.3181818181818182, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.35294117647058826, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1508}, "2": {"facts_total": 66, "facts_retained": 21, "retention_rate": 0.3181818181818182, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.35294117647058826, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 1, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1498}}}, "reference_key_id": "keys/synthetic/syn_03024_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "dbc18f4d960624946c884bb20c9368119b30b62a2cad122ffe8ba81f0d70bc39", "output_hash": "30b20d9aae4d7b0587edeefdc2a9a159f73faf6e2570f1d408a415fe92cbf45e"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "de38db94279fbc7af8908fe2", "attempt_id": "5add0cc8521f", "run_group_id": "E4_core:A4|repairable|qwen3b:h2", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:25:05.520929Z", "task_family": "synthetic", "document_id": "syn_03024_2000", "document_hash": "ea10cd11ad08f01c50b0712bc481668857c9d2ddc8e6423968f2a31eeb97ac55", "question_id": "q05_multi_hop", "split": "test", "cluster_id": "syn_03024_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 2, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 2068, "query_tokens": 605, "message_tokens": 512, "message_utf8_bytes": 1498, "cache_tensor_bytes": null, "serialized_payload_bytes": 1498, "side_information_bytes": 0, "fetched_bytes": 1169, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 13474.19674997218, "hops": {"1": 6945.187625009567, "2": 6529.009124962613}, "receiver_prefill_ms": 227.1708749467507, "ttft_ms": 11.961457901634276, "decode_ms": 11.95499999448657}, "total_wall_ms": 13725.830999901518, "ttft_ms": 11.961457901634276, "input_usage": {"lookups": 2}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 66, "facts_retained": 21, "retention_rate": 0.3181818181818182, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.35294117647058826, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 1, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1498}, "ledger_by_hop": {"1": {"facts_total": 66, "facts_retained": 21, "retention_rate": 0.3181818181818182, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.35294117647058826, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1508}, "2": {"facts_total": 66, "facts_retained": 21, "retention_rate": 0.3181818181818182, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.35294117647058826, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 1, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1498}}}, "reference_key_id": "keys/synthetic/syn_03024_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "dbc18f4d960624946c884bb20c9368119b30b62a2cad122ffe8ba81f0d70bc39", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "1649ac1afc6a9df21ecf9663", "attempt_id": "7aec67e491fc", "run_group_id": "E4_core:A4|repairable|qwen3b:h2", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:25:05.788330Z", "task_family": "synthetic", "document_id": "syn_03024_2000", "document_hash": "ea10cd11ad08f01c50b0712bc481668857c9d2ddc8e6423968f2a31eeb97ac55", "question_id": "q06_negation", "split": "test", "cluster_id": "syn_03024_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 2, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 2068, "query_tokens": 594, "message_tokens": 512, "message_utf8_bytes": 1498, "cache_tensor_bytes": null, "serialized_payload_bytes": 1498, "side_information_bytes": 0, "fetched_bytes": 1169, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 13474.19674997218, "hops": {"1": 6945.187625009567, "2": 6529.009124962613}, "receiver_prefill_ms": 226.8016659654677, "ttft_ms": 12.11179094389081, "decode_ms": 11.98212499730289}, "total_wall_ms": 13725.765916984528, "ttft_ms": 12.11179094389081, "input_usage": {"lookups": 2}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "No", "evidence_artifact": {"ledger": {"facts_total": 66, "facts_retained": 21, "retention_rate": 0.3181818181818182, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.35294117647058826, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 1, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1498}, "ledger_by_hop": {"1": {"facts_total": 66, "facts_retained": 21, "retention_rate": 0.3181818181818182, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.35294117647058826, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1508}, "2": {"facts_total": 66, "facts_retained": 21, "retention_rate": 0.3181818181818182, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.35294117647058826, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 1, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1498}}}, "reference_key_id": "keys/synthetic/syn_03024_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "dbc18f4d960624946c884bb20c9368119b30b62a2cad122ffe8ba81f0d70bc39", "output_hash": "1ea442a134b2a184bd5d40104401f2a37fbc09ccf3f4bc9da161c6099be3691d"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "699107816d669f7e0a3e23bf", "attempt_id": "cf01b5e3dfa8", "run_group_id": "E4_core:A4|repairable|qwen3b:h2", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:25:06.055667Z", "task_family": "synthetic", "document_id": "syn_03024_2000", "document_hash": "ea10cd11ad08f01c50b0712bc481668857c9d2ddc8e6423968f2a31eeb97ac55", "question_id": "q07_negation", "split": "test", "cluster_id": "syn_03024_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 2, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 2068, "query_tokens": 596, "message_tokens": 512, "message_utf8_bytes": 1498, "cache_tensor_bytes": null, "serialized_payload_bytes": 1498, "side_information_bytes": 0, "fetched_bytes": 1169, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 13474.19674997218, "hops": {"1": 6945.187625009567, "2": 6529.009124962613}, "receiver_prefill_ms": 227.28991694748402, "ttft_ms": 11.97016704827547, "decode_ms": 11.920665972866118}, "total_wall_ms": 13725.915541988797, "ttft_ms": 11.97016704827547, "input_usage": {"lookups": 2}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "no", "evidence_artifact": {"ledger": {"facts_total": 66, "facts_retained": 21, "retention_rate": 0.3181818181818182, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.35294117647058826, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 1, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1498}, "ledger_by_hop": {"1": {"facts_total": 66, "facts_retained": 21, "retention_rate": 0.3181818181818182, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.35294117647058826, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1508}, "2": {"facts_total": 66, "facts_retained": 21, "retention_rate": 0.3181818181818182, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.35294117647058826, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 1, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1498}}}, "reference_key_id": "keys/synthetic/syn_03024_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "dbc18f4d960624946c884bb20c9368119b30b62a2cad122ffe8ba81f0d70bc39", "output_hash": "9390298f3fb0c5b160498935d79cb139aef28e1c47358b4bbba61862b9c26e59"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "b0a2af98e6361b232020b4c4", "attempt_id": "46a607b531bf", "run_group_id": "E4_core:A4|repairable|qwen3b:h2", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:25:06.323338Z", "task_family": "synthetic", "document_id": "syn_03024_2000", "document_hash": "ea10cd11ad08f01c50b0712bc481668857c9d2ddc8e6423968f2a31eeb97ac55", "question_id": "q08_revision", "split": "test", "cluster_id": "syn_03024_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 2, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 2068, "query_tokens": 612, "message_tokens": 512, "message_utf8_bytes": 1498, "cache_tensor_bytes": null, "serialized_payload_bytes": 1498, "side_information_bytes": 0, "fetched_bytes": 1169, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 13474.19674997218, "hops": {"1": 6945.187625009567, "2": 6529.009124962613}, "receiver_prefill_ms": 228.967540897429, "ttft_ms": 12.09699991159141, "decode_ms": 11.913041002117097}, "total_wall_ms": 13727.846957975999, "ttft_ms": 12.09699991159141, "input_usage": {"lookups": 2}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 66, "facts_retained": 21, "retention_rate": 0.3181818181818182, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.35294117647058826, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 1, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1498}, "ledger_by_hop": {"1": {"facts_total": 66, "facts_retained": 21, "retention_rate": 0.3181818181818182, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.35294117647058826, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1508}, "2": {"facts_total": 66, "facts_retained": 21, "retention_rate": 0.3181818181818182, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.35294117647058826, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 1, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1498}}}, "reference_key_id": "keys/synthetic/syn_03024_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "dbc18f4d960624946c884bb20c9368119b30b62a2cad122ffe8ba81f0d70bc39", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "697645ece476866d837fffc0", "attempt_id": "99cace8fee14", "run_group_id": "E4_core:A4|repairable|qwen3b:h2", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:25:06.592848Z", "task_family": "synthetic", "document_id": "syn_03024_2000", "document_hash": "ea10cd11ad08f01c50b0712bc481668857c9d2ddc8e6423968f2a31eeb97ac55", "question_id": "q09_revision", "split": "test", "cluster_id": "syn_03024_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 2, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 2068, "query_tokens": 603, "message_tokens": 512, "message_utf8_bytes": 1498, "cache_tensor_bytes": null, "serialized_payload_bytes": 1498, "side_information_bytes": 0, "fetched_bytes": 1169, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 13474.19674997218, "hops": {"1": 6945.187625009567, "2": 6529.009124962613}, "receiver_prefill_ms": 227.57745801936835, "ttft_ms": 11.894499999471009, "decode_ms": 11.974791996181011}, "total_wall_ms": 13726.162582985125, "ttft_ms": 11.894499999471009, "input_usage": {"lookups": 2}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 66, "facts_retained": 21, "retention_rate": 0.3181818181818182, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.35294117647058826, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 1, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1498}, "ledger_by_hop": {"1": {"facts_total": 66, "facts_retained": 21, "retention_rate": 0.3181818181818182, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.35294117647058826, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1508}, "2": {"facts_total": 66, "facts_retained": 21, "retention_rate": 0.3181818181818182, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.35294117647058826, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 1, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1498}}}, "reference_key_id": "keys/synthetic/syn_03024_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "dbc18f4d960624946c884bb20c9368119b30b62a2cad122ffe8ba81f0d70bc39", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "f5ac5245051f19929cba1156", "attempt_id": "4b7586d36fe7", "run_group_id": "E4_core:A4|repairable|qwen3b:h2", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:25:06.860923Z", "task_family": "synthetic", "document_id": "syn_03024_2000", "document_hash": "ea10cd11ad08f01c50b0712bc481668857c9d2ddc8e6423968f2a31eeb97ac55", "question_id": "q10_temporal", "split": "test", "cluster_id": "syn_03024_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 2, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 2068, "query_tokens": 613, "message_tokens": 512, "message_utf8_bytes": 1498, "cache_tensor_bytes": null, "serialized_payload_bytes": 1498, "side_information_bytes": 0, "fetched_bytes": 1169, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 13474.19674997218, "hops": {"1": 6945.187625009567, "2": 6529.009124962613}, "receiver_prefill_ms": 229.3671660590917, "ttft_ms": 12.044333037920296, "decode_ms": 11.911208974197507}, "total_wall_ms": 13728.087250026874, "ttft_ms": 12.044333037920296, "input_usage": {"lookups": 2}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 66, "facts_retained": 21, "retention_rate": 0.3181818181818182, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.35294117647058826, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 1, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1498}, "ledger_by_hop": {"1": {"facts_total": 66, "facts_retained": 21, "retention_rate": 0.3181818181818182, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.35294117647058826, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1508}, "2": {"facts_total": 66, "facts_retained": 21, "retention_rate": 0.3181818181818182, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.35294117647058826, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 1, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1498}}}, "reference_key_id": "keys/synthetic/syn_03024_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "dbc18f4d960624946c884bb20c9368119b30b62a2cad122ffe8ba81f0d70bc39", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "77d547e0da4a676a516b25c6", "attempt_id": "51a62505c708", "run_group_id": "E4_core:A4|repairable|qwen3b:h2", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:25:07.130124Z", "task_family": "synthetic", "document_id": "syn_03024_2000", "document_hash": "ea10cd11ad08f01c50b0712bc481668857c9d2ddc8e6423968f2a31eeb97ac55", "question_id": "q11_aggregation", "split": "test", "cluster_id": "syn_03024_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 2, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 2068, "query_tokens": 606, "message_tokens": 512, "message_utf8_bytes": 1498, "cache_tensor_bytes": null, "serialized_payload_bytes": 1498, "side_information_bytes": 0, "fetched_bytes": 1169, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 13474.19674997218, "hops": {"1": 6945.187625009567, "2": 6529.009124962613}, "receiver_prefill_ms": 227.38770802970976, "ttft_ms": 11.881542042829096, "decode_ms": 118.59470896888524}, "total_wall_ms": 13832.722624996677, "ttft_ms": 11.881542042829096, "input_usage": {"lookups": 2}, "output_usage": {"output_tokens": 10}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "27,310,344", "evidence_artifact": {"ledger": {"facts_total": 66, "facts_retained": 21, "retention_rate": 0.3181818181818182, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.35294117647058826, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 1, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1498}, "ledger_by_hop": {"1": {"facts_total": 66, "facts_retained": 21, "retention_rate": 0.3181818181818182, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.35294117647058826, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1508}, "2": {"facts_total": 66, "facts_retained": 21, "retention_rate": 0.3181818181818182, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.35294117647058826, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 1, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1498}}}, "reference_key_id": "keys/synthetic/syn_03024_2000.json", "task_scores": {"correct": 0.0, "abstained": 0.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "dbc18f4d960624946c884bb20c9368119b30b62a2cad122ffe8ba81f0d70bc39", "output_hash": "57d43335b2f7697d49fa5c54ca28131a174ade7ed50cefb8f9e04bda80bbc15f"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "edfe6182ed4c363a784dd7c1", "attempt_id": "65819bfa9800", "run_group_id": "E4_core:A4|repairable|qwen3b:h2", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:25:07.504845Z", "task_family": "synthetic", "document_id": "syn_03024_2000", "document_hash": "ea10cd11ad08f01c50b0712bc481668857c9d2ddc8e6423968f2a31eeb97ac55", "question_id": "q12_unit_conversion", "split": "test", "cluster_id": "syn_03024_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 2, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 2068, "query_tokens": 603, "message_tokens": 512, "message_utf8_bytes": 1498, "cache_tensor_bytes": null, "serialized_payload_bytes": 1498, "side_information_bytes": 0, "fetched_bytes": 1169, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 13474.19674997218, "hops": {"1": 6945.187625009567, "2": 6529.009124962613}, "receiver_prefill_ms": 226.63245897274464, "ttft_ms": 12.079125037416816, "decode_ms": 11.959875002503395}, "total_wall_ms": 13725.523082888685, "ttft_ms": 12.079125037416816, "input_usage": {"lookups": 2}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 66, "facts_retained": 21, "retention_rate": 0.3181818181818182, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.35294117647058826, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 1, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1498}, "ledger_by_hop": {"1": {"facts_total": 66, "facts_retained": 21, "retention_rate": 0.3181818181818182, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.35294117647058826, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1508}, "2": {"facts_total": 66, "facts_retained": 21, "retention_rate": 0.3181818181818182, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.35294117647058826, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 1, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1498}}}, "reference_key_id": "keys/synthetic/syn_03024_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "dbc18f4d960624946c884bb20c9368119b30b62a2cad122ffe8ba81f0d70bc39", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "d36780001f0d8386b12bcfac", "attempt_id": "e8e4161722b4", "run_group_id": "E4_core:A4|repairable|qwen3b:h2", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:25:07.772870Z", "task_family": "synthetic", "document_id": "syn_03024_2000", "document_hash": "ea10cd11ad08f01c50b0712bc481668857c9d2ddc8e6423968f2a31eeb97ac55", "question_id": "q13_unanswerable", "split": "test", "cluster_id": "syn_03024_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 2, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 2068, "query_tokens": 607, "message_tokens": 512, "message_utf8_bytes": 1498, "cache_tensor_bytes": null, "serialized_payload_bytes": 1498, "side_information_bytes": 0, "fetched_bytes": 1169, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 13474.19674997218, "hops": {"1": 6945.187625009567, "2": 6529.009124962613}, "receiver_prefill_ms": 226.955582969822, "ttft_ms": 12.005541939288378, "decode_ms": 11.948833940550685}, "total_wall_ms": 13725.715999957174, "ttft_ms": 12.005541939288378, "input_usage": {"lookups": 2}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 66, "facts_retained": 21, "retention_rate": 0.3181818181818182, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.35294117647058826, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 1, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1498}, "ledger_by_hop": {"1": {"facts_total": 66, "facts_retained": 21, "retention_rate": 0.3181818181818182, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.35294117647058826, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1508}, "2": {"facts_total": 66, "facts_retained": 21, "retention_rate": 0.3181818181818182, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.35294117647058826, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 1, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1498}}}, "reference_key_id": "keys/synthetic/syn_03024_2000.json", "task_scores": {"correct": 1.0, "abstained": 1.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "dbc18f4d960624946c884bb20c9368119b30b62a2cad122ffe8ba81f0d70bc39", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "0a217c7fd0a62d0e3230eb17", "attempt_id": "c7eec7c6324c", "run_group_id": "E4_core:A4|repairable|qwen3b:h2", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:25:08.040485Z", "task_family": "synthetic", "document_id": "syn_03024_2000", "document_hash": "ea10cd11ad08f01c50b0712bc481668857c9d2ddc8e6423968f2a31eeb97ac55", "question_id": "q14_unanswerable", "split": "test", "cluster_id": "syn_03024_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 2, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 2068, "query_tokens": 608, "message_tokens": 512, "message_utf8_bytes": 1498, "cache_tensor_bytes": null, "serialized_payload_bytes": 1498, "side_information_bytes": 0, "fetched_bytes": 1169, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 13474.19674997218, "hops": {"1": 6945.187625009567, "2": 6529.009124962613}, "receiver_prefill_ms": 226.89295804593712, "ttft_ms": 11.957416078075767, "decode_ms": 12.158999918028712}, "total_wall_ms": 13725.748124998063, "ttft_ms": 11.957416078075767, "input_usage": {"lookups": 2}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 66, "facts_retained": 21, "retention_rate": 0.3181818181818182, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.35294117647058826, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 1, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1498}, "ledger_by_hop": {"1": {"facts_total": 66, "facts_retained": 21, "retention_rate": 0.3181818181818182, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.35294117647058826, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1508}, "2": {"facts_total": 66, "facts_retained": 21, "retention_rate": 0.3181818181818182, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.35294117647058826, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 1, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1498}}}, "reference_key_id": "keys/synthetic/syn_03024_2000.json", "task_scores": {"correct": 1.0, "abstained": 1.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "dbc18f4d960624946c884bb20c9368119b30b62a2cad122ffe8ba81f0d70bc39", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "379d5b5538ae30643f5249ce", "attempt_id": "4e381c78dd68", "run_group_id": "E4_core:A4|repairable|qwen3b:h4", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:25:20.167322Z", "task_family": "synthetic", "document_id": "syn_03024_2000", "document_hash": "ea10cd11ad08f01c50b0712bc481668857c9d2ddc8e6423968f2a31eeb97ac55", "question_id": "q00_single_fact", "split": "test", "cluster_id": "syn_03024_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 4, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 2068, "query_tokens": 545, "message_tokens": 463, "message_utf8_bytes": 1435, "cache_tensor_bytes": null, "serialized_payload_bytes": 1435, "side_information_bytes": 0, "fetched_bytes": 3507, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 25330.884958035313, "hops": {"1": 6945.187625009567, "2": 6529.009124962613, "3": 5943.714833003469, "4": 5912.973375059664}, "receiver_prefill_ms": 208.06799991987646, "ttft_ms": 11.886875028721988, "decode_ms": 11.92337495740503}, "total_wall_ms": 25563.275250140578, "ttft_ms": 11.886875028721988, "input_usage": {"lookups": 6}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 66, "facts_retained": 21, "retention_rate": 0.3181818181818182, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.35294117647058826, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 1, "unsupported_identifiers": 0, "distorted_sentences": 1, "message_bytes": 1435}, "ledger_by_hop": {"1": {"facts_total": 66, "facts_retained": 21, "retention_rate": 0.3181818181818182, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.35294117647058826, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1508}, "2": {"facts_total": 66, "facts_retained": 21, "retention_rate": 0.3181818181818182, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.35294117647058826, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 1, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1498}, "3": {"facts_total": 66, "facts_retained": 21, "retention_rate": 0.3181818181818182, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.35294117647058826, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 1, "unsupported_identifiers": 0, "distorted_sentences": 1, "message_bytes": 1435}, "4": {"facts_total": 66, "facts_retained": 21, "retention_rate": 0.3181818181818182, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.35294117647058826, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 1, "unsupported_identifiers": 0, "distorted_sentences": 1, "message_bytes": 1435}}}, "reference_key_id": "keys/synthetic/syn_03024_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "f946ccf31a9f007db2f33763c10d0ace160259ffb2b94381d1019bf2461c8a0a", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "2c7b1e8d752a7e243d1cff05", "attempt_id": "824e30cca4ae", "run_group_id": "E4_core:A4|repairable|qwen3b:h4", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:25:20.416122Z", "task_family": "synthetic", "document_id": "syn_03024_2000", "document_hash": "ea10cd11ad08f01c50b0712bc481668857c9d2ddc8e6423968f2a31eeb97ac55", "question_id": "q01_single_fact", "split": "test", "cluster_id": "syn_03024_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 4, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 2068, "query_tokens": 550, "message_tokens": 463, "message_utf8_bytes": 1435, "cache_tensor_bytes": null, "serialized_payload_bytes": 1435, "side_information_bytes": 0, "fetched_bytes": 3507, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 25330.884958035313, "hops": {"1": 6945.187625009567, "2": 6529.009124962613, "3": 5943.714833003469, "4": 5912.973375059664}, "receiver_prefill_ms": 210.17608302645385, "ttft_ms": 11.901749996468425, "decode_ms": 12.00404204428196}, "total_wall_ms": 25565.47687505372, "ttft_ms": 11.901749996468425, "input_usage": {"lookups": 6}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 66, "facts_retained": 21, "retention_rate": 0.3181818181818182, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.35294117647058826, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 1, "unsupported_identifiers": 0, "distorted_sentences": 1, "message_bytes": 1435}, "ledger_by_hop": {"1": {"facts_total": 66, "facts_retained": 21, "retention_rate": 0.3181818181818182, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.35294117647058826, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1508}, "2": {"facts_total": 66, "facts_retained": 21, "retention_rate": 0.3181818181818182, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.35294117647058826, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 1, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1498}, "3": {"facts_total": 66, "facts_retained": 21, "retention_rate": 0.3181818181818182, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.35294117647058826, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 1, "unsupported_identifiers": 0, "distorted_sentences": 1, "message_bytes": 1435}, "4": {"facts_total": 66, "facts_retained": 21, "retention_rate": 0.3181818181818182, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.35294117647058826, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 1, "unsupported_identifiers": 0, "distorted_sentences": 1, "message_bytes": 1435}}}, "reference_key_id": "keys/synthetic/syn_03024_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "f946ccf31a9f007db2f33763c10d0ace160259ffb2b94381d1019bf2461c8a0a", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "6812f7a2ba1cd68f1088eebc", "attempt_id": "74b109fd16e8", "run_group_id": "E4_core:A4|repairable|qwen3b:h4", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:25:20.666282Z", "task_family": "synthetic", "document_id": "syn_03024_2000", "document_hash": "ea10cd11ad08f01c50b0712bc481668857c9d2ddc8e6423968f2a31eeb97ac55", "question_id": "q02_exact_copy", "split": "test", "cluster_id": "syn_03024_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 4, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 2068, "query_tokens": 550, "message_tokens": 463, "message_utf8_bytes": 1435, "cache_tensor_bytes": null, "serialized_payload_bytes": 1435, "side_information_bytes": 0, "fetched_bytes": 3507, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 25330.884958035313, "hops": {"1": 6945.187625009567, "2": 6529.009124962613, "3": 5943.714833003469, "4": 5912.973375059664}, "receiver_prefill_ms": 210.53391590248793, "ttft_ms": 11.850791983306408, "decode_ms": 82.68633403349668}, "total_wall_ms": 25636.444958043285, "ttft_ms": 11.850791983306408, "input_usage": {"lookups": 6}, "output_usage": {"output_tokens": 7}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "30.0224", "evidence_artifact": {"ledger": {"facts_total": 66, "facts_retained": 21, "retention_rate": 0.3181818181818182, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.35294117647058826, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 1, "unsupported_identifiers": 0, "distorted_sentences": 1, "message_bytes": 1435}, "ledger_by_hop": {"1": {"facts_total": 66, "facts_retained": 21, "retention_rate": 0.3181818181818182, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.35294117647058826, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1508}, "2": {"facts_total": 66, "facts_retained": 21, "retention_rate": 0.3181818181818182, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.35294117647058826, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 1, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1498}, "3": {"facts_total": 66, "facts_retained": 21, "retention_rate": 0.3181818181818182, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.35294117647058826, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 1, "unsupported_identifiers": 0, "distorted_sentences": 1, "message_bytes": 1435}, "4": {"facts_total": 66, "facts_retained": 21, "retention_rate": 0.3181818181818182, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.35294117647058826, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 1, "unsupported_identifiers": 0, "distorted_sentences": 1, "message_bytes": 1435}}}, "reference_key_id": "keys/synthetic/syn_03024_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "f946ccf31a9f007db2f33763c10d0ace160259ffb2b94381d1019bf2461c8a0a", "output_hash": "e28a78bc4deb3cec8beab4bf5b87297d7cc55a3e6f839b43fe6408735d464923"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "dd5d7c7d7fad46dc6f565997", "attempt_id": "7839c04733a4", "run_group_id": "E4_core:A4|repairable|qwen3b:h4", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:25:20.986940Z", "task_family": "synthetic", "document_id": "syn_03024_2000", "document_hash": "ea10cd11ad08f01c50b0712bc481668857c9d2ddc8e6423968f2a31eeb97ac55", "question_id": "q03_exact_copy", "split": "test", "cluster_id": "syn_03024_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 4, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 2068, "query_tokens": 545, "message_tokens": 463, "message_utf8_bytes": 1435, "cache_tensor_bytes": null, "serialized_payload_bytes": 1435, "side_information_bytes": 0, "fetched_bytes": 3507, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 25330.884958035313, "hops": {"1": 6945.187625009567, "2": 6529.009124962613, "3": 5943.714833003469, "4": 5912.973375059664}, "receiver_prefill_ms": 208.44499999657273, "ttft_ms": 11.90079189836979, "decode_ms": 11.803834000602365}, "total_wall_ms": 25563.497915980406, "ttft_ms": 11.90079189836979, "input_usage": {"lookups": 6}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 66, "facts_retained": 21, "retention_rate": 0.3181818181818182, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.35294117647058826, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 1, "unsupported_identifiers": 0, "distorted_sentences": 1, "message_bytes": 1435}, "ledger_by_hop": {"1": {"facts_total": 66, "facts_retained": 21, "retention_rate": 0.3181818181818182, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.35294117647058826, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1508}, "2": {"facts_total": 66, "facts_retained": 21, "retention_rate": 0.3181818181818182, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.35294117647058826, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 1, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1498}, "3": {"facts_total": 66, "facts_retained": 21, "retention_rate": 0.3181818181818182, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.35294117647058826, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 1, "unsupported_identifiers": 0, "distorted_sentences": 1, "message_bytes": 1435}, "4": {"facts_total": 66, "facts_retained": 21, "retention_rate": 0.3181818181818182, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.35294117647058826, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 1, "unsupported_identifiers": 0, "distorted_sentences": 1, "message_bytes": 1435}}}, "reference_key_id": "keys/synthetic/syn_03024_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "f946ccf31a9f007db2f33763c10d0ace160259ffb2b94381d1019bf2461c8a0a", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "4904ba7a72aa51f181f35b71", "attempt_id": "f3940cc5f46a", "run_group_id": "E4_core:A4|repairable|qwen3b:h4", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:25:21.234778Z", "task_family": "synthetic", "document_id": "syn_03024_2000", "document_hash": "ea10cd11ad08f01c50b0712bc481668857c9d2ddc8e6423968f2a31eeb97ac55", "question_id": "q04_multi_hop", "split": "test", "cluster_id": "syn_03024_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 4, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 2068, "query_tokens": 551, "message_tokens": 463, "message_utf8_bytes": 1435, "cache_tensor_bytes": null, "serialized_payload_bytes": 1435, "side_information_bytes": 0, "fetched_bytes": 3507, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 25330.884958035313, "hops": {"1": 6945.187625009567, "2": 6529.009124962613, "3": 5943.714833003469, "4": 5912.973375059664}, "receiver_prefill_ms": 210.2336250245571, "ttft_ms": 11.869499925523996, "decode_ms": 83.0089170485735}, "total_wall_ms": 25636.500083026476, "ttft_ms": 11.869499925523996, "input_usage": {"lookups": 6}, "output_usage": {"output_tokens": 7}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "6009480", "evidence_artifact": {"ledger": {"facts_total": 66, "facts_retained": 21, "retention_rate": 0.3181818181818182, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.35294117647058826, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 1, "unsupported_identifiers": 0, "distorted_sentences": 1, "message_bytes": 1435}, "ledger_by_hop": {"1": {"facts_total": 66, "facts_retained": 21, "retention_rate": 0.3181818181818182, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.35294117647058826, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1508}, "2": {"facts_total": 66, "facts_retained": 21, "retention_rate": 0.3181818181818182, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.35294117647058826, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 1, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1498}, "3": {"facts_total": 66, "facts_retained": 21, "retention_rate": 0.3181818181818182, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.35294117647058826, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 1, "unsupported_identifiers": 0, "distorted_sentences": 1, "message_bytes": 1435}, "4": {"facts_total": 66, "facts_retained": 21, "retention_rate": 0.3181818181818182, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.35294117647058826, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 1, "unsupported_identifiers": 0, "distorted_sentences": 1, "message_bytes": 1435}}}, "reference_key_id": "keys/synthetic/syn_03024_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "f946ccf31a9f007db2f33763c10d0ace160259ffb2b94381d1019bf2461c8a0a", "output_hash": "925ebdb42a0311d11ea73517fd780f133bed1e6fad2f9f6d66408fe5dba57a45"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "7aff21c6616007f66536f0bc", "attempt_id": "95fa376ca579", "run_group_id": "E4_core:A4|repairable|qwen3b:h4", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:25:21.555236Z", "task_family": "synthetic", "document_id": "syn_03024_2000", "document_hash": "ea10cd11ad08f01c50b0712bc481668857c9d2ddc8e6423968f2a31eeb97ac55", "question_id": "q05_multi_hop", "split": "test", "cluster_id": "syn_03024_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 4, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 2068, "query_tokens": 555, "message_tokens": 463, "message_utf8_bytes": 1435, "cache_tensor_bytes": null, "serialized_payload_bytes": 1435, "side_information_bytes": 0, "fetched_bytes": 3507, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 25330.884958035313, "hops": {"1": 6945.187625009567, "2": 6529.009124962613, "3": 5943.714833003469, "4": 5912.973375059664}, "receiver_prefill_ms": 210.08454205002636, "ttft_ms": 12.00154097750783, "decode_ms": 11.829583090730011}, "total_wall_ms": 25565.407915972173, "ttft_ms": 12.00154097750783, "input_usage": {"lookups": 6}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 66, "facts_retained": 21, "retention_rate": 0.3181818181818182, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.35294117647058826, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 1, "unsupported_identifiers": 0, "distorted_sentences": 1, "message_bytes": 1435}, "ledger_by_hop": {"1": {"facts_total": 66, "facts_retained": 21, "retention_rate": 0.3181818181818182, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.35294117647058826, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1508}, "2": {"facts_total": 66, "facts_retained": 21, "retention_rate": 0.3181818181818182, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.35294117647058826, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 1, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1498}, "3": {"facts_total": 66, "facts_retained": 21, "retention_rate": 0.3181818181818182, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.35294117647058826, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 1, "unsupported_identifiers": 0, "distorted_sentences": 1, "message_bytes": 1435}, "4": {"facts_total": 66, "facts_retained": 21, "retention_rate": 0.3181818181818182, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.35294117647058826, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 1, "unsupported_identifiers": 0, "distorted_sentences": 1, "message_bytes": 1435}}}, "reference_key_id": "keys/synthetic/syn_03024_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "f946ccf31a9f007db2f33763c10d0ace160259ffb2b94381d1019bf2461c8a0a", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "fbe8fa88274efe34defa866d", "attempt_id": "c91fc282df52", "run_group_id": "E4_core:A4|repairable|qwen3b:h4", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:25:21.804851Z", "task_family": "synthetic", "document_id": "syn_03024_2000", "document_hash": "ea10cd11ad08f01c50b0712bc481668857c9d2ddc8e6423968f2a31eeb97ac55", "question_id": "q06_negation", "split": "test", "cluster_id": "syn_03024_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 4, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 2068, "query_tokens": 544, "message_tokens": 463, "message_utf8_bytes": 1435, "cache_tensor_bytes": null, "serialized_payload_bytes": 1435, "side_information_bytes": 0, "fetched_bytes": 3507, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 25330.884958035313, "hops": {"1": 6945.187625009567, "2": 6529.009124962613, "3": 5943.714833003469, "4": 5912.973375059664}, "receiver_prefill_ms": 208.07570801116526, "ttft_ms": 11.908499989658594, "decode_ms": 11.832666001282632}, "total_wall_ms": 25563.19179199636, "ttft_ms": 11.908499989658594, "input_usage": {"lookups": 6}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "No", "evidence_artifact": {"ledger": {"facts_total": 66, "facts_retained": 21, "retention_rate": 0.3181818181818182, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.35294117647058826, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 1, "unsupported_identifiers": 0, "distorted_sentences": 1, "message_bytes": 1435}, "ledger_by_hop": {"1": {"facts_total": 66, "facts_retained": 21, "retention_rate": 0.3181818181818182, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.35294117647058826, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1508}, "2": {"facts_total": 66, "facts_retained": 21, "retention_rate": 0.3181818181818182, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.35294117647058826, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 1, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1498}, "3": {"facts_total": 66, "facts_retained": 21, "retention_rate": 0.3181818181818182, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.35294117647058826, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 1, "unsupported_identifiers": 0, "distorted_sentences": 1, "message_bytes": 1435}, "4": {"facts_total": 66, "facts_retained": 21, "retention_rate": 0.3181818181818182, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.35294117647058826, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 1, "unsupported_identifiers": 0, "distorted_sentences": 1, "message_bytes": 1435}}}, "reference_key_id": "keys/synthetic/syn_03024_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "f946ccf31a9f007db2f33763c10d0ace160259ffb2b94381d1019bf2461c8a0a", "output_hash": "1ea442a134b2a184bd5d40104401f2a37fbc09ccf3f4bc9da161c6099be3691d"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "f5a9c19b242f372562c2c431", "attempt_id": "88073ca7b4d6", "run_group_id": "E4_core:A4|repairable|qwen3b:h4", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:25:22.053383Z", "task_family": "synthetic", "document_id": "syn_03024_2000", "document_hash": "ea10cd11ad08f01c50b0712bc481668857c9d2ddc8e6423968f2a31eeb97ac55", "question_id": "q07_negation", "split": "test", "cluster_id": "syn_03024_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 4, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 2068, "query_tokens": 546, "message_tokens": 463, "message_utf8_bytes": 1435, "cache_tensor_bytes": null, "serialized_payload_bytes": 1435, "side_information_bytes": 0, "fetched_bytes": 3507, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 25330.884958035313, "hops": {"1": 6945.187625009567, "2": 6529.009124962613, "3": 5943.714833003469, "4": 5912.973375059664}, "receiver_prefill_ms": 209.97550000902265, "ttft_ms": 11.934541980735958, "decode_ms": 11.87483302783221}, "total_wall_ms": 25565.17512514256, "ttft_ms": 11.934541980735958, "input_usage": {"lookups": 6}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "no", "evidence_artifact": {"ledger": {"facts_total": 66, "facts_retained": 21, "retention_rate": 0.3181818181818182, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.35294117647058826, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 1, "unsupported_identifiers": 0, "distorted_sentences": 1, "message_bytes": 1435}, "ledger_by_hop": {"1": {"facts_total": 66, "facts_retained": 21, "retention_rate": 0.3181818181818182, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.35294117647058826, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1508}, "2": {"facts_total": 66, "facts_retained": 21, "retention_rate": 0.3181818181818182, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.35294117647058826, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 1, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1498}, "3": {"facts_total": 66, "facts_retained": 21, "retention_rate": 0.3181818181818182, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.35294117647058826, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 1, "unsupported_identifiers": 0, "distorted_sentences": 1, "message_bytes": 1435}, "4": {"facts_total": 66, "facts_retained": 21, "retention_rate": 0.3181818181818182, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.35294117647058826, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 1, "unsupported_identifiers": 0, "distorted_sentences": 1, "message_bytes": 1435}}}, "reference_key_id": "keys/synthetic/syn_03024_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "f946ccf31a9f007db2f33763c10d0ace160259ffb2b94381d1019bf2461c8a0a", "output_hash": "9390298f3fb0c5b160498935d79cb139aef28e1c47358b4bbba61862b9c26e59"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "96fd33520dbff2c8001b3994", "attempt_id": "a8507bd06d42", "run_group_id": "E4_core:A4|repairable|qwen3b:h4", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:25:22.303034Z", "task_family": "synthetic", "document_id": "syn_03024_2000", "document_hash": "ea10cd11ad08f01c50b0712bc481668857c9d2ddc8e6423968f2a31eeb97ac55", "question_id": "q08_revision", "split": "test", "cluster_id": "syn_03024_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 4, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 2068, "query_tokens": 562, "message_tokens": 463, "message_utf8_bytes": 1435, "cache_tensor_bytes": null, "serialized_payload_bytes": 1435, "side_information_bytes": 0, "fetched_bytes": 3507, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 25330.884958035313, "hops": {"1": 6945.187625009567, "2": 6529.009124962613, "3": 5943.714833003469, "4": 5912.973375059664}, "receiver_prefill_ms": 210.991624975577, "ttft_ms": 11.858333018608391, "decode_ms": 11.896957992576063}, "total_wall_ms": 25566.133375046775, "ttft_ms": 11.858333018608391, "input_usage": {"lookups": 6}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 66, "facts_retained": 21, "retention_rate": 0.3181818181818182, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.35294117647058826, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 1, "unsupported_identifiers": 0, "distorted_sentences": 1, "message_bytes": 1435}, "ledger_by_hop": {"1": {"facts_total": 66, "facts_retained": 21, "retention_rate": 0.3181818181818182, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.35294117647058826, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1508}, "2": {"facts_total": 66, "facts_retained": 21, "retention_rate": 0.3181818181818182, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.35294117647058826, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 1, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1498}, "3": {"facts_total": 66, "facts_retained": 21, "retention_rate": 0.3181818181818182, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.35294117647058826, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 1, "unsupported_identifiers": 0, "distorted_sentences": 1, "message_bytes": 1435}, "4": {"facts_total": 66, "facts_retained": 21, "retention_rate": 0.3181818181818182, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.35294117647058826, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 1, "unsupported_identifiers": 0, "distorted_sentences": 1, "message_bytes": 1435}}}, "reference_key_id": "keys/synthetic/syn_03024_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "f946ccf31a9f007db2f33763c10d0ace160259ffb2b94381d1019bf2461c8a0a", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "c23ef4c124ac921da98151db", "attempt_id": "536dd7bb37ab", "run_group_id": "E4_core:A4|repairable|qwen3b:h4", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:25:22.553423Z", "task_family": "synthetic", "document_id": "syn_03024_2000", "document_hash": "ea10cd11ad08f01c50b0712bc481668857c9d2ddc8e6423968f2a31eeb97ac55", "question_id": "q09_revision", "split": "test", "cluster_id": "syn_03024_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 4, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 2068, "query_tokens": 553, "message_tokens": 463, "message_utf8_bytes": 1435, "cache_tensor_bytes": null, "serialized_payload_bytes": 1435, "side_information_bytes": 0, "fetched_bytes": 3507, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 25330.884958035313, "hops": {"1": 6945.187625009567, "2": 6529.009124962613, "3": 5943.714833003469, "4": 5912.973375059664}, "receiver_prefill_ms": 210.20912495441735, "ttft_ms": 11.930332984775305, "decode_ms": 11.827250011265278}, "total_wall_ms": 25565.332373953424, "ttft_ms": 11.930332984775305, "input_usage": {"lookups": 6}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 66, "facts_retained": 21, "retention_rate": 0.3181818181818182, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.35294117647058826, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 1, "unsupported_identifiers": 0, "distorted_sentences": 1, "message_bytes": 1435}, "ledger_by_hop": {"1": {"facts_total": 66, "facts_retained": 21, "retention_rate": 0.3181818181818182, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.35294117647058826, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1508}, "2": {"facts_total": 66, "facts_retained": 21, "retention_rate": 0.3181818181818182, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.35294117647058826, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 1, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1498}, "3": {"facts_total": 66, "facts_retained": 21, "retention_rate": 0.3181818181818182, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.35294117647058826, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 1, "unsupported_identifiers": 0, "distorted_sentences": 1, "message_bytes": 1435}, "4": {"facts_total": 66, "facts_retained": 21, "retention_rate": 0.3181818181818182, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.35294117647058826, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 1, "unsupported_identifiers": 0, "distorted_sentences": 1, "message_bytes": 1435}}}, "reference_key_id": "keys/synthetic/syn_03024_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "f946ccf31a9f007db2f33763c10d0ace160259ffb2b94381d1019bf2461c8a0a", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "36ba732ea4cfc41cc1d732ce", "attempt_id": "1e5a75f88d73", "run_group_id": "E4_core:A4|repairable|qwen3b:h4", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:25:22.803404Z", "task_family": "synthetic", "document_id": "syn_03024_2000", "document_hash": "ea10cd11ad08f01c50b0712bc481668857c9d2ddc8e6423968f2a31eeb97ac55", "question_id": "q10_temporal", "split": "test", "cluster_id": "syn_03024_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 4, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 2068, "query_tokens": 563, "message_tokens": 463, "message_utf8_bytes": 1435, "cache_tensor_bytes": null, "serialized_payload_bytes": 1435, "side_information_bytes": 0, "fetched_bytes": 3507, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 25330.884958035313, "hops": {"1": 6945.187625009567, "2": 6529.009124962613, "3": 5943.714833003469, "4": 5912.973375059664}, "receiver_prefill_ms": 210.91000002343208, "ttft_ms": 11.837874888442457, "decode_ms": 71.11466606147587}, "total_wall_ms": 25625.217791064642, "ttft_ms": 11.837874888442457, "input_usage": {"lookups": 6}, "output_usage": {"output_tokens": 6}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "Quilun-16", "evidence_artifact": {"ledger": {"facts_total": 66, "facts_retained": 21, "retention_rate": 0.3181818181818182, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.35294117647058826, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 1, "unsupported_identifiers": 0, "distorted_sentences": 1, "message_bytes": 1435}, "ledger_by_hop": {"1": {"facts_total": 66, "facts_retained": 21, "retention_rate": 0.3181818181818182, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.35294117647058826, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1508}, "2": {"facts_total": 66, "facts_retained": 21, "retention_rate": 0.3181818181818182, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.35294117647058826, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 1, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1498}, "3": {"facts_total": 66, "facts_retained": 21, "retention_rate": 0.3181818181818182, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.35294117647058826, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 1, "unsupported_identifiers": 0, "distorted_sentences": 1, "message_bytes": 1435}, "4": {"facts_total": 66, "facts_retained": 21, "retention_rate": 0.3181818181818182, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.35294117647058826, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 1, "unsupported_identifiers": 0, "distorted_sentences": 1, "message_bytes": 1435}}}, "reference_key_id": "keys/synthetic/syn_03024_2000.json", "task_scores": {"correct": 0.0, "abstained": 0.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "f946ccf31a9f007db2f33763c10d0ace160259ffb2b94381d1019bf2461c8a0a", "output_hash": "34a7c0e11c93ba75eb4b20d3d3f5303b417856d89a232e8b16fef75c4336afc6"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "f0b5796cbe429684200f5e2e", "attempt_id": "7371af8583c7", "run_group_id": "E4_core:A4|repairable|qwen3b:h4", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:25:23.113146Z", "task_family": "synthetic", "document_id": "syn_03024_2000", "document_hash": "ea10cd11ad08f01c50b0712bc481668857c9d2ddc8e6423968f2a31eeb97ac55", "question_id": "q11_aggregation", "split": "test", "cluster_id": "syn_03024_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 4, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 2068, "query_tokens": 556, "message_tokens": 463, "message_utf8_bytes": 1435, "cache_tensor_bytes": null, "serialized_payload_bytes": 1435, "side_information_bytes": 0, "fetched_bytes": 3507, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 25330.884958035313, "hops": {"1": 6945.187625009567, "2": 6529.009124962613, "3": 5943.714833003469, "4": 5912.973375059664}, "receiver_prefill_ms": 210.468832985498, "ttft_ms": 11.858290992677212, "decode_ms": 118.48804203327745}, "total_wall_ms": 25672.169915982522, "ttft_ms": 11.858290992677212, "input_usage": {"lookups": 6}, "output_usage": {"output_tokens": 10}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "37,154,259", "evidence_artifact": {"ledger": {"facts_total": 66, "facts_retained": 21, "retention_rate": 0.3181818181818182, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.35294117647058826, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 1, "unsupported_identifiers": 0, "distorted_sentences": 1, "message_bytes": 1435}, "ledger_by_hop": {"1": {"facts_total": 66, "facts_retained": 21, "retention_rate": 0.3181818181818182, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.35294117647058826, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1508}, "2": {"facts_total": 66, "facts_retained": 21, "retention_rate": 0.3181818181818182, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.35294117647058826, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 1, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1498}, "3": {"facts_total": 66, "facts_retained": 21, "retention_rate": 0.3181818181818182, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.35294117647058826, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 1, "unsupported_identifiers": 0, "distorted_sentences": 1, "message_bytes": 1435}, "4": {"facts_total": 66, "facts_retained": 21, "retention_rate": 0.3181818181818182, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.35294117647058826, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 1, "unsupported_identifiers": 0, "distorted_sentences": 1, "message_bytes": 1435}}}, "reference_key_id": "keys/synthetic/syn_03024_2000.json", "task_scores": {"correct": 0.0, "abstained": 0.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "f946ccf31a9f007db2f33763c10d0ace160259ffb2b94381d1019bf2461c8a0a", "output_hash": "475e1c30b81f429f1ab8d86eb02313598a196f4663b8300f83f9721827c72fa2"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "9711e0131a0cd3d701c0574b", "attempt_id": "7b1bf9d26ee2", "run_group_id": "E4_core:A4|repairable|qwen3b:h4", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:25:23.469497Z", "task_family": "synthetic", "document_id": "syn_03024_2000", "document_hash": "ea10cd11ad08f01c50b0712bc481668857c9d2ddc8e6423968f2a31eeb97ac55", "question_id": "q12_unit_conversion", "split": "test", "cluster_id": "syn_03024_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 4, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 2068, "query_tokens": 553, "message_tokens": 463, "message_utf8_bytes": 1435, "cache_tensor_bytes": null, "serialized_payload_bytes": 1435, "side_information_bytes": 0, "fetched_bytes": 3507, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 25330.884958035313, "hops": {"1": 6945.187625009567, "2": 6529.009124962613, "3": 5943.714833003469, "4": 5912.973375059664}, "receiver_prefill_ms": 210.56908403988928, "ttft_ms": 11.872458970174193, "decode_ms": 11.849624919705093}, "total_wall_ms": 25565.627333009616, "ttft_ms": 11.872458970174193, "input_usage": {"lookups": 6}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 66, "facts_retained": 21, "retention_rate": 0.3181818181818182, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.35294117647058826, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 1, "unsupported_identifiers": 0, "distorted_sentences": 1, "message_bytes": 1435}, "ledger_by_hop": {"1": {"facts_total": 66, "facts_retained": 21, "retention_rate": 0.3181818181818182, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.35294117647058826, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1508}, "2": {"facts_total": 66, "facts_retained": 21, "retention_rate": 0.3181818181818182, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.35294117647058826, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 1, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1498}, "3": {"facts_total": 66, "facts_retained": 21, "retention_rate": 0.3181818181818182, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.35294117647058826, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 1, "unsupported_identifiers": 0, "distorted_sentences": 1, "message_bytes": 1435}, "4": {"facts_total": 66, "facts_retained": 21, "retention_rate": 0.3181818181818182, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.35294117647058826, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 1, "unsupported_identifiers": 0, "distorted_sentences": 1, "message_bytes": 1435}}}, "reference_key_id": "keys/synthetic/syn_03024_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "f946ccf31a9f007db2f33763c10d0ace160259ffb2b94381d1019bf2461c8a0a", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "d0b7c8d4571db28f7ac5b1e1", "attempt_id": "f50266f7a39b", "run_group_id": "E4_core:A4|repairable|qwen3b:h4", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:25:23.719579Z", "task_family": "synthetic", "document_id": "syn_03024_2000", "document_hash": "ea10cd11ad08f01c50b0712bc481668857c9d2ddc8e6423968f2a31eeb97ac55", "question_id": "q13_unanswerable", "split": "test", "cluster_id": "syn_03024_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 4, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 2068, "query_tokens": 557, "message_tokens": 463, "message_utf8_bytes": 1435, "cache_tensor_bytes": null, "serialized_payload_bytes": 1435, "side_information_bytes": 0, "fetched_bytes": 3507, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 25330.884958035313, "hops": {"1": 6945.187625009567, "2": 6529.009124962613, "3": 5943.714833003469, "4": 5912.973375059664}, "receiver_prefill_ms": 210.87249997071922, "ttft_ms": 11.962375021539629, "decode_ms": 11.809041956439614}, "total_wall_ms": 25565.994290984236, "ttft_ms": 11.962375021539629, "input_usage": {"lookups": 6}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 66, "facts_retained": 21, "retention_rate": 0.3181818181818182, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.35294117647058826, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 1, "unsupported_identifiers": 0, "distorted_sentences": 1, "message_bytes": 1435}, "ledger_by_hop": {"1": {"facts_total": 66, "facts_retained": 21, "retention_rate": 0.3181818181818182, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.35294117647058826, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1508}, "2": {"facts_total": 66, "facts_retained": 21, "retention_rate": 0.3181818181818182, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.35294117647058826, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 1, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1498}, "3": {"facts_total": 66, "facts_retained": 21, "retention_rate": 0.3181818181818182, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.35294117647058826, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 1, "unsupported_identifiers": 0, "distorted_sentences": 1, "message_bytes": 1435}, "4": {"facts_total": 66, "facts_retained": 21, "retention_rate": 0.3181818181818182, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.35294117647058826, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 1, "unsupported_identifiers": 0, "distorted_sentences": 1, "message_bytes": 1435}}}, "reference_key_id": "keys/synthetic/syn_03024_2000.json", "task_scores": {"correct": 1.0, "abstained": 1.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "f946ccf31a9f007db2f33763c10d0ace160259ffb2b94381d1019bf2461c8a0a", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "adb01e79b0bbb1677e43af85", "attempt_id": "bf93f950cae7", "run_group_id": "E4_core:A4|repairable|qwen3b:h4", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:25:23.970052Z", "task_family": "synthetic", "document_id": "syn_03024_2000", "document_hash": "ea10cd11ad08f01c50b0712bc481668857c9d2ddc8e6423968f2a31eeb97ac55", "question_id": "q14_unanswerable", "split": "test", "cluster_id": "syn_03024_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 4, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 2068, "query_tokens": 558, "message_tokens": 463, "message_utf8_bytes": 1435, "cache_tensor_bytes": null, "serialized_payload_bytes": 1435, "side_information_bytes": 0, "fetched_bytes": 3507, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 25330.884958035313, "hops": {"1": 6945.187625009567, "2": 6529.009124962613, "3": 5943.714833003469, "4": 5912.973375059664}, "receiver_prefill_ms": 210.50145791377872, "ttft_ms": 11.901709018275142, "decode_ms": 11.903250007890165}, "total_wall_ms": 25565.673999954015, "ttft_ms": 11.901709018275142, "input_usage": {"lookups": 6}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 66, "facts_retained": 21, "retention_rate": 0.3181818181818182, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.35294117647058826, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 1, "unsupported_identifiers": 0, "distorted_sentences": 1, "message_bytes": 1435}, "ledger_by_hop": {"1": {"facts_total": 66, "facts_retained": 21, "retention_rate": 0.3181818181818182, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.35294117647058826, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1508}, "2": {"facts_total": 66, "facts_retained": 21, "retention_rate": 0.3181818181818182, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.35294117647058826, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 1, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1498}, "3": {"facts_total": 66, "facts_retained": 21, "retention_rate": 0.3181818181818182, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.35294117647058826, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 1, "unsupported_identifiers": 0, "distorted_sentences": 1, "message_bytes": 1435}, "4": {"facts_total": 66, "facts_retained": 21, "retention_rate": 0.3181818181818182, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.35294117647058826, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 1, "unsupported_identifiers": 0, "distorted_sentences": 1, "message_bytes": 1435}}}, "reference_key_id": "keys/synthetic/syn_03024_2000.json", "task_scores": {"correct": 1.0, "abstained": 1.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "f946ccf31a9f007db2f33763c10d0ace160259ffb2b94381d1019bf2461c8a0a", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "ea9d7975f67b5aeb7c411aa4", "attempt_id": "63f6e7b49bf5", "run_group_id": "E4_core:A4|repairable|qwen3b:h8", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:25:47.857837Z", "task_family": "synthetic", "document_id": "syn_03024_2000", "document_hash": "ea10cd11ad08f01c50b0712bc481668857c9d2ddc8e6423968f2a31eeb97ac55", "question_id": "q00_single_fact", "split": "test", "cluster_id": "syn_03024_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 8, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 2068, "query_tokens": 545, "message_tokens": 463, "message_utf8_bytes": 1435, "cache_tensor_bytes": null, "serialized_payload_bytes": 1435, "side_information_bytes": 0, "fetched_bytes": 8183, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 48962.82533311751, "hops": {"1": 6945.187625009567, "2": 6529.009124962613, "3": 5943.714833003469, "4": 5912.973375059664, "5": 5908.367124968208, "6": 5911.119500058703, "7": 5908.995958045125, "8": 5903.457792010158}, "receiver_prefill_ms": 208.17083399742842, "ttft_ms": 11.87329099047929, "decode_ms": 11.856459081172943}, "total_wall_ms": 49195.238958112895, "ttft_ms": 11.87329099047929, "input_usage": {"lookups": 14}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 66, "facts_retained": 21, "retention_rate": 0.3181818181818182, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.35294117647058826, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 1, "unsupported_identifiers": 0, "distorted_sentences": 1, "message_bytes": 1435}, "ledger_by_hop": {"1": {"facts_total": 66, "facts_retained": 21, "retention_rate": 0.3181818181818182, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.35294117647058826, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1508}, "2": {"facts_total": 66, "facts_retained": 21, "retention_rate": 0.3181818181818182, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.35294117647058826, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 1, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1498}, "3": {"facts_total": 66, "facts_retained": 21, "retention_rate": 0.3181818181818182, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.35294117647058826, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 1, "unsupported_identifiers": 0, "distorted_sentences": 1, "message_bytes": 1435}, "4": {"facts_total": 66, "facts_retained": 21, "retention_rate": 0.3181818181818182, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.35294117647058826, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 1, "unsupported_identifiers": 0, "distorted_sentences": 1, "message_bytes": 1435}, "5": {"facts_total": 66, "facts_retained": 21, "retention_rate": 0.3181818181818182, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.35294117647058826, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 1, "unsupported_identifiers": 0, "distorted_sentences": 1, "message_bytes": 1435}, "6": {"facts_total": 66, "facts_retained": 21, "retention_rate": 0.3181818181818182, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.35294117647058826, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 1, "unsupported_identifiers": 0, "distorted_sentences": 1, "message_bytes": 1435}, "7": {"facts_total": 66, "facts_retained": 21, "retention_rate": 0.3181818181818182, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.35294117647058826, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 1, "unsupported_identifiers": 0, "distorted_sentences": 1, "message_bytes": 1435}, "8": {"facts_total": 66, "facts_retained": 21, "retention_rate": 0.3181818181818182, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.35294117647058826, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 1, "unsupported_identifiers": 0, "distorted_sentences": 1, "message_bytes": 1435}}}, "reference_key_id": "keys/synthetic/syn_03024_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "f946ccf31a9f007db2f33763c10d0ace160259ffb2b94381d1019bf2461c8a0a", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "4a6ccde7d05e2f611db92b53", "attempt_id": "a67b54cc6126", "run_group_id": "E4_core:A4|repairable|qwen3b:h8", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:25:48.106543Z", "task_family": "synthetic", "document_id": "syn_03024_2000", "document_hash": "ea10cd11ad08f01c50b0712bc481668857c9d2ddc8e6423968f2a31eeb97ac55", "question_id": "q01_single_fact", "split": "test", "cluster_id": "syn_03024_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 8, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 2068, "query_tokens": 550, "message_tokens": 463, "message_utf8_bytes": 1435, "cache_tensor_bytes": null, "serialized_payload_bytes": 1435, "side_information_bytes": 0, "fetched_bytes": 8183, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 48962.82533311751, "hops": {"1": 6945.187625009567, "2": 6529.009124962613, "3": 5943.714833003469, "4": 5912.973375059664, "5": 5908.367124968208, "6": 5911.119500058703, "7": 5908.995958045125, "8": 5903.457792010158}, "receiver_prefill_ms": 210.02850006334484, "ttft_ms": 11.856041965074837, "decode_ms": 11.861832928843796}, "total_wall_ms": 49197.0619171625, "ttft_ms": 11.856041965074837, "input_usage": {"lookups": 14}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 66, "facts_retained": 21, "retention_rate": 0.3181818181818182, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.35294117647058826, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 1, "unsupported_identifiers": 0, "distorted_sentences": 1, "message_bytes": 1435}, "ledger_by_hop": {"1": {"facts_total": 66, "facts_retained": 21, "retention_rate": 0.3181818181818182, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.35294117647058826, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1508}, "2": {"facts_total": 66, "facts_retained": 21, "retention_rate": 0.3181818181818182, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.35294117647058826, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 1, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1498}, "3": {"facts_total": 66, "facts_retained": 21, "retention_rate": 0.3181818181818182, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.35294117647058826, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 1, "unsupported_identifiers": 0, "distorted_sentences": 1, "message_bytes": 1435}, "4": {"facts_total": 66, "facts_retained": 21, "retention_rate": 0.3181818181818182, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.35294117647058826, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 1, "unsupported_identifiers": 0, "distorted_sentences": 1, "message_bytes": 1435}, "5": {"facts_total": 66, "facts_retained": 21, "retention_rate": 0.3181818181818182, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.35294117647058826, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 1, "unsupported_identifiers": 0, "distorted_sentences": 1, "message_bytes": 1435}, "6": {"facts_total": 66, "facts_retained": 21, "retention_rate": 0.3181818181818182, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.35294117647058826, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 1, "unsupported_identifiers": 0, "distorted_sentences": 1, "message_bytes": 1435}, "7": {"facts_total": 66, "facts_retained": 21, "retention_rate": 0.3181818181818182, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.35294117647058826, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 1, "unsupported_identifiers": 0, "distorted_sentences": 1, "message_bytes": 1435}, "8": {"facts_total": 66, "facts_retained": 21, "retention_rate": 0.3181818181818182, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.35294117647058826, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 1, "unsupported_identifiers": 0, "distorted_sentences": 1, "message_bytes": 1435}}}, "reference_key_id": "keys/synthetic/syn_03024_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "f946ccf31a9f007db2f33763c10d0ace160259ffb2b94381d1019bf2461c8a0a", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "07212d7317ff3732c632195e", "attempt_id": "4fa869a32a56", "run_group_id": "E4_core:A4|repairable|qwen3b:h8", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:25:48.356141Z", "task_family": "synthetic", "document_id": "syn_03024_2000", "document_hash": "ea10cd11ad08f01c50b0712bc481668857c9d2ddc8e6423968f2a31eeb97ac55", "question_id": "q02_exact_copy", "split": "test", "cluster_id": "syn_03024_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 8, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 2068, "query_tokens": 550, "message_tokens": 463, "message_utf8_bytes": 1435, "cache_tensor_bytes": null, "serialized_payload_bytes": 1435, "side_information_bytes": 0, "fetched_bytes": 8183, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 48962.82533311751, "hops": {"1": 6945.187625009567, "2": 6529.009124962613, "3": 5943.714833003469, "4": 5912.973375059664, "5": 5908.367124968208, "6": 5911.119500058703, "7": 5908.995958045125, "8": 5903.457792010158}, "receiver_prefill_ms": 209.67208303045481, "ttft_ms": 12.14137498755008, "decode_ms": 82.86954194772989}, "total_wall_ms": 49268.13470805064, "ttft_ms": 12.14137498755008, "input_usage": {"lookups": 14}, "output_usage": {"output_tokens": 7}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "30.0224", "evidence_artifact": {"ledger": {"facts_total": 66, "facts_retained": 21, "retention_rate": 0.3181818181818182, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.35294117647058826, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 1, "unsupported_identifiers": 0, "distorted_sentences": 1, "message_bytes": 1435}, "ledger_by_hop": {"1": {"facts_total": 66, "facts_retained": 21, "retention_rate": 0.3181818181818182, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.35294117647058826, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1508}, "2": {"facts_total": 66, "facts_retained": 21, "retention_rate": 0.3181818181818182, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.35294117647058826, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 1, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1498}, "3": {"facts_total": 66, "facts_retained": 21, "retention_rate": 0.3181818181818182, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.35294117647058826, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 1, "unsupported_identifiers": 0, "distorted_sentences": 1, "message_bytes": 1435}, "4": {"facts_total": 66, "facts_retained": 21, "retention_rate": 0.3181818181818182, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.35294117647058826, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 1, "unsupported_identifiers": 0, "distorted_sentences": 1, "message_bytes": 1435}, "5": {"facts_total": 66, "facts_retained": 21, "retention_rate": 0.3181818181818182, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.35294117647058826, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 1, "unsupported_identifiers": 0, "distorted_sentences": 1, "message_bytes": 1435}, "6": {"facts_total": 66, "facts_retained": 21, "retention_rate": 0.3181818181818182, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.35294117647058826, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 1, "unsupported_identifiers": 0, "distorted_sentences": 1, "message_bytes": 1435}, "7": {"facts_total": 66, "facts_retained": 21, "retention_rate": 0.3181818181818182, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.35294117647058826, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 1, "unsupported_identifiers": 0, "distorted_sentences": 1, "message_bytes": 1435}, "8": {"facts_total": 66, "facts_retained": 21, "retention_rate": 0.3181818181818182, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.35294117647058826, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 1, "unsupported_identifiers": 0, "distorted_sentences": 1, "message_bytes": 1435}}}, "reference_key_id": "keys/synthetic/syn_03024_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "f946ccf31a9f007db2f33763c10d0ace160259ffb2b94381d1019bf2461c8a0a", "output_hash": "e28a78bc4deb3cec8beab4bf5b87297d7cc55a3e6f839b43fe6408735d464923"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "36371940b36eea642b4e07dc", "attempt_id": "6cfe80fcc392", "run_group_id": "E4_core:A4|repairable|qwen3b:h8", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:25:48.676719Z", "task_family": "synthetic", "document_id": "syn_03024_2000", "document_hash": "ea10cd11ad08f01c50b0712bc481668857c9d2ddc8e6423968f2a31eeb97ac55", "question_id": "q03_exact_copy", "split": "test", "cluster_id": "syn_03024_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 8, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 2068, "query_tokens": 545, "message_tokens": 463, "message_utf8_bytes": 1435, "cache_tensor_bytes": null, "serialized_payload_bytes": 1435, "side_information_bytes": 0, "fetched_bytes": 8183, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 48962.82533311751, "hops": {"1": 6945.187625009567, "2": 6529.009124962613, "3": 5943.714833003469, "4": 5912.973375059664, "5": 5908.367124968208, "6": 5911.119500058703, "7": 5908.995958045125, "8": 5903.457792010158}, "receiver_prefill_ms": 208.0312919570133, "ttft_ms": 11.740374960936606, "decode_ms": 11.79654209408909}, "total_wall_ms": 49194.9811251834, "ttft_ms": 11.740374960936606, "input_usage": {"lookups": 14}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 66, "facts_retained": 21, "retention_rate": 0.3181818181818182, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.35294117647058826, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 1, "unsupported_identifiers": 0, "distorted_sentences": 1, "message_bytes": 1435}, "ledger_by_hop": {"1": {"facts_total": 66, "facts_retained": 21, "retention_rate": 0.3181818181818182, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.35294117647058826, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1508}, "2": {"facts_total": 66, "facts_retained": 21, "retention_rate": 0.3181818181818182, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.35294117647058826, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 1, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1498}, "3": {"facts_total": 66, "facts_retained": 21, "retention_rate": 0.3181818181818182, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.35294117647058826, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 1, "unsupported_identifiers": 0, "distorted_sentences": 1, "message_bytes": 1435}, "4": {"facts_total": 66, "facts_retained": 21, "retention_rate": 0.3181818181818182, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.35294117647058826, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 1, "unsupported_identifiers": 0, "distorted_sentences": 1, "message_bytes": 1435}, "5": {"facts_total": 66, "facts_retained": 21, "retention_rate": 0.3181818181818182, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.35294117647058826, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 1, "unsupported_identifiers": 0, "distorted_sentences": 1, "message_bytes": 1435}, "6": {"facts_total": 66, "facts_retained": 21, "retention_rate": 0.3181818181818182, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.35294117647058826, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 1, "unsupported_identifiers": 0, "distorted_sentences": 1, "message_bytes": 1435}, "7": {"facts_total": 66, "facts_retained": 21, "retention_rate": 0.3181818181818182, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.35294117647058826, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 1, "unsupported_identifiers": 0, "distorted_sentences": 1, "message_bytes": 1435}, "8": {"facts_total": 66, "facts_retained": 21, "retention_rate": 0.3181818181818182, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.35294117647058826, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 1, "unsupported_identifiers": 0, "distorted_sentences": 1, "message_bytes": 1435}}}, "reference_key_id": "keys/synthetic/syn_03024_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "f946ccf31a9f007db2f33763c10d0ace160259ffb2b94381d1019bf2461c8a0a", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "f8745caef4dfc82a81e2f64d", "attempt_id": "0604157bcc5a", "run_group_id": "E4_core:A4|repairable|qwen3b:h8", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:25:48.924806Z", "task_family": "synthetic", "document_id": "syn_03024_2000", "document_hash": "ea10cd11ad08f01c50b0712bc481668857c9d2ddc8e6423968f2a31eeb97ac55", "question_id": "q04_multi_hop", "split": "test", "cluster_id": "syn_03024_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 8, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 2068, "query_tokens": 551, "message_tokens": 463, "message_utf8_bytes": 1435, "cache_tensor_bytes": null, "serialized_payload_bytes": 1435, "side_information_bytes": 0, "fetched_bytes": 8183, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 48962.82533311751, "hops": {"1": 6945.187625009567, "2": 6529.009124962613, "3": 5943.714833003469, "4": 5912.973375059664, "5": 5908.367124968208, "6": 5911.119500058703, "7": 5908.995958045125, "8": 5903.457792010158}, "receiver_prefill_ms": 209.74558300804347, "ttft_ms": 12.021249975077808, "decode_ms": 83.20454100612551}, "total_wall_ms": 49268.38666712865, "ttft_ms": 12.021249975077808, "input_usage": {"lookups": 14}, "output_usage": {"output_tokens": 7}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "6009480", "evidence_artifact": {"ledger": {"facts_total": 66, "facts_retained": 21, "retention_rate": 0.3181818181818182, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.35294117647058826, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 1, "unsupported_identifiers": 0, "distorted_sentences": 1, "message_bytes": 1435}, "ledger_by_hop": {"1": {"facts_total": 66, "facts_retained": 21, "retention_rate": 0.3181818181818182, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.35294117647058826, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1508}, "2": {"facts_total": 66, "facts_retained": 21, "retention_rate": 0.3181818181818182, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.35294117647058826, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 1, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1498}, "3": {"facts_total": 66, "facts_retained": 21, "retention_rate": 0.3181818181818182, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.35294117647058826, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 1, "unsupported_identifiers": 0, "distorted_sentences": 1, "message_bytes": 1435}, "4": {"facts_total": 66, "facts_retained": 21, "retention_rate": 0.3181818181818182, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.35294117647058826, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 1, "unsupported_identifiers": 0, "distorted_sentences": 1, "message_bytes": 1435}, "5": {"facts_total": 66, "facts_retained": 21, "retention_rate": 0.3181818181818182, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.35294117647058826, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 1, "unsupported_identifiers": 0, "distorted_sentences": 1, "message_bytes": 1435}, "6": {"facts_total": 66, "facts_retained": 21, "retention_rate": 0.3181818181818182, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.35294117647058826, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 1, "unsupported_identifiers": 0, "distorted_sentences": 1, "message_bytes": 1435}, "7": {"facts_total": 66, "facts_retained": 21, "retention_rate": 0.3181818181818182, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.35294117647058826, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 1, "unsupported_identifiers": 0, "distorted_sentences": 1, "message_bytes": 1435}, "8": {"facts_total": 66, "facts_retained": 21, "retention_rate": 0.3181818181818182, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.35294117647058826, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 1, "unsupported_identifiers": 0, "distorted_sentences": 1, "message_bytes": 1435}}}, "reference_key_id": "keys/synthetic/syn_03024_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "f946ccf31a9f007db2f33763c10d0ace160259ffb2b94381d1019bf2461c8a0a", "output_hash": "925ebdb42a0311d11ea73517fd780f133bed1e6fad2f9f6d66408fe5dba57a45"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "7a59db27d10289109f0328a0", "attempt_id": "88fa2190d8c9", "run_group_id": "E4_core:A4|repairable|qwen3b:h8", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:25:49.247229Z", "task_family": "synthetic", "document_id": "syn_03024_2000", "document_hash": "ea10cd11ad08f01c50b0712bc481668857c9d2ddc8e6423968f2a31eeb97ac55", "question_id": "q05_multi_hop", "split": "test", "cluster_id": "syn_03024_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 8, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 2068, "query_tokens": 555, "message_tokens": 463, "message_utf8_bytes": 1435, "cache_tensor_bytes": null, "serialized_payload_bytes": 1435, "side_information_bytes": 0, "fetched_bytes": 8183, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 48962.82533311751, "hops": {"1": 6945.187625009567, "2": 6529.009124962613, "3": 5943.714833003469, "4": 5912.973375059664, "5": 5908.367124968208, "6": 5911.119500058703, "7": 5908.995958045125, "8": 5903.457792010158}, "receiver_prefill_ms": 210.0943330442533, "ttft_ms": 12.061125016771257, "decode_ms": 11.999916983768344}, "total_wall_ms": 49197.596250101924, "ttft_ms": 12.061125016771257, "input_usage": {"lookups": 14}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 66, "facts_retained": 21, "retention_rate": 0.3181818181818182, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.35294117647058826, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 1, "unsupported_identifiers": 0, "distorted_sentences": 1, "message_bytes": 1435}, "ledger_by_hop": {"1": {"facts_total": 66, "facts_retained": 21, "retention_rate": 0.3181818181818182, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.35294117647058826, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1508}, "2": {"facts_total": 66, "facts_retained": 21, "retention_rate": 0.3181818181818182, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.35294117647058826, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 1, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1498}, "3": {"facts_total": 66, "facts_retained": 21, "retention_rate": 0.3181818181818182, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.35294117647058826, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 1, "unsupported_identifiers": 0, "distorted_sentences": 1, "message_bytes": 1435}, "4": {"facts_total": 66, "facts_retained": 21, "retention_rate": 0.3181818181818182, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.35294117647058826, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 1, "unsupported_identifiers": 0, "distorted_sentences": 1, "message_bytes": 1435}, "5": {"facts_total": 66, "facts_retained": 21, "retention_rate": 0.3181818181818182, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.35294117647058826, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 1, "unsupported_identifiers": 0, "distorted_sentences": 1, "message_bytes": 1435}, "6": {"facts_total": 66, "facts_retained": 21, "retention_rate": 0.3181818181818182, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.35294117647058826, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 1, "unsupported_identifiers": 0, "distorted_sentences": 1, "message_bytes": 1435}, "7": {"facts_total": 66, "facts_retained": 21, "retention_rate": 0.3181818181818182, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.35294117647058826, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 1, "unsupported_identifiers": 0, "distorted_sentences": 1, "message_bytes": 1435}, "8": {"facts_total": 66, "facts_retained": 21, "retention_rate": 0.3181818181818182, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.35294117647058826, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 1, "unsupported_identifiers": 0, "distorted_sentences": 1, "message_bytes": 1435}}}, "reference_key_id": "keys/synthetic/syn_03024_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "f946ccf31a9f007db2f33763c10d0ace160259ffb2b94381d1019bf2461c8a0a", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "ef45120307fc2266b11d8af7", "attempt_id": "825a9c93c631", "run_group_id": "E4_core:A4|repairable|qwen3b:h8", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:25:49.498147Z", "task_family": "synthetic", "document_id": "syn_03024_2000", "document_hash": "ea10cd11ad08f01c50b0712bc481668857c9d2ddc8e6423968f2a31eeb97ac55", "question_id": "q06_negation", "split": "test", "cluster_id": "syn_03024_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 8, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 2068, "query_tokens": 544, "message_tokens": 463, "message_utf8_bytes": 1435, "cache_tensor_bytes": null, "serialized_payload_bytes": 1435, "side_information_bytes": 0, "fetched_bytes": 8183, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 48962.82533311751, "hops": {"1": 6945.187625009567, "2": 6529.009124962613, "3": 5943.714833003469, "4": 5912.973375059664, "5": 5908.367124968208, "6": 5911.119500058703, "7": 5908.995958045125, "8": 5903.457792010158}, "receiver_prefill_ms": 208.67200009524822, "ttft_ms": 11.841083061881363, "decode_ms": 11.896749958395958}, "total_wall_ms": 49195.72095805779, "ttft_ms": 11.841083061881363, "input_usage": {"lookups": 14}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "No", "evidence_artifact": {"ledger": {"facts_total": 66, "facts_retained": 21, "retention_rate": 0.3181818181818182, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.35294117647058826, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 1, "unsupported_identifiers": 0, "distorted_sentences": 1, "message_bytes": 1435}, "ledger_by_hop": {"1": {"facts_total": 66, "facts_retained": 21, "retention_rate": 0.3181818181818182, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.35294117647058826, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1508}, "2": {"facts_total": 66, "facts_retained": 21, "retention_rate": 0.3181818181818182, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.35294117647058826, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 1, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1498}, "3": {"facts_total": 66, "facts_retained": 21, "retention_rate": 0.3181818181818182, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.35294117647058826, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 1, "unsupported_identifiers": 0, "distorted_sentences": 1, "message_bytes": 1435}, "4": {"facts_total": 66, "facts_retained": 21, "retention_rate": 0.3181818181818182, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.35294117647058826, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 1, "unsupported_identifiers": 0, "distorted_sentences": 1, "message_bytes": 1435}, "5": {"facts_total": 66, "facts_retained": 21, "retention_rate": 0.3181818181818182, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.35294117647058826, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 1, "unsupported_identifiers": 0, "distorted_sentences": 1, "message_bytes": 1435}, "6": {"facts_total": 66, "facts_retained": 21, "retention_rate": 0.3181818181818182, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.35294117647058826, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 1, "unsupported_identifiers": 0, "distorted_sentences": 1, "message_bytes": 1435}, "7": {"facts_total": 66, "facts_retained": 21, "retention_rate": 0.3181818181818182, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.35294117647058826, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 1, "unsupported_identifiers": 0, "distorted_sentences": 1, "message_bytes": 1435}, "8": {"facts_total": 66, "facts_retained": 21, "retention_rate": 0.3181818181818182, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.35294117647058826, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 1, "unsupported_identifiers": 0, "distorted_sentences": 1, "message_bytes": 1435}}}, "reference_key_id": "keys/synthetic/syn_03024_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "f946ccf31a9f007db2f33763c10d0ace160259ffb2b94381d1019bf2461c8a0a", "output_hash": "1ea442a134b2a184bd5d40104401f2a37fbc09ccf3f4bc9da161c6099be3691d"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "c42dad7e95de16e07f29fb97", "attempt_id": "40addd9b02ed", "run_group_id": "E4_core:A4|repairable|qwen3b:h8", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:25:49.746514Z", "task_family": "synthetic", "document_id": "syn_03024_2000", "document_hash": "ea10cd11ad08f01c50b0712bc481668857c9d2ddc8e6423968f2a31eeb97ac55", "question_id": "q07_negation", "split": "test", "cluster_id": "syn_03024_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 8, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 2068, "query_tokens": 546, "message_tokens": 463, "message_utf8_bytes": 1435, "cache_tensor_bytes": null, "serialized_payload_bytes": 1435, "side_information_bytes": 0, "fetched_bytes": 8183, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 48962.82533311751, "hops": {"1": 6945.187625009567, "2": 6529.009124962613, "3": 5943.714833003469, "4": 5912.973375059664, "5": 5908.367124968208, "6": 5911.119500058703, "7": 5908.995958045125, "8": 5903.457792010158}, "receiver_prefill_ms": 209.4325409270823, "ttft_ms": 11.94883300922811, "decode_ms": 11.900707962922752}, "total_wall_ms": 49196.7768331524, "ttft_ms": 11.94883300922811, "input_usage": {"lookups": 14}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "no", "evidence_artifact": {"ledger": {"facts_total": 66, "facts_retained": 21, "retention_rate": 0.3181818181818182, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.35294117647058826, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 1, "unsupported_identifiers": 0, "distorted_sentences": 1, "message_bytes": 1435}, "ledger_by_hop": {"1": {"facts_total": 66, "facts_retained": 21, "retention_rate": 0.3181818181818182, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.35294117647058826, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1508}, "2": {"facts_total": 66, "facts_retained": 21, "retention_rate": 0.3181818181818182, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.35294117647058826, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 1, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1498}, "3": {"facts_total": 66, "facts_retained": 21, "retention_rate": 0.3181818181818182, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.35294117647058826, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 1, "unsupported_identifiers": 0, "distorted_sentences": 1, "message_bytes": 1435}, "4": {"facts_total": 66, "facts_retained": 21, "retention_rate": 0.3181818181818182, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.35294117647058826, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 1, "unsupported_identifiers": 0, "distorted_sentences": 1, "message_bytes": 1435}, "5": {"facts_total": 66, "facts_retained": 21, "retention_rate": 0.3181818181818182, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.35294117647058826, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 1, "unsupported_identifiers": 0, "distorted_sentences": 1, "message_bytes": 1435}, "6": {"facts_total": 66, "facts_retained": 21, "retention_rate": 0.3181818181818182, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.35294117647058826, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 1, "unsupported_identifiers": 0, "distorted_sentences": 1, "message_bytes": 1435}, "7": {"facts_total": 66, "facts_retained": 21, "retention_rate": 0.3181818181818182, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.35294117647058826, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 1, "unsupported_identifiers": 0, "distorted_sentences": 1, "message_bytes": 1435}, "8": {"facts_total": 66, "facts_retained": 21, "retention_rate": 0.3181818181818182, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.35294117647058826, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 1, "unsupported_identifiers": 0, "distorted_sentences": 1, "message_bytes": 1435}}}, "reference_key_id": "keys/synthetic/syn_03024_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "f946ccf31a9f007db2f33763c10d0ace160259ffb2b94381d1019bf2461c8a0a", "output_hash": "9390298f3fb0c5b160498935d79cb139aef28e1c47358b4bbba61862b9c26e59"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "7f198d3f2f095fad3afbaf7f", "attempt_id": "cab9b981316d", "run_group_id": "E4_core:A4|repairable|qwen3b:h8", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:25:49.995401Z", "task_family": "synthetic", "document_id": "syn_03024_2000", "document_hash": "ea10cd11ad08f01c50b0712bc481668857c9d2ddc8e6423968f2a31eeb97ac55", "question_id": "q08_revision", "split": "test", "cluster_id": "syn_03024_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 8, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 2068, "query_tokens": 562, "message_tokens": 463, "message_utf8_bytes": 1435, "cache_tensor_bytes": null, "serialized_payload_bytes": 1435, "side_information_bytes": 0, "fetched_bytes": 8183, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 48962.82533311751, "hops": {"1": 6945.187625009567, "2": 6529.009124962613, "3": 5943.714833003469, "4": 5912.973375059664, "5": 5908.367124968208, "6": 5911.119500058703, "7": 5908.995958045125, "8": 5903.457792010158}, "receiver_prefill_ms": 210.71041701361537, "ttft_ms": 11.958374991081655, "decode_ms": 11.919458978809416}, "total_wall_ms": 49198.026416124776, "ttft_ms": 11.958374991081655, "input_usage": {"lookups": 14}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 66, "facts_retained": 21, "retention_rate": 0.3181818181818182, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.35294117647058826, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 1, "unsupported_identifiers": 0, "distorted_sentences": 1, "message_bytes": 1435}, "ledger_by_hop": {"1": {"facts_total": 66, "facts_retained": 21, "retention_rate": 0.3181818181818182, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.35294117647058826, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1508}, "2": {"facts_total": 66, "facts_retained": 21, "retention_rate": 0.3181818181818182, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.35294117647058826, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 1, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1498}, "3": {"facts_total": 66, "facts_retained": 21, "retention_rate": 0.3181818181818182, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.35294117647058826, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 1, "unsupported_identifiers": 0, "distorted_sentences": 1, "message_bytes": 1435}, "4": {"facts_total": 66, "facts_retained": 21, "retention_rate": 0.3181818181818182, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.35294117647058826, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 1, "unsupported_identifiers": 0, "distorted_sentences": 1, "message_bytes": 1435}, "5": {"facts_total": 66, "facts_retained": 21, "retention_rate": 0.3181818181818182, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.35294117647058826, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 1, "unsupported_identifiers": 0, "distorted_sentences": 1, "message_bytes": 1435}, "6": {"facts_total": 66, "facts_retained": 21, "retention_rate": 0.3181818181818182, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.35294117647058826, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 1, "unsupported_identifiers": 0, "distorted_sentences": 1, "message_bytes": 1435}, "7": {"facts_total": 66, "facts_retained": 21, "retention_rate": 0.3181818181818182, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.35294117647058826, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 1, "unsupported_identifiers": 0, "distorted_sentences": 1, "message_bytes": 1435}, "8": {"facts_total": 66, "facts_retained": 21, "retention_rate": 0.3181818181818182, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.35294117647058826, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 1, "unsupported_identifiers": 0, "distorted_sentences": 1, "message_bytes": 1435}}}, "reference_key_id": "keys/synthetic/syn_03024_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "f946ccf31a9f007db2f33763c10d0ace160259ffb2b94381d1019bf2461c8a0a", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "5d4d9d8d1c6fd7c1ad0c4ed9", "attempt_id": "2f9274536f77", "run_group_id": "E4_core:A4|repairable|qwen3b:h8", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:25:50.247543Z", "task_family": "synthetic", "document_id": "syn_03024_2000", "document_hash": "ea10cd11ad08f01c50b0712bc481668857c9d2ddc8e6423968f2a31eeb97ac55", "question_id": "q09_revision", "split": "test", "cluster_id": "syn_03024_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 8, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 2068, "query_tokens": 553, "message_tokens": 463, "message_utf8_bytes": 1435, "cache_tensor_bytes": null, "serialized_payload_bytes": 1435, "side_information_bytes": 0, "fetched_bytes": 8183, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 48962.82533311751, "hops": {"1": 6945.187625009567, "2": 6529.009124962613, "3": 5943.714833003469, "4": 5912.973375059664, "5": 5908.367124968208, "6": 5911.119500058703, "7": 5908.995958045125, "8": 5903.457792010158}, "receiver_prefill_ms": 210.1943331072107, "ttft_ms": 11.87558297533542, "decode_ms": 11.915415991097689}, "total_wall_ms": 49197.29508308228, "ttft_ms": 11.87558297533542, "input_usage": {"lookups": 14}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 66, "facts_retained": 21, "retention_rate": 0.3181818181818182, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.35294117647058826, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 1, "unsupported_identifiers": 0, "distorted_sentences": 1, "message_bytes": 1435}, "ledger_by_hop": {"1": {"facts_total": 66, "facts_retained": 21, "retention_rate": 0.3181818181818182, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.35294117647058826, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1508}, "2": {"facts_total": 66, "facts_retained": 21, "retention_rate": 0.3181818181818182, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.35294117647058826, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 1, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1498}, "3": {"facts_total": 66, "facts_retained": 21, "retention_rate": 0.3181818181818182, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.35294117647058826, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 1, "unsupported_identifiers": 0, "distorted_sentences": 1, "message_bytes": 1435}, "4": {"facts_total": 66, "facts_retained": 21, "retention_rate": 0.3181818181818182, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.35294117647058826, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 1, "unsupported_identifiers": 0, "distorted_sentences": 1, "message_bytes": 1435}, "5": {"facts_total": 66, "facts_retained": 21, "retention_rate": 0.3181818181818182, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.35294117647058826, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 1, "unsupported_identifiers": 0, "distorted_sentences": 1, "message_bytes": 1435}, "6": {"facts_total": 66, "facts_retained": 21, "retention_rate": 0.3181818181818182, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.35294117647058826, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 1, "unsupported_identifiers": 0, "distorted_sentences": 1, "message_bytes": 1435}, "7": {"facts_total": 66, "facts_retained": 21, "retention_rate": 0.3181818181818182, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.35294117647058826, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 1, "unsupported_identifiers": 0, "distorted_sentences": 1, "message_bytes": 1435}, "8": {"facts_total": 66, "facts_retained": 21, "retention_rate": 0.3181818181818182, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.35294117647058826, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 1, "unsupported_identifiers": 0, "distorted_sentences": 1, "message_bytes": 1435}}}, "reference_key_id": "keys/synthetic/syn_03024_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "f946ccf31a9f007db2f33763c10d0ace160259ffb2b94381d1019bf2461c8a0a", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "df9b49b6ec627d80b0d980e4", "attempt_id": "47faad5bae03", "run_group_id": "E4_core:A4|repairable|qwen3b:h8", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:25:50.497767Z", "task_family": "synthetic", "document_id": "syn_03024_2000", "document_hash": "ea10cd11ad08f01c50b0712bc481668857c9d2ddc8e6423968f2a31eeb97ac55", "question_id": "q10_temporal", "split": "test", "cluster_id": "syn_03024_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 8, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 2068, "query_tokens": 563, "message_tokens": 463, "message_utf8_bytes": 1435, "cache_tensor_bytes": null, "serialized_payload_bytes": 1435, "side_information_bytes": 0, "fetched_bytes": 8183, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 48962.82533311751, "hops": {"1": 6945.187625009567, "2": 6529.009124962613, "3": 5943.714833003469, "4": 5912.973375059664, "5": 5908.367124968208, "6": 5911.119500058703, "7": 5908.995958045125, "8": 5903.457792010158}, "receiver_prefill_ms": 210.931958979927, "ttft_ms": 11.847249930724502, "decode_ms": 71.27329206559807}, "total_wall_ms": 49257.371041108854, "ttft_ms": 11.847249930724502, "input_usage": {"lookups": 14}, "output_usage": {"output_tokens": 6}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "Quilun-16", "evidence_artifact": {"ledger": {"facts_total": 66, "facts_retained": 21, "retention_rate": 0.3181818181818182, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.35294117647058826, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 1, "unsupported_identifiers": 0, "distorted_sentences": 1, "message_bytes": 1435}, "ledger_by_hop": {"1": {"facts_total": 66, "facts_retained": 21, "retention_rate": 0.3181818181818182, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.35294117647058826, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1508}, "2": {"facts_total": 66, "facts_retained": 21, "retention_rate": 0.3181818181818182, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.35294117647058826, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 1, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1498}, "3": {"facts_total": 66, "facts_retained": 21, "retention_rate": 0.3181818181818182, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.35294117647058826, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 1, "unsupported_identifiers": 0, "distorted_sentences": 1, "message_bytes": 1435}, "4": {"facts_total": 66, "facts_retained": 21, "retention_rate": 0.3181818181818182, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.35294117647058826, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 1, "unsupported_identifiers": 0, "distorted_sentences": 1, "message_bytes": 1435}, "5": {"facts_total": 66, "facts_retained": 21, "retention_rate": 0.3181818181818182, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.35294117647058826, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 1, "unsupported_identifiers": 0, "distorted_sentences": 1, "message_bytes": 1435}, "6": {"facts_total": 66, "facts_retained": 21, "retention_rate": 0.3181818181818182, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.35294117647058826, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 1, "unsupported_identifiers": 0, "distorted_sentences": 1, "message_bytes": 1435}, "7": {"facts_total": 66, "facts_retained": 21, "retention_rate": 0.3181818181818182, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.35294117647058826, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 1, "unsupported_identifiers": 0, "distorted_sentences": 1, "message_bytes": 1435}, "8": {"facts_total": 66, "facts_retained": 21, "retention_rate": 0.3181818181818182, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.35294117647058826, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 1, "unsupported_identifiers": 0, "distorted_sentences": 1, "message_bytes": 1435}}}, "reference_key_id": "keys/synthetic/syn_03024_2000.json", "task_scores": {"correct": 0.0, "abstained": 0.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "f946ccf31a9f007db2f33763c10d0ace160259ffb2b94381d1019bf2461c8a0a", "output_hash": "34a7c0e11c93ba75eb4b20d3d3f5303b417856d89a232e8b16fef75c4336afc6"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "0745a7b15a0f41255c0e8386", "attempt_id": "a745c3d814c7", "run_group_id": "E4_core:A4|repairable|qwen3b:h8", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:25:50.807671Z", "task_family": "synthetic", "document_id": "syn_03024_2000", "document_hash": "ea10cd11ad08f01c50b0712bc481668857c9d2ddc8e6423968f2a31eeb97ac55", "question_id": "q11_aggregation", "split": "test", "cluster_id": "syn_03024_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 8, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 2068, "query_tokens": 556, "message_tokens": 463, "message_utf8_bytes": 1435, "cache_tensor_bytes": null, "serialized_payload_bytes": 1435, "side_information_bytes": 0, "fetched_bytes": 8183, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 48962.82533311751, "hops": {"1": 6945.187625009567, "2": 6529.009124962613, "3": 5943.714833003469, "4": 5912.973375059664, "5": 5908.367124968208, "6": 5911.119500058703, "7": 5908.995958045125, "8": 5903.457792010158}, "receiver_prefill_ms": 210.0283340550959, "ttft_ms": 11.9330829475075, "decode_ms": 82.78916694689542}, "total_wall_ms": 49268.15895806067, "ttft_ms": 11.9330829475075, "input_usage": {"lookups": 14}, "output_usage": {"output_tokens": 7}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "3354484", "evidence_artifact": {"ledger": {"facts_total": 66, "facts_retained": 21, "retention_rate": 0.3181818181818182, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.35294117647058826, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 1, "unsupported_identifiers": 0, "distorted_sentences": 1, "message_bytes": 1435}, "ledger_by_hop": {"1": {"facts_total": 66, "facts_retained": 21, "retention_rate": 0.3181818181818182, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.35294117647058826, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1508}, "2": {"facts_total": 66, "facts_retained": 21, "retention_rate": 0.3181818181818182, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.35294117647058826, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 1, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1498}, "3": {"facts_total": 66, "facts_retained": 21, "retention_rate": 0.3181818181818182, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.35294117647058826, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 1, "unsupported_identifiers": 0, "distorted_sentences": 1, "message_bytes": 1435}, "4": {"facts_total": 66, "facts_retained": 21, "retention_rate": 0.3181818181818182, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.35294117647058826, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 1, "unsupported_identifiers": 0, "distorted_sentences": 1, "message_bytes": 1435}, "5": {"facts_total": 66, "facts_retained": 21, "retention_rate": 0.3181818181818182, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.35294117647058826, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 1, "unsupported_identifiers": 0, "distorted_sentences": 1, "message_bytes": 1435}, "6": {"facts_total": 66, "facts_retained": 21, "retention_rate": 0.3181818181818182, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.35294117647058826, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 1, "unsupported_identifiers": 0, "distorted_sentences": 1, "message_bytes": 1435}, "7": {"facts_total": 66, "facts_retained": 21, "retention_rate": 0.3181818181818182, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.35294117647058826, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 1, "unsupported_identifiers": 0, "distorted_sentences": 1, "message_bytes": 1435}, "8": {"facts_total": 66, "facts_retained": 21, "retention_rate": 0.3181818181818182, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.35294117647058826, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 1, "unsupported_identifiers": 0, "distorted_sentences": 1, "message_bytes": 1435}}}, "reference_key_id": "keys/synthetic/syn_03024_2000.json", "task_scores": {"correct": 0.0, "abstained": 0.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "f946ccf31a9f007db2f33763c10d0ace160259ffb2b94381d1019bf2461c8a0a", "output_hash": "5bf31ed3819d2b67352d6e22ea7f16de2ebd589c8281b4393305795b432156e3"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "7c25e5cad91c4ab6c6c925d0", "attempt_id": "b401e3571008", "run_group_id": "E4_core:A4|repairable|qwen3b:h8", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:25:51.128413Z", "task_family": "synthetic", "document_id": "syn_03024_2000", "document_hash": "ea10cd11ad08f01c50b0712bc481668857c9d2ddc8e6423968f2a31eeb97ac55", "question_id": "q12_unit_conversion", "split": "test", "cluster_id": "syn_03024_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 8, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 2068, "query_tokens": 553, "message_tokens": 463, "message_utf8_bytes": 1435, "cache_tensor_bytes": null, "serialized_payload_bytes": 1435, "side_information_bytes": 0, "fetched_bytes": 8183, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 48962.82533311751, "hops": {"1": 6945.187625009567, "2": 6529.009124962613, "3": 5943.714833003469, "4": 5912.973375059664, "5": 5908.367124968208, "6": 5911.119500058703, "7": 5908.995958045125, "8": 5903.457792010158}, "receiver_prefill_ms": 210.57270793244243, "ttft_ms": 11.816957965493202, "decode_ms": 11.756874970160425}, "total_wall_ms": 49197.45633320417, "ttft_ms": 11.816957965493202, "input_usage": {"lookups": 14}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 66, "facts_retained": 21, "retention_rate": 0.3181818181818182, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.35294117647058826, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 1, "unsupported_identifiers": 0, "distorted_sentences": 1, "message_bytes": 1435}, "ledger_by_hop": {"1": {"facts_total": 66, "facts_retained": 21, "retention_rate": 0.3181818181818182, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.35294117647058826, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1508}, "2": {"facts_total": 66, "facts_retained": 21, "retention_rate": 0.3181818181818182, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.35294117647058826, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 1, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1498}, "3": {"facts_total": 66, "facts_retained": 21, "retention_rate": 0.3181818181818182, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.35294117647058826, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 1, "unsupported_identifiers": 0, "distorted_sentences": 1, "message_bytes": 1435}, "4": {"facts_total": 66, "facts_retained": 21, "retention_rate": 0.3181818181818182, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.35294117647058826, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 1, "unsupported_identifiers": 0, "distorted_sentences": 1, "message_bytes": 1435}, "5": {"facts_total": 66, "facts_retained": 21, "retention_rate": 0.3181818181818182, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.35294117647058826, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 1, "unsupported_identifiers": 0, "distorted_sentences": 1, "message_bytes": 1435}, "6": {"facts_total": 66, "facts_retained": 21, "retention_rate": 0.3181818181818182, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.35294117647058826, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 1, "unsupported_identifiers": 0, "distorted_sentences": 1, "message_bytes": 1435}, "7": {"facts_total": 66, "facts_retained": 21, "retention_rate": 0.3181818181818182, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.35294117647058826, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 1, "unsupported_identifiers": 0, "distorted_sentences": 1, "message_bytes": 1435}, "8": {"facts_total": 66, "facts_retained": 21, "retention_rate": 0.3181818181818182, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.35294117647058826, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 1, "unsupported_identifiers": 0, "distorted_sentences": 1, "message_bytes": 1435}}}, "reference_key_id": "keys/synthetic/syn_03024_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "f946ccf31a9f007db2f33763c10d0ace160259ffb2b94381d1019bf2461c8a0a", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "20b59930dcbddc00065ce674", "attempt_id": "3a1820a0b83c", "run_group_id": "E4_core:A4|repairable|qwen3b:h8", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:25:51.378907Z", "task_family": "synthetic", "document_id": "syn_03024_2000", "document_hash": "ea10cd11ad08f01c50b0712bc481668857c9d2ddc8e6423968f2a31eeb97ac55", "question_id": "q13_unanswerable", "split": "test", "cluster_id": "syn_03024_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 8, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 2068, "query_tokens": 557, "message_tokens": 463, "message_utf8_bytes": 1435, "cache_tensor_bytes": null, "serialized_payload_bytes": 1435, "side_information_bytes": 0, "fetched_bytes": 8183, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 48962.82533311751, "hops": {"1": 6945.187625009567, "2": 6529.009124962613, "3": 5943.714833003469, "4": 5912.973375059664, "5": 5908.367124968208, "6": 5911.119500058703, "7": 5908.995958045125, "8": 5903.457792010158}, "receiver_prefill_ms": 210.54037509020418, "ttft_ms": 11.821750085800886, "decode_ms": 11.873625102452934}, "total_wall_ms": 49197.53104215488, "ttft_ms": 11.821750085800886, "input_usage": {"lookups": 14}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 66, "facts_retained": 21, "retention_rate": 0.3181818181818182, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.35294117647058826, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 1, "unsupported_identifiers": 0, "distorted_sentences": 1, "message_bytes": 1435}, "ledger_by_hop": {"1": {"facts_total": 66, "facts_retained": 21, "retention_rate": 0.3181818181818182, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.35294117647058826, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1508}, "2": {"facts_total": 66, "facts_retained": 21, "retention_rate": 0.3181818181818182, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.35294117647058826, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 1, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1498}, "3": {"facts_total": 66, "facts_retained": 21, "retention_rate": 0.3181818181818182, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.35294117647058826, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 1, "unsupported_identifiers": 0, "distorted_sentences": 1, "message_bytes": 1435}, "4": {"facts_total": 66, "facts_retained": 21, "retention_rate": 0.3181818181818182, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.35294117647058826, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 1, "unsupported_identifiers": 0, "distorted_sentences": 1, "message_bytes": 1435}, "5": {"facts_total": 66, "facts_retained": 21, "retention_rate": 0.3181818181818182, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.35294117647058826, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 1, "unsupported_identifiers": 0, "distorted_sentences": 1, "message_bytes": 1435}, "6": {"facts_total": 66, "facts_retained": 21, "retention_rate": 0.3181818181818182, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.35294117647058826, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 1, "unsupported_identifiers": 0, "distorted_sentences": 1, "message_bytes": 1435}, "7": {"facts_total": 66, "facts_retained": 21, "retention_rate": 0.3181818181818182, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.35294117647058826, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 1, "unsupported_identifiers": 0, "distorted_sentences": 1, "message_bytes": 1435}, "8": {"facts_total": 66, "facts_retained": 21, "retention_rate": 0.3181818181818182, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.35294117647058826, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 1, "unsupported_identifiers": 0, "distorted_sentences": 1, "message_bytes": 1435}}}, "reference_key_id": "keys/synthetic/syn_03024_2000.json", "task_scores": {"correct": 1.0, "abstained": 1.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "f946ccf31a9f007db2f33763c10d0ace160259ffb2b94381d1019bf2461c8a0a", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "c558c630043a4780c8416796", "attempt_id": "8b2e9c1fd8ce", "run_group_id": "E4_core:A4|repairable|qwen3b:h8", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:25:51.628891Z", "task_family": "synthetic", "document_id": "syn_03024_2000", "document_hash": "ea10cd11ad08f01c50b0712bc481668857c9d2ddc8e6423968f2a31eeb97ac55", "question_id": "q14_unanswerable", "split": "test", "cluster_id": "syn_03024_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 8, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 2068, "query_tokens": 558, "message_tokens": 463, "message_utf8_bytes": 1435, "cache_tensor_bytes": null, "serialized_payload_bytes": 1435, "side_information_bytes": 0, "fetched_bytes": 8183, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 48962.82533311751, "hops": {"1": 6945.187625009567, "2": 6529.009124962613, "3": 5943.714833003469, "4": 5912.973375059664, "5": 5908.367124968208, "6": 5911.119500058703, "7": 5908.995958045125, "8": 5903.457792010158}, "receiver_prefill_ms": 210.47225000802428, "ttft_ms": 11.869207955896854, "decode_ms": 11.876833974383771}, "total_wall_ms": 49197.51333317254, "ttft_ms": 11.869207955896854, "input_usage": {"lookups": 14}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 66, "facts_retained": 21, "retention_rate": 0.3181818181818182, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.35294117647058826, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 1, "unsupported_identifiers": 0, "distorted_sentences": 1, "message_bytes": 1435}, "ledger_by_hop": {"1": {"facts_total": 66, "facts_retained": 21, "retention_rate": 0.3181818181818182, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.35294117647058826, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1508}, "2": {"facts_total": 66, "facts_retained": 21, "retention_rate": 0.3181818181818182, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.35294117647058826, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 1, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1498}, "3": {"facts_total": 66, "facts_retained": 21, "retention_rate": 0.3181818181818182, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.35294117647058826, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 1, "unsupported_identifiers": 0, "distorted_sentences": 1, "message_bytes": 1435}, "4": {"facts_total": 66, "facts_retained": 21, "retention_rate": 0.3181818181818182, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.35294117647058826, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 1, "unsupported_identifiers": 0, "distorted_sentences": 1, "message_bytes": 1435}, "5": {"facts_total": 66, "facts_retained": 21, "retention_rate": 0.3181818181818182, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.35294117647058826, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 1, "unsupported_identifiers": 0, "distorted_sentences": 1, "message_bytes": 1435}, "6": {"facts_total": 66, "facts_retained": 21, "retention_rate": 0.3181818181818182, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.35294117647058826, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 1, "unsupported_identifiers": 0, "distorted_sentences": 1, "message_bytes": 1435}, "7": {"facts_total": 66, "facts_retained": 21, "retention_rate": 0.3181818181818182, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.35294117647058826, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 1, "unsupported_identifiers": 0, "distorted_sentences": 1, "message_bytes": 1435}, "8": {"facts_total": 66, "facts_retained": 21, "retention_rate": 0.3181818181818182, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.35294117647058826, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 1, "unsupported_identifiers": 0, "distorted_sentences": 1, "message_bytes": 1435}}}, "reference_key_id": "keys/synthetic/syn_03024_2000.json", "task_scores": {"correct": 1.0, "abstained": 1.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "f946ccf31a9f007db2f33763c10d0ace160259ffb2b94381d1019bf2461c8a0a", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "03c853b60429fefd8c52404b", "attempt_id": "f3083fa0756a", "run_group_id": "E4_core:A5|repairable|qwen3b:h1", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:25:58.819090Z", "task_family": "synthetic", "document_id": "syn_03024_2000", "document_hash": "ea10cd11ad08f01c50b0712bc481668857c9d2ddc8e6423968f2a31eeb97ac55", "question_id": "q00_single_fact", "split": "test", "cluster_id": "syn_03024_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 1, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 2068, "query_tokens": 440, "message_tokens": 357, "message_utf8_bytes": 818, "cache_tensor_bytes": null, "serialized_payload_bytes": 818, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 6938.470332999714, "hops": {"1": 6938.470332999714}, "receiver_prefill_ms": 164.517541998066, "ttft_ms": 11.817041086032987, "decode_ms": 11.78620895370841}, "total_wall_ms": 7127.00237496756, "ttft_ms": 11.817041086032987, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 66, "facts_retained": 6, "retention_rate": 0.09090909090909091, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.17647058823529413, "status": 0.0, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 754}, "ledger_by_hop": {"1": {"facts_total": 66, "facts_retained": 6, "retention_rate": 0.09090909090909091, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.17647058823529413, "status": 0.0, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 754}}}, "reference_key_id": "keys/synthetic/syn_03024_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "3c07df4f32c3dffe23905714a167378f1096d7812f8c55ff7cf92b920f04befa", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "659546d3c6c8e91434e91120", "attempt_id": "8693fef81a0f", "run_group_id": "E4_core:A5|repairable|qwen3b:h1", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:25:59.024065Z", "task_family": "synthetic", "document_id": "syn_03024_2000", "document_hash": "ea10cd11ad08f01c50b0712bc481668857c9d2ddc8e6423968f2a31eeb97ac55", "question_id": "q01_single_fact", "split": "test", "cluster_id": "syn_03024_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 1, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 2068, "query_tokens": 445, "message_tokens": 357, "message_utf8_bytes": 818, "cache_tensor_bytes": null, "serialized_payload_bytes": 818, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 6938.470332999714, "hops": {"1": 6938.470332999714}, "receiver_prefill_ms": 165.25012499187142, "ttft_ms": 11.77537499461323, "decode_ms": 129.4018329354003}, "total_wall_ms": 7245.301707996987, "ttft_ms": 11.77537499461323, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 11}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "Lot 568-G in Mont-Laurier", "evidence_artifact": {"ledger": {"facts_total": 66, "facts_retained": 6, "retention_rate": 0.09090909090909091, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.17647058823529413, "status": 0.0, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 754}, "ledger_by_hop": {"1": {"facts_total": 66, "facts_retained": 6, "retention_rate": 0.09090909090909091, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.17647058823529413, "status": 0.0, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 754}}}, "reference_key_id": "keys/synthetic/syn_03024_2000.json", "task_scores": {"correct": 0.0, "abstained": 0.0, "f1": 0.5}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "3c07df4f32c3dffe23905714a167378f1096d7812f8c55ff7cf92b920f04befa", "output_hash": "98e007faa5a10742ad74f2778a64e3ea0d6b72bec1c0392200d475073d7bf082"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "bc2dfe32753ea89d44e1cfdb", "attempt_id": "decac8927b3c", "run_group_id": "E4_core:A5|repairable|qwen3b:h1", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:25:59.345970Z", "task_family": "synthetic", "document_id": "syn_03024_2000", "document_hash": "ea10cd11ad08f01c50b0712bc481668857c9d2ddc8e6423968f2a31eeb97ac55", "question_id": "q02_exact_copy", "split": "test", "cluster_id": "syn_03024_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 1, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 2068, "query_tokens": 445, "message_tokens": 357, "message_utf8_bytes": 818, "cache_tensor_bytes": null, "serialized_payload_bytes": 818, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 6938.470332999714, "hops": {"1": 6938.470332999714}, "receiver_prefill_ms": 163.9936660649255, "ttft_ms": 12.05391704570502, "decode_ms": 11.81683293543756}, "total_wall_ms": 7126.827500062063, "ttft_ms": 12.05391704570502, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 66, "facts_retained": 6, "retention_rate": 0.09090909090909091, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.17647058823529413, "status": 0.0, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 754}, "ledger_by_hop": {"1": {"facts_total": 66, "facts_retained": 6, "retention_rate": 0.09090909090909091, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.17647058823529413, "status": 0.0, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 754}}}, "reference_key_id": "keys/synthetic/syn_03024_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "3c07df4f32c3dffe23905714a167378f1096d7812f8c55ff7cf92b920f04befa", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "004f79a48f8bb5ef4b629de6", "attempt_id": "d3f7582f8d23", "run_group_id": "E4_core:A5|repairable|qwen3b:h1", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:25:59.549814Z", "task_family": "synthetic", "document_id": "syn_03024_2000", "document_hash": "ea10cd11ad08f01c50b0712bc481668857c9d2ddc8e6423968f2a31eeb97ac55", "question_id": "q03_exact_copy", "split": "test", "cluster_id": "syn_03024_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 1, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 2068, "query_tokens": 440, "message_tokens": 357, "message_utf8_bytes": 818, "cache_tensor_bytes": null, "serialized_payload_bytes": 818, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 6938.470332999714, "hops": {"1": 6938.470332999714}, "receiver_prefill_ms": 164.43112504202873, "ttft_ms": 11.818542028777301, "decode_ms": 11.775457998737693}, "total_wall_ms": 7126.878416980617, "ttft_ms": 11.818542028777301, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 66, "facts_retained": 6, "retention_rate": 0.09090909090909091, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.17647058823529413, "status": 0.0, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 754}, "ledger_by_hop": {"1": {"facts_total": 66, "facts_retained": 6, "retention_rate": 0.09090909090909091, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.17647058823529413, "status": 0.0, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 754}}}, "reference_key_id": "keys/synthetic/syn_03024_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "3c07df4f32c3dffe23905714a167378f1096d7812f8c55ff7cf92b920f04befa", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "ef81e1358ba9280b1dea01fd", "attempt_id": "8f1be55433c9", "run_group_id": "E4_core:A5|repairable|qwen3b:h1", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:25:59.753823Z", "task_family": "synthetic", "document_id": "syn_03024_2000", "document_hash": "ea10cd11ad08f01c50b0712bc481668857c9d2ddc8e6423968f2a31eeb97ac55", "question_id": "q04_multi_hop", "split": "test", "cluster_id": "syn_03024_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 1, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 2068, "query_tokens": 446, "message_tokens": 357, "message_utf8_bytes": 818, "cache_tensor_bytes": null, "serialized_payload_bytes": 818, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 6938.470332999714, "hops": {"1": 6938.470332999714}, "receiver_prefill_ms": 165.45770806260407, "ttft_ms": 11.775291990488768, "decode_ms": 11.769125005230308}, "total_wall_ms": 7127.935083000921, "ttft_ms": 11.775291990488768, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 66, "facts_retained": 6, "retention_rate": 0.09090909090909091, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.17647058823529413, "status": 0.0, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 754}, "ledger_by_hop": {"1": {"facts_total": 66, "facts_retained": 6, "retention_rate": 0.09090909090909091, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.17647058823529413, "status": 0.0, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 754}}}, "reference_key_id": "keys/synthetic/syn_03024_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "3c07df4f32c3dffe23905714a167378f1096d7812f8c55ff7cf92b920f04befa", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "be1e71ee029391c89409175a", "attempt_id": "3253db3564f0", "run_group_id": "E4_core:A5|repairable|qwen3b:h1", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:25:59.958096Z", "task_family": "synthetic", "document_id": "syn_03024_2000", "document_hash": "ea10cd11ad08f01c50b0712bc481668857c9d2ddc8e6423968f2a31eeb97ac55", "question_id": "q05_multi_hop", "split": "test", "cluster_id": "syn_03024_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 1, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 2068, "query_tokens": 450, "message_tokens": 357, "message_utf8_bytes": 818, "cache_tensor_bytes": null, "serialized_payload_bytes": 818, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 6938.470332999714, "hops": {"1": 6938.470332999714}, "receiver_prefill_ms": 184.9020830122754, "ttft_ms": 11.921457946300507, "decode_ms": 11.848208028823137}, "total_wall_ms": 7147.635500063188, "ttft_ms": 11.921457946300507, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 66, "facts_retained": 6, "retention_rate": 0.09090909090909091, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.17647058823529413, "status": 0.0, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 754}, "ledger_by_hop": {"1": {"facts_total": 66, "facts_retained": 6, "retention_rate": 0.09090909090909091, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.17647058823529413, "status": 0.0, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 754}}}, "reference_key_id": "keys/synthetic/syn_03024_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "3c07df4f32c3dffe23905714a167378f1096d7812f8c55ff7cf92b920f04befa", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "67a19ff324af232bb43044d8", "attempt_id": "717ea252564c", "run_group_id": "E4_core:A5|repairable|qwen3b:h1", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:26:00.183591Z", "task_family": "synthetic", "document_id": "syn_03024_2000", "document_hash": "ea10cd11ad08f01c50b0712bc481668857c9d2ddc8e6423968f2a31eeb97ac55", "question_id": "q06_negation", "split": "test", "cluster_id": "syn_03024_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 1, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 2068, "query_tokens": 439, "message_tokens": 357, "message_utf8_bytes": 818, "cache_tensor_bytes": null, "serialized_payload_bytes": 818, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 6938.470332999714, "hops": {"1": 6938.470332999714}, "receiver_prefill_ms": 164.2949590459466, "ttft_ms": 11.728540994226933, "decode_ms": 11.765125091187656}, "total_wall_ms": 7126.6367910429835, "ttft_ms": 11.728540994226933, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "no", "evidence_artifact": {"ledger": {"facts_total": 66, "facts_retained": 6, "retention_rate": 0.09090909090909091, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.17647058823529413, "status": 0.0, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 754}, "ledger_by_hop": {"1": {"facts_total": 66, "facts_retained": 6, "retention_rate": 0.09090909090909091, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.17647058823529413, "status": 0.0, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 754}}}, "reference_key_id": "keys/synthetic/syn_03024_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "3c07df4f32c3dffe23905714a167378f1096d7812f8c55ff7cf92b920f04befa", "output_hash": "9390298f3fb0c5b160498935d79cb139aef28e1c47358b4bbba61862b9c26e59"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "0b75b646c7ae9de3e5b1d391", "attempt_id": "1e949e438fa2", "run_group_id": "E4_core:A5|repairable|qwen3b:h1", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:26:00.387686Z", "task_family": "synthetic", "document_id": "syn_03024_2000", "document_hash": "ea10cd11ad08f01c50b0712bc481668857c9d2ddc8e6423968f2a31eeb97ac55", "question_id": "q07_negation", "split": "test", "cluster_id": "syn_03024_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 1, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 2068, "query_tokens": 441, "message_tokens": 357, "message_utf8_bytes": 818, "cache_tensor_bytes": null, "serialized_payload_bytes": 818, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 6938.470332999714, "hops": {"1": 6938.470332999714}, "receiver_prefill_ms": 164.01404200587422, "ttft_ms": 11.757125030271709, "decode_ms": 12.009124970063567}, "total_wall_ms": 7126.621874049306, "ttft_ms": 11.757125030271709, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "no", "evidence_artifact": {"ledger": {"facts_total": 66, "facts_retained": 6, "retention_rate": 0.09090909090909091, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.17647058823529413, "status": 0.0, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 754}, "ledger_by_hop": {"1": {"facts_total": 66, "facts_retained": 6, "retention_rate": 0.09090909090909091, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.17647058823529413, "status": 0.0, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 754}}}, "reference_key_id": "keys/synthetic/syn_03024_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "3c07df4f32c3dffe23905714a167378f1096d7812f8c55ff7cf92b920f04befa", "output_hash": "9390298f3fb0c5b160498935d79cb139aef28e1c47358b4bbba61862b9c26e59"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "9764778f1bc324f5bda62993", "attempt_id": "b6d39c828995", "run_group_id": "E4_core:A5|repairable|qwen3b:h1", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:26:00.590799Z", "task_family": "synthetic", "document_id": "syn_03024_2000", "document_hash": "ea10cd11ad08f01c50b0712bc481668857c9d2ddc8e6423968f2a31eeb97ac55", "question_id": "q08_revision", "split": "test", "cluster_id": "syn_03024_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 1, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 2068, "query_tokens": 457, "message_tokens": 357, "message_utf8_bytes": 818, "cache_tensor_bytes": null, "serialized_payload_bytes": 818, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 6938.470332999714, "hops": {"1": 6938.470332999714}, "receiver_prefill_ms": 185.68837502971292, "ttft_ms": 11.77166705019772, "decode_ms": 11.99179096147418}, "total_wall_ms": 7148.304958012886, "ttft_ms": 11.77166705019772, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 66, "facts_retained": 6, "retention_rate": 0.09090909090909091, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.17647058823529413, "status": 0.0, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 754}, "ledger_by_hop": {"1": {"facts_total": 66, "facts_retained": 6, "retention_rate": 0.09090909090909091, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.17647058823529413, "status": 0.0, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 754}}}, "reference_key_id": "keys/synthetic/syn_03024_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "3c07df4f32c3dffe23905714a167378f1096d7812f8c55ff7cf92b920f04befa", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "a97c4ba1dcaebb932af7be3a", "attempt_id": "c40ddb710540", "run_group_id": "E4_core:A5|repairable|qwen3b:h1", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:26:00.815836Z", "task_family": "synthetic", "document_id": "syn_03024_2000", "document_hash": "ea10cd11ad08f01c50b0712bc481668857c9d2ddc8e6423968f2a31eeb97ac55", "question_id": "q09_revision", "split": "test", "cluster_id": "syn_03024_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 1, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 2068, "query_tokens": 448, "message_tokens": 357, "message_utf8_bytes": 818, "cache_tensor_bytes": null, "serialized_payload_bytes": 818, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 6938.470332999714, "hops": {"1": 6938.470332999714}, "receiver_prefill_ms": 165.8008329104632, "ttft_ms": 11.785916984081268, "decode_ms": 82.51291699707508}, "total_wall_ms": 7198.933623963967, "ttft_ms": 11.785916984081268, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 7}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "9134435", "evidence_artifact": {"ledger": {"facts_total": 66, "facts_retained": 6, "retention_rate": 0.09090909090909091, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.17647058823529413, "status": 0.0, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 754}, "ledger_by_hop": {"1": {"facts_total": 66, "facts_retained": 6, "retention_rate": 0.09090909090909091, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.17647058823529413, "status": 0.0, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 754}}}, "reference_key_id": "keys/synthetic/syn_03024_2000.json", "task_scores": {"correct": 0.0, "abstained": 0.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "3c07df4f32c3dffe23905714a167378f1096d7812f8c55ff7cf92b920f04befa", "output_hash": "f7384892cb4f9a38f6115f91da5fec3f0d7e24e47877737b0f34e2b4a6af7d5a"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "fffda64207b098973f304634", "attempt_id": "67c02ab41338", "run_group_id": "E4_core:A5|repairable|qwen3b:h1", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:26:01.091262Z", "task_family": "synthetic", "document_id": "syn_03024_2000", "document_hash": "ea10cd11ad08f01c50b0712bc481668857c9d2ddc8e6423968f2a31eeb97ac55", "question_id": "q10_temporal", "split": "test", "cluster_id": "syn_03024_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 1, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 2068, "query_tokens": 458, "message_tokens": 357, "message_utf8_bytes": 818, "cache_tensor_bytes": null, "serialized_payload_bytes": 818, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 6938.470332999714, "hops": {"1": 6938.470332999714}, "receiver_prefill_ms": 185.10133295785636, "ttft_ms": 11.834957986138761, "decode_ms": 11.790541000664234}, "total_wall_ms": 7147.569333086722, "ttft_ms": 11.834957986138761, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 66, "facts_retained": 6, "retention_rate": 0.09090909090909091, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.17647058823529413, "status": 0.0, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 754}, "ledger_by_hop": {"1": {"facts_total": 66, "facts_retained": 6, "retention_rate": 0.09090909090909091, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.17647058823529413, "status": 0.0, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 754}}}, "reference_key_id": "keys/synthetic/syn_03024_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "3c07df4f32c3dffe23905714a167378f1096d7812f8c55ff7cf92b920f04befa", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "e91412e5d9c632cab392f9a9", "attempt_id": "29e2b1f997f6", "run_group_id": "E4_core:A5|repairable|qwen3b:h1", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:26:01.315207Z", "task_family": "synthetic", "document_id": "syn_03024_2000", "document_hash": "ea10cd11ad08f01c50b0712bc481668857c9d2ddc8e6423968f2a31eeb97ac55", "question_id": "q11_aggregation", "split": "test", "cluster_id": "syn_03024_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 1, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 2068, "query_tokens": 451, "message_tokens": 357, "message_utf8_bytes": 818, "cache_tensor_bytes": null, "serialized_payload_bytes": 818, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 6938.470332999714, "hops": {"1": 6938.470332999714}, "receiver_prefill_ms": 185.14366704039276, "ttft_ms": 11.891166912391782, "decode_ms": 82.28991692885756}, "total_wall_ms": 7218.1846658932045, "ttft_ms": 11.891166912391782, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 7}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "8066734", "evidence_artifact": {"ledger": {"facts_total": 66, "facts_retained": 6, "retention_rate": 0.09090909090909091, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.17647058823529413, "status": 0.0, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 754}, "ledger_by_hop": {"1": {"facts_total": 66, "facts_retained": 6, "retention_rate": 0.09090909090909091, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.17647058823529413, "status": 0.0, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 754}}}, "reference_key_id": "keys/synthetic/syn_03024_2000.json", "task_scores": {"correct": 0.0, "abstained": 0.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "3c07df4f32c3dffe23905714a167378f1096d7812f8c55ff7cf92b920f04befa", "output_hash": "167c75fdd6a41fca64228797ab8f3a1a3f994475246df0bffb2e060d5d2e4ff0"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "febf4ecb58a79ecb171e348b", "attempt_id": "ee03eea820ea", "run_group_id": "E4_core:A5|repairable|qwen3b:h1", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:26:01.610181Z", "task_family": "synthetic", "document_id": "syn_03024_2000", "document_hash": "ea10cd11ad08f01c50b0712bc481668857c9d2ddc8e6423968f2a31eeb97ac55", "question_id": "q12_unit_conversion", "split": "test", "cluster_id": "syn_03024_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 1, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 2068, "query_tokens": 448, "message_tokens": 357, "message_utf8_bytes": 818, "cache_tensor_bytes": null, "serialized_payload_bytes": 818, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 6938.470332999714, "hops": {"1": 6938.470332999714}, "receiver_prefill_ms": 165.15287500806153, "ttft_ms": 11.796292033977807, "decode_ms": 11.700666975229979}, "total_wall_ms": 7127.4879990378395, "ttft_ms": 11.796292033977807, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 66, "facts_retained": 6, "retention_rate": 0.09090909090909091, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.17647058823529413, "status": 0.0, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 754}, "ledger_by_hop": {"1": {"facts_total": 66, "facts_retained": 6, "retention_rate": 0.09090909090909091, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.17647058823529413, "status": 0.0, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 754}}}, "reference_key_id": "keys/synthetic/syn_03024_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "3c07df4f32c3dffe23905714a167378f1096d7812f8c55ff7cf92b920f04befa", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "350140ff476429db88de04c7", "attempt_id": "23321660e55c", "run_group_id": "E4_core:A5|repairable|qwen3b:h1", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:26:01.814094Z", "task_family": "synthetic", "document_id": "syn_03024_2000", "document_hash": "ea10cd11ad08f01c50b0712bc481668857c9d2ddc8e6423968f2a31eeb97ac55", "question_id": "q13_unanswerable", "split": "test", "cluster_id": "syn_03024_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 1, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 2068, "query_tokens": 452, "message_tokens": 357, "message_utf8_bytes": 818, "cache_tensor_bytes": null, "serialized_payload_bytes": 818, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 6938.470332999714, "hops": {"1": 6938.470332999714}, "receiver_prefill_ms": 184.93266601581126, "ttft_ms": 11.85204100329429, "decode_ms": 11.755957966670394}, "total_wall_ms": 7147.374207968824, "ttft_ms": 11.85204100329429, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 66, "facts_retained": 6, "retention_rate": 0.09090909090909091, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.17647058823529413, "status": 0.0, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 754}, "ledger_by_hop": {"1": {"facts_total": 66, "facts_retained": 6, "retention_rate": 0.09090909090909091, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.17647058823529413, "status": 0.0, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 754}}}, "reference_key_id": "keys/synthetic/syn_03024_2000.json", "task_scores": {"correct": 1.0, "abstained": 1.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "3c07df4f32c3dffe23905714a167378f1096d7812f8c55ff7cf92b920f04befa", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "ac0be4f793cfcc69f6eb9d91", "attempt_id": "560c357d500b", "run_group_id": "E4_core:A5|repairable|qwen3b:h1", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:26:02.037931Z", "task_family": "synthetic", "document_id": "syn_03024_2000", "document_hash": "ea10cd11ad08f01c50b0712bc481668857c9d2ddc8e6423968f2a31eeb97ac55", "question_id": "q14_unanswerable", "split": "test", "cluster_id": "syn_03024_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 1, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 2068, "query_tokens": 453, "message_tokens": 357, "message_utf8_bytes": 818, "cache_tensor_bytes": null, "serialized_payload_bytes": 818, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 6938.470332999714, "hops": {"1": 6938.470332999714}, "receiver_prefill_ms": 185.35120808519423, "ttft_ms": 11.818500002846122, "decode_ms": 11.868750094436109}, "total_wall_ms": 7147.877166979015, "ttft_ms": 11.818500002846122, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 66, "facts_retained": 6, "retention_rate": 0.09090909090909091, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.17647058823529413, "status": 0.0, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 754}, "ledger_by_hop": {"1": {"facts_total": 66, "facts_retained": 6, "retention_rate": 0.09090909090909091, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.17647058823529413, "status": 0.0, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 754}}}, "reference_key_id": "keys/synthetic/syn_03024_2000.json", "task_scores": {"correct": 1.0, "abstained": 1.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "3c07df4f32c3dffe23905714a167378f1096d7812f8c55ff7cf92b920f04befa", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "424b07858953312ca1a64299", "attempt_id": "9ee8b429de25", "run_group_id": "E4_core:A5|repairable|qwen3b:h2", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:26:06.965140Z", "task_family": "synthetic", "document_id": "syn_03024_2000", "document_hash": "ea10cd11ad08f01c50b0712bc481668857c9d2ddc8e6423968f2a31eeb97ac55", "question_id": "q00_single_fact", "split": "test", "cluster_id": "syn_03024_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 2, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 2068, "query_tokens": 449, "message_tokens": 367, "message_utf8_bytes": 847, "cache_tensor_bytes": null, "serialized_payload_bytes": 847, "side_information_bytes": 0, "fetched_bytes": 582, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 11639.434623997658, "hops": {"1": 6938.470332999714, "2": 4700.964290997945}, "receiver_prefill_ms": 164.64054200332612, "ttft_ms": 11.968332924880087, "decode_ms": 117.66037496272475}, "total_wall_ms": 11934.194498928264, "ttft_ms": 11.968332924880087, "input_usage": {"lookups": 1}, "output_usage": {"output_tokens": 10}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "BU-55G-3254", "evidence_artifact": {"ledger": {"facts_total": 66, "facts_retained": 8, "retention_rate": 0.12121212121212122, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.17647058823529413, "status": 0.16666666666666666, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 783}, "ledger_by_hop": {"1": {"facts_total": 66, "facts_retained": 6, "retention_rate": 0.09090909090909091, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.17647058823529413, "status": 0.0, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 754}, "2": {"facts_total": 66, "facts_retained": 8, "retention_rate": 0.12121212121212122, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.17647058823529413, "status": 0.16666666666666666, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 783}}}, "reference_key_id": "keys/synthetic/syn_03024_2000.json", "task_scores": {"correct": 0.0, "abstained": 0.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "3cb4552bd39acfb9b1ca04c5d04cccae183051dc06ba450042dd9b360768202c", "output_hash": "98227e32c2121a7856e6d2b32b74598ce6f4183c2ec3eec7ce4b9b0b3f66b98b"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "90531576983c730896a68ce8", "attempt_id": "70f45377c44e", "run_group_id": "E4_core:A5|repairable|qwen3b:h2", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:26:07.277076Z", "task_family": "synthetic", "document_id": "syn_03024_2000", "document_hash": "ea10cd11ad08f01c50b0712bc481668857c9d2ddc8e6423968f2a31eeb97ac55", "question_id": "q01_single_fact", "split": "test", "cluster_id": "syn_03024_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 2, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 2068, "query_tokens": 454, "message_tokens": 367, "message_utf8_bytes": 847, "cache_tensor_bytes": null, "serialized_payload_bytes": 847, "side_information_bytes": 0, "fetched_bytes": 582, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 11639.434623997658, "hops": {"1": 6938.470332999714, "2": 4700.964290997945}, "receiver_prefill_ms": 185.25383400265127, "ttft_ms": 11.858458979986608, "decode_ms": 129.4203749857843}, "total_wall_ms": 11966.401290032081, "ttft_ms": 11.858458979986608, "input_usage": {"lookups": 1}, "output_usage": {"output_tokens": 11}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "Lot 568-G in Mont-Laurier", "evidence_artifact": {"ledger": {"facts_total": 66, "facts_retained": 8, "retention_rate": 0.12121212121212122, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.17647058823529413, "status": 0.16666666666666666, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 783}, "ledger_by_hop": {"1": {"facts_total": 66, "facts_retained": 6, "retention_rate": 0.09090909090909091, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.17647058823529413, "status": 0.0, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 754}, "2": {"facts_total": 66, "facts_retained": 8, "retention_rate": 0.12121212121212122, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.17647058823529413, "status": 0.16666666666666666, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 783}}}, "reference_key_id": "keys/synthetic/syn_03024_2000.json", "task_scores": {"correct": 0.0, "abstained": 0.0, "f1": 0.5}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "3cb4552bd39acfb9b1ca04c5d04cccae183051dc06ba450042dd9b360768202c", "output_hash": "98e007faa5a10742ad74f2778a64e3ea0d6b72bec1c0392200d475073d7bf082"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "751f2ecaa99dfeda54c9e58c", "attempt_id": "ce660dedb179", "run_group_id": "E4_core:A5|repairable|qwen3b:h2", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:26:07.619905Z", "task_family": "synthetic", "document_id": "syn_03024_2000", "document_hash": "ea10cd11ad08f01c50b0712bc481668857c9d2ddc8e6423968f2a31eeb97ac55", "question_id": "q02_exact_copy", "split": "test", "cluster_id": "syn_03024_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 2, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 2068, "query_tokens": 454, "message_tokens": 367, "message_utf8_bytes": 847, "cache_tensor_bytes": null, "serialized_payload_bytes": 847, "side_information_bytes": 0, "fetched_bytes": 582, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 11639.434623997658, "hops": {"1": 6938.470332999714, "2": 4700.964290997945}, "receiver_prefill_ms": 185.57770806364715, "ttft_ms": 11.873041046783328, "decode_ms": 11.823041015304625}, "total_wall_ms": 11849.075665930286, "ttft_ms": 11.873041046783328, "input_usage": {"lookups": 1}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 66, "facts_retained": 8, "retention_rate": 0.12121212121212122, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.17647058823529413, "status": 0.16666666666666666, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 783}, "ledger_by_hop": {"1": {"facts_total": 66, "facts_retained": 6, "retention_rate": 0.09090909090909091, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.17647058823529413, "status": 0.0, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 754}, "2": {"facts_total": 66, "facts_retained": 8, "retention_rate": 0.12121212121212122, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.17647058823529413, "status": 0.16666666666666666, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 783}}}, "reference_key_id": "keys/synthetic/syn_03024_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "3cb4552bd39acfb9b1ca04c5d04cccae183051dc06ba450042dd9b360768202c", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "ebaec1a7648c85c400d6e572", "attempt_id": "40f1875644a1", "run_group_id": "E4_core:A5|repairable|qwen3b:h2", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:26:07.844850Z", "task_family": "synthetic", "document_id": "syn_03024_2000", "document_hash": "ea10cd11ad08f01c50b0712bc481668857c9d2ddc8e6423968f2a31eeb97ac55", "question_id": "q03_exact_copy", "split": "test", "cluster_id": "syn_03024_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 2, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 2068, "query_tokens": 449, "message_tokens": 367, "message_utf8_bytes": 847, "cache_tensor_bytes": null, "serialized_payload_bytes": 847, "side_information_bytes": 0, "fetched_bytes": 582, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 11639.434623997658, "hops": {"1": 6938.470332999714, "2": 4700.964290997945}, "receiver_prefill_ms": 164.29262491874397, "ttft_ms": 11.786874965764582, "decode_ms": 11.796750011853874}, "total_wall_ms": 11827.676290995441, "ttft_ms": 11.786874965764582, "input_usage": {"lookups": 1}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 66, "facts_retained": 8, "retention_rate": 0.12121212121212122, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.17647058823529413, "status": 0.16666666666666666, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 783}, "ledger_by_hop": {"1": {"facts_total": 66, "facts_retained": 6, "retention_rate": 0.09090909090909091, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.17647058823529413, "status": 0.0, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 754}, "2": {"facts_total": 66, "facts_retained": 8, "retention_rate": 0.12121212121212122, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.17647058823529413, "status": 0.16666666666666666, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 783}}}, "reference_key_id": "keys/synthetic/syn_03024_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "3cb4552bd39acfb9b1ca04c5d04cccae183051dc06ba450042dd9b360768202c", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "81891eb09c27b34cffe6a7a9", "attempt_id": "6add956523f0", "run_group_id": "E4_core:A5|repairable|qwen3b:h2", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:26:08.048199Z", "task_family": "synthetic", "document_id": "syn_03024_2000", "document_hash": "ea10cd11ad08f01c50b0712bc481668857c9d2ddc8e6423968f2a31eeb97ac55", "question_id": "q04_multi_hop", "split": "test", "cluster_id": "syn_03024_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 2, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 2068, "query_tokens": 455, "message_tokens": 367, "message_utf8_bytes": 847, "cache_tensor_bytes": null, "serialized_payload_bytes": 847, "side_information_bytes": 0, "fetched_bytes": 582, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 11639.434623997658, "hops": {"1": 6938.470332999714, "2": 4700.964290997945}, "receiver_prefill_ms": 185.66891690716147, "ttft_ms": 11.844040942378342, "decode_ms": 11.776958010159433}, "total_wall_ms": 11849.090290954337, "ttft_ms": 11.844040942378342, "input_usage": {"lookups": 1}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 66, "facts_retained": 8, "retention_rate": 0.12121212121212122, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.17647058823529413, "status": 0.16666666666666666, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 783}, "ledger_by_hop": {"1": {"facts_total": 66, "facts_retained": 6, "retention_rate": 0.09090909090909091, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.17647058823529413, "status": 0.0, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 754}, "2": {"facts_total": 66, "facts_retained": 8, "retention_rate": 0.12121212121212122, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.17647058823529413, "status": 0.16666666666666666, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 783}}}, "reference_key_id": "keys/synthetic/syn_03024_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "3cb4552bd39acfb9b1ca04c5d04cccae183051dc06ba450042dd9b360768202c", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "c6169ee6a443339438e23875", "attempt_id": "a57ee2e3f66d", "run_group_id": "E4_core:A5|repairable|qwen3b:h2", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:26:08.272554Z", "task_family": "synthetic", "document_id": "syn_03024_2000", "document_hash": "ea10cd11ad08f01c50b0712bc481668857c9d2ddc8e6423968f2a31eeb97ac55", "question_id": "q05_multi_hop", "split": "test", "cluster_id": "syn_03024_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 2, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 2068, "query_tokens": 459, "message_tokens": 367, "message_utf8_bytes": 847, "cache_tensor_bytes": null, "serialized_payload_bytes": 847, "side_information_bytes": 0, "fetched_bytes": 582, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 11639.434623997658, "hops": {"1": 6938.470332999714, "2": 4700.964290997945}, "receiver_prefill_ms": 185.2607079781592, "ttft_ms": 12.044250033795834, "decode_ms": 11.8002500385046}, "total_wall_ms": 11849.064249079674, "ttft_ms": 12.044250033795834, "input_usage": {"lookups": 1}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 66, "facts_retained": 8, "retention_rate": 0.12121212121212122, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.17647058823529413, "status": 0.16666666666666666, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 783}, "ledger_by_hop": {"1": {"facts_total": 66, "facts_retained": 6, "retention_rate": 0.09090909090909091, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.17647058823529413, "status": 0.0, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 754}, "2": {"facts_total": 66, "facts_retained": 8, "retention_rate": 0.12121212121212122, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.17647058823529413, "status": 0.16666666666666666, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 783}}}, "reference_key_id": "keys/synthetic/syn_03024_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "3cb4552bd39acfb9b1ca04c5d04cccae183051dc06ba450042dd9b360768202c", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "f0aa0f2ff17d00ee0b5b8816", "attempt_id": "b609613785cd", "run_group_id": "E4_core:A5|repairable|qwen3b:h2", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:26:08.497533Z", "task_family": "synthetic", "document_id": "syn_03024_2000", "document_hash": "ea10cd11ad08f01c50b0712bc481668857c9d2ddc8e6423968f2a31eeb97ac55", "question_id": "q06_negation", "split": "test", "cluster_id": "syn_03024_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 2, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 2068, "query_tokens": 448, "message_tokens": 367, "message_utf8_bytes": 847, "cache_tensor_bytes": null, "serialized_payload_bytes": 847, "side_information_bytes": 0, "fetched_bytes": 582, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 11639.434623997658, "hops": {"1": 6938.470332999714, "2": 4700.964290997945}, "receiver_prefill_ms": 165.24495801422745, "ttft_ms": 11.977125075645745, "decode_ms": 11.791958939284086}, "total_wall_ms": 11828.95158289466, "ttft_ms": 11.977125075645745, "input_usage": {"lookups": 1}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "no", "evidence_artifact": {"ledger": {"facts_total": 66, "facts_retained": 8, "retention_rate": 0.12121212121212122, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.17647058823529413, "status": 0.16666666666666666, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 783}, "ledger_by_hop": {"1": {"facts_total": 66, "facts_retained": 6, "retention_rate": 0.09090909090909091, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.17647058823529413, "status": 0.0, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 754}, "2": {"facts_total": 66, "facts_retained": 8, "retention_rate": 0.12121212121212122, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.17647058823529413, "status": 0.16666666666666666, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 783}}}, "reference_key_id": "keys/synthetic/syn_03024_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "3cb4552bd39acfb9b1ca04c5d04cccae183051dc06ba450042dd9b360768202c", "output_hash": "9390298f3fb0c5b160498935d79cb139aef28e1c47358b4bbba61862b9c26e59"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "e0899cf79f0f7cd52286d1f5", "attempt_id": "cc06d91f470c", "run_group_id": "E4_core:A5|repairable|qwen3b:h2", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:26:08.703516Z", "task_family": "synthetic", "document_id": "syn_03024_2000", "document_hash": "ea10cd11ad08f01c50b0712bc481668857c9d2ddc8e6423968f2a31eeb97ac55", "question_id": "q07_negation", "split": "test", "cluster_id": "syn_03024_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 2, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 2068, "query_tokens": 450, "message_tokens": 367, "message_utf8_bytes": 847, "cache_tensor_bytes": null, "serialized_payload_bytes": 847, "side_information_bytes": 0, "fetched_bytes": 582, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 11639.434623997658, "hops": {"1": 6938.470332999714, "2": 4700.964290997945}, "receiver_prefill_ms": 184.92145801428705, "ttft_ms": 11.808958952315152, "decode_ms": 11.843875050544739}, "total_wall_ms": 11848.410123959184, "ttft_ms": 11.808958952315152, "input_usage": {"lookups": 1}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "no", "evidence_artifact": {"ledger": {"facts_total": 66, "facts_retained": 8, "retention_rate": 0.12121212121212122, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.17647058823529413, "status": 0.16666666666666666, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 783}, "ledger_by_hop": {"1": {"facts_total": 66, "facts_retained": 6, "retention_rate": 0.09090909090909091, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.17647058823529413, "status": 0.0, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 754}, "2": {"facts_total": 66, "facts_retained": 8, "retention_rate": 0.12121212121212122, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.17647058823529413, "status": 0.16666666666666666, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 783}}}, "reference_key_id": "keys/synthetic/syn_03024_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "3cb4552bd39acfb9b1ca04c5d04cccae183051dc06ba450042dd9b360768202c", "output_hash": "9390298f3fb0c5b160498935d79cb139aef28e1c47358b4bbba61862b9c26e59"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "4afaca2b78a5a2420a6246ac", "attempt_id": "ed1e17cb3579", "run_group_id": "E4_core:A5|repairable|qwen3b:h2", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:26:08.928038Z", "task_family": "synthetic", "document_id": "syn_03024_2000", "document_hash": "ea10cd11ad08f01c50b0712bc481668857c9d2ddc8e6423968f2a31eeb97ac55", "question_id": "q08_revision", "split": "test", "cluster_id": "syn_03024_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 2, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 2068, "query_tokens": 466, "message_tokens": 367, "message_utf8_bytes": 847, "cache_tensor_bytes": null, "serialized_payload_bytes": 847, "side_information_bytes": 0, "fetched_bytes": 582, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 11639.434623997658, "hops": {"1": 6938.470332999714, "2": 4700.964290997945}, "receiver_prefill_ms": 185.9617909649387, "ttft_ms": 11.853833915665746, "decode_ms": 11.820040992461145}, "total_wall_ms": 11849.50516500976, "ttft_ms": 11.853833915665746, "input_usage": {"lookups": 1}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 66, "facts_retained": 8, "retention_rate": 0.12121212121212122, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.17647058823529413, "status": 0.16666666666666666, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 783}, "ledger_by_hop": {"1": {"facts_total": 66, "facts_retained": 6, "retention_rate": 0.09090909090909091, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.17647058823529413, "status": 0.0, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 754}, "2": {"facts_total": 66, "facts_retained": 8, "retention_rate": 0.12121212121212122, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.17647058823529413, "status": 0.16666666666666666, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 783}}}, "reference_key_id": "keys/synthetic/syn_03024_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "3cb4552bd39acfb9b1ca04c5d04cccae183051dc06ba450042dd9b360768202c", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "45e5d22c0f88b0f7dafad5f5", "attempt_id": "aa199fc05bbc", "run_group_id": "E4_core:A5|repairable|qwen3b:h2", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:26:09.153414Z", "task_family": "synthetic", "document_id": "syn_03024_2000", "document_hash": "ea10cd11ad08f01c50b0712bc481668857c9d2ddc8e6423968f2a31eeb97ac55", "question_id": "q09_revision", "split": "test", "cluster_id": "syn_03024_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 2, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 2068, "query_tokens": 457, "message_tokens": 367, "message_utf8_bytes": 847, "cache_tensor_bytes": null, "serialized_payload_bytes": 847, "side_information_bytes": 0, "fetched_bytes": 582, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 11639.434623997658, "hops": {"1": 6938.470332999714, "2": 4700.964290997945}, "receiver_prefill_ms": 185.0972919492051, "ttft_ms": 11.72175002284348, "decode_ms": 11.691333027556539}, "total_wall_ms": 11848.329706932418, "ttft_ms": 11.72175002284348, "input_usage": {"lookups": 1}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 66, "facts_retained": 8, "retention_rate": 0.12121212121212122, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.17647058823529413, "status": 0.16666666666666666, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 783}, "ledger_by_hop": {"1": {"facts_total": 66, "facts_retained": 6, "retention_rate": 0.09090909090909091, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.17647058823529413, "status": 0.0, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 754}, "2": {"facts_total": 66, "facts_retained": 8, "retention_rate": 0.12121212121212122, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.17647058823529413, "status": 0.16666666666666666, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 783}}}, "reference_key_id": "keys/synthetic/syn_03024_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "3cb4552bd39acfb9b1ca04c5d04cccae183051dc06ba450042dd9b360768202c", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "ad152b0fbec499eb226af5ba", "attempt_id": "d80390d54ed3", "run_group_id": "E4_core:A5|repairable|qwen3b:h2", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:26:09.377257Z", "task_family": "synthetic", "document_id": "syn_03024_2000", "document_hash": "ea10cd11ad08f01c50b0712bc481668857c9d2ddc8e6423968f2a31eeb97ac55", "question_id": "q10_temporal", "split": "test", "cluster_id": "syn_03024_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 2, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 2068, "query_tokens": 467, "message_tokens": 367, "message_utf8_bytes": 847, "cache_tensor_bytes": null, "serialized_payload_bytes": 847, "side_information_bytes": 0, "fetched_bytes": 582, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 11639.434623997658, "hops": {"1": 6938.470332999714, "2": 4700.964290997945}, "receiver_prefill_ms": 185.96912501379848, "ttft_ms": 11.843541986308992, "decode_ms": 11.839832994155586}, "total_wall_ms": 11849.508290062658, "ttft_ms": 11.843541986308992, "input_usage": {"lookups": 1}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 66, "facts_retained": 8, "retention_rate": 0.12121212121212122, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.17647058823529413, "status": 0.16666666666666666, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 783}, "ledger_by_hop": {"1": {"facts_total": 66, "facts_retained": 6, "retention_rate": 0.09090909090909091, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.17647058823529413, "status": 0.0, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 754}, "2": {"facts_total": 66, "facts_retained": 8, "retention_rate": 0.12121212121212122, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.17647058823529413, "status": 0.16666666666666666, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 783}}}, "reference_key_id": "keys/synthetic/syn_03024_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "3cb4552bd39acfb9b1ca04c5d04cccae183051dc06ba450042dd9b360768202c", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "ea7dfe800281f7663543ee0e", "attempt_id": "e222ce7868c4", "run_group_id": "E4_core:A5|repairable|qwen3b:h2", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:26:09.602842Z", "task_family": "synthetic", "document_id": "syn_03024_2000", "document_hash": "ea10cd11ad08f01c50b0712bc481668857c9d2ddc8e6423968f2a31eeb97ac55", "question_id": "q11_aggregation", "split": "test", "cluster_id": "syn_03024_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 2, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 2068, "query_tokens": 460, "message_tokens": 367, "message_utf8_bytes": 847, "cache_tensor_bytes": null, "serialized_payload_bytes": 847, "side_information_bytes": 0, "fetched_bytes": 582, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 11639.434623997658, "hops": {"1": 6938.470332999714, "2": 4700.964290997945}, "receiver_prefill_ms": 186.07433408033103, "ttft_ms": 11.826290981844068, "decode_ms": 82.48495811130852}, "total_wall_ms": 11920.229956973344, "ttft_ms": 11.826290981844068, "input_usage": {"lookups": 1}, "output_usage": {"output_tokens": 7}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "8066734", "evidence_artifact": {"ledger": {"facts_total": 66, "facts_retained": 8, "retention_rate": 0.12121212121212122, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.17647058823529413, "status": 0.16666666666666666, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 783}, "ledger_by_hop": {"1": {"facts_total": 66, "facts_retained": 6, "retention_rate": 0.09090909090909091, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.17647058823529413, "status": 0.0, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 754}, "2": {"facts_total": 66, "facts_retained": 8, "retention_rate": 0.12121212121212122, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.17647058823529413, "status": 0.16666666666666666, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 783}}}, "reference_key_id": "keys/synthetic/syn_03024_2000.json", "task_scores": {"correct": 0.0, "abstained": 0.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "3cb4552bd39acfb9b1ca04c5d04cccae183051dc06ba450042dd9b360768202c", "output_hash": "167c75fdd6a41fca64228797ab8f3a1a3f994475246df0bffb2e060d5d2e4ff0"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "949a8dab89a6758d64b67831", "attempt_id": "58d5af2be7cb", "run_group_id": "E4_core:A5|repairable|qwen3b:h2", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:26:09.898871Z", "task_family": "synthetic", "document_id": "syn_03024_2000", "document_hash": "ea10cd11ad08f01c50b0712bc481668857c9d2ddc8e6423968f2a31eeb97ac55", "question_id": "q12_unit_conversion", "split": "test", "cluster_id": "syn_03024_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 2, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 2068, "query_tokens": 457, "message_tokens": 367, "message_utf8_bytes": 847, "cache_tensor_bytes": null, "serialized_payload_bytes": 847, "side_information_bytes": 0, "fetched_bytes": 582, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 11639.434623997658, "hops": {"1": 6938.470332999714, "2": 4700.964290997945}, "receiver_prefill_ms": 185.84016605746, "ttft_ms": 11.794457910582423, "decode_ms": 11.997958994470537}, "total_wall_ms": 11849.45574903395, "ttft_ms": 11.794457910582423, "input_usage": {"lookups": 1}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 66, "facts_retained": 8, "retention_rate": 0.12121212121212122, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.17647058823529413, "status": 0.16666666666666666, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 783}, "ledger_by_hop": {"1": {"facts_total": 66, "facts_retained": 6, "retention_rate": 0.09090909090909091, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.17647058823529413, "status": 0.0, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 754}, "2": {"facts_total": 66, "facts_retained": 8, "retention_rate": 0.12121212121212122, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.17647058823529413, "status": 0.16666666666666666, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 783}}}, "reference_key_id": "keys/synthetic/syn_03024_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "3cb4552bd39acfb9b1ca04c5d04cccae183051dc06ba450042dd9b360768202c", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "bc008dc614648e16e1588613", "attempt_id": "92f96266df72", "run_group_id": "E4_core:A5|repairable|qwen3b:h2", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:26:10.124159Z", "task_family": "synthetic", "document_id": "syn_03024_2000", "document_hash": "ea10cd11ad08f01c50b0712bc481668857c9d2ddc8e6423968f2a31eeb97ac55", "question_id": "q13_unanswerable", "split": "test", "cluster_id": "syn_03024_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 2, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 2068, "query_tokens": 461, "message_tokens": 367, "message_utf8_bytes": 847, "cache_tensor_bytes": null, "serialized_payload_bytes": 847, "side_information_bytes": 0, "fetched_bytes": 582, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 11639.434623997658, "hops": {"1": 6938.470332999714, "2": 4700.964290997945}, "receiver_prefill_ms": 185.82870799582452, "ttft_ms": 11.764083988964558, "decode_ms": 11.943458928726614}, "total_wall_ms": 11849.347415962256, "ttft_ms": 11.764083988964558, "input_usage": {"lookups": 1}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 66, "facts_retained": 8, "retention_rate": 0.12121212121212122, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.17647058823529413, "status": 0.16666666666666666, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 783}, "ledger_by_hop": {"1": {"facts_total": 66, "facts_retained": 6, "retention_rate": 0.09090909090909091, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.17647058823529413, "status": 0.0, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 754}, "2": {"facts_total": 66, "facts_retained": 8, "retention_rate": 0.12121212121212122, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.17647058823529413, "status": 0.16666666666666666, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 783}}}, "reference_key_id": "keys/synthetic/syn_03024_2000.json", "task_scores": {"correct": 1.0, "abstained": 1.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "3cb4552bd39acfb9b1ca04c5d04cccae183051dc06ba450042dd9b360768202c", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "16af9a53046028a4496ab413", "attempt_id": "28f6de5fcc12", "run_group_id": "E4_core:A5|repairable|qwen3b:h2", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:26:10.349260Z", "task_family": "synthetic", "document_id": "syn_03024_2000", "document_hash": "ea10cd11ad08f01c50b0712bc481668857c9d2ddc8e6423968f2a31eeb97ac55", "question_id": "q14_unanswerable", "split": "test", "cluster_id": "syn_03024_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 2, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 2068, "query_tokens": 462, "message_tokens": 367, "message_utf8_bytes": 847, "cache_tensor_bytes": null, "serialized_payload_bytes": 847, "side_information_bytes": 0, "fetched_bytes": 582, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 11639.434623997658, "hops": {"1": 6938.470332999714, "2": 4700.964290997945}, "receiver_prefill_ms": 185.67920906934887, "ttft_ms": 12.047957978211343, "decode_ms": 11.850040988065302}, "total_wall_ms": 11849.513498949818, "ttft_ms": 12.047957978211343, "input_usage": {"lookups": 1}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 66, "facts_retained": 8, "retention_rate": 0.12121212121212122, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.17647058823529413, "status": 0.16666666666666666, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 783}, "ledger_by_hop": {"1": {"facts_total": 66, "facts_retained": 6, "retention_rate": 0.09090909090909091, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.17647058823529413, "status": 0.0, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 754}, "2": {"facts_total": 66, "facts_retained": 8, "retention_rate": 0.12121212121212122, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.17647058823529413, "status": 0.16666666666666666, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 783}}}, "reference_key_id": "keys/synthetic/syn_03024_2000.json", "task_scores": {"correct": 1.0, "abstained": 1.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "3cb4552bd39acfb9b1ca04c5d04cccae183051dc06ba450042dd9b360768202c", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "0e665ad4e80089ed6441be87", "attempt_id": "52c4c4ea9cc7", "run_group_id": "E4_core:A5|repairable|qwen3b:h4", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:26:19.994474Z", "task_family": "synthetic", "document_id": "syn_03024_2000", "document_hash": "ea10cd11ad08f01c50b0712bc481668857c9d2ddc8e6423968f2a31eeb97ac55", "question_id": "q00_single_fact", "split": "test", "cluster_id": "syn_03024_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 4, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 2068, "query_tokens": 449, "message_tokens": 367, "message_utf8_bytes": 847, "cache_tensor_bytes": null, "serialized_payload_bytes": 847, "side_information_bytes": 0, "fetched_bytes": 1746, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 21056.896539987065, "hops": {"1": 6938.470332999714, "2": 4700.964290997945, "3": 4701.423165970482, "4": 4716.0387500189245}, "receiver_prefill_ms": 164.56274990923703, "ttft_ms": 11.853333096951246, "decode_ms": 117.99962492659688}, "total_wall_ms": 21351.748873945326, "ttft_ms": 11.853333096951246, "input_usage": {"lookups": 3}, "output_usage": {"output_tokens": 10}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "BU-55G-3254", "evidence_artifact": {"ledger": {"facts_total": 66, "facts_retained": 8, "retention_rate": 0.12121212121212122, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.17647058823529413, "status": 0.16666666666666666, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 783}, "ledger_by_hop": {"1": {"facts_total": 66, "facts_retained": 6, "retention_rate": 0.09090909090909091, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.17647058823529413, "status": 0.0, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 754}, "2": {"facts_total": 66, "facts_retained": 8, "retention_rate": 0.12121212121212122, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.17647058823529413, "status": 0.16666666666666666, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 783}, "3": {"facts_total": 66, "facts_retained": 8, "retention_rate": 0.12121212121212122, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.17647058823529413, "status": 0.16666666666666666, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 783}, "4": {"facts_total": 66, "facts_retained": 8, "retention_rate": 0.12121212121212122, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.17647058823529413, "status": 0.16666666666666666, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 783}}}, "reference_key_id": "keys/synthetic/syn_03024_2000.json", "task_scores": {"correct": 0.0, "abstained": 0.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "3cb4552bd39acfb9b1ca04c5d04cccae183051dc06ba450042dd9b360768202c", "output_hash": "98227e32c2121a7856e6d2b32b74598ce6f4183c2ec3eec7ce4b9b0b3f66b98b"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "2df70b38d477858500554113", "attempt_id": "57411c6174b6", "run_group_id": "E4_core:A5|repairable|qwen3b:h4", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:26:20.305605Z", "task_family": "synthetic", "document_id": "syn_03024_2000", "document_hash": "ea10cd11ad08f01c50b0712bc481668857c9d2ddc8e6423968f2a31eeb97ac55", "question_id": "q01_single_fact", "split": "test", "cluster_id": "syn_03024_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 4, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 2068, "query_tokens": 454, "message_tokens": 367, "message_utf8_bytes": 847, "cache_tensor_bytes": null, "serialized_payload_bytes": 847, "side_information_bytes": 0, "fetched_bytes": 1746, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 21056.896539987065, "hops": {"1": 6938.470332999714, "2": 4700.964290997945, "3": 4701.423165970482, "4": 4716.0387500189245}, "receiver_prefill_ms": 184.66441705822945, "ttft_ms": 12.011292041279376, "decode_ms": 129.9014170654118}, "total_wall_ms": 21383.978081983514, "ttft_ms": 12.011292041279376, "input_usage": {"lookups": 3}, "output_usage": {"output_tokens": 11}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "Lot 568-G in Mont-Laurier", "evidence_artifact": {"ledger": {"facts_total": 66, "facts_retained": 8, "retention_rate": 0.12121212121212122, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.17647058823529413, "status": 0.16666666666666666, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 783}, "ledger_by_hop": {"1": {"facts_total": 66, "facts_retained": 6, "retention_rate": 0.09090909090909091, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.17647058823529413, "status": 0.0, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 754}, "2": {"facts_total": 66, "facts_retained": 8, "retention_rate": 0.12121212121212122, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.17647058823529413, "status": 0.16666666666666666, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 783}, "3": {"facts_total": 66, "facts_retained": 8, "retention_rate": 0.12121212121212122, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.17647058823529413, "status": 0.16666666666666666, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 783}, "4": {"facts_total": 66, "facts_retained": 8, "retention_rate": 0.12121212121212122, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.17647058823529413, "status": 0.16666666666666666, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 783}}}, "reference_key_id": "keys/synthetic/syn_03024_2000.json", "task_scores": {"correct": 0.0, "abstained": 0.0, "f1": 0.5}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "3cb4552bd39acfb9b1ca04c5d04cccae183051dc06ba450042dd9b360768202c", "output_hash": "98e007faa5a10742ad74f2778a64e3ea0d6b72bec1c0392200d475073d7bf082"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "0687e8c45124b05171238763", "attempt_id": "80743994c668", "run_group_id": "E4_core:A5|repairable|qwen3b:h4", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:26:20.648131Z", "task_family": "synthetic", "document_id": "syn_03024_2000", "document_hash": "ea10cd11ad08f01c50b0712bc481668857c9d2ddc8e6423968f2a31eeb97ac55", "question_id": "q02_exact_copy", "split": "test", "cluster_id": "syn_03024_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 4, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 2068, "query_tokens": 454, "message_tokens": 367, "message_utf8_bytes": 847, "cache_tensor_bytes": null, "serialized_payload_bytes": 847, "side_information_bytes": 0, "fetched_bytes": 1746, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 21056.896539987065, "hops": {"1": 6938.470332999714, "2": 4700.964290997945, "3": 4701.423165970482, "4": 4716.0387500189245}, "receiver_prefill_ms": 185.12658297549933, "ttft_ms": 11.88695803284645, "decode_ms": 11.830750037916005}, "total_wall_ms": 21266.123705892824, "ttft_ms": 11.88695803284645, "input_usage": {"lookups": 3}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 66, "facts_retained": 8, "retention_rate": 0.12121212121212122, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.17647058823529413, "status": 0.16666666666666666, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 783}, "ledger_by_hop": {"1": {"facts_total": 66, "facts_retained": 6, "retention_rate": 0.09090909090909091, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.17647058823529413, "status": 0.0, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 754}, "2": {"facts_total": 66, "facts_retained": 8, "retention_rate": 0.12121212121212122, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.17647058823529413, "status": 0.16666666666666666, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 783}, "3": {"facts_total": 66, "facts_retained": 8, "retention_rate": 0.12121212121212122, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.17647058823529413, "status": 0.16666666666666666, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 783}, "4": {"facts_total": 66, "facts_retained": 8, "retention_rate": 0.12121212121212122, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.17647058823529413, "status": 0.16666666666666666, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 783}}}, "reference_key_id": "keys/synthetic/syn_03024_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "3cb4552bd39acfb9b1ca04c5d04cccae183051dc06ba450042dd9b360768202c", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "a6271a37f0050d5fc481cbf1", "attempt_id": "3cfcf48721b7", "run_group_id": "E4_core:A5|repairable|qwen3b:h4", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:26:20.872912Z", "task_family": "synthetic", "document_id": "syn_03024_2000", "document_hash": "ea10cd11ad08f01c50b0712bc481668857c9d2ddc8e6423968f2a31eeb97ac55", "question_id": "q03_exact_copy", "split": "test", "cluster_id": "syn_03024_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 4, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 2068, "query_tokens": 449, "message_tokens": 367, "message_utf8_bytes": 847, "cache_tensor_bytes": null, "serialized_payload_bytes": 847, "side_information_bytes": 0, "fetched_bytes": 1746, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 21056.896539987065, "hops": {"1": 6938.470332999714, "2": 4700.964290997945, "3": 4701.423165970482, "4": 4716.0387500189245}, "receiver_prefill_ms": 163.80154201760888, "ttft_ms": 11.907457956112921, "decode_ms": 11.8259999435395}, "total_wall_ms": 21244.801664957777, "ttft_ms": 11.907457956112921, "input_usage": {"lookups": 3}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 66, "facts_retained": 8, "retention_rate": 0.12121212121212122, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.17647058823529413, "status": 0.16666666666666666, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 783}, "ledger_by_hop": {"1": {"facts_total": 66, "facts_retained": 6, "retention_rate": 0.09090909090909091, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.17647058823529413, "status": 0.0, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 754}, "2": {"facts_total": 66, "facts_retained": 8, "retention_rate": 0.12121212121212122, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.17647058823529413, "status": 0.16666666666666666, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 783}, "3": {"facts_total": 66, "facts_retained": 8, "retention_rate": 0.12121212121212122, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.17647058823529413, "status": 0.16666666666666666, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 783}, "4": {"facts_total": 66, "facts_retained": 8, "retention_rate": 0.12121212121212122, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.17647058823529413, "status": 0.16666666666666666, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 783}}}, "reference_key_id": "keys/synthetic/syn_03024_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "3cb4552bd39acfb9b1ca04c5d04cccae183051dc06ba450042dd9b360768202c", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "43e8188e6fcbbaa714559065", "attempt_id": "5c0d580b8ef0", "run_group_id": "E4_core:A5|repairable|qwen3b:h4", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:26:21.075420Z", "task_family": "synthetic", "document_id": "syn_03024_2000", "document_hash": "ea10cd11ad08f01c50b0712bc481668857c9d2ddc8e6423968f2a31eeb97ac55", "question_id": "q04_multi_hop", "split": "test", "cluster_id": "syn_03024_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 4, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 2068, "query_tokens": 455, "message_tokens": 367, "message_utf8_bytes": 847, "cache_tensor_bytes": null, "serialized_payload_bytes": 847, "side_information_bytes": 0, "fetched_bytes": 1746, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 21056.896539987065, "hops": {"1": 6938.470332999714, "2": 4700.964290997945, "3": 4701.423165970482, "4": 4716.0387500189245}, "receiver_prefill_ms": 184.74191706627607, "ttft_ms": 11.820500018075109, "decode_ms": 12.04025000333786}, "total_wall_ms": 21265.862830914557, "ttft_ms": 11.820500018075109, "input_usage": {"lookups": 3}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 66, "facts_retained": 8, "retention_rate": 0.12121212121212122, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.17647058823529413, "status": 0.16666666666666666, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 783}, "ledger_by_hop": {"1": {"facts_total": 66, "facts_retained": 6, "retention_rate": 0.09090909090909091, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.17647058823529413, "status": 0.0, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 754}, "2": {"facts_total": 66, "facts_retained": 8, "retention_rate": 0.12121212121212122, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.17647058823529413, "status": 0.16666666666666666, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 783}, "3": {"facts_total": 66, "facts_retained": 8, "retention_rate": 0.12121212121212122, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.17647058823529413, "status": 0.16666666666666666, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 783}, "4": {"facts_total": 66, "facts_retained": 8, "retention_rate": 0.12121212121212122, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.17647058823529413, "status": 0.16666666666666666, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 783}}}, "reference_key_id": "keys/synthetic/syn_03024_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "3cb4552bd39acfb9b1ca04c5d04cccae183051dc06ba450042dd9b360768202c", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "c9dc205e7109900d187a806d", "attempt_id": "cf0c96dcba30", "run_group_id": "E4_core:A5|repairable|qwen3b:h4", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:26:21.299314Z", "task_family": "synthetic", "document_id": "syn_03024_2000", "document_hash": "ea10cd11ad08f01c50b0712bc481668857c9d2ddc8e6423968f2a31eeb97ac55", "question_id": "q05_multi_hop", "split": "test", "cluster_id": "syn_03024_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 4, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 2068, "query_tokens": 459, "message_tokens": 367, "message_utf8_bytes": 847, "cache_tensor_bytes": null, "serialized_payload_bytes": 847, "side_information_bytes": 0, "fetched_bytes": 1746, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 21056.896539987065, "hops": {"1": 6938.470332999714, "2": 4700.964290997945, "3": 4701.423165970482, "4": 4716.0387500189245}, "receiver_prefill_ms": 185.26295793708414, "ttft_ms": 12.03054201323539, "decode_ms": 12.195541989058256}, "total_wall_ms": 21266.860874020495, "ttft_ms": 12.03054201323539, "input_usage": {"lookups": 3}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 66, "facts_retained": 8, "retention_rate": 0.12121212121212122, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.17647058823529413, "status": 0.16666666666666666, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 783}, "ledger_by_hop": {"1": {"facts_total": 66, "facts_retained": 6, "retention_rate": 0.09090909090909091, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.17647058823529413, "status": 0.0, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 754}, "2": {"facts_total": 66, "facts_retained": 8, "retention_rate": 0.12121212121212122, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.17647058823529413, "status": 0.16666666666666666, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 783}, "3": {"facts_total": 66, "facts_retained": 8, "retention_rate": 0.12121212121212122, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.17647058823529413, "status": 0.16666666666666666, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 783}, "4": {"facts_total": 66, "facts_retained": 8, "retention_rate": 0.12121212121212122, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.17647058823529413, "status": 0.16666666666666666, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 783}}}, "reference_key_id": "keys/synthetic/syn_03024_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "3cb4552bd39acfb9b1ca04c5d04cccae183051dc06ba450042dd9b360768202c", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "3ab9372f4939a85724fe460d", "attempt_id": "735ed43fda60", "run_group_id": "E4_core:A5|repairable|qwen3b:h4", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:26:21.524660Z", "task_family": "synthetic", "document_id": "syn_03024_2000", "document_hash": "ea10cd11ad08f01c50b0712bc481668857c9d2ddc8e6423968f2a31eeb97ac55", "question_id": "q06_negation", "split": "test", "cluster_id": "syn_03024_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 4, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 2068, "query_tokens": 448, "message_tokens": 367, "message_utf8_bytes": 847, "cache_tensor_bytes": null, "serialized_payload_bytes": 847, "side_information_bytes": 0, "fetched_bytes": 1746, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 21056.896539987065, "hops": {"1": 6938.470332999714, "2": 4700.964290997945, "3": 4701.423165970482, "4": 4716.0387500189245}, "receiver_prefill_ms": 164.92141596972942, "ttft_ms": 11.728417011909187, "decode_ms": 11.774874990805984}, "total_wall_ms": 21245.71341497358, "ttft_ms": 11.728417011909187, "input_usage": {"lookups": 3}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "no", "evidence_artifact": {"ledger": {"facts_total": 66, "facts_retained": 8, "retention_rate": 0.12121212121212122, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.17647058823529413, "status": 0.16666666666666666, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 783}, "ledger_by_hop": {"1": {"facts_total": 66, "facts_retained": 6, "retention_rate": 0.09090909090909091, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.17647058823529413, "status": 0.0, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 754}, "2": {"facts_total": 66, "facts_retained": 8, "retention_rate": 0.12121212121212122, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.17647058823529413, "status": 0.16666666666666666, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 783}, "3": {"facts_total": 66, "facts_retained": 8, "retention_rate": 0.12121212121212122, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.17647058823529413, "status": 0.16666666666666666, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 783}, "4": {"facts_total": 66, "facts_retained": 8, "retention_rate": 0.12121212121212122, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.17647058823529413, "status": 0.16666666666666666, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 783}}}, "reference_key_id": "keys/synthetic/syn_03024_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "3cb4552bd39acfb9b1ca04c5d04cccae183051dc06ba450042dd9b360768202c", "output_hash": "9390298f3fb0c5b160498935d79cb139aef28e1c47358b4bbba61862b9c26e59"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "f5e3e1f4699c2e1eba0af4a2", "attempt_id": "de9c015cf9e1", "run_group_id": "E4_core:A5|repairable|qwen3b:h4", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:26:21.729308Z", "task_family": "synthetic", "document_id": "syn_03024_2000", "document_hash": "ea10cd11ad08f01c50b0712bc481668857c9d2ddc8e6423968f2a31eeb97ac55", "question_id": "q07_negation", "split": "test", "cluster_id": "syn_03024_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 4, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 2068, "query_tokens": 450, "message_tokens": 367, "message_utf8_bytes": 847, "cache_tensor_bytes": null, "serialized_payload_bytes": 847, "side_information_bytes": 0, "fetched_bytes": 1746, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 21056.896539987065, "hops": {"1": 6938.470332999714, "2": 4700.964290997945, "3": 4701.423165970482, "4": 4716.0387500189245}, "receiver_prefill_ms": 184.78320795111358, "ttft_ms": 11.876291944645345, "decode_ms": 11.811333009973168}, "total_wall_ms": 21265.739165013656, "ttft_ms": 11.876291944645345, "input_usage": {"lookups": 3}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "no", "evidence_artifact": {"ledger": {"facts_total": 66, "facts_retained": 8, "retention_rate": 0.12121212121212122, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.17647058823529413, "status": 0.16666666666666666, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 783}, "ledger_by_hop": {"1": {"facts_total": 66, "facts_retained": 6, "retention_rate": 0.09090909090909091, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.17647058823529413, "status": 0.0, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 754}, "2": {"facts_total": 66, "facts_retained": 8, "retention_rate": 0.12121212121212122, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.17647058823529413, "status": 0.16666666666666666, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 783}, "3": {"facts_total": 66, "facts_retained": 8, "retention_rate": 0.12121212121212122, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.17647058823529413, "status": 0.16666666666666666, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 783}, "4": {"facts_total": 66, "facts_retained": 8, "retention_rate": 0.12121212121212122, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.17647058823529413, "status": 0.16666666666666666, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 783}}}, "reference_key_id": "keys/synthetic/syn_03024_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "3cb4552bd39acfb9b1ca04c5d04cccae183051dc06ba450042dd9b360768202c", "output_hash": "9390298f3fb0c5b160498935d79cb139aef28e1c47358b4bbba61862b9c26e59"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "fc2f863d93ab99742e6425c2", "attempt_id": "c82ad5dcf1f1", "run_group_id": "E4_core:A5|repairable|qwen3b:h4", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:26:21.952962Z", "task_family": "synthetic", "document_id": "syn_03024_2000", "document_hash": "ea10cd11ad08f01c50b0712bc481668857c9d2ddc8e6423968f2a31eeb97ac55", "question_id": "q08_revision", "split": "test", "cluster_id": "syn_03024_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 4, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 2068, "query_tokens": 466, "message_tokens": 367, "message_utf8_bytes": 847, "cache_tensor_bytes": null, "serialized_payload_bytes": 847, "side_information_bytes": 0, "fetched_bytes": 1746, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 21056.896539987065, "hops": {"1": 6938.470332999714, "2": 4700.964290997945, "3": 4701.423165970482, "4": 4716.0387500189245}, "receiver_prefill_ms": 185.4307500179857, "ttft_ms": 12.07654201425612, "decode_ms": 11.884083040058613}, "total_wall_ms": 21266.66474796366, "ttft_ms": 12.07654201425612, "input_usage": {"lookups": 3}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 66, "facts_retained": 8, "retention_rate": 0.12121212121212122, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.17647058823529413, "status": 0.16666666666666666, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 783}, "ledger_by_hop": {"1": {"facts_total": 66, "facts_retained": 6, "retention_rate": 0.09090909090909091, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.17647058823529413, "status": 0.0, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 754}, "2": {"facts_total": 66, "facts_retained": 8, "retention_rate": 0.12121212121212122, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.17647058823529413, "status": 0.16666666666666666, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 783}, "3": {"facts_total": 66, "facts_retained": 8, "retention_rate": 0.12121212121212122, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.17647058823529413, "status": 0.16666666666666666, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 783}, "4": {"facts_total": 66, "facts_retained": 8, "retention_rate": 0.12121212121212122, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.17647058823529413, "status": 0.16666666666666666, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 783}}}, "reference_key_id": "keys/synthetic/syn_03024_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "3cb4552bd39acfb9b1ca04c5d04cccae183051dc06ba450042dd9b360768202c", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "1baa789d177e91815ae7c290", "attempt_id": "4977b3e56415", "run_group_id": "E4_core:A5|repairable|qwen3b:h4", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:26:22.177851Z", "task_family": "synthetic", "document_id": "syn_03024_2000", "document_hash": "ea10cd11ad08f01c50b0712bc481668857c9d2ddc8e6423968f2a31eeb97ac55", "question_id": "q09_revision", "split": "test", "cluster_id": "syn_03024_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 4, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 2068, "query_tokens": 457, "message_tokens": 367, "message_utf8_bytes": 847, "cache_tensor_bytes": null, "serialized_payload_bytes": 847, "side_information_bytes": 0, "fetched_bytes": 1746, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 21056.896539987065, "hops": {"1": 6938.470332999714, "2": 4700.964290997945, "3": 4701.423165970482, "4": 4716.0387500189245}, "receiver_prefill_ms": 185.31995895318687, "ttft_ms": 11.825166991911829, "decode_ms": 11.875541997142136}, "total_wall_ms": 21266.290581901558, "ttft_ms": 11.825166991911829, "input_usage": {"lookups": 3}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 66, "facts_retained": 8, "retention_rate": 0.12121212121212122, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.17647058823529413, "status": 0.16666666666666666, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 783}, "ledger_by_hop": {"1": {"facts_total": 66, "facts_retained": 6, "retention_rate": 0.09090909090909091, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.17647058823529413, "status": 0.0, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 754}, "2": {"facts_total": 66, "facts_retained": 8, "retention_rate": 0.12121212121212122, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.17647058823529413, "status": 0.16666666666666666, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 783}, "3": {"facts_total": 66, "facts_retained": 8, "retention_rate": 0.12121212121212122, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.17647058823529413, "status": 0.16666666666666666, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 783}, "4": {"facts_total": 66, "facts_retained": 8, "retention_rate": 0.12121212121212122, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.17647058823529413, "status": 0.16666666666666666, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 783}}}, "reference_key_id": "keys/synthetic/syn_03024_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "3cb4552bd39acfb9b1ca04c5d04cccae183051dc06ba450042dd9b360768202c", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "6880837c1ad977807c81a8c7", "attempt_id": "795d5a0d3d3c", "run_group_id": "E4_core:A5|repairable|qwen3b:h4", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:26:22.402353Z", "task_family": "synthetic", "document_id": "syn_03024_2000", "document_hash": "ea10cd11ad08f01c50b0712bc481668857c9d2ddc8e6423968f2a31eeb97ac55", "question_id": "q10_temporal", "split": "test", "cluster_id": "syn_03024_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 4, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 2068, "query_tokens": 467, "message_tokens": 367, "message_utf8_bytes": 847, "cache_tensor_bytes": null, "serialized_payload_bytes": 847, "side_information_bytes": 0, "fetched_bytes": 1746, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 21056.896539987065, "hops": {"1": 6938.470332999714, "2": 4700.964290997945, "3": 4701.423165970482, "4": 4716.0387500189245}, "receiver_prefill_ms": 185.88358291890472, "ttft_ms": 11.947875027544796, "decode_ms": 11.9317090138793}, "total_wall_ms": 21267.132830922492, "ttft_ms": 11.947875027544796, "input_usage": {"lookups": 3}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 66, "facts_retained": 8, "retention_rate": 0.12121212121212122, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.17647058823529413, "status": 0.16666666666666666, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 783}, "ledger_by_hop": {"1": {"facts_total": 66, "facts_retained": 6, "retention_rate": 0.09090909090909091, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.17647058823529413, "status": 0.0, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 754}, "2": {"facts_total": 66, "facts_retained": 8, "retention_rate": 0.12121212121212122, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.17647058823529413, "status": 0.16666666666666666, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 783}, "3": {"facts_total": 66, "facts_retained": 8, "retention_rate": 0.12121212121212122, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.17647058823529413, "status": 0.16666666666666666, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 783}, "4": {"facts_total": 66, "facts_retained": 8, "retention_rate": 0.12121212121212122, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.17647058823529413, "status": 0.16666666666666666, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 783}}}, "reference_key_id": "keys/synthetic/syn_03024_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "3cb4552bd39acfb9b1ca04c5d04cccae183051dc06ba450042dd9b360768202c", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "2901c2cc1c8400aed5c10162", "attempt_id": "7434807679c6", "run_group_id": "E4_core:A5|repairable|qwen3b:h4", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:26:22.627525Z", "task_family": "synthetic", "document_id": "syn_03024_2000", "document_hash": "ea10cd11ad08f01c50b0712bc481668857c9d2ddc8e6423968f2a31eeb97ac55", "question_id": "q11_aggregation", "split": "test", "cluster_id": "syn_03024_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 4, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 2068, "query_tokens": 460, "message_tokens": 367, "message_utf8_bytes": 847, "cache_tensor_bytes": null, "serialized_payload_bytes": 847, "side_information_bytes": 0, "fetched_bytes": 1746, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 21056.896539987065, "hops": {"1": 6938.470332999714, "2": 4700.964290997945, "3": 4701.423165970482, "4": 4716.0387500189245}, "receiver_prefill_ms": 185.49416703172028, "ttft_ms": 11.954375077039003, "decode_ms": 82.17991702258587}, "total_wall_ms": 21336.92470600363, "ttft_ms": 11.954375077039003, "input_usage": {"lookups": 3}, "output_usage": {"output_tokens": 7}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "8066734", "evidence_artifact": {"ledger": {"facts_total": 66, "facts_retained": 8, "retention_rate": 0.12121212121212122, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.17647058823529413, "status": 0.16666666666666666, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 783}, "ledger_by_hop": {"1": {"facts_total": 66, "facts_retained": 6, "retention_rate": 0.09090909090909091, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.17647058823529413, "status": 0.0, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 754}, "2": {"facts_total": 66, "facts_retained": 8, "retention_rate": 0.12121212121212122, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.17647058823529413, "status": 0.16666666666666666, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 783}, "3": {"facts_total": 66, "facts_retained": 8, "retention_rate": 0.12121212121212122, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.17647058823529413, "status": 0.16666666666666666, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 783}, "4": {"facts_total": 66, "facts_retained": 8, "retention_rate": 0.12121212121212122, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.17647058823529413, "status": 0.16666666666666666, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 783}}}, "reference_key_id": "keys/synthetic/syn_03024_2000.json", "task_scores": {"correct": 0.0, "abstained": 0.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "3cb4552bd39acfb9b1ca04c5d04cccae183051dc06ba450042dd9b360768202c", "output_hash": "167c75fdd6a41fca64228797ab8f3a1a3f994475246df0bffb2e060d5d2e4ff0"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "3aa487a1dd3a5c87a1d77098", "attempt_id": "3db19c9fbe5b", "run_group_id": "E4_core:A5|repairable|qwen3b:h4", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:26:22.923126Z", "task_family": "synthetic", "document_id": "syn_03024_2000", "document_hash": "ea10cd11ad08f01c50b0712bc481668857c9d2ddc8e6423968f2a31eeb97ac55", "question_id": "q12_unit_conversion", "split": "test", "cluster_id": "syn_03024_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 4, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 2068, "query_tokens": 457, "message_tokens": 367, "message_utf8_bytes": 847, "cache_tensor_bytes": null, "serialized_payload_bytes": 847, "side_information_bytes": 0, "fetched_bytes": 1746, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 21056.896539987065, "hops": {"1": 6938.470332999714, "2": 4700.964290997945, "3": 4701.423165970482, "4": 4716.0387500189245}, "receiver_prefill_ms": 185.62016694340855, "ttft_ms": 11.85858401004225, "decode_ms": 11.833833996206522}, "total_wall_ms": 21266.592290019616, "ttft_ms": 11.85858401004225, "input_usage": {"lookups": 3}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 66, "facts_retained": 8, "retention_rate": 0.12121212121212122, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.17647058823529413, "status": 0.16666666666666666, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 783}, "ledger_by_hop": {"1": {"facts_total": 66, "facts_retained": 6, "retention_rate": 0.09090909090909091, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.17647058823529413, "status": 0.0, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 754}, "2": {"facts_total": 66, "facts_retained": 8, "retention_rate": 0.12121212121212122, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.17647058823529413, "status": 0.16666666666666666, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 783}, "3": {"facts_total": 66, "facts_retained": 8, "retention_rate": 0.12121212121212122, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.17647058823529413, "status": 0.16666666666666666, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 783}, "4": {"facts_total": 66, "facts_retained": 8, "retention_rate": 0.12121212121212122, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.17647058823529413, "status": 0.16666666666666666, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 783}}}, "reference_key_id": "keys/synthetic/syn_03024_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "3cb4552bd39acfb9b1ca04c5d04cccae183051dc06ba450042dd9b360768202c", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "4442e52b34025e6a191afe2f", "attempt_id": "e81310278a02", "run_group_id": "E4_core:A5|repairable|qwen3b:h4", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:26:23.148148Z", "task_family": "synthetic", "document_id": "syn_03024_2000", "document_hash": "ea10cd11ad08f01c50b0712bc481668857c9d2ddc8e6423968f2a31eeb97ac55", "question_id": "q13_unanswerable", "split": "test", "cluster_id": "syn_03024_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 4, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 2068, "query_tokens": 461, "message_tokens": 367, "message_utf8_bytes": 847, "cache_tensor_bytes": null, "serialized_payload_bytes": 847, "side_information_bytes": 0, "fetched_bytes": 1746, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 21056.896539987065, "hops": {"1": 6938.470332999714, "2": 4700.964290997945, "3": 4701.423165970482, "4": 4716.0387500189245}, "receiver_prefill_ms": 185.344333993271, "ttft_ms": 11.912125046364963, "decode_ms": 11.801417102105916}, "total_wall_ms": 21266.318249050528, "ttft_ms": 11.912125046364963, "input_usage": {"lookups": 3}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 66, "facts_retained": 8, "retention_rate": 0.12121212121212122, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.17647058823529413, "status": 0.16666666666666666, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 783}, "ledger_by_hop": {"1": {"facts_total": 66, "facts_retained": 6, "retention_rate": 0.09090909090909091, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.17647058823529413, "status": 0.0, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 754}, "2": {"facts_total": 66, "facts_retained": 8, "retention_rate": 0.12121212121212122, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.17647058823529413, "status": 0.16666666666666666, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 783}, "3": {"facts_total": 66, "facts_retained": 8, "retention_rate": 0.12121212121212122, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.17647058823529413, "status": 0.16666666666666666, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 783}, "4": {"facts_total": 66, "facts_retained": 8, "retention_rate": 0.12121212121212122, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.17647058823529413, "status": 0.16666666666666666, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 783}}}, "reference_key_id": "keys/synthetic/syn_03024_2000.json", "task_scores": {"correct": 1.0, "abstained": 1.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "3cb4552bd39acfb9b1ca04c5d04cccae183051dc06ba450042dd9b360768202c", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "c9f045d2e63d93e8d6a99d2f", "attempt_id": "b4a5e56399cf", "run_group_id": "E4_core:A5|repairable|qwen3b:h4", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:26:23.372326Z", "task_family": "synthetic", "document_id": "syn_03024_2000", "document_hash": "ea10cd11ad08f01c50b0712bc481668857c9d2ddc8e6423968f2a31eeb97ac55", "question_id": "q14_unanswerable", "split": "test", "cluster_id": "syn_03024_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 4, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 2068, "query_tokens": 462, "message_tokens": 367, "message_utf8_bytes": 847, "cache_tensor_bytes": null, "serialized_payload_bytes": 847, "side_information_bytes": 0, "fetched_bytes": 1746, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 21056.896539987065, "hops": {"1": 6938.470332999714, "2": 4700.964290997945, "3": 4701.423165970482, "4": 4716.0387500189245}, "receiver_prefill_ms": 185.00958301592618, "ttft_ms": 12.10662501398474, "decode_ms": 11.93462498486042}, "total_wall_ms": 21266.433289973065, "ttft_ms": 12.10662501398474, "input_usage": {"lookups": 3}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 66, "facts_retained": 8, "retention_rate": 0.12121212121212122, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.17647058823529413, "status": 0.16666666666666666, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 783}, "ledger_by_hop": {"1": {"facts_total": 66, "facts_retained": 6, "retention_rate": 0.09090909090909091, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.17647058823529413, "status": 0.0, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 754}, "2": {"facts_total": 66, "facts_retained": 8, "retention_rate": 0.12121212121212122, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.17647058823529413, "status": 0.16666666666666666, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 783}, "3": {"facts_total": 66, "facts_retained": 8, "retention_rate": 0.12121212121212122, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.17647058823529413, "status": 0.16666666666666666, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 783}, "4": {"facts_total": 66, "facts_retained": 8, "retention_rate": 0.12121212121212122, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.17647058823529413, "status": 0.16666666666666666, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 783}}}, "reference_key_id": "keys/synthetic/syn_03024_2000.json", "task_scores": {"correct": 1.0, "abstained": 1.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "3cb4552bd39acfb9b1ca04c5d04cccae183051dc06ba450042dd9b360768202c", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "8fc64bc3cad06c9588215dad", "attempt_id": "5e8695b9cc3d", "run_group_id": "E4_core:A5|repairable|qwen3b:h8", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:26:42.448490Z", "task_family": "synthetic", "document_id": "syn_03024_2000", "document_hash": "ea10cd11ad08f01c50b0712bc481668857c9d2ddc8e6423968f2a31eeb97ac55", "question_id": "q00_single_fact", "split": "test", "cluster_id": "syn_03024_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 8, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 2068, "query_tokens": 449, "message_tokens": 367, "message_utf8_bytes": 847, "cache_tensor_bytes": null, "serialized_payload_bytes": 847, "side_information_bytes": 0, "fetched_bytes": 4074, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 39903.3381229965, "hops": {"1": 6938.470332999714, "2": 4700.964290997945, "3": 4701.423165970482, "4": 4716.0387500189245, "5": 4711.475750082172, "6": 4709.1882498934865, "7": 4716.373875038698, "8": 4709.4037079950795}, "receiver_prefill_ms": 164.31158292107284, "ttft_ms": 11.830041999928653, "decode_ms": 118.30420908518136}, "total_wall_ms": 40198.22749809828, "ttft_ms": 11.830041999928653, "input_usage": {"lookups": 7}, "output_usage": {"output_tokens": 10}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "BU-55G-3254", "evidence_artifact": {"ledger": {"facts_total": 66, "facts_retained": 8, "retention_rate": 0.12121212121212122, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.17647058823529413, "status": 0.16666666666666666, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 783}, "ledger_by_hop": {"1": {"facts_total": 66, "facts_retained": 6, "retention_rate": 0.09090909090909091, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.17647058823529413, "status": 0.0, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 754}, "2": {"facts_total": 66, "facts_retained": 8, "retention_rate": 0.12121212121212122, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.17647058823529413, "status": 0.16666666666666666, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 783}, "3": {"facts_total": 66, "facts_retained": 8, "retention_rate": 0.12121212121212122, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.17647058823529413, "status": 0.16666666666666666, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 783}, "4": {"facts_total": 66, "facts_retained": 8, "retention_rate": 0.12121212121212122, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.17647058823529413, "status": 0.16666666666666666, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 783}, "5": {"facts_total": 66, "facts_retained": 8, "retention_rate": 0.12121212121212122, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.17647058823529413, "status": 0.16666666666666666, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 783}, "6": {"facts_total": 66, "facts_retained": 8, "retention_rate": 0.12121212121212122, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.17647058823529413, "status": 0.16666666666666666, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 783}, "7": {"facts_total": 66, "facts_retained": 8, "retention_rate": 0.12121212121212122, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.17647058823529413, "status": 0.16666666666666666, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 783}, "8": {"facts_total": 66, "facts_retained": 8, "retention_rate": 0.12121212121212122, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.17647058823529413, "status": 0.16666666666666666, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 783}}}, "reference_key_id": "keys/synthetic/syn_03024_2000.json", "task_scores": {"correct": 0.0, "abstained": 0.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "3cb4552bd39acfb9b1ca04c5d04cccae183051dc06ba450042dd9b360768202c", "output_hash": "98227e32c2121a7856e6d2b32b74598ce6f4183c2ec3eec7ce4b9b0b3f66b98b"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "4ca4383ae11ed99ab4be24fd", "attempt_id": "9f3af4dbc6ca", "run_group_id": "E4_core:A5|repairable|qwen3b:h8", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:26:42.759966Z", "task_family": "synthetic", "document_id": "syn_03024_2000", "document_hash": "ea10cd11ad08f01c50b0712bc481668857c9d2ddc8e6423968f2a31eeb97ac55", "question_id": "q01_single_fact", "split": "test", "cluster_id": "syn_03024_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 8, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 2068, "query_tokens": 454, "message_tokens": 367, "message_utf8_bytes": 847, "cache_tensor_bytes": null, "serialized_payload_bytes": 847, "side_information_bytes": 0, "fetched_bytes": 4074, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 39903.3381229965, "hops": {"1": 6938.470332999714, "2": 4700.964290997945, "3": 4701.423165970482, "4": 4716.0387500189245, "5": 4711.475750082172, "6": 4709.1882498934865, "7": 4716.373875038698, "8": 4709.4037079950795}, "receiver_prefill_ms": 185.30508305411786, "ttft_ms": 11.870500049553812, "decode_ms": 11.98899990413338}, "total_wall_ms": 40112.903831060976, "ttft_ms": 11.870500049553812, "input_usage": {"lookups": 7}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 66, "facts_retained": 8, "retention_rate": 0.12121212121212122, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.17647058823529413, "status": 0.16666666666666666, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 783}, "ledger_by_hop": {"1": {"facts_total": 66, "facts_retained": 6, "retention_rate": 0.09090909090909091, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.17647058823529413, "status": 0.0, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 754}, "2": {"facts_total": 66, "facts_retained": 8, "retention_rate": 0.12121212121212122, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.17647058823529413, "status": 0.16666666666666666, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 783}, "3": {"facts_total": 66, "facts_retained": 8, "retention_rate": 0.12121212121212122, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.17647058823529413, "status": 0.16666666666666666, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 783}, "4": {"facts_total": 66, "facts_retained": 8, "retention_rate": 0.12121212121212122, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.17647058823529413, "status": 0.16666666666666666, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 783}, "5": {"facts_total": 66, "facts_retained": 8, "retention_rate": 0.12121212121212122, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.17647058823529413, "status": 0.16666666666666666, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 783}, "6": {"facts_total": 66, "facts_retained": 8, "retention_rate": 0.12121212121212122, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.17647058823529413, "status": 0.16666666666666666, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 783}, "7": {"facts_total": 66, "facts_retained": 8, "retention_rate": 0.12121212121212122, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.17647058823529413, "status": 0.16666666666666666, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 783}, "8": {"facts_total": 66, "facts_retained": 8, "retention_rate": 0.12121212121212122, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.17647058823529413, "status": 0.16666666666666666, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 783}}}, "reference_key_id": "keys/synthetic/syn_03024_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "3cb4552bd39acfb9b1ca04c5d04cccae183051dc06ba450042dd9b360768202c", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "d9af521fe603265dcd61405b", "attempt_id": "a4e31c2666a5", "run_group_id": "E4_core:A5|repairable|qwen3b:h8", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:26:42.985024Z", "task_family": "synthetic", "document_id": "syn_03024_2000", "document_hash": "ea10cd11ad08f01c50b0712bc481668857c9d2ddc8e6423968f2a31eeb97ac55", "question_id": "q02_exact_copy", "split": "test", "cluster_id": "syn_03024_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 8, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 2068, "query_tokens": 454, "message_tokens": 367, "message_utf8_bytes": 847, "cache_tensor_bytes": null, "serialized_payload_bytes": 847, "side_information_bytes": 0, "fetched_bytes": 4074, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 39903.3381229965, "hops": {"1": 6938.470332999714, "2": 4700.964290997945, "3": 4701.423165970482, "4": 4716.0387500189245, "5": 4711.475750082172, "6": 4709.1882498934865, "7": 4716.373875038698, "8": 4709.4037079950795}, "receiver_prefill_ms": 185.3665839880705, "ttft_ms": 11.911417008377612, "decode_ms": 12.037124950438738}, "total_wall_ms": 40113.028290099464, "ttft_ms": 11.911417008377612, "input_usage": {"lookups": 7}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 66, "facts_retained": 8, "retention_rate": 0.12121212121212122, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.17647058823529413, "status": 0.16666666666666666, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 783}, "ledger_by_hop": {"1": {"facts_total": 66, "facts_retained": 6, "retention_rate": 0.09090909090909091, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.17647058823529413, "status": 0.0, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 754}, "2": {"facts_total": 66, "facts_retained": 8, "retention_rate": 0.12121212121212122, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.17647058823529413, "status": 0.16666666666666666, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 783}, "3": {"facts_total": 66, "facts_retained": 8, "retention_rate": 0.12121212121212122, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.17647058823529413, "status": 0.16666666666666666, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 783}, "4": {"facts_total": 66, "facts_retained": 8, "retention_rate": 0.12121212121212122, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.17647058823529413, "status": 0.16666666666666666, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 783}, "5": {"facts_total": 66, "facts_retained": 8, "retention_rate": 0.12121212121212122, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.17647058823529413, "status": 0.16666666666666666, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 783}, "6": {"facts_total": 66, "facts_retained": 8, "retention_rate": 0.12121212121212122, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.17647058823529413, "status": 0.16666666666666666, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 783}, "7": {"facts_total": 66, "facts_retained": 8, "retention_rate": 0.12121212121212122, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.17647058823529413, "status": 0.16666666666666666, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 783}, "8": {"facts_total": 66, "facts_retained": 8, "retention_rate": 0.12121212121212122, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.17647058823529413, "status": 0.16666666666666666, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 783}}}, "reference_key_id": "keys/synthetic/syn_03024_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "3cb4552bd39acfb9b1ca04c5d04cccae183051dc06ba450042dd9b360768202c", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "d644cd219138f195496a6d51", "attempt_id": "8366d6394f45", "run_group_id": "E4_core:A5|repairable|qwen3b:h8", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:26:43.210258Z", "task_family": "synthetic", "document_id": "syn_03024_2000", "document_hash": "ea10cd11ad08f01c50b0712bc481668857c9d2ddc8e6423968f2a31eeb97ac55", "question_id": "q03_exact_copy", "split": "test", "cluster_id": "syn_03024_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 8, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 2068, "query_tokens": 449, "message_tokens": 367, "message_utf8_bytes": 847, "cache_tensor_bytes": null, "serialized_payload_bytes": 847, "side_information_bytes": 0, "fetched_bytes": 4074, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 39903.3381229965, "hops": {"1": 6938.470332999714, "2": 4700.964290997945, "3": 4701.423165970482, "4": 4716.0387500189245, "5": 4711.475750082172, "6": 4709.1882498934865, "7": 4716.373875038698, "8": 4709.4037079950795}, "receiver_prefill_ms": 164.26441702060401, "ttft_ms": 11.79575000423938, "decode_ms": 11.690332903526723}, "total_wall_ms": 40091.45541500766, "ttft_ms": 11.79575000423938, "input_usage": {"lookups": 7}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 66, "facts_retained": 8, "retention_rate": 0.12121212121212122, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.17647058823529413, "status": 0.16666666666666666, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 783}, "ledger_by_hop": {"1": {"facts_total": 66, "facts_retained": 6, "retention_rate": 0.09090909090909091, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.17647058823529413, "status": 0.0, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 754}, "2": {"facts_total": 66, "facts_retained": 8, "retention_rate": 0.12121212121212122, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.17647058823529413, "status": 0.16666666666666666, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 783}, "3": {"facts_total": 66, "facts_retained": 8, "retention_rate": 0.12121212121212122, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.17647058823529413, "status": 0.16666666666666666, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 783}, "4": {"facts_total": 66, "facts_retained": 8, "retention_rate": 0.12121212121212122, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.17647058823529413, "status": 0.16666666666666666, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 783}, "5": {"facts_total": 66, "facts_retained": 8, "retention_rate": 0.12121212121212122, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.17647058823529413, "status": 0.16666666666666666, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 783}, "6": {"facts_total": 66, "facts_retained": 8, "retention_rate": 0.12121212121212122, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.17647058823529413, "status": 0.16666666666666666, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 783}, "7": {"facts_total": 66, "facts_retained": 8, "retention_rate": 0.12121212121212122, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.17647058823529413, "status": 0.16666666666666666, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 783}, "8": {"facts_total": 66, "facts_retained": 8, "retention_rate": 0.12121212121212122, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.17647058823529413, "status": 0.16666666666666666, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 783}}}, "reference_key_id": "keys/synthetic/syn_03024_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "3cb4552bd39acfb9b1ca04c5d04cccae183051dc06ba450042dd9b360768202c", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "967619c4f4f2ca8f1dcd00f8", "attempt_id": "d11cd46bc111", "run_group_id": "E4_core:A5|repairable|qwen3b:h8", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:26:43.413397Z", "task_family": "synthetic", "document_id": "syn_03024_2000", "document_hash": "ea10cd11ad08f01c50b0712bc481668857c9d2ddc8e6423968f2a31eeb97ac55", "question_id": "q04_multi_hop", "split": "test", "cluster_id": "syn_03024_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 8, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 2068, "query_tokens": 455, "message_tokens": 367, "message_utf8_bytes": 847, "cache_tensor_bytes": null, "serialized_payload_bytes": 847, "side_information_bytes": 0, "fetched_bytes": 4074, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 39903.3381229965, "hops": {"1": 6938.470332999714, "2": 4700.964290997945, "3": 4701.423165970482, "4": 4716.0387500189245, "5": 4711.475750082172, "6": 4709.1882498934865, "7": 4716.373875038698, "8": 4709.4037079950795}, "receiver_prefill_ms": 185.41454209480435, "ttft_ms": 11.846874956972897, "decode_ms": 11.894208961166441}, "total_wall_ms": 40112.84574808087, "ttft_ms": 11.846874956972897, "input_usage": {"lookups": 7}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 66, "facts_retained": 8, "retention_rate": 0.12121212121212122, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.17647058823529413, "status": 0.16666666666666666, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 783}, "ledger_by_hop": {"1": {"facts_total": 66, "facts_retained": 6, "retention_rate": 0.09090909090909091, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.17647058823529413, "status": 0.0, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 754}, "2": {"facts_total": 66, "facts_retained": 8, "retention_rate": 0.12121212121212122, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.17647058823529413, "status": 0.16666666666666666, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 783}, "3": {"facts_total": 66, "facts_retained": 8, "retention_rate": 0.12121212121212122, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.17647058823529413, "status": 0.16666666666666666, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 783}, "4": {"facts_total": 66, "facts_retained": 8, "retention_rate": 0.12121212121212122, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.17647058823529413, "status": 0.16666666666666666, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 783}, "5": {"facts_total": 66, "facts_retained": 8, "retention_rate": 0.12121212121212122, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.17647058823529413, "status": 0.16666666666666666, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 783}, "6": {"facts_total": 66, "facts_retained": 8, "retention_rate": 0.12121212121212122, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.17647058823529413, "status": 0.16666666666666666, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 783}, "7": {"facts_total": 66, "facts_retained": 8, "retention_rate": 0.12121212121212122, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.17647058823529413, "status": 0.16666666666666666, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 783}, "8": {"facts_total": 66, "facts_retained": 8, "retention_rate": 0.12121212121212122, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.17647058823529413, "status": 0.16666666666666666, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 783}}}, "reference_key_id": "keys/synthetic/syn_03024_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "3cb4552bd39acfb9b1ca04c5d04cccae183051dc06ba450042dd9b360768202c", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "5831b593dd59b0101bd7e0ba", "attempt_id": "1466ef0c5242", "run_group_id": "E4_core:A5|repairable|qwen3b:h8", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:26:43.638039Z", "task_family": "synthetic", "document_id": "syn_03024_2000", "document_hash": "ea10cd11ad08f01c50b0712bc481668857c9d2ddc8e6423968f2a31eeb97ac55", "question_id": "q05_multi_hop", "split": "test", "cluster_id": "syn_03024_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 8, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 2068, "query_tokens": 459, "message_tokens": 367, "message_utf8_bytes": 847, "cache_tensor_bytes": null, "serialized_payload_bytes": 847, "side_information_bytes": 0, "fetched_bytes": 4074, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 39903.3381229965, "hops": {"1": 6938.470332999714, "2": 4700.964290997945, "3": 4701.423165970482, "4": 4716.0387500189245, "5": 4711.475750082172, "6": 4709.1882498934865, "7": 4716.373875038698, "8": 4709.4037079950795}, "receiver_prefill_ms": 185.5804999358952, "ttft_ms": 11.820833082310855, "decode_ms": 11.707291007041931}, "total_wall_ms": 40112.84791503567, "ttft_ms": 11.820833082310855, "input_usage": {"lookups": 7}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 66, "facts_retained": 8, "retention_rate": 0.12121212121212122, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.17647058823529413, "status": 0.16666666666666666, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 783}, "ledger_by_hop": {"1": {"facts_total": 66, "facts_retained": 6, "retention_rate": 0.09090909090909091, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.17647058823529413, "status": 0.0, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 754}, "2": {"facts_total": 66, "facts_retained": 8, "retention_rate": 0.12121212121212122, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.17647058823529413, "status": 0.16666666666666666, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 783}, "3": {"facts_total": 66, "facts_retained": 8, "retention_rate": 0.12121212121212122, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.17647058823529413, "status": 0.16666666666666666, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 783}, "4": {"facts_total": 66, "facts_retained": 8, "retention_rate": 0.12121212121212122, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.17647058823529413, "status": 0.16666666666666666, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 783}, "5": {"facts_total": 66, "facts_retained": 8, "retention_rate": 0.12121212121212122, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.17647058823529413, "status": 0.16666666666666666, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 783}, "6": {"facts_total": 66, "facts_retained": 8, "retention_rate": 0.12121212121212122, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.17647058823529413, "status": 0.16666666666666666, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 783}, "7": {"facts_total": 66, "facts_retained": 8, "retention_rate": 0.12121212121212122, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.17647058823529413, "status": 0.16666666666666666, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 783}, "8": {"facts_total": 66, "facts_retained": 8, "retention_rate": 0.12121212121212122, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.17647058823529413, "status": 0.16666666666666666, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 783}}}, "reference_key_id": "keys/synthetic/syn_03024_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "3cb4552bd39acfb9b1ca04c5d04cccae183051dc06ba450042dd9b360768202c", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "997ad129bc8fd46175b9fbb6", "attempt_id": "b60ff342f297", "run_group_id": "E4_core:A5|repairable|qwen3b:h8", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:26:43.863054Z", "task_family": "synthetic", "document_id": "syn_03024_2000", "document_hash": "ea10cd11ad08f01c50b0712bc481668857c9d2ddc8e6423968f2a31eeb97ac55", "question_id": "q06_negation", "split": "test", "cluster_id": "syn_03024_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 8, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 2068, "query_tokens": 448, "message_tokens": 367, "message_utf8_bytes": 847, "cache_tensor_bytes": null, "serialized_payload_bytes": 847, "side_information_bytes": 0, "fetched_bytes": 4074, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 39903.3381229965, "hops": {"1": 6938.470332999714, "2": 4700.964290997945, "3": 4701.423165970482, "4": 4716.0387500189245, "5": 4711.475750082172, "6": 4709.1882498934865, "7": 4716.373875038698, "8": 4709.4037079950795}, "receiver_prefill_ms": 163.71025005355477, "ttft_ms": 12.047083000652492, "decode_ms": 11.986915953457355}, "total_wall_ms": 40091.58279001713, "ttft_ms": 12.047083000652492, "input_usage": {"lookups": 7}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "no", "evidence_artifact": {"ledger": {"facts_total": 66, "facts_retained": 8, "retention_rate": 0.12121212121212122, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.17647058823529413, "status": 0.16666666666666666, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 783}, "ledger_by_hop": {"1": {"facts_total": 66, "facts_retained": 6, "retention_rate": 0.09090909090909091, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.17647058823529413, "status": 0.0, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 754}, "2": {"facts_total": 66, "facts_retained": 8, "retention_rate": 0.12121212121212122, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.17647058823529413, "status": 0.16666666666666666, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 783}, "3": {"facts_total": 66, "facts_retained": 8, "retention_rate": 0.12121212121212122, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.17647058823529413, "status": 0.16666666666666666, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 783}, "4": {"facts_total": 66, "facts_retained": 8, "retention_rate": 0.12121212121212122, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.17647058823529413, "status": 0.16666666666666666, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 783}, "5": {"facts_total": 66, "facts_retained": 8, "retention_rate": 0.12121212121212122, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.17647058823529413, "status": 0.16666666666666666, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 783}, "6": {"facts_total": 66, "facts_retained": 8, "retention_rate": 0.12121212121212122, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.17647058823529413, "status": 0.16666666666666666, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 783}, "7": {"facts_total": 66, "facts_retained": 8, "retention_rate": 0.12121212121212122, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.17647058823529413, "status": 0.16666666666666666, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 783}, "8": {"facts_total": 66, "facts_retained": 8, "retention_rate": 0.12121212121212122, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.17647058823529413, "status": 0.16666666666666666, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 783}}}, "reference_key_id": "keys/synthetic/syn_03024_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "3cb4552bd39acfb9b1ca04c5d04cccae183051dc06ba450042dd9b360768202c", "output_hash": "9390298f3fb0c5b160498935d79cb139aef28e1c47358b4bbba61862b9c26e59"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "bc715339cd3a5b092ab0917c", "attempt_id": "132a9cf5c14a", "run_group_id": "E4_core:A5|repairable|qwen3b:h8", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:26:44.066533Z", "task_family": "synthetic", "document_id": "syn_03024_2000", "document_hash": "ea10cd11ad08f01c50b0712bc481668857c9d2ddc8e6423968f2a31eeb97ac55", "question_id": "q07_negation", "split": "test", "cluster_id": "syn_03024_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 8, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 2068, "query_tokens": 450, "message_tokens": 367, "message_utf8_bytes": 847, "cache_tensor_bytes": null, "serialized_payload_bytes": 847, "side_information_bytes": 0, "fetched_bytes": 4074, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 39903.3381229965, "hops": {"1": 6938.470332999714, "2": 4700.964290997945, "3": 4701.423165970482, "4": 4716.0387500189245, "5": 4711.475750082172, "6": 4709.1882498934865, "7": 4716.373875038698, "8": 4709.4037079950795}, "receiver_prefill_ms": 184.9652089877054, "ttft_ms": 11.876665987074375, "decode_ms": 11.847666930407286}, "total_wall_ms": 40112.41283197887, "ttft_ms": 11.876665987074375, "input_usage": {"lookups": 7}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "no", "evidence_artifact": {"ledger": {"facts_total": 66, "facts_retained": 8, "retention_rate": 0.12121212121212122, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.17647058823529413, "status": 0.16666666666666666, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 783}, "ledger_by_hop": {"1": {"facts_total": 66, "facts_retained": 6, "retention_rate": 0.09090909090909091, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.17647058823529413, "status": 0.0, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 754}, "2": {"facts_total": 66, "facts_retained": 8, "retention_rate": 0.12121212121212122, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.17647058823529413, "status": 0.16666666666666666, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 783}, "3": {"facts_total": 66, "facts_retained": 8, "retention_rate": 0.12121212121212122, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.17647058823529413, "status": 0.16666666666666666, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 783}, "4": {"facts_total": 66, "facts_retained": 8, "retention_rate": 0.12121212121212122, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.17647058823529413, "status": 0.16666666666666666, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 783}, "5": {"facts_total": 66, "facts_retained": 8, "retention_rate": 0.12121212121212122, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.17647058823529413, "status": 0.16666666666666666, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 783}, "6": {"facts_total": 66, "facts_retained": 8, "retention_rate": 0.12121212121212122, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.17647058823529413, "status": 0.16666666666666666, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 783}, "7": {"facts_total": 66, "facts_retained": 8, "retention_rate": 0.12121212121212122, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.17647058823529413, "status": 0.16666666666666666, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 783}, "8": {"facts_total": 66, "facts_retained": 8, "retention_rate": 0.12121212121212122, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.17647058823529413, "status": 0.16666666666666666, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 783}}}, "reference_key_id": "keys/synthetic/syn_03024_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "3cb4552bd39acfb9b1ca04c5d04cccae183051dc06ba450042dd9b360768202c", "output_hash": "9390298f3fb0c5b160498935d79cb139aef28e1c47358b4bbba61862b9c26e59"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "289389570fdead364afe57cd", "attempt_id": "8215859f24aa", "run_group_id": "E4_core:A5|repairable|qwen3b:h8", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:26:44.291414Z", "task_family": "synthetic", "document_id": "syn_03024_2000", "document_hash": "ea10cd11ad08f01c50b0712bc481668857c9d2ddc8e6423968f2a31eeb97ac55", "question_id": "q08_revision", "split": "test", "cluster_id": "syn_03024_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 8, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 2068, "query_tokens": 466, "message_tokens": 367, "message_utf8_bytes": 847, "cache_tensor_bytes": null, "serialized_payload_bytes": 847, "side_information_bytes": 0, "fetched_bytes": 4074, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 39903.3381229965, "hops": {"1": 6938.470332999714, "2": 4700.964290997945, "3": 4701.423165970482, "4": 4716.0387500189245, "5": 4711.475750082172, "6": 4709.1882498934865, "7": 4716.373875038698, "8": 4709.4037079950795}, "receiver_prefill_ms": 186.13395793363452, "ttft_ms": 11.890332913026214, "decode_ms": 11.867624940350652}, "total_wall_ms": 40113.636623020284, "ttft_ms": 11.890332913026214, "input_usage": {"lookups": 7}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 66, "facts_retained": 8, "retention_rate": 0.12121212121212122, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.17647058823529413, "status": 0.16666666666666666, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 783}, "ledger_by_hop": {"1": {"facts_total": 66, "facts_retained": 6, "retention_rate": 0.09090909090909091, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.17647058823529413, "status": 0.0, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 754}, "2": {"facts_total": 66, "facts_retained": 8, "retention_rate": 0.12121212121212122, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.17647058823529413, "status": 0.16666666666666666, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 783}, "3": {"facts_total": 66, "facts_retained": 8, "retention_rate": 0.12121212121212122, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.17647058823529413, "status": 0.16666666666666666, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 783}, "4": {"facts_total": 66, "facts_retained": 8, "retention_rate": 0.12121212121212122, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.17647058823529413, "status": 0.16666666666666666, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 783}, "5": {"facts_total": 66, "facts_retained": 8, "retention_rate": 0.12121212121212122, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.17647058823529413, "status": 0.16666666666666666, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 783}, "6": {"facts_total": 66, "facts_retained": 8, "retention_rate": 0.12121212121212122, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.17647058823529413, "status": 0.16666666666666666, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 783}, "7": {"facts_total": 66, "facts_retained": 8, "retention_rate": 0.12121212121212122, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.17647058823529413, "status": 0.16666666666666666, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 783}, "8": {"facts_total": 66, "facts_retained": 8, "retention_rate": 0.12121212121212122, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.17647058823529413, "status": 0.16666666666666666, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 783}}}, "reference_key_id": "keys/synthetic/syn_03024_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "3cb4552bd39acfb9b1ca04c5d04cccae183051dc06ba450042dd9b360768202c", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "1a75f72cbdf1789e1a4b0bea", "attempt_id": "7918716ca426", "run_group_id": "E4_core:A5|repairable|qwen3b:h8", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:26:44.516912Z", "task_family": "synthetic", "document_id": "syn_03024_2000", "document_hash": "ea10cd11ad08f01c50b0712bc481668857c9d2ddc8e6423968f2a31eeb97ac55", "question_id": "q09_revision", "split": "test", "cluster_id": "syn_03024_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 8, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 2068, "query_tokens": 457, "message_tokens": 367, "message_utf8_bytes": 847, "cache_tensor_bytes": null, "serialized_payload_bytes": 847, "side_information_bytes": 0, "fetched_bytes": 4074, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 39903.3381229965, "hops": {"1": 6938.470332999714, "2": 4700.964290997945, "3": 4701.423165970482, "4": 4716.0387500189245, "5": 4711.475750082172, "6": 4709.1882498934865, "7": 4716.373875038698, "8": 4709.4037079950795}, "receiver_prefill_ms": 185.54324994329363, "ttft_ms": 11.868958012200892, "decode_ms": 11.753374943509698}, "total_wall_ms": 40112.881915061735, "ttft_ms": 11.868958012200892, "input_usage": {"lookups": 7}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 66, "facts_retained": 8, "retention_rate": 0.12121212121212122, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.17647058823529413, "status": 0.16666666666666666, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 783}, "ledger_by_hop": {"1": {"facts_total": 66, "facts_retained": 6, "retention_rate": 0.09090909090909091, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.17647058823529413, "status": 0.0, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 754}, "2": {"facts_total": 66, "facts_retained": 8, "retention_rate": 0.12121212121212122, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.17647058823529413, "status": 0.16666666666666666, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 783}, "3": {"facts_total": 66, "facts_retained": 8, "retention_rate": 0.12121212121212122, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.17647058823529413, "status": 0.16666666666666666, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 783}, "4": {"facts_total": 66, "facts_retained": 8, "retention_rate": 0.12121212121212122, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.17647058823529413, "status": 0.16666666666666666, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 783}, "5": {"facts_total": 66, "facts_retained": 8, "retention_rate": 0.12121212121212122, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.17647058823529413, "status": 0.16666666666666666, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 783}, "6": {"facts_total": 66, "facts_retained": 8, "retention_rate": 0.12121212121212122, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.17647058823529413, "status": 0.16666666666666666, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 783}, "7": {"facts_total": 66, "facts_retained": 8, "retention_rate": 0.12121212121212122, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.17647058823529413, "status": 0.16666666666666666, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 783}, "8": {"facts_total": 66, "facts_retained": 8, "retention_rate": 0.12121212121212122, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.17647058823529413, "status": 0.16666666666666666, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 783}}}, "reference_key_id": "keys/synthetic/syn_03024_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "3cb4552bd39acfb9b1ca04c5d04cccae183051dc06ba450042dd9b360768202c", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "7b3c24aefef39fcd15ed4f6b", "attempt_id": "5a8d1c87703f", "run_group_id": "E4_core:A5|repairable|qwen3b:h8", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:26:44.741923Z", "task_family": "synthetic", "document_id": "syn_03024_2000", "document_hash": "ea10cd11ad08f01c50b0712bc481668857c9d2ddc8e6423968f2a31eeb97ac55", "question_id": "q10_temporal", "split": "test", "cluster_id": "syn_03024_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 8, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 2068, "query_tokens": 467, "message_tokens": 367, "message_utf8_bytes": 847, "cache_tensor_bytes": null, "serialized_payload_bytes": 847, "side_information_bytes": 0, "fetched_bytes": 4074, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 39903.3381229965, "hops": {"1": 6938.470332999714, "2": 4700.964290997945, "3": 4701.423165970482, "4": 4716.0387500189245, "5": 4711.475750082172, "6": 4709.1882498934865, "7": 4716.373875038698, "8": 4709.4037079950795}, "receiver_prefill_ms": 185.2469160221517, "ttft_ms": 11.851832969114184, "decode_ms": 11.884917039424181}, "total_wall_ms": 40112.6968319295, "ttft_ms": 11.851832969114184, "input_usage": {"lookups": 7}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 66, "facts_retained": 8, "retention_rate": 0.12121212121212122, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.17647058823529413, "status": 0.16666666666666666, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 783}, "ledger_by_hop": {"1": {"facts_total": 66, "facts_retained": 6, "retention_rate": 0.09090909090909091, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.17647058823529413, "status": 0.0, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 754}, "2": {"facts_total": 66, "facts_retained": 8, "retention_rate": 0.12121212121212122, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.17647058823529413, "status": 0.16666666666666666, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 783}, "3": {"facts_total": 66, "facts_retained": 8, "retention_rate": 0.12121212121212122, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.17647058823529413, "status": 0.16666666666666666, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 783}, "4": {"facts_total": 66, "facts_retained": 8, "retention_rate": 0.12121212121212122, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.17647058823529413, "status": 0.16666666666666666, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 783}, "5": {"facts_total": 66, "facts_retained": 8, "retention_rate": 0.12121212121212122, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.17647058823529413, "status": 0.16666666666666666, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 783}, "6": {"facts_total": 66, "facts_retained": 8, "retention_rate": 0.12121212121212122, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.17647058823529413, "status": 0.16666666666666666, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 783}, "7": {"facts_total": 66, "facts_retained": 8, "retention_rate": 0.12121212121212122, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.17647058823529413, "status": 0.16666666666666666, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 783}, "8": {"facts_total": 66, "facts_retained": 8, "retention_rate": 0.12121212121212122, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.17647058823529413, "status": 0.16666666666666666, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 783}}}, "reference_key_id": "keys/synthetic/syn_03024_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "3cb4552bd39acfb9b1ca04c5d04cccae183051dc06ba450042dd9b360768202c", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "c1ef69fa9d14f6018e17a8a5", "attempt_id": "d62e988f09e2", "run_group_id": "E4_core:A5|repairable|qwen3b:h8", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:26:44.966402Z", "task_family": "synthetic", "document_id": "syn_03024_2000", "document_hash": "ea10cd11ad08f01c50b0712bc481668857c9d2ddc8e6423968f2a31eeb97ac55", "question_id": "q11_aggregation", "split": "test", "cluster_id": "syn_03024_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 8, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 2068, "query_tokens": 460, "message_tokens": 367, "message_utf8_bytes": 847, "cache_tensor_bytes": null, "serialized_payload_bytes": 847, "side_information_bytes": 0, "fetched_bytes": 4074, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 39903.3381229965, "hops": {"1": 6938.470332999714, "2": 4700.964290997945, "3": 4701.423165970482, "4": 4716.0387500189245, "5": 4711.475750082172, "6": 4709.1882498934865, "7": 4716.373875038698, "8": 4709.4037079950795}, "receiver_prefill_ms": 185.5697090504691, "ttft_ms": 11.85687503311783, "decode_ms": 82.77320803608745}, "total_wall_ms": 40183.931623003446, "ttft_ms": 11.85687503311783, "input_usage": {"lookups": 7}, "output_usage": {"output_tokens": 7}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "8066734", "evidence_artifact": {"ledger": {"facts_total": 66, "facts_retained": 8, "retention_rate": 0.12121212121212122, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.17647058823529413, "status": 0.16666666666666666, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 783}, "ledger_by_hop": {"1": {"facts_total": 66, "facts_retained": 6, "retention_rate": 0.09090909090909091, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.17647058823529413, "status": 0.0, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 754}, "2": {"facts_total": 66, "facts_retained": 8, "retention_rate": 0.12121212121212122, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.17647058823529413, "status": 0.16666666666666666, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 783}, "3": {"facts_total": 66, "facts_retained": 8, "retention_rate": 0.12121212121212122, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.17647058823529413, "status": 0.16666666666666666, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 783}, "4": {"facts_total": 66, "facts_retained": 8, "retention_rate": 0.12121212121212122, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.17647058823529413, "status": 0.16666666666666666, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 783}, "5": {"facts_total": 66, "facts_retained": 8, "retention_rate": 0.12121212121212122, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.17647058823529413, "status": 0.16666666666666666, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 783}, "6": {"facts_total": 66, "facts_retained": 8, "retention_rate": 0.12121212121212122, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.17647058823529413, "status": 0.16666666666666666, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 783}, "7": {"facts_total": 66, "facts_retained": 8, "retention_rate": 0.12121212121212122, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.17647058823529413, "status": 0.16666666666666666, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 783}, "8": {"facts_total": 66, "facts_retained": 8, "retention_rate": 0.12121212121212122, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.17647058823529413, "status": 0.16666666666666666, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 783}}}, "reference_key_id": "keys/synthetic/syn_03024_2000.json", "task_scores": {"correct": 0.0, "abstained": 0.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "3cb4552bd39acfb9b1ca04c5d04cccae183051dc06ba450042dd9b360768202c", "output_hash": "167c75fdd6a41fca64228797ab8f3a1a3f994475246df0bffb2e060d5d2e4ff0"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "8e993e2b93f3d7d252e07ac7", "attempt_id": "9af155bbb6c5", "run_group_id": "E4_core:A5|repairable|qwen3b:h8", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:26:45.262135Z", "task_family": "synthetic", "document_id": "syn_03024_2000", "document_hash": "ea10cd11ad08f01c50b0712bc481668857c9d2ddc8e6423968f2a31eeb97ac55", "question_id": "q12_unit_conversion", "split": "test", "cluster_id": "syn_03024_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 8, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 2068, "query_tokens": 457, "message_tokens": 367, "message_utf8_bytes": 847, "cache_tensor_bytes": null, "serialized_payload_bytes": 847, "side_information_bytes": 0, "fetched_bytes": 4074, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 39903.3381229965, "hops": {"1": 6938.470332999714, "2": 4700.964290997945, "3": 4701.423165970482, "4": 4716.0387500189245, "5": 4711.475750082172, "6": 4709.1882498934865, "7": 4716.373875038698, "8": 4709.4037079950795}, "receiver_prefill_ms": 185.13104098383337, "ttft_ms": 11.781290988437831, "decode_ms": 11.864875094033778}, "total_wall_ms": 40112.47958091553, "ttft_ms": 11.781290988437831, "input_usage": {"lookups": 7}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 66, "facts_retained": 8, "retention_rate": 0.12121212121212122, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.17647058823529413, "status": 0.16666666666666666, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 783}, "ledger_by_hop": {"1": {"facts_total": 66, "facts_retained": 6, "retention_rate": 0.09090909090909091, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.17647058823529413, "status": 0.0, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 754}, "2": {"facts_total": 66, "facts_retained": 8, "retention_rate": 0.12121212121212122, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.17647058823529413, "status": 0.16666666666666666, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 783}, "3": {"facts_total": 66, "facts_retained": 8, "retention_rate": 0.12121212121212122, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.17647058823529413, "status": 0.16666666666666666, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 783}, "4": {"facts_total": 66, "facts_retained": 8, "retention_rate": 0.12121212121212122, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.17647058823529413, "status": 0.16666666666666666, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 783}, "5": {"facts_total": 66, "facts_retained": 8, "retention_rate": 0.12121212121212122, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.17647058823529413, "status": 0.16666666666666666, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 783}, "6": {"facts_total": 66, "facts_retained": 8, "retention_rate": 0.12121212121212122, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.17647058823529413, "status": 0.16666666666666666, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 783}, "7": {"facts_total": 66, "facts_retained": 8, "retention_rate": 0.12121212121212122, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.17647058823529413, "status": 0.16666666666666666, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 783}, "8": {"facts_total": 66, "facts_retained": 8, "retention_rate": 0.12121212121212122, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.17647058823529413, "status": 0.16666666666666666, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 783}}}, "reference_key_id": "keys/synthetic/syn_03024_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "3cb4552bd39acfb9b1ca04c5d04cccae183051dc06ba450042dd9b360768202c", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "c9802cdde1eaf14dab20233b", "attempt_id": "307bd52ae976", "run_group_id": "E4_core:A5|repairable|qwen3b:h8", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:26:45.486687Z", "task_family": "synthetic", "document_id": "syn_03024_2000", "document_hash": "ea10cd11ad08f01c50b0712bc481668857c9d2ddc8e6423968f2a31eeb97ac55", "question_id": "q13_unanswerable", "split": "test", "cluster_id": "syn_03024_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 8, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 2068, "query_tokens": 461, "message_tokens": 367, "message_utf8_bytes": 847, "cache_tensor_bytes": null, "serialized_payload_bytes": 847, "side_information_bytes": 0, "fetched_bytes": 4074, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 39903.3381229965, "hops": {"1": 6938.470332999714, "2": 4700.964290997945, "3": 4701.423165970482, "4": 4716.0387500189245, "5": 4711.475750082172, "6": 4709.1882498934865, "7": 4716.373875038698, "8": 4709.4037079950795}, "receiver_prefill_ms": 185.66183291841298, "ttft_ms": 11.656958027742803, "decode_ms": 11.801082990132272}, "total_wall_ms": 40112.91949788574, "ttft_ms": 11.656958027742803, "input_usage": {"lookups": 7}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 66, "facts_retained": 8, "retention_rate": 0.12121212121212122, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.17647058823529413, "status": 0.16666666666666666, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 783}, "ledger_by_hop": {"1": {"facts_total": 66, "facts_retained": 6, "retention_rate": 0.09090909090909091, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.17647058823529413, "status": 0.0, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 754}, "2": {"facts_total": 66, "facts_retained": 8, "retention_rate": 0.12121212121212122, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.17647058823529413, "status": 0.16666666666666666, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 783}, "3": {"facts_total": 66, "facts_retained": 8, "retention_rate": 0.12121212121212122, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.17647058823529413, "status": 0.16666666666666666, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 783}, "4": {"facts_total": 66, "facts_retained": 8, "retention_rate": 0.12121212121212122, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.17647058823529413, "status": 0.16666666666666666, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 783}, "5": {"facts_total": 66, "facts_retained": 8, "retention_rate": 0.12121212121212122, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.17647058823529413, "status": 0.16666666666666666, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 783}, "6": {"facts_total": 66, "facts_retained": 8, "retention_rate": 0.12121212121212122, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.17647058823529413, "status": 0.16666666666666666, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 783}, "7": {"facts_total": 66, "facts_retained": 8, "retention_rate": 0.12121212121212122, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.17647058823529413, "status": 0.16666666666666666, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 783}, "8": {"facts_total": 66, "facts_retained": 8, "retention_rate": 0.12121212121212122, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.17647058823529413, "status": 0.16666666666666666, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 783}}}, "reference_key_id": "keys/synthetic/syn_03024_2000.json", "task_scores": {"correct": 1.0, "abstained": 1.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "3cb4552bd39acfb9b1ca04c5d04cccae183051dc06ba450042dd9b360768202c", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "852f08d7cdd2d42c1f2c6935", "attempt_id": "e9a0318599ce", "run_group_id": "E4_core:A5|repairable|qwen3b:h8", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:26:45.711271Z", "task_family": "synthetic", "document_id": "syn_03024_2000", "document_hash": "ea10cd11ad08f01c50b0712bc481668857c9d2ddc8e6423968f2a31eeb97ac55", "question_id": "q14_unanswerable", "split": "test", "cluster_id": "syn_03024_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 8, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 2068, "query_tokens": 462, "message_tokens": 367, "message_utf8_bytes": 847, "cache_tensor_bytes": null, "serialized_payload_bytes": 847, "side_information_bytes": 0, "fetched_bytes": 4074, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 39903.3381229965, "hops": {"1": 6938.470332999714, "2": 4700.964290997945, "3": 4701.423165970482, "4": 4716.0387500189245, "5": 4711.475750082172, "6": 4709.1882498934865, "7": 4716.373875038698, "8": 4709.4037079950795}, "receiver_prefill_ms": 185.55879197083414, "ttft_ms": 12.172917020507157, "decode_ms": 11.823417036794126}, "total_wall_ms": 40113.387331948616, "ttft_ms": 12.172917020507157, "input_usage": {"lookups": 7}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 66, "facts_retained": 8, "retention_rate": 0.12121212121212122, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.17647058823529413, "status": 0.16666666666666666, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 783}, "ledger_by_hop": {"1": {"facts_total": 66, "facts_retained": 6, "retention_rate": 0.09090909090909091, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.17647058823529413, "status": 0.0, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 754}, "2": {"facts_total": 66, "facts_retained": 8, "retention_rate": 0.12121212121212122, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.17647058823529413, "status": 0.16666666666666666, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 783}, "3": {"facts_total": 66, "facts_retained": 8, "retention_rate": 0.12121212121212122, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.17647058823529413, "status": 0.16666666666666666, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 783}, "4": {"facts_total": 66, "facts_retained": 8, "retention_rate": 0.12121212121212122, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.17647058823529413, "status": 0.16666666666666666, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 783}, "5": {"facts_total": 66, "facts_retained": 8, "retention_rate": 0.12121212121212122, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.17647058823529413, "status": 0.16666666666666666, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 783}, "6": {"facts_total": 66, "facts_retained": 8, "retention_rate": 0.12121212121212122, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.17647058823529413, "status": 0.16666666666666666, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 783}, "7": {"facts_total": 66, "facts_retained": 8, "retention_rate": 0.12121212121212122, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.17647058823529413, "status": 0.16666666666666666, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 783}, "8": {"facts_total": 66, "facts_retained": 8, "retention_rate": 0.12121212121212122, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.17647058823529413, "status": 0.16666666666666666, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 783}}}, "reference_key_id": "keys/synthetic/syn_03024_2000.json", "task_scores": {"correct": 1.0, "abstained": 1.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "3cb4552bd39acfb9b1ca04c5d04cccae183051dc06ba450042dd9b360768202c", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "68f50df0be63c2da43fbe79a", "attempt_id": "4db60b8ac599", "run_group_id": "E4_core:A4|closed|qwen3b-llama3b:h1", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:26:52.884291Z", "task_family": "synthetic", "document_id": "syn_03024_2000", "document_hash": "ea10cd11ad08f01c50b0712bc481668857c9d2ddc8e6423968f2a31eeb97ac55", "question_id": "q00_single_fact", "split": "test", "cluster_id": "syn_03024_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 1, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 2068, "query_tokens": 595, "message_tokens": 512, "message_utf8_bytes": 1508, "cache_tensor_bytes": null, "serialized_payload_bytes": 1508, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 6944.753459072672, "hops": {"1": 6944.753459072672}, "receiver_prefill_ms": 227.1524170646444, "ttft_ms": 11.938499985262752, "decode_ms": 11.892291018739343}, "total_wall_ms": 7196.284751058556, "ttft_ms": 11.938499985262752, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 66, "facts_retained": 21, "retention_rate": 0.3181818181818182, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.35294117647058826, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1508}, "ledger_by_hop": {"1": {"facts_total": 66, "facts_retained": 21, "retention_rate": 0.3181818181818182, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.35294117647058826, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1508}}}, "reference_key_id": "keys/synthetic/syn_03024_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "34fcc1e23e5a866b5b7d7e222f0380ba771c4005d3732e9747f026c211840d43", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "e11e0d2619d789e65d5058f3", "attempt_id": "6cfd25bdbfc0", "run_group_id": "E4_core:A4|closed|qwen3b-llama3b:h1", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:26:53.152589Z", "task_family": "synthetic", "document_id": "syn_03024_2000", "document_hash": "ea10cd11ad08f01c50b0712bc481668857c9d2ddc8e6423968f2a31eeb97ac55", "question_id": "q01_single_fact", "split": "test", "cluster_id": "syn_03024_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 1, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 2068, "query_tokens": 600, "message_tokens": 512, "message_utf8_bytes": 1508, "cache_tensor_bytes": null, "serialized_payload_bytes": 1508, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 6944.753459072672, "hops": {"1": 6944.753459072672}, "receiver_prefill_ms": 227.1518340567127, "ttft_ms": 11.90512499306351, "decode_ms": 11.911417008377612}, "total_wall_ms": 7196.398084051907, "ttft_ms": 11.90512499306351, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 66, "facts_retained": 21, "retention_rate": 0.3181818181818182, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.35294117647058826, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1508}, "ledger_by_hop": {"1": {"facts_total": 66, "facts_retained": 21, "retention_rate": 0.3181818181818182, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.35294117647058826, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1508}}}, "reference_key_id": "keys/synthetic/syn_03024_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "34fcc1e23e5a866b5b7d7e222f0380ba771c4005d3732e9747f026c211840d43", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "fae165009be3b7cb6f9d6d32", "attempt_id": "219d24d19ec8", "run_group_id": "E4_core:A4|closed|qwen3b-llama3b:h1", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:26:53.420308Z", "task_family": "synthetic", "document_id": "syn_03024_2000", "document_hash": "ea10cd11ad08f01c50b0712bc481668857c9d2ddc8e6423968f2a31eeb97ac55", "question_id": "q02_exact_copy", "split": "test", "cluster_id": "syn_03024_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 1, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 2068, "query_tokens": 600, "message_tokens": 512, "message_utf8_bytes": 1508, "cache_tensor_bytes": null, "serialized_payload_bytes": 1508, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 6944.753459072672, "hops": {"1": 6944.753459072672}, "receiver_prefill_ms": 227.42758295498788, "ttft_ms": 11.96104206610471, "decode_ms": 83.00912496633828}, "total_wall_ms": 7267.676292103715, "ttft_ms": 11.96104206610471, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 7}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "30.0224", "evidence_artifact": {"ledger": {"facts_total": 66, "facts_retained": 21, "retention_rate": 0.3181818181818182, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.35294117647058826, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1508}, "ledger_by_hop": {"1": {"facts_total": 66, "facts_retained": 21, "retention_rate": 0.3181818181818182, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.35294117647058826, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1508}}}, "reference_key_id": "keys/synthetic/syn_03024_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "34fcc1e23e5a866b5b7d7e222f0380ba771c4005d3732e9747f026c211840d43", "output_hash": "e28a78bc4deb3cec8beab4bf5b87297d7cc55a3e6f839b43fe6408735d464923"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "02dc215c8204ba577fbbea51", "attempt_id": "857f65976221", "run_group_id": "E4_core:A4|closed|qwen3b-llama3b:h1", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:26:53.759086Z", "task_family": "synthetic", "document_id": "syn_03024_2000", "document_hash": "ea10cd11ad08f01c50b0712bc481668857c9d2ddc8e6423968f2a31eeb97ac55", "question_id": "q03_exact_copy", "split": "test", "cluster_id": "syn_03024_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 1, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 2068, "query_tokens": 595, "message_tokens": 512, "message_utf8_bytes": 1508, "cache_tensor_bytes": null, "serialized_payload_bytes": 1508, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 6944.753459072672, "hops": {"1": 6944.753459072672}, "receiver_prefill_ms": 226.56595904845744, "ttft_ms": 12.16012507211417, "decode_ms": 11.987749952822924}, "total_wall_ms": 7196.175834047608, "ttft_ms": 12.16012507211417, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 66, "facts_retained": 21, "retention_rate": 0.3181818181818182, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.35294117647058826, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1508}, "ledger_by_hop": {"1": {"facts_total": 66, "facts_retained": 21, "retention_rate": 0.3181818181818182, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.35294117647058826, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1508}}}, "reference_key_id": "keys/synthetic/syn_03024_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "34fcc1e23e5a866b5b7d7e222f0380ba771c4005d3732e9747f026c211840d43", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "8c511232ddd04769672c226c", "attempt_id": "68804eda82b6", "run_group_id": "E4_core:A4|closed|qwen3b-llama3b:h1", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:26:54.026161Z", "task_family": "synthetic", "document_id": "syn_03024_2000", "document_hash": "ea10cd11ad08f01c50b0712bc481668857c9d2ddc8e6423968f2a31eeb97ac55", "question_id": "q04_multi_hop", "split": "test", "cluster_id": "syn_03024_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 1, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 2068, "query_tokens": 601, "message_tokens": 512, "message_utf8_bytes": 1508, "cache_tensor_bytes": null, "serialized_payload_bytes": 1508, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 6944.753459072672, "hops": {"1": 6944.753459072672}, "receiver_prefill_ms": 227.2182919550687, "ttft_ms": 11.998250032775104, "decode_ms": 106.68470803648233}, "total_wall_ms": 7291.227417066693, "ttft_ms": 11.998250032775104, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 9}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "6,009,480", "evidence_artifact": {"ledger": {"facts_total": 66, "facts_retained": 21, "retention_rate": 0.3181818181818182, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.35294117647058826, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1508}, "ledger_by_hop": {"1": {"facts_total": 66, "facts_retained": 21, "retention_rate": 0.3181818181818182, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.35294117647058826, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1508}}}, "reference_key_id": "keys/synthetic/syn_03024_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "34fcc1e23e5a866b5b7d7e222f0380ba771c4005d3732e9747f026c211840d43", "output_hash": "30b20d9aae4d7b0587edeefdc2a9a159f73faf6e2570f1d408a415fe92cbf45e"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "00ac6ff047f1da2dba0f40f9", "attempt_id": "c52537b35467", "run_group_id": "E4_core:A4|closed|qwen3b-llama3b:h1", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:26:54.388737Z", "task_family": "synthetic", "document_id": "syn_03024_2000", "document_hash": "ea10cd11ad08f01c50b0712bc481668857c9d2ddc8e6423968f2a31eeb97ac55", "question_id": "q05_multi_hop", "split": "test", "cluster_id": "syn_03024_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 1, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 2068, "query_tokens": 605, "message_tokens": 512, "message_utf8_bytes": 1508, "cache_tensor_bytes": null, "serialized_payload_bytes": 1508, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 6944.753459072672, "hops": {"1": 6944.753459072672}, "receiver_prefill_ms": 227.1752500673756, "ttft_ms": 12.066584080457687, "decode_ms": 12.00175005942583}, "total_wall_ms": 7196.650042082183, "ttft_ms": 12.066584080457687, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 66, "facts_retained": 21, "retention_rate": 0.3181818181818182, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.35294117647058826, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1508}, "ledger_by_hop": {"1": {"facts_total": 66, "facts_retained": 21, "retention_rate": 0.3181818181818182, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.35294117647058826, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1508}}}, "reference_key_id": "keys/synthetic/syn_03024_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "34fcc1e23e5a866b5b7d7e222f0380ba771c4005d3732e9747f026c211840d43", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "39c502bfca2daed29daa04dc", "attempt_id": "7506e57ceb8b", "run_group_id": "E4_core:A4|closed|qwen3b-llama3b:h1", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:26:54.657417Z", "task_family": "synthetic", "document_id": "syn_03024_2000", "document_hash": "ea10cd11ad08f01c50b0712bc481668857c9d2ddc8e6423968f2a31eeb97ac55", "question_id": "q06_negation", "split": "test", "cluster_id": "syn_03024_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 1, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 2068, "query_tokens": 594, "message_tokens": 512, "message_utf8_bytes": 1508, "cache_tensor_bytes": null, "serialized_payload_bytes": 1508, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 6944.753459072672, "hops": {"1": 6944.753459072672}, "receiver_prefill_ms": 227.0451249787584, "ttft_ms": 11.883500032126904, "decode_ms": 11.969916056841612}, "total_wall_ms": 7196.165043045767, "ttft_ms": 11.883500032126904, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "No", "evidence_artifact": {"ledger": {"facts_total": 66, "facts_retained": 21, "retention_rate": 0.3181818181818182, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.35294117647058826, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1508}, "ledger_by_hop": {"1": {"facts_total": 66, "facts_retained": 21, "retention_rate": 0.3181818181818182, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.35294117647058826, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1508}}}, "reference_key_id": "keys/synthetic/syn_03024_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "34fcc1e23e5a866b5b7d7e222f0380ba771c4005d3732e9747f026c211840d43", "output_hash": "1ea442a134b2a184bd5d40104401f2a37fbc09ccf3f4bc9da161c6099be3691d"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "406514a3b1fcdf8c764b2cf5", "attempt_id": "e203d9c2cdda", "run_group_id": "E4_core:A4|closed|qwen3b-llama3b:h1", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:26:54.925033Z", "task_family": "synthetic", "document_id": "syn_03024_2000", "document_hash": "ea10cd11ad08f01c50b0712bc481668857c9d2ddc8e6423968f2a31eeb97ac55", "question_id": "q07_negation", "split": "test", "cluster_id": "syn_03024_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 1, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 2068, "query_tokens": 596, "message_tokens": 512, "message_utf8_bytes": 1508, "cache_tensor_bytes": null, "serialized_payload_bytes": 1508, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 6944.753459072672, "hops": {"1": 6944.753459072672}, "receiver_prefill_ms": 227.03204106073827, "ttft_ms": 11.973832966759801, "decode_ms": 11.963958968408406}, "total_wall_ms": 7196.251251152717, "ttft_ms": 11.973832966759801, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "no", "evidence_artifact": {"ledger": {"facts_total": 66, "facts_retained": 21, "retention_rate": 0.3181818181818182, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.35294117647058826, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1508}, "ledger_by_hop": {"1": {"facts_total": 66, "facts_retained": 21, "retention_rate": 0.3181818181818182, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.35294117647058826, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1508}}}, "reference_key_id": "keys/synthetic/syn_03024_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "34fcc1e23e5a866b5b7d7e222f0380ba771c4005d3732e9747f026c211840d43", "output_hash": "9390298f3fb0c5b160498935d79cb139aef28e1c47358b4bbba61862b9c26e59"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "34866c010d028255202e6678", "attempt_id": "1be1746ce157", "run_group_id": "E4_core:A4|closed|qwen3b-llama3b:h1", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:26:55.192486Z", "task_family": "synthetic", "document_id": "syn_03024_2000", "document_hash": "ea10cd11ad08f01c50b0712bc481668857c9d2ddc8e6423968f2a31eeb97ac55", "question_id": "q08_revision", "split": "test", "cluster_id": "syn_03024_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 1, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 2068, "query_tokens": 612, "message_tokens": 512, "message_utf8_bytes": 1508, "cache_tensor_bytes": null, "serialized_payload_bytes": 1508, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 6944.753459072672, "hops": {"1": 6944.753459072672}, "receiver_prefill_ms": 229.4199160533026, "ttft_ms": 11.845832923427224, "decode_ms": 11.970583000220358}, "total_wall_ms": 7198.51283414755, "ttft_ms": 11.845832923427224, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 66, "facts_retained": 21, "retention_rate": 0.3181818181818182, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.35294117647058826, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1508}, "ledger_by_hop": {"1": {"facts_total": 66, "facts_retained": 21, "retention_rate": 0.3181818181818182, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.35294117647058826, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1508}}}, "reference_key_id": "keys/synthetic/syn_03024_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "34fcc1e23e5a866b5b7d7e222f0380ba771c4005d3732e9747f026c211840d43", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "385e7e889a365c58210ad056", "attempt_id": "2f0b0c70941d", "run_group_id": "E4_core:A4|closed|qwen3b-llama3b:h1", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:26:55.461693Z", "task_family": "synthetic", "document_id": "syn_03024_2000", "document_hash": "ea10cd11ad08f01c50b0712bc481668857c9d2ddc8e6423968f2a31eeb97ac55", "question_id": "q09_revision", "split": "test", "cluster_id": "syn_03024_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 1, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 2068, "query_tokens": 603, "message_tokens": 512, "message_utf8_bytes": 1508, "cache_tensor_bytes": null, "serialized_payload_bytes": 1508, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 6944.753459072672, "hops": {"1": 6944.753459072672}, "receiver_prefill_ms": 227.41287492681295, "ttft_ms": 11.94062503054738, "decode_ms": 11.902958038263023}, "total_wall_ms": 7196.541083976626, "ttft_ms": 11.94062503054738, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 66, "facts_retained": 21, "retention_rate": 0.3181818181818182, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.35294117647058826, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1508}, "ledger_by_hop": {"1": {"facts_total": 66, "facts_retained": 21, "retention_rate": 0.3181818181818182, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.35294117647058826, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1508}}}, "reference_key_id": "keys/synthetic/syn_03024_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "34fcc1e23e5a866b5b7d7e222f0380ba771c4005d3732e9747f026c211840d43", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "babe670b47b3449042e1dba4", "attempt_id": "c0a4dd05c363", "run_group_id": "E4_core:A4|closed|qwen3b-llama3b:h1", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:26:55.728948Z", "task_family": "synthetic", "document_id": "syn_03024_2000", "document_hash": "ea10cd11ad08f01c50b0712bc481668857c9d2ddc8e6423968f2a31eeb97ac55", "question_id": "q10_temporal", "split": "test", "cluster_id": "syn_03024_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 1, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 2068, "query_tokens": 613, "message_tokens": 512, "message_utf8_bytes": 1508, "cache_tensor_bytes": null, "serialized_payload_bytes": 1508, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 6944.753459072672, "hops": {"1": 6944.753459072672}, "receiver_prefill_ms": 229.5377500122413, "ttft_ms": 11.927249957807362, "decode_ms": 12.015708023682237}, "total_wall_ms": 7198.722709086724, "ttft_ms": 11.927249957807362, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 66, "facts_retained": 21, "retention_rate": 0.3181818181818182, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.35294117647058826, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1508}, "ledger_by_hop": {"1": {"facts_total": 66, "facts_retained": 21, "retention_rate": 0.3181818181818182, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.35294117647058826, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1508}}}, "reference_key_id": "keys/synthetic/syn_03024_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "34fcc1e23e5a866b5b7d7e222f0380ba771c4005d3732e9747f026c211840d43", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "0dafcf322c7b111ebba73e59", "attempt_id": "9c044929c5b6", "run_group_id": "E4_core:A4|closed|qwen3b-llama3b:h1", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:26:55.998335Z", "task_family": "synthetic", "document_id": "syn_03024_2000", "document_hash": "ea10cd11ad08f01c50b0712bc481668857c9d2ddc8e6423968f2a31eeb97ac55", "question_id": "q11_aggregation", "split": "test", "cluster_id": "syn_03024_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 1, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 2068, "query_tokens": 606, "message_tokens": 512, "message_utf8_bytes": 1508, "cache_tensor_bytes": null, "serialized_payload_bytes": 1508, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 6944.753459072672, "hops": {"1": 6944.753459072672}, "receiver_prefill_ms": 227.59179095737636, "ttft_ms": 11.944000027142465, "decode_ms": 83.36770895402879}, "total_wall_ms": 7268.157125101425, "ttft_ms": 11.944000027142465, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 7}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "2707654", "evidence_artifact": {"ledger": {"facts_total": 66, "facts_retained": 21, "retention_rate": 0.3181818181818182, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.35294117647058826, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1508}, "ledger_by_hop": {"1": {"facts_total": 66, "facts_retained": 21, "retention_rate": 0.3181818181818182, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.35294117647058826, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1508}}}, "reference_key_id": "keys/synthetic/syn_03024_2000.json", "task_scores": {"correct": 0.0, "abstained": 0.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "34fcc1e23e5a866b5b7d7e222f0380ba771c4005d3732e9747f026c211840d43", "output_hash": "7799e805760fa6568574360f61d4c7cc9e162842191ff434dfed71076922855e"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "6ce601a2052216f87b89273b", "attempt_id": "359803814e4a", "run_group_id": "E4_core:A4|closed|qwen3b-llama3b:h1", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:26:56.336875Z", "task_family": "synthetic", "document_id": "syn_03024_2000", "document_hash": "ea10cd11ad08f01c50b0712bc481668857c9d2ddc8e6423968f2a31eeb97ac55", "question_id": "q12_unit_conversion", "split": "test", "cluster_id": "syn_03024_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 1, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 2068, "query_tokens": 603, "message_tokens": 512, "message_utf8_bytes": 1508, "cache_tensor_bytes": null, "serialized_payload_bytes": 1508, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 6944.753459072672, "hops": {"1": 6944.753459072672}, "receiver_prefill_ms": 227.23837499506772, "ttft_ms": 12.079291976988316, "decode_ms": 11.961207957938313}, "total_wall_ms": 7196.67954207398, "ttft_ms": 12.079291976988316, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 66, "facts_retained": 21, "retention_rate": 0.3181818181818182, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.35294117647058826, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1508}, "ledger_by_hop": {"1": {"facts_total": 66, "facts_retained": 21, "retention_rate": 0.3181818181818182, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.35294117647058826, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1508}}}, "reference_key_id": "keys/synthetic/syn_03024_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "34fcc1e23e5a866b5b7d7e222f0380ba771c4005d3732e9747f026c211840d43", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "71d9edf8102fdd8fa05fdfb0", "attempt_id": "dcfff703ac93", "run_group_id": "E4_core:A4|closed|qwen3b-llama3b:h1", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:26:56.604501Z", "task_family": "synthetic", "document_id": "syn_03024_2000", "document_hash": "ea10cd11ad08f01c50b0712bc481668857c9d2ddc8e6423968f2a31eeb97ac55", "question_id": "q13_unanswerable", "split": "test", "cluster_id": "syn_03024_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 1, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 2068, "query_tokens": 607, "message_tokens": 512, "message_utf8_bytes": 1508, "cache_tensor_bytes": null, "serialized_payload_bytes": 1508, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 6944.753459072672, "hops": {"1": 6944.753459072672}, "receiver_prefill_ms": 227.17162501066923, "ttft_ms": 11.984833050519228, "decode_ms": 11.980416951701045}, "total_wall_ms": 7196.417959057726, "ttft_ms": 11.984833050519228, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 66, "facts_retained": 21, "retention_rate": 0.3181818181818182, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.35294117647058826, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1508}, "ledger_by_hop": {"1": {"facts_total": 66, "facts_retained": 21, "retention_rate": 0.3181818181818182, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.35294117647058826, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1508}}}, "reference_key_id": "keys/synthetic/syn_03024_2000.json", "task_scores": {"correct": 1.0, "abstained": 1.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "34fcc1e23e5a866b5b7d7e222f0380ba771c4005d3732e9747f026c211840d43", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "3e80711c1c4b49dbcdf9a792", "attempt_id": "4a51d726041c", "run_group_id": "E4_core:A4|closed|qwen3b-llama3b:h1", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:26:56.871886Z", "task_family": "synthetic", "document_id": "syn_03024_2000", "document_hash": "ea10cd11ad08f01c50b0712bc481668857c9d2ddc8e6423968f2a31eeb97ac55", "question_id": "q14_unanswerable", "split": "test", "cluster_id": "syn_03024_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 1, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 2068, "query_tokens": 608, "message_tokens": 512, "message_utf8_bytes": 1508, "cache_tensor_bytes": null, "serialized_payload_bytes": 1508, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 6944.753459072672, "hops": {"1": 6944.753459072672}, "receiver_prefill_ms": 226.47537500597537, "ttft_ms": 12.374832993373275, "decode_ms": 12.116165948100388}, "total_wall_ms": 7196.464334032498, "ttft_ms": 12.374832993373275, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 66, "facts_retained": 21, "retention_rate": 0.3181818181818182, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.35294117647058826, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1508}, "ledger_by_hop": {"1": {"facts_total": 66, "facts_retained": 21, "retention_rate": 0.3181818181818182, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.35294117647058826, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1508}}}, "reference_key_id": "keys/synthetic/syn_03024_2000.json", "task_scores": {"correct": 1.0, "abstained": 1.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "34fcc1e23e5a866b5b7d7e222f0380ba771c4005d3732e9747f026c211840d43", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "b4f324d6da35d9081f7dd04a", "attempt_id": "02572e4cbbc2", "run_group_id": "E4_core:A4|closed|qwen3b-llama3b:h2", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:27:02.701137Z", "task_family": "synthetic", "document_id": "syn_03024_2000", "document_hash": "ea10cd11ad08f01c50b0712bc481668857c9d2ddc8e6423968f2a31eeb97ac55", "question_id": "q00_single_fact", "split": "test", "cluster_id": "syn_03024_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "tokenizer_hashes": {"receiver": "b9d8872cf3a99da99e1fc4b0f71232a451632dc49ca4c6771549b4efa5c1c763"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 2, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1833, "query_tokens": 559, "message_tokens": 456, "message_utf8_bytes": 1579, "cache_tensor_bytes": null, "serialized_payload_bytes": 1579, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 12504.99008409679, "hops": {"1": 6944.753459072672, "2": 5560.2366250241175}, "receiver_prefill_ms": 211.97016700170934, "ttft_ms": 11.616958072409034, "decode_ms": 80.87270800024271}, "total_wall_ms": 12810.002376092598, "ttft_ms": 11.616958072409034, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 7}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "CT-24D-4130", "evidence_artifact": {"ledger": {"facts_total": 66, "facts_retained": 21, "retention_rate": 0.3181818181818182, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.35294117647058826, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1579}, "ledger_by_hop": {"1": {"facts_total": 66, "facts_retained": 21, "retention_rate": 0.3181818181818182, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.35294117647058826, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1508}, "2": {"facts_total": 66, "facts_retained": 21, "retention_rate": 0.3181818181818182, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.35294117647058826, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1579}}}, "reference_key_id": "keys/synthetic/syn_03024_2000.json", "task_scores": {"correct": 0.0, "abstained": 0.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "7105614ab75d6c721015e472329580cee0345d85cb536748ebca5e81e8dbca22", "output_hash": "ef4c922649b61c51f72c7c8c9d9c9941a9d5c51879072478dde9d6a044f0fdc5"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "49bbf7e6cd5b317d2dee3c10", "attempt_id": "6f549b7c583f", "run_group_id": "E4_core:A4|closed|qwen3b-llama3b:h2", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:27:03.022454Z", "task_family": "synthetic", "document_id": "syn_03024_2000", "document_hash": "ea10cd11ad08f01c50b0712bc481668857c9d2ddc8e6423968f2a31eeb97ac55", "question_id": "q01_single_fact", "split": "test", "cluster_id": "syn_03024_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "tokenizer_hashes": {"receiver": "b9d8872cf3a99da99e1fc4b0f71232a451632dc49ca4c6771549b4efa5c1c763"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 2, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1833, "query_tokens": 564, "message_tokens": 456, "message_utf8_bytes": 1579, "cache_tensor_bytes": null, "serialized_payload_bytes": 1579, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 12504.99008409679, "hops": {"1": 6944.753459072672, "2": 5560.2366250241175}, "receiver_prefill_ms": 212.2890829341486, "ttft_ms": 11.661082971841097, "decode_ms": 11.755208019167185}, "total_wall_ms": 12741.194042144343, "ttft_ms": 11.661082971841097, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 66, "facts_retained": 21, "retention_rate": 0.3181818181818182, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.35294117647058826, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1579}, "ledger_by_hop": {"1": {"facts_total": 66, "facts_retained": 21, "retention_rate": 0.3181818181818182, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.35294117647058826, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1508}, "2": {"facts_total": 66, "facts_retained": 21, "retention_rate": 0.3181818181818182, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.35294117647058826, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1579}}}, "reference_key_id": "keys/synthetic/syn_03024_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "7105614ab75d6c721015e472329580cee0345d85cb536748ebca5e81e8dbca22", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "97c4b9a4cae7feb9f4d996f6", "attempt_id": "99544a096fd0", "run_group_id": "E4_core:A4|closed|qwen3b-llama3b:h2", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:27:03.274132Z", "task_family": "synthetic", "document_id": "syn_03024_2000", "document_hash": "ea10cd11ad08f01c50b0712bc481668857c9d2ddc8e6423968f2a31eeb97ac55", "question_id": "q02_exact_copy", "split": "test", "cluster_id": "syn_03024_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "tokenizer_hashes": {"receiver": "b9d8872cf3a99da99e1fc4b0f71232a451632dc49ca4c6771549b4efa5c1c763"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 2, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1833, "query_tokens": 564, "message_tokens": 456, "message_utf8_bytes": 1579, "cache_tensor_bytes": null, "serialized_payload_bytes": 1579, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 12504.99008409679, "hops": {"1": 6944.753459072672, "2": 5560.2366250241175}, "receiver_prefill_ms": 211.0874579520896, "ttft_ms": 11.825416004285216, "decode_ms": 46.2122909957543}, "total_wall_ms": 12774.739751126617, "ttft_ms": 11.825416004285216, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 4}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "30.0224", "evidence_artifact": {"ledger": {"facts_total": 66, "facts_retained": 21, "retention_rate": 0.3181818181818182, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.35294117647058826, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1579}, "ledger_by_hop": {"1": {"facts_total": 66, "facts_retained": 21, "retention_rate": 0.3181818181818182, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.35294117647058826, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1508}, "2": {"facts_total": 66, "facts_retained": 21, "retention_rate": 0.3181818181818182, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.35294117647058826, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1579}}}, "reference_key_id": "keys/synthetic/syn_03024_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "7105614ab75d6c721015e472329580cee0345d85cb536748ebca5e81e8dbca22", "output_hash": "e28a78bc4deb3cec8beab4bf5b87297d7cc55a3e6f839b43fe6408735d464923"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "5487615749e711fcdc3eb6bf", "attempt_id": "04d459727032", "run_group_id": "E4_core:A4|closed|qwen3b-llama3b:h2", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:27:03.559092Z", "task_family": "synthetic", "document_id": "syn_03024_2000", "document_hash": "ea10cd11ad08f01c50b0712bc481668857c9d2ddc8e6423968f2a31eeb97ac55", "question_id": "q03_exact_copy", "split": "test", "cluster_id": "syn_03024_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "tokenizer_hashes": {"receiver": "b9d8872cf3a99da99e1fc4b0f71232a451632dc49ca4c6771549b4efa5c1c763"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 2, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1833, "query_tokens": 559, "message_tokens": 456, "message_utf8_bytes": 1579, "cache_tensor_bytes": null, "serialized_payload_bytes": 1579, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 12504.99008409679, "hops": {"1": 6944.753459072672, "2": 5560.2366250241175}, "receiver_prefill_ms": 211.9234170531854, "ttft_ms": 11.812791926786304, "decode_ms": 11.560500017367303}, "total_wall_ms": 12740.752417128533, "ttft_ms": 11.812791926786304, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 66, "facts_retained": 21, "retention_rate": 0.3181818181818182, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.35294117647058826, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1579}, "ledger_by_hop": {"1": {"facts_total": 66, "facts_retained": 21, "retention_rate": 0.3181818181818182, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.35294117647058826, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1508}, "2": {"facts_total": 66, "facts_retained": 21, "retention_rate": 0.3181818181818182, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.35294117647058826, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1579}}}, "reference_key_id": "keys/synthetic/syn_03024_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "7105614ab75d6c721015e472329580cee0345d85cb536748ebca5e81e8dbca22", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "8b26fe7c0df9a4efff0c4c74", "attempt_id": "2d7916462ea4", "run_group_id": "E4_core:A4|closed|qwen3b-llama3b:h2", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:27:03.810604Z", "task_family": "synthetic", "document_id": "syn_03024_2000", "document_hash": "ea10cd11ad08f01c50b0712bc481668857c9d2ddc8e6423968f2a31eeb97ac55", "question_id": "q04_multi_hop", "split": "test", "cluster_id": "syn_03024_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "tokenizer_hashes": {"receiver": "b9d8872cf3a99da99e1fc4b0f71232a451632dc49ca4c6771549b4efa5c1c763"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 2, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1833, "query_tokens": 566, "message_tokens": 456, "message_utf8_bytes": 1579, "cache_tensor_bytes": null, "serialized_payload_bytes": 1579, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 12504.99008409679, "hops": {"1": 6944.753459072672, "2": 5560.2366250241175}, "receiver_prefill_ms": 212.04395906534046, "ttft_ms": 11.588540975935757, "decode_ms": 57.86379100754857}, "total_wall_ms": 12786.96691710502, "ttft_ms": 11.588540975935757, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 5}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "6,009,480", "evidence_artifact": {"ledger": {"facts_total": 66, "facts_retained": 21, "retention_rate": 0.3181818181818182, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.35294117647058826, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1579}, "ledger_by_hop": {"1": {"facts_total": 66, "facts_retained": 21, "retention_rate": 0.3181818181818182, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.35294117647058826, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1508}, "2": {"facts_total": 66, "facts_retained": 21, "retention_rate": 0.3181818181818182, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.35294117647058826, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1579}}}, "reference_key_id": "keys/synthetic/syn_03024_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "7105614ab75d6c721015e472329580cee0345d85cb536748ebca5e81e8dbca22", "output_hash": "30b20d9aae4d7b0587edeefdc2a9a159f73faf6e2570f1d408a415fe92cbf45e"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "f8b9c028b74ba3f70ca5b894", "attempt_id": "5ce45c38bf17", "run_group_id": "E4_core:A4|closed|qwen3b-llama3b:h2", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:27:04.108005Z", "task_family": "synthetic", "document_id": "syn_03024_2000", "document_hash": "ea10cd11ad08f01c50b0712bc481668857c9d2ddc8e6423968f2a31eeb97ac55", "question_id": "q05_multi_hop", "split": "test", "cluster_id": "syn_03024_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "tokenizer_hashes": {"receiver": "b9d8872cf3a99da99e1fc4b0f71232a451632dc49ca4c6771549b4efa5c1c763"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 2, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1833, "query_tokens": 568, "message_tokens": 456, "message_utf8_bytes": 1579, "cache_tensor_bytes": null, "serialized_payload_bytes": 1579, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 12504.99008409679, "hops": {"1": 6944.753459072672, "2": 5560.2366250241175}, "receiver_prefill_ms": 212.4286670004949, "ttft_ms": 11.623249971307814, "decode_ms": 11.489375028759241}, "total_wall_ms": 12740.999334026128, "ttft_ms": 11.623249971307814, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 66, "facts_retained": 21, "retention_rate": 0.3181818181818182, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.35294117647058826, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1579}, "ledger_by_hop": {"1": {"facts_total": 66, "facts_retained": 21, "retention_rate": 0.3181818181818182, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.35294117647058826, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1508}, "2": {"facts_total": 66, "facts_retained": 21, "retention_rate": 0.3181818181818182, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.35294117647058826, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1579}}}, "reference_key_id": "keys/synthetic/syn_03024_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "7105614ab75d6c721015e472329580cee0345d85cb536748ebca5e81e8dbca22", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "32dc7df69e7d376f96bb6054", "attempt_id": "e4411a4ecf69", "run_group_id": "E4_core:A4|closed|qwen3b-llama3b:h2", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:27:04.359303Z", "task_family": "synthetic", "document_id": "syn_03024_2000", "document_hash": "ea10cd11ad08f01c50b0712bc481668857c9d2ddc8e6423968f2a31eeb97ac55", "question_id": "q06_negation", "split": "test", "cluster_id": "syn_03024_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "tokenizer_hashes": {"receiver": "b9d8872cf3a99da99e1fc4b0f71232a451632dc49ca4c6771549b4efa5c1c763"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 2, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1833, "query_tokens": 559, "message_tokens": 456, "message_utf8_bytes": 1579, "cache_tensor_bytes": null, "serialized_payload_bytes": 1579, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 12504.99008409679, "hops": {"1": 6944.753459072672, "2": 5560.2366250241175}, "receiver_prefill_ms": 211.47954103071243, "ttft_ms": 11.694958084262908, "decode_ms": 23.376333992928267}, "total_wall_ms": 12752.191834151745, "ttft_ms": 11.694958084262908, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 2}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "No.", "evidence_artifact": {"ledger": {"facts_total": 66, "facts_retained": 21, "retention_rate": 0.3181818181818182, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.35294117647058826, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1579}, "ledger_by_hop": {"1": {"facts_total": 66, "facts_retained": 21, "retention_rate": 0.3181818181818182, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.35294117647058826, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1508}, "2": {"facts_total": 66, "facts_retained": 21, "retention_rate": 0.3181818181818182, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.35294117647058826, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1579}}}, "reference_key_id": "keys/synthetic/syn_03024_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "7105614ab75d6c721015e472329580cee0345d85cb536748ebca5e81e8dbca22", "output_hash": "38178a20b470cfd18299fb1593dd4ba706f1a83321dac051d01086efd8b7a96f"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "f9d7762f9ac9b03675c3120e", "attempt_id": "7e3fe8f8181f", "run_group_id": "E4_core:A4|closed|qwen3b-llama3b:h2", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:27:04.621781Z", "task_family": "synthetic", "document_id": "syn_03024_2000", "document_hash": "ea10cd11ad08f01c50b0712bc481668857c9d2ddc8e6423968f2a31eeb97ac55", "question_id": "q07_negation", "split": "test", "cluster_id": "syn_03024_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "tokenizer_hashes": {"receiver": "b9d8872cf3a99da99e1fc4b0f71232a451632dc49ca4c6771549b4efa5c1c763"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 2, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1833, "query_tokens": 560, "message_tokens": 456, "message_utf8_bytes": 1579, "cache_tensor_bytes": null, "serialized_payload_bytes": 1579, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 12504.99008409679, "hops": {"1": 6944.753459072672, "2": 5560.2366250241175}, "receiver_prefill_ms": 211.97699999902397, "ttft_ms": 11.56116696074605, "decode_ms": 23.316249949857593}, "total_wall_ms": 12752.484542084858, "ttft_ms": 11.56116696074605, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 2}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "No.", "evidence_artifact": {"ledger": {"facts_total": 66, "facts_retained": 21, "retention_rate": 0.3181818181818182, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.35294117647058826, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1579}, "ledger_by_hop": {"1": {"facts_total": 66, "facts_retained": 21, "retention_rate": 0.3181818181818182, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.35294117647058826, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1508}, "2": {"facts_total": 66, "facts_retained": 21, "retention_rate": 0.3181818181818182, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.35294117647058826, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1579}}}, "reference_key_id": "keys/synthetic/syn_03024_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "7105614ab75d6c721015e472329580cee0345d85cb536748ebca5e81e8dbca22", "output_hash": "38178a20b470cfd18299fb1593dd4ba706f1a83321dac051d01086efd8b7a96f"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "649f4a52bd99f4b73bf23100", "attempt_id": "70692f9e60e7", "run_group_id": "E4_core:A4|closed|qwen3b-llama3b:h2", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:27:04.885108Z", "task_family": "synthetic", "document_id": "syn_03024_2000", "document_hash": "ea10cd11ad08f01c50b0712bc481668857c9d2ddc8e6423968f2a31eeb97ac55", "question_id": "q08_revision", "split": "test", "cluster_id": "syn_03024_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "tokenizer_hashes": {"receiver": "b9d8872cf3a99da99e1fc4b0f71232a451632dc49ca4c6771549b4efa5c1c763"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 2, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1833, "query_tokens": 576, "message_tokens": 456, "message_utf8_bytes": 1579, "cache_tensor_bytes": null, "serialized_payload_bytes": 1579, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 12504.99008409679, "hops": {"1": 6944.753459072672, "2": 5560.2366250241175}, "receiver_prefill_ms": 212.21516700461507, "ttft_ms": 12.015625019557774, "decode_ms": 11.65220804978162}, "total_wall_ms": 12741.531834122725, "ttft_ms": 12.015625019557774, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 66, "facts_retained": 21, "retention_rate": 0.3181818181818182, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.35294117647058826, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1579}, "ledger_by_hop": {"1": {"facts_total": 66, "facts_retained": 21, "retention_rate": 0.3181818181818182, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.35294117647058826, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1508}, "2": {"facts_total": 66, "facts_retained": 21, "retention_rate": 0.3181818181818182, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.35294117647058826, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1579}}}, "reference_key_id": "keys/synthetic/syn_03024_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "7105614ab75d6c721015e472329580cee0345d85cb536748ebca5e81e8dbca22", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "b43245129a659e905753ecfc", "attempt_id": "1916b8d91973", "run_group_id": "E4_core:A4|closed|qwen3b-llama3b:h2", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:27:05.137704Z", "task_family": "synthetic", "document_id": "syn_03024_2000", "document_hash": "ea10cd11ad08f01c50b0712bc481668857c9d2ddc8e6423968f2a31eeb97ac55", "question_id": "q09_revision", "split": "test", "cluster_id": "syn_03024_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "tokenizer_hashes": {"receiver": "b9d8872cf3a99da99e1fc4b0f71232a451632dc49ca4c6771549b4efa5c1c763"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 2, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1833, "query_tokens": 567, "message_tokens": 456, "message_utf8_bytes": 1579, "cache_tensor_bytes": null, "serialized_payload_bytes": 1579, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 12504.99008409679, "hops": {"1": 6944.753459072672, "2": 5560.2366250241175}, "receiver_prefill_ms": 212.25241606589407, "ttft_ms": 11.684166034683585, "decode_ms": 11.535666999407113}, "total_wall_ms": 12740.972751169465, "ttft_ms": 11.684166034683585, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "Unknown", "evidence_artifact": {"ledger": {"facts_total": 66, "facts_retained": 21, "retention_rate": 0.3181818181818182, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.35294117647058826, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1579}, "ledger_by_hop": {"1": {"facts_total": 66, "facts_retained": 21, "retention_rate": 0.3181818181818182, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.35294117647058826, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1508}, "2": {"facts_total": 66, "facts_retained": 21, "retention_rate": 0.3181818181818182, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.35294117647058826, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1579}}}, "reference_key_id": "keys/synthetic/syn_03024_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "7105614ab75d6c721015e472329580cee0345d85cb536748ebca5e81e8dbca22", "output_hash": "b764cdc0eab7137467211272fa539f1260d1bf2e71bcf6ff3bdc960f5c16aa14"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "1527f0659f51f503c5cfa7c9", "attempt_id": "cd2354957210", "run_group_id": "E4_core:A4|closed|qwen3b-llama3b:h2", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:27:05.389510Z", "task_family": "synthetic", "document_id": "syn_03024_2000", "document_hash": "ea10cd11ad08f01c50b0712bc481668857c9d2ddc8e6423968f2a31eeb97ac55", "question_id": "q10_temporal", "split": "test", "cluster_id": "syn_03024_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "tokenizer_hashes": {"receiver": "b9d8872cf3a99da99e1fc4b0f71232a451632dc49ca4c6771549b4efa5c1c763"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 2, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1833, "query_tokens": 576, "message_tokens": 456, "message_utf8_bytes": 1579, "cache_tensor_bytes": null, "serialized_payload_bytes": 1579, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 12504.99008409679, "hops": {"1": 6944.753459072672, "2": 5560.2366250241175}, "receiver_prefill_ms": 212.76787505485117, "ttft_ms": 11.658875038847327, "decode_ms": 11.655832990072668}, "total_wall_ms": 12741.532542160712, "ttft_ms": 11.658875038847327, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 66, "facts_retained": 21, "retention_rate": 0.3181818181818182, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.35294117647058826, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1579}, "ledger_by_hop": {"1": {"facts_total": 66, "facts_retained": 21, "retention_rate": 0.3181818181818182, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.35294117647058826, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1508}, "2": {"facts_total": 66, "facts_retained": 21, "retention_rate": 0.3181818181818182, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.35294117647058826, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1579}}}, "reference_key_id": "keys/synthetic/syn_03024_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "7105614ab75d6c721015e472329580cee0345d85cb536748ebca5e81e8dbca22", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "f82a48787694dc7a1a39bd4f", "attempt_id": "120e47cec02b", "run_group_id": "E4_core:A4|closed|qwen3b-llama3b:h2", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:27:05.641281Z", "task_family": "synthetic", "document_id": "syn_03024_2000", "document_hash": "ea10cd11ad08f01c50b0712bc481668857c9d2ddc8e6423968f2a31eeb97ac55", "question_id": "q11_aggregation", "split": "test", "cluster_id": "syn_03024_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "tokenizer_hashes": {"receiver": "b9d8872cf3a99da99e1fc4b0f71232a451632dc49ca4c6771549b4efa5c1c763"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 2, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1833, "query_tokens": 571, "message_tokens": 456, "message_utf8_bytes": 1579, "cache_tensor_bytes": null, "serialized_payload_bytes": 1579, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 12504.99008409679, "hops": {"1": 6944.753459072672, "2": 5560.2366250241175}, "receiver_prefill_ms": 212.16933405958116, "ttft_ms": 11.637625051662326, "decode_ms": 57.92537494562566}, "total_wall_ms": 12787.278209114447, "ttft_ms": 11.637625051662326, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 5}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "11,085,310", "evidence_artifact": {"ledger": {"facts_total": 66, "facts_retained": 21, "retention_rate": 0.3181818181818182, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.35294117647058826, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1579}, "ledger_by_hop": {"1": {"facts_total": 66, "facts_retained": 21, "retention_rate": 0.3181818181818182, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.35294117647058826, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1508}, "2": {"facts_total": 66, "facts_retained": 21, "retention_rate": 0.3181818181818182, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.35294117647058826, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1579}}}, "reference_key_id": "keys/synthetic/syn_03024_2000.json", "task_scores": {"correct": 0.0, "abstained": 0.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "7105614ab75d6c721015e472329580cee0345d85cb536748ebca5e81e8dbca22", "output_hash": "759234a32a22ec346567fe3fddc1cee453d2284580e2c63d21796dc4d97af3f0"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "0e7ed76be5dda33f02ba3dc8", "attempt_id": "c748e3d56eef", "run_group_id": "E4_core:A4|closed|qwen3b-llama3b:h2", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:27:05.938877Z", "task_family": "synthetic", "document_id": "syn_03024_2000", "document_hash": "ea10cd11ad08f01c50b0712bc481668857c9d2ddc8e6423968f2a31eeb97ac55", "question_id": "q12_unit_conversion", "split": "test", "cluster_id": "syn_03024_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "tokenizer_hashes": {"receiver": "b9d8872cf3a99da99e1fc4b0f71232a451632dc49ca4c6771549b4efa5c1c763"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 2, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1833, "query_tokens": 567, "message_tokens": 456, "message_utf8_bytes": 1579, "cache_tensor_bytes": null, "serialized_payload_bytes": 1579, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 12504.99008409679, "hops": {"1": 6944.753459072672, "2": 5560.2366250241175}, "receiver_prefill_ms": 211.69579098932445, "ttft_ms": 11.639332980848849, "decode_ms": 11.567125096917152}, "total_wall_ms": 12740.377584123053, "ttft_ms": 11.639332980848849, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 66, "facts_retained": 21, "retention_rate": 0.3181818181818182, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.35294117647058826, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1579}, "ledger_by_hop": {"1": {"facts_total": 66, "facts_retained": 21, "retention_rate": 0.3181818181818182, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.35294117647058826, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1508}, "2": {"facts_total": 66, "facts_retained": 21, "retention_rate": 0.3181818181818182, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.35294117647058826, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1579}}}, "reference_key_id": "keys/synthetic/syn_03024_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "7105614ab75d6c721015e472329580cee0345d85cb536748ebca5e81e8dbca22", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "90473dad2993e92aed0169de", "attempt_id": "b486bb8cc8e5", "run_group_id": "E4_core:A4|closed|qwen3b-llama3b:h2", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:27:06.189964Z", "task_family": "synthetic", "document_id": "syn_03024_2000", "document_hash": "ea10cd11ad08f01c50b0712bc481668857c9d2ddc8e6423968f2a31eeb97ac55", "question_id": "q13_unanswerable", "split": "test", "cluster_id": "syn_03024_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "tokenizer_hashes": {"receiver": "b9d8872cf3a99da99e1fc4b0f71232a451632dc49ca4c6771549b4efa5c1c763"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 2, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1833, "query_tokens": 571, "message_tokens": 456, "message_utf8_bytes": 1579, "cache_tensor_bytes": null, "serialized_payload_bytes": 1579, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 12504.99008409679, "hops": {"1": 6944.753459072672, "2": 5560.2366250241175}, "receiver_prefill_ms": 212.50716701615602, "ttft_ms": 11.632375069893897, "decode_ms": 11.580291902646422}, "total_wall_ms": 12741.362876142375, "ttft_ms": 11.632375069893897, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 66, "facts_retained": 21, "retention_rate": 0.3181818181818182, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.35294117647058826, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1579}, "ledger_by_hop": {"1": {"facts_total": 66, "facts_retained": 21, "retention_rate": 0.3181818181818182, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.35294117647058826, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1508}, "2": {"facts_total": 66, "facts_retained": 21, "retention_rate": 0.3181818181818182, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.35294117647058826, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1579}}}, "reference_key_id": "keys/synthetic/syn_03024_2000.json", "task_scores": {"correct": 1.0, "abstained": 1.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "7105614ab75d6c721015e472329580cee0345d85cb536748ebca5e81e8dbca22", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "80b9cd275f630db236ee1cd0", "attempt_id": "d5757000dff5", "run_group_id": "E4_core:A4|closed|qwen3b-llama3b:h2", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:27:06.441617Z", "task_family": "synthetic", "document_id": "syn_03024_2000", "document_hash": "ea10cd11ad08f01c50b0712bc481668857c9d2ddc8e6423968f2a31eeb97ac55", "question_id": "q14_unanswerable", "split": "test", "cluster_id": "syn_03024_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "tokenizer_hashes": {"receiver": "b9d8872cf3a99da99e1fc4b0f71232a451632dc49ca4c6771549b4efa5c1c763"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 2, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1833, "query_tokens": 573, "message_tokens": 456, "message_utf8_bytes": 1579, "cache_tensor_bytes": null, "serialized_payload_bytes": 1579, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 12504.99008409679, "hops": {"1": 6944.753459072672, "2": 5560.2366250241175}, "receiver_prefill_ms": 212.03270799014717, "ttft_ms": 11.574291973374784, "decode_ms": 11.784041998907924}, "total_wall_ms": 12740.904167061672, "ttft_ms": 11.574291973374784, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 66, "facts_retained": 21, "retention_rate": 0.3181818181818182, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.35294117647058826, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1579}, "ledger_by_hop": {"1": {"facts_total": 66, "facts_retained": 21, "retention_rate": 0.3181818181818182, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.35294117647058826, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1508}, "2": {"facts_total": 66, "facts_retained": 21, "retention_rate": 0.3181818181818182, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.35294117647058826, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1579}}}, "reference_key_id": "keys/synthetic/syn_03024_2000.json", "task_scores": {"correct": 1.0, "abstained": 1.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "7105614ab75d6c721015e472329580cee0345d85cb536748ebca5e81e8dbca22", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "d2cf26d2ac7fdf55d6d09cc2", "attempt_id": "e9d255dcf895", "run_group_id": "E4_core:A4|closed|qwen3b-llama3b:h4", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:27:18.110008Z", "task_family": "synthetic", "document_id": "syn_03024_2000", "document_hash": "ea10cd11ad08f01c50b0712bc481668857c9d2ddc8e6423968f2a31eeb97ac55", "question_id": "q00_single_fact", "split": "test", "cluster_id": "syn_03024_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "tokenizer_hashes": {"receiver": "b9d8872cf3a99da99e1fc4b0f71232a451632dc49ca4c6771549b4efa5c1c763"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 4, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1833, "query_tokens": 514, "message_tokens": 411, "message_utf8_bytes": 1606, "cache_tensor_bytes": null, "serialized_payload_bytes": 1606, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 23918.322250130586, "hops": {"1": 6944.753459072672, "2": 5560.2366250241175, "3": 6389.902791008353, "4": 5023.429375025444}, "receiver_prefill_ms": 209.95479193516076, "ttft_ms": 11.591459042392671, "decode_ms": 11.725915945135057}, "total_wall_ms": 24152.113250107504, "ttft_ms": 11.591459042392671, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 66, "facts_retained": 20, "retention_rate": 0.30303030303030304, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.0, "amount": 0.35294117647058826, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1606}, "ledger_by_hop": {"1": {"facts_total": 66, "facts_retained": 21, "retention_rate": 0.3181818181818182, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.35294117647058826, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1508}, "2": {"facts_total": 66, "facts_retained": 21, "retention_rate": 0.3181818181818182, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.35294117647058826, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1579}, "3": {"facts_total": 66, "facts_retained": 21, "retention_rate": 0.3181818181818182, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.35294117647058826, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1514}, "4": {"facts_total": 66, "facts_retained": 20, "retention_rate": 0.30303030303030304, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.0, "amount": 0.35294117647058826, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1606}}}, "reference_key_id": "keys/synthetic/syn_03024_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "9df441bcc254691aca50ef729117ceda7fcec91596ba108d61042763177cb3a7", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "c50016ca784113b8411cd237", "attempt_id": "a88fe76f2f5f", "run_group_id": "E4_core:A4|closed|qwen3b-llama3b:h4", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:27:18.359935Z", "task_family": "synthetic", "document_id": "syn_03024_2000", "document_hash": "ea10cd11ad08f01c50b0712bc481668857c9d2ddc8e6423968f2a31eeb97ac55", "question_id": "q01_single_fact", "split": "test", "cluster_id": "syn_03024_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "tokenizer_hashes": {"receiver": "b9d8872cf3a99da99e1fc4b0f71232a451632dc49ca4c6771549b4efa5c1c763"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 4, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1833, "query_tokens": 519, "message_tokens": 411, "message_utf8_bytes": 1606, "cache_tensor_bytes": null, "serialized_payload_bytes": 1606, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 23918.322250130586, "hops": {"1": 6944.753459072672, "2": 5560.2366250241175, "3": 6389.902791008353, "4": 5023.429375025444}, "receiver_prefill_ms": 209.8256249446422, "ttft_ms": 11.622958001680672, "decode_ms": 11.49333396460861}, "total_wall_ms": 24151.864167070016, "ttft_ms": 11.622958001680672, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 66, "facts_retained": 20, "retention_rate": 0.30303030303030304, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.0, "amount": 0.35294117647058826, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1606}, "ledger_by_hop": {"1": {"facts_total": 66, "facts_retained": 21, "retention_rate": 0.3181818181818182, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.35294117647058826, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1508}, "2": {"facts_total": 66, "facts_retained": 21, "retention_rate": 0.3181818181818182, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.35294117647058826, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1579}, "3": {"facts_total": 66, "facts_retained": 21, "retention_rate": 0.3181818181818182, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.35294117647058826, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1514}, "4": {"facts_total": 66, "facts_retained": 20, "retention_rate": 0.30303030303030304, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.0, "amount": 0.35294117647058826, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1606}}}, "reference_key_id": "keys/synthetic/syn_03024_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "9df441bcc254691aca50ef729117ceda7fcec91596ba108d61042763177cb3a7", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "e0ca350d858a89d79fd390fe", "attempt_id": "8720f9b8db35", "run_group_id": "E4_core:A4|closed|qwen3b-llama3b:h4", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:27:18.609003Z", "task_family": "synthetic", "document_id": "syn_03024_2000", "document_hash": "ea10cd11ad08f01c50b0712bc481668857c9d2ddc8e6423968f2a31eeb97ac55", "question_id": "q02_exact_copy", "split": "test", "cluster_id": "syn_03024_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "tokenizer_hashes": {"receiver": "b9d8872cf3a99da99e1fc4b0f71232a451632dc49ca4c6771549b4efa5c1c763"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 4, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1833, "query_tokens": 519, "message_tokens": 411, "message_utf8_bytes": 1606, "cache_tensor_bytes": null, "serialized_payload_bytes": 1606, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 23918.322250130586, "hops": {"1": 6944.753459072672, "2": 5560.2366250241175, "3": 6389.902791008353, "4": 5023.429375025444}, "receiver_prefill_ms": 209.50975001323968, "ttft_ms": 11.60995801910758, "decode_ms": 46.00933298934251}, "total_wall_ms": 24185.910208150744, "ttft_ms": 11.60995801910758, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 4}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "30.0224", "evidence_artifact": {"ledger": {"facts_total": 66, "facts_retained": 20, "retention_rate": 0.30303030303030304, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.0, "amount": 0.35294117647058826, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1606}, "ledger_by_hop": {"1": {"facts_total": 66, "facts_retained": 21, "retention_rate": 0.3181818181818182, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.35294117647058826, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1508}, "2": {"facts_total": 66, "facts_retained": 21, "retention_rate": 0.3181818181818182, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.35294117647058826, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1579}, "3": {"facts_total": 66, "facts_retained": 21, "retention_rate": 0.3181818181818182, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.35294117647058826, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1514}, "4": {"facts_total": 66, "facts_retained": 20, "retention_rate": 0.30303030303030304, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.0, "amount": 0.35294117647058826, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1606}}}, "reference_key_id": "keys/synthetic/syn_03024_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "9df441bcc254691aca50ef729117ceda7fcec91596ba108d61042763177cb3a7", "output_hash": "e28a78bc4deb3cec8beab4bf5b87297d7cc55a3e6f839b43fe6408735d464923"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "eaf153d40e5dd83ab2253030", "attempt_id": "38b1d9b250ce", "run_group_id": "E4_core:A4|closed|qwen3b-llama3b:h4", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:27:18.892280Z", "task_family": "synthetic", "document_id": "syn_03024_2000", "document_hash": "ea10cd11ad08f01c50b0712bc481668857c9d2ddc8e6423968f2a31eeb97ac55", "question_id": "q03_exact_copy", "split": "test", "cluster_id": "syn_03024_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "tokenizer_hashes": {"receiver": "b9d8872cf3a99da99e1fc4b0f71232a451632dc49ca4c6771549b4efa5c1c763"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 4, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1833, "query_tokens": 514, "message_tokens": 411, "message_utf8_bytes": 1606, "cache_tensor_bytes": null, "serialized_payload_bytes": 1606, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 23918.322250130586, "hops": {"1": 6944.753459072672, "2": 5560.2366250241175, "3": 6389.902791008353, "4": 5023.429375025444}, "receiver_prefill_ms": 209.6272500930354, "ttft_ms": 11.58116606529802, "decode_ms": 11.570165981538594}, "total_wall_ms": 24151.551791117527, "ttft_ms": 11.58116606529802, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 66, "facts_retained": 20, "retention_rate": 0.30303030303030304, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.0, "amount": 0.35294117647058826, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1606}, "ledger_by_hop": {"1": {"facts_total": 66, "facts_retained": 21, "retention_rate": 0.3181818181818182, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.35294117647058826, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1508}, "2": {"facts_total": 66, "facts_retained": 21, "retention_rate": 0.3181818181818182, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.35294117647058826, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1579}, "3": {"facts_total": 66, "facts_retained": 21, "retention_rate": 0.3181818181818182, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.35294117647058826, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1514}, "4": {"facts_total": 66, "facts_retained": 20, "retention_rate": 0.30303030303030304, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.0, "amount": 0.35294117647058826, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1606}}}, "reference_key_id": "keys/synthetic/syn_03024_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "9df441bcc254691aca50ef729117ceda7fcec91596ba108d61042763177cb3a7", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "4395b20aed0c24c72eb49f25", "attempt_id": "70a5f9836700", "run_group_id": "E4_core:A4|closed|qwen3b-llama3b:h4", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:27:19.140476Z", "task_family": "synthetic", "document_id": "syn_03024_2000", "document_hash": "ea10cd11ad08f01c50b0712bc481668857c9d2ddc8e6423968f2a31eeb97ac55", "question_id": "q04_multi_hop", "split": "test", "cluster_id": "syn_03024_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "tokenizer_hashes": {"receiver": "b9d8872cf3a99da99e1fc4b0f71232a451632dc49ca4c6771549b4efa5c1c763"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 4, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1833, "query_tokens": 521, "message_tokens": 411, "message_utf8_bytes": 1606, "cache_tensor_bytes": null, "serialized_payload_bytes": 1606, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 23918.322250130586, "hops": {"1": 6944.753459072672, "2": 5560.2366250241175, "3": 6389.902791008353, "4": 5023.429375025444}, "receiver_prefill_ms": 209.93962499778718, "ttft_ms": 11.500249966047704, "decode_ms": 57.71816708147526}, "total_wall_ms": 24198.06879211683, "ttft_ms": 11.500249966047704, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 5}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "6,009,480", "evidence_artifact": {"ledger": {"facts_total": 66, "facts_retained": 20, "retention_rate": 0.30303030303030304, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.0, "amount": 0.35294117647058826, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1606}, "ledger_by_hop": {"1": {"facts_total": 66, "facts_retained": 21, "retention_rate": 0.3181818181818182, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.35294117647058826, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1508}, "2": {"facts_total": 66, "facts_retained": 21, "retention_rate": 0.3181818181818182, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.35294117647058826, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1579}, "3": {"facts_total": 66, "facts_retained": 21, "retention_rate": 0.3181818181818182, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.35294117647058826, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1514}, "4": {"facts_total": 66, "facts_retained": 20, "retention_rate": 0.30303030303030304, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.0, "amount": 0.35294117647058826, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1606}}}, "reference_key_id": "keys/synthetic/syn_03024_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "9df441bcc254691aca50ef729117ceda7fcec91596ba108d61042763177cb3a7", "output_hash": "30b20d9aae4d7b0587edeefdc2a9a159f73faf6e2570f1d408a415fe92cbf45e"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "0156396c92732a83f57119da", "attempt_id": "7f2b29fe2eae", "run_group_id": "E4_core:A4|closed|qwen3b-llama3b:h4", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:27:19.435268Z", "task_family": "synthetic", "document_id": "syn_03024_2000", "document_hash": "ea10cd11ad08f01c50b0712bc481668857c9d2ddc8e6423968f2a31eeb97ac55", "question_id": "q05_multi_hop", "split": "test", "cluster_id": "syn_03024_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "tokenizer_hashes": {"receiver": "b9d8872cf3a99da99e1fc4b0f71232a451632dc49ca4c6771549b4efa5c1c763"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 4, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1833, "query_tokens": 523, "message_tokens": 411, "message_utf8_bytes": 1606, "cache_tensor_bytes": null, "serialized_payload_bytes": 1606, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 23918.322250130586, "hops": {"1": 6944.753459072672, "2": 5560.2366250241175, "3": 6389.902791008353, "4": 5023.429375025444}, "receiver_prefill_ms": 210.1938750129193, "ttft_ms": 11.526582995429635, "decode_ms": 11.78062497638166}, "total_wall_ms": 24152.32833311893, "ttft_ms": 11.526582995429635, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 66, "facts_retained": 20, "retention_rate": 0.30303030303030304, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.0, "amount": 0.35294117647058826, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1606}, "ledger_by_hop": {"1": {"facts_total": 66, "facts_retained": 21, "retention_rate": 0.3181818181818182, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.35294117647058826, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1508}, "2": {"facts_total": 66, "facts_retained": 21, "retention_rate": 0.3181818181818182, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.35294117647058826, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1579}, "3": {"facts_total": 66, "facts_retained": 21, "retention_rate": 0.3181818181818182, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.35294117647058826, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1514}, "4": {"facts_total": 66, "facts_retained": 20, "retention_rate": 0.30303030303030304, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.0, "amount": 0.35294117647058826, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1606}}}, "reference_key_id": "keys/synthetic/syn_03024_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "9df441bcc254691aca50ef729117ceda7fcec91596ba108d61042763177cb3a7", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "d48a71040ca2b766ec3dfcf3", "attempt_id": "0b2aba828063", "run_group_id": "E4_core:A4|closed|qwen3b-llama3b:h4", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:27:19.684908Z", "task_family": "synthetic", "document_id": "syn_03024_2000", "document_hash": "ea10cd11ad08f01c50b0712bc481668857c9d2ddc8e6423968f2a31eeb97ac55", "question_id": "q06_negation", "split": "test", "cluster_id": "syn_03024_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "tokenizer_hashes": {"receiver": "b9d8872cf3a99da99e1fc4b0f71232a451632dc49ca4c6771549b4efa5c1c763"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 4, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1833, "query_tokens": 514, "message_tokens": 411, "message_utf8_bytes": 1606, "cache_tensor_bytes": null, "serialized_payload_bytes": 1606, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 23918.322250130586, "hops": {"1": 6944.753459072672, "2": 5560.2366250241175, "3": 6389.902791008353, "4": 5023.429375025444}, "receiver_prefill_ms": 209.58245801739395, "ttft_ms": 11.653708992525935, "decode_ms": 22.99008402042091}, "total_wall_ms": 24162.9817080684, "ttft_ms": 11.653708992525935, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 2}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "No.", "evidence_artifact": {"ledger": {"facts_total": 66, "facts_retained": 20, "retention_rate": 0.30303030303030304, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.0, "amount": 0.35294117647058826, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1606}, "ledger_by_hop": {"1": {"facts_total": 66, "facts_retained": 21, "retention_rate": 0.3181818181818182, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.35294117647058826, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1508}, "2": {"facts_total": 66, "facts_retained": 21, "retention_rate": 0.3181818181818182, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.35294117647058826, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1579}, "3": {"facts_total": 66, "facts_retained": 21, "retention_rate": 0.3181818181818182, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.35294117647058826, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1514}, "4": {"facts_total": 66, "facts_retained": 20, "retention_rate": 0.30303030303030304, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.0, "amount": 0.35294117647058826, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1606}}}, "reference_key_id": "keys/synthetic/syn_03024_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "9df441bcc254691aca50ef729117ceda7fcec91596ba108d61042763177cb3a7", "output_hash": "38178a20b470cfd18299fb1593dd4ba706f1a83321dac051d01086efd8b7a96f"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "c9bc382494a3a0be370a1b8e", "attempt_id": "7d777366268d", "run_group_id": "E4_core:A4|closed|qwen3b-llama3b:h4", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:27:19.944599Z", "task_family": "synthetic", "document_id": "syn_03024_2000", "document_hash": "ea10cd11ad08f01c50b0712bc481668857c9d2ddc8e6423968f2a31eeb97ac55", "question_id": "q07_negation", "split": "test", "cluster_id": "syn_03024_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "tokenizer_hashes": {"receiver": "b9d8872cf3a99da99e1fc4b0f71232a451632dc49ca4c6771549b4efa5c1c763"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 4, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1833, "query_tokens": 515, "message_tokens": 411, "message_utf8_bytes": 1606, "cache_tensor_bytes": null, "serialized_payload_bytes": 1606, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 23918.322250130586, "hops": {"1": 6944.753459072672, "2": 5560.2366250241175, "3": 6389.902791008353, "4": 5023.429375025444}, "receiver_prefill_ms": 209.60525004193187, "ttft_ms": 11.536874924786389, "decode_ms": 23.005667026154697}, "total_wall_ms": 24162.907750112936, "ttft_ms": 11.536874924786389, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 2}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "No.", "evidence_artifact": {"ledger": {"facts_total": 66, "facts_retained": 20, "retention_rate": 0.30303030303030304, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.0, "amount": 0.35294117647058826, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1606}, "ledger_by_hop": {"1": {"facts_total": 66, "facts_retained": 21, "retention_rate": 0.3181818181818182, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.35294117647058826, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1508}, "2": {"facts_total": 66, "facts_retained": 21, "retention_rate": 0.3181818181818182, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.35294117647058826, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1579}, "3": {"facts_total": 66, "facts_retained": 21, "retention_rate": 0.3181818181818182, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.35294117647058826, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1514}, "4": {"facts_total": 66, "facts_retained": 20, "retention_rate": 0.30303030303030304, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.0, "amount": 0.35294117647058826, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1606}}}, "reference_key_id": "keys/synthetic/syn_03024_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "9df441bcc254691aca50ef729117ceda7fcec91596ba108d61042763177cb3a7", "output_hash": "38178a20b470cfd18299fb1593dd4ba706f1a83321dac051d01086efd8b7a96f"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "43091c788a0fb7c5874a16cc", "attempt_id": "771c5dc51338", "run_group_id": "E4_core:A4|closed|qwen3b-llama3b:h4", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:27:20.204158Z", "task_family": "synthetic", "document_id": "syn_03024_2000", "document_hash": "ea10cd11ad08f01c50b0712bc481668857c9d2ddc8e6423968f2a31eeb97ac55", "question_id": "q08_revision", "split": "test", "cluster_id": "syn_03024_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "tokenizer_hashes": {"receiver": "b9d8872cf3a99da99e1fc4b0f71232a451632dc49ca4c6771549b4efa5c1c763"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 4, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1833, "query_tokens": 531, "message_tokens": 411, "message_utf8_bytes": 1606, "cache_tensor_bytes": null, "serialized_payload_bytes": 1606, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 23918.322250130586, "hops": {"1": 6944.753459072672, "2": 5560.2366250241175, "3": 6389.902791008353, "4": 5023.429375025444}, "receiver_prefill_ms": 210.17629199195653, "ttft_ms": 11.611875030212104, "decode_ms": 11.541125015355647}, "total_wall_ms": 24152.276833192445, "ttft_ms": 11.611875030212104, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 66, "facts_retained": 20, "retention_rate": 0.30303030303030304, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.0, "amount": 0.35294117647058826, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1606}, "ledger_by_hop": {"1": {"facts_total": 66, "facts_retained": 21, "retention_rate": 0.3181818181818182, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.35294117647058826, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1508}, "2": {"facts_total": 66, "facts_retained": 21, "retention_rate": 0.3181818181818182, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.35294117647058826, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1579}, "3": {"facts_total": 66, "facts_retained": 21, "retention_rate": 0.3181818181818182, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.35294117647058826, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1514}, "4": {"facts_total": 66, "facts_retained": 20, "retention_rate": 0.30303030303030304, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.0, "amount": 0.35294117647058826, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1606}}}, "reference_key_id": "keys/synthetic/syn_03024_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "9df441bcc254691aca50ef729117ceda7fcec91596ba108d61042763177cb3a7", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "7cab3c73da43ba26f7bcfb93", "attempt_id": "4ceefbea26ee", "run_group_id": "E4_core:A4|closed|qwen3b-llama3b:h4", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:27:20.452913Z", "task_family": "synthetic", "document_id": "syn_03024_2000", "document_hash": "ea10cd11ad08f01c50b0712bc481668857c9d2ddc8e6423968f2a31eeb97ac55", "question_id": "q09_revision", "split": "test", "cluster_id": "syn_03024_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "tokenizer_hashes": {"receiver": "b9d8872cf3a99da99e1fc4b0f71232a451632dc49ca4c6771549b4efa5c1c763"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 4, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1833, "query_tokens": 522, "message_tokens": 411, "message_utf8_bytes": 1606, "cache_tensor_bytes": null, "serialized_payload_bytes": 1606, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 23918.322250130586, "hops": {"1": 6944.753459072672, "2": 5560.2366250241175, "3": 6389.902791008353, "4": 5023.429375025444}, "receiver_prefill_ms": 209.83304106630385, "ttft_ms": 11.75800000783056, "decode_ms": 11.580749996937811}, "total_wall_ms": 24152.088042115793, "ttft_ms": 11.75800000783056, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 66, "facts_retained": 20, "retention_rate": 0.30303030303030304, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.0, "amount": 0.35294117647058826, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1606}, "ledger_by_hop": {"1": {"facts_total": 66, "facts_retained": 21, "retention_rate": 0.3181818181818182, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.35294117647058826, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1508}, "2": {"facts_total": 66, "facts_retained": 21, "retention_rate": 0.3181818181818182, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.35294117647058826, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1579}, "3": {"facts_total": 66, "facts_retained": 21, "retention_rate": 0.3181818181818182, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.35294117647058826, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1514}, "4": {"facts_total": 66, "facts_retained": 20, "retention_rate": 0.30303030303030304, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.0, "amount": 0.35294117647058826, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1606}}}, "reference_key_id": "keys/synthetic/syn_03024_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "9df441bcc254691aca50ef729117ceda7fcec91596ba108d61042763177cb3a7", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "162b350a6b5fb319bd25108c", "attempt_id": "3940956f32a5", "run_group_id": "E4_core:A4|closed|qwen3b-llama3b:h4", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:27:20.703231Z", "task_family": "synthetic", "document_id": "syn_03024_2000", "document_hash": "ea10cd11ad08f01c50b0712bc481668857c9d2ddc8e6423968f2a31eeb97ac55", "question_id": "q10_temporal", "split": "test", "cluster_id": "syn_03024_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "tokenizer_hashes": {"receiver": "b9d8872cf3a99da99e1fc4b0f71232a451632dc49ca4c6771549b4efa5c1c763"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 4, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1833, "query_tokens": 531, "message_tokens": 411, "message_utf8_bytes": 1606, "cache_tensor_bytes": null, "serialized_payload_bytes": 1606, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 23918.322250130586, "hops": {"1": 6944.753459072672, "2": 5560.2366250241175, "3": 6389.902791008353, "4": 5023.429375025444}, "receiver_prefill_ms": 210.2346250321716, "ttft_ms": 11.699000024236739, "decode_ms": 11.551374918781221}, "total_wall_ms": 24152.424250147305, "ttft_ms": 11.699000024236739, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 66, "facts_retained": 20, "retention_rate": 0.30303030303030304, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.0, "amount": 0.35294117647058826, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1606}, "ledger_by_hop": {"1": {"facts_total": 66, "facts_retained": 21, "retention_rate": 0.3181818181818182, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.35294117647058826, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1508}, "2": {"facts_total": 66, "facts_retained": 21, "retention_rate": 0.3181818181818182, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.35294117647058826, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1579}, "3": {"facts_total": 66, "facts_retained": 21, "retention_rate": 0.3181818181818182, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.35294117647058826, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1514}, "4": {"facts_total": 66, "facts_retained": 20, "retention_rate": 0.30303030303030304, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.0, "amount": 0.35294117647058826, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1606}}}, "reference_key_id": "keys/synthetic/syn_03024_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "9df441bcc254691aca50ef729117ceda7fcec91596ba108d61042763177cb3a7", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "1eb5af42cd4bff7c0fe3db72", "attempt_id": "f038a2d9d3ba", "run_group_id": "E4_core:A4|closed|qwen3b-llama3b:h4", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:27:20.952821Z", "task_family": "synthetic", "document_id": "syn_03024_2000", "document_hash": "ea10cd11ad08f01c50b0712bc481668857c9d2ddc8e6423968f2a31eeb97ac55", "question_id": "q11_aggregation", "split": "test", "cluster_id": "syn_03024_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "tokenizer_hashes": {"receiver": "b9d8872cf3a99da99e1fc4b0f71232a451632dc49ca4c6771549b4efa5c1c763"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 4, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1833, "query_tokens": 526, "message_tokens": 411, "message_utf8_bytes": 1606, "cache_tensor_bytes": null, "serialized_payload_bytes": 1606, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 23918.322250130586, "hops": {"1": 6944.753459072672, "2": 5560.2366250241175, "3": 6389.902791008353, "4": 5023.429375025444}, "receiver_prefill_ms": 210.15895798336715, "ttft_ms": 11.71933300793171, "decode_ms": 57.41829099133611}, "total_wall_ms": 24198.201042134315, "ttft_ms": 11.71933300793171, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 5}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "11,120,310", "evidence_artifact": {"ledger": {"facts_total": 66, "facts_retained": 20, "retention_rate": 0.30303030303030304, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.0, "amount": 0.35294117647058826, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1606}, "ledger_by_hop": {"1": {"facts_total": 66, "facts_retained": 21, "retention_rate": 0.3181818181818182, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.35294117647058826, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1508}, "2": {"facts_total": 66, "facts_retained": 21, "retention_rate": 0.3181818181818182, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.35294117647058826, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1579}, "3": {"facts_total": 66, "facts_retained": 21, "retention_rate": 0.3181818181818182, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.35294117647058826, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1514}, "4": {"facts_total": 66, "facts_retained": 20, "retention_rate": 0.30303030303030304, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.0, "amount": 0.35294117647058826, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1606}}}, "reference_key_id": "keys/synthetic/syn_03024_2000.json", "task_scores": {"correct": 0.0, "abstained": 0.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "9df441bcc254691aca50ef729117ceda7fcec91596ba108d61042763177cb3a7", "output_hash": "c4e2d208e7fb67c45a0af9f9a6d4964665fdcf54126bdc7d4d62a5c42e59adf2"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "a5d26c83ba82f55989600b2c", "attempt_id": "4c9b6e48be4d", "run_group_id": "E4_core:A4|closed|qwen3b-llama3b:h4", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:27:21.248874Z", "task_family": "synthetic", "document_id": "syn_03024_2000", "document_hash": "ea10cd11ad08f01c50b0712bc481668857c9d2ddc8e6423968f2a31eeb97ac55", "question_id": "q12_unit_conversion", "split": "test", "cluster_id": "syn_03024_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "tokenizer_hashes": {"receiver": "b9d8872cf3a99da99e1fc4b0f71232a451632dc49ca4c6771549b4efa5c1c763"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 4, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1833, "query_tokens": 522, "message_tokens": 411, "message_utf8_bytes": 1606, "cache_tensor_bytes": null, "serialized_payload_bytes": 1606, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 23918.322250130586, "hops": {"1": 6944.753459072672, "2": 5560.2366250241175, "3": 6389.902791008353, "4": 5023.429375025444}, "receiver_prefill_ms": 209.70904105342925, "ttft_ms": 11.660916963592172, "decode_ms": 11.569375055842102}, "total_wall_ms": 24151.899458141997, "ttft_ms": 11.660916963592172, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 66, "facts_retained": 20, "retention_rate": 0.30303030303030304, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.0, "amount": 0.35294117647058826, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1606}, "ledger_by_hop": {"1": {"facts_total": 66, "facts_retained": 21, "retention_rate": 0.3181818181818182, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.35294117647058826, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1508}, "2": {"facts_total": 66, "facts_retained": 21, "retention_rate": 0.3181818181818182, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.35294117647058826, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1579}, "3": {"facts_total": 66, "facts_retained": 21, "retention_rate": 0.3181818181818182, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.35294117647058826, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1514}, "4": {"facts_total": 66, "facts_retained": 20, "retention_rate": 0.30303030303030304, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.0, "amount": 0.35294117647058826, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1606}}}, "reference_key_id": "keys/synthetic/syn_03024_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "9df441bcc254691aca50ef729117ceda7fcec91596ba108d61042763177cb3a7", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "d203749cb9f3ead5f700a8fb", "attempt_id": "3bbb9b694eb9", "run_group_id": "E4_core:A4|closed|qwen3b-llama3b:h4", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:27:21.498526Z", "task_family": "synthetic", "document_id": "syn_03024_2000", "document_hash": "ea10cd11ad08f01c50b0712bc481668857c9d2ddc8e6423968f2a31eeb97ac55", "question_id": "q13_unanswerable", "split": "test", "cluster_id": "syn_03024_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "tokenizer_hashes": {"receiver": "b9d8872cf3a99da99e1fc4b0f71232a451632dc49ca4c6771549b4efa5c1c763"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 4, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1833, "query_tokens": 526, "message_tokens": 411, "message_utf8_bytes": 1606, "cache_tensor_bytes": null, "serialized_payload_bytes": 1606, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 23918.322250130586, "hops": {"1": 6944.753459072672, "2": 5560.2366250241175, "3": 6389.902791008353, "4": 5023.429375025444}, "receiver_prefill_ms": 209.973624907434, "ttft_ms": 11.686792015098035, "decode_ms": 11.549124959856272}, "total_wall_ms": 24152.119250036776, "ttft_ms": 11.686792015098035, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 66, "facts_retained": 20, "retention_rate": 0.30303030303030304, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.0, "amount": 0.35294117647058826, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1606}, "ledger_by_hop": {"1": {"facts_total": 66, "facts_retained": 21, "retention_rate": 0.3181818181818182, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.35294117647058826, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1508}, "2": {"facts_total": 66, "facts_retained": 21, "retention_rate": 0.3181818181818182, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.35294117647058826, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1579}, "3": {"facts_total": 66, "facts_retained": 21, "retention_rate": 0.3181818181818182, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.35294117647058826, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1514}, "4": {"facts_total": 66, "facts_retained": 20, "retention_rate": 0.30303030303030304, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.0, "amount": 0.35294117647058826, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1606}}}, "reference_key_id": "keys/synthetic/syn_03024_2000.json", "task_scores": {"correct": 1.0, "abstained": 1.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "9df441bcc254691aca50ef729117ceda7fcec91596ba108d61042763177cb3a7", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "50b01c048336574be4196ea1", "attempt_id": "c063cd9223f8", "run_group_id": "E4_core:A4|closed|qwen3b-llama3b:h4", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:27:21.748984Z", "task_family": "synthetic", "document_id": "syn_03024_2000", "document_hash": "ea10cd11ad08f01c50b0712bc481668857c9d2ddc8e6423968f2a31eeb97ac55", "question_id": "q14_unanswerable", "split": "test", "cluster_id": "syn_03024_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "tokenizer_hashes": {"receiver": "b9d8872cf3a99da99e1fc4b0f71232a451632dc49ca4c6771549b4efa5c1c763"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 4, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1833, "query_tokens": 528, "message_tokens": 411, "message_utf8_bytes": 1606, "cache_tensor_bytes": null, "serialized_payload_bytes": 1606, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 23918.322250130586, "hops": {"1": 6944.753459072672, "2": 5560.2366250241175, "3": 6389.902791008353, "4": 5023.429375025444}, "receiver_prefill_ms": 210.39549994748086, "ttft_ms": 11.564792017452419, "decode_ms": 11.556208948604763}, "total_wall_ms": 24152.326250099577, "ttft_ms": 11.564792017452419, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 66, "facts_retained": 20, "retention_rate": 0.30303030303030304, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.0, "amount": 0.35294117647058826, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1606}, "ledger_by_hop": {"1": {"facts_total": 66, "facts_retained": 21, "retention_rate": 0.3181818181818182, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.35294117647058826, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1508}, "2": {"facts_total": 66, "facts_retained": 21, "retention_rate": 0.3181818181818182, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.35294117647058826, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1579}, "3": {"facts_total": 66, "facts_retained": 21, "retention_rate": 0.3181818181818182, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.35294117647058826, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1514}, "4": {"facts_total": 66, "facts_retained": 20, "retention_rate": 0.30303030303030304, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.0, "amount": 0.35294117647058826, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1606}}}, "reference_key_id": "keys/synthetic/syn_03024_2000.json", "task_scores": {"correct": 1.0, "abstained": 1.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "9df441bcc254691aca50ef729117ceda7fcec91596ba108d61042763177cb3a7", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "0a53eb29c116facf1120564f", "attempt_id": "2318f6be076a", "run_group_id": "E4_core:A4|closed|qwen3b-llama3b:h8", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:27:43.733317Z", "task_family": "synthetic", "document_id": "syn_03024_2000", "document_hash": "ea10cd11ad08f01c50b0712bc481668857c9d2ddc8e6423968f2a31eeb97ac55", "question_id": "q00_single_fact", "split": "test", "cluster_id": "syn_03024_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "tokenizer_hashes": {"receiver": "b9d8872cf3a99da99e1fc4b0f71232a451632dc49ca4c6771549b4efa5c1c763"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 8, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1833, "query_tokens": 514, "message_tokens": 411, "message_utf8_bytes": 1572, "cache_tensor_bytes": null, "serialized_payload_bytes": 1572, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 45646.34845801629, "hops": {"1": 6944.753459072672, "2": 5560.2366250241175, "3": 6389.902791008353, "4": 5023.429375025444, "5": 5870.618999935687, "6": 5081.656874972396, "7": 5788.924874970689, "8": 4986.825458006933}, "receiver_prefill_ms": 209.4774170545861, "ttft_ms": 11.706708930432796, "decode_ms": 11.513833072967827}, "total_wall_ms": 45879.62020805571, "ttft_ms": 11.706708930432796, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 66, "facts_retained": 20, "retention_rate": 0.30303030303030304, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.0, "amount": 0.35294117647058826, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1572}, "ledger_by_hop": {"1": {"facts_total": 66, "facts_retained": 21, "retention_rate": 0.3181818181818182, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.35294117647058826, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1508}, "2": {"facts_total": 66, "facts_retained": 21, "retention_rate": 0.3181818181818182, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.35294117647058826, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1579}, "3": {"facts_total": 66, "facts_retained": 21, "retention_rate": 0.3181818181818182, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.35294117647058826, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1514}, "4": {"facts_total": 66, "facts_retained": 20, "retention_rate": 0.30303030303030304, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.0, "amount": 0.35294117647058826, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1606}, "5": {"facts_total": 66, "facts_retained": 20, "retention_rate": 0.30303030303030304, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.0, "amount": 0.35294117647058826, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1580}, "6": {"facts_total": 66, "facts_retained": 20, "retention_rate": 0.30303030303030304, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.0, "amount": 0.35294117647058826, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1614}, "7": {"facts_total": 66, "facts_retained": 20, "retention_rate": 0.30303030303030304, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.0, "amount": 0.35294117647058826, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1539}, "8": {"facts_total": 66, "facts_retained": 20, "retention_rate": 0.30303030303030304, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.0, "amount": 0.35294117647058826, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1572}}}, "reference_key_id": "keys/synthetic/syn_03024_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "e5f91a90ae14ae9b8b6fe51726c465074152fb5d1c9595ff18bd7d10fb4e9aa4", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "d19a9e872122659e4354d6b1", "attempt_id": "0fa981dd2b36", "run_group_id": "E4_core:A4|closed|qwen3b-llama3b:h8", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:27:43.982805Z", "task_family": "synthetic", "document_id": "syn_03024_2000", "document_hash": "ea10cd11ad08f01c50b0712bc481668857c9d2ddc8e6423968f2a31eeb97ac55", "question_id": "q01_single_fact", "split": "test", "cluster_id": "syn_03024_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "tokenizer_hashes": {"receiver": "b9d8872cf3a99da99e1fc4b0f71232a451632dc49ca4c6771549b4efa5c1c763"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 8, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1833, "query_tokens": 519, "message_tokens": 411, "message_utf8_bytes": 1572, "cache_tensor_bytes": null, "serialized_payload_bytes": 1572, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 45646.34845801629, "hops": {"1": 6944.753459072672, "2": 5560.2366250241175, "3": 6389.902791008353, "4": 5023.429375025444, "5": 5870.618999935687, "6": 5081.656874972396, "7": 5788.924874970689, "8": 4986.825458006933}, "receiver_prefill_ms": 209.87370796501637, "ttft_ms": 11.523750028572977, "decode_ms": 11.485249968245625}, "total_wall_ms": 45879.67462395318, "ttft_ms": 11.523750028572977, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 66, "facts_retained": 20, "retention_rate": 0.30303030303030304, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.0, "amount": 0.35294117647058826, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1572}, "ledger_by_hop": {"1": {"facts_total": 66, "facts_retained": 21, "retention_rate": 0.3181818181818182, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.35294117647058826, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1508}, "2": {"facts_total": 66, "facts_retained": 21, "retention_rate": 0.3181818181818182, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.35294117647058826, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1579}, "3": {"facts_total": 66, "facts_retained": 21, "retention_rate": 0.3181818181818182, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.35294117647058826, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1514}, "4": {"facts_total": 66, "facts_retained": 20, "retention_rate": 0.30303030303030304, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.0, "amount": 0.35294117647058826, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1606}, "5": {"facts_total": 66, "facts_retained": 20, "retention_rate": 0.30303030303030304, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.0, "amount": 0.35294117647058826, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1580}, "6": {"facts_total": 66, "facts_retained": 20, "retention_rate": 0.30303030303030304, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.0, "amount": 0.35294117647058826, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1614}, "7": {"facts_total": 66, "facts_retained": 20, "retention_rate": 0.30303030303030304, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.0, "amount": 0.35294117647058826, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1539}, "8": {"facts_total": 66, "facts_retained": 20, "retention_rate": 0.30303030303030304, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.0, "amount": 0.35294117647058826, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1572}}}, "reference_key_id": "keys/synthetic/syn_03024_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "e5f91a90ae14ae9b8b6fe51726c465074152fb5d1c9595ff18bd7d10fb4e9aa4", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "4a042778f42cc7d5dcc8b8b5", "attempt_id": "e632c4b929b4", "run_group_id": "E4_core:A4|closed|qwen3b-llama3b:h8", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:27:44.231979Z", "task_family": "synthetic", "document_id": "syn_03024_2000", "document_hash": "ea10cd11ad08f01c50b0712bc481668857c9d2ddc8e6423968f2a31eeb97ac55", "question_id": "q02_exact_copy", "split": "test", "cluster_id": "syn_03024_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "tokenizer_hashes": {"receiver": "b9d8872cf3a99da99e1fc4b0f71232a451632dc49ca4c6771549b4efa5c1c763"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 8, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1833, "query_tokens": 519, "message_tokens": 411, "message_utf8_bytes": 1572, "cache_tensor_bytes": null, "serialized_payload_bytes": 1572, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 45646.34845801629, "hops": {"1": 6944.753459072672, "2": 5560.2366250241175, "3": 6389.902791008353, "4": 5023.429375025444, "5": 5870.618999935687, "6": 5081.656874972396, "7": 5788.924874970689, "8": 4986.825458006933}, "receiver_prefill_ms": 210.03379102330655, "ttft_ms": 11.518332990817726, "decode_ms": 45.96875002607703}, "total_wall_ms": 45914.309291983955, "ttft_ms": 11.518332990817726, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 4}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "30.0224", "evidence_artifact": {"ledger": {"facts_total": 66, "facts_retained": 20, "retention_rate": 0.30303030303030304, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.0, "amount": 0.35294117647058826, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1572}, "ledger_by_hop": {"1": {"facts_total": 66, "facts_retained": 21, "retention_rate": 0.3181818181818182, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.35294117647058826, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1508}, "2": {"facts_total": 66, "facts_retained": 21, "retention_rate": 0.3181818181818182, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.35294117647058826, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1579}, "3": {"facts_total": 66, "facts_retained": 21, "retention_rate": 0.3181818181818182, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.35294117647058826, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1514}, "4": {"facts_total": 66, "facts_retained": 20, "retention_rate": 0.30303030303030304, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.0, "amount": 0.35294117647058826, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1606}, "5": {"facts_total": 66, "facts_retained": 20, "retention_rate": 0.30303030303030304, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.0, "amount": 0.35294117647058826, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1580}, "6": {"facts_total": 66, "facts_retained": 20, "retention_rate": 0.30303030303030304, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.0, "amount": 0.35294117647058826, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1614}, "7": {"facts_total": 66, "facts_retained": 20, "retention_rate": 0.30303030303030304, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.0, "amount": 0.35294117647058826, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1539}, "8": {"facts_total": 66, "facts_retained": 20, "retention_rate": 0.30303030303030304, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.0, "amount": 0.35294117647058826, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1572}}}, "reference_key_id": "keys/synthetic/syn_03024_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "e5f91a90ae14ae9b8b6fe51726c465074152fb5d1c9595ff18bd7d10fb4e9aa4", "output_hash": "e28a78bc4deb3cec8beab4bf5b87297d7cc55a3e6f839b43fe6408735d464923"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "46afbea4c60193baf8eea7a5", "attempt_id": "319089f7e0f3", "run_group_id": "E4_core:A4|closed|qwen3b-llama3b:h8", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:27:44.515038Z", "task_family": "synthetic", "document_id": "syn_03024_2000", "document_hash": "ea10cd11ad08f01c50b0712bc481668857c9d2ddc8e6423968f2a31eeb97ac55", "question_id": "q03_exact_copy", "split": "test", "cluster_id": "syn_03024_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "tokenizer_hashes": {"receiver": "b9d8872cf3a99da99e1fc4b0f71232a451632dc49ca4c6771549b4efa5c1c763"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 8, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1833, "query_tokens": 514, "message_tokens": 411, "message_utf8_bytes": 1572, "cache_tensor_bytes": null, "serialized_payload_bytes": 1572, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 45646.34845801629, "hops": {"1": 6944.753459072672, "2": 5560.2366250241175, "3": 6389.902791008353, "4": 5023.429375025444, "5": 5870.618999935687, "6": 5081.656874972396, "7": 5788.924874970689, "8": 4986.825458006933}, "receiver_prefill_ms": 210.16345801763237, "ttft_ms": 11.467749951407313, "decode_ms": 11.508374940603971}, "total_wall_ms": 45879.936957964674, "ttft_ms": 11.467749951407313, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 66, "facts_retained": 20, "retention_rate": 0.30303030303030304, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.0, "amount": 0.35294117647058826, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1572}, "ledger_by_hop": {"1": {"facts_total": 66, "facts_retained": 21, "retention_rate": 0.3181818181818182, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.35294117647058826, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1508}, "2": {"facts_total": 66, "facts_retained": 21, "retention_rate": 0.3181818181818182, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.35294117647058826, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1579}, "3": {"facts_total": 66, "facts_retained": 21, "retention_rate": 0.3181818181818182, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.35294117647058826, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1514}, "4": {"facts_total": 66, "facts_retained": 20, "retention_rate": 0.30303030303030304, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.0, "amount": 0.35294117647058826, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1606}, "5": {"facts_total": 66, "facts_retained": 20, "retention_rate": 0.30303030303030304, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.0, "amount": 0.35294117647058826, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1580}, "6": {"facts_total": 66, "facts_retained": 20, "retention_rate": 0.30303030303030304, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.0, "amount": 0.35294117647058826, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1614}, "7": {"facts_total": 66, "facts_retained": 20, "retention_rate": 0.30303030303030304, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.0, "amount": 0.35294117647058826, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1539}, "8": {"facts_total": 66, "facts_retained": 20, "retention_rate": 0.30303030303030304, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.0, "amount": 0.35294117647058826, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1572}}}, "reference_key_id": "keys/synthetic/syn_03024_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "e5f91a90ae14ae9b8b6fe51726c465074152fb5d1c9595ff18bd7d10fb4e9aa4", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "3312eb392bc296e3ac10ab47", "attempt_id": "d5f2e26ccdde", "run_group_id": "E4_core:A4|closed|qwen3b-llama3b:h8", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:27:44.763625Z", "task_family": "synthetic", "document_id": "syn_03024_2000", "document_hash": "ea10cd11ad08f01c50b0712bc481668857c9d2ddc8e6423968f2a31eeb97ac55", "question_id": "q04_multi_hop", "split": "test", "cluster_id": "syn_03024_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "tokenizer_hashes": {"receiver": "b9d8872cf3a99da99e1fc4b0f71232a451632dc49ca4c6771549b4efa5c1c763"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 8, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1833, "query_tokens": 521, "message_tokens": 411, "message_utf8_bytes": 1572, "cache_tensor_bytes": null, "serialized_payload_bytes": 1572, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 45646.34845801629, "hops": {"1": 6944.753459072672, "2": 5560.2366250241175, "3": 6389.902791008353, "4": 5023.429375025444, "5": 5870.618999935687, "6": 5081.656874972396, "7": 5788.924874970689, "8": 4986.825458006933}, "receiver_prefill_ms": 209.93829099461436, "ttft_ms": 11.683167074806988, "decode_ms": 57.34379100613296}, "total_wall_ms": 45925.88841600809, "ttft_ms": 11.683167074806988, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 5}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "6,009,480", "evidence_artifact": {"ledger": {"facts_total": 66, "facts_retained": 20, "retention_rate": 0.30303030303030304, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.0, "amount": 0.35294117647058826, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1572}, "ledger_by_hop": {"1": {"facts_total": 66, "facts_retained": 21, "retention_rate": 0.3181818181818182, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.35294117647058826, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1508}, "2": {"facts_total": 66, "facts_retained": 21, "retention_rate": 0.3181818181818182, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.35294117647058826, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1579}, "3": {"facts_total": 66, "facts_retained": 21, "retention_rate": 0.3181818181818182, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.35294117647058826, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1514}, "4": {"facts_total": 66, "facts_retained": 20, "retention_rate": 0.30303030303030304, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.0, "amount": 0.35294117647058826, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1606}, "5": {"facts_total": 66, "facts_retained": 20, "retention_rate": 0.30303030303030304, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.0, "amount": 0.35294117647058826, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1580}, "6": {"facts_total": 66, "facts_retained": 20, "retention_rate": 0.30303030303030304, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.0, "amount": 0.35294117647058826, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1614}, "7": {"facts_total": 66, "facts_retained": 20, "retention_rate": 0.30303030303030304, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.0, "amount": 0.35294117647058826, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1539}, "8": {"facts_total": 66, "facts_retained": 20, "retention_rate": 0.30303030303030304, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.0, "amount": 0.35294117647058826, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1572}}}, "reference_key_id": "keys/synthetic/syn_03024_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "e5f91a90ae14ae9b8b6fe51726c465074152fb5d1c9595ff18bd7d10fb4e9aa4", "output_hash": "30b20d9aae4d7b0587edeefdc2a9a159f73faf6e2570f1d408a415fe92cbf45e"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "e256a24eea71bbf4a881837b", "attempt_id": "36f2f13fcba4", "run_group_id": "E4_core:A4|closed|qwen3b-llama3b:h8", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:27:45.058104Z", "task_family": "synthetic", "document_id": "syn_03024_2000", "document_hash": "ea10cd11ad08f01c50b0712bc481668857c9d2ddc8e6423968f2a31eeb97ac55", "question_id": "q05_multi_hop", "split": "test", "cluster_id": "syn_03024_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "tokenizer_hashes": {"receiver": "b9d8872cf3a99da99e1fc4b0f71232a451632dc49ca4c6771549b4efa5c1c763"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 8, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1833, "query_tokens": 523, "message_tokens": 411, "message_utf8_bytes": 1572, "cache_tensor_bytes": null, "serialized_payload_bytes": 1572, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 45646.34845801629, "hops": {"1": 6944.753459072672, "2": 5560.2366250241175, "3": 6389.902791008353, "4": 5023.429375025444, "5": 5870.618999935687, "6": 5081.656874972396, "7": 5788.924874970689, "8": 4986.825458006933}, "receiver_prefill_ms": 210.05954209249467, "ttft_ms": 11.571915936656296, "decode_ms": 11.591000016778708}, "total_wall_ms": 45880.02587505616, "ttft_ms": 11.571915936656296, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 66, "facts_retained": 20, "retention_rate": 0.30303030303030304, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.0, "amount": 0.35294117647058826, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1572}, "ledger_by_hop": {"1": {"facts_total": 66, "facts_retained": 21, "retention_rate": 0.3181818181818182, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.35294117647058826, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1508}, "2": {"facts_total": 66, "facts_retained": 21, "retention_rate": 0.3181818181818182, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.35294117647058826, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1579}, "3": {"facts_total": 66, "facts_retained": 21, "retention_rate": 0.3181818181818182, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.35294117647058826, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1514}, "4": {"facts_total": 66, "facts_retained": 20, "retention_rate": 0.30303030303030304, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.0, "amount": 0.35294117647058826, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1606}, "5": {"facts_total": 66, "facts_retained": 20, "retention_rate": 0.30303030303030304, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.0, "amount": 0.35294117647058826, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1580}, "6": {"facts_total": 66, "facts_retained": 20, "retention_rate": 0.30303030303030304, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.0, "amount": 0.35294117647058826, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1614}, "7": {"facts_total": 66, "facts_retained": 20, "retention_rate": 0.30303030303030304, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.0, "amount": 0.35294117647058826, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1539}, "8": {"facts_total": 66, "facts_retained": 20, "retention_rate": 0.30303030303030304, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.0, "amount": 0.35294117647058826, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1572}}}, "reference_key_id": "keys/synthetic/syn_03024_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "e5f91a90ae14ae9b8b6fe51726c465074152fb5d1c9595ff18bd7d10fb4e9aa4", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "75d7c208ecc0cbc1720dbeff", "attempt_id": "99ba0830220b", "run_group_id": "E4_core:A4|closed|qwen3b-llama3b:h8", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:27:45.307632Z", "task_family": "synthetic", "document_id": "syn_03024_2000", "document_hash": "ea10cd11ad08f01c50b0712bc481668857c9d2ddc8e6423968f2a31eeb97ac55", "question_id": "q06_negation", "split": "test", "cluster_id": "syn_03024_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "tokenizer_hashes": {"receiver": "b9d8872cf3a99da99e1fc4b0f71232a451632dc49ca4c6771549b4efa5c1c763"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 8, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1833, "query_tokens": 514, "message_tokens": 411, "message_utf8_bytes": 1572, "cache_tensor_bytes": null, "serialized_payload_bytes": 1572, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 45646.34845801629, "hops": {"1": 6944.753459072672, "2": 5560.2366250241175, "3": 6389.902791008353, "4": 5023.429375025444, "5": 5870.618999935687, "6": 5081.656874972396, "7": 5788.924874970689, "8": 4986.825458006933}, "receiver_prefill_ms": 208.7915410520509, "ttft_ms": 11.58149994444102, "decode_ms": 23.20962504018098}, "total_wall_ms": 45890.458708046935, "ttft_ms": 11.58149994444102, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 2}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "No.", "evidence_artifact": {"ledger": {"facts_total": 66, "facts_retained": 20, "retention_rate": 0.30303030303030304, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.0, "amount": 0.35294117647058826, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1572}, "ledger_by_hop": {"1": {"facts_total": 66, "facts_retained": 21, "retention_rate": 0.3181818181818182, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.35294117647058826, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1508}, "2": {"facts_total": 66, "facts_retained": 21, "retention_rate": 0.3181818181818182, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.35294117647058826, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1579}, "3": {"facts_total": 66, "facts_retained": 21, "retention_rate": 0.3181818181818182, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.35294117647058826, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1514}, "4": {"facts_total": 66, "facts_retained": 20, "retention_rate": 0.30303030303030304, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.0, "amount": 0.35294117647058826, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1606}, "5": {"facts_total": 66, "facts_retained": 20, "retention_rate": 0.30303030303030304, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.0, "amount": 0.35294117647058826, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1580}, "6": {"facts_total": 66, "facts_retained": 20, "retention_rate": 0.30303030303030304, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.0, "amount": 0.35294117647058826, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1614}, "7": {"facts_total": 66, "facts_retained": 20, "retention_rate": 0.30303030303030304, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.0, "amount": 0.35294117647058826, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1539}, "8": {"facts_total": 66, "facts_retained": 20, "retention_rate": 0.30303030303030304, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.0, "amount": 0.35294117647058826, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1572}}}, "reference_key_id": "keys/synthetic/syn_03024_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "e5f91a90ae14ae9b8b6fe51726c465074152fb5d1c9595ff18bd7d10fb4e9aa4", "output_hash": "38178a20b470cfd18299fb1593dd4ba706f1a83321dac051d01086efd8b7a96f"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "7ae6ce946411a5c966c8886a", "attempt_id": "6283746f5e33", "run_group_id": "E4_core:A4|closed|qwen3b-llama3b:h8", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:27:45.566820Z", "task_family": "synthetic", "document_id": "syn_03024_2000", "document_hash": "ea10cd11ad08f01c50b0712bc481668857c9d2ddc8e6423968f2a31eeb97ac55", "question_id": "q07_negation", "split": "test", "cluster_id": "syn_03024_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "tokenizer_hashes": {"receiver": "b9d8872cf3a99da99e1fc4b0f71232a451632dc49ca4c6771549b4efa5c1c763"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 8, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1833, "query_tokens": 515, "message_tokens": 411, "message_utf8_bytes": 1572, "cache_tensor_bytes": null, "serialized_payload_bytes": 1572, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 45646.34845801629, "hops": {"1": 6944.753459072672, "2": 5560.2366250241175, "3": 6389.902791008353, "4": 5023.429375025444, "5": 5870.618999935687, "6": 5081.656874972396, "7": 5788.924874970689, "8": 4986.825458006933}, "receiver_prefill_ms": 209.80845799203962, "ttft_ms": 11.496959021314979, "decode_ms": 23.185333935543895}, "total_wall_ms": 45891.29208307713, "ttft_ms": 11.496959021314979, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 2}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "No.", "evidence_artifact": {"ledger": {"facts_total": 66, "facts_retained": 20, "retention_rate": 0.30303030303030304, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.0, "amount": 0.35294117647058826, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1572}, "ledger_by_hop": {"1": {"facts_total": 66, "facts_retained": 21, "retention_rate": 0.3181818181818182, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.35294117647058826, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1508}, "2": {"facts_total": 66, "facts_retained": 21, "retention_rate": 0.3181818181818182, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.35294117647058826, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1579}, "3": {"facts_total": 66, "facts_retained": 21, "retention_rate": 0.3181818181818182, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.35294117647058826, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1514}, "4": {"facts_total": 66, "facts_retained": 20, "retention_rate": 0.30303030303030304, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.0, "amount": 0.35294117647058826, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1606}, "5": {"facts_total": 66, "facts_retained": 20, "retention_rate": 0.30303030303030304, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.0, "amount": 0.35294117647058826, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1580}, "6": {"facts_total": 66, "facts_retained": 20, "retention_rate": 0.30303030303030304, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.0, "amount": 0.35294117647058826, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1614}, "7": {"facts_total": 66, "facts_retained": 20, "retention_rate": 0.30303030303030304, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.0, "amount": 0.35294117647058826, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1539}, "8": {"facts_total": 66, "facts_retained": 20, "retention_rate": 0.30303030303030304, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.0, "amount": 0.35294117647058826, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1572}}}, "reference_key_id": "keys/synthetic/syn_03024_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "e5f91a90ae14ae9b8b6fe51726c465074152fb5d1c9595ff18bd7d10fb4e9aa4", "output_hash": "38178a20b470cfd18299fb1593dd4ba706f1a83321dac051d01086efd8b7a96f"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "fb7a0f959b5b9d9b4c6eb180", "attempt_id": "f453d08aedb8", "run_group_id": "E4_core:A4|closed|qwen3b-llama3b:h8", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:27:45.826887Z", "task_family": "synthetic", "document_id": "syn_03024_2000", "document_hash": "ea10cd11ad08f01c50b0712bc481668857c9d2ddc8e6423968f2a31eeb97ac55", "question_id": "q08_revision", "split": "test", "cluster_id": "syn_03024_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "tokenizer_hashes": {"receiver": "b9d8872cf3a99da99e1fc4b0f71232a451632dc49ca4c6771549b4efa5c1c763"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 8, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1833, "query_tokens": 531, "message_tokens": 411, "message_utf8_bytes": 1572, "cache_tensor_bytes": null, "serialized_payload_bytes": 1572, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 45646.34845801629, "hops": {"1": 6944.753459072672, "2": 5560.2366250241175, "3": 6389.902791008353, "4": 5023.429375025444, "5": 5870.618999935687, "6": 5081.656874972396, "7": 5788.924874970689, "8": 4986.825458006933}, "receiver_prefill_ms": 210.21349995862693, "ttft_ms": 11.583082959987223, "decode_ms": 11.523167020641267}, "total_wall_ms": 45880.12187497225, "ttft_ms": 11.583082959987223, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 66, "facts_retained": 20, "retention_rate": 0.30303030303030304, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.0, "amount": 0.35294117647058826, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1572}, "ledger_by_hop": {"1": {"facts_total": 66, "facts_retained": 21, "retention_rate": 0.3181818181818182, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.35294117647058826, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1508}, "2": {"facts_total": 66, "facts_retained": 21, "retention_rate": 0.3181818181818182, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.35294117647058826, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1579}, "3": {"facts_total": 66, "facts_retained": 21, "retention_rate": 0.3181818181818182, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.35294117647058826, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1514}, "4": {"facts_total": 66, "facts_retained": 20, "retention_rate": 0.30303030303030304, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.0, "amount": 0.35294117647058826, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1606}, "5": {"facts_total": 66, "facts_retained": 20, "retention_rate": 0.30303030303030304, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.0, "amount": 0.35294117647058826, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1580}, "6": {"facts_total": 66, "facts_retained": 20, "retention_rate": 0.30303030303030304, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.0, "amount": 0.35294117647058826, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1614}, "7": {"facts_total": 66, "facts_retained": 20, "retention_rate": 0.30303030303030304, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.0, "amount": 0.35294117647058826, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1539}, "8": {"facts_total": 66, "facts_retained": 20, "retention_rate": 0.30303030303030304, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.0, "amount": 0.35294117647058826, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1572}}}, "reference_key_id": "keys/synthetic/syn_03024_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "e5f91a90ae14ae9b8b6fe51726c465074152fb5d1c9595ff18bd7d10fb4e9aa4", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "7a3e75e7ff0bdb49ad0e315c", "attempt_id": "d27eb5a714b2", "run_group_id": "E4_core:A4|closed|qwen3b-llama3b:h8", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:27:46.075907Z", "task_family": "synthetic", "document_id": "syn_03024_2000", "document_hash": "ea10cd11ad08f01c50b0712bc481668857c9d2ddc8e6423968f2a31eeb97ac55", "question_id": "q09_revision", "split": "test", "cluster_id": "syn_03024_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "tokenizer_hashes": {"receiver": "b9d8872cf3a99da99e1fc4b0f71232a451632dc49ca4c6771549b4efa5c1c763"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 8, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1833, "query_tokens": 522, "message_tokens": 411, "message_utf8_bytes": 1572, "cache_tensor_bytes": null, "serialized_payload_bytes": 1572, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 45646.34845801629, "hops": {"1": 6944.753459072672, "2": 5560.2366250241175, "3": 6389.902791008353, "4": 5023.429375025444, "5": 5870.618999935687, "6": 5081.656874972396, "7": 5788.924874970689, "8": 4986.825458006933}, "receiver_prefill_ms": 209.37204209621996, "ttft_ms": 11.538374936208129, "decode_ms": 11.743542039766908}, "total_wall_ms": 45879.43687499501, "ttft_ms": 11.538374936208129, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 66, "facts_retained": 20, "retention_rate": 0.30303030303030304, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.0, "amount": 0.35294117647058826, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1572}, "ledger_by_hop": {"1": {"facts_total": 66, "facts_retained": 21, "retention_rate": 0.3181818181818182, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.35294117647058826, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1508}, "2": {"facts_total": 66, "facts_retained": 21, "retention_rate": 0.3181818181818182, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.35294117647058826, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1579}, "3": {"facts_total": 66, "facts_retained": 21, "retention_rate": 0.3181818181818182, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.35294117647058826, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1514}, "4": {"facts_total": 66, "facts_retained": 20, "retention_rate": 0.30303030303030304, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.0, "amount": 0.35294117647058826, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1606}, "5": {"facts_total": 66, "facts_retained": 20, "retention_rate": 0.30303030303030304, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.0, "amount": 0.35294117647058826, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1580}, "6": {"facts_total": 66, "facts_retained": 20, "retention_rate": 0.30303030303030304, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.0, "amount": 0.35294117647058826, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1614}, "7": {"facts_total": 66, "facts_retained": 20, "retention_rate": 0.30303030303030304, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.0, "amount": 0.35294117647058826, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1539}, "8": {"facts_total": 66, "facts_retained": 20, "retention_rate": 0.30303030303030304, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.0, "amount": 0.35294117647058826, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1572}}}, "reference_key_id": "keys/synthetic/syn_03024_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "e5f91a90ae14ae9b8b6fe51726c465074152fb5d1c9595ff18bd7d10fb4e9aa4", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "369505830f6d0af79312068e", "attempt_id": "190ce7eef685", "run_group_id": "E4_core:A4|closed|qwen3b-llama3b:h8", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:27:46.323965Z", "task_family": "synthetic", "document_id": "syn_03024_2000", "document_hash": "ea10cd11ad08f01c50b0712bc481668857c9d2ddc8e6423968f2a31eeb97ac55", "question_id": "q10_temporal", "split": "test", "cluster_id": "syn_03024_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "tokenizer_hashes": {"receiver": "b9d8872cf3a99da99e1fc4b0f71232a451632dc49ca4c6771549b4efa5c1c763"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 8, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1833, "query_tokens": 531, "message_tokens": 411, "message_utf8_bytes": 1572, "cache_tensor_bytes": null, "serialized_payload_bytes": 1572, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 45646.34845801629, "hops": {"1": 6944.753459072672, "2": 5560.2366250241175, "3": 6389.902791008353, "4": 5023.429375025444, "5": 5870.618999935687, "6": 5081.656874972396, "7": 5788.924874970689, "8": 4986.825458006933}, "receiver_prefill_ms": 210.2324590086937, "ttft_ms": 11.696124915033579, "decode_ms": 11.458582943305373}, "total_wall_ms": 45880.2797079552, "ttft_ms": 11.696124915033579, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 66, "facts_retained": 20, "retention_rate": 0.30303030303030304, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.0, "amount": 0.35294117647058826, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1572}, "ledger_by_hop": {"1": {"facts_total": 66, "facts_retained": 21, "retention_rate": 0.3181818181818182, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.35294117647058826, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1508}, "2": {"facts_total": 66, "facts_retained": 21, "retention_rate": 0.3181818181818182, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.35294117647058826, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1579}, "3": {"facts_total": 66, "facts_retained": 21, "retention_rate": 0.3181818181818182, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.35294117647058826, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1514}, "4": {"facts_total": 66, "facts_retained": 20, "retention_rate": 0.30303030303030304, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.0, "amount": 0.35294117647058826, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1606}, "5": {"facts_total": 66, "facts_retained": 20, "retention_rate": 0.30303030303030304, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.0, "amount": 0.35294117647058826, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1580}, "6": {"facts_total": 66, "facts_retained": 20, "retention_rate": 0.30303030303030304, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.0, "amount": 0.35294117647058826, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1614}, "7": {"facts_total": 66, "facts_retained": 20, "retention_rate": 0.30303030303030304, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.0, "amount": 0.35294117647058826, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1539}, "8": {"facts_total": 66, "facts_retained": 20, "retention_rate": 0.30303030303030304, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.0, "amount": 0.35294117647058826, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1572}}}, "reference_key_id": "keys/synthetic/syn_03024_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "e5f91a90ae14ae9b8b6fe51726c465074152fb5d1c9595ff18bd7d10fb4e9aa4", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "2a987097d1761c6472c2cf5b", "attempt_id": "fc19e8c1e03b", "run_group_id": "E4_core:A4|closed|qwen3b-llama3b:h8", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:27:46.572734Z", "task_family": "synthetic", "document_id": "syn_03024_2000", "document_hash": "ea10cd11ad08f01c50b0712bc481668857c9d2ddc8e6423968f2a31eeb97ac55", "question_id": "q11_aggregation", "split": "test", "cluster_id": "syn_03024_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "tokenizer_hashes": {"receiver": "b9d8872cf3a99da99e1fc4b0f71232a451632dc49ca4c6771549b4efa5c1c763"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 8, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1833, "query_tokens": 526, "message_tokens": 411, "message_utf8_bytes": 1572, "cache_tensor_bytes": null, "serialized_payload_bytes": 1572, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 45646.34845801629, "hops": {"1": 6944.753459072672, "2": 5560.2366250241175, "3": 6389.902791008353, "4": 5023.429375025444, "5": 5870.618999935687, "6": 5081.656874972396, "7": 5788.924874970689, "8": 4986.825458006933}, "receiver_prefill_ms": 209.68958304729313, "ttft_ms": 11.5634590620175, "decode_ms": 57.24787502549589}, "total_wall_ms": 45925.434957956895, "ttft_ms": 11.5634590620175, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 5}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "11,190,269", "evidence_artifact": {"ledger": {"facts_total": 66, "facts_retained": 20, "retention_rate": 0.30303030303030304, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.0, "amount": 0.35294117647058826, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1572}, "ledger_by_hop": {"1": {"facts_total": 66, "facts_retained": 21, "retention_rate": 0.3181818181818182, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.35294117647058826, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1508}, "2": {"facts_total": 66, "facts_retained": 21, "retention_rate": 0.3181818181818182, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.35294117647058826, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1579}, "3": {"facts_total": 66, "facts_retained": 21, "retention_rate": 0.3181818181818182, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.35294117647058826, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1514}, "4": {"facts_total": 66, "facts_retained": 20, "retention_rate": 0.30303030303030304, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.0, "amount": 0.35294117647058826, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1606}, "5": {"facts_total": 66, "facts_retained": 20, "retention_rate": 0.30303030303030304, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.0, "amount": 0.35294117647058826, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1580}, "6": {"facts_total": 66, "facts_retained": 20, "retention_rate": 0.30303030303030304, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.0, "amount": 0.35294117647058826, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1614}, "7": {"facts_total": 66, "facts_retained": 20, "retention_rate": 0.30303030303030304, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.0, "amount": 0.35294117647058826, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1539}, "8": {"facts_total": 66, "facts_retained": 20, "retention_rate": 0.30303030303030304, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.0, "amount": 0.35294117647058826, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1572}}}, "reference_key_id": "keys/synthetic/syn_03024_2000.json", "task_scores": {"correct": 0.0, "abstained": 0.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "e5f91a90ae14ae9b8b6fe51726c465074152fb5d1c9595ff18bd7d10fb4e9aa4", "output_hash": "21023e2c5262695db33c4b4dfbf2ce40c8795742ceadf13db29509b3a088ad29"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "ef14f9fff8fa5f6f50021e7a", "attempt_id": "d425bcc14047", "run_group_id": "E4_core:A4|closed|qwen3b-llama3b:h8", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:27:46.867226Z", "task_family": "synthetic", "document_id": "syn_03024_2000", "document_hash": "ea10cd11ad08f01c50b0712bc481668857c9d2ddc8e6423968f2a31eeb97ac55", "question_id": "q12_unit_conversion", "split": "test", "cluster_id": "syn_03024_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "tokenizer_hashes": {"receiver": "b9d8872cf3a99da99e1fc4b0f71232a451632dc49ca4c6771549b4efa5c1c763"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 8, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1833, "query_tokens": 522, "message_tokens": 411, "message_utf8_bytes": 1572, "cache_tensor_bytes": null, "serialized_payload_bytes": 1572, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 45646.34845801629, "hops": {"1": 6944.753459072672, "2": 5560.2366250241175, "3": 6389.902791008353, "4": 5023.429375025444, "5": 5870.618999935687, "6": 5081.656874972396, "7": 5788.924874970689, "8": 4986.825458006933}, "receiver_prefill_ms": 210.1102079031989, "ttft_ms": 11.63962495047599, "decode_ms": 11.495582992210984}, "total_wall_ms": 45880.12345810421, "ttft_ms": 11.63962495047599, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 66, "facts_retained": 20, "retention_rate": 0.30303030303030304, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.0, "amount": 0.35294117647058826, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1572}, "ledger_by_hop": {"1": {"facts_total": 66, "facts_retained": 21, "retention_rate": 0.3181818181818182, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.35294117647058826, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1508}, "2": {"facts_total": 66, "facts_retained": 21, "retention_rate": 0.3181818181818182, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.35294117647058826, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1579}, "3": {"facts_total": 66, "facts_retained": 21, "retention_rate": 0.3181818181818182, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.35294117647058826, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1514}, "4": {"facts_total": 66, "facts_retained": 20, "retention_rate": 0.30303030303030304, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.0, "amount": 0.35294117647058826, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1606}, "5": {"facts_total": 66, "facts_retained": 20, "retention_rate": 0.30303030303030304, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.0, "amount": 0.35294117647058826, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1580}, "6": {"facts_total": 66, "facts_retained": 20, "retention_rate": 0.30303030303030304, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.0, "amount": 0.35294117647058826, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1614}, "7": {"facts_total": 66, "facts_retained": 20, "retention_rate": 0.30303030303030304, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.0, "amount": 0.35294117647058826, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1539}, "8": {"facts_total": 66, "facts_retained": 20, "retention_rate": 0.30303030303030304, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.0, "amount": 0.35294117647058826, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1572}}}, "reference_key_id": "keys/synthetic/syn_03024_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "e5f91a90ae14ae9b8b6fe51726c465074152fb5d1c9595ff18bd7d10fb4e9aa4", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "18cbbe38a8da5c961b6bd6a7", "attempt_id": "4e99f3b95af7", "run_group_id": "E4_core:A4|closed|qwen3b-llama3b:h8", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:27:47.117226Z", "task_family": "synthetic", "document_id": "syn_03024_2000", "document_hash": "ea10cd11ad08f01c50b0712bc481668857c9d2ddc8e6423968f2a31eeb97ac55", "question_id": "q13_unanswerable", "split": "test", "cluster_id": "syn_03024_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "tokenizer_hashes": {"receiver": "b9d8872cf3a99da99e1fc4b0f71232a451632dc49ca4c6771549b4efa5c1c763"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 8, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1833, "query_tokens": 526, "message_tokens": 411, "message_utf8_bytes": 1572, "cache_tensor_bytes": null, "serialized_payload_bytes": 1572, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 45646.34845801629, "hops": {"1": 6944.753459072672, "2": 5560.2366250241175, "3": 6389.902791008353, "4": 5023.429375025444, "5": 5870.618999935687, "6": 5081.656874972396, "7": 5788.924874970689, "8": 4986.825458006933}, "receiver_prefill_ms": 210.20649990532547, "ttft_ms": 11.656583985313773, "decode_ms": 11.594834038987756}, "total_wall_ms": 45880.25770802051, "ttft_ms": 11.656583985313773, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 66, "facts_retained": 20, "retention_rate": 0.30303030303030304, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.0, "amount": 0.35294117647058826, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1572}, "ledger_by_hop": {"1": {"facts_total": 66, "facts_retained": 21, "retention_rate": 0.3181818181818182, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.35294117647058826, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1508}, "2": {"facts_total": 66, "facts_retained": 21, "retention_rate": 0.3181818181818182, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.35294117647058826, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1579}, "3": {"facts_total": 66, "facts_retained": 21, "retention_rate": 0.3181818181818182, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.35294117647058826, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1514}, "4": {"facts_total": 66, "facts_retained": 20, "retention_rate": 0.30303030303030304, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.0, "amount": 0.35294117647058826, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1606}, "5": {"facts_total": 66, "facts_retained": 20, "retention_rate": 0.30303030303030304, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.0, "amount": 0.35294117647058826, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1580}, "6": {"facts_total": 66, "facts_retained": 20, "retention_rate": 0.30303030303030304, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.0, "amount": 0.35294117647058826, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1614}, "7": {"facts_total": 66, "facts_retained": 20, "retention_rate": 0.30303030303030304, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.0, "amount": 0.35294117647058826, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1539}, "8": {"facts_total": 66, "facts_retained": 20, "retention_rate": 0.30303030303030304, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.0, "amount": 0.35294117647058826, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1572}}}, "reference_key_id": "keys/synthetic/syn_03024_2000.json", "task_scores": {"correct": 1.0, "abstained": 1.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "e5f91a90ae14ae9b8b6fe51726c465074152fb5d1c9595ff18bd7d10fb4e9aa4", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "5282996b8e943c58145fdb3b", "attempt_id": "d438d80378fa", "run_group_id": "E4_core:A4|closed|qwen3b-llama3b:h8", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:27:47.366819Z", "task_family": "synthetic", "document_id": "syn_03024_2000", "document_hash": "ea10cd11ad08f01c50b0712bc481668857c9d2ddc8e6423968f2a31eeb97ac55", "question_id": "q14_unanswerable", "split": "test", "cluster_id": "syn_03024_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "tokenizer_hashes": {"receiver": "b9d8872cf3a99da99e1fc4b0f71232a451632dc49ca4c6771549b4efa5c1c763"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 8, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1833, "query_tokens": 528, "message_tokens": 411, "message_utf8_bytes": 1572, "cache_tensor_bytes": null, "serialized_payload_bytes": 1572, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 45646.34845801629, "hops": {"1": 6944.753459072672, "2": 5560.2366250241175, "3": 6389.902791008353, "4": 5023.429375025444, "5": 5870.618999935687, "6": 5081.656874972396, "7": 5788.924874970689, "8": 4986.825458006933}, "receiver_prefill_ms": 210.3344580391422, "ttft_ms": 11.54904195573181, "decode_ms": 11.54974999371916}, "total_wall_ms": 45880.225665983744, "ttft_ms": 11.54904195573181, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 66, "facts_retained": 20, "retention_rate": 0.30303030303030304, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.0, "amount": 0.35294117647058826, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1572}, "ledger_by_hop": {"1": {"facts_total": 66, "facts_retained": 21, "retention_rate": 0.3181818181818182, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.35294117647058826, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1508}, "2": {"facts_total": 66, "facts_retained": 21, "retention_rate": 0.3181818181818182, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.35294117647058826, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1579}, "3": {"facts_total": 66, "facts_retained": 21, "retention_rate": 0.3181818181818182, "retention_by_kind": {"person": 0.5, "date": 0.0, "amount": 0.35294117647058826, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1514}, "4": {"facts_total": 66, "facts_retained": 20, "retention_rate": 0.30303030303030304, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.0, "amount": 0.35294117647058826, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1606}, "5": {"facts_total": 66, "facts_retained": 20, "retention_rate": 0.30303030303030304, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.0, "amount": 0.35294117647058826, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1580}, "6": {"facts_total": 66, "facts_retained": 20, "retention_rate": 0.30303030303030304, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.0, "amount": 0.35294117647058826, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1614}, "7": {"facts_total": 66, "facts_retained": 20, "retention_rate": 0.30303030303030304, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.0, "amount": 0.35294117647058826, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1539}, "8": {"facts_total": 66, "facts_retained": 20, "retention_rate": 0.30303030303030304, "retention_by_kind": {"person": 0.3333333333333333, "date": 0.0, "amount": 0.35294117647058826, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1572}}}, "reference_key_id": "keys/synthetic/syn_03024_2000.json", "task_scores": {"correct": 1.0, "abstained": 1.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "e5f91a90ae14ae9b8b6fe51726c465074152fb5d1c9595ff18bd7d10fb4e9aa4", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "7ea68c21988e6c3e307fe98a", "attempt_id": "61760789fca4", "run_group_id": "E4_core:A5|closed|qwen3b-llama3b:h1", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:27:54.560642Z", "task_family": "synthetic", "document_id": "syn_03024_2000", "document_hash": "ea10cd11ad08f01c50b0712bc481668857c9d2ddc8e6423968f2a31eeb97ac55", "question_id": "q00_single_fact", "split": "test", "cluster_id": "syn_03024_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 1, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 2068, "query_tokens": 440, "message_tokens": 357, "message_utf8_bytes": 818, "cache_tensor_bytes": null, "serialized_payload_bytes": 818, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 6942.490917048417, "hops": {"1": 6942.490917048417}, "receiver_prefill_ms": 164.17279106099159, "ttft_ms": 11.82345801498741, "decode_ms": 11.827292037196457}, "total_wall_ms": 7130.719625158235, "ttft_ms": 11.82345801498741, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 66, "facts_retained": 6, "retention_rate": 0.09090909090909091, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.17647058823529413, "status": 0.0, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 754}, "ledger_by_hop": {"1": {"facts_total": 66, "facts_retained": 6, "retention_rate": 0.09090909090909091, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.17647058823529413, "status": 0.0, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 754}}}, "reference_key_id": "keys/synthetic/syn_03024_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "3c07df4f32c3dffe23905714a167378f1096d7812f8c55ff7cf92b920f04befa", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "b50913f2e4efe90df7866026", "attempt_id": "82e4e40bbb30", "run_group_id": "E4_core:A5|closed|qwen3b-llama3b:h1", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:27:54.765287Z", "task_family": "synthetic", "document_id": "syn_03024_2000", "document_hash": "ea10cd11ad08f01c50b0712bc481668857c9d2ddc8e6423968f2a31eeb97ac55", "question_id": "q01_single_fact", "split": "test", "cluster_id": "syn_03024_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 1, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 2068, "query_tokens": 445, "message_tokens": 357, "message_utf8_bytes": 818, "cache_tensor_bytes": null, "serialized_payload_bytes": 818, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 6942.490917048417, "hops": {"1": 6942.490917048417}, "receiver_prefill_ms": 164.5389589248225, "ttft_ms": 11.915499926544726, "decode_ms": 129.21375001315027}, "total_wall_ms": 7248.639875091612, "ttft_ms": 11.915499926544726, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 11}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "Lot 568-G in Mont-Laurier", "evidence_artifact": {"ledger": {"facts_total": 66, "facts_retained": 6, "retention_rate": 0.09090909090909091, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.17647058823529413, "status": 0.0, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 754}, "ledger_by_hop": {"1": {"facts_total": 66, "facts_retained": 6, "retention_rate": 0.09090909090909091, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.17647058823529413, "status": 0.0, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 754}}}, "reference_key_id": "keys/synthetic/syn_03024_2000.json", "task_scores": {"correct": 0.0, "abstained": 0.0, "f1": 0.5}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "3c07df4f32c3dffe23905714a167378f1096d7812f8c55ff7cf92b920f04befa", "output_hash": "98e007faa5a10742ad74f2778a64e3ea0d6b72bec1c0392200d475073d7bf082"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "40ff4ba3875193852fcc2869", "attempt_id": "e84b4dbfc5db", "run_group_id": "E4_core:A5|closed|qwen3b-llama3b:h1", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:27:55.087199Z", "task_family": "synthetic", "document_id": "syn_03024_2000", "document_hash": "ea10cd11ad08f01c50b0712bc481668857c9d2ddc8e6423968f2a31eeb97ac55", "question_id": "q02_exact_copy", "split": "test", "cluster_id": "syn_03024_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 1, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 2068, "query_tokens": 445, "message_tokens": 357, "message_utf8_bytes": 818, "cache_tensor_bytes": null, "serialized_payload_bytes": 818, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 6942.490917048417, "hops": {"1": 6942.490917048417}, "receiver_prefill_ms": 164.58516696002334, "ttft_ms": 11.787749943323433, "decode_ms": 11.765792034566402}, "total_wall_ms": 7131.01237511728, "ttft_ms": 11.787749943323433, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 66, "facts_retained": 6, "retention_rate": 0.09090909090909091, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.17647058823529413, "status": 0.0, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 754}, "ledger_by_hop": {"1": {"facts_total": 66, "facts_retained": 6, "retention_rate": 0.09090909090909091, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.17647058823529413, "status": 0.0, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 754}}}, "reference_key_id": "keys/synthetic/syn_03024_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "3c07df4f32c3dffe23905714a167378f1096d7812f8c55ff7cf92b920f04befa", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "cdf26439619bebeaab3e4c6e", "attempt_id": "54c0db6df4e6", "run_group_id": "E4_core:A5|closed|qwen3b-llama3b:h1", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:27:55.291190Z", "task_family": "synthetic", "document_id": "syn_03024_2000", "document_hash": "ea10cd11ad08f01c50b0712bc481668857c9d2ddc8e6423968f2a31eeb97ac55", "question_id": "q03_exact_copy", "split": "test", "cluster_id": "syn_03024_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 1, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 2068, "query_tokens": 440, "message_tokens": 357, "message_utf8_bytes": 818, "cache_tensor_bytes": null, "serialized_payload_bytes": 818, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 6942.490917048417, "hops": {"1": 6942.490917048417}, "receiver_prefill_ms": 164.17929204180837, "ttft_ms": 12.010416015982628, "decode_ms": 11.781290988437831}, "total_wall_ms": 7130.971959093586, "ttft_ms": 12.010416015982628, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 66, "facts_retained": 6, "retention_rate": 0.09090909090909091, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.17647058823529413, "status": 0.0, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 754}, "ledger_by_hop": {"1": {"facts_total": 66, "facts_retained": 6, "retention_rate": 0.09090909090909091, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.17647058823529413, "status": 0.0, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 754}}}, "reference_key_id": "keys/synthetic/syn_03024_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "3c07df4f32c3dffe23905714a167378f1096d7812f8c55ff7cf92b920f04befa", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "23b9f866e5e5c1b8cbc29935", "attempt_id": "359d82465c6e", "run_group_id": "E4_core:A5|closed|qwen3b-llama3b:h1", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:27:55.495093Z", "task_family": "synthetic", "document_id": "syn_03024_2000", "document_hash": "ea10cd11ad08f01c50b0712bc481668857c9d2ddc8e6423968f2a31eeb97ac55", "question_id": "q04_multi_hop", "split": "test", "cluster_id": "syn_03024_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 1, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 2068, "query_tokens": 446, "message_tokens": 357, "message_utf8_bytes": 818, "cache_tensor_bytes": null, "serialized_payload_bytes": 818, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 6942.490917048417, "hops": {"1": 6942.490917048417}, "receiver_prefill_ms": 165.33824999351054, "ttft_ms": 11.931874905712903, "decode_ms": 11.804083944298327}, "total_wall_ms": 7132.061625015922, "ttft_ms": 11.931874905712903, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 66, "facts_retained": 6, "retention_rate": 0.09090909090909091, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.17647058823529413, "status": 0.0, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 754}, "ledger_by_hop": {"1": {"facts_total": 66, "facts_retained": 6, "retention_rate": 0.09090909090909091, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.17647058823529413, "status": 0.0, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 754}}}, "reference_key_id": "keys/synthetic/syn_03024_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "3c07df4f32c3dffe23905714a167378f1096d7812f8c55ff7cf92b920f04befa", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "1930788d821a20e147f1ef40", "attempt_id": "5beaf27ef6dc", "run_group_id": "E4_core:A5|closed|qwen3b-llama3b:h1", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:27:55.701211Z", "task_family": "synthetic", "document_id": "syn_03024_2000", "document_hash": "ea10cd11ad08f01c50b0712bc481668857c9d2ddc8e6423968f2a31eeb97ac55", "question_id": "q05_multi_hop", "split": "test", "cluster_id": "syn_03024_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 1, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 2068, "query_tokens": 450, "message_tokens": 357, "message_utf8_bytes": 818, "cache_tensor_bytes": null, "serialized_payload_bytes": 818, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 6942.490917048417, "hops": {"1": 6942.490917048417}, "receiver_prefill_ms": 185.3592919651419, "ttft_ms": 11.888915905728936, "decode_ms": 11.97108393535018}, "total_wall_ms": 7152.126542059705, "ttft_ms": 11.888915905728936, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 66, "facts_retained": 6, "retention_rate": 0.09090909090909091, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.17647058823529413, "status": 0.0, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 754}, "ledger_by_hop": {"1": {"facts_total": 66, "facts_retained": 6, "retention_rate": 0.09090909090909091, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.17647058823529413, "status": 0.0, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 754}}}, "reference_key_id": "keys/synthetic/syn_03024_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "3c07df4f32c3dffe23905714a167378f1096d7812f8c55ff7cf92b920f04befa", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "0c0133b9c689e9d857481420", "attempt_id": "fb008591de2f", "run_group_id": "E4_core:A5|closed|qwen3b-llama3b:h1", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:27:55.926782Z", "task_family": "synthetic", "document_id": "syn_03024_2000", "document_hash": "ea10cd11ad08f01c50b0712bc481668857c9d2ddc8e6423968f2a31eeb97ac55", "question_id": "q06_negation", "split": "test", "cluster_id": "syn_03024_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 1, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 2068, "query_tokens": 439, "message_tokens": 357, "message_utf8_bytes": 818, "cache_tensor_bytes": null, "serialized_payload_bytes": 818, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 6942.490917048417, "hops": {"1": 6942.490917048417}, "receiver_prefill_ms": 164.45625002961606, "ttft_ms": 11.807958013378084, "decode_ms": 11.83654204942286}, "total_wall_ms": 7130.97204209771, "ttft_ms": 11.807958013378084, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "no", "evidence_artifact": {"ledger": {"facts_total": 66, "facts_retained": 6, "retention_rate": 0.09090909090909091, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.17647058823529413, "status": 0.0, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 754}, "ledger_by_hop": {"1": {"facts_total": 66, "facts_retained": 6, "retention_rate": 0.09090909090909091, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.17647058823529413, "status": 0.0, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 754}}}, "reference_key_id": "keys/synthetic/syn_03024_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "3c07df4f32c3dffe23905714a167378f1096d7812f8c55ff7cf92b920f04befa", "output_hash": "9390298f3fb0c5b160498935d79cb139aef28e1c47358b4bbba61862b9c26e59"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "54df357d5a8f2d4e93433bfa", "attempt_id": "4bcd26fae1b5", "run_group_id": "E4_core:A5|closed|qwen3b-llama3b:h1", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:27:56.130611Z", "task_family": "synthetic", "document_id": "syn_03024_2000", "document_hash": "ea10cd11ad08f01c50b0712bc481668857c9d2ddc8e6423968f2a31eeb97ac55", "question_id": "q07_negation", "split": "test", "cluster_id": "syn_03024_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 1, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 2068, "query_tokens": 441, "message_tokens": 357, "message_utf8_bytes": 818, "cache_tensor_bytes": null, "serialized_payload_bytes": 818, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 6942.490917048417, "hops": {"1": 6942.490917048417}, "receiver_prefill_ms": 164.52458291314542, "ttft_ms": 11.794665944762528, "decode_ms": 11.808041017502546}, "total_wall_ms": 7130.982834030874, "ttft_ms": 11.794665944762528, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "no", "evidence_artifact": {"ledger": {"facts_total": 66, "facts_retained": 6, "retention_rate": 0.09090909090909091, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.17647058823529413, "status": 0.0, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 754}, "ledger_by_hop": {"1": {"facts_total": 66, "facts_retained": 6, "retention_rate": 0.09090909090909091, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.17647058823529413, "status": 0.0, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 754}}}, "reference_key_id": "keys/synthetic/syn_03024_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "3c07df4f32c3dffe23905714a167378f1096d7812f8c55ff7cf92b920f04befa", "output_hash": "9390298f3fb0c5b160498935d79cb139aef28e1c47358b4bbba61862b9c26e59"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "5d70de1e7a817eb1353daa2a", "attempt_id": "832f12926bc9", "run_group_id": "E4_core:A5|closed|qwen3b-llama3b:h1", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:27:56.334131Z", "task_family": "synthetic", "document_id": "syn_03024_2000", "document_hash": "ea10cd11ad08f01c50b0712bc481668857c9d2ddc8e6423968f2a31eeb97ac55", "question_id": "q08_revision", "split": "test", "cluster_id": "syn_03024_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 1, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 2068, "query_tokens": 457, "message_tokens": 357, "message_utf8_bytes": 818, "cache_tensor_bytes": null, "serialized_payload_bytes": 818, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 6942.490917048417, "hops": {"1": 6942.490917048417}, "receiver_prefill_ms": 185.8098329976201, "ttft_ms": 11.822375003248453, "decode_ms": 11.8561249691993}, "total_wall_ms": 7152.3516670567915, "ttft_ms": 11.822375003248453, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 66, "facts_retained": 6, "retention_rate": 0.09090909090909091, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.17647058823529413, "status": 0.0, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 754}, "ledger_by_hop": {"1": {"facts_total": 66, "facts_retained": 6, "retention_rate": 0.09090909090909091, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.17647058823529413, "status": 0.0, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 754}}}, "reference_key_id": "keys/synthetic/syn_03024_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "3c07df4f32c3dffe23905714a167378f1096d7812f8c55ff7cf92b920f04befa", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "e362aa75542bd746776f0c73", "attempt_id": "410b5f7e4587", "run_group_id": "E4_core:A5|closed|qwen3b-llama3b:h1", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:27:56.558788Z", "task_family": "synthetic", "document_id": "syn_03024_2000", "document_hash": "ea10cd11ad08f01c50b0712bc481668857c9d2ddc8e6423968f2a31eeb97ac55", "question_id": "q09_revision", "split": "test", "cluster_id": "syn_03024_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 1, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 2068, "query_tokens": 448, "message_tokens": 357, "message_utf8_bytes": 818, "cache_tensor_bytes": null, "serialized_payload_bytes": 818, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 6942.490917048417, "hops": {"1": 6942.490917048417}, "receiver_prefill_ms": 165.33612494822592, "ttft_ms": 11.832375079393387, "decode_ms": 82.35804201103747}, "total_wall_ms": 7202.369041973725, "ttft_ms": 11.832375079393387, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 7}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "9134435", "evidence_artifact": {"ledger": {"facts_total": 66, "facts_retained": 6, "retention_rate": 0.09090909090909091, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.17647058823529413, "status": 0.0, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 754}, "ledger_by_hop": {"1": {"facts_total": 66, "facts_retained": 6, "retention_rate": 0.09090909090909091, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.17647058823529413, "status": 0.0, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 754}}}, "reference_key_id": "keys/synthetic/syn_03024_2000.json", "task_scores": {"correct": 0.0, "abstained": 0.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "3c07df4f32c3dffe23905714a167378f1096d7812f8c55ff7cf92b920f04befa", "output_hash": "f7384892cb4f9a38f6115f91da5fec3f0d7e24e47877737b0f34e2b4a6af7d5a"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "c98d98199120f34f8827f16d", "attempt_id": "d490e7b827d5", "run_group_id": "E4_core:A5|closed|qwen3b-llama3b:h1", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:27:56.833980Z", "task_family": "synthetic", "document_id": "syn_03024_2000", "document_hash": "ea10cd11ad08f01c50b0712bc481668857c9d2ddc8e6423968f2a31eeb97ac55", "question_id": "q10_temporal", "split": "test", "cluster_id": "syn_03024_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 1, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 2068, "query_tokens": 458, "message_tokens": 357, "message_utf8_bytes": 818, "cache_tensor_bytes": null, "serialized_payload_bytes": 818, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 6942.490917048417, "hops": {"1": 6942.490917048417}, "receiver_prefill_ms": 185.3366669965908, "ttft_ms": 12.078124913387, "decode_ms": 11.815457954071462}, "total_wall_ms": 7152.1960840327665, "ttft_ms": 12.078124913387, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 66, "facts_retained": 6, "retention_rate": 0.09090909090909091, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.17647058823529413, "status": 0.0, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 754}, "ledger_by_hop": {"1": {"facts_total": 66, "facts_retained": 6, "retention_rate": 0.09090909090909091, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.17647058823529413, "status": 0.0, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 754}}}, "reference_key_id": "keys/synthetic/syn_03024_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "3c07df4f32c3dffe23905714a167378f1096d7812f8c55ff7cf92b920f04befa", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "7f1663034de9f6aba686ce64", "attempt_id": "2e1ae2e964a8", "run_group_id": "E4_core:A5|closed|qwen3b-llama3b:h1", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:27:57.059192Z", "task_family": "synthetic", "document_id": "syn_03024_2000", "document_hash": "ea10cd11ad08f01c50b0712bc481668857c9d2ddc8e6423968f2a31eeb97ac55", "question_id": "q11_aggregation", "split": "test", "cluster_id": "syn_03024_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 1, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 2068, "query_tokens": 451, "message_tokens": 357, "message_utf8_bytes": 818, "cache_tensor_bytes": null, "serialized_payload_bytes": 818, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 6942.490917048417, "hops": {"1": 6942.490917048417}, "receiver_prefill_ms": 185.3935410035774, "ttft_ms": 11.88008300960064, "decode_ms": 82.08512503188103}, "total_wall_ms": 7222.34874998685, "ttft_ms": 11.88008300960064, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 7}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "8066734", "evidence_artifact": {"ledger": {"facts_total": 66, "facts_retained": 6, "retention_rate": 0.09090909090909091, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.17647058823529413, "status": 0.0, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 754}, "ledger_by_hop": {"1": {"facts_total": 66, "facts_retained": 6, "retention_rate": 0.09090909090909091, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.17647058823529413, "status": 0.0, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 754}}}, "reference_key_id": "keys/synthetic/syn_03024_2000.json", "task_scores": {"correct": 0.0, "abstained": 0.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "3c07df4f32c3dffe23905714a167378f1096d7812f8c55ff7cf92b920f04befa", "output_hash": "167c75fdd6a41fca64228797ab8f3a1a3f994475246df0bffb2e060d5d2e4ff0"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "04c8079f8fa3824a466c287a", "attempt_id": "d5cbd328b5ff", "run_group_id": "E4_core:A5|closed|qwen3b-llama3b:h1", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:27:57.354591Z", "task_family": "synthetic", "document_id": "syn_03024_2000", "document_hash": "ea10cd11ad08f01c50b0712bc481668857c9d2ddc8e6423968f2a31eeb97ac55", "question_id": "q12_unit_conversion", "split": "test", "cluster_id": "syn_03024_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 1, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 2068, "query_tokens": 448, "message_tokens": 357, "message_utf8_bytes": 818, "cache_tensor_bytes": null, "serialized_payload_bytes": 818, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 6942.490917048417, "hops": {"1": 6942.490917048417}, "receiver_prefill_ms": 164.56816694699228, "ttft_ms": 12.100499938242137, "decode_ms": 11.922874953597784}, "total_wall_ms": 7131.640583975241, "ttft_ms": 12.100499938242137, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 66, "facts_retained": 6, "retention_rate": 0.09090909090909091, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.17647058823529413, "status": 0.0, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 754}, "ledger_by_hop": {"1": {"facts_total": 66, "facts_retained": 6, "retention_rate": 0.09090909090909091, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.17647058823529413, "status": 0.0, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 754}}}, "reference_key_id": "keys/synthetic/syn_03024_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "3c07df4f32c3dffe23905714a167378f1096d7812f8c55ff7cf92b920f04befa", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "417e157d94e0e3154c9c7cd6", "attempt_id": "bd2ab830ed00", "run_group_id": "E4_core:A5|closed|qwen3b-llama3b:h1", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:27:57.559883Z", "task_family": "synthetic", "document_id": "syn_03024_2000", "document_hash": "ea10cd11ad08f01c50b0712bc481668857c9d2ddc8e6423968f2a31eeb97ac55", "question_id": "q13_unanswerable", "split": "test", "cluster_id": "syn_03024_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 1, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 2068, "query_tokens": 452, "message_tokens": 357, "message_utf8_bytes": 818, "cache_tensor_bytes": null, "serialized_payload_bytes": 818, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 6942.490917048417, "hops": {"1": 6942.490917048417}, "receiver_prefill_ms": 185.2195409592241, "ttft_ms": 12.060333974659443, "decode_ms": 11.880666948854923}, "total_wall_ms": 7152.249751030467, "ttft_ms": 12.060333974659443, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 66, "facts_retained": 6, "retention_rate": 0.09090909090909091, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.17647058823529413, "status": 0.0, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 754}, "ledger_by_hop": {"1": {"facts_total": 66, "facts_retained": 6, "retention_rate": 0.09090909090909091, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.17647058823529413, "status": 0.0, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 754}}}, "reference_key_id": "keys/synthetic/syn_03024_2000.json", "task_scores": {"correct": 1.0, "abstained": 1.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "3c07df4f32c3dffe23905714a167378f1096d7812f8c55ff7cf92b920f04befa", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "b840b5a8a2bea799e19e3f8f", "attempt_id": "9a3f0b9d305d", "run_group_id": "E4_core:A5|closed|qwen3b-llama3b:h1", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:27:57.786959Z", "task_family": "synthetic", "document_id": "syn_03024_2000", "document_hash": "ea10cd11ad08f01c50b0712bc481668857c9d2ddc8e6423968f2a31eeb97ac55", "question_id": "q14_unanswerable", "split": "test", "cluster_id": "syn_03024_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 1, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 2068, "query_tokens": 453, "message_tokens": 357, "message_utf8_bytes": 818, "cache_tensor_bytes": null, "serialized_payload_bytes": 818, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 6942.490917048417, "hops": {"1": 6942.490917048417}, "receiver_prefill_ms": 185.63199997879565, "ttft_ms": 11.94083399605006, "decode_ms": 11.851874995045364}, "total_wall_ms": 7152.419749996625, "ttft_ms": 11.94083399605006, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 66, "facts_retained": 6, "retention_rate": 0.09090909090909091, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.17647058823529413, "status": 0.0, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 754}, "ledger_by_hop": {"1": {"facts_total": 66, "facts_retained": 6, "retention_rate": 0.09090909090909091, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.17647058823529413, "status": 0.0, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 754}}}, "reference_key_id": "keys/synthetic/syn_03024_2000.json", "task_scores": {"correct": 1.0, "abstained": 1.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "3c07df4f32c3dffe23905714a167378f1096d7812f8c55ff7cf92b920f04befa", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "3765388b04f031b28489e005", "attempt_id": "a71aa6b7eb29", "run_group_id": "E4_core:A5|closed|qwen3b-llama3b:h2", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:28:01.971893Z", "task_family": "synthetic", "document_id": "syn_03024_2000", "document_hash": "ea10cd11ad08f01c50b0712bc481668857c9d2ddc8e6423968f2a31eeb97ac55", "question_id": "q00_single_fact", "split": "test", "cluster_id": "syn_03024_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "tokenizer_hashes": {"receiver": "b9d8872cf3a99da99e1fc4b0f71232a451632dc49ca4c6771549b4efa5c1c763"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 2, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1833, "query_tokens": 426, "message_tokens": 323, "message_utf8_bytes": 919, "cache_tensor_bytes": null, "serialized_payload_bytes": 919, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 10900.45654203277, "hops": {"1": 6942.490917048417, "2": 3957.965624984354}, "receiver_prefill_ms": 183.02712496370077, "ttft_ms": 11.487290961667895, "decode_ms": 11.496416991576552}, "total_wall_ms": 11106.89208295662, "ttft_ms": 11.487290961667895, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 66, "facts_retained": 8, "retention_rate": 0.12121212121212122, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.17647058823529413, "status": 0.16666666666666666, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 919}, "ledger_by_hop": {"1": {"facts_total": 66, "facts_retained": 6, "retention_rate": 0.09090909090909091, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.17647058823529413, "status": 0.0, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 754}, "2": {"facts_total": 66, "facts_retained": 8, "retention_rate": 0.12121212121212122, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.17647058823529413, "status": 0.16666666666666666, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 919}}}, "reference_key_id": "keys/synthetic/syn_03024_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "ec4605a52403f5abd0fa044d6604914a51692b197ccf5eac20a7175fe08f4ea8", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "02167f383d66e8cf77d7fdb9", "attempt_id": "3d5e86716d1f", "run_group_id": "E4_core:A5|closed|qwen3b-llama3b:h2", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:28:02.194376Z", "task_family": "synthetic", "document_id": "syn_03024_2000", "document_hash": "ea10cd11ad08f01c50b0712bc481668857c9d2ddc8e6423968f2a31eeb97ac55", "question_id": "q01_single_fact", "split": "test", "cluster_id": "syn_03024_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "tokenizer_hashes": {"receiver": "b9d8872cf3a99da99e1fc4b0f71232a451632dc49ca4c6771549b4efa5c1c763"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 2, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1833, "query_tokens": 431, "message_tokens": 323, "message_utf8_bytes": 919, "cache_tensor_bytes": null, "serialized_payload_bytes": 919, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 10900.45654203277, "hops": {"1": 6942.490917048417, "2": 3957.965624984354}, "receiver_prefill_ms": 183.27483301982284, "ttft_ms": 11.460541049018502, "decode_ms": 11.465542018413544}, "total_wall_ms": 11107.04320797231, "ttft_ms": 11.460541049018502, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 66, "facts_retained": 8, "retention_rate": 0.12121212121212122, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.17647058823529413, "status": 0.16666666666666666, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 919}, "ledger_by_hop": {"1": {"facts_total": 66, "facts_retained": 6, "retention_rate": 0.09090909090909091, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.17647058823529413, "status": 0.0, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 754}, "2": {"facts_total": 66, "facts_retained": 8, "retention_rate": 0.12121212121212122, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.17647058823529413, "status": 0.16666666666666666, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 919}}}, "reference_key_id": "keys/synthetic/syn_03024_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "ec4605a52403f5abd0fa044d6604914a51692b197ccf5eac20a7175fe08f4ea8", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "f52f453ce41971920a155afc", "attempt_id": "1a4648b7f9c8", "run_group_id": "E4_core:A5|closed|qwen3b-llama3b:h2", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:28:02.416316Z", "task_family": "synthetic", "document_id": "syn_03024_2000", "document_hash": "ea10cd11ad08f01c50b0712bc481668857c9d2ddc8e6423968f2a31eeb97ac55", "question_id": "q02_exact_copy", "split": "test", "cluster_id": "syn_03024_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "tokenizer_hashes": {"receiver": "b9d8872cf3a99da99e1fc4b0f71232a451632dc49ca4c6771549b4efa5c1c763"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 2, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1833, "query_tokens": 431, "message_tokens": 323, "message_utf8_bytes": 919, "cache_tensor_bytes": null, "serialized_payload_bytes": 919, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 10900.45654203277, "hops": {"1": 6942.490917048417, "2": 3957.965624984354}, "receiver_prefill_ms": 187.6737920101732, "ttft_ms": 11.563832988031209, "decode_ms": 11.674916953779757}, "total_wall_ms": 11111.748209106736, "ttft_ms": 11.563832988031209, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 66, "facts_retained": 8, "retention_rate": 0.12121212121212122, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.17647058823529413, "status": 0.16666666666666666, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 919}, "ledger_by_hop": {"1": {"facts_total": 66, "facts_retained": 6, "retention_rate": 0.09090909090909091, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.17647058823529413, "status": 0.0, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 754}, "2": {"facts_total": 66, "facts_retained": 8, "retention_rate": 0.12121212121212122, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.17647058823529413, "status": 0.16666666666666666, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 919}}}, "reference_key_id": "keys/synthetic/syn_03024_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "ec4605a52403f5abd0fa044d6604914a51692b197ccf5eac20a7175fe08f4ea8", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "f94e6b903be994776c463660", "attempt_id": "a4e81ef944e2", "run_group_id": "E4_core:A5|closed|qwen3b-llama3b:h2", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:28:02.643020Z", "task_family": "synthetic", "document_id": "syn_03024_2000", "document_hash": "ea10cd11ad08f01c50b0712bc481668857c9d2ddc8e6423968f2a31eeb97ac55", "question_id": "q03_exact_copy", "split": "test", "cluster_id": "syn_03024_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "tokenizer_hashes": {"receiver": "b9d8872cf3a99da99e1fc4b0f71232a451632dc49ca4c6771549b4efa5c1c763"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 2, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1833, "query_tokens": 426, "message_tokens": 323, "message_utf8_bytes": 919, "cache_tensor_bytes": null, "serialized_payload_bytes": 919, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 10900.45654203277, "hops": {"1": 6942.490917048417, "2": 3957.965624984354}, "receiver_prefill_ms": 181.05399992782623, "ttft_ms": 11.5665839985013, "decode_ms": 11.42991695087403}, "total_wall_ms": 11104.8718750244, "ttft_ms": 11.5665839985013, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 66, "facts_retained": 8, "retention_rate": 0.12121212121212122, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.17647058823529413, "status": 0.16666666666666666, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 919}, "ledger_by_hop": {"1": {"facts_total": 66, "facts_retained": 6, "retention_rate": 0.09090909090909091, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.17647058823529413, "status": 0.0, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 754}, "2": {"facts_total": 66, "facts_retained": 8, "retention_rate": 0.12121212121212122, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.17647058823529413, "status": 0.16666666666666666, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 919}}}, "reference_key_id": "keys/synthetic/syn_03024_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "ec4605a52403f5abd0fa044d6604914a51692b197ccf5eac20a7175fe08f4ea8", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "2c82fa120f3bc17680c45643", "attempt_id": "a2dcd8d19210", "run_group_id": "E4_core:A5|closed|qwen3b-llama3b:h2", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:28:02.862162Z", "task_family": "synthetic", "document_id": "syn_03024_2000", "document_hash": "ea10cd11ad08f01c50b0712bc481668857c9d2ddc8e6423968f2a31eeb97ac55", "question_id": "q04_multi_hop", "split": "test", "cluster_id": "syn_03024_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "tokenizer_hashes": {"receiver": "b9d8872cf3a99da99e1fc4b0f71232a451632dc49ca4c6771549b4efa5c1c763"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 2, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1833, "query_tokens": 433, "message_tokens": 323, "message_utf8_bytes": 919, "cache_tensor_bytes": null, "serialized_payload_bytes": 919, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 10900.45654203277, "hops": {"1": 6942.490917048417, "2": 3957.965624984354}, "receiver_prefill_ms": 186.2979590659961, "ttft_ms": 11.589249945245683, "decode_ms": 11.464667040854692}, "total_wall_ms": 11110.232583014295, "ttft_ms": 11.589249945245683, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 66, "facts_retained": 8, "retention_rate": 0.12121212121212122, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.17647058823529413, "status": 0.16666666666666666, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 919}, "ledger_by_hop": {"1": {"facts_total": 66, "facts_retained": 6, "retention_rate": 0.09090909090909091, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.17647058823529413, "status": 0.0, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 754}, "2": {"facts_total": 66, "facts_retained": 8, "retention_rate": 0.12121212121212122, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.17647058823529413, "status": 0.16666666666666666, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 919}}}, "reference_key_id": "keys/synthetic/syn_03024_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "ec4605a52403f5abd0fa044d6604914a51692b197ccf5eac20a7175fe08f4ea8", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "0b6911b723a0a6d66f439ed0", "attempt_id": "38f74865ed67", "run_group_id": "E4_core:A5|closed|qwen3b-llama3b:h2", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:28:03.087265Z", "task_family": "synthetic", "document_id": "syn_03024_2000", "document_hash": "ea10cd11ad08f01c50b0712bc481668857c9d2ddc8e6423968f2a31eeb97ac55", "question_id": "q05_multi_hop", "split": "test", "cluster_id": "syn_03024_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "tokenizer_hashes": {"receiver": "b9d8872cf3a99da99e1fc4b0f71232a451632dc49ca4c6771549b4efa5c1c763"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 2, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1833, "query_tokens": 435, "message_tokens": 323, "message_utf8_bytes": 919, "cache_tensor_bytes": null, "serialized_payload_bytes": 919, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 10900.45654203277, "hops": {"1": 6942.490917048417, "2": 3957.965624984354}, "receiver_prefill_ms": 188.05520795285702, "ttft_ms": 11.52770803309977, "decode_ms": 11.488832999020815}, "total_wall_ms": 11111.909626051784, "ttft_ms": 11.52770803309977, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 66, "facts_retained": 8, "retention_rate": 0.12121212121212122, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.17647058823529413, "status": 0.16666666666666666, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 919}, "ledger_by_hop": {"1": {"facts_total": 66, "facts_retained": 6, "retention_rate": 0.09090909090909091, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.17647058823529413, "status": 0.0, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 754}, "2": {"facts_total": 66, "facts_retained": 8, "retention_rate": 0.12121212121212122, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.17647058823529413, "status": 0.16666666666666666, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 919}}}, "reference_key_id": "keys/synthetic/syn_03024_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "ec4605a52403f5abd0fa044d6604914a51692b197ccf5eac20a7175fe08f4ea8", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "74408db948005691fd67cf2f", "attempt_id": "179a25fa3c47", "run_group_id": "E4_core:A5|closed|qwen3b-llama3b:h2", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:28:03.314314Z", "task_family": "synthetic", "document_id": "syn_03024_2000", "document_hash": "ea10cd11ad08f01c50b0712bc481668857c9d2ddc8e6423968f2a31eeb97ac55", "question_id": "q06_negation", "split": "test", "cluster_id": "syn_03024_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "tokenizer_hashes": {"receiver": "b9d8872cf3a99da99e1fc4b0f71232a451632dc49ca4c6771549b4efa5c1c763"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 2, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1833, "query_tokens": 426, "message_tokens": 323, "message_utf8_bytes": 919, "cache_tensor_bytes": null, "serialized_payload_bytes": 919, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 10900.45654203277, "hops": {"1": 6942.490917048417, "2": 3957.965624984354}, "receiver_prefill_ms": 182.77116701938212, "ttft_ms": 11.77541702054441, "decode_ms": 11.436625034548342}, "total_wall_ms": 11106.801541987807, "ttft_ms": 11.77541702054441, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "No", "evidence_artifact": {"ledger": {"facts_total": 66, "facts_retained": 8, "retention_rate": 0.12121212121212122, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.17647058823529413, "status": 0.16666666666666666, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 919}, "ledger_by_hop": {"1": {"facts_total": 66, "facts_retained": 6, "retention_rate": 0.09090909090909091, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.17647058823529413, "status": 0.0, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 754}, "2": {"facts_total": 66, "facts_retained": 8, "retention_rate": 0.12121212121212122, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.17647058823529413, "status": 0.16666666666666666, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 919}}}, "reference_key_id": "keys/synthetic/syn_03024_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "ec4605a52403f5abd0fa044d6604914a51692b197ccf5eac20a7175fe08f4ea8", "output_hash": "1ea442a134b2a184bd5d40104401f2a37fbc09ccf3f4bc9da161c6099be3691d"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "a46ba1327729fe935236eda5", "attempt_id": "1dded563a49c", "run_group_id": "E4_core:A5|closed|qwen3b-llama3b:h2", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:28:03.535427Z", "task_family": "synthetic", "document_id": "syn_03024_2000", "document_hash": "ea10cd11ad08f01c50b0712bc481668857c9d2ddc8e6423968f2a31eeb97ac55", "question_id": "q07_negation", "split": "test", "cluster_id": "syn_03024_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "tokenizer_hashes": {"receiver": "b9d8872cf3a99da99e1fc4b0f71232a451632dc49ca4c6771549b4efa5c1c763"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 2, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1833, "query_tokens": 427, "message_tokens": 323, "message_utf8_bytes": 919, "cache_tensor_bytes": null, "serialized_payload_bytes": 919, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 10900.45654203277, "hops": {"1": 6942.490917048417, "2": 3957.965624984354}, "receiver_prefill_ms": 183.2695419434458, "ttft_ms": 11.47595897782594, "decode_ms": 11.704165954142809}, "total_wall_ms": 11107.27700102143, "ttft_ms": 11.47595897782594, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "No", "evidence_artifact": {"ledger": {"facts_total": 66, "facts_retained": 8, "retention_rate": 0.12121212121212122, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.17647058823529413, "status": 0.16666666666666666, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 919}, "ledger_by_hop": {"1": {"facts_total": 66, "facts_retained": 6, "retention_rate": 0.09090909090909091, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.17647058823529413, "status": 0.0, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 754}, "2": {"facts_total": 66, "facts_retained": 8, "retention_rate": 0.12121212121212122, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.17647058823529413, "status": 0.16666666666666666, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 919}}}, "reference_key_id": "keys/synthetic/syn_03024_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "ec4605a52403f5abd0fa044d6604914a51692b197ccf5eac20a7175fe08f4ea8", "output_hash": "1ea442a134b2a184bd5d40104401f2a37fbc09ccf3f4bc9da161c6099be3691d"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "3110b0e89708c09b55470ecf", "attempt_id": "36a6c8ab9f4d", "run_group_id": "E4_core:A5|closed|qwen3b-llama3b:h2", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:28:03.757378Z", "task_family": "synthetic", "document_id": "syn_03024_2000", "document_hash": "ea10cd11ad08f01c50b0712bc481668857c9d2ddc8e6423968f2a31eeb97ac55", "question_id": "q08_revision", "split": "test", "cluster_id": "syn_03024_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "tokenizer_hashes": {"receiver": "b9d8872cf3a99da99e1fc4b0f71232a451632dc49ca4c6771549b4efa5c1c763"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 2, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1833, "query_tokens": 443, "message_tokens": 323, "message_utf8_bytes": 919, "cache_tensor_bytes": null, "serialized_payload_bytes": 919, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 10900.45654203277, "hops": {"1": 6942.490917048417, "2": 3957.965624984354}, "receiver_prefill_ms": 186.68641697149724, "ttft_ms": 11.72691595274955, "decode_ms": 11.568082962185144}, "total_wall_ms": 11110.898167011328, "ttft_ms": 11.72691595274955, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 66, "facts_retained": 8, "retention_rate": 0.12121212121212122, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.17647058823529413, "status": 0.16666666666666666, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 919}, "ledger_by_hop": {"1": {"facts_total": 66, "facts_retained": 6, "retention_rate": 0.09090909090909091, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.17647058823529413, "status": 0.0, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 754}, "2": {"facts_total": 66, "facts_retained": 8, "retention_rate": 0.12121212121212122, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.17647058823529413, "status": 0.16666666666666666, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 919}}}, "reference_key_id": "keys/synthetic/syn_03024_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "ec4605a52403f5abd0fa044d6604914a51692b197ccf5eac20a7175fe08f4ea8", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "80771fd8fc01a8494cf3e917", "attempt_id": "865cf39c2408", "run_group_id": "E4_core:A5|closed|qwen3b-llama3b:h2", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:28:03.982572Z", "task_family": "synthetic", "document_id": "syn_03024_2000", "document_hash": "ea10cd11ad08f01c50b0712bc481668857c9d2ddc8e6423968f2a31eeb97ac55", "question_id": "q09_revision", "split": "test", "cluster_id": "syn_03024_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "tokenizer_hashes": {"receiver": "b9d8872cf3a99da99e1fc4b0f71232a451632dc49ca4c6771549b4efa5c1c763"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 2, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1833, "query_tokens": 434, "message_tokens": 323, "message_utf8_bytes": 919, "cache_tensor_bytes": null, "serialized_payload_bytes": 919, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 10900.45654203277, "hops": {"1": 6942.490917048417, "2": 3957.965624984354}, "receiver_prefill_ms": 185.10508292820305, "ttft_ms": 11.534375022165477, "decode_ms": 11.476874933578074}, "total_wall_ms": 11108.943209052086, "ttft_ms": 11.534375022165477, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 66, "facts_retained": 8, "retention_rate": 0.12121212121212122, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.17647058823529413, "status": 0.16666666666666666, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 919}, "ledger_by_hop": {"1": {"facts_total": 66, "facts_retained": 6, "retention_rate": 0.09090909090909091, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.17647058823529413, "status": 0.0, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 754}, "2": {"facts_total": 66, "facts_retained": 8, "retention_rate": 0.12121212121212122, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.17647058823529413, "status": 0.16666666666666666, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 919}}}, "reference_key_id": "keys/synthetic/syn_03024_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "ec4605a52403f5abd0fa044d6604914a51692b197ccf5eac20a7175fe08f4ea8", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "ab9c0812bc1e5975c3077779", "attempt_id": "82966a7c33a8", "run_group_id": "E4_core:A5|closed|qwen3b-llama3b:h2", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:28:04.206904Z", "task_family": "synthetic", "document_id": "syn_03024_2000", "document_hash": "ea10cd11ad08f01c50b0712bc481668857c9d2ddc8e6423968f2a31eeb97ac55", "question_id": "q10_temporal", "split": "test", "cluster_id": "syn_03024_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "tokenizer_hashes": {"receiver": "b9d8872cf3a99da99e1fc4b0f71232a451632dc49ca4c6771549b4efa5c1c763"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 2, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1833, "query_tokens": 443, "message_tokens": 323, "message_utf8_bytes": 919, "cache_tensor_bytes": null, "serialized_payload_bytes": 919, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 10900.45654203277, "hops": {"1": 6942.490917048417, "2": 3957.965624984354}, "receiver_prefill_ms": 184.13154198788106, "ttft_ms": 11.771833058446646, "decode_ms": 11.547999922186136}, "total_wall_ms": 11108.296750113368, "ttft_ms": 11.771833058446646, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 66, "facts_retained": 8, "retention_rate": 0.12121212121212122, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.17647058823529413, "status": 0.16666666666666666, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 919}, "ledger_by_hop": {"1": {"facts_total": 66, "facts_retained": 6, "retention_rate": 0.09090909090909091, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.17647058823529413, "status": 0.0, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 754}, "2": {"facts_total": 66, "facts_retained": 8, "retention_rate": 0.12121212121212122, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.17647058823529413, "status": 0.16666666666666666, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 919}}}, "reference_key_id": "keys/synthetic/syn_03024_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "ec4605a52403f5abd0fa044d6604914a51692b197ccf5eac20a7175fe08f4ea8", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "2313839625baddfb110a28b1", "attempt_id": "9835423d86ce", "run_group_id": "E4_core:A5|closed|qwen3b-llama3b:h2", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:28:04.429592Z", "task_family": "synthetic", "document_id": "syn_03024_2000", "document_hash": "ea10cd11ad08f01c50b0712bc481668857c9d2ddc8e6423968f2a31eeb97ac55", "question_id": "q11_aggregation", "split": "test", "cluster_id": "syn_03024_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "tokenizer_hashes": {"receiver": "b9d8872cf3a99da99e1fc4b0f71232a451632dc49ca4c6771549b4efa5c1c763"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 2, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1833, "query_tokens": 438, "message_tokens": 323, "message_utf8_bytes": 919, "cache_tensor_bytes": null, "serialized_payload_bytes": 919, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 10900.45654203277, "hops": {"1": 6942.490917048417, "2": 3957.965624984354}, "receiver_prefill_ms": 183.85466700419784, "ttft_ms": 11.49316702503711, "decode_ms": 57.294458034448326}, "total_wall_ms": 11153.48612505477, "ttft_ms": 11.49316702503711, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 5}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "8,066,734", "evidence_artifact": {"ledger": {"facts_total": 66, "facts_retained": 8, "retention_rate": 0.12121212121212122, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.17647058823529413, "status": 0.16666666666666666, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 919}, "ledger_by_hop": {"1": {"facts_total": 66, "facts_retained": 6, "retention_rate": 0.09090909090909091, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.17647058823529413, "status": 0.0, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 754}, "2": {"facts_total": 66, "facts_retained": 8, "retention_rate": 0.12121212121212122, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.17647058823529413, "status": 0.16666666666666666, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 919}}}, "reference_key_id": "keys/synthetic/syn_03024_2000.json", "task_scores": {"correct": 0.0, "abstained": 0.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "ec4605a52403f5abd0fa044d6604914a51692b197ccf5eac20a7175fe08f4ea8", "output_hash": "dc9cab51923d7bfcae06d683e0d005a1575296aa981b1594b358ff8e12cd8aa2"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "34a7ad34fb6e6579a2a680ba", "attempt_id": "7db5b64893bb", "run_group_id": "E4_core:A5|closed|qwen3b-llama3b:h2", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:28:04.697315Z", "task_family": "synthetic", "document_id": "syn_03024_2000", "document_hash": "ea10cd11ad08f01c50b0712bc481668857c9d2ddc8e6423968f2a31eeb97ac55", "question_id": "q12_unit_conversion", "split": "test", "cluster_id": "syn_03024_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "tokenizer_hashes": {"receiver": "b9d8872cf3a99da99e1fc4b0f71232a451632dc49ca4c6771549b4efa5c1c763"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 2, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1833, "query_tokens": 434, "message_tokens": 323, "message_utf8_bytes": 919, "cache_tensor_bytes": null, "serialized_payload_bytes": 919, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 10900.45654203277, "hops": {"1": 6942.490917048417, "2": 3957.965624984354}, "receiver_prefill_ms": 181.80583301000297, "ttft_ms": 11.380832991562784, "decode_ms": 11.359915952198207}, "total_wall_ms": 11105.435084085912, "ttft_ms": 11.380832991562784, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 66, "facts_retained": 8, "retention_rate": 0.12121212121212122, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.17647058823529413, "status": 0.16666666666666666, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 919}, "ledger_by_hop": {"1": {"facts_total": 66, "facts_retained": 6, "retention_rate": 0.09090909090909091, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.17647058823529413, "status": 0.0, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 754}, "2": {"facts_total": 66, "facts_retained": 8, "retention_rate": 0.12121212121212122, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.17647058823529413, "status": 0.16666666666666666, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 919}}}, "reference_key_id": "keys/synthetic/syn_03024_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "ec4605a52403f5abd0fa044d6604914a51692b197ccf5eac20a7175fe08f4ea8", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "5cab9bb8fcc30736863eaa8a", "attempt_id": "adbce4bc2b9a", "run_group_id": "E4_core:A5|closed|qwen3b-llama3b:h2", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:28:04.917535Z", "task_family": "synthetic", "document_id": "syn_03024_2000", "document_hash": "ea10cd11ad08f01c50b0712bc481668857c9d2ddc8e6423968f2a31eeb97ac55", "question_id": "q13_unanswerable", "split": "test", "cluster_id": "syn_03024_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "tokenizer_hashes": {"receiver": "b9d8872cf3a99da99e1fc4b0f71232a451632dc49ca4c6771549b4efa5c1c763"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 2, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1833, "query_tokens": 438, "message_tokens": 323, "message_utf8_bytes": 919, "cache_tensor_bytes": null, "serialized_payload_bytes": 919, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 10900.45654203277, "hops": {"1": 6942.490917048417, "2": 3957.965624984354}, "receiver_prefill_ms": 183.6892500286922, "ttft_ms": 11.75812492147088, "decode_ms": 11.59737491980195}, "total_wall_ms": 11108.054375974461, "ttft_ms": 11.75812492147088, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 66, "facts_retained": 8, "retention_rate": 0.12121212121212122, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.17647058823529413, "status": 0.16666666666666666, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 919}, "ledger_by_hop": {"1": {"facts_total": 66, "facts_retained": 6, "retention_rate": 0.09090909090909091, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.17647058823529413, "status": 0.0, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 754}, "2": {"facts_total": 66, "facts_retained": 8, "retention_rate": 0.12121212121212122, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.17647058823529413, "status": 0.16666666666666666, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 919}}}, "reference_key_id": "keys/synthetic/syn_03024_2000.json", "task_scores": {"correct": 1.0, "abstained": 1.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "ec4605a52403f5abd0fa044d6604914a51692b197ccf5eac20a7175fe08f4ea8", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "3af432d259e3bca11790a66c", "attempt_id": "0388347f07e1", "run_group_id": "E4_core:A5|closed|qwen3b-llama3b:h2", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:28:05.141625Z", "task_family": "synthetic", "document_id": "syn_03024_2000", "document_hash": "ea10cd11ad08f01c50b0712bc481668857c9d2ddc8e6423968f2a31eeb97ac55", "question_id": "q14_unanswerable", "split": "test", "cluster_id": "syn_03024_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "tokenizer_hashes": {"receiver": "b9d8872cf3a99da99e1fc4b0f71232a451632dc49ca4c6771549b4efa5c1c763"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 2, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1833, "query_tokens": 440, "message_tokens": 323, "message_utf8_bytes": 919, "cache_tensor_bytes": null, "serialized_payload_bytes": 919, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 10900.45654203277, "hops": {"1": 6942.490917048417, "2": 3957.965624984354}, "receiver_prefill_ms": 184.25258295610547, "ttft_ms": 11.788749950937927, "decode_ms": 11.493749916553497}, "total_wall_ms": 11108.51729195565, "ttft_ms": 11.788749950937927, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 66, "facts_retained": 8, "retention_rate": 0.12121212121212122, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.17647058823529413, "status": 0.16666666666666666, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 919}, "ledger_by_hop": {"1": {"facts_total": 66, "facts_retained": 6, "retention_rate": 0.09090909090909091, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.17647058823529413, "status": 0.0, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 754}, "2": {"facts_total": 66, "facts_retained": 8, "retention_rate": 0.12121212121212122, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.17647058823529413, "status": 0.16666666666666666, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 919}}}, "reference_key_id": "keys/synthetic/syn_03024_2000.json", "task_scores": {"correct": 1.0, "abstained": 1.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "ec4605a52403f5abd0fa044d6604914a51692b197ccf5eac20a7175fe08f4ea8", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "6e3620e341e5f19190427c2a", "attempt_id": "c589a63df051", "run_group_id": "E4_core:A5|closed|qwen3b-llama3b:h4", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:28:13.983520Z", "task_family": "synthetic", "document_id": "syn_03024_2000", "document_hash": "ea10cd11ad08f01c50b0712bc481668857c9d2ddc8e6423968f2a31eeb97ac55", "question_id": "q00_single_fact", "split": "test", "cluster_id": "syn_03024_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "tokenizer_hashes": {"receiver": "b9d8872cf3a99da99e1fc4b0f71232a451632dc49ca4c6771549b4efa5c1c763"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 4, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1833, "query_tokens": 426, "message_tokens": 323, "message_utf8_bytes": 919, "cache_tensor_bytes": null, "serialized_payload_bytes": 919, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 19514.309624908492, "hops": {"1": 6942.490917048417, "2": 3957.965624984354, "3": 4647.70749991294, "4": 3966.145582962781}, "receiver_prefill_ms": 184.1999579919502, "ttft_ms": 11.594541021622717, "decode_ms": 11.508165975101292}, "total_wall_ms": 19722.014916944318, "ttft_ms": 11.594541021622717, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 66, "facts_retained": 8, "retention_rate": 0.12121212121212122, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.17647058823529413, "status": 0.16666666666666666, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 919}, "ledger_by_hop": {"1": {"facts_total": 66, "facts_retained": 6, "retention_rate": 0.09090909090909091, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.17647058823529413, "status": 0.0, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 754}, "2": {"facts_total": 66, "facts_retained": 8, "retention_rate": 0.12121212121212122, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.17647058823529413, "status": 0.16666666666666666, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 919}, "3": {"facts_total": 66, "facts_retained": 8, "retention_rate": 0.12121212121212122, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.17647058823529413, "status": 0.16666666666666666, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 919}, "4": {"facts_total": 66, "facts_retained": 8, "retention_rate": 0.12121212121212122, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.17647058823529413, "status": 0.16666666666666666, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 919}}}, "reference_key_id": "keys/synthetic/syn_03024_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "ec4605a52403f5abd0fa044d6604914a51692b197ccf5eac20a7175fe08f4ea8", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "341398af21254299fd05432c", "attempt_id": "e87eb141ab39", "run_group_id": "E4_core:A5|closed|qwen3b-llama3b:h4", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:28:14.207464Z", "task_family": "synthetic", "document_id": "syn_03024_2000", "document_hash": "ea10cd11ad08f01c50b0712bc481668857c9d2ddc8e6423968f2a31eeb97ac55", "question_id": "q01_single_fact", "split": "test", "cluster_id": "syn_03024_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "tokenizer_hashes": {"receiver": "b9d8872cf3a99da99e1fc4b0f71232a451632dc49ca4c6771549b4efa5c1c763"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 4, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1833, "query_tokens": 431, "message_tokens": 323, "message_utf8_bytes": 919, "cache_tensor_bytes": null, "serialized_payload_bytes": 919, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 19514.309624908492, "hops": {"1": 6942.490917048417, "2": 3957.965624984354, "3": 4647.70749991294, "4": 3966.145582962781}, "receiver_prefill_ms": 181.44158297218382, "ttft_ms": 11.374584049917758, "decode_ms": 11.39879203401506}, "total_wall_ms": 19718.946166918613, "ttft_ms": 11.374584049917758, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 66, "facts_retained": 8, "retention_rate": 0.12121212121212122, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.17647058823529413, "status": 0.16666666666666666, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 919}, "ledger_by_hop": {"1": {"facts_total": 66, "facts_retained": 6, "retention_rate": 0.09090909090909091, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.17647058823529413, "status": 0.0, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 754}, "2": {"facts_total": 66, "facts_retained": 8, "retention_rate": 0.12121212121212122, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.17647058823529413, "status": 0.16666666666666666, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 919}, "3": {"facts_total": 66, "facts_retained": 8, "retention_rate": 0.12121212121212122, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.17647058823529413, "status": 0.16666666666666666, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 919}, "4": {"facts_total": 66, "facts_retained": 8, "retention_rate": 0.12121212121212122, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.17647058823529413, "status": 0.16666666666666666, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 919}}}, "reference_key_id": "keys/synthetic/syn_03024_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "ec4605a52403f5abd0fa044d6604914a51692b197ccf5eac20a7175fe08f4ea8", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "8ea130f623cb39142ed4abc7", "attempt_id": "426194313c55", "run_group_id": "E4_core:A5|closed|qwen3b-llama3b:h4", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:28:14.427675Z", "task_family": "synthetic", "document_id": "syn_03024_2000", "document_hash": "ea10cd11ad08f01c50b0712bc481668857c9d2ddc8e6423968f2a31eeb97ac55", "question_id": "q02_exact_copy", "split": "test", "cluster_id": "syn_03024_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "tokenizer_hashes": {"receiver": "b9d8872cf3a99da99e1fc4b0f71232a451632dc49ca4c6771549b4efa5c1c763"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 4, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1833, "query_tokens": 431, "message_tokens": 323, "message_utf8_bytes": 919, "cache_tensor_bytes": null, "serialized_payload_bytes": 919, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 19514.309624908492, "hops": {"1": 6942.490917048417, "2": 3957.965624984354, "3": 4647.70749991294, "4": 3966.145582962781}, "receiver_prefill_ms": 182.22395901102573, "ttft_ms": 11.67129201348871, "decode_ms": 11.588084045797586}, "total_wall_ms": 19720.284540904686, "ttft_ms": 11.67129201348871, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 66, "facts_retained": 8, "retention_rate": 0.12121212121212122, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.17647058823529413, "status": 0.16666666666666666, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 919}, "ledger_by_hop": {"1": {"facts_total": 66, "facts_retained": 6, "retention_rate": 0.09090909090909091, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.17647058823529413, "status": 0.0, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 754}, "2": {"facts_total": 66, "facts_retained": 8, "retention_rate": 0.12121212121212122, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.17647058823529413, "status": 0.16666666666666666, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 919}, "3": {"facts_total": 66, "facts_retained": 8, "retention_rate": 0.12121212121212122, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.17647058823529413, "status": 0.16666666666666666, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 919}, "4": {"facts_total": 66, "facts_retained": 8, "retention_rate": 0.12121212121212122, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.17647058823529413, "status": 0.16666666666666666, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 919}}}, "reference_key_id": "keys/synthetic/syn_03024_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "ec4605a52403f5abd0fa044d6604914a51692b197ccf5eac20a7175fe08f4ea8", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "483c4c81d9e0a199d7040350", "attempt_id": "fabb6e48b401", "run_group_id": "E4_core:A5|closed|qwen3b-llama3b:h4", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:28:14.650090Z", "task_family": "synthetic", "document_id": "syn_03024_2000", "document_hash": "ea10cd11ad08f01c50b0712bc481668857c9d2ddc8e6423968f2a31eeb97ac55", "question_id": "q03_exact_copy", "split": "test", "cluster_id": "syn_03024_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "tokenizer_hashes": {"receiver": "b9d8872cf3a99da99e1fc4b0f71232a451632dc49ca4c6771549b4efa5c1c763"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 4, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1833, "query_tokens": 426, "message_tokens": 323, "message_utf8_bytes": 919, "cache_tensor_bytes": null, "serialized_payload_bytes": 919, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 19514.309624908492, "hops": {"1": 6942.490917048417, "2": 3957.965624984354, "3": 4647.70749991294, "4": 3966.145582962781}, "receiver_prefill_ms": 183.7006249697879, "ttft_ms": 11.6478749550879, "decode_ms": 11.421707924455404}, "total_wall_ms": 19721.45666682627, "ttft_ms": 11.6478749550879, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 66, "facts_retained": 8, "retention_rate": 0.12121212121212122, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.17647058823529413, "status": 0.16666666666666666, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 919}, "ledger_by_hop": {"1": {"facts_total": 66, "facts_retained": 6, "retention_rate": 0.09090909090909091, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.17647058823529413, "status": 0.0, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 754}, "2": {"facts_total": 66, "facts_retained": 8, "retention_rate": 0.12121212121212122, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.17647058823529413, "status": 0.16666666666666666, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 919}, "3": {"facts_total": 66, "facts_retained": 8, "retention_rate": 0.12121212121212122, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.17647058823529413, "status": 0.16666666666666666, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 919}, "4": {"facts_total": 66, "facts_retained": 8, "retention_rate": 0.12121212121212122, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.17647058823529413, "status": 0.16666666666666666, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 919}}}, "reference_key_id": "keys/synthetic/syn_03024_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "ec4605a52403f5abd0fa044d6604914a51692b197ccf5eac20a7175fe08f4ea8", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "8ac08b5146a1444ed44eceed", "attempt_id": "739d48ad990f", "run_group_id": "E4_core:A5|closed|qwen3b-llama3b:h4", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:28:14.872852Z", "task_family": "synthetic", "document_id": "syn_03024_2000", "document_hash": "ea10cd11ad08f01c50b0712bc481668857c9d2ddc8e6423968f2a31eeb97ac55", "question_id": "q04_multi_hop", "split": "test", "cluster_id": "syn_03024_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "tokenizer_hashes": {"receiver": "b9d8872cf3a99da99e1fc4b0f71232a451632dc49ca4c6771549b4efa5c1c763"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 4, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1833, "query_tokens": 433, "message_tokens": 323, "message_utf8_bytes": 919, "cache_tensor_bytes": null, "serialized_payload_bytes": 919, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 19514.309624908492, "hops": {"1": 6942.490917048417, "2": 3957.965624984354, "3": 4647.70749991294, "4": 3966.145582962781}, "receiver_prefill_ms": 184.29208302404732, "ttft_ms": 11.762207956053317, "decode_ms": 11.552499956451356}, "total_wall_ms": 19722.280749934725, "ttft_ms": 11.762207956053317, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 66, "facts_retained": 8, "retention_rate": 0.12121212121212122, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.17647058823529413, "status": 0.16666666666666666, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 919}, "ledger_by_hop": {"1": {"facts_total": 66, "facts_retained": 6, "retention_rate": 0.09090909090909091, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.17647058823529413, "status": 0.0, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 754}, "2": {"facts_total": 66, "facts_retained": 8, "retention_rate": 0.12121212121212122, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.17647058823529413, "status": 0.16666666666666666, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 919}, "3": {"facts_total": 66, "facts_retained": 8, "retention_rate": 0.12121212121212122, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.17647058823529413, "status": 0.16666666666666666, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 919}, "4": {"facts_total": 66, "facts_retained": 8, "retention_rate": 0.12121212121212122, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.17647058823529413, "status": 0.16666666666666666, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 919}}}, "reference_key_id": "keys/synthetic/syn_03024_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "ec4605a52403f5abd0fa044d6604914a51692b197ccf5eac20a7175fe08f4ea8", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "5a0be8a9e3fcf17d9d150608", "attempt_id": "cdcb8c8c491c", "run_group_id": "E4_core:A5|closed|qwen3b-llama3b:h4", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:28:15.096165Z", "task_family": "synthetic", "document_id": "syn_03024_2000", "document_hash": "ea10cd11ad08f01c50b0712bc481668857c9d2ddc8e6423968f2a31eeb97ac55", "question_id": "q05_multi_hop", "split": "test", "cluster_id": "syn_03024_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "tokenizer_hashes": {"receiver": "b9d8872cf3a99da99e1fc4b0f71232a451632dc49ca4c6771549b4efa5c1c763"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 4, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1833, "query_tokens": 435, "message_tokens": 323, "message_utf8_bytes": 919, "cache_tensor_bytes": null, "serialized_payload_bytes": 919, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 19514.309624908492, "hops": {"1": 6942.490917048417, "2": 3957.965624984354, "3": 4647.70749991294, "4": 3966.145582962781}, "receiver_prefill_ms": 183.27404104638845, "ttft_ms": 11.543708038516343, "decode_ms": 11.734125087969005}, "total_wall_ms": 19721.227041911334, "ttft_ms": 11.543708038516343, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 66, "facts_retained": 8, "retention_rate": 0.12121212121212122, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.17647058823529413, "status": 0.16666666666666666, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 919}, "ledger_by_hop": {"1": {"facts_total": 66, "facts_retained": 6, "retention_rate": 0.09090909090909091, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.17647058823529413, "status": 0.0, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 754}, "2": {"facts_total": 66, "facts_retained": 8, "retention_rate": 0.12121212121212122, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.17647058823529413, "status": 0.16666666666666666, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 919}, "3": {"facts_total": 66, "facts_retained": 8, "retention_rate": 0.12121212121212122, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.17647058823529413, "status": 0.16666666666666666, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 919}, "4": {"facts_total": 66, "facts_retained": 8, "retention_rate": 0.12121212121212122, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.17647058823529413, "status": 0.16666666666666666, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 919}}}, "reference_key_id": "keys/synthetic/syn_03024_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "ec4605a52403f5abd0fa044d6604914a51692b197ccf5eac20a7175fe08f4ea8", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "5f743a995d01fdf85b468378", "attempt_id": "e09d04d580e2", "run_group_id": "E4_core:A5|closed|qwen3b-llama3b:h4", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:28:15.318041Z", "task_family": "synthetic", "document_id": "syn_03024_2000", "document_hash": "ea10cd11ad08f01c50b0712bc481668857c9d2ddc8e6423968f2a31eeb97ac55", "question_id": "q06_negation", "split": "test", "cluster_id": "syn_03024_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "tokenizer_hashes": {"receiver": "b9d8872cf3a99da99e1fc4b0f71232a451632dc49ca4c6771549b4efa5c1c763"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 4, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1833, "query_tokens": 426, "message_tokens": 323, "message_utf8_bytes": 919, "cache_tensor_bytes": null, "serialized_payload_bytes": 919, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 19514.309624908492, "hops": {"1": 6942.490917048417, "2": 3957.965624984354, "3": 4647.70749991294, "4": 3966.145582962781}, "receiver_prefill_ms": 182.8124999301508, "ttft_ms": 11.591291055083275, "decode_ms": 11.483666021376848}, "total_wall_ms": 19720.546958851628, "ttft_ms": 11.591291055083275, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "No", "evidence_artifact": {"ledger": {"facts_total": 66, "facts_retained": 8, "retention_rate": 0.12121212121212122, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.17647058823529413, "status": 0.16666666666666666, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 919}, "ledger_by_hop": {"1": {"facts_total": 66, "facts_retained": 6, "retention_rate": 0.09090909090909091, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.17647058823529413, "status": 0.0, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 754}, "2": {"facts_total": 66, "facts_retained": 8, "retention_rate": 0.12121212121212122, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.17647058823529413, "status": 0.16666666666666666, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 919}, "3": {"facts_total": 66, "facts_retained": 8, "retention_rate": 0.12121212121212122, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.17647058823529413, "status": 0.16666666666666666, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 919}, "4": {"facts_total": 66, "facts_retained": 8, "retention_rate": 0.12121212121212122, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.17647058823529413, "status": 0.16666666666666666, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 919}}}, "reference_key_id": "keys/synthetic/syn_03024_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "ec4605a52403f5abd0fa044d6604914a51692b197ccf5eac20a7175fe08f4ea8", "output_hash": "1ea442a134b2a184bd5d40104401f2a37fbc09ccf3f4bc9da161c6099be3691d"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "927b79ffdb1963ca4e10d658", "attempt_id": "44aabd13aa1a", "run_group_id": "E4_core:A5|closed|qwen3b-llama3b:h4", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:28:15.539613Z", "task_family": "synthetic", "document_id": "syn_03024_2000", "document_hash": "ea10cd11ad08f01c50b0712bc481668857c9d2ddc8e6423968f2a31eeb97ac55", "question_id": "q07_negation", "split": "test", "cluster_id": "syn_03024_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "tokenizer_hashes": {"receiver": "b9d8872cf3a99da99e1fc4b0f71232a451632dc49ca4c6771549b4efa5c1c763"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 4, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1833, "query_tokens": 427, "message_tokens": 323, "message_utf8_bytes": 919, "cache_tensor_bytes": null, "serialized_payload_bytes": 919, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 19514.309624908492, "hops": {"1": 6942.490917048417, "2": 3957.965624984354, "3": 4647.70749991294, "4": 3966.145582962781}, "receiver_prefill_ms": 183.50520904641598, "ttft_ms": 11.576459044590592, "decode_ms": 11.547417030669749}, "total_wall_ms": 19721.285083913244, "ttft_ms": 11.576459044590592, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "No", "evidence_artifact": {"ledger": {"facts_total": 66, "facts_retained": 8, "retention_rate": 0.12121212121212122, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.17647058823529413, "status": 0.16666666666666666, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 919}, "ledger_by_hop": {"1": {"facts_total": 66, "facts_retained": 6, "retention_rate": 0.09090909090909091, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.17647058823529413, "status": 0.0, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 754}, "2": {"facts_total": 66, "facts_retained": 8, "retention_rate": 0.12121212121212122, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.17647058823529413, "status": 0.16666666666666666, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 919}, "3": {"facts_total": 66, "facts_retained": 8, "retention_rate": 0.12121212121212122, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.17647058823529413, "status": 0.16666666666666666, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 919}, "4": {"facts_total": 66, "facts_retained": 8, "retention_rate": 0.12121212121212122, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.17647058823529413, "status": 0.16666666666666666, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 919}}}, "reference_key_id": "keys/synthetic/syn_03024_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "ec4605a52403f5abd0fa044d6604914a51692b197ccf5eac20a7175fe08f4ea8", "output_hash": "1ea442a134b2a184bd5d40104401f2a37fbc09ccf3f4bc9da161c6099be3691d"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "57c7f9d0e47ccd7281b791a9", "attempt_id": "49e820a8cbe8", "run_group_id": "E4_core:A5|closed|qwen3b-llama3b:h4", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:28:15.760995Z", "task_family": "synthetic", "document_id": "syn_03024_2000", "document_hash": "ea10cd11ad08f01c50b0712bc481668857c9d2ddc8e6423968f2a31eeb97ac55", "question_id": "q08_revision", "split": "test", "cluster_id": "syn_03024_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "tokenizer_hashes": {"receiver": "b9d8872cf3a99da99e1fc4b0f71232a451632dc49ca4c6771549b4efa5c1c763"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 4, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1833, "query_tokens": 443, "message_tokens": 323, "message_utf8_bytes": 919, "cache_tensor_bytes": null, "serialized_payload_bytes": 919, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 19514.309624908492, "hops": {"1": 6942.490917048417, "2": 3957.965624984354, "3": 4647.70749991294, "4": 3966.145582962781}, "receiver_prefill_ms": 185.54966698866338, "ttft_ms": 11.717541958205402, "decode_ms": 11.486334027722478}, "total_wall_ms": 19723.56299997773, "ttft_ms": 11.717541958205402, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 66, "facts_retained": 8, "retention_rate": 0.12121212121212122, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.17647058823529413, "status": 0.16666666666666666, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 919}, "ledger_by_hop": {"1": {"facts_total": 66, "facts_retained": 6, "retention_rate": 0.09090909090909091, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.17647058823529413, "status": 0.0, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 754}, "2": {"facts_total": 66, "facts_retained": 8, "retention_rate": 0.12121212121212122, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.17647058823529413, "status": 0.16666666666666666, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 919}, "3": {"facts_total": 66, "facts_retained": 8, "retention_rate": 0.12121212121212122, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.17647058823529413, "status": 0.16666666666666666, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 919}, "4": {"facts_total": 66, "facts_retained": 8, "retention_rate": 0.12121212121212122, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.17647058823529413, "status": 0.16666666666666666, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 919}}}, "reference_key_id": "keys/synthetic/syn_03024_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "ec4605a52403f5abd0fa044d6604914a51692b197ccf5eac20a7175fe08f4ea8", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "ecad42e6593cb2bb3039820e", "attempt_id": "80eb3587ba88", "run_group_id": "E4_core:A5|closed|qwen3b-llama3b:h4", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:28:15.984777Z", "task_family": "synthetic", "document_id": "syn_03024_2000", "document_hash": "ea10cd11ad08f01c50b0712bc481668857c9d2ddc8e6423968f2a31eeb97ac55", "question_id": "q09_revision", "split": "test", "cluster_id": "syn_03024_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "tokenizer_hashes": {"receiver": "b9d8872cf3a99da99e1fc4b0f71232a451632dc49ca4c6771549b4efa5c1c763"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 4, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1833, "query_tokens": 434, "message_tokens": 323, "message_utf8_bytes": 919, "cache_tensor_bytes": null, "serialized_payload_bytes": 919, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 19514.309624908492, "hops": {"1": 6942.490917048417, "2": 3957.965624984354, "3": 4647.70749991294, "4": 3966.145582962781}, "receiver_prefill_ms": 183.88687493279576, "ttft_ms": 11.714499909430742, "decode_ms": 11.591916088946164}, "total_wall_ms": 19721.97991691064, "ttft_ms": 11.714499909430742, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 66, "facts_retained": 8, "retention_rate": 0.12121212121212122, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.17647058823529413, "status": 0.16666666666666666, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 919}, "ledger_by_hop": {"1": {"facts_total": 66, "facts_retained": 6, "retention_rate": 0.09090909090909091, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.17647058823529413, "status": 0.0, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 754}, "2": {"facts_total": 66, "facts_retained": 8, "retention_rate": 0.12121212121212122, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.17647058823529413, "status": 0.16666666666666666, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 919}, "3": {"facts_total": 66, "facts_retained": 8, "retention_rate": 0.12121212121212122, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.17647058823529413, "status": 0.16666666666666666, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 919}, "4": {"facts_total": 66, "facts_retained": 8, "retention_rate": 0.12121212121212122, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.17647058823529413, "status": 0.16666666666666666, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 919}}}, "reference_key_id": "keys/synthetic/syn_03024_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "ec4605a52403f5abd0fa044d6604914a51692b197ccf5eac20a7175fe08f4ea8", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "55580b0dd5cc4d278dd95af6", "attempt_id": "a2a15e8a51f4", "run_group_id": "E4_core:A5|closed|qwen3b-llama3b:h4", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:28:16.208789Z", "task_family": "synthetic", "document_id": "syn_03024_2000", "document_hash": "ea10cd11ad08f01c50b0712bc481668857c9d2ddc8e6423968f2a31eeb97ac55", "question_id": "q10_temporal", "split": "test", "cluster_id": "syn_03024_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "tokenizer_hashes": {"receiver": "b9d8872cf3a99da99e1fc4b0f71232a451632dc49ca4c6771549b4efa5c1c763"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 4, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1833, "query_tokens": 443, "message_tokens": 323, "message_utf8_bytes": 919, "cache_tensor_bytes": null, "serialized_payload_bytes": 919, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 19514.309624908492, "hops": {"1": 6942.490917048417, "2": 3957.965624984354, "3": 4647.70749991294, "4": 3966.145582962781}, "receiver_prefill_ms": 183.63045796286315, "ttft_ms": 11.513874982483685, "decode_ms": 11.528540984727442}, "total_wall_ms": 19721.35595791042, "ttft_ms": 11.513874982483685, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 66, "facts_retained": 8, "retention_rate": 0.12121212121212122, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.17647058823529413, "status": 0.16666666666666666, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 919}, "ledger_by_hop": {"1": {"facts_total": 66, "facts_retained": 6, "retention_rate": 0.09090909090909091, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.17647058823529413, "status": 0.0, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 754}, "2": {"facts_total": 66, "facts_retained": 8, "retention_rate": 0.12121212121212122, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.17647058823529413, "status": 0.16666666666666666, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 919}, "3": {"facts_total": 66, "facts_retained": 8, "retention_rate": 0.12121212121212122, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.17647058823529413, "status": 0.16666666666666666, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 919}, "4": {"facts_total": 66, "facts_retained": 8, "retention_rate": 0.12121212121212122, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.17647058823529413, "status": 0.16666666666666666, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 919}}}, "reference_key_id": "keys/synthetic/syn_03024_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "ec4605a52403f5abd0fa044d6604914a51692b197ccf5eac20a7175fe08f4ea8", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "8f08cbdd34b9bc51dbbee7b9", "attempt_id": "ee87677406ad", "run_group_id": "E4_core:A5|closed|qwen3b-llama3b:h4", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:28:16.431465Z", "task_family": "synthetic", "document_id": "syn_03024_2000", "document_hash": "ea10cd11ad08f01c50b0712bc481668857c9d2ddc8e6423968f2a31eeb97ac55", "question_id": "q11_aggregation", "split": "test", "cluster_id": "syn_03024_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "tokenizer_hashes": {"receiver": "b9d8872cf3a99da99e1fc4b0f71232a451632dc49ca4c6771549b4efa5c1c763"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 4, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1833, "query_tokens": 438, "message_tokens": 323, "message_utf8_bytes": 919, "cache_tensor_bytes": null, "serialized_payload_bytes": 919, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 19514.309624908492, "hops": {"1": 6942.490917048417, "2": 3957.965624984354, "3": 4647.70749991294, "4": 3966.145582962781}, "receiver_prefill_ms": 184.42091590259224, "ttft_ms": 11.530417017638683, "decode_ms": 57.694667018949986}, "total_wall_ms": 19768.310999847017, "ttft_ms": 11.530417017638683, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 5}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "8,066,734", "evidence_artifact": {"ledger": {"facts_total": 66, "facts_retained": 8, "retention_rate": 0.12121212121212122, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.17647058823529413, "status": 0.16666666666666666, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 919}, "ledger_by_hop": {"1": {"facts_total": 66, "facts_retained": 6, "retention_rate": 0.09090909090909091, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.17647058823529413, "status": 0.0, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 754}, "2": {"facts_total": 66, "facts_retained": 8, "retention_rate": 0.12121212121212122, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.17647058823529413, "status": 0.16666666666666666, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 919}, "3": {"facts_total": 66, "facts_retained": 8, "retention_rate": 0.12121212121212122, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.17647058823529413, "status": 0.16666666666666666, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 919}, "4": {"facts_total": 66, "facts_retained": 8, "retention_rate": 0.12121212121212122, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.17647058823529413, "status": 0.16666666666666666, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 919}}}, "reference_key_id": "keys/synthetic/syn_03024_2000.json", "task_scores": {"correct": 0.0, "abstained": 0.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "ec4605a52403f5abd0fa044d6604914a51692b197ccf5eac20a7175fe08f4ea8", "output_hash": "dc9cab51923d7bfcae06d683e0d005a1575296aa981b1594b358ff8e12cd8aa2"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "6a29544c45d697374848293b", "attempt_id": "6b5c885dec96", "run_group_id": "E4_core:A5|closed|qwen3b-llama3b:h4", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:28:16.700159Z", "task_family": "synthetic", "document_id": "syn_03024_2000", "document_hash": "ea10cd11ad08f01c50b0712bc481668857c9d2ddc8e6423968f2a31eeb97ac55", "question_id": "q12_unit_conversion", "split": "test", "cluster_id": "syn_03024_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "tokenizer_hashes": {"receiver": "b9d8872cf3a99da99e1fc4b0f71232a451632dc49ca4c6771549b4efa5c1c763"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 4, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1833, "query_tokens": 434, "message_tokens": 323, "message_utf8_bytes": 919, "cache_tensor_bytes": null, "serialized_payload_bytes": 919, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 19514.309624908492, "hops": {"1": 6942.490917048417, "2": 3957.965624984354, "3": 4647.70749991294, "4": 3966.145582962781}, "receiver_prefill_ms": 184.90699992980808, "ttft_ms": 11.957417009398341, "decode_ms": 11.740165995433927}, "total_wall_ms": 19723.47149986308, "ttft_ms": 11.957417009398341, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 66, "facts_retained": 8, "retention_rate": 0.12121212121212122, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.17647058823529413, "status": 0.16666666666666666, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 919}, "ledger_by_hop": {"1": {"facts_total": 66, "facts_retained": 6, "retention_rate": 0.09090909090909091, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.17647058823529413, "status": 0.0, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 754}, "2": {"facts_total": 66, "facts_retained": 8, "retention_rate": 0.12121212121212122, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.17647058823529413, "status": 0.16666666666666666, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 919}, "3": {"facts_total": 66, "facts_retained": 8, "retention_rate": 0.12121212121212122, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.17647058823529413, "status": 0.16666666666666666, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 919}, "4": {"facts_total": 66, "facts_retained": 8, "retention_rate": 0.12121212121212122, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.17647058823529413, "status": 0.16666666666666666, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 919}}}, "reference_key_id": "keys/synthetic/syn_03024_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "ec4605a52403f5abd0fa044d6604914a51692b197ccf5eac20a7175fe08f4ea8", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "37b13692712397bfb9969bdd", "attempt_id": "4ee39ef4a927", "run_group_id": "E4_core:A5|closed|qwen3b-llama3b:h4", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:28:16.924776Z", "task_family": "synthetic", "document_id": "syn_03024_2000", "document_hash": "ea10cd11ad08f01c50b0712bc481668857c9d2ddc8e6423968f2a31eeb97ac55", "question_id": "q13_unanswerable", "split": "test", "cluster_id": "syn_03024_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "tokenizer_hashes": {"receiver": "b9d8872cf3a99da99e1fc4b0f71232a451632dc49ca4c6771549b4efa5c1c763"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 4, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1833, "query_tokens": 438, "message_tokens": 323, "message_utf8_bytes": 919, "cache_tensor_bytes": null, "serialized_payload_bytes": 919, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 19514.309624908492, "hops": {"1": 6942.490917048417, "2": 3957.965624984354, "3": 4647.70749991294, "4": 3966.145582962781}, "receiver_prefill_ms": 181.35991704184562, "ttft_ms": 11.548041016794741, "decode_ms": 11.508374940603971}, "total_wall_ms": 19719.093291903846, "ttft_ms": 11.548041016794741, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 66, "facts_retained": 8, "retention_rate": 0.12121212121212122, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.17647058823529413, "status": 0.16666666666666666, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 919}, "ledger_by_hop": {"1": {"facts_total": 66, "facts_retained": 6, "retention_rate": 0.09090909090909091, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.17647058823529413, "status": 0.0, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 754}, "2": {"facts_total": 66, "facts_retained": 8, "retention_rate": 0.12121212121212122, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.17647058823529413, "status": 0.16666666666666666, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 919}, "3": {"facts_total": 66, "facts_retained": 8, "retention_rate": 0.12121212121212122, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.17647058823529413, "status": 0.16666666666666666, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 919}, "4": {"facts_total": 66, "facts_retained": 8, "retention_rate": 0.12121212121212122, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.17647058823529413, "status": 0.16666666666666666, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 919}}}, "reference_key_id": "keys/synthetic/syn_03024_2000.json", "task_scores": {"correct": 1.0, "abstained": 1.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "ec4605a52403f5abd0fa044d6604914a51692b197ccf5eac20a7175fe08f4ea8", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "0c07d95ce1c0fdcfcc593ccb", "attempt_id": "15207700b686", "run_group_id": "E4_core:A5|closed|qwen3b-llama3b:h4", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:28:17.146310Z", "task_family": "synthetic", "document_id": "syn_03024_2000", "document_hash": "ea10cd11ad08f01c50b0712bc481668857c9d2ddc8e6423968f2a31eeb97ac55", "question_id": "q14_unanswerable", "split": "test", "cluster_id": "syn_03024_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "tokenizer_hashes": {"receiver": "b9d8872cf3a99da99e1fc4b0f71232a451632dc49ca4c6771549b4efa5c1c763"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 4, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1833, "query_tokens": 440, "message_tokens": 323, "message_utf8_bytes": 919, "cache_tensor_bytes": null, "serialized_payload_bytes": 919, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 19514.309624908492, "hops": {"1": 6942.490917048417, "2": 3957.965624984354, "3": 4647.70749991294, "4": 3966.145582962781}, "receiver_prefill_ms": 184.65470790397376, "ttft_ms": 11.542166932485998, "decode_ms": 11.513334000483155}, "total_wall_ms": 19722.375583834946, "ttft_ms": 11.542166932485998, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 66, "facts_retained": 8, "retention_rate": 0.12121212121212122, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.17647058823529413, "status": 0.16666666666666666, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 919}, "ledger_by_hop": {"1": {"facts_total": 66, "facts_retained": 6, "retention_rate": 0.09090909090909091, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.17647058823529413, "status": 0.0, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 754}, "2": {"facts_total": 66, "facts_retained": 8, "retention_rate": 0.12121212121212122, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.17647058823529413, "status": 0.16666666666666666, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 919}, "3": {"facts_total": 66, "facts_retained": 8, "retention_rate": 0.12121212121212122, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.17647058823529413, "status": 0.16666666666666666, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 919}, "4": {"facts_total": 66, "facts_retained": 8, "retention_rate": 0.12121212121212122, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.17647058823529413, "status": 0.16666666666666666, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 919}}}, "reference_key_id": "keys/synthetic/syn_03024_2000.json", "task_scores": {"correct": 1.0, "abstained": 1.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "ec4605a52403f5abd0fa044d6604914a51692b197ccf5eac20a7175fe08f4ea8", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "39e2a6f143e6938c0c3db118", "attempt_id": "09f2a992287b", "run_group_id": "E4_core:A5|closed|qwen3b-llama3b:h8", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:28:34.649674Z", "task_family": "synthetic", "document_id": "syn_03024_2000", "document_hash": "ea10cd11ad08f01c50b0712bc481668857c9d2ddc8e6423968f2a31eeb97ac55", "question_id": "q00_single_fact", "split": "test", "cluster_id": "syn_03024_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "tokenizer_hashes": {"receiver": "b9d8872cf3a99da99e1fc4b0f71232a451632dc49ca4c6771549b4efa5c1c763"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 8, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1833, "query_tokens": 426, "message_tokens": 323, "message_utf8_bytes": 919, "cache_tensor_bytes": null, "serialized_payload_bytes": 919, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 36789.773125899956, "hops": {"1": 6942.490917048417, "2": 3957.965624984354, "3": 4647.70749991294, "4": 3966.145582962781, "5": 4663.1102920509875, "6": 3976.8799169687554, "7": 4664.292625035159, "8": 3971.1806669365615}, "receiver_prefill_ms": 183.08550002984703, "ttft_ms": 11.601291946135461, "decode_ms": 11.484125046990812}, "total_wall_ms": 36996.3697087951, "ttft_ms": 11.601291946135461, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 66, "facts_retained": 8, "retention_rate": 0.12121212121212122, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.17647058823529413, "status": 0.16666666666666666, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 919}, "ledger_by_hop": {"1": {"facts_total": 66, "facts_retained": 6, "retention_rate": 0.09090909090909091, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.17647058823529413, "status": 0.0, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 754}, "2": {"facts_total": 66, "facts_retained": 8, "retention_rate": 0.12121212121212122, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.17647058823529413, "status": 0.16666666666666666, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 919}, "3": {"facts_total": 66, "facts_retained": 8, "retention_rate": 0.12121212121212122, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.17647058823529413, "status": 0.16666666666666666, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 919}, "4": {"facts_total": 66, "facts_retained": 8, "retention_rate": 0.12121212121212122, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.17647058823529413, "status": 0.16666666666666666, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 919}, "5": {"facts_total": 66, "facts_retained": 8, "retention_rate": 0.12121212121212122, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.17647058823529413, "status": 0.16666666666666666, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 919}, "6": {"facts_total": 66, "facts_retained": 8, "retention_rate": 0.12121212121212122, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.17647058823529413, "status": 0.16666666666666666, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 919}, "7": {"facts_total": 66, "facts_retained": 8, "retention_rate": 0.12121212121212122, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.17647058823529413, "status": 0.16666666666666666, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 919}, "8": {"facts_total": 66, "facts_retained": 8, "retention_rate": 0.12121212121212122, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.17647058823529413, "status": 0.16666666666666666, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 919}}}, "reference_key_id": "keys/synthetic/syn_03024_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "ec4605a52403f5abd0fa044d6604914a51692b197ccf5eac20a7175fe08f4ea8", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "899c3af9a862ddf9b1c9041f", "attempt_id": "50e48687a91f", "run_group_id": "E4_core:A5|closed|qwen3b-llama3b:h8", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:28:34.872670Z", "task_family": "synthetic", "document_id": "syn_03024_2000", "document_hash": "ea10cd11ad08f01c50b0712bc481668857c9d2ddc8e6423968f2a31eeb97ac55", "question_id": "q01_single_fact", "split": "test", "cluster_id": "syn_03024_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "tokenizer_hashes": {"receiver": "b9d8872cf3a99da99e1fc4b0f71232a451632dc49ca4c6771549b4efa5c1c763"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 8, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1833, "query_tokens": 431, "message_tokens": 323, "message_utf8_bytes": 919, "cache_tensor_bytes": null, "serialized_payload_bytes": 919, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 36789.773125899956, "hops": {"1": 6942.490917048417, "2": 3957.965624984354, "3": 4647.70749991294, "4": 3966.145582962781, "5": 4663.1102920509875, "6": 3976.8799169687554, "7": 4664.292625035159, "8": 3971.1806669365615}, "receiver_prefill_ms": 182.90829204488546, "ttft_ms": 11.730083962902427, "decode_ms": 11.43295899964869}, "total_wall_ms": 36996.33025098592, "ttft_ms": 11.730083962902427, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 66, "facts_retained": 8, "retention_rate": 0.12121212121212122, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.17647058823529413, "status": 0.16666666666666666, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 919}, "ledger_by_hop": {"1": {"facts_total": 66, "facts_retained": 6, "retention_rate": 0.09090909090909091, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.17647058823529413, "status": 0.0, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 754}, "2": {"facts_total": 66, "facts_retained": 8, "retention_rate": 0.12121212121212122, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.17647058823529413, "status": 0.16666666666666666, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 919}, "3": {"facts_total": 66, "facts_retained": 8, "retention_rate": 0.12121212121212122, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.17647058823529413, "status": 0.16666666666666666, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 919}, "4": {"facts_total": 66, "facts_retained": 8, "retention_rate": 0.12121212121212122, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.17647058823529413, "status": 0.16666666666666666, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 919}, "5": {"facts_total": 66, "facts_retained": 8, "retention_rate": 0.12121212121212122, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.17647058823529413, "status": 0.16666666666666666, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 919}, "6": {"facts_total": 66, "facts_retained": 8, "retention_rate": 0.12121212121212122, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.17647058823529413, "status": 0.16666666666666666, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 919}, "7": {"facts_total": 66, "facts_retained": 8, "retention_rate": 0.12121212121212122, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.17647058823529413, "status": 0.16666666666666666, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 919}, "8": {"facts_total": 66, "facts_retained": 8, "retention_rate": 0.12121212121212122, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.17647058823529413, "status": 0.16666666666666666, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 919}}}, "reference_key_id": "keys/synthetic/syn_03024_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "ec4605a52403f5abd0fa044d6604914a51692b197ccf5eac20a7175fe08f4ea8", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "8b2082f56337184cddfcab15", "attempt_id": "81493b7edde7", "run_group_id": "E4_core:A5|closed|qwen3b-llama3b:h8", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:28:35.094787Z", "task_family": "synthetic", "document_id": "syn_03024_2000", "document_hash": "ea10cd11ad08f01c50b0712bc481668857c9d2ddc8e6423968f2a31eeb97ac55", "question_id": "q02_exact_copy", "split": "test", "cluster_id": "syn_03024_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "tokenizer_hashes": {"receiver": "b9d8872cf3a99da99e1fc4b0f71232a451632dc49ca4c6771549b4efa5c1c763"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 8, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1833, "query_tokens": 431, "message_tokens": 323, "message_utf8_bytes": 919, "cache_tensor_bytes": null, "serialized_payload_bytes": 919, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 36789.773125899956, "hops": {"1": 6942.490917048417, "2": 3957.965624984354, "3": 4647.70749991294, "4": 3966.145582962781, "5": 4663.1102920509875, "6": 3976.8799169687554, "7": 4664.292625035159, "8": 3971.1806669365615}, "receiver_prefill_ms": 185.24195801001042, "ttft_ms": 11.776624945923686, "decode_ms": 11.558542028069496}, "total_wall_ms": 36998.90954291914, "ttft_ms": 11.776624945923686, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 66, "facts_retained": 8, "retention_rate": 0.12121212121212122, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.17647058823529413, "status": 0.16666666666666666, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 919}, "ledger_by_hop": {"1": {"facts_total": 66, "facts_retained": 6, "retention_rate": 0.09090909090909091, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.17647058823529413, "status": 0.0, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 754}, "2": {"facts_total": 66, "facts_retained": 8, "retention_rate": 0.12121212121212122, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.17647058823529413, "status": 0.16666666666666666, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 919}, "3": {"facts_total": 66, "facts_retained": 8, "retention_rate": 0.12121212121212122, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.17647058823529413, "status": 0.16666666666666666, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 919}, "4": {"facts_total": 66, "facts_retained": 8, "retention_rate": 0.12121212121212122, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.17647058823529413, "status": 0.16666666666666666, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 919}, "5": {"facts_total": 66, "facts_retained": 8, "retention_rate": 0.12121212121212122, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.17647058823529413, "status": 0.16666666666666666, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 919}, "6": {"facts_total": 66, "facts_retained": 8, "retention_rate": 0.12121212121212122, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.17647058823529413, "status": 0.16666666666666666, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 919}, "7": {"facts_total": 66, "facts_retained": 8, "retention_rate": 0.12121212121212122, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.17647058823529413, "status": 0.16666666666666666, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 919}, "8": {"facts_total": 66, "facts_retained": 8, "retention_rate": 0.12121212121212122, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.17647058823529413, "status": 0.16666666666666666, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 919}}}, "reference_key_id": "keys/synthetic/syn_03024_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "ec4605a52403f5abd0fa044d6604914a51692b197ccf5eac20a7175fe08f4ea8", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "bef343ba48fd744c7b5dbff5", "attempt_id": "1d8241ac7869", "run_group_id": "E4_core:A5|closed|qwen3b-llama3b:h8", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:28:35.319566Z", "task_family": "synthetic", "document_id": "syn_03024_2000", "document_hash": "ea10cd11ad08f01c50b0712bc481668857c9d2ddc8e6423968f2a31eeb97ac55", "question_id": "q03_exact_copy", "split": "test", "cluster_id": "syn_03024_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "tokenizer_hashes": {"receiver": "b9d8872cf3a99da99e1fc4b0f71232a451632dc49ca4c6771549b4efa5c1c763"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 8, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1833, "query_tokens": 426, "message_tokens": 323, "message_utf8_bytes": 919, "cache_tensor_bytes": null, "serialized_payload_bytes": 919, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 36789.773125899956, "hops": {"1": 6942.490917048417, "2": 3957.965624984354, "3": 4647.70749991294, "4": 3966.145582962781, "5": 4663.1102920509875, "6": 3976.8799169687554, "7": 4664.292625035159, "8": 3971.1806669365615}, "receiver_prefill_ms": 184.03991602826864, "ttft_ms": 11.676999973133206, "decode_ms": 11.541791958734393}, "total_wall_ms": 36997.526583843865, "ttft_ms": 11.676999973133206, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 66, "facts_retained": 8, "retention_rate": 0.12121212121212122, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.17647058823529413, "status": 0.16666666666666666, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 919}, "ledger_by_hop": {"1": {"facts_total": 66, "facts_retained": 6, "retention_rate": 0.09090909090909091, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.17647058823529413, "status": 0.0, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 754}, "2": {"facts_total": 66, "facts_retained": 8, "retention_rate": 0.12121212121212122, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.17647058823529413, "status": 0.16666666666666666, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 919}, "3": {"facts_total": 66, "facts_retained": 8, "retention_rate": 0.12121212121212122, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.17647058823529413, "status": 0.16666666666666666, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 919}, "4": {"facts_total": 66, "facts_retained": 8, "retention_rate": 0.12121212121212122, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.17647058823529413, "status": 0.16666666666666666, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 919}, "5": {"facts_total": 66, "facts_retained": 8, "retention_rate": 0.12121212121212122, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.17647058823529413, "status": 0.16666666666666666, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 919}, "6": {"facts_total": 66, "facts_retained": 8, "retention_rate": 0.12121212121212122, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.17647058823529413, "status": 0.16666666666666666, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 919}, "7": {"facts_total": 66, "facts_retained": 8, "retention_rate": 0.12121212121212122, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.17647058823529413, "status": 0.16666666666666666, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 919}, "8": {"facts_total": 66, "facts_retained": 8, "retention_rate": 0.12121212121212122, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.17647058823529413, "status": 0.16666666666666666, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 919}}}, "reference_key_id": "keys/synthetic/syn_03024_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "ec4605a52403f5abd0fa044d6604914a51692b197ccf5eac20a7175fe08f4ea8", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "443e892f11ce691c38d53866", "attempt_id": "0e754a5883fc", "run_group_id": "E4_core:A5|closed|qwen3b-llama3b:h8", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:28:35.544813Z", "task_family": "synthetic", "document_id": "syn_03024_2000", "document_hash": "ea10cd11ad08f01c50b0712bc481668857c9d2ddc8e6423968f2a31eeb97ac55", "question_id": "q04_multi_hop", "split": "test", "cluster_id": "syn_03024_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "tokenizer_hashes": {"receiver": "b9d8872cf3a99da99e1fc4b0f71232a451632dc49ca4c6771549b4efa5c1c763"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 8, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1833, "query_tokens": 433, "message_tokens": 323, "message_utf8_bytes": 919, "cache_tensor_bytes": null, "serialized_payload_bytes": 919, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 36789.773125899956, "hops": {"1": 6942.490917048417, "2": 3957.965624984354, "3": 4647.70749991294, "4": 3966.145582962781, "5": 4663.1102920509875, "6": 3976.8799169687554, "7": 4664.292625035159, "8": 3971.1806669365615}, "receiver_prefill_ms": 185.23162510246038, "ttft_ms": 11.700915987603366, "decode_ms": 11.60245796199888}, "total_wall_ms": 36998.68741782848, "ttft_ms": 11.700915987603366, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 66, "facts_retained": 8, "retention_rate": 0.12121212121212122, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.17647058823529413, "status": 0.16666666666666666, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 919}, "ledger_by_hop": {"1": {"facts_total": 66, "facts_retained": 6, "retention_rate": 0.09090909090909091, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.17647058823529413, "status": 0.0, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 754}, "2": {"facts_total": 66, "facts_retained": 8, "retention_rate": 0.12121212121212122, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.17647058823529413, "status": 0.16666666666666666, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 919}, "3": {"facts_total": 66, "facts_retained": 8, "retention_rate": 0.12121212121212122, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.17647058823529413, "status": 0.16666666666666666, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 919}, "4": {"facts_total": 66, "facts_retained": 8, "retention_rate": 0.12121212121212122, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.17647058823529413, "status": 0.16666666666666666, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 919}, "5": {"facts_total": 66, "facts_retained": 8, "retention_rate": 0.12121212121212122, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.17647058823529413, "status": 0.16666666666666666, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 919}, "6": {"facts_total": 66, "facts_retained": 8, "retention_rate": 0.12121212121212122, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.17647058823529413, "status": 0.16666666666666666, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 919}, "7": {"facts_total": 66, "facts_retained": 8, "retention_rate": 0.12121212121212122, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.17647058823529413, "status": 0.16666666666666666, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 919}, "8": {"facts_total": 66, "facts_retained": 8, "retention_rate": 0.12121212121212122, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.17647058823529413, "status": 0.16666666666666666, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 919}}}, "reference_key_id": "keys/synthetic/syn_03024_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "ec4605a52403f5abd0fa044d6604914a51692b197ccf5eac20a7175fe08f4ea8", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "b9c8f02f284f8c808bf7cb73", "attempt_id": "85dee5f358ed", "run_group_id": "E4_core:A5|closed|qwen3b-llama3b:h8", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:28:35.768737Z", "task_family": "synthetic", "document_id": "syn_03024_2000", "document_hash": "ea10cd11ad08f01c50b0712bc481668857c9d2ddc8e6423968f2a31eeb97ac55", "question_id": "q05_multi_hop", "split": "test", "cluster_id": "syn_03024_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "tokenizer_hashes": {"receiver": "b9d8872cf3a99da99e1fc4b0f71232a451632dc49ca4c6771549b4efa5c1c763"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 8, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1833, "query_tokens": 435, "message_tokens": 323, "message_utf8_bytes": 919, "cache_tensor_bytes": null, "serialized_payload_bytes": 919, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 36789.773125899956, "hops": {"1": 6942.490917048417, "2": 3957.965624984354, "3": 4647.70749991294, "4": 3966.145582962781, "5": 4663.1102920509875, "6": 3976.8799169687554, "7": 4664.292625035159, "8": 3971.1806669365615}, "receiver_prefill_ms": 184.88633399829268, "ttft_ms": 11.612833011895418, "decode_ms": 11.551167001016438}, "total_wall_ms": 36998.181041912176, "ttft_ms": 11.612833011895418, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 66, "facts_retained": 8, "retention_rate": 0.12121212121212122, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.17647058823529413, "status": 0.16666666666666666, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 919}, "ledger_by_hop": {"1": {"facts_total": 66, "facts_retained": 6, "retention_rate": 0.09090909090909091, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.17647058823529413, "status": 0.0, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 754}, "2": {"facts_total": 66, "facts_retained": 8, "retention_rate": 0.12121212121212122, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.17647058823529413, "status": 0.16666666666666666, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 919}, "3": {"facts_total": 66, "facts_retained": 8, "retention_rate": 0.12121212121212122, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.17647058823529413, "status": 0.16666666666666666, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 919}, "4": {"facts_total": 66, "facts_retained": 8, "retention_rate": 0.12121212121212122, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.17647058823529413, "status": 0.16666666666666666, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 919}, "5": {"facts_total": 66, "facts_retained": 8, "retention_rate": 0.12121212121212122, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.17647058823529413, "status": 0.16666666666666666, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 919}, "6": {"facts_total": 66, "facts_retained": 8, "retention_rate": 0.12121212121212122, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.17647058823529413, "status": 0.16666666666666666, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 919}, "7": {"facts_total": 66, "facts_retained": 8, "retention_rate": 0.12121212121212122, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.17647058823529413, "status": 0.16666666666666666, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 919}, "8": {"facts_total": 66, "facts_retained": 8, "retention_rate": 0.12121212121212122, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.17647058823529413, "status": 0.16666666666666666, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 919}}}, "reference_key_id": "keys/synthetic/syn_03024_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "ec4605a52403f5abd0fa044d6604914a51692b197ccf5eac20a7175fe08f4ea8", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "31988c7fcdb04eae184ef6e0", "attempt_id": "244405cfe9ab", "run_group_id": "E4_core:A5|closed|qwen3b-llama3b:h8", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:28:35.992030Z", "task_family": "synthetic", "document_id": "syn_03024_2000", "document_hash": "ea10cd11ad08f01c50b0712bc481668857c9d2ddc8e6423968f2a31eeb97ac55", "question_id": "q06_negation", "split": "test", "cluster_id": "syn_03024_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "tokenizer_hashes": {"receiver": "b9d8872cf3a99da99e1fc4b0f71232a451632dc49ca4c6771549b4efa5c1c763"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 8, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1833, "query_tokens": 426, "message_tokens": 323, "message_utf8_bytes": 919, "cache_tensor_bytes": null, "serialized_payload_bytes": 919, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 36789.773125899956, "hops": {"1": 6942.490917048417, "2": 3957.965624984354, "3": 4647.70749991294, "4": 3966.145582962781, "5": 4663.1102920509875, "6": 3976.8799169687554, "7": 4664.292625035159, "8": 3971.1806669365615}, "receiver_prefill_ms": 183.257416007109, "ttft_ms": 11.537457932718098, "decode_ms": 11.523457942530513}, "total_wall_ms": 36996.444708900526, "ttft_ms": 11.537457932718098, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "No", "evidence_artifact": {"ledger": {"facts_total": 66, "facts_retained": 8, "retention_rate": 0.12121212121212122, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.17647058823529413, "status": 0.16666666666666666, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 919}, "ledger_by_hop": {"1": {"facts_total": 66, "facts_retained": 6, "retention_rate": 0.09090909090909091, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.17647058823529413, "status": 0.0, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 754}, "2": {"facts_total": 66, "facts_retained": 8, "retention_rate": 0.12121212121212122, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.17647058823529413, "status": 0.16666666666666666, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 919}, "3": {"facts_total": 66, "facts_retained": 8, "retention_rate": 0.12121212121212122, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.17647058823529413, "status": 0.16666666666666666, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 919}, "4": {"facts_total": 66, "facts_retained": 8, "retention_rate": 0.12121212121212122, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.17647058823529413, "status": 0.16666666666666666, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 919}, "5": {"facts_total": 66, "facts_retained": 8, "retention_rate": 0.12121212121212122, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.17647058823529413, "status": 0.16666666666666666, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 919}, "6": {"facts_total": 66, "facts_retained": 8, "retention_rate": 0.12121212121212122, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.17647058823529413, "status": 0.16666666666666666, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 919}, "7": {"facts_total": 66, "facts_retained": 8, "retention_rate": 0.12121212121212122, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.17647058823529413, "status": 0.16666666666666666, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 919}, "8": {"facts_total": 66, "facts_retained": 8, "retention_rate": 0.12121212121212122, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.17647058823529413, "status": 0.16666666666666666, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 919}}}, "reference_key_id": "keys/synthetic/syn_03024_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "ec4605a52403f5abd0fa044d6604914a51692b197ccf5eac20a7175fe08f4ea8", "output_hash": "1ea442a134b2a184bd5d40104401f2a37fbc09ccf3f4bc9da161c6099be3691d"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "dd4f984fd246a4b0a43721c4", "attempt_id": "ece20b68d2c4", "run_group_id": "E4_core:A5|closed|qwen3b-llama3b:h8", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:28:36.213603Z", "task_family": "synthetic", "document_id": "syn_03024_2000", "document_hash": "ea10cd11ad08f01c50b0712bc481668857c9d2ddc8e6423968f2a31eeb97ac55", "question_id": "q07_negation", "split": "test", "cluster_id": "syn_03024_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "tokenizer_hashes": {"receiver": "b9d8872cf3a99da99e1fc4b0f71232a451632dc49ca4c6771549b4efa5c1c763"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 8, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1833, "query_tokens": 427, "message_tokens": 323, "message_utf8_bytes": 919, "cache_tensor_bytes": null, "serialized_payload_bytes": 919, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 36789.773125899956, "hops": {"1": 6942.490917048417, "2": 3957.965624984354, "3": 4647.70749991294, "4": 3966.145582962781, "5": 4663.1102920509875, "6": 3976.8799169687554, "7": 4664.292625035159, "8": 3971.1806669365615}, "receiver_prefill_ms": 183.06658300571144, "ttft_ms": 11.994833941571414, "decode_ms": 11.570833972655237}, "total_wall_ms": 36996.86645891052, "ttft_ms": 11.994833941571414, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "No", "evidence_artifact": {"ledger": {"facts_total": 66, "facts_retained": 8, "retention_rate": 0.12121212121212122, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.17647058823529413, "status": 0.16666666666666666, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 919}, "ledger_by_hop": {"1": {"facts_total": 66, "facts_retained": 6, "retention_rate": 0.09090909090909091, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.17647058823529413, "status": 0.0, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 754}, "2": {"facts_total": 66, "facts_retained": 8, "retention_rate": 0.12121212121212122, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.17647058823529413, "status": 0.16666666666666666, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 919}, "3": {"facts_total": 66, "facts_retained": 8, "retention_rate": 0.12121212121212122, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.17647058823529413, "status": 0.16666666666666666, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 919}, "4": {"facts_total": 66, "facts_retained": 8, "retention_rate": 0.12121212121212122, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.17647058823529413, "status": 0.16666666666666666, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 919}, "5": {"facts_total": 66, "facts_retained": 8, "retention_rate": 0.12121212121212122, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.17647058823529413, "status": 0.16666666666666666, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 919}, "6": {"facts_total": 66, "facts_retained": 8, "retention_rate": 0.12121212121212122, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.17647058823529413, "status": 0.16666666666666666, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 919}, "7": {"facts_total": 66, "facts_retained": 8, "retention_rate": 0.12121212121212122, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.17647058823529413, "status": 0.16666666666666666, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 919}, "8": {"facts_total": 66, "facts_retained": 8, "retention_rate": 0.12121212121212122, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.17647058823529413, "status": 0.16666666666666666, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 919}}}, "reference_key_id": "keys/synthetic/syn_03024_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "ec4605a52403f5abd0fa044d6604914a51692b197ccf5eac20a7175fe08f4ea8", "output_hash": "1ea442a134b2a184bd5d40104401f2a37fbc09ccf3f4bc9da161c6099be3691d"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "820d76ff89aac2a575b498a1", "attempt_id": "2e04488afea4", "run_group_id": "E4_core:A5|closed|qwen3b-llama3b:h8", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:28:36.435516Z", "task_family": "synthetic", "document_id": "syn_03024_2000", "document_hash": "ea10cd11ad08f01c50b0712bc481668857c9d2ddc8e6423968f2a31eeb97ac55", "question_id": "q08_revision", "split": "test", "cluster_id": "syn_03024_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "tokenizer_hashes": {"receiver": "b9d8872cf3a99da99e1fc4b0f71232a451632dc49ca4c6771549b4efa5c1c763"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 8, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1833, "query_tokens": 443, "message_tokens": 323, "message_utf8_bytes": 919, "cache_tensor_bytes": null, "serialized_payload_bytes": 919, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 36789.773125899956, "hops": {"1": 6942.490917048417, "2": 3957.965624984354, "3": 4647.70749991294, "4": 3966.145582962781, "5": 4663.1102920509875, "6": 3976.8799169687554, "7": 4664.292625035159, "8": 3971.1806669365615}, "receiver_prefill_ms": 183.62854199949652, "ttft_ms": 11.598541983403265, "decode_ms": 11.552541982382536}, "total_wall_ms": 36996.93225091323, "ttft_ms": 11.598541983403265, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 66, "facts_retained": 8, "retention_rate": 0.12121212121212122, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.17647058823529413, "status": 0.16666666666666666, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 919}, "ledger_by_hop": {"1": {"facts_total": 66, "facts_retained": 6, "retention_rate": 0.09090909090909091, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.17647058823529413, "status": 0.0, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 754}, "2": {"facts_total": 66, "facts_retained": 8, "retention_rate": 0.12121212121212122, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.17647058823529413, "status": 0.16666666666666666, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 919}, "3": {"facts_total": 66, "facts_retained": 8, "retention_rate": 0.12121212121212122, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.17647058823529413, "status": 0.16666666666666666, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 919}, "4": {"facts_total": 66, "facts_retained": 8, "retention_rate": 0.12121212121212122, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.17647058823529413, "status": 0.16666666666666666, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 919}, "5": {"facts_total": 66, "facts_retained": 8, "retention_rate": 0.12121212121212122, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.17647058823529413, "status": 0.16666666666666666, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 919}, "6": {"facts_total": 66, "facts_retained": 8, "retention_rate": 0.12121212121212122, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.17647058823529413, "status": 0.16666666666666666, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 919}, "7": {"facts_total": 66, "facts_retained": 8, "retention_rate": 0.12121212121212122, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.17647058823529413, "status": 0.16666666666666666, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 919}, "8": {"facts_total": 66, "facts_retained": 8, "retention_rate": 0.12121212121212122, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.17647058823529413, "status": 0.16666666666666666, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 919}}}, "reference_key_id": "keys/synthetic/syn_03024_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "ec4605a52403f5abd0fa044d6604914a51692b197ccf5eac20a7175fe08f4ea8", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "b00e6a1b7f83f6c0def836b4", "attempt_id": "a54bd0e3c2f1", "run_group_id": "E4_core:A5|closed|qwen3b-llama3b:h8", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:28:36.658381Z", "task_family": "synthetic", "document_id": "syn_03024_2000", "document_hash": "ea10cd11ad08f01c50b0712bc481668857c9d2ddc8e6423968f2a31eeb97ac55", "question_id": "q09_revision", "split": "test", "cluster_id": "syn_03024_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "tokenizer_hashes": {"receiver": "b9d8872cf3a99da99e1fc4b0f71232a451632dc49ca4c6771549b4efa5c1c763"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 8, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1833, "query_tokens": 434, "message_tokens": 323, "message_utf8_bytes": 919, "cache_tensor_bytes": null, "serialized_payload_bytes": 919, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 36789.773125899956, "hops": {"1": 6942.490917048417, "2": 3957.965624984354, "3": 4647.70749991294, "4": 3966.145582962781, "5": 4663.1102920509875, "6": 3976.8799169687554, "7": 4664.292625035159, "8": 3971.1806669365615}, "receiver_prefill_ms": 185.11045794002712, "ttft_ms": 11.474458035081625, "decode_ms": 11.570999980904162}, "total_wall_ms": 36998.308541835286, "ttft_ms": 11.474458035081625, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 66, "facts_retained": 8, "retention_rate": 0.12121212121212122, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.17647058823529413, "status": 0.16666666666666666, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 919}, "ledger_by_hop": {"1": {"facts_total": 66, "facts_retained": 6, "retention_rate": 0.09090909090909091, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.17647058823529413, "status": 0.0, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 754}, "2": {"facts_total": 66, "facts_retained": 8, "retention_rate": 0.12121212121212122, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.17647058823529413, "status": 0.16666666666666666, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 919}, "3": {"facts_total": 66, "facts_retained": 8, "retention_rate": 0.12121212121212122, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.17647058823529413, "status": 0.16666666666666666, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 919}, "4": {"facts_total": 66, "facts_retained": 8, "retention_rate": 0.12121212121212122, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.17647058823529413, "status": 0.16666666666666666, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 919}, "5": {"facts_total": 66, "facts_retained": 8, "retention_rate": 0.12121212121212122, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.17647058823529413, "status": 0.16666666666666666, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 919}, "6": {"facts_total": 66, "facts_retained": 8, "retention_rate": 0.12121212121212122, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.17647058823529413, "status": 0.16666666666666666, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 919}, "7": {"facts_total": 66, "facts_retained": 8, "retention_rate": 0.12121212121212122, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.17647058823529413, "status": 0.16666666666666666, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 919}, "8": {"facts_total": 66, "facts_retained": 8, "retention_rate": 0.12121212121212122, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.17647058823529413, "status": 0.16666666666666666, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 919}}}, "reference_key_id": "keys/synthetic/syn_03024_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "ec4605a52403f5abd0fa044d6604914a51692b197ccf5eac20a7175fe08f4ea8", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "dfcd2d188fe0e7d04babc79c", "attempt_id": "0fffbdbf5ba7", "run_group_id": "E4_core:A5|closed|qwen3b-llama3b:h8", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:28:36.882133Z", "task_family": "synthetic", "document_id": "syn_03024_2000", "document_hash": "ea10cd11ad08f01c50b0712bc481668857c9d2ddc8e6423968f2a31eeb97ac55", "question_id": "q10_temporal", "split": "test", "cluster_id": "syn_03024_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "tokenizer_hashes": {"receiver": "b9d8872cf3a99da99e1fc4b0f71232a451632dc49ca4c6771549b4efa5c1c763"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 8, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1833, "query_tokens": 443, "message_tokens": 323, "message_utf8_bytes": 919, "cache_tensor_bytes": null, "serialized_payload_bytes": 919, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 36789.773125899956, "hops": {"1": 6942.490917048417, "2": 3957.965624984354, "3": 4647.70749991294, "4": 3966.145582962781, "5": 4663.1102920509875, "6": 3976.8799169687554, "7": 4664.292625035159, "8": 3971.1806669365615}, "receiver_prefill_ms": 185.32691604923457, "ttft_ms": 11.628582957200706, "decode_ms": 11.593833100050688}, "total_wall_ms": 36998.779916903004, "ttft_ms": 11.628582957200706, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 66, "facts_retained": 8, "retention_rate": 0.12121212121212122, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.17647058823529413, "status": 0.16666666666666666, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 919}, "ledger_by_hop": {"1": {"facts_total": 66, "facts_retained": 6, "retention_rate": 0.09090909090909091, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.17647058823529413, "status": 0.0, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 754}, "2": {"facts_total": 66, "facts_retained": 8, "retention_rate": 0.12121212121212122, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.17647058823529413, "status": 0.16666666666666666, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 919}, "3": {"facts_total": 66, "facts_retained": 8, "retention_rate": 0.12121212121212122, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.17647058823529413, "status": 0.16666666666666666, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 919}, "4": {"facts_total": 66, "facts_retained": 8, "retention_rate": 0.12121212121212122, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.17647058823529413, "status": 0.16666666666666666, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 919}, "5": {"facts_total": 66, "facts_retained": 8, "retention_rate": 0.12121212121212122, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.17647058823529413, "status": 0.16666666666666666, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 919}, "6": {"facts_total": 66, "facts_retained": 8, "retention_rate": 0.12121212121212122, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.17647058823529413, "status": 0.16666666666666666, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 919}, "7": {"facts_total": 66, "facts_retained": 8, "retention_rate": 0.12121212121212122, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.17647058823529413, "status": 0.16666666666666666, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 919}, "8": {"facts_total": 66, "facts_retained": 8, "retention_rate": 0.12121212121212122, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.17647058823529413, "status": 0.16666666666666666, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 919}}}, "reference_key_id": "keys/synthetic/syn_03024_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "ec4605a52403f5abd0fa044d6604914a51692b197ccf5eac20a7175fe08f4ea8", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "7675354915ddcee787617f38", "attempt_id": "addda394f4c3", "run_group_id": "E4_core:A5|closed|qwen3b-llama3b:h8", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:28:37.105901Z", "task_family": "synthetic", "document_id": "syn_03024_2000", "document_hash": "ea10cd11ad08f01c50b0712bc481668857c9d2ddc8e6423968f2a31eeb97ac55", "question_id": "q11_aggregation", "split": "test", "cluster_id": "syn_03024_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "tokenizer_hashes": {"receiver": "b9d8872cf3a99da99e1fc4b0f71232a451632dc49ca4c6771549b4efa5c1c763"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 8, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1833, "query_tokens": 438, "message_tokens": 323, "message_utf8_bytes": 919, "cache_tensor_bytes": null, "serialized_payload_bytes": 919, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 36789.773125899956, "hops": {"1": 6942.490917048417, "2": 3957.965624984354, "3": 4647.70749991294, "4": 3966.145582962781, "5": 4663.1102920509875, "6": 3976.8799169687554, "7": 4664.292625035159, "8": 3971.1806669365615}, "receiver_prefill_ms": 183.45025007147342, "ttft_ms": 11.538333958014846, "decode_ms": 57.53012502100319}, "total_wall_ms": 37042.65787592158, "ttft_ms": 11.538333958014846, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 5}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "8,066,734", "evidence_artifact": {"ledger": {"facts_total": 66, "facts_retained": 8, "retention_rate": 0.12121212121212122, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.17647058823529413, "status": 0.16666666666666666, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 919}, "ledger_by_hop": {"1": {"facts_total": 66, "facts_retained": 6, "retention_rate": 0.09090909090909091, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.17647058823529413, "status": 0.0, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 754}, "2": {"facts_total": 66, "facts_retained": 8, "retention_rate": 0.12121212121212122, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.17647058823529413, "status": 0.16666666666666666, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 919}, "3": {"facts_total": 66, "facts_retained": 8, "retention_rate": 0.12121212121212122, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.17647058823529413, "status": 0.16666666666666666, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 919}, "4": {"facts_total": 66, "facts_retained": 8, "retention_rate": 0.12121212121212122, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.17647058823529413, "status": 0.16666666666666666, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 919}, "5": {"facts_total": 66, "facts_retained": 8, "retention_rate": 0.12121212121212122, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.17647058823529413, "status": 0.16666666666666666, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 919}, "6": {"facts_total": 66, "facts_retained": 8, "retention_rate": 0.12121212121212122, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.17647058823529413, "status": 0.16666666666666666, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 919}, "7": {"facts_total": 66, "facts_retained": 8, "retention_rate": 0.12121212121212122, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.17647058823529413, "status": 0.16666666666666666, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 919}, "8": {"facts_total": 66, "facts_retained": 8, "retention_rate": 0.12121212121212122, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.17647058823529413, "status": 0.16666666666666666, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 919}}}, "reference_key_id": "keys/synthetic/syn_03024_2000.json", "task_scores": {"correct": 0.0, "abstained": 0.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "ec4605a52403f5abd0fa044d6604914a51692b197ccf5eac20a7175fe08f4ea8", "output_hash": "dc9cab51923d7bfcae06d683e0d005a1575296aa981b1594b358ff8e12cd8aa2"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "a4702798a9616e99b3f21f59", "attempt_id": "0a5ee6adbe1f", "run_group_id": "E4_core:A5|closed|qwen3b-llama3b:h8", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:28:37.374216Z", "task_family": "synthetic", "document_id": "syn_03024_2000", "document_hash": "ea10cd11ad08f01c50b0712bc481668857c9d2ddc8e6423968f2a31eeb97ac55", "question_id": "q12_unit_conversion", "split": "test", "cluster_id": "syn_03024_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "tokenizer_hashes": {"receiver": "b9d8872cf3a99da99e1fc4b0f71232a451632dc49ca4c6771549b4efa5c1c763"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 8, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1833, "query_tokens": 434, "message_tokens": 323, "message_utf8_bytes": 919, "cache_tensor_bytes": null, "serialized_payload_bytes": 919, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 36789.773125899956, "hops": {"1": 6942.490917048417, "2": 3957.965624984354, "3": 4647.70749991294, "4": 3966.145582962781, "5": 4663.1102920509875, "6": 3976.8799169687554, "7": 4664.292625035159, "8": 3971.1806669365615}, "receiver_prefill_ms": 182.42191604804248, "ttft_ms": 11.514583951793611, "decode_ms": 11.424958007410169}, "total_wall_ms": 36995.47787592746, "ttft_ms": 11.514583951793611, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 66, "facts_retained": 8, "retention_rate": 0.12121212121212122, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.17647058823529413, "status": 0.16666666666666666, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 919}, "ledger_by_hop": {"1": {"facts_total": 66, "facts_retained": 6, "retention_rate": 0.09090909090909091, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.17647058823529413, "status": 0.0, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 754}, "2": {"facts_total": 66, "facts_retained": 8, "retention_rate": 0.12121212121212122, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.17647058823529413, "status": 0.16666666666666666, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 919}, "3": {"facts_total": 66, "facts_retained": 8, "retention_rate": 0.12121212121212122, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.17647058823529413, "status": 0.16666666666666666, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 919}, "4": {"facts_total": 66, "facts_retained": 8, "retention_rate": 0.12121212121212122, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.17647058823529413, "status": 0.16666666666666666, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 919}, "5": {"facts_total": 66, "facts_retained": 8, "retention_rate": 0.12121212121212122, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.17647058823529413, "status": 0.16666666666666666, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 919}, "6": {"facts_total": 66, "facts_retained": 8, "retention_rate": 0.12121212121212122, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.17647058823529413, "status": 0.16666666666666666, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 919}, "7": {"facts_total": 66, "facts_retained": 8, "retention_rate": 0.12121212121212122, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.17647058823529413, "status": 0.16666666666666666, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 919}, "8": {"facts_total": 66, "facts_retained": 8, "retention_rate": 0.12121212121212122, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.17647058823529413, "status": 0.16666666666666666, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 919}}}, "reference_key_id": "keys/synthetic/syn_03024_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "ec4605a52403f5abd0fa044d6604914a51692b197ccf5eac20a7175fe08f4ea8", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "63bb4d730073f8a8136a1dc0", "attempt_id": "957543e372b8", "run_group_id": "E4_core:A5|closed|qwen3b-llama3b:h8", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:28:37.595047Z", "task_family": "synthetic", "document_id": "syn_03024_2000", "document_hash": "ea10cd11ad08f01c50b0712bc481668857c9d2ddc8e6423968f2a31eeb97ac55", "question_id": "q13_unanswerable", "split": "test", "cluster_id": "syn_03024_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "tokenizer_hashes": {"receiver": "b9d8872cf3a99da99e1fc4b0f71232a451632dc49ca4c6771549b4efa5c1c763"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 8, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1833, "query_tokens": 438, "message_tokens": 323, "message_utf8_bytes": 919, "cache_tensor_bytes": null, "serialized_payload_bytes": 919, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 36789.773125899956, "hops": {"1": 6942.490917048417, "2": 3957.965624984354, "3": 4647.70749991294, "4": 3966.145582962781, "5": 4663.1102920509875, "6": 3976.8799169687554, "7": 4664.292625035159, "8": 3971.1806669365615}, "receiver_prefill_ms": 181.93808302748948, "ttft_ms": 12.016542023047805, "decode_ms": 11.811875039711595}, "total_wall_ms": 36996.139541850425, "ttft_ms": 12.016542023047805, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 66, "facts_retained": 8, "retention_rate": 0.12121212121212122, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.17647058823529413, "status": 0.16666666666666666, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 919}, "ledger_by_hop": {"1": {"facts_total": 66, "facts_retained": 6, "retention_rate": 0.09090909090909091, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.17647058823529413, "status": 0.0, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 754}, "2": {"facts_total": 66, "facts_retained": 8, "retention_rate": 0.12121212121212122, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.17647058823529413, "status": 0.16666666666666666, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 919}, "3": {"facts_total": 66, "facts_retained": 8, "retention_rate": 0.12121212121212122, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.17647058823529413, "status": 0.16666666666666666, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 919}, "4": {"facts_total": 66, "facts_retained": 8, "retention_rate": 0.12121212121212122, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.17647058823529413, "status": 0.16666666666666666, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 919}, "5": {"facts_total": 66, "facts_retained": 8, "retention_rate": 0.12121212121212122, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.17647058823529413, "status": 0.16666666666666666, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 919}, "6": {"facts_total": 66, "facts_retained": 8, "retention_rate": 0.12121212121212122, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.17647058823529413, "status": 0.16666666666666666, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 919}, "7": {"facts_total": 66, "facts_retained": 8, "retention_rate": 0.12121212121212122, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.17647058823529413, "status": 0.16666666666666666, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 919}, "8": {"facts_total": 66, "facts_retained": 8, "retention_rate": 0.12121212121212122, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.17647058823529413, "status": 0.16666666666666666, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 919}}}, "reference_key_id": "keys/synthetic/syn_03024_2000.json", "task_scores": {"correct": 1.0, "abstained": 1.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "ec4605a52403f5abd0fa044d6604914a51692b197ccf5eac20a7175fe08f4ea8", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "225263431e070ab46a7e57fd", "attempt_id": "ce874528b8ff", "run_group_id": "E4_core:A5|closed|qwen3b-llama3b:h8", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:28:37.816687Z", "task_family": "synthetic", "document_id": "syn_03024_2000", "document_hash": "ea10cd11ad08f01c50b0712bc481668857c9d2ddc8e6423968f2a31eeb97ac55", "question_id": "q14_unanswerable", "split": "test", "cluster_id": "syn_03024_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "tokenizer_hashes": {"receiver": "b9d8872cf3a99da99e1fc4b0f71232a451632dc49ca4c6771549b4efa5c1c763"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 8, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1833, "query_tokens": 440, "message_tokens": 323, "message_utf8_bytes": 919, "cache_tensor_bytes": null, "serialized_payload_bytes": 919, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 36789.773125899956, "hops": {"1": 6942.490917048417, "2": 3957.965624984354, "3": 4647.70749991294, "4": 3966.145582962781, "5": 4663.1102920509875, "6": 3976.8799169687554, "7": 4664.292625035159, "8": 3971.1806669365615}, "receiver_prefill_ms": 185.19450002349913, "ttft_ms": 11.919625103473663, "decode_ms": 11.719666887074709}, "total_wall_ms": 36999.187958892435, "ttft_ms": 11.919625103473663, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 66, "facts_retained": 8, "retention_rate": 0.12121212121212122, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.17647058823529413, "status": 0.16666666666666666, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 919}, "ledger_by_hop": {"1": {"facts_total": 66, "facts_retained": 6, "retention_rate": 0.09090909090909091, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.17647058823529413, "status": 0.0, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 754}, "2": {"facts_total": 66, "facts_retained": 8, "retention_rate": 0.12121212121212122, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.17647058823529413, "status": 0.16666666666666666, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 919}, "3": {"facts_total": 66, "facts_retained": 8, "retention_rate": 0.12121212121212122, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.17647058823529413, "status": 0.16666666666666666, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 919}, "4": {"facts_total": 66, "facts_retained": 8, "retention_rate": 0.12121212121212122, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.17647058823529413, "status": 0.16666666666666666, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 919}, "5": {"facts_total": 66, "facts_retained": 8, "retention_rate": 0.12121212121212122, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.17647058823529413, "status": 0.16666666666666666, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 919}, "6": {"facts_total": 66, "facts_retained": 8, "retention_rate": 0.12121212121212122, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.17647058823529413, "status": 0.16666666666666666, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 919}, "7": {"facts_total": 66, "facts_retained": 8, "retention_rate": 0.12121212121212122, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.17647058823529413, "status": 0.16666666666666666, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 919}, "8": {"facts_total": 66, "facts_retained": 8, "retention_rate": 0.12121212121212122, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.0, "amount": 0.17647058823529413, "status": 0.16666666666666666, "place": 0.16666666666666666, "identifier": 0.08333333333333333, "quantity": 0.0}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 919}}}, "reference_key_id": "keys/synthetic/syn_03024_2000.json", "task_scores": {"correct": 1.0, "abstained": 1.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "ec4605a52403f5abd0fa044d6604914a51692b197ccf5eac20a7175fe08f4ea8", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "66572425c7034fc4285d7b51", "attempt_id": "2e86bc484468", "run_group_id": "E4_core:A4|closed|qwen3b:h1", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:28:45.020626Z", "task_family": "synthetic", "document_id": "syn_03027_2000", "document_hash": "10326080979f71b625e4110c7c19767e5a752e267de5034f7e37d98cb6e06249", "question_id": "q00_single_fact", "split": "test", "cluster_id": "syn_03027_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 1, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1955, "query_tokens": 594, "message_tokens": 512, "message_utf8_bytes": 1087, "cache_tensor_bytes": null, "serialized_payload_bytes": 1087, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 6896.277834079228, "hops": {"1": 6896.277834079228}, "receiver_prefill_ms": 227.18162497039884, "ttft_ms": 11.884292005561292, "decode_ms": 11.968541075475514}, "total_wall_ms": 7147.845167084597, "ttft_ms": 11.884292005561292, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 63, "facts_retained": 27, "retention_rate": 0.42857142857142855, "retention_by_kind": {"person": 0.5, "date": 0.125, "amount": 0.6153846153846154, "status": 0.75, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1087}, "ledger_by_hop": {"1": {"facts_total": 63, "facts_retained": 27, "retention_rate": 0.42857142857142855, "retention_by_kind": {"person": 0.5, "date": 0.125, "amount": 0.6153846153846154, "status": 0.75, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1087}}}, "reference_key_id": "keys/synthetic/syn_03027_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "1f88f0e78c63504e418dea843641b99a98555e15d47b02cd5cb2f57da3003dea", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "a505875f1d4297785d2b1c75", "attempt_id": "0feb6491173b", "run_group_id": "E4_core:A4|closed|qwen3b:h1", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:28:45.288112Z", "task_family": "synthetic", "document_id": "syn_03027_2000", "document_hash": "10326080979f71b625e4110c7c19767e5a752e267de5034f7e37d98cb6e06249", "question_id": "q01_single_fact", "split": "test", "cluster_id": "syn_03027_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 1, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1955, "query_tokens": 601, "message_tokens": 512, "message_utf8_bytes": 1087, "cache_tensor_bytes": null, "serialized_payload_bytes": 1087, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 6896.277834079228, "hops": {"1": 6896.277834079228}, "receiver_prefill_ms": 227.2848329739645, "ttft_ms": 11.879250057972968, "decode_ms": 143.36750004440546}, "total_wall_ms": 7279.33433407452, "ttft_ms": 11.879250057972968, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 12}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "Lot 247-S, Baie-Comeau", "evidence_artifact": {"ledger": {"facts_total": 63, "facts_retained": 27, "retention_rate": 0.42857142857142855, "retention_by_kind": {"person": 0.5, "date": 0.125, "amount": 0.6153846153846154, "status": 0.75, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1087}, "ledger_by_hop": {"1": {"facts_total": 63, "facts_retained": 27, "retention_rate": 0.42857142857142855, "retention_by_kind": {"person": 0.5, "date": 0.125, "amount": 0.6153846153846154, "status": 0.75, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1087}}}, "reference_key_id": "keys/synthetic/syn_03027_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 0.8571428571428571}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "1f88f0e78c63504e418dea843641b99a98555e15d47b02cd5cb2f57da3003dea", "output_hash": "77b7da1d43e7d8815e9e9159ebbd735045bc7f668260ba30b93674f70622b8ec"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "2e82ba4be7c3551d233bae4a", "attempt_id": "fd9ec5061d83", "run_group_id": "E4_core:A4|closed|qwen3b:h1", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:28:45.686474Z", "task_family": "synthetic", "document_id": "syn_03027_2000", "document_hash": "10326080979f71b625e4110c7c19767e5a752e267de5034f7e37d98cb6e06249", "question_id": "q02_exact_copy", "split": "test", "cluster_id": "syn_03027_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 1, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1955, "query_tokens": 600, "message_tokens": 512, "message_utf8_bytes": 1087, "cache_tensor_bytes": null, "serialized_payload_bytes": 1087, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 6896.277834079228, "hops": {"1": 6896.277834079228}, "receiver_prefill_ms": 226.71499999705702, "ttft_ms": 12.005083030089736, "decode_ms": 83.47604202572256}, "total_wall_ms": 7219.073334126733, "ttft_ms": 12.005083030089736, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 7}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "15.0054", "evidence_artifact": {"ledger": {"facts_total": 63, "facts_retained": 27, "retention_rate": 0.42857142857142855, "retention_by_kind": {"person": 0.5, "date": 0.125, "amount": 0.6153846153846154, "status": 0.75, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1087}, "ledger_by_hop": {"1": {"facts_total": 63, "facts_retained": 27, "retention_rate": 0.42857142857142855, "retention_by_kind": {"person": 0.5, "date": 0.125, "amount": 0.6153846153846154, "status": 0.75, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1087}}}, "reference_key_id": "keys/synthetic/syn_03027_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "1f88f0e78c63504e418dea843641b99a98555e15d47b02cd5cb2f57da3003dea", "output_hash": "d13adba73068c847c850237059e0202fc5b91ccacf2e792e9ce5e3b90556501a"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "2663feb6b3806e15db462141", "attempt_id": "4a68814f482c", "run_group_id": "E4_core:A4|closed|qwen3b:h1", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:28:46.024752Z", "task_family": "synthetic", "document_id": "syn_03027_2000", "document_hash": "10326080979f71b625e4110c7c19767e5a752e267de5034f7e37d98cb6e06249", "question_id": "q03_exact_copy", "split": "test", "cluster_id": "syn_03027_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 1, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1955, "query_tokens": 595, "message_tokens": 512, "message_utf8_bytes": 1087, "cache_tensor_bytes": null, "serialized_payload_bytes": 1087, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 6896.277834079228, "hops": {"1": 6896.277834079228}, "receiver_prefill_ms": 227.24654199555516, "ttft_ms": 11.966542107984424, "decode_ms": 107.16783395037055}, "total_wall_ms": 7243.166626081802, "ttft_ms": 11.966542107984424, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 9}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "Priyamvada Sørlie", "evidence_artifact": {"ledger": {"facts_total": 63, "facts_retained": 27, "retention_rate": 0.42857142857142855, "retention_by_kind": {"person": 0.5, "date": 0.125, "amount": 0.6153846153846154, "status": 0.75, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1087}, "ledger_by_hop": {"1": {"facts_total": 63, "facts_retained": 27, "retention_rate": 0.42857142857142855, "retention_by_kind": {"person": 0.5, "date": 0.125, "amount": 0.6153846153846154, "status": 0.75, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1087}}}, "reference_key_id": "keys/synthetic/syn_03027_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "1f88f0e78c63504e418dea843641b99a98555e15d47b02cd5cb2f57da3003dea", "output_hash": "c5f554196ed23dd71ad995ec89f7142099d524dbb6534cc10d94a0c89d13e792"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "1958547f71b8ddf37f656406", "attempt_id": "1a46ec016340", "run_group_id": "E4_core:A4|closed|qwen3b:h1", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:28:46.387524Z", "task_family": "synthetic", "document_id": "syn_03027_2000", "document_hash": "10326080979f71b625e4110c7c19767e5a752e267de5034f7e37d98cb6e06249", "question_id": "q04_multi_hop", "split": "test", "cluster_id": "syn_03027_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 1, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1955, "query_tokens": 606, "message_tokens": 512, "message_utf8_bytes": 1087, "cache_tensor_bytes": null, "serialized_payload_bytes": 1087, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 6896.277834079228, "hops": {"1": 6896.277834079228}, "receiver_prefill_ms": 227.36379189882427, "ttft_ms": 12.000042013823986, "decode_ms": 11.930583976209164}, "total_wall_ms": 7148.120459052734, "ttft_ms": 12.000042013823986, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 63, "facts_retained": 27, "retention_rate": 0.42857142857142855, "retention_by_kind": {"person": 0.5, "date": 0.125, "amount": 0.6153846153846154, "status": 0.75, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1087}, "ledger_by_hop": {"1": {"facts_total": 63, "facts_retained": 27, "retention_rate": 0.42857142857142855, "retention_by_kind": {"person": 0.5, "date": 0.125, "amount": 0.6153846153846154, "status": 0.75, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1087}}}, "reference_key_id": "keys/synthetic/syn_03027_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "1f88f0e78c63504e418dea843641b99a98555e15d47b02cd5cb2f57da3003dea", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "902f6c62c96e48069f943197", "attempt_id": "b1ccca241cdd", "run_group_id": "E4_core:A4|closed|qwen3b:h1", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:28:46.654816Z", "task_family": "synthetic", "document_id": "syn_03027_2000", "document_hash": "10326080979f71b625e4110c7c19767e5a752e267de5034f7e37d98cb6e06249", "question_id": "q05_multi_hop", "split": "test", "cluster_id": "syn_03027_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 1, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1955, "query_tokens": 603, "message_tokens": 512, "message_utf8_bytes": 1087, "cache_tensor_bytes": null, "serialized_payload_bytes": 1087, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 6896.277834079228, "hops": {"1": 6896.277834079228}, "receiver_prefill_ms": 227.08666697144508, "ttft_ms": 11.93825004156679, "decode_ms": 11.94079197011888}, "total_wall_ms": 7147.7413760731, "ttft_ms": 11.93825004156679, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 63, "facts_retained": 27, "retention_rate": 0.42857142857142855, "retention_by_kind": {"person": 0.5, "date": 0.125, "amount": 0.6153846153846154, "status": 0.75, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1087}, "ledger_by_hop": {"1": {"facts_total": 63, "facts_retained": 27, "retention_rate": 0.42857142857142855, "retention_by_kind": {"person": 0.5, "date": 0.125, "amount": 0.6153846153846154, "status": 0.75, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1087}}}, "reference_key_id": "keys/synthetic/syn_03027_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "1f88f0e78c63504e418dea843641b99a98555e15d47b02cd5cb2f57da3003dea", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "576887d7bdeaf2bc4e7b4398", "attempt_id": "8dca5f79eb36", "run_group_id": "E4_core:A4|closed|qwen3b:h1", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:28:46.921605Z", "task_family": "synthetic", "document_id": "syn_03027_2000", "document_hash": "10326080979f71b625e4110c7c19767e5a752e267de5034f7e37d98cb6e06249", "question_id": "q06_negation", "split": "test", "cluster_id": "syn_03027_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 1, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1955, "query_tokens": 595, "message_tokens": 512, "message_utf8_bytes": 1087, "cache_tensor_bytes": null, "serialized_payload_bytes": 1087, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 6896.277834079228, "hops": {"1": 6896.277834079228}, "receiver_prefill_ms": 226.9028329756111, "ttft_ms": 11.929624946787953, "decode_ms": 11.902374913915992}, "total_wall_ms": 7147.443876019679, "ttft_ms": 11.929624946787953, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 63, "facts_retained": 27, "retention_rate": 0.42857142857142855, "retention_by_kind": {"person": 0.5, "date": 0.125, "amount": 0.6153846153846154, "status": 0.75, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1087}, "ledger_by_hop": {"1": {"facts_total": 63, "facts_retained": 27, "retention_rate": 0.42857142857142855, "retention_by_kind": {"person": 0.5, "date": 0.125, "amount": 0.6153846153846154, "status": 0.75, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1087}}}, "reference_key_id": "keys/synthetic/syn_03027_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "1f88f0e78c63504e418dea843641b99a98555e15d47b02cd5cb2f57da3003dea", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "de4b312c2dede396a1a34597", "attempt_id": "1f8bac3b7bd1", "run_group_id": "E4_core:A4|closed|qwen3b:h1", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:28:47.187621Z", "task_family": "synthetic", "document_id": "syn_03027_2000", "document_hash": "10326080979f71b625e4110c7c19767e5a752e267de5034f7e37d98cb6e06249", "question_id": "q07_negation", "split": "test", "cluster_id": "syn_03027_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 1, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1955, "query_tokens": 596, "message_tokens": 512, "message_utf8_bytes": 1087, "cache_tensor_bytes": null, "serialized_payload_bytes": 1087, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 6896.277834079228, "hops": {"1": 6896.277834079228}, "receiver_prefill_ms": 227.014834061265, "ttft_ms": 12.189500033855438, "decode_ms": 12.008458958007395}, "total_wall_ms": 7148.10291712638, "ttft_ms": 12.189500033855438, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "no", "evidence_artifact": {"ledger": {"facts_total": 63, "facts_retained": 27, "retention_rate": 0.42857142857142855, "retention_by_kind": {"person": 0.5, "date": 0.125, "amount": 0.6153846153846154, "status": 0.75, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1087}, "ledger_by_hop": {"1": {"facts_total": 63, "facts_retained": 27, "retention_rate": 0.42857142857142855, "retention_by_kind": {"person": 0.5, "date": 0.125, "amount": 0.6153846153846154, "status": 0.75, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1087}}}, "reference_key_id": "keys/synthetic/syn_03027_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "1f88f0e78c63504e418dea843641b99a98555e15d47b02cd5cb2f57da3003dea", "output_hash": "9390298f3fb0c5b160498935d79cb139aef28e1c47358b4bbba61862b9c26e59"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "af04aeaebaa092508d102f44", "attempt_id": "36e81c844352", "run_group_id": "E4_core:A4|closed|qwen3b:h1", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:28:47.454980Z", "task_family": "synthetic", "document_id": "syn_03027_2000", "document_hash": "10326080979f71b625e4110c7c19767e5a752e267de5034f7e37d98cb6e06249", "question_id": "q08_revision", "split": "test", "cluster_id": "syn_03027_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 1, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1955, "query_tokens": 610, "message_tokens": 512, "message_utf8_bytes": 1087, "cache_tensor_bytes": null, "serialized_payload_bytes": 1087, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 6896.277834079228, "hops": {"1": 6896.277834079228}, "receiver_prefill_ms": 229.73904199898243, "ttft_ms": 11.996667017228901, "decode_ms": 83.51000002585351}, "total_wall_ms": 7222.080167150125, "ttft_ms": 11.996667017228901, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 7}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "8010205", "evidence_artifact": {"ledger": {"facts_total": 63, "facts_retained": 27, "retention_rate": 0.42857142857142855, "retention_by_kind": {"person": 0.5, "date": 0.125, "amount": 0.6153846153846154, "status": 0.75, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1087}, "ledger_by_hop": {"1": {"facts_total": 63, "facts_retained": 27, "retention_rate": 0.42857142857142855, "retention_by_kind": {"person": 0.5, "date": 0.125, "amount": 0.6153846153846154, "status": 0.75, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1087}}}, "reference_key_id": "keys/synthetic/syn_03027_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "1f88f0e78c63504e418dea843641b99a98555e15d47b02cd5cb2f57da3003dea", "output_hash": "6d1ceb36132f38bd77e8f46c23ea21dd9bcff78bc80bd8a9eaf2d7f2f3bacf46"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "c0ee52d6b9cc2e3be3a2d06d", "attempt_id": "d2437f21e50a", "run_group_id": "E4_core:A4|closed|qwen3b:h1", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:28:47.796422Z", "task_family": "synthetic", "document_id": "syn_03027_2000", "document_hash": "10326080979f71b625e4110c7c19767e5a752e267de5034f7e37d98cb6e06249", "question_id": "q09_revision", "split": "test", "cluster_id": "syn_03027_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 1, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1955, "query_tokens": 603, "message_tokens": 512, "message_utf8_bytes": 1087, "cache_tensor_bytes": null, "serialized_payload_bytes": 1087, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 6896.277834079228, "hops": {"1": 6896.277834079228}, "receiver_prefill_ms": 226.97950003203005, "ttft_ms": 11.93766703363508, "decode_ms": 107.30429191607982}, "total_wall_ms": 7242.995709064417, "ttft_ms": 11.93766703363508, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 9}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "6,056,150", "evidence_artifact": {"ledger": {"facts_total": 63, "facts_retained": 27, "retention_rate": 0.42857142857142855, "retention_by_kind": {"person": 0.5, "date": 0.125, "amount": 0.6153846153846154, "status": 0.75, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1087}, "ledger_by_hop": {"1": {"facts_total": 63, "facts_retained": 27, "retention_rate": 0.42857142857142855, "retention_by_kind": {"person": 0.5, "date": 0.125, "amount": 0.6153846153846154, "status": 0.75, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1087}}}, "reference_key_id": "keys/synthetic/syn_03027_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "1f88f0e78c63504e418dea843641b99a98555e15d47b02cd5cb2f57da3003dea", "output_hash": "8262889a8478c7e0ee965f159d9849a3849b412e276f5b979476e2da5b320249"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "59b9ff8b13d556d357c09d32", "attempt_id": "4dd5657b5038", "run_group_id": "E4_core:A4|closed|qwen3b:h1", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:28:48.158347Z", "task_family": "synthetic", "document_id": "syn_03027_2000", "document_hash": "10326080979f71b625e4110c7c19767e5a752e267de5034f7e37d98cb6e06249", "question_id": "q10_temporal", "split": "test", "cluster_id": "syn_03027_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 1, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1955, "query_tokens": 613, "message_tokens": 512, "message_utf8_bytes": 1087, "cache_tensor_bytes": null, "serialized_payload_bytes": 1087, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 6896.277834079228, "hops": {"1": 6896.277834079228}, "receiver_prefill_ms": 229.18437502812594, "ttft_ms": 11.881874990649521, "decode_ms": 83.33941700402647}, "total_wall_ms": 7221.206209156662, "ttft_ms": 11.881874990649521, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 7}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "Lunlith-13", "evidence_artifact": {"ledger": {"facts_total": 63, "facts_retained": 27, "retention_rate": 0.42857142857142855, "retention_by_kind": {"person": 0.5, "date": 0.125, "amount": 0.6153846153846154, "status": 0.75, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1087}, "ledger_by_hop": {"1": {"facts_total": 63, "facts_retained": 27, "retention_rate": 0.42857142857142855, "retention_by_kind": {"person": 0.5, "date": 0.125, "amount": 0.6153846153846154, "status": 0.75, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1087}}}, "reference_key_id": "keys/synthetic/syn_03027_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "1f88f0e78c63504e418dea843641b99a98555e15d47b02cd5cb2f57da3003dea", "output_hash": "c45f000afe6d865c8167954eebdff761141225817d4f5b26c962cd50cd75d391"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "e44219fd3fa6f41965c4458a", "attempt_id": "ad4dbca8b42d", "run_group_id": "E4_core:A4|closed|qwen3b:h1", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:28:48.498213Z", "task_family": "synthetic", "document_id": "syn_03027_2000", "document_hash": "10326080979f71b625e4110c7c19767e5a752e267de5034f7e37d98cb6e06249", "question_id": "q11_aggregation", "split": "test", "cluster_id": "syn_03027_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 1, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1955, "query_tokens": 606, "message_tokens": 512, "message_utf8_bytes": 1087, "cache_tensor_bytes": null, "serialized_payload_bytes": 1087, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 6896.277834079228, "hops": {"1": 6896.277834079228}, "receiver_prefill_ms": 227.49158297665417, "ttft_ms": 11.954542016610503, "decode_ms": 119.40641701221466}, "total_wall_ms": 7255.645459052175, "ttft_ms": 11.954542016610503, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 10}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "23,450,077", "evidence_artifact": {"ledger": {"facts_total": 63, "facts_retained": 27, "retention_rate": 0.42857142857142855, "retention_by_kind": {"person": 0.5, "date": 0.125, "amount": 0.6153846153846154, "status": 0.75, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1087}, "ledger_by_hop": {"1": {"facts_total": 63, "facts_retained": 27, "retention_rate": 0.42857142857142855, "retention_by_kind": {"person": 0.5, "date": 0.125, "amount": 0.6153846153846154, "status": 0.75, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1087}}}, "reference_key_id": "keys/synthetic/syn_03027_2000.json", "task_scores": {"correct": 0.0, "abstained": 0.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "1f88f0e78c63504e418dea843641b99a98555e15d47b02cd5cb2f57da3003dea", "output_hash": "2044d7c4afd32892ac22705680e35174db83308c139b02a1494275cc5e24aff3"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "94e171be9d7fba83905b3a10", "attempt_id": "406fb3f0385a", "run_group_id": "E4_core:A4|closed|qwen3b:h1", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:28:48.873002Z", "task_family": "synthetic", "document_id": "syn_03027_2000", "document_hash": "10326080979f71b625e4110c7c19767e5a752e267de5034f7e37d98cb6e06249", "question_id": "q12_unit_conversion", "split": "test", "cluster_id": "syn_03027_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 1, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1955, "query_tokens": 603, "message_tokens": 512, "message_utf8_bytes": 1087, "cache_tensor_bytes": null, "serialized_payload_bytes": 1087, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 6896.277834079228, "hops": {"1": 6896.277834079228}, "receiver_prefill_ms": 226.21345799416304, "ttft_ms": 12.138666934333742, "decode_ms": 11.895625037141144}, "total_wall_ms": 7147.149959113449, "ttft_ms": 12.138666934333742, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 63, "facts_retained": 27, "retention_rate": 0.42857142857142855, "retention_by_kind": {"person": 0.5, "date": 0.125, "amount": 0.6153846153846154, "status": 0.75, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1087}, "ledger_by_hop": {"1": {"facts_total": 63, "facts_retained": 27, "retention_rate": 0.42857142857142855, "retention_by_kind": {"person": 0.5, "date": 0.125, "amount": 0.6153846153846154, "status": 0.75, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1087}}}, "reference_key_id": "keys/synthetic/syn_03027_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "1f88f0e78c63504e418dea843641b99a98555e15d47b02cd5cb2f57da3003dea", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "1bf57ae1ad924ff10170ec6c", "attempt_id": "a03914bdc8b1", "run_group_id": "E4_core:A4|closed|qwen3b:h1", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:28:49.139135Z", "task_family": "synthetic", "document_id": "syn_03027_2000", "document_hash": "10326080979f71b625e4110c7c19767e5a752e267de5034f7e37d98cb6e06249", "question_id": "q13_unanswerable", "split": "test", "cluster_id": "syn_03027_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 1, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1955, "query_tokens": 606, "message_tokens": 512, "message_utf8_bytes": 1087, "cache_tensor_bytes": null, "serialized_payload_bytes": 1087, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 6896.277834079228, "hops": {"1": 6896.277834079228}, "receiver_prefill_ms": 226.58466605935246, "ttft_ms": 12.026333017274737, "decode_ms": 11.8740409379825}, "total_wall_ms": 7147.256001015194, "ttft_ms": 12.026333017274737, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 63, "facts_retained": 27, "retention_rate": 0.42857142857142855, "retention_by_kind": {"person": 0.5, "date": 0.125, "amount": 0.6153846153846154, "status": 0.75, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1087}, "ledger_by_hop": {"1": {"facts_total": 63, "facts_retained": 27, "retention_rate": 0.42857142857142855, "retention_by_kind": {"person": 0.5, "date": 0.125, "amount": 0.6153846153846154, "status": 0.75, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1087}}}, "reference_key_id": "keys/synthetic/syn_03027_2000.json", "task_scores": {"correct": 1.0, "abstained": 1.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "1f88f0e78c63504e418dea843641b99a98555e15d47b02cd5cb2f57da3003dea", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "8903790c81721835f22e455b", "attempt_id": "753fac8761e6", "run_group_id": "E4_core:A4|closed|qwen3b:h1", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:28:49.405855Z", "task_family": "synthetic", "document_id": "syn_03027_2000", "document_hash": "10326080979f71b625e4110c7c19767e5a752e267de5034f7e37d98cb6e06249", "question_id": "q14_unanswerable", "split": "test", "cluster_id": "syn_03027_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 1, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1955, "query_tokens": 608, "message_tokens": 512, "message_utf8_bytes": 1087, "cache_tensor_bytes": null, "serialized_payload_bytes": 1087, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 6896.277834079228, "hops": {"1": 6896.277834079228}, "receiver_prefill_ms": 226.44899995066226, "ttft_ms": 12.065957998856902, "decode_ms": 12.056792038492858}, "total_wall_ms": 7147.501792060211, "ttft_ms": 12.065957998856902, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 63, "facts_retained": 27, "retention_rate": 0.42857142857142855, "retention_by_kind": {"person": 0.5, "date": 0.125, "amount": 0.6153846153846154, "status": 0.75, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1087}, "ledger_by_hop": {"1": {"facts_total": 63, "facts_retained": 27, "retention_rate": 0.42857142857142855, "retention_by_kind": {"person": 0.5, "date": 0.125, "amount": 0.6153846153846154, "status": 0.75, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1087}}}, "reference_key_id": "keys/synthetic/syn_03027_2000.json", "task_scores": {"correct": 1.0, "abstained": 1.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "1f88f0e78c63504e418dea843641b99a98555e15d47b02cd5cb2f57da3003dea", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "380461c99b1dd78a93896e5a", "attempt_id": "183d164b05e1", "run_group_id": "E4_core:A4|closed|qwen3b:h2", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:28:56.046438Z", "task_family": "synthetic", "document_id": "syn_03027_2000", "document_hash": "10326080979f71b625e4110c7c19767e5a752e267de5034f7e37d98cb6e06249", "question_id": "q00_single_fact", "split": "test", "cluster_id": "syn_03027_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 2, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1955, "query_tokens": 594, "message_tokens": 512, "message_utf8_bytes": 1087, "cache_tensor_bytes": null, "serialized_payload_bytes": 1087, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 13268.967709154822, "hops": {"1": 6896.277834079228, "2": 6372.689875075594}, "receiver_prefill_ms": 226.353875012137, "ttft_ms": 11.964041041210294, "decode_ms": 11.987666017375886}, "total_wall_ms": 13519.761501112953, "ttft_ms": 11.964041041210294, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 63, "facts_retained": 27, "retention_rate": 0.42857142857142855, "retention_by_kind": {"person": 0.5, "date": 0.125, "amount": 0.6153846153846154, "status": 0.75, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1087}, "ledger_by_hop": {"1": {"facts_total": 63, "facts_retained": 27, "retention_rate": 0.42857142857142855, "retention_by_kind": {"person": 0.5, "date": 0.125, "amount": 0.6153846153846154, "status": 0.75, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1087}, "2": {"facts_total": 63, "facts_retained": 27, "retention_rate": 0.42857142857142855, "retention_by_kind": {"person": 0.5, "date": 0.125, "amount": 0.6153846153846154, "status": 0.75, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1087}}}, "reference_key_id": "keys/synthetic/syn_03027_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "1f88f0e78c63504e418dea843641b99a98555e15d47b02cd5cb2f57da3003dea", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "bb672f0d17e1dce0f73b2bfe", "attempt_id": "bc487cb70976", "run_group_id": "E4_core:A4|closed|qwen3b:h2", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:28:56.313252Z", "task_family": "synthetic", "document_id": "syn_03027_2000", "document_hash": "10326080979f71b625e4110c7c19767e5a752e267de5034f7e37d98cb6e06249", "question_id": "q01_single_fact", "split": "test", "cluster_id": "syn_03027_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 2, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1955, "query_tokens": 601, "message_tokens": 512, "message_utf8_bytes": 1087, "cache_tensor_bytes": null, "serialized_payload_bytes": 1087, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 13268.967709154822, "hops": {"1": 6896.277834079228, "2": 6372.689875075594}, "receiver_prefill_ms": 226.6314160078764, "ttft_ms": 11.946083046495914, "decode_ms": 142.85629207734019}, "total_wall_ms": 13650.942834210582, "ttft_ms": 11.946083046495914, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 12}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "Lot 247-S, Baie-Comeau", "evidence_artifact": {"ledger": {"facts_total": 63, "facts_retained": 27, "retention_rate": 0.42857142857142855, "retention_by_kind": {"person": 0.5, "date": 0.125, "amount": 0.6153846153846154, "status": 0.75, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1087}, "ledger_by_hop": {"1": {"facts_total": 63, "facts_retained": 27, "retention_rate": 0.42857142857142855, "retention_by_kind": {"person": 0.5, "date": 0.125, "amount": 0.6153846153846154, "status": 0.75, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1087}, "2": {"facts_total": 63, "facts_retained": 27, "retention_rate": 0.42857142857142855, "retention_by_kind": {"person": 0.5, "date": 0.125, "amount": 0.6153846153846154, "status": 0.75, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1087}}}, "reference_key_id": "keys/synthetic/syn_03027_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 0.8571428571428571}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "1f88f0e78c63504e418dea843641b99a98555e15d47b02cd5cb2f57da3003dea", "output_hash": "77b7da1d43e7d8815e9e9159ebbd735045bc7f668260ba30b93674f70622b8ec"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "2f42ac5d2c84d2ded643b7b8", "attempt_id": "1ff8956d4f1e", "run_group_id": "E4_core:A4|closed|qwen3b:h2", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:28:56.710422Z", "task_family": "synthetic", "document_id": "syn_03027_2000", "document_hash": "10326080979f71b625e4110c7c19767e5a752e267de5034f7e37d98cb6e06249", "question_id": "q02_exact_copy", "split": "test", "cluster_id": "syn_03027_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 2, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1955, "query_tokens": 600, "message_tokens": 512, "message_utf8_bytes": 1087, "cache_tensor_bytes": null, "serialized_payload_bytes": 1087, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 13268.967709154822, "hops": {"1": 6896.277834079228, "2": 6372.689875075594}, "receiver_prefill_ms": 226.8895839806646, "ttft_ms": 11.947875027544796, "decode_ms": 83.42229202389717}, "total_wall_ms": 13591.716709081084, "ttft_ms": 11.947875027544796, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 7}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "15.0054", "evidence_artifact": {"ledger": {"facts_total": 63, "facts_retained": 27, "retention_rate": 0.42857142857142855, "retention_by_kind": {"person": 0.5, "date": 0.125, "amount": 0.6153846153846154, "status": 0.75, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1087}, "ledger_by_hop": {"1": {"facts_total": 63, "facts_retained": 27, "retention_rate": 0.42857142857142855, "retention_by_kind": {"person": 0.5, "date": 0.125, "amount": 0.6153846153846154, "status": 0.75, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1087}, "2": {"facts_total": 63, "facts_retained": 27, "retention_rate": 0.42857142857142855, "retention_by_kind": {"person": 0.5, "date": 0.125, "amount": 0.6153846153846154, "status": 0.75, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1087}}}, "reference_key_id": "keys/synthetic/syn_03027_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "1f88f0e78c63504e418dea843641b99a98555e15d47b02cd5cb2f57da3003dea", "output_hash": "d13adba73068c847c850237059e0202fc5b91ccacf2e792e9ce5e3b90556501a"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "96f79ccf45bf13aa8e20f958", "attempt_id": "207e91b6d37a", "run_group_id": "E4_core:A4|closed|qwen3b:h2", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:28:57.049391Z", "task_family": "synthetic", "document_id": "syn_03027_2000", "document_hash": "10326080979f71b625e4110c7c19767e5a752e267de5034f7e37d98cb6e06249", "question_id": "q03_exact_copy", "split": "test", "cluster_id": "syn_03027_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 2, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1955, "query_tokens": 595, "message_tokens": 512, "message_utf8_bytes": 1087, "cache_tensor_bytes": null, "serialized_payload_bytes": 1087, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 13268.967709154822, "hops": {"1": 6896.277834079228, "2": 6372.689875075594}, "receiver_prefill_ms": 226.73004202079028, "ttft_ms": 11.906833969987929, "decode_ms": 106.80929094087332}, "total_wall_ms": 13614.928168128245, "ttft_ms": 11.906833969987929, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 9}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "Priyamvada Sørlie", "evidence_artifact": {"ledger": {"facts_total": 63, "facts_retained": 27, "retention_rate": 0.42857142857142855, "retention_by_kind": {"person": 0.5, "date": 0.125, "amount": 0.6153846153846154, "status": 0.75, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1087}, "ledger_by_hop": {"1": {"facts_total": 63, "facts_retained": 27, "retention_rate": 0.42857142857142855, "retention_by_kind": {"person": 0.5, "date": 0.125, "amount": 0.6153846153846154, "status": 0.75, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1087}, "2": {"facts_total": 63, "facts_retained": 27, "retention_rate": 0.42857142857142855, "retention_by_kind": {"person": 0.5, "date": 0.125, "amount": 0.6153846153846154, "status": 0.75, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1087}}}, "reference_key_id": "keys/synthetic/syn_03027_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "1f88f0e78c63504e418dea843641b99a98555e15d47b02cd5cb2f57da3003dea", "output_hash": "c5f554196ed23dd71ad995ec89f7142099d524dbb6534cc10d94a0c89d13e792"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "02be1f189f0e83e046ada8a0", "attempt_id": "67e0ed26a09d", "run_group_id": "E4_core:A4|closed|qwen3b:h2", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:28:57.410889Z", "task_family": "synthetic", "document_id": "syn_03027_2000", "document_hash": "10326080979f71b625e4110c7c19767e5a752e267de5034f7e37d98cb6e06249", "question_id": "q04_multi_hop", "split": "test", "cluster_id": "syn_03027_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 2, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1955, "query_tokens": 606, "message_tokens": 512, "message_utf8_bytes": 1087, "cache_tensor_bytes": null, "serialized_payload_bytes": 1087, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 13268.967709154822, "hops": {"1": 6896.277834079228, "2": 6372.689875075594}, "receiver_prefill_ms": 226.68625006917864, "ttft_ms": 12.129915994592011, "decode_ms": 11.946541024371982}, "total_wall_ms": 13520.36462619435, "ttft_ms": 12.129915994592011, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 63, "facts_retained": 27, "retention_rate": 0.42857142857142855, "retention_by_kind": {"person": 0.5, "date": 0.125, "amount": 0.6153846153846154, "status": 0.75, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1087}, "ledger_by_hop": {"1": {"facts_total": 63, "facts_retained": 27, "retention_rate": 0.42857142857142855, "retention_by_kind": {"person": 0.5, "date": 0.125, "amount": 0.6153846153846154, "status": 0.75, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1087}, "2": {"facts_total": 63, "facts_retained": 27, "retention_rate": 0.42857142857142855, "retention_by_kind": {"person": 0.5, "date": 0.125, "amount": 0.6153846153846154, "status": 0.75, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1087}}}, "reference_key_id": "keys/synthetic/syn_03027_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "1f88f0e78c63504e418dea843641b99a98555e15d47b02cd5cb2f57da3003dea", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "4f9b895e9fe6ba03ced05886", "attempt_id": "821443996dd8", "run_group_id": "E4_core:A4|closed|qwen3b:h2", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:28:57.678651Z", "task_family": "synthetic", "document_id": "syn_03027_2000", "document_hash": "10326080979f71b625e4110c7c19767e5a752e267de5034f7e37d98cb6e06249", "question_id": "q05_multi_hop", "split": "test", "cluster_id": "syn_03027_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 2, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1955, "query_tokens": 603, "message_tokens": 512, "message_utf8_bytes": 1087, "cache_tensor_bytes": null, "serialized_payload_bytes": 1087, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 13268.967709154822, "hops": {"1": 6896.277834079228, "2": 6372.689875075594}, "receiver_prefill_ms": 226.64520796388388, "ttft_ms": 12.175749987363815, "decode_ms": 12.064957991242409}, "total_wall_ms": 13520.53808409255, "ttft_ms": 12.175749987363815, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 63, "facts_retained": 27, "retention_rate": 0.42857142857142855, "retention_by_kind": {"person": 0.5, "date": 0.125, "amount": 0.6153846153846154, "status": 0.75, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1087}, "ledger_by_hop": {"1": {"facts_total": 63, "facts_retained": 27, "retention_rate": 0.42857142857142855, "retention_by_kind": {"person": 0.5, "date": 0.125, "amount": 0.6153846153846154, "status": 0.75, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1087}, "2": {"facts_total": 63, "facts_retained": 27, "retention_rate": 0.42857142857142855, "retention_by_kind": {"person": 0.5, "date": 0.125, "amount": 0.6153846153846154, "status": 0.75, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1087}}}, "reference_key_id": "keys/synthetic/syn_03027_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "1f88f0e78c63504e418dea843641b99a98555e15d47b02cd5cb2f57da3003dea", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "b6b3ac448590ef7597cbad0e", "attempt_id": "98cb0c7a8ad5", "run_group_id": "E4_core:A4|closed|qwen3b:h2", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:28:57.946886Z", "task_family": "synthetic", "document_id": "syn_03027_2000", "document_hash": "10326080979f71b625e4110c7c19767e5a752e267de5034f7e37d98cb6e06249", "question_id": "q06_negation", "split": "test", "cluster_id": "syn_03027_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 2, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1955, "query_tokens": 595, "message_tokens": 512, "message_utf8_bytes": 1087, "cache_tensor_bytes": null, "serialized_payload_bytes": 1087, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 13268.967709154822, "hops": {"1": 6896.277834079228, "2": 6372.689875075594}, "receiver_prefill_ms": 226.30995803046972, "ttft_ms": 12.314041960053146, "decode_ms": 12.070958036929369}, "total_wall_ms": 13520.389126148075, "ttft_ms": 12.314041960053146, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 63, "facts_retained": 27, "retention_rate": 0.42857142857142855, "retention_by_kind": {"person": 0.5, "date": 0.125, "amount": 0.6153846153846154, "status": 0.75, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1087}, "ledger_by_hop": {"1": {"facts_total": 63, "facts_retained": 27, "retention_rate": 0.42857142857142855, "retention_by_kind": {"person": 0.5, "date": 0.125, "amount": 0.6153846153846154, "status": 0.75, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1087}, "2": {"facts_total": 63, "facts_retained": 27, "retention_rate": 0.42857142857142855, "retention_by_kind": {"person": 0.5, "date": 0.125, "amount": 0.6153846153846154, "status": 0.75, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1087}}}, "reference_key_id": "keys/synthetic/syn_03027_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "1f88f0e78c63504e418dea843641b99a98555e15d47b02cd5cb2f57da3003dea", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "0d345796a81846aaa98c201c", "attempt_id": "46761864a207", "run_group_id": "E4_core:A4|closed|qwen3b:h2", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:28:58.215165Z", "task_family": "synthetic", "document_id": "syn_03027_2000", "document_hash": "10326080979f71b625e4110c7c19767e5a752e267de5034f7e37d98cb6e06249", "question_id": "q07_negation", "split": "test", "cluster_id": "syn_03027_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 2, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1955, "query_tokens": 596, "message_tokens": 512, "message_utf8_bytes": 1087, "cache_tensor_bytes": null, "serialized_payload_bytes": 1087, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 13268.967709154822, "hops": {"1": 6896.277834079228, "2": 6372.689875075594}, "receiver_prefill_ms": 226.8997090868652, "ttft_ms": 12.095415964722633, "decode_ms": 11.946999933570623}, "total_wall_ms": 13520.565334125422, "ttft_ms": 12.095415964722633, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "no", "evidence_artifact": {"ledger": {"facts_total": 63, "facts_retained": 27, "retention_rate": 0.42857142857142855, "retention_by_kind": {"person": 0.5, "date": 0.125, "amount": 0.6153846153846154, "status": 0.75, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1087}, "ledger_by_hop": {"1": {"facts_total": 63, "facts_retained": 27, "retention_rate": 0.42857142857142855, "retention_by_kind": {"person": 0.5, "date": 0.125, "amount": 0.6153846153846154, "status": 0.75, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1087}, "2": {"facts_total": 63, "facts_retained": 27, "retention_rate": 0.42857142857142855, "retention_by_kind": {"person": 0.5, "date": 0.125, "amount": 0.6153846153846154, "status": 0.75, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1087}}}, "reference_key_id": "keys/synthetic/syn_03027_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "1f88f0e78c63504e418dea843641b99a98555e15d47b02cd5cb2f57da3003dea", "output_hash": "9390298f3fb0c5b160498935d79cb139aef28e1c47358b4bbba61862b9c26e59"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "4bdf6b31732546e18ddc60f3", "attempt_id": "cd86d3586881", "run_group_id": "E4_core:A4|closed|qwen3b:h2", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:28:58.482109Z", "task_family": "synthetic", "document_id": "syn_03027_2000", "document_hash": "10326080979f71b625e4110c7c19767e5a752e267de5034f7e37d98cb6e06249", "question_id": "q08_revision", "split": "test", "cluster_id": "syn_03027_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 2, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1955, "query_tokens": 610, "message_tokens": 512, "message_utf8_bytes": 1087, "cache_tensor_bytes": null, "serialized_payload_bytes": 1087, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 13268.967709154822, "hops": {"1": 6896.277834079228, "2": 6372.689875075594}, "receiver_prefill_ms": 229.23491708934307, "ttft_ms": 12.094041099771857, "decode_ms": 83.74658308457583}, "total_wall_ms": 13594.683792092837, "ttft_ms": 12.094041099771857, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 7}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "8010205", "evidence_artifact": {"ledger": {"facts_total": 63, "facts_retained": 27, "retention_rate": 0.42857142857142855, "retention_by_kind": {"person": 0.5, "date": 0.125, "amount": 0.6153846153846154, "status": 0.75, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1087}, "ledger_by_hop": {"1": {"facts_total": 63, "facts_retained": 27, "retention_rate": 0.42857142857142855, "retention_by_kind": {"person": 0.5, "date": 0.125, "amount": 0.6153846153846154, "status": 0.75, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1087}, "2": {"facts_total": 63, "facts_retained": 27, "retention_rate": 0.42857142857142855, "retention_by_kind": {"person": 0.5, "date": 0.125, "amount": 0.6153846153846154, "status": 0.75, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1087}}}, "reference_key_id": "keys/synthetic/syn_03027_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "1f88f0e78c63504e418dea843641b99a98555e15d47b02cd5cb2f57da3003dea", "output_hash": "6d1ceb36132f38bd77e8f46c23ea21dd9bcff78bc80bd8a9eaf2d7f2f3bacf46"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "2856af02e43ff323d8c69a7b", "attempt_id": "74113a5e69fa", "run_group_id": "E4_core:A4|closed|qwen3b:h2", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:28:58.824450Z", "task_family": "synthetic", "document_id": "syn_03027_2000", "document_hash": "10326080979f71b625e4110c7c19767e5a752e267de5034f7e37d98cb6e06249", "question_id": "q09_revision", "split": "test", "cluster_id": "syn_03027_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 2, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1955, "query_tokens": 603, "message_tokens": 512, "message_utf8_bytes": 1087, "cache_tensor_bytes": null, "serialized_payload_bytes": 1087, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 13268.967709154822, "hops": {"1": 6896.277834079228, "2": 6372.689875075594}, "receiver_prefill_ms": 227.23045805469155, "ttft_ms": 11.929041938856244, "decode_ms": 83.53291696403176}, "total_wall_ms": 13592.285459162667, "ttft_ms": 11.929041938856244, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 7}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "6056150", "evidence_artifact": {"ledger": {"facts_total": 63, "facts_retained": 27, "retention_rate": 0.42857142857142855, "retention_by_kind": {"person": 0.5, "date": 0.125, "amount": 0.6153846153846154, "status": 0.75, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1087}, "ledger_by_hop": {"1": {"facts_total": 63, "facts_retained": 27, "retention_rate": 0.42857142857142855, "retention_by_kind": {"person": 0.5, "date": 0.125, "amount": 0.6153846153846154, "status": 0.75, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1087}, "2": {"facts_total": 63, "facts_retained": 27, "retention_rate": 0.42857142857142855, "retention_by_kind": {"person": 0.5, "date": 0.125, "amount": 0.6153846153846154, "status": 0.75, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1087}}}, "reference_key_id": "keys/synthetic/syn_03027_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "1f88f0e78c63504e418dea843641b99a98555e15d47b02cd5cb2f57da3003dea", "output_hash": "f016f8b5fc26708e6c1ff75ff8e98659082c41a168f273819ac48b88feea8d3f"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "541ad99c626c3a41b2c3175d", "attempt_id": "5659a2252616", "run_group_id": "E4_core:A4|closed|qwen3b:h2", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:28:59.163426Z", "task_family": "synthetic", "document_id": "syn_03027_2000", "document_hash": "10326080979f71b625e4110c7c19767e5a752e267de5034f7e37d98cb6e06249", "question_id": "q10_temporal", "split": "test", "cluster_id": "syn_03027_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 2, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1955, "query_tokens": 613, "message_tokens": 512, "message_utf8_bytes": 1087, "cache_tensor_bytes": null, "serialized_payload_bytes": 1087, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 13268.967709154822, "hops": {"1": 6896.277834079228, "2": 6372.689875075594}, "receiver_prefill_ms": 229.40008400473744, "ttft_ms": 11.954874964430928, "decode_ms": 83.58804101590067}, "total_wall_ms": 13594.433375168592, "ttft_ms": 11.954874964430928, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 7}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "Lunlith-13", "evidence_artifact": {"ledger": {"facts_total": 63, "facts_retained": 27, "retention_rate": 0.42857142857142855, "retention_by_kind": {"person": 0.5, "date": 0.125, "amount": 0.6153846153846154, "status": 0.75, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1087}, "ledger_by_hop": {"1": {"facts_total": 63, "facts_retained": 27, "retention_rate": 0.42857142857142855, "retention_by_kind": {"person": 0.5, "date": 0.125, "amount": 0.6153846153846154, "status": 0.75, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1087}, "2": {"facts_total": 63, "facts_retained": 27, "retention_rate": 0.42857142857142855, "retention_by_kind": {"person": 0.5, "date": 0.125, "amount": 0.6153846153846154, "status": 0.75, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1087}}}, "reference_key_id": "keys/synthetic/syn_03027_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "1f88f0e78c63504e418dea843641b99a98555e15d47b02cd5cb2f57da3003dea", "output_hash": "c45f000afe6d865c8167954eebdff761141225817d4f5b26c962cd50cd75d391"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "b6f096479c1c3767ecfd79af", "attempt_id": "bedd18d26377", "run_group_id": "E4_core:A4|closed|qwen3b:h2", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:28:59.504837Z", "task_family": "synthetic", "document_id": "syn_03027_2000", "document_hash": "10326080979f71b625e4110c7c19767e5a752e267de5034f7e37d98cb6e06249", "question_id": "q11_aggregation", "split": "test", "cluster_id": "syn_03027_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 2, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1955, "query_tokens": 606, "message_tokens": 512, "message_utf8_bytes": 1087, "cache_tensor_bytes": null, "serialized_payload_bytes": 1087, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 13268.967709154822, "hops": {"1": 6896.277834079228, "2": 6372.689875075594}, "receiver_prefill_ms": 226.7585409572348, "ttft_ms": 11.888833018019795, "decode_ms": 107.57604101672769}, "total_wall_ms": 13615.696668159217, "ttft_ms": 11.888833018019795, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 9}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "23,450165", "evidence_artifact": {"ledger": {"facts_total": 63, "facts_retained": 27, "retention_rate": 0.42857142857142855, "retention_by_kind": {"person": 0.5, "date": 0.125, "amount": 0.6153846153846154, "status": 0.75, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1087}, "ledger_by_hop": {"1": {"facts_total": 63, "facts_retained": 27, "retention_rate": 0.42857142857142855, "retention_by_kind": {"person": 0.5, "date": 0.125, "amount": 0.6153846153846154, "status": 0.75, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1087}, "2": {"facts_total": 63, "facts_retained": 27, "retention_rate": 0.42857142857142855, "retention_by_kind": {"person": 0.5, "date": 0.125, "amount": 0.6153846153846154, "status": 0.75, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1087}}}, "reference_key_id": "keys/synthetic/syn_03027_2000.json", "task_scores": {"correct": 0.0, "abstained": 0.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "1f88f0e78c63504e418dea843641b99a98555e15d47b02cd5cb2f57da3003dea", "output_hash": "a929e5c5097069e202c39470de0d2861ac8d456564e72788db30daadb9f19129"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "03a625a270456957c9a5260d", "attempt_id": "d2753c6723dc", "run_group_id": "E4_core:A4|closed|qwen3b:h2", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:28:59.866982Z", "task_family": "synthetic", "document_id": "syn_03027_2000", "document_hash": "10326080979f71b625e4110c7c19767e5a752e267de5034f7e37d98cb6e06249", "question_id": "q12_unit_conversion", "split": "test", "cluster_id": "syn_03027_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 2, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1955, "query_tokens": 603, "message_tokens": 512, "message_utf8_bytes": 1087, "cache_tensor_bytes": null, "serialized_payload_bytes": 1087, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 13268.967709154822, "hops": {"1": 6896.277834079228, "2": 6372.689875075594}, "receiver_prefill_ms": 227.17066702898592, "ttft_ms": 12.123832944780588, "decode_ms": 11.935041984543204}, "total_wall_ms": 13520.759376115166, "ttft_ms": 12.123832944780588, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 63, "facts_retained": 27, "retention_rate": 0.42857142857142855, "retention_by_kind": {"person": 0.5, "date": 0.125, "amount": 0.6153846153846154, "status": 0.75, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1087}, "ledger_by_hop": {"1": {"facts_total": 63, "facts_retained": 27, "retention_rate": 0.42857142857142855, "retention_by_kind": {"person": 0.5, "date": 0.125, "amount": 0.6153846153846154, "status": 0.75, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1087}, "2": {"facts_total": 63, "facts_retained": 27, "retention_rate": 0.42857142857142855, "retention_by_kind": {"person": 0.5, "date": 0.125, "amount": 0.6153846153846154, "status": 0.75, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1087}}}, "reference_key_id": "keys/synthetic/syn_03027_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "1f88f0e78c63504e418dea843641b99a98555e15d47b02cd5cb2f57da3003dea", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "82286b4876452534535fc3f7", "attempt_id": "fc32f95bc122", "run_group_id": "E4_core:A4|closed|qwen3b:h2", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:29:00.134157Z", "task_family": "synthetic", "document_id": "syn_03027_2000", "document_hash": "10326080979f71b625e4110c7c19767e5a752e267de5034f7e37d98cb6e06249", "question_id": "q13_unanswerable", "split": "test", "cluster_id": "syn_03027_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 2, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1955, "query_tokens": 606, "message_tokens": 512, "message_utf8_bytes": 1087, "cache_tensor_bytes": null, "serialized_payload_bytes": 1087, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 13268.967709154822, "hops": {"1": 6896.277834079228, "2": 6372.689875075594}, "receiver_prefill_ms": 227.18558309134096, "ttft_ms": 11.929249973036349, "decode_ms": 11.930458946153522}, "total_wall_ms": 13520.495167118497, "ttft_ms": 11.929249973036349, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 63, "facts_retained": 27, "retention_rate": 0.42857142857142855, "retention_by_kind": {"person": 0.5, "date": 0.125, "amount": 0.6153846153846154, "status": 0.75, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1087}, "ledger_by_hop": {"1": {"facts_total": 63, "facts_retained": 27, "retention_rate": 0.42857142857142855, "retention_by_kind": {"person": 0.5, "date": 0.125, "amount": 0.6153846153846154, "status": 0.75, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1087}, "2": {"facts_total": 63, "facts_retained": 27, "retention_rate": 0.42857142857142855, "retention_by_kind": {"person": 0.5, "date": 0.125, "amount": 0.6153846153846154, "status": 0.75, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1087}}}, "reference_key_id": "keys/synthetic/syn_03027_2000.json", "task_scores": {"correct": 1.0, "abstained": 1.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "1f88f0e78c63504e418dea843641b99a98555e15d47b02cd5cb2f57da3003dea", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "ecb5609d72e2010bf8a46dd6", "attempt_id": "b3f114190f70", "run_group_id": "E4_core:A4|closed|qwen3b:h2", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:29:00.401203Z", "task_family": "synthetic", "document_id": "syn_03027_2000", "document_hash": "10326080979f71b625e4110c7c19767e5a752e267de5034f7e37d98cb6e06249", "question_id": "q14_unanswerable", "split": "test", "cluster_id": "syn_03027_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 2, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1955, "query_tokens": 608, "message_tokens": 512, "message_utf8_bytes": 1087, "cache_tensor_bytes": null, "serialized_payload_bytes": 1087, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 13268.967709154822, "hops": {"1": 6896.277834079228, "2": 6372.689875075594}, "receiver_prefill_ms": 227.34008298721164, "ttft_ms": 11.943458928726614, "decode_ms": 12.19241600483656}, "total_wall_ms": 13520.941001246683, "ttft_ms": 11.943458928726614, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 63, "facts_retained": 27, "retention_rate": 0.42857142857142855, "retention_by_kind": {"person": 0.5, "date": 0.125, "amount": 0.6153846153846154, "status": 0.75, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1087}, "ledger_by_hop": {"1": {"facts_total": 63, "facts_retained": 27, "retention_rate": 0.42857142857142855, "retention_by_kind": {"person": 0.5, "date": 0.125, "amount": 0.6153846153846154, "status": 0.75, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1087}, "2": {"facts_total": 63, "facts_retained": 27, "retention_rate": 0.42857142857142855, "retention_by_kind": {"person": 0.5, "date": 0.125, "amount": 0.6153846153846154, "status": 0.75, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1087}}}, "reference_key_id": "keys/synthetic/syn_03027_2000.json", "task_scores": {"correct": 1.0, "abstained": 1.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "1f88f0e78c63504e418dea843641b99a98555e15d47b02cd5cb2f57da3003dea", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "ff1006e28b6f6c0c9566fe34", "attempt_id": "a51fca6b3e37", "run_group_id": "E4_core:A4|closed|qwen3b:h4", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:29:13.407941Z", "task_family": "synthetic", "document_id": "syn_03027_2000", "document_hash": "10326080979f71b625e4110c7c19767e5a752e267de5034f7e37d98cb6e06249", "question_id": "q00_single_fact", "split": "test", "cluster_id": "syn_03027_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 4, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1955, "query_tokens": 594, "message_tokens": 512, "message_utf8_bytes": 1087, "cache_tensor_bytes": null, "serialized_payload_bytes": 1087, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 26005.745126167312, "hops": {"1": 6896.277834079228, "2": 6372.689875075594, "3": 6377.503250027075, "4": 6359.274166985415}, "receiver_prefill_ms": 226.7156249145046, "ttft_ms": 12.033959035761654, "decode_ms": 11.925957980565727}, "total_wall_ms": 26257.028710097075, "ttft_ms": 12.033959035761654, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 63, "facts_retained": 27, "retention_rate": 0.42857142857142855, "retention_by_kind": {"person": 0.5, "date": 0.125, "amount": 0.6153846153846154, "status": 0.75, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1087}, "ledger_by_hop": {"1": {"facts_total": 63, "facts_retained": 27, "retention_rate": 0.42857142857142855, "retention_by_kind": {"person": 0.5, "date": 0.125, "amount": 0.6153846153846154, "status": 0.75, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1087}, "2": {"facts_total": 63, "facts_retained": 27, "retention_rate": 0.42857142857142855, "retention_by_kind": {"person": 0.5, "date": 0.125, "amount": 0.6153846153846154, "status": 0.75, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1087}, "3": {"facts_total": 63, "facts_retained": 27, "retention_rate": 0.42857142857142855, "retention_by_kind": {"person": 0.5, "date": 0.125, "amount": 0.6153846153846154, "status": 0.75, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1087}, "4": {"facts_total": 63, "facts_retained": 27, "retention_rate": 0.42857142857142855, "retention_by_kind": {"person": 0.5, "date": 0.125, "amount": 0.6153846153846154, "status": 0.75, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1087}}}, "reference_key_id": "keys/synthetic/syn_03027_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "1f88f0e78c63504e418dea843641b99a98555e15d47b02cd5cb2f57da3003dea", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "0f34eb76b4cc9238155f6bda", "attempt_id": "53b937192b1b", "run_group_id": "E4_core:A4|closed|qwen3b:h4", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:29:13.675623Z", "task_family": "synthetic", "document_id": "syn_03027_2000", "document_hash": "10326080979f71b625e4110c7c19767e5a752e267de5034f7e37d98cb6e06249", "question_id": "q01_single_fact", "split": "test", "cluster_id": "syn_03027_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 4, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1955, "query_tokens": 601, "message_tokens": 512, "message_utf8_bytes": 1087, "cache_tensor_bytes": null, "serialized_payload_bytes": 1087, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 26005.745126167312, "hops": {"1": 6896.277834079228, "2": 6372.689875075594, "3": 6377.503250027075, "4": 6359.274166985415}, "receiver_prefill_ms": 226.34779091458768, "ttft_ms": 12.206749990582466, "decode_ms": 142.62641698587686}, "total_wall_ms": 26387.52383517567, "ttft_ms": 12.206749990582466, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 12}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "Lot 247-S, Baie-Comeau", "evidence_artifact": {"ledger": {"facts_total": 63, "facts_retained": 27, "retention_rate": 0.42857142857142855, "retention_by_kind": {"person": 0.5, "date": 0.125, "amount": 0.6153846153846154, "status": 0.75, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1087}, "ledger_by_hop": {"1": {"facts_total": 63, "facts_retained": 27, "retention_rate": 0.42857142857142855, "retention_by_kind": {"person": 0.5, "date": 0.125, "amount": 0.6153846153846154, "status": 0.75, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1087}, "2": {"facts_total": 63, "facts_retained": 27, "retention_rate": 0.42857142857142855, "retention_by_kind": {"person": 0.5, "date": 0.125, "amount": 0.6153846153846154, "status": 0.75, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1087}, "3": {"facts_total": 63, "facts_retained": 27, "retention_rate": 0.42857142857142855, "retention_by_kind": {"person": 0.5, "date": 0.125, "amount": 0.6153846153846154, "status": 0.75, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1087}, "4": {"facts_total": 63, "facts_retained": 27, "retention_rate": 0.42857142857142855, "retention_by_kind": {"person": 0.5, "date": 0.125, "amount": 0.6153846153846154, "status": 0.75, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1087}}}, "reference_key_id": "keys/synthetic/syn_03027_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 0.8571428571428571}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "1f88f0e78c63504e418dea843641b99a98555e15d47b02cd5cb2f57da3003dea", "output_hash": "77b7da1d43e7d8815e9e9159ebbd735045bc7f668260ba30b93674f70622b8ec"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "d865c8871bae28443662716e", "attempt_id": "e05e91e03c9d", "run_group_id": "E4_core:A4|closed|qwen3b:h4", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:29:14.073210Z", "task_family": "synthetic", "document_id": "syn_03027_2000", "document_hash": "10326080979f71b625e4110c7c19767e5a752e267de5034f7e37d98cb6e06249", "question_id": "q02_exact_copy", "split": "test", "cluster_id": "syn_03027_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 4, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1955, "query_tokens": 600, "message_tokens": 512, "message_utf8_bytes": 1087, "cache_tensor_bytes": null, "serialized_payload_bytes": 1087, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 26005.745126167312, "hops": {"1": 6896.277834079228, "2": 6372.689875075594, "3": 6377.503250027075, "4": 6359.274166985415}, "receiver_prefill_ms": 226.63799999281764, "ttft_ms": 11.91820902749896, "decode_ms": 82.98595796804875}, "total_wall_ms": 26327.778584207408, "ttft_ms": 11.91820902749896, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 7}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "15.0054", "evidence_artifact": {"ledger": {"facts_total": 63, "facts_retained": 27, "retention_rate": 0.42857142857142855, "retention_by_kind": {"person": 0.5, "date": 0.125, "amount": 0.6153846153846154, "status": 0.75, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1087}, "ledger_by_hop": {"1": {"facts_total": 63, "facts_retained": 27, "retention_rate": 0.42857142857142855, "retention_by_kind": {"person": 0.5, "date": 0.125, "amount": 0.6153846153846154, "status": 0.75, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1087}, "2": {"facts_total": 63, "facts_retained": 27, "retention_rate": 0.42857142857142855, "retention_by_kind": {"person": 0.5, "date": 0.125, "amount": 0.6153846153846154, "status": 0.75, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1087}, "3": {"facts_total": 63, "facts_retained": 27, "retention_rate": 0.42857142857142855, "retention_by_kind": {"person": 0.5, "date": 0.125, "amount": 0.6153846153846154, "status": 0.75, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1087}, "4": {"facts_total": 63, "facts_retained": 27, "retention_rate": 0.42857142857142855, "retention_by_kind": {"person": 0.5, "date": 0.125, "amount": 0.6153846153846154, "status": 0.75, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1087}}}, "reference_key_id": "keys/synthetic/syn_03027_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "1f88f0e78c63504e418dea843641b99a98555e15d47b02cd5cb2f57da3003dea", "output_hash": "d13adba73068c847c850237059e0202fc5b91ccacf2e792e9ce5e3b90556501a"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "6816ded36c88edd66b898984", "attempt_id": "19c5a78e1d25", "run_group_id": "E4_core:A4|closed|qwen3b:h4", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:29:14.410499Z", "task_family": "synthetic", "document_id": "syn_03027_2000", "document_hash": "10326080979f71b625e4110c7c19767e5a752e267de5034f7e37d98cb6e06249", "question_id": "q03_exact_copy", "split": "test", "cluster_id": "syn_03027_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 4, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1955, "query_tokens": 595, "message_tokens": 512, "message_utf8_bytes": 1087, "cache_tensor_bytes": null, "serialized_payload_bytes": 1087, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 26005.745126167312, "hops": {"1": 6896.277834079228, "2": 6372.689875075594, "3": 6377.503250027075, "4": 6359.274166985415}, "receiver_prefill_ms": 226.96537501178682, "ttft_ms": 12.162000057287514, "decode_ms": 106.8377080373466}, "total_wall_ms": 26352.300918195397, "ttft_ms": 12.162000057287514, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 9}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "Priyamvada Sørlie", "evidence_artifact": {"ledger": {"facts_total": 63, "facts_retained": 27, "retention_rate": 0.42857142857142855, "retention_by_kind": {"person": 0.5, "date": 0.125, "amount": 0.6153846153846154, "status": 0.75, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1087}, "ledger_by_hop": {"1": {"facts_total": 63, "facts_retained": 27, "retention_rate": 0.42857142857142855, "retention_by_kind": {"person": 0.5, "date": 0.125, "amount": 0.6153846153846154, "status": 0.75, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1087}, "2": {"facts_total": 63, "facts_retained": 27, "retention_rate": 0.42857142857142855, "retention_by_kind": {"person": 0.5, "date": 0.125, "amount": 0.6153846153846154, "status": 0.75, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1087}, "3": {"facts_total": 63, "facts_retained": 27, "retention_rate": 0.42857142857142855, "retention_by_kind": {"person": 0.5, "date": 0.125, "amount": 0.6153846153846154, "status": 0.75, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1087}, "4": {"facts_total": 63, "facts_retained": 27, "retention_rate": 0.42857142857142855, "retention_by_kind": {"person": 0.5, "date": 0.125, "amount": 0.6153846153846154, "status": 0.75, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1087}}}, "reference_key_id": "keys/synthetic/syn_03027_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "1f88f0e78c63504e418dea843641b99a98555e15d47b02cd5cb2f57da3003dea", "output_hash": "c5f554196ed23dd71ad995ec89f7142099d524dbb6534cc10d94a0c89d13e792"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "c301dd766530faf62455a168", "attempt_id": "81b901526295", "run_group_id": "E4_core:A4|closed|qwen3b:h4", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:29:14.772878Z", "task_family": "synthetic", "document_id": "syn_03027_2000", "document_hash": "10326080979f71b625e4110c7c19767e5a752e267de5034f7e37d98cb6e06249", "question_id": "q04_multi_hop", "split": "test", "cluster_id": "syn_03027_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 4, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1955, "query_tokens": 606, "message_tokens": 512, "message_utf8_bytes": 1087, "cache_tensor_bytes": null, "serialized_payload_bytes": 1087, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 26005.745126167312, "hops": {"1": 6896.277834079228, "2": 6372.689875075594, "3": 6377.503250027075, "4": 6359.274166985415}, "receiver_prefill_ms": 227.0164160290733, "ttft_ms": 12.081584078259766, "decode_ms": 11.892499984242022}, "total_wall_ms": 26257.335210219026, "ttft_ms": 12.081584078259766, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 63, "facts_retained": 27, "retention_rate": 0.42857142857142855, "retention_by_kind": {"person": 0.5, "date": 0.125, "amount": 0.6153846153846154, "status": 0.75, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1087}, "ledger_by_hop": {"1": {"facts_total": 63, "facts_retained": 27, "retention_rate": 0.42857142857142855, "retention_by_kind": {"person": 0.5, "date": 0.125, "amount": 0.6153846153846154, "status": 0.75, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1087}, "2": {"facts_total": 63, "facts_retained": 27, "retention_rate": 0.42857142857142855, "retention_by_kind": {"person": 0.5, "date": 0.125, "amount": 0.6153846153846154, "status": 0.75, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1087}, "3": {"facts_total": 63, "facts_retained": 27, "retention_rate": 0.42857142857142855, "retention_by_kind": {"person": 0.5, "date": 0.125, "amount": 0.6153846153846154, "status": 0.75, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1087}, "4": {"facts_total": 63, "facts_retained": 27, "retention_rate": 0.42857142857142855, "retention_by_kind": {"person": 0.5, "date": 0.125, "amount": 0.6153846153846154, "status": 0.75, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1087}}}, "reference_key_id": "keys/synthetic/syn_03027_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "1f88f0e78c63504e418dea843641b99a98555e15d47b02cd5cb2f57da3003dea", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "515b9ba7e270b411ea70bebc", "attempt_id": "22dcd59bf4de", "run_group_id": "E4_core:A4|closed|qwen3b:h4", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:29:15.040142Z", "task_family": "synthetic", "document_id": "syn_03027_2000", "document_hash": "10326080979f71b625e4110c7c19767e5a752e267de5034f7e37d98cb6e06249", "question_id": "q05_multi_hop", "split": "test", "cluster_id": "syn_03027_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 4, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1955, "query_tokens": 603, "message_tokens": 512, "message_utf8_bytes": 1087, "cache_tensor_bytes": null, "serialized_payload_bytes": 1087, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 26005.745126167312, "hops": {"1": 6896.277834079228, "2": 6372.689875075594, "3": 6377.503250027075, "4": 6359.274166985415}, "receiver_prefill_ms": 227.2122090216726, "ttft_ms": 11.936416034586728, "decode_ms": 12.089541996829212}, "total_wall_ms": 26257.46879214421, "ttft_ms": 11.936416034586728, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 63, "facts_retained": 27, "retention_rate": 0.42857142857142855, "retention_by_kind": {"person": 0.5, "date": 0.125, "amount": 0.6153846153846154, "status": 0.75, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1087}, "ledger_by_hop": {"1": {"facts_total": 63, "facts_retained": 27, "retention_rate": 0.42857142857142855, "retention_by_kind": {"person": 0.5, "date": 0.125, "amount": 0.6153846153846154, "status": 0.75, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1087}, "2": {"facts_total": 63, "facts_retained": 27, "retention_rate": 0.42857142857142855, "retention_by_kind": {"person": 0.5, "date": 0.125, "amount": 0.6153846153846154, "status": 0.75, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1087}, "3": {"facts_total": 63, "facts_retained": 27, "retention_rate": 0.42857142857142855, "retention_by_kind": {"person": 0.5, "date": 0.125, "amount": 0.6153846153846154, "status": 0.75, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1087}, "4": {"facts_total": 63, "facts_retained": 27, "retention_rate": 0.42857142857142855, "retention_by_kind": {"person": 0.5, "date": 0.125, "amount": 0.6153846153846154, "status": 0.75, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1087}}}, "reference_key_id": "keys/synthetic/syn_03027_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "1f88f0e78c63504e418dea843641b99a98555e15d47b02cd5cb2f57da3003dea", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "d1511681c18cdabba4457282", "attempt_id": "034bd2cbc782", "run_group_id": "E4_core:A4|closed|qwen3b:h4", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:29:15.307282Z", "task_family": "synthetic", "document_id": "syn_03027_2000", "document_hash": "10326080979f71b625e4110c7c19767e5a752e267de5034f7e37d98cb6e06249", "question_id": "q06_negation", "split": "test", "cluster_id": "syn_03027_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 4, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1955, "query_tokens": 595, "message_tokens": 512, "message_utf8_bytes": 1087, "cache_tensor_bytes": null, "serialized_payload_bytes": 1087, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 26005.745126167312, "hops": {"1": 6896.277834079228, "2": 6372.689875075594, "3": 6377.503250027075, "4": 6359.274166985415}, "receiver_prefill_ms": 226.56787501182407, "ttft_ms": 12.073084013536572, "decode_ms": 11.860958067700267}, "total_wall_ms": 26256.818293244578, "ttft_ms": 12.073084013536572, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 63, "facts_retained": 27, "retention_rate": 0.42857142857142855, "retention_by_kind": {"person": 0.5, "date": 0.125, "amount": 0.6153846153846154, "status": 0.75, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1087}, "ledger_by_hop": {"1": {"facts_total": 63, "facts_retained": 27, "retention_rate": 0.42857142857142855, "retention_by_kind": {"person": 0.5, "date": 0.125, "amount": 0.6153846153846154, "status": 0.75, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1087}, "2": {"facts_total": 63, "facts_retained": 27, "retention_rate": 0.42857142857142855, "retention_by_kind": {"person": 0.5, "date": 0.125, "amount": 0.6153846153846154, "status": 0.75, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1087}, "3": {"facts_total": 63, "facts_retained": 27, "retention_rate": 0.42857142857142855, "retention_by_kind": {"person": 0.5, "date": 0.125, "amount": 0.6153846153846154, "status": 0.75, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1087}, "4": {"facts_total": 63, "facts_retained": 27, "retention_rate": 0.42857142857142855, "retention_by_kind": {"person": 0.5, "date": 0.125, "amount": 0.6153846153846154, "status": 0.75, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1087}}}, "reference_key_id": "keys/synthetic/syn_03027_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "1f88f0e78c63504e418dea843641b99a98555e15d47b02cd5cb2f57da3003dea", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "2ec7fa37799fc68a66dde9e2", "attempt_id": "a4a75d106f20", "run_group_id": "E4_core:A4|closed|qwen3b:h4", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:29:15.573209Z", "task_family": "synthetic", "document_id": "syn_03027_2000", "document_hash": "10326080979f71b625e4110c7c19767e5a752e267de5034f7e37d98cb6e06249", "question_id": "q07_negation", "split": "test", "cluster_id": "syn_03027_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 4, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1955, "query_tokens": 596, "message_tokens": 512, "message_utf8_bytes": 1087, "cache_tensor_bytes": null, "serialized_payload_bytes": 1087, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 26005.745126167312, "hops": {"1": 6896.277834079228, "2": 6372.689875075594, "3": 6377.503250027075, "4": 6359.274166985415}, "receiver_prefill_ms": 226.69395804405212, "ttft_ms": 11.937333038076758, "decode_ms": 11.821958003565669}, "total_wall_ms": 26256.68870913796, "ttft_ms": 11.937333038076758, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "no", "evidence_artifact": {"ledger": {"facts_total": 63, "facts_retained": 27, "retention_rate": 0.42857142857142855, "retention_by_kind": {"person": 0.5, "date": 0.125, "amount": 0.6153846153846154, "status": 0.75, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1087}, "ledger_by_hop": {"1": {"facts_total": 63, "facts_retained": 27, "retention_rate": 0.42857142857142855, "retention_by_kind": {"person": 0.5, "date": 0.125, "amount": 0.6153846153846154, "status": 0.75, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1087}, "2": {"facts_total": 63, "facts_retained": 27, "retention_rate": 0.42857142857142855, "retention_by_kind": {"person": 0.5, "date": 0.125, "amount": 0.6153846153846154, "status": 0.75, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1087}, "3": {"facts_total": 63, "facts_retained": 27, "retention_rate": 0.42857142857142855, "retention_by_kind": {"person": 0.5, "date": 0.125, "amount": 0.6153846153846154, "status": 0.75, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1087}, "4": {"facts_total": 63, "facts_retained": 27, "retention_rate": 0.42857142857142855, "retention_by_kind": {"person": 0.5, "date": 0.125, "amount": 0.6153846153846154, "status": 0.75, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1087}}}, "reference_key_id": "keys/synthetic/syn_03027_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "1f88f0e78c63504e418dea843641b99a98555e15d47b02cd5cb2f57da3003dea", "output_hash": "9390298f3fb0c5b160498935d79cb139aef28e1c47358b4bbba61862b9c26e59"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "d27ec1b890aacecdb5689e44", "attempt_id": "9e1a0fbc23ca", "run_group_id": "E4_core:A4|closed|qwen3b:h4", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:29:15.839519Z", "task_family": "synthetic", "document_id": "syn_03027_2000", "document_hash": "10326080979f71b625e4110c7c19767e5a752e267de5034f7e37d98cb6e06249", "question_id": "q08_revision", "split": "test", "cluster_id": "syn_03027_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 4, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1955, "query_tokens": 610, "message_tokens": 512, "message_utf8_bytes": 1087, "cache_tensor_bytes": null, "serialized_payload_bytes": 1087, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 26005.745126167312, "hops": {"1": 6896.277834079228, "2": 6372.689875075594, "3": 6377.503250027075, "4": 6359.274166985415}, "receiver_prefill_ms": 228.81687502376735, "ttft_ms": 12.114583048969507, "decode_ms": 83.16391706466675}, "total_wall_ms": 26330.437876167707, "ttft_ms": 12.114583048969507, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 7}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "8010205", "evidence_artifact": {"ledger": {"facts_total": 63, "facts_retained": 27, "retention_rate": 0.42857142857142855, "retention_by_kind": {"person": 0.5, "date": 0.125, "amount": 0.6153846153846154, "status": 0.75, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1087}, "ledger_by_hop": {"1": {"facts_total": 63, "facts_retained": 27, "retention_rate": 0.42857142857142855, "retention_by_kind": {"person": 0.5, "date": 0.125, "amount": 0.6153846153846154, "status": 0.75, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1087}, "2": {"facts_total": 63, "facts_retained": 27, "retention_rate": 0.42857142857142855, "retention_by_kind": {"person": 0.5, "date": 0.125, "amount": 0.6153846153846154, "status": 0.75, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1087}, "3": {"facts_total": 63, "facts_retained": 27, "retention_rate": 0.42857142857142855, "retention_by_kind": {"person": 0.5, "date": 0.125, "amount": 0.6153846153846154, "status": 0.75, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1087}, "4": {"facts_total": 63, "facts_retained": 27, "retention_rate": 0.42857142857142855, "retention_by_kind": {"person": 0.5, "date": 0.125, "amount": 0.6153846153846154, "status": 0.75, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1087}}}, "reference_key_id": "keys/synthetic/syn_03027_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "1f88f0e78c63504e418dea843641b99a98555e15d47b02cd5cb2f57da3003dea", "output_hash": "6d1ceb36132f38bd77e8f46c23ea21dd9bcff78bc80bd8a9eaf2d7f2f3bacf46"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "9a44dbfe50a76668bb7d8098", "attempt_id": "25b5e424bc5e", "run_group_id": "E4_core:A4|closed|qwen3b:h4", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:29:16.179154Z", "task_family": "synthetic", "document_id": "syn_03027_2000", "document_hash": "10326080979f71b625e4110c7c19767e5a752e267de5034f7e37d98cb6e06249", "question_id": "q09_revision", "split": "test", "cluster_id": "syn_03027_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 4, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1955, "query_tokens": 603, "message_tokens": 512, "message_utf8_bytes": 1087, "cache_tensor_bytes": null, "serialized_payload_bytes": 1087, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 26005.745126167312, "hops": {"1": 6896.277834079228, "2": 6372.689875075594, "3": 6377.503250027075, "4": 6359.274166985415}, "receiver_prefill_ms": 227.1562909008935, "ttft_ms": 11.805207934230566, "decode_ms": 82.96349993906915}, "total_wall_ms": 26328.26362620108, "ttft_ms": 11.805207934230566, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 7}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "6056150", "evidence_artifact": {"ledger": {"facts_total": 63, "facts_retained": 27, "retention_rate": 0.42857142857142855, "retention_by_kind": {"person": 0.5, "date": 0.125, "amount": 0.6153846153846154, "status": 0.75, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1087}, "ledger_by_hop": {"1": {"facts_total": 63, "facts_retained": 27, "retention_rate": 0.42857142857142855, "retention_by_kind": {"person": 0.5, "date": 0.125, "amount": 0.6153846153846154, "status": 0.75, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1087}, "2": {"facts_total": 63, "facts_retained": 27, "retention_rate": 0.42857142857142855, "retention_by_kind": {"person": 0.5, "date": 0.125, "amount": 0.6153846153846154, "status": 0.75, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1087}, "3": {"facts_total": 63, "facts_retained": 27, "retention_rate": 0.42857142857142855, "retention_by_kind": {"person": 0.5, "date": 0.125, "amount": 0.6153846153846154, "status": 0.75, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1087}, "4": {"facts_total": 63, "facts_retained": 27, "retention_rate": 0.42857142857142855, "retention_by_kind": {"person": 0.5, "date": 0.125, "amount": 0.6153846153846154, "status": 0.75, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1087}}}, "reference_key_id": "keys/synthetic/syn_03027_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "1f88f0e78c63504e418dea843641b99a98555e15d47b02cd5cb2f57da3003dea", "output_hash": "f016f8b5fc26708e6c1ff75ff8e98659082c41a168f273819ac48b88feea8d3f"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "ce4a8ce342fe2b4d9d75cef4", "attempt_id": "3145a04bf5b4", "run_group_id": "E4_core:A4|closed|qwen3b:h4", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:29:16.517038Z", "task_family": "synthetic", "document_id": "syn_03027_2000", "document_hash": "10326080979f71b625e4110c7c19767e5a752e267de5034f7e37d98cb6e06249", "question_id": "q10_temporal", "split": "test", "cluster_id": "syn_03027_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 4, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1955, "query_tokens": 613, "message_tokens": 512, "message_utf8_bytes": 1087, "cache_tensor_bytes": null, "serialized_payload_bytes": 1087, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 26005.745126167312, "hops": {"1": 6896.277834079228, "2": 6372.689875075594, "3": 6377.503250027075, "4": 6359.274166985415}, "receiver_prefill_ms": 229.53029198106378, "ttft_ms": 11.93220796994865, "decode_ms": 83.03158299531788}, "total_wall_ms": 26330.745167215355, "ttft_ms": 11.93220796994865, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 7}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "Lunlith-13", "evidence_artifact": {"ledger": {"facts_total": 63, "facts_retained": 27, "retention_rate": 0.42857142857142855, "retention_by_kind": {"person": 0.5, "date": 0.125, "amount": 0.6153846153846154, "status": 0.75, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1087}, "ledger_by_hop": {"1": {"facts_total": 63, "facts_retained": 27, "retention_rate": 0.42857142857142855, "retention_by_kind": {"person": 0.5, "date": 0.125, "amount": 0.6153846153846154, "status": 0.75, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1087}, "2": {"facts_total": 63, "facts_retained": 27, "retention_rate": 0.42857142857142855, "retention_by_kind": {"person": 0.5, "date": 0.125, "amount": 0.6153846153846154, "status": 0.75, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1087}, "3": {"facts_total": 63, "facts_retained": 27, "retention_rate": 0.42857142857142855, "retention_by_kind": {"person": 0.5, "date": 0.125, "amount": 0.6153846153846154, "status": 0.75, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1087}, "4": {"facts_total": 63, "facts_retained": 27, "retention_rate": 0.42857142857142855, "retention_by_kind": {"person": 0.5, "date": 0.125, "amount": 0.6153846153846154, "status": 0.75, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1087}}}, "reference_key_id": "keys/synthetic/syn_03027_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "1f88f0e78c63504e418dea843641b99a98555e15d47b02cd5cb2f57da3003dea", "output_hash": "c45f000afe6d865c8167954eebdff761141225817d4f5b26c962cd50cd75d391"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "cf43bac325be2e3c6cdbe538", "attempt_id": "02116ce0e197", "run_group_id": "E4_core:A4|closed|qwen3b:h4", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:29:16.857399Z", "task_family": "synthetic", "document_id": "syn_03027_2000", "document_hash": "10326080979f71b625e4110c7c19767e5a752e267de5034f7e37d98cb6e06249", "question_id": "q11_aggregation", "split": "test", "cluster_id": "syn_03027_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 4, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1955, "query_tokens": 606, "message_tokens": 512, "message_utf8_bytes": 1087, "cache_tensor_bytes": null, "serialized_payload_bytes": 1087, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 26005.745126167312, "hops": {"1": 6896.277834079228, "2": 6372.689875075594, "3": 6377.503250027075, "4": 6359.274166985415}, "receiver_prefill_ms": 227.23541699815542, "ttft_ms": 11.96366606745869, "decode_ms": 106.80866707116365}, "total_wall_ms": 26352.34383516945, "ttft_ms": 11.96366606745869, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 9}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "23,450165", "evidence_artifact": {"ledger": {"facts_total": 63, "facts_retained": 27, "retention_rate": 0.42857142857142855, "retention_by_kind": {"person": 0.5, "date": 0.125, "amount": 0.6153846153846154, "status": 0.75, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1087}, "ledger_by_hop": {"1": {"facts_total": 63, "facts_retained": 27, "retention_rate": 0.42857142857142855, "retention_by_kind": {"person": 0.5, "date": 0.125, "amount": 0.6153846153846154, "status": 0.75, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1087}, "2": {"facts_total": 63, "facts_retained": 27, "retention_rate": 0.42857142857142855, "retention_by_kind": {"person": 0.5, "date": 0.125, "amount": 0.6153846153846154, "status": 0.75, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1087}, "3": {"facts_total": 63, "facts_retained": 27, "retention_rate": 0.42857142857142855, "retention_by_kind": {"person": 0.5, "date": 0.125, "amount": 0.6153846153846154, "status": 0.75, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1087}, "4": {"facts_total": 63, "facts_retained": 27, "retention_rate": 0.42857142857142855, "retention_by_kind": {"person": 0.5, "date": 0.125, "amount": 0.6153846153846154, "status": 0.75, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1087}}}, "reference_key_id": "keys/synthetic/syn_03027_2000.json", "task_scores": {"correct": 0.0, "abstained": 0.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "1f88f0e78c63504e418dea843641b99a98555e15d47b02cd5cb2f57da3003dea", "output_hash": "a929e5c5097069e202c39470de0d2861ac8d456564e72788db30daadb9f19129"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "cb444dc416a2807482ddba59", "attempt_id": "4bd944d4aad0", "run_group_id": "E4_core:A4|closed|qwen3b:h4", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:29:17.219825Z", "task_family": "synthetic", "document_id": "syn_03027_2000", "document_hash": "10326080979f71b625e4110c7c19767e5a752e267de5034f7e37d98cb6e06249", "question_id": "q12_unit_conversion", "split": "test", "cluster_id": "syn_03027_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 4, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1955, "query_tokens": 603, "message_tokens": 512, "message_utf8_bytes": 1087, "cache_tensor_bytes": null, "serialized_payload_bytes": 1087, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 26005.745126167312, "hops": {"1": 6896.277834079228, "2": 6372.689875075594, "3": 6377.503250027075, "4": 6359.274166985415}, "receiver_prefill_ms": 227.01216698624194, "ttft_ms": 12.129625072702765, "decode_ms": 11.959999916143715}, "total_wall_ms": 26257.44529312942, "ttft_ms": 12.129625072702765, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 63, "facts_retained": 27, "retention_rate": 0.42857142857142855, "retention_by_kind": {"person": 0.5, "date": 0.125, "amount": 0.6153846153846154, "status": 0.75, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1087}, "ledger_by_hop": {"1": {"facts_total": 63, "facts_retained": 27, "retention_rate": 0.42857142857142855, "retention_by_kind": {"person": 0.5, "date": 0.125, "amount": 0.6153846153846154, "status": 0.75, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1087}, "2": {"facts_total": 63, "facts_retained": 27, "retention_rate": 0.42857142857142855, "retention_by_kind": {"person": 0.5, "date": 0.125, "amount": 0.6153846153846154, "status": 0.75, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1087}, "3": {"facts_total": 63, "facts_retained": 27, "retention_rate": 0.42857142857142855, "retention_by_kind": {"person": 0.5, "date": 0.125, "amount": 0.6153846153846154, "status": 0.75, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1087}, "4": {"facts_total": 63, "facts_retained": 27, "retention_rate": 0.42857142857142855, "retention_by_kind": {"person": 0.5, "date": 0.125, "amount": 0.6153846153846154, "status": 0.75, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1087}}}, "reference_key_id": "keys/synthetic/syn_03027_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "1f88f0e78c63504e418dea843641b99a98555e15d47b02cd5cb2f57da3003dea", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "9cd4e7c12cea665f50ab676a", "attempt_id": "6946650fb94e", "run_group_id": "E4_core:A4|closed|qwen3b:h4", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:29:17.487458Z", "task_family": "synthetic", "document_id": "syn_03027_2000", "document_hash": "10326080979f71b625e4110c7c19767e5a752e267de5034f7e37d98cb6e06249", "question_id": "q13_unanswerable", "split": "test", "cluster_id": "syn_03027_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 4, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1955, "query_tokens": 606, "message_tokens": 512, "message_utf8_bytes": 1087, "cache_tensor_bytes": null, "serialized_payload_bytes": 1087, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 26005.745126167312, "hops": {"1": 6896.277834079228, "2": 6372.689875075594, "3": 6377.503250027075, "4": 6359.274166985415}, "receiver_prefill_ms": 227.3828339530155, "ttft_ms": 12.002791976556182, "decode_ms": 11.87199994456023}, "total_wall_ms": 26257.582126068883, "ttft_ms": 12.002791976556182, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 63, "facts_retained": 27, "retention_rate": 0.42857142857142855, "retention_by_kind": {"person": 0.5, "date": 0.125, "amount": 0.6153846153846154, "status": 0.75, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1087}, "ledger_by_hop": {"1": {"facts_total": 63, "facts_retained": 27, "retention_rate": 0.42857142857142855, "retention_by_kind": {"person": 0.5, "date": 0.125, "amount": 0.6153846153846154, "status": 0.75, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1087}, "2": {"facts_total": 63, "facts_retained": 27, "retention_rate": 0.42857142857142855, "retention_by_kind": {"person": 0.5, "date": 0.125, "amount": 0.6153846153846154, "status": 0.75, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1087}, "3": {"facts_total": 63, "facts_retained": 27, "retention_rate": 0.42857142857142855, "retention_by_kind": {"person": 0.5, "date": 0.125, "amount": 0.6153846153846154, "status": 0.75, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1087}, "4": {"facts_total": 63, "facts_retained": 27, "retention_rate": 0.42857142857142855, "retention_by_kind": {"person": 0.5, "date": 0.125, "amount": 0.6153846153846154, "status": 0.75, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1087}}}, "reference_key_id": "keys/synthetic/syn_03027_2000.json", "task_scores": {"correct": 1.0, "abstained": 1.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "1f88f0e78c63504e418dea843641b99a98555e15d47b02cd5cb2f57da3003dea", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "91adf0fd87f698d5fc4ccc26", "attempt_id": "9ce3f9c45474", "run_group_id": "E4_core:A4|closed|qwen3b:h4", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:29:17.754665Z", "task_family": "synthetic", "document_id": "syn_03027_2000", "document_hash": "10326080979f71b625e4110c7c19767e5a752e267de5034f7e37d98cb6e06249", "question_id": "q14_unanswerable", "split": "test", "cluster_id": "syn_03027_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 4, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1955, "query_tokens": 608, "message_tokens": 512, "message_utf8_bytes": 1087, "cache_tensor_bytes": null, "serialized_payload_bytes": 1087, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 26005.745126167312, "hops": {"1": 6896.277834079228, "2": 6372.689875075594, "3": 6377.503250027075, "4": 6359.274166985415}, "receiver_prefill_ms": 227.2061250405386, "ttft_ms": 12.478624936193228, "decode_ms": 12.182833044789732}, "total_wall_ms": 26258.310334174894, "ttft_ms": 12.478624936193228, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 63, "facts_retained": 27, "retention_rate": 0.42857142857142855, "retention_by_kind": {"person": 0.5, "date": 0.125, "amount": 0.6153846153846154, "status": 0.75, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1087}, "ledger_by_hop": {"1": {"facts_total": 63, "facts_retained": 27, "retention_rate": 0.42857142857142855, "retention_by_kind": {"person": 0.5, "date": 0.125, "amount": 0.6153846153846154, "status": 0.75, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1087}, "2": {"facts_total": 63, "facts_retained": 27, "retention_rate": 0.42857142857142855, "retention_by_kind": {"person": 0.5, "date": 0.125, "amount": 0.6153846153846154, "status": 0.75, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1087}, "3": {"facts_total": 63, "facts_retained": 27, "retention_rate": 0.42857142857142855, "retention_by_kind": {"person": 0.5, "date": 0.125, "amount": 0.6153846153846154, "status": 0.75, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1087}, "4": {"facts_total": 63, "facts_retained": 27, "retention_rate": 0.42857142857142855, "retention_by_kind": {"person": 0.5, "date": 0.125, "amount": 0.6153846153846154, "status": 0.75, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1087}}}, "reference_key_id": "keys/synthetic/syn_03027_2000.json", "task_scores": {"correct": 1.0, "abstained": 1.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "1f88f0e78c63504e418dea843641b99a98555e15d47b02cd5cb2f57da3003dea", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "37df68d8055d79fbfd23563b", "attempt_id": "ca9e93fa7e21", "run_group_id": "E4_core:A4|closed|qwen3b:h8", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:29:43.427164Z", "task_family": "synthetic", "document_id": "syn_03027_2000", "document_hash": "10326080979f71b625e4110c7c19767e5a752e267de5034f7e37d98cb6e06249", "question_id": "q00_single_fact", "split": "test", "cluster_id": "syn_03027_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 8, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1955, "query_tokens": 594, "message_tokens": 512, "message_utf8_bytes": 1087, "cache_tensor_bytes": null, "serialized_payload_bytes": 1087, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 51405.68041824736, "hops": {"1": 6896.277834079228, "2": 6372.689875075594, "3": 6377.503250027075, "4": 6359.274166985415, "5": 6348.349208012223, "6": 6350.234291981906, "7": 6348.7960420316085, "8": 6352.555750054307}, "receiver_prefill_ms": 227.07891697064042, "ttft_ms": 11.851249961182475, "decode_ms": 11.946334037929773}, "total_wall_ms": 51657.05141820945, "ttft_ms": 11.851249961182475, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 63, "facts_retained": 27, "retention_rate": 0.42857142857142855, "retention_by_kind": {"person": 0.5, "date": 0.125, "amount": 0.6153846153846154, "status": 0.75, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1087}, "ledger_by_hop": {"1": {"facts_total": 63, "facts_retained": 27, "retention_rate": 0.42857142857142855, "retention_by_kind": {"person": 0.5, "date": 0.125, "amount": 0.6153846153846154, "status": 0.75, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1087}, "2": {"facts_total": 63, "facts_retained": 27, "retention_rate": 0.42857142857142855, "retention_by_kind": {"person": 0.5, "date": 0.125, "amount": 0.6153846153846154, "status": 0.75, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1087}, "3": {"facts_total": 63, "facts_retained": 27, "retention_rate": 0.42857142857142855, "retention_by_kind": {"person": 0.5, "date": 0.125, "amount": 0.6153846153846154, "status": 0.75, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1087}, "4": {"facts_total": 63, "facts_retained": 27, "retention_rate": 0.42857142857142855, "retention_by_kind": {"person": 0.5, "date": 0.125, "amount": 0.6153846153846154, "status": 0.75, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1087}, "5": {"facts_total": 63, "facts_retained": 27, "retention_rate": 0.42857142857142855, "retention_by_kind": {"person": 0.5, "date": 0.125, "amount": 0.6153846153846154, "status": 0.75, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1087}, "6": {"facts_total": 63, "facts_retained": 27, "retention_rate": 0.42857142857142855, "retention_by_kind": {"person": 0.5, "date": 0.125, "amount": 0.6153846153846154, "status": 0.75, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1087}, "7": {"facts_total": 63, "facts_retained": 27, "retention_rate": 0.42857142857142855, "retention_by_kind": {"person": 0.5, "date": 0.125, "amount": 0.6153846153846154, "status": 0.75, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1087}, "8": {"facts_total": 63, "facts_retained": 27, "retention_rate": 0.42857142857142855, "retention_by_kind": {"person": 0.5, "date": 0.125, "amount": 0.6153846153846154, "status": 0.75, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1087}}}, "reference_key_id": "keys/synthetic/syn_03027_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "1f88f0e78c63504e418dea843641b99a98555e15d47b02cd5cb2f57da3003dea", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "3fbb5dbcefb601f59b61329c", "attempt_id": "d116a2b33bc3", "run_group_id": "E4_core:A4|closed|qwen3b:h8", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:29:43.694527Z", "task_family": "synthetic", "document_id": "syn_03027_2000", "document_hash": "10326080979f71b625e4110c7c19767e5a752e267de5034f7e37d98cb6e06249", "question_id": "q01_single_fact", "split": "test", "cluster_id": "syn_03027_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 8, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1955, "query_tokens": 601, "message_tokens": 512, "message_utf8_bytes": 1087, "cache_tensor_bytes": null, "serialized_payload_bytes": 1087, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 51405.68041824736, "hops": {"1": 6896.277834079228, "2": 6372.689875075594, "3": 6377.503250027075, "4": 6359.274166985415, "5": 6348.349208012223, "6": 6350.234291981906, "7": 6348.7960420316085, "8": 6352.555750054307}, "receiver_prefill_ms": 227.05824999138713, "ttft_ms": 11.875666910782456, "decode_ms": 142.63566699810326}, "total_wall_ms": 51787.73366822861, "ttft_ms": 11.875666910782456, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 12}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "Lot 247-S, Baie-Comeau", "evidence_artifact": {"ledger": {"facts_total": 63, "facts_retained": 27, "retention_rate": 0.42857142857142855, "retention_by_kind": {"person": 0.5, "date": 0.125, "amount": 0.6153846153846154, "status": 0.75, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1087}, "ledger_by_hop": {"1": {"facts_total": 63, "facts_retained": 27, "retention_rate": 0.42857142857142855, "retention_by_kind": {"person": 0.5, "date": 0.125, "amount": 0.6153846153846154, "status": 0.75, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1087}, "2": {"facts_total": 63, "facts_retained": 27, "retention_rate": 0.42857142857142855, "retention_by_kind": {"person": 0.5, "date": 0.125, "amount": 0.6153846153846154, "status": 0.75, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1087}, "3": {"facts_total": 63, "facts_retained": 27, "retention_rate": 0.42857142857142855, "retention_by_kind": {"person": 0.5, "date": 0.125, "amount": 0.6153846153846154, "status": 0.75, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1087}, "4": {"facts_total": 63, "facts_retained": 27, "retention_rate": 0.42857142857142855, "retention_by_kind": {"person": 0.5, "date": 0.125, "amount": 0.6153846153846154, "status": 0.75, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1087}, "5": {"facts_total": 63, "facts_retained": 27, "retention_rate": 0.42857142857142855, "retention_by_kind": {"person": 0.5, "date": 0.125, "amount": 0.6153846153846154, "status": 0.75, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1087}, "6": {"facts_total": 63, "facts_retained": 27, "retention_rate": 0.42857142857142855, "retention_by_kind": {"person": 0.5, "date": 0.125, "amount": 0.6153846153846154, "status": 0.75, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1087}, "7": {"facts_total": 63, "facts_retained": 27, "retention_rate": 0.42857142857142855, "retention_by_kind": {"person": 0.5, "date": 0.125, "amount": 0.6153846153846154, "status": 0.75, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1087}, "8": {"facts_total": 63, "facts_retained": 27, "retention_rate": 0.42857142857142855, "retention_by_kind": {"person": 0.5, "date": 0.125, "amount": 0.6153846153846154, "status": 0.75, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1087}}}, "reference_key_id": "keys/synthetic/syn_03027_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 0.8571428571428571}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "1f88f0e78c63504e418dea843641b99a98555e15d47b02cd5cb2f57da3003dea", "output_hash": "77b7da1d43e7d8815e9e9159ebbd735045bc7f668260ba30b93674f70622b8ec"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "5cee5fdde1034c1da3597c7c", "attempt_id": "b6cadff16730", "run_group_id": "E4_core:A4|closed|qwen3b:h8", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:29:44.091872Z", "task_family": "synthetic", "document_id": "syn_03027_2000", "document_hash": "10326080979f71b625e4110c7c19767e5a752e267de5034f7e37d98cb6e06249", "question_id": "q02_exact_copy", "split": "test", "cluster_id": "syn_03027_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 8, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1955, "query_tokens": 600, "message_tokens": 512, "message_utf8_bytes": 1087, "cache_tensor_bytes": null, "serialized_payload_bytes": 1087, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 51405.68041824736, "hops": {"1": 6896.277834079228, "2": 6372.689875075594, "3": 6377.503250027075, "4": 6359.274166985415, "5": 6348.349208012223, "6": 6350.234291981906, "7": 6348.7960420316085, "8": 6352.555750054307}, "receiver_prefill_ms": 227.16091701295227, "ttft_ms": 11.90966705325991, "decode_ms": 83.14320794306695}, "total_wall_ms": 51728.349834214896, "ttft_ms": 11.90966705325991, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 7}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "15.0054", "evidence_artifact": {"ledger": {"facts_total": 63, "facts_retained": 27, "retention_rate": 0.42857142857142855, "retention_by_kind": {"person": 0.5, "date": 0.125, "amount": 0.6153846153846154, "status": 0.75, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1087}, "ledger_by_hop": {"1": {"facts_total": 63, "facts_retained": 27, "retention_rate": 0.42857142857142855, "retention_by_kind": {"person": 0.5, "date": 0.125, "amount": 0.6153846153846154, "status": 0.75, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1087}, "2": {"facts_total": 63, "facts_retained": 27, "retention_rate": 0.42857142857142855, "retention_by_kind": {"person": 0.5, "date": 0.125, "amount": 0.6153846153846154, "status": 0.75, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1087}, "3": {"facts_total": 63, "facts_retained": 27, "retention_rate": 0.42857142857142855, "retention_by_kind": {"person": 0.5, "date": 0.125, "amount": 0.6153846153846154, "status": 0.75, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1087}, "4": {"facts_total": 63, "facts_retained": 27, "retention_rate": 0.42857142857142855, "retention_by_kind": {"person": 0.5, "date": 0.125, "amount": 0.6153846153846154, "status": 0.75, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1087}, "5": {"facts_total": 63, "facts_retained": 27, "retention_rate": 0.42857142857142855, "retention_by_kind": {"person": 0.5, "date": 0.125, "amount": 0.6153846153846154, "status": 0.75, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1087}, "6": {"facts_total": 63, "facts_retained": 27, "retention_rate": 0.42857142857142855, "retention_by_kind": {"person": 0.5, "date": 0.125, "amount": 0.6153846153846154, "status": 0.75, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1087}, "7": {"facts_total": 63, "facts_retained": 27, "retention_rate": 0.42857142857142855, "retention_by_kind": {"person": 0.5, "date": 0.125, "amount": 0.6153846153846154, "status": 0.75, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1087}, "8": {"facts_total": 63, "facts_retained": 27, "retention_rate": 0.42857142857142855, "retention_by_kind": {"person": 0.5, "date": 0.125, "amount": 0.6153846153846154, "status": 0.75, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1087}}}, "reference_key_id": "keys/synthetic/syn_03027_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "1f88f0e78c63504e418dea843641b99a98555e15d47b02cd5cb2f57da3003dea", "output_hash": "d13adba73068c847c850237059e0202fc5b91ccacf2e792e9ce5e3b90556501a"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "7258afe9605849d00393b6f2", "attempt_id": "29885195cd24", "run_group_id": "E4_core:A4|closed|qwen3b:h8", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:29:44.429461Z", "task_family": "synthetic", "document_id": "syn_03027_2000", "document_hash": "10326080979f71b625e4110c7c19767e5a752e267de5034f7e37d98cb6e06249", "question_id": "q03_exact_copy", "split": "test", "cluster_id": "syn_03027_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 8, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1955, "query_tokens": 595, "message_tokens": 512, "message_utf8_bytes": 1087, "cache_tensor_bytes": null, "serialized_payload_bytes": 1087, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 51405.68041824736, "hops": {"1": 6896.277834079228, "2": 6372.689875075594, "3": 6377.503250027075, "4": 6359.274166985415, "5": 6348.349208012223, "6": 6350.234291981906, "7": 6348.7960420316085, "8": 6352.555750054307}, "receiver_prefill_ms": 227.08812507335097, "ttft_ms": 11.908416985534132, "decode_ms": 106.7351249512285}, "total_wall_ms": 51751.89341825899, "ttft_ms": 11.908416985534132, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 9}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "Priyamvada Sørlie", "evidence_artifact": {"ledger": {"facts_total": 63, "facts_retained": 27, "retention_rate": 0.42857142857142855, "retention_by_kind": {"person": 0.5, "date": 0.125, "amount": 0.6153846153846154, "status": 0.75, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1087}, "ledger_by_hop": {"1": {"facts_total": 63, "facts_retained": 27, "retention_rate": 0.42857142857142855, "retention_by_kind": {"person": 0.5, "date": 0.125, "amount": 0.6153846153846154, "status": 0.75, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1087}, "2": {"facts_total": 63, "facts_retained": 27, "retention_rate": 0.42857142857142855, "retention_by_kind": {"person": 0.5, "date": 0.125, "amount": 0.6153846153846154, "status": 0.75, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1087}, "3": {"facts_total": 63, "facts_retained": 27, "retention_rate": 0.42857142857142855, "retention_by_kind": {"person": 0.5, "date": 0.125, "amount": 0.6153846153846154, "status": 0.75, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1087}, "4": {"facts_total": 63, "facts_retained": 27, "retention_rate": 0.42857142857142855, "retention_by_kind": {"person": 0.5, "date": 0.125, "amount": 0.6153846153846154, "status": 0.75, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1087}, "5": {"facts_total": 63, "facts_retained": 27, "retention_rate": 0.42857142857142855, "retention_by_kind": {"person": 0.5, "date": 0.125, "amount": 0.6153846153846154, "status": 0.75, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1087}, "6": {"facts_total": 63, "facts_retained": 27, "retention_rate": 0.42857142857142855, "retention_by_kind": {"person": 0.5, "date": 0.125, "amount": 0.6153846153846154, "status": 0.75, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1087}, "7": {"facts_total": 63, "facts_retained": 27, "retention_rate": 0.42857142857142855, "retention_by_kind": {"person": 0.5, "date": 0.125, "amount": 0.6153846153846154, "status": 0.75, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1087}, "8": {"facts_total": 63, "facts_retained": 27, "retention_rate": 0.42857142857142855, "retention_by_kind": {"person": 0.5, "date": 0.125, "amount": 0.6153846153846154, "status": 0.75, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1087}}}, "reference_key_id": "keys/synthetic/syn_03027_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "1f88f0e78c63504e418dea843641b99a98555e15d47b02cd5cb2f57da3003dea", "output_hash": "c5f554196ed23dd71ad995ec89f7142099d524dbb6534cc10d94a0c89d13e792"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "6ed83324d3a81cb598bdfe59", "attempt_id": "17cdac03bc1e", "run_group_id": "E4_core:A4|closed|qwen3b:h8", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:29:44.790918Z", "task_family": "synthetic", "document_id": "syn_03027_2000", "document_hash": "10326080979f71b625e4110c7c19767e5a752e267de5034f7e37d98cb6e06249", "question_id": "q04_multi_hop", "split": "test", "cluster_id": "syn_03027_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 8, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1955, "query_tokens": 606, "message_tokens": 512, "message_utf8_bytes": 1087, "cache_tensor_bytes": null, "serialized_payload_bytes": 1087, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 51405.68041824736, "hops": {"1": 6896.277834079228, "2": 6372.689875075594, "3": 6377.503250027075, "4": 6359.274166985415, "5": 6348.349208012223, "6": 6350.234291981906, "7": 6348.7960420316085, "8": 6352.555750054307}, "receiver_prefill_ms": 227.51779097598046, "ttft_ms": 11.86512503772974, "decode_ms": 11.907125008292496}, "total_wall_ms": 51657.44196018204, "ttft_ms": 11.86512503772974, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 63, "facts_retained": 27, "retention_rate": 0.42857142857142855, "retention_by_kind": {"person": 0.5, "date": 0.125, "amount": 0.6153846153846154, "status": 0.75, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1087}, "ledger_by_hop": {"1": {"facts_total": 63, "facts_retained": 27, "retention_rate": 0.42857142857142855, "retention_by_kind": {"person": 0.5, "date": 0.125, "amount": 0.6153846153846154, "status": 0.75, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1087}, "2": {"facts_total": 63, "facts_retained": 27, "retention_rate": 0.42857142857142855, "retention_by_kind": {"person": 0.5, "date": 0.125, "amount": 0.6153846153846154, "status": 0.75, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1087}, "3": {"facts_total": 63, "facts_retained": 27, "retention_rate": 0.42857142857142855, "retention_by_kind": {"person": 0.5, "date": 0.125, "amount": 0.6153846153846154, "status": 0.75, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1087}, "4": {"facts_total": 63, "facts_retained": 27, "retention_rate": 0.42857142857142855, "retention_by_kind": {"person": 0.5, "date": 0.125, "amount": 0.6153846153846154, "status": 0.75, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1087}, "5": {"facts_total": 63, "facts_retained": 27, "retention_rate": 0.42857142857142855, "retention_by_kind": {"person": 0.5, "date": 0.125, "amount": 0.6153846153846154, "status": 0.75, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1087}, "6": {"facts_total": 63, "facts_retained": 27, "retention_rate": 0.42857142857142855, "retention_by_kind": {"person": 0.5, "date": 0.125, "amount": 0.6153846153846154, "status": 0.75, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1087}, "7": {"facts_total": 63, "facts_retained": 27, "retention_rate": 0.42857142857142855, "retention_by_kind": {"person": 0.5, "date": 0.125, "amount": 0.6153846153846154, "status": 0.75, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1087}, "8": {"facts_total": 63, "facts_retained": 27, "retention_rate": 0.42857142857142855, "retention_by_kind": {"person": 0.5, "date": 0.125, "amount": 0.6153846153846154, "status": 0.75, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1087}}}, "reference_key_id": "keys/synthetic/syn_03027_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "1f88f0e78c63504e418dea843641b99a98555e15d47b02cd5cb2f57da3003dea", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "be09b3aa2cf0ce21e167c50f", "attempt_id": "5eda559b81c8", "run_group_id": "E4_core:A4|closed|qwen3b:h8", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:29:45.057736Z", "task_family": "synthetic", "document_id": "syn_03027_2000", "document_hash": "10326080979f71b625e4110c7c19767e5a752e267de5034f7e37d98cb6e06249", "question_id": "q05_multi_hop", "split": "test", "cluster_id": "syn_03027_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 8, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1955, "query_tokens": 603, "message_tokens": 512, "message_utf8_bytes": 1087, "cache_tensor_bytes": null, "serialized_payload_bytes": 1087, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 51405.68041824736, "hops": {"1": 6896.277834079228, "2": 6372.689875075594, "3": 6377.503250027075, "4": 6359.274166985415, "5": 6348.349208012223, "6": 6350.234291981906, "7": 6348.7960420316085, "8": 6352.555750054307}, "receiver_prefill_ms": 227.63350000604987, "ttft_ms": 11.872583068907261, "decode_ms": 11.936583905480802}, "total_wall_ms": 51657.599751255475, "ttft_ms": 11.872583068907261, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 63, "facts_retained": 27, "retention_rate": 0.42857142857142855, "retention_by_kind": {"person": 0.5, "date": 0.125, "amount": 0.6153846153846154, "status": 0.75, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1087}, "ledger_by_hop": {"1": {"facts_total": 63, "facts_retained": 27, "retention_rate": 0.42857142857142855, "retention_by_kind": {"person": 0.5, "date": 0.125, "amount": 0.6153846153846154, "status": 0.75, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1087}, "2": {"facts_total": 63, "facts_retained": 27, "retention_rate": 0.42857142857142855, "retention_by_kind": {"person": 0.5, "date": 0.125, "amount": 0.6153846153846154, "status": 0.75, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1087}, "3": {"facts_total": 63, "facts_retained": 27, "retention_rate": 0.42857142857142855, "retention_by_kind": {"person": 0.5, "date": 0.125, "amount": 0.6153846153846154, "status": 0.75, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1087}, "4": {"facts_total": 63, "facts_retained": 27, "retention_rate": 0.42857142857142855, "retention_by_kind": {"person": 0.5, "date": 0.125, "amount": 0.6153846153846154, "status": 0.75, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1087}, "5": {"facts_total": 63, "facts_retained": 27, "retention_rate": 0.42857142857142855, "retention_by_kind": {"person": 0.5, "date": 0.125, "amount": 0.6153846153846154, "status": 0.75, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1087}, "6": {"facts_total": 63, "facts_retained": 27, "retention_rate": 0.42857142857142855, "retention_by_kind": {"person": 0.5, "date": 0.125, "amount": 0.6153846153846154, "status": 0.75, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1087}, "7": {"facts_total": 63, "facts_retained": 27, "retention_rate": 0.42857142857142855, "retention_by_kind": {"person": 0.5, "date": 0.125, "amount": 0.6153846153846154, "status": 0.75, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1087}, "8": {"facts_total": 63, "facts_retained": 27, "retention_rate": 0.42857142857142855, "retention_by_kind": {"person": 0.5, "date": 0.125, "amount": 0.6153846153846154, "status": 0.75, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1087}}}, "reference_key_id": "keys/synthetic/syn_03027_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "1f88f0e78c63504e418dea843641b99a98555e15d47b02cd5cb2f57da3003dea", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "1ff9d580e18b146759c5708a", "attempt_id": "808c256cbac4", "run_group_id": "E4_core:A4|closed|qwen3b:h8", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:29:45.324773Z", "task_family": "synthetic", "document_id": "syn_03027_2000", "document_hash": "10326080979f71b625e4110c7c19767e5a752e267de5034f7e37d98cb6e06249", "question_id": "q06_negation", "split": "test", "cluster_id": "syn_03027_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 8, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1955, "query_tokens": 595, "message_tokens": 512, "message_utf8_bytes": 1087, "cache_tensor_bytes": null, "serialized_payload_bytes": 1087, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 51405.68041824736, "hops": {"1": 6896.277834079228, "2": 6372.689875075594, "3": 6377.503250027075, "4": 6359.274166985415, "5": 6348.349208012223, "6": 6350.234291981906, "7": 6348.7960420316085, "8": 6352.555750054307}, "receiver_prefill_ms": 226.95679205935448, "ttft_ms": 11.836083023808897, "decode_ms": 11.92099996842444}, "total_wall_ms": 51656.8319182843, "ttft_ms": 11.836083023808897, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 63, "facts_retained": 27, "retention_rate": 0.42857142857142855, "retention_by_kind": {"person": 0.5, "date": 0.125, "amount": 0.6153846153846154, "status": 0.75, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1087}, "ledger_by_hop": {"1": {"facts_total": 63, "facts_retained": 27, "retention_rate": 0.42857142857142855, "retention_by_kind": {"person": 0.5, "date": 0.125, "amount": 0.6153846153846154, "status": 0.75, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1087}, "2": {"facts_total": 63, "facts_retained": 27, "retention_rate": 0.42857142857142855, "retention_by_kind": {"person": 0.5, "date": 0.125, "amount": 0.6153846153846154, "status": 0.75, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1087}, "3": {"facts_total": 63, "facts_retained": 27, "retention_rate": 0.42857142857142855, "retention_by_kind": {"person": 0.5, "date": 0.125, "amount": 0.6153846153846154, "status": 0.75, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1087}, "4": {"facts_total": 63, "facts_retained": 27, "retention_rate": 0.42857142857142855, "retention_by_kind": {"person": 0.5, "date": 0.125, "amount": 0.6153846153846154, "status": 0.75, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1087}, "5": {"facts_total": 63, "facts_retained": 27, "retention_rate": 0.42857142857142855, "retention_by_kind": {"person": 0.5, "date": 0.125, "amount": 0.6153846153846154, "status": 0.75, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1087}, "6": {"facts_total": 63, "facts_retained": 27, "retention_rate": 0.42857142857142855, "retention_by_kind": {"person": 0.5, "date": 0.125, "amount": 0.6153846153846154, "status": 0.75, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1087}, "7": {"facts_total": 63, "facts_retained": 27, "retention_rate": 0.42857142857142855, "retention_by_kind": {"person": 0.5, "date": 0.125, "amount": 0.6153846153846154, "status": 0.75, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1087}, "8": {"facts_total": 63, "facts_retained": 27, "retention_rate": 0.42857142857142855, "retention_by_kind": {"person": 0.5, "date": 0.125, "amount": 0.6153846153846154, "status": 0.75, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1087}}}, "reference_key_id": "keys/synthetic/syn_03027_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "1f88f0e78c63504e418dea843641b99a98555e15d47b02cd5cb2f57da3003dea", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "5f0340a743fad487b3bf3c62", "attempt_id": "2423134f87ea", "run_group_id": "E4_core:A4|closed|qwen3b:h8", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:29:45.590986Z", "task_family": "synthetic", "document_id": "syn_03027_2000", "document_hash": "10326080979f71b625e4110c7c19767e5a752e267de5034f7e37d98cb6e06249", "question_id": "q07_negation", "split": "test", "cluster_id": "syn_03027_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 8, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1955, "query_tokens": 596, "message_tokens": 512, "message_utf8_bytes": 1087, "cache_tensor_bytes": null, "serialized_payload_bytes": 1087, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 51405.68041824736, "hops": {"1": 6896.277834079228, "2": 6372.689875075594, "3": 6377.503250027075, "4": 6359.274166985415, "5": 6348.349208012223, "6": 6350.234291981906, "7": 6348.7960420316085, "8": 6352.555750054307}, "receiver_prefill_ms": 227.29758301284164, "ttft_ms": 11.906791012734175, "decode_ms": 11.86133292503655}, "total_wall_ms": 51657.20420924481, "ttft_ms": 11.906791012734175, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "no", "evidence_artifact": {"ledger": {"facts_total": 63, "facts_retained": 27, "retention_rate": 0.42857142857142855, "retention_by_kind": {"person": 0.5, "date": 0.125, "amount": 0.6153846153846154, "status": 0.75, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1087}, "ledger_by_hop": {"1": {"facts_total": 63, "facts_retained": 27, "retention_rate": 0.42857142857142855, "retention_by_kind": {"person": 0.5, "date": 0.125, "amount": 0.6153846153846154, "status": 0.75, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1087}, "2": {"facts_total": 63, "facts_retained": 27, "retention_rate": 0.42857142857142855, "retention_by_kind": {"person": 0.5, "date": 0.125, "amount": 0.6153846153846154, "status": 0.75, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1087}, "3": {"facts_total": 63, "facts_retained": 27, "retention_rate": 0.42857142857142855, "retention_by_kind": {"person": 0.5, "date": 0.125, "amount": 0.6153846153846154, "status": 0.75, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1087}, "4": {"facts_total": 63, "facts_retained": 27, "retention_rate": 0.42857142857142855, "retention_by_kind": {"person": 0.5, "date": 0.125, "amount": 0.6153846153846154, "status": 0.75, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1087}, "5": {"facts_total": 63, "facts_retained": 27, "retention_rate": 0.42857142857142855, "retention_by_kind": {"person": 0.5, "date": 0.125, "amount": 0.6153846153846154, "status": 0.75, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1087}, "6": {"facts_total": 63, "facts_retained": 27, "retention_rate": 0.42857142857142855, "retention_by_kind": {"person": 0.5, "date": 0.125, "amount": 0.6153846153846154, "status": 0.75, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1087}, "7": {"facts_total": 63, "facts_retained": 27, "retention_rate": 0.42857142857142855, "retention_by_kind": {"person": 0.5, "date": 0.125, "amount": 0.6153846153846154, "status": 0.75, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1087}, "8": {"facts_total": 63, "facts_retained": 27, "retention_rate": 0.42857142857142855, "retention_by_kind": {"person": 0.5, "date": 0.125, "amount": 0.6153846153846154, "status": 0.75, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1087}}}, "reference_key_id": "keys/synthetic/syn_03027_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "1f88f0e78c63504e418dea843641b99a98555e15d47b02cd5cb2f57da3003dea", "output_hash": "9390298f3fb0c5b160498935d79cb139aef28e1c47358b4bbba61862b9c26e59"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "1cb6280d20f6891b6924da71", "attempt_id": "ab8085c940ca", "run_group_id": "E4_core:A4|closed|qwen3b:h8", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:29:45.857087Z", "task_family": "synthetic", "document_id": "syn_03027_2000", "document_hash": "10326080979f71b625e4110c7c19767e5a752e267de5034f7e37d98cb6e06249", "question_id": "q08_revision", "split": "test", "cluster_id": "syn_03027_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 8, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1955, "query_tokens": 610, "message_tokens": 512, "message_utf8_bytes": 1087, "cache_tensor_bytes": null, "serialized_payload_bytes": 1087, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 51405.68041824736, "hops": {"1": 6896.277834079228, "2": 6372.689875075594, "3": 6377.503250027075, "4": 6359.274166985415, "5": 6348.349208012223, "6": 6350.234291981906, "7": 6348.7960420316085, "8": 6352.555750054307}, "receiver_prefill_ms": 229.32608297560364, "ttft_ms": 11.926500010304153, "decode_ms": 82.94041699264199}, "total_wall_ms": 51730.34229315817, "ttft_ms": 11.926500010304153, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 7}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "8010205", "evidence_artifact": {"ledger": {"facts_total": 63, "facts_retained": 27, "retention_rate": 0.42857142857142855, "retention_by_kind": {"person": 0.5, "date": 0.125, "amount": 0.6153846153846154, "status": 0.75, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1087}, "ledger_by_hop": {"1": {"facts_total": 63, "facts_retained": 27, "retention_rate": 0.42857142857142855, "retention_by_kind": {"person": 0.5, "date": 0.125, "amount": 0.6153846153846154, "status": 0.75, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1087}, "2": {"facts_total": 63, "facts_retained": 27, "retention_rate": 0.42857142857142855, "retention_by_kind": {"person": 0.5, "date": 0.125, "amount": 0.6153846153846154, "status": 0.75, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1087}, "3": {"facts_total": 63, "facts_retained": 27, "retention_rate": 0.42857142857142855, "retention_by_kind": {"person": 0.5, "date": 0.125, "amount": 0.6153846153846154, "status": 0.75, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1087}, "4": {"facts_total": 63, "facts_retained": 27, "retention_rate": 0.42857142857142855, "retention_by_kind": {"person": 0.5, "date": 0.125, "amount": 0.6153846153846154, "status": 0.75, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1087}, "5": {"facts_total": 63, "facts_retained": 27, "retention_rate": 0.42857142857142855, "retention_by_kind": {"person": 0.5, "date": 0.125, "amount": 0.6153846153846154, "status": 0.75, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1087}, "6": {"facts_total": 63, "facts_retained": 27, "retention_rate": 0.42857142857142855, "retention_by_kind": {"person": 0.5, "date": 0.125, "amount": 0.6153846153846154, "status": 0.75, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1087}, "7": {"facts_total": 63, "facts_retained": 27, "retention_rate": 0.42857142857142855, "retention_by_kind": {"person": 0.5, "date": 0.125, "amount": 0.6153846153846154, "status": 0.75, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1087}, "8": {"facts_total": 63, "facts_retained": 27, "retention_rate": 0.42857142857142855, "retention_by_kind": {"person": 0.5, "date": 0.125, "amount": 0.6153846153846154, "status": 0.75, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1087}}}, "reference_key_id": "keys/synthetic/syn_03027_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "1f88f0e78c63504e418dea843641b99a98555e15d47b02cd5cb2f57da3003dea", "output_hash": "6d1ceb36132f38bd77e8f46c23ea21dd9bcff78bc80bd8a9eaf2d7f2f3bacf46"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "b27be43fe49ed2b2f7c41115", "attempt_id": "dfd418fb5e0a", "run_group_id": "E4_core:A4|closed|qwen3b:h8", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:29:46.196742Z", "task_family": "synthetic", "document_id": "syn_03027_2000", "document_hash": "10326080979f71b625e4110c7c19767e5a752e267de5034f7e37d98cb6e06249", "question_id": "q09_revision", "split": "test", "cluster_id": "syn_03027_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 8, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1955, "query_tokens": 603, "message_tokens": 512, "message_utf8_bytes": 1087, "cache_tensor_bytes": null, "serialized_payload_bytes": 1087, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 51405.68041824736, "hops": {"1": 6896.277834079228, "2": 6372.689875075594, "3": 6377.503250027075, "4": 6359.274166985415, "5": 6348.349208012223, "6": 6350.234291981906, "7": 6348.7960420316085, "8": 6352.555750054307}, "receiver_prefill_ms": 227.38179203588516, "ttft_ms": 12.031542020849884, "decode_ms": 82.88679202087224}, "total_wall_ms": 51728.450918220915, "ttft_ms": 12.031542020849884, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 7}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "6056150", "evidence_artifact": {"ledger": {"facts_total": 63, "facts_retained": 27, "retention_rate": 0.42857142857142855, "retention_by_kind": {"person": 0.5, "date": 0.125, "amount": 0.6153846153846154, "status": 0.75, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1087}, "ledger_by_hop": {"1": {"facts_total": 63, "facts_retained": 27, "retention_rate": 0.42857142857142855, "retention_by_kind": {"person": 0.5, "date": 0.125, "amount": 0.6153846153846154, "status": 0.75, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1087}, "2": {"facts_total": 63, "facts_retained": 27, "retention_rate": 0.42857142857142855, "retention_by_kind": {"person": 0.5, "date": 0.125, "amount": 0.6153846153846154, "status": 0.75, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1087}, "3": {"facts_total": 63, "facts_retained": 27, "retention_rate": 0.42857142857142855, "retention_by_kind": {"person": 0.5, "date": 0.125, "amount": 0.6153846153846154, "status": 0.75, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1087}, "4": {"facts_total": 63, "facts_retained": 27, "retention_rate": 0.42857142857142855, "retention_by_kind": {"person": 0.5, "date": 0.125, "amount": 0.6153846153846154, "status": 0.75, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1087}, "5": {"facts_total": 63, "facts_retained": 27, "retention_rate": 0.42857142857142855, "retention_by_kind": {"person": 0.5, "date": 0.125, "amount": 0.6153846153846154, "status": 0.75, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1087}, "6": {"facts_total": 63, "facts_retained": 27, "retention_rate": 0.42857142857142855, "retention_by_kind": {"person": 0.5, "date": 0.125, "amount": 0.6153846153846154, "status": 0.75, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1087}, "7": {"facts_total": 63, "facts_retained": 27, "retention_rate": 0.42857142857142855, "retention_by_kind": {"person": 0.5, "date": 0.125, "amount": 0.6153846153846154, "status": 0.75, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1087}, "8": {"facts_total": 63, "facts_retained": 27, "retention_rate": 0.42857142857142855, "retention_by_kind": {"person": 0.5, "date": 0.125, "amount": 0.6153846153846154, "status": 0.75, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1087}}}, "reference_key_id": "keys/synthetic/syn_03027_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "1f88f0e78c63504e418dea843641b99a98555e15d47b02cd5cb2f57da3003dea", "output_hash": "f016f8b5fc26708e6c1ff75ff8e98659082c41a168f273819ac48b88feea8d3f"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "57775a8b8c086ba5469d6edc", "attempt_id": "a7a503e3100a", "run_group_id": "E4_core:A4|closed|qwen3b:h8", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:29:46.534662Z", "task_family": "synthetic", "document_id": "syn_03027_2000", "document_hash": "10326080979f71b625e4110c7c19767e5a752e267de5034f7e37d98cb6e06249", "question_id": "q10_temporal", "split": "test", "cluster_id": "syn_03027_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 8, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1955, "query_tokens": 613, "message_tokens": 512, "message_utf8_bytes": 1087, "cache_tensor_bytes": null, "serialized_payload_bytes": 1087, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 51405.68041824736, "hops": {"1": 6896.277834079228, "2": 6372.689875075594, "3": 6377.503250027075, "4": 6359.274166985415, "5": 6348.349208012223, "6": 6350.234291981906, "7": 6348.7960420316085, "8": 6352.555750054307}, "receiver_prefill_ms": 229.35337503440678, "ttft_ms": 11.958708055317402, "decode_ms": 83.04466598201543}, "total_wall_ms": 51730.515543255024, "ttft_ms": 11.958708055317402, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 7}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "Lunlith-13", "evidence_artifact": {"ledger": {"facts_total": 63, "facts_retained": 27, "retention_rate": 0.42857142857142855, "retention_by_kind": {"person": 0.5, "date": 0.125, "amount": 0.6153846153846154, "status": 0.75, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1087}, "ledger_by_hop": {"1": {"facts_total": 63, "facts_retained": 27, "retention_rate": 0.42857142857142855, "retention_by_kind": {"person": 0.5, "date": 0.125, "amount": 0.6153846153846154, "status": 0.75, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1087}, "2": {"facts_total": 63, "facts_retained": 27, "retention_rate": 0.42857142857142855, "retention_by_kind": {"person": 0.5, "date": 0.125, "amount": 0.6153846153846154, "status": 0.75, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1087}, "3": {"facts_total": 63, "facts_retained": 27, "retention_rate": 0.42857142857142855, "retention_by_kind": {"person": 0.5, "date": 0.125, "amount": 0.6153846153846154, "status": 0.75, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1087}, "4": {"facts_total": 63, "facts_retained": 27, "retention_rate": 0.42857142857142855, "retention_by_kind": {"person": 0.5, "date": 0.125, "amount": 0.6153846153846154, "status": 0.75, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1087}, "5": {"facts_total": 63, "facts_retained": 27, "retention_rate": 0.42857142857142855, "retention_by_kind": {"person": 0.5, "date": 0.125, "amount": 0.6153846153846154, "status": 0.75, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1087}, "6": {"facts_total": 63, "facts_retained": 27, "retention_rate": 0.42857142857142855, "retention_by_kind": {"person": 0.5, "date": 0.125, "amount": 0.6153846153846154, "status": 0.75, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1087}, "7": {"facts_total": 63, "facts_retained": 27, "retention_rate": 0.42857142857142855, "retention_by_kind": {"person": 0.5, "date": 0.125, "amount": 0.6153846153846154, "status": 0.75, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1087}, "8": {"facts_total": 63, "facts_retained": 27, "retention_rate": 0.42857142857142855, "retention_by_kind": {"person": 0.5, "date": 0.125, "amount": 0.6153846153846154, "status": 0.75, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1087}}}, "reference_key_id": "keys/synthetic/syn_03027_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "1f88f0e78c63504e418dea843641b99a98555e15d47b02cd5cb2f57da3003dea", "output_hash": "c45f000afe6d865c8167954eebdff761141225817d4f5b26c962cd50cd75d391"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "07f9e2f958596f66b857e98e", "attempt_id": "0f41a48e3b3f", "run_group_id": "E4_core:A4|closed|qwen3b:h8", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:29:46.874530Z", "task_family": "synthetic", "document_id": "syn_03027_2000", "document_hash": "10326080979f71b625e4110c7c19767e5a752e267de5034f7e37d98cb6e06249", "question_id": "q11_aggregation", "split": "test", "cluster_id": "syn_03027_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 8, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1955, "query_tokens": 606, "message_tokens": 512, "message_utf8_bytes": 1087, "cache_tensor_bytes": null, "serialized_payload_bytes": 1087, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 51405.68041824736, "hops": {"1": 6896.277834079228, "2": 6372.689875075594, "3": 6377.503250027075, "4": 6359.274166985415, "5": 6348.349208012223, "6": 6350.234291981906, "7": 6348.7960420316085, "8": 6352.555750054307}, "receiver_prefill_ms": 226.87454207334667, "ttft_ms": 11.904707993380725, "decode_ms": 119.04387490358204}, "total_wall_ms": 51763.96646024659, "ttft_ms": 11.904707993380725, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 10}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "23,450,777", "evidence_artifact": {"ledger": {"facts_total": 63, "facts_retained": 27, "retention_rate": 0.42857142857142855, "retention_by_kind": {"person": 0.5, "date": 0.125, "amount": 0.6153846153846154, "status": 0.75, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1087}, "ledger_by_hop": {"1": {"facts_total": 63, "facts_retained": 27, "retention_rate": 0.42857142857142855, "retention_by_kind": {"person": 0.5, "date": 0.125, "amount": 0.6153846153846154, "status": 0.75, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1087}, "2": {"facts_total": 63, "facts_retained": 27, "retention_rate": 0.42857142857142855, "retention_by_kind": {"person": 0.5, "date": 0.125, "amount": 0.6153846153846154, "status": 0.75, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1087}, "3": {"facts_total": 63, "facts_retained": 27, "retention_rate": 0.42857142857142855, "retention_by_kind": {"person": 0.5, "date": 0.125, "amount": 0.6153846153846154, "status": 0.75, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1087}, "4": {"facts_total": 63, "facts_retained": 27, "retention_rate": 0.42857142857142855, "retention_by_kind": {"person": 0.5, "date": 0.125, "amount": 0.6153846153846154, "status": 0.75, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1087}, "5": {"facts_total": 63, "facts_retained": 27, "retention_rate": 0.42857142857142855, "retention_by_kind": {"person": 0.5, "date": 0.125, "amount": 0.6153846153846154, "status": 0.75, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1087}, "6": {"facts_total": 63, "facts_retained": 27, "retention_rate": 0.42857142857142855, "retention_by_kind": {"person": 0.5, "date": 0.125, "amount": 0.6153846153846154, "status": 0.75, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1087}, "7": {"facts_total": 63, "facts_retained": 27, "retention_rate": 0.42857142857142855, "retention_by_kind": {"person": 0.5, "date": 0.125, "amount": 0.6153846153846154, "status": 0.75, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1087}, "8": {"facts_total": 63, "facts_retained": 27, "retention_rate": 0.42857142857142855, "retention_by_kind": {"person": 0.5, "date": 0.125, "amount": 0.6153846153846154, "status": 0.75, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1087}}}, "reference_key_id": "keys/synthetic/syn_03027_2000.json", "task_scores": {"correct": 0.0, "abstained": 0.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "1f88f0e78c63504e418dea843641b99a98555e15d47b02cd5cb2f57da3003dea", "output_hash": "3894c3c39e5c76c297dfb1859a5e1724e2454039e072357510c841bc907d01a5"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "edc488dc9231b9ba0fcc4ff0", "attempt_id": "ddc061e2b281", "run_group_id": "E4_core:A4|closed|qwen3b:h8", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:29:47.247702Z", "task_family": "synthetic", "document_id": "syn_03027_2000", "document_hash": "10326080979f71b625e4110c7c19767e5a752e267de5034f7e37d98cb6e06249", "question_id": "q12_unit_conversion", "split": "test", "cluster_id": "syn_03027_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 8, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1955, "query_tokens": 603, "message_tokens": 512, "message_utf8_bytes": 1087, "cache_tensor_bytes": null, "serialized_payload_bytes": 1087, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 51405.68041824736, "hops": {"1": 6896.277834079228, "2": 6372.689875075594, "3": 6377.503250027075, "4": 6359.274166985415, "5": 6348.349208012223, "6": 6350.234291981906, "7": 6348.7960420316085, "8": 6352.555750054307}, "receiver_prefill_ms": 227.00095898471773, "ttft_ms": 11.933500063605607, "decode_ms": 11.876834090799093}, "total_wall_ms": 51656.951210228726, "ttft_ms": 11.933500063605607, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 63, "facts_retained": 27, "retention_rate": 0.42857142857142855, "retention_by_kind": {"person": 0.5, "date": 0.125, "amount": 0.6153846153846154, "status": 0.75, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1087}, "ledger_by_hop": {"1": {"facts_total": 63, "facts_retained": 27, "retention_rate": 0.42857142857142855, "retention_by_kind": {"person": 0.5, "date": 0.125, "amount": 0.6153846153846154, "status": 0.75, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1087}, "2": {"facts_total": 63, "facts_retained": 27, "retention_rate": 0.42857142857142855, "retention_by_kind": {"person": 0.5, "date": 0.125, "amount": 0.6153846153846154, "status": 0.75, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1087}, "3": {"facts_total": 63, "facts_retained": 27, "retention_rate": 0.42857142857142855, "retention_by_kind": {"person": 0.5, "date": 0.125, "amount": 0.6153846153846154, "status": 0.75, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1087}, "4": {"facts_total": 63, "facts_retained": 27, "retention_rate": 0.42857142857142855, "retention_by_kind": {"person": 0.5, "date": 0.125, "amount": 0.6153846153846154, "status": 0.75, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1087}, "5": {"facts_total": 63, "facts_retained": 27, "retention_rate": 0.42857142857142855, "retention_by_kind": {"person": 0.5, "date": 0.125, "amount": 0.6153846153846154, "status": 0.75, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1087}, "6": {"facts_total": 63, "facts_retained": 27, "retention_rate": 0.42857142857142855, "retention_by_kind": {"person": 0.5, "date": 0.125, "amount": 0.6153846153846154, "status": 0.75, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1087}, "7": {"facts_total": 63, "facts_retained": 27, "retention_rate": 0.42857142857142855, "retention_by_kind": {"person": 0.5, "date": 0.125, "amount": 0.6153846153846154, "status": 0.75, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1087}, "8": {"facts_total": 63, "facts_retained": 27, "retention_rate": 0.42857142857142855, "retention_by_kind": {"person": 0.5, "date": 0.125, "amount": 0.6153846153846154, "status": 0.75, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1087}}}, "reference_key_id": "keys/synthetic/syn_03027_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "1f88f0e78c63504e418dea843641b99a98555e15d47b02cd5cb2f57da3003dea", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "7fb0d639e291adf520c80ab8", "attempt_id": "a2f2e9a8fbfe", "run_group_id": "E4_core:A4|closed|qwen3b:h8", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:29:47.514050Z", "task_family": "synthetic", "document_id": "syn_03027_2000", "document_hash": "10326080979f71b625e4110c7c19767e5a752e267de5034f7e37d98cb6e06249", "question_id": "q13_unanswerable", "split": "test", "cluster_id": "syn_03027_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 8, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1955, "query_tokens": 606, "message_tokens": 512, "message_utf8_bytes": 1087, "cache_tensor_bytes": null, "serialized_payload_bytes": 1087, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 51405.68041824736, "hops": {"1": 6896.277834079228, "2": 6372.689875075594, "3": 6377.503250027075, "4": 6359.274166985415, "5": 6348.349208012223, "6": 6350.234291981906, "7": 6348.7960420316085, "8": 6352.555750054307}, "receiver_prefill_ms": 227.1774170221761, "ttft_ms": 11.896083015017211, "decode_ms": 11.803666944615543}, "total_wall_ms": 51657.00679330621, "ttft_ms": 11.896083015017211, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 63, "facts_retained": 27, "retention_rate": 0.42857142857142855, "retention_by_kind": {"person": 0.5, "date": 0.125, "amount": 0.6153846153846154, "status": 0.75, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1087}, "ledger_by_hop": {"1": {"facts_total": 63, "facts_retained": 27, "retention_rate": 0.42857142857142855, "retention_by_kind": {"person": 0.5, "date": 0.125, "amount": 0.6153846153846154, "status": 0.75, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1087}, "2": {"facts_total": 63, "facts_retained": 27, "retention_rate": 0.42857142857142855, "retention_by_kind": {"person": 0.5, "date": 0.125, "amount": 0.6153846153846154, "status": 0.75, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1087}, "3": {"facts_total": 63, "facts_retained": 27, "retention_rate": 0.42857142857142855, "retention_by_kind": {"person": 0.5, "date": 0.125, "amount": 0.6153846153846154, "status": 0.75, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1087}, "4": {"facts_total": 63, "facts_retained": 27, "retention_rate": 0.42857142857142855, "retention_by_kind": {"person": 0.5, "date": 0.125, "amount": 0.6153846153846154, "status": 0.75, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1087}, "5": {"facts_total": 63, "facts_retained": 27, "retention_rate": 0.42857142857142855, "retention_by_kind": {"person": 0.5, "date": 0.125, "amount": 0.6153846153846154, "status": 0.75, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1087}, "6": {"facts_total": 63, "facts_retained": 27, "retention_rate": 0.42857142857142855, "retention_by_kind": {"person": 0.5, "date": 0.125, "amount": 0.6153846153846154, "status": 0.75, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1087}, "7": {"facts_total": 63, "facts_retained": 27, "retention_rate": 0.42857142857142855, "retention_by_kind": {"person": 0.5, "date": 0.125, "amount": 0.6153846153846154, "status": 0.75, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1087}, "8": {"facts_total": 63, "facts_retained": 27, "retention_rate": 0.42857142857142855, "retention_by_kind": {"person": 0.5, "date": 0.125, "amount": 0.6153846153846154, "status": 0.75, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1087}}}, "reference_key_id": "keys/synthetic/syn_03027_2000.json", "task_scores": {"correct": 1.0, "abstained": 1.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "1f88f0e78c63504e418dea843641b99a98555e15d47b02cd5cb2f57da3003dea", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "78503e57a1b0783db6da36c3", "attempt_id": "2dd8e8d64a86", "run_group_id": "E4_core:A4|closed|qwen3b:h8", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:29:47.780586Z", "task_family": "synthetic", "document_id": "syn_03027_2000", "document_hash": "10326080979f71b625e4110c7c19767e5a752e267de5034f7e37d98cb6e06249", "question_id": "q14_unanswerable", "split": "test", "cluster_id": "syn_03027_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 8, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1955, "query_tokens": 608, "message_tokens": 512, "message_utf8_bytes": 1087, "cache_tensor_bytes": null, "serialized_payload_bytes": 1087, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 51405.68041824736, "hops": {"1": 6896.277834079228, "2": 6372.689875075594, "3": 6377.503250027075, "4": 6359.274166985415, "5": 6348.349208012223, "6": 6350.234291981906, "7": 6348.7960420316085, "8": 6352.555750054307}, "receiver_prefill_ms": 226.98229097295552, "ttft_ms": 11.855292017571628, "decode_ms": 11.927249957807362}, "total_wall_ms": 51656.907084281556, "ttft_ms": 11.855292017571628, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 63, "facts_retained": 27, "retention_rate": 0.42857142857142855, "retention_by_kind": {"person": 0.5, "date": 0.125, "amount": 0.6153846153846154, "status": 0.75, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1087}, "ledger_by_hop": {"1": {"facts_total": 63, "facts_retained": 27, "retention_rate": 0.42857142857142855, "retention_by_kind": {"person": 0.5, "date": 0.125, "amount": 0.6153846153846154, "status": 0.75, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1087}, "2": {"facts_total": 63, "facts_retained": 27, "retention_rate": 0.42857142857142855, "retention_by_kind": {"person": 0.5, "date": 0.125, "amount": 0.6153846153846154, "status": 0.75, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1087}, "3": {"facts_total": 63, "facts_retained": 27, "retention_rate": 0.42857142857142855, "retention_by_kind": {"person": 0.5, "date": 0.125, "amount": 0.6153846153846154, "status": 0.75, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1087}, "4": {"facts_total": 63, "facts_retained": 27, "retention_rate": 0.42857142857142855, "retention_by_kind": {"person": 0.5, "date": 0.125, "amount": 0.6153846153846154, "status": 0.75, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1087}, "5": {"facts_total": 63, "facts_retained": 27, "retention_rate": 0.42857142857142855, "retention_by_kind": {"person": 0.5, "date": 0.125, "amount": 0.6153846153846154, "status": 0.75, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1087}, "6": {"facts_total": 63, "facts_retained": 27, "retention_rate": 0.42857142857142855, "retention_by_kind": {"person": 0.5, "date": 0.125, "amount": 0.6153846153846154, "status": 0.75, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1087}, "7": {"facts_total": 63, "facts_retained": 27, "retention_rate": 0.42857142857142855, "retention_by_kind": {"person": 0.5, "date": 0.125, "amount": 0.6153846153846154, "status": 0.75, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1087}, "8": {"facts_total": 63, "facts_retained": 27, "retention_rate": 0.42857142857142855, "retention_by_kind": {"person": 0.5, "date": 0.125, "amount": 0.6153846153846154, "status": 0.75, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1087}}}, "reference_key_id": "keys/synthetic/syn_03027_2000.json", "task_scores": {"correct": 1.0, "abstained": 1.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "1f88f0e78c63504e418dea843641b99a98555e15d47b02cd5cb2f57da3003dea", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "dc91bf3256611aa501924736", "attempt_id": "f8fd82f647e2", "run_group_id": "E4_core:A5|closed|qwen3b:h1", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:29:54.929608Z", "task_family": "synthetic", "document_id": "syn_03027_2000", "document_hash": "10326080979f71b625e4110c7c19767e5a752e267de5034f7e37d98cb6e06249", "question_id": "q00_single_fact", "split": "test", "cluster_id": "syn_03027_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 1, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1955, "query_tokens": 411, "message_tokens": 329, "message_utf8_bytes": 827, "cache_tensor_bytes": null, "serialized_payload_bytes": 827, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 6880.990542005748, "hops": {"1": 6880.990542005748}, "receiver_prefill_ms": 161.8083749199286, "ttft_ms": 11.755542014725506, "decode_ms": 11.817999999038875}, "total_wall_ms": 7066.811042022891, "ttft_ms": 11.755542014725506, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 63, "facts_retained": 9, "retention_rate": 0.14285714285714285, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.25, "amount": 0.07692307692307693, "status": 0.25, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 787}, "ledger_by_hop": {"1": {"facts_total": 63, "facts_retained": 9, "retention_rate": 0.14285714285714285, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.25, "amount": 0.07692307692307693, "status": 0.25, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 787}}}, "reference_key_id": "keys/synthetic/syn_03027_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "a3bc07ae34a4f72dd11a6738ade8463df3fa03c063d19d58651b6ead2a13f850", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "6cbe80d42f9714060a2991d0", "attempt_id": "8c6a5a6dbe22", "run_group_id": "E4_core:A5|closed|qwen3b:h1", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:29:55.130994Z", "task_family": "synthetic", "document_id": "syn_03027_2000", "document_hash": "10326080979f71b625e4110c7c19767e5a752e267de5034f7e37d98cb6e06249", "question_id": "q01_single_fact", "split": "test", "cluster_id": "syn_03027_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 1, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1955, "query_tokens": 418, "message_tokens": 329, "message_utf8_bytes": 827, "cache_tensor_bytes": null, "serialized_payload_bytes": 827, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 6880.990542005748, "hops": {"1": 6880.990542005748}, "receiver_prefill_ms": 163.06329204235226, "ttft_ms": 11.834666947834194, "decode_ms": 11.84266700875014}, "total_wall_ms": 7068.1651250924915, "ttft_ms": 11.834666947834194, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 63, "facts_retained": 9, "retention_rate": 0.14285714285714285, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.25, "amount": 0.07692307692307693, "status": 0.25, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 787}, "ledger_by_hop": {"1": {"facts_total": 63, "facts_retained": 9, "retention_rate": 0.14285714285714285, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.25, "amount": 0.07692307692307693, "status": 0.25, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 787}}}, "reference_key_id": "keys/synthetic/syn_03027_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "a3bc07ae34a4f72dd11a6738ade8463df3fa03c063d19d58651b6ead2a13f850", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "b0ad1d43e019fa53ed4f0173", "attempt_id": "478a77e6a2c0", "run_group_id": "E4_core:A5|closed|qwen3b:h1", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:29:55.333579Z", "task_family": "synthetic", "document_id": "syn_03027_2000", "document_hash": "10326080979f71b625e4110c7c19767e5a752e267de5034f7e37d98cb6e06249", "question_id": "q02_exact_copy", "split": "test", "cluster_id": "syn_03027_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 1, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1955, "query_tokens": 417, "message_tokens": 329, "message_utf8_bytes": 827, "cache_tensor_bytes": null, "serialized_payload_bytes": 827, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 6880.990542005748, "hops": {"1": 6880.990542005748}, "receiver_prefill_ms": 162.7220829250291, "ttft_ms": 11.770875076763332, "decode_ms": 11.774334008805454}, "total_wall_ms": 7067.642375943251, "ttft_ms": 11.770875076763332, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 63, "facts_retained": 9, "retention_rate": 0.14285714285714285, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.25, "amount": 0.07692307692307693, "status": 0.25, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 787}, "ledger_by_hop": {"1": {"facts_total": 63, "facts_retained": 9, "retention_rate": 0.14285714285714285, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.25, "amount": 0.07692307692307693, "status": 0.25, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 787}}}, "reference_key_id": "keys/synthetic/syn_03027_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "a3bc07ae34a4f72dd11a6738ade8463df3fa03c063d19d58651b6ead2a13f850", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "788a17db9e1537b34287249a", "attempt_id": "6631850234e7", "run_group_id": "E4_core:A5|closed|qwen3b:h1", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:29:55.535125Z", "task_family": "synthetic", "document_id": "syn_03027_2000", "document_hash": "10326080979f71b625e4110c7c19767e5a752e267de5034f7e37d98cb6e06249", "question_id": "q03_exact_copy", "split": "test", "cluster_id": "syn_03027_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 1, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1955, "query_tokens": 412, "message_tokens": 329, "message_utf8_bytes": 827, "cache_tensor_bytes": null, "serialized_payload_bytes": 827, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 6880.990542005748, "hops": {"1": 6880.990542005748}, "receiver_prefill_ms": 162.14337502606213, "ttft_ms": 11.827750015072525, "decode_ms": 105.74670799542218}, "total_wall_ms": 7161.081625963561, "ttft_ms": 11.827750015072525, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 9}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "Priyamvada Sørlie", "evidence_artifact": {"ledger": {"facts_total": 63, "facts_retained": 9, "retention_rate": 0.14285714285714285, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.25, "amount": 0.07692307692307693, "status": 0.25, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 787}, "ledger_by_hop": {"1": {"facts_total": 63, "facts_retained": 9, "retention_rate": 0.14285714285714285, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.25, "amount": 0.07692307692307693, "status": 0.25, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 787}}}, "reference_key_id": "keys/synthetic/syn_03027_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "a3bc07ae34a4f72dd11a6738ade8463df3fa03c063d19d58651b6ead2a13f850", "output_hash": "c5f554196ed23dd71ad995ec89f7142099d524dbb6534cc10d94a0c89d13e792"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "9460e4679fa38b776c8c592f", "attempt_id": "5022e9728bd7", "run_group_id": "E4_core:A5|closed|qwen3b:h1", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:29:55.830167Z", "task_family": "synthetic", "document_id": "syn_03027_2000", "document_hash": "10326080979f71b625e4110c7c19767e5a752e267de5034f7e37d98cb6e06249", "question_id": "q04_multi_hop", "split": "test", "cluster_id": "syn_03027_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 1, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1955, "query_tokens": 423, "message_tokens": 329, "message_utf8_bytes": 827, "cache_tensor_bytes": null, "serialized_payload_bytes": 827, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 6880.990542005748, "hops": {"1": 6880.990542005748}, "receiver_prefill_ms": 163.51045900955796, "ttft_ms": 11.79112505633384, "decode_ms": 11.737541994079947}, "total_wall_ms": 7068.429584032856, "ttft_ms": 11.79112505633384, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 63, "facts_retained": 9, "retention_rate": 0.14285714285714285, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.25, "amount": 0.07692307692307693, "status": 0.25, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 787}, "ledger_by_hop": {"1": {"facts_total": 63, "facts_retained": 9, "retention_rate": 0.14285714285714285, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.25, "amount": 0.07692307692307693, "status": 0.25, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 787}}}, "reference_key_id": "keys/synthetic/syn_03027_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "a3bc07ae34a4f72dd11a6738ade8463df3fa03c063d19d58651b6ead2a13f850", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "d7b5715c41d1321977f4f986", "attempt_id": "624a3ba78665", "run_group_id": "E4_core:A5|closed|qwen3b:h1", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:29:56.032452Z", "task_family": "synthetic", "document_id": "syn_03027_2000", "document_hash": "10326080979f71b625e4110c7c19767e5a752e267de5034f7e37d98cb6e06249", "question_id": "q05_multi_hop", "split": "test", "cluster_id": "syn_03027_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 1, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1955, "query_tokens": 420, "message_tokens": 329, "message_utf8_bytes": 827, "cache_tensor_bytes": null, "serialized_payload_bytes": 827, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 6880.990542005748, "hops": {"1": 6880.990542005748}, "receiver_prefill_ms": 163.4532919852063, "ttft_ms": 11.797582963481545, "decode_ms": 117.62612499296665}, "total_wall_ms": 7174.235957907513, "ttft_ms": 11.797582963481545, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 10}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "WT-26M-9227", "evidence_artifact": {"ledger": {"facts_total": 63, "facts_retained": 9, "retention_rate": 0.14285714285714285, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.25, "amount": 0.07692307692307693, "status": 0.25, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 787}, "ledger_by_hop": {"1": {"facts_total": 63, "facts_retained": 9, "retention_rate": 0.14285714285714285, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.25, "amount": 0.07692307692307693, "status": 0.25, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 787}}}, "reference_key_id": "keys/synthetic/syn_03027_2000.json", "task_scores": {"correct": 0.0, "abstained": 0.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "a3bc07ae34a4f72dd11a6738ade8463df3fa03c063d19d58651b6ead2a13f850", "output_hash": "ef2302e0ab6f5a6951b11463fea177ba766b9d438afcaf6e6101ebfa3ba38aba"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "52c9e7d825e2534692398267", "attempt_id": "173908bff61e", "run_group_id": "E4_core:A5|closed|qwen3b:h1", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:29:56.340583Z", "task_family": "synthetic", "document_id": "syn_03027_2000", "document_hash": "10326080979f71b625e4110c7c19767e5a752e267de5034f7e37d98cb6e06249", "question_id": "q06_negation", "split": "test", "cluster_id": "syn_03027_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 1, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1955, "query_tokens": 412, "message_tokens": 329, "message_utf8_bytes": 827, "cache_tensor_bytes": null, "serialized_payload_bytes": 827, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 6880.990542005748, "hops": {"1": 6880.990542005748}, "receiver_prefill_ms": 161.7492079967633, "ttft_ms": 11.77383295726031, "decode_ms": 11.793666984885931}, "total_wall_ms": 7066.660376032814, "ttft_ms": 11.77383295726031, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 63, "facts_retained": 9, "retention_rate": 0.14285714285714285, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.25, "amount": 0.07692307692307693, "status": 0.25, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 787}, "ledger_by_hop": {"1": {"facts_total": 63, "facts_retained": 9, "retention_rate": 0.14285714285714285, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.25, "amount": 0.07692307692307693, "status": 0.25, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 787}}}, "reference_key_id": "keys/synthetic/syn_03027_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "a3bc07ae34a4f72dd11a6738ade8463df3fa03c063d19d58651b6ead2a13f850", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "15721e70ebc62d40ce737d3c", "attempt_id": "0b9700866239", "run_group_id": "E4_core:A5|closed|qwen3b:h1", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:29:56.541097Z", "task_family": "synthetic", "document_id": "syn_03027_2000", "document_hash": "10326080979f71b625e4110c7c19767e5a752e267de5034f7e37d98cb6e06249", "question_id": "q07_negation", "split": "test", "cluster_id": "syn_03027_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 1, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1955, "query_tokens": 413, "message_tokens": 329, "message_utf8_bytes": 827, "cache_tensor_bytes": null, "serialized_payload_bytes": 827, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 6880.990542005748, "hops": {"1": 6880.990542005748}, "receiver_prefill_ms": 162.26837504655123, "ttft_ms": 11.79854094516486, "decode_ms": 11.771459016017616}, "total_wall_ms": 7067.187792039476, "ttft_ms": 11.79854094516486, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 63, "facts_retained": 9, "retention_rate": 0.14285714285714285, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.25, "amount": 0.07692307692307693, "status": 0.25, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 787}, "ledger_by_hop": {"1": {"facts_total": 63, "facts_retained": 9, "retention_rate": 0.14285714285714285, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.25, "amount": 0.07692307692307693, "status": 0.25, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 787}}}, "reference_key_id": "keys/synthetic/syn_03027_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "a3bc07ae34a4f72dd11a6738ade8463df3fa03c063d19d58651b6ead2a13f850", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "5b731c948e203da928543b78", "attempt_id": "ac3bf59ea017", "run_group_id": "E4_core:A5|closed|qwen3b:h1", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:29:56.742292Z", "task_family": "synthetic", "document_id": "syn_03027_2000", "document_hash": "10326080979f71b625e4110c7c19767e5a752e267de5034f7e37d98cb6e06249", "question_id": "q08_revision", "split": "test", "cluster_id": "syn_03027_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 1, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1955, "query_tokens": 427, "message_tokens": 329, "message_utf8_bytes": 827, "cache_tensor_bytes": null, "serialized_payload_bytes": 827, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 6880.990542005748, "hops": {"1": 6880.990542005748}, "receiver_prefill_ms": 164.2026249319315, "ttft_ms": 11.838082922622561, "decode_ms": 105.69224995560944}, "total_wall_ms": 7163.160209078342, "ttft_ms": 11.838082922622561, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 9}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "6,914,937", "evidence_artifact": {"ledger": {"facts_total": 63, "facts_retained": 9, "retention_rate": 0.14285714285714285, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.25, "amount": 0.07692307692307693, "status": 0.25, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 787}, "ledger_by_hop": {"1": {"facts_total": 63, "facts_retained": 9, "retention_rate": 0.14285714285714285, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.25, "amount": 0.07692307692307693, "status": 0.25, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 787}}}, "reference_key_id": "keys/synthetic/syn_03027_2000.json", "task_scores": {"correct": 0.0, "abstained": 0.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "a3bc07ae34a4f72dd11a6738ade8463df3fa03c063d19d58651b6ead2a13f850", "output_hash": "c56851abfdbf8851226a1e71c4911c616a691c490a883eb754e61bac551d004c"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "9f91a40ae69006ed1e5cd02d", "attempt_id": "f6502e223df8", "run_group_id": "E4_core:A5|closed|qwen3b:h1", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:29:57.039446Z", "task_family": "synthetic", "document_id": "syn_03027_2000", "document_hash": "10326080979f71b625e4110c7c19767e5a752e267de5034f7e37d98cb6e06249", "question_id": "q09_revision", "split": "test", "cluster_id": "syn_03027_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 1, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1955, "query_tokens": 420, "message_tokens": 329, "message_utf8_bytes": 827, "cache_tensor_bytes": null, "serialized_payload_bytes": 827, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 6880.990542005748, "hops": {"1": 6880.990542005748}, "receiver_prefill_ms": 162.7134169684723, "ttft_ms": 11.790916905738413, "decode_ms": 94.00495805311948}, "total_wall_ms": 7149.857000098564, "ttft_ms": 11.790916905738413, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 8}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "15828866", "evidence_artifact": {"ledger": {"facts_total": 63, "facts_retained": 9, "retention_rate": 0.14285714285714285, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.25, "amount": 0.07692307692307693, "status": 0.25, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 787}, "ledger_by_hop": {"1": {"facts_total": 63, "facts_retained": 9, "retention_rate": 0.14285714285714285, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.25, "amount": 0.07692307692307693, "status": 0.25, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 787}}}, "reference_key_id": "keys/synthetic/syn_03027_2000.json", "task_scores": {"correct": 0.0, "abstained": 0.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "a3bc07ae34a4f72dd11a6738ade8463df3fa03c063d19d58651b6ead2a13f850", "output_hash": "c9d49be6f876880a6dfe30589ff0b746c048ebaf1842118a883f5503a349a8b6"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "4cf09811dbc2279993f1a833", "attempt_id": "5b7de7e763f7", "run_group_id": "E4_core:A5|closed|qwen3b:h1", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:29:57.323554Z", "task_family": "synthetic", "document_id": "syn_03027_2000", "document_hash": "10326080979f71b625e4110c7c19767e5a752e267de5034f7e37d98cb6e06249", "question_id": "q10_temporal", "split": "test", "cluster_id": "syn_03027_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 1, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1955, "query_tokens": 430, "message_tokens": 329, "message_utf8_bytes": 827, "cache_tensor_bytes": null, "serialized_payload_bytes": 827, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 6880.990542005748, "hops": {"1": 6880.990542005748}, "receiver_prefill_ms": 164.0333749819547, "ttft_ms": 11.753374943509698, "decode_ms": 11.779749998822808}, "total_wall_ms": 7068.9538340084255, "ttft_ms": 11.753374943509698, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 63, "facts_retained": 9, "retention_rate": 0.14285714285714285, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.25, "amount": 0.07692307692307693, "status": 0.25, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 787}, "ledger_by_hop": {"1": {"facts_total": 63, "facts_retained": 9, "retention_rate": 0.14285714285714285, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.25, "amount": 0.07692307692307693, "status": 0.25, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 787}}}, "reference_key_id": "keys/synthetic/syn_03027_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "a3bc07ae34a4f72dd11a6738ade8463df3fa03c063d19d58651b6ead2a13f850", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "59a500566bdd55db455ad8b4", "attempt_id": "794842ffcd57", "run_group_id": "E4_core:A5|closed|qwen3b:h1", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:29:57.526547Z", "task_family": "synthetic", "document_id": "syn_03027_2000", "document_hash": "10326080979f71b625e4110c7c19767e5a752e267de5034f7e37d98cb6e06249", "question_id": "q11_aggregation", "split": "test", "cluster_id": "syn_03027_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 1, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1955, "query_tokens": 423, "message_tokens": 329, "message_utf8_bytes": 827, "cache_tensor_bytes": null, "serialized_payload_bytes": 827, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 6880.990542005748, "hops": {"1": 6880.990542005748}, "receiver_prefill_ms": 163.1590829929337, "ttft_ms": 11.918083066120744, "decode_ms": 82.08591700531542}, "total_wall_ms": 7138.619584031403, "ttft_ms": 11.918083066120744, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 7}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "6914937", "evidence_artifact": {"ledger": {"facts_total": 63, "facts_retained": 9, "retention_rate": 0.14285714285714285, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.25, "amount": 0.07692307692307693, "status": 0.25, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 787}, "ledger_by_hop": {"1": {"facts_total": 63, "facts_retained": 9, "retention_rate": 0.14285714285714285, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.25, "amount": 0.07692307692307693, "status": 0.25, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 787}}}, "reference_key_id": "keys/synthetic/syn_03027_2000.json", "task_scores": {"correct": 0.0, "abstained": 0.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "a3bc07ae34a4f72dd11a6738ade8463df3fa03c063d19d58651b6ead2a13f850", "output_hash": "534d40ded32f784c68307dd2386faed2a27bf39dcc27321ee6985e42ccd285e5"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "48670f301a79e0bb5d3c71b0", "attempt_id": "e89bed80944f", "run_group_id": "E4_core:A5|closed|qwen3b:h1", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:29:57.799468Z", "task_family": "synthetic", "document_id": "syn_03027_2000", "document_hash": "10326080979f71b625e4110c7c19767e5a752e267de5034f7e37d98cb6e06249", "question_id": "q12_unit_conversion", "split": "test", "cluster_id": "syn_03027_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 1, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1955, "query_tokens": 420, "message_tokens": 329, "message_utf8_bytes": 827, "cache_tensor_bytes": null, "serialized_payload_bytes": 827, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 6880.990542005748, "hops": {"1": 6880.990542005748}, "receiver_prefill_ms": 163.70604105759412, "ttft_ms": 11.711874976754189, "decode_ms": 11.718790978193283}, "total_wall_ms": 7068.602624931373, "ttft_ms": 11.711874976754189, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 63, "facts_retained": 9, "retention_rate": 0.14285714285714285, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.25, "amount": 0.07692307692307693, "status": 0.25, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 787}, "ledger_by_hop": {"1": {"facts_total": 63, "facts_retained": 9, "retention_rate": 0.14285714285714285, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.25, "amount": 0.07692307692307693, "status": 0.25, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 787}}}, "reference_key_id": "keys/synthetic/syn_03027_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "a3bc07ae34a4f72dd11a6738ade8463df3fa03c063d19d58651b6ead2a13f850", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "4646412da99cbc2f36971c9c", "attempt_id": "ef266b3c327b", "run_group_id": "E4_core:A5|closed|qwen3b:h1", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:29:58.002612Z", "task_family": "synthetic", "document_id": "syn_03027_2000", "document_hash": "10326080979f71b625e4110c7c19767e5a752e267de5034f7e37d98cb6e06249", "question_id": "q13_unanswerable", "split": "test", "cluster_id": "syn_03027_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 1, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1955, "query_tokens": 423, "message_tokens": 329, "message_utf8_bytes": 827, "cache_tensor_bytes": null, "serialized_payload_bytes": 827, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 6880.990542005748, "hops": {"1": 6880.990542005748}, "receiver_prefill_ms": 163.35641697514802, "ttft_ms": 12.127833091653883, "decode_ms": 11.905333027243614}, "total_wall_ms": 7068.938417010941, "ttft_ms": 12.127833091653883, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 63, "facts_retained": 9, "retention_rate": 0.14285714285714285, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.25, "amount": 0.07692307692307693, "status": 0.25, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 787}, "ledger_by_hop": {"1": {"facts_total": 63, "facts_retained": 9, "retention_rate": 0.14285714285714285, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.25, "amount": 0.07692307692307693, "status": 0.25, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 787}}}, "reference_key_id": "keys/synthetic/syn_03027_2000.json", "task_scores": {"correct": 1.0, "abstained": 1.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "a3bc07ae34a4f72dd11a6738ade8463df3fa03c063d19d58651b6ead2a13f850", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "9d26bae2abb3869b62d84b4d", "attempt_id": "5a2f5b547565", "run_group_id": "E4_core:A5|closed|qwen3b:h1", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:29:58.205623Z", "task_family": "synthetic", "document_id": "syn_03027_2000", "document_hash": "10326080979f71b625e4110c7c19767e5a752e267de5034f7e37d98cb6e06249", "question_id": "q14_unanswerable", "split": "test", "cluster_id": "syn_03027_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 1, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1955, "query_tokens": 425, "message_tokens": 329, "message_utf8_bytes": 827, "cache_tensor_bytes": null, "serialized_payload_bytes": 827, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 6880.990542005748, "hops": {"1": 6880.990542005748}, "receiver_prefill_ms": 162.71916695404798, "ttft_ms": 12.210833956487477, "decode_ms": 11.97233295533806}, "total_wall_ms": 7068.44333303161, "ttft_ms": 12.210833956487477, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 63, "facts_retained": 9, "retention_rate": 0.14285714285714285, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.25, "amount": 0.07692307692307693, "status": 0.25, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 787}, "ledger_by_hop": {"1": {"facts_total": 63, "facts_retained": 9, "retention_rate": 0.14285714285714285, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.25, "amount": 0.07692307692307693, "status": 0.25, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 787}}}, "reference_key_id": "keys/synthetic/syn_03027_2000.json", "task_scores": {"correct": 1.0, "abstained": 1.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "a3bc07ae34a4f72dd11a6738ade8463df3fa03c063d19d58651b6ead2a13f850", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "a215fb56bdd5d9dcc24ee100", "attempt_id": "6872ee6a37c9", "run_group_id": "E4_core:A5|closed|qwen3b:h2", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:30:02.286724Z", "task_family": "synthetic", "document_id": "syn_03027_2000", "document_hash": "10326080979f71b625e4110c7c19767e5a752e267de5034f7e37d98cb6e06249", "question_id": "q00_single_fact", "split": "test", "cluster_id": "syn_03027_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 2, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1955, "query_tokens": 376, "message_tokens": 294, "message_utf8_bytes": 736, "cache_tensor_bytes": null, "serialized_payload_bytes": 736, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 10756.447292049415, "hops": {"1": 6880.990542005748, "2": 3875.456750043668}, "receiver_prefill_ms": 143.5628749895841, "ttft_ms": 11.809916002675891, "decode_ms": 11.738666100427508}, "total_wall_ms": 10923.97341702599, "ttft_ms": 11.809916002675891, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 63, "facts_retained": 8, "retention_rate": 0.12698412698412698, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.125, "amount": 0.07692307692307693, "status": 0.25, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 696}, "ledger_by_hop": {"1": {"facts_total": 63, "facts_retained": 9, "retention_rate": 0.14285714285714285, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.25, "amount": 0.07692307692307693, "status": 0.25, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 787}, "2": {"facts_total": 63, "facts_retained": 8, "retention_rate": 0.12698412698412698, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.125, "amount": 0.07692307692307693, "status": 0.25, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 696}}}, "reference_key_id": "keys/synthetic/syn_03027_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "5de8bbf0fdf02bbceb67207371e6f5c0ec86d2b30877ec832e254d4b22f40895", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "80f81157a4dd9708955d1999", "attempt_id": "35b9f0e90c25", "run_group_id": "E4_core:A5|closed|qwen3b:h2", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:30:02.470318Z", "task_family": "synthetic", "document_id": "syn_03027_2000", "document_hash": "10326080979f71b625e4110c7c19767e5a752e267de5034f7e37d98cb6e06249", "question_id": "q01_single_fact", "split": "test", "cluster_id": "syn_03027_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 2, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1955, "query_tokens": 383, "message_tokens": 294, "message_utf8_bytes": 736, "cache_tensor_bytes": null, "serialized_payload_bytes": 736, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 10756.447292049415, "hops": {"1": 6880.990542005748, "2": 3875.456750043668}, "receiver_prefill_ms": 144.05679190531373, "ttft_ms": 11.773000005632639, "decode_ms": 11.773083009757102}, "total_wall_ms": 10924.424584023654, "ttft_ms": 11.773000005632639, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 63, "facts_retained": 8, "retention_rate": 0.12698412698412698, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.125, "amount": 0.07692307692307693, "status": 0.25, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 696}, "ledger_by_hop": {"1": {"facts_total": 63, "facts_retained": 9, "retention_rate": 0.14285714285714285, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.25, "amount": 0.07692307692307693, "status": 0.25, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 787}, "2": {"facts_total": 63, "facts_retained": 8, "retention_rate": 0.12698412698412698, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.125, "amount": 0.07692307692307693, "status": 0.25, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 696}}}, "reference_key_id": "keys/synthetic/syn_03027_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "5de8bbf0fdf02bbceb67207371e6f5c0ec86d2b30877ec832e254d4b22f40895", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "3c193555e08b06c0c29574bd", "attempt_id": "84b5076b1f98", "run_group_id": "E4_core:A5|closed|qwen3b:h2", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:30:02.653790Z", "task_family": "synthetic", "document_id": "syn_03027_2000", "document_hash": "10326080979f71b625e4110c7c19767e5a752e267de5034f7e37d98cb6e06249", "question_id": "q02_exact_copy", "split": "test", "cluster_id": "syn_03027_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 2, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1955, "query_tokens": 382, "message_tokens": 294, "message_utf8_bytes": 736, "cache_tensor_bytes": null, "serialized_payload_bytes": 736, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 10756.447292049415, "hops": {"1": 6880.990542005748, "2": 3875.456750043668}, "receiver_prefill_ms": 143.37716600857675, "ttft_ms": 11.903790989890695, "decode_ms": 11.75787509419024}, "total_wall_ms": 10923.935167025775, "ttft_ms": 11.903790989890695, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 63, "facts_retained": 8, "retention_rate": 0.12698412698412698, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.125, "amount": 0.07692307692307693, "status": 0.25, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 696}, "ledger_by_hop": {"1": {"facts_total": 63, "facts_retained": 9, "retention_rate": 0.14285714285714285, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.25, "amount": 0.07692307692307693, "status": 0.25, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 787}, "2": {"facts_total": 63, "facts_retained": 8, "retention_rate": 0.12698412698412698, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.125, "amount": 0.07692307692307693, "status": 0.25, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 696}}}, "reference_key_id": "keys/synthetic/syn_03027_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "5de8bbf0fdf02bbceb67207371e6f5c0ec86d2b30877ec832e254d4b22f40895", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "a6c018aceca5a88492bcdb91", "attempt_id": "e42cad0c552b", "run_group_id": "E4_core:A5|closed|qwen3b:h2", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:30:02.837343Z", "task_family": "synthetic", "document_id": "syn_03027_2000", "document_hash": "10326080979f71b625e4110c7c19767e5a752e267de5034f7e37d98cb6e06249", "question_id": "q03_exact_copy", "split": "test", "cluster_id": "syn_03027_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 2, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1955, "query_tokens": 377, "message_tokens": 294, "message_utf8_bytes": 736, "cache_tensor_bytes": null, "serialized_payload_bytes": 736, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 10756.447292049415, "hops": {"1": 6880.990542005748, "2": 3875.456750043668}, "receiver_prefill_ms": 143.55033403262496, "ttft_ms": 11.759332963265479, "decode_ms": 105.43054109439254}, "total_wall_ms": 11017.573041957803, "ttft_ms": 11.759332963265479, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 9}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "Priyamvada Sørlie", "evidence_artifact": {"ledger": {"facts_total": 63, "facts_retained": 8, "retention_rate": 0.12698412698412698, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.125, "amount": 0.07692307692307693, "status": 0.25, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 696}, "ledger_by_hop": {"1": {"facts_total": 63, "facts_retained": 9, "retention_rate": 0.14285714285714285, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.25, "amount": 0.07692307692307693, "status": 0.25, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 787}, "2": {"facts_total": 63, "facts_retained": 8, "retention_rate": 0.12698412698412698, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.125, "amount": 0.07692307692307693, "status": 0.25, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 696}}}, "reference_key_id": "keys/synthetic/syn_03027_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "5de8bbf0fdf02bbceb67207371e6f5c0ec86d2b30877ec832e254d4b22f40895", "output_hash": "c5f554196ed23dd71ad995ec89f7142099d524dbb6534cc10d94a0c89d13e792"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "583deff437bb60efb1b6c8f8", "attempt_id": "63ee471a94da", "run_group_id": "E4_core:A5|closed|qwen3b:h2", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:30:03.114265Z", "task_family": "synthetic", "document_id": "syn_03027_2000", "document_hash": "10326080979f71b625e4110c7c19767e5a752e267de5034f7e37d98cb6e06249", "question_id": "q04_multi_hop", "split": "test", "cluster_id": "syn_03027_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 2, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1955, "query_tokens": 388, "message_tokens": 294, "message_utf8_bytes": 736, "cache_tensor_bytes": null, "serialized_payload_bytes": 736, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 10756.447292049415, "hops": {"1": 6880.990542005748, "2": 3875.456750043668}, "receiver_prefill_ms": 161.0157919349149, "ttft_ms": 11.84691593516618, "decode_ms": 11.685207951813936}, "total_wall_ms": 10941.439876100048, "ttft_ms": 11.84691593516618, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 63, "facts_retained": 8, "retention_rate": 0.12698412698412698, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.125, "amount": 0.07692307692307693, "status": 0.25, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 696}, "ledger_by_hop": {"1": {"facts_total": 63, "facts_retained": 9, "retention_rate": 0.14285714285714285, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.25, "amount": 0.07692307692307693, "status": 0.25, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 787}, "2": {"facts_total": 63, "facts_retained": 8, "retention_rate": 0.12698412698412698, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.125, "amount": 0.07692307692307693, "status": 0.25, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 696}}}, "reference_key_id": "keys/synthetic/syn_03027_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "5de8bbf0fdf02bbceb67207371e6f5c0ec86d2b30877ec832e254d4b22f40895", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "a8f600170f94579ec9585d34", "attempt_id": "6b9c20cf7d3f", "run_group_id": "E4_core:A5|closed|qwen3b:h2", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:30:03.314364Z", "task_family": "synthetic", "document_id": "syn_03027_2000", "document_hash": "10326080979f71b625e4110c7c19767e5a752e267de5034f7e37d98cb6e06249", "question_id": "q05_multi_hop", "split": "test", "cluster_id": "syn_03027_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 2, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1955, "query_tokens": 385, "message_tokens": 294, "message_utf8_bytes": 736, "cache_tensor_bytes": null, "serialized_payload_bytes": 736, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 10756.447292049415, "hops": {"1": 6880.990542005748, "2": 3875.456750043668}, "receiver_prefill_ms": 142.33524992596358, "ttft_ms": 11.749166995286942, "decode_ms": 116.95554200559855}, "total_wall_ms": 11027.872417122126, "ttft_ms": 11.749166995286942, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 10}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "WT-26M-9227", "evidence_artifact": {"ledger": {"facts_total": 63, "facts_retained": 8, "retention_rate": 0.12698412698412698, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.125, "amount": 0.07692307692307693, "status": 0.25, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 696}, "ledger_by_hop": {"1": {"facts_total": 63, "facts_retained": 9, "retention_rate": 0.14285714285714285, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.25, "amount": 0.07692307692307693, "status": 0.25, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 787}, "2": {"facts_total": 63, "facts_retained": 8, "retention_rate": 0.12698412698412698, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.125, "amount": 0.07692307692307693, "status": 0.25, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 696}}}, "reference_key_id": "keys/synthetic/syn_03027_2000.json", "task_scores": {"correct": 0.0, "abstained": 0.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "5de8bbf0fdf02bbceb67207371e6f5c0ec86d2b30877ec832e254d4b22f40895", "output_hash": "ef2302e0ab6f5a6951b11463fea177ba766b9d438afcaf6e6101ebfa3ba38aba"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "bc8cecc338f4d774bf3db62a", "attempt_id": "850bc39cf5ff", "run_group_id": "E4_core:A5|closed|qwen3b:h2", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:30:03.601080Z", "task_family": "synthetic", "document_id": "syn_03027_2000", "document_hash": "10326080979f71b625e4110c7c19767e5a752e267de5034f7e37d98cb6e06249", "question_id": "q06_negation", "split": "test", "cluster_id": "syn_03027_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 2, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1955, "query_tokens": 377, "message_tokens": 294, "message_utf8_bytes": 736, "cache_tensor_bytes": null, "serialized_payload_bytes": 736, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 10756.447292049415, "hops": {"1": 6880.990542005748, "2": 3875.456750043668}, "receiver_prefill_ms": 143.05870793759823, "ttft_ms": 12.021500035189092, "decode_ms": 11.887499946169555}, "total_wall_ms": 10923.935708007775, "ttft_ms": 12.021500035189092, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 63, "facts_retained": 8, "retention_rate": 0.12698412698412698, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.125, "amount": 0.07692307692307693, "status": 0.25, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 696}, "ledger_by_hop": {"1": {"facts_total": 63, "facts_retained": 9, "retention_rate": 0.14285714285714285, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.25, "amount": 0.07692307692307693, "status": 0.25, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 787}, "2": {"facts_total": 63, "facts_retained": 8, "retention_rate": 0.12698412698412698, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.125, "amount": 0.07692307692307693, "status": 0.25, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 696}}}, "reference_key_id": "keys/synthetic/syn_03027_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "5de8bbf0fdf02bbceb67207371e6f5c0ec86d2b30877ec832e254d4b22f40895", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "cdef368123aba3a0bbe72e54", "attempt_id": "364d8e70244c", "run_group_id": "E4_core:A5|closed|qwen3b:h2", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:30:03.783684Z", "task_family": "synthetic", "document_id": "syn_03027_2000", "document_hash": "10326080979f71b625e4110c7c19767e5a752e267de5034f7e37d98cb6e06249", "question_id": "q07_negation", "split": "test", "cluster_id": "syn_03027_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 2, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1955, "query_tokens": 378, "message_tokens": 294, "message_utf8_bytes": 736, "cache_tensor_bytes": null, "serialized_payload_bytes": 736, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 10756.447292049415, "hops": {"1": 6880.990542005748, "2": 3875.456750043668}, "receiver_prefill_ms": 142.8997089387849, "ttft_ms": 12.004165910184383, "decode_ms": 11.800624895840883}, "total_wall_ms": 10923.662667046301, "ttft_ms": 12.004165910184383, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 63, "facts_retained": 8, "retention_rate": 0.12698412698412698, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.125, "amount": 0.07692307692307693, "status": 0.25, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 696}, "ledger_by_hop": {"1": {"facts_total": 63, "facts_retained": 9, "retention_rate": 0.14285714285714285, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.25, "amount": 0.07692307692307693, "status": 0.25, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 787}, "2": {"facts_total": 63, "facts_retained": 8, "retention_rate": 0.12698412698412698, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.125, "amount": 0.07692307692307693, "status": 0.25, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 696}}}, "reference_key_id": "keys/synthetic/syn_03027_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "5de8bbf0fdf02bbceb67207371e6f5c0ec86d2b30877ec832e254d4b22f40895", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "6481b95b4e7bb581fbe16fe3", "attempt_id": "4236d1c6f117", "run_group_id": "E4_core:A5|closed|qwen3b:h2", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:30:03.968321Z", "task_family": "synthetic", "document_id": "syn_03027_2000", "document_hash": "10326080979f71b625e4110c7c19767e5a752e267de5034f7e37d98cb6e06249", "question_id": "q08_revision", "split": "test", "cluster_id": "syn_03027_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 2, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1955, "query_tokens": 392, "message_tokens": 294, "message_utf8_bytes": 736, "cache_tensor_bytes": null, "serialized_payload_bytes": 736, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 10756.447292049415, "hops": {"1": 6880.990542005748, "2": 3875.456750043668}, "receiver_prefill_ms": 161.11570899374783, "ttft_ms": 12.195707997307181, "decode_ms": 106.42774996813387}, "total_wall_ms": 11036.710750078782, "ttft_ms": 12.195707997307181, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 9}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "6,914,937", "evidence_artifact": {"ledger": {"facts_total": 63, "facts_retained": 8, "retention_rate": 0.12698412698412698, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.125, "amount": 0.07692307692307693, "status": 0.25, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 696}, "ledger_by_hop": {"1": {"facts_total": 63, "facts_retained": 9, "retention_rate": 0.14285714285714285, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.25, "amount": 0.07692307692307693, "status": 0.25, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 787}, "2": {"facts_total": 63, "facts_retained": 8, "retention_rate": 0.12698412698412698, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.125, "amount": 0.07692307692307693, "status": 0.25, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 696}}}, "reference_key_id": "keys/synthetic/syn_03027_2000.json", "task_scores": {"correct": 0.0, "abstained": 0.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "5de8bbf0fdf02bbceb67207371e6f5c0ec86d2b30877ec832e254d4b22f40895", "output_hash": "c56851abfdbf8851226a1e71c4911c616a691c490a883eb754e61bac551d004c"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "1544335c4a8e6a5ff2c7427e", "attempt_id": "27cfaf6f0bd2", "run_group_id": "E4_core:A5|closed|qwen3b:h2", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:30:04.266219Z", "task_family": "synthetic", "document_id": "syn_03027_2000", "document_hash": "10326080979f71b625e4110c7c19767e5a752e267de5034f7e37d98cb6e06249", "question_id": "q09_revision", "split": "test", "cluster_id": "syn_03027_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 2, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1955, "query_tokens": 385, "message_tokens": 294, "message_utf8_bytes": 736, "cache_tensor_bytes": null, "serialized_payload_bytes": 736, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 10756.447292049415, "hops": {"1": 6880.990542005748, "2": 3875.456750043668}, "receiver_prefill_ms": 142.22429203800857, "ttft_ms": 12.08758307620883, "decode_ms": 82.41825003642589}, "total_wall_ms": 10993.69737599045, "ttft_ms": 12.08758307620883, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 7}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "6914937", "evidence_artifact": {"ledger": {"facts_total": 63, "facts_retained": 8, "retention_rate": 0.12698412698412698, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.125, "amount": 0.07692307692307693, "status": 0.25, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 696}, "ledger_by_hop": {"1": {"facts_total": 63, "facts_retained": 9, "retention_rate": 0.14285714285714285, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.25, "amount": 0.07692307692307693, "status": 0.25, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 787}, "2": {"facts_total": 63, "facts_retained": 8, "retention_rate": 0.12698412698412698, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.125, "amount": 0.07692307692307693, "status": 0.25, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 696}}}, "reference_key_id": "keys/synthetic/syn_03027_2000.json", "task_scores": {"correct": 0.0, "abstained": 0.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "5de8bbf0fdf02bbceb67207371e6f5c0ec86d2b30877ec832e254d4b22f40895", "output_hash": "534d40ded32f784c68307dd2386faed2a27bf39dcc27321ee6985e42ccd285e5"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "8edca167e09f3375bc616adf", "attempt_id": "bc297b7098fc", "run_group_id": "E4_core:A5|closed|qwen3b:h2", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:30:04.520972Z", "task_family": "synthetic", "document_id": "syn_03027_2000", "document_hash": "10326080979f71b625e4110c7c19767e5a752e267de5034f7e37d98cb6e06249", "question_id": "q10_temporal", "split": "test", "cluster_id": "syn_03027_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 2, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1955, "query_tokens": 395, "message_tokens": 294, "message_utf8_bytes": 736, "cache_tensor_bytes": null, "serialized_payload_bytes": 736, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 10756.447292049415, "hops": {"1": 6880.990542005748, "2": 3875.456750043668}, "receiver_prefill_ms": 161.3340419717133, "ttft_ms": 12.025582953356206, "decode_ms": 11.966958991251886}, "total_wall_ms": 10942.332791979425, "ttft_ms": 12.025582953356206, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 63, "facts_retained": 8, "retention_rate": 0.12698412698412698, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.125, "amount": 0.07692307692307693, "status": 0.25, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 696}, "ledger_by_hop": {"1": {"facts_total": 63, "facts_retained": 9, "retention_rate": 0.14285714285714285, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.25, "amount": 0.07692307692307693, "status": 0.25, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 787}, "2": {"facts_total": 63, "facts_retained": 8, "retention_rate": 0.12698412698412698, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.125, "amount": 0.07692307692307693, "status": 0.25, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 696}}}, "reference_key_id": "keys/synthetic/syn_03027_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "5de8bbf0fdf02bbceb67207371e6f5c0ec86d2b30877ec832e254d4b22f40895", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "3dee557e0c4e200d2b200a6f", "attempt_id": "d4b530b0131e", "run_group_id": "E4_core:A5|closed|qwen3b:h2", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:30:04.724024Z", "task_family": "synthetic", "document_id": "syn_03027_2000", "document_hash": "10326080979f71b625e4110c7c19767e5a752e267de5034f7e37d98cb6e06249", "question_id": "q11_aggregation", "split": "test", "cluster_id": "syn_03027_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 2, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1955, "query_tokens": 388, "message_tokens": 294, "message_utf8_bytes": 736, "cache_tensor_bytes": null, "serialized_payload_bytes": 736, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 10756.447292049415, "hops": {"1": 6880.990542005748, "2": 3875.456750043668}, "receiver_prefill_ms": 160.82124994136393, "ttft_ms": 12.035874999128282, "decode_ms": 82.46141695417464}, "total_wall_ms": 11012.295959051698, "ttft_ms": 12.035874999128282, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 7}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "6914937", "evidence_artifact": {"ledger": {"facts_total": 63, "facts_retained": 8, "retention_rate": 0.12698412698412698, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.125, "amount": 0.07692307692307693, "status": 0.25, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 696}, "ledger_by_hop": {"1": {"facts_total": 63, "facts_retained": 9, "retention_rate": 0.14285714285714285, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.25, "amount": 0.07692307692307693, "status": 0.25, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 787}, "2": {"facts_total": 63, "facts_retained": 8, "retention_rate": 0.12698412698412698, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.125, "amount": 0.07692307692307693, "status": 0.25, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 696}}}, "reference_key_id": "keys/synthetic/syn_03027_2000.json", "task_scores": {"correct": 0.0, "abstained": 0.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "5de8bbf0fdf02bbceb67207371e6f5c0ec86d2b30877ec832e254d4b22f40895", "output_hash": "534d40ded32f784c68307dd2386faed2a27bf39dcc27321ee6985e42ccd285e5"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "a1ba19e6806d21bca12e1e30", "attempt_id": "db44ca22e817", "run_group_id": "E4_core:A5|closed|qwen3b:h2", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:30:04.997627Z", "task_family": "synthetic", "document_id": "syn_03027_2000", "document_hash": "10326080979f71b625e4110c7c19767e5a752e267de5034f7e37d98cb6e06249", "question_id": "q12_unit_conversion", "split": "test", "cluster_id": "syn_03027_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 2, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1955, "query_tokens": 385, "message_tokens": 294, "message_utf8_bytes": 736, "cache_tensor_bytes": null, "serialized_payload_bytes": 736, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 10756.447292049415, "hops": {"1": 6880.990542005748, "2": 3875.456750043668}, "receiver_prefill_ms": 142.9430419811979, "ttft_ms": 12.168082990683615, "decode_ms": 11.824416927993298}, "total_wall_ms": 10923.923125024885, "ttft_ms": 12.168082990683615, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 63, "facts_retained": 8, "retention_rate": 0.12698412698412698, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.125, "amount": 0.07692307692307693, "status": 0.25, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 696}, "ledger_by_hop": {"1": {"facts_total": 63, "facts_retained": 9, "retention_rate": 0.14285714285714285, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.25, "amount": 0.07692307692307693, "status": 0.25, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 787}, "2": {"facts_total": 63, "facts_retained": 8, "retention_rate": 0.12698412698412698, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.125, "amount": 0.07692307692307693, "status": 0.25, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 696}}}, "reference_key_id": "keys/synthetic/syn_03027_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "5de8bbf0fdf02bbceb67207371e6f5c0ec86d2b30877ec832e254d4b22f40895", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "49c0f71b6f88463fac3f55ca", "attempt_id": "90543c38e07d", "run_group_id": "E4_core:A5|closed|qwen3b:h2", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:30:05.181995Z", "task_family": "synthetic", "document_id": "syn_03027_2000", "document_hash": "10326080979f71b625e4110c7c19767e5a752e267de5034f7e37d98cb6e06249", "question_id": "q13_unanswerable", "split": "test", "cluster_id": "syn_03027_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 2, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1955, "query_tokens": 388, "message_tokens": 294, "message_utf8_bytes": 736, "cache_tensor_bytes": null, "serialized_payload_bytes": 736, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 10756.447292049415, "hops": {"1": 6880.990542005748, "2": 3875.456750043668}, "receiver_prefill_ms": 160.85291700437665, "ttft_ms": 12.146541033871472, "decode_ms": 11.815583100542426}, "total_wall_ms": 10941.807334078476, "ttft_ms": 12.146541033871472, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 63, "facts_retained": 8, "retention_rate": 0.12698412698412698, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.125, "amount": 0.07692307692307693, "status": 0.25, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 696}, "ledger_by_hop": {"1": {"facts_total": 63, "facts_retained": 9, "retention_rate": 0.14285714285714285, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.25, "amount": 0.07692307692307693, "status": 0.25, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 787}, "2": {"facts_total": 63, "facts_retained": 8, "retention_rate": 0.12698412698412698, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.125, "amount": 0.07692307692307693, "status": 0.25, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 696}}}, "reference_key_id": "keys/synthetic/syn_03027_2000.json", "task_scores": {"correct": 1.0, "abstained": 1.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "5de8bbf0fdf02bbceb67207371e6f5c0ec86d2b30877ec832e254d4b22f40895", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "9dbf30fd230eb6add33b7d79", "attempt_id": "08bdc60456e6", "run_group_id": "E4_core:A5|closed|qwen3b:h2", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:30:05.385180Z", "task_family": "synthetic", "document_id": "syn_03027_2000", "document_hash": "10326080979f71b625e4110c7c19767e5a752e267de5034f7e37d98cb6e06249", "question_id": "q14_unanswerable", "split": "test", "cluster_id": "syn_03027_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 2, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1955, "query_tokens": 390, "message_tokens": 294, "message_utf8_bytes": 736, "cache_tensor_bytes": null, "serialized_payload_bytes": 736, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 10756.447292049415, "hops": {"1": 6880.990542005748, "2": 3875.456750043668}, "receiver_prefill_ms": 161.5862079197541, "ttft_ms": 11.996874934993684, "decode_ms": 11.793666053563356}, "total_wall_ms": 10942.224792088382, "ttft_ms": 11.996874934993684, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 63, "facts_retained": 8, "retention_rate": 0.12698412698412698, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.125, "amount": 0.07692307692307693, "status": 0.25, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 696}, "ledger_by_hop": {"1": {"facts_total": 63, "facts_retained": 9, "retention_rate": 0.14285714285714285, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.25, "amount": 0.07692307692307693, "status": 0.25, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 787}, "2": {"facts_total": 63, "facts_retained": 8, "retention_rate": 0.12698412698412698, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.125, "amount": 0.07692307692307693, "status": 0.25, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 696}}}, "reference_key_id": "keys/synthetic/syn_03027_2000.json", "task_scores": {"correct": 1.0, "abstained": 1.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "5de8bbf0fdf02bbceb67207371e6f5c0ec86d2b30877ec832e254d4b22f40895", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "4f5d4177a9dc907602e5718c", "attempt_id": "9c75de4c2aba", "run_group_id": "E4_core:A5|closed|qwen3b:h4", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:30:13.261493Z", "task_family": "synthetic", "document_id": "syn_03027_2000", "document_hash": "10326080979f71b625e4110c7c19767e5a752e267de5034f7e37d98cb6e06249", "question_id": "q00_single_fact", "split": "test", "cluster_id": "syn_03027_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 4, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1955, "query_tokens": 376, "message_tokens": 294, "message_utf8_bytes": 736, "cache_tensor_bytes": null, "serialized_payload_bytes": 736, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 18427.254083915614, "hops": {"1": 6880.990542005748, "2": 3875.456750043668, "3": 3842.124249902554, "4": 3828.6825419636443}, "receiver_prefill_ms": 143.47737503703684, "ttft_ms": 11.812917073257267, "decode_ms": 11.73145801294595}, "total_wall_ms": 18594.766833935864, "ttft_ms": 11.812917073257267, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 63, "facts_retained": 8, "retention_rate": 0.12698412698412698, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.125, "amount": 0.07692307692307693, "status": 0.25, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 696}, "ledger_by_hop": {"1": {"facts_total": 63, "facts_retained": 9, "retention_rate": 0.14285714285714285, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.25, "amount": 0.07692307692307693, "status": 0.25, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 787}, "2": {"facts_total": 63, "facts_retained": 8, "retention_rate": 0.12698412698412698, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.125, "amount": 0.07692307692307693, "status": 0.25, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 696}, "3": {"facts_total": 63, "facts_retained": 8, "retention_rate": 0.12698412698412698, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.125, "amount": 0.07692307692307693, "status": 0.25, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 696}, "4": {"facts_total": 63, "facts_retained": 8, "retention_rate": 0.12698412698412698, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.125, "amount": 0.07692307692307693, "status": 0.25, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 696}}}, "reference_key_id": "keys/synthetic/syn_03027_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "5de8bbf0fdf02bbceb67207371e6f5c0ec86d2b30877ec832e254d4b22f40895", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "ef0d4761efbfeb625f4e83a0", "attempt_id": "800fcb539538", "run_group_id": "E4_core:A5|closed|qwen3b:h4", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:30:13.446274Z", "task_family": "synthetic", "document_id": "syn_03027_2000", "document_hash": "10326080979f71b625e4110c7c19767e5a752e267de5034f7e37d98cb6e06249", "question_id": "q01_single_fact", "split": "test", "cluster_id": "syn_03027_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 4, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1955, "query_tokens": 383, "message_tokens": 294, "message_utf8_bytes": 736, "cache_tensor_bytes": null, "serialized_payload_bytes": 736, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 18427.254083915614, "hops": {"1": 6880.990542005748, "2": 3875.456750043668, "3": 3842.124249902554, "4": 3828.6825419636443}, "receiver_prefill_ms": 142.933415947482, "ttft_ms": 11.707625002600253, "decode_ms": 11.721999966539443}, "total_wall_ms": 18593.97570893634, "ttft_ms": 11.707625002600253, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 63, "facts_retained": 8, "retention_rate": 0.12698412698412698, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.125, "amount": 0.07692307692307693, "status": 0.25, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 696}, "ledger_by_hop": {"1": {"facts_total": 63, "facts_retained": 9, "retention_rate": 0.14285714285714285, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.25, "amount": 0.07692307692307693, "status": 0.25, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 787}, "2": {"facts_total": 63, "facts_retained": 8, "retention_rate": 0.12698412698412698, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.125, "amount": 0.07692307692307693, "status": 0.25, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 696}, "3": {"facts_total": 63, "facts_retained": 8, "retention_rate": 0.12698412698412698, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.125, "amount": 0.07692307692307693, "status": 0.25, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 696}, "4": {"facts_total": 63, "facts_retained": 8, "retention_rate": 0.12698412698412698, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.125, "amount": 0.07692307692307693, "status": 0.25, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 696}}}, "reference_key_id": "keys/synthetic/syn_03027_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "5de8bbf0fdf02bbceb67207371e6f5c0ec86d2b30877ec832e254d4b22f40895", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "acd7c806044d2865c211698c", "attempt_id": "09d0aa57434a", "run_group_id": "E4_core:A5|closed|qwen3b:h4", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:30:13.628620Z", "task_family": "synthetic", "document_id": "syn_03027_2000", "document_hash": "10326080979f71b625e4110c7c19767e5a752e267de5034f7e37d98cb6e06249", "question_id": "q02_exact_copy", "split": "test", "cluster_id": "syn_03027_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 4, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1955, "query_tokens": 382, "message_tokens": 294, "message_utf8_bytes": 736, "cache_tensor_bytes": null, "serialized_payload_bytes": 736, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 18427.254083915614, "hops": {"1": 6880.990542005748, "2": 3875.456750043668, "3": 3842.124249902554, "4": 3828.6825419636443}, "receiver_prefill_ms": 143.11829104553908, "ttft_ms": 11.913125053979456, "decode_ms": 11.759917018935084}, "total_wall_ms": 18594.487750902772, "ttft_ms": 11.913125053979456, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 63, "facts_retained": 8, "retention_rate": 0.12698412698412698, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.125, "amount": 0.07692307692307693, "status": 0.25, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 696}, "ledger_by_hop": {"1": {"facts_total": 63, "facts_retained": 9, "retention_rate": 0.14285714285714285, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.25, "amount": 0.07692307692307693, "status": 0.25, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 787}, "2": {"facts_total": 63, "facts_retained": 8, "retention_rate": 0.12698412698412698, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.125, "amount": 0.07692307692307693, "status": 0.25, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 696}, "3": {"facts_total": 63, "facts_retained": 8, "retention_rate": 0.12698412698412698, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.125, "amount": 0.07692307692307693, "status": 0.25, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 696}, "4": {"facts_total": 63, "facts_retained": 8, "retention_rate": 0.12698412698412698, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.125, "amount": 0.07692307692307693, "status": 0.25, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 696}}}, "reference_key_id": "keys/synthetic/syn_03027_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "5de8bbf0fdf02bbceb67207371e6f5c0ec86d2b30877ec832e254d4b22f40895", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "26e86c5173d8d85f6653ab36", "attempt_id": "889ee085ddd1", "run_group_id": "E4_core:A5|closed|qwen3b:h4", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:30:13.810524Z", "task_family": "synthetic", "document_id": "syn_03027_2000", "document_hash": "10326080979f71b625e4110c7c19767e5a752e267de5034f7e37d98cb6e06249", "question_id": "q03_exact_copy", "split": "test", "cluster_id": "syn_03027_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 4, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1955, "query_tokens": 377, "message_tokens": 294, "message_utf8_bytes": 736, "cache_tensor_bytes": null, "serialized_payload_bytes": 736, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 18427.254083915614, "hops": {"1": 6880.990542005748, "2": 3875.456750043668, "3": 3842.124249902554, "4": 3828.6825419636443}, "receiver_prefill_ms": 142.85570802167058, "ttft_ms": 11.718833004124463, "decode_ms": 105.01845797989517}, "total_wall_ms": 18687.208874849603, "ttft_ms": 11.718833004124463, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 9}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "Priyamvada Sørlie", "evidence_artifact": {"ledger": {"facts_total": 63, "facts_retained": 8, "retention_rate": 0.12698412698412698, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.125, "amount": 0.07692307692307693, "status": 0.25, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 696}, "ledger_by_hop": {"1": {"facts_total": 63, "facts_retained": 9, "retention_rate": 0.14285714285714285, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.25, "amount": 0.07692307692307693, "status": 0.25, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 787}, "2": {"facts_total": 63, "facts_retained": 8, "retention_rate": 0.12698412698412698, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.125, "amount": 0.07692307692307693, "status": 0.25, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 696}, "3": {"facts_total": 63, "facts_retained": 8, "retention_rate": 0.12698412698412698, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.125, "amount": 0.07692307692307693, "status": 0.25, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 696}, "4": {"facts_total": 63, "facts_retained": 8, "retention_rate": 0.12698412698412698, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.125, "amount": 0.07692307692307693, "status": 0.25, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 696}}}, "reference_key_id": "keys/synthetic/syn_03027_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "5de8bbf0fdf02bbceb67207371e6f5c0ec86d2b30877ec832e254d4b22f40895", "output_hash": "c5f554196ed23dd71ad995ec89f7142099d524dbb6534cc10d94a0c89d13e792"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "7ffaa203ef1d4b65f1d36afd", "attempt_id": "2ee3b872449e", "run_group_id": "E4_core:A5|closed|qwen3b:h4", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:30:14.085957Z", "task_family": "synthetic", "document_id": "syn_03027_2000", "document_hash": "10326080979f71b625e4110c7c19767e5a752e267de5034f7e37d98cb6e06249", "question_id": "q04_multi_hop", "split": "test", "cluster_id": "syn_03027_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 4, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1955, "query_tokens": 388, "message_tokens": 294, "message_utf8_bytes": 736, "cache_tensor_bytes": null, "serialized_payload_bytes": 736, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 18427.254083915614, "hops": {"1": 6880.990542005748, "2": 3875.456750043668, "3": 3842.124249902554, "4": 3828.6825419636443}, "receiver_prefill_ms": 160.48441699240357, "ttft_ms": 11.946541955694556, "decode_ms": 12.141457991674542}, "total_wall_ms": 18612.320083891973, "ttft_ms": 11.946541955694556, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 63, "facts_retained": 8, "retention_rate": 0.12698412698412698, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.125, "amount": 0.07692307692307693, "status": 0.25, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 696}, "ledger_by_hop": {"1": {"facts_total": 63, "facts_retained": 9, "retention_rate": 0.14285714285714285, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.25, "amount": 0.07692307692307693, "status": 0.25, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 787}, "2": {"facts_total": 63, "facts_retained": 8, "retention_rate": 0.12698412698412698, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.125, "amount": 0.07692307692307693, "status": 0.25, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 696}, "3": {"facts_total": 63, "facts_retained": 8, "retention_rate": 0.12698412698412698, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.125, "amount": 0.07692307692307693, "status": 0.25, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 696}, "4": {"facts_total": 63, "facts_retained": 8, "retention_rate": 0.12698412698412698, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.125, "amount": 0.07692307692307693, "status": 0.25, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 696}}}, "reference_key_id": "keys/synthetic/syn_03027_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "5de8bbf0fdf02bbceb67207371e6f5c0ec86d2b30877ec832e254d4b22f40895", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "ad4e51d3cdda20f19eaa8b88", "attempt_id": "c91a46c19aa2", "run_group_id": "E4_core:A5|closed|qwen3b:h4", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:30:14.287565Z", "task_family": "synthetic", "document_id": "syn_03027_2000", "document_hash": "10326080979f71b625e4110c7c19767e5a752e267de5034f7e37d98cb6e06249", "question_id": "q05_multi_hop", "split": "test", "cluster_id": "syn_03027_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 4, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1955, "query_tokens": 385, "message_tokens": 294, "message_utf8_bytes": 736, "cache_tensor_bytes": null, "serialized_payload_bytes": 736, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 18427.254083915614, "hops": {"1": 6880.990542005748, "2": 3875.456750043668, "3": 3842.124249902554, "4": 3828.6825419636443}, "receiver_prefill_ms": 142.1923750312999, "ttft_ms": 11.8681670865044, "decode_ms": 116.82800005655736}, "total_wall_ms": 18698.578042909503, "ttft_ms": 11.8681670865044, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 10}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "WT-26M-9227", "evidence_artifact": {"ledger": {"facts_total": 63, "facts_retained": 8, "retention_rate": 0.12698412698412698, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.125, "amount": 0.07692307692307693, "status": 0.25, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 696}, "ledger_by_hop": {"1": {"facts_total": 63, "facts_retained": 9, "retention_rate": 0.14285714285714285, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.25, "amount": 0.07692307692307693, "status": 0.25, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 787}, "2": {"facts_total": 63, "facts_retained": 8, "retention_rate": 0.12698412698412698, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.125, "amount": 0.07692307692307693, "status": 0.25, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 696}, "3": {"facts_total": 63, "facts_retained": 8, "retention_rate": 0.12698412698412698, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.125, "amount": 0.07692307692307693, "status": 0.25, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 696}, "4": {"facts_total": 63, "facts_retained": 8, "retention_rate": 0.12698412698412698, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.125, "amount": 0.07692307692307693, "status": 0.25, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 696}}}, "reference_key_id": "keys/synthetic/syn_03027_2000.json", "task_scores": {"correct": 0.0, "abstained": 0.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "5de8bbf0fdf02bbceb67207371e6f5c0ec86d2b30877ec832e254d4b22f40895", "output_hash": "ef2302e0ab6f5a6951b11463fea177ba766b9d438afcaf6e6101ebfa3ba38aba"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "c668cf3ab0edfcf4abfa601b", "attempt_id": "8d70f5c596d9", "run_group_id": "E4_core:A5|closed|qwen3b:h4", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:30:14.574961Z", "task_family": "synthetic", "document_id": "syn_03027_2000", "document_hash": "10326080979f71b625e4110c7c19767e5a752e267de5034f7e37d98cb6e06249", "question_id": "q06_negation", "split": "test", "cluster_id": "syn_03027_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 4, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1955, "query_tokens": 377, "message_tokens": 294, "message_utf8_bytes": 736, "cache_tensor_bytes": null, "serialized_payload_bytes": 736, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 18427.254083915614, "hops": {"1": 6880.990542005748, "2": 3875.456750043668, "3": 3842.124249902554, "4": 3828.6825419636443}, "receiver_prefill_ms": 142.60345802176744, "ttft_ms": 11.67291693855077, "decode_ms": 11.709416983649135}, "total_wall_ms": 18593.607750954106, "ttft_ms": 11.67291693855077, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 63, "facts_retained": 8, "retention_rate": 0.12698412698412698, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.125, "amount": 0.07692307692307693, "status": 0.25, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 696}, "ledger_by_hop": {"1": {"facts_total": 63, "facts_retained": 9, "retention_rate": 0.14285714285714285, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.25, "amount": 0.07692307692307693, "status": 0.25, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 787}, "2": {"facts_total": 63, "facts_retained": 8, "retention_rate": 0.12698412698412698, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.125, "amount": 0.07692307692307693, "status": 0.25, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 696}, "3": {"facts_total": 63, "facts_retained": 8, "retention_rate": 0.12698412698412698, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.125, "amount": 0.07692307692307693, "status": 0.25, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 696}, "4": {"facts_total": 63, "facts_retained": 8, "retention_rate": 0.12698412698412698, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.125, "amount": 0.07692307692307693, "status": 0.25, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 696}}}, "reference_key_id": "keys/synthetic/syn_03027_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "5de8bbf0fdf02bbceb67207371e6f5c0ec86d2b30877ec832e254d4b22f40895", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "0b51667ccdaeeb0ae4660d84", "attempt_id": "8eb52c29d750", "run_group_id": "E4_core:A5|closed|qwen3b:h4", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:30:14.756334Z", "task_family": "synthetic", "document_id": "syn_03027_2000", "document_hash": "10326080979f71b625e4110c7c19767e5a752e267de5034f7e37d98cb6e06249", "question_id": "q07_negation", "split": "test", "cluster_id": "syn_03027_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 4, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1955, "query_tokens": 378, "message_tokens": 294, "message_utf8_bytes": 736, "cache_tensor_bytes": null, "serialized_payload_bytes": 736, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 18427.254083915614, "hops": {"1": 6880.990542005748, "2": 3875.456750043668, "3": 3842.124249902554, "4": 3828.6825419636443}, "receiver_prefill_ms": 143.4065829962492, "ttft_ms": 11.7488750256598, "decode_ms": 11.724249925464392}, "total_wall_ms": 18594.483333989047, "ttft_ms": 11.7488750256598, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 63, "facts_retained": 8, "retention_rate": 0.12698412698412698, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.125, "amount": 0.07692307692307693, "status": 0.25, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 696}, "ledger_by_hop": {"1": {"facts_total": 63, "facts_retained": 9, "retention_rate": 0.14285714285714285, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.25, "amount": 0.07692307692307693, "status": 0.25, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 787}, "2": {"facts_total": 63, "facts_retained": 8, "retention_rate": 0.12698412698412698, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.125, "amount": 0.07692307692307693, "status": 0.25, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 696}, "3": {"facts_total": 63, "facts_retained": 8, "retention_rate": 0.12698412698412698, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.125, "amount": 0.07692307692307693, "status": 0.25, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 696}, "4": {"facts_total": 63, "facts_retained": 8, "retention_rate": 0.12698412698412698, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.125, "amount": 0.07692307692307693, "status": 0.25, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 696}}}, "reference_key_id": "keys/synthetic/syn_03027_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "5de8bbf0fdf02bbceb67207371e6f5c0ec86d2b30877ec832e254d4b22f40895", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "f4776219be502d00b1023137", "attempt_id": "5897f745b2d8", "run_group_id": "E4_core:A5|closed|qwen3b:h4", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:30:14.938340Z", "task_family": "synthetic", "document_id": "syn_03027_2000", "document_hash": "10326080979f71b625e4110c7c19767e5a752e267de5034f7e37d98cb6e06249", "question_id": "q08_revision", "split": "test", "cluster_id": "syn_03027_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 4, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1955, "query_tokens": 392, "message_tokens": 294, "message_utf8_bytes": 736, "cache_tensor_bytes": null, "serialized_payload_bytes": 736, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 18427.254083915614, "hops": {"1": 6880.990542005748, "2": 3875.456750043668, "3": 3842.124249902554, "4": 3828.6825419636443}, "receiver_prefill_ms": 161.03537497110665, "ttft_ms": 11.804500012658536, "decode_ms": 105.1398329436779}, "total_wall_ms": 18705.592874903232, "ttft_ms": 11.804500012658536, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 9}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "6,914,937", "evidence_artifact": {"ledger": {"facts_total": 63, "facts_retained": 8, "retention_rate": 0.12698412698412698, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.125, "amount": 0.07692307692307693, "status": 0.25, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 696}, "ledger_by_hop": {"1": {"facts_total": 63, "facts_retained": 9, "retention_rate": 0.14285714285714285, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.25, "amount": 0.07692307692307693, "status": 0.25, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 787}, "2": {"facts_total": 63, "facts_retained": 8, "retention_rate": 0.12698412698412698, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.125, "amount": 0.07692307692307693, "status": 0.25, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 696}, "3": {"facts_total": 63, "facts_retained": 8, "retention_rate": 0.12698412698412698, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.125, "amount": 0.07692307692307693, "status": 0.25, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 696}, "4": {"facts_total": 63, "facts_retained": 8, "retention_rate": 0.12698412698412698, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.125, "amount": 0.07692307692307693, "status": 0.25, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 696}}}, "reference_key_id": "keys/synthetic/syn_03027_2000.json", "task_scores": {"correct": 0.0, "abstained": 0.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "5de8bbf0fdf02bbceb67207371e6f5c0ec86d2b30877ec832e254d4b22f40895", "output_hash": "c56851abfdbf8851226a1e71c4911c616a691c490a883eb754e61bac551d004c"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "1b6ef64ee6a56ca1233d7517", "attempt_id": "02efce2f7fe0", "run_group_id": "E4_core:A5|closed|qwen3b:h4", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:30:15.231234Z", "task_family": "synthetic", "document_id": "syn_03027_2000", "document_hash": "10326080979f71b625e4110c7c19767e5a752e267de5034f7e37d98cb6e06249", "question_id": "q09_revision", "split": "test", "cluster_id": "syn_03027_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 4, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1955, "query_tokens": 385, "message_tokens": 294, "message_utf8_bytes": 736, "cache_tensor_bytes": null, "serialized_payload_bytes": 736, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 18427.254083915614, "hops": {"1": 6880.990542005748, "2": 3875.456750043668, "3": 3842.124249902554, "4": 3828.6825419636443}, "receiver_prefill_ms": 142.64912507496774, "ttft_ms": 11.904458049684763, "decode_ms": 81.76762494258583}, "total_wall_ms": 18663.90858392697, "ttft_ms": 11.904458049684763, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 7}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "6914937", "evidence_artifact": {"ledger": {"facts_total": 63, "facts_retained": 8, "retention_rate": 0.12698412698412698, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.125, "amount": 0.07692307692307693, "status": 0.25, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 696}, "ledger_by_hop": {"1": {"facts_total": 63, "facts_retained": 9, "retention_rate": 0.14285714285714285, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.25, "amount": 0.07692307692307693, "status": 0.25, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 787}, "2": {"facts_total": 63, "facts_retained": 8, "retention_rate": 0.12698412698412698, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.125, "amount": 0.07692307692307693, "status": 0.25, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 696}, "3": {"facts_total": 63, "facts_retained": 8, "retention_rate": 0.12698412698412698, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.125, "amount": 0.07692307692307693, "status": 0.25, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 696}, "4": {"facts_total": 63, "facts_retained": 8, "retention_rate": 0.12698412698412698, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.125, "amount": 0.07692307692307693, "status": 0.25, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 696}}}, "reference_key_id": "keys/synthetic/syn_03027_2000.json", "task_scores": {"correct": 0.0, "abstained": 0.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "5de8bbf0fdf02bbceb67207371e6f5c0ec86d2b30877ec832e254d4b22f40895", "output_hash": "534d40ded32f784c68307dd2386faed2a27bf39dcc27321ee6985e42ccd285e5"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "196de3234c02bae78e1c224e", "attempt_id": "90583394cd53", "run_group_id": "E4_core:A5|closed|qwen3b:h4", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:30:15.482626Z", "task_family": "synthetic", "document_id": "syn_03027_2000", "document_hash": "10326080979f71b625e4110c7c19767e5a752e267de5034f7e37d98cb6e06249", "question_id": "q10_temporal", "split": "test", "cluster_id": "syn_03027_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 4, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1955, "query_tokens": 395, "message_tokens": 294, "message_utf8_bytes": 736, "cache_tensor_bytes": null, "serialized_payload_bytes": 736, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 18427.254083915614, "hops": {"1": 6880.990542005748, "2": 3875.456750043668, "3": 3842.124249902554, "4": 3828.6825419636443}, "receiver_prefill_ms": 161.09754191711545, "ttft_ms": 11.719250003807247, "decode_ms": 11.691750027239323}, "total_wall_ms": 18612.112749833614, "ttft_ms": 11.719250003807247, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 63, "facts_retained": 8, "retention_rate": 0.12698412698412698, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.125, "amount": 0.07692307692307693, "status": 0.25, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 696}, "ledger_by_hop": {"1": {"facts_total": 63, "facts_retained": 9, "retention_rate": 0.14285714285714285, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.25, "amount": 0.07692307692307693, "status": 0.25, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 787}, "2": {"facts_total": 63, "facts_retained": 8, "retention_rate": 0.12698412698412698, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.125, "amount": 0.07692307692307693, "status": 0.25, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 696}, "3": {"facts_total": 63, "facts_retained": 8, "retention_rate": 0.12698412698412698, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.125, "amount": 0.07692307692307693, "status": 0.25, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 696}, "4": {"facts_total": 63, "facts_retained": 8, "retention_rate": 0.12698412698412698, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.125, "amount": 0.07692307692307693, "status": 0.25, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 696}}}, "reference_key_id": "keys/synthetic/syn_03027_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "5de8bbf0fdf02bbceb67207371e6f5c0ec86d2b30877ec832e254d4b22f40895", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "ae88c3e3f1a5578f7a126f8f", "attempt_id": "11cfa484a26b", "run_group_id": "E4_core:A5|closed|qwen3b:h4", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:30:15.682528Z", "task_family": "synthetic", "document_id": "syn_03027_2000", "document_hash": "10326080979f71b625e4110c7c19767e5a752e267de5034f7e37d98cb6e06249", "question_id": "q11_aggregation", "split": "test", "cluster_id": "syn_03027_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 4, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1955, "query_tokens": 388, "message_tokens": 294, "message_utf8_bytes": 736, "cache_tensor_bytes": null, "serialized_payload_bytes": 736, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 18427.254083915614, "hops": {"1": 6880.990542005748, "2": 3875.456750043668, "3": 3842.124249902554, "4": 3828.6825419636443}, "receiver_prefill_ms": 160.5674580205232, "ttft_ms": 11.716125067323446, "decode_ms": 81.88541699200869}, "total_wall_ms": 18681.780041893944, "ttft_ms": 11.716125067323446, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 7}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "6914937", "evidence_artifact": {"ledger": {"facts_total": 63, "facts_retained": 8, "retention_rate": 0.12698412698412698, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.125, "amount": 0.07692307692307693, "status": 0.25, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 696}, "ledger_by_hop": {"1": {"facts_total": 63, "facts_retained": 9, "retention_rate": 0.14285714285714285, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.25, "amount": 0.07692307692307693, "status": 0.25, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 787}, "2": {"facts_total": 63, "facts_retained": 8, "retention_rate": 0.12698412698412698, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.125, "amount": 0.07692307692307693, "status": 0.25, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 696}, "3": {"facts_total": 63, "facts_retained": 8, "retention_rate": 0.12698412698412698, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.125, "amount": 0.07692307692307693, "status": 0.25, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 696}, "4": {"facts_total": 63, "facts_retained": 8, "retention_rate": 0.12698412698412698, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.125, "amount": 0.07692307692307693, "status": 0.25, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 696}}}, "reference_key_id": "keys/synthetic/syn_03027_2000.json", "task_scores": {"correct": 0.0, "abstained": 0.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "5de8bbf0fdf02bbceb67207371e6f5c0ec86d2b30877ec832e254d4b22f40895", "output_hash": "534d40ded32f784c68307dd2386faed2a27bf39dcc27321ee6985e42ccd285e5"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "ddd746d82af591115750f48f", "attempt_id": "e072aa26f2ad", "run_group_id": "E4_core:A5|closed|qwen3b:h4", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:30:15.952026Z", "task_family": "synthetic", "document_id": "syn_03027_2000", "document_hash": "10326080979f71b625e4110c7c19767e5a752e267de5034f7e37d98cb6e06249", "question_id": "q12_unit_conversion", "split": "test", "cluster_id": "syn_03027_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 4, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1955, "query_tokens": 385, "message_tokens": 294, "message_utf8_bytes": 736, "cache_tensor_bytes": null, "serialized_payload_bytes": 736, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 18427.254083915614, "hops": {"1": 6880.990542005748, "2": 3875.456750043668, "3": 3842.124249902554, "4": 3828.6825419636443}, "receiver_prefill_ms": 142.90795801207423, "ttft_ms": 11.818624916486442, "decode_ms": 11.799416039139032}, "total_wall_ms": 18594.124916940928, "ttft_ms": 11.818624916486442, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 63, "facts_retained": 8, "retention_rate": 0.12698412698412698, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.125, "amount": 0.07692307692307693, "status": 0.25, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 696}, "ledger_by_hop": {"1": {"facts_total": 63, "facts_retained": 9, "retention_rate": 0.14285714285714285, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.25, "amount": 0.07692307692307693, "status": 0.25, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 787}, "2": {"facts_total": 63, "facts_retained": 8, "retention_rate": 0.12698412698412698, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.125, "amount": 0.07692307692307693, "status": 0.25, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 696}, "3": {"facts_total": 63, "facts_retained": 8, "retention_rate": 0.12698412698412698, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.125, "amount": 0.07692307692307693, "status": 0.25, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 696}, "4": {"facts_total": 63, "facts_retained": 8, "retention_rate": 0.12698412698412698, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.125, "amount": 0.07692307692307693, "status": 0.25, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 696}}}, "reference_key_id": "keys/synthetic/syn_03027_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "5de8bbf0fdf02bbceb67207371e6f5c0ec86d2b30877ec832e254d4b22f40895", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "ce8bd9c159c29e14f8bf882d", "attempt_id": "e1ccebf41b6f", "run_group_id": "E4_core:A5|closed|qwen3b:h4", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:30:16.133793Z", "task_family": "synthetic", "document_id": "syn_03027_2000", "document_hash": "10326080979f71b625e4110c7c19767e5a752e267de5034f7e37d98cb6e06249", "question_id": "q13_unanswerable", "split": "test", "cluster_id": "syn_03027_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 4, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1955, "query_tokens": 388, "message_tokens": 294, "message_utf8_bytes": 736, "cache_tensor_bytes": null, "serialized_payload_bytes": 736, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 18427.254083915614, "hops": {"1": 6880.990542005748, "2": 3875.456750043668, "3": 3842.124249902554, "4": 3828.6825419636443}, "receiver_prefill_ms": 161.0071249306202, "ttft_ms": 11.830665986053646, "decode_ms": 11.767374933697283}, "total_wall_ms": 18612.18333384022, "ttft_ms": 11.830665986053646, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 63, "facts_retained": 8, "retention_rate": 0.12698412698412698, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.125, "amount": 0.07692307692307693, "status": 0.25, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 696}, "ledger_by_hop": {"1": {"facts_total": 63, "facts_retained": 9, "retention_rate": 0.14285714285714285, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.25, "amount": 0.07692307692307693, "status": 0.25, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 787}, "2": {"facts_total": 63, "facts_retained": 8, "retention_rate": 0.12698412698412698, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.125, "amount": 0.07692307692307693, "status": 0.25, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 696}, "3": {"facts_total": 63, "facts_retained": 8, "retention_rate": 0.12698412698412698, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.125, "amount": 0.07692307692307693, "status": 0.25, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 696}, "4": {"facts_total": 63, "facts_retained": 8, "retention_rate": 0.12698412698412698, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.125, "amount": 0.07692307692307693, "status": 0.25, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 696}}}, "reference_key_id": "keys/synthetic/syn_03027_2000.json", "task_scores": {"correct": 1.0, "abstained": 1.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "5de8bbf0fdf02bbceb67207371e6f5c0ec86d2b30877ec832e254d4b22f40895", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "fc70f0d4f80b35e40c9b8113", "attempt_id": "77a828a66882", "run_group_id": "E4_core:A5|closed|qwen3b:h4", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:30:16.333405Z", "task_family": "synthetic", "document_id": "syn_03027_2000", "document_hash": "10326080979f71b625e4110c7c19767e5a752e267de5034f7e37d98cb6e06249", "question_id": "q14_unanswerable", "split": "test", "cluster_id": "syn_03027_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 4, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1955, "query_tokens": 390, "message_tokens": 294, "message_utf8_bytes": 736, "cache_tensor_bytes": null, "serialized_payload_bytes": 736, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 18427.254083915614, "hops": {"1": 6880.990542005748, "2": 3875.456750043668, "3": 3842.124249902554, "4": 3828.6825419636443}, "receiver_prefill_ms": 161.34524997323751, "ttft_ms": 11.964292032644153, "decode_ms": 11.853417032398283}, "total_wall_ms": 18612.765374942683, "ttft_ms": 11.964292032644153, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 63, "facts_retained": 8, "retention_rate": 0.12698412698412698, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.125, "amount": 0.07692307692307693, "status": 0.25, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 696}, "ledger_by_hop": {"1": {"facts_total": 63, "facts_retained": 9, "retention_rate": 0.14285714285714285, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.25, "amount": 0.07692307692307693, "status": 0.25, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 787}, "2": {"facts_total": 63, "facts_retained": 8, "retention_rate": 0.12698412698412698, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.125, "amount": 0.07692307692307693, "status": 0.25, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 696}, "3": {"facts_total": 63, "facts_retained": 8, "retention_rate": 0.12698412698412698, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.125, "amount": 0.07692307692307693, "status": 0.25, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 696}, "4": {"facts_total": 63, "facts_retained": 8, "retention_rate": 0.12698412698412698, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.125, "amount": 0.07692307692307693, "status": 0.25, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 696}}}, "reference_key_id": "keys/synthetic/syn_03027_2000.json", "task_scores": {"correct": 1.0, "abstained": 1.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "5de8bbf0fdf02bbceb67207371e6f5c0ec86d2b30877ec832e254d4b22f40895", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "b173341cbe379c0d0a7b28b3", "attempt_id": "d975d197e761", "run_group_id": "E4_core:A5|closed|qwen3b:h8", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:30:31.882605Z", "task_family": "synthetic", "document_id": "syn_03027_2000", "document_hash": "10326080979f71b625e4110c7c19767e5a752e267de5034f7e37d98cb6e06249", "question_id": "q00_single_fact", "split": "test", "cluster_id": "syn_03027_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 8, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1955, "query_tokens": 376, "message_tokens": 294, "message_utf8_bytes": 736, "cache_tensor_bytes": null, "serialized_payload_bytes": 736, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 33771.53900091071, "hops": {"1": 6880.990542005748, "2": 3875.456750043668, "3": 3842.124249902554, "4": 3828.6825419636443, "5": 3843.770332983695, "6": 3833.0124999629334, "7": 3827.791125047952, "8": 3839.710959000513}, "receiver_prefill_ms": 143.2176660746336, "ttft_ms": 11.783457943238318, "decode_ms": 11.821625055745244}, "total_wall_ms": 33938.74595896341, "ttft_ms": 11.783457943238318, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 63, "facts_retained": 8, "retention_rate": 0.12698412698412698, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.125, "amount": 0.07692307692307693, "status": 0.25, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 696}, "ledger_by_hop": {"1": {"facts_total": 63, "facts_retained": 9, "retention_rate": 0.14285714285714285, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.25, "amount": 0.07692307692307693, "status": 0.25, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 787}, "2": {"facts_total": 63, "facts_retained": 8, "retention_rate": 0.12698412698412698, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.125, "amount": 0.07692307692307693, "status": 0.25, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 696}, "3": {"facts_total": 63, "facts_retained": 8, "retention_rate": 0.12698412698412698, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.125, "amount": 0.07692307692307693, "status": 0.25, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 696}, "4": {"facts_total": 63, "facts_retained": 8, "retention_rate": 0.12698412698412698, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.125, "amount": 0.07692307692307693, "status": 0.25, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 696}, "5": {"facts_total": 63, "facts_retained": 8, "retention_rate": 0.12698412698412698, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.125, "amount": 0.07692307692307693, "status": 0.25, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 696}, "6": {"facts_total": 63, "facts_retained": 8, "retention_rate": 0.12698412698412698, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.125, "amount": 0.07692307692307693, "status": 0.25, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 696}, "7": {"facts_total": 63, "facts_retained": 8, "retention_rate": 0.12698412698412698, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.125, "amount": 0.07692307692307693, "status": 0.25, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 696}, "8": {"facts_total": 63, "facts_retained": 8, "retention_rate": 0.12698412698412698, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.125, "amount": 0.07692307692307693, "status": 0.25, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 696}}}, "reference_key_id": "keys/synthetic/syn_03027_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "5de8bbf0fdf02bbceb67207371e6f5c0ec86d2b30877ec832e254d4b22f40895", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "44220c9c7d7054b6eaf60f0c", "attempt_id": "940c762f4e53", "run_group_id": "E4_core:A5|closed|qwen3b:h8", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:30:32.065894Z", "task_family": "synthetic", "document_id": "syn_03027_2000", "document_hash": "10326080979f71b625e4110c7c19767e5a752e267de5034f7e37d98cb6e06249", "question_id": "q01_single_fact", "split": "test", "cluster_id": "syn_03027_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 8, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1955, "query_tokens": 383, "message_tokens": 294, "message_utf8_bytes": 736, "cache_tensor_bytes": null, "serialized_payload_bytes": 736, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 33771.53900091071, "hops": {"1": 6880.990542005748, "2": 3875.456750043668, "3": 3842.124249902554, "4": 3828.6825419636443, "5": 3843.770332983695, "6": 3833.0124999629334, "7": 3827.791125047952, "8": 3839.710959000513}, "receiver_prefill_ms": 143.8692919909954, "ttft_ms": 11.777416919358075, "decode_ms": 11.827833019196987}, "total_wall_ms": 33939.367875806056, "ttft_ms": 11.777416919358075, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 63, "facts_retained": 8, "retention_rate": 0.12698412698412698, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.125, "amount": 0.07692307692307693, "status": 0.25, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 696}, "ledger_by_hop": {"1": {"facts_total": 63, "facts_retained": 9, "retention_rate": 0.14285714285714285, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.25, "amount": 0.07692307692307693, "status": 0.25, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 787}, "2": {"facts_total": 63, "facts_retained": 8, "retention_rate": 0.12698412698412698, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.125, "amount": 0.07692307692307693, "status": 0.25, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 696}, "3": {"facts_total": 63, "facts_retained": 8, "retention_rate": 0.12698412698412698, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.125, "amount": 0.07692307692307693, "status": 0.25, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 696}, "4": {"facts_total": 63, "facts_retained": 8, "retention_rate": 0.12698412698412698, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.125, "amount": 0.07692307692307693, "status": 0.25, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 696}, "5": {"facts_total": 63, "facts_retained": 8, "retention_rate": 0.12698412698412698, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.125, "amount": 0.07692307692307693, "status": 0.25, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 696}, "6": {"facts_total": 63, "facts_retained": 8, "retention_rate": 0.12698412698412698, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.125, "amount": 0.07692307692307693, "status": 0.25, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 696}, "7": {"facts_total": 63, "facts_retained": 8, "retention_rate": 0.12698412698412698, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.125, "amount": 0.07692307692307693, "status": 0.25, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 696}, "8": {"facts_total": 63, "facts_retained": 8, "retention_rate": 0.12698412698412698, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.125, "amount": 0.07692307692307693, "status": 0.25, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 696}}}, "reference_key_id": "keys/synthetic/syn_03027_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "5de8bbf0fdf02bbceb67207371e6f5c0ec86d2b30877ec832e254d4b22f40895", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "86f81738fcaa1956a8116c38", "attempt_id": "e3b7dddcba8e", "run_group_id": "E4_core:A5|closed|qwen3b:h8", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:30:32.249212Z", "task_family": "synthetic", "document_id": "syn_03027_2000", "document_hash": "10326080979f71b625e4110c7c19767e5a752e267de5034f7e37d98cb6e06249", "question_id": "q02_exact_copy", "split": "test", "cluster_id": "syn_03027_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 8, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1955, "query_tokens": 382, "message_tokens": 294, "message_utf8_bytes": 736, "cache_tensor_bytes": null, "serialized_payload_bytes": 736, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 33771.53900091071, "hops": {"1": 6880.990542005748, "2": 3875.456750043668, "3": 3842.124249902554, "4": 3828.6825419636443, "5": 3843.770332983695, "6": 3833.0124999629334, "7": 3827.791125047952, "8": 3839.710959000513}, "receiver_prefill_ms": 143.66883295588195, "ttft_ms": 11.724208015948534, "decode_ms": 12.064500013366342}, "total_wall_ms": 33939.35116683133, "ttft_ms": 11.724208015948534, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 63, "facts_retained": 8, "retention_rate": 0.12698412698412698, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.125, "amount": 0.07692307692307693, "status": 0.25, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 696}, "ledger_by_hop": {"1": {"facts_total": 63, "facts_retained": 9, "retention_rate": 0.14285714285714285, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.25, "amount": 0.07692307692307693, "status": 0.25, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 787}, "2": {"facts_total": 63, "facts_retained": 8, "retention_rate": 0.12698412698412698, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.125, "amount": 0.07692307692307693, "status": 0.25, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 696}, "3": {"facts_total": 63, "facts_retained": 8, "retention_rate": 0.12698412698412698, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.125, "amount": 0.07692307692307693, "status": 0.25, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 696}, "4": {"facts_total": 63, "facts_retained": 8, "retention_rate": 0.12698412698412698, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.125, "amount": 0.07692307692307693, "status": 0.25, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 696}, "5": {"facts_total": 63, "facts_retained": 8, "retention_rate": 0.12698412698412698, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.125, "amount": 0.07692307692307693, "status": 0.25, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 696}, "6": {"facts_total": 63, "facts_retained": 8, "retention_rate": 0.12698412698412698, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.125, "amount": 0.07692307692307693, "status": 0.25, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 696}, "7": {"facts_total": 63, "facts_retained": 8, "retention_rate": 0.12698412698412698, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.125, "amount": 0.07692307692307693, "status": 0.25, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 696}, "8": {"facts_total": 63, "facts_retained": 8, "retention_rate": 0.12698412698412698, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.125, "amount": 0.07692307692307693, "status": 0.25, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 696}}}, "reference_key_id": "keys/synthetic/syn_03027_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "5de8bbf0fdf02bbceb67207371e6f5c0ec86d2b30877ec832e254d4b22f40895", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "a8f9525849fb21de402671dc", "attempt_id": "c8d0a59299ba", "run_group_id": "E4_core:A5|closed|qwen3b:h8", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:30:32.432388Z", "task_family": "synthetic", "document_id": "syn_03027_2000", "document_hash": "10326080979f71b625e4110c7c19767e5a752e267de5034f7e37d98cb6e06249", "question_id": "q03_exact_copy", "split": "test", "cluster_id": "syn_03027_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 8, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1955, "query_tokens": 377, "message_tokens": 294, "message_utf8_bytes": 736, "cache_tensor_bytes": null, "serialized_payload_bytes": 736, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 33771.53900091071, "hops": {"1": 6880.990542005748, "2": 3875.456750043668, "3": 3842.124249902554, "4": 3828.6825419636443, "5": 3843.770332983695, "6": 3833.0124999629334, "7": 3827.791125047952, "8": 3839.710959000513}, "receiver_prefill_ms": 143.6260410118848, "ttft_ms": 11.769667034968734, "decode_ms": 106.04074993170798}, "total_wall_ms": 34033.34854298737, "ttft_ms": 11.769667034968734, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 9}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "Priyamvada Sørlie", "evidence_artifact": {"ledger": {"facts_total": 63, "facts_retained": 8, "retention_rate": 0.12698412698412698, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.125, "amount": 0.07692307692307693, "status": 0.25, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 696}, "ledger_by_hop": {"1": {"facts_total": 63, "facts_retained": 9, "retention_rate": 0.14285714285714285, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.25, "amount": 0.07692307692307693, "status": 0.25, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 787}, "2": {"facts_total": 63, "facts_retained": 8, "retention_rate": 0.12698412698412698, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.125, "amount": 0.07692307692307693, "status": 0.25, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 696}, "3": {"facts_total": 63, "facts_retained": 8, "retention_rate": 0.12698412698412698, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.125, "amount": 0.07692307692307693, "status": 0.25, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 696}, "4": {"facts_total": 63, "facts_retained": 8, "retention_rate": 0.12698412698412698, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.125, "amount": 0.07692307692307693, "status": 0.25, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 696}, "5": {"facts_total": 63, "facts_retained": 8, "retention_rate": 0.12698412698412698, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.125, "amount": 0.07692307692307693, "status": 0.25, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 696}, "6": {"facts_total": 63, "facts_retained": 8, "retention_rate": 0.12698412698412698, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.125, "amount": 0.07692307692307693, "status": 0.25, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 696}, "7": {"facts_total": 63, "facts_retained": 8, "retention_rate": 0.12698412698412698, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.125, "amount": 0.07692307692307693, "status": 0.25, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 696}, "8": {"facts_total": 63, "facts_retained": 8, "retention_rate": 0.12698412698412698, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.125, "amount": 0.07692307692307693, "status": 0.25, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 696}}}, "reference_key_id": "keys/synthetic/syn_03027_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "5de8bbf0fdf02bbceb67207371e6f5c0ec86d2b30877ec832e254d4b22f40895", "output_hash": "c5f554196ed23dd71ad995ec89f7142099d524dbb6534cc10d94a0c89d13e792"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "beae63f4a02dac2ef15d56ca", "attempt_id": "10b189f7899b", "run_group_id": "E4_core:A5|closed|qwen3b:h8", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:30:32.709415Z", "task_family": "synthetic", "document_id": "syn_03027_2000", "document_hash": "10326080979f71b625e4110c7c19767e5a752e267de5034f7e37d98cb6e06249", "question_id": "q04_multi_hop", "split": "test", "cluster_id": "syn_03027_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 8, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1955, "query_tokens": 388, "message_tokens": 294, "message_utf8_bytes": 736, "cache_tensor_bytes": null, "serialized_payload_bytes": 736, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 33771.53900091071, "hops": {"1": 6880.990542005748, "2": 3875.456750043668, "3": 3842.124249902554, "4": 3828.6825419636443, "5": 3843.770332983695, "6": 3833.0124999629334, "7": 3827.791125047952, "8": 3839.710959000513}, "receiver_prefill_ms": 160.79454205464572, "ttft_ms": 11.829916038550436, "decode_ms": 11.93750009406358}, "total_wall_ms": 33956.46762591787, "ttft_ms": 11.829916038550436, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 63, "facts_retained": 8, "retention_rate": 0.12698412698412698, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.125, "amount": 0.07692307692307693, "status": 0.25, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 696}, "ledger_by_hop": {"1": {"facts_total": 63, "facts_retained": 9, "retention_rate": 0.14285714285714285, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.25, "amount": 0.07692307692307693, "status": 0.25, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 787}, "2": {"facts_total": 63, "facts_retained": 8, "retention_rate": 0.12698412698412698, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.125, "amount": 0.07692307692307693, "status": 0.25, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 696}, "3": {"facts_total": 63, "facts_retained": 8, "retention_rate": 0.12698412698412698, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.125, "amount": 0.07692307692307693, "status": 0.25, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 696}, "4": {"facts_total": 63, "facts_retained": 8, "retention_rate": 0.12698412698412698, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.125, "amount": 0.07692307692307693, "status": 0.25, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 696}, "5": {"facts_total": 63, "facts_retained": 8, "retention_rate": 0.12698412698412698, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.125, "amount": 0.07692307692307693, "status": 0.25, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 696}, "6": {"facts_total": 63, "facts_retained": 8, "retention_rate": 0.12698412698412698, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.125, "amount": 0.07692307692307693, "status": 0.25, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 696}, "7": {"facts_total": 63, "facts_retained": 8, "retention_rate": 0.12698412698412698, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.125, "amount": 0.07692307692307693, "status": 0.25, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 696}, "8": {"facts_total": 63, "facts_retained": 8, "retention_rate": 0.12698412698412698, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.125, "amount": 0.07692307692307693, "status": 0.25, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 696}}}, "reference_key_id": "keys/synthetic/syn_03027_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "5de8bbf0fdf02bbceb67207371e6f5c0ec86d2b30877ec832e254d4b22f40895", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "dd92271a33962db624b25f6b", "attempt_id": "1484da0a2689", "run_group_id": "E4_core:A5|closed|qwen3b:h8", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:30:32.909683Z", "task_family": "synthetic", "document_id": "syn_03027_2000", "document_hash": "10326080979f71b625e4110c7c19767e5a752e267de5034f7e37d98cb6e06249", "question_id": "q05_multi_hop", "split": "test", "cluster_id": "syn_03027_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 8, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1955, "query_tokens": 385, "message_tokens": 294, "message_utf8_bytes": 736, "cache_tensor_bytes": null, "serialized_payload_bytes": 736, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 33771.53900091071, "hops": {"1": 6880.990542005748, "2": 3875.456750043668, "3": 3842.124249902554, "4": 3828.6825419636443, "5": 3843.770332983695, "6": 3833.0124999629334, "7": 3827.791125047952, "8": 3839.710959000513}, "receiver_prefill_ms": 142.5096249440685, "ttft_ms": 11.806208989582956, "decode_ms": 118.01516695413738}, "total_wall_ms": 34044.2333339015, "ttft_ms": 11.806208989582956, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 10}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "WT-26M-9227", "evidence_artifact": {"ledger": {"facts_total": 63, "facts_retained": 8, "retention_rate": 0.12698412698412698, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.125, "amount": 0.07692307692307693, "status": 0.25, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 696}, "ledger_by_hop": {"1": {"facts_total": 63, "facts_retained": 9, "retention_rate": 0.14285714285714285, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.25, "amount": 0.07692307692307693, "status": 0.25, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 787}, "2": {"facts_total": 63, "facts_retained": 8, "retention_rate": 0.12698412698412698, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.125, "amount": 0.07692307692307693, "status": 0.25, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 696}, "3": {"facts_total": 63, "facts_retained": 8, "retention_rate": 0.12698412698412698, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.125, "amount": 0.07692307692307693, "status": 0.25, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 696}, "4": {"facts_total": 63, "facts_retained": 8, "retention_rate": 0.12698412698412698, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.125, "amount": 0.07692307692307693, "status": 0.25, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 696}, "5": {"facts_total": 63, "facts_retained": 8, "retention_rate": 0.12698412698412698, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.125, "amount": 0.07692307692307693, "status": 0.25, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 696}, "6": {"facts_total": 63, "facts_retained": 8, "retention_rate": 0.12698412698412698, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.125, "amount": 0.07692307692307693, "status": 0.25, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 696}, "7": {"facts_total": 63, "facts_retained": 8, "retention_rate": 0.12698412698412698, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.125, "amount": 0.07692307692307693, "status": 0.25, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 696}, "8": {"facts_total": 63, "facts_retained": 8, "retention_rate": 0.12698412698412698, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.125, "amount": 0.07692307692307693, "status": 0.25, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 696}}}, "reference_key_id": "keys/synthetic/syn_03027_2000.json", "task_scores": {"correct": 0.0, "abstained": 0.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "5de8bbf0fdf02bbceb67207371e6f5c0ec86d2b30877ec832e254d4b22f40895", "output_hash": "ef2302e0ab6f5a6951b11463fea177ba766b9d438afcaf6e6101ebfa3ba38aba"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "6ce8546e294d327883c78944", "attempt_id": "f1521c6475b2", "run_group_id": "E4_core:A5|closed|qwen3b:h8", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:30:33.197715Z", "task_family": "synthetic", "document_id": "syn_03027_2000", "document_hash": "10326080979f71b625e4110c7c19767e5a752e267de5034f7e37d98cb6e06249", "question_id": "q06_negation", "split": "test", "cluster_id": "syn_03027_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 8, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1955, "query_tokens": 377, "message_tokens": 294, "message_utf8_bytes": 736, "cache_tensor_bytes": null, "serialized_payload_bytes": 736, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 33771.53900091071, "hops": {"1": 6880.990542005748, "2": 3875.456750043668, "3": 3842.124249902554, "4": 3828.6825419636443, "5": 3843.770332983695, "6": 3833.0124999629334, "7": 3827.791125047952, "8": 3839.710959000513}, "receiver_prefill_ms": 143.50408292375505, "ttft_ms": 11.811125092208385, "decode_ms": 11.785084032453597}, "total_wall_ms": 33938.9966679737, "ttft_ms": 11.811125092208385, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 63, "facts_retained": 8, "retention_rate": 0.12698412698412698, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.125, "amount": 0.07692307692307693, "status": 0.25, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 696}, "ledger_by_hop": {"1": {"facts_total": 63, "facts_retained": 9, "retention_rate": 0.14285714285714285, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.25, "amount": 0.07692307692307693, "status": 0.25, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 787}, "2": {"facts_total": 63, "facts_retained": 8, "retention_rate": 0.12698412698412698, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.125, "amount": 0.07692307692307693, "status": 0.25, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 696}, "3": {"facts_total": 63, "facts_retained": 8, "retention_rate": 0.12698412698412698, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.125, "amount": 0.07692307692307693, "status": 0.25, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 696}, "4": {"facts_total": 63, "facts_retained": 8, "retention_rate": 0.12698412698412698, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.125, "amount": 0.07692307692307693, "status": 0.25, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 696}, "5": {"facts_total": 63, "facts_retained": 8, "retention_rate": 0.12698412698412698, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.125, "amount": 0.07692307692307693, "status": 0.25, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 696}, "6": {"facts_total": 63, "facts_retained": 8, "retention_rate": 0.12698412698412698, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.125, "amount": 0.07692307692307693, "status": 0.25, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 696}, "7": {"facts_total": 63, "facts_retained": 8, "retention_rate": 0.12698412698412698, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.125, "amount": 0.07692307692307693, "status": 0.25, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 696}, "8": {"facts_total": 63, "facts_retained": 8, "retention_rate": 0.12698412698412698, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.125, "amount": 0.07692307692307693, "status": 0.25, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 696}}}, "reference_key_id": "keys/synthetic/syn_03027_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "5de8bbf0fdf02bbceb67207371e6f5c0ec86d2b30877ec832e254d4b22f40895", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "14e2316bae62e3150e5da7ed", "attempt_id": "f22028da1af3", "run_group_id": "E4_core:A5|closed|qwen3b:h8", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:30:33.380636Z", "task_family": "synthetic", "document_id": "syn_03027_2000", "document_hash": "10326080979f71b625e4110c7c19767e5a752e267de5034f7e37d98cb6e06249", "question_id": "q07_negation", "split": "test", "cluster_id": "syn_03027_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 8, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1955, "query_tokens": 378, "message_tokens": 294, "message_utf8_bytes": 736, "cache_tensor_bytes": null, "serialized_payload_bytes": 736, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 33771.53900091071, "hops": {"1": 6880.990542005748, "2": 3875.456750043668, "3": 3842.124249902554, "4": 3828.6825419636443, "5": 3843.770332983695, "6": 3833.0124999629334, "7": 3827.791125047952, "8": 3839.710959000513}, "receiver_prefill_ms": 142.95812498312443, "ttft_ms": 11.813041986897588, "decode_ms": 11.816457961685956}, "total_wall_ms": 33938.47612594254, "ttft_ms": 11.813041986897588, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 63, "facts_retained": 8, "retention_rate": 0.12698412698412698, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.125, "amount": 0.07692307692307693, "status": 0.25, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 696}, "ledger_by_hop": {"1": {"facts_total": 63, "facts_retained": 9, "retention_rate": 0.14285714285714285, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.25, "amount": 0.07692307692307693, "status": 0.25, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 787}, "2": {"facts_total": 63, "facts_retained": 8, "retention_rate": 0.12698412698412698, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.125, "amount": 0.07692307692307693, "status": 0.25, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 696}, "3": {"facts_total": 63, "facts_retained": 8, "retention_rate": 0.12698412698412698, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.125, "amount": 0.07692307692307693, "status": 0.25, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 696}, "4": {"facts_total": 63, "facts_retained": 8, "retention_rate": 0.12698412698412698, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.125, "amount": 0.07692307692307693, "status": 0.25, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 696}, "5": {"facts_total": 63, "facts_retained": 8, "retention_rate": 0.12698412698412698, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.125, "amount": 0.07692307692307693, "status": 0.25, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 696}, "6": {"facts_total": 63, "facts_retained": 8, "retention_rate": 0.12698412698412698, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.125, "amount": 0.07692307692307693, "status": 0.25, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 696}, "7": {"facts_total": 63, "facts_retained": 8, "retention_rate": 0.12698412698412698, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.125, "amount": 0.07692307692307693, "status": 0.25, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 696}, "8": {"facts_total": 63, "facts_retained": 8, "retention_rate": 0.12698412698412698, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.125, "amount": 0.07692307692307693, "status": 0.25, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 696}}}, "reference_key_id": "keys/synthetic/syn_03027_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "5de8bbf0fdf02bbceb67207371e6f5c0ec86d2b30877ec832e254d4b22f40895", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "190ff5864d9c0a9fa6431bc6", "attempt_id": "af2a960bdb29", "run_group_id": "E4_core:A5|closed|qwen3b:h8", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:30:33.562613Z", "task_family": "synthetic", "document_id": "syn_03027_2000", "document_hash": "10326080979f71b625e4110c7c19767e5a752e267de5034f7e37d98cb6e06249", "question_id": "q08_revision", "split": "test", "cluster_id": "syn_03027_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 8, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1955, "query_tokens": 392, "message_tokens": 294, "message_utf8_bytes": 736, "cache_tensor_bytes": null, "serialized_payload_bytes": 736, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 33771.53900091071, "hops": {"1": 6880.990542005748, "2": 3875.456750043668, "3": 3842.124249902554, "4": 3828.6825419636443, "5": 3843.770332983695, "6": 3833.0124999629334, "7": 3827.791125047952, "8": 3839.710959000513}, "receiver_prefill_ms": 161.10620892141014, "ttft_ms": 11.814457946456969, "decode_ms": 105.62295897398144}, "total_wall_ms": 34050.455209915526, "ttft_ms": 11.814457946456969, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 9}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "6,914,937", "evidence_artifact": {"ledger": {"facts_total": 63, "facts_retained": 8, "retention_rate": 0.12698412698412698, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.125, "amount": 0.07692307692307693, "status": 0.25, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 696}, "ledger_by_hop": {"1": {"facts_total": 63, "facts_retained": 9, "retention_rate": 0.14285714285714285, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.25, "amount": 0.07692307692307693, "status": 0.25, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 787}, "2": {"facts_total": 63, "facts_retained": 8, "retention_rate": 0.12698412698412698, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.125, "amount": 0.07692307692307693, "status": 0.25, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 696}, "3": {"facts_total": 63, "facts_retained": 8, "retention_rate": 0.12698412698412698, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.125, "amount": 0.07692307692307693, "status": 0.25, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 696}, "4": {"facts_total": 63, "facts_retained": 8, "retention_rate": 0.12698412698412698, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.125, "amount": 0.07692307692307693, "status": 0.25, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 696}, "5": {"facts_total": 63, "facts_retained": 8, "retention_rate": 0.12698412698412698, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.125, "amount": 0.07692307692307693, "status": 0.25, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 696}, "6": {"facts_total": 63, "facts_retained": 8, "retention_rate": 0.12698412698412698, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.125, "amount": 0.07692307692307693, "status": 0.25, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 696}, "7": {"facts_total": 63, "facts_retained": 8, "retention_rate": 0.12698412698412698, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.125, "amount": 0.07692307692307693, "status": 0.25, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 696}, "8": {"facts_total": 63, "facts_retained": 8, "retention_rate": 0.12698412698412698, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.125, "amount": 0.07692307692307693, "status": 0.25, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 696}}}, "reference_key_id": "keys/synthetic/syn_03027_2000.json", "task_scores": {"correct": 0.0, "abstained": 0.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "5de8bbf0fdf02bbceb67207371e6f5c0ec86d2b30877ec832e254d4b22f40895", "output_hash": "c56851abfdbf8851226a1e71c4911c616a691c490a883eb754e61bac551d004c"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "78b59f8fc9ce016fc86219b8", "attempt_id": "536c3afe262c", "run_group_id": "E4_core:A5|closed|qwen3b:h8", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:30:33.856274Z", "task_family": "synthetic", "document_id": "syn_03027_2000", "document_hash": "10326080979f71b625e4110c7c19767e5a752e267de5034f7e37d98cb6e06249", "question_id": "q09_revision", "split": "test", "cluster_id": "syn_03027_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 8, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1955, "query_tokens": 385, "message_tokens": 294, "message_utf8_bytes": 736, "cache_tensor_bytes": null, "serialized_payload_bytes": 736, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 33771.53900091071, "hops": {"1": 6880.990542005748, "2": 3875.456750043668, "3": 3842.124249902554, "4": 3828.6825419636443, "5": 3843.770332983695, "6": 3833.0124999629334, "7": 3827.791125047952, "8": 3839.710959000513}, "receiver_prefill_ms": 142.65770895872265, "ttft_ms": 11.740584042854607, "decode_ms": 81.82291698176414}, "total_wall_ms": 34008.11458390672, "ttft_ms": 11.740584042854607, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 7}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "6914937", "evidence_artifact": {"ledger": {"facts_total": 63, "facts_retained": 8, "retention_rate": 0.12698412698412698, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.125, "amount": 0.07692307692307693, "status": 0.25, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 696}, "ledger_by_hop": {"1": {"facts_total": 63, "facts_retained": 9, "retention_rate": 0.14285714285714285, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.25, "amount": 0.07692307692307693, "status": 0.25, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 787}, "2": {"facts_total": 63, "facts_retained": 8, "retention_rate": 0.12698412698412698, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.125, "amount": 0.07692307692307693, "status": 0.25, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 696}, "3": {"facts_total": 63, "facts_retained": 8, "retention_rate": 0.12698412698412698, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.125, "amount": 0.07692307692307693, "status": 0.25, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 696}, "4": {"facts_total": 63, "facts_retained": 8, "retention_rate": 0.12698412698412698, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.125, "amount": 0.07692307692307693, "status": 0.25, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 696}, "5": {"facts_total": 63, "facts_retained": 8, "retention_rate": 0.12698412698412698, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.125, "amount": 0.07692307692307693, "status": 0.25, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 696}, "6": {"facts_total": 63, "facts_retained": 8, "retention_rate": 0.12698412698412698, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.125, "amount": 0.07692307692307693, "status": 0.25, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 696}, "7": {"facts_total": 63, "facts_retained": 8, "retention_rate": 0.12698412698412698, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.125, "amount": 0.07692307692307693, "status": 0.25, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 696}, "8": {"facts_total": 63, "facts_retained": 8, "retention_rate": 0.12698412698412698, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.125, "amount": 0.07692307692307693, "status": 0.25, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 696}}}, "reference_key_id": "keys/synthetic/syn_03027_2000.json", "task_scores": {"correct": 0.0, "abstained": 0.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "5de8bbf0fdf02bbceb67207371e6f5c0ec86d2b30877ec832e254d4b22f40895", "output_hash": "534d40ded32f784c68307dd2386faed2a27bf39dcc27321ee6985e42ccd285e5"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "8d16664747a264fac994528e", "attempt_id": "6aada17c4c14", "run_group_id": "E4_core:A5|closed|qwen3b:h8", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:30:34.108050Z", "task_family": "synthetic", "document_id": "syn_03027_2000", "document_hash": "10326080979f71b625e4110c7c19767e5a752e267de5034f7e37d98cb6e06249", "question_id": "q10_temporal", "split": "test", "cluster_id": "syn_03027_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 8, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1955, "query_tokens": 395, "message_tokens": 294, "message_utf8_bytes": 736, "cache_tensor_bytes": null, "serialized_payload_bytes": 736, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 33771.53900091071, "hops": {"1": 6880.990542005748, "2": 3875.456750043668, "3": 3842.124249902554, "4": 3828.6825419636443, "5": 3843.770332983695, "6": 3833.0124999629334, "7": 3827.791125047952, "8": 3839.710959000513}, "receiver_prefill_ms": 160.55770800448954, "ttft_ms": 11.701749986968935, "decode_ms": 11.696540983393788}, "total_wall_ms": 33955.845542950556, "ttft_ms": 11.701749986968935, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 63, "facts_retained": 8, "retention_rate": 0.12698412698412698, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.125, "amount": 0.07692307692307693, "status": 0.25, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 696}, "ledger_by_hop": {"1": {"facts_total": 63, "facts_retained": 9, "retention_rate": 0.14285714285714285, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.25, "amount": 0.07692307692307693, "status": 0.25, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 787}, "2": {"facts_total": 63, "facts_retained": 8, "retention_rate": 0.12698412698412698, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.125, "amount": 0.07692307692307693, "status": 0.25, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 696}, "3": {"facts_total": 63, "facts_retained": 8, "retention_rate": 0.12698412698412698, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.125, "amount": 0.07692307692307693, "status": 0.25, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 696}, "4": {"facts_total": 63, "facts_retained": 8, "retention_rate": 0.12698412698412698, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.125, "amount": 0.07692307692307693, "status": 0.25, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 696}, "5": {"facts_total": 63, "facts_retained": 8, "retention_rate": 0.12698412698412698, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.125, "amount": 0.07692307692307693, "status": 0.25, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 696}, "6": {"facts_total": 63, "facts_retained": 8, "retention_rate": 0.12698412698412698, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.125, "amount": 0.07692307692307693, "status": 0.25, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 696}, "7": {"facts_total": 63, "facts_retained": 8, "retention_rate": 0.12698412698412698, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.125, "amount": 0.07692307692307693, "status": 0.25, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 696}, "8": {"facts_total": 63, "facts_retained": 8, "retention_rate": 0.12698412698412698, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.125, "amount": 0.07692307692307693, "status": 0.25, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 696}}}, "reference_key_id": "keys/synthetic/syn_03027_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "5de8bbf0fdf02bbceb67207371e6f5c0ec86d2b30877ec832e254d4b22f40895", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "f66ab10c7154687f9b902c04", "attempt_id": "3f680ff9bd41", "run_group_id": "E4_core:A5|closed|qwen3b:h8", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:30:34.307316Z", "task_family": "synthetic", "document_id": "syn_03027_2000", "document_hash": "10326080979f71b625e4110c7c19767e5a752e267de5034f7e37d98cb6e06249", "question_id": "q11_aggregation", "split": "test", "cluster_id": "syn_03027_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 8, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1955, "query_tokens": 388, "message_tokens": 294, "message_utf8_bytes": 736, "cache_tensor_bytes": null, "serialized_payload_bytes": 736, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 33771.53900091071, "hops": {"1": 6880.990542005748, "2": 3875.456750043668, "3": 3842.124249902554, "4": 3828.6825419636443, "5": 3843.770332983695, "6": 3833.0124999629334, "7": 3827.791125047952, "8": 3839.710959000513}, "receiver_prefill_ms": 160.3999580256641, "ttft_ms": 11.728040990419686, "decode_ms": 81.81229105684906}, "total_wall_ms": 34025.84295999259, "ttft_ms": 11.728040990419686, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 7}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "6914937", "evidence_artifact": {"ledger": {"facts_total": 63, "facts_retained": 8, "retention_rate": 0.12698412698412698, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.125, "amount": 0.07692307692307693, "status": 0.25, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 696}, "ledger_by_hop": {"1": {"facts_total": 63, "facts_retained": 9, "retention_rate": 0.14285714285714285, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.25, "amount": 0.07692307692307693, "status": 0.25, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 787}, "2": {"facts_total": 63, "facts_retained": 8, "retention_rate": 0.12698412698412698, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.125, "amount": 0.07692307692307693, "status": 0.25, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 696}, "3": {"facts_total": 63, "facts_retained": 8, "retention_rate": 0.12698412698412698, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.125, "amount": 0.07692307692307693, "status": 0.25, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 696}, "4": {"facts_total": 63, "facts_retained": 8, "retention_rate": 0.12698412698412698, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.125, "amount": 0.07692307692307693, "status": 0.25, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 696}, "5": {"facts_total": 63, "facts_retained": 8, "retention_rate": 0.12698412698412698, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.125, "amount": 0.07692307692307693, "status": 0.25, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 696}, "6": {"facts_total": 63, "facts_retained": 8, "retention_rate": 0.12698412698412698, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.125, "amount": 0.07692307692307693, "status": 0.25, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 696}, "7": {"facts_total": 63, "facts_retained": 8, "retention_rate": 0.12698412698412698, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.125, "amount": 0.07692307692307693, "status": 0.25, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 696}, "8": {"facts_total": 63, "facts_retained": 8, "retention_rate": 0.12698412698412698, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.125, "amount": 0.07692307692307693, "status": 0.25, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 696}}}, "reference_key_id": "keys/synthetic/syn_03027_2000.json", "task_scores": {"correct": 0.0, "abstained": 0.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "5de8bbf0fdf02bbceb67207371e6f5c0ec86d2b30877ec832e254d4b22f40895", "output_hash": "534d40ded32f784c68307dd2386faed2a27bf39dcc27321ee6985e42ccd285e5"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "8d705cddf11cf741f5df60e0", "attempt_id": "d0314846e27e", "run_group_id": "E4_core:A5|closed|qwen3b:h8", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:30:34.576471Z", "task_family": "synthetic", "document_id": "syn_03027_2000", "document_hash": "10326080979f71b625e4110c7c19767e5a752e267de5034f7e37d98cb6e06249", "question_id": "q12_unit_conversion", "split": "test", "cluster_id": "syn_03027_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 8, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1955, "query_tokens": 385, "message_tokens": 294, "message_utf8_bytes": 736, "cache_tensor_bytes": null, "serialized_payload_bytes": 736, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 33771.53900091071, "hops": {"1": 6880.990542005748, "2": 3875.456750043668, "3": 3842.124249902554, "4": 3828.6825419636443, "5": 3843.770332983695, "6": 3833.0124999629334, "7": 3827.791125047952, "8": 3839.710959000513}, "receiver_prefill_ms": 142.28320808615535, "ttft_ms": 11.696208966895938, "decode_ms": 11.715666973032057}, "total_wall_ms": 33937.57891689893, "ttft_ms": 11.696208966895938, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 63, "facts_retained": 8, "retention_rate": 0.12698412698412698, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.125, "amount": 0.07692307692307693, "status": 0.25, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 696}, "ledger_by_hop": {"1": {"facts_total": 63, "facts_retained": 9, "retention_rate": 0.14285714285714285, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.25, "amount": 0.07692307692307693, "status": 0.25, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 787}, "2": {"facts_total": 63, "facts_retained": 8, "retention_rate": 0.12698412698412698, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.125, "amount": 0.07692307692307693, "status": 0.25, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 696}, "3": {"facts_total": 63, "facts_retained": 8, "retention_rate": 0.12698412698412698, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.125, "amount": 0.07692307692307693, "status": 0.25, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 696}, "4": {"facts_total": 63, "facts_retained": 8, "retention_rate": 0.12698412698412698, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.125, "amount": 0.07692307692307693, "status": 0.25, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 696}, "5": {"facts_total": 63, "facts_retained": 8, "retention_rate": 0.12698412698412698, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.125, "amount": 0.07692307692307693, "status": 0.25, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 696}, "6": {"facts_total": 63, "facts_retained": 8, "retention_rate": 0.12698412698412698, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.125, "amount": 0.07692307692307693, "status": 0.25, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 696}, "7": {"facts_total": 63, "facts_retained": 8, "retention_rate": 0.12698412698412698, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.125, "amount": 0.07692307692307693, "status": 0.25, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 696}, "8": {"facts_total": 63, "facts_retained": 8, "retention_rate": 0.12698412698412698, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.125, "amount": 0.07692307692307693, "status": 0.25, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 696}}}, "reference_key_id": "keys/synthetic/syn_03027_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "5de8bbf0fdf02bbceb67207371e6f5c0ec86d2b30877ec832e254d4b22f40895", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "f0d319a0a906287feb1d2ce9", "attempt_id": "f7535372e763", "run_group_id": "E4_core:A5|closed|qwen3b:h8", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:30:34.757606Z", "task_family": "synthetic", "document_id": "syn_03027_2000", "document_hash": "10326080979f71b625e4110c7c19767e5a752e267de5034f7e37d98cb6e06249", "question_id": "q13_unanswerable", "split": "test", "cluster_id": "syn_03027_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 8, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1955, "query_tokens": 388, "message_tokens": 294, "message_utf8_bytes": 736, "cache_tensor_bytes": null, "serialized_payload_bytes": 736, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 33771.53900091071, "hops": {"1": 6880.990542005748, "2": 3875.456750043668, "3": 3842.124249902554, "4": 3828.6825419636443, "5": 3843.770332983695, "6": 3833.0124999629334, "7": 3827.791125047952, "8": 3839.710959000513}, "receiver_prefill_ms": 160.82045796792954, "ttft_ms": 11.703249998390675, "decode_ms": 11.678499984554946}, "total_wall_ms": 33956.093625864014, "ttft_ms": 11.703249998390675, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 63, "facts_retained": 8, "retention_rate": 0.12698412698412698, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.125, "amount": 0.07692307692307693, "status": 0.25, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 696}, "ledger_by_hop": {"1": {"facts_total": 63, "facts_retained": 9, "retention_rate": 0.14285714285714285, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.25, "amount": 0.07692307692307693, "status": 0.25, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 787}, "2": {"facts_total": 63, "facts_retained": 8, "retention_rate": 0.12698412698412698, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.125, "amount": 0.07692307692307693, "status": 0.25, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 696}, "3": {"facts_total": 63, "facts_retained": 8, "retention_rate": 0.12698412698412698, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.125, "amount": 0.07692307692307693, "status": 0.25, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 696}, "4": {"facts_total": 63, "facts_retained": 8, "retention_rate": 0.12698412698412698, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.125, "amount": 0.07692307692307693, "status": 0.25, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 696}, "5": {"facts_total": 63, "facts_retained": 8, "retention_rate": 0.12698412698412698, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.125, "amount": 0.07692307692307693, "status": 0.25, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 696}, "6": {"facts_total": 63, "facts_retained": 8, "retention_rate": 0.12698412698412698, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.125, "amount": 0.07692307692307693, "status": 0.25, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 696}, "7": {"facts_total": 63, "facts_retained": 8, "retention_rate": 0.12698412698412698, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.125, "amount": 0.07692307692307693, "status": 0.25, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 696}, "8": {"facts_total": 63, "facts_retained": 8, "retention_rate": 0.12698412698412698, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.125, "amount": 0.07692307692307693, "status": 0.25, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 696}}}, "reference_key_id": "keys/synthetic/syn_03027_2000.json", "task_scores": {"correct": 1.0, "abstained": 1.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "5de8bbf0fdf02bbceb67207371e6f5c0ec86d2b30877ec832e254d4b22f40895", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "5d3eaf196ab4168c641de924", "attempt_id": "553728977050", "run_group_id": "E4_core:A5|closed|qwen3b:h8", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:30:34.957283Z", "task_family": "synthetic", "document_id": "syn_03027_2000", "document_hash": "10326080979f71b625e4110c7c19767e5a752e267de5034f7e37d98cb6e06249", "question_id": "q14_unanswerable", "split": "test", "cluster_id": "syn_03027_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 8, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1955, "query_tokens": 390, "message_tokens": 294, "message_utf8_bytes": 736, "cache_tensor_bytes": null, "serialized_payload_bytes": 736, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 33771.53900091071, "hops": {"1": 6880.990542005748, "2": 3875.456750043668, "3": 3842.124249902554, "4": 3828.6825419636443, "5": 3843.770332983695, "6": 3833.0124999629334, "7": 3827.791125047952, "8": 3839.710959000513}, "receiver_prefill_ms": 160.95841699279845, "ttft_ms": 11.682874988764524, "decode_ms": 11.715583968907595}, "total_wall_ms": 33956.26037591137, "ttft_ms": 11.682874988764524, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 63, "facts_retained": 8, "retention_rate": 0.12698412698412698, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.125, "amount": 0.07692307692307693, "status": 0.25, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 696}, "ledger_by_hop": {"1": {"facts_total": 63, "facts_retained": 9, "retention_rate": 0.14285714285714285, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.25, "amount": 0.07692307692307693, "status": 0.25, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 787}, "2": {"facts_total": 63, "facts_retained": 8, "retention_rate": 0.12698412698412698, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.125, "amount": 0.07692307692307693, "status": 0.25, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 696}, "3": {"facts_total": 63, "facts_retained": 8, "retention_rate": 0.12698412698412698, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.125, "amount": 0.07692307692307693, "status": 0.25, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 696}, "4": {"facts_total": 63, "facts_retained": 8, "retention_rate": 0.12698412698412698, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.125, "amount": 0.07692307692307693, "status": 0.25, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 696}, "5": {"facts_total": 63, "facts_retained": 8, "retention_rate": 0.12698412698412698, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.125, "amount": 0.07692307692307693, "status": 0.25, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 696}, "6": {"facts_total": 63, "facts_retained": 8, "retention_rate": 0.12698412698412698, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.125, "amount": 0.07692307692307693, "status": 0.25, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 696}, "7": {"facts_total": 63, "facts_retained": 8, "retention_rate": 0.12698412698412698, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.125, "amount": 0.07692307692307693, "status": 0.25, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 696}, "8": {"facts_total": 63, "facts_retained": 8, "retention_rate": 0.12698412698412698, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.125, "amount": 0.07692307692307693, "status": 0.25, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 696}}}, "reference_key_id": "keys/synthetic/syn_03027_2000.json", "task_scores": {"correct": 1.0, "abstained": 1.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "5de8bbf0fdf02bbceb67207371e6f5c0ec86d2b30877ec832e254d4b22f40895", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "6e6fd989c552471cceede09d", "attempt_id": "0c3d90351834", "run_group_id": "E4_core:A4|repairable|qwen3b:h1", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:30:42.072957Z", "task_family": "synthetic", "document_id": "syn_03027_2000", "document_hash": "10326080979f71b625e4110c7c19767e5a752e267de5034f7e37d98cb6e06249", "question_id": "q00_single_fact", "split": "test", "cluster_id": "syn_03027_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 1, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1955, "query_tokens": 594, "message_tokens": 512, "message_utf8_bytes": 1087, "cache_tensor_bytes": null, "serialized_payload_bytes": 1087, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 6913.723333040252, "hops": {"1": 6913.723333040252}, "receiver_prefill_ms": 225.96166701987386, "ttft_ms": 12.14233401697129, "decode_ms": 12.142083025537431}, "total_wall_ms": 7164.618417038582, "ttft_ms": 12.14233401697129, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 63, "facts_retained": 27, "retention_rate": 0.42857142857142855, "retention_by_kind": {"person": 0.5, "date": 0.125, "amount": 0.6153846153846154, "status": 0.75, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1087}, "ledger_by_hop": {"1": {"facts_total": 63, "facts_retained": 27, "retention_rate": 0.42857142857142855, "retention_by_kind": {"person": 0.5, "date": 0.125, "amount": 0.6153846153846154, "status": 0.75, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1087}}}, "reference_key_id": "keys/synthetic/syn_03027_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "1f88f0e78c63504e418dea843641b99a98555e15d47b02cd5cb2f57da3003dea", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "b985e7ee755f590f19dd7ced", "attempt_id": "1f364c6119a6", "run_group_id": "E4_core:A4|repairable|qwen3b:h1", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:30:42.340858Z", "task_family": "synthetic", "document_id": "syn_03027_2000", "document_hash": "10326080979f71b625e4110c7c19767e5a752e267de5034f7e37d98cb6e06249", "question_id": "q01_single_fact", "split": "test", "cluster_id": "syn_03027_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 1, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1955, "query_tokens": 601, "message_tokens": 512, "message_utf8_bytes": 1087, "cache_tensor_bytes": null, "serialized_payload_bytes": 1087, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 6913.723333040252, "hops": {"1": 6913.723333040252}, "receiver_prefill_ms": 226.10537498258054, "ttft_ms": 12.183125014416873, "decode_ms": 143.83104199077934}, "total_wall_ms": 7296.495040995069, "ttft_ms": 12.183125014416873, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 12}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "Lot 247-S, Baie-Comeau", "evidence_artifact": {"ledger": {"facts_total": 63, "facts_retained": 27, "retention_rate": 0.42857142857142855, "retention_by_kind": {"person": 0.5, "date": 0.125, "amount": 0.6153846153846154, "status": 0.75, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1087}, "ledger_by_hop": {"1": {"facts_total": 63, "facts_retained": 27, "retention_rate": 0.42857142857142855, "retention_by_kind": {"person": 0.5, "date": 0.125, "amount": 0.6153846153846154, "status": 0.75, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1087}}}, "reference_key_id": "keys/synthetic/syn_03027_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 0.8571428571428571}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "1f88f0e78c63504e418dea843641b99a98555e15d47b02cd5cb2f57da3003dea", "output_hash": "77b7da1d43e7d8815e9e9159ebbd735045bc7f668260ba30b93674f70622b8ec"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "548a8796c91c2ecf2521bbd7", "attempt_id": "5e01c238a318", "run_group_id": "E4_core:A4|repairable|qwen3b:h1", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:30:42.740299Z", "task_family": "synthetic", "document_id": "syn_03027_2000", "document_hash": "10326080979f71b625e4110c7c19767e5a752e267de5034f7e37d98cb6e06249", "question_id": "q02_exact_copy", "split": "test", "cluster_id": "syn_03027_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 1, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1955, "query_tokens": 600, "message_tokens": 512, "message_utf8_bytes": 1087, "cache_tensor_bytes": null, "serialized_payload_bytes": 1087, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 6913.723333040252, "hops": {"1": 6913.723333040252}, "receiver_prefill_ms": 226.22108296491206, "ttft_ms": 12.253916938789189, "decode_ms": 84.06479202676564}, "total_wall_ms": 7236.934624030255, "ttft_ms": 12.253916938789189, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 7}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "15.0054", "evidence_artifact": {"ledger": {"facts_total": 63, "facts_retained": 27, "retention_rate": 0.42857142857142855, "retention_by_kind": {"person": 0.5, "date": 0.125, "amount": 0.6153846153846154, "status": 0.75, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1087}, "ledger_by_hop": {"1": {"facts_total": 63, "facts_retained": 27, "retention_rate": 0.42857142857142855, "retention_by_kind": {"person": 0.5, "date": 0.125, "amount": 0.6153846153846154, "status": 0.75, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1087}}}, "reference_key_id": "keys/synthetic/syn_03027_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "1f88f0e78c63504e418dea843641b99a98555e15d47b02cd5cb2f57da3003dea", "output_hash": "d13adba73068c847c850237059e0202fc5b91ccacf2e792e9ce5e3b90556501a"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "39c511551d0929cbd8c30526", "attempt_id": "9aa07bc3c245", "run_group_id": "E4_core:A4|repairable|qwen3b:h1", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:30:43.080421Z", "task_family": "synthetic", "document_id": "syn_03027_2000", "document_hash": "10326080979f71b625e4110c7c19767e5a752e267de5034f7e37d98cb6e06249", "question_id": "q03_exact_copy", "split": "test", "cluster_id": "syn_03027_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 1, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1955, "query_tokens": 595, "message_tokens": 512, "message_utf8_bytes": 1087, "cache_tensor_bytes": null, "serialized_payload_bytes": 1087, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 6913.723333040252, "hops": {"1": 6913.723333040252}, "receiver_prefill_ms": 226.4239169890061, "ttft_ms": 12.130500050261617, "decode_ms": 107.62237501330674}, "total_wall_ms": 7260.570583050139, "ttft_ms": 12.130500050261617, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 9}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "Priyamvada Sørlie", "evidence_artifact": {"ledger": {"facts_total": 63, "facts_retained": 27, "retention_rate": 0.42857142857142855, "retention_by_kind": {"person": 0.5, "date": 0.125, "amount": 0.6153846153846154, "status": 0.75, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1087}, "ledger_by_hop": {"1": {"facts_total": 63, "facts_retained": 27, "retention_rate": 0.42857142857142855, "retention_by_kind": {"person": 0.5, "date": 0.125, "amount": 0.6153846153846154, "status": 0.75, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1087}}}, "reference_key_id": "keys/synthetic/syn_03027_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "1f88f0e78c63504e418dea843641b99a98555e15d47b02cd5cb2f57da3003dea", "output_hash": "c5f554196ed23dd71ad995ec89f7142099d524dbb6534cc10d94a0c89d13e792"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "6b2b4122f63542287d3f9fba", "attempt_id": "7471bc7cb223", "run_group_id": "E4_core:A4|repairable|qwen3b:h1", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:30:43.444163Z", "task_family": "synthetic", "document_id": "syn_03027_2000", "document_hash": "10326080979f71b625e4110c7c19767e5a752e267de5034f7e37d98cb6e06249", "question_id": "q04_multi_hop", "split": "test", "cluster_id": "syn_03027_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 1, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1955, "query_tokens": 606, "message_tokens": 512, "message_utf8_bytes": 1087, "cache_tensor_bytes": null, "serialized_payload_bytes": 1087, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 6913.723333040252, "hops": {"1": 6913.723333040252}, "receiver_prefill_ms": 226.8315830733627, "ttft_ms": 12.146459077484906, "decode_ms": 11.996458051726222}, "total_wall_ms": 7165.376124088652, "ttft_ms": 12.146459077484906, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 63, "facts_retained": 27, "retention_rate": 0.42857142857142855, "retention_by_kind": {"person": 0.5, "date": 0.125, "amount": 0.6153846153846154, "status": 0.75, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1087}, "ledger_by_hop": {"1": {"facts_total": 63, "facts_retained": 27, "retention_rate": 0.42857142857142855, "retention_by_kind": {"person": 0.5, "date": 0.125, "amount": 0.6153846153846154, "status": 0.75, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1087}}}, "reference_key_id": "keys/synthetic/syn_03027_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "1f88f0e78c63504e418dea843641b99a98555e15d47b02cd5cb2f57da3003dea", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "0a1bb68080696d7ada6bcc03", "attempt_id": "15265438a15d", "run_group_id": "E4_core:A4|repairable|qwen3b:h1", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:30:43.713002Z", "task_family": "synthetic", "document_id": "syn_03027_2000", "document_hash": "10326080979f71b625e4110c7c19767e5a752e267de5034f7e37d98cb6e06249", "question_id": "q05_multi_hop", "split": "test", "cluster_id": "syn_03027_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 1, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1955, "query_tokens": 603, "message_tokens": 512, "message_utf8_bytes": 1087, "cache_tensor_bytes": null, "serialized_payload_bytes": 1087, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 6913.723333040252, "hops": {"1": 6913.723333040252}, "receiver_prefill_ms": 226.76500002853572, "ttft_ms": 12.177166994661093, "decode_ms": 11.972832959145308}, "total_wall_ms": 7165.297083090991, "ttft_ms": 12.177166994661093, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 63, "facts_retained": 27, "retention_rate": 0.42857142857142855, "retention_by_kind": {"person": 0.5, "date": 0.125, "amount": 0.6153846153846154, "status": 0.75, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1087}, "ledger_by_hop": {"1": {"facts_total": 63, "facts_retained": 27, "retention_rate": 0.42857142857142855, "retention_by_kind": {"person": 0.5, "date": 0.125, "amount": 0.6153846153846154, "status": 0.75, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1087}}}, "reference_key_id": "keys/synthetic/syn_03027_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "1f88f0e78c63504e418dea843641b99a98555e15d47b02cd5cb2f57da3003dea", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "9e4325b1939b80f9c577f774", "attempt_id": "29f482ef35a7", "run_group_id": "E4_core:A4|repairable|qwen3b:h1", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:30:43.981236Z", "task_family": "synthetic", "document_id": "syn_03027_2000", "document_hash": "10326080979f71b625e4110c7c19767e5a752e267de5034f7e37d98cb6e06249", "question_id": "q06_negation", "split": "test", "cluster_id": "syn_03027_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 1, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1955, "query_tokens": 595, "message_tokens": 512, "message_utf8_bytes": 1087, "cache_tensor_bytes": null, "serialized_payload_bytes": 1087, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 6913.723333040252, "hops": {"1": 6913.723333040252}, "receiver_prefill_ms": 226.59629094414413, "ttft_ms": 12.120499974116683, "decode_ms": 11.998625006526709}, "total_wall_ms": 7165.1322490070015, "ttft_ms": 12.120499974116683, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 63, "facts_retained": 27, "retention_rate": 0.42857142857142855, "retention_by_kind": {"person": 0.5, "date": 0.125, "amount": 0.6153846153846154, "status": 0.75, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1087}, "ledger_by_hop": {"1": {"facts_total": 63, "facts_retained": 27, "retention_rate": 0.42857142857142855, "retention_by_kind": {"person": 0.5, "date": 0.125, "amount": 0.6153846153846154, "status": 0.75, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1087}}}, "reference_key_id": "keys/synthetic/syn_03027_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "1f88f0e78c63504e418dea843641b99a98555e15d47b02cd5cb2f57da3003dea", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "327c40a35449264958ac7ec0", "attempt_id": "f6ffd1b7f38b", "run_group_id": "E4_core:A4|repairable|qwen3b:h1", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:30:44.249810Z", "task_family": "synthetic", "document_id": "syn_03027_2000", "document_hash": "10326080979f71b625e4110c7c19767e5a752e267de5034f7e37d98cb6e06249", "question_id": "q07_negation", "split": "test", "cluster_id": "syn_03027_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 1, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1955, "query_tokens": 596, "message_tokens": 512, "message_utf8_bytes": 1087, "cache_tensor_bytes": null, "serialized_payload_bytes": 1087, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 6913.723333040252, "hops": {"1": 6913.723333040252}, "receiver_prefill_ms": 226.79187497124076, "ttft_ms": 12.165124993771315, "decode_ms": 11.956791044212878}, "total_wall_ms": 7165.306583046913, "ttft_ms": 12.165124993771315, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "no", "evidence_artifact": {"ledger": {"facts_total": 63, "facts_retained": 27, "retention_rate": 0.42857142857142855, "retention_by_kind": {"person": 0.5, "date": 0.125, "amount": 0.6153846153846154, "status": 0.75, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1087}, "ledger_by_hop": {"1": {"facts_total": 63, "facts_retained": 27, "retention_rate": 0.42857142857142855, "retention_by_kind": {"person": 0.5, "date": 0.125, "amount": 0.6153846153846154, "status": 0.75, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1087}}}, "reference_key_id": "keys/synthetic/syn_03027_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "1f88f0e78c63504e418dea843641b99a98555e15d47b02cd5cb2f57da3003dea", "output_hash": "9390298f3fb0c5b160498935d79cb139aef28e1c47358b4bbba61862b9c26e59"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "62ad7b40251cf5a894e0e5f5", "attempt_id": "f599a6e76ed3", "run_group_id": "E4_core:A4|repairable|qwen3b:h1", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:30:44.519082Z", "task_family": "synthetic", "document_id": "syn_03027_2000", "document_hash": "10326080979f71b625e4110c7c19767e5a752e267de5034f7e37d98cb6e06249", "question_id": "q08_revision", "split": "test", "cluster_id": "syn_03027_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 1, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1955, "query_tokens": 610, "message_tokens": 512, "message_utf8_bytes": 1087, "cache_tensor_bytes": null, "serialized_payload_bytes": 1087, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 6913.723333040252, "hops": {"1": 6913.723333040252}, "receiver_prefill_ms": 229.277583071962, "ttft_ms": 12.159582925960422, "decode_ms": 84.08250007778406}, "total_wall_ms": 7239.922000095248, "ttft_ms": 12.159582925960422, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 7}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "8010205", "evidence_artifact": {"ledger": {"facts_total": 63, "facts_retained": 27, "retention_rate": 0.42857142857142855, "retention_by_kind": {"person": 0.5, "date": 0.125, "amount": 0.6153846153846154, "status": 0.75, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1087}, "ledger_by_hop": {"1": {"facts_total": 63, "facts_retained": 27, "retention_rate": 0.42857142857142855, "retention_by_kind": {"person": 0.5, "date": 0.125, "amount": 0.6153846153846154, "status": 0.75, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1087}}}, "reference_key_id": "keys/synthetic/syn_03027_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "1f88f0e78c63504e418dea843641b99a98555e15d47b02cd5cb2f57da3003dea", "output_hash": "6d1ceb36132f38bd77e8f46c23ea21dd9bcff78bc80bd8a9eaf2d7f2f3bacf46"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "757649df8e9bbc3b12bc3ed2", "attempt_id": "e0ea438959ef", "run_group_id": "E4_core:A4|repairable|qwen3b:h1", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:30:44.862223Z", "task_family": "synthetic", "document_id": "syn_03027_2000", "document_hash": "10326080979f71b625e4110c7c19767e5a752e267de5034f7e37d98cb6e06249", "question_id": "q09_revision", "split": "test", "cluster_id": "syn_03027_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 1, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1955, "query_tokens": 603, "message_tokens": 512, "message_utf8_bytes": 1087, "cache_tensor_bytes": null, "serialized_payload_bytes": 1087, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 6913.723333040252, "hops": {"1": 6913.723333040252}, "receiver_prefill_ms": 226.67512495536357, "ttft_ms": 12.085333000868559, "decode_ms": 107.8914999961853}, "total_wall_ms": 7261.088541010395, "ttft_ms": 12.085333000868559, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 9}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "6,056,150", "evidence_artifact": {"ledger": {"facts_total": 63, "facts_retained": 27, "retention_rate": 0.42857142857142855, "retention_by_kind": {"person": 0.5, "date": 0.125, "amount": 0.6153846153846154, "status": 0.75, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1087}, "ledger_by_hop": {"1": {"facts_total": 63, "facts_retained": 27, "retention_rate": 0.42857142857142855, "retention_by_kind": {"person": 0.5, "date": 0.125, "amount": 0.6153846153846154, "status": 0.75, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1087}}}, "reference_key_id": "keys/synthetic/syn_03027_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "1f88f0e78c63504e418dea843641b99a98555e15d47b02cd5cb2f57da3003dea", "output_hash": "8262889a8478c7e0ee965f159d9849a3849b412e276f5b979476e2da5b320249"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "bdcbc28dec2dfaa66c061550", "attempt_id": "72a15ea591a0", "run_group_id": "E4_core:A4|repairable|qwen3b:h1", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:30:45.226549Z", "task_family": "synthetic", "document_id": "syn_03027_2000", "document_hash": "10326080979f71b625e4110c7c19767e5a752e267de5034f7e37d98cb6e06249", "question_id": "q10_temporal", "split": "test", "cluster_id": "syn_03027_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 1, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1955, "query_tokens": 613, "message_tokens": 512, "message_utf8_bytes": 1087, "cache_tensor_bytes": null, "serialized_payload_bytes": 1087, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 6913.723333040252, "hops": {"1": 6913.723333040252}, "receiver_prefill_ms": 228.9500420447439, "ttft_ms": 12.183125014416873, "decode_ms": 83.30258296336979}, "total_wall_ms": 7238.800166058354, "ttft_ms": 12.183125014416873, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 7}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "Lunlith-13", "evidence_artifact": {"ledger": {"facts_total": 63, "facts_retained": 27, "retention_rate": 0.42857142857142855, "retention_by_kind": {"person": 0.5, "date": 0.125, "amount": 0.6153846153846154, "status": 0.75, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1087}, "ledger_by_hop": {"1": {"facts_total": 63, "facts_retained": 27, "retention_rate": 0.42857142857142855, "retention_by_kind": {"person": 0.5, "date": 0.125, "amount": 0.6153846153846154, "status": 0.75, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1087}}}, "reference_key_id": "keys/synthetic/syn_03027_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "1f88f0e78c63504e418dea843641b99a98555e15d47b02cd5cb2f57da3003dea", "output_hash": "c45f000afe6d865c8167954eebdff761141225817d4f5b26c962cd50cd75d391"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "b6d5f31a9df67f6e222bc2ae", "attempt_id": "d538ccfd6487", "run_group_id": "E4_core:A4|repairable|qwen3b:h1", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:30:45.568474Z", "task_family": "synthetic", "document_id": "syn_03027_2000", "document_hash": "10326080979f71b625e4110c7c19767e5a752e267de5034f7e37d98cb6e06249", "question_id": "q11_aggregation", "split": "test", "cluster_id": "syn_03027_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 1, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1955, "query_tokens": 606, "message_tokens": 512, "message_utf8_bytes": 1087, "cache_tensor_bytes": null, "serialized_payload_bytes": 1087, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 6913.723333040252, "hops": {"1": 6913.723333040252}, "receiver_prefill_ms": 226.82812495622784, "ttft_ms": 12.169958092272282, "decode_ms": 118.85733297094703}, "total_wall_ms": 7272.266249055974, "ttft_ms": 12.169958092272282, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 10}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "23,450,077", "evidence_artifact": {"ledger": {"facts_total": 63, "facts_retained": 27, "retention_rate": 0.42857142857142855, "retention_by_kind": {"person": 0.5, "date": 0.125, "amount": 0.6153846153846154, "status": 0.75, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1087}, "ledger_by_hop": {"1": {"facts_total": 63, "facts_retained": 27, "retention_rate": 0.42857142857142855, "retention_by_kind": {"person": 0.5, "date": 0.125, "amount": 0.6153846153846154, "status": 0.75, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1087}}}, "reference_key_id": "keys/synthetic/syn_03027_2000.json", "task_scores": {"correct": 0.0, "abstained": 0.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "1f88f0e78c63504e418dea843641b99a98555e15d47b02cd5cb2f57da3003dea", "output_hash": "2044d7c4afd32892ac22705680e35174db83308c139b02a1494275cc5e24aff3"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "b9a6db463aabbfd242e621b6", "attempt_id": "b3d8766da95d", "run_group_id": "E4_core:A4|repairable|qwen3b:h1", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:30:45.943582Z", "task_family": "synthetic", "document_id": "syn_03027_2000", "document_hash": "10326080979f71b625e4110c7c19767e5a752e267de5034f7e37d98cb6e06249", "question_id": "q12_unit_conversion", "split": "test", "cluster_id": "syn_03027_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 1, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1955, "query_tokens": 603, "message_tokens": 512, "message_utf8_bytes": 1087, "cache_tensor_bytes": null, "serialized_payload_bytes": 1087, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 6913.723333040252, "hops": {"1": 6913.723333040252}, "receiver_prefill_ms": 226.58933291677386, "ttft_ms": 12.23166601266712, "decode_ms": 12.016292079351842}, "total_wall_ms": 7165.241083013825, "ttft_ms": 12.23166601266712, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 63, "facts_retained": 27, "retention_rate": 0.42857142857142855, "retention_by_kind": {"person": 0.5, "date": 0.125, "amount": 0.6153846153846154, "status": 0.75, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1087}, "ledger_by_hop": {"1": {"facts_total": 63, "facts_retained": 27, "retention_rate": 0.42857142857142855, "retention_by_kind": {"person": 0.5, "date": 0.125, "amount": 0.6153846153846154, "status": 0.75, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1087}}}, "reference_key_id": "keys/synthetic/syn_03027_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "1f88f0e78c63504e418dea843641b99a98555e15d47b02cd5cb2f57da3003dea", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "64a3724f1f53f0a670e85247", "attempt_id": "11b0e4adb5b3", "run_group_id": "E4_core:A4|repairable|qwen3b:h1", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:30:46.212212Z", "task_family": "synthetic", "document_id": "syn_03027_2000", "document_hash": "10326080979f71b625e4110c7c19767e5a752e267de5034f7e37d98cb6e06249", "question_id": "q13_unanswerable", "split": "test", "cluster_id": "syn_03027_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 1, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1955, "query_tokens": 606, "message_tokens": 512, "message_utf8_bytes": 1087, "cache_tensor_bytes": null, "serialized_payload_bytes": 1087, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 6913.723333040252, "hops": {"1": 6913.723333040252}, "receiver_prefill_ms": 227.10920800454915, "ttft_ms": 12.291125021874905, "decode_ms": 12.066417024470866}, "total_wall_ms": 7165.863624075428, "ttft_ms": 12.291125021874905, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 63, "facts_retained": 27, "retention_rate": 0.42857142857142855, "retention_by_kind": {"person": 0.5, "date": 0.125, "amount": 0.6153846153846154, "status": 0.75, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1087}, "ledger_by_hop": {"1": {"facts_total": 63, "facts_retained": 27, "retention_rate": 0.42857142857142855, "retention_by_kind": {"person": 0.5, "date": 0.125, "amount": 0.6153846153846154, "status": 0.75, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1087}}}, "reference_key_id": "keys/synthetic/syn_03027_2000.json", "task_scores": {"correct": 1.0, "abstained": 1.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "1f88f0e78c63504e418dea843641b99a98555e15d47b02cd5cb2f57da3003dea", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "9cef6c4a91ca520c14ebc0fc", "attempt_id": "146173a7e154", "run_group_id": "E4_core:A4|repairable|qwen3b:h1", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:30:46.481119Z", "task_family": "synthetic", "document_id": "syn_03027_2000", "document_hash": "10326080979f71b625e4110c7c19767e5a752e267de5034f7e37d98cb6e06249", "question_id": "q14_unanswerable", "split": "test", "cluster_id": "syn_03027_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 1, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1955, "query_tokens": 608, "message_tokens": 512, "message_utf8_bytes": 1087, "cache_tensor_bytes": null, "serialized_payload_bytes": 1087, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 6913.723333040252, "hops": {"1": 6913.723333040252}, "receiver_prefill_ms": 226.86308296397328, "ttft_ms": 12.225582962855697, "decode_ms": 12.079791980795562}, "total_wall_ms": 7165.599124040455, "ttft_ms": 12.225582962855697, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 63, "facts_retained": 27, "retention_rate": 0.42857142857142855, "retention_by_kind": {"person": 0.5, "date": 0.125, "amount": 0.6153846153846154, "status": 0.75, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1087}, "ledger_by_hop": {"1": {"facts_total": 63, "facts_retained": 27, "retention_rate": 0.42857142857142855, "retention_by_kind": {"person": 0.5, "date": 0.125, "amount": 0.6153846153846154, "status": 0.75, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1087}}}, "reference_key_id": "keys/synthetic/syn_03027_2000.json", "task_scores": {"correct": 1.0, "abstained": 1.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "1f88f0e78c63504e418dea843641b99a98555e15d47b02cd5cb2f57da3003dea", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "df0dbdc3889b2a6c345d19f2", "attempt_id": "350b58766205", "run_group_id": "E4_core:A4|repairable|qwen3b:h2", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:30:53.289390Z", "task_family": "synthetic", "document_id": "syn_03027_2000", "document_hash": "10326080979f71b625e4110c7c19767e5a752e267de5034f7e37d98cb6e06249", "question_id": "q00_single_fact", "split": "test", "cluster_id": "syn_03027_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 2, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1955, "query_tokens": 594, "message_tokens": 512, "message_utf8_bytes": 1087, "cache_tensor_bytes": null, "serialized_payload_bytes": 1087, "side_information_bytes": 0, "fetched_bytes": 854, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 13451.244542025961, "hops": {"1": 6913.723333040252, "2": 6537.521208985709}, "receiver_prefill_ms": 226.35254205670208, "ttft_ms": 12.28716701734811, "decode_ms": 12.075082981027663}, "total_wall_ms": 13702.642832999118, "ttft_ms": 12.28716701734811, "input_usage": {"lookups": 4}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 63, "facts_retained": 27, "retention_rate": 0.42857142857142855, "retention_by_kind": {"person": 0.5, "date": 0.125, "amount": 0.6153846153846154, "status": 0.75, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1087}, "ledger_by_hop": {"1": {"facts_total": 63, "facts_retained": 27, "retention_rate": 0.42857142857142855, "retention_by_kind": {"person": 0.5, "date": 0.125, "amount": 0.6153846153846154, "status": 0.75, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1087}, "2": {"facts_total": 63, "facts_retained": 27, "retention_rate": 0.42857142857142855, "retention_by_kind": {"person": 0.5, "date": 0.125, "amount": 0.6153846153846154, "status": 0.75, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1087}}}, "reference_key_id": "keys/synthetic/syn_03027_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "1f88f0e78c63504e418dea843641b99a98555e15d47b02cd5cb2f57da3003dea", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "5cf7b6c029ac4818104d7115", "attempt_id": "a6d7f037b37e", "run_group_id": "E4_core:A4|repairable|qwen3b:h2", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:30:53.558294Z", "task_family": "synthetic", "document_id": "syn_03027_2000", "document_hash": "10326080979f71b625e4110c7c19767e5a752e267de5034f7e37d98cb6e06249", "question_id": "q01_single_fact", "split": "test", "cluster_id": "syn_03027_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 2, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1955, "query_tokens": 601, "message_tokens": 512, "message_utf8_bytes": 1087, "cache_tensor_bytes": null, "serialized_payload_bytes": 1087, "side_information_bytes": 0, "fetched_bytes": 854, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 13451.244542025961, "hops": {"1": 6913.723333040252, "2": 6537.521208985709}, "receiver_prefill_ms": 226.64000000804663, "ttft_ms": 12.293500010855496, "decode_ms": 143.57829198706895}, "total_wall_ms": 13834.429916925728, "ttft_ms": 12.293500010855496, "input_usage": {"lookups": 4}, "output_usage": {"output_tokens": 12}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "Lot 247-S, Baie-Comeau", "evidence_artifact": {"ledger": {"facts_total": 63, "facts_retained": 27, "retention_rate": 0.42857142857142855, "retention_by_kind": {"person": 0.5, "date": 0.125, "amount": 0.6153846153846154, "status": 0.75, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1087}, "ledger_by_hop": {"1": {"facts_total": 63, "facts_retained": 27, "retention_rate": 0.42857142857142855, "retention_by_kind": {"person": 0.5, "date": 0.125, "amount": 0.6153846153846154, "status": 0.75, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1087}, "2": {"facts_total": 63, "facts_retained": 27, "retention_rate": 0.42857142857142855, "retention_by_kind": {"person": 0.5, "date": 0.125, "amount": 0.6153846153846154, "status": 0.75, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1087}}}, "reference_key_id": "keys/synthetic/syn_03027_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 0.8571428571428571}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "1f88f0e78c63504e418dea843641b99a98555e15d47b02cd5cb2f57da3003dea", "output_hash": "77b7da1d43e7d8815e9e9159ebbd735045bc7f668260ba30b93674f70622b8ec"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "08a8bc058ab123854ed02cdc", "attempt_id": "b9ab57669002", "run_group_id": "E4_core:A4|repairable|qwen3b:h2", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:30:53.958453Z", "task_family": "synthetic", "document_id": "syn_03027_2000", "document_hash": "10326080979f71b625e4110c7c19767e5a752e267de5034f7e37d98cb6e06249", "question_id": "q02_exact_copy", "split": "test", "cluster_id": "syn_03027_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 2, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1955, "query_tokens": 600, "message_tokens": 512, "message_utf8_bytes": 1087, "cache_tensor_bytes": null, "serialized_payload_bytes": 1087, "side_information_bytes": 0, "fetched_bytes": 854, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 13451.244542025961, "hops": {"1": 6913.723333040252, "2": 6537.521208985709}, "receiver_prefill_ms": 226.69533302541822, "ttft_ms": 12.066500028595328, "decode_ms": 83.11524998862296}, "total_wall_ms": 13773.747792001814, "ttft_ms": 12.066500028595328, "input_usage": {"lookups": 4}, "output_usage": {"output_tokens": 7}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "15.0054", "evidence_artifact": {"ledger": {"facts_total": 63, "facts_retained": 27, "retention_rate": 0.42857142857142855, "retention_by_kind": {"person": 0.5, "date": 0.125, "amount": 0.6153846153846154, "status": 0.75, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1087}, "ledger_by_hop": {"1": {"facts_total": 63, "facts_retained": 27, "retention_rate": 0.42857142857142855, "retention_by_kind": {"person": 0.5, "date": 0.125, "amount": 0.6153846153846154, "status": 0.75, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1087}, "2": {"facts_total": 63, "facts_retained": 27, "retention_rate": 0.42857142857142855, "retention_by_kind": {"person": 0.5, "date": 0.125, "amount": 0.6153846153846154, "status": 0.75, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1087}}}, "reference_key_id": "keys/synthetic/syn_03027_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "1f88f0e78c63504e418dea843641b99a98555e15d47b02cd5cb2f57da3003dea", "output_hash": "d13adba73068c847c850237059e0202fc5b91ccacf2e792e9ce5e3b90556501a"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "d08ecc1335c91e9ce6267148", "attempt_id": "726638d270fd", "run_group_id": "E4_core:A4|repairable|qwen3b:h2", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:30:54.297839Z", "task_family": "synthetic", "document_id": "syn_03027_2000", "document_hash": "10326080979f71b625e4110c7c19767e5a752e267de5034f7e37d98cb6e06249", "question_id": "q03_exact_copy", "split": "test", "cluster_id": "syn_03027_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 2, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1955, "query_tokens": 595, "message_tokens": 512, "message_utf8_bytes": 1087, "cache_tensor_bytes": null, "serialized_payload_bytes": 1087, "side_information_bytes": 0, "fetched_bytes": 854, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 13451.244542025961, "hops": {"1": 6913.723333040252, "2": 6537.521208985709}, "receiver_prefill_ms": 226.5240829437971, "ttft_ms": 11.990500031970441, "decode_ms": 106.53254191856831}, "total_wall_ms": 13796.93720897194, "ttft_ms": 11.990500031970441, "input_usage": {"lookups": 4}, "output_usage": {"output_tokens": 9}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "Priyamvada Sørlie", "evidence_artifact": {"ledger": {"facts_total": 63, "facts_retained": 27, "retention_rate": 0.42857142857142855, "retention_by_kind": {"person": 0.5, "date": 0.125, "amount": 0.6153846153846154, "status": 0.75, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1087}, "ledger_by_hop": {"1": {"facts_total": 63, "facts_retained": 27, "retention_rate": 0.42857142857142855, "retention_by_kind": {"person": 0.5, "date": 0.125, "amount": 0.6153846153846154, "status": 0.75, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1087}, "2": {"facts_total": 63, "facts_retained": 27, "retention_rate": 0.42857142857142855, "retention_by_kind": {"person": 0.5, "date": 0.125, "amount": 0.6153846153846154, "status": 0.75, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1087}}}, "reference_key_id": "keys/synthetic/syn_03027_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "1f88f0e78c63504e418dea843641b99a98555e15d47b02cd5cb2f57da3003dea", "output_hash": "c5f554196ed23dd71ad995ec89f7142099d524dbb6534cc10d94a0c89d13e792"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "b06892989a5155729f89fa32", "attempt_id": "5f63ac8f803c", "run_group_id": "E4_core:A4|repairable|qwen3b:h2", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:30:54.660596Z", "task_family": "synthetic", "document_id": "syn_03027_2000", "document_hash": "10326080979f71b625e4110c7c19767e5a752e267de5034f7e37d98cb6e06249", "question_id": "q04_multi_hop", "split": "test", "cluster_id": "syn_03027_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 2, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1955, "query_tokens": 606, "message_tokens": 512, "message_utf8_bytes": 1087, "cache_tensor_bytes": null, "serialized_payload_bytes": 1087, "side_information_bytes": 0, "fetched_bytes": 854, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 13451.244542025961, "hops": {"1": 6913.723333040252, "2": 6537.521208985709}, "receiver_prefill_ms": 226.84458293952048, "ttft_ms": 12.060915934853256, "decode_ms": 11.918332893401384}, "total_wall_ms": 13702.739875065163, "ttft_ms": 12.060915934853256, "input_usage": {"lookups": 4}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 63, "facts_retained": 27, "retention_rate": 0.42857142857142855, "retention_by_kind": {"person": 0.5, "date": 0.125, "amount": 0.6153846153846154, "status": 0.75, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1087}, "ledger_by_hop": {"1": {"facts_total": 63, "facts_retained": 27, "retention_rate": 0.42857142857142855, "retention_by_kind": {"person": 0.5, "date": 0.125, "amount": 0.6153846153846154, "status": 0.75, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1087}, "2": {"facts_total": 63, "facts_retained": 27, "retention_rate": 0.42857142857142855, "retention_by_kind": {"person": 0.5, "date": 0.125, "amount": 0.6153846153846154, "status": 0.75, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1087}}}, "reference_key_id": "keys/synthetic/syn_03027_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "1f88f0e78c63504e418dea843641b99a98555e15d47b02cd5cb2f57da3003dea", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "fff346f880d82842e6b4cbea", "attempt_id": "0dc009303f2f", "run_group_id": "E4_core:A4|repairable|qwen3b:h2", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:30:54.929017Z", "task_family": "synthetic", "document_id": "syn_03027_2000", "document_hash": "10326080979f71b625e4110c7c19767e5a752e267de5034f7e37d98cb6e06249", "question_id": "q05_multi_hop", "split": "test", "cluster_id": "syn_03027_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 2, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1955, "query_tokens": 603, "message_tokens": 512, "message_utf8_bytes": 1087, "cache_tensor_bytes": null, "serialized_payload_bytes": 1087, "side_information_bytes": 0, "fetched_bytes": 854, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 13451.244542025961, "hops": {"1": 6913.723333040252, "2": 6537.521208985709}, "receiver_prefill_ms": 226.62308299914002, "ttft_ms": 12.006666045635939, "decode_ms": 11.951415915973485}, "total_wall_ms": 13702.495709061623, "ttft_ms": 12.006666045635939, "input_usage": {"lookups": 4}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 63, "facts_retained": 27, "retention_rate": 0.42857142857142855, "retention_by_kind": {"person": 0.5, "date": 0.125, "amount": 0.6153846153846154, "status": 0.75, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1087}, "ledger_by_hop": {"1": {"facts_total": 63, "facts_retained": 27, "retention_rate": 0.42857142857142855, "retention_by_kind": {"person": 0.5, "date": 0.125, "amount": 0.6153846153846154, "status": 0.75, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1087}, "2": {"facts_total": 63, "facts_retained": 27, "retention_rate": 0.42857142857142855, "retention_by_kind": {"person": 0.5, "date": 0.125, "amount": 0.6153846153846154, "status": 0.75, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1087}}}, "reference_key_id": "keys/synthetic/syn_03027_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "1f88f0e78c63504e418dea843641b99a98555e15d47b02cd5cb2f57da3003dea", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "d62e55029a0ba2cd7e09c719", "attempt_id": "41e92ce0fe55", "run_group_id": "E4_core:A4|repairable|qwen3b:h2", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:30:55.197097Z", "task_family": "synthetic", "document_id": "syn_03027_2000", "document_hash": "10326080979f71b625e4110c7c19767e5a752e267de5034f7e37d98cb6e06249", "question_id": "q06_negation", "split": "test", "cluster_id": "syn_03027_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 2, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1955, "query_tokens": 595, "message_tokens": 512, "message_utf8_bytes": 1087, "cache_tensor_bytes": null, "serialized_payload_bytes": 1087, "side_information_bytes": 0, "fetched_bytes": 854, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 13451.244542025961, "hops": {"1": 6913.723333040252, "2": 6537.521208985709}, "receiver_prefill_ms": 226.68587497901171, "ttft_ms": 11.990415980108082, "decode_ms": 11.908792075701058}, "total_wall_ms": 13702.452500117943, "ttft_ms": 11.990415980108082, "input_usage": {"lookups": 4}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 63, "facts_retained": 27, "retention_rate": 0.42857142857142855, "retention_by_kind": {"person": 0.5, "date": 0.125, "amount": 0.6153846153846154, "status": 0.75, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1087}, "ledger_by_hop": {"1": {"facts_total": 63, "facts_retained": 27, "retention_rate": 0.42857142857142855, "retention_by_kind": {"person": 0.5, "date": 0.125, "amount": 0.6153846153846154, "status": 0.75, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1087}, "2": {"facts_total": 63, "facts_retained": 27, "retention_rate": 0.42857142857142855, "retention_by_kind": {"person": 0.5, "date": 0.125, "amount": 0.6153846153846154, "status": 0.75, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1087}}}, "reference_key_id": "keys/synthetic/syn_03027_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "1f88f0e78c63504e418dea843641b99a98555e15d47b02cd5cb2f57da3003dea", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "6da39d4d588b902897721ebe", "attempt_id": "7c8707a7d761", "run_group_id": "E4_core:A4|repairable|qwen3b:h2", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:30:55.465343Z", "task_family": "synthetic", "document_id": "syn_03027_2000", "document_hash": "10326080979f71b625e4110c7c19767e5a752e267de5034f7e37d98cb6e06249", "question_id": "q07_negation", "split": "test", "cluster_id": "syn_03027_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 2, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1955, "query_tokens": 596, "message_tokens": 512, "message_utf8_bytes": 1087, "cache_tensor_bytes": null, "serialized_payload_bytes": 1087, "side_information_bytes": 0, "fetched_bytes": 854, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 13451.244542025961, "hops": {"1": 6913.723333040252, "2": 6537.521208985709}, "receiver_prefill_ms": 226.7853330122307, "ttft_ms": 11.99666690081358, "decode_ms": 11.910333065316081}, "total_wall_ms": 13702.584250015207, "ttft_ms": 11.99666690081358, "input_usage": {"lookups": 4}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "no", "evidence_artifact": {"ledger": {"facts_total": 63, "facts_retained": 27, "retention_rate": 0.42857142857142855, "retention_by_kind": {"person": 0.5, "date": 0.125, "amount": 0.6153846153846154, "status": 0.75, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1087}, "ledger_by_hop": {"1": {"facts_total": 63, "facts_retained": 27, "retention_rate": 0.42857142857142855, "retention_by_kind": {"person": 0.5, "date": 0.125, "amount": 0.6153846153846154, "status": 0.75, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1087}, "2": {"facts_total": 63, "facts_retained": 27, "retention_rate": 0.42857142857142855, "retention_by_kind": {"person": 0.5, "date": 0.125, "amount": 0.6153846153846154, "status": 0.75, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1087}}}, "reference_key_id": "keys/synthetic/syn_03027_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "1f88f0e78c63504e418dea843641b99a98555e15d47b02cd5cb2f57da3003dea", "output_hash": "9390298f3fb0c5b160498935d79cb139aef28e1c47358b4bbba61862b9c26e59"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "a83ba17574baa40d711dab88", "attempt_id": "0c93cab1f249", "run_group_id": "E4_core:A4|repairable|qwen3b:h2", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:30:55.733341Z", "task_family": "synthetic", "document_id": "syn_03027_2000", "document_hash": "10326080979f71b625e4110c7c19767e5a752e267de5034f7e37d98cb6e06249", "question_id": "q08_revision", "split": "test", "cluster_id": "syn_03027_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 2, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1955, "query_tokens": 610, "message_tokens": 512, "message_utf8_bytes": 1087, "cache_tensor_bytes": null, "serialized_payload_bytes": 1087, "side_information_bytes": 0, "fetched_bytes": 854, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 13451.244542025961, "hops": {"1": 6913.723333040252, "2": 6537.521208985709}, "receiver_prefill_ms": 228.2851659692824, "ttft_ms": 12.088999967090786, "decode_ms": 83.26541702263057}, "total_wall_ms": 13775.558584020473, "ttft_ms": 12.088999967090786, "input_usage": {"lookups": 4}, "output_usage": {"output_tokens": 7}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "8010205", "evidence_artifact": {"ledger": {"facts_total": 63, "facts_retained": 27, "retention_rate": 0.42857142857142855, "retention_by_kind": {"person": 0.5, "date": 0.125, "amount": 0.6153846153846154, "status": 0.75, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1087}, "ledger_by_hop": {"1": {"facts_total": 63, "facts_retained": 27, "retention_rate": 0.42857142857142855, "retention_by_kind": {"person": 0.5, "date": 0.125, "amount": 0.6153846153846154, "status": 0.75, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1087}, "2": {"facts_total": 63, "facts_retained": 27, "retention_rate": 0.42857142857142855, "retention_by_kind": {"person": 0.5, "date": 0.125, "amount": 0.6153846153846154, "status": 0.75, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1087}}}, "reference_key_id": "keys/synthetic/syn_03027_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "1f88f0e78c63504e418dea843641b99a98555e15d47b02cd5cb2f57da3003dea", "output_hash": "6d1ceb36132f38bd77e8f46c23ea21dd9bcff78bc80bd8a9eaf2d7f2f3bacf46"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "33c1495d050ae225bec74528", "attempt_id": "041173017e3f", "run_group_id": "E4_core:A4|repairable|qwen3b:h2", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:30:56.074276Z", "task_family": "synthetic", "document_id": "syn_03027_2000", "document_hash": "10326080979f71b625e4110c7c19767e5a752e267de5034f7e37d98cb6e06249", "question_id": "q09_revision", "split": "test", "cluster_id": "syn_03027_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 2, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1955, "query_tokens": 603, "message_tokens": 512, "message_utf8_bytes": 1087, "cache_tensor_bytes": null, "serialized_payload_bytes": 1087, "side_information_bytes": 0, "fetched_bytes": 854, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 13451.244542025961, "hops": {"1": 6913.723333040252, "2": 6537.521208985709}, "receiver_prefill_ms": 226.19133302941918, "ttft_ms": 12.269999948330224, "decode_ms": 82.75495795533061}, "total_wall_ms": 13773.125833948143, "ttft_ms": 12.269999948330224, "input_usage": {"lookups": 4}, "output_usage": {"output_tokens": 7}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "6056150", "evidence_artifact": {"ledger": {"facts_total": 63, "facts_retained": 27, "retention_rate": 0.42857142857142855, "retention_by_kind": {"person": 0.5, "date": 0.125, "amount": 0.6153846153846154, "status": 0.75, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1087}, "ledger_by_hop": {"1": {"facts_total": 63, "facts_retained": 27, "retention_rate": 0.42857142857142855, "retention_by_kind": {"person": 0.5, "date": 0.125, "amount": 0.6153846153846154, "status": 0.75, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1087}, "2": {"facts_total": 63, "facts_retained": 27, "retention_rate": 0.42857142857142855, "retention_by_kind": {"person": 0.5, "date": 0.125, "amount": 0.6153846153846154, "status": 0.75, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1087}}}, "reference_key_id": "keys/synthetic/syn_03027_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "1f88f0e78c63504e418dea843641b99a98555e15d47b02cd5cb2f57da3003dea", "output_hash": "f016f8b5fc26708e6c1ff75ff8e98659082c41a168f273819ac48b88feea8d3f"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "c7cdc4b421e72469d18bd94f", "attempt_id": "394a75b23025", "run_group_id": "E4_core:A4|repairable|qwen3b:h2", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:30:56.412546Z", "task_family": "synthetic", "document_id": "syn_03027_2000", "document_hash": "10326080979f71b625e4110c7c19767e5a752e267de5034f7e37d98cb6e06249", "question_id": "q10_temporal", "split": "test", "cluster_id": "syn_03027_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 2, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1955, "query_tokens": 613, "message_tokens": 512, "message_utf8_bytes": 1087, "cache_tensor_bytes": null, "serialized_payload_bytes": 1087, "side_information_bytes": 0, "fetched_bytes": 854, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 13451.244542025961, "hops": {"1": 6913.723333040252, "2": 6537.521208985709}, "receiver_prefill_ms": 228.50770794320852, "ttft_ms": 12.111791060306132, "decode_ms": 82.78241602238268}, "total_wall_ms": 13775.284833973274, "ttft_ms": 12.111791060306132, "input_usage": {"lookups": 4}, "output_usage": {"output_tokens": 7}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "Lunlith-13", "evidence_artifact": {"ledger": {"facts_total": 63, "facts_retained": 27, "retention_rate": 0.42857142857142855, "retention_by_kind": {"person": 0.5, "date": 0.125, "amount": 0.6153846153846154, "status": 0.75, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1087}, "ledger_by_hop": {"1": {"facts_total": 63, "facts_retained": 27, "retention_rate": 0.42857142857142855, "retention_by_kind": {"person": 0.5, "date": 0.125, "amount": 0.6153846153846154, "status": 0.75, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1087}, "2": {"facts_total": 63, "facts_retained": 27, "retention_rate": 0.42857142857142855, "retention_by_kind": {"person": 0.5, "date": 0.125, "amount": 0.6153846153846154, "status": 0.75, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1087}}}, "reference_key_id": "keys/synthetic/syn_03027_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "1f88f0e78c63504e418dea843641b99a98555e15d47b02cd5cb2f57da3003dea", "output_hash": "c45f000afe6d865c8167954eebdff761141225817d4f5b26c962cd50cd75d391"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "d9de66d05b9d383c900d69a5", "attempt_id": "3c26a3fa35cb", "run_group_id": "E4_core:A4|repairable|qwen3b:h2", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:30:56.753059Z", "task_family": "synthetic", "document_id": "syn_03027_2000", "document_hash": "10326080979f71b625e4110c7c19767e5a752e267de5034f7e37d98cb6e06249", "question_id": "q11_aggregation", "split": "test", "cluster_id": "syn_03027_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 2, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1955, "query_tokens": 606, "message_tokens": 512, "message_utf8_bytes": 1087, "cache_tensor_bytes": null, "serialized_payload_bytes": 1087, "side_information_bytes": 0, "fetched_bytes": 854, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 13451.244542025961, "hops": {"1": 6913.723333040252, "2": 6537.521208985709}, "receiver_prefill_ms": 226.88262502197176, "ttft_ms": 12.022458016872406, "decode_ms": 106.70112492516637}, "total_wall_ms": 13797.50537511427, "ttft_ms": 12.022458016872406, "input_usage": {"lookups": 4}, "output_usage": {"output_tokens": 9}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "23,450165", "evidence_artifact": {"ledger": {"facts_total": 63, "facts_retained": 27, "retention_rate": 0.42857142857142855, "retention_by_kind": {"person": 0.5, "date": 0.125, "amount": 0.6153846153846154, "status": 0.75, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1087}, "ledger_by_hop": {"1": {"facts_total": 63, "facts_retained": 27, "retention_rate": 0.42857142857142855, "retention_by_kind": {"person": 0.5, "date": 0.125, "amount": 0.6153846153846154, "status": 0.75, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1087}, "2": {"facts_total": 63, "facts_retained": 27, "retention_rate": 0.42857142857142855, "retention_by_kind": {"person": 0.5, "date": 0.125, "amount": 0.6153846153846154, "status": 0.75, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1087}}}, "reference_key_id": "keys/synthetic/syn_03027_2000.json", "task_scores": {"correct": 0.0, "abstained": 0.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "1f88f0e78c63504e418dea843641b99a98555e15d47b02cd5cb2f57da3003dea", "output_hash": "a929e5c5097069e202c39470de0d2861ac8d456564e72788db30daadb9f19129"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "57b4ec7e0d27a3d7fc0cf8e4", "attempt_id": "02b853b2d02e", "run_group_id": "E4_core:A4|repairable|qwen3b:h2", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:30:57.115881Z", "task_family": "synthetic", "document_id": "syn_03027_2000", "document_hash": "10326080979f71b625e4110c7c19767e5a752e267de5034f7e37d98cb6e06249", "question_id": "q12_unit_conversion", "split": "test", "cluster_id": "syn_03027_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 2, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1955, "query_tokens": 603, "message_tokens": 512, "message_utf8_bytes": 1087, "cache_tensor_bytes": null, "serialized_payload_bytes": 1087, "side_information_bytes": 0, "fetched_bytes": 854, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 13451.244542025961, "hops": {"1": 6913.723333040252, "2": 6537.521208985709}, "receiver_prefill_ms": 226.58875002525747, "ttft_ms": 12.043667025864124, "decode_ms": 11.88120897859335}, "total_wall_ms": 13702.396042062901, "ttft_ms": 12.043667025864124, "input_usage": {"lookups": 4}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 63, "facts_retained": 27, "retention_rate": 0.42857142857142855, "retention_by_kind": {"person": 0.5, "date": 0.125, "amount": 0.6153846153846154, "status": 0.75, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1087}, "ledger_by_hop": {"1": {"facts_total": 63, "facts_retained": 27, "retention_rate": 0.42857142857142855, "retention_by_kind": {"person": 0.5, "date": 0.125, "amount": 0.6153846153846154, "status": 0.75, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1087}, "2": {"facts_total": 63, "facts_retained": 27, "retention_rate": 0.42857142857142855, "retention_by_kind": {"person": 0.5, "date": 0.125, "amount": 0.6153846153846154, "status": 0.75, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1087}}}, "reference_key_id": "keys/synthetic/syn_03027_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "1f88f0e78c63504e418dea843641b99a98555e15d47b02cd5cb2f57da3003dea", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "b696778534835305efdbfb84", "attempt_id": "5f15c6112c61", "run_group_id": "E4_core:A4|repairable|qwen3b:h2", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:30:57.383842Z", "task_family": "synthetic", "document_id": "syn_03027_2000", "document_hash": "10326080979f71b625e4110c7c19767e5a752e267de5034f7e37d98cb6e06249", "question_id": "q13_unanswerable", "split": "test", "cluster_id": "syn_03027_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 2, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1955, "query_tokens": 606, "message_tokens": 512, "message_utf8_bytes": 1087, "cache_tensor_bytes": null, "serialized_payload_bytes": 1087, "side_information_bytes": 0, "fetched_bytes": 854, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 13451.244542025961, "hops": {"1": 6913.723333040252, "2": 6537.521208985709}, "receiver_prefill_ms": 226.85708303470165, "ttft_ms": 12.016250053420663, "decode_ms": 11.925583006814122}, "total_wall_ms": 13702.708334079944, "ttft_ms": 12.016250053420663, "input_usage": {"lookups": 4}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 63, "facts_retained": 27, "retention_rate": 0.42857142857142855, "retention_by_kind": {"person": 0.5, "date": 0.125, "amount": 0.6153846153846154, "status": 0.75, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1087}, "ledger_by_hop": {"1": {"facts_total": 63, "facts_retained": 27, "retention_rate": 0.42857142857142855, "retention_by_kind": {"person": 0.5, "date": 0.125, "amount": 0.6153846153846154, "status": 0.75, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1087}, "2": {"facts_total": 63, "facts_retained": 27, "retention_rate": 0.42857142857142855, "retention_by_kind": {"person": 0.5, "date": 0.125, "amount": 0.6153846153846154, "status": 0.75, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1087}}}, "reference_key_id": "keys/synthetic/syn_03027_2000.json", "task_scores": {"correct": 1.0, "abstained": 1.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "1f88f0e78c63504e418dea843641b99a98555e15d47b02cd5cb2f57da3003dea", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "0a76d42e6f7af180e2ae75f9", "attempt_id": "1ea029929a79", "run_group_id": "E4_core:A4|repairable|qwen3b:h2", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:30:57.652009Z", "task_family": "synthetic", "document_id": "syn_03027_2000", "document_hash": "10326080979f71b625e4110c7c19767e5a752e267de5034f7e37d98cb6e06249", "question_id": "q14_unanswerable", "split": "test", "cluster_id": "syn_03027_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 2, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1955, "query_tokens": 608, "message_tokens": 512, "message_utf8_bytes": 1087, "cache_tensor_bytes": null, "serialized_payload_bytes": 1087, "side_information_bytes": 0, "fetched_bytes": 854, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 13451.244542025961, "hops": {"1": 6913.723333040252, "2": 6537.521208985709}, "receiver_prefill_ms": 227.03512501902878, "ttft_ms": 12.07654201425612, "decode_ms": 11.895833071321249}, "total_wall_ms": 13702.917042071931, "ttft_ms": 12.07654201425612, "input_usage": {"lookups": 4}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 63, "facts_retained": 27, "retention_rate": 0.42857142857142855, "retention_by_kind": {"person": 0.5, "date": 0.125, "amount": 0.6153846153846154, "status": 0.75, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1087}, "ledger_by_hop": {"1": {"facts_total": 63, "facts_retained": 27, "retention_rate": 0.42857142857142855, "retention_by_kind": {"person": 0.5, "date": 0.125, "amount": 0.6153846153846154, "status": 0.75, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1087}, "2": {"facts_total": 63, "facts_retained": 27, "retention_rate": 0.42857142857142855, "retention_by_kind": {"person": 0.5, "date": 0.125, "amount": 0.6153846153846154, "status": 0.75, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1087}}}, "reference_key_id": "keys/synthetic/syn_03027_2000.json", "task_scores": {"correct": 1.0, "abstained": 1.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "1f88f0e78c63504e418dea843641b99a98555e15d47b02cd5cb2f57da3003dea", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "d9edfad527989a186700f618", "attempt_id": "9ab32be73dd6", "run_group_id": "E4_core:A4|repairable|qwen3b:h4", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:31:10.898438Z", "task_family": "synthetic", "document_id": "syn_03027_2000", "document_hash": "10326080979f71b625e4110c7c19767e5a752e267de5034f7e37d98cb6e06249", "question_id": "q00_single_fact", "split": "test", "cluster_id": "syn_03027_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 4, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1955, "query_tokens": 594, "message_tokens": 512, "message_utf8_bytes": 1087, "cache_tensor_bytes": null, "serialized_payload_bytes": 1087, "side_information_bytes": 0, "fetched_bytes": 2562, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 26426.132791908458, "hops": {"1": 6913.723333040252, "2": 6537.521208985709, "3": 6488.541791914031, "4": 6486.346457968466}, "receiver_prefill_ms": 226.20100004132837, "ttft_ms": 12.043249909766018, "decode_ms": 11.869083042256534}, "total_wall_ms": 26676.909374888055, "ttft_ms": 12.043249909766018, "input_usage": {"lookups": 12}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 63, "facts_retained": 27, "retention_rate": 0.42857142857142855, "retention_by_kind": {"person": 0.5, "date": 0.125, "amount": 0.6153846153846154, "status": 0.75, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1087}, "ledger_by_hop": {"1": {"facts_total": 63, "facts_retained": 27, "retention_rate": 0.42857142857142855, "retention_by_kind": {"person": 0.5, "date": 0.125, "amount": 0.6153846153846154, "status": 0.75, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1087}, "2": {"facts_total": 63, "facts_retained": 27, "retention_rate": 0.42857142857142855, "retention_by_kind": {"person": 0.5, "date": 0.125, "amount": 0.6153846153846154, "status": 0.75, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1087}, "3": {"facts_total": 63, "facts_retained": 27, "retention_rate": 0.42857142857142855, "retention_by_kind": {"person": 0.5, "date": 0.125, "amount": 0.6153846153846154, "status": 0.75, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1087}, "4": {"facts_total": 63, "facts_retained": 27, "retention_rate": 0.42857142857142855, "retention_by_kind": {"person": 0.5, "date": 0.125, "amount": 0.6153846153846154, "status": 0.75, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1087}}}, "reference_key_id": "keys/synthetic/syn_03027_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "1f88f0e78c63504e418dea843641b99a98555e15d47b02cd5cb2f57da3003dea", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "89cc6c4231d348a92ae1c360", "attempt_id": "554978bdb979", "run_group_id": "E4_core:A4|repairable|qwen3b:h4", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:31:11.166060Z", "task_family": "synthetic", "document_id": "syn_03027_2000", "document_hash": "10326080979f71b625e4110c7c19767e5a752e267de5034f7e37d98cb6e06249", "question_id": "q01_single_fact", "split": "test", "cluster_id": "syn_03027_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 4, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1955, "query_tokens": 601, "message_tokens": 512, "message_utf8_bytes": 1087, "cache_tensor_bytes": null, "serialized_payload_bytes": 1087, "side_information_bytes": 0, "fetched_bytes": 2562, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 26426.132791908458, "hops": {"1": 6913.723333040252, "2": 6537.521208985709, "3": 6488.541791914031, "4": 6486.346457968466}, "receiver_prefill_ms": 226.74266702961177, "ttft_ms": 12.103458982892334, "decode_ms": 142.3419580096379}, "total_wall_ms": 26807.97545891255, "ttft_ms": 12.103458982892334, "input_usage": {"lookups": 12}, "output_usage": {"output_tokens": 12}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "Lot 247-S, Baie-Comeau", "evidence_artifact": {"ledger": {"facts_total": 63, "facts_retained": 27, "retention_rate": 0.42857142857142855, "retention_by_kind": {"person": 0.5, "date": 0.125, "amount": 0.6153846153846154, "status": 0.75, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1087}, "ledger_by_hop": {"1": {"facts_total": 63, "facts_retained": 27, "retention_rate": 0.42857142857142855, "retention_by_kind": {"person": 0.5, "date": 0.125, "amount": 0.6153846153846154, "status": 0.75, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1087}, "2": {"facts_total": 63, "facts_retained": 27, "retention_rate": 0.42857142857142855, "retention_by_kind": {"person": 0.5, "date": 0.125, "amount": 0.6153846153846154, "status": 0.75, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1087}, "3": {"facts_total": 63, "facts_retained": 27, "retention_rate": 0.42857142857142855, "retention_by_kind": {"person": 0.5, "date": 0.125, "amount": 0.6153846153846154, "status": 0.75, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1087}, "4": {"facts_total": 63, "facts_retained": 27, "retention_rate": 0.42857142857142855, "retention_by_kind": {"person": 0.5, "date": 0.125, "amount": 0.6153846153846154, "status": 0.75, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1087}}}, "reference_key_id": "keys/synthetic/syn_03027_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 0.8571428571428571}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "1f88f0e78c63504e418dea843641b99a98555e15d47b02cd5cb2f57da3003dea", "output_hash": "77b7da1d43e7d8815e9e9159ebbd735045bc7f668260ba30b93674f70622b8ec"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "7cb051c1c46a0bcf7a727733", "attempt_id": "1aa4eb2e9bf4", "run_group_id": "E4_core:A4|repairable|qwen3b:h4", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:31:11.564427Z", "task_family": "synthetic", "document_id": "syn_03027_2000", "document_hash": "10326080979f71b625e4110c7c19767e5a752e267de5034f7e37d98cb6e06249", "question_id": "q02_exact_copy", "split": "test", "cluster_id": "syn_03027_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 4, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1955, "query_tokens": 600, "message_tokens": 512, "message_utf8_bytes": 1087, "cache_tensor_bytes": null, "serialized_payload_bytes": 1087, "side_information_bytes": 0, "fetched_bytes": 2562, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 26426.132791908458, "hops": {"1": 6913.723333040252, "2": 6537.521208985709, "3": 6488.541791914031, "4": 6486.346457968466}, "receiver_prefill_ms": 226.67437500786036, "ttft_ms": 11.96850009728223, "decode_ms": 84.03270808048546}, "total_wall_ms": 26749.359791981988, "ttft_ms": 11.96850009728223, "input_usage": {"lookups": 12}, "output_usage": {"output_tokens": 7}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "15.0054", "evidence_artifact": {"ledger": {"facts_total": 63, "facts_retained": 27, "retention_rate": 0.42857142857142855, "retention_by_kind": {"person": 0.5, "date": 0.125, "amount": 0.6153846153846154, "status": 0.75, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1087}, "ledger_by_hop": {"1": {"facts_total": 63, "facts_retained": 27, "retention_rate": 0.42857142857142855, "retention_by_kind": {"person": 0.5, "date": 0.125, "amount": 0.6153846153846154, "status": 0.75, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1087}, "2": {"facts_total": 63, "facts_retained": 27, "retention_rate": 0.42857142857142855, "retention_by_kind": {"person": 0.5, "date": 0.125, "amount": 0.6153846153846154, "status": 0.75, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1087}, "3": {"facts_total": 63, "facts_retained": 27, "retention_rate": 0.42857142857142855, "retention_by_kind": {"person": 0.5, "date": 0.125, "amount": 0.6153846153846154, "status": 0.75, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1087}, "4": {"facts_total": 63, "facts_retained": 27, "retention_rate": 0.42857142857142855, "retention_by_kind": {"person": 0.5, "date": 0.125, "amount": 0.6153846153846154, "status": 0.75, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1087}}}, "reference_key_id": "keys/synthetic/syn_03027_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "1f88f0e78c63504e418dea843641b99a98555e15d47b02cd5cb2f57da3003dea", "output_hash": "d13adba73068c847c850237059e0202fc5b91ccacf2e792e9ce5e3b90556501a"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "6d9f808101708f68c960c3aa", "attempt_id": "447a22f1d5ab", "run_group_id": "E4_core:A4|repairable|qwen3b:h4", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:31:11.904778Z", "task_family": "synthetic", "document_id": "syn_03027_2000", "document_hash": "10326080979f71b625e4110c7c19767e5a752e267de5034f7e37d98cb6e06249", "question_id": "q03_exact_copy", "split": "test", "cluster_id": "syn_03027_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 4, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1955, "query_tokens": 595, "message_tokens": 512, "message_utf8_bytes": 1087, "cache_tensor_bytes": null, "serialized_payload_bytes": 1087, "side_information_bytes": 0, "fetched_bytes": 2562, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 26426.132791908458, "hops": {"1": 6913.723333040252, "2": 6537.521208985709, "3": 6488.541791914031, "4": 6486.346457968466}, "receiver_prefill_ms": 226.30774998106062, "ttft_ms": 12.088541989214718, "decode_ms": 106.47579201031476}, "total_wall_ms": 26771.668457891792, "ttft_ms": 12.088541989214718, "input_usage": {"lookups": 12}, "output_usage": {"output_tokens": 9}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "Priyamvada Sørlie", "evidence_artifact": {"ledger": {"facts_total": 63, "facts_retained": 27, "retention_rate": 0.42857142857142855, "retention_by_kind": {"person": 0.5, "date": 0.125, "amount": 0.6153846153846154, "status": 0.75, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1087}, "ledger_by_hop": {"1": {"facts_total": 63, "facts_retained": 27, "retention_rate": 0.42857142857142855, "retention_by_kind": {"person": 0.5, "date": 0.125, "amount": 0.6153846153846154, "status": 0.75, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1087}, "2": {"facts_total": 63, "facts_retained": 27, "retention_rate": 0.42857142857142855, "retention_by_kind": {"person": 0.5, "date": 0.125, "amount": 0.6153846153846154, "status": 0.75, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1087}, "3": {"facts_total": 63, "facts_retained": 27, "retention_rate": 0.42857142857142855, "retention_by_kind": {"person": 0.5, "date": 0.125, "amount": 0.6153846153846154, "status": 0.75, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1087}, "4": {"facts_total": 63, "facts_retained": 27, "retention_rate": 0.42857142857142855, "retention_by_kind": {"person": 0.5, "date": 0.125, "amount": 0.6153846153846154, "status": 0.75, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1087}}}, "reference_key_id": "keys/synthetic/syn_03027_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "1f88f0e78c63504e418dea843641b99a98555e15d47b02cd5cb2f57da3003dea", "output_hash": "c5f554196ed23dd71ad995ec89f7142099d524dbb6534cc10d94a0c89d13e792"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "7fbef4fe984176467b6c5f17", "attempt_id": "738c12f40b4c", "run_group_id": "E4_core:A4|repairable|qwen3b:h4", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:31:12.266795Z", "task_family": "synthetic", "document_id": "syn_03027_2000", "document_hash": "10326080979f71b625e4110c7c19767e5a752e267de5034f7e37d98cb6e06249", "question_id": "q04_multi_hop", "split": "test", "cluster_id": "syn_03027_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 4, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1955, "query_tokens": 606, "message_tokens": 512, "message_utf8_bytes": 1087, "cache_tensor_bytes": null, "serialized_payload_bytes": 1087, "side_information_bytes": 0, "fetched_bytes": 2562, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 26426.132791908458, "hops": {"1": 6913.723333040252, "2": 6537.521208985709, "3": 6488.541791914031, "4": 6486.346457968466}, "receiver_prefill_ms": 226.79329197853804, "ttft_ms": 12.054541031830013, "decode_ms": 12.051792000420392}, "total_wall_ms": 26677.702624816447, "ttft_ms": 12.054541031830013, "input_usage": {"lookups": 12}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 63, "facts_retained": 27, "retention_rate": 0.42857142857142855, "retention_by_kind": {"person": 0.5, "date": 0.125, "amount": 0.6153846153846154, "status": 0.75, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1087}, "ledger_by_hop": {"1": {"facts_total": 63, "facts_retained": 27, "retention_rate": 0.42857142857142855, "retention_by_kind": {"person": 0.5, "date": 0.125, "amount": 0.6153846153846154, "status": 0.75, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1087}, "2": {"facts_total": 63, "facts_retained": 27, "retention_rate": 0.42857142857142855, "retention_by_kind": {"person": 0.5, "date": 0.125, "amount": 0.6153846153846154, "status": 0.75, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1087}, "3": {"facts_total": 63, "facts_retained": 27, "retention_rate": 0.42857142857142855, "retention_by_kind": {"person": 0.5, "date": 0.125, "amount": 0.6153846153846154, "status": 0.75, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1087}, "4": {"facts_total": 63, "facts_retained": 27, "retention_rate": 0.42857142857142855, "retention_by_kind": {"person": 0.5, "date": 0.125, "amount": 0.6153846153846154, "status": 0.75, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1087}}}, "reference_key_id": "keys/synthetic/syn_03027_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "1f88f0e78c63504e418dea843641b99a98555e15d47b02cd5cb2f57da3003dea", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "c7814552da275c5d5d653991", "attempt_id": "a2a35a8c7a74", "run_group_id": "E4_core:A4|repairable|qwen3b:h4", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:31:12.534918Z", "task_family": "synthetic", "document_id": "syn_03027_2000", "document_hash": "10326080979f71b625e4110c7c19767e5a752e267de5034f7e37d98cb6e06249", "question_id": "q05_multi_hop", "split": "test", "cluster_id": "syn_03027_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 4, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1955, "query_tokens": 603, "message_tokens": 512, "message_utf8_bytes": 1087, "cache_tensor_bytes": null, "serialized_payload_bytes": 1087, "side_information_bytes": 0, "fetched_bytes": 2562, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 26426.132791908458, "hops": {"1": 6913.723333040252, "2": 6537.521208985709, "3": 6488.541791914031, "4": 6486.346457968466}, "receiver_prefill_ms": 226.98087489698082, "ttft_ms": 12.00774998869747, "decode_ms": 11.901167104952037}, "total_wall_ms": 26677.675499930046, "ttft_ms": 12.00774998869747, "input_usage": {"lookups": 12}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 63, "facts_retained": 27, "retention_rate": 0.42857142857142855, "retention_by_kind": {"person": 0.5, "date": 0.125, "amount": 0.6153846153846154, "status": 0.75, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1087}, "ledger_by_hop": {"1": {"facts_total": 63, "facts_retained": 27, "retention_rate": 0.42857142857142855, "retention_by_kind": {"person": 0.5, "date": 0.125, "amount": 0.6153846153846154, "status": 0.75, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1087}, "2": {"facts_total": 63, "facts_retained": 27, "retention_rate": 0.42857142857142855, "retention_by_kind": {"person": 0.5, "date": 0.125, "amount": 0.6153846153846154, "status": 0.75, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1087}, "3": {"facts_total": 63, "facts_retained": 27, "retention_rate": 0.42857142857142855, "retention_by_kind": {"person": 0.5, "date": 0.125, "amount": 0.6153846153846154, "status": 0.75, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1087}, "4": {"facts_total": 63, "facts_retained": 27, "retention_rate": 0.42857142857142855, "retention_by_kind": {"person": 0.5, "date": 0.125, "amount": 0.6153846153846154, "status": 0.75, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1087}}}, "reference_key_id": "keys/synthetic/syn_03027_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "1f88f0e78c63504e418dea843641b99a98555e15d47b02cd5cb2f57da3003dea", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "c2cb6639cb073103e140adc2", "attempt_id": "e8fa8930ff14", "run_group_id": "E4_core:A4|repairable|qwen3b:h4", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:31:12.803172Z", "task_family": "synthetic", "document_id": "syn_03027_2000", "document_hash": "10326080979f71b625e4110c7c19767e5a752e267de5034f7e37d98cb6e06249", "question_id": "q06_negation", "split": "test", "cluster_id": "syn_03027_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 4, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1955, "query_tokens": 595, "message_tokens": 512, "message_utf8_bytes": 1087, "cache_tensor_bytes": null, "serialized_payload_bytes": 1087, "side_information_bytes": 0, "fetched_bytes": 2562, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 26426.132791908458, "hops": {"1": 6913.723333040252, "2": 6537.521208985709, "3": 6488.541791914031, "4": 6486.346457968466}, "receiver_prefill_ms": 226.3444580603391, "ttft_ms": 12.020500027574599, "decode_ms": 12.003917014226317}, "total_wall_ms": 26677.15237592347, "ttft_ms": 12.020500027574599, "input_usage": {"lookups": 12}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 63, "facts_retained": 27, "retention_rate": 0.42857142857142855, "retention_by_kind": {"person": 0.5, "date": 0.125, "amount": 0.6153846153846154, "status": 0.75, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1087}, "ledger_by_hop": {"1": {"facts_total": 63, "facts_retained": 27, "retention_rate": 0.42857142857142855, "retention_by_kind": {"person": 0.5, "date": 0.125, "amount": 0.6153846153846154, "status": 0.75, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1087}, "2": {"facts_total": 63, "facts_retained": 27, "retention_rate": 0.42857142857142855, "retention_by_kind": {"person": 0.5, "date": 0.125, "amount": 0.6153846153846154, "status": 0.75, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1087}, "3": {"facts_total": 63, "facts_retained": 27, "retention_rate": 0.42857142857142855, "retention_by_kind": {"person": 0.5, "date": 0.125, "amount": 0.6153846153846154, "status": 0.75, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1087}, "4": {"facts_total": 63, "facts_retained": 27, "retention_rate": 0.42857142857142855, "retention_by_kind": {"person": 0.5, "date": 0.125, "amount": 0.6153846153846154, "status": 0.75, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1087}}}, "reference_key_id": "keys/synthetic/syn_03027_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "1f88f0e78c63504e418dea843641b99a98555e15d47b02cd5cb2f57da3003dea", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "a9e5768ece59f38e62e92ef4", "attempt_id": "7e8dd2c472ce", "run_group_id": "E4_core:A4|repairable|qwen3b:h4", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:31:13.071146Z", "task_family": "synthetic", "document_id": "syn_03027_2000", "document_hash": "10326080979f71b625e4110c7c19767e5a752e267de5034f7e37d98cb6e06249", "question_id": "q07_negation", "split": "test", "cluster_id": "syn_03027_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 4, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1955, "query_tokens": 596, "message_tokens": 512, "message_utf8_bytes": 1087, "cache_tensor_bytes": null, "serialized_payload_bytes": 1087, "side_information_bytes": 0, "fetched_bytes": 2562, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 26426.132791908458, "hops": {"1": 6913.723333040252, "2": 6537.521208985709, "3": 6488.541791914031, "4": 6486.346457968466}, "receiver_prefill_ms": 226.0023329872638, "ttft_ms": 12.39254092797637, "decode_ms": 12.073875055648386}, "total_wall_ms": 26677.256583818235, "ttft_ms": 12.39254092797637, "input_usage": {"lookups": 12}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "no", "evidence_artifact": {"ledger": {"facts_total": 63, "facts_retained": 27, "retention_rate": 0.42857142857142855, "retention_by_kind": {"person": 0.5, "date": 0.125, "amount": 0.6153846153846154, "status": 0.75, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1087}, "ledger_by_hop": {"1": {"facts_total": 63, "facts_retained": 27, "retention_rate": 0.42857142857142855, "retention_by_kind": {"person": 0.5, "date": 0.125, "amount": 0.6153846153846154, "status": 0.75, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1087}, "2": {"facts_total": 63, "facts_retained": 27, "retention_rate": 0.42857142857142855, "retention_by_kind": {"person": 0.5, "date": 0.125, "amount": 0.6153846153846154, "status": 0.75, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1087}, "3": {"facts_total": 63, "facts_retained": 27, "retention_rate": 0.42857142857142855, "retention_by_kind": {"person": 0.5, "date": 0.125, "amount": 0.6153846153846154, "status": 0.75, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1087}, "4": {"facts_total": 63, "facts_retained": 27, "retention_rate": 0.42857142857142855, "retention_by_kind": {"person": 0.5, "date": 0.125, "amount": 0.6153846153846154, "status": 0.75, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1087}}}, "reference_key_id": "keys/synthetic/syn_03027_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "1f88f0e78c63504e418dea843641b99a98555e15d47b02cd5cb2f57da3003dea", "output_hash": "9390298f3fb0c5b160498935d79cb139aef28e1c47358b4bbba61862b9c26e59"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "16081d619c57b203592d22f9", "attempt_id": "63674f47490a", "run_group_id": "E4_core:A4|repairable|qwen3b:h4", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:31:13.338648Z", "task_family": "synthetic", "document_id": "syn_03027_2000", "document_hash": "10326080979f71b625e4110c7c19767e5a752e267de5034f7e37d98cb6e06249", "question_id": "q08_revision", "split": "test", "cluster_id": "syn_03027_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 4, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1955, "query_tokens": 610, "message_tokens": 512, "message_utf8_bytes": 1087, "cache_tensor_bytes": null, "serialized_payload_bytes": 1087, "side_information_bytes": 0, "fetched_bytes": 2562, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 26426.132791908458, "hops": {"1": 6913.723333040252, "2": 6537.521208985709, "3": 6488.541791914031, "4": 6486.346457968466}, "receiver_prefill_ms": 228.54979196563363, "ttft_ms": 12.095917016267776, "decode_ms": 83.30291707534343}, "total_wall_ms": 26750.76162489131, "ttft_ms": 12.095917016267776, "input_usage": {"lookups": 12}, "output_usage": {"output_tokens": 7}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "8010205", "evidence_artifact": {"ledger": {"facts_total": 63, "facts_retained": 27, "retention_rate": 0.42857142857142855, "retention_by_kind": {"person": 0.5, "date": 0.125, "amount": 0.6153846153846154, "status": 0.75, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1087}, "ledger_by_hop": {"1": {"facts_total": 63, "facts_retained": 27, "retention_rate": 0.42857142857142855, "retention_by_kind": {"person": 0.5, "date": 0.125, "amount": 0.6153846153846154, "status": 0.75, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1087}, "2": {"facts_total": 63, "facts_retained": 27, "retention_rate": 0.42857142857142855, "retention_by_kind": {"person": 0.5, "date": 0.125, "amount": 0.6153846153846154, "status": 0.75, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1087}, "3": {"facts_total": 63, "facts_retained": 27, "retention_rate": 0.42857142857142855, "retention_by_kind": {"person": 0.5, "date": 0.125, "amount": 0.6153846153846154, "status": 0.75, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1087}, "4": {"facts_total": 63, "facts_retained": 27, "retention_rate": 0.42857142857142855, "retention_by_kind": {"person": 0.5, "date": 0.125, "amount": 0.6153846153846154, "status": 0.75, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1087}}}, "reference_key_id": "keys/synthetic/syn_03027_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "1f88f0e78c63504e418dea843641b99a98555e15d47b02cd5cb2f57da3003dea", "output_hash": "6d1ceb36132f38bd77e8f46c23ea21dd9bcff78bc80bd8a9eaf2d7f2f3bacf46"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "2547156712741e8aac6a6cdf", "attempt_id": "60a208b3fe78", "run_group_id": "E4_core:A4|repairable|qwen3b:h4", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:31:13.680175Z", "task_family": "synthetic", "document_id": "syn_03027_2000", "document_hash": "10326080979f71b625e4110c7c19767e5a752e267de5034f7e37d98cb6e06249", "question_id": "q09_revision", "split": "test", "cluster_id": "syn_03027_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 4, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1955, "query_tokens": 603, "message_tokens": 512, "message_utf8_bytes": 1087, "cache_tensor_bytes": null, "serialized_payload_bytes": 1087, "side_information_bytes": 0, "fetched_bytes": 2562, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 26426.132791908458, "hops": {"1": 6913.723333040252, "2": 6537.521208985709, "3": 6488.541791914031, "4": 6486.346457968466}, "receiver_prefill_ms": 226.7534170532599, "ttft_ms": 12.13912502862513, "decode_ms": 83.0200829077512}, "total_wall_ms": 26748.7223749049, "ttft_ms": 12.13912502862513, "input_usage": {"lookups": 12}, "output_usage": {"output_tokens": 7}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "6056150", "evidence_artifact": {"ledger": {"facts_total": 63, "facts_retained": 27, "retention_rate": 0.42857142857142855, "retention_by_kind": {"person": 0.5, "date": 0.125, "amount": 0.6153846153846154, "status": 0.75, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1087}, "ledger_by_hop": {"1": {"facts_total": 63, "facts_retained": 27, "retention_rate": 0.42857142857142855, "retention_by_kind": {"person": 0.5, "date": 0.125, "amount": 0.6153846153846154, "status": 0.75, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1087}, "2": {"facts_total": 63, "facts_retained": 27, "retention_rate": 0.42857142857142855, "retention_by_kind": {"person": 0.5, "date": 0.125, "amount": 0.6153846153846154, "status": 0.75, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1087}, "3": {"facts_total": 63, "facts_retained": 27, "retention_rate": 0.42857142857142855, "retention_by_kind": {"person": 0.5, "date": 0.125, "amount": 0.6153846153846154, "status": 0.75, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1087}, "4": {"facts_total": 63, "facts_retained": 27, "retention_rate": 0.42857142857142855, "retention_by_kind": {"person": 0.5, "date": 0.125, "amount": 0.6153846153846154, "status": 0.75, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1087}}}, "reference_key_id": "keys/synthetic/syn_03027_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "1f88f0e78c63504e418dea843641b99a98555e15d47b02cd5cb2f57da3003dea", "output_hash": "f016f8b5fc26708e6c1ff75ff8e98659082c41a168f273819ac48b88feea8d3f"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "cba452fd0ef4d0ad5eee6d55", "attempt_id": "f113f24d7ee1", "run_group_id": "E4_core:A4|repairable|qwen3b:h4", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:31:14.019455Z", "task_family": "synthetic", "document_id": "syn_03027_2000", "document_hash": "10326080979f71b625e4110c7c19767e5a752e267de5034f7e37d98cb6e06249", "question_id": "q10_temporal", "split": "test", "cluster_id": "syn_03027_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 4, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1955, "query_tokens": 613, "message_tokens": 512, "message_utf8_bytes": 1087, "cache_tensor_bytes": null, "serialized_payload_bytes": 1087, "side_information_bytes": 0, "fetched_bytes": 2562, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 26426.132791908458, "hops": {"1": 6913.723333040252, "2": 6537.521208985709, "3": 6488.541791914031, "4": 6486.346457968466}, "receiver_prefill_ms": 228.5613330313936, "ttft_ms": 12.18979200348258, "decode_ms": 83.81041698157787}, "total_wall_ms": 26751.38504197821, "ttft_ms": 12.18979200348258, "input_usage": {"lookups": 12}, "output_usage": {"output_tokens": 7}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "Lunlith-13", "evidence_artifact": {"ledger": {"facts_total": 63, "facts_retained": 27, "retention_rate": 0.42857142857142855, "retention_by_kind": {"person": 0.5, "date": 0.125, "amount": 0.6153846153846154, "status": 0.75, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1087}, "ledger_by_hop": {"1": {"facts_total": 63, "facts_retained": 27, "retention_rate": 0.42857142857142855, "retention_by_kind": {"person": 0.5, "date": 0.125, "amount": 0.6153846153846154, "status": 0.75, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1087}, "2": {"facts_total": 63, "facts_retained": 27, "retention_rate": 0.42857142857142855, "retention_by_kind": {"person": 0.5, "date": 0.125, "amount": 0.6153846153846154, "status": 0.75, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1087}, "3": {"facts_total": 63, "facts_retained": 27, "retention_rate": 0.42857142857142855, "retention_by_kind": {"person": 0.5, "date": 0.125, "amount": 0.6153846153846154, "status": 0.75, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1087}, "4": {"facts_total": 63, "facts_retained": 27, "retention_rate": 0.42857142857142855, "retention_by_kind": {"person": 0.5, "date": 0.125, "amount": 0.6153846153846154, "status": 0.75, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1087}}}, "reference_key_id": "keys/synthetic/syn_03027_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "1f88f0e78c63504e418dea843641b99a98555e15d47b02cd5cb2f57da3003dea", "output_hash": "c45f000afe6d865c8167954eebdff761141225817d4f5b26c962cd50cd75d391"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "77c228a0255d05bd3e716f7a", "attempt_id": "8dca5e3ee55f", "run_group_id": "E4_core:A4|repairable|qwen3b:h4", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:31:14.361333Z", "task_family": "synthetic", "document_id": "syn_03027_2000", "document_hash": "10326080979f71b625e4110c7c19767e5a752e267de5034f7e37d98cb6e06249", "question_id": "q11_aggregation", "split": "test", "cluster_id": "syn_03027_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 4, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1955, "query_tokens": 606, "message_tokens": 512, "message_utf8_bytes": 1087, "cache_tensor_bytes": null, "serialized_payload_bytes": 1087, "side_information_bytes": 0, "fetched_bytes": 2562, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 26426.132791908458, "hops": {"1": 6913.723333040252, "2": 6537.521208985709, "3": 6488.541791914031, "4": 6486.346457968466}, "receiver_prefill_ms": 227.13570797350258, "ttft_ms": 12.029916979372501, "decode_ms": 106.78820800967515}, "total_wall_ms": 26772.779416874982, "ttft_ms": 12.029916979372501, "input_usage": {"lookups": 12}, "output_usage": {"output_tokens": 9}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "23,450165", "evidence_artifact": {"ledger": {"facts_total": 63, "facts_retained": 27, "retention_rate": 0.42857142857142855, "retention_by_kind": {"person": 0.5, "date": 0.125, "amount": 0.6153846153846154, "status": 0.75, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1087}, "ledger_by_hop": {"1": {"facts_total": 63, "facts_retained": 27, "retention_rate": 0.42857142857142855, "retention_by_kind": {"person": 0.5, "date": 0.125, "amount": 0.6153846153846154, "status": 0.75, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1087}, "2": {"facts_total": 63, "facts_retained": 27, "retention_rate": 0.42857142857142855, "retention_by_kind": {"person": 0.5, "date": 0.125, "amount": 0.6153846153846154, "status": 0.75, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1087}, "3": {"facts_total": 63, "facts_retained": 27, "retention_rate": 0.42857142857142855, "retention_by_kind": {"person": 0.5, "date": 0.125, "amount": 0.6153846153846154, "status": 0.75, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1087}, "4": {"facts_total": 63, "facts_retained": 27, "retention_rate": 0.42857142857142855, "retention_by_kind": {"person": 0.5, "date": 0.125, "amount": 0.6153846153846154, "status": 0.75, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1087}}}, "reference_key_id": "keys/synthetic/syn_03027_2000.json", "task_scores": {"correct": 0.0, "abstained": 0.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "1f88f0e78c63504e418dea843641b99a98555e15d47b02cd5cb2f57da3003dea", "output_hash": "a929e5c5097069e202c39470de0d2861ac8d456564e72788db30daadb9f19129"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "90f071cd7a7512b245abfaf4", "attempt_id": "7fc026131e9f", "run_group_id": "E4_core:A4|repairable|qwen3b:h4", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:31:14.725338Z", "task_family": "synthetic", "document_id": "syn_03027_2000", "document_hash": "10326080979f71b625e4110c7c19767e5a752e267de5034f7e37d98cb6e06249", "question_id": "q12_unit_conversion", "split": "test", "cluster_id": "syn_03027_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 4, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1955, "query_tokens": 603, "message_tokens": 512, "message_utf8_bytes": 1087, "cache_tensor_bytes": null, "serialized_payload_bytes": 1087, "side_information_bytes": 0, "fetched_bytes": 2562, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 26426.132791908458, "hops": {"1": 6913.723333040252, "2": 6537.521208985709, "3": 6488.541791914031, "4": 6486.346457968466}, "receiver_prefill_ms": 226.80549998767674, "ttft_ms": 11.990292114205658, "decode_ms": 11.960500036366284}, "total_wall_ms": 26677.562749944627, "ttft_ms": 11.990292114205658, "input_usage": {"lookups": 12}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 63, "facts_retained": 27, "retention_rate": 0.42857142857142855, "retention_by_kind": {"person": 0.5, "date": 0.125, "amount": 0.6153846153846154, "status": 0.75, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1087}, "ledger_by_hop": {"1": {"facts_total": 63, "facts_retained": 27, "retention_rate": 0.42857142857142855, "retention_by_kind": {"person": 0.5, "date": 0.125, "amount": 0.6153846153846154, "status": 0.75, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1087}, "2": {"facts_total": 63, "facts_retained": 27, "retention_rate": 0.42857142857142855, "retention_by_kind": {"person": 0.5, "date": 0.125, "amount": 0.6153846153846154, "status": 0.75, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1087}, "3": {"facts_total": 63, "facts_retained": 27, "retention_rate": 0.42857142857142855, "retention_by_kind": {"person": 0.5, "date": 0.125, "amount": 0.6153846153846154, "status": 0.75, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1087}, "4": {"facts_total": 63, "facts_retained": 27, "retention_rate": 0.42857142857142855, "retention_by_kind": {"person": 0.5, "date": 0.125, "amount": 0.6153846153846154, "status": 0.75, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1087}}}, "reference_key_id": "keys/synthetic/syn_03027_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "1f88f0e78c63504e418dea843641b99a98555e15d47b02cd5cb2f57da3003dea", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "be95d7f7e3a3c4a6c32f37c2", "attempt_id": "a19119267cf5", "run_group_id": "E4_core:A4|repairable|qwen3b:h4", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:31:14.993732Z", "task_family": "synthetic", "document_id": "syn_03027_2000", "document_hash": "10326080979f71b625e4110c7c19767e5a752e267de5034f7e37d98cb6e06249", "question_id": "q13_unanswerable", "split": "test", "cluster_id": "syn_03027_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 4, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1955, "query_tokens": 606, "message_tokens": 512, "message_utf8_bytes": 1087, "cache_tensor_bytes": null, "serialized_payload_bytes": 1087, "side_information_bytes": 0, "fetched_bytes": 2562, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 26426.132791908458, "hops": {"1": 6913.723333040252, "2": 6537.521208985709, "3": 6488.541791914031, "4": 6486.346457968466}, "receiver_prefill_ms": 227.09733399096876, "ttft_ms": 12.04025000333786, "decode_ms": 12.06674997229129}, "total_wall_ms": 26678.004707908258, "ttft_ms": 12.04025000333786, "input_usage": {"lookups": 12}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 63, "facts_retained": 27, "retention_rate": 0.42857142857142855, "retention_by_kind": {"person": 0.5, "date": 0.125, "amount": 0.6153846153846154, "status": 0.75, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1087}, "ledger_by_hop": {"1": {"facts_total": 63, "facts_retained": 27, "retention_rate": 0.42857142857142855, "retention_by_kind": {"person": 0.5, "date": 0.125, "amount": 0.6153846153846154, "status": 0.75, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1087}, "2": {"facts_total": 63, "facts_retained": 27, "retention_rate": 0.42857142857142855, "retention_by_kind": {"person": 0.5, "date": 0.125, "amount": 0.6153846153846154, "status": 0.75, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1087}, "3": {"facts_total": 63, "facts_retained": 27, "retention_rate": 0.42857142857142855, "retention_by_kind": {"person": 0.5, "date": 0.125, "amount": 0.6153846153846154, "status": 0.75, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1087}, "4": {"facts_total": 63, "facts_retained": 27, "retention_rate": 0.42857142857142855, "retention_by_kind": {"person": 0.5, "date": 0.125, "amount": 0.6153846153846154, "status": 0.75, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1087}}}, "reference_key_id": "keys/synthetic/syn_03027_2000.json", "task_scores": {"correct": 1.0, "abstained": 1.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "1f88f0e78c63504e418dea843641b99a98555e15d47b02cd5cb2f57da3003dea", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "f8c86339a3eb4ef398fe6e60", "attempt_id": "ed9390d91aec", "run_group_id": "E4_core:A4|repairable|qwen3b:h4", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:31:15.262618Z", "task_family": "synthetic", "document_id": "syn_03027_2000", "document_hash": "10326080979f71b625e4110c7c19767e5a752e267de5034f7e37d98cb6e06249", "question_id": "q14_unanswerable", "split": "test", "cluster_id": "syn_03027_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 4, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1955, "query_tokens": 608, "message_tokens": 512, "message_utf8_bytes": 1087, "cache_tensor_bytes": null, "serialized_payload_bytes": 1087, "side_information_bytes": 0, "fetched_bytes": 2562, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 26426.132791908458, "hops": {"1": 6913.723333040252, "2": 6537.521208985709, "3": 6488.541791914031, "4": 6486.346457968466}, "receiver_prefill_ms": 226.85412492137402, "ttft_ms": 12.025041971355677, "decode_ms": 11.947166989557445}, "total_wall_ms": 26677.633916959167, "ttft_ms": 12.025041971355677, "input_usage": {"lookups": 12}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 63, "facts_retained": 27, "retention_rate": 0.42857142857142855, "retention_by_kind": {"person": 0.5, "date": 0.125, "amount": 0.6153846153846154, "status": 0.75, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1087}, "ledger_by_hop": {"1": {"facts_total": 63, "facts_retained": 27, "retention_rate": 0.42857142857142855, "retention_by_kind": {"person": 0.5, "date": 0.125, "amount": 0.6153846153846154, "status": 0.75, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1087}, "2": {"facts_total": 63, "facts_retained": 27, "retention_rate": 0.42857142857142855, "retention_by_kind": {"person": 0.5, "date": 0.125, "amount": 0.6153846153846154, "status": 0.75, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1087}, "3": {"facts_total": 63, "facts_retained": 27, "retention_rate": 0.42857142857142855, "retention_by_kind": {"person": 0.5, "date": 0.125, "amount": 0.6153846153846154, "status": 0.75, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1087}, "4": {"facts_total": 63, "facts_retained": 27, "retention_rate": 0.42857142857142855, "retention_by_kind": {"person": 0.5, "date": 0.125, "amount": 0.6153846153846154, "status": 0.75, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1087}}}, "reference_key_id": "keys/synthetic/syn_03027_2000.json", "task_scores": {"correct": 1.0, "abstained": 1.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "1f88f0e78c63504e418dea843641b99a98555e15d47b02cd5cb2f57da3003dea", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "1aeb5f54937e2da6c8994065", "attempt_id": "84421481d492", "run_group_id": "E4_core:A4|repairable|qwen3b:h8", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:31:41.492059Z", "task_family": "synthetic", "document_id": "syn_03027_2000", "document_hash": "10326080979f71b625e4110c7c19767e5a752e267de5034f7e37d98cb6e06249", "question_id": "q00_single_fact", "split": "test", "cluster_id": "syn_03027_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 8, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1955, "query_tokens": 594, "message_tokens": 512, "message_utf8_bytes": 1087, "cache_tensor_bytes": null, "serialized_payload_bytes": 1087, "side_information_bytes": 0, "fetched_bytes": 5978, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 52381.81649975013, "hops": {"1": 6913.723333040252, "2": 6537.521208985709, "3": 6488.541791914031, "4": 6486.346457968466, "5": 6492.82487493474, "6": 6482.172666001134, "7": 6484.1609169961885, "8": 6496.52524990961}, "receiver_prefill_ms": 226.67104098945856, "ttft_ms": 12.146624969318509, "decode_ms": 11.900875018909574}, "total_wall_ms": 52633.19920876529, "ttft_ms": 12.146624969318509, "input_usage": {"lookups": 28}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 63, "facts_retained": 27, "retention_rate": 0.42857142857142855, "retention_by_kind": {"person": 0.5, "date": 0.125, "amount": 0.6153846153846154, "status": 0.75, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1087}, "ledger_by_hop": {"1": {"facts_total": 63, "facts_retained": 27, "retention_rate": 0.42857142857142855, "retention_by_kind": {"person": 0.5, "date": 0.125, "amount": 0.6153846153846154, "status": 0.75, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1087}, "2": {"facts_total": 63, "facts_retained": 27, "retention_rate": 0.42857142857142855, "retention_by_kind": {"person": 0.5, "date": 0.125, "amount": 0.6153846153846154, "status": 0.75, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1087}, "3": {"facts_total": 63, "facts_retained": 27, "retention_rate": 0.42857142857142855, "retention_by_kind": {"person": 0.5, "date": 0.125, "amount": 0.6153846153846154, "status": 0.75, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1087}, "4": {"facts_total": 63, "facts_retained": 27, "retention_rate": 0.42857142857142855, "retention_by_kind": {"person": 0.5, "date": 0.125, "amount": 0.6153846153846154, "status": 0.75, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1087}, "5": {"facts_total": 63, "facts_retained": 27, "retention_rate": 0.42857142857142855, "retention_by_kind": {"person": 0.5, "date": 0.125, "amount": 0.6153846153846154, "status": 0.75, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1087}, "6": {"facts_total": 63, "facts_retained": 27, "retention_rate": 0.42857142857142855, "retention_by_kind": {"person": 0.5, "date": 0.125, "amount": 0.6153846153846154, "status": 0.75, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1087}, "7": {"facts_total": 63, "facts_retained": 27, "retention_rate": 0.42857142857142855, "retention_by_kind": {"person": 0.5, "date": 0.125, "amount": 0.6153846153846154, "status": 0.75, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1087}, "8": {"facts_total": 63, "facts_retained": 27, "retention_rate": 0.42857142857142855, "retention_by_kind": {"person": 0.5, "date": 0.125, "amount": 0.6153846153846154, "status": 0.75, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1087}}}, "reference_key_id": "keys/synthetic/syn_03027_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "1f88f0e78c63504e418dea843641b99a98555e15d47b02cd5cb2f57da3003dea", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "7f9bf270ea8291e919e4a1a5", "attempt_id": "ad561d8fbe8c", "run_group_id": "E4_core:A4|repairable|qwen3b:h8", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:31:41.760404Z", "task_family": "synthetic", "document_id": "syn_03027_2000", "document_hash": "10326080979f71b625e4110c7c19767e5a752e267de5034f7e37d98cb6e06249", "question_id": "q01_single_fact", "split": "test", "cluster_id": "syn_03027_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 8, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1955, "query_tokens": 601, "message_tokens": 512, "message_utf8_bytes": 1087, "cache_tensor_bytes": null, "serialized_payload_bytes": 1087, "side_information_bytes": 0, "fetched_bytes": 5978, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 52381.81649975013, "hops": {"1": 6913.723333040252, "2": 6537.521208985709, "3": 6488.541791914031, "4": 6486.346457968466, "5": 6492.82487493474, "6": 6482.172666001134, "7": 6484.1609169961885, "8": 6496.52524990961}, "receiver_prefill_ms": 226.53450001962483, "ttft_ms": 12.05608295276761, "decode_ms": 142.4653750145808}, "total_wall_ms": 52763.53699970059, "ttft_ms": 12.05608295276761, "input_usage": {"lookups": 28}, "output_usage": {"output_tokens": 12}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "Lot 247-S, Baie-Comeau", "evidence_artifact": {"ledger": {"facts_total": 63, "facts_retained": 27, "retention_rate": 0.42857142857142855, "retention_by_kind": {"person": 0.5, "date": 0.125, "amount": 0.6153846153846154, "status": 0.75, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1087}, "ledger_by_hop": {"1": {"facts_total": 63, "facts_retained": 27, "retention_rate": 0.42857142857142855, "retention_by_kind": {"person": 0.5, "date": 0.125, "amount": 0.6153846153846154, "status": 0.75, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1087}, "2": {"facts_total": 63, "facts_retained": 27, "retention_rate": 0.42857142857142855, "retention_by_kind": {"person": 0.5, "date": 0.125, "amount": 0.6153846153846154, "status": 0.75, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1087}, "3": {"facts_total": 63, "facts_retained": 27, "retention_rate": 0.42857142857142855, "retention_by_kind": {"person": 0.5, "date": 0.125, "amount": 0.6153846153846154, "status": 0.75, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1087}, "4": {"facts_total": 63, "facts_retained": 27, "retention_rate": 0.42857142857142855, "retention_by_kind": {"person": 0.5, "date": 0.125, "amount": 0.6153846153846154, "status": 0.75, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1087}, "5": {"facts_total": 63, "facts_retained": 27, "retention_rate": 0.42857142857142855, "retention_by_kind": {"person": 0.5, "date": 0.125, "amount": 0.6153846153846154, "status": 0.75, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1087}, "6": {"facts_total": 63, "facts_retained": 27, "retention_rate": 0.42857142857142855, "retention_by_kind": {"person": 0.5, "date": 0.125, "amount": 0.6153846153846154, "status": 0.75, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1087}, "7": {"facts_total": 63, "facts_retained": 27, "retention_rate": 0.42857142857142855, "retention_by_kind": {"person": 0.5, "date": 0.125, "amount": 0.6153846153846154, "status": 0.75, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1087}, "8": {"facts_total": 63, "facts_retained": 27, "retention_rate": 0.42857142857142855, "retention_by_kind": {"person": 0.5, "date": 0.125, "amount": 0.6153846153846154, "status": 0.75, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1087}}}, "reference_key_id": "keys/synthetic/syn_03027_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 0.8571428571428571}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "1f88f0e78c63504e418dea843641b99a98555e15d47b02cd5cb2f57da3003dea", "output_hash": "77b7da1d43e7d8815e9e9159ebbd735045bc7f668260ba30b93674f70622b8ec"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "02f5591b35570cef9a088d74", "attempt_id": "515f7bc01037", "run_group_id": "E4_core:A4|repairable|qwen3b:h8", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:31:42.158505Z", "task_family": "synthetic", "document_id": "syn_03027_2000", "document_hash": "10326080979f71b625e4110c7c19767e5a752e267de5034f7e37d98cb6e06249", "question_id": "q02_exact_copy", "split": "test", "cluster_id": "syn_03027_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 8, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1955, "query_tokens": 600, "message_tokens": 512, "message_utf8_bytes": 1087, "cache_tensor_bytes": null, "serialized_payload_bytes": 1087, "side_information_bytes": 0, "fetched_bytes": 5978, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 52381.81649975013, "hops": {"1": 6913.723333040252, "2": 6537.521208985709, "3": 6488.541791914031, "4": 6486.346457968466, "5": 6492.82487493474, "6": 6482.172666001134, "7": 6484.1609169961885, "8": 6496.52524990961}, "receiver_prefill_ms": 226.54391604010016, "ttft_ms": 12.045416980981827, "decode_ms": 83.09945906512439}, "total_wall_ms": 52704.16529173963, "ttft_ms": 12.045416980981827, "input_usage": {"lookups": 28}, "output_usage": {"output_tokens": 7}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "15.0054", "evidence_artifact": {"ledger": {"facts_total": 63, "facts_retained": 27, "retention_rate": 0.42857142857142855, "retention_by_kind": {"person": 0.5, "date": 0.125, "amount": 0.6153846153846154, "status": 0.75, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1087}, "ledger_by_hop": {"1": {"facts_total": 63, "facts_retained": 27, "retention_rate": 0.42857142857142855, "retention_by_kind": {"person": 0.5, "date": 0.125, "amount": 0.6153846153846154, "status": 0.75, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1087}, "2": {"facts_total": 63, "facts_retained": 27, "retention_rate": 0.42857142857142855, "retention_by_kind": {"person": 0.5, "date": 0.125, "amount": 0.6153846153846154, "status": 0.75, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1087}, "3": {"facts_total": 63, "facts_retained": 27, "retention_rate": 0.42857142857142855, "retention_by_kind": {"person": 0.5, "date": 0.125, "amount": 0.6153846153846154, "status": 0.75, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1087}, "4": {"facts_total": 63, "facts_retained": 27, "retention_rate": 0.42857142857142855, "retention_by_kind": {"person": 0.5, "date": 0.125, "amount": 0.6153846153846154, "status": 0.75, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1087}, "5": {"facts_total": 63, "facts_retained": 27, "retention_rate": 0.42857142857142855, "retention_by_kind": {"person": 0.5, "date": 0.125, "amount": 0.6153846153846154, "status": 0.75, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1087}, "6": {"facts_total": 63, "facts_retained": 27, "retention_rate": 0.42857142857142855, "retention_by_kind": {"person": 0.5, "date": 0.125, "amount": 0.6153846153846154, "status": 0.75, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1087}, "7": {"facts_total": 63, "facts_retained": 27, "retention_rate": 0.42857142857142855, "retention_by_kind": {"person": 0.5, "date": 0.125, "amount": 0.6153846153846154, "status": 0.75, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1087}, "8": {"facts_total": 63, "facts_retained": 27, "retention_rate": 0.42857142857142855, "retention_by_kind": {"person": 0.5, "date": 0.125, "amount": 0.6153846153846154, "status": 0.75, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1087}}}, "reference_key_id": "keys/synthetic/syn_03027_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "1f88f0e78c63504e418dea843641b99a98555e15d47b02cd5cb2f57da3003dea", "output_hash": "d13adba73068c847c850237059e0202fc5b91ccacf2e792e9ce5e3b90556501a"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "3e6868a843a0607f2e10955b", "attempt_id": "4feec602560f", "run_group_id": "E4_core:A4|repairable|qwen3b:h8", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:31:42.497689Z", "task_family": "synthetic", "document_id": "syn_03027_2000", "document_hash": "10326080979f71b625e4110c7c19767e5a752e267de5034f7e37d98cb6e06249", "question_id": "q03_exact_copy", "split": "test", "cluster_id": "syn_03027_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 8, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1955, "query_tokens": 595, "message_tokens": 512, "message_utf8_bytes": 1087, "cache_tensor_bytes": null, "serialized_payload_bytes": 1087, "side_information_bytes": 0, "fetched_bytes": 5978, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 52381.81649975013, "hops": {"1": 6913.723333040252, "2": 6537.521208985709, "3": 6488.541791914031, "4": 6486.346457968466, "5": 6492.82487493474, "6": 6482.172666001134, "7": 6484.1609169961885, "8": 6496.52524990961}, "receiver_prefill_ms": 226.34745901450515, "ttft_ms": 12.016375083476305, "decode_ms": 106.95979208685458}, "total_wall_ms": 52727.82749973703, "ttft_ms": 12.016375083476305, "input_usage": {"lookups": 28}, "output_usage": {"output_tokens": 9}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "Priyamvada Sørlie", "evidence_artifact": {"ledger": {"facts_total": 63, "facts_retained": 27, "retention_rate": 0.42857142857142855, "retention_by_kind": {"person": 0.5, "date": 0.125, "amount": 0.6153846153846154, "status": 0.75, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1087}, "ledger_by_hop": {"1": {"facts_total": 63, "facts_retained": 27, "retention_rate": 0.42857142857142855, "retention_by_kind": {"person": 0.5, "date": 0.125, "amount": 0.6153846153846154, "status": 0.75, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1087}, "2": {"facts_total": 63, "facts_retained": 27, "retention_rate": 0.42857142857142855, "retention_by_kind": {"person": 0.5, "date": 0.125, "amount": 0.6153846153846154, "status": 0.75, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1087}, "3": {"facts_total": 63, "facts_retained": 27, "retention_rate": 0.42857142857142855, "retention_by_kind": {"person": 0.5, "date": 0.125, "amount": 0.6153846153846154, "status": 0.75, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1087}, "4": {"facts_total": 63, "facts_retained": 27, "retention_rate": 0.42857142857142855, "retention_by_kind": {"person": 0.5, "date": 0.125, "amount": 0.6153846153846154, "status": 0.75, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1087}, "5": {"facts_total": 63, "facts_retained": 27, "retention_rate": 0.42857142857142855, "retention_by_kind": {"person": 0.5, "date": 0.125, "amount": 0.6153846153846154, "status": 0.75, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1087}, "6": {"facts_total": 63, "facts_retained": 27, "retention_rate": 0.42857142857142855, "retention_by_kind": {"person": 0.5, "date": 0.125, "amount": 0.6153846153846154, "status": 0.75, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1087}, "7": {"facts_total": 63, "facts_retained": 27, "retention_rate": 0.42857142857142855, "retention_by_kind": {"person": 0.5, "date": 0.125, "amount": 0.6153846153846154, "status": 0.75, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1087}, "8": {"facts_total": 63, "facts_retained": 27, "retention_rate": 0.42857142857142855, "retention_by_kind": {"person": 0.5, "date": 0.125, "amount": 0.6153846153846154, "status": 0.75, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1087}}}, "reference_key_id": "keys/synthetic/syn_03027_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "1f88f0e78c63504e418dea843641b99a98555e15d47b02cd5cb2f57da3003dea", "output_hash": "c5f554196ed23dd71ad995ec89f7142099d524dbb6534cc10d94a0c89d13e792"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "c690b3481e12ef11a37e8bd8", "attempt_id": "2b32d4dc5ecb", "run_group_id": "E4_core:A4|repairable|qwen3b:h8", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:31:42.860218Z", "task_family": "synthetic", "document_id": "syn_03027_2000", "document_hash": "10326080979f71b625e4110c7c19767e5a752e267de5034f7e37d98cb6e06249", "question_id": "q04_multi_hop", "split": "test", "cluster_id": "syn_03027_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 8, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1955, "query_tokens": 606, "message_tokens": 512, "message_utf8_bytes": 1087, "cache_tensor_bytes": null, "serialized_payload_bytes": 1087, "side_information_bytes": 0, "fetched_bytes": 5978, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 52381.81649975013, "hops": {"1": 6913.723333040252, "2": 6537.521208985709, "3": 6488.541791914031, "4": 6486.346457968466, "5": 6492.82487493474, "6": 6482.172666001134, "7": 6484.1609169961885, "8": 6496.52524990961}, "receiver_prefill_ms": 226.99391597416252, "ttft_ms": 12.069249991327524, "decode_ms": 11.948042083531618}, "total_wall_ms": 52633.48758383654, "ttft_ms": 12.069249991327524, "input_usage": {"lookups": 28}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 63, "facts_retained": 27, "retention_rate": 0.42857142857142855, "retention_by_kind": {"person": 0.5, "date": 0.125, "amount": 0.6153846153846154, "status": 0.75, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1087}, "ledger_by_hop": {"1": {"facts_total": 63, "facts_retained": 27, "retention_rate": 0.42857142857142855, "retention_by_kind": {"person": 0.5, "date": 0.125, "amount": 0.6153846153846154, "status": 0.75, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1087}, "2": {"facts_total": 63, "facts_retained": 27, "retention_rate": 0.42857142857142855, "retention_by_kind": {"person": 0.5, "date": 0.125, "amount": 0.6153846153846154, "status": 0.75, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1087}, "3": {"facts_total": 63, "facts_retained": 27, "retention_rate": 0.42857142857142855, "retention_by_kind": {"person": 0.5, "date": 0.125, "amount": 0.6153846153846154, "status": 0.75, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1087}, "4": {"facts_total": 63, "facts_retained": 27, "retention_rate": 0.42857142857142855, "retention_by_kind": {"person": 0.5, "date": 0.125, "amount": 0.6153846153846154, "status": 0.75, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1087}, "5": {"facts_total": 63, "facts_retained": 27, "retention_rate": 0.42857142857142855, "retention_by_kind": {"person": 0.5, "date": 0.125, "amount": 0.6153846153846154, "status": 0.75, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1087}, "6": {"facts_total": 63, "facts_retained": 27, "retention_rate": 0.42857142857142855, "retention_by_kind": {"person": 0.5, "date": 0.125, "amount": 0.6153846153846154, "status": 0.75, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1087}, "7": {"facts_total": 63, "facts_retained": 27, "retention_rate": 0.42857142857142855, "retention_by_kind": {"person": 0.5, "date": 0.125, "amount": 0.6153846153846154, "status": 0.75, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1087}, "8": {"facts_total": 63, "facts_retained": 27, "retention_rate": 0.42857142857142855, "retention_by_kind": {"person": 0.5, "date": 0.125, "amount": 0.6153846153846154, "status": 0.75, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1087}}}, "reference_key_id": "keys/synthetic/syn_03027_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "1f88f0e78c63504e418dea843641b99a98555e15d47b02cd5cb2f57da3003dea", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "d10c2a97d11f73b5ef257795", "attempt_id": "382b8e78bdc2", "run_group_id": "E4_core:A4|repairable|qwen3b:h8", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:31:43.128682Z", "task_family": "synthetic", "document_id": "syn_03027_2000", "document_hash": "10326080979f71b625e4110c7c19767e5a752e267de5034f7e37d98cb6e06249", "question_id": "q05_multi_hop", "split": "test", "cluster_id": "syn_03027_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 8, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1955, "query_tokens": 603, "message_tokens": 512, "message_utf8_bytes": 1087, "cache_tensor_bytes": null, "serialized_payload_bytes": 1087, "side_information_bytes": 0, "fetched_bytes": 5978, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 52381.81649975013, "hops": {"1": 6913.723333040252, "2": 6537.521208985709, "3": 6488.541791914031, "4": 6486.346457968466, "5": 6492.82487493474, "6": 6482.172666001134, "7": 6484.1609169961885, "8": 6496.52524990961}, "receiver_prefill_ms": 226.90041607711464, "ttft_ms": 12.111415970139205, "decode_ms": 11.93616702221334}, "total_wall_ms": 52633.430541842245, "ttft_ms": 12.111415970139205, "input_usage": {"lookups": 28}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 63, "facts_retained": 27, "retention_rate": 0.42857142857142855, "retention_by_kind": {"person": 0.5, "date": 0.125, "amount": 0.6153846153846154, "status": 0.75, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1087}, "ledger_by_hop": {"1": {"facts_total": 63, "facts_retained": 27, "retention_rate": 0.42857142857142855, "retention_by_kind": {"person": 0.5, "date": 0.125, "amount": 0.6153846153846154, "status": 0.75, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1087}, "2": {"facts_total": 63, "facts_retained": 27, "retention_rate": 0.42857142857142855, "retention_by_kind": {"person": 0.5, "date": 0.125, "amount": 0.6153846153846154, "status": 0.75, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1087}, "3": {"facts_total": 63, "facts_retained": 27, "retention_rate": 0.42857142857142855, "retention_by_kind": {"person": 0.5, "date": 0.125, "amount": 0.6153846153846154, "status": 0.75, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1087}, "4": {"facts_total": 63, "facts_retained": 27, "retention_rate": 0.42857142857142855, "retention_by_kind": {"person": 0.5, "date": 0.125, "amount": 0.6153846153846154, "status": 0.75, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1087}, "5": {"facts_total": 63, "facts_retained": 27, "retention_rate": 0.42857142857142855, "retention_by_kind": {"person": 0.5, "date": 0.125, "amount": 0.6153846153846154, "status": 0.75, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1087}, "6": {"facts_total": 63, "facts_retained": 27, "retention_rate": 0.42857142857142855, "retention_by_kind": {"person": 0.5, "date": 0.125, "amount": 0.6153846153846154, "status": 0.75, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1087}, "7": {"facts_total": 63, "facts_retained": 27, "retention_rate": 0.42857142857142855, "retention_by_kind": {"person": 0.5, "date": 0.125, "amount": 0.6153846153846154, "status": 0.75, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1087}, "8": {"facts_total": 63, "facts_retained": 27, "retention_rate": 0.42857142857142855, "retention_by_kind": {"person": 0.5, "date": 0.125, "amount": 0.6153846153846154, "status": 0.75, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1087}}}, "reference_key_id": "keys/synthetic/syn_03027_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "1f88f0e78c63504e418dea843641b99a98555e15d47b02cd5cb2f57da3003dea", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "e404a7952c0b16c741688be7", "attempt_id": "c8df5a819e29", "run_group_id": "E4_core:A4|repairable|qwen3b:h8", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:31:43.397029Z", "task_family": "synthetic", "document_id": "syn_03027_2000", "document_hash": "10326080979f71b625e4110c7c19767e5a752e267de5034f7e37d98cb6e06249", "question_id": "q06_negation", "split": "test", "cluster_id": "syn_03027_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 8, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1955, "query_tokens": 595, "message_tokens": 512, "message_utf8_bytes": 1087, "cache_tensor_bytes": null, "serialized_payload_bytes": 1087, "side_information_bytes": 0, "fetched_bytes": 5978, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 52381.81649975013, "hops": {"1": 6913.723333040252, "2": 6537.521208985709, "3": 6488.541791914031, "4": 6486.346457968466, "5": 6492.82487493474, "6": 6482.172666001134, "7": 6484.1609169961885, "8": 6496.52524990961}, "receiver_prefill_ms": 226.5007080277428, "ttft_ms": 12.092125019989908, "decode_ms": 11.913749971427023}, "total_wall_ms": 52632.952208747156, "ttft_ms": 12.092125019989908, "input_usage": {"lookups": 28}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 63, "facts_retained": 27, "retention_rate": 0.42857142857142855, "retention_by_kind": {"person": 0.5, "date": 0.125, "amount": 0.6153846153846154, "status": 0.75, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1087}, "ledger_by_hop": {"1": {"facts_total": 63, "facts_retained": 27, "retention_rate": 0.42857142857142855, "retention_by_kind": {"person": 0.5, "date": 0.125, "amount": 0.6153846153846154, "status": 0.75, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1087}, "2": {"facts_total": 63, "facts_retained": 27, "retention_rate": 0.42857142857142855, "retention_by_kind": {"person": 0.5, "date": 0.125, "amount": 0.6153846153846154, "status": 0.75, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1087}, "3": {"facts_total": 63, "facts_retained": 27, "retention_rate": 0.42857142857142855, "retention_by_kind": {"person": 0.5, "date": 0.125, "amount": 0.6153846153846154, "status": 0.75, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1087}, "4": {"facts_total": 63, "facts_retained": 27, "retention_rate": 0.42857142857142855, "retention_by_kind": {"person": 0.5, "date": 0.125, "amount": 0.6153846153846154, "status": 0.75, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1087}, "5": {"facts_total": 63, "facts_retained": 27, "retention_rate": 0.42857142857142855, "retention_by_kind": {"person": 0.5, "date": 0.125, "amount": 0.6153846153846154, "status": 0.75, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1087}, "6": {"facts_total": 63, "facts_retained": 27, "retention_rate": 0.42857142857142855, "retention_by_kind": {"person": 0.5, "date": 0.125, "amount": 0.6153846153846154, "status": 0.75, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1087}, "7": {"facts_total": 63, "facts_retained": 27, "retention_rate": 0.42857142857142855, "retention_by_kind": {"person": 0.5, "date": 0.125, "amount": 0.6153846153846154, "status": 0.75, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1087}, "8": {"facts_total": 63, "facts_retained": 27, "retention_rate": 0.42857142857142855, "retention_by_kind": {"person": 0.5, "date": 0.125, "amount": 0.6153846153846154, "status": 0.75, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1087}}}, "reference_key_id": "keys/synthetic/syn_03027_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "1f88f0e78c63504e418dea843641b99a98555e15d47b02cd5cb2f57da3003dea", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "9472f66c3be2613e05c931a8", "attempt_id": "020415f015e0", "run_group_id": "E4_core:A4|repairable|qwen3b:h8", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:31:43.664623Z", "task_family": "synthetic", "document_id": "syn_03027_2000", "document_hash": "10326080979f71b625e4110c7c19767e5a752e267de5034f7e37d98cb6e06249", "question_id": "q07_negation", "split": "test", "cluster_id": "syn_03027_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 8, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1955, "query_tokens": 596, "message_tokens": 512, "message_utf8_bytes": 1087, "cache_tensor_bytes": null, "serialized_payload_bytes": 1087, "side_information_bytes": 0, "fetched_bytes": 5978, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 52381.81649975013, "hops": {"1": 6913.723333040252, "2": 6537.521208985709, "3": 6488.541791914031, "4": 6486.346457968466, "5": 6492.82487493474, "6": 6482.172666001134, "7": 6484.1609169961885, "8": 6496.52524990961}, "receiver_prefill_ms": 226.55254194978625, "ttft_ms": 12.070374912582338, "decode_ms": 11.949707986786962}, "total_wall_ms": 52633.06937471498, "ttft_ms": 12.070374912582338, "input_usage": {"lookups": 28}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "no", "evidence_artifact": {"ledger": {"facts_total": 63, "facts_retained": 27, "retention_rate": 0.42857142857142855, "retention_by_kind": {"person": 0.5, "date": 0.125, "amount": 0.6153846153846154, "status": 0.75, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1087}, "ledger_by_hop": {"1": {"facts_total": 63, "facts_retained": 27, "retention_rate": 0.42857142857142855, "retention_by_kind": {"person": 0.5, "date": 0.125, "amount": 0.6153846153846154, "status": 0.75, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1087}, "2": {"facts_total": 63, "facts_retained": 27, "retention_rate": 0.42857142857142855, "retention_by_kind": {"person": 0.5, "date": 0.125, "amount": 0.6153846153846154, "status": 0.75, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1087}, "3": {"facts_total": 63, "facts_retained": 27, "retention_rate": 0.42857142857142855, "retention_by_kind": {"person": 0.5, "date": 0.125, "amount": 0.6153846153846154, "status": 0.75, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1087}, "4": {"facts_total": 63, "facts_retained": 27, "retention_rate": 0.42857142857142855, "retention_by_kind": {"person": 0.5, "date": 0.125, "amount": 0.6153846153846154, "status": 0.75, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1087}, "5": {"facts_total": 63, "facts_retained": 27, "retention_rate": 0.42857142857142855, "retention_by_kind": {"person": 0.5, "date": 0.125, "amount": 0.6153846153846154, "status": 0.75, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1087}, "6": {"facts_total": 63, "facts_retained": 27, "retention_rate": 0.42857142857142855, "retention_by_kind": {"person": 0.5, "date": 0.125, "amount": 0.6153846153846154, "status": 0.75, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1087}, "7": {"facts_total": 63, "facts_retained": 27, "retention_rate": 0.42857142857142855, "retention_by_kind": {"person": 0.5, "date": 0.125, "amount": 0.6153846153846154, "status": 0.75, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1087}, "8": {"facts_total": 63, "facts_retained": 27, "retention_rate": 0.42857142857142855, "retention_by_kind": {"person": 0.5, "date": 0.125, "amount": 0.6153846153846154, "status": 0.75, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1087}}}, "reference_key_id": "keys/synthetic/syn_03027_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "1f88f0e78c63504e418dea843641b99a98555e15d47b02cd5cb2f57da3003dea", "output_hash": "9390298f3fb0c5b160498935d79cb139aef28e1c47358b4bbba61862b9c26e59"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "bbabcc96cd5a72fe8435034a", "attempt_id": "4c3758da9756", "run_group_id": "E4_core:A4|repairable|qwen3b:h8", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:31:43.932245Z", "task_family": "synthetic", "document_id": "syn_03027_2000", "document_hash": "10326080979f71b625e4110c7c19767e5a752e267de5034f7e37d98cb6e06249", "question_id": "q08_revision", "split": "test", "cluster_id": "syn_03027_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 8, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1955, "query_tokens": 610, "message_tokens": 512, "message_utf8_bytes": 1087, "cache_tensor_bytes": null, "serialized_payload_bytes": 1087, "side_information_bytes": 0, "fetched_bytes": 5978, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 52381.81649975013, "hops": {"1": 6913.723333040252, "2": 6537.521208985709, "3": 6488.541791914031, "4": 6486.346457968466, "5": 6492.82487493474, "6": 6482.172666001134, "7": 6484.1609169961885, "8": 6496.52524990961}, "receiver_prefill_ms": 228.38616697117686, "ttft_ms": 12.151625007390976, "decode_ms": 83.05100002326071}, "total_wall_ms": 52706.074374727905, "ttft_ms": 12.151625007390976, "input_usage": {"lookups": 28}, "output_usage": {"output_tokens": 7}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "8010205", "evidence_artifact": {"ledger": {"facts_total": 63, "facts_retained": 27, "retention_rate": 0.42857142857142855, "retention_by_kind": {"person": 0.5, "date": 0.125, "amount": 0.6153846153846154, "status": 0.75, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1087}, "ledger_by_hop": {"1": {"facts_total": 63, "facts_retained": 27, "retention_rate": 0.42857142857142855, "retention_by_kind": {"person": 0.5, "date": 0.125, "amount": 0.6153846153846154, "status": 0.75, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1087}, "2": {"facts_total": 63, "facts_retained": 27, "retention_rate": 0.42857142857142855, "retention_by_kind": {"person": 0.5, "date": 0.125, "amount": 0.6153846153846154, "status": 0.75, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1087}, "3": {"facts_total": 63, "facts_retained": 27, "retention_rate": 0.42857142857142855, "retention_by_kind": {"person": 0.5, "date": 0.125, "amount": 0.6153846153846154, "status": 0.75, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1087}, "4": {"facts_total": 63, "facts_retained": 27, "retention_rate": 0.42857142857142855, "retention_by_kind": {"person": 0.5, "date": 0.125, "amount": 0.6153846153846154, "status": 0.75, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1087}, "5": {"facts_total": 63, "facts_retained": 27, "retention_rate": 0.42857142857142855, "retention_by_kind": {"person": 0.5, "date": 0.125, "amount": 0.6153846153846154, "status": 0.75, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1087}, "6": {"facts_total": 63, "facts_retained": 27, "retention_rate": 0.42857142857142855, "retention_by_kind": {"person": 0.5, "date": 0.125, "amount": 0.6153846153846154, "status": 0.75, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1087}, "7": {"facts_total": 63, "facts_retained": 27, "retention_rate": 0.42857142857142855, "retention_by_kind": {"person": 0.5, "date": 0.125, "amount": 0.6153846153846154, "status": 0.75, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1087}, "8": {"facts_total": 63, "facts_retained": 27, "retention_rate": 0.42857142857142855, "retention_by_kind": {"person": 0.5, "date": 0.125, "amount": 0.6153846153846154, "status": 0.75, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1087}}}, "reference_key_id": "keys/synthetic/syn_03027_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "1f88f0e78c63504e418dea843641b99a98555e15d47b02cd5cb2f57da3003dea", "output_hash": "6d1ceb36132f38bd77e8f46c23ea21dd9bcff78bc80bd8a9eaf2d7f2f3bacf46"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "034ef50cc7813b185f117602", "attempt_id": "90e5c712bda0", "run_group_id": "E4_core:A4|repairable|qwen3b:h8", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:31:44.273141Z", "task_family": "synthetic", "document_id": "syn_03027_2000", "document_hash": "10326080979f71b625e4110c7c19767e5a752e267de5034f7e37d98cb6e06249", "question_id": "q09_revision", "split": "test", "cluster_id": "syn_03027_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 8, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1955, "query_tokens": 603, "message_tokens": 512, "message_utf8_bytes": 1087, "cache_tensor_bytes": null, "serialized_payload_bytes": 1087, "side_information_bytes": 0, "fetched_bytes": 5978, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 52381.81649975013, "hops": {"1": 6913.723333040252, "2": 6537.521208985709, "3": 6488.541791914031, "4": 6486.346457968466, "5": 6492.82487493474, "6": 6482.172666001134, "7": 6484.1609169961885, "8": 6496.52524990961}, "receiver_prefill_ms": 226.02399997413158, "ttft_ms": 12.103083077818155, "decode_ms": 83.07608298491687}, "total_wall_ms": 52703.70483270381, "ttft_ms": 12.103083077818155, "input_usage": {"lookups": 28}, "output_usage": {"output_tokens": 7}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "6056150", "evidence_artifact": {"ledger": {"facts_total": 63, "facts_retained": 27, "retention_rate": 0.42857142857142855, "retention_by_kind": {"person": 0.5, "date": 0.125, "amount": 0.6153846153846154, "status": 0.75, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1087}, "ledger_by_hop": {"1": {"facts_total": 63, "facts_retained": 27, "retention_rate": 0.42857142857142855, "retention_by_kind": {"person": 0.5, "date": 0.125, "amount": 0.6153846153846154, "status": 0.75, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1087}, "2": {"facts_total": 63, "facts_retained": 27, "retention_rate": 0.42857142857142855, "retention_by_kind": {"person": 0.5, "date": 0.125, "amount": 0.6153846153846154, "status": 0.75, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1087}, "3": {"facts_total": 63, "facts_retained": 27, "retention_rate": 0.42857142857142855, "retention_by_kind": {"person": 0.5, "date": 0.125, "amount": 0.6153846153846154, "status": 0.75, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1087}, "4": {"facts_total": 63, "facts_retained": 27, "retention_rate": 0.42857142857142855, "retention_by_kind": {"person": 0.5, "date": 0.125, "amount": 0.6153846153846154, "status": 0.75, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1087}, "5": {"facts_total": 63, "facts_retained": 27, "retention_rate": 0.42857142857142855, "retention_by_kind": {"person": 0.5, "date": 0.125, "amount": 0.6153846153846154, "status": 0.75, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1087}, "6": {"facts_total": 63, "facts_retained": 27, "retention_rate": 0.42857142857142855, "retention_by_kind": {"person": 0.5, "date": 0.125, "amount": 0.6153846153846154, "status": 0.75, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1087}, "7": {"facts_total": 63, "facts_retained": 27, "retention_rate": 0.42857142857142855, "retention_by_kind": {"person": 0.5, "date": 0.125, "amount": 0.6153846153846154, "status": 0.75, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1087}, "8": {"facts_total": 63, "facts_retained": 27, "retention_rate": 0.42857142857142855, "retention_by_kind": {"person": 0.5, "date": 0.125, "amount": 0.6153846153846154, "status": 0.75, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1087}}}, "reference_key_id": "keys/synthetic/syn_03027_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "1f88f0e78c63504e418dea843641b99a98555e15d47b02cd5cb2f57da3003dea", "output_hash": "f016f8b5fc26708e6c1ff75ff8e98659082c41a168f273819ac48b88feea8d3f"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "0b31d55d64cec5592c667b90", "attempt_id": "92a51ed65cb9", "run_group_id": "E4_core:A4|repairable|qwen3b:h8", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:31:44.612304Z", "task_family": "synthetic", "document_id": "syn_03027_2000", "document_hash": "10326080979f71b625e4110c7c19767e5a752e267de5034f7e37d98cb6e06249", "question_id": "q10_temporal", "split": "test", "cluster_id": "syn_03027_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 8, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1955, "query_tokens": 613, "message_tokens": 512, "message_utf8_bytes": 1087, "cache_tensor_bytes": null, "serialized_payload_bytes": 1087, "side_information_bytes": 0, "fetched_bytes": 5978, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 52381.81649975013, "hops": {"1": 6913.723333040252, "2": 6537.521208985709, "3": 6488.541791914031, "4": 6486.346457968466, "5": 6492.82487493474, "6": 6482.172666001134, "7": 6484.1609169961885, "8": 6496.52524990961}, "receiver_prefill_ms": 228.62429195083678, "ttft_ms": 12.161041027866304, "decode_ms": 83.1940839998424}, "total_wall_ms": 52706.45424979739, "ttft_ms": 12.161041027866304, "input_usage": {"lookups": 28}, "output_usage": {"output_tokens": 7}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "Lunlith-13", "evidence_artifact": {"ledger": {"facts_total": 63, "facts_retained": 27, "retention_rate": 0.42857142857142855, "retention_by_kind": {"person": 0.5, "date": 0.125, "amount": 0.6153846153846154, "status": 0.75, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1087}, "ledger_by_hop": {"1": {"facts_total": 63, "facts_retained": 27, "retention_rate": 0.42857142857142855, "retention_by_kind": {"person": 0.5, "date": 0.125, "amount": 0.6153846153846154, "status": 0.75, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1087}, "2": {"facts_total": 63, "facts_retained": 27, "retention_rate": 0.42857142857142855, "retention_by_kind": {"person": 0.5, "date": 0.125, "amount": 0.6153846153846154, "status": 0.75, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1087}, "3": {"facts_total": 63, "facts_retained": 27, "retention_rate": 0.42857142857142855, "retention_by_kind": {"person": 0.5, "date": 0.125, "amount": 0.6153846153846154, "status": 0.75, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1087}, "4": {"facts_total": 63, "facts_retained": 27, "retention_rate": 0.42857142857142855, "retention_by_kind": {"person": 0.5, "date": 0.125, "amount": 0.6153846153846154, "status": 0.75, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1087}, "5": {"facts_total": 63, "facts_retained": 27, "retention_rate": 0.42857142857142855, "retention_by_kind": {"person": 0.5, "date": 0.125, "amount": 0.6153846153846154, "status": 0.75, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1087}, "6": {"facts_total": 63, "facts_retained": 27, "retention_rate": 0.42857142857142855, "retention_by_kind": {"person": 0.5, "date": 0.125, "amount": 0.6153846153846154, "status": 0.75, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1087}, "7": {"facts_total": 63, "facts_retained": 27, "retention_rate": 0.42857142857142855, "retention_by_kind": {"person": 0.5, "date": 0.125, "amount": 0.6153846153846154, "status": 0.75, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1087}, "8": {"facts_total": 63, "facts_retained": 27, "retention_rate": 0.42857142857142855, "retention_by_kind": {"person": 0.5, "date": 0.125, "amount": 0.6153846153846154, "status": 0.75, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1087}}}, "reference_key_id": "keys/synthetic/syn_03027_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "1f88f0e78c63504e418dea843641b99a98555e15d47b02cd5cb2f57da3003dea", "output_hash": "c45f000afe6d865c8167954eebdff761141225817d4f5b26c962cd50cd75d391"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "ad6f532b3b569a830dcf7911", "attempt_id": "22e400367940", "run_group_id": "E4_core:A4|repairable|qwen3b:h8", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:31:44.953782Z", "task_family": "synthetic", "document_id": "syn_03027_2000", "document_hash": "10326080979f71b625e4110c7c19767e5a752e267de5034f7e37d98cb6e06249", "question_id": "q11_aggregation", "split": "test", "cluster_id": "syn_03027_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 8, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1955, "query_tokens": 606, "message_tokens": 512, "message_utf8_bytes": 1087, "cache_tensor_bytes": null, "serialized_payload_bytes": 1087, "side_information_bytes": 0, "fetched_bytes": 5978, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 52381.81649975013, "hops": {"1": 6913.723333040252, "2": 6537.521208985709, "3": 6488.541791914031, "4": 6486.346457968466, "5": 6492.82487493474, "6": 6482.172666001134, "7": 6484.1609169961885, "8": 6496.52524990961}, "receiver_prefill_ms": 226.94208403117955, "ttft_ms": 12.143999920226634, "decode_ms": 119.12737495731562}, "total_wall_ms": 52740.74666574597, "ttft_ms": 12.143999920226634, "input_usage": {"lookups": 28}, "output_usage": {"output_tokens": 10}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "23,450,777", "evidence_artifact": {"ledger": {"facts_total": 63, "facts_retained": 27, "retention_rate": 0.42857142857142855, "retention_by_kind": {"person": 0.5, "date": 0.125, "amount": 0.6153846153846154, "status": 0.75, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1087}, "ledger_by_hop": {"1": {"facts_total": 63, "facts_retained": 27, "retention_rate": 0.42857142857142855, "retention_by_kind": {"person": 0.5, "date": 0.125, "amount": 0.6153846153846154, "status": 0.75, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1087}, "2": {"facts_total": 63, "facts_retained": 27, "retention_rate": 0.42857142857142855, "retention_by_kind": {"person": 0.5, "date": 0.125, "amount": 0.6153846153846154, "status": 0.75, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1087}, "3": {"facts_total": 63, "facts_retained": 27, "retention_rate": 0.42857142857142855, "retention_by_kind": {"person": 0.5, "date": 0.125, "amount": 0.6153846153846154, "status": 0.75, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1087}, "4": {"facts_total": 63, "facts_retained": 27, "retention_rate": 0.42857142857142855, "retention_by_kind": {"person": 0.5, "date": 0.125, "amount": 0.6153846153846154, "status": 0.75, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1087}, "5": {"facts_total": 63, "facts_retained": 27, "retention_rate": 0.42857142857142855, "retention_by_kind": {"person": 0.5, "date": 0.125, "amount": 0.6153846153846154, "status": 0.75, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1087}, "6": {"facts_total": 63, "facts_retained": 27, "retention_rate": 0.42857142857142855, "retention_by_kind": {"person": 0.5, "date": 0.125, "amount": 0.6153846153846154, "status": 0.75, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1087}, "7": {"facts_total": 63, "facts_retained": 27, "retention_rate": 0.42857142857142855, "retention_by_kind": {"person": 0.5, "date": 0.125, "amount": 0.6153846153846154, "status": 0.75, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1087}, "8": {"facts_total": 63, "facts_retained": 27, "retention_rate": 0.42857142857142855, "retention_by_kind": {"person": 0.5, "date": 0.125, "amount": 0.6153846153846154, "status": 0.75, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1087}}}, "reference_key_id": "keys/synthetic/syn_03027_2000.json", "task_scores": {"correct": 0.0, "abstained": 0.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "1f88f0e78c63504e418dea843641b99a98555e15d47b02cd5cb2f57da3003dea", "output_hash": "3894c3c39e5c76c297dfb1859a5e1724e2454039e072357510c841bc907d01a5"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "b84b69525c6fcba469e0f1b8", "attempt_id": "6a67ca0887d2", "run_group_id": "E4_core:A4|repairable|qwen3b:h8", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:31:45.330111Z", "task_family": "synthetic", "document_id": "syn_03027_2000", "document_hash": "10326080979f71b625e4110c7c19767e5a752e267de5034f7e37d98cb6e06249", "question_id": "q12_unit_conversion", "split": "test", "cluster_id": "syn_03027_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 8, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1955, "query_tokens": 603, "message_tokens": 512, "message_utf8_bytes": 1087, "cache_tensor_bytes": null, "serialized_payload_bytes": 1087, "side_information_bytes": 0, "fetched_bytes": 5978, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 52381.81649975013, "hops": {"1": 6913.723333040252, "2": 6537.521208985709, "3": 6488.541791914031, "4": 6486.346457968466, "5": 6492.82487493474, "6": 6482.172666001134, "7": 6484.1609169961885, "8": 6496.52524990961}, "receiver_prefill_ms": 226.91083303652704, "ttft_ms": 12.10962503682822, "decode_ms": 12.047167052514851}, "total_wall_ms": 52633.554916712455, "ttft_ms": 12.10962503682822, "input_usage": {"lookups": 28}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 63, "facts_retained": 27, "retention_rate": 0.42857142857142855, "retention_by_kind": {"person": 0.5, "date": 0.125, "amount": 0.6153846153846154, "status": 0.75, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1087}, "ledger_by_hop": {"1": {"facts_total": 63, "facts_retained": 27, "retention_rate": 0.42857142857142855, "retention_by_kind": {"person": 0.5, "date": 0.125, "amount": 0.6153846153846154, "status": 0.75, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1087}, "2": {"facts_total": 63, "facts_retained": 27, "retention_rate": 0.42857142857142855, "retention_by_kind": {"person": 0.5, "date": 0.125, "amount": 0.6153846153846154, "status": 0.75, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1087}, "3": {"facts_total": 63, "facts_retained": 27, "retention_rate": 0.42857142857142855, "retention_by_kind": {"person": 0.5, "date": 0.125, "amount": 0.6153846153846154, "status": 0.75, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1087}, "4": {"facts_total": 63, "facts_retained": 27, "retention_rate": 0.42857142857142855, "retention_by_kind": {"person": 0.5, "date": 0.125, "amount": 0.6153846153846154, "status": 0.75, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1087}, "5": {"facts_total": 63, "facts_retained": 27, "retention_rate": 0.42857142857142855, "retention_by_kind": {"person": 0.5, "date": 0.125, "amount": 0.6153846153846154, "status": 0.75, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1087}, "6": {"facts_total": 63, "facts_retained": 27, "retention_rate": 0.42857142857142855, "retention_by_kind": {"person": 0.5, "date": 0.125, "amount": 0.6153846153846154, "status": 0.75, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1087}, "7": {"facts_total": 63, "facts_retained": 27, "retention_rate": 0.42857142857142855, "retention_by_kind": {"person": 0.5, "date": 0.125, "amount": 0.6153846153846154, "status": 0.75, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1087}, "8": {"facts_total": 63, "facts_retained": 27, "retention_rate": 0.42857142857142855, "retention_by_kind": {"person": 0.5, "date": 0.125, "amount": 0.6153846153846154, "status": 0.75, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1087}}}, "reference_key_id": "keys/synthetic/syn_03027_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "1f88f0e78c63504e418dea843641b99a98555e15d47b02cd5cb2f57da3003dea", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "3a6ef133feda9ecc7a8e4e63", "attempt_id": "3e87f416fec5", "run_group_id": "E4_core:A4|repairable|qwen3b:h8", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:31:45.599182Z", "task_family": "synthetic", "document_id": "syn_03027_2000", "document_hash": "10326080979f71b625e4110c7c19767e5a752e267de5034f7e37d98cb6e06249", "question_id": "q13_unanswerable", "split": "test", "cluster_id": "syn_03027_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 8, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1955, "query_tokens": 606, "message_tokens": 512, "message_utf8_bytes": 1087, "cache_tensor_bytes": null, "serialized_payload_bytes": 1087, "side_information_bytes": 0, "fetched_bytes": 5978, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 52381.81649975013, "hops": {"1": 6913.723333040252, "2": 6537.521208985709, "3": 6488.541791914031, "4": 6486.346457968466, "5": 6492.82487493474, "6": 6482.172666001134, "7": 6484.1609169961885, "8": 6496.52524990961}, "receiver_prefill_ms": 226.62904195021838, "ttft_ms": 12.135457945987582, "decode_ms": 11.944000027142465}, "total_wall_ms": 52633.18341667764, "ttft_ms": 12.135457945987582, "input_usage": {"lookups": 28}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 63, "facts_retained": 27, "retention_rate": 0.42857142857142855, "retention_by_kind": {"person": 0.5, "date": 0.125, "amount": 0.6153846153846154, "status": 0.75, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1087}, "ledger_by_hop": {"1": {"facts_total": 63, "facts_retained": 27, "retention_rate": 0.42857142857142855, "retention_by_kind": {"person": 0.5, "date": 0.125, "amount": 0.6153846153846154, "status": 0.75, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1087}, "2": {"facts_total": 63, "facts_retained": 27, "retention_rate": 0.42857142857142855, "retention_by_kind": {"person": 0.5, "date": 0.125, "amount": 0.6153846153846154, "status": 0.75, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1087}, "3": {"facts_total": 63, "facts_retained": 27, "retention_rate": 0.42857142857142855, "retention_by_kind": {"person": 0.5, "date": 0.125, "amount": 0.6153846153846154, "status": 0.75, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1087}, "4": {"facts_total": 63, "facts_retained": 27, "retention_rate": 0.42857142857142855, "retention_by_kind": {"person": 0.5, "date": 0.125, "amount": 0.6153846153846154, "status": 0.75, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1087}, "5": {"facts_total": 63, "facts_retained": 27, "retention_rate": 0.42857142857142855, "retention_by_kind": {"person": 0.5, "date": 0.125, "amount": 0.6153846153846154, "status": 0.75, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1087}, "6": {"facts_total": 63, "facts_retained": 27, "retention_rate": 0.42857142857142855, "retention_by_kind": {"person": 0.5, "date": 0.125, "amount": 0.6153846153846154, "status": 0.75, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1087}, "7": {"facts_total": 63, "facts_retained": 27, "retention_rate": 0.42857142857142855, "retention_by_kind": {"person": 0.5, "date": 0.125, "amount": 0.6153846153846154, "status": 0.75, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1087}, "8": {"facts_total": 63, "facts_retained": 27, "retention_rate": 0.42857142857142855, "retention_by_kind": {"person": 0.5, "date": 0.125, "amount": 0.6153846153846154, "status": 0.75, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1087}}}, "reference_key_id": "keys/synthetic/syn_03027_2000.json", "task_scores": {"correct": 1.0, "abstained": 1.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "1f88f0e78c63504e418dea843641b99a98555e15d47b02cd5cb2f57da3003dea", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "4a334d56e7944cbea659c31e", "attempt_id": "d308556439b7", "run_group_id": "E4_core:A4|repairable|qwen3b:h8", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:31:45.867476Z", "task_family": "synthetic", "document_id": "syn_03027_2000", "document_hash": "10326080979f71b625e4110c7c19767e5a752e267de5034f7e37d98cb6e06249", "question_id": "q14_unanswerable", "split": "test", "cluster_id": "syn_03027_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 8, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1955, "query_tokens": 608, "message_tokens": 512, "message_utf8_bytes": 1087, "cache_tensor_bytes": null, "serialized_payload_bytes": 1087, "side_information_bytes": 0, "fetched_bytes": 5978, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 52381.81649975013, "hops": {"1": 6913.723333040252, "2": 6537.521208985709, "3": 6488.541791914031, "4": 6486.346457968466, "5": 6492.82487493474, "6": 6482.172666001134, "7": 6484.1609169961885, "8": 6496.52524990961}, "receiver_prefill_ms": 226.80845798458904, "ttft_ms": 12.105750036425889, "decode_ms": 11.993582942523062}, "total_wall_ms": 52633.39083374012, "ttft_ms": 12.105750036425889, "input_usage": {"lookups": 28}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 63, "facts_retained": 27, "retention_rate": 0.42857142857142855, "retention_by_kind": {"person": 0.5, "date": 0.125, "amount": 0.6153846153846154, "status": 0.75, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1087}, "ledger_by_hop": {"1": {"facts_total": 63, "facts_retained": 27, "retention_rate": 0.42857142857142855, "retention_by_kind": {"person": 0.5, "date": 0.125, "amount": 0.6153846153846154, "status": 0.75, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1087}, "2": {"facts_total": 63, "facts_retained": 27, "retention_rate": 0.42857142857142855, "retention_by_kind": {"person": 0.5, "date": 0.125, "amount": 0.6153846153846154, "status": 0.75, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1087}, "3": {"facts_total": 63, "facts_retained": 27, "retention_rate": 0.42857142857142855, "retention_by_kind": {"person": 0.5, "date": 0.125, "amount": 0.6153846153846154, "status": 0.75, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1087}, "4": {"facts_total": 63, "facts_retained": 27, "retention_rate": 0.42857142857142855, "retention_by_kind": {"person": 0.5, "date": 0.125, "amount": 0.6153846153846154, "status": 0.75, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1087}, "5": {"facts_total": 63, "facts_retained": 27, "retention_rate": 0.42857142857142855, "retention_by_kind": {"person": 0.5, "date": 0.125, "amount": 0.6153846153846154, "status": 0.75, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1087}, "6": {"facts_total": 63, "facts_retained": 27, "retention_rate": 0.42857142857142855, "retention_by_kind": {"person": 0.5, "date": 0.125, "amount": 0.6153846153846154, "status": 0.75, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1087}, "7": {"facts_total": 63, "facts_retained": 27, "retention_rate": 0.42857142857142855, "retention_by_kind": {"person": 0.5, "date": 0.125, "amount": 0.6153846153846154, "status": 0.75, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1087}, "8": {"facts_total": 63, "facts_retained": 27, "retention_rate": 0.42857142857142855, "retention_by_kind": {"person": 0.5, "date": 0.125, "amount": 0.6153846153846154, "status": 0.75, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1087}}}, "reference_key_id": "keys/synthetic/syn_03027_2000.json", "task_scores": {"correct": 1.0, "abstained": 1.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "1f88f0e78c63504e418dea843641b99a98555e15d47b02cd5cb2f57da3003dea", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "dab64d0a547352a5de2a9544", "attempt_id": "13793f9a59d3", "run_group_id": "E4_core:A5|repairable|qwen3b:h1", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:31:53.041817Z", "task_family": "synthetic", "document_id": "syn_03027_2000", "document_hash": "10326080979f71b625e4110c7c19767e5a752e267de5034f7e37d98cb6e06249", "question_id": "q00_single_fact", "split": "test", "cluster_id": "syn_03027_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 1, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1955, "query_tokens": 411, "message_tokens": 329, "message_utf8_bytes": 827, "cache_tensor_bytes": null, "serialized_payload_bytes": 827, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 6904.5422920025885, "hops": {"1": 6904.5422920025885}, "receiver_prefill_ms": 161.35308309458196, "ttft_ms": 11.968999984674156, "decode_ms": 11.747209005989134}, "total_wall_ms": 7090.143167064525, "ttft_ms": 11.968999984674156, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 63, "facts_retained": 9, "retention_rate": 0.14285714285714285, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.25, "amount": 0.07692307692307693, "status": 0.25, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 787}, "ledger_by_hop": {"1": {"facts_total": 63, "facts_retained": 9, "retention_rate": 0.14285714285714285, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.25, "amount": 0.07692307692307693, "status": 0.25, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 787}}}, "reference_key_id": "keys/synthetic/syn_03027_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "a3bc07ae34a4f72dd11a6738ade8463df3fa03c063d19d58651b6ead2a13f850", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "98e33e69fcd96548462d24e8", "attempt_id": "1f644926e934", "run_group_id": "E4_core:A5|repairable|qwen3b:h1", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:31:53.244476Z", "task_family": "synthetic", "document_id": "syn_03027_2000", "document_hash": "10326080979f71b625e4110c7c19767e5a752e267de5034f7e37d98cb6e06249", "question_id": "q01_single_fact", "split": "test", "cluster_id": "syn_03027_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 1, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1955, "query_tokens": 418, "message_tokens": 329, "message_utf8_bytes": 827, "cache_tensor_bytes": null, "serialized_payload_bytes": 827, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 6904.5422920025885, "hops": {"1": 6904.5422920025885}, "receiver_prefill_ms": 162.33229194767773, "ttft_ms": 12.009500060230494, "decode_ms": 11.792459059506655}, "total_wall_ms": 7091.230208054185, "ttft_ms": 12.009500060230494, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 63, "facts_retained": 9, "retention_rate": 0.14285714285714285, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.25, "amount": 0.07692307692307693, "status": 0.25, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 787}, "ledger_by_hop": {"1": {"facts_total": 63, "facts_retained": 9, "retention_rate": 0.14285714285714285, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.25, "amount": 0.07692307692307693, "status": 0.25, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 787}}}, "reference_key_id": "keys/synthetic/syn_03027_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "a3bc07ae34a4f72dd11a6738ade8463df3fa03c063d19d58651b6ead2a13f850", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "5fc6b89d06b24bae9d1c3f3a", "attempt_id": "af09ad849a74", "run_group_id": "E4_core:A5|repairable|qwen3b:h1", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:31:53.448073Z", "task_family": "synthetic", "document_id": "syn_03027_2000", "document_hash": "10326080979f71b625e4110c7c19767e5a752e267de5034f7e37d98cb6e06249", "question_id": "q02_exact_copy", "split": "test", "cluster_id": "syn_03027_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 1, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1955, "query_tokens": 417, "message_tokens": 329, "message_utf8_bytes": 827, "cache_tensor_bytes": null, "serialized_payload_bytes": 827, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 6904.5422920025885, "hops": {"1": 6904.5422920025885}, "receiver_prefill_ms": 162.48287493363023, "ttft_ms": 11.940457974560559, "decode_ms": 11.760167079046369}, "total_wall_ms": 7091.2466669688, "ttft_ms": 11.940457974560559, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 63, "facts_retained": 9, "retention_rate": 0.14285714285714285, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.25, "amount": 0.07692307692307693, "status": 0.25, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 787}, "ledger_by_hop": {"1": {"facts_total": 63, "facts_retained": 9, "retention_rate": 0.14285714285714285, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.25, "amount": 0.07692307692307693, "status": 0.25, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 787}}}, "reference_key_id": "keys/synthetic/syn_03027_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "a3bc07ae34a4f72dd11a6738ade8463df3fa03c063d19d58651b6ead2a13f850", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "aa2a802f302d6af100ca2314", "attempt_id": "949a3ae10a75", "run_group_id": "E4_core:A5|repairable|qwen3b:h1", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:31:53.651464Z", "task_family": "synthetic", "document_id": "syn_03027_2000", "document_hash": "10326080979f71b625e4110c7c19767e5a752e267de5034f7e37d98cb6e06249", "question_id": "q03_exact_copy", "split": "test", "cluster_id": "syn_03027_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 1, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1955, "query_tokens": 412, "message_tokens": 329, "message_utf8_bytes": 827, "cache_tensor_bytes": null, "serialized_payload_bytes": 827, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 6904.5422920025885, "hops": {"1": 6904.5422920025885}, "receiver_prefill_ms": 161.80162504315376, "ttft_ms": 12.081459048204124, "decode_ms": 106.64941591676325}, "total_wall_ms": 7185.678042005748, "ttft_ms": 12.081459048204124, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 9}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "Priyamvada Sørlie", "evidence_artifact": {"ledger": {"facts_total": 63, "facts_retained": 9, "retention_rate": 0.14285714285714285, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.25, "amount": 0.07692307692307693, "status": 0.25, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 787}, "ledger_by_hop": {"1": {"facts_total": 63, "facts_retained": 9, "retention_rate": 0.14285714285714285, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.25, "amount": 0.07692307692307693, "status": 0.25, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 787}}}, "reference_key_id": "keys/synthetic/syn_03027_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "a3bc07ae34a4f72dd11a6738ade8463df3fa03c063d19d58651b6ead2a13f850", "output_hash": "c5f554196ed23dd71ad995ec89f7142099d524dbb6534cc10d94a0c89d13e792"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "83714a2aa1eb09dc4279d2d0", "attempt_id": "bc10c9c8df81", "run_group_id": "E4_core:A5|repairable|qwen3b:h1", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:31:53.949155Z", "task_family": "synthetic", "document_id": "syn_03027_2000", "document_hash": "10326080979f71b625e4110c7c19767e5a752e267de5034f7e37d98cb6e06249", "question_id": "q04_multi_hop", "split": "test", "cluster_id": "syn_03027_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 1, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1955, "query_tokens": 423, "message_tokens": 329, "message_utf8_bytes": 827, "cache_tensor_bytes": null, "serialized_payload_bytes": 827, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 6904.5422920025885, "hops": {"1": 6904.5422920025885}, "receiver_prefill_ms": 162.68145805224776, "ttft_ms": 11.984875076450408, "decode_ms": 11.788874980993569}, "total_wall_ms": 7091.526416945271, "ttft_ms": 11.984875076450408, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 63, "facts_retained": 9, "retention_rate": 0.14285714285714285, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.25, "amount": 0.07692307692307693, "status": 0.25, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 787}, "ledger_by_hop": {"1": {"facts_total": 63, "facts_retained": 9, "retention_rate": 0.14285714285714285, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.25, "amount": 0.07692307692307693, "status": 0.25, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 787}}}, "reference_key_id": "keys/synthetic/syn_03027_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "a3bc07ae34a4f72dd11a6738ade8463df3fa03c063d19d58651b6ead2a13f850", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "6bd30c8f10b52862d5810f45", "attempt_id": "7d2fe4959243", "run_group_id": "E4_core:A5|repairable|qwen3b:h1", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:31:54.152836Z", "task_family": "synthetic", "document_id": "syn_03027_2000", "document_hash": "10326080979f71b625e4110c7c19767e5a752e267de5034f7e37d98cb6e06249", "question_id": "q05_multi_hop", "split": "test", "cluster_id": "syn_03027_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 1, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1955, "query_tokens": 420, "message_tokens": 329, "message_utf8_bytes": 827, "cache_tensor_bytes": null, "serialized_payload_bytes": 827, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 6904.5422920025885, "hops": {"1": 6904.5422920025885}, "receiver_prefill_ms": 162.5219580018893, "ttft_ms": 12.112334021367133, "decode_ms": 118.49020794034004}, "total_wall_ms": 7198.215292068198, "ttft_ms": 12.112334021367133, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 10}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "WT-26M-9227", "evidence_artifact": {"ledger": {"facts_total": 63, "facts_retained": 9, "retention_rate": 0.14285714285714285, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.25, "amount": 0.07692307692307693, "status": 0.25, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 787}, "ledger_by_hop": {"1": {"facts_total": 63, "facts_retained": 9, "retention_rate": 0.14285714285714285, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.25, "amount": 0.07692307692307693, "status": 0.25, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 787}}}, "reference_key_id": "keys/synthetic/syn_03027_2000.json", "task_scores": {"correct": 0.0, "abstained": 0.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "a3bc07ae34a4f72dd11a6738ade8463df3fa03c063d19d58651b6ead2a13f850", "output_hash": "ef2302e0ab6f5a6951b11463fea177ba766b9d438afcaf6e6101ebfa3ba38aba"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "b7896adeafc7d849a4c3a0af", "attempt_id": "11e6a2cd6e04", "run_group_id": "E4_core:A5|repairable|qwen3b:h1", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:31:54.463119Z", "task_family": "synthetic", "document_id": "syn_03027_2000", "document_hash": "10326080979f71b625e4110c7c19767e5a752e267de5034f7e37d98cb6e06249", "question_id": "q06_negation", "split": "test", "cluster_id": "syn_03027_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 1, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1955, "query_tokens": 412, "message_tokens": 329, "message_utf8_bytes": 827, "cache_tensor_bytes": null, "serialized_payload_bytes": 827, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 6904.5422920025885, "hops": {"1": 6904.5422920025885}, "receiver_prefill_ms": 162.01400000136346, "ttft_ms": 11.932500055991113, "decode_ms": 11.78358297329396}, "total_wall_ms": 7090.810250956565, "ttft_ms": 11.932500055991113, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 63, "facts_retained": 9, "retention_rate": 0.14285714285714285, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.25, "amount": 0.07692307692307693, "status": 0.25, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 787}, "ledger_by_hop": {"1": {"facts_total": 63, "facts_retained": 9, "retention_rate": 0.14285714285714285, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.25, "amount": 0.07692307692307693, "status": 0.25, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 787}}}, "reference_key_id": "keys/synthetic/syn_03027_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "a3bc07ae34a4f72dd11a6738ade8463df3fa03c063d19d58651b6ead2a13f850", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "62d9ca57dfb4018dc86169d0", "attempt_id": "72f06d99d7df", "run_group_id": "E4_core:A5|repairable|qwen3b:h1", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:31:54.665921Z", "task_family": "synthetic", "document_id": "syn_03027_2000", "document_hash": "10326080979f71b625e4110c7c19767e5a752e267de5034f7e37d98cb6e06249", "question_id": "q07_negation", "split": "test", "cluster_id": "syn_03027_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 1, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1955, "query_tokens": 413, "message_tokens": 329, "message_utf8_bytes": 827, "cache_tensor_bytes": null, "serialized_payload_bytes": 827, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 6904.5422920025885, "hops": {"1": 6904.5422920025885}, "receiver_prefill_ms": 161.93608404137194, "ttft_ms": 11.971125029958785, "decode_ms": 11.868417030200362}, "total_wall_ms": 7090.858459006995, "ttft_ms": 11.971125029958785, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 63, "facts_retained": 9, "retention_rate": 0.14285714285714285, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.25, "amount": 0.07692307692307693, "status": 0.25, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 787}, "ledger_by_hop": {"1": {"facts_total": 63, "facts_retained": 9, "retention_rate": 0.14285714285714285, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.25, "amount": 0.07692307692307693, "status": 0.25, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 787}}}, "reference_key_id": "keys/synthetic/syn_03027_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "a3bc07ae34a4f72dd11a6738ade8463df3fa03c063d19d58651b6ead2a13f850", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "bc3895362ef22e676128c686", "attempt_id": "60e76ad9438f", "run_group_id": "E4_core:A5|repairable|qwen3b:h1", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:31:54.868576Z", "task_family": "synthetic", "document_id": "syn_03027_2000", "document_hash": "10326080979f71b625e4110c7c19767e5a752e267de5034f7e37d98cb6e06249", "question_id": "q08_revision", "split": "test", "cluster_id": "syn_03027_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 1, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1955, "query_tokens": 427, "message_tokens": 329, "message_utf8_bytes": 827, "cache_tensor_bytes": null, "serialized_payload_bytes": 827, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 6904.5422920025885, "hops": {"1": 6904.5422920025885}, "receiver_prefill_ms": 162.84087498206645, "ttft_ms": 11.946416925638914, "decode_ms": 105.67383305169642}, "total_wall_ms": 7185.53083401639, "ttft_ms": 11.946416925638914, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 9}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "6,914,937", "evidence_artifact": {"ledger": {"facts_total": 63, "facts_retained": 9, "retention_rate": 0.14285714285714285, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.25, "amount": 0.07692307692307693, "status": 0.25, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 787}, "ledger_by_hop": {"1": {"facts_total": 63, "facts_retained": 9, "retention_rate": 0.14285714285714285, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.25, "amount": 0.07692307692307693, "status": 0.25, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 787}}}, "reference_key_id": "keys/synthetic/syn_03027_2000.json", "task_scores": {"correct": 0.0, "abstained": 0.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "a3bc07ae34a4f72dd11a6738ade8463df3fa03c063d19d58651b6ead2a13f850", "output_hash": "c56851abfdbf8851226a1e71c4911c616a691c490a883eb754e61bac551d004c"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "9e7f5534461c015a02fa459f", "attempt_id": "dfdd77e8b190", "run_group_id": "E4_core:A5|repairable|qwen3b:h1", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:31:55.165960Z", "task_family": "synthetic", "document_id": "syn_03027_2000", "document_hash": "10326080979f71b625e4110c7c19767e5a752e267de5034f7e37d98cb6e06249", "question_id": "q09_revision", "split": "test", "cluster_id": "syn_03027_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 1, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1955, "query_tokens": 420, "message_tokens": 329, "message_utf8_bytes": 827, "cache_tensor_bytes": null, "serialized_payload_bytes": 827, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 6904.5422920025885, "hops": {"1": 6904.5422920025885}, "receiver_prefill_ms": 162.5372909475118, "ttft_ms": 11.957375099882483, "decode_ms": 93.99795904755592}, "total_wall_ms": 7173.577666981146, "ttft_ms": 11.957375099882483, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 8}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "15828866", "evidence_artifact": {"ledger": {"facts_total": 63, "facts_retained": 9, "retention_rate": 0.14285714285714285, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.25, "amount": 0.07692307692307693, "status": 0.25, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 787}, "ledger_by_hop": {"1": {"facts_total": 63, "facts_retained": 9, "retention_rate": 0.14285714285714285, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.25, "amount": 0.07692307692307693, "status": 0.25, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 787}}}, "reference_key_id": "keys/synthetic/syn_03027_2000.json", "task_scores": {"correct": 0.0, "abstained": 0.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "a3bc07ae34a4f72dd11a6738ade8463df3fa03c063d19d58651b6ead2a13f850", "output_hash": "c9d49be6f876880a6dfe30589ff0b746c048ebaf1842118a883f5503a349a8b6"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "9bea21ab7d2e05318774b882", "attempt_id": "c24f4d354a96", "run_group_id": "E4_core:A5|repairable|qwen3b:h1", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:31:55.451771Z", "task_family": "synthetic", "document_id": "syn_03027_2000", "document_hash": "10326080979f71b625e4110c7c19767e5a752e267de5034f7e37d98cb6e06249", "question_id": "q10_temporal", "split": "test", "cluster_id": "syn_03027_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 1, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1955, "query_tokens": 430, "message_tokens": 329, "message_utf8_bytes": 827, "cache_tensor_bytes": null, "serialized_payload_bytes": 827, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 6904.5422920025885, "hops": {"1": 6904.5422920025885}, "receiver_prefill_ms": 162.81020897440612, "ttft_ms": 11.880208039656281, "decode_ms": 11.8079170351848}, "total_wall_ms": 7091.553583042696, "ttft_ms": 11.880208039656281, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 63, "facts_retained": 9, "retention_rate": 0.14285714285714285, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.25, "amount": 0.07692307692307693, "status": 0.25, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 787}, "ledger_by_hop": {"1": {"facts_total": 63, "facts_retained": 9, "retention_rate": 0.14285714285714285, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.25, "amount": 0.07692307692307693, "status": 0.25, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 787}}}, "reference_key_id": "keys/synthetic/syn_03027_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "a3bc07ae34a4f72dd11a6738ade8463df3fa03c063d19d58651b6ead2a13f850", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "805673cfffe8889e3fd2bff6", "attempt_id": "424674c91996", "run_group_id": "E4_core:A5|repairable|qwen3b:h1", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:31:55.655480Z", "task_family": "synthetic", "document_id": "syn_03027_2000", "document_hash": "10326080979f71b625e4110c7c19767e5a752e267de5034f7e37d98cb6e06249", "question_id": "q11_aggregation", "split": "test", "cluster_id": "syn_03027_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 1, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1955, "query_tokens": 423, "message_tokens": 329, "message_utf8_bytes": 827, "cache_tensor_bytes": null, "serialized_payload_bytes": 827, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 6904.5422920025885, "hops": {"1": 6904.5422920025885}, "receiver_prefill_ms": 162.78024995699525, "ttft_ms": 11.979625094681978, "decode_ms": 82.71695801522583}, "total_wall_ms": 7162.57908300031, "ttft_ms": 11.979625094681978, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 7}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "6914937", "evidence_artifact": {"ledger": {"facts_total": 63, "facts_retained": 9, "retention_rate": 0.14285714285714285, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.25, "amount": 0.07692307692307693, "status": 0.25, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 787}, "ledger_by_hop": {"1": {"facts_total": 63, "facts_retained": 9, "retention_rate": 0.14285714285714285, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.25, "amount": 0.07692307692307693, "status": 0.25, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 787}}}, "reference_key_id": "keys/synthetic/syn_03027_2000.json", "task_scores": {"correct": 0.0, "abstained": 0.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "a3bc07ae34a4f72dd11a6738ade8463df3fa03c063d19d58651b6ead2a13f850", "output_hash": "534d40ded32f784c68307dd2386faed2a27bf39dcc27321ee6985e42ccd285e5"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "69eab1984a9641b97adf6fc1", "attempt_id": "0ef1215be621", "run_group_id": "E4_core:A5|repairable|qwen3b:h1", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:31:55.930212Z", "task_family": "synthetic", "document_id": "syn_03027_2000", "document_hash": "10326080979f71b625e4110c7c19767e5a752e267de5034f7e37d98cb6e06249", "question_id": "q12_unit_conversion", "split": "test", "cluster_id": "syn_03027_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 1, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1955, "query_tokens": 420, "message_tokens": 329, "message_utf8_bytes": 827, "cache_tensor_bytes": null, "serialized_payload_bytes": 827, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 6904.5422920025885, "hops": {"1": 6904.5422920025885}, "receiver_prefill_ms": 162.80720895156264, "ttft_ms": 11.959749972447753, "decode_ms": 11.858667014166713}, "total_wall_ms": 7091.693707974628, "ttft_ms": 11.959749972447753, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 63, "facts_retained": 9, "retention_rate": 0.14285714285714285, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.25, "amount": 0.07692307692307693, "status": 0.25, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 787}, "ledger_by_hop": {"1": {"facts_total": 63, "facts_retained": 9, "retention_rate": 0.14285714285714285, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.25, "amount": 0.07692307692307693, "status": 0.25, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 787}}}, "reference_key_id": "keys/synthetic/syn_03027_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "a3bc07ae34a4f72dd11a6738ade8463df3fa03c063d19d58651b6ead2a13f850", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "b11dc7bbca785c64eba7d265", "attempt_id": "6d626ab87bb7", "run_group_id": "E4_core:A5|repairable|qwen3b:h1", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:31:56.133825Z", "task_family": "synthetic", "document_id": "syn_03027_2000", "document_hash": "10326080979f71b625e4110c7c19767e5a752e267de5034f7e37d98cb6e06249", "question_id": "q13_unanswerable", "split": "test", "cluster_id": "syn_03027_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 1, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1955, "query_tokens": 423, "message_tokens": 329, "message_utf8_bytes": 827, "cache_tensor_bytes": null, "serialized_payload_bytes": 827, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 6904.5422920025885, "hops": {"1": 6904.5422920025885}, "receiver_prefill_ms": 163.12254196964204, "ttft_ms": 12.071083998307586, "decode_ms": 11.853625066578388}, "total_wall_ms": 7092.146209091879, "ttft_ms": 12.071083998307586, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 63, "facts_retained": 9, "retention_rate": 0.14285714285714285, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.25, "amount": 0.07692307692307693, "status": 0.25, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 787}, "ledger_by_hop": {"1": {"facts_total": 63, "facts_retained": 9, "retention_rate": 0.14285714285714285, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.25, "amount": 0.07692307692307693, "status": 0.25, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 787}}}, "reference_key_id": "keys/synthetic/syn_03027_2000.json", "task_scores": {"correct": 1.0, "abstained": 1.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "a3bc07ae34a4f72dd11a6738ade8463df3fa03c063d19d58651b6ead2a13f850", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "03714abd5e5ec0e00f0eb703", "attempt_id": "6f4a2578c06b", "run_group_id": "E4_core:A5|repairable|qwen3b:h1", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:31:56.337725Z", "task_family": "synthetic", "document_id": "syn_03027_2000", "document_hash": "10326080979f71b625e4110c7c19767e5a752e267de5034f7e37d98cb6e06249", "question_id": "q14_unanswerable", "split": "test", "cluster_id": "syn_03027_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 1, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1955, "query_tokens": 425, "message_tokens": 329, "message_utf8_bytes": 827, "cache_tensor_bytes": null, "serialized_payload_bytes": 827, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 6904.5422920025885, "hops": {"1": 6904.5422920025885}, "receiver_prefill_ms": 162.3170409584418, "ttft_ms": 11.97616697754711, "decode_ms": 11.79329096339643}, "total_wall_ms": 7091.1500009242445, "ttft_ms": 11.97616697754711, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 63, "facts_retained": 9, "retention_rate": 0.14285714285714285, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.25, "amount": 0.07692307692307693, "status": 0.25, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 787}, "ledger_by_hop": {"1": {"facts_total": 63, "facts_retained": 9, "retention_rate": 0.14285714285714285, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.25, "amount": 0.07692307692307693, "status": 0.25, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 787}}}, "reference_key_id": "keys/synthetic/syn_03027_2000.json", "task_scores": {"correct": 1.0, "abstained": 1.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "a3bc07ae34a4f72dd11a6738ade8463df3fa03c063d19d58651b6ead2a13f850", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "e775fc3bab68ed83ead8cea7", "attempt_id": "0ae83bca10c1", "run_group_id": "E4_core:A5|repairable|qwen3b:h2", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:32:00.922722Z", "task_family": "synthetic", "document_id": "syn_03027_2000", "document_hash": "10326080979f71b625e4110c7c19767e5a752e267de5034f7e37d98cb6e06249", "question_id": "q00_single_fact", "split": "test", "cluster_id": "syn_03027_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 2, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1955, "query_tokens": 420, "message_tokens": 339, "message_utf8_bytes": 858, "cache_tensor_bytes": null, "serialized_payload_bytes": 858, "side_information_bytes": 0, "fetched_bytes": 602, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 11284.805292030796, "hops": {"1": 6904.5422920025885, "2": 4380.263000028208}, "receiver_prefill_ms": 162.09245810750872, "ttft_ms": 11.985916993580759, "decode_ms": 12.057916959747672}, "total_wall_ms": 11471.509834052995, "ttft_ms": 11.985916993580759, "input_usage": {"lookups": 1}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 63, "facts_retained": 12, "retention_rate": 0.19047619047619047, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.25, "amount": 0.07692307692307693, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 818}, "ledger_by_hop": {"1": {"facts_total": 63, "facts_retained": 9, "retention_rate": 0.14285714285714285, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.25, "amount": 0.07692307692307693, "status": 0.25, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 787}, "2": {"facts_total": 63, "facts_retained": 12, "retention_rate": 0.19047619047619047, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.25, "amount": 0.07692307692307693, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 818}}}, "reference_key_id": "keys/synthetic/syn_03027_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "7ab33a7e50920be6dd2887eea3e354ca9b7ea53ad5cd929f38c58204c3e3af7e", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "1696aaf6c268c3e6658bf492", "attempt_id": "127a32e44648", "run_group_id": "E4_core:A5|repairable|qwen3b:h2", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:32:01.126583Z", "task_family": "synthetic", "document_id": "syn_03027_2000", "document_hash": "10326080979f71b625e4110c7c19767e5a752e267de5034f7e37d98cb6e06249", "question_id": "q01_single_fact", "split": "test", "cluster_id": "syn_03027_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 2, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1955, "query_tokens": 427, "message_tokens": 339, "message_utf8_bytes": 858, "cache_tensor_bytes": null, "serialized_payload_bytes": 858, "side_information_bytes": 0, "fetched_bytes": 602, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 11284.805292030796, "hops": {"1": 6904.5422920025885, "2": 4380.263000028208}, "receiver_prefill_ms": 162.84387500490993, "ttft_ms": 11.996875051409006, "decode_ms": 11.861791950650513}, "total_wall_ms": 11472.064876114018, "ttft_ms": 11.996875051409006, "input_usage": {"lookups": 1}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 63, "facts_retained": 12, "retention_rate": 0.19047619047619047, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.25, "amount": 0.07692307692307693, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 818}, "ledger_by_hop": {"1": {"facts_total": 63, "facts_retained": 9, "retention_rate": 0.14285714285714285, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.25, "amount": 0.07692307692307693, "status": 0.25, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 787}, "2": {"facts_total": 63, "facts_retained": 12, "retention_rate": 0.19047619047619047, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.25, "amount": 0.07692307692307693, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 818}}}, "reference_key_id": "keys/synthetic/syn_03027_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "7ab33a7e50920be6dd2887eea3e354ca9b7ea53ad5cd929f38c58204c3e3af7e", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "0d07e34dce806c0e6cb7bed6", "attempt_id": "832687ce30a6", "run_group_id": "E4_core:A5|repairable|qwen3b:h2", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:32:01.330487Z", "task_family": "synthetic", "document_id": "syn_03027_2000", "document_hash": "10326080979f71b625e4110c7c19767e5a752e267de5034f7e37d98cb6e06249", "question_id": "q02_exact_copy", "split": "test", "cluster_id": "syn_03027_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 2, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1955, "query_tokens": 426, "message_tokens": 339, "message_utf8_bytes": 858, "cache_tensor_bytes": null, "serialized_payload_bytes": 858, "side_information_bytes": 0, "fetched_bytes": 602, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 11284.805292030796, "hops": {"1": 6904.5422920025885, "2": 4380.263000028208}, "receiver_prefill_ms": 162.6890420448035, "ttft_ms": 12.013040948659182, "decode_ms": 11.781834065914154}, "total_wall_ms": 11471.826125052758, "ttft_ms": 12.013040948659182, "input_usage": {"lookups": 1}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 63, "facts_retained": 12, "retention_rate": 0.19047619047619047, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.25, "amount": 0.07692307692307693, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 818}, "ledger_by_hop": {"1": {"facts_total": 63, "facts_retained": 9, "retention_rate": 0.14285714285714285, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.25, "amount": 0.07692307692307693, "status": 0.25, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 787}, "2": {"facts_total": 63, "facts_retained": 12, "retention_rate": 0.19047619047619047, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.25, "amount": 0.07692307692307693, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 818}}}, "reference_key_id": "keys/synthetic/syn_03027_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "7ab33a7e50920be6dd2887eea3e354ca9b7ea53ad5cd929f38c58204c3e3af7e", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "d8968a0a4346de1fecc8dcb5", "attempt_id": "fe0b5ef8e6a2", "run_group_id": "E4_core:A5|repairable|qwen3b:h2", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:32:01.533918Z", "task_family": "synthetic", "document_id": "syn_03027_2000", "document_hash": "10326080979f71b625e4110c7c19767e5a752e267de5034f7e37d98cb6e06249", "question_id": "q03_exact_copy", "split": "test", "cluster_id": "syn_03027_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 2, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1955, "query_tokens": 421, "message_tokens": 339, "message_utf8_bytes": 858, "cache_tensor_bytes": null, "serialized_payload_bytes": 858, "side_information_bytes": 0, "fetched_bytes": 602, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 11284.805292030796, "hops": {"1": 6904.5422920025885, "2": 4380.263000028208}, "receiver_prefill_ms": 162.48783294577152, "ttft_ms": 12.087165960110724, "decode_ms": 106.29158304072917}, "total_wall_ms": 11566.244917106815, "ttft_ms": 12.087165960110724, "input_usage": {"lookups": 1}, "output_usage": {"output_tokens": 9}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "Priyamvada Sørlie", "evidence_artifact": {"ledger": {"facts_total": 63, "facts_retained": 12, "retention_rate": 0.19047619047619047, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.25, "amount": 0.07692307692307693, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 818}, "ledger_by_hop": {"1": {"facts_total": 63, "facts_retained": 9, "retention_rate": 0.14285714285714285, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.25, "amount": 0.07692307692307693, "status": 0.25, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 787}, "2": {"facts_total": 63, "facts_retained": 12, "retention_rate": 0.19047619047619047, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.25, "amount": 0.07692307692307693, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 818}}}, "reference_key_id": "keys/synthetic/syn_03027_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "7ab33a7e50920be6dd2887eea3e354ca9b7ea53ad5cd929f38c58204c3e3af7e", "output_hash": "c5f554196ed23dd71ad995ec89f7142099d524dbb6534cc10d94a0c89d13e792"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "d7c55ef06bfb85aedf69d997", "attempt_id": "e1929096b231", "run_group_id": "E4_core:A5|repairable|qwen3b:h2", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:32:01.831743Z", "task_family": "synthetic", "document_id": "syn_03027_2000", "document_hash": "10326080979f71b625e4110c7c19767e5a752e267de5034f7e37d98cb6e06249", "question_id": "q04_multi_hop", "split": "test", "cluster_id": "syn_03027_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 2, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1955, "query_tokens": 432, "message_tokens": 339, "message_utf8_bytes": 858, "cache_tensor_bytes": null, "serialized_payload_bytes": 858, "side_information_bytes": 0, "fetched_bytes": 602, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 11284.805292030796, "hops": {"1": 6904.5422920025885, "2": 4380.263000028208}, "receiver_prefill_ms": 162.69545897375792, "ttft_ms": 12.013084022328258, "decode_ms": 11.807791073806584}, "total_wall_ms": 11471.873625996523, "ttft_ms": 12.013084022328258, "input_usage": {"lookups": 1}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 63, "facts_retained": 12, "retention_rate": 0.19047619047619047, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.25, "amount": 0.07692307692307693, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 818}, "ledger_by_hop": {"1": {"facts_total": 63, "facts_retained": 9, "retention_rate": 0.14285714285714285, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.25, "amount": 0.07692307692307693, "status": 0.25, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 787}, "2": {"facts_total": 63, "facts_retained": 12, "retention_rate": 0.19047619047619047, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.25, "amount": 0.07692307692307693, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 818}}}, "reference_key_id": "keys/synthetic/syn_03027_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "7ab33a7e50920be6dd2887eea3e354ca9b7ea53ad5cd929f38c58204c3e3af7e", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "17e55ae417f32d2b1cfa701e", "attempt_id": "fca1fcbc54f1", "run_group_id": "E4_core:A5|repairable|qwen3b:h2", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:32:02.035823Z", "task_family": "synthetic", "document_id": "syn_03027_2000", "document_hash": "10326080979f71b625e4110c7c19767e5a752e267de5034f7e37d98cb6e06249", "question_id": "q05_multi_hop", "split": "test", "cluster_id": "syn_03027_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 2, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1955, "query_tokens": 429, "message_tokens": 339, "message_utf8_bytes": 858, "cache_tensor_bytes": null, "serialized_payload_bytes": 858, "side_information_bytes": 0, "fetched_bytes": 602, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 11284.805292030796, "hops": {"1": 6904.5422920025885, "2": 4380.263000028208}, "receiver_prefill_ms": 163.11737499199808, "ttft_ms": 11.956166941672564, "decode_ms": 117.67824995331466}, "total_wall_ms": 11578.09308404103, "ttft_ms": 11.956166941672564, "input_usage": {"lookups": 1}, "output_usage": {"output_tokens": 10}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "WT-26M-9227", "evidence_artifact": {"ledger": {"facts_total": 63, "facts_retained": 12, "retention_rate": 0.19047619047619047, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.25, "amount": 0.07692307692307693, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 818}, "ledger_by_hop": {"1": {"facts_total": 63, "facts_retained": 9, "retention_rate": 0.14285714285714285, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.25, "amount": 0.07692307692307693, "status": 0.25, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 787}, "2": {"facts_total": 63, "facts_retained": 12, "retention_rate": 0.19047619047619047, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.25, "amount": 0.07692307692307693, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 818}}}, "reference_key_id": "keys/synthetic/syn_03027_2000.json", "task_scores": {"correct": 0.0, "abstained": 0.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "7ab33a7e50920be6dd2887eea3e354ca9b7ea53ad5cd929f38c58204c3e3af7e", "output_hash": "ef2302e0ab6f5a6951b11463fea177ba766b9d438afcaf6e6101ebfa3ba38aba"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "ab5af495ee733eb9022b0f44", "attempt_id": "9889aead7b6e", "run_group_id": "E4_core:A5|repairable|qwen3b:h2", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:32:02.345805Z", "task_family": "synthetic", "document_id": "syn_03027_2000", "document_hash": "10326080979f71b625e4110c7c19767e5a752e267de5034f7e37d98cb6e06249", "question_id": "q06_negation", "split": "test", "cluster_id": "syn_03027_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 2, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1955, "query_tokens": 421, "message_tokens": 339, "message_utf8_bytes": 858, "cache_tensor_bytes": null, "serialized_payload_bytes": 858, "side_information_bytes": 0, "fetched_bytes": 602, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 11284.805292030796, "hops": {"1": 6904.5422920025885, "2": 4380.263000028208}, "receiver_prefill_ms": 162.54537494387478, "ttft_ms": 11.936415918171406, "decode_ms": 11.806041933596134}, "total_wall_ms": 11471.638292074203, "ttft_ms": 11.936415918171406, "input_usage": {"lookups": 1}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 63, "facts_retained": 12, "retention_rate": 0.19047619047619047, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.25, "amount": 0.07692307692307693, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 818}, "ledger_by_hop": {"1": {"facts_total": 63, "facts_retained": 9, "retention_rate": 0.14285714285714285, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.25, "amount": 0.07692307692307693, "status": 0.25, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 787}, "2": {"facts_total": 63, "facts_retained": 12, "retention_rate": 0.19047619047619047, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.25, "amount": 0.07692307692307693, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 818}}}, "reference_key_id": "keys/synthetic/syn_03027_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "7ab33a7e50920be6dd2887eea3e354ca9b7ea53ad5cd929f38c58204c3e3af7e", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "8dbd1b624ce6756619d5ba71", "attempt_id": "c01f4066c5d1", "run_group_id": "E4_core:A5|repairable|qwen3b:h2", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:32:02.548884Z", "task_family": "synthetic", "document_id": "syn_03027_2000", "document_hash": "10326080979f71b625e4110c7c19767e5a752e267de5034f7e37d98cb6e06249", "question_id": "q07_negation", "split": "test", "cluster_id": "syn_03027_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 2, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1955, "query_tokens": 422, "message_tokens": 339, "message_utf8_bytes": 858, "cache_tensor_bytes": null, "serialized_payload_bytes": 858, "side_information_bytes": 0, "fetched_bytes": 602, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 11284.805292030796, "hops": {"1": 6904.5422920025885, "2": 4380.263000028208}, "receiver_prefill_ms": 163.00804202910513, "ttft_ms": 11.888957931660116, "decode_ms": 11.774999904446304}, "total_wall_ms": 11472.005667048506, "ttft_ms": 11.888957931660116, "input_usage": {"lookups": 1}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 63, "facts_retained": 12, "retention_rate": 0.19047619047619047, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.25, "amount": 0.07692307692307693, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 818}, "ledger_by_hop": {"1": {"facts_total": 63, "facts_retained": 9, "retention_rate": 0.14285714285714285, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.25, "amount": 0.07692307692307693, "status": 0.25, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 787}, "2": {"facts_total": 63, "facts_retained": 12, "retention_rate": 0.19047619047619047, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.25, "amount": 0.07692307692307693, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 818}}}, "reference_key_id": "keys/synthetic/syn_03027_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "7ab33a7e50920be6dd2887eea3e354ca9b7ea53ad5cd929f38c58204c3e3af7e", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "3e771d59b7a17f49bb2df650", "attempt_id": "7f3d6a008293", "run_group_id": "E4_core:A5|repairable|qwen3b:h2", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:32:02.752438Z", "task_family": "synthetic", "document_id": "syn_03027_2000", "document_hash": "10326080979f71b625e4110c7c19767e5a752e267de5034f7e37d98cb6e06249", "question_id": "q08_revision", "split": "test", "cluster_id": "syn_03027_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 2, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1955, "query_tokens": 436, "message_tokens": 339, "message_utf8_bytes": 858, "cache_tensor_bytes": null, "serialized_payload_bytes": 858, "side_information_bytes": 0, "fetched_bytes": 602, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 11284.805292030796, "hops": {"1": 6904.5422920025885, "2": 4380.263000028208}, "receiver_prefill_ms": 163.1523750256747, "ttft_ms": 12.022958020679653, "decode_ms": 106.00737505592406}, "total_wall_ms": 11566.529459087178, "ttft_ms": 12.022958020679653, "input_usage": {"lookups": 1}, "output_usage": {"output_tokens": 9}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "6,914,937", "evidence_artifact": {"ledger": {"facts_total": 63, "facts_retained": 12, "retention_rate": 0.19047619047619047, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.25, "amount": 0.07692307692307693, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 818}, "ledger_by_hop": {"1": {"facts_total": 63, "facts_retained": 9, "retention_rate": 0.14285714285714285, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.25, "amount": 0.07692307692307693, "status": 0.25, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 787}, "2": {"facts_total": 63, "facts_retained": 12, "retention_rate": 0.19047619047619047, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.25, "amount": 0.07692307692307693, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 818}}}, "reference_key_id": "keys/synthetic/syn_03027_2000.json", "task_scores": {"correct": 0.0, "abstained": 0.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "7ab33a7e50920be6dd2887eea3e354ca9b7ea53ad5cd929f38c58204c3e3af7e", "output_hash": "c56851abfdbf8851226a1e71c4911c616a691c490a883eb754e61bac551d004c"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "cdd456e1214b18e16a30fac5", "attempt_id": "5e7a25bfad26", "run_group_id": "E4_core:A5|repairable|qwen3b:h2", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:32:03.050493Z", "task_family": "synthetic", "document_id": "syn_03027_2000", "document_hash": "10326080979f71b625e4110c7c19767e5a752e267de5034f7e37d98cb6e06249", "question_id": "q09_revision", "split": "test", "cluster_id": "syn_03027_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 2, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1955, "query_tokens": 429, "message_tokens": 339, "message_utf8_bytes": 858, "cache_tensor_bytes": null, "serialized_payload_bytes": 858, "side_information_bytes": 0, "fetched_bytes": 602, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 11284.805292030796, "hops": {"1": 6904.5422920025885, "2": 4380.263000028208}, "receiver_prefill_ms": 163.13912498299032, "ttft_ms": 12.226958060637116, "decode_ms": 94.86566693522036}, "total_wall_ms": 11555.596334044822, "ttft_ms": 12.226958060637116, "input_usage": {"lookups": 1}, "output_usage": {"output_tokens": 8}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "15828866", "evidence_artifact": {"ledger": {"facts_total": 63, "facts_retained": 12, "retention_rate": 0.19047619047619047, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.25, "amount": 0.07692307692307693, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 818}, "ledger_by_hop": {"1": {"facts_total": 63, "facts_retained": 9, "retention_rate": 0.14285714285714285, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.25, "amount": 0.07692307692307693, "status": 0.25, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 787}, "2": {"facts_total": 63, "facts_retained": 12, "retention_rate": 0.19047619047619047, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.25, "amount": 0.07692307692307693, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 818}}}, "reference_key_id": "keys/synthetic/syn_03027_2000.json", "task_scores": {"correct": 0.0, "abstained": 0.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "7ab33a7e50920be6dd2887eea3e354ca9b7ea53ad5cd929f38c58204c3e3af7e", "output_hash": "c9d49be6f876880a6dfe30589ff0b746c048ebaf1842118a883f5503a349a8b6"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "d5fa1ff6cab1fe04d163a98f", "attempt_id": "e3add4802be9", "run_group_id": "E4_core:A5|repairable|qwen3b:h2", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:32:03.338111Z", "task_family": "synthetic", "document_id": "syn_03027_2000", "document_hash": "10326080979f71b625e4110c7c19767e5a752e267de5034f7e37d98cb6e06249", "question_id": "q10_temporal", "split": "test", "cluster_id": "syn_03027_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 2, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1955, "query_tokens": 439, "message_tokens": 339, "message_utf8_bytes": 858, "cache_tensor_bytes": null, "serialized_payload_bytes": 858, "side_information_bytes": 0, "fetched_bytes": 602, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 11284.805292030796, "hops": {"1": 6904.5422920025885, "2": 4380.263000028208}, "receiver_prefill_ms": 163.31266600172967, "ttft_ms": 11.974875000305474, "decode_ms": 11.80491701234132}, "total_wall_ms": 11472.441207966767, "ttft_ms": 11.974875000305474, "input_usage": {"lookups": 1}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 63, "facts_retained": 12, "retention_rate": 0.19047619047619047, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.25, "amount": 0.07692307692307693, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 818}, "ledger_by_hop": {"1": {"facts_total": 63, "facts_retained": 9, "retention_rate": 0.14285714285714285, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.25, "amount": 0.07692307692307693, "status": 0.25, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 787}, "2": {"facts_total": 63, "facts_retained": 12, "retention_rate": 0.19047619047619047, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.25, "amount": 0.07692307692307693, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 818}}}, "reference_key_id": "keys/synthetic/syn_03027_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "7ab33a7e50920be6dd2887eea3e354ca9b7ea53ad5cd929f38c58204c3e3af7e", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "4886b86ef15b303367702cf0", "attempt_id": "ad0f77ee89ed", "run_group_id": "E4_core:A5|repairable|qwen3b:h2", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:32:03.542500Z", "task_family": "synthetic", "document_id": "syn_03027_2000", "document_hash": "10326080979f71b625e4110c7c19767e5a752e267de5034f7e37d98cb6e06249", "question_id": "q11_aggregation", "split": "test", "cluster_id": "syn_03027_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 2, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1955, "query_tokens": 432, "message_tokens": 339, "message_utf8_bytes": 858, "cache_tensor_bytes": null, "serialized_payload_bytes": 858, "side_information_bytes": 0, "fetched_bytes": 602, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 11284.805292030796, "hops": {"1": 6904.5422920025885, "2": 4380.263000028208}, "receiver_prefill_ms": 162.80037490651011, "ttft_ms": 12.014084029942751, "decode_ms": 82.52329204697162}, "total_wall_ms": 11542.69691708032, "ttft_ms": 12.014084029942751, "input_usage": {"lookups": 1}, "output_usage": {"output_tokens": 7}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "6914937", "evidence_artifact": {"ledger": {"facts_total": 63, "facts_retained": 12, "retention_rate": 0.19047619047619047, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.25, "amount": 0.07692307692307693, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 818}, "ledger_by_hop": {"1": {"facts_total": 63, "facts_retained": 9, "retention_rate": 0.14285714285714285, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.25, "amount": 0.07692307692307693, "status": 0.25, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 787}, "2": {"facts_total": 63, "facts_retained": 12, "retention_rate": 0.19047619047619047, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.25, "amount": 0.07692307692307693, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 818}}}, "reference_key_id": "keys/synthetic/syn_03027_2000.json", "task_scores": {"correct": 0.0, "abstained": 0.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "7ab33a7e50920be6dd2887eea3e354ca9b7ea53ad5cd929f38c58204c3e3af7e", "output_hash": "534d40ded32f784c68307dd2386faed2a27bf39dcc27321ee6985e42ccd285e5"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "8e825058dd772199c9e5477d", "attempt_id": "1bfcc1f9fe83", "run_group_id": "E4_core:A5|repairable|qwen3b:h2", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:32:03.816780Z", "task_family": "synthetic", "document_id": "syn_03027_2000", "document_hash": "10326080979f71b625e4110c7c19767e5a752e267de5034f7e37d98cb6e06249", "question_id": "q12_unit_conversion", "split": "test", "cluster_id": "syn_03027_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 2, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1955, "query_tokens": 429, "message_tokens": 339, "message_utf8_bytes": 858, "cache_tensor_bytes": null, "serialized_payload_bytes": 858, "side_information_bytes": 0, "fetched_bytes": 602, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 11284.805292030796, "hops": {"1": 6904.5422920025885, "2": 4380.263000028208}, "receiver_prefill_ms": 162.76987502351403, "ttft_ms": 11.977499932982028, "decode_ms": 11.774958111345768}, "total_wall_ms": 11471.862250007689, "ttft_ms": 11.977499932982028, "input_usage": {"lookups": 1}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 63, "facts_retained": 12, "retention_rate": 0.19047619047619047, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.25, "amount": 0.07692307692307693, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 818}, "ledger_by_hop": {"1": {"facts_total": 63, "facts_retained": 9, "retention_rate": 0.14285714285714285, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.25, "amount": 0.07692307692307693, "status": 0.25, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 787}, "2": {"facts_total": 63, "facts_retained": 12, "retention_rate": 0.19047619047619047, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.25, "amount": 0.07692307692307693, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 818}}}, "reference_key_id": "keys/synthetic/syn_03027_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "7ab33a7e50920be6dd2887eea3e354ca9b7ea53ad5cd929f38c58204c3e3af7e", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "de6ad348ea47793b0c2e5cbc", "attempt_id": "7cdfd5890c82", "run_group_id": "E4_core:A5|repairable|qwen3b:h2", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:32:04.020291Z", "task_family": "synthetic", "document_id": "syn_03027_2000", "document_hash": "10326080979f71b625e4110c7c19767e5a752e267de5034f7e37d98cb6e06249", "question_id": "q13_unanswerable", "split": "test", "cluster_id": "syn_03027_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 2, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1955, "query_tokens": 432, "message_tokens": 339, "message_utf8_bytes": 858, "cache_tensor_bytes": null, "serialized_payload_bytes": 858, "side_information_bytes": 0, "fetched_bytes": 602, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 11284.805292030796, "hops": {"1": 6904.5422920025885, "2": 4380.263000028208}, "receiver_prefill_ms": 162.82299999147654, "ttft_ms": 11.96270901709795, "decode_ms": 11.970332940109074}, "total_wall_ms": 11472.077584010549, "ttft_ms": 11.96270901709795, "input_usage": {"lookups": 1}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 63, "facts_retained": 12, "retention_rate": 0.19047619047619047, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.25, "amount": 0.07692307692307693, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 818}, "ledger_by_hop": {"1": {"facts_total": 63, "facts_retained": 9, "retention_rate": 0.14285714285714285, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.25, "amount": 0.07692307692307693, "status": 0.25, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 787}, "2": {"facts_total": 63, "facts_retained": 12, "retention_rate": 0.19047619047619047, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.25, "amount": 0.07692307692307693, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 818}}}, "reference_key_id": "keys/synthetic/syn_03027_2000.json", "task_scores": {"correct": 1.0, "abstained": 1.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "7ab33a7e50920be6dd2887eea3e354ca9b7ea53ad5cd929f38c58204c3e3af7e", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "4908d3841e4aa6f2b500f61c", "attempt_id": "8d77966a0093", "run_group_id": "E4_core:A5|repairable|qwen3b:h2", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:32:04.223968Z", "task_family": "synthetic", "document_id": "syn_03027_2000", "document_hash": "10326080979f71b625e4110c7c19767e5a752e267de5034f7e37d98cb6e06249", "question_id": "q14_unanswerable", "split": "test", "cluster_id": "syn_03027_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 2, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1955, "query_tokens": 434, "message_tokens": 339, "message_utf8_bytes": 858, "cache_tensor_bytes": null, "serialized_payload_bytes": 858, "side_information_bytes": 0, "fetched_bytes": 602, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 11284.805292030796, "hops": {"1": 6904.5422920025885, "2": 4380.263000028208}, "receiver_prefill_ms": 163.1097919307649, "ttft_ms": 11.991541017778218, "decode_ms": 11.808917042799294}, "total_wall_ms": 11472.263500094414, "ttft_ms": 11.991541017778218, "input_usage": {"lookups": 1}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 63, "facts_retained": 12, "retention_rate": 0.19047619047619047, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.25, "amount": 0.07692307692307693, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 818}, "ledger_by_hop": {"1": {"facts_total": 63, "facts_retained": 9, "retention_rate": 0.14285714285714285, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.25, "amount": 0.07692307692307693, "status": 0.25, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 787}, "2": {"facts_total": 63, "facts_retained": 12, "retention_rate": 0.19047619047619047, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.25, "amount": 0.07692307692307693, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 818}}}, "reference_key_id": "keys/synthetic/syn_03027_2000.json", "task_scores": {"correct": 1.0, "abstained": 1.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "7ab33a7e50920be6dd2887eea3e354ca9b7ea53ad5cd929f38c58204c3e3af7e", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "9c21f8bb1ddde17384c00141", "attempt_id": "fc6bc5a2e662", "run_group_id": "E4_core:A5|repairable|qwen3b:h4", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:32:13.176836Z", "task_family": "synthetic", "document_id": "syn_03027_2000", "document_hash": "10326080979f71b625e4110c7c19767e5a752e267de5034f7e37d98cb6e06249", "question_id": "q00_single_fact", "split": "test", "cluster_id": "syn_03027_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 4, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1955, "query_tokens": 420, "message_tokens": 339, "message_utf8_bytes": 858, "cache_tensor_bytes": null, "serialized_payload_bytes": 858, "side_information_bytes": 0, "fetched_bytes": 1806, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 20031.605917145498, "hops": {"1": 6904.5422920025885, "2": 4380.263000028208, "3": 4375.677625066601, "4": 4371.123000048101}, "receiver_prefill_ms": 162.25183405913413, "ttft_ms": 11.997958994470537, "decode_ms": 11.784625006839633}, "total_wall_ms": 20218.177042203024, "ttft_ms": 11.997958994470537, "input_usage": {"lookups": 3}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 63, "facts_retained": 12, "retention_rate": 0.19047619047619047, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.25, "amount": 0.07692307692307693, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 818}, "ledger_by_hop": {"1": {"facts_total": 63, "facts_retained": 9, "retention_rate": 0.14285714285714285, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.25, "amount": 0.07692307692307693, "status": 0.25, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 787}, "2": {"facts_total": 63, "facts_retained": 12, "retention_rate": 0.19047619047619047, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.25, "amount": 0.07692307692307693, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 818}, "3": {"facts_total": 63, "facts_retained": 12, "retention_rate": 0.19047619047619047, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.25, "amount": 0.07692307692307693, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 818}, "4": {"facts_total": 63, "facts_retained": 12, "retention_rate": 0.19047619047619047, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.25, "amount": 0.07692307692307693, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 818}}}, "reference_key_id": "keys/synthetic/syn_03027_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "7ab33a7e50920be6dd2887eea3e354ca9b7ea53ad5cd929f38c58204c3e3af7e", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "751f48bf49f0ad6f27c2923c", "attempt_id": "0d3c69cf0d1a", "run_group_id": "E4_core:A5|repairable|qwen3b:h4", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:32:13.380555Z", "task_family": "synthetic", "document_id": "syn_03027_2000", "document_hash": "10326080979f71b625e4110c7c19767e5a752e267de5034f7e37d98cb6e06249", "question_id": "q01_single_fact", "split": "test", "cluster_id": "syn_03027_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 4, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1955, "query_tokens": 427, "message_tokens": 339, "message_utf8_bytes": 858, "cache_tensor_bytes": null, "serialized_payload_bytes": 858, "side_information_bytes": 0, "fetched_bytes": 1806, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 20031.605917145498, "hops": {"1": 6904.5422920025885, "2": 4380.263000028208, "3": 4375.677625066601, "4": 4371.123000048101}, "receiver_prefill_ms": 162.82395797315985, "ttft_ms": 11.958041926845908, "decode_ms": 11.7972920415923}, "total_wall_ms": 20218.719209195115, "ttft_ms": 11.958041926845908, "input_usage": {"lookups": 3}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 63, "facts_retained": 12, "retention_rate": 0.19047619047619047, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.25, "amount": 0.07692307692307693, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 818}, "ledger_by_hop": {"1": {"facts_total": 63, "facts_retained": 9, "retention_rate": 0.14285714285714285, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.25, "amount": 0.07692307692307693, "status": 0.25, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 787}, "2": {"facts_total": 63, "facts_retained": 12, "retention_rate": 0.19047619047619047, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.25, "amount": 0.07692307692307693, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 818}, "3": {"facts_total": 63, "facts_retained": 12, "retention_rate": 0.19047619047619047, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.25, "amount": 0.07692307692307693, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 818}, "4": {"facts_total": 63, "facts_retained": 12, "retention_rate": 0.19047619047619047, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.25, "amount": 0.07692307692307693, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 818}}}, "reference_key_id": "keys/synthetic/syn_03027_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "7ab33a7e50920be6dd2887eea3e354ca9b7ea53ad5cd929f38c58204c3e3af7e", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "17756b5049a9ffcad090e3dc", "attempt_id": "2b4b553f47d8", "run_group_id": "E4_core:A5|repairable|qwen3b:h4", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:32:13.584240Z", "task_family": "synthetic", "document_id": "syn_03027_2000", "document_hash": "10326080979f71b625e4110c7c19767e5a752e267de5034f7e37d98cb6e06249", "question_id": "q02_exact_copy", "split": "test", "cluster_id": "syn_03027_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 4, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1955, "query_tokens": 426, "message_tokens": 339, "message_utf8_bytes": 858, "cache_tensor_bytes": null, "serialized_payload_bytes": 858, "side_information_bytes": 0, "fetched_bytes": 1806, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 20031.605917145498, "hops": {"1": 6904.5422920025885, "2": 4380.263000028208, "3": 4375.677625066601, "4": 4371.123000048101}, "receiver_prefill_ms": 162.5250829383731, "ttft_ms": 11.995665961876512, "decode_ms": 11.941708973608911}, "total_wall_ms": 20218.64683413878, "ttft_ms": 11.995665961876512, "input_usage": {"lookups": 3}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 63, "facts_retained": 12, "retention_rate": 0.19047619047619047, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.25, "amount": 0.07692307692307693, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 818}, "ledger_by_hop": {"1": {"facts_total": 63, "facts_retained": 9, "retention_rate": 0.14285714285714285, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.25, "amount": 0.07692307692307693, "status": 0.25, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 787}, "2": {"facts_total": 63, "facts_retained": 12, "retention_rate": 0.19047619047619047, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.25, "amount": 0.07692307692307693, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 818}, "3": {"facts_total": 63, "facts_retained": 12, "retention_rate": 0.19047619047619047, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.25, "amount": 0.07692307692307693, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 818}, "4": {"facts_total": 63, "facts_retained": 12, "retention_rate": 0.19047619047619047, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.25, "amount": 0.07692307692307693, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 818}}}, "reference_key_id": "keys/synthetic/syn_03027_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "7ab33a7e50920be6dd2887eea3e354ca9b7ea53ad5cd929f38c58204c3e3af7e", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "4fb0caa5a627939e76f5829e", "attempt_id": "c3f6dce83c68", "run_group_id": "E4_core:A5|repairable|qwen3b:h4", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:32:13.787718Z", "task_family": "synthetic", "document_id": "syn_03027_2000", "document_hash": "10326080979f71b625e4110c7c19767e5a752e267de5034f7e37d98cb6e06249", "question_id": "q03_exact_copy", "split": "test", "cluster_id": "syn_03027_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 4, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1955, "query_tokens": 421, "message_tokens": 339, "message_utf8_bytes": 858, "cache_tensor_bytes": null, "serialized_payload_bytes": 858, "side_information_bytes": 0, "fetched_bytes": 1806, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 20031.605917145498, "hops": {"1": 6904.5422920025885, "2": 4380.263000028208, "3": 4375.677625066601, "4": 4371.123000048101}, "receiver_prefill_ms": 162.5159999821335, "ttft_ms": 11.918375035747886, "decode_ms": 105.98783404566348}, "total_wall_ms": 20312.572292168625, "ttft_ms": 11.918375035747886, "input_usage": {"lookups": 3}, "output_usage": {"output_tokens": 9}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "Priyamvada Sørlie", "evidence_artifact": {"ledger": {"facts_total": 63, "facts_retained": 12, "retention_rate": 0.19047619047619047, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.25, "amount": 0.07692307692307693, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 818}, "ledger_by_hop": {"1": {"facts_total": 63, "facts_retained": 9, "retention_rate": 0.14285714285714285, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.25, "amount": 0.07692307692307693, "status": 0.25, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 787}, "2": {"facts_total": 63, "facts_retained": 12, "retention_rate": 0.19047619047619047, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.25, "amount": 0.07692307692307693, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 818}, "3": {"facts_total": 63, "facts_retained": 12, "retention_rate": 0.19047619047619047, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.25, "amount": 0.07692307692307693, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 818}, "4": {"facts_total": 63, "facts_retained": 12, "retention_rate": 0.19047619047619047, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.25, "amount": 0.07692307692307693, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 818}}}, "reference_key_id": "keys/synthetic/syn_03027_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "7ab33a7e50920be6dd2887eea3e354ca9b7ea53ad5cd929f38c58204c3e3af7e", "output_hash": "c5f554196ed23dd71ad995ec89f7142099d524dbb6534cc10d94a0c89d13e792"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "5d8d5b194d65cc28b41287ea", "attempt_id": "49fb1f6cdb05", "run_group_id": "E4_core:A5|repairable|qwen3b:h4", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:32:14.084874Z", "task_family": "synthetic", "document_id": "syn_03027_2000", "document_hash": "10326080979f71b625e4110c7c19767e5a752e267de5034f7e37d98cb6e06249", "question_id": "q04_multi_hop", "split": "test", "cluster_id": "syn_03027_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 4, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1955, "query_tokens": 432, "message_tokens": 339, "message_utf8_bytes": 858, "cache_tensor_bytes": null, "serialized_payload_bytes": 858, "side_information_bytes": 0, "fetched_bytes": 1806, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 20031.605917145498, "hops": {"1": 6904.5422920025885, "2": 4380.263000028208, "3": 4375.677625066601, "4": 4371.123000048101}, "receiver_prefill_ms": 162.71654202137142, "ttft_ms": 11.935874936170876, "decode_ms": 11.876665987074375}, "total_wall_ms": 20218.70729222428, "ttft_ms": 11.935874936170876, "input_usage": {"lookups": 3}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 63, "facts_retained": 12, "retention_rate": 0.19047619047619047, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.25, "amount": 0.07692307692307693, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 818}, "ledger_by_hop": {"1": {"facts_total": 63, "facts_retained": 9, "retention_rate": 0.14285714285714285, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.25, "amount": 0.07692307692307693, "status": 0.25, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 787}, "2": {"facts_total": 63, "facts_retained": 12, "retention_rate": 0.19047619047619047, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.25, "amount": 0.07692307692307693, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 818}, "3": {"facts_total": 63, "facts_retained": 12, "retention_rate": 0.19047619047619047, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.25, "amount": 0.07692307692307693, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 818}, "4": {"facts_total": 63, "facts_retained": 12, "retention_rate": 0.19047619047619047, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.25, "amount": 0.07692307692307693, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 818}}}, "reference_key_id": "keys/synthetic/syn_03027_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "7ab33a7e50920be6dd2887eea3e354ca9b7ea53ad5cd929f38c58204c3e3af7e", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "e9b2d46179221737e4c2d757", "attempt_id": "143eb2af9a23", "run_group_id": "E4_core:A5|repairable|qwen3b:h4", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:32:14.288549Z", "task_family": "synthetic", "document_id": "syn_03027_2000", "document_hash": "10326080979f71b625e4110c7c19767e5a752e267de5034f7e37d98cb6e06249", "question_id": "q05_multi_hop", "split": "test", "cluster_id": "syn_03027_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 4, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1955, "query_tokens": 429, "message_tokens": 339, "message_utf8_bytes": 858, "cache_tensor_bytes": null, "serialized_payload_bytes": 858, "side_information_bytes": 0, "fetched_bytes": 1806, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 20031.605917145498, "hops": {"1": 6904.5422920025885, "2": 4380.263000028208, "3": 4375.677625066601, "4": 4371.123000048101}, "receiver_prefill_ms": 163.1305000046268, "ttft_ms": 12.034083949401975, "decode_ms": 117.69745801575482}, "total_wall_ms": 20325.0216251472, "ttft_ms": 12.034083949401975, "input_usage": {"lookups": 3}, "output_usage": {"output_tokens": 10}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "WT-26M-9227", "evidence_artifact": {"ledger": {"facts_total": 63, "facts_retained": 12, "retention_rate": 0.19047619047619047, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.25, "amount": 0.07692307692307693, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 818}, "ledger_by_hop": {"1": {"facts_total": 63, "facts_retained": 9, "retention_rate": 0.14285714285714285, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.25, "amount": 0.07692307692307693, "status": 0.25, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 787}, "2": {"facts_total": 63, "facts_retained": 12, "retention_rate": 0.19047619047619047, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.25, "amount": 0.07692307692307693, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 818}, "3": {"facts_total": 63, "facts_retained": 12, "retention_rate": 0.19047619047619047, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.25, "amount": 0.07692307692307693, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 818}, "4": {"facts_total": 63, "facts_retained": 12, "retention_rate": 0.19047619047619047, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.25, "amount": 0.07692307692307693, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 818}}}, "reference_key_id": "keys/synthetic/syn_03027_2000.json", "task_scores": {"correct": 0.0, "abstained": 0.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "7ab33a7e50920be6dd2887eea3e354ca9b7ea53ad5cd929f38c58204c3e3af7e", "output_hash": "ef2302e0ab6f5a6951b11463fea177ba766b9d438afcaf6e6101ebfa3ba38aba"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "f5048c85a91773bb28f02db4", "attempt_id": "c73e1c7bda90", "run_group_id": "E4_core:A5|repairable|qwen3b:h4", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:32:14.599141Z", "task_family": "synthetic", "document_id": "syn_03027_2000", "document_hash": "10326080979f71b625e4110c7c19767e5a752e267de5034f7e37d98cb6e06249", "question_id": "q06_negation", "split": "test", "cluster_id": "syn_03027_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 4, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1955, "query_tokens": 421, "message_tokens": 339, "message_utf8_bytes": 858, "cache_tensor_bytes": null, "serialized_payload_bytes": 858, "side_information_bytes": 0, "fetched_bytes": 1806, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 20031.605917145498, "hops": {"1": 6904.5422920025885, "2": 4380.263000028208, "3": 4375.677625066601, "4": 4371.123000048101}, "receiver_prefill_ms": 162.68974996637553, "ttft_ms": 11.981167015619576, "decode_ms": 11.93391706328839}, "total_wall_ms": 20218.790667131543, "ttft_ms": 11.981167015619576, "input_usage": {"lookups": 3}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 63, "facts_retained": 12, "retention_rate": 0.19047619047619047, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.25, "amount": 0.07692307692307693, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 818}, "ledger_by_hop": {"1": {"facts_total": 63, "facts_retained": 9, "retention_rate": 0.14285714285714285, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.25, "amount": 0.07692307692307693, "status": 0.25, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 787}, "2": {"facts_total": 63, "facts_retained": 12, "retention_rate": 0.19047619047619047, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.25, "amount": 0.07692307692307693, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 818}, "3": {"facts_total": 63, "facts_retained": 12, "retention_rate": 0.19047619047619047, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.25, "amount": 0.07692307692307693, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 818}, "4": {"facts_total": 63, "facts_retained": 12, "retention_rate": 0.19047619047619047, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.25, "amount": 0.07692307692307693, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 818}}}, "reference_key_id": "keys/synthetic/syn_03027_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "7ab33a7e50920be6dd2887eea3e354ca9b7ea53ad5cd929f38c58204c3e3af7e", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "f8a1cf03bd27bfbb71932f5e", "attempt_id": "019f9ca3f15b", "run_group_id": "E4_core:A5|repairable|qwen3b:h4", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:32:14.802881Z", "task_family": "synthetic", "document_id": "syn_03027_2000", "document_hash": "10326080979f71b625e4110c7c19767e5a752e267de5034f7e37d98cb6e06249", "question_id": "q07_negation", "split": "test", "cluster_id": "syn_03027_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 4, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1955, "query_tokens": 422, "message_tokens": 339, "message_utf8_bytes": 858, "cache_tensor_bytes": null, "serialized_payload_bytes": 858, "side_information_bytes": 0, "fetched_bytes": 1806, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 20031.605917145498, "hops": {"1": 6904.5422920025885, "2": 4380.263000028208, "3": 4375.677625066601, "4": 4371.123000048101}, "receiver_prefill_ms": 162.5451250001788, "ttft_ms": 11.957417009398341, "decode_ms": 11.788124917075038}, "total_wall_ms": 20218.4524590848, "ttft_ms": 11.957417009398341, "input_usage": {"lookups": 3}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 63, "facts_retained": 12, "retention_rate": 0.19047619047619047, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.25, "amount": 0.07692307692307693, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 818}, "ledger_by_hop": {"1": {"facts_total": 63, "facts_retained": 9, "retention_rate": 0.14285714285714285, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.25, "amount": 0.07692307692307693, "status": 0.25, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 787}, "2": {"facts_total": 63, "facts_retained": 12, "retention_rate": 0.19047619047619047, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.25, "amount": 0.07692307692307693, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 818}, "3": {"facts_total": 63, "facts_retained": 12, "retention_rate": 0.19047619047619047, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.25, "amount": 0.07692307692307693, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 818}, "4": {"facts_total": 63, "facts_retained": 12, "retention_rate": 0.19047619047619047, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.25, "amount": 0.07692307692307693, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 818}}}, "reference_key_id": "keys/synthetic/syn_03027_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "7ab33a7e50920be6dd2887eea3e354ca9b7ea53ad5cd929f38c58204c3e3af7e", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "c000a339087b6ea282931afe", "attempt_id": "27a4a925085e", "run_group_id": "E4_core:A5|repairable|qwen3b:h4", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:32:15.006497Z", "task_family": "synthetic", "document_id": "syn_03027_2000", "document_hash": "10326080979f71b625e4110c7c19767e5a752e267de5034f7e37d98cb6e06249", "question_id": "q08_revision", "split": "test", "cluster_id": "syn_03027_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 4, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1955, "query_tokens": 436, "message_tokens": 339, "message_utf8_bytes": 858, "cache_tensor_bytes": null, "serialized_payload_bytes": 858, "side_information_bytes": 0, "fetched_bytes": 1806, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 20031.605917145498, "hops": {"1": 6904.5422920025885, "2": 4380.263000028208, "3": 4375.677625066601, "4": 4371.123000048101}, "receiver_prefill_ms": 163.19950006436557, "ttft_ms": 12.186832958832383, "decode_ms": 105.9744170634076}, "total_wall_ms": 20313.508458202705, "ttft_ms": 12.186832958832383, "input_usage": {"lookups": 3}, "output_usage": {"output_tokens": 9}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "6,914,937", "evidence_artifact": {"ledger": {"facts_total": 63, "facts_retained": 12, "retention_rate": 0.19047619047619047, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.25, "amount": 0.07692307692307693, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 818}, "ledger_by_hop": {"1": {"facts_total": 63, "facts_retained": 9, "retention_rate": 0.14285714285714285, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.25, "amount": 0.07692307692307693, "status": 0.25, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 787}, "2": {"facts_total": 63, "facts_retained": 12, "retention_rate": 0.19047619047619047, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.25, "amount": 0.07692307692307693, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 818}, "3": {"facts_total": 63, "facts_retained": 12, "retention_rate": 0.19047619047619047, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.25, "amount": 0.07692307692307693, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 818}, "4": {"facts_total": 63, "facts_retained": 12, "retention_rate": 0.19047619047619047, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.25, "amount": 0.07692307692307693, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 818}}}, "reference_key_id": "keys/synthetic/syn_03027_2000.json", "task_scores": {"correct": 0.0, "abstained": 0.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "7ab33a7e50920be6dd2887eea3e354ca9b7ea53ad5cd929f38c58204c3e3af7e", "output_hash": "c56851abfdbf8851226a1e71c4911c616a691c490a883eb754e61bac551d004c"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "2d448014a3e584e8cfd601c7", "attempt_id": "af2cbca5a0f9", "run_group_id": "E4_core:A5|repairable|qwen3b:h4", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:32:15.304998Z", "task_family": "synthetic", "document_id": "syn_03027_2000", "document_hash": "10326080979f71b625e4110c7c19767e5a752e267de5034f7e37d98cb6e06249", "question_id": "q09_revision", "split": "test", "cluster_id": "syn_03027_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 4, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1955, "query_tokens": 429, "message_tokens": 339, "message_utf8_bytes": 858, "cache_tensor_bytes": null, "serialized_payload_bytes": 858, "side_information_bytes": 0, "fetched_bytes": 1806, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 20031.605917145498, "hops": {"1": 6904.5422920025885, "2": 4380.263000028208, "3": 4375.677625066601, "4": 4371.123000048101}, "receiver_prefill_ms": 162.2902089729905, "ttft_ms": 11.997041990980506, "decode_ms": 94.16520793456584}, "total_wall_ms": 20300.61170924455, "ttft_ms": 11.997041990980506, "input_usage": {"lookups": 3}, "output_usage": {"output_tokens": 8}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "15828865", "evidence_artifact": {"ledger": {"facts_total": 63, "facts_retained": 12, "retention_rate": 0.19047619047619047, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.25, "amount": 0.07692307692307693, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 818}, "ledger_by_hop": {"1": {"facts_total": 63, "facts_retained": 9, "retention_rate": 0.14285714285714285, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.25, "amount": 0.07692307692307693, "status": 0.25, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 787}, "2": {"facts_total": 63, "facts_retained": 12, "retention_rate": 0.19047619047619047, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.25, "amount": 0.07692307692307693, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 818}, "3": {"facts_total": 63, "facts_retained": 12, "retention_rate": 0.19047619047619047, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.25, "amount": 0.07692307692307693, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 818}, "4": {"facts_total": 63, "facts_retained": 12, "retention_rate": 0.19047619047619047, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.25, "amount": 0.07692307692307693, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 818}}}, "reference_key_id": "keys/synthetic/syn_03027_2000.json", "task_scores": {"correct": 0.0, "abstained": 0.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "7ab33a7e50920be6dd2887eea3e354ca9b7ea53ad5cd929f38c58204c3e3af7e", "output_hash": "b38715dfc60d964ebea3bdcc90d47ca28d803114815818233d093f08214df6ae"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "7746e3e1fe1d98dd566c83cc", "attempt_id": "d7eb50fd6d38", "run_group_id": "E4_core:A5|repairable|qwen3b:h4", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:32:15.591636Z", "task_family": "synthetic", "document_id": "syn_03027_2000", "document_hash": "10326080979f71b625e4110c7c19767e5a752e267de5034f7e37d98cb6e06249", "question_id": "q10_temporal", "split": "test", "cluster_id": "syn_03027_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 4, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1955, "query_tokens": 439, "message_tokens": 339, "message_utf8_bytes": 858, "cache_tensor_bytes": null, "serialized_payload_bytes": 858, "side_information_bytes": 0, "fetched_bytes": 1806, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 20031.605917145498, "hops": {"1": 6904.5422920025885, "2": 4380.263000028208, "3": 4375.677625066601, "4": 4371.123000048101}, "receiver_prefill_ms": 163.38491695933044, "ttft_ms": 11.889249901287258, "decode_ms": 11.825541034340858}, "total_wall_ms": 20219.235292170197, "ttft_ms": 11.889249901287258, "input_usage": {"lookups": 3}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 63, "facts_retained": 12, "retention_rate": 0.19047619047619047, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.25, "amount": 0.07692307692307693, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 818}, "ledger_by_hop": {"1": {"facts_total": 63, "facts_retained": 9, "retention_rate": 0.14285714285714285, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.25, "amount": 0.07692307692307693, "status": 0.25, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 787}, "2": {"facts_total": 63, "facts_retained": 12, "retention_rate": 0.19047619047619047, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.25, "amount": 0.07692307692307693, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 818}, "3": {"facts_total": 63, "facts_retained": 12, "retention_rate": 0.19047619047619047, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.25, "amount": 0.07692307692307693, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 818}, "4": {"facts_total": 63, "facts_retained": 12, "retention_rate": 0.19047619047619047, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.25, "amount": 0.07692307692307693, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 818}}}, "reference_key_id": "keys/synthetic/syn_03027_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "7ab33a7e50920be6dd2887eea3e354ca9b7ea53ad5cd929f38c58204c3e3af7e", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "d34ecd59b92be4ad062bcd17", "attempt_id": "ad42a4f2d6bf", "run_group_id": "E4_core:A5|repairable|qwen3b:h4", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:32:15.795807Z", "task_family": "synthetic", "document_id": "syn_03027_2000", "document_hash": "10326080979f71b625e4110c7c19767e5a752e267de5034f7e37d98cb6e06249", "question_id": "q11_aggregation", "split": "test", "cluster_id": "syn_03027_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 4, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1955, "query_tokens": 432, "message_tokens": 339, "message_utf8_bytes": 858, "cache_tensor_bytes": null, "serialized_payload_bytes": 858, "side_information_bytes": 0, "fetched_bytes": 1806, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 20031.605917145498, "hops": {"1": 6904.5422920025885, "2": 4380.263000028208, "3": 4375.677625066601, "4": 4371.123000048101}, "receiver_prefill_ms": 163.17116608843207, "ttft_ms": 12.047125026583672, "decode_ms": 82.41575001738966}, "total_wall_ms": 20289.788750116713, "ttft_ms": 12.047125026583672, "input_usage": {"lookups": 3}, "output_usage": {"output_tokens": 7}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "6914937", "evidence_artifact": {"ledger": {"facts_total": 63, "facts_retained": 12, "retention_rate": 0.19047619047619047, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.25, "amount": 0.07692307692307693, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 818}, "ledger_by_hop": {"1": {"facts_total": 63, "facts_retained": 9, "retention_rate": 0.14285714285714285, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.25, "amount": 0.07692307692307693, "status": 0.25, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 787}, "2": {"facts_total": 63, "facts_retained": 12, "retention_rate": 0.19047619047619047, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.25, "amount": 0.07692307692307693, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 818}, "3": {"facts_total": 63, "facts_retained": 12, "retention_rate": 0.19047619047619047, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.25, "amount": 0.07692307692307693, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 818}, "4": {"facts_total": 63, "facts_retained": 12, "retention_rate": 0.19047619047619047, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.25, "amount": 0.07692307692307693, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 818}}}, "reference_key_id": "keys/synthetic/syn_03027_2000.json", "task_scores": {"correct": 0.0, "abstained": 0.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "7ab33a7e50920be6dd2887eea3e354ca9b7ea53ad5cd929f38c58204c3e3af7e", "output_hash": "534d40ded32f784c68307dd2386faed2a27bf39dcc27321ee6985e42ccd285e5"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "0d09153a036bf6cc557aec7b", "attempt_id": "55c108237e98", "run_group_id": "E4_core:A5|repairable|qwen3b:h4", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:32:16.071138Z", "task_family": "synthetic", "document_id": "syn_03027_2000", "document_hash": "10326080979f71b625e4110c7c19767e5a752e267de5034f7e37d98cb6e06249", "question_id": "q12_unit_conversion", "split": "test", "cluster_id": "syn_03027_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 4, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1955, "query_tokens": 429, "message_tokens": 339, "message_utf8_bytes": 858, "cache_tensor_bytes": null, "serialized_payload_bytes": 858, "side_information_bytes": 0, "fetched_bytes": 1806, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 20031.605917145498, "hops": {"1": 6904.5422920025885, "2": 4380.263000028208, "3": 4375.677625066601, "4": 4371.123000048101}, "receiver_prefill_ms": 162.26541600190103, "ttft_ms": 11.967791011556983, "decode_ms": 11.867208988405764}, "total_wall_ms": 20218.269708100706, "ttft_ms": 11.967791011556983, "input_usage": {"lookups": 3}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 63, "facts_retained": 12, "retention_rate": 0.19047619047619047, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.25, "amount": 0.07692307692307693, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 818}, "ledger_by_hop": {"1": {"facts_total": 63, "facts_retained": 9, "retention_rate": 0.14285714285714285, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.25, "amount": 0.07692307692307693, "status": 0.25, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 787}, "2": {"facts_total": 63, "facts_retained": 12, "retention_rate": 0.19047619047619047, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.25, "amount": 0.07692307692307693, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 818}, "3": {"facts_total": 63, "facts_retained": 12, "retention_rate": 0.19047619047619047, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.25, "amount": 0.07692307692307693, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 818}, "4": {"facts_total": 63, "facts_retained": 12, "retention_rate": 0.19047619047619047, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.25, "amount": 0.07692307692307693, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 818}}}, "reference_key_id": "keys/synthetic/syn_03027_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "7ab33a7e50920be6dd2887eea3e354ca9b7ea53ad5cd929f38c58204c3e3af7e", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "e7df7cfd0881c4bd8729c5fa", "attempt_id": "f89a2eb1f0b5", "run_group_id": "E4_core:A5|repairable|qwen3b:h4", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:32:16.274252Z", "task_family": "synthetic", "document_id": "syn_03027_2000", "document_hash": "10326080979f71b625e4110c7c19767e5a752e267de5034f7e37d98cb6e06249", "question_id": "q13_unanswerable", "split": "test", "cluster_id": "syn_03027_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 4, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1955, "query_tokens": 432, "message_tokens": 339, "message_utf8_bytes": 858, "cache_tensor_bytes": null, "serialized_payload_bytes": 858, "side_information_bytes": 0, "fetched_bytes": 1806, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 20031.605917145498, "hops": {"1": 6904.5422920025885, "2": 4380.263000028208, "3": 4375.677625066601, "4": 4371.123000048101}, "receiver_prefill_ms": 162.85170800983906, "ttft_ms": 11.99487503618002, "decode_ms": 11.803624918684363}, "total_wall_ms": 20218.79883413203, "ttft_ms": 11.99487503618002, "input_usage": {"lookups": 3}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 63, "facts_retained": 12, "retention_rate": 0.19047619047619047, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.25, "amount": 0.07692307692307693, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 818}, "ledger_by_hop": {"1": {"facts_total": 63, "facts_retained": 9, "retention_rate": 0.14285714285714285, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.25, "amount": 0.07692307692307693, "status": 0.25, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 787}, "2": {"facts_total": 63, "facts_retained": 12, "retention_rate": 0.19047619047619047, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.25, "amount": 0.07692307692307693, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 818}, "3": {"facts_total": 63, "facts_retained": 12, "retention_rate": 0.19047619047619047, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.25, "amount": 0.07692307692307693, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 818}, "4": {"facts_total": 63, "facts_retained": 12, "retention_rate": 0.19047619047619047, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.25, "amount": 0.07692307692307693, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 818}}}, "reference_key_id": "keys/synthetic/syn_03027_2000.json", "task_scores": {"correct": 1.0, "abstained": 1.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "7ab33a7e50920be6dd2887eea3e354ca9b7ea53ad5cd929f38c58204c3e3af7e", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "71464fbbcc1692cd9e7f4cce", "attempt_id": "6cbe67c15efd", "run_group_id": "E4_core:A5|repairable|qwen3b:h4", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:32:16.478275Z", "task_family": "synthetic", "document_id": "syn_03027_2000", "document_hash": "10326080979f71b625e4110c7c19767e5a752e267de5034f7e37d98cb6e06249", "question_id": "q14_unanswerable", "split": "test", "cluster_id": "syn_03027_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 4, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1955, "query_tokens": 434, "message_tokens": 339, "message_utf8_bytes": 858, "cache_tensor_bytes": null, "serialized_payload_bytes": 858, "side_information_bytes": 0, "fetched_bytes": 1806, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 20031.605917145498, "hops": {"1": 6904.5422920025885, "2": 4380.263000028208, "3": 4375.677625066601, "4": 4371.123000048101}, "receiver_prefill_ms": 163.19333296269178, "ttft_ms": 11.967458995059133, "decode_ms": 11.754250037483871}, "total_wall_ms": 20219.068792182952, "ttft_ms": 11.967458995059133, "input_usage": {"lookups": 3}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 63, "facts_retained": 12, "retention_rate": 0.19047619047619047, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.25, "amount": 0.07692307692307693, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 818}, "ledger_by_hop": {"1": {"facts_total": 63, "facts_retained": 9, "retention_rate": 0.14285714285714285, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.25, "amount": 0.07692307692307693, "status": 0.25, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 787}, "2": {"facts_total": 63, "facts_retained": 12, "retention_rate": 0.19047619047619047, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.25, "amount": 0.07692307692307693, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 818}, "3": {"facts_total": 63, "facts_retained": 12, "retention_rate": 0.19047619047619047, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.25, "amount": 0.07692307692307693, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 818}, "4": {"facts_total": 63, "facts_retained": 12, "retention_rate": 0.19047619047619047, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.25, "amount": 0.07692307692307693, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 818}}}, "reference_key_id": "keys/synthetic/syn_03027_2000.json", "task_scores": {"correct": 1.0, "abstained": 1.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "7ab33a7e50920be6dd2887eea3e354ca9b7ea53ad5cd929f38c58204c3e3af7e", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "b4afafb68040b44b9c1fc82c", "attempt_id": "5d6eb1e3f4eb", "run_group_id": "E4_core:A5|repairable|qwen3b:h8", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:32:34.206915Z", "task_family": "synthetic", "document_id": "syn_03027_2000", "document_hash": "10326080979f71b625e4110c7c19767e5a752e267de5034f7e37d98cb6e06249", "question_id": "q00_single_fact", "split": "test", "cluster_id": "syn_03027_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 8, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1955, "query_tokens": 420, "message_tokens": 339, "message_utf8_bytes": 858, "cache_tensor_bytes": null, "serialized_payload_bytes": 858, "side_information_bytes": 0, "fetched_bytes": 4214, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 37551.62508401554, "hops": {"1": 6904.5422920025885, "2": 4380.263000028208, "3": 4375.677625066601, "4": 4371.123000048101, "5": 4391.168999951333, "6": 4377.536708954722, "7": 4374.012624961324, "8": 4377.300833002664}, "receiver_prefill_ms": 162.8488750429824, "ttft_ms": 11.996084009297192, "decode_ms": 11.838959064334631}, "total_wall_ms": 37738.89483406674, "ttft_ms": 11.996084009297192, "input_usage": {"lookups": 7}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 63, "facts_retained": 12, "retention_rate": 0.19047619047619047, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.25, "amount": 0.07692307692307693, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 818}, "ledger_by_hop": {"1": {"facts_total": 63, "facts_retained": 9, "retention_rate": 0.14285714285714285, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.25, "amount": 0.07692307692307693, "status": 0.25, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 787}, "2": {"facts_total": 63, "facts_retained": 12, "retention_rate": 0.19047619047619047, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.25, "amount": 0.07692307692307693, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 818}, "3": {"facts_total": 63, "facts_retained": 12, "retention_rate": 0.19047619047619047, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.25, "amount": 0.07692307692307693, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 818}, "4": {"facts_total": 63, "facts_retained": 12, "retention_rate": 0.19047619047619047, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.25, "amount": 0.07692307692307693, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 818}, "5": {"facts_total": 63, "facts_retained": 12, "retention_rate": 0.19047619047619047, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.25, "amount": 0.07692307692307693, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 818}, "6": {"facts_total": 63, "facts_retained": 12, "retention_rate": 0.19047619047619047, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.25, "amount": 0.07692307692307693, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 818}, "7": {"facts_total": 63, "facts_retained": 12, "retention_rate": 0.19047619047619047, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.25, "amount": 0.07692307692307693, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 818}, "8": {"facts_total": 63, "facts_retained": 12, "retention_rate": 0.19047619047619047, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.25, "amount": 0.07692307692307693, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 818}}}, "reference_key_id": "keys/synthetic/syn_03027_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "7ab33a7e50920be6dd2887eea3e354ca9b7ea53ad5cd929f38c58204c3e3af7e", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "d6f948964e92b522b70f6569", "attempt_id": "23ad64ba0e56", "run_group_id": "E4_core:A5|repairable|qwen3b:h8", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:32:34.411332Z", "task_family": "synthetic", "document_id": "syn_03027_2000", "document_hash": "10326080979f71b625e4110c7c19767e5a752e267de5034f7e37d98cb6e06249", "question_id": "q01_single_fact", "split": "test", "cluster_id": "syn_03027_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 8, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1955, "query_tokens": 427, "message_tokens": 339, "message_utf8_bytes": 858, "cache_tensor_bytes": null, "serialized_payload_bytes": 858, "side_information_bytes": 0, "fetched_bytes": 4214, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 37551.62508401554, "hops": {"1": 6904.5422920025885, "2": 4380.263000028208, "3": 4375.677625066601, "4": 4371.123000048101, "5": 4391.168999951333, "6": 4377.536708954722, "7": 4374.012624961324, "8": 4377.300833002664}, "receiver_prefill_ms": 162.86229202523828, "ttft_ms": 11.990666971541941, "decode_ms": 11.968291946686804}, "total_wall_ms": 37738.99479303509, "ttft_ms": 11.990666971541941, "input_usage": {"lookups": 7}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 63, "facts_retained": 12, "retention_rate": 0.19047619047619047, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.25, "amount": 0.07692307692307693, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 818}, "ledger_by_hop": {"1": {"facts_total": 63, "facts_retained": 9, "retention_rate": 0.14285714285714285, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.25, "amount": 0.07692307692307693, "status": 0.25, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 787}, "2": {"facts_total": 63, "facts_retained": 12, "retention_rate": 0.19047619047619047, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.25, "amount": 0.07692307692307693, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 818}, "3": {"facts_total": 63, "facts_retained": 12, "retention_rate": 0.19047619047619047, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.25, "amount": 0.07692307692307693, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 818}, "4": {"facts_total": 63, "facts_retained": 12, "retention_rate": 0.19047619047619047, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.25, "amount": 0.07692307692307693, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 818}, "5": {"facts_total": 63, "facts_retained": 12, "retention_rate": 0.19047619047619047, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.25, "amount": 0.07692307692307693, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 818}, "6": {"facts_total": 63, "facts_retained": 12, "retention_rate": 0.19047619047619047, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.25, "amount": 0.07692307692307693, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 818}, "7": {"facts_total": 63, "facts_retained": 12, "retention_rate": 0.19047619047619047, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.25, "amount": 0.07692307692307693, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 818}, "8": {"facts_total": 63, "facts_retained": 12, "retention_rate": 0.19047619047619047, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.25, "amount": 0.07692307692307693, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 818}}}, "reference_key_id": "keys/synthetic/syn_03027_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "7ab33a7e50920be6dd2887eea3e354ca9b7ea53ad5cd929f38c58204c3e3af7e", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "c00aac3fe9fd50be18cae024", "attempt_id": "8a4b5312840e", "run_group_id": "E4_core:A5|repairable|qwen3b:h8", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:32:34.615554Z", "task_family": "synthetic", "document_id": "syn_03027_2000", "document_hash": "10326080979f71b625e4110c7c19767e5a752e267de5034f7e37d98cb6e06249", "question_id": "q02_exact_copy", "split": "test", "cluster_id": "syn_03027_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 8, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1955, "query_tokens": 426, "message_tokens": 339, "message_utf8_bytes": 858, "cache_tensor_bytes": null, "serialized_payload_bytes": 858, "side_information_bytes": 0, "fetched_bytes": 4214, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 37551.62508401554, "hops": {"1": 6904.5422920025885, "2": 4380.263000028208, "3": 4375.677625066601, "4": 4371.123000048101, "5": 4391.168999951333, "6": 4377.536708954722, "7": 4374.012624961324, "8": 4377.300833002664}, "receiver_prefill_ms": 162.7461250172928, "ttft_ms": 12.085292022675276, "decode_ms": 11.794500052928925}, "total_wall_ms": 37738.796918070875, "ttft_ms": 12.085292022675276, "input_usage": {"lookups": 7}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 63, "facts_retained": 12, "retention_rate": 0.19047619047619047, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.25, "amount": 0.07692307692307693, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 818}, "ledger_by_hop": {"1": {"facts_total": 63, "facts_retained": 9, "retention_rate": 0.14285714285714285, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.25, "amount": 0.07692307692307693, "status": 0.25, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 787}, "2": {"facts_total": 63, "facts_retained": 12, "retention_rate": 0.19047619047619047, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.25, "amount": 0.07692307692307693, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 818}, "3": {"facts_total": 63, "facts_retained": 12, "retention_rate": 0.19047619047619047, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.25, "amount": 0.07692307692307693, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 818}, "4": {"facts_total": 63, "facts_retained": 12, "retention_rate": 0.19047619047619047, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.25, "amount": 0.07692307692307693, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 818}, "5": {"facts_total": 63, "facts_retained": 12, "retention_rate": 0.19047619047619047, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.25, "amount": 0.07692307692307693, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 818}, "6": {"facts_total": 63, "facts_retained": 12, "retention_rate": 0.19047619047619047, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.25, "amount": 0.07692307692307693, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 818}, "7": {"facts_total": 63, "facts_retained": 12, "retention_rate": 0.19047619047619047, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.25, "amount": 0.07692307692307693, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 818}, "8": {"facts_total": 63, "facts_retained": 12, "retention_rate": 0.19047619047619047, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.25, "amount": 0.07692307692307693, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 818}}}, "reference_key_id": "keys/synthetic/syn_03027_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "7ab33a7e50920be6dd2887eea3e354ca9b7ea53ad5cd929f38c58204c3e3af7e", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "ac28b052817fd71b7c34046d", "attempt_id": "6853feec4d2e", "run_group_id": "E4_core:A5|repairable|qwen3b:h8", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:32:34.819212Z", "task_family": "synthetic", "document_id": "syn_03027_2000", "document_hash": "10326080979f71b625e4110c7c19767e5a752e267de5034f7e37d98cb6e06249", "question_id": "q03_exact_copy", "split": "test", "cluster_id": "syn_03027_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 8, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1955, "query_tokens": 421, "message_tokens": 339, "message_utf8_bytes": 858, "cache_tensor_bytes": null, "serialized_payload_bytes": 858, "side_information_bytes": 0, "fetched_bytes": 4214, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 37551.62508401554, "hops": {"1": 6904.5422920025885, "2": 4380.263000028208, "3": 4375.677625066601, "4": 4371.123000048101, "5": 4391.168999951333, "6": 4377.536708954722, "7": 4374.012624961324, "8": 4377.300833002664}, "receiver_prefill_ms": 162.26975002791733, "ttft_ms": 11.933750007301569, "decode_ms": 105.8265829924494}, "total_wall_ms": 37832.18812500127, "ttft_ms": 11.933750007301569, "input_usage": {"lookups": 7}, "output_usage": {"output_tokens": 9}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "Priyamvada Sørlie", "evidence_artifact": {"ledger": {"facts_total": 63, "facts_retained": 12, "retention_rate": 0.19047619047619047, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.25, "amount": 0.07692307692307693, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 818}, "ledger_by_hop": {"1": {"facts_total": 63, "facts_retained": 9, "retention_rate": 0.14285714285714285, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.25, "amount": 0.07692307692307693, "status": 0.25, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 787}, "2": {"facts_total": 63, "facts_retained": 12, "retention_rate": 0.19047619047619047, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.25, "amount": 0.07692307692307693, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 818}, "3": {"facts_total": 63, "facts_retained": 12, "retention_rate": 0.19047619047619047, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.25, "amount": 0.07692307692307693, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 818}, "4": {"facts_total": 63, "facts_retained": 12, "retention_rate": 0.19047619047619047, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.25, "amount": 0.07692307692307693, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 818}, "5": {"facts_total": 63, "facts_retained": 12, "retention_rate": 0.19047619047619047, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.25, "amount": 0.07692307692307693, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 818}, "6": {"facts_total": 63, "facts_retained": 12, "retention_rate": 0.19047619047619047, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.25, "amount": 0.07692307692307693, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 818}, "7": {"facts_total": 63, "facts_retained": 12, "retention_rate": 0.19047619047619047, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.25, "amount": 0.07692307692307693, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 818}, "8": {"facts_total": 63, "facts_retained": 12, "retention_rate": 0.19047619047619047, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.25, "amount": 0.07692307692307693, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 818}}}, "reference_key_id": "keys/synthetic/syn_03027_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "7ab33a7e50920be6dd2887eea3e354ca9b7ea53ad5cd929f38c58204c3e3af7e", "output_hash": "c5f554196ed23dd71ad995ec89f7142099d524dbb6534cc10d94a0c89d13e792"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "9cdbccae53bba2d9a4d44a84", "attempt_id": "f203b1c63fa8", "run_group_id": "E4_core:A5|repairable|qwen3b:h8", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:32:35.116097Z", "task_family": "synthetic", "document_id": "syn_03027_2000", "document_hash": "10326080979f71b625e4110c7c19767e5a752e267de5034f7e37d98cb6e06249", "question_id": "q04_multi_hop", "split": "test", "cluster_id": "syn_03027_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 8, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1955, "query_tokens": 432, "message_tokens": 339, "message_utf8_bytes": 858, "cache_tensor_bytes": null, "serialized_payload_bytes": 858, "side_information_bytes": 0, "fetched_bytes": 4214, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 37551.62508401554, "hops": {"1": 6904.5422920025885, "2": 4380.263000028208, "3": 4375.677625066601, "4": 4371.123000048101, "5": 4391.168999951333, "6": 4377.536708954722, "7": 4374.012624961324, "8": 4377.300833002664}, "receiver_prefill_ms": 162.9758341005072, "ttft_ms": 12.005583965219557, "decode_ms": 11.819457984529436}, "total_wall_ms": 37738.981500966474, "ttft_ms": 12.005583965219557, "input_usage": {"lookups": 7}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 63, "facts_retained": 12, "retention_rate": 0.19047619047619047, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.25, "amount": 0.07692307692307693, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 818}, "ledger_by_hop": {"1": {"facts_total": 63, "facts_retained": 9, "retention_rate": 0.14285714285714285, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.25, "amount": 0.07692307692307693, "status": 0.25, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 787}, "2": {"facts_total": 63, "facts_retained": 12, "retention_rate": 0.19047619047619047, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.25, "amount": 0.07692307692307693, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 818}, "3": {"facts_total": 63, "facts_retained": 12, "retention_rate": 0.19047619047619047, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.25, "amount": 0.07692307692307693, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 818}, "4": {"facts_total": 63, "facts_retained": 12, "retention_rate": 0.19047619047619047, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.25, "amount": 0.07692307692307693, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 818}, "5": {"facts_total": 63, "facts_retained": 12, "retention_rate": 0.19047619047619047, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.25, "amount": 0.07692307692307693, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 818}, "6": {"facts_total": 63, "facts_retained": 12, "retention_rate": 0.19047619047619047, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.25, "amount": 0.07692307692307693, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 818}, "7": {"facts_total": 63, "facts_retained": 12, "retention_rate": 0.19047619047619047, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.25, "amount": 0.07692307692307693, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 818}, "8": {"facts_total": 63, "facts_retained": 12, "retention_rate": 0.19047619047619047, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.25, "amount": 0.07692307692307693, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 818}}}, "reference_key_id": "keys/synthetic/syn_03027_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "7ab33a7e50920be6dd2887eea3e354ca9b7ea53ad5cd929f38c58204c3e3af7e", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "acad1d07d8826bbc007c28f0", "attempt_id": "d5cd60bdd5c8", "run_group_id": "E4_core:A5|repairable|qwen3b:h8", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:32:35.320506Z", "task_family": "synthetic", "document_id": "syn_03027_2000", "document_hash": "10326080979f71b625e4110c7c19767e5a752e267de5034f7e37d98cb6e06249", "question_id": "q05_multi_hop", "split": "test", "cluster_id": "syn_03027_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 8, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1955, "query_tokens": 429, "message_tokens": 339, "message_utf8_bytes": 858, "cache_tensor_bytes": null, "serialized_payload_bytes": 858, "side_information_bytes": 0, "fetched_bytes": 4214, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 37551.62508401554, "hops": {"1": 6904.5422920025885, "2": 4380.263000028208, "3": 4375.677625066601, "4": 4371.123000048101, "5": 4391.168999951333, "6": 4377.536708954722, "7": 4374.012624961324, "8": 4377.300833002664}, "receiver_prefill_ms": 162.78112505096942, "ttft_ms": 11.994667001999915, "decode_ms": 117.7488750545308}, "total_wall_ms": 37844.68225098681, "ttft_ms": 11.994667001999915, "input_usage": {"lookups": 7}, "output_usage": {"output_tokens": 10}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "WT-26M-9227", "evidence_artifact": {"ledger": {"facts_total": 63, "facts_retained": 12, "retention_rate": 0.19047619047619047, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.25, "amount": 0.07692307692307693, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 818}, "ledger_by_hop": {"1": {"facts_total": 63, "facts_retained": 9, "retention_rate": 0.14285714285714285, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.25, "amount": 0.07692307692307693, "status": 0.25, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 787}, "2": {"facts_total": 63, "facts_retained": 12, "retention_rate": 0.19047619047619047, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.25, "amount": 0.07692307692307693, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 818}, "3": {"facts_total": 63, "facts_retained": 12, "retention_rate": 0.19047619047619047, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.25, "amount": 0.07692307692307693, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 818}, "4": {"facts_total": 63, "facts_retained": 12, "retention_rate": 0.19047619047619047, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.25, "amount": 0.07692307692307693, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 818}, "5": {"facts_total": 63, "facts_retained": 12, "retention_rate": 0.19047619047619047, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.25, "amount": 0.07692307692307693, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 818}, "6": {"facts_total": 63, "facts_retained": 12, "retention_rate": 0.19047619047619047, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.25, "amount": 0.07692307692307693, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 818}, "7": {"facts_total": 63, "facts_retained": 12, "retention_rate": 0.19047619047619047, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.25, "amount": 0.07692307692307693, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 818}, "8": {"facts_total": 63, "facts_retained": 12, "retention_rate": 0.19047619047619047, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.25, "amount": 0.07692307692307693, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 818}}}, "reference_key_id": "keys/synthetic/syn_03027_2000.json", "task_scores": {"correct": 0.0, "abstained": 0.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "7ab33a7e50920be6dd2887eea3e354ca9b7ea53ad5cd929f38c58204c3e3af7e", "output_hash": "ef2302e0ab6f5a6951b11463fea177ba766b9d438afcaf6e6101ebfa3ba38aba"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "7a240f56fa53b0d50b39a871", "attempt_id": "52484533d07c", "run_group_id": "E4_core:A5|repairable|qwen3b:h8", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:32:35.630298Z", "task_family": "synthetic", "document_id": "syn_03027_2000", "document_hash": "10326080979f71b625e4110c7c19767e5a752e267de5034f7e37d98cb6e06249", "question_id": "q06_negation", "split": "test", "cluster_id": "syn_03027_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 8, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1955, "query_tokens": 421, "message_tokens": 339, "message_utf8_bytes": 858, "cache_tensor_bytes": null, "serialized_payload_bytes": 858, "side_information_bytes": 0, "fetched_bytes": 4214, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 37551.62508401554, "hops": {"1": 6904.5422920025885, "2": 4380.263000028208, "3": 4375.677625066601, "4": 4371.123000048101, "5": 4391.168999951333, "6": 4377.536708954722, "7": 4374.012624961324, "8": 4377.300833002664}, "receiver_prefill_ms": 162.44033304974437, "ttft_ms": 11.959042050875723, "decode_ms": 11.80325006134808}, "total_wall_ms": 37738.37541707326, "ttft_ms": 11.959042050875723, "input_usage": {"lookups": 7}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 63, "facts_retained": 12, "retention_rate": 0.19047619047619047, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.25, "amount": 0.07692307692307693, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 818}, "ledger_by_hop": {"1": {"facts_total": 63, "facts_retained": 9, "retention_rate": 0.14285714285714285, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.25, "amount": 0.07692307692307693, "status": 0.25, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 787}, "2": {"facts_total": 63, "facts_retained": 12, "retention_rate": 0.19047619047619047, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.25, "amount": 0.07692307692307693, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 818}, "3": {"facts_total": 63, "facts_retained": 12, "retention_rate": 0.19047619047619047, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.25, "amount": 0.07692307692307693, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 818}, "4": {"facts_total": 63, "facts_retained": 12, "retention_rate": 0.19047619047619047, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.25, "amount": 0.07692307692307693, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 818}, "5": {"facts_total": 63, "facts_retained": 12, "retention_rate": 0.19047619047619047, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.25, "amount": 0.07692307692307693, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 818}, "6": {"facts_total": 63, "facts_retained": 12, "retention_rate": 0.19047619047619047, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.25, "amount": 0.07692307692307693, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 818}, "7": {"facts_total": 63, "facts_retained": 12, "retention_rate": 0.19047619047619047, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.25, "amount": 0.07692307692307693, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 818}, "8": {"facts_total": 63, "facts_retained": 12, "retention_rate": 0.19047619047619047, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.25, "amount": 0.07692307692307693, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 818}}}, "reference_key_id": "keys/synthetic/syn_03027_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "7ab33a7e50920be6dd2887eea3e354ca9b7ea53ad5cd929f38c58204c3e3af7e", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "74c23190343c10ac03dbf635", "attempt_id": "df1d707151eb", "run_group_id": "E4_core:A5|repairable|qwen3b:h8", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:32:35.833524Z", "task_family": "synthetic", "document_id": "syn_03027_2000", "document_hash": "10326080979f71b625e4110c7c19767e5a752e267de5034f7e37d98cb6e06249", "question_id": "q07_negation", "split": "test", "cluster_id": "syn_03027_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 8, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1955, "query_tokens": 422, "message_tokens": 339, "message_utf8_bytes": 858, "cache_tensor_bytes": null, "serialized_payload_bytes": 858, "side_information_bytes": 0, "fetched_bytes": 4214, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 37551.62508401554, "hops": {"1": 6904.5422920025885, "2": 4380.263000028208, "3": 4375.677625066601, "4": 4371.123000048101, "5": 4391.168999951333, "6": 4377.536708954722, "7": 4374.012624961324, "8": 4377.300833002664}, "receiver_prefill_ms": 162.72825002670288, "ttft_ms": 12.043875060044229, "decode_ms": 11.784875066950917}, "total_wall_ms": 37738.72566805221, "ttft_ms": 12.043875060044229, "input_usage": {"lookups": 7}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 63, "facts_retained": 12, "retention_rate": 0.19047619047619047, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.25, "amount": 0.07692307692307693, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 818}, "ledger_by_hop": {"1": {"facts_total": 63, "facts_retained": 9, "retention_rate": 0.14285714285714285, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.25, "amount": 0.07692307692307693, "status": 0.25, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 787}, "2": {"facts_total": 63, "facts_retained": 12, "retention_rate": 0.19047619047619047, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.25, "amount": 0.07692307692307693, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 818}, "3": {"facts_total": 63, "facts_retained": 12, "retention_rate": 0.19047619047619047, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.25, "amount": 0.07692307692307693, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 818}, "4": {"facts_total": 63, "facts_retained": 12, "retention_rate": 0.19047619047619047, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.25, "amount": 0.07692307692307693, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 818}, "5": {"facts_total": 63, "facts_retained": 12, "retention_rate": 0.19047619047619047, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.25, "amount": 0.07692307692307693, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 818}, "6": {"facts_total": 63, "facts_retained": 12, "retention_rate": 0.19047619047619047, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.25, "amount": 0.07692307692307693, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 818}, "7": {"facts_total": 63, "facts_retained": 12, "retention_rate": 0.19047619047619047, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.25, "amount": 0.07692307692307693, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 818}, "8": {"facts_total": 63, "facts_retained": 12, "retention_rate": 0.19047619047619047, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.25, "amount": 0.07692307692307693, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 818}}}, "reference_key_id": "keys/synthetic/syn_03027_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "7ab33a7e50920be6dd2887eea3e354ca9b7ea53ad5cd929f38c58204c3e3af7e", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "6ee2d494446dd486ca015d17", "attempt_id": "856c43d4ce76", "run_group_id": "E4_core:A5|repairable|qwen3b:h8", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:32:36.036757Z", "task_family": "synthetic", "document_id": "syn_03027_2000", "document_hash": "10326080979f71b625e4110c7c19767e5a752e267de5034f7e37d98cb6e06249", "question_id": "q08_revision", "split": "test", "cluster_id": "syn_03027_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 8, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1955, "query_tokens": 436, "message_tokens": 339, "message_utf8_bytes": 858, "cache_tensor_bytes": null, "serialized_payload_bytes": 858, "side_information_bytes": 0, "fetched_bytes": 4214, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 37551.62508401554, "hops": {"1": 6904.5422920025885, "2": 4380.263000028208, "3": 4375.677625066601, "4": 4371.123000048101, "5": 4391.168999951333, "6": 4377.536708954722, "7": 4374.012624961324, "8": 4377.300833002664}, "receiver_prefill_ms": 162.82924998085946, "ttft_ms": 12.080707936547697, "decode_ms": 106.78008291870356}, "total_wall_ms": 37833.86616699863, "ttft_ms": 12.080707936547697, "input_usage": {"lookups": 7}, "output_usage": {"output_tokens": 9}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "6,914,937", "evidence_artifact": {"ledger": {"facts_total": 63, "facts_retained": 12, "retention_rate": 0.19047619047619047, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.25, "amount": 0.07692307692307693, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 818}, "ledger_by_hop": {"1": {"facts_total": 63, "facts_retained": 9, "retention_rate": 0.14285714285714285, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.25, "amount": 0.07692307692307693, "status": 0.25, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 787}, "2": {"facts_total": 63, "facts_retained": 12, "retention_rate": 0.19047619047619047, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.25, "amount": 0.07692307692307693, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 818}, "3": {"facts_total": 63, "facts_retained": 12, "retention_rate": 0.19047619047619047, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.25, "amount": 0.07692307692307693, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 818}, "4": {"facts_total": 63, "facts_retained": 12, "retention_rate": 0.19047619047619047, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.25, "amount": 0.07692307692307693, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 818}, "5": {"facts_total": 63, "facts_retained": 12, "retention_rate": 0.19047619047619047, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.25, "amount": 0.07692307692307693, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 818}, "6": {"facts_total": 63, "facts_retained": 12, "retention_rate": 0.19047619047619047, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.25, "amount": 0.07692307692307693, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 818}, "7": {"facts_total": 63, "facts_retained": 12, "retention_rate": 0.19047619047619047, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.25, "amount": 0.07692307692307693, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 818}, "8": {"facts_total": 63, "facts_retained": 12, "retention_rate": 0.19047619047619047, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.25, "amount": 0.07692307692307693, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 818}}}, "reference_key_id": "keys/synthetic/syn_03027_2000.json", "task_scores": {"correct": 0.0, "abstained": 0.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "7ab33a7e50920be6dd2887eea3e354ca9b7ea53ad5cd929f38c58204c3e3af7e", "output_hash": "c56851abfdbf8851226a1e71c4911c616a691c490a883eb754e61bac551d004c"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "fb8baac8f08eb7f253855b90", "attempt_id": "7e86b02eb9a8", "run_group_id": "E4_core:A5|repairable|qwen3b:h8", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:32:36.335315Z", "task_family": "synthetic", "document_id": "syn_03027_2000", "document_hash": "10326080979f71b625e4110c7c19767e5a752e267de5034f7e37d98cb6e06249", "question_id": "q09_revision", "split": "test", "cluster_id": "syn_03027_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 8, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1955, "query_tokens": 429, "message_tokens": 339, "message_utf8_bytes": 858, "cache_tensor_bytes": null, "serialized_payload_bytes": 858, "side_information_bytes": 0, "fetched_bytes": 4214, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 37551.62508401554, "hops": {"1": 6904.5422920025885, "2": 4380.263000028208, "3": 4375.677625066601, "4": 4371.123000048101, "5": 4391.168999951333, "6": 4377.536708954722, "7": 4374.012624961324, "8": 4377.300833002664}, "receiver_prefill_ms": 163.19829202257097, "ttft_ms": 12.103290995582938, "decode_ms": 94.25379196181893}, "total_wall_ms": 37821.74595899414, "ttft_ms": 12.103290995582938, "input_usage": {"lookups": 7}, "output_usage": {"output_tokens": 8}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "15828869", "evidence_artifact": {"ledger": {"facts_total": 63, "facts_retained": 12, "retention_rate": 0.19047619047619047, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.25, "amount": 0.07692307692307693, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 818}, "ledger_by_hop": {"1": {"facts_total": 63, "facts_retained": 9, "retention_rate": 0.14285714285714285, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.25, "amount": 0.07692307692307693, "status": 0.25, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 787}, "2": {"facts_total": 63, "facts_retained": 12, "retention_rate": 0.19047619047619047, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.25, "amount": 0.07692307692307693, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 818}, "3": {"facts_total": 63, "facts_retained": 12, "retention_rate": 0.19047619047619047, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.25, "amount": 0.07692307692307693, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 818}, "4": {"facts_total": 63, "facts_retained": 12, "retention_rate": 0.19047619047619047, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.25, "amount": 0.07692307692307693, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 818}, "5": {"facts_total": 63, "facts_retained": 12, "retention_rate": 0.19047619047619047, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.25, "amount": 0.07692307692307693, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 818}, "6": {"facts_total": 63, "facts_retained": 12, "retention_rate": 0.19047619047619047, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.25, "amount": 0.07692307692307693, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 818}, "7": {"facts_total": 63, "facts_retained": 12, "retention_rate": 0.19047619047619047, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.25, "amount": 0.07692307692307693, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 818}, "8": {"facts_total": 63, "facts_retained": 12, "retention_rate": 0.19047619047619047, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.25, "amount": 0.07692307692307693, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 818}}}, "reference_key_id": "keys/synthetic/syn_03027_2000.json", "task_scores": {"correct": 0.0, "abstained": 0.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "7ab33a7e50920be6dd2887eea3e354ca9b7ea53ad5cd929f38c58204c3e3af7e", "output_hash": "c37a6a24f6b43a5a4c626354b69a44fd62c2a91de72daee7f63e18e6519e339c"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "f983dd5cc660adb3a0944dc3", "attempt_id": "3428f9ab3c0d", "run_group_id": "E4_core:A5|repairable|qwen3b:h8", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:32:36.621764Z", "task_family": "synthetic", "document_id": "syn_03027_2000", "document_hash": "10326080979f71b625e4110c7c19767e5a752e267de5034f7e37d98cb6e06249", "question_id": "q10_temporal", "split": "test", "cluster_id": "syn_03027_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 8, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1955, "query_tokens": 439, "message_tokens": 339, "message_utf8_bytes": 858, "cache_tensor_bytes": null, "serialized_payload_bytes": 858, "side_information_bytes": 0, "fetched_bytes": 4214, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 37551.62508401554, "hops": {"1": 6904.5422920025885, "2": 4380.263000028208, "3": 4375.677625066601, "4": 4371.123000048101, "5": 4391.168999951333, "6": 4377.536708954722, "7": 4374.012624961324, "8": 4377.300833002664}, "receiver_prefill_ms": 163.37683296296746, "ttft_ms": 12.100375024601817, "decode_ms": 11.785792070440948}, "total_wall_ms": 37739.43362594582, "ttft_ms": 12.100375024601817, "input_usage": {"lookups": 7}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 63, "facts_retained": 12, "retention_rate": 0.19047619047619047, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.25, "amount": 0.07692307692307693, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 818}, "ledger_by_hop": {"1": {"facts_total": 63, "facts_retained": 9, "retention_rate": 0.14285714285714285, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.25, "amount": 0.07692307692307693, "status": 0.25, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 787}, "2": {"facts_total": 63, "facts_retained": 12, "retention_rate": 0.19047619047619047, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.25, "amount": 0.07692307692307693, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 818}, "3": {"facts_total": 63, "facts_retained": 12, "retention_rate": 0.19047619047619047, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.25, "amount": 0.07692307692307693, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 818}, "4": {"facts_total": 63, "facts_retained": 12, "retention_rate": 0.19047619047619047, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.25, "amount": 0.07692307692307693, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 818}, "5": {"facts_total": 63, "facts_retained": 12, "retention_rate": 0.19047619047619047, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.25, "amount": 0.07692307692307693, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 818}, "6": {"facts_total": 63, "facts_retained": 12, "retention_rate": 0.19047619047619047, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.25, "amount": 0.07692307692307693, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 818}, "7": {"facts_total": 63, "facts_retained": 12, "retention_rate": 0.19047619047619047, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.25, "amount": 0.07692307692307693, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 818}, "8": {"facts_total": 63, "facts_retained": 12, "retention_rate": 0.19047619047619047, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.25, "amount": 0.07692307692307693, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 818}}}, "reference_key_id": "keys/synthetic/syn_03027_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "7ab33a7e50920be6dd2887eea3e354ca9b7ea53ad5cd929f38c58204c3e3af7e", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "5a3b938433b7325380210ba5", "attempt_id": "d16eade7a533", "run_group_id": "E4_core:A5|repairable|qwen3b:h8", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:32:36.826294Z", "task_family": "synthetic", "document_id": "syn_03027_2000", "document_hash": "10326080979f71b625e4110c7c19767e5a752e267de5034f7e37d98cb6e06249", "question_id": "q11_aggregation", "split": "test", "cluster_id": "syn_03027_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 8, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1955, "query_tokens": 432, "message_tokens": 339, "message_utf8_bytes": 858, "cache_tensor_bytes": null, "serialized_payload_bytes": 858, "side_information_bytes": 0, "fetched_bytes": 4214, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 37551.62508401554, "hops": {"1": 6904.5422920025885, "2": 4380.263000028208, "3": 4375.677625066601, "4": 4371.123000048101, "5": 4391.168999951333, "6": 4377.536708954722, "7": 4374.012624961324, "8": 4377.300833002664}, "receiver_prefill_ms": 163.18254102952778, "ttft_ms": 11.924582999199629, "decode_ms": 82.65087497420609}, "total_wall_ms": 37809.9230841035, "ttft_ms": 11.924582999199629, "input_usage": {"lookups": 7}, "output_usage": {"output_tokens": 7}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "6914937", "evidence_artifact": {"ledger": {"facts_total": 63, "facts_retained": 12, "retention_rate": 0.19047619047619047, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.25, "amount": 0.07692307692307693, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 818}, "ledger_by_hop": {"1": {"facts_total": 63, "facts_retained": 9, "retention_rate": 0.14285714285714285, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.25, "amount": 0.07692307692307693, "status": 0.25, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 787}, "2": {"facts_total": 63, "facts_retained": 12, "retention_rate": 0.19047619047619047, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.25, "amount": 0.07692307692307693, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 818}, "3": {"facts_total": 63, "facts_retained": 12, "retention_rate": 0.19047619047619047, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.25, "amount": 0.07692307692307693, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 818}, "4": {"facts_total": 63, "facts_retained": 12, "retention_rate": 0.19047619047619047, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.25, "amount": 0.07692307692307693, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 818}, "5": {"facts_total": 63, "facts_retained": 12, "retention_rate": 0.19047619047619047, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.25, "amount": 0.07692307692307693, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 818}, "6": {"facts_total": 63, "facts_retained": 12, "retention_rate": 0.19047619047619047, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.25, "amount": 0.07692307692307693, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 818}, "7": {"facts_total": 63, "facts_retained": 12, "retention_rate": 0.19047619047619047, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.25, "amount": 0.07692307692307693, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 818}, "8": {"facts_total": 63, "facts_retained": 12, "retention_rate": 0.19047619047619047, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.25, "amount": 0.07692307692307693, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 818}}}, "reference_key_id": "keys/synthetic/syn_03027_2000.json", "task_scores": {"correct": 0.0, "abstained": 0.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "7ab33a7e50920be6dd2887eea3e354ca9b7ea53ad5cd929f38c58204c3e3af7e", "output_hash": "534d40ded32f784c68307dd2386faed2a27bf39dcc27321ee6985e42ccd285e5"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "ae97db2ca3bf2b857c0c4853", "attempt_id": "855320fc4edc", "run_group_id": "E4_core:A5|repairable|qwen3b:h8", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:32:37.100907Z", "task_family": "synthetic", "document_id": "syn_03027_2000", "document_hash": "10326080979f71b625e4110c7c19767e5a752e267de5034f7e37d98cb6e06249", "question_id": "q12_unit_conversion", "split": "test", "cluster_id": "syn_03027_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 8, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1955, "query_tokens": 429, "message_tokens": 339, "message_utf8_bytes": 858, "cache_tensor_bytes": null, "serialized_payload_bytes": 858, "side_information_bytes": 0, "fetched_bytes": 4214, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 37551.62508401554, "hops": {"1": 6904.5422920025885, "2": 4380.263000028208, "3": 4375.677625066601, "4": 4371.123000048101, "5": 4391.168999951333, "6": 4377.536708954722, "7": 4374.012624961324, "8": 4377.300833002664}, "receiver_prefill_ms": 163.13762497156858, "ttft_ms": 12.296458939090371, "decode_ms": 12.01220799703151}, "total_wall_ms": 37739.60337601602, "ttft_ms": 12.296458939090371, "input_usage": {"lookups": 7}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 63, "facts_retained": 12, "retention_rate": 0.19047619047619047, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.25, "amount": 0.07692307692307693, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 818}, "ledger_by_hop": {"1": {"facts_total": 63, "facts_retained": 9, "retention_rate": 0.14285714285714285, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.25, "amount": 0.07692307692307693, "status": 0.25, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 787}, "2": {"facts_total": 63, "facts_retained": 12, "retention_rate": 0.19047619047619047, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.25, "amount": 0.07692307692307693, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 818}, "3": {"facts_total": 63, "facts_retained": 12, "retention_rate": 0.19047619047619047, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.25, "amount": 0.07692307692307693, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 818}, "4": {"facts_total": 63, "facts_retained": 12, "retention_rate": 0.19047619047619047, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.25, "amount": 0.07692307692307693, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 818}, "5": {"facts_total": 63, "facts_retained": 12, "retention_rate": 0.19047619047619047, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.25, "amount": 0.07692307692307693, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 818}, "6": {"facts_total": 63, "facts_retained": 12, "retention_rate": 0.19047619047619047, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.25, "amount": 0.07692307692307693, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 818}, "7": {"facts_total": 63, "facts_retained": 12, "retention_rate": 0.19047619047619047, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.25, "amount": 0.07692307692307693, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 818}, "8": {"facts_total": 63, "facts_retained": 12, "retention_rate": 0.19047619047619047, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.25, "amount": 0.07692307692307693, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 818}}}, "reference_key_id": "keys/synthetic/syn_03027_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "7ab33a7e50920be6dd2887eea3e354ca9b7ea53ad5cd929f38c58204c3e3af7e", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "e5f61921fb6e86702125797a", "attempt_id": "7806cd14f4fb", "run_group_id": "E4_core:A5|repairable|qwen3b:h8", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:32:37.305026Z", "task_family": "synthetic", "document_id": "syn_03027_2000", "document_hash": "10326080979f71b625e4110c7c19767e5a752e267de5034f7e37d98cb6e06249", "question_id": "q13_unanswerable", "split": "test", "cluster_id": "syn_03027_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 8, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1955, "query_tokens": 432, "message_tokens": 339, "message_utf8_bytes": 858, "cache_tensor_bytes": null, "serialized_payload_bytes": 858, "side_information_bytes": 0, "fetched_bytes": 4214, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 37551.62508401554, "hops": {"1": 6904.5422920025885, "2": 4380.263000028208, "3": 4375.677625066601, "4": 4371.123000048101, "5": 4391.168999951333, "6": 4377.536708954722, "7": 4374.012624961324, "8": 4377.300833002664}, "receiver_prefill_ms": 162.32495789881796, "ttft_ms": 11.958708055317402, "decode_ms": 11.77216600626707}, "total_wall_ms": 37738.21370897349, "ttft_ms": 11.958708055317402, "input_usage": {"lookups": 7}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 63, "facts_retained": 12, "retention_rate": 0.19047619047619047, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.25, "amount": 0.07692307692307693, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 818}, "ledger_by_hop": {"1": {"facts_total": 63, "facts_retained": 9, "retention_rate": 0.14285714285714285, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.25, "amount": 0.07692307692307693, "status": 0.25, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 787}, "2": {"facts_total": 63, "facts_retained": 12, "retention_rate": 0.19047619047619047, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.25, "amount": 0.07692307692307693, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 818}, "3": {"facts_total": 63, "facts_retained": 12, "retention_rate": 0.19047619047619047, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.25, "amount": 0.07692307692307693, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 818}, "4": {"facts_total": 63, "facts_retained": 12, "retention_rate": 0.19047619047619047, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.25, "amount": 0.07692307692307693, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 818}, "5": {"facts_total": 63, "facts_retained": 12, "retention_rate": 0.19047619047619047, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.25, "amount": 0.07692307692307693, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 818}, "6": {"facts_total": 63, "facts_retained": 12, "retention_rate": 0.19047619047619047, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.25, "amount": 0.07692307692307693, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 818}, "7": {"facts_total": 63, "facts_retained": 12, "retention_rate": 0.19047619047619047, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.25, "amount": 0.07692307692307693, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 818}, "8": {"facts_total": 63, "facts_retained": 12, "retention_rate": 0.19047619047619047, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.25, "amount": 0.07692307692307693, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 818}}}, "reference_key_id": "keys/synthetic/syn_03027_2000.json", "task_scores": {"correct": 1.0, "abstained": 1.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "7ab33a7e50920be6dd2887eea3e354ca9b7ea53ad5cd929f38c58204c3e3af7e", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "2164eaf55adec160350afcea", "attempt_id": "4028e3ae5bf4", "run_group_id": "E4_core:A5|repairable|qwen3b:h8", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:32:37.508467Z", "task_family": "synthetic", "document_id": "syn_03027_2000", "document_hash": "10326080979f71b625e4110c7c19767e5a752e267de5034f7e37d98cb6e06249", "question_id": "q14_unanswerable", "split": "test", "cluster_id": "syn_03027_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "repairable", "agents": ["qwen3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_repairable", "handoff_index": 8, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1955, "query_tokens": 434, "message_tokens": 339, "message_utf8_bytes": 858, "cache_tensor_bytes": null, "serialized_payload_bytes": 858, "side_information_bytes": 0, "fetched_bytes": 4214, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 37551.62508401554, "hops": {"1": 6904.5422920025885, "2": 4380.263000028208, "3": 4375.677625066601, "4": 4371.123000048101, "5": 4391.168999951333, "6": 4377.536708954722, "7": 4374.012624961324, "8": 4377.300833002664}, "receiver_prefill_ms": 163.11120800673962, "ttft_ms": 12.010792037472129, "decode_ms": 11.776332976296544}, "total_wall_ms": 37739.052542019635, "ttft_ms": 12.010792037472129, "input_usage": {"lookups": 7}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 63, "facts_retained": 12, "retention_rate": 0.19047619047619047, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.25, "amount": 0.07692307692307693, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 818}, "ledger_by_hop": {"1": {"facts_total": 63, "facts_retained": 9, "retention_rate": 0.14285714285714285, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.25, "amount": 0.07692307692307693, "status": 0.25, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 787}, "2": {"facts_total": 63, "facts_retained": 12, "retention_rate": 0.19047619047619047, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.25, "amount": 0.07692307692307693, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 818}, "3": {"facts_total": 63, "facts_retained": 12, "retention_rate": 0.19047619047619047, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.25, "amount": 0.07692307692307693, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 818}, "4": {"facts_total": 63, "facts_retained": 12, "retention_rate": 0.19047619047619047, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.25, "amount": 0.07692307692307693, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 818}, "5": {"facts_total": 63, "facts_retained": 12, "retention_rate": 0.19047619047619047, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.25, "amount": 0.07692307692307693, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 818}, "6": {"facts_total": 63, "facts_retained": 12, "retention_rate": 0.19047619047619047, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.25, "amount": 0.07692307692307693, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 818}, "7": {"facts_total": 63, "facts_retained": 12, "retention_rate": 0.19047619047619047, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.25, "amount": 0.07692307692307693, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 818}, "8": {"facts_total": 63, "facts_retained": 12, "retention_rate": 0.19047619047619047, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.25, "amount": 0.07692307692307693, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 818}}}, "reference_key_id": "keys/synthetic/syn_03027_2000.json", "task_scores": {"correct": 1.0, "abstained": 1.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "7ab33a7e50920be6dd2887eea3e354ca9b7ea53ad5cd929f38c58204c3e3af7e", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "ecf378f18f83e91a4a2bb009", "attempt_id": "a840f58ffae3", "run_group_id": "E4_core:A4|closed|qwen3b-llama3b:h1", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:32:44.586850Z", "task_family": "synthetic", "document_id": "syn_03027_2000", "document_hash": "10326080979f71b625e4110c7c19767e5a752e267de5034f7e37d98cb6e06249", "question_id": "q00_single_fact", "split": "test", "cluster_id": "syn_03027_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 1, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1955, "query_tokens": 594, "message_tokens": 512, "message_utf8_bytes": 1087, "cache_tensor_bytes": null, "serialized_payload_bytes": 1087, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 6872.302875039168, "hops": {"1": 6872.302875039168}, "receiver_prefill_ms": 226.39816696755588, "ttft_ms": 12.008749996311963, "decode_ms": 11.808292008936405}, "total_wall_ms": 7123.195458087139, "ttft_ms": 12.008749996311963, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 63, "facts_retained": 27, "retention_rate": 0.42857142857142855, "retention_by_kind": {"person": 0.5, "date": 0.125, "amount": 0.6153846153846154, "status": 0.75, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1087}, "ledger_by_hop": {"1": {"facts_total": 63, "facts_retained": 27, "retention_rate": 0.42857142857142855, "retention_by_kind": {"person": 0.5, "date": 0.125, "amount": 0.6153846153846154, "status": 0.75, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1087}}}, "reference_key_id": "keys/synthetic/syn_03027_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "1f88f0e78c63504e418dea843641b99a98555e15d47b02cd5cb2f57da3003dea", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "3717a5487ca301c1b90e712a", "attempt_id": "18e238d8b598", "run_group_id": "E4_core:A4|closed|qwen3b-llama3b:h1", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:32:44.854300Z", "task_family": "synthetic", "document_id": "syn_03027_2000", "document_hash": "10326080979f71b625e4110c7c19767e5a752e267de5034f7e37d98cb6e06249", "question_id": "q01_single_fact", "split": "test", "cluster_id": "syn_03027_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 1, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1955, "query_tokens": 601, "message_tokens": 512, "message_utf8_bytes": 1087, "cache_tensor_bytes": null, "serialized_payload_bytes": 1087, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 6872.302875039168, "hops": {"1": 6872.302875039168}, "receiver_prefill_ms": 226.32675000932068, "ttft_ms": 12.204624945297837, "decode_ms": 141.75216702278703}, "total_wall_ms": 7253.243750077672, "ttft_ms": 12.204624945297837, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 12}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "Lot 247-S, Baie-Comeau", "evidence_artifact": {"ledger": {"facts_total": 63, "facts_retained": 27, "retention_rate": 0.42857142857142855, "retention_by_kind": {"person": 0.5, "date": 0.125, "amount": 0.6153846153846154, "status": 0.75, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1087}, "ledger_by_hop": {"1": {"facts_total": 63, "facts_retained": 27, "retention_rate": 0.42857142857142855, "retention_by_kind": {"person": 0.5, "date": 0.125, "amount": 0.6153846153846154, "status": 0.75, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1087}}}, "reference_key_id": "keys/synthetic/syn_03027_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 0.8571428571428571}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "1f88f0e78c63504e418dea843641b99a98555e15d47b02cd5cb2f57da3003dea", "output_hash": "77b7da1d43e7d8815e9e9159ebbd735045bc7f668260ba30b93674f70622b8ec"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "f39015b7f4b92e8aadb79155", "attempt_id": "5821318a51f3", "run_group_id": "E4_core:A4|closed|qwen3b-llama3b:h1", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:32:45.251472Z", "task_family": "synthetic", "document_id": "syn_03027_2000", "document_hash": "10326080979f71b625e4110c7c19767e5a752e267de5034f7e37d98cb6e06249", "question_id": "q02_exact_copy", "split": "test", "cluster_id": "syn_03027_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 1, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1955, "query_tokens": 600, "message_tokens": 512, "message_utf8_bytes": 1087, "cache_tensor_bytes": null, "serialized_payload_bytes": 1087, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 6872.302875039168, "hops": {"1": 6872.302875039168}, "receiver_prefill_ms": 226.2632919009775, "ttft_ms": 12.079749954864383, "decode_ms": 82.49475003685802}, "total_wall_ms": 7193.786250078119, "ttft_ms": 12.079749954864383, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 7}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "15.0054", "evidence_artifact": {"ledger": {"facts_total": 63, "facts_retained": 27, "retention_rate": 0.42857142857142855, "retention_by_kind": {"person": 0.5, "date": 0.125, "amount": 0.6153846153846154, "status": 0.75, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1087}, "ledger_by_hop": {"1": {"facts_total": 63, "facts_retained": 27, "retention_rate": 0.42857142857142855, "retention_by_kind": {"person": 0.5, "date": 0.125, "amount": 0.6153846153846154, "status": 0.75, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1087}}}, "reference_key_id": "keys/synthetic/syn_03027_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "1f88f0e78c63504e418dea843641b99a98555e15d47b02cd5cb2f57da3003dea", "output_hash": "d13adba73068c847c850237059e0202fc5b91ccacf2e792e9ce5e3b90556501a"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "89afbd65374758573e3bf235", "attempt_id": "993379ddccbd", "run_group_id": "E4_core:A4|closed|qwen3b-llama3b:h1", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:32:45.589886Z", "task_family": "synthetic", "document_id": "syn_03027_2000", "document_hash": "10326080979f71b625e4110c7c19767e5a752e267de5034f7e37d98cb6e06249", "question_id": "q03_exact_copy", "split": "test", "cluster_id": "syn_03027_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 1, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1955, "query_tokens": 595, "message_tokens": 512, "message_utf8_bytes": 1087, "cache_tensor_bytes": null, "serialized_payload_bytes": 1087, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 6872.302875039168, "hops": {"1": 6872.302875039168}, "receiver_prefill_ms": 226.6386670526117, "ttft_ms": 12.205207953229547, "decode_ms": 107.5730420416221}, "total_wall_ms": 7219.407292082906, "ttft_ms": 12.205207953229547, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 9}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "Priyamvada Sørlie", "evidence_artifact": {"ledger": {"facts_total": 63, "facts_retained": 27, "retention_rate": 0.42857142857142855, "retention_by_kind": {"person": 0.5, "date": 0.125, "amount": 0.6153846153846154, "status": 0.75, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1087}, "ledger_by_hop": {"1": {"facts_total": 63, "facts_retained": 27, "retention_rate": 0.42857142857142855, "retention_by_kind": {"person": 0.5, "date": 0.125, "amount": 0.6153846153846154, "status": 0.75, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1087}}}, "reference_key_id": "keys/synthetic/syn_03027_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "1f88f0e78c63504e418dea843641b99a98555e15d47b02cd5cb2f57da3003dea", "output_hash": "c5f554196ed23dd71ad995ec89f7142099d524dbb6534cc10d94a0c89d13e792"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "a6ceb38c7a36d3b76b121ce6", "attempt_id": "558191f219e4", "run_group_id": "E4_core:A4|closed|qwen3b-llama3b:h1", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:32:45.953968Z", "task_family": "synthetic", "document_id": "syn_03027_2000", "document_hash": "10326080979f71b625e4110c7c19767e5a752e267de5034f7e37d98cb6e06249", "question_id": "q04_multi_hop", "split": "test", "cluster_id": "syn_03027_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 1, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1955, "query_tokens": 606, "message_tokens": 512, "message_utf8_bytes": 1087, "cache_tensor_bytes": null, "serialized_payload_bytes": 1087, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 6872.302875039168, "hops": {"1": 6872.302875039168}, "receiver_prefill_ms": 226.708208094351, "ttft_ms": 12.044125003740191, "decode_ms": 11.885791085660458}, "total_wall_ms": 7123.605582979508, "ttft_ms": 12.044125003740191, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 63, "facts_retained": 27, "retention_rate": 0.42857142857142855, "retention_by_kind": {"person": 0.5, "date": 0.125, "amount": 0.6153846153846154, "status": 0.75, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1087}, "ledger_by_hop": {"1": {"facts_total": 63, "facts_retained": 27, "retention_rate": 0.42857142857142855, "retention_by_kind": {"person": 0.5, "date": 0.125, "amount": 0.6153846153846154, "status": 0.75, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1087}}}, "reference_key_id": "keys/synthetic/syn_03027_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "1f88f0e78c63504e418dea843641b99a98555e15d47b02cd5cb2f57da3003dea", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "d55940ad814ea51f66aa670d", "attempt_id": "a0400bdb8e95", "run_group_id": "E4_core:A4|closed|qwen3b-llama3b:h1", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:32:46.222757Z", "task_family": "synthetic", "document_id": "syn_03027_2000", "document_hash": "10326080979f71b625e4110c7c19767e5a752e267de5034f7e37d98cb6e06249", "question_id": "q05_multi_hop", "split": "test", "cluster_id": "syn_03027_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 1, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1955, "query_tokens": 603, "message_tokens": 512, "message_utf8_bytes": 1087, "cache_tensor_bytes": null, "serialized_payload_bytes": 1087, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 6872.302875039168, "hops": {"1": 6872.302875039168}, "receiver_prefill_ms": 227.11304191034287, "ttft_ms": 12.050083023495972, "decode_ms": 11.975583038292825}, "total_wall_ms": 7124.084834009409, "ttft_ms": 12.050083023495972, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 63, "facts_retained": 27, "retention_rate": 0.42857142857142855, "retention_by_kind": {"person": 0.5, "date": 0.125, "amount": 0.6153846153846154, "status": 0.75, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1087}, "ledger_by_hop": {"1": {"facts_total": 63, "facts_retained": 27, "retention_rate": 0.42857142857142855, "retention_by_kind": {"person": 0.5, "date": 0.125, "amount": 0.6153846153846154, "status": 0.75, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1087}}}, "reference_key_id": "keys/synthetic/syn_03027_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "1f88f0e78c63504e418dea843641b99a98555e15d47b02cd5cb2f57da3003dea", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "5f64637debfcb3abed126c29", "attempt_id": "9d1ee3a7170e", "run_group_id": "E4_core:A4|closed|qwen3b-llama3b:h1", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:32:46.490727Z", "task_family": "synthetic", "document_id": "syn_03027_2000", "document_hash": "10326080979f71b625e4110c7c19767e5a752e267de5034f7e37d98cb6e06249", "question_id": "q06_negation", "split": "test", "cluster_id": "syn_03027_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 1, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1955, "query_tokens": 595, "message_tokens": 512, "message_utf8_bytes": 1087, "cache_tensor_bytes": null, "serialized_payload_bytes": 1087, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 6872.302875039168, "hops": {"1": 6872.302875039168}, "receiver_prefill_ms": 226.7370419576764, "ttft_ms": 12.006000033579767, "decode_ms": 11.910791974514723}, "total_wall_ms": 7123.604124994017, "ttft_ms": 12.006000033579767, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 63, "facts_retained": 27, "retention_rate": 0.42857142857142855, "retention_by_kind": {"person": 0.5, "date": 0.125, "amount": 0.6153846153846154, "status": 0.75, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1087}, "ledger_by_hop": {"1": {"facts_total": 63, "facts_retained": 27, "retention_rate": 0.42857142857142855, "retention_by_kind": {"person": 0.5, "date": 0.125, "amount": 0.6153846153846154, "status": 0.75, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1087}}}, "reference_key_id": "keys/synthetic/syn_03027_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "1f88f0e78c63504e418dea843641b99a98555e15d47b02cd5cb2f57da3003dea", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "00dcee01a631ac4074851cfd", "attempt_id": "ed9d5d16718b", "run_group_id": "E4_core:A4|closed|qwen3b-llama3b:h1", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:32:46.758253Z", "task_family": "synthetic", "document_id": "syn_03027_2000", "document_hash": "10326080979f71b625e4110c7c19767e5a752e267de5034f7e37d98cb6e06249", "question_id": "q07_negation", "split": "test", "cluster_id": "syn_03027_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 1, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1955, "query_tokens": 596, "message_tokens": 512, "message_utf8_bytes": 1087, "cache_tensor_bytes": null, "serialized_payload_bytes": 1087, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 6872.302875039168, "hops": {"1": 6872.302875039168}, "receiver_prefill_ms": 226.4244999969378, "ttft_ms": 12.102290987968445, "decode_ms": 11.87124999705702}, "total_wall_ms": 7123.323500039987, "ttft_ms": 12.102290987968445, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "no", "evidence_artifact": {"ledger": {"facts_total": 63, "facts_retained": 27, "retention_rate": 0.42857142857142855, "retention_by_kind": {"person": 0.5, "date": 0.125, "amount": 0.6153846153846154, "status": 0.75, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1087}, "ledger_by_hop": {"1": {"facts_total": 63, "facts_retained": 27, "retention_rate": 0.42857142857142855, "retention_by_kind": {"person": 0.5, "date": 0.125, "amount": 0.6153846153846154, "status": 0.75, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1087}}}, "reference_key_id": "keys/synthetic/syn_03027_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "1f88f0e78c63504e418dea843641b99a98555e15d47b02cd5cb2f57da3003dea", "output_hash": "9390298f3fb0c5b160498935d79cb139aef28e1c47358b4bbba61862b9c26e59"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "dff5d1c7f0b819f9c7b485ba", "attempt_id": "d5715475cd55", "run_group_id": "E4_core:A4|closed|qwen3b-llama3b:h1", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:32:47.025362Z", "task_family": "synthetic", "document_id": "syn_03027_2000", "document_hash": "10326080979f71b625e4110c7c19767e5a752e267de5034f7e37d98cb6e06249", "question_id": "q08_revision", "split": "test", "cluster_id": "syn_03027_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 1, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1955, "query_tokens": 610, "message_tokens": 512, "message_utf8_bytes": 1087, "cache_tensor_bytes": null, "serialized_payload_bytes": 1087, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 6872.302875039168, "hops": {"1": 6872.302875039168}, "receiver_prefill_ms": 228.43862499576062, "ttft_ms": 11.999665992334485, "decode_ms": 82.97087496612221}, "total_wall_ms": 7196.34866702836, "ttft_ms": 11.999665992334485, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 7}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "8010205", "evidence_artifact": {"ledger": {"facts_total": 63, "facts_retained": 27, "retention_rate": 0.42857142857142855, "retention_by_kind": {"person": 0.5, "date": 0.125, "amount": 0.6153846153846154, "status": 0.75, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1087}, "ledger_by_hop": {"1": {"facts_total": 63, "facts_retained": 27, "retention_rate": 0.42857142857142855, "retention_by_kind": {"person": 0.5, "date": 0.125, "amount": 0.6153846153846154, "status": 0.75, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1087}}}, "reference_key_id": "keys/synthetic/syn_03027_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "1f88f0e78c63504e418dea843641b99a98555e15d47b02cd5cb2f57da3003dea", "output_hash": "6d1ceb36132f38bd77e8f46c23ea21dd9bcff78bc80bd8a9eaf2d7f2f3bacf46"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "4d5374393eb6553718c49068", "attempt_id": "b5aa76e038db", "run_group_id": "E4_core:A4|closed|qwen3b-llama3b:h1", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:32:47.366144Z", "task_family": "synthetic", "document_id": "syn_03027_2000", "document_hash": "10326080979f71b625e4110c7c19767e5a752e267de5034f7e37d98cb6e06249", "question_id": "q09_revision", "split": "test", "cluster_id": "syn_03027_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 1, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1955, "query_tokens": 603, "message_tokens": 512, "message_utf8_bytes": 1087, "cache_tensor_bytes": null, "serialized_payload_bytes": 1087, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 6872.302875039168, "hops": {"1": 6872.302875039168}, "receiver_prefill_ms": 227.09895798470825, "ttft_ms": 12.103792047128081, "decode_ms": 106.72749998047948}, "total_wall_ms": 7218.880292028189, "ttft_ms": 12.103792047128081, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 9}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "6,056,150", "evidence_artifact": {"ledger": {"facts_total": 63, "facts_retained": 27, "retention_rate": 0.42857142857142855, "retention_by_kind": {"person": 0.5, "date": 0.125, "amount": 0.6153846153846154, "status": 0.75, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1087}, "ledger_by_hop": {"1": {"facts_total": 63, "facts_retained": 27, "retention_rate": 0.42857142857142855, "retention_by_kind": {"person": 0.5, "date": 0.125, "amount": 0.6153846153846154, "status": 0.75, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1087}}}, "reference_key_id": "keys/synthetic/syn_03027_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "1f88f0e78c63504e418dea843641b99a98555e15d47b02cd5cb2f57da3003dea", "output_hash": "8262889a8478c7e0ee965f159d9849a3849b412e276f5b979476e2da5b320249"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "dda8b0ccc53ee42e5fc7832f", "attempt_id": "891dc21a2a8e", "run_group_id": "E4_core:A4|closed|qwen3b-llama3b:h1", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:32:47.729631Z", "task_family": "synthetic", "document_id": "syn_03027_2000", "document_hash": "10326080979f71b625e4110c7c19767e5a752e267de5034f7e37d98cb6e06249", "question_id": "q10_temporal", "split": "test", "cluster_id": "syn_03027_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 1, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1955, "query_tokens": 613, "message_tokens": 512, "message_utf8_bytes": 1087, "cache_tensor_bytes": null, "serialized_payload_bytes": 1087, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 6872.302875039168, "hops": {"1": 6872.302875039168}, "receiver_prefill_ms": 228.44483295921236, "ttft_ms": 12.022083974443376, "decode_ms": 82.89099996909499}, "total_wall_ms": 7196.3205829961225, "ttft_ms": 12.022083974443376, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 7}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "Lunlith-13", "evidence_artifact": {"ledger": {"facts_total": 63, "facts_retained": 27, "retention_rate": 0.42857142857142855, "retention_by_kind": {"person": 0.5, "date": 0.125, "amount": 0.6153846153846154, "status": 0.75, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1087}, "ledger_by_hop": {"1": {"facts_total": 63, "facts_retained": 27, "retention_rate": 0.42857142857142855, "retention_by_kind": {"person": 0.5, "date": 0.125, "amount": 0.6153846153846154, "status": 0.75, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1087}}}, "reference_key_id": "keys/synthetic/syn_03027_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "1f88f0e78c63504e418dea843641b99a98555e15d47b02cd5cb2f57da3003dea", "output_hash": "c45f000afe6d865c8167954eebdff761141225817d4f5b26c962cd50cd75d391"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "59f470b9d69dff39cae28df5", "attempt_id": "d64745a6af00", "run_group_id": "E4_core:A4|closed|qwen3b-llama3b:h1", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:32:48.070402Z", "task_family": "synthetic", "document_id": "syn_03027_2000", "document_hash": "10326080979f71b625e4110c7c19767e5a752e267de5034f7e37d98cb6e06249", "question_id": "q11_aggregation", "split": "test", "cluster_id": "syn_03027_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 1, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1955, "query_tokens": 606, "message_tokens": 512, "message_utf8_bytes": 1087, "cache_tensor_bytes": null, "serialized_payload_bytes": 1087, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 6872.302875039168, "hops": {"1": 6872.302875039168}, "receiver_prefill_ms": 226.62149998359382, "ttft_ms": 12.313459068536758, "decode_ms": 118.08500008191913}, "total_wall_ms": 7229.986250051297, "ttft_ms": 12.313459068536758, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 10}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "23,450,077", "evidence_artifact": {"ledger": {"facts_total": 63, "facts_retained": 27, "retention_rate": 0.42857142857142855, "retention_by_kind": {"person": 0.5, "date": 0.125, "amount": 0.6153846153846154, "status": 0.75, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1087}, "ledger_by_hop": {"1": {"facts_total": 63, "facts_retained": 27, "retention_rate": 0.42857142857142855, "retention_by_kind": {"person": 0.5, "date": 0.125, "amount": 0.6153846153846154, "status": 0.75, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1087}}}, "reference_key_id": "keys/synthetic/syn_03027_2000.json", "task_scores": {"correct": 0.0, "abstained": 0.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "1f88f0e78c63504e418dea843641b99a98555e15d47b02cd5cb2f57da3003dea", "output_hash": "2044d7c4afd32892ac22705680e35174db83308c139b02a1494275cc5e24aff3"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "c7c9cca67d997a388d204358", "attempt_id": "d513539d765d", "run_group_id": "E4_core:A4|closed|qwen3b-llama3b:h1", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:32:48.444654Z", "task_family": "synthetic", "document_id": "syn_03027_2000", "document_hash": "10326080979f71b625e4110c7c19767e5a752e267de5034f7e37d98cb6e06249", "question_id": "q12_unit_conversion", "split": "test", "cluster_id": "syn_03027_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 1, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1955, "query_tokens": 603, "message_tokens": 512, "message_utf8_bytes": 1087, "cache_tensor_bytes": null, "serialized_payload_bytes": 1087, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 6872.302875039168, "hops": {"1": 6872.302875039168}, "receiver_prefill_ms": 226.81825002655387, "ttft_ms": 11.991707957349718, "decode_ms": 11.848459020256996}, "total_wall_ms": 7123.618375044316, "ttft_ms": 11.991707957349718, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 63, "facts_retained": 27, "retention_rate": 0.42857142857142855, "retention_by_kind": {"person": 0.5, "date": 0.125, "amount": 0.6153846153846154, "status": 0.75, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1087}, "ledger_by_hop": {"1": {"facts_total": 63, "facts_retained": 27, "retention_rate": 0.42857142857142855, "retention_by_kind": {"person": 0.5, "date": 0.125, "amount": 0.6153846153846154, "status": 0.75, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1087}}}, "reference_key_id": "keys/synthetic/syn_03027_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "1f88f0e78c63504e418dea843641b99a98555e15d47b02cd5cb2f57da3003dea", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "55e681c17db92b0fcb649758", "attempt_id": "93a9f185f633", "run_group_id": "E4_core:A4|closed|qwen3b-llama3b:h1", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:32:48.712735Z", "task_family": "synthetic", "document_id": "syn_03027_2000", "document_hash": "10326080979f71b625e4110c7c19767e5a752e267de5034f7e37d98cb6e06249", "question_id": "q13_unanswerable", "split": "test", "cluster_id": "syn_03027_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 1, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1955, "query_tokens": 606, "message_tokens": 512, "message_utf8_bytes": 1087, "cache_tensor_bytes": null, "serialized_payload_bytes": 1087, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 6872.302875039168, "hops": {"1": 6872.302875039168}, "receiver_prefill_ms": 226.541250012815, "ttft_ms": 12.040166067890823, "decode_ms": 11.896249954588711}, "total_wall_ms": 7123.438459006138, "ttft_ms": 12.040166067890823, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 63, "facts_retained": 27, "retention_rate": 0.42857142857142855, "retention_by_kind": {"person": 0.5, "date": 0.125, "amount": 0.6153846153846154, "status": 0.75, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1087}, "ledger_by_hop": {"1": {"facts_total": 63, "facts_retained": 27, "retention_rate": 0.42857142857142855, "retention_by_kind": {"person": 0.5, "date": 0.125, "amount": 0.6153846153846154, "status": 0.75, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1087}}}, "reference_key_id": "keys/synthetic/syn_03027_2000.json", "task_scores": {"correct": 1.0, "abstained": 1.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "1f88f0e78c63504e418dea843641b99a98555e15d47b02cd5cb2f57da3003dea", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "3a2c56a8ebbc43adf3d95555", "attempt_id": "3e06879f4692", "run_group_id": "E4_core:A4|closed|qwen3b-llama3b:h1", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:32:48.980309Z", "task_family": "synthetic", "document_id": "syn_03027_2000", "document_hash": "10326080979f71b625e4110c7c19767e5a752e267de5034f7e37d98cb6e06249", "question_id": "q14_unanswerable", "split": "test", "cluster_id": "syn_03027_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 1, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1955, "query_tokens": 608, "message_tokens": 512, "message_utf8_bytes": 1087, "cache_tensor_bytes": null, "serialized_payload_bytes": 1087, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 6872.302875039168, "hops": {"1": 6872.302875039168}, "receiver_prefill_ms": 227.02700004447252, "ttft_ms": 11.992000043392181, "decode_ms": 11.835499899461865}, "total_wall_ms": 7123.790000099689, "ttft_ms": 11.992000043392181, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 63, "facts_retained": 27, "retention_rate": 0.42857142857142855, "retention_by_kind": {"person": 0.5, "date": 0.125, "amount": 0.6153846153846154, "status": 0.75, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1087}, "ledger_by_hop": {"1": {"facts_total": 63, "facts_retained": 27, "retention_rate": 0.42857142857142855, "retention_by_kind": {"person": 0.5, "date": 0.125, "amount": 0.6153846153846154, "status": 0.75, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1087}}}, "reference_key_id": "keys/synthetic/syn_03027_2000.json", "task_scores": {"correct": 1.0, "abstained": 1.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "1f88f0e78c63504e418dea843641b99a98555e15d47b02cd5cb2f57da3003dea", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "292f2d35ea7fc966ba58c3f3", "attempt_id": "5b0fd0a05039", "run_group_id": "E4_core:A4|closed|qwen3b-llama3b:h2", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:32:54.292130Z", "task_family": "synthetic", "document_id": "syn_03027_2000", "document_hash": "10326080979f71b625e4110c7c19767e5a752e267de5034f7e37d98cb6e06249", "question_id": "q00_single_fact", "split": "test", "cluster_id": "syn_03027_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "tokenizer_hashes": {"receiver": "b9d8872cf3a99da99e1fc4b0f71232a451632dc49ca4c6771549b4efa5c1c763"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 2, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1760, "query_tokens": 519, "message_tokens": 416, "message_utf8_bytes": 1063, "cache_tensor_bytes": null, "serialized_payload_bytes": 1063, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 11915.002124966122, "hops": {"1": 6872.302875039168, "2": 5042.6992499269545}, "receiver_prefill_ms": 209.1570000629872, "ttft_ms": 11.686292011290789, "decode_ms": 11.701290961354971}, "total_wall_ms": 12148.171665961854, "ttft_ms": 11.686292011290789, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 63, "facts_retained": 24, "retention_rate": 0.38095238095238093, "retention_by_kind": {"person": 0.5, "date": 0.125, "amount": 0.6153846153846154, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1063}, "ledger_by_hop": {"1": {"facts_total": 63, "facts_retained": 27, "retention_rate": 0.42857142857142855, "retention_by_kind": {"person": 0.5, "date": 0.125, "amount": 0.6153846153846154, "status": 0.75, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1087}, "2": {"facts_total": 63, "facts_retained": 24, "retention_rate": 0.38095238095238093, "retention_by_kind": {"person": 0.5, "date": 0.125, "amount": 0.6153846153846154, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1063}}}, "reference_key_id": "keys/synthetic/syn_03027_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "a84a1aff1e9050ab45ed7ed2bd9f5306446a866f8b9ff0cedf4ce4f983883642", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "429d9dd81e60b3d3340970fe", "attempt_id": "c3fe10eaf7c7", "run_group_id": "E4_core:A4|closed|qwen3b-llama3b:h2", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:32:54.542884Z", "task_family": "synthetic", "document_id": "syn_03027_2000", "document_hash": "10326080979f71b625e4110c7c19767e5a752e267de5034f7e37d98cb6e06249", "question_id": "q01_single_fact", "split": "test", "cluster_id": "syn_03027_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "tokenizer_hashes": {"receiver": "b9d8872cf3a99da99e1fc4b0f71232a451632dc49ca4c6771549b4efa5c1c763"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 2, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1760, "query_tokens": 526, "message_tokens": 416, "message_utf8_bytes": 1063, "cache_tensor_bytes": null, "serialized_payload_bytes": 1063, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 11915.002124966122, "hops": {"1": 6872.302875039168, "2": 5042.6992499269545}, "receiver_prefill_ms": 209.60116700734943, "ttft_ms": 11.759374989196658, "decode_ms": 114.80199999641627}, "total_wall_ms": 12251.792415976524, "ttft_ms": 11.759374989196658, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 10}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "Lot 247-S, Baie-Comeau", "evidence_artifact": {"ledger": {"facts_total": 63, "facts_retained": 24, "retention_rate": 0.38095238095238093, "retention_by_kind": {"person": 0.5, "date": 0.125, "amount": 0.6153846153846154, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1063}, "ledger_by_hop": {"1": {"facts_total": 63, "facts_retained": 27, "retention_rate": 0.42857142857142855, "retention_by_kind": {"person": 0.5, "date": 0.125, "amount": 0.6153846153846154, "status": 0.75, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1087}, "2": {"facts_total": 63, "facts_retained": 24, "retention_rate": 0.38095238095238093, "retention_by_kind": {"person": 0.5, "date": 0.125, "amount": 0.6153846153846154, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1063}}}, "reference_key_id": "keys/synthetic/syn_03027_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 0.8571428571428571}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "a84a1aff1e9050ab45ed7ed2bd9f5306446a866f8b9ff0cedf4ce4f983883642", "output_hash": "77b7da1d43e7d8815e9e9159ebbd735045bc7f668260ba30b93674f70622b8ec"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "3afedd96345ac9c9ea4879fa", "attempt_id": "e8c232196903", "run_group_id": "E4_core:A4|closed|qwen3b-llama3b:h2", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:32:54.896436Z", "task_family": "synthetic", "document_id": "syn_03027_2000", "document_hash": "10326080979f71b625e4110c7c19767e5a752e267de5034f7e37d98cb6e06249", "question_id": "q02_exact_copy", "split": "test", "cluster_id": "syn_03027_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "tokenizer_hashes": {"receiver": "b9d8872cf3a99da99e1fc4b0f71232a451632dc49ca4c6771549b4efa5c1c763"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 2, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1760, "query_tokens": 525, "message_tokens": 416, "message_utf8_bytes": 1063, "cache_tensor_bytes": null, "serialized_payload_bytes": 1063, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 11915.002124966122, "hops": {"1": 6872.302875039168, "2": 5042.6992499269545}, "receiver_prefill_ms": 209.69383302144706, "ttft_ms": 11.725250049494207, "decode_ms": 45.76870799064636}, "total_wall_ms": 12182.807583012618, "ttft_ms": 11.725250049494207, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 4}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "15.0054", "evidence_artifact": {"ledger": {"facts_total": 63, "facts_retained": 24, "retention_rate": 0.38095238095238093, "retention_by_kind": {"person": 0.5, "date": 0.125, "amount": 0.6153846153846154, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1063}, "ledger_by_hop": {"1": {"facts_total": 63, "facts_retained": 27, "retention_rate": 0.42857142857142855, "retention_by_kind": {"person": 0.5, "date": 0.125, "amount": 0.6153846153846154, "status": 0.75, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1087}, "2": {"facts_total": 63, "facts_retained": 24, "retention_rate": 0.38095238095238093, "retention_by_kind": {"person": 0.5, "date": 0.125, "amount": 0.6153846153846154, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1063}}}, "reference_key_id": "keys/synthetic/syn_03027_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "a84a1aff1e9050ab45ed7ed2bd9f5306446a866f8b9ff0cedf4ce4f983883642", "output_hash": "d13adba73068c847c850237059e0202fc5b91ccacf2e792e9ce5e3b90556501a"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "ec7e80efbd61a7c2cc8f41a2", "attempt_id": "5ae3f837d63c", "run_group_id": "E4_core:A4|closed|qwen3b-llama3b:h2", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:32:55.180655Z", "task_family": "synthetic", "document_id": "syn_03027_2000", "document_hash": "10326080979f71b625e4110c7c19767e5a752e267de5034f7e37d98cb6e06249", "question_id": "q03_exact_copy", "split": "test", "cluster_id": "syn_03027_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "tokenizer_hashes": {"receiver": "b9d8872cf3a99da99e1fc4b0f71232a451632dc49ca4c6771549b4efa5c1c763"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 2, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1760, "query_tokens": 521, "message_tokens": 416, "message_utf8_bytes": 1063, "cache_tensor_bytes": null, "serialized_payload_bytes": 1063, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 11915.002124966122, "hops": {"1": 6872.302875039168, "2": 5042.6992499269545}, "receiver_prefill_ms": 209.19991703704, "ttft_ms": 11.640792014077306, "decode_ms": 103.30366692505777}, "total_wall_ms": 12239.772083004937, "ttft_ms": 11.640792014077306, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 9}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "Priyamvada Sørlie", "evidence_artifact": {"ledger": {"facts_total": 63, "facts_retained": 24, "retention_rate": 0.38095238095238093, "retention_by_kind": {"person": 0.5, "date": 0.125, "amount": 0.6153846153846154, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1063}, "ledger_by_hop": {"1": {"facts_total": 63, "facts_retained": 27, "retention_rate": 0.42857142857142855, "retention_by_kind": {"person": 0.5, "date": 0.125, "amount": 0.6153846153846154, "status": 0.75, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1087}, "2": {"facts_total": 63, "facts_retained": 24, "retention_rate": 0.38095238095238093, "retention_by_kind": {"person": 0.5, "date": 0.125, "amount": 0.6153846153846154, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1063}}}, "reference_key_id": "keys/synthetic/syn_03027_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "a84a1aff1e9050ab45ed7ed2bd9f5306446a866f8b9ff0cedf4ce4f983883642", "output_hash": "c5f554196ed23dd71ad995ec89f7142099d524dbb6534cc10d94a0c89d13e792"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "4bbc2899b863a572a5b9185e", "attempt_id": "02c13bf379fd", "run_group_id": "E4_core:A4|closed|qwen3b-llama3b:h2", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:32:55.521554Z", "task_family": "synthetic", "document_id": "syn_03027_2000", "document_hash": "10326080979f71b625e4110c7c19767e5a752e267de5034f7e37d98cb6e06249", "question_id": "q04_multi_hop", "split": "test", "cluster_id": "syn_03027_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "tokenizer_hashes": {"receiver": "b9d8872cf3a99da99e1fc4b0f71232a451632dc49ca4c6771549b4efa5c1c763"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 2, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1760, "query_tokens": 532, "message_tokens": 416, "message_utf8_bytes": 1063, "cache_tensor_bytes": null, "serialized_payload_bytes": 1063, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 11915.002124966122, "hops": {"1": 6872.302875039168, "2": 5042.6992499269545}, "receiver_prefill_ms": 210.0380000192672, "ttft_ms": 11.629624990746379, "decode_ms": 11.488749994896352}, "total_wall_ms": 12148.786542005837, "ttft_ms": 11.629624990746379, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 63, "facts_retained": 24, "retention_rate": 0.38095238095238093, "retention_by_kind": {"person": 0.5, "date": 0.125, "amount": 0.6153846153846154, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1063}, "ledger_by_hop": {"1": {"facts_total": 63, "facts_retained": 27, "retention_rate": 0.42857142857142855, "retention_by_kind": {"person": 0.5, "date": 0.125, "amount": 0.6153846153846154, "status": 0.75, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1087}, "2": {"facts_total": 63, "facts_retained": 24, "retention_rate": 0.38095238095238093, "retention_by_kind": {"person": 0.5, "date": 0.125, "amount": 0.6153846153846154, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1063}}}, "reference_key_id": "keys/synthetic/syn_03027_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "a84a1aff1e9050ab45ed7ed2bd9f5306446a866f8b9ff0cedf4ce4f983883642", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "ca5a2d032fdbe4fb5cd4d564", "attempt_id": "c422f312da6d", "run_group_id": "E4_core:A4|closed|qwen3b-llama3b:h2", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:32:55.771576Z", "task_family": "synthetic", "document_id": "syn_03027_2000", "document_hash": "10326080979f71b625e4110c7c19767e5a752e267de5034f7e37d98cb6e06249", "question_id": "q05_multi_hop", "split": "test", "cluster_id": "syn_03027_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "tokenizer_hashes": {"receiver": "b9d8872cf3a99da99e1fc4b0f71232a451632dc49ca4c6771549b4efa5c1c763"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 2, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1760, "query_tokens": 527, "message_tokens": 416, "message_utf8_bytes": 1063, "cache_tensor_bytes": null, "serialized_payload_bytes": 1063, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 11915.002124966122, "hops": {"1": 6872.302875039168, "2": 5042.6992499269545}, "receiver_prefill_ms": 209.37150006648153, "ttft_ms": 11.71729201450944, "decode_ms": 11.523833964020014}, "total_wall_ms": 12148.233084008098, "ttft_ms": 11.71729201450944, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 63, "facts_retained": 24, "retention_rate": 0.38095238095238093, "retention_by_kind": {"person": 0.5, "date": 0.125, "amount": 0.6153846153846154, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1063}, "ledger_by_hop": {"1": {"facts_total": 63, "facts_retained": 27, "retention_rate": 0.42857142857142855, "retention_by_kind": {"person": 0.5, "date": 0.125, "amount": 0.6153846153846154, "status": 0.75, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1087}, "2": {"facts_total": 63, "facts_retained": 24, "retention_rate": 0.38095238095238093, "retention_by_kind": {"person": 0.5, "date": 0.125, "amount": 0.6153846153846154, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1063}}}, "reference_key_id": "keys/synthetic/syn_03027_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "a84a1aff1e9050ab45ed7ed2bd9f5306446a866f8b9ff0cedf4ce4f983883642", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "751968142b8b3167a2b7f5d4", "attempt_id": "bb63684a4bb5", "run_group_id": "E4_core:A4|closed|qwen3b-llama3b:h2", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:32:56.021535Z", "task_family": "synthetic", "document_id": "syn_03027_2000", "document_hash": "10326080979f71b625e4110c7c19767e5a752e267de5034f7e37d98cb6e06249", "question_id": "q06_negation", "split": "test", "cluster_id": "syn_03027_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "tokenizer_hashes": {"receiver": "b9d8872cf3a99da99e1fc4b0f71232a451632dc49ca4c6771549b4efa5c1c763"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 2, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1760, "query_tokens": 521, "message_tokens": 416, "message_utf8_bytes": 1063, "cache_tensor_bytes": null, "serialized_payload_bytes": 1063, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 11915.002124966122, "hops": {"1": 6872.302875039168, "2": 5042.6992499269545}, "receiver_prefill_ms": 209.30079196114093, "ttft_ms": 11.983166914433241, "decode_ms": 22.998959058895707}, "total_wall_ms": 12159.89487490151, "ttft_ms": 11.983166914433241, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 2}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "No.", "evidence_artifact": {"ledger": {"facts_total": 63, "facts_retained": 24, "retention_rate": 0.38095238095238093, "retention_by_kind": {"person": 0.5, "date": 0.125, "amount": 0.6153846153846154, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1063}, "ledger_by_hop": {"1": {"facts_total": 63, "facts_retained": 27, "retention_rate": 0.42857142857142855, "retention_by_kind": {"person": 0.5, "date": 0.125, "amount": 0.6153846153846154, "status": 0.75, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1087}, "2": {"facts_total": 63, "facts_retained": 24, "retention_rate": 0.38095238095238093, "retention_by_kind": {"person": 0.5, "date": 0.125, "amount": 0.6153846153846154, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1063}}}, "reference_key_id": "keys/synthetic/syn_03027_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "a84a1aff1e9050ab45ed7ed2bd9f5306446a866f8b9ff0cedf4ce4f983883642", "output_hash": "38178a20b470cfd18299fb1593dd4ba706f1a83321dac051d01086efd8b7a96f"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "d3935d5524047cfa8f9b2bf0", "attempt_id": "f11c6fb95282", "run_group_id": "E4_core:A4|closed|qwen3b-llama3b:h2", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:32:56.282596Z", "task_family": "synthetic", "document_id": "syn_03027_2000", "document_hash": "10326080979f71b625e4110c7c19767e5a752e267de5034f7e37d98cb6e06249", "question_id": "q07_negation", "split": "test", "cluster_id": "syn_03027_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "tokenizer_hashes": {"receiver": "b9d8872cf3a99da99e1fc4b0f71232a451632dc49ca4c6771549b4efa5c1c763"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 2, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1760, "query_tokens": 521, "message_tokens": 416, "message_utf8_bytes": 1063, "cache_tensor_bytes": null, "serialized_payload_bytes": 1063, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 11915.002124966122, "hops": {"1": 6872.302875039168, "2": 5042.6992499269545}, "receiver_prefill_ms": 209.30437499191612, "ttft_ms": 11.904875049367547, "decode_ms": 22.978166001848876}, "total_wall_ms": 12159.794874954969, "ttft_ms": 11.904875049367547, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 2}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "No.", "evidence_artifact": {"ledger": {"facts_total": 63, "facts_retained": 24, "retention_rate": 0.38095238095238093, "retention_by_kind": {"person": 0.5, "date": 0.125, "amount": 0.6153846153846154, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1063}, "ledger_by_hop": {"1": {"facts_total": 63, "facts_retained": 27, "retention_rate": 0.42857142857142855, "retention_by_kind": {"person": 0.5, "date": 0.125, "amount": 0.6153846153846154, "status": 0.75, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1087}, "2": {"facts_total": 63, "facts_retained": 24, "retention_rate": 0.38095238095238093, "retention_by_kind": {"person": 0.5, "date": 0.125, "amount": 0.6153846153846154, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1063}}}, "reference_key_id": "keys/synthetic/syn_03027_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "a84a1aff1e9050ab45ed7ed2bd9f5306446a866f8b9ff0cedf4ce4f983883642", "output_hash": "38178a20b470cfd18299fb1593dd4ba706f1a83321dac051d01086efd8b7a96f"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "72e3acf0e01234d3aeab6cee", "attempt_id": "99801e1fe749", "run_group_id": "E4_core:A4|closed|qwen3b-llama3b:h2", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:32:56.544602Z", "task_family": "synthetic", "document_id": "syn_03027_2000", "document_hash": "10326080979f71b625e4110c7c19767e5a752e267de5034f7e37d98cb6e06249", "question_id": "q08_revision", "split": "test", "cluster_id": "syn_03027_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "tokenizer_hashes": {"receiver": "b9d8872cf3a99da99e1fc4b0f71232a451632dc49ca4c6771549b4efa5c1c763"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 2, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1760, "query_tokens": 536, "message_tokens": 416, "message_utf8_bytes": 1063, "cache_tensor_bytes": null, "serialized_payload_bytes": 1063, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 11915.002124966122, "hops": {"1": 6872.302875039168, "2": 5042.6992499269545}, "receiver_prefill_ms": 209.90879100281745, "ttft_ms": 11.64100004825741, "decode_ms": 57.76754196267575}, "total_wall_ms": 12194.930416997522, "ttft_ms": 11.64100004825741, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 5}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "8,010,205", "evidence_artifact": {"ledger": {"facts_total": 63, "facts_retained": 24, "retention_rate": 0.38095238095238093, "retention_by_kind": {"person": 0.5, "date": 0.125, "amount": 0.6153846153846154, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1063}, "ledger_by_hop": {"1": {"facts_total": 63, "facts_retained": 27, "retention_rate": 0.42857142857142855, "retention_by_kind": {"person": 0.5, "date": 0.125, "amount": 0.6153846153846154, "status": 0.75, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1087}, "2": {"facts_total": 63, "facts_retained": 24, "retention_rate": 0.38095238095238093, "retention_by_kind": {"person": 0.5, "date": 0.125, "amount": 0.6153846153846154, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1063}}}, "reference_key_id": "keys/synthetic/syn_03027_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "a84a1aff1e9050ab45ed7ed2bd9f5306446a866f8b9ff0cedf4ce4f983883642", "output_hash": "12facbb746e46f5872a48eea64f64dc2cbd9610b1548522078108253d03a7904"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "780c38ae1e7a859fb34e6815", "attempt_id": "1280f2628e32", "run_group_id": "E4_core:A4|closed|qwen3b-llama3b:h2", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:32:56.840885Z", "task_family": "synthetic", "document_id": "syn_03027_2000", "document_hash": "10326080979f71b625e4110c7c19767e5a752e267de5034f7e37d98cb6e06249", "question_id": "q09_revision", "split": "test", "cluster_id": "syn_03027_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "tokenizer_hashes": {"receiver": "b9d8872cf3a99da99e1fc4b0f71232a451632dc49ca4c6771549b4efa5c1c763"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 2, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1760, "query_tokens": 529, "message_tokens": 416, "message_utf8_bytes": 1063, "cache_tensor_bytes": null, "serialized_payload_bytes": 1063, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 11915.002124966122, "hops": {"1": 6872.302875039168, "2": 5042.6992499269545}, "receiver_prefill_ms": 209.10687500145286, "ttft_ms": 11.756792082451284, "decode_ms": 57.42233304772526}, "total_wall_ms": 12193.888499983586, "ttft_ms": 11.756792082451284, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 5}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "6,056,150", "evidence_artifact": {"ledger": {"facts_total": 63, "facts_retained": 24, "retention_rate": 0.38095238095238093, "retention_by_kind": {"person": 0.5, "date": 0.125, "amount": 0.6153846153846154, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1063}, "ledger_by_hop": {"1": {"facts_total": 63, "facts_retained": 27, "retention_rate": 0.42857142857142855, "retention_by_kind": {"person": 0.5, "date": 0.125, "amount": 0.6153846153846154, "status": 0.75, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1087}, "2": {"facts_total": 63, "facts_retained": 24, "retention_rate": 0.38095238095238093, "retention_by_kind": {"person": 0.5, "date": 0.125, "amount": 0.6153846153846154, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1063}}}, "reference_key_id": "keys/synthetic/syn_03027_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "a84a1aff1e9050ab45ed7ed2bd9f5306446a866f8b9ff0cedf4ce4f983883642", "output_hash": "8262889a8478c7e0ee965f159d9849a3849b412e276f5b979476e2da5b320249"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "be639b2a148fb27410da43bb", "attempt_id": "5f0e4c640208", "run_group_id": "E4_core:A4|closed|qwen3b-llama3b:h2", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:32:57.136256Z", "task_family": "synthetic", "document_id": "syn_03027_2000", "document_hash": "10326080979f71b625e4110c7c19767e5a752e267de5034f7e37d98cb6e06249", "question_id": "q10_temporal", "split": "test", "cluster_id": "syn_03027_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "tokenizer_hashes": {"receiver": "b9d8872cf3a99da99e1fc4b0f71232a451632dc49ca4c6771549b4efa5c1c763"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 2, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1760, "query_tokens": 537, "message_tokens": 416, "message_utf8_bytes": 1063, "cache_tensor_bytes": null, "serialized_payload_bytes": 1063, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 11915.002124966122, "hops": {"1": 6872.302875039168, "2": 5042.6992499269545}, "receiver_prefill_ms": 210.32570803072304, "ttft_ms": 11.808958020992577, "decode_ms": 11.568875052034855}, "total_wall_ms": 12149.332833010703, "ttft_ms": 11.808958020992577, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 63, "facts_retained": 24, "retention_rate": 0.38095238095238093, "retention_by_kind": {"person": 0.5, "date": 0.125, "amount": 0.6153846153846154, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1063}, "ledger_by_hop": {"1": {"facts_total": 63, "facts_retained": 27, "retention_rate": 0.42857142857142855, "retention_by_kind": {"person": 0.5, "date": 0.125, "amount": 0.6153846153846154, "status": 0.75, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1087}, "2": {"facts_total": 63, "facts_retained": 24, "retention_rate": 0.38095238095238093, "retention_by_kind": {"person": 0.5, "date": 0.125, "amount": 0.6153846153846154, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1063}}}, "reference_key_id": "keys/synthetic/syn_03027_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "a84a1aff1e9050ab45ed7ed2bd9f5306446a866f8b9ff0cedf4ce4f983883642", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "ee4d154591dcac9396307c9e", "attempt_id": "2c4ed2618709", "run_group_id": "E4_core:A4|closed|qwen3b-llama3b:h2", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:32:57.386562Z", "task_family": "synthetic", "document_id": "syn_03027_2000", "document_hash": "10326080979f71b625e4110c7c19767e5a752e267de5034f7e37d98cb6e06249", "question_id": "q11_aggregation", "split": "test", "cluster_id": "syn_03027_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "tokenizer_hashes": {"receiver": "b9d8872cf3a99da99e1fc4b0f71232a451632dc49ca4c6771549b4efa5c1c763"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 2, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1760, "query_tokens": 532, "message_tokens": 416, "message_utf8_bytes": 1063, "cache_tensor_bytes": null, "serialized_payload_bytes": 1063, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 11915.002124966122, "hops": {"1": 6872.302875039168, "2": 5042.6992499269545}, "receiver_prefill_ms": 210.01262497156858, "ttft_ms": 11.720290989615023, "decode_ms": 298.07845898903906}, "total_wall_ms": 12435.455665923655, "ttft_ms": 11.720290989615023, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 26}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "6,914,937 + 1,787,610 + 8,278,225 = 16,980,772", "evidence_artifact": {"ledger": {"facts_total": 63, "facts_retained": 24, "retention_rate": 0.38095238095238093, "retention_by_kind": {"person": 0.5, "date": 0.125, "amount": 0.6153846153846154, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1063}, "ledger_by_hop": {"1": {"facts_total": 63, "facts_retained": 27, "retention_rate": 0.42857142857142855, "retention_by_kind": {"person": 0.5, "date": 0.125, "amount": 0.6153846153846154, "status": 0.75, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1087}, "2": {"facts_total": 63, "facts_retained": 24, "retention_rate": 0.38095238095238093, "retention_by_kind": {"person": 0.5, "date": 0.125, "amount": 0.6153846153846154, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1063}}}, "reference_key_id": "keys/synthetic/syn_03027_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 0.4}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "a84a1aff1e9050ab45ed7ed2bd9f5306446a866f8b9ff0cedf4ce4f983883642", "output_hash": "c5fb5d9b76b7a02439f0acb83233430b95eef4aa96f8df88b9152c3b3e51343c"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "8e0eb7bbee465867fbabf295", "attempt_id": "524a6272f179", "run_group_id": "E4_core:A4|closed|qwen3b-llama3b:h2", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:32:57.924621Z", "task_family": "synthetic", "document_id": "syn_03027_2000", "document_hash": "10326080979f71b625e4110c7c19767e5a752e267de5034f7e37d98cb6e06249", "question_id": "q12_unit_conversion", "split": "test", "cluster_id": "syn_03027_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "tokenizer_hashes": {"receiver": "b9d8872cf3a99da99e1fc4b0f71232a451632dc49ca4c6771549b4efa5c1c763"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 2, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1760, "query_tokens": 528, "message_tokens": 416, "message_utf8_bytes": 1063, "cache_tensor_bytes": null, "serialized_payload_bytes": 1063, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 11915.002124966122, "hops": {"1": 6872.302875039168, "2": 5042.6992499269545}, "receiver_prefill_ms": 209.4906660495326, "ttft_ms": 11.656083050183952, "decode_ms": 11.460999958217144}, "total_wall_ms": 12148.191374959424, "ttft_ms": 11.656083050183952, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 63, "facts_retained": 24, "retention_rate": 0.38095238095238093, "retention_by_kind": {"person": 0.5, "date": 0.125, "amount": 0.6153846153846154, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1063}, "ledger_by_hop": {"1": {"facts_total": 63, "facts_retained": 27, "retention_rate": 0.42857142857142855, "retention_by_kind": {"person": 0.5, "date": 0.125, "amount": 0.6153846153846154, "status": 0.75, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1087}, "2": {"facts_total": 63, "facts_retained": 24, "retention_rate": 0.38095238095238093, "retention_by_kind": {"person": 0.5, "date": 0.125, "amount": 0.6153846153846154, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1063}}}, "reference_key_id": "keys/synthetic/syn_03027_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "a84a1aff1e9050ab45ed7ed2bd9f5306446a866f8b9ff0cedf4ce4f983883642", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "83b00003e3fa4b248f69e837", "attempt_id": "d716a71b2147", "run_group_id": "E4_core:A4|closed|qwen3b-llama3b:h2", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:32:58.174663Z", "task_family": "synthetic", "document_id": "syn_03027_2000", "document_hash": "10326080979f71b625e4110c7c19767e5a752e267de5034f7e37d98cb6e06249", "question_id": "q13_unanswerable", "split": "test", "cluster_id": "syn_03027_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "tokenizer_hashes": {"receiver": "b9d8872cf3a99da99e1fc4b0f71232a451632dc49ca4c6771549b4efa5c1c763"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 2, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1760, "query_tokens": 532, "message_tokens": 416, "message_utf8_bytes": 1063, "cache_tensor_bytes": null, "serialized_payload_bytes": 1063, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 11915.002124966122, "hops": {"1": 6872.302875039168, "2": 5042.6992499269545}, "receiver_prefill_ms": 209.80624994263053, "ttft_ms": 11.868958012200892, "decode_ms": 23.161166929639876}, "total_wall_ms": 12160.468666930683, "ttft_ms": 11.868958012200892, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 2}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "Unknown.", "evidence_artifact": {"ledger": {"facts_total": 63, "facts_retained": 24, "retention_rate": 0.38095238095238093, "retention_by_kind": {"person": 0.5, "date": 0.125, "amount": 0.6153846153846154, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1063}, "ledger_by_hop": {"1": {"facts_total": 63, "facts_retained": 27, "retention_rate": 0.42857142857142855, "retention_by_kind": {"person": 0.5, "date": 0.125, "amount": 0.6153846153846154, "status": 0.75, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1087}, "2": {"facts_total": 63, "facts_retained": 24, "retention_rate": 0.38095238095238093, "retention_by_kind": {"person": 0.5, "date": 0.125, "amount": 0.6153846153846154, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1063}}}, "reference_key_id": "keys/synthetic/syn_03027_2000.json", "task_scores": {"correct": 1.0, "abstained": 1.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "a84a1aff1e9050ab45ed7ed2bd9f5306446a866f8b9ff0cedf4ce4f983883642", "output_hash": "9cf9f50925b2a7803c73826a305235d4215dc428721c731072a9f2c0f5d6ae5a"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "db09be4d7a69be2eb46431c3", "attempt_id": "f2e89c99e080", "run_group_id": "E4_core:A4|closed|qwen3b-llama3b:h2", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:32:58.436190Z", "task_family": "synthetic", "document_id": "syn_03027_2000", "document_hash": "10326080979f71b625e4110c7c19767e5a752e267de5034f7e37d98cb6e06249", "question_id": "q14_unanswerable", "split": "test", "cluster_id": "syn_03027_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "tokenizer_hashes": {"receiver": "b9d8872cf3a99da99e1fc4b0f71232a451632dc49ca4c6771549b4efa5c1c763"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 2, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1760, "query_tokens": 534, "message_tokens": 416, "message_utf8_bytes": 1063, "cache_tensor_bytes": null, "serialized_payload_bytes": 1063, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 11915.002124966122, "hops": {"1": 6872.302875039168, "2": 5042.6992499269545}, "receiver_prefill_ms": 209.62866698391736, "ttft_ms": 11.73908298369497, "decode_ms": 11.487790965475142}, "total_wall_ms": 12148.470208980143, "ttft_ms": 11.73908298369497, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 63, "facts_retained": 24, "retention_rate": 0.38095238095238093, "retention_by_kind": {"person": 0.5, "date": 0.125, "amount": 0.6153846153846154, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1063}, "ledger_by_hop": {"1": {"facts_total": 63, "facts_retained": 27, "retention_rate": 0.42857142857142855, "retention_by_kind": {"person": 0.5, "date": 0.125, "amount": 0.6153846153846154, "status": 0.75, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1087}, "2": {"facts_total": 63, "facts_retained": 24, "retention_rate": 0.38095238095238093, "retention_by_kind": {"person": 0.5, "date": 0.125, "amount": 0.6153846153846154, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1063}}}, "reference_key_id": "keys/synthetic/syn_03027_2000.json", "task_scores": {"correct": 1.0, "abstained": 1.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "a84a1aff1e9050ab45ed7ed2bd9f5306446a866f8b9ff0cedf4ce4f983883642", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "dfabe703c66f3f8e1bd3d719", "attempt_id": "4861d6afec1b", "run_group_id": "E4_core:A4|closed|qwen3b-llama3b:h4", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:33:09.758074Z", "task_family": "synthetic", "document_id": "syn_03027_2000", "document_hash": "10326080979f71b625e4110c7c19767e5a752e267de5034f7e37d98cb6e06249", "question_id": "q00_single_fact", "split": "test", "cluster_id": "syn_03027_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "tokenizer_hashes": {"receiver": "b9d8872cf3a99da99e1fc4b0f71232a451632dc49ca4c6771549b4efa5c1c763"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 4, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1760, "query_tokens": 490, "message_tokens": 387, "message_utf8_bytes": 1023, "cache_tensor_bytes": null, "serialized_payload_bytes": 1023, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 22983.29725000076, "hops": {"1": 6872.302875039168, "2": 5042.6992499269545, "3": 6354.470000020228, "4": 4713.825125014409}, "receiver_prefill_ms": 206.24066703021526, "ttft_ms": 11.743957991711795, "decode_ms": 11.539915925823152}, "total_wall_ms": 23213.434666977264, "ttft_ms": 11.743957991711795, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 63, "facts_retained": 24, "retention_rate": 0.38095238095238093, "retention_by_kind": {"person": 0.5, "date": 0.125, "amount": 0.6153846153846154, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1023}, "ledger_by_hop": {"1": {"facts_total": 63, "facts_retained": 27, "retention_rate": 0.42857142857142855, "retention_by_kind": {"person": 0.5, "date": 0.125, "amount": 0.6153846153846154, "status": 0.75, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1087}, "2": {"facts_total": 63, "facts_retained": 24, "retention_rate": 0.38095238095238093, "retention_by_kind": {"person": 0.5, "date": 0.125, "amount": 0.6153846153846154, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1063}, "3": {"facts_total": 63, "facts_retained": 24, "retention_rate": 0.38095238095238093, "retention_by_kind": {"person": 0.5, "date": 0.125, "amount": 0.6153846153846154, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1062}, "4": {"facts_total": 63, "facts_retained": 24, "retention_rate": 0.38095238095238093, "retention_by_kind": {"person": 0.5, "date": 0.125, "amount": 0.6153846153846154, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1023}}}, "reference_key_id": "keys/synthetic/syn_03027_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "9597ef7dd5408dec598b05726488fa7f8924bdb0755c45a0837fd9d429f51e78", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "e4e2f414adcf3890df45d844", "attempt_id": "79f5b35dc816", "run_group_id": "E4_core:A4|closed|qwen3b-llama3b:h4", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:33:10.004910Z", "task_family": "synthetic", "document_id": "syn_03027_2000", "document_hash": "10326080979f71b625e4110c7c19767e5a752e267de5034f7e37d98cb6e06249", "question_id": "q01_single_fact", "split": "test", "cluster_id": "syn_03027_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "tokenizer_hashes": {"receiver": "b9d8872cf3a99da99e1fc4b0f71232a451632dc49ca4c6771549b4efa5c1c763"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 4, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1760, "query_tokens": 497, "message_tokens": 387, "message_utf8_bytes": 1023, "cache_tensor_bytes": null, "serialized_payload_bytes": 1023, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 22983.29725000076, "hops": {"1": 6872.302875039168, "2": 5042.6992499269545, "3": 6354.470000020228, "4": 4713.825125014409}, "receiver_prefill_ms": 207.9006660496816, "ttft_ms": 11.733249993994832, "decode_ms": 115.8215410541743}, "total_wall_ms": 23319.396708044223, "ttft_ms": 11.733249993994832, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 10}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "Lot 247-S, Baie-Comeau", "evidence_artifact": {"ledger": {"facts_total": 63, "facts_retained": 24, "retention_rate": 0.38095238095238093, "retention_by_kind": {"person": 0.5, "date": 0.125, "amount": 0.6153846153846154, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1023}, "ledger_by_hop": {"1": {"facts_total": 63, "facts_retained": 27, "retention_rate": 0.42857142857142855, "retention_by_kind": {"person": 0.5, "date": 0.125, "amount": 0.6153846153846154, "status": 0.75, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1087}, "2": {"facts_total": 63, "facts_retained": 24, "retention_rate": 0.38095238095238093, "retention_by_kind": {"person": 0.5, "date": 0.125, "amount": 0.6153846153846154, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1063}, "3": {"facts_total": 63, "facts_retained": 24, "retention_rate": 0.38095238095238093, "retention_by_kind": {"person": 0.5, "date": 0.125, "amount": 0.6153846153846154, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1062}, "4": {"facts_total": 63, "facts_retained": 24, "retention_rate": 0.38095238095238093, "retention_by_kind": {"person": 0.5, "date": 0.125, "amount": 0.6153846153846154, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1023}}}, "reference_key_id": "keys/synthetic/syn_03027_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 0.8571428571428571}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "9597ef7dd5408dec598b05726488fa7f8924bdb0755c45a0837fd9d429f51e78", "output_hash": "77b7da1d43e7d8815e9e9159ebbd735045bc7f668260ba30b93674f70622b8ec"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "a13b47f56009315605b80552", "attempt_id": "f8ddb8dd3241", "run_group_id": "E4_core:A4|closed|qwen3b-llama3b:h4", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:33:10.357057Z", "task_family": "synthetic", "document_id": "syn_03027_2000", "document_hash": "10326080979f71b625e4110c7c19767e5a752e267de5034f7e37d98cb6e06249", "question_id": "q02_exact_copy", "split": "test", "cluster_id": "syn_03027_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "tokenizer_hashes": {"receiver": "b9d8872cf3a99da99e1fc4b0f71232a451632dc49ca4c6771549b4efa5c1c763"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 4, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1760, "query_tokens": 496, "message_tokens": 387, "message_utf8_bytes": 1023, "cache_tensor_bytes": null, "serialized_payload_bytes": 1023, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 22983.29725000076, "hops": {"1": 6872.302875039168, "2": 5042.6992499269545, "3": 6354.470000020228, "4": 4713.825125014409}, "receiver_prefill_ms": 205.87050006724894, "ttft_ms": 11.806832975707948, "decode_ms": 46.47175001446158}, "total_wall_ms": 23248.114499961957, "ttft_ms": 11.806832975707948, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 4}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "15.0054", "evidence_artifact": {"ledger": {"facts_total": 63, "facts_retained": 24, "retention_rate": 0.38095238095238093, "retention_by_kind": {"person": 0.5, "date": 0.125, "amount": 0.6153846153846154, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1023}, "ledger_by_hop": {"1": {"facts_total": 63, "facts_retained": 27, "retention_rate": 0.42857142857142855, "retention_by_kind": {"person": 0.5, "date": 0.125, "amount": 0.6153846153846154, "status": 0.75, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1087}, "2": {"facts_total": 63, "facts_retained": 24, "retention_rate": 0.38095238095238093, "retention_by_kind": {"person": 0.5, "date": 0.125, "amount": 0.6153846153846154, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1063}, "3": {"facts_total": 63, "facts_retained": 24, "retention_rate": 0.38095238095238093, "retention_by_kind": {"person": 0.5, "date": 0.125, "amount": 0.6153846153846154, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1062}, "4": {"facts_total": 63, "facts_retained": 24, "retention_rate": 0.38095238095238093, "retention_by_kind": {"person": 0.5, "date": 0.125, "amount": 0.6153846153846154, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1023}}}, "reference_key_id": "keys/synthetic/syn_03027_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "9597ef7dd5408dec598b05726488fa7f8924bdb0755c45a0837fd9d429f51e78", "output_hash": "d13adba73068c847c850237059e0202fc5b91ccacf2e792e9ce5e3b90556501a"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "8626319698b2c2156c13f0a7", "attempt_id": "0baf19023d48", "run_group_id": "E4_core:A4|closed|qwen3b-llama3b:h4", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:33:10.638756Z", "task_family": "synthetic", "document_id": "syn_03027_2000", "document_hash": "10326080979f71b625e4110c7c19767e5a752e267de5034f7e37d98cb6e06249", "question_id": "q03_exact_copy", "split": "test", "cluster_id": "syn_03027_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "tokenizer_hashes": {"receiver": "b9d8872cf3a99da99e1fc4b0f71232a451632dc49ca4c6771549b4efa5c1c763"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 4, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1760, "query_tokens": 492, "message_tokens": 387, "message_utf8_bytes": 1023, "cache_tensor_bytes": null, "serialized_payload_bytes": 1023, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 22983.29725000076, "hops": {"1": 6872.302875039168, "2": 5042.6992499269545, "3": 6354.470000020228, "4": 4713.825125014409}, "receiver_prefill_ms": 206.41062501817942, "ttft_ms": 11.7896250449121, "decode_ms": 104.55258307047188}, "total_wall_ms": 23306.675791041926, "ttft_ms": 11.7896250449121, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 9}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "Priyamvada Sørlie", "evidence_artifact": {"ledger": {"facts_total": 63, "facts_retained": 24, "retention_rate": 0.38095238095238093, "retention_by_kind": {"person": 0.5, "date": 0.125, "amount": 0.6153846153846154, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1023}, "ledger_by_hop": {"1": {"facts_total": 63, "facts_retained": 27, "retention_rate": 0.42857142857142855, "retention_by_kind": {"person": 0.5, "date": 0.125, "amount": 0.6153846153846154, "status": 0.75, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1087}, "2": {"facts_total": 63, "facts_retained": 24, "retention_rate": 0.38095238095238093, "retention_by_kind": {"person": 0.5, "date": 0.125, "amount": 0.6153846153846154, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1063}, "3": {"facts_total": 63, "facts_retained": 24, "retention_rate": 0.38095238095238093, "retention_by_kind": {"person": 0.5, "date": 0.125, "amount": 0.6153846153846154, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1062}, "4": {"facts_total": 63, "facts_retained": 24, "retention_rate": 0.38095238095238093, "retention_by_kind": {"person": 0.5, "date": 0.125, "amount": 0.6153846153846154, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1023}}}, "reference_key_id": "keys/synthetic/syn_03027_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "9597ef7dd5408dec598b05726488fa7f8924bdb0755c45a0837fd9d429f51e78", "output_hash": "c5f554196ed23dd71ad995ec89f7142099d524dbb6534cc10d94a0c89d13e792"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "a9c34da8bf935186ca88ad1b", "attempt_id": "14b5fe2801a6", "run_group_id": "E4_core:A4|closed|qwen3b-llama3b:h4", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:33:10.978972Z", "task_family": "synthetic", "document_id": "syn_03027_2000", "document_hash": "10326080979f71b625e4110c7c19767e5a752e267de5034f7e37d98cb6e06249", "question_id": "q04_multi_hop", "split": "test", "cluster_id": "syn_03027_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "tokenizer_hashes": {"receiver": "b9d8872cf3a99da99e1fc4b0f71232a451632dc49ca4c6771549b4efa5c1c763"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 4, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1760, "query_tokens": 503, "message_tokens": 387, "message_utf8_bytes": 1023, "cache_tensor_bytes": null, "serialized_payload_bytes": 1023, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 22983.29725000076, "hops": {"1": 6872.302875039168, "2": 5042.6992499269545, "3": 6354.470000020228, "4": 4713.825125014409}, "receiver_prefill_ms": 208.06566695682704, "ttft_ms": 11.687041958793998, "decode_ms": 11.644500074908137}, "total_wall_ms": 23215.29337496031, "ttft_ms": 11.687041958793998, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 63, "facts_retained": 24, "retention_rate": 0.38095238095238093, "retention_by_kind": {"person": 0.5, "date": 0.125, "amount": 0.6153846153846154, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1023}, "ledger_by_hop": {"1": {"facts_total": 63, "facts_retained": 27, "retention_rate": 0.42857142857142855, "retention_by_kind": {"person": 0.5, "date": 0.125, "amount": 0.6153846153846154, "status": 0.75, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1087}, "2": {"facts_total": 63, "facts_retained": 24, "retention_rate": 0.38095238095238093, "retention_by_kind": {"person": 0.5, "date": 0.125, "amount": 0.6153846153846154, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1063}, "3": {"facts_total": 63, "facts_retained": 24, "retention_rate": 0.38095238095238093, "retention_by_kind": {"person": 0.5, "date": 0.125, "amount": 0.6153846153846154, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1062}, "4": {"facts_total": 63, "facts_retained": 24, "retention_rate": 0.38095238095238093, "retention_by_kind": {"person": 0.5, "date": 0.125, "amount": 0.6153846153846154, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1023}}}, "reference_key_id": "keys/synthetic/syn_03027_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "9597ef7dd5408dec598b05726488fa7f8924bdb0755c45a0837fd9d429f51e78", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "9dafde273e18c4ee3e7069eb", "attempt_id": "50f610781441", "run_group_id": "E4_core:A4|closed|qwen3b-llama3b:h4", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:33:11.227879Z", "task_family": "synthetic", "document_id": "syn_03027_2000", "document_hash": "10326080979f71b625e4110c7c19767e5a752e267de5034f7e37d98cb6e06249", "question_id": "q05_multi_hop", "split": "test", "cluster_id": "syn_03027_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "tokenizer_hashes": {"receiver": "b9d8872cf3a99da99e1fc4b0f71232a451632dc49ca4c6771549b4efa5c1c763"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 4, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1760, "query_tokens": 498, "message_tokens": 387, "message_utf8_bytes": 1023, "cache_tensor_bytes": null, "serialized_payload_bytes": 1023, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 22983.29725000076, "hops": {"1": 6872.302875039168, "2": 5042.6992499269545, "3": 6354.470000020228, "4": 4713.825125014409}, "receiver_prefill_ms": 207.67320797313005, "ttft_ms": 11.876165983267128, "decode_ms": 11.814334080554545}, "total_wall_ms": 23215.294416993856, "ttft_ms": 11.876165983267128, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 63, "facts_retained": 24, "retention_rate": 0.38095238095238093, "retention_by_kind": {"person": 0.5, "date": 0.125, "amount": 0.6153846153846154, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1023}, "ledger_by_hop": {"1": {"facts_total": 63, "facts_retained": 27, "retention_rate": 0.42857142857142855, "retention_by_kind": {"person": 0.5, "date": 0.125, "amount": 0.6153846153846154, "status": 0.75, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1087}, "2": {"facts_total": 63, "facts_retained": 24, "retention_rate": 0.38095238095238093, "retention_by_kind": {"person": 0.5, "date": 0.125, "amount": 0.6153846153846154, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1063}, "3": {"facts_total": 63, "facts_retained": 24, "retention_rate": 0.38095238095238093, "retention_by_kind": {"person": 0.5, "date": 0.125, "amount": 0.6153846153846154, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1062}, "4": {"facts_total": 63, "facts_retained": 24, "retention_rate": 0.38095238095238093, "retention_by_kind": {"person": 0.5, "date": 0.125, "amount": 0.6153846153846154, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1023}}}, "reference_key_id": "keys/synthetic/syn_03027_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "9597ef7dd5408dec598b05726488fa7f8924bdb0755c45a0837fd9d429f51e78", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "aaa7a7add60ce5e11a1441e7", "attempt_id": "bba9fd272f81", "run_group_id": "E4_core:A4|closed|qwen3b-llama3b:h4", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:33:11.476430Z", "task_family": "synthetic", "document_id": "syn_03027_2000", "document_hash": "10326080979f71b625e4110c7c19767e5a752e267de5034f7e37d98cb6e06249", "question_id": "q06_negation", "split": "test", "cluster_id": "syn_03027_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "tokenizer_hashes": {"receiver": "b9d8872cf3a99da99e1fc4b0f71232a451632dc49ca4c6771549b4efa5c1c763"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 4, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1760, "query_tokens": 492, "message_tokens": 387, "message_utf8_bytes": 1023, "cache_tensor_bytes": null, "serialized_payload_bytes": 1023, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 22983.29725000076, "hops": {"1": 6872.302875039168, "2": 5042.6992499269545, "3": 6354.470000020228, "4": 4713.825125014409}, "receiver_prefill_ms": 205.95729094929993, "ttft_ms": 11.706916033290327, "decode_ms": 23.29354197718203}, "total_wall_ms": 23224.83874997124, "ttft_ms": 11.706916033290327, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 2}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "No.", "evidence_artifact": {"ledger": {"facts_total": 63, "facts_retained": 24, "retention_rate": 0.38095238095238093, "retention_by_kind": {"person": 0.5, "date": 0.125, "amount": 0.6153846153846154, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1023}, "ledger_by_hop": {"1": {"facts_total": 63, "facts_retained": 27, "retention_rate": 0.42857142857142855, "retention_by_kind": {"person": 0.5, "date": 0.125, "amount": 0.6153846153846154, "status": 0.75, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1087}, "2": {"facts_total": 63, "facts_retained": 24, "retention_rate": 0.38095238095238093, "retention_by_kind": {"person": 0.5, "date": 0.125, "amount": 0.6153846153846154, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1063}, "3": {"facts_total": 63, "facts_retained": 24, "retention_rate": 0.38095238095238093, "retention_by_kind": {"person": 0.5, "date": 0.125, "amount": 0.6153846153846154, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1062}, "4": {"facts_total": 63, "facts_retained": 24, "retention_rate": 0.38095238095238093, "retention_by_kind": {"person": 0.5, "date": 0.125, "amount": 0.6153846153846154, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1023}}}, "reference_key_id": "keys/synthetic/syn_03027_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "9597ef7dd5408dec598b05726488fa7f8924bdb0755c45a0837fd9d429f51e78", "output_hash": "38178a20b470cfd18299fb1593dd4ba706f1a83321dac051d01086efd8b7a96f"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "a7eac567267d564d1e795b39", "attempt_id": "f698f696c65b", "run_group_id": "E4_core:A4|closed|qwen3b-llama3b:h4", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:33:11.734874Z", "task_family": "synthetic", "document_id": "syn_03027_2000", "document_hash": "10326080979f71b625e4110c7c19767e5a752e267de5034f7e37d98cb6e06249", "question_id": "q07_negation", "split": "test", "cluster_id": "syn_03027_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "tokenizer_hashes": {"receiver": "b9d8872cf3a99da99e1fc4b0f71232a451632dc49ca4c6771549b4efa5c1c763"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 4, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1760, "query_tokens": 492, "message_tokens": 387, "message_utf8_bytes": 1023, "cache_tensor_bytes": null, "serialized_payload_bytes": 1023, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 22983.29725000076, "hops": {"1": 6872.302875039168, "2": 5042.6992499269545, "3": 6354.470000020228, "4": 4713.825125014409}, "receiver_prefill_ms": 207.91820797603577, "ttft_ms": 11.799959000200033, "decode_ms": 23.308750009164214}, "total_wall_ms": 23226.958292070776, "ttft_ms": 11.799959000200033, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 2}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "No.", "evidence_artifact": {"ledger": {"facts_total": 63, "facts_retained": 24, "retention_rate": 0.38095238095238093, "retention_by_kind": {"person": 0.5, "date": 0.125, "amount": 0.6153846153846154, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1023}, "ledger_by_hop": {"1": {"facts_total": 63, "facts_retained": 27, "retention_rate": 0.42857142857142855, "retention_by_kind": {"person": 0.5, "date": 0.125, "amount": 0.6153846153846154, "status": 0.75, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1087}, "2": {"facts_total": 63, "facts_retained": 24, "retention_rate": 0.38095238095238093, "retention_by_kind": {"person": 0.5, "date": 0.125, "amount": 0.6153846153846154, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1063}, "3": {"facts_total": 63, "facts_retained": 24, "retention_rate": 0.38095238095238093, "retention_by_kind": {"person": 0.5, "date": 0.125, "amount": 0.6153846153846154, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1062}, "4": {"facts_total": 63, "facts_retained": 24, "retention_rate": 0.38095238095238093, "retention_by_kind": {"person": 0.5, "date": 0.125, "amount": 0.6153846153846154, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1023}}}, "reference_key_id": "keys/synthetic/syn_03027_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "9597ef7dd5408dec598b05726488fa7f8924bdb0755c45a0837fd9d429f51e78", "output_hash": "38178a20b470cfd18299fb1593dd4ba706f1a83321dac051d01086efd8b7a96f"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "8358207eadeb8126475457b5", "attempt_id": "74feeb97cfa4", "run_group_id": "E4_core:A4|closed|qwen3b-llama3b:h4", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:33:11.995903Z", "task_family": "synthetic", "document_id": "syn_03027_2000", "document_hash": "10326080979f71b625e4110c7c19767e5a752e267de5034f7e37d98cb6e06249", "question_id": "q08_revision", "split": "test", "cluster_id": "syn_03027_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "tokenizer_hashes": {"receiver": "b9d8872cf3a99da99e1fc4b0f71232a451632dc49ca4c6771549b4efa5c1c763"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 4, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1760, "query_tokens": 507, "message_tokens": 387, "message_utf8_bytes": 1023, "cache_tensor_bytes": null, "serialized_payload_bytes": 1023, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 22983.29725000076, "hops": {"1": 6872.302875039168, "2": 5042.6992499269545, "3": 6354.470000020228, "4": 4713.825125014409}, "receiver_prefill_ms": 207.12370902765542, "ttft_ms": 11.836000019684434, "decode_ms": 58.3128749858588}, "total_wall_ms": 23261.240166961215, "ttft_ms": 11.836000019684434, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 5}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "8,010,205", "evidence_artifact": {"ledger": {"facts_total": 63, "facts_retained": 24, "retention_rate": 0.38095238095238093, "retention_by_kind": {"person": 0.5, "date": 0.125, "amount": 0.6153846153846154, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1023}, "ledger_by_hop": {"1": {"facts_total": 63, "facts_retained": 27, "retention_rate": 0.42857142857142855, "retention_by_kind": {"person": 0.5, "date": 0.125, "amount": 0.6153846153846154, "status": 0.75, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1087}, "2": {"facts_total": 63, "facts_retained": 24, "retention_rate": 0.38095238095238093, "retention_by_kind": {"person": 0.5, "date": 0.125, "amount": 0.6153846153846154, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1063}, "3": {"facts_total": 63, "facts_retained": 24, "retention_rate": 0.38095238095238093, "retention_by_kind": {"person": 0.5, "date": 0.125, "amount": 0.6153846153846154, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1062}, "4": {"facts_total": 63, "facts_retained": 24, "retention_rate": 0.38095238095238093, "retention_by_kind": {"person": 0.5, "date": 0.125, "amount": 0.6153846153846154, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1023}}}, "reference_key_id": "keys/synthetic/syn_03027_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "9597ef7dd5408dec598b05726488fa7f8924bdb0755c45a0837fd9d429f51e78", "output_hash": "12facbb746e46f5872a48eea64f64dc2cbd9610b1548522078108253d03a7904"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "df29ce9d3ec2ca091c6a5f91", "attempt_id": "338bb16699d3", "run_group_id": "E4_core:A4|closed|qwen3b-llama3b:h4", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:33:12.291250Z", "task_family": "synthetic", "document_id": "syn_03027_2000", "document_hash": "10326080979f71b625e4110c7c19767e5a752e267de5034f7e37d98cb6e06249", "question_id": "q09_revision", "split": "test", "cluster_id": "syn_03027_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "tokenizer_hashes": {"receiver": "b9d8872cf3a99da99e1fc4b0f71232a451632dc49ca4c6771549b4efa5c1c763"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 4, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1760, "query_tokens": 500, "message_tokens": 387, "message_utf8_bytes": 1023, "cache_tensor_bytes": null, "serialized_payload_bytes": 1023, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 22983.29725000076, "hops": {"1": 6872.302875039168, "2": 5042.6992499269545, "3": 6354.470000020228, "4": 4713.825125014409}, "receiver_prefill_ms": 206.97824994567782, "ttft_ms": 11.8665830232203, "decode_ms": 58.1426250282675}, "total_wall_ms": 23260.930624906905, "ttft_ms": 11.8665830232203, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 5}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "6,056,150", "evidence_artifact": {"ledger": {"facts_total": 63, "facts_retained": 24, "retention_rate": 0.38095238095238093, "retention_by_kind": {"person": 0.5, "date": 0.125, "amount": 0.6153846153846154, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1023}, "ledger_by_hop": {"1": {"facts_total": 63, "facts_retained": 27, "retention_rate": 0.42857142857142855, "retention_by_kind": {"person": 0.5, "date": 0.125, "amount": 0.6153846153846154, "status": 0.75, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1087}, "2": {"facts_total": 63, "facts_retained": 24, "retention_rate": 0.38095238095238093, "retention_by_kind": {"person": 0.5, "date": 0.125, "amount": 0.6153846153846154, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1063}, "3": {"facts_total": 63, "facts_retained": 24, "retention_rate": 0.38095238095238093, "retention_by_kind": {"person": 0.5, "date": 0.125, "amount": 0.6153846153846154, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1062}, "4": {"facts_total": 63, "facts_retained": 24, "retention_rate": 0.38095238095238093, "retention_by_kind": {"person": 0.5, "date": 0.125, "amount": 0.6153846153846154, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1023}}}, "reference_key_id": "keys/synthetic/syn_03027_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "9597ef7dd5408dec598b05726488fa7f8924bdb0755c45a0837fd9d429f51e78", "output_hash": "8262889a8478c7e0ee965f159d9849a3849b412e276f5b979476e2da5b320249"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "c5a9c203da8d3c85f53bee8c", "attempt_id": "ddeef272b1c2", "run_group_id": "E4_core:A4|closed|qwen3b-llama3b:h4", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:33:12.585910Z", "task_family": "synthetic", "document_id": "syn_03027_2000", "document_hash": "10326080979f71b625e4110c7c19767e5a752e267de5034f7e37d98cb6e06249", "question_id": "q10_temporal", "split": "test", "cluster_id": "syn_03027_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "tokenizer_hashes": {"receiver": "b9d8872cf3a99da99e1fc4b0f71232a451632dc49ca4c6771549b4efa5c1c763"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 4, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1760, "query_tokens": 508, "message_tokens": 387, "message_utf8_bytes": 1023, "cache_tensor_bytes": null, "serialized_payload_bytes": 1023, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 22983.29725000076, "hops": {"1": 6872.302875039168, "2": 5042.6992499269545, "3": 6354.470000020228, "4": 4713.825125014409}, "receiver_prefill_ms": 207.73366698995233, "ttft_ms": 11.777165927924216, "decode_ms": 11.898291064426303}, "total_wall_ms": 23215.30525002163, "ttft_ms": 11.777165927924216, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 63, "facts_retained": 24, "retention_rate": 0.38095238095238093, "retention_by_kind": {"person": 0.5, "date": 0.125, "amount": 0.6153846153846154, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1023}, "ledger_by_hop": {"1": {"facts_total": 63, "facts_retained": 27, "retention_rate": 0.42857142857142855, "retention_by_kind": {"person": 0.5, "date": 0.125, "amount": 0.6153846153846154, "status": 0.75, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1087}, "2": {"facts_total": 63, "facts_retained": 24, "retention_rate": 0.38095238095238093, "retention_by_kind": {"person": 0.5, "date": 0.125, "amount": 0.6153846153846154, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1063}, "3": {"facts_total": 63, "facts_retained": 24, "retention_rate": 0.38095238095238093, "retention_by_kind": {"person": 0.5, "date": 0.125, "amount": 0.6153846153846154, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1062}, "4": {"facts_total": 63, "facts_retained": 24, "retention_rate": 0.38095238095238093, "retention_by_kind": {"person": 0.5, "date": 0.125, "amount": 0.6153846153846154, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1023}}}, "reference_key_id": "keys/synthetic/syn_03027_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "9597ef7dd5408dec598b05726488fa7f8924bdb0755c45a0837fd9d429f51e78", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "df206db021827fbaf90b5d73", "attempt_id": "12a67fd78404", "run_group_id": "E4_core:A4|closed|qwen3b-llama3b:h4", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:33:12.834295Z", "task_family": "synthetic", "document_id": "syn_03027_2000", "document_hash": "10326080979f71b625e4110c7c19767e5a752e267de5034f7e37d98cb6e06249", "question_id": "q11_aggregation", "split": "test", "cluster_id": "syn_03027_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "tokenizer_hashes": {"receiver": "b9d8872cf3a99da99e1fc4b0f71232a451632dc49ca4c6771549b4efa5c1c763"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 4, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1760, "query_tokens": 503, "message_tokens": 387, "message_utf8_bytes": 1023, "cache_tensor_bytes": null, "serialized_payload_bytes": 1023, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 22983.29725000076, "hops": {"1": 6872.302875039168, "2": 5042.6992499269545, "3": 6354.470000020228, "4": 4713.825125014409}, "receiver_prefill_ms": 207.91666698642075, "ttft_ms": 11.728040990419686, "decode_ms": 301.89700005576015}, "total_wall_ms": 23505.46812498942, "ttft_ms": 11.728040990419686, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 26}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "6,914,937 + 1,787,610 + 8,278,225 = 16,980,772", "evidence_artifact": {"ledger": {"facts_total": 63, "facts_retained": 24, "retention_rate": 0.38095238095238093, "retention_by_kind": {"person": 0.5, "date": 0.125, "amount": 0.6153846153846154, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1023}, "ledger_by_hop": {"1": {"facts_total": 63, "facts_retained": 27, "retention_rate": 0.42857142857142855, "retention_by_kind": {"person": 0.5, "date": 0.125, "amount": 0.6153846153846154, "status": 0.75, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1087}, "2": {"facts_total": 63, "facts_retained": 24, "retention_rate": 0.38095238095238093, "retention_by_kind": {"person": 0.5, "date": 0.125, "amount": 0.6153846153846154, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1063}, "3": {"facts_total": 63, "facts_retained": 24, "retention_rate": 0.38095238095238093, "retention_by_kind": {"person": 0.5, "date": 0.125, "amount": 0.6153846153846154, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1062}, "4": {"facts_total": 63, "facts_retained": 24, "retention_rate": 0.38095238095238093, "retention_by_kind": {"person": 0.5, "date": 0.125, "amount": 0.6153846153846154, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1023}}}, "reference_key_id": "keys/synthetic/syn_03027_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 0.4}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "9597ef7dd5408dec598b05726488fa7f8924bdb0755c45a0837fd9d429f51e78", "output_hash": "c5fb5d9b76b7a02439f0acb83233430b95eef4aa96f8df88b9152c3b3e51343c"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "a25e6f29980ac10c153c7c60", "attempt_id": "3441fa81c60f", "run_group_id": "E4_core:A4|closed|qwen3b-llama3b:h4", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:33:13.373097Z", "task_family": "synthetic", "document_id": "syn_03027_2000", "document_hash": "10326080979f71b625e4110c7c19767e5a752e267de5034f7e37d98cb6e06249", "question_id": "q12_unit_conversion", "split": "test", "cluster_id": "syn_03027_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "tokenizer_hashes": {"receiver": "b9d8872cf3a99da99e1fc4b0f71232a451632dc49ca4c6771549b4efa5c1c763"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 4, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1760, "query_tokens": 499, "message_tokens": 387, "message_utf8_bytes": 1023, "cache_tensor_bytes": null, "serialized_payload_bytes": 1023, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 22983.29725000076, "hops": {"1": 6872.302875039168, "2": 5042.6992499269545, "3": 6354.470000020228, "4": 4713.825125014409}, "receiver_prefill_ms": 207.03924994450063, "ttft_ms": 11.80079102050513, "decode_ms": 11.682207928970456}, "total_wall_ms": 23214.439749950543, "ttft_ms": 11.80079102050513, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 63, "facts_retained": 24, "retention_rate": 0.38095238095238093, "retention_by_kind": {"person": 0.5, "date": 0.125, "amount": 0.6153846153846154, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1023}, "ledger_by_hop": {"1": {"facts_total": 63, "facts_retained": 27, "retention_rate": 0.42857142857142855, "retention_by_kind": {"person": 0.5, "date": 0.125, "amount": 0.6153846153846154, "status": 0.75, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1087}, "2": {"facts_total": 63, "facts_retained": 24, "retention_rate": 0.38095238095238093, "retention_by_kind": {"person": 0.5, "date": 0.125, "amount": 0.6153846153846154, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1063}, "3": {"facts_total": 63, "facts_retained": 24, "retention_rate": 0.38095238095238093, "retention_by_kind": {"person": 0.5, "date": 0.125, "amount": 0.6153846153846154, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1062}, "4": {"facts_total": 63, "facts_retained": 24, "retention_rate": 0.38095238095238093, "retention_by_kind": {"person": 0.5, "date": 0.125, "amount": 0.6153846153846154, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1023}}}, "reference_key_id": "keys/synthetic/syn_03027_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "9597ef7dd5408dec598b05726488fa7f8924bdb0755c45a0837fd9d429f51e78", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "4ab18cd59050673e42c6dbf5", "attempt_id": "c4501c8859f7", "run_group_id": "E4_core:A4|closed|qwen3b-llama3b:h4", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:33:13.620762Z", "task_family": "synthetic", "document_id": "syn_03027_2000", "document_hash": "10326080979f71b625e4110c7c19767e5a752e267de5034f7e37d98cb6e06249", "question_id": "q13_unanswerable", "split": "test", "cluster_id": "syn_03027_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "tokenizer_hashes": {"receiver": "b9d8872cf3a99da99e1fc4b0f71232a451632dc49ca4c6771549b4efa5c1c763"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 4, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1760, "query_tokens": 503, "message_tokens": 387, "message_utf8_bytes": 1023, "cache_tensor_bytes": null, "serialized_payload_bytes": 1023, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 22983.29725000076, "hops": {"1": 6872.302875039168, "2": 5042.6992499269545, "3": 6354.470000020228, "4": 4713.825125014409}, "receiver_prefill_ms": 209.02758301235735, "ttft_ms": 11.82587502989918, "decode_ms": 11.626291903667152}, "total_wall_ms": 23216.419583070092, "ttft_ms": 11.82587502989918, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 63, "facts_retained": 24, "retention_rate": 0.38095238095238093, "retention_by_kind": {"person": 0.5, "date": 0.125, "amount": 0.6153846153846154, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1023}, "ledger_by_hop": {"1": {"facts_total": 63, "facts_retained": 27, "retention_rate": 0.42857142857142855, "retention_by_kind": {"person": 0.5, "date": 0.125, "amount": 0.6153846153846154, "status": 0.75, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1087}, "2": {"facts_total": 63, "facts_retained": 24, "retention_rate": 0.38095238095238093, "retention_by_kind": {"person": 0.5, "date": 0.125, "amount": 0.6153846153846154, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1063}, "3": {"facts_total": 63, "facts_retained": 24, "retention_rate": 0.38095238095238093, "retention_by_kind": {"person": 0.5, "date": 0.125, "amount": 0.6153846153846154, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1062}, "4": {"facts_total": 63, "facts_retained": 24, "retention_rate": 0.38095238095238093, "retention_by_kind": {"person": 0.5, "date": 0.125, "amount": 0.6153846153846154, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1023}}}, "reference_key_id": "keys/synthetic/syn_03027_2000.json", "task_scores": {"correct": 1.0, "abstained": 1.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "9597ef7dd5408dec598b05726488fa7f8924bdb0755c45a0837fd9d429f51e78", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "f865d1300eef0a30fee6d154", "attempt_id": "cdc13db857d0", "run_group_id": "E4_core:A4|closed|qwen3b-llama3b:h4", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:33:13.870869Z", "task_family": "synthetic", "document_id": "syn_03027_2000", "document_hash": "10326080979f71b625e4110c7c19767e5a752e267de5034f7e37d98cb6e06249", "question_id": "q14_unanswerable", "split": "test", "cluster_id": "syn_03027_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "tokenizer_hashes": {"receiver": "b9d8872cf3a99da99e1fc4b0f71232a451632dc49ca4c6771549b4efa5c1c763"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 4, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1760, "query_tokens": 505, "message_tokens": 387, "message_utf8_bytes": 1023, "cache_tensor_bytes": null, "serialized_payload_bytes": 1023, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 22983.29725000076, "hops": {"1": 6872.302875039168, "2": 5042.6992499269545, "3": 6354.470000020228, "4": 4713.825125014409}, "receiver_prefill_ms": 208.67758302483708, "ttft_ms": 11.671332991681993, "decode_ms": 11.490499950014055}, "total_wall_ms": 23215.73445899412, "ttft_ms": 11.671332991681993, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 63, "facts_retained": 24, "retention_rate": 0.38095238095238093, "retention_by_kind": {"person": 0.5, "date": 0.125, "amount": 0.6153846153846154, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1023}, "ledger_by_hop": {"1": {"facts_total": 63, "facts_retained": 27, "retention_rate": 0.42857142857142855, "retention_by_kind": {"person": 0.5, "date": 0.125, "amount": 0.6153846153846154, "status": 0.75, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1087}, "2": {"facts_total": 63, "facts_retained": 24, "retention_rate": 0.38095238095238093, "retention_by_kind": {"person": 0.5, "date": 0.125, "amount": 0.6153846153846154, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1063}, "3": {"facts_total": 63, "facts_retained": 24, "retention_rate": 0.38095238095238093, "retention_by_kind": {"person": 0.5, "date": 0.125, "amount": 0.6153846153846154, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1062}, "4": {"facts_total": 63, "facts_retained": 24, "retention_rate": 0.38095238095238093, "retention_by_kind": {"person": 0.5, "date": 0.125, "amount": 0.6153846153846154, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1023}}}, "reference_key_id": "keys/synthetic/syn_03027_2000.json", "task_scores": {"correct": 1.0, "abstained": 1.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "9597ef7dd5408dec598b05726488fa7f8924bdb0755c45a0837fd9d429f51e78", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "5bea7e952ef3124857ebf1e5", "attempt_id": "2bab77b0c2af", "run_group_id": "E4_core:A4|closed|qwen3b-llama3b:h8", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:33:35.766587Z", "task_family": "synthetic", "document_id": "syn_03027_2000", "document_hash": "10326080979f71b625e4110c7c19767e5a752e267de5034f7e37d98cb6e06249", "question_id": "q00_single_fact", "split": "test", "cluster_id": "syn_03027_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "tokenizer_hashes": {"receiver": "b9d8872cf3a99da99e1fc4b0f71232a451632dc49ca4c6771549b4efa5c1c763"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 8, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1760, "query_tokens": 490, "message_tokens": 387, "message_utf8_bytes": 1022, "cache_tensor_bytes": null, "serialized_payload_bytes": 1022, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 44624.60495799314, "hops": {"1": 6872.302875039168, "2": 5042.6992499269545, "3": 6354.470000020228, "4": 4713.825125014409, "5": 6335.791665944271, "6": 4672.110749990679, "7": 5924.405542085879, "8": 4708.999749971554}, "receiver_prefill_ms": 205.54070791695267, "ttft_ms": 11.606125044636428, "decode_ms": 11.500791064463556}, "total_wall_ms": 44853.72995806392, "ttft_ms": 11.606125044636428, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 63, "facts_retained": 24, "retention_rate": 0.38095238095238093, "retention_by_kind": {"person": 0.5, "date": 0.125, "amount": 0.6153846153846154, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1022}, "ledger_by_hop": {"1": {"facts_total": 63, "facts_retained": 27, "retention_rate": 0.42857142857142855, "retention_by_kind": {"person": 0.5, "date": 0.125, "amount": 0.6153846153846154, "status": 0.75, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1087}, "2": {"facts_total": 63, "facts_retained": 24, "retention_rate": 0.38095238095238093, "retention_by_kind": {"person": 0.5, "date": 0.125, "amount": 0.6153846153846154, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1063}, "3": {"facts_total": 63, "facts_retained": 24, "retention_rate": 0.38095238095238093, "retention_by_kind": {"person": 0.5, "date": 0.125, "amount": 0.6153846153846154, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1062}, "4": {"facts_total": 63, "facts_retained": 24, "retention_rate": 0.38095238095238093, "retention_by_kind": {"person": 0.5, "date": 0.125, "amount": 0.6153846153846154, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1023}, "5": {"facts_total": 63, "facts_retained": 24, "retention_rate": 0.38095238095238093, "retention_by_kind": {"person": 0.5, "date": 0.125, "amount": 0.6153846153846154, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1050}, "6": {"facts_total": 63, "facts_retained": 24, "retention_rate": 0.38095238095238093, "retention_by_kind": {"person": 0.5, "date": 0.125, "amount": 0.6153846153846154, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1010}, "7": {"facts_total": 63, "facts_retained": 24, "retention_rate": 0.38095238095238093, "retention_by_kind": {"person": 0.5, "date": 0.125, "amount": 0.6153846153846154, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 992}, "8": {"facts_total": 63, "facts_retained": 24, "retention_rate": 0.38095238095238093, "retention_by_kind": {"person": 0.5, "date": 0.125, "amount": 0.6153846153846154, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1022}}}, "reference_key_id": "keys/synthetic/syn_03027_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "e5864c0b88a688a045e4eb1d1985303e81b16f7e80826e759450dac2bb9d2e48", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "6d04997b4e7393c1c4d422f7", "attempt_id": "0ca8e559b3c2", "run_group_id": "E4_core:A4|closed|qwen3b-llama3b:h8", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:33:36.012329Z", "task_family": "synthetic", "document_id": "syn_03027_2000", "document_hash": "10326080979f71b625e4110c7c19767e5a752e267de5034f7e37d98cb6e06249", "question_id": "q01_single_fact", "split": "test", "cluster_id": "syn_03027_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "tokenizer_hashes": {"receiver": "b9d8872cf3a99da99e1fc4b0f71232a451632dc49ca4c6771549b4efa5c1c763"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 8, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1760, "query_tokens": 497, "message_tokens": 387, "message_utf8_bytes": 1022, "cache_tensor_bytes": null, "serialized_payload_bytes": 1022, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 44624.60495799314, "hops": {"1": 6872.302875039168, "2": 5042.6992499269545, "3": 6354.470000020228, "4": 4713.825125014409, "5": 6335.791665944271, "6": 4672.110749990679, "7": 5924.405542085879, "8": 4708.999749971554}, "receiver_prefill_ms": 205.6505000218749, "ttft_ms": 11.552124982699752, "decode_ms": 114.7153329802677}, "total_wall_ms": 44956.97683293838, "ttft_ms": 11.552124982699752, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 10}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "Lot 247-S, Baie-Comeau", "evidence_artifact": {"ledger": {"facts_total": 63, "facts_retained": 24, "retention_rate": 0.38095238095238093, "retention_by_kind": {"person": 0.5, "date": 0.125, "amount": 0.6153846153846154, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1022}, "ledger_by_hop": {"1": {"facts_total": 63, "facts_retained": 27, "retention_rate": 0.42857142857142855, "retention_by_kind": {"person": 0.5, "date": 0.125, "amount": 0.6153846153846154, "status": 0.75, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1087}, "2": {"facts_total": 63, "facts_retained": 24, "retention_rate": 0.38095238095238093, "retention_by_kind": {"person": 0.5, "date": 0.125, "amount": 0.6153846153846154, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1063}, "3": {"facts_total": 63, "facts_retained": 24, "retention_rate": 0.38095238095238093, "retention_by_kind": {"person": 0.5, "date": 0.125, "amount": 0.6153846153846154, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1062}, "4": {"facts_total": 63, "facts_retained": 24, "retention_rate": 0.38095238095238093, "retention_by_kind": {"person": 0.5, "date": 0.125, "amount": 0.6153846153846154, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1023}, "5": {"facts_total": 63, "facts_retained": 24, "retention_rate": 0.38095238095238093, "retention_by_kind": {"person": 0.5, "date": 0.125, "amount": 0.6153846153846154, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1050}, "6": {"facts_total": 63, "facts_retained": 24, "retention_rate": 0.38095238095238093, "retention_by_kind": {"person": 0.5, "date": 0.125, "amount": 0.6153846153846154, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1010}, "7": {"facts_total": 63, "facts_retained": 24, "retention_rate": 0.38095238095238093, "retention_by_kind": {"person": 0.5, "date": 0.125, "amount": 0.6153846153846154, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 992}, "8": {"facts_total": 63, "facts_retained": 24, "retention_rate": 0.38095238095238093, "retention_by_kind": {"person": 0.5, "date": 0.125, "amount": 0.6153846153846154, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1022}}}, "reference_key_id": "keys/synthetic/syn_03027_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 0.8571428571428571}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "e5864c0b88a688a045e4eb1d1985303e81b16f7e80826e759450dac2bb9d2e48", "output_hash": "77b7da1d43e7d8815e9e9159ebbd735045bc7f668260ba30b93674f70622b8ec"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "5513ed92fbb9a62ad2e3e670", "attempt_id": "0b1358da766f", "run_group_id": "E4_core:A4|closed|qwen3b-llama3b:h8", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:33:36.360524Z", "task_family": "synthetic", "document_id": "syn_03027_2000", "document_hash": "10326080979f71b625e4110c7c19767e5a752e267de5034f7e37d98cb6e06249", "question_id": "q02_exact_copy", "split": "test", "cluster_id": "syn_03027_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "tokenizer_hashes": {"receiver": "b9d8872cf3a99da99e1fc4b0f71232a451632dc49ca4c6771549b4efa5c1c763"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 8, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1760, "query_tokens": 496, "message_tokens": 387, "message_utf8_bytes": 1022, "cache_tensor_bytes": null, "serialized_payload_bytes": 1022, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 44624.60495799314, "hops": {"1": 6872.302875039168, "2": 5042.6992499269545, "3": 6354.470000020228, "4": 4713.825125014409, "5": 6335.791665944271, "6": 4672.110749990679, "7": 5924.405542085879, "8": 4708.999749971554}, "receiver_prefill_ms": 205.58270800393075, "ttft_ms": 11.726291966624558, "decode_ms": 46.06037505436689}, "total_wall_ms": 44888.58349993825, "ttft_ms": 11.726291966624558, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 4}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "15.0054", "evidence_artifact": {"ledger": {"facts_total": 63, "facts_retained": 24, "retention_rate": 0.38095238095238093, "retention_by_kind": {"person": 0.5, "date": 0.125, "amount": 0.6153846153846154, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1022}, "ledger_by_hop": {"1": {"facts_total": 63, "facts_retained": 27, "retention_rate": 0.42857142857142855, "retention_by_kind": {"person": 0.5, "date": 0.125, "amount": 0.6153846153846154, "status": 0.75, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1087}, "2": {"facts_total": 63, "facts_retained": 24, "retention_rate": 0.38095238095238093, "retention_by_kind": {"person": 0.5, "date": 0.125, "amount": 0.6153846153846154, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1063}, "3": {"facts_total": 63, "facts_retained": 24, "retention_rate": 0.38095238095238093, "retention_by_kind": {"person": 0.5, "date": 0.125, "amount": 0.6153846153846154, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1062}, "4": {"facts_total": 63, "facts_retained": 24, "retention_rate": 0.38095238095238093, "retention_by_kind": {"person": 0.5, "date": 0.125, "amount": 0.6153846153846154, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1023}, "5": {"facts_total": 63, "facts_retained": 24, "retention_rate": 0.38095238095238093, "retention_by_kind": {"person": 0.5, "date": 0.125, "amount": 0.6153846153846154, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1050}, "6": {"facts_total": 63, "facts_retained": 24, "retention_rate": 0.38095238095238093, "retention_by_kind": {"person": 0.5, "date": 0.125, "amount": 0.6153846153846154, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1010}, "7": {"facts_total": 63, "facts_retained": 24, "retention_rate": 0.38095238095238093, "retention_by_kind": {"person": 0.5, "date": 0.125, "amount": 0.6153846153846154, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 992}, "8": {"facts_total": 63, "facts_retained": 24, "retention_rate": 0.38095238095238093, "retention_by_kind": {"person": 0.5, "date": 0.125, "amount": 0.6153846153846154, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1022}}}, "reference_key_id": "keys/synthetic/syn_03027_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "e5864c0b88a688a045e4eb1d1985303e81b16f7e80826e759450dac2bb9d2e48", "output_hash": "d13adba73068c847c850237059e0202fc5b91ccacf2e792e9ce5e3b90556501a"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "342492acda34dd62c8927c7a", "attempt_id": "82e71c2e27fb", "run_group_id": "E4_core:A4|closed|qwen3b-llama3b:h8", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:33:36.640029Z", "task_family": "synthetic", "document_id": "syn_03027_2000", "document_hash": "10326080979f71b625e4110c7c19767e5a752e267de5034f7e37d98cb6e06249", "question_id": "q03_exact_copy", "split": "test", "cluster_id": "syn_03027_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "tokenizer_hashes": {"receiver": "b9d8872cf3a99da99e1fc4b0f71232a451632dc49ca4c6771549b4efa5c1c763"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 8, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1760, "query_tokens": 492, "message_tokens": 387, "message_utf8_bytes": 1022, "cache_tensor_bytes": null, "serialized_payload_bytes": 1022, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 44624.60495799314, "hops": {"1": 6872.302875039168, "2": 5042.6992499269545, "3": 6354.470000020228, "4": 4713.825125014409, "5": 6335.791665944271, "6": 4672.110749990679, "7": 5924.405542085879, "8": 4708.999749971554}, "receiver_prefill_ms": 205.49395796842873, "ttft_ms": 11.706040939316154, "decode_ms": 103.56912505812943}, "total_wall_ms": 44945.98799897358, "ttft_ms": 11.706040939316154, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 9}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "Priyamvada Sørlie", "evidence_artifact": {"ledger": {"facts_total": 63, "facts_retained": 24, "retention_rate": 0.38095238095238093, "retention_by_kind": {"person": 0.5, "date": 0.125, "amount": 0.6153846153846154, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1022}, "ledger_by_hop": {"1": {"facts_total": 63, "facts_retained": 27, "retention_rate": 0.42857142857142855, "retention_by_kind": {"person": 0.5, "date": 0.125, "amount": 0.6153846153846154, "status": 0.75, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1087}, "2": {"facts_total": 63, "facts_retained": 24, "retention_rate": 0.38095238095238093, "retention_by_kind": {"person": 0.5, "date": 0.125, "amount": 0.6153846153846154, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1063}, "3": {"facts_total": 63, "facts_retained": 24, "retention_rate": 0.38095238095238093, "retention_by_kind": {"person": 0.5, "date": 0.125, "amount": 0.6153846153846154, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1062}, "4": {"facts_total": 63, "facts_retained": 24, "retention_rate": 0.38095238095238093, "retention_by_kind": {"person": 0.5, "date": 0.125, "amount": 0.6153846153846154, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1023}, "5": {"facts_total": 63, "facts_retained": 24, "retention_rate": 0.38095238095238093, "retention_by_kind": {"person": 0.5, "date": 0.125, "amount": 0.6153846153846154, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1050}, "6": {"facts_total": 63, "facts_retained": 24, "retention_rate": 0.38095238095238093, "retention_by_kind": {"person": 0.5, "date": 0.125, "amount": 0.6153846153846154, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1010}, "7": {"facts_total": 63, "facts_retained": 24, "retention_rate": 0.38095238095238093, "retention_by_kind": {"person": 0.5, "date": 0.125, "amount": 0.6153846153846154, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 992}, "8": {"facts_total": 63, "facts_retained": 24, "retention_rate": 0.38095238095238093, "retention_by_kind": {"person": 0.5, "date": 0.125, "amount": 0.6153846153846154, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1022}}}, "reference_key_id": "keys/synthetic/syn_03027_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "e5864c0b88a688a045e4eb1d1985303e81b16f7e80826e759450dac2bb9d2e48", "output_hash": "c5f554196ed23dd71ad995ec89f7142099d524dbb6534cc10d94a0c89d13e792"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "fb7b0cba5561957abf2f9e7d", "attempt_id": "ff7eb9ea83c8", "run_group_id": "E4_core:A4|closed|qwen3b-llama3b:h8", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:33:36.977139Z", "task_family": "synthetic", "document_id": "syn_03027_2000", "document_hash": "10326080979f71b625e4110c7c19767e5a752e267de5034f7e37d98cb6e06249", "question_id": "q04_multi_hop", "split": "test", "cluster_id": "syn_03027_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "tokenizer_hashes": {"receiver": "b9d8872cf3a99da99e1fc4b0f71232a451632dc49ca4c6771549b4efa5c1c763"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 8, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1760, "query_tokens": 503, "message_tokens": 387, "message_utf8_bytes": 1022, "cache_tensor_bytes": null, "serialized_payload_bytes": 1022, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 44624.60495799314, "hops": {"1": 6872.302875039168, "2": 5042.6992499269545, "3": 6354.470000020228, "4": 4713.825125014409, "5": 6335.791665944271, "6": 4672.110749990679, "7": 5924.405542085879, "8": 4708.999749971554}, "receiver_prefill_ms": 207.89683295879513, "ttft_ms": 11.536833946593106, "decode_ms": 11.517417035065591}, "total_wall_ms": 44856.00374999922, "ttft_ms": 11.536833946593106, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 63, "facts_retained": 24, "retention_rate": 0.38095238095238093, "retention_by_kind": {"person": 0.5, "date": 0.125, "amount": 0.6153846153846154, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1022}, "ledger_by_hop": {"1": {"facts_total": 63, "facts_retained": 27, "retention_rate": 0.42857142857142855, "retention_by_kind": {"person": 0.5, "date": 0.125, "amount": 0.6153846153846154, "status": 0.75, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1087}, "2": {"facts_total": 63, "facts_retained": 24, "retention_rate": 0.38095238095238093, "retention_by_kind": {"person": 0.5, "date": 0.125, "amount": 0.6153846153846154, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1063}, "3": {"facts_total": 63, "facts_retained": 24, "retention_rate": 0.38095238095238093, "retention_by_kind": {"person": 0.5, "date": 0.125, "amount": 0.6153846153846154, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1062}, "4": {"facts_total": 63, "facts_retained": 24, "retention_rate": 0.38095238095238093, "retention_by_kind": {"person": 0.5, "date": 0.125, "amount": 0.6153846153846154, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1023}, "5": {"facts_total": 63, "facts_retained": 24, "retention_rate": 0.38095238095238093, "retention_by_kind": {"person": 0.5, "date": 0.125, "amount": 0.6153846153846154, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1050}, "6": {"facts_total": 63, "facts_retained": 24, "retention_rate": 0.38095238095238093, "retention_by_kind": {"person": 0.5, "date": 0.125, "amount": 0.6153846153846154, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1010}, "7": {"facts_total": 63, "facts_retained": 24, "retention_rate": 0.38095238095238093, "retention_by_kind": {"person": 0.5, "date": 0.125, "amount": 0.6153846153846154, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 992}, "8": {"facts_total": 63, "facts_retained": 24, "retention_rate": 0.38095238095238093, "retention_by_kind": {"person": 0.5, "date": 0.125, "amount": 0.6153846153846154, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1022}}}, "reference_key_id": "keys/synthetic/syn_03027_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "e5864c0b88a688a045e4eb1d1985303e81b16f7e80826e759450dac2bb9d2e48", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "ddae4fd2ea0f6622042da8a1", "attempt_id": "923a18f829ab", "run_group_id": "E4_core:A4|closed|qwen3b-llama3b:h8", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:33:37.225005Z", "task_family": "synthetic", "document_id": "syn_03027_2000", "document_hash": "10326080979f71b625e4110c7c19767e5a752e267de5034f7e37d98cb6e06249", "question_id": "q05_multi_hop", "split": "test", "cluster_id": "syn_03027_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "tokenizer_hashes": {"receiver": "b9d8872cf3a99da99e1fc4b0f71232a451632dc49ca4c6771549b4efa5c1c763"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 8, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1760, "query_tokens": 498, "message_tokens": 387, "message_utf8_bytes": 1022, "cache_tensor_bytes": null, "serialized_payload_bytes": 1022, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 44624.60495799314, "hops": {"1": 6872.302875039168, "2": 5042.6992499269545, "3": 6354.470000020228, "4": 4713.825125014409, "5": 6335.791665944271, "6": 4672.110749990679, "7": 5924.405542085879, "8": 4708.999749971554}, "receiver_prefill_ms": 206.6919170320034, "ttft_ms": 11.615708004683256, "decode_ms": 11.492042103782296}, "total_wall_ms": 44854.840541025624, "ttft_ms": 11.615708004683256, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 63, "facts_retained": 24, "retention_rate": 0.38095238095238093, "retention_by_kind": {"person": 0.5, "date": 0.125, "amount": 0.6153846153846154, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1022}, "ledger_by_hop": {"1": {"facts_total": 63, "facts_retained": 27, "retention_rate": 0.42857142857142855, "retention_by_kind": {"person": 0.5, "date": 0.125, "amount": 0.6153846153846154, "status": 0.75, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1087}, "2": {"facts_total": 63, "facts_retained": 24, "retention_rate": 0.38095238095238093, "retention_by_kind": {"person": 0.5, "date": 0.125, "amount": 0.6153846153846154, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1063}, "3": {"facts_total": 63, "facts_retained": 24, "retention_rate": 0.38095238095238093, "retention_by_kind": {"person": 0.5, "date": 0.125, "amount": 0.6153846153846154, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1062}, "4": {"facts_total": 63, "facts_retained": 24, "retention_rate": 0.38095238095238093, "retention_by_kind": {"person": 0.5, "date": 0.125, "amount": 0.6153846153846154, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1023}, "5": {"facts_total": 63, "facts_retained": 24, "retention_rate": 0.38095238095238093, "retention_by_kind": {"person": 0.5, "date": 0.125, "amount": 0.6153846153846154, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1050}, "6": {"facts_total": 63, "facts_retained": 24, "retention_rate": 0.38095238095238093, "retention_by_kind": {"person": 0.5, "date": 0.125, "amount": 0.6153846153846154, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1010}, "7": {"facts_total": 63, "facts_retained": 24, "retention_rate": 0.38095238095238093, "retention_by_kind": {"person": 0.5, "date": 0.125, "amount": 0.6153846153846154, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 992}, "8": {"facts_total": 63, "facts_retained": 24, "retention_rate": 0.38095238095238093, "retention_by_kind": {"person": 0.5, "date": 0.125, "amount": 0.6153846153846154, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1022}}}, "reference_key_id": "keys/synthetic/syn_03027_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "e5864c0b88a688a045e4eb1d1985303e81b16f7e80826e759450dac2bb9d2e48", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "e99444682b2647a9d1f225ef", "attempt_id": "1dc4f37b0538", "run_group_id": "E4_core:A4|closed|qwen3b-llama3b:h8", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:33:37.470529Z", "task_family": "synthetic", "document_id": "syn_03027_2000", "document_hash": "10326080979f71b625e4110c7c19767e5a752e267de5034f7e37d98cb6e06249", "question_id": "q06_negation", "split": "test", "cluster_id": "syn_03027_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "tokenizer_hashes": {"receiver": "b9d8872cf3a99da99e1fc4b0f71232a451632dc49ca4c6771549b4efa5c1c763"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 8, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1760, "query_tokens": 492, "message_tokens": 387, "message_utf8_bytes": 1022, "cache_tensor_bytes": null, "serialized_payload_bytes": 1022, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 44624.60495799314, "hops": {"1": 6872.302875039168, "2": 5042.6992499269545, "3": 6354.470000020228, "4": 4713.825125014409, "5": 6335.791665944271, "6": 4672.110749990679, "7": 5924.405542085879, "8": 4708.999749971554}, "receiver_prefill_ms": 206.48579101543874, "ttft_ms": 11.523665976710618, "decode_ms": 22.95541705098003}, "total_wall_ms": 44865.99916697014, "ttft_ms": 11.523665976710618, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 2}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "No.", "evidence_artifact": {"ledger": {"facts_total": 63, "facts_retained": 24, "retention_rate": 0.38095238095238093, "retention_by_kind": {"person": 0.5, "date": 0.125, "amount": 0.6153846153846154, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1022}, "ledger_by_hop": {"1": {"facts_total": 63, "facts_retained": 27, "retention_rate": 0.42857142857142855, "retention_by_kind": {"person": 0.5, "date": 0.125, "amount": 0.6153846153846154, "status": 0.75, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1087}, "2": {"facts_total": 63, "facts_retained": 24, "retention_rate": 0.38095238095238093, "retention_by_kind": {"person": 0.5, "date": 0.125, "amount": 0.6153846153846154, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1063}, "3": {"facts_total": 63, "facts_retained": 24, "retention_rate": 0.38095238095238093, "retention_by_kind": {"person": 0.5, "date": 0.125, "amount": 0.6153846153846154, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1062}, "4": {"facts_total": 63, "facts_retained": 24, "retention_rate": 0.38095238095238093, "retention_by_kind": {"person": 0.5, "date": 0.125, "amount": 0.6153846153846154, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1023}, "5": {"facts_total": 63, "facts_retained": 24, "retention_rate": 0.38095238095238093, "retention_by_kind": {"person": 0.5, "date": 0.125, "amount": 0.6153846153846154, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1050}, "6": {"facts_total": 63, "facts_retained": 24, "retention_rate": 0.38095238095238093, "retention_by_kind": {"person": 0.5, "date": 0.125, "amount": 0.6153846153846154, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1010}, "7": {"facts_total": 63, "facts_retained": 24, "retention_rate": 0.38095238095238093, "retention_by_kind": {"person": 0.5, "date": 0.125, "amount": 0.6153846153846154, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 992}, "8": {"facts_total": 63, "facts_retained": 24, "retention_rate": 0.38095238095238093, "retention_by_kind": {"person": 0.5, "date": 0.125, "amount": 0.6153846153846154, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1022}}}, "reference_key_id": "keys/synthetic/syn_03027_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "e5864c0b88a688a045e4eb1d1985303e81b16f7e80826e759450dac2bb9d2e48", "output_hash": "38178a20b470cfd18299fb1593dd4ba706f1a83321dac051d01086efd8b7a96f"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "c82159bb6d1b2693879010ce", "attempt_id": "9cfc68fe24a1", "run_group_id": "E4_core:A4|closed|qwen3b-llama3b:h8", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:33:37.727520Z", "task_family": "synthetic", "document_id": "syn_03027_2000", "document_hash": "10326080979f71b625e4110c7c19767e5a752e267de5034f7e37d98cb6e06249", "question_id": "q07_negation", "split": "test", "cluster_id": "syn_03027_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "tokenizer_hashes": {"receiver": "b9d8872cf3a99da99e1fc4b0f71232a451632dc49ca4c6771549b4efa5c1c763"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 8, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1760, "query_tokens": 492, "message_tokens": 387, "message_utf8_bytes": 1022, "cache_tensor_bytes": null, "serialized_payload_bytes": 1022, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 44624.60495799314, "hops": {"1": 6872.302875039168, "2": 5042.6992499269545, "3": 6354.470000020228, "4": 4713.825125014409, "5": 6335.791665944271, "6": 4672.110749990679, "7": 5924.405542085879, "8": 4708.999749971554}, "receiver_prefill_ms": 205.37658303510398, "ttft_ms": 11.629833024926484, "decode_ms": 23.068250040523708}, "total_wall_ms": 44865.222748951055, "ttft_ms": 11.629833024926484, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 2}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "No.", "evidence_artifact": {"ledger": {"facts_total": 63, "facts_retained": 24, "retention_rate": 0.38095238095238093, "retention_by_kind": {"person": 0.5, "date": 0.125, "amount": 0.6153846153846154, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1022}, "ledger_by_hop": {"1": {"facts_total": 63, "facts_retained": 27, "retention_rate": 0.42857142857142855, "retention_by_kind": {"person": 0.5, "date": 0.125, "amount": 0.6153846153846154, "status": 0.75, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1087}, "2": {"facts_total": 63, "facts_retained": 24, "retention_rate": 0.38095238095238093, "retention_by_kind": {"person": 0.5, "date": 0.125, "amount": 0.6153846153846154, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1063}, "3": {"facts_total": 63, "facts_retained": 24, "retention_rate": 0.38095238095238093, "retention_by_kind": {"person": 0.5, "date": 0.125, "amount": 0.6153846153846154, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1062}, "4": {"facts_total": 63, "facts_retained": 24, "retention_rate": 0.38095238095238093, "retention_by_kind": {"person": 0.5, "date": 0.125, "amount": 0.6153846153846154, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1023}, "5": {"facts_total": 63, "facts_retained": 24, "retention_rate": 0.38095238095238093, "retention_by_kind": {"person": 0.5, "date": 0.125, "amount": 0.6153846153846154, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1050}, "6": {"facts_total": 63, "facts_retained": 24, "retention_rate": 0.38095238095238093, "retention_by_kind": {"person": 0.5, "date": 0.125, "amount": 0.6153846153846154, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1010}, "7": {"facts_total": 63, "facts_retained": 24, "retention_rate": 0.38095238095238093, "retention_by_kind": {"person": 0.5, "date": 0.125, "amount": 0.6153846153846154, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 992}, "8": {"facts_total": 63, "facts_retained": 24, "retention_rate": 0.38095238095238093, "retention_by_kind": {"person": 0.5, "date": 0.125, "amount": 0.6153846153846154, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1022}}}, "reference_key_id": "keys/synthetic/syn_03027_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "e5864c0b88a688a045e4eb1d1985303e81b16f7e80826e759450dac2bb9d2e48", "output_hash": "38178a20b470cfd18299fb1593dd4ba706f1a83321dac051d01086efd8b7a96f"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "a2a8eaa3f395a27f4ab5cedd", "attempt_id": "75fe32a71ebd", "run_group_id": "E4_core:A4|closed|qwen3b-llama3b:h8", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:33:37.983494Z", "task_family": "synthetic", "document_id": "syn_03027_2000", "document_hash": "10326080979f71b625e4110c7c19767e5a752e267de5034f7e37d98cb6e06249", "question_id": "q08_revision", "split": "test", "cluster_id": "syn_03027_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "tokenizer_hashes": {"receiver": "b9d8872cf3a99da99e1fc4b0f71232a451632dc49ca4c6771549b4efa5c1c763"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 8, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1760, "query_tokens": 507, "message_tokens": 387, "message_utf8_bytes": 1022, "cache_tensor_bytes": null, "serialized_payload_bytes": 1022, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 44624.60495799314, "hops": {"1": 6872.302875039168, "2": 5042.6992499269545, "3": 6354.470000020228, "4": 4713.825125014409, "5": 6335.791665944271, "6": 4672.110749990679, "7": 5924.405542085879, "8": 4708.999749971554}, "receiver_prefill_ms": 207.95070903841406, "ttft_ms": 11.485417024232447, "decode_ms": 57.63437494169921}, "total_wall_ms": 44902.11654105224, "ttft_ms": 11.485417024232447, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 5}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "8,010,205", "evidence_artifact": {"ledger": {"facts_total": 63, "facts_retained": 24, "retention_rate": 0.38095238095238093, "retention_by_kind": {"person": 0.5, "date": 0.125, "amount": 0.6153846153846154, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1022}, "ledger_by_hop": {"1": {"facts_total": 63, "facts_retained": 27, "retention_rate": 0.42857142857142855, "retention_by_kind": {"person": 0.5, "date": 0.125, "amount": 0.6153846153846154, "status": 0.75, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1087}, "2": {"facts_total": 63, "facts_retained": 24, "retention_rate": 0.38095238095238093, "retention_by_kind": {"person": 0.5, "date": 0.125, "amount": 0.6153846153846154, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1063}, "3": {"facts_total": 63, "facts_retained": 24, "retention_rate": 0.38095238095238093, "retention_by_kind": {"person": 0.5, "date": 0.125, "amount": 0.6153846153846154, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1062}, "4": {"facts_total": 63, "facts_retained": 24, "retention_rate": 0.38095238095238093, "retention_by_kind": {"person": 0.5, "date": 0.125, "amount": 0.6153846153846154, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1023}, "5": {"facts_total": 63, "facts_retained": 24, "retention_rate": 0.38095238095238093, "retention_by_kind": {"person": 0.5, "date": 0.125, "amount": 0.6153846153846154, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1050}, "6": {"facts_total": 63, "facts_retained": 24, "retention_rate": 0.38095238095238093, "retention_by_kind": {"person": 0.5, "date": 0.125, "amount": 0.6153846153846154, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1010}, "7": {"facts_total": 63, "facts_retained": 24, "retention_rate": 0.38095238095238093, "retention_by_kind": {"person": 0.5, "date": 0.125, "amount": 0.6153846153846154, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 992}, "8": {"facts_total": 63, "facts_retained": 24, "retention_rate": 0.38095238095238093, "retention_by_kind": {"person": 0.5, "date": 0.125, "amount": 0.6153846153846154, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1022}}}, "reference_key_id": "keys/synthetic/syn_03027_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "e5864c0b88a688a045e4eb1d1985303e81b16f7e80826e759450dac2bb9d2e48", "output_hash": "12facbb746e46f5872a48eea64f64dc2cbd9610b1548522078108253d03a7904"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "6f5cf23ff573451d905002c2", "attempt_id": "a49aebdd83cb", "run_group_id": "E4_core:A4|closed|qwen3b-llama3b:h8", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:33:38.276726Z", "task_family": "synthetic", "document_id": "syn_03027_2000", "document_hash": "10326080979f71b625e4110c7c19767e5a752e267de5034f7e37d98cb6e06249", "question_id": "q09_revision", "split": "test", "cluster_id": "syn_03027_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "tokenizer_hashes": {"receiver": "b9d8872cf3a99da99e1fc4b0f71232a451632dc49ca4c6771549b4efa5c1c763"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 8, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1760, "query_tokens": 500, "message_tokens": 387, "message_utf8_bytes": 1022, "cache_tensor_bytes": null, "serialized_payload_bytes": 1022, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 44624.60495799314, "hops": {"1": 6872.302875039168, "2": 5042.6992499269545, "3": 6354.470000020228, "4": 4713.825125014409, "5": 6335.791665944271, "6": 4672.110749990679, "7": 5924.405542085879, "8": 4708.999749971554}, "receiver_prefill_ms": 206.10175002366304, "ttft_ms": 11.502541950903833, "decode_ms": 57.40062496624887}, "total_wall_ms": 44900.05395794287, "ttft_ms": 11.502541950903833, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 5}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "6,056,150", "evidence_artifact": {"ledger": {"facts_total": 63, "facts_retained": 24, "retention_rate": 0.38095238095238093, "retention_by_kind": {"person": 0.5, "date": 0.125, "amount": 0.6153846153846154, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1022}, "ledger_by_hop": {"1": {"facts_total": 63, "facts_retained": 27, "retention_rate": 0.42857142857142855, "retention_by_kind": {"person": 0.5, "date": 0.125, "amount": 0.6153846153846154, "status": 0.75, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1087}, "2": {"facts_total": 63, "facts_retained": 24, "retention_rate": 0.38095238095238093, "retention_by_kind": {"person": 0.5, "date": 0.125, "amount": 0.6153846153846154, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1063}, "3": {"facts_total": 63, "facts_retained": 24, "retention_rate": 0.38095238095238093, "retention_by_kind": {"person": 0.5, "date": 0.125, "amount": 0.6153846153846154, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1062}, "4": {"facts_total": 63, "facts_retained": 24, "retention_rate": 0.38095238095238093, "retention_by_kind": {"person": 0.5, "date": 0.125, "amount": 0.6153846153846154, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1023}, "5": {"facts_total": 63, "facts_retained": 24, "retention_rate": 0.38095238095238093, "retention_by_kind": {"person": 0.5, "date": 0.125, "amount": 0.6153846153846154, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1050}, "6": {"facts_total": 63, "facts_retained": 24, "retention_rate": 0.38095238095238093, "retention_by_kind": {"person": 0.5, "date": 0.125, "amount": 0.6153846153846154, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1010}, "7": {"facts_total": 63, "facts_retained": 24, "retention_rate": 0.38095238095238093, "retention_by_kind": {"person": 0.5, "date": 0.125, "amount": 0.6153846153846154, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 992}, "8": {"facts_total": 63, "facts_retained": 24, "retention_rate": 0.38095238095238093, "retention_by_kind": {"person": 0.5, "date": 0.125, "amount": 0.6153846153846154, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1022}}}, "reference_key_id": "keys/synthetic/syn_03027_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "e5864c0b88a688a045e4eb1d1985303e81b16f7e80826e759450dac2bb9d2e48", "output_hash": "8262889a8478c7e0ee965f159d9849a3849b412e276f5b979476e2da5b320249"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "0660a5cf6123369c78d0112f", "attempt_id": "b83f846951b2", "run_group_id": "E4_core:A4|closed|qwen3b-llama3b:h8", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:33:38.567754Z", "task_family": "synthetic", "document_id": "syn_03027_2000", "document_hash": "10326080979f71b625e4110c7c19767e5a752e267de5034f7e37d98cb6e06249", "question_id": "q10_temporal", "split": "test", "cluster_id": "syn_03027_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "tokenizer_hashes": {"receiver": "b9d8872cf3a99da99e1fc4b0f71232a451632dc49ca4c6771549b4efa5c1c763"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 8, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1760, "query_tokens": 508, "message_tokens": 387, "message_utf8_bytes": 1022, "cache_tensor_bytes": null, "serialized_payload_bytes": 1022, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 44624.60495799314, "hops": {"1": 6872.302875039168, "2": 5042.6992499269545, "3": 6354.470000020228, "4": 4713.825125014409, "5": 6335.791665944271, "6": 4672.110749990679, "7": 5924.405542085879, "8": 4708.999749971554}, "receiver_prefill_ms": 208.3208749536425, "ttft_ms": 11.555084027349949, "decode_ms": 11.706332908943295}, "total_wall_ms": 44856.603916035965, "ttft_ms": 11.555084027349949, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 63, "facts_retained": 24, "retention_rate": 0.38095238095238093, "retention_by_kind": {"person": 0.5, "date": 0.125, "amount": 0.6153846153846154, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1022}, "ledger_by_hop": {"1": {"facts_total": 63, "facts_retained": 27, "retention_rate": 0.42857142857142855, "retention_by_kind": {"person": 0.5, "date": 0.125, "amount": 0.6153846153846154, "status": 0.75, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1087}, "2": {"facts_total": 63, "facts_retained": 24, "retention_rate": 0.38095238095238093, "retention_by_kind": {"person": 0.5, "date": 0.125, "amount": 0.6153846153846154, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1063}, "3": {"facts_total": 63, "facts_retained": 24, "retention_rate": 0.38095238095238093, "retention_by_kind": {"person": 0.5, "date": 0.125, "amount": 0.6153846153846154, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1062}, "4": {"facts_total": 63, "facts_retained": 24, "retention_rate": 0.38095238095238093, "retention_by_kind": {"person": 0.5, "date": 0.125, "amount": 0.6153846153846154, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1023}, "5": {"facts_total": 63, "facts_retained": 24, "retention_rate": 0.38095238095238093, "retention_by_kind": {"person": 0.5, "date": 0.125, "amount": 0.6153846153846154, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1050}, "6": {"facts_total": 63, "facts_retained": 24, "retention_rate": 0.38095238095238093, "retention_by_kind": {"person": 0.5, "date": 0.125, "amount": 0.6153846153846154, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1010}, "7": {"facts_total": 63, "facts_retained": 24, "retention_rate": 0.38095238095238093, "retention_by_kind": {"person": 0.5, "date": 0.125, "amount": 0.6153846153846154, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 992}, "8": {"facts_total": 63, "facts_retained": 24, "retention_rate": 0.38095238095238093, "retention_by_kind": {"person": 0.5, "date": 0.125, "amount": 0.6153846153846154, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1022}}}, "reference_key_id": "keys/synthetic/syn_03027_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "e5864c0b88a688a045e4eb1d1985303e81b16f7e80826e759450dac2bb9d2e48", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "8ff8a9e150f4804461a35ecf", "attempt_id": "944d46eb4f12", "run_group_id": "E4_core:A4|closed|qwen3b-llama3b:h8", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:33:38.815297Z", "task_family": "synthetic", "document_id": "syn_03027_2000", "document_hash": "10326080979f71b625e4110c7c19767e5a752e267de5034f7e37d98cb6e06249", "question_id": "q11_aggregation", "split": "test", "cluster_id": "syn_03027_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "tokenizer_hashes": {"receiver": "b9d8872cf3a99da99e1fc4b0f71232a451632dc49ca4c6771549b4efa5c1c763"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 8, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1760, "query_tokens": 503, "message_tokens": 387, "message_utf8_bytes": 1022, "cache_tensor_bytes": null, "serialized_payload_bytes": 1022, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 44624.60495799314, "hops": {"1": 6872.302875039168, "2": 5042.6992499269545, "3": 6354.470000020228, "4": 4713.825125014409, "5": 6335.791665944271, "6": 4672.110749990679, "7": 5924.405542085879, "8": 4708.999749971554}, "receiver_prefill_ms": 208.48258293699473, "ttft_ms": 11.795250000432134, "decode_ms": 298.95991703961045}, "total_wall_ms": 45144.31745803449, "ttft_ms": 11.795250000432134, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 26}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "6,914,937 + 1,787,610 + 8,278,225 = 16,980,772", "evidence_artifact": {"ledger": {"facts_total": 63, "facts_retained": 24, "retention_rate": 0.38095238095238093, "retention_by_kind": {"person": 0.5, "date": 0.125, "amount": 0.6153846153846154, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1022}, "ledger_by_hop": {"1": {"facts_total": 63, "facts_retained": 27, "retention_rate": 0.42857142857142855, "retention_by_kind": {"person": 0.5, "date": 0.125, "amount": 0.6153846153846154, "status": 0.75, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1087}, "2": {"facts_total": 63, "facts_retained": 24, "retention_rate": 0.38095238095238093, "retention_by_kind": {"person": 0.5, "date": 0.125, "amount": 0.6153846153846154, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1063}, "3": {"facts_total": 63, "facts_retained": 24, "retention_rate": 0.38095238095238093, "retention_by_kind": {"person": 0.5, "date": 0.125, "amount": 0.6153846153846154, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1062}, "4": {"facts_total": 63, "facts_retained": 24, "retention_rate": 0.38095238095238093, "retention_by_kind": {"person": 0.5, "date": 0.125, "amount": 0.6153846153846154, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1023}, "5": {"facts_total": 63, "facts_retained": 24, "retention_rate": 0.38095238095238093, "retention_by_kind": {"person": 0.5, "date": 0.125, "amount": 0.6153846153846154, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1050}, "6": {"facts_total": 63, "facts_retained": 24, "retention_rate": 0.38095238095238093, "retention_by_kind": {"person": 0.5, "date": 0.125, "amount": 0.6153846153846154, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1010}, "7": {"facts_total": 63, "facts_retained": 24, "retention_rate": 0.38095238095238093, "retention_by_kind": {"person": 0.5, "date": 0.125, "amount": 0.6153846153846154, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 992}, "8": {"facts_total": 63, "facts_retained": 24, "retention_rate": 0.38095238095238093, "retention_by_kind": {"person": 0.5, "date": 0.125, "amount": 0.6153846153846154, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1022}}}, "reference_key_id": "keys/synthetic/syn_03027_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 0.4}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "e5864c0b88a688a045e4eb1d1985303e81b16f7e80826e759450dac2bb9d2e48", "output_hash": "c5fb5d9b76b7a02439f0acb83233430b95eef4aa96f8df88b9152c3b3e51343c"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "74e98c76a3302f7e1645398b", "attempt_id": "730775e135cc", "run_group_id": "E4_core:A4|closed|qwen3b-llama3b:h8", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:33:39.350078Z", "task_family": "synthetic", "document_id": "syn_03027_2000", "document_hash": "10326080979f71b625e4110c7c19767e5a752e267de5034f7e37d98cb6e06249", "question_id": "q12_unit_conversion", "split": "test", "cluster_id": "syn_03027_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "tokenizer_hashes": {"receiver": "b9d8872cf3a99da99e1fc4b0f71232a451632dc49ca4c6771549b4efa5c1c763"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 8, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1760, "query_tokens": 499, "message_tokens": 387, "message_utf8_bytes": 1022, "cache_tensor_bytes": null, "serialized_payload_bytes": 1022, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 44624.60495799314, "hops": {"1": 6872.302875039168, "2": 5042.6992499269545, "3": 6354.470000020228, "4": 4713.825125014409, "5": 6335.791665944271, "6": 4672.110749990679, "7": 5924.405542085879, "8": 4708.999749971554}, "receiver_prefill_ms": 207.86812505684793, "ttft_ms": 11.652125045657158, "decode_ms": 11.538667022250593}, "total_wall_ms": 44856.26437398605, "ttft_ms": 11.652125045657158, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 63, "facts_retained": 24, "retention_rate": 0.38095238095238093, "retention_by_kind": {"person": 0.5, "date": 0.125, "amount": 0.6153846153846154, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1022}, "ledger_by_hop": {"1": {"facts_total": 63, "facts_retained": 27, "retention_rate": 0.42857142857142855, "retention_by_kind": {"person": 0.5, "date": 0.125, "amount": 0.6153846153846154, "status": 0.75, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1087}, "2": {"facts_total": 63, "facts_retained": 24, "retention_rate": 0.38095238095238093, "retention_by_kind": {"person": 0.5, "date": 0.125, "amount": 0.6153846153846154, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1063}, "3": {"facts_total": 63, "facts_retained": 24, "retention_rate": 0.38095238095238093, "retention_by_kind": {"person": 0.5, "date": 0.125, "amount": 0.6153846153846154, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1062}, "4": {"facts_total": 63, "facts_retained": 24, "retention_rate": 0.38095238095238093, "retention_by_kind": {"person": 0.5, "date": 0.125, "amount": 0.6153846153846154, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1023}, "5": {"facts_total": 63, "facts_retained": 24, "retention_rate": 0.38095238095238093, "retention_by_kind": {"person": 0.5, "date": 0.125, "amount": 0.6153846153846154, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1050}, "6": {"facts_total": 63, "facts_retained": 24, "retention_rate": 0.38095238095238093, "retention_by_kind": {"person": 0.5, "date": 0.125, "amount": 0.6153846153846154, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1010}, "7": {"facts_total": 63, "facts_retained": 24, "retention_rate": 0.38095238095238093, "retention_by_kind": {"person": 0.5, "date": 0.125, "amount": 0.6153846153846154, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 992}, "8": {"facts_total": 63, "facts_retained": 24, "retention_rate": 0.38095238095238093, "retention_by_kind": {"person": 0.5, "date": 0.125, "amount": 0.6153846153846154, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1022}}}, "reference_key_id": "keys/synthetic/syn_03027_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "e5864c0b88a688a045e4eb1d1985303e81b16f7e80826e759450dac2bb9d2e48", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "30ef5b3e969734698441069d", "attempt_id": "d2f6f26e05e1", "run_group_id": "E4_core:A4|closed|qwen3b-llama3b:h8", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:33:39.597157Z", "task_family": "synthetic", "document_id": "syn_03027_2000", "document_hash": "10326080979f71b625e4110c7c19767e5a752e267de5034f7e37d98cb6e06249", "question_id": "q13_unanswerable", "split": "test", "cluster_id": "syn_03027_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "tokenizer_hashes": {"receiver": "b9d8872cf3a99da99e1fc4b0f71232a451632dc49ca4c6771549b4efa5c1c763"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 8, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1760, "query_tokens": 503, "message_tokens": 387, "message_utf8_bytes": 1022, "cache_tensor_bytes": null, "serialized_payload_bytes": 1022, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 44624.60495799314, "hops": {"1": 6872.302875039168, "2": 5042.6992499269545, "3": 6354.470000020228, "4": 4713.825125014409, "5": 6335.791665944271, "6": 4672.110749990679, "7": 5924.405542085879, "8": 4708.999749971554}, "receiver_prefill_ms": 208.42470810748637, "ttft_ms": 11.685249977745116, "decode_ms": 11.51199999731034}, "total_wall_ms": 44856.84912500437, "ttft_ms": 11.685249977745116, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 63, "facts_retained": 24, "retention_rate": 0.38095238095238093, "retention_by_kind": {"person": 0.5, "date": 0.125, "amount": 0.6153846153846154, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1022}, "ledger_by_hop": {"1": {"facts_total": 63, "facts_retained": 27, "retention_rate": 0.42857142857142855, "retention_by_kind": {"person": 0.5, "date": 0.125, "amount": 0.6153846153846154, "status": 0.75, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1087}, "2": {"facts_total": 63, "facts_retained": 24, "retention_rate": 0.38095238095238093, "retention_by_kind": {"person": 0.5, "date": 0.125, "amount": 0.6153846153846154, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1063}, "3": {"facts_total": 63, "facts_retained": 24, "retention_rate": 0.38095238095238093, "retention_by_kind": {"person": 0.5, "date": 0.125, "amount": 0.6153846153846154, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1062}, "4": {"facts_total": 63, "facts_retained": 24, "retention_rate": 0.38095238095238093, "retention_by_kind": {"person": 0.5, "date": 0.125, "amount": 0.6153846153846154, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1023}, "5": {"facts_total": 63, "facts_retained": 24, "retention_rate": 0.38095238095238093, "retention_by_kind": {"person": 0.5, "date": 0.125, "amount": 0.6153846153846154, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1050}, "6": {"facts_total": 63, "facts_retained": 24, "retention_rate": 0.38095238095238093, "retention_by_kind": {"person": 0.5, "date": 0.125, "amount": 0.6153846153846154, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1010}, "7": {"facts_total": 63, "facts_retained": 24, "retention_rate": 0.38095238095238093, "retention_by_kind": {"person": 0.5, "date": 0.125, "amount": 0.6153846153846154, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 992}, "8": {"facts_total": 63, "facts_retained": 24, "retention_rate": 0.38095238095238093, "retention_by_kind": {"person": 0.5, "date": 0.125, "amount": 0.6153846153846154, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1022}}}, "reference_key_id": "keys/synthetic/syn_03027_2000.json", "task_scores": {"correct": 1.0, "abstained": 1.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "e5864c0b88a688a045e4eb1d1985303e81b16f7e80826e759450dac2bb9d2e48", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "25715c96bfa5e7120815644e", "attempt_id": "c8b533206d63", "run_group_id": "E4_core:A4|closed|qwen3b-llama3b:h8", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:33:39.845092Z", "task_family": "synthetic", "document_id": "syn_03027_2000", "document_hash": "10326080979f71b625e4110c7c19767e5a752e267de5034f7e37d98cb6e06249", "question_id": "q14_unanswerable", "split": "test", "cluster_id": "syn_03027_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "tokenizer_hashes": {"receiver": "b9d8872cf3a99da99e1fc4b0f71232a451632dc49ca4c6771549b4efa5c1c763"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A4", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 8, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1760, "query_tokens": 505, "message_tokens": 387, "message_utf8_bytes": 1022, "cache_tensor_bytes": null, "serialized_payload_bytes": 1022, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 44624.60495799314, "hops": {"1": 6872.302875039168, "2": 5042.6992499269545, "3": 6354.470000020228, "4": 4713.825125014409, "5": 6335.791665944271, "6": 4672.110749990679, "7": 5924.405542085879, "8": 4708.999749971554}, "receiver_prefill_ms": 207.27558399084955, "ttft_ms": 11.788583011366427, "decode_ms": 11.573042022064328}, "total_wall_ms": 44855.82375002559, "ttft_ms": 11.788583011366427, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 63, "facts_retained": 24, "retention_rate": 0.38095238095238093, "retention_by_kind": {"person": 0.5, "date": 0.125, "amount": 0.6153846153846154, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1022}, "ledger_by_hop": {"1": {"facts_total": 63, "facts_retained": 27, "retention_rate": 0.42857142857142855, "retention_by_kind": {"person": 0.5, "date": 0.125, "amount": 0.6153846153846154, "status": 0.75, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1087}, "2": {"facts_total": 63, "facts_retained": 24, "retention_rate": 0.38095238095238093, "retention_by_kind": {"person": 0.5, "date": 0.125, "amount": 0.6153846153846154, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1063}, "3": {"facts_total": 63, "facts_retained": 24, "retention_rate": 0.38095238095238093, "retention_by_kind": {"person": 0.5, "date": 0.125, "amount": 0.6153846153846154, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1062}, "4": {"facts_total": 63, "facts_retained": 24, "retention_rate": 0.38095238095238093, "retention_by_kind": {"person": 0.5, "date": 0.125, "amount": 0.6153846153846154, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1023}, "5": {"facts_total": 63, "facts_retained": 24, "retention_rate": 0.38095238095238093, "retention_by_kind": {"person": 0.5, "date": 0.125, "amount": 0.6153846153846154, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1050}, "6": {"facts_total": 63, "facts_retained": 24, "retention_rate": 0.38095238095238093, "retention_by_kind": {"person": 0.5, "date": 0.125, "amount": 0.6153846153846154, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1010}, "7": {"facts_total": 63, "facts_retained": 24, "retention_rate": 0.38095238095238093, "retention_by_kind": {"person": 0.5, "date": 0.125, "amount": 0.6153846153846154, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 992}, "8": {"facts_total": 63, "facts_retained": 24, "retention_rate": 0.38095238095238093, "retention_by_kind": {"person": 0.5, "date": 0.125, "amount": 0.6153846153846154, "status": 0.5, "place": 0.0, "identifier": 0.3333333333333333, "quantity": 0.3333333333333333}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 1022}}}, "reference_key_id": "keys/synthetic/syn_03027_2000.json", "task_scores": {"correct": 1.0, "abstained": 1.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "e5864c0b88a688a045e4eb1d1985303e81b16f7e80826e759450dac2bb9d2e48", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "5b8ad3e54d8c46a7a3708aaf", "attempt_id": "dfd63c4ee7a1", "run_group_id": "E4_core:A5|closed|qwen3b-llama3b:h1", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:33:46.997516Z", "task_family": "synthetic", "document_id": "syn_03027_2000", "document_hash": "10326080979f71b625e4110c7c19767e5a752e267de5034f7e37d98cb6e06249", "question_id": "q00_single_fact", "split": "test", "cluster_id": "syn_03027_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 1, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1955, "query_tokens": 411, "message_tokens": 329, "message_utf8_bytes": 827, "cache_tensor_bytes": null, "serialized_payload_bytes": 827, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 6902.532499982044, "hops": {"1": 6902.532499982044}, "receiver_prefill_ms": 162.03166695777327, "ttft_ms": 11.918416945263743, "decode_ms": 11.814957950264215}, "total_wall_ms": 7088.69258291088, "ttft_ms": 11.918416945263743, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 63, "facts_retained": 9, "retention_rate": 0.14285714285714285, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.25, "amount": 0.07692307692307693, "status": 0.25, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 787}, "ledger_by_hop": {"1": {"facts_total": 63, "facts_retained": 9, "retention_rate": 0.14285714285714285, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.25, "amount": 0.07692307692307693, "status": 0.25, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 787}}}, "reference_key_id": "keys/synthetic/syn_03027_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "a3bc07ae34a4f72dd11a6738ade8463df3fa03c063d19d58651b6ead2a13f850", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "25c983ff6d01a61fd9bc3a10", "attempt_id": "82a6c4a15660", "run_group_id": "E4_core:A5|closed|qwen3b-llama3b:h1", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:33:47.200062Z", "task_family": "synthetic", "document_id": "syn_03027_2000", "document_hash": "10326080979f71b625e4110c7c19767e5a752e267de5034f7e37d98cb6e06249", "question_id": "q01_single_fact", "split": "test", "cluster_id": "syn_03027_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 1, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1955, "query_tokens": 418, "message_tokens": 329, "message_utf8_bytes": 827, "cache_tensor_bytes": null, "serialized_payload_bytes": 827, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 6902.532499982044, "hops": {"1": 6902.532499982044}, "receiver_prefill_ms": 163.37533295154572, "ttft_ms": 11.79183297790587, "decode_ms": 11.748916003853083}, "total_wall_ms": 7089.861541986465, "ttft_ms": 11.79183297790587, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 63, "facts_retained": 9, "retention_rate": 0.14285714285714285, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.25, "amount": 0.07692307692307693, "status": 0.25, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 787}, "ledger_by_hop": {"1": {"facts_total": 63, "facts_retained": 9, "retention_rate": 0.14285714285714285, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.25, "amount": 0.07692307692307693, "status": 0.25, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 787}}}, "reference_key_id": "keys/synthetic/syn_03027_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "a3bc07ae34a4f72dd11a6738ade8463df3fa03c063d19d58651b6ead2a13f850", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "e666bc2522ce9a46111cef56", "attempt_id": "914b684296ff", "run_group_id": "E4_core:A5|closed|qwen3b-llama3b:h1", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:33:47.403041Z", "task_family": "synthetic", "document_id": "syn_03027_2000", "document_hash": "10326080979f71b625e4110c7c19767e5a752e267de5034f7e37d98cb6e06249", "question_id": "q02_exact_copy", "split": "test", "cluster_id": "syn_03027_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 1, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1955, "query_tokens": 417, "message_tokens": 329, "message_utf8_bytes": 827, "cache_tensor_bytes": null, "serialized_payload_bytes": 827, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 6902.532499982044, "hops": {"1": 6902.532499982044}, "receiver_prefill_ms": 162.71820897236466, "ttft_ms": 11.816457961685956, "decode_ms": 11.783291003666818}, "total_wall_ms": 7089.226332958788, "ttft_ms": 11.816457961685956, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 63, "facts_retained": 9, "retention_rate": 0.14285714285714285, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.25, "amount": 0.07692307692307693, "status": 0.25, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 787}, "ledger_by_hop": {"1": {"facts_total": 63, "facts_retained": 9, "retention_rate": 0.14285714285714285, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.25, "amount": 0.07692307692307693, "status": 0.25, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 787}}}, "reference_key_id": "keys/synthetic/syn_03027_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "a3bc07ae34a4f72dd11a6738ade8463df3fa03c063d19d58651b6ead2a13f850", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "c99549ec9e7f764bfd39fcae", "attempt_id": "82c1267cbf76", "run_group_id": "E4_core:A5|closed|qwen3b-llama3b:h1", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:33:47.605341Z", "task_family": "synthetic", "document_id": "syn_03027_2000", "document_hash": "10326080979f71b625e4110c7c19767e5a752e267de5034f7e37d98cb6e06249", "question_id": "q03_exact_copy", "split": "test", "cluster_id": "syn_03027_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 1, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1955, "query_tokens": 412, "message_tokens": 329, "message_utf8_bytes": 827, "cache_tensor_bytes": null, "serialized_payload_bytes": 827, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 6902.532499982044, "hops": {"1": 6902.532499982044}, "receiver_prefill_ms": 162.0434580836445, "ttft_ms": 11.916000046767294, "decode_ms": 106.1162919504568}, "total_wall_ms": 7183.022542041726, "ttft_ms": 11.916000046767294, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 9}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "Priyamvada Sørlie", "evidence_artifact": {"ledger": {"facts_total": 63, "facts_retained": 9, "retention_rate": 0.14285714285714285, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.25, "amount": 0.07692307692307693, "status": 0.25, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 787}, "ledger_by_hop": {"1": {"facts_total": 63, "facts_retained": 9, "retention_rate": 0.14285714285714285, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.25, "amount": 0.07692307692307693, "status": 0.25, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 787}}}, "reference_key_id": "keys/synthetic/syn_03027_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "a3bc07ae34a4f72dd11a6738ade8463df3fa03c063d19d58651b6ead2a13f850", "output_hash": "c5f554196ed23dd71ad995ec89f7142099d524dbb6534cc10d94a0c89d13e792"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "e30614e72986a0e897a531b7", "attempt_id": "c0ddc23ab826", "run_group_id": "E4_core:A5|closed|qwen3b-llama3b:h1", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:33:47.901315Z", "task_family": "synthetic", "document_id": "syn_03027_2000", "document_hash": "10326080979f71b625e4110c7c19767e5a752e267de5034f7e37d98cb6e06249", "question_id": "q04_multi_hop", "split": "test", "cluster_id": "syn_03027_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 1, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1955, "query_tokens": 423, "message_tokens": 329, "message_utf8_bytes": 827, "cache_tensor_bytes": null, "serialized_payload_bytes": 827, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 6902.532499982044, "hops": {"1": 6902.532499982044}, "receiver_prefill_ms": 163.66183292120695, "ttft_ms": 11.917000054381788, "decode_ms": 11.772415949963033}, "total_wall_ms": 7090.299208997749, "ttft_ms": 11.917000054381788, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 63, "facts_retained": 9, "retention_rate": 0.14285714285714285, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.25, "amount": 0.07692307692307693, "status": 0.25, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 787}, "ledger_by_hop": {"1": {"facts_total": 63, "facts_retained": 9, "retention_rate": 0.14285714285714285, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.25, "amount": 0.07692307692307693, "status": 0.25, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 787}}}, "reference_key_id": "keys/synthetic/syn_03027_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "a3bc07ae34a4f72dd11a6738ade8463df3fa03c063d19d58651b6ead2a13f850", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "809760b2e49d08da3c32504f", "attempt_id": "bc6711c52254", "run_group_id": "E4_core:A5|closed|qwen3b-llama3b:h1", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:33:48.104427Z", "task_family": "synthetic", "document_id": "syn_03027_2000", "document_hash": "10326080979f71b625e4110c7c19767e5a752e267de5034f7e37d98cb6e06249", "question_id": "q05_multi_hop", "split": "test", "cluster_id": "syn_03027_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 1, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1955, "query_tokens": 420, "message_tokens": 329, "message_utf8_bytes": 827, "cache_tensor_bytes": null, "serialized_payload_bytes": 827, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 6902.532499982044, "hops": {"1": 6902.532499982044}, "receiver_prefill_ms": 163.67008304223418, "ttft_ms": 11.832458898425102, "decode_ms": 117.6555419806391}, "total_wall_ms": 7196.079499903135, "ttft_ms": 11.832458898425102, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 10}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "WT-26M-9227", "evidence_artifact": {"ledger": {"facts_total": 63, "facts_retained": 9, "retention_rate": 0.14285714285714285, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.25, "amount": 0.07692307692307693, "status": 0.25, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 787}, "ledger_by_hop": {"1": {"facts_total": 63, "facts_retained": 9, "retention_rate": 0.14285714285714285, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.25, "amount": 0.07692307692307693, "status": 0.25, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 787}}}, "reference_key_id": "keys/synthetic/syn_03027_2000.json", "task_scores": {"correct": 0.0, "abstained": 0.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "a3bc07ae34a4f72dd11a6738ade8463df3fa03c063d19d58651b6ead2a13f850", "output_hash": "ef2302e0ab6f5a6951b11463fea177ba766b9d438afcaf6e6101ebfa3ba38aba"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "91115ee512e65260a24be95a", "attempt_id": "11bcf2b8f656", "run_group_id": "E4_core:A5|closed|qwen3b-llama3b:h1", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:33:48.413022Z", "task_family": "synthetic", "document_id": "syn_03027_2000", "document_hash": "10326080979f71b625e4110c7c19767e5a752e267de5034f7e37d98cb6e06249", "question_id": "q06_negation", "split": "test", "cluster_id": "syn_03027_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 1, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1955, "query_tokens": 412, "message_tokens": 329, "message_utf8_bytes": 827, "cache_tensor_bytes": null, "serialized_payload_bytes": 827, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 6902.532499982044, "hops": {"1": 6902.532499982044}, "receiver_prefill_ms": 162.16120903845876, "ttft_ms": 11.766124982386827, "decode_ms": 11.783291003666818}, "total_wall_ms": 7088.605041964911, "ttft_ms": 11.766124982386827, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 63, "facts_retained": 9, "retention_rate": 0.14285714285714285, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.25, "amount": 0.07692307692307693, "status": 0.25, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 787}, "ledger_by_hop": {"1": {"facts_total": 63, "facts_retained": 9, "retention_rate": 0.14285714285714285, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.25, "amount": 0.07692307692307693, "status": 0.25, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 787}}}, "reference_key_id": "keys/synthetic/syn_03027_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "a3bc07ae34a4f72dd11a6738ade8463df3fa03c063d19d58651b6ead2a13f850", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "3f4d8191f8a1a79816532c58", "attempt_id": "b46fb5cfe4cc", "run_group_id": "E4_core:A5|closed|qwen3b-llama3b:h1", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:33:48.614653Z", "task_family": "synthetic", "document_id": "syn_03027_2000", "document_hash": "10326080979f71b625e4110c7c19767e5a752e267de5034f7e37d98cb6e06249", "question_id": "q07_negation", "split": "test", "cluster_id": "syn_03027_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 1, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1955, "query_tokens": 413, "message_tokens": 329, "message_utf8_bytes": 827, "cache_tensor_bytes": null, "serialized_payload_bytes": 827, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 6902.532499982044, "hops": {"1": 6902.532499982044}, "receiver_prefill_ms": 162.3319589998573, "ttft_ms": 11.778834043070674, "decode_ms": 11.786999995820224}, "total_wall_ms": 7088.82975007873, "ttft_ms": 11.778834043070674, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 63, "facts_retained": 9, "retention_rate": 0.14285714285714285, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.25, "amount": 0.07692307692307693, "status": 0.25, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 787}, "ledger_by_hop": {"1": {"facts_total": 63, "facts_retained": 9, "retention_rate": 0.14285714285714285, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.25, "amount": 0.07692307692307693, "status": 0.25, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 787}}}, "reference_key_id": "keys/synthetic/syn_03027_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "a3bc07ae34a4f72dd11a6738ade8463df3fa03c063d19d58651b6ead2a13f850", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "e55da20ffcdcfedbbd152b0f", "attempt_id": "20baf1b80cb6", "run_group_id": "E4_core:A5|closed|qwen3b-llama3b:h1", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:33:48.816194Z", "task_family": "synthetic", "document_id": "syn_03027_2000", "document_hash": "10326080979f71b625e4110c7c19767e5a752e267de5034f7e37d98cb6e06249", "question_id": "q08_revision", "split": "test", "cluster_id": "syn_03027_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 1, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1955, "query_tokens": 427, "message_tokens": 329, "message_utf8_bytes": 827, "cache_tensor_bytes": null, "serialized_payload_bytes": 827, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 6902.532499982044, "hops": {"1": 6902.532499982044}, "receiver_prefill_ms": 163.74241700395942, "ttft_ms": 11.76379097159952, "decode_ms": 105.97279097419232}, "total_wall_ms": 7184.409208944999, "ttft_ms": 11.76379097159952, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 9}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "6,914,937", "evidence_artifact": {"ledger": {"facts_total": 63, "facts_retained": 9, "retention_rate": 0.14285714285714285, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.25, "amount": 0.07692307692307693, "status": 0.25, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 787}, "ledger_by_hop": {"1": {"facts_total": 63, "facts_retained": 9, "retention_rate": 0.14285714285714285, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.25, "amount": 0.07692307692307693, "status": 0.25, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 787}}}, "reference_key_id": "keys/synthetic/syn_03027_2000.json", "task_scores": {"correct": 0.0, "abstained": 0.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "a3bc07ae34a4f72dd11a6738ade8463df3fa03c063d19d58651b6ead2a13f850", "output_hash": "c56851abfdbf8851226a1e71c4911c616a691c490a883eb754e61bac551d004c"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "36267248c59b4bf66952492c", "attempt_id": "2d803970785d", "run_group_id": "E4_core:A5|closed|qwen3b-llama3b:h1", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:33:49.113299Z", "task_family": "synthetic", "document_id": "syn_03027_2000", "document_hash": "10326080979f71b625e4110c7c19767e5a752e267de5034f7e37d98cb6e06249", "question_id": "q09_revision", "split": "test", "cluster_id": "syn_03027_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 1, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1955, "query_tokens": 420, "message_tokens": 329, "message_utf8_bytes": 827, "cache_tensor_bytes": null, "serialized_payload_bytes": 827, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 6902.532499982044, "hops": {"1": 6902.532499982044}, "receiver_prefill_ms": 163.74362504575402, "ttft_ms": 11.74829201772809, "decode_ms": 94.459000043571}, "total_wall_ms": 7172.85324994009, "ttft_ms": 11.74829201772809, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 8}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "15828866", "evidence_artifact": {"ledger": {"facts_total": 63, "facts_retained": 9, "retention_rate": 0.14285714285714285, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.25, "amount": 0.07692307692307693, "status": 0.25, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 787}, "ledger_by_hop": {"1": {"facts_total": 63, "facts_retained": 9, "retention_rate": 0.14285714285714285, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.25, "amount": 0.07692307692307693, "status": 0.25, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 787}}}, "reference_key_id": "keys/synthetic/syn_03027_2000.json", "task_scores": {"correct": 0.0, "abstained": 0.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "a3bc07ae34a4f72dd11a6738ade8463df3fa03c063d19d58651b6ead2a13f850", "output_hash": "c9d49be6f876880a6dfe30589ff0b746c048ebaf1842118a883f5503a349a8b6"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "455749589724001613fb9977", "attempt_id": "edf9e1a9d292", "run_group_id": "E4_core:A5|closed|qwen3b-llama3b:h1", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:33:49.398808Z", "task_family": "synthetic", "document_id": "syn_03027_2000", "document_hash": "10326080979f71b625e4110c7c19767e5a752e267de5034f7e37d98cb6e06249", "question_id": "q10_temporal", "split": "test", "cluster_id": "syn_03027_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 1, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1955, "query_tokens": 430, "message_tokens": 329, "message_utf8_bytes": 827, "cache_tensor_bytes": null, "serialized_payload_bytes": 827, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 6902.532499982044, "hops": {"1": 6902.532499982044}, "receiver_prefill_ms": 163.61391602549702, "ttft_ms": 11.799250030890107, "decode_ms": 11.838750098831952}, "total_wall_ms": 7090.168667025864, "ttft_ms": 11.799250030890107, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 63, "facts_retained": 9, "retention_rate": 0.14285714285714285, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.25, "amount": 0.07692307692307693, "status": 0.25, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 787}, "ledger_by_hop": {"1": {"facts_total": 63, "facts_retained": 9, "retention_rate": 0.14285714285714285, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.25, "amount": 0.07692307692307693, "status": 0.25, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 787}}}, "reference_key_id": "keys/synthetic/syn_03027_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "a3bc07ae34a4f72dd11a6738ade8463df3fa03c063d19d58651b6ead2a13f850", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "930fdfaeba90af289a75cb42", "attempt_id": "210167b42167", "run_group_id": "E4_core:A5|closed|qwen3b-llama3b:h1", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:33:49.601819Z", "task_family": "synthetic", "document_id": "syn_03027_2000", "document_hash": "10326080979f71b625e4110c7c19767e5a752e267de5034f7e37d98cb6e06249", "question_id": "q11_aggregation", "split": "test", "cluster_id": "syn_03027_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 1, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1955, "query_tokens": 423, "message_tokens": 329, "message_utf8_bytes": 827, "cache_tensor_bytes": null, "serialized_payload_bytes": 827, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 6902.532499982044, "hops": {"1": 6902.532499982044}, "receiver_prefill_ms": 163.16475009080023, "ttft_ms": 12.144459062255919, "decode_ms": 82.49591593630612}, "total_wall_ms": 7160.804915940389, "ttft_ms": 12.144459062255919, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 7}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "6914937", "evidence_artifact": {"ledger": {"facts_total": 63, "facts_retained": 9, "retention_rate": 0.14285714285714285, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.25, "amount": 0.07692307692307693, "status": 0.25, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 787}, "ledger_by_hop": {"1": {"facts_total": 63, "facts_retained": 9, "retention_rate": 0.14285714285714285, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.25, "amount": 0.07692307692307693, "status": 0.25, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 787}}}, "reference_key_id": "keys/synthetic/syn_03027_2000.json", "task_scores": {"correct": 0.0, "abstained": 0.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "a3bc07ae34a4f72dd11a6738ade8463df3fa03c063d19d58651b6ead2a13f850", "output_hash": "534d40ded32f784c68307dd2386faed2a27bf39dcc27321ee6985e42ccd285e5"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "c83c1e1d1b2d5fc9ed136f8b", "attempt_id": "b24ebe69d32d", "run_group_id": "E4_core:A5|closed|qwen3b-llama3b:h1", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:33:49.875452Z", "task_family": "synthetic", "document_id": "syn_03027_2000", "document_hash": "10326080979f71b625e4110c7c19767e5a752e267de5034f7e37d98cb6e06249", "question_id": "q12_unit_conversion", "split": "test", "cluster_id": "syn_03027_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 1, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1955, "query_tokens": 420, "message_tokens": 329, "message_utf8_bytes": 827, "cache_tensor_bytes": null, "serialized_payload_bytes": 827, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 6902.532499982044, "hops": {"1": 6902.532499982044}, "receiver_prefill_ms": 163.4248330956325, "ttft_ms": 11.745458003133535, "decode_ms": 11.822083033621311}, "total_wall_ms": 7089.925166917965, "ttft_ms": 11.745458003133535, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 63, "facts_retained": 9, "retention_rate": 0.14285714285714285, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.25, "amount": 0.07692307692307693, "status": 0.25, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 787}, "ledger_by_hop": {"1": {"facts_total": 63, "facts_retained": 9, "retention_rate": 0.14285714285714285, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.25, "amount": 0.07692307692307693, "status": 0.25, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 787}}}, "reference_key_id": "keys/synthetic/syn_03027_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "a3bc07ae34a4f72dd11a6738ade8463df3fa03c063d19d58651b6ead2a13f850", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "a413e661e3f29a3499aa4cef", "attempt_id": "14096fe6c9a7", "run_group_id": "E4_core:A5|closed|qwen3b-llama3b:h1", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:33:50.078611Z", "task_family": "synthetic", "document_id": "syn_03027_2000", "document_hash": "10326080979f71b625e4110c7c19767e5a752e267de5034f7e37d98cb6e06249", "question_id": "q13_unanswerable", "split": "test", "cluster_id": "syn_03027_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 1, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1955, "query_tokens": 423, "message_tokens": 329, "message_utf8_bytes": 827, "cache_tensor_bytes": null, "serialized_payload_bytes": 827, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 6902.532499982044, "hops": {"1": 6902.532499982044}, "receiver_prefill_ms": 162.6319580245763, "ttft_ms": 12.003249954432249, "decode_ms": 11.87183300498873}, "total_wall_ms": 7089.54525005538, "ttft_ms": 12.003249954432249, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 63, "facts_retained": 9, "retention_rate": 0.14285714285714285, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.25, "amount": 0.07692307692307693, "status": 0.25, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 787}, "ledger_by_hop": {"1": {"facts_total": 63, "facts_retained": 9, "retention_rate": 0.14285714285714285, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.25, "amount": 0.07692307692307693, "status": 0.25, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 787}}}, "reference_key_id": "keys/synthetic/syn_03027_2000.json", "task_scores": {"correct": 1.0, "abstained": 1.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "a3bc07ae34a4f72dd11a6738ade8463df3fa03c063d19d58651b6ead2a13f850", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "7d67252d68708916ce139c40", "attempt_id": "8b7a0833495b", "run_group_id": "E4_core:A5|closed|qwen3b-llama3b:h1", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:33:50.281170Z", "task_family": "synthetic", "document_id": "syn_03027_2000", "document_hash": "10326080979f71b625e4110c7c19767e5a752e267de5034f7e37d98cb6e06249", "question_id": "q14_unanswerable", "split": "test", "cluster_id": "syn_03027_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11", "tokenizer_hashes": {"receiver": "5c5a439ba2677fe28e1dea9b22371b270ecd5e34acbc849bc619689605f9547b"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 1, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1955, "query_tokens": 425, "message_tokens": 329, "message_utf8_bytes": 827, "cache_tensor_bytes": null, "serialized_payload_bytes": 827, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 6902.532499982044, "hops": {"1": 6902.532499982044}, "receiver_prefill_ms": 163.75304199755192, "ttft_ms": 11.849457980133593, "decode_ms": 11.84250006917864}, "total_wall_ms": 7090.372124919668, "ttft_ms": 11.849457980133593, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 63, "facts_retained": 9, "retention_rate": 0.14285714285714285, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.25, "amount": 0.07692307692307693, "status": 0.25, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 787}, "ledger_by_hop": {"1": {"facts_total": 63, "facts_retained": 9, "retention_rate": 0.14285714285714285, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.25, "amount": 0.07692307692307693, "status": 0.25, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 787}}}, "reference_key_id": "keys/synthetic/syn_03027_2000.json", "task_scores": {"correct": 1.0, "abstained": 1.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "a3bc07ae34a4f72dd11a6738ade8463df3fa03c063d19d58651b6ead2a13f850", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "32f0aea542e323974c082c53", "attempt_id": "56b00936af11", "run_group_id": "E4_core:A5|closed|qwen3b-llama3b:h2", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:33:54.207317Z", "task_family": "synthetic", "document_id": "syn_03027_2000", "document_hash": "10326080979f71b625e4110c7c19767e5a752e267de5034f7e37d98cb6e06249", "question_id": "q00_single_fact", "split": "test", "cluster_id": "syn_03027_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "tokenizer_hashes": {"receiver": "b9d8872cf3a99da99e1fc4b0f71232a451632dc49ca4c6771549b4efa5c1c763"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 2, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1760, "query_tokens": 401, "message_tokens": 299, "message_utf8_bytes": 856, "cache_tensor_bytes": null, "serialized_payload_bytes": 856, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 10623.747832956724, "hops": {"1": 6902.532499982044, "2": 3721.2153329746798}, "receiver_prefill_ms": 185.11791701894253, "ttft_ms": 11.538624996319413, "decode_ms": 11.575874988920987}, "total_wall_ms": 10832.374625024386, "ttft_ms": 11.538624996319413, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 63, "facts_retained": 12, "retention_rate": 0.19047619047619047, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.25, "amount": 0.07692307692307693, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 816}, "ledger_by_hop": {"1": {"facts_total": 63, "facts_retained": 9, "retention_rate": 0.14285714285714285, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.25, "amount": 0.07692307692307693, "status": 0.25, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 787}, "2": {"facts_total": 63, "facts_retained": 12, "retention_rate": 0.19047619047619047, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.25, "amount": 0.07692307692307693, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 816}}}, "reference_key_id": "keys/synthetic/syn_03027_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "ba14cf3ffcb23a84821d65b144b85efdb6c26697d433b2af58366827b4c8eba4", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "52e00003bf89a2dd0db49ee0", "attempt_id": "6c3e2fed0be6", "run_group_id": "E4_core:A5|closed|qwen3b-llama3b:h2", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:33:54.432105Z", "task_family": "synthetic", "document_id": "syn_03027_2000", "document_hash": "10326080979f71b625e4110c7c19767e5a752e267de5034f7e37d98cb6e06249", "question_id": "q01_single_fact", "split": "test", "cluster_id": "syn_03027_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "tokenizer_hashes": {"receiver": "b9d8872cf3a99da99e1fc4b0f71232a451632dc49ca4c6771549b4efa5c1c763"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 2, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1760, "query_tokens": 408, "message_tokens": 299, "message_utf8_bytes": 856, "cache_tensor_bytes": null, "serialized_payload_bytes": 856, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 10623.747832956724, "hops": {"1": 6902.532499982044, "2": 3721.2153329746798}, "receiver_prefill_ms": 187.07454099785537, "ttft_ms": 11.601374950259924, "decode_ms": 11.515165911987424}, "total_wall_ms": 10834.40662396606, "ttft_ms": 11.601374950259924, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 63, "facts_retained": 12, "retention_rate": 0.19047619047619047, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.25, "amount": 0.07692307692307693, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 816}, "ledger_by_hop": {"1": {"facts_total": 63, "facts_retained": 9, "retention_rate": 0.14285714285714285, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.25, "amount": 0.07692307692307693, "status": 0.25, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 787}, "2": {"facts_total": 63, "facts_retained": 12, "retention_rate": 0.19047619047619047, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.25, "amount": 0.07692307692307693, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 816}}}, "reference_key_id": "keys/synthetic/syn_03027_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "ba14cf3ffcb23a84821d65b144b85efdb6c26697d433b2af58366827b4c8eba4", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "076477e095094cf954531adc", "attempt_id": "6feeefafeacf", "run_group_id": "E4_core:A5|closed|qwen3b-llama3b:h2", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:33:54.658401Z", "task_family": "synthetic", "document_id": "syn_03027_2000", "document_hash": "10326080979f71b625e4110c7c19767e5a752e267de5034f7e37d98cb6e06249", "question_id": "q02_exact_copy", "split": "test", "cluster_id": "syn_03027_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "tokenizer_hashes": {"receiver": "b9d8872cf3a99da99e1fc4b0f71232a451632dc49ca4c6771549b4efa5c1c763"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 2, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1760, "query_tokens": 407, "message_tokens": 299, "message_utf8_bytes": 856, "cache_tensor_bytes": null, "serialized_payload_bytes": 856, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 10623.747832956724, "hops": {"1": 6902.532499982044, "2": 3721.2153329746798}, "receiver_prefill_ms": 185.49241696018726, "ttft_ms": 11.523708933964372, "decode_ms": 11.476584011688828}, "total_wall_ms": 10832.613540929742, "ttft_ms": 11.523708933964372, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 63, "facts_retained": 12, "retention_rate": 0.19047619047619047, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.25, "amount": 0.07692307692307693, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 816}, "ledger_by_hop": {"1": {"facts_total": 63, "facts_retained": 9, "retention_rate": 0.14285714285714285, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.25, "amount": 0.07692307692307693, "status": 0.25, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 787}, "2": {"facts_total": 63, "facts_retained": 12, "retention_rate": 0.19047619047619047, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.25, "amount": 0.07692307692307693, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 816}}}, "reference_key_id": "keys/synthetic/syn_03027_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "ba14cf3ffcb23a84821d65b144b85efdb6c26697d433b2af58366827b4c8eba4", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "80df95069eef4bba0c71069b", "attempt_id": "9441a5d15ba2", "run_group_id": "E4_core:A5|closed|qwen3b-llama3b:h2", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:33:54.883030Z", "task_family": "synthetic", "document_id": "syn_03027_2000", "document_hash": "10326080979f71b625e4110c7c19767e5a752e267de5034f7e37d98cb6e06249", "question_id": "q03_exact_copy", "split": "test", "cluster_id": "syn_03027_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "tokenizer_hashes": {"receiver": "b9d8872cf3a99da99e1fc4b0f71232a451632dc49ca4c6771549b4efa5c1c763"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 2, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1760, "query_tokens": 403, "message_tokens": 299, "message_utf8_bytes": 856, "cache_tensor_bytes": null, "serialized_payload_bytes": 856, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 10623.747832956724, "hops": {"1": 6902.532499982044, "2": 3721.2153329746798}, "receiver_prefill_ms": 185.39558397606015, "ttft_ms": 11.498625040985644, "decode_ms": 103.44258300028741}, "total_wall_ms": 10924.560333020054, "ttft_ms": 11.498625040985644, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 9}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "Priyamvada Sørlie", "evidence_artifact": {"ledger": {"facts_total": 63, "facts_retained": 12, "retention_rate": 0.19047619047619047, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.25, "amount": 0.07692307692307693, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 816}, "ledger_by_hop": {"1": {"facts_total": 63, "facts_retained": 9, "retention_rate": 0.14285714285714285, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.25, "amount": 0.07692307692307693, "status": 0.25, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 787}, "2": {"facts_total": 63, "facts_retained": 12, "retention_rate": 0.19047619047619047, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.25, "amount": 0.07692307692307693, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 816}}}, "reference_key_id": "keys/synthetic/syn_03027_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "ba14cf3ffcb23a84821d65b144b85efdb6c26697d433b2af58366827b4c8eba4", "output_hash": "c5f554196ed23dd71ad995ec89f7142099d524dbb6534cc10d94a0c89d13e792"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "f7ef7c35ce0bb2b23a36107c", "attempt_id": "f85e1169a0c5", "run_group_id": "E4_core:A5|closed|qwen3b-llama3b:h2", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:33:55.198965Z", "task_family": "synthetic", "document_id": "syn_03027_2000", "document_hash": "10326080979f71b625e4110c7c19767e5a752e267de5034f7e37d98cb6e06249", "question_id": "q04_multi_hop", "split": "test", "cluster_id": "syn_03027_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "tokenizer_hashes": {"receiver": "b9d8872cf3a99da99e1fc4b0f71232a451632dc49ca4c6771549b4efa5c1c763"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 2, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1760, "query_tokens": 414, "message_tokens": 299, "message_utf8_bytes": 856, "cache_tensor_bytes": null, "serialized_payload_bytes": 856, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 10623.747832956724, "hops": {"1": 6902.532499982044, "2": 3721.2153329746798}, "receiver_prefill_ms": 182.9623329686001, "ttft_ms": 11.450292076915503, "decode_ms": 11.449708021245897}, "total_wall_ms": 10829.991040984169, "ttft_ms": 11.450292076915503, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 63, "facts_retained": 12, "retention_rate": 0.19047619047619047, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.25, "amount": 0.07692307692307693, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 816}, "ledger_by_hop": {"1": {"facts_total": 63, "facts_retained": 9, "retention_rate": 0.14285714285714285, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.25, "amount": 0.07692307692307693, "status": 0.25, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 787}, "2": {"facts_total": 63, "facts_retained": 12, "retention_rate": 0.19047619047619047, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.25, "amount": 0.07692307692307693, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 816}}}, "reference_key_id": "keys/synthetic/syn_03027_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "ba14cf3ffcb23a84821d65b144b85efdb6c26697d433b2af58366827b4c8eba4", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "47ba67bdea77540ff06ab780", "attempt_id": "ca09a65cbbf2", "run_group_id": "E4_core:A5|closed|qwen3b-llama3b:h2", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:33:55.420757Z", "task_family": "synthetic", "document_id": "syn_03027_2000", "document_hash": "10326080979f71b625e4110c7c19767e5a752e267de5034f7e37d98cb6e06249", "question_id": "q05_multi_hop", "split": "test", "cluster_id": "syn_03027_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "tokenizer_hashes": {"receiver": "b9d8872cf3a99da99e1fc4b0f71232a451632dc49ca4c6771549b4efa5c1c763"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 2, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1760, "query_tokens": 409, "message_tokens": 299, "message_utf8_bytes": 856, "cache_tensor_bytes": null, "serialized_payload_bytes": 856, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 10623.747832956724, "hops": {"1": 6902.532499982044, "2": 3721.2153329746798}, "receiver_prefill_ms": 183.7443329859525, "ttft_ms": 11.527374968864024, "decode_ms": 11.427958030253649}, "total_wall_ms": 10830.80004202202, "ttft_ms": 11.527374968864024, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 63, "facts_retained": 12, "retention_rate": 0.19047619047619047, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.25, "amount": 0.07692307692307693, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 816}, "ledger_by_hop": {"1": {"facts_total": 63, "facts_retained": 9, "retention_rate": 0.14285714285714285, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.25, "amount": 0.07692307692307693, "status": 0.25, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 787}, "2": {"facts_total": 63, "facts_retained": 12, "retention_rate": 0.19047619047619047, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.25, "amount": 0.07692307692307693, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 816}}}, "reference_key_id": "keys/synthetic/syn_03027_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "ba14cf3ffcb23a84821d65b144b85efdb6c26697d433b2af58366827b4c8eba4", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "b341bb9b289354fbbef778a0", "attempt_id": "efc9f2707e87", "run_group_id": "E4_core:A5|closed|qwen3b-llama3b:h2", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:33:55.642982Z", "task_family": "synthetic", "document_id": "syn_03027_2000", "document_hash": "10326080979f71b625e4110c7c19767e5a752e267de5034f7e37d98cb6e06249", "question_id": "q06_negation", "split": "test", "cluster_id": "syn_03027_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "tokenizer_hashes": {"receiver": "b9d8872cf3a99da99e1fc4b0f71232a451632dc49ca4c6771549b4efa5c1c763"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 2, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1760, "query_tokens": 403, "message_tokens": 299, "message_utf8_bytes": 856, "cache_tensor_bytes": null, "serialized_payload_bytes": 856, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 10623.747832956724, "hops": {"1": 6902.532499982044, "2": 3721.2153329746798}, "receiver_prefill_ms": 184.20104193501174, "ttft_ms": 11.518582934513688, "decode_ms": 11.543250060640275}, "total_wall_ms": 10831.360082956962, "ttft_ms": 11.518582934513688, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 63, "facts_retained": 12, "retention_rate": 0.19047619047619047, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.25, "amount": 0.07692307692307693, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 816}, "ledger_by_hop": {"1": {"facts_total": 63, "facts_retained": 9, "retention_rate": 0.14285714285714285, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.25, "amount": 0.07692307692307693, "status": 0.25, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 787}, "2": {"facts_total": 63, "facts_retained": 12, "retention_rate": 0.19047619047619047, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.25, "amount": 0.07692307692307693, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 816}}}, "reference_key_id": "keys/synthetic/syn_03027_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "ba14cf3ffcb23a84821d65b144b85efdb6c26697d433b2af58366827b4c8eba4", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "808fe4967e57b1088dbbd2fb", "attempt_id": "ff20fe126819", "run_group_id": "E4_core:A5|closed|qwen3b-llama3b:h2", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:33:55.865367Z", "task_family": "synthetic", "document_id": "syn_03027_2000", "document_hash": "10326080979f71b625e4110c7c19767e5a752e267de5034f7e37d98cb6e06249", "question_id": "q07_negation", "split": "test", "cluster_id": "syn_03027_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "tokenizer_hashes": {"receiver": "b9d8872cf3a99da99e1fc4b0f71232a451632dc49ca4c6771549b4efa5c1c763"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 2, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1760, "query_tokens": 403, "message_tokens": 299, "message_utf8_bytes": 856, "cache_tensor_bytes": null, "serialized_payload_bytes": 856, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 10623.747832956724, "hops": {"1": 6902.532499982044, "2": 3721.2153329746798}, "receiver_prefill_ms": 186.58191699068993, "ttft_ms": 11.780000058934093, "decode_ms": 11.600499972701073}, "total_wall_ms": 10834.17795796413, "ttft_ms": 11.780000058934093, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "No", "evidence_artifact": {"ledger": {"facts_total": 63, "facts_retained": 12, "retention_rate": 0.19047619047619047, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.25, "amount": 0.07692307692307693, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 816}, "ledger_by_hop": {"1": {"facts_total": 63, "facts_retained": 9, "retention_rate": 0.14285714285714285, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.25, "amount": 0.07692307692307693, "status": 0.25, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 787}, "2": {"facts_total": 63, "facts_retained": 12, "retention_rate": 0.19047619047619047, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.25, "amount": 0.07692307692307693, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 816}}}, "reference_key_id": "keys/synthetic/syn_03027_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "ba14cf3ffcb23a84821d65b144b85efdb6c26697d433b2af58366827b4c8eba4", "output_hash": "1ea442a134b2a184bd5d40104401f2a37fbc09ccf3f4bc9da161c6099be3691d"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "f3fc6c39901c8b347be8d999", "attempt_id": "f48ff9ebe6c6", "run_group_id": "E4_core:A5|closed|qwen3b-llama3b:h2", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:33:56.090508Z", "task_family": "synthetic", "document_id": "syn_03027_2000", "document_hash": "10326080979f71b625e4110c7c19767e5a752e267de5034f7e37d98cb6e06249", "question_id": "q08_revision", "split": "test", "cluster_id": "syn_03027_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "tokenizer_hashes": {"receiver": "b9d8872cf3a99da99e1fc4b0f71232a451632dc49ca4c6771549b4efa5c1c763"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 2, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1760, "query_tokens": 418, "message_tokens": 299, "message_utf8_bytes": 856, "cache_tensor_bytes": null, "serialized_payload_bytes": 856, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 10623.747832956724, "hops": {"1": 6902.532499982044, "2": 3721.2153329746798}, "receiver_prefill_ms": 184.23929205164313, "ttft_ms": 11.577625060454011, "decode_ms": 57.389084016904235}, "total_wall_ms": 10877.343916916288, "ttft_ms": 11.577625060454011, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 5}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "6,914,937", "evidence_artifact": {"ledger": {"facts_total": 63, "facts_retained": 12, "retention_rate": 0.19047619047619047, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.25, "amount": 0.07692307692307693, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 816}, "ledger_by_hop": {"1": {"facts_total": 63, "facts_retained": 9, "retention_rate": 0.14285714285714285, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.25, "amount": 0.07692307692307693, "status": 0.25, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 787}, "2": {"facts_total": 63, "facts_retained": 12, "retention_rate": 0.19047619047619047, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.25, "amount": 0.07692307692307693, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 816}}}, "reference_key_id": "keys/synthetic/syn_03027_2000.json", "task_scores": {"correct": 0.0, "abstained": 0.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "ba14cf3ffcb23a84821d65b144b85efdb6c26697d433b2af58366827b4c8eba4", "output_hash": "c56851abfdbf8851226a1e71c4911c616a691c490a883eb754e61bac551d004c"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "0fd7e8fcf4255c723aee3f42", "attempt_id": "271530ae39d2", "run_group_id": "E4_core:A5|closed|qwen3b-llama3b:h2", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:33:56.359377Z", "task_family": "synthetic", "document_id": "syn_03027_2000", "document_hash": "10326080979f71b625e4110c7c19767e5a752e267de5034f7e37d98cb6e06249", "question_id": "q09_revision", "split": "test", "cluster_id": "syn_03027_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "tokenizer_hashes": {"receiver": "b9d8872cf3a99da99e1fc4b0f71232a451632dc49ca4c6771549b4efa5c1c763"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 2, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1760, "query_tokens": 411, "message_tokens": 299, "message_utf8_bytes": 856, "cache_tensor_bytes": null, "serialized_payload_bytes": 856, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 10623.747832956724, "hops": {"1": 6902.532499982044, "2": 3721.2153329746798}, "receiver_prefill_ms": 187.27079196833074, "ttft_ms": 11.509082978591323, "decode_ms": 57.34191706869751}, "total_wall_ms": 10880.241040955298, "ttft_ms": 11.509082978591323, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 5}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "7,000,000", "evidence_artifact": {"ledger": {"facts_total": 63, "facts_retained": 12, "retention_rate": 0.19047619047619047, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.25, "amount": 0.07692307692307693, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 816}, "ledger_by_hop": {"1": {"facts_total": 63, "facts_retained": 9, "retention_rate": 0.14285714285714285, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.25, "amount": 0.07692307692307693, "status": 0.25, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 787}, "2": {"facts_total": 63, "facts_retained": 12, "retention_rate": 0.19047619047619047, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.25, "amount": 0.07692307692307693, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 816}}}, "reference_key_id": "keys/synthetic/syn_03027_2000.json", "task_scores": {"correct": 0.0, "abstained": 0.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "ba14cf3ffcb23a84821d65b144b85efdb6c26697d433b2af58366827b4c8eba4", "output_hash": "a8f19157bfa97ae9789ef71ec65944fe20e98491b7b0c2aa047d68bf92a3e7fd"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "301fe2a9b9d7f0292e06b499", "attempt_id": "29911531c148", "run_group_id": "E4_core:A5|closed|qwen3b-llama3b:h2", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:33:56.630954Z", "task_family": "synthetic", "document_id": "syn_03027_2000", "document_hash": "10326080979f71b625e4110c7c19767e5a752e267de5034f7e37d98cb6e06249", "question_id": "q10_temporal", "split": "test", "cluster_id": "syn_03027_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "tokenizer_hashes": {"receiver": "b9d8872cf3a99da99e1fc4b0f71232a451632dc49ca4c6771549b4efa5c1c763"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 2, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1760, "query_tokens": 419, "message_tokens": 299, "message_utf8_bytes": 856, "cache_tensor_bytes": null, "serialized_payload_bytes": 856, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 10623.747832956724, "hops": {"1": 6902.532499982044, "2": 3721.2153329746798}, "receiver_prefill_ms": 181.3983329338953, "ttft_ms": 11.585332918912172, "decode_ms": 11.511708959005773}, "total_wall_ms": 10828.584041912109, "ttft_ms": 11.585332918912172, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 63, "facts_retained": 12, "retention_rate": 0.19047619047619047, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.25, "amount": 0.07692307692307693, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 816}, "ledger_by_hop": {"1": {"facts_total": 63, "facts_retained": 9, "retention_rate": 0.14285714285714285, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.25, "amount": 0.07692307692307693, "status": 0.25, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 787}, "2": {"facts_total": 63, "facts_retained": 12, "retention_rate": 0.19047619047619047, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.25, "amount": 0.07692307692307693, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 816}}}, "reference_key_id": "keys/synthetic/syn_03027_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "ba14cf3ffcb23a84821d65b144b85efdb6c26697d433b2af58366827b4c8eba4", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "d7600423fe9bb67750d9c6eb", "attempt_id": "736405ccda0f", "run_group_id": "E4_core:A5|closed|qwen3b-llama3b:h2", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:33:56.851089Z", "task_family": "synthetic", "document_id": "syn_03027_2000", "document_hash": "10326080979f71b625e4110c7c19767e5a752e267de5034f7e37d98cb6e06249", "question_id": "q11_aggregation", "split": "test", "cluster_id": "syn_03027_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "tokenizer_hashes": {"receiver": "b9d8872cf3a99da99e1fc4b0f71232a451632dc49ca4c6771549b4efa5c1c763"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 2, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1760, "query_tokens": 414, "message_tokens": 299, "message_utf8_bytes": 856, "cache_tensor_bytes": null, "serialized_payload_bytes": 856, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 10623.747832956724, "hops": {"1": 6902.532499982044, "2": 3721.2153329746798}, "receiver_prefill_ms": 184.55691705457866, "ttft_ms": 11.523167020641267, "decode_ms": 57.48795799445361}, "total_wall_ms": 10877.674916875549, "ttft_ms": 11.523167020641267, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 5}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "6,914,937", "evidence_artifact": {"ledger": {"facts_total": 63, "facts_retained": 12, "retention_rate": 0.19047619047619047, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.25, "amount": 0.07692307692307693, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 816}, "ledger_by_hop": {"1": {"facts_total": 63, "facts_retained": 9, "retention_rate": 0.14285714285714285, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.25, "amount": 0.07692307692307693, "status": 0.25, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 787}, "2": {"facts_total": 63, "facts_retained": 12, "retention_rate": 0.19047619047619047, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.25, "amount": 0.07692307692307693, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 816}}}, "reference_key_id": "keys/synthetic/syn_03027_2000.json", "task_scores": {"correct": 0.0, "abstained": 0.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "ba14cf3ffcb23a84821d65b144b85efdb6c26697d433b2af58366827b4c8eba4", "output_hash": "c56851abfdbf8851226a1e71c4911c616a691c490a883eb754e61bac551d004c"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "d8a893b6b7b40dcd1750232b", "attempt_id": "087350c8c1a0", "run_group_id": "E4_core:A5|closed|qwen3b-llama3b:h2", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:33:57.120063Z", "task_family": "synthetic", "document_id": "syn_03027_2000", "document_hash": "10326080979f71b625e4110c7c19767e5a752e267de5034f7e37d98cb6e06249", "question_id": "q12_unit_conversion", "split": "test", "cluster_id": "syn_03027_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "tokenizer_hashes": {"receiver": "b9d8872cf3a99da99e1fc4b0f71232a451632dc49ca4c6771549b4efa5c1c763"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 2, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1760, "query_tokens": 410, "message_tokens": 299, "message_utf8_bytes": 856, "cache_tensor_bytes": null, "serialized_payload_bytes": 856, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 10623.747832956724, "hops": {"1": 6902.532499982044, "2": 3721.2153329746798}, "receiver_prefill_ms": 185.09566702414304, "ttft_ms": 11.490083066746593, "decode_ms": 11.453374987468123}, "total_wall_ms": 10832.114374963567, "ttft_ms": 11.490083066746593, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 63, "facts_retained": 12, "retention_rate": 0.19047619047619047, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.25, "amount": 0.07692307692307693, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 816}, "ledger_by_hop": {"1": {"facts_total": 63, "facts_retained": 9, "retention_rate": 0.14285714285714285, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.25, "amount": 0.07692307692307693, "status": 0.25, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 787}, "2": {"facts_total": 63, "facts_retained": 12, "retention_rate": 0.19047619047619047, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.25, "amount": 0.07692307692307693, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 816}}}, "reference_key_id": "keys/synthetic/syn_03027_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "ba14cf3ffcb23a84821d65b144b85efdb6c26697d433b2af58366827b4c8eba4", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "e26ea3805eb9a9309fe7295b", "attempt_id": "f92eeca785e2", "run_group_id": "E4_core:A5|closed|qwen3b-llama3b:h2", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:33:57.343543Z", "task_family": "synthetic", "document_id": "syn_03027_2000", "document_hash": "10326080979f71b625e4110c7c19767e5a752e267de5034f7e37d98cb6e06249", "question_id": "q13_unanswerable", "split": "test", "cluster_id": "syn_03027_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "tokenizer_hashes": {"receiver": "b9d8872cf3a99da99e1fc4b0f71232a451632dc49ca4c6771549b4efa5c1c763"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 2, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1760, "query_tokens": 414, "message_tokens": 299, "message_utf8_bytes": 856, "cache_tensor_bytes": null, "serialized_payload_bytes": 856, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 10623.747832956724, "hops": {"1": 6902.532499982044, "2": 3721.2153329746798}, "receiver_prefill_ms": 184.01579197961837, "ttft_ms": 11.505834059789777, "decode_ms": 11.486624949611723}, "total_wall_ms": 10831.102123949677, "ttft_ms": 11.505834059789777, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 63, "facts_retained": 12, "retention_rate": 0.19047619047619047, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.25, "amount": 0.07692307692307693, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 816}, "ledger_by_hop": {"1": {"facts_total": 63, "facts_retained": 9, "retention_rate": 0.14285714285714285, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.25, "amount": 0.07692307692307693, "status": 0.25, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 787}, "2": {"facts_total": 63, "facts_retained": 12, "retention_rate": 0.19047619047619047, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.25, "amount": 0.07692307692307693, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 816}}}, "reference_key_id": "keys/synthetic/syn_03027_2000.json", "task_scores": {"correct": 1.0, "abstained": 1.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "ba14cf3ffcb23a84821d65b144b85efdb6c26697d433b2af58366827b4c8eba4", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "8b7d6b742cbb2bab36003ae9", "attempt_id": "90cb0f935966", "run_group_id": "E4_core:A5|closed|qwen3b-llama3b:h2", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:33:57.565991Z", "task_family": "synthetic", "document_id": "syn_03027_2000", "document_hash": "10326080979f71b625e4110c7c19767e5a752e267de5034f7e37d98cb6e06249", "question_id": "q14_unanswerable", "split": "test", "cluster_id": "syn_03027_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "tokenizer_hashes": {"receiver": "b9d8872cf3a99da99e1fc4b0f71232a451632dc49ca4c6771549b4efa5c1c763"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 2, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1760, "query_tokens": 416, "message_tokens": 299, "message_utf8_bytes": 856, "cache_tensor_bytes": null, "serialized_payload_bytes": 856, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 10623.747832956724, "hops": {"1": 6902.532499982044, "2": 3721.2153329746798}, "receiver_prefill_ms": 186.47879210766405, "ttft_ms": 11.604208033531904, "decode_ms": 11.550666997209191}, "total_wall_ms": 10833.764332928695, "ttft_ms": 11.604208033531904, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 63, "facts_retained": 12, "retention_rate": 0.19047619047619047, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.25, "amount": 0.07692307692307693, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 816}, "ledger_by_hop": {"1": {"facts_total": 63, "facts_retained": 9, "retention_rate": 0.14285714285714285, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.25, "amount": 0.07692307692307693, "status": 0.25, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 787}, "2": {"facts_total": 63, "facts_retained": 12, "retention_rate": 0.19047619047619047, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.25, "amount": 0.07692307692307693, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 816}}}, "reference_key_id": "keys/synthetic/syn_03027_2000.json", "task_scores": {"correct": 1.0, "abstained": 1.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "ba14cf3ffcb23a84821d65b144b85efdb6c26697d433b2af58366827b4c8eba4", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "282648cc1ad692181a94033b", "attempt_id": "f3a7c9017338", "run_group_id": "E4_core:A5|closed|qwen3b-llama3b:h4", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:34:06.145574Z", "task_family": "synthetic", "document_id": "syn_03027_2000", "document_hash": "10326080979f71b625e4110c7c19767e5a752e267de5034f7e37d98cb6e06249", "question_id": "q00_single_fact", "split": "test", "cluster_id": "syn_03027_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "tokenizer_hashes": {"receiver": "b9d8872cf3a99da99e1fc4b0f71232a451632dc49ca4c6771549b4efa5c1c763"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 4, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1760, "query_tokens": 427, "message_tokens": 325, "message_utf8_bytes": 923, "cache_tensor_bytes": null, "serialized_payload_bytes": 923, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 18975.824540830217, "hops": {"1": 6902.532499982044, "2": 3721.2153329746798, "3": 4287.0670419652015, "4": 4065.009665908292}, "receiver_prefill_ms": 183.1518328981474, "ttft_ms": 11.601875070482492, "decode_ms": 11.466708034276962}, "total_wall_ms": 19182.455581845716, "ttft_ms": 11.601875070482492, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 63, "facts_retained": 12, "retention_rate": 0.19047619047619047, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.25, "amount": 0.07692307692307693, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 923}, "ledger_by_hop": {"1": {"facts_total": 63, "facts_retained": 9, "retention_rate": 0.14285714285714285, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.25, "amount": 0.07692307692307693, "status": 0.25, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 787}, "2": {"facts_total": 63, "facts_retained": 12, "retention_rate": 0.19047619047619047, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.25, "amount": 0.07692307692307693, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 816}, "3": {"facts_total": 63, "facts_retained": 12, "retention_rate": 0.19047619047619047, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.25, "amount": 0.07692307692307693, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 816}, "4": {"facts_total": 63, "facts_retained": 12, "retention_rate": 0.19047619047619047, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.25, "amount": 0.07692307692307693, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 923}}}, "reference_key_id": "keys/synthetic/syn_03027_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "3e25d9c439f55699f2305c8554c1cca0a9a3f39bbfef22b0c8d64ebbde2b63e5", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "2efd215d6f379713b19ef1c5", "attempt_id": "aac35836d7d5", "run_group_id": "E4_core:A5|closed|qwen3b-llama3b:h4", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:34:06.368425Z", "task_family": "synthetic", "document_id": "syn_03027_2000", "document_hash": "10326080979f71b625e4110c7c19767e5a752e267de5034f7e37d98cb6e06249", "question_id": "q01_single_fact", "split": "test", "cluster_id": "syn_03027_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "tokenizer_hashes": {"receiver": "b9d8872cf3a99da99e1fc4b0f71232a451632dc49ca4c6771549b4efa5c1c763"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 4, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1760, "query_tokens": 434, "message_tokens": 325, "message_utf8_bytes": 923, "cache_tensor_bytes": null, "serialized_payload_bytes": 923, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 18975.824540830217, "hops": {"1": 6902.532499982044, "2": 3721.2153329746798, "3": 4287.0670419652015, "4": 4065.009665908292}, "receiver_prefill_ms": 184.1881669824943, "ttft_ms": 11.533332988619804, "decode_ms": 11.421707924455404}, "total_wall_ms": 19183.356457855552, "ttft_ms": 11.533332988619804, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 63, "facts_retained": 12, "retention_rate": 0.19047619047619047, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.25, "amount": 0.07692307692307693, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 923}, "ledger_by_hop": {"1": {"facts_total": 63, "facts_retained": 9, "retention_rate": 0.14285714285714285, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.25, "amount": 0.07692307692307693, "status": 0.25, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 787}, "2": {"facts_total": 63, "facts_retained": 12, "retention_rate": 0.19047619047619047, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.25, "amount": 0.07692307692307693, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 816}, "3": {"facts_total": 63, "facts_retained": 12, "retention_rate": 0.19047619047619047, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.25, "amount": 0.07692307692307693, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 816}, "4": {"facts_total": 63, "facts_retained": 12, "retention_rate": 0.19047619047619047, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.25, "amount": 0.07692307692307693, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 923}}}, "reference_key_id": "keys/synthetic/syn_03027_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "3e25d9c439f55699f2305c8554c1cca0a9a3f39bbfef22b0c8d64ebbde2b63e5", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "a199f98122c4ea0436580d6e", "attempt_id": "b61ed8163389", "run_group_id": "E4_core:A5|closed|qwen3b-llama3b:h4", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:34:06.591094Z", "task_family": "synthetic", "document_id": "syn_03027_2000", "document_hash": "10326080979f71b625e4110c7c19767e5a752e267de5034f7e37d98cb6e06249", "question_id": "q02_exact_copy", "split": "test", "cluster_id": "syn_03027_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "tokenizer_hashes": {"receiver": "b9d8872cf3a99da99e1fc4b0f71232a451632dc49ca4c6771549b4efa5c1c763"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 4, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1760, "query_tokens": 433, "message_tokens": 325, "message_utf8_bytes": 923, "cache_tensor_bytes": null, "serialized_payload_bytes": 923, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 18975.824540830217, "hops": {"1": 6902.532499982044, "2": 3721.2153329746798, "3": 4287.0670419652015, "4": 4065.009665908292}, "receiver_prefill_ms": 183.53512499015778, "ttft_ms": 11.808916926383972, "decode_ms": 11.57833298202604}, "total_wall_ms": 19183.29666578211, "ttft_ms": 11.808916926383972, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 63, "facts_retained": 12, "retention_rate": 0.19047619047619047, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.25, "amount": 0.07692307692307693, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 923}, "ledger_by_hop": {"1": {"facts_total": 63, "facts_retained": 9, "retention_rate": 0.14285714285714285, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.25, "amount": 0.07692307692307693, "status": 0.25, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 787}, "2": {"facts_total": 63, "facts_retained": 12, "retention_rate": 0.19047619047619047, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.25, "amount": 0.07692307692307693, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 816}, "3": {"facts_total": 63, "facts_retained": 12, "retention_rate": 0.19047619047619047, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.25, "amount": 0.07692307692307693, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 816}, "4": {"facts_total": 63, "facts_retained": 12, "retention_rate": 0.19047619047619047, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.25, "amount": 0.07692307692307693, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 923}}}, "reference_key_id": "keys/synthetic/syn_03027_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "3e25d9c439f55699f2305c8554c1cca0a9a3f39bbfef22b0c8d64ebbde2b63e5", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "ea875e8491bf9ad2111e80b0", "attempt_id": "a2079ddcb751", "run_group_id": "E4_core:A5|closed|qwen3b-llama3b:h4", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:34:06.814222Z", "task_family": "synthetic", "document_id": "syn_03027_2000", "document_hash": "10326080979f71b625e4110c7c19767e5a752e267de5034f7e37d98cb6e06249", "question_id": "q03_exact_copy", "split": "test", "cluster_id": "syn_03027_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "tokenizer_hashes": {"receiver": "b9d8872cf3a99da99e1fc4b0f71232a451632dc49ca4c6771549b4efa5c1c763"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 4, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1760, "query_tokens": 429, "message_tokens": 325, "message_utf8_bytes": 923, "cache_tensor_bytes": null, "serialized_payload_bytes": 923, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 18975.824540830217, "hops": {"1": 6902.532499982044, "2": 3721.2153329746798, "3": 4287.0670419652015, "4": 4065.009665908292}, "receiver_prefill_ms": 183.72191698290408, "ttft_ms": 11.70645805541426, "decode_ms": 103.66066603455693}, "total_wall_ms": 19275.43687389698, "ttft_ms": 11.70645805541426, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 9}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "Priyamvada Sørlie", "evidence_artifact": {"ledger": {"facts_total": 63, "facts_retained": 12, "retention_rate": 0.19047619047619047, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.25, "amount": 0.07692307692307693, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 923}, "ledger_by_hop": {"1": {"facts_total": 63, "facts_retained": 9, "retention_rate": 0.14285714285714285, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.25, "amount": 0.07692307692307693, "status": 0.25, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 787}, "2": {"facts_total": 63, "facts_retained": 12, "retention_rate": 0.19047619047619047, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.25, "amount": 0.07692307692307693, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 816}, "3": {"facts_total": 63, "facts_retained": 12, "retention_rate": 0.19047619047619047, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.25, "amount": 0.07692307692307693, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 816}, "4": {"facts_total": 63, "facts_retained": 12, "retention_rate": 0.19047619047619047, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.25, "amount": 0.07692307692307693, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 923}}}, "reference_key_id": "keys/synthetic/syn_03027_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "3e25d9c439f55699f2305c8554c1cca0a9a3f39bbfef22b0c8d64ebbde2b63e5", "output_hash": "c5f554196ed23dd71ad995ec89f7142099d524dbb6534cc10d94a0c89d13e792"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "d1146bf9fbcb72926ca626e1", "attempt_id": "e942b88622b5", "run_group_id": "E4_core:A5|closed|qwen3b-llama3b:h4", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:34:07.130065Z", "task_family": "synthetic", "document_id": "syn_03027_2000", "document_hash": "10326080979f71b625e4110c7c19767e5a752e267de5034f7e37d98cb6e06249", "question_id": "q04_multi_hop", "split": "test", "cluster_id": "syn_03027_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "tokenizer_hashes": {"receiver": "b9d8872cf3a99da99e1fc4b0f71232a451632dc49ca4c6771549b4efa5c1c763"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 4, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1760, "query_tokens": 440, "message_tokens": 325, "message_utf8_bytes": 923, "cache_tensor_bytes": null, "serialized_payload_bytes": 923, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 18975.824540830217, "hops": {"1": 6902.532499982044, "2": 3721.2153329746798, "3": 4287.0670419652015, "4": 4065.009665908292}, "receiver_prefill_ms": 184.64200000744313, "ttft_ms": 11.569666909053922, "decode_ms": 11.474541970528662}, "total_wall_ms": 19183.910374878906, "ttft_ms": 11.569666909053922, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 63, "facts_retained": 12, "retention_rate": 0.19047619047619047, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.25, "amount": 0.07692307692307693, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 923}, "ledger_by_hop": {"1": {"facts_total": 63, "facts_retained": 9, "retention_rate": 0.14285714285714285, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.25, "amount": 0.07692307692307693, "status": 0.25, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 787}, "2": {"facts_total": 63, "facts_retained": 12, "retention_rate": 0.19047619047619047, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.25, "amount": 0.07692307692307693, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 816}, "3": {"facts_total": 63, "facts_retained": 12, "retention_rate": 0.19047619047619047, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.25, "amount": 0.07692307692307693, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 816}, "4": {"facts_total": 63, "facts_retained": 12, "retention_rate": 0.19047619047619047, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.25, "amount": 0.07692307692307693, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 923}}}, "reference_key_id": "keys/synthetic/syn_03027_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "3e25d9c439f55699f2305c8554c1cca0a9a3f39bbfef22b0c8d64ebbde2b63e5", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "c424cebb58afa5480b073e7a", "attempt_id": "8dbcf864c9ef", "run_group_id": "E4_core:A5|closed|qwen3b-llama3b:h4", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:34:07.353743Z", "task_family": "synthetic", "document_id": "syn_03027_2000", "document_hash": "10326080979f71b625e4110c7c19767e5a752e267de5034f7e37d98cb6e06249", "question_id": "q05_multi_hop", "split": "test", "cluster_id": "syn_03027_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "tokenizer_hashes": {"receiver": "b9d8872cf3a99da99e1fc4b0f71232a451632dc49ca4c6771549b4efa5c1c763"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 4, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1760, "query_tokens": 435, "message_tokens": 325, "message_utf8_bytes": 923, "cache_tensor_bytes": null, "serialized_payload_bytes": 923, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 18975.824540830217, "hops": {"1": 6902.532499982044, "2": 3721.2153329746798, "3": 4287.0670419652015, "4": 4065.009665908292}, "receiver_prefill_ms": 185.20074989646673, "ttft_ms": 11.636749957688153, "decode_ms": 11.452458100393414}, "total_wall_ms": 19184.493873850442, "ttft_ms": 11.636749957688153, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 63, "facts_retained": 12, "retention_rate": 0.19047619047619047, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.25, "amount": 0.07692307692307693, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 923}, "ledger_by_hop": {"1": {"facts_total": 63, "facts_retained": 9, "retention_rate": 0.14285714285714285, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.25, "amount": 0.07692307692307693, "status": 0.25, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 787}, "2": {"facts_total": 63, "facts_retained": 12, "retention_rate": 0.19047619047619047, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.25, "amount": 0.07692307692307693, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 816}, "3": {"facts_total": 63, "facts_retained": 12, "retention_rate": 0.19047619047619047, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.25, "amount": 0.07692307692307693, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 816}, "4": {"facts_total": 63, "facts_retained": 12, "retention_rate": 0.19047619047619047, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.25, "amount": 0.07692307692307693, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 923}}}, "reference_key_id": "keys/synthetic/syn_03027_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "3e25d9c439f55699f2305c8554c1cca0a9a3f39bbfef22b0c8d64ebbde2b63e5", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "0cf9fa8bf456be126d1a70ac", "attempt_id": "5bf132d61906", "run_group_id": "E4_core:A5|closed|qwen3b-llama3b:h4", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:34:07.577705Z", "task_family": "synthetic", "document_id": "syn_03027_2000", "document_hash": "10326080979f71b625e4110c7c19767e5a752e267de5034f7e37d98cb6e06249", "question_id": "q06_negation", "split": "test", "cluster_id": "syn_03027_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "tokenizer_hashes": {"receiver": "b9d8872cf3a99da99e1fc4b0f71232a451632dc49ca4c6771549b4efa5c1c763"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 4, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1760, "query_tokens": 429, "message_tokens": 325, "message_utf8_bytes": 923, "cache_tensor_bytes": null, "serialized_payload_bytes": 923, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 18975.824540830217, "hops": {"1": 6902.532499982044, "2": 3721.2153329746798, "3": 4287.0670419652015, "4": 4065.009665908292}, "receiver_prefill_ms": 183.41195897664875, "ttft_ms": 11.66979200206697, "decode_ms": 11.54587499331683}, "total_wall_ms": 19182.796790846623, "ttft_ms": 11.66979200206697, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 63, "facts_retained": 12, "retention_rate": 0.19047619047619047, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.25, "amount": 0.07692307692307693, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 923}, "ledger_by_hop": {"1": {"facts_total": 63, "facts_retained": 9, "retention_rate": 0.14285714285714285, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.25, "amount": 0.07692307692307693, "status": 0.25, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 787}, "2": {"facts_total": 63, "facts_retained": 12, "retention_rate": 0.19047619047619047, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.25, "amount": 0.07692307692307693, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 816}, "3": {"facts_total": 63, "facts_retained": 12, "retention_rate": 0.19047619047619047, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.25, "amount": 0.07692307692307693, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 816}, "4": {"facts_total": 63, "facts_retained": 12, "retention_rate": 0.19047619047619047, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.25, "amount": 0.07692307692307693, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 923}}}, "reference_key_id": "keys/synthetic/syn_03027_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "3e25d9c439f55699f2305c8554c1cca0a9a3f39bbfef22b0c8d64ebbde2b63e5", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "877dbce9aad6b105b61fbc39", "attempt_id": "7e1897b6ffb6", "run_group_id": "E4_core:A5|closed|qwen3b-llama3b:h4", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:34:07.799523Z", "task_family": "synthetic", "document_id": "syn_03027_2000", "document_hash": "10326080979f71b625e4110c7c19767e5a752e267de5034f7e37d98cb6e06249", "question_id": "q07_negation", "split": "test", "cluster_id": "syn_03027_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "tokenizer_hashes": {"receiver": "b9d8872cf3a99da99e1fc4b0f71232a451632dc49ca4c6771549b4efa5c1c763"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 4, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1760, "query_tokens": 429, "message_tokens": 325, "message_utf8_bytes": 923, "cache_tensor_bytes": null, "serialized_payload_bytes": 923, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 18975.824540830217, "hops": {"1": 6902.532499982044, "2": 3721.2153329746798, "3": 4287.0670419652015, "4": 4065.009665908292}, "receiver_prefill_ms": 185.62191596720368, "ttft_ms": 11.542417109012604, "decode_ms": 11.489874916151166}, "total_wall_ms": 19184.848331846297, "ttft_ms": 11.542417109012604, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "No", "evidence_artifact": {"ledger": {"facts_total": 63, "facts_retained": 12, "retention_rate": 0.19047619047619047, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.25, "amount": 0.07692307692307693, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 923}, "ledger_by_hop": {"1": {"facts_total": 63, "facts_retained": 9, "retention_rate": 0.14285714285714285, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.25, "amount": 0.07692307692307693, "status": 0.25, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 787}, "2": {"facts_total": 63, "facts_retained": 12, "retention_rate": 0.19047619047619047, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.25, "amount": 0.07692307692307693, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 816}, "3": {"facts_total": 63, "facts_retained": 12, "retention_rate": 0.19047619047619047, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.25, "amount": 0.07692307692307693, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 816}, "4": {"facts_total": 63, "facts_retained": 12, "retention_rate": 0.19047619047619047, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.25, "amount": 0.07692307692307693, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 923}}}, "reference_key_id": "keys/synthetic/syn_03027_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "3e25d9c439f55699f2305c8554c1cca0a9a3f39bbfef22b0c8d64ebbde2b63e5", "output_hash": "1ea442a134b2a184bd5d40104401f2a37fbc09ccf3f4bc9da161c6099be3691d"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "9bf21bf5e9cd7b14dbf52c10", "attempt_id": "0a036e322eb1", "run_group_id": "E4_core:A5|closed|qwen3b-llama3b:h4", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:34:08.023148Z", "task_family": "synthetic", "document_id": "syn_03027_2000", "document_hash": "10326080979f71b625e4110c7c19767e5a752e267de5034f7e37d98cb6e06249", "question_id": "q08_revision", "split": "test", "cluster_id": "syn_03027_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "tokenizer_hashes": {"receiver": "b9d8872cf3a99da99e1fc4b0f71232a451632dc49ca4c6771549b4efa5c1c763"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 4, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1760, "query_tokens": 444, "message_tokens": 325, "message_utf8_bytes": 923, "cache_tensor_bytes": null, "serialized_payload_bytes": 923, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 18975.824540830217, "hops": {"1": 6902.532499982044, "2": 3721.2153329746798, "3": 4287.0670419652015, "4": 4065.009665908292}, "receiver_prefill_ms": 184.39075001515448, "ttft_ms": 11.536792037077248, "decode_ms": 57.62087507173419}, "total_wall_ms": 19229.782623820938, "ttft_ms": 11.536792037077248, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 5}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "6,914,937", "evidence_artifact": {"ledger": {"facts_total": 63, "facts_retained": 12, "retention_rate": 0.19047619047619047, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.25, "amount": 0.07692307692307693, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 923}, "ledger_by_hop": {"1": {"facts_total": 63, "facts_retained": 9, "retention_rate": 0.14285714285714285, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.25, "amount": 0.07692307692307693, "status": 0.25, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 787}, "2": {"facts_total": 63, "facts_retained": 12, "retention_rate": 0.19047619047619047, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.25, "amount": 0.07692307692307693, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 816}, "3": {"facts_total": 63, "facts_retained": 12, "retention_rate": 0.19047619047619047, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.25, "amount": 0.07692307692307693, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 816}, "4": {"facts_total": 63, "facts_retained": 12, "retention_rate": 0.19047619047619047, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.25, "amount": 0.07692307692307693, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 923}}}, "reference_key_id": "keys/synthetic/syn_03027_2000.json", "task_scores": {"correct": 0.0, "abstained": 0.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "3e25d9c439f55699f2305c8554c1cca0a9a3f39bbfef22b0c8d64ebbde2b63e5", "output_hash": "c56851abfdbf8851226a1e71c4911c616a691c490a883eb754e61bac551d004c"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "4991735f0c6d669ddb410cc9", "attempt_id": "8c0d7c661a06", "run_group_id": "E4_core:A5|closed|qwen3b-llama3b:h4", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:34:08.292353Z", "task_family": "synthetic", "document_id": "syn_03027_2000", "document_hash": "10326080979f71b625e4110c7c19767e5a752e267de5034f7e37d98cb6e06249", "question_id": "q09_revision", "split": "test", "cluster_id": "syn_03027_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "tokenizer_hashes": {"receiver": "b9d8872cf3a99da99e1fc4b0f71232a451632dc49ca4c6771549b4efa5c1c763"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 4, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1760, "query_tokens": 437, "message_tokens": 325, "message_utf8_bytes": 923, "cache_tensor_bytes": null, "serialized_payload_bytes": 923, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 18975.824540830217, "hops": {"1": 6902.532499982044, "2": 3721.2153329746798, "3": 4287.0670419652015, "4": 4065.009665908292}, "receiver_prefill_ms": 183.92350000794977, "ttft_ms": 11.530791060067713, "decode_ms": 57.63629195280373}, "total_wall_ms": 19229.3029158609, "ttft_ms": 11.530791060067713, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 5}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "7,000,000", "evidence_artifact": {"ledger": {"facts_total": 63, "facts_retained": 12, "retention_rate": 0.19047619047619047, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.25, "amount": 0.07692307692307693, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 923}, "ledger_by_hop": {"1": {"facts_total": 63, "facts_retained": 9, "retention_rate": 0.14285714285714285, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.25, "amount": 0.07692307692307693, "status": 0.25, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 787}, "2": {"facts_total": 63, "facts_retained": 12, "retention_rate": 0.19047619047619047, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.25, "amount": 0.07692307692307693, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 816}, "3": {"facts_total": 63, "facts_retained": 12, "retention_rate": 0.19047619047619047, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.25, "amount": 0.07692307692307693, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 816}, "4": {"facts_total": 63, "facts_retained": 12, "retention_rate": 0.19047619047619047, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.25, "amount": 0.07692307692307693, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 923}}}, "reference_key_id": "keys/synthetic/syn_03027_2000.json", "task_scores": {"correct": 0.0, "abstained": 0.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "3e25d9c439f55699f2305c8554c1cca0a9a3f39bbfef22b0c8d64ebbde2b63e5", "output_hash": "a8f19157bfa97ae9789ef71ec65944fe20e98491b7b0c2aa047d68bf92a3e7fd"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "c47c24168ce645996e58b291", "attempt_id": "46f84613dd6d", "run_group_id": "E4_core:A5|closed|qwen3b-llama3b:h4", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:34:08.561144Z", "task_family": "synthetic", "document_id": "syn_03027_2000", "document_hash": "10326080979f71b625e4110c7c19767e5a752e267de5034f7e37d98cb6e06249", "question_id": "q10_temporal", "split": "test", "cluster_id": "syn_03027_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "tokenizer_hashes": {"receiver": "b9d8872cf3a99da99e1fc4b0f71232a451632dc49ca4c6771549b4efa5c1c763"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 4, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1760, "query_tokens": 445, "message_tokens": 325, "message_utf8_bytes": 923, "cache_tensor_bytes": null, "serialized_payload_bytes": 923, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 18975.824540830217, "hops": {"1": 6902.532499982044, "2": 3721.2153329746798, "3": 4287.0670419652015, "4": 4065.009665908292}, "receiver_prefill_ms": 185.29229098930955, "ttft_ms": 11.526333983056247, "decode_ms": 11.494959006085992}, "total_wall_ms": 19184.505624813028, "ttft_ms": 11.526333983056247, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 63, "facts_retained": 12, "retention_rate": 0.19047619047619047, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.25, "amount": 0.07692307692307693, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 923}, "ledger_by_hop": {"1": {"facts_total": 63, "facts_retained": 9, "retention_rate": 0.14285714285714285, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.25, "amount": 0.07692307692307693, "status": 0.25, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 787}, "2": {"facts_total": 63, "facts_retained": 12, "retention_rate": 0.19047619047619047, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.25, "amount": 0.07692307692307693, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 816}, "3": {"facts_total": 63, "facts_retained": 12, "retention_rate": 0.19047619047619047, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.25, "amount": 0.07692307692307693, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 816}, "4": {"facts_total": 63, "facts_retained": 12, "retention_rate": 0.19047619047619047, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.25, "amount": 0.07692307692307693, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 923}}}, "reference_key_id": "keys/synthetic/syn_03027_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "3e25d9c439f55699f2305c8554c1cca0a9a3f39bbfef22b0c8d64ebbde2b63e5", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "6393fcd54e83bd5591ea952e", "attempt_id": "b729ecb000cf", "run_group_id": "E4_core:A5|closed|qwen3b-llama3b:h4", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:34:08.784875Z", "task_family": "synthetic", "document_id": "syn_03027_2000", "document_hash": "10326080979f71b625e4110c7c19767e5a752e267de5034f7e37d98cb6e06249", "question_id": "q11_aggregation", "split": "test", "cluster_id": "syn_03027_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "tokenizer_hashes": {"receiver": "b9d8872cf3a99da99e1fc4b0f71232a451632dc49ca4c6771549b4efa5c1c763"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 4, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1760, "query_tokens": 440, "message_tokens": 325, "message_utf8_bytes": 923, "cache_tensor_bytes": null, "serialized_payload_bytes": 923, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 18975.824540830217, "hops": {"1": 6902.532499982044, "2": 3721.2153329746798, "3": 4287.0670419652015, "4": 4065.009665908292}, "receiver_prefill_ms": 184.37124998308718, "ttft_ms": 11.541667045094073, "decode_ms": 57.602541986852884}, "total_wall_ms": 19229.72504084464, "ttft_ms": 11.541667045094073, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 5}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "6,914,937", "evidence_artifact": {"ledger": {"facts_total": 63, "facts_retained": 12, "retention_rate": 0.19047619047619047, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.25, "amount": 0.07692307692307693, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 923}, "ledger_by_hop": {"1": {"facts_total": 63, "facts_retained": 9, "retention_rate": 0.14285714285714285, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.25, "amount": 0.07692307692307693, "status": 0.25, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 787}, "2": {"facts_total": 63, "facts_retained": 12, "retention_rate": 0.19047619047619047, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.25, "amount": 0.07692307692307693, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 816}, "3": {"facts_total": 63, "facts_retained": 12, "retention_rate": 0.19047619047619047, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.25, "amount": 0.07692307692307693, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 816}, "4": {"facts_total": 63, "facts_retained": 12, "retention_rate": 0.19047619047619047, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.25, "amount": 0.07692307692307693, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 923}}}, "reference_key_id": "keys/synthetic/syn_03027_2000.json", "task_scores": {"correct": 0.0, "abstained": 0.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "3e25d9c439f55699f2305c8554c1cca0a9a3f39bbfef22b0c8d64ebbde2b63e5", "output_hash": "c56851abfdbf8851226a1e71c4911c616a691c490a883eb754e61bac551d004c"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "9a916e02221f5f654f3ea10a", "attempt_id": "773b8ba96a14", "run_group_id": "E4_core:A5|closed|qwen3b-llama3b:h4", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:34:09.054030Z", "task_family": "synthetic", "document_id": "syn_03027_2000", "document_hash": "10326080979f71b625e4110c7c19767e5a752e267de5034f7e37d98cb6e06249", "question_id": "q12_unit_conversion", "split": "test", "cluster_id": "syn_03027_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "tokenizer_hashes": {"receiver": "b9d8872cf3a99da99e1fc4b0f71232a451632dc49ca4c6771549b4efa5c1c763"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 4, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1760, "query_tokens": 436, "message_tokens": 325, "message_utf8_bytes": 923, "cache_tensor_bytes": null, "serialized_payload_bytes": 923, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 18975.824540830217, "hops": {"1": 6902.532499982044, "2": 3721.2153329746798, "3": 4287.0670419652015, "4": 4065.009665908292}, "receiver_prefill_ms": 185.26408297475427, "ttft_ms": 11.60491700284183, "decode_ms": 11.488249991089106}, "total_wall_ms": 19184.685706859455, "ttft_ms": 11.60491700284183, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 63, "facts_retained": 12, "retention_rate": 0.19047619047619047, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.25, "amount": 0.07692307692307693, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 923}, "ledger_by_hop": {"1": {"facts_total": 63, "facts_retained": 9, "retention_rate": 0.14285714285714285, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.25, "amount": 0.07692307692307693, "status": 0.25, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 787}, "2": {"facts_total": 63, "facts_retained": 12, "retention_rate": 0.19047619047619047, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.25, "amount": 0.07692307692307693, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 816}, "3": {"facts_total": 63, "facts_retained": 12, "retention_rate": 0.19047619047619047, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.25, "amount": 0.07692307692307693, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 816}, "4": {"facts_total": 63, "facts_retained": 12, "retention_rate": 0.19047619047619047, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.25, "amount": 0.07692307692307693, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 923}}}, "reference_key_id": "keys/synthetic/syn_03027_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "3e25d9c439f55699f2305c8554c1cca0a9a3f39bbfef22b0c8d64ebbde2b63e5", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "fabb394ca93a9fb5207b7189", "attempt_id": "a9469949540a", "run_group_id": "E4_core:A5|closed|qwen3b-llama3b:h4", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:34:09.278464Z", "task_family": "synthetic", "document_id": "syn_03027_2000", "document_hash": "10326080979f71b625e4110c7c19767e5a752e267de5034f7e37d98cb6e06249", "question_id": "q13_unanswerable", "split": "test", "cluster_id": "syn_03027_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "tokenizer_hashes": {"receiver": "b9d8872cf3a99da99e1fc4b0f71232a451632dc49ca4c6771549b4efa5c1c763"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 4, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1760, "query_tokens": 440, "message_tokens": 325, "message_utf8_bytes": 923, "cache_tensor_bytes": null, "serialized_payload_bytes": 923, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 18975.824540830217, "hops": {"1": 6902.532499982044, "2": 3721.2153329746798, "3": 4287.0670419652015, "4": 4065.009665908292}, "receiver_prefill_ms": 184.82229195069522, "ttft_ms": 11.583750019781291, "decode_ms": 11.56625000294298}, "total_wall_ms": 19184.16024884209, "ttft_ms": 11.583750019781291, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 63, "facts_retained": 12, "retention_rate": 0.19047619047619047, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.25, "amount": 0.07692307692307693, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 923}, "ledger_by_hop": {"1": {"facts_total": 63, "facts_retained": 9, "retention_rate": 0.14285714285714285, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.25, "amount": 0.07692307692307693, "status": 0.25, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 787}, "2": {"facts_total": 63, "facts_retained": 12, "retention_rate": 0.19047619047619047, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.25, "amount": 0.07692307692307693, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 816}, "3": {"facts_total": 63, "facts_retained": 12, "retention_rate": 0.19047619047619047, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.25, "amount": 0.07692307692307693, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 816}, "4": {"facts_total": 63, "facts_retained": 12, "retention_rate": 0.19047619047619047, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.25, "amount": 0.07692307692307693, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 923}}}, "reference_key_id": "keys/synthetic/syn_03027_2000.json", "task_scores": {"correct": 1.0, "abstained": 1.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "3e25d9c439f55699f2305c8554c1cca0a9a3f39bbfef22b0c8d64ebbde2b63e5", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "13d5741a3d04d8a6501b3f92", "attempt_id": "e7f662333103", "run_group_id": "E4_core:A5|closed|qwen3b-llama3b:h4", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:34:09.502472Z", "task_family": "synthetic", "document_id": "syn_03027_2000", "document_hash": "10326080979f71b625e4110c7c19767e5a752e267de5034f7e37d98cb6e06249", "question_id": "q14_unanswerable", "split": "test", "cluster_id": "syn_03027_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "tokenizer_hashes": {"receiver": "b9d8872cf3a99da99e1fc4b0f71232a451632dc49ca4c6771549b4efa5c1c763"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 4, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1760, "query_tokens": 442, "message_tokens": 325, "message_utf8_bytes": 923, "cache_tensor_bytes": null, "serialized_payload_bytes": 923, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 18975.824540830217, "hops": {"1": 6902.532499982044, "2": 3721.2153329746798, "3": 4287.0670419652015, "4": 4065.009665908292}, "receiver_prefill_ms": 184.81195799540728, "ttft_ms": 11.778416926972568, "decode_ms": 11.68158301152289}, "total_wall_ms": 19184.613832854666, "ttft_ms": 11.778416926972568, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 63, "facts_retained": 12, "retention_rate": 0.19047619047619047, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.25, "amount": 0.07692307692307693, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 923}, "ledger_by_hop": {"1": {"facts_total": 63, "facts_retained": 9, "retention_rate": 0.14285714285714285, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.25, "amount": 0.07692307692307693, "status": 0.25, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 787}, "2": {"facts_total": 63, "facts_retained": 12, "retention_rate": 0.19047619047619047, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.25, "amount": 0.07692307692307693, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 816}, "3": {"facts_total": 63, "facts_retained": 12, "retention_rate": 0.19047619047619047, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.25, "amount": 0.07692307692307693, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 816}, "4": {"facts_total": 63, "facts_retained": 12, "retention_rate": 0.19047619047619047, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.25, "amount": 0.07692307692307693, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 923}}}, "reference_key_id": "keys/synthetic/syn_03027_2000.json", "task_scores": {"correct": 1.0, "abstained": 1.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "3e25d9c439f55699f2305c8554c1cca0a9a3f39bbfef22b0c8d64ebbde2b63e5", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "986b57137b9ce8fe0a232066", "attempt_id": "70d8121f81eb", "run_group_id": "E4_core:A5|closed|qwen3b-llama3b:h8", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:34:27.075314Z", "task_family": "synthetic", "document_id": "syn_03027_2000", "document_hash": "10326080979f71b625e4110c7c19767e5a752e267de5034f7e37d98cb6e06249", "question_id": "q00_single_fact", "split": "test", "cluster_id": "syn_03027_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "tokenizer_hashes": {"receiver": "b9d8872cf3a99da99e1fc4b0f71232a451632dc49ca4c6771549b4efa5c1c763"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 8, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1760, "query_tokens": 427, "message_tokens": 325, "message_utf8_bytes": 923, "cache_tensor_bytes": null, "serialized_payload_bytes": 923, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 36320.24874887429, "hops": {"1": 6902.532499982044, "2": 3721.2153329746798, "3": 4287.0670419652015, "4": 4065.009665908292, "5": 4598.389083053917, "6": 4072.139750001952, "7": 4599.793374887668, "8": 4074.102000100538}, "receiver_prefill_ms": 182.95579100959003, "ttft_ms": 11.527625028975308, "decode_ms": 11.49016700219363}, "total_wall_ms": 36526.62808180321, "ttft_ms": 11.527625028975308, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 63, "facts_retained": 12, "retention_rate": 0.19047619047619047, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.25, "amount": 0.07692307692307693, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 923}, "ledger_by_hop": {"1": {"facts_total": 63, "facts_retained": 9, "retention_rate": 0.14285714285714285, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.25, "amount": 0.07692307692307693, "status": 0.25, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 787}, "2": {"facts_total": 63, "facts_retained": 12, "retention_rate": 0.19047619047619047, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.25, "amount": 0.07692307692307693, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 816}, "3": {"facts_total": 63, "facts_retained": 12, "retention_rate": 0.19047619047619047, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.25, "amount": 0.07692307692307693, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 816}, "4": {"facts_total": 63, "facts_retained": 12, "retention_rate": 0.19047619047619047, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.25, "amount": 0.07692307692307693, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 923}, "5": {"facts_total": 63, "facts_retained": 12, "retention_rate": 0.19047619047619047, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.25, "amount": 0.07692307692307693, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 923}, "6": {"facts_total": 63, "facts_retained": 12, "retention_rate": 0.19047619047619047, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.25, "amount": 0.07692307692307693, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 923}, "7": {"facts_total": 63, "facts_retained": 12, "retention_rate": 0.19047619047619047, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.25, "amount": 0.07692307692307693, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 923}, "8": {"facts_total": 63, "facts_retained": 12, "retention_rate": 0.19047619047619047, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.25, "amount": 0.07692307692307693, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 923}}}, "reference_key_id": "keys/synthetic/syn_03027_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "3e25d9c439f55699f2305c8554c1cca0a9a3f39bbfef22b0c8d64ebbde2b63e5", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "68fc80b4320f08c4739af3c0", "attempt_id": "a65a82a1bf8f", "run_group_id": "E4_core:A5|closed|qwen3b-llama3b:h8", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:34:27.297865Z", "task_family": "synthetic", "document_id": "syn_03027_2000", "document_hash": "10326080979f71b625e4110c7c19767e5a752e267de5034f7e37d98cb6e06249", "question_id": "q01_single_fact", "split": "test", "cluster_id": "syn_03027_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "tokenizer_hashes": {"receiver": "b9d8872cf3a99da99e1fc4b0f71232a451632dc49ca4c6771549b4efa5c1c763"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 8, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1760, "query_tokens": 434, "message_tokens": 325, "message_utf8_bytes": 923, "cache_tensor_bytes": null, "serialized_payload_bytes": 923, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 36320.24874887429, "hops": {"1": 6902.532499982044, "2": 3721.2153329746798, "3": 4287.0670419652015, "4": 4065.009665908292, "5": 4598.389083053917, "6": 4072.139750001952, "7": 4599.793374887668, "8": 4074.102000100538}, "receiver_prefill_ms": 182.6196250040084, "ttft_ms": 11.574708973057568, "decode_ms": 11.532500036992133}, "total_wall_ms": 36526.34958282579, "ttft_ms": 11.574708973057568, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 63, "facts_retained": 12, "retention_rate": 0.19047619047619047, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.25, "amount": 0.07692307692307693, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 923}, "ledger_by_hop": {"1": {"facts_total": 63, "facts_retained": 9, "retention_rate": 0.14285714285714285, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.25, "amount": 0.07692307692307693, "status": 0.25, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 787}, "2": {"facts_total": 63, "facts_retained": 12, "retention_rate": 0.19047619047619047, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.25, "amount": 0.07692307692307693, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 816}, "3": {"facts_total": 63, "facts_retained": 12, "retention_rate": 0.19047619047619047, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.25, "amount": 0.07692307692307693, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 816}, "4": {"facts_total": 63, "facts_retained": 12, "retention_rate": 0.19047619047619047, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.25, "amount": 0.07692307692307693, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 923}, "5": {"facts_total": 63, "facts_retained": 12, "retention_rate": 0.19047619047619047, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.25, "amount": 0.07692307692307693, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 923}, "6": {"facts_total": 63, "facts_retained": 12, "retention_rate": 0.19047619047619047, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.25, "amount": 0.07692307692307693, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 923}, "7": {"facts_total": 63, "facts_retained": 12, "retention_rate": 0.19047619047619047, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.25, "amount": 0.07692307692307693, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 923}, "8": {"facts_total": 63, "facts_retained": 12, "retention_rate": 0.19047619047619047, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.25, "amount": 0.07692307692307693, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 923}}}, "reference_key_id": "keys/synthetic/syn_03027_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "3e25d9c439f55699f2305c8554c1cca0a9a3f39bbfef22b0c8d64ebbde2b63e5", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "af761bf2968d73dc291319e3", "attempt_id": "d4da182e198a", "run_group_id": "E4_core:A5|closed|qwen3b-llama3b:h8", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:34:27.519735Z", "task_family": "synthetic", "document_id": "syn_03027_2000", "document_hash": "10326080979f71b625e4110c7c19767e5a752e267de5034f7e37d98cb6e06249", "question_id": "q02_exact_copy", "split": "test", "cluster_id": "syn_03027_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "tokenizer_hashes": {"receiver": "b9d8872cf3a99da99e1fc4b0f71232a451632dc49ca4c6771549b4efa5c1c763"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 8, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1760, "query_tokens": 433, "message_tokens": 325, "message_utf8_bytes": 923, "cache_tensor_bytes": null, "serialized_payload_bytes": 923, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 36320.24874887429, "hops": {"1": 6902.532499982044, "2": 3721.2153329746798, "3": 4287.0670419652015, "4": 4065.009665908292, "5": 4598.389083053917, "6": 4072.139750001952, "7": 4599.793374887668, "8": 4074.102000100538}, "receiver_prefill_ms": 183.0052910372615, "ttft_ms": 11.547958012670279, "decode_ms": 11.56937493942678}, "total_wall_ms": 36526.74520784058, "ttft_ms": 11.547958012670279, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 63, "facts_retained": 12, "retention_rate": 0.19047619047619047, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.25, "amount": 0.07692307692307693, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 923}, "ledger_by_hop": {"1": {"facts_total": 63, "facts_retained": 9, "retention_rate": 0.14285714285714285, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.25, "amount": 0.07692307692307693, "status": 0.25, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 787}, "2": {"facts_total": 63, "facts_retained": 12, "retention_rate": 0.19047619047619047, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.25, "amount": 0.07692307692307693, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 816}, "3": {"facts_total": 63, "facts_retained": 12, "retention_rate": 0.19047619047619047, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.25, "amount": 0.07692307692307693, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 816}, "4": {"facts_total": 63, "facts_retained": 12, "retention_rate": 0.19047619047619047, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.25, "amount": 0.07692307692307693, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 923}, "5": {"facts_total": 63, "facts_retained": 12, "retention_rate": 0.19047619047619047, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.25, "amount": 0.07692307692307693, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 923}, "6": {"facts_total": 63, "facts_retained": 12, "retention_rate": 0.19047619047619047, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.25, "amount": 0.07692307692307693, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 923}, "7": {"facts_total": 63, "facts_retained": 12, "retention_rate": 0.19047619047619047, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.25, "amount": 0.07692307692307693, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 923}, "8": {"facts_total": 63, "facts_retained": 12, "retention_rate": 0.19047619047619047, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.25, "amount": 0.07692307692307693, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 923}}}, "reference_key_id": "keys/synthetic/syn_03027_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "3e25d9c439f55699f2305c8554c1cca0a9a3f39bbfef22b0c8d64ebbde2b63e5", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "31f6324d4005a81bb6e87b23", "attempt_id": "f5958669933b", "run_group_id": "E4_core:A5|closed|qwen3b-llama3b:h8", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:34:27.742047Z", "task_family": "synthetic", "document_id": "syn_03027_2000", "document_hash": "10326080979f71b625e4110c7c19767e5a752e267de5034f7e37d98cb6e06249", "question_id": "q03_exact_copy", "split": "test", "cluster_id": "syn_03027_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "tokenizer_hashes": {"receiver": "b9d8872cf3a99da99e1fc4b0f71232a451632dc49ca4c6771549b4efa5c1c763"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 8, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1760, "query_tokens": 429, "message_tokens": 325, "message_utf8_bytes": 923, "cache_tensor_bytes": null, "serialized_payload_bytes": 923, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 36320.24874887429, "hops": {"1": 6902.532499982044, "2": 3721.2153329746798, "3": 4287.0670419652015, "4": 4065.009665908292, "5": 4598.389083053917, "6": 4072.139750001952, "7": 4599.793374887668, "8": 4074.102000100538}, "receiver_prefill_ms": 184.25354093778878, "ttft_ms": 11.528333998285234, "decode_ms": 103.85308298282325}, "total_wall_ms": 36620.2802908374, "ttft_ms": 11.528333998285234, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 9}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "Priyamvada Sørlie", "evidence_artifact": {"ledger": {"facts_total": 63, "facts_retained": 12, "retention_rate": 0.19047619047619047, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.25, "amount": 0.07692307692307693, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 923}, "ledger_by_hop": {"1": {"facts_total": 63, "facts_retained": 9, "retention_rate": 0.14285714285714285, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.25, "amount": 0.07692307692307693, "status": 0.25, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 787}, "2": {"facts_total": 63, "facts_retained": 12, "retention_rate": 0.19047619047619047, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.25, "amount": 0.07692307692307693, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 816}, "3": {"facts_total": 63, "facts_retained": 12, "retention_rate": 0.19047619047619047, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.25, "amount": 0.07692307692307693, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 816}, "4": {"facts_total": 63, "facts_retained": 12, "retention_rate": 0.19047619047619047, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.25, "amount": 0.07692307692307693, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 923}, "5": {"facts_total": 63, "facts_retained": 12, "retention_rate": 0.19047619047619047, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.25, "amount": 0.07692307692307693, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 923}, "6": {"facts_total": 63, "facts_retained": 12, "retention_rate": 0.19047619047619047, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.25, "amount": 0.07692307692307693, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 923}, "7": {"facts_total": 63, "facts_retained": 12, "retention_rate": 0.19047619047619047, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.25, "amount": 0.07692307692307693, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 923}, "8": {"facts_total": 63, "facts_retained": 12, "retention_rate": 0.19047619047619047, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.25, "amount": 0.07692307692307693, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 923}}}, "reference_key_id": "keys/synthetic/syn_03027_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "3e25d9c439f55699f2305c8554c1cca0a9a3f39bbfef22b0c8d64ebbde2b63e5", "output_hash": "c5f554196ed23dd71ad995ec89f7142099d524dbb6534cc10d94a0c89d13e792"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "a23ad7ecff07967e17a704de", "attempt_id": "20564f11cdaa", "run_group_id": "E4_core:A5|closed|qwen3b-llama3b:h8", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:34:28.057282Z", "task_family": "synthetic", "document_id": "syn_03027_2000", "document_hash": "10326080979f71b625e4110c7c19767e5a752e267de5034f7e37d98cb6e06249", "question_id": "q04_multi_hop", "split": "test", "cluster_id": "syn_03027_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "tokenizer_hashes": {"receiver": "b9d8872cf3a99da99e1fc4b0f71232a451632dc49ca4c6771549b4efa5c1c763"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 8, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1760, "query_tokens": 440, "message_tokens": 325, "message_utf8_bytes": 923, "cache_tensor_bytes": null, "serialized_payload_bytes": 923, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 36320.24874887429, "hops": {"1": 6902.532499982044, "2": 3721.2153329746798, "3": 4287.0670419652015, "4": 4065.009665908292, "5": 4598.389083053917, "6": 4072.139750001952, "7": 4599.793374887668, "8": 4074.102000100538}, "receiver_prefill_ms": 184.3003750545904, "ttft_ms": 11.527125025168061, "decode_ms": 11.529542040079832}, "total_wall_ms": 36527.97053998802, "ttft_ms": 11.527125025168061, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 63, "facts_retained": 12, "retention_rate": 0.19047619047619047, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.25, "amount": 0.07692307692307693, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 923}, "ledger_by_hop": {"1": {"facts_total": 63, "facts_retained": 9, "retention_rate": 0.14285714285714285, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.25, "amount": 0.07692307692307693, "status": 0.25, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 787}, "2": {"facts_total": 63, "facts_retained": 12, "retention_rate": 0.19047619047619047, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.25, "amount": 0.07692307692307693, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 816}, "3": {"facts_total": 63, "facts_retained": 12, "retention_rate": 0.19047619047619047, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.25, "amount": 0.07692307692307693, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 816}, "4": {"facts_total": 63, "facts_retained": 12, "retention_rate": 0.19047619047619047, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.25, "amount": 0.07692307692307693, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 923}, "5": {"facts_total": 63, "facts_retained": 12, "retention_rate": 0.19047619047619047, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.25, "amount": 0.07692307692307693, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 923}, "6": {"facts_total": 63, "facts_retained": 12, "retention_rate": 0.19047619047619047, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.25, "amount": 0.07692307692307693, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 923}, "7": {"facts_total": 63, "facts_retained": 12, "retention_rate": 0.19047619047619047, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.25, "amount": 0.07692307692307693, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 923}, "8": {"facts_total": 63, "facts_retained": 12, "retention_rate": 0.19047619047619047, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.25, "amount": 0.07692307692307693, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 923}}}, "reference_key_id": "keys/synthetic/syn_03027_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "3e25d9c439f55699f2305c8554c1cca0a9a3f39bbfef22b0c8d64ebbde2b63e5", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "0030acdb6dbf81d0d87ea3c8", "attempt_id": "aff3b9c61732", "run_group_id": "E4_core:A5|closed|qwen3b-llama3b:h8", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:34:28.280243Z", "task_family": "synthetic", "document_id": "syn_03027_2000", "document_hash": "10326080979f71b625e4110c7c19767e5a752e267de5034f7e37d98cb6e06249", "question_id": "q05_multi_hop", "split": "test", "cluster_id": "syn_03027_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "tokenizer_hashes": {"receiver": "b9d8872cf3a99da99e1fc4b0f71232a451632dc49ca4c6771549b4efa5c1c763"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 8, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1760, "query_tokens": 435, "message_tokens": 325, "message_utf8_bytes": 923, "cache_tensor_bytes": null, "serialized_payload_bytes": 923, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 36320.24874887429, "hops": {"1": 6902.532499982044, "2": 3721.2153329746798, "3": 4287.0670419652015, "4": 4065.009665908292, "5": 4598.389083053917, "6": 4072.139750001952, "7": 4599.793374887668, "8": 4074.102000100538}, "receiver_prefill_ms": 186.0141250072047, "ttft_ms": 11.580625083297491, "decode_ms": 11.531542055308819}, "total_wall_ms": 36529.73383292556, "ttft_ms": 11.580625083297491, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 63, "facts_retained": 12, "retention_rate": 0.19047619047619047, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.25, "amount": 0.07692307692307693, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 923}, "ledger_by_hop": {"1": {"facts_total": 63, "facts_retained": 9, "retention_rate": 0.14285714285714285, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.25, "amount": 0.07692307692307693, "status": 0.25, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 787}, "2": {"facts_total": 63, "facts_retained": 12, "retention_rate": 0.19047619047619047, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.25, "amount": 0.07692307692307693, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 816}, "3": {"facts_total": 63, "facts_retained": 12, "retention_rate": 0.19047619047619047, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.25, "amount": 0.07692307692307693, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 816}, "4": {"facts_total": 63, "facts_retained": 12, "retention_rate": 0.19047619047619047, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.25, "amount": 0.07692307692307693, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 923}, "5": {"facts_total": 63, "facts_retained": 12, "retention_rate": 0.19047619047619047, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.25, "amount": 0.07692307692307693, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 923}, "6": {"facts_total": 63, "facts_retained": 12, "retention_rate": 0.19047619047619047, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.25, "amount": 0.07692307692307693, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 923}, "7": {"facts_total": 63, "facts_retained": 12, "retention_rate": 0.19047619047619047, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.25, "amount": 0.07692307692307693, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 923}, "8": {"facts_total": 63, "facts_retained": 12, "retention_rate": 0.19047619047619047, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.25, "amount": 0.07692307692307693, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 923}}}, "reference_key_id": "keys/synthetic/syn_03027_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "3e25d9c439f55699f2305c8554c1cca0a9a3f39bbfef22b0c8d64ebbde2b63e5", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "3c1b90a9dc1fe77b57325b3f", "attempt_id": "de5809ce296a", "run_group_id": "E4_core:A5|closed|qwen3b-llama3b:h8", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:34:28.505046Z", "task_family": "synthetic", "document_id": "syn_03027_2000", "document_hash": "10326080979f71b625e4110c7c19767e5a752e267de5034f7e37d98cb6e06249", "question_id": "q06_negation", "split": "test", "cluster_id": "syn_03027_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "tokenizer_hashes": {"receiver": "b9d8872cf3a99da99e1fc4b0f71232a451632dc49ca4c6771549b4efa5c1c763"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 8, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1760, "query_tokens": 429, "message_tokens": 325, "message_utf8_bytes": 923, "cache_tensor_bytes": null, "serialized_payload_bytes": 923, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 36320.24874887429, "hops": {"1": 6902.532499982044, "2": 3721.2153329746798, "3": 4287.0670419652015, "4": 4065.009665908292, "5": 4598.389083053917, "6": 4072.139750001952, "7": 4599.793374887668, "8": 4074.102000100538}, "receiver_prefill_ms": 183.01004194654524, "ttft_ms": 11.83879200834781, "decode_ms": 11.49150007404387}, "total_wall_ms": 36526.95266494993, "ttft_ms": 11.83879200834781, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 63, "facts_retained": 12, "retention_rate": 0.19047619047619047, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.25, "amount": 0.07692307692307693, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 923}, "ledger_by_hop": {"1": {"facts_total": 63, "facts_retained": 9, "retention_rate": 0.14285714285714285, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.25, "amount": 0.07692307692307693, "status": 0.25, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 787}, "2": {"facts_total": 63, "facts_retained": 12, "retention_rate": 0.19047619047619047, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.25, "amount": 0.07692307692307693, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 816}, "3": {"facts_total": 63, "facts_retained": 12, "retention_rate": 0.19047619047619047, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.25, "amount": 0.07692307692307693, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 816}, "4": {"facts_total": 63, "facts_retained": 12, "retention_rate": 0.19047619047619047, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.25, "amount": 0.07692307692307693, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 923}, "5": {"facts_total": 63, "facts_retained": 12, "retention_rate": 0.19047619047619047, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.25, "amount": 0.07692307692307693, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 923}, "6": {"facts_total": 63, "facts_retained": 12, "retention_rate": 0.19047619047619047, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.25, "amount": 0.07692307692307693, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 923}, "7": {"facts_total": 63, "facts_retained": 12, "retention_rate": 0.19047619047619047, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.25, "amount": 0.07692307692307693, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 923}, "8": {"facts_total": 63, "facts_retained": 12, "retention_rate": 0.19047619047619047, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.25, "amount": 0.07692307692307693, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 923}}}, "reference_key_id": "keys/synthetic/syn_03027_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "3e25d9c439f55699f2305c8554c1cca0a9a3f39bbfef22b0c8d64ebbde2b63e5", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "702bd4b1cb5dcffaa41e0b10", "attempt_id": "e37327299fcf", "run_group_id": "E4_core:A5|closed|qwen3b-llama3b:h8", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:34:28.726653Z", "task_family": "synthetic", "document_id": "syn_03027_2000", "document_hash": "10326080979f71b625e4110c7c19767e5a752e267de5034f7e37d98cb6e06249", "question_id": "q07_negation", "split": "test", "cluster_id": "syn_03027_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "tokenizer_hashes": {"receiver": "b9d8872cf3a99da99e1fc4b0f71232a451632dc49ca4c6771549b4efa5c1c763"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 8, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1760, "query_tokens": 429, "message_tokens": 325, "message_utf8_bytes": 923, "cache_tensor_bytes": null, "serialized_payload_bytes": 923, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 36320.24874887429, "hops": {"1": 6902.532499982044, "2": 3721.2153329746798, "3": 4287.0670419652015, "4": 4065.009665908292, "5": 4598.389083053917, "6": 4072.139750001952, "7": 4599.793374887668, "8": 4074.102000100538}, "receiver_prefill_ms": 183.0625410657376, "ttft_ms": 11.526082991622388, "decode_ms": 11.536041973158717}, "total_wall_ms": 36526.72833192628, "ttft_ms": 11.526082991622388, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "No", "evidence_artifact": {"ledger": {"facts_total": 63, "facts_retained": 12, "retention_rate": 0.19047619047619047, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.25, "amount": 0.07692307692307693, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 923}, "ledger_by_hop": {"1": {"facts_total": 63, "facts_retained": 9, "retention_rate": 0.14285714285714285, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.25, "amount": 0.07692307692307693, "status": 0.25, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 787}, "2": {"facts_total": 63, "facts_retained": 12, "retention_rate": 0.19047619047619047, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.25, "amount": 0.07692307692307693, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 816}, "3": {"facts_total": 63, "facts_retained": 12, "retention_rate": 0.19047619047619047, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.25, "amount": 0.07692307692307693, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 816}, "4": {"facts_total": 63, "facts_retained": 12, "retention_rate": 0.19047619047619047, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.25, "amount": 0.07692307692307693, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 923}, "5": {"facts_total": 63, "facts_retained": 12, "retention_rate": 0.19047619047619047, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.25, "amount": 0.07692307692307693, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 923}, "6": {"facts_total": 63, "facts_retained": 12, "retention_rate": 0.19047619047619047, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.25, "amount": 0.07692307692307693, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 923}, "7": {"facts_total": 63, "facts_retained": 12, "retention_rate": 0.19047619047619047, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.25, "amount": 0.07692307692307693, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 923}, "8": {"facts_total": 63, "facts_retained": 12, "retention_rate": 0.19047619047619047, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.25, "amount": 0.07692307692307693, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 923}}}, "reference_key_id": "keys/synthetic/syn_03027_2000.json", "task_scores": {"correct": 1.0, "abstained": 0.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "3e25d9c439f55699f2305c8554c1cca0a9a3f39bbfef22b0c8d64ebbde2b63e5", "output_hash": "1ea442a134b2a184bd5d40104401f2a37fbc09ccf3f4bc9da161c6099be3691d"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "f42418df841e1cee7a2ba592", "attempt_id": "41382625009e", "run_group_id": "E4_core:A5|closed|qwen3b-llama3b:h8", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:34:28.947685Z", "task_family": "synthetic", "document_id": "syn_03027_2000", "document_hash": "10326080979f71b625e4110c7c19767e5a752e267de5034f7e37d98cb6e06249", "question_id": "q08_revision", "split": "test", "cluster_id": "syn_03027_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "tokenizer_hashes": {"receiver": "b9d8872cf3a99da99e1fc4b0f71232a451632dc49ca4c6771549b4efa5c1c763"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 8, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1760, "query_tokens": 444, "message_tokens": 325, "message_utf8_bytes": 923, "cache_tensor_bytes": null, "serialized_payload_bytes": 923, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 36320.24874887429, "hops": {"1": 6902.532499982044, "2": 3721.2153329746798, "3": 4287.0670419652015, "4": 4065.009665908292, "5": 4598.389083053917, "6": 4072.139750001952, "7": 4599.793374887668, "8": 4074.102000100538}, "receiver_prefill_ms": 184.81733393855393, "ttft_ms": 11.634916998445988, "decode_ms": 57.519542053341866}, "total_wall_ms": 36574.61024890654, "ttft_ms": 11.634916998445988, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 5}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "6,914,937", "evidence_artifact": {"ledger": {"facts_total": 63, "facts_retained": 12, "retention_rate": 0.19047619047619047, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.25, "amount": 0.07692307692307693, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 923}, "ledger_by_hop": {"1": {"facts_total": 63, "facts_retained": 9, "retention_rate": 0.14285714285714285, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.25, "amount": 0.07692307692307693, "status": 0.25, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 787}, "2": {"facts_total": 63, "facts_retained": 12, "retention_rate": 0.19047619047619047, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.25, "amount": 0.07692307692307693, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 816}, "3": {"facts_total": 63, "facts_retained": 12, "retention_rate": 0.19047619047619047, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.25, "amount": 0.07692307692307693, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 816}, "4": {"facts_total": 63, "facts_retained": 12, "retention_rate": 0.19047619047619047, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.25, "amount": 0.07692307692307693, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 923}, "5": {"facts_total": 63, "facts_retained": 12, "retention_rate": 0.19047619047619047, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.25, "amount": 0.07692307692307693, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 923}, "6": {"facts_total": 63, "facts_retained": 12, "retention_rate": 0.19047619047619047, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.25, "amount": 0.07692307692307693, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 923}, "7": {"facts_total": 63, "facts_retained": 12, "retention_rate": 0.19047619047619047, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.25, "amount": 0.07692307692307693, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 923}, "8": {"facts_total": 63, "facts_retained": 12, "retention_rate": 0.19047619047619047, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.25, "amount": 0.07692307692307693, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 923}}}, "reference_key_id": "keys/synthetic/syn_03027_2000.json", "task_scores": {"correct": 0.0, "abstained": 0.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "3e25d9c439f55699f2305c8554c1cca0a9a3f39bbfef22b0c8d64ebbde2b63e5", "output_hash": "c56851abfdbf8851226a1e71c4911c616a691c490a883eb754e61bac551d004c"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "9ff142be453fac0f188c51ee", "attempt_id": "6a004181a6b0", "run_group_id": "E4_core:A5|closed|qwen3b-llama3b:h8", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:34:29.216944Z", "task_family": "synthetic", "document_id": "syn_03027_2000", "document_hash": "10326080979f71b625e4110c7c19767e5a752e267de5034f7e37d98cb6e06249", "question_id": "q09_revision", "split": "test", "cluster_id": "syn_03027_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "tokenizer_hashes": {"receiver": "b9d8872cf3a99da99e1fc4b0f71232a451632dc49ca4c6771549b4efa5c1c763"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 8, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1760, "query_tokens": 437, "message_tokens": 325, "message_utf8_bytes": 923, "cache_tensor_bytes": null, "serialized_payload_bytes": 923, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 36320.24874887429, "hops": {"1": 6902.532499982044, "2": 3721.2153329746798, "3": 4287.0670419652015, "4": 4065.009665908292, "5": 4598.389083053917, "6": 4072.139750001952, "7": 4599.793374887668, "8": 4074.102000100538}, "receiver_prefill_ms": 184.62754203937948, "ttft_ms": 11.469374992884696, "decode_ms": 57.3348329635337}, "total_wall_ms": 36574.054081807844, "ttft_ms": 11.469374992884696, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 5}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "7,000,000", "evidence_artifact": {"ledger": {"facts_total": 63, "facts_retained": 12, "retention_rate": 0.19047619047619047, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.25, "amount": 0.07692307692307693, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 923}, "ledger_by_hop": {"1": {"facts_total": 63, "facts_retained": 9, "retention_rate": 0.14285714285714285, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.25, "amount": 0.07692307692307693, "status": 0.25, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 787}, "2": {"facts_total": 63, "facts_retained": 12, "retention_rate": 0.19047619047619047, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.25, "amount": 0.07692307692307693, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 816}, "3": {"facts_total": 63, "facts_retained": 12, "retention_rate": 0.19047619047619047, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.25, "amount": 0.07692307692307693, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 816}, "4": {"facts_total": 63, "facts_retained": 12, "retention_rate": 0.19047619047619047, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.25, "amount": 0.07692307692307693, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 923}, "5": {"facts_total": 63, "facts_retained": 12, "retention_rate": 0.19047619047619047, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.25, "amount": 0.07692307692307693, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 923}, "6": {"facts_total": 63, "facts_retained": 12, "retention_rate": 0.19047619047619047, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.25, "amount": 0.07692307692307693, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 923}, "7": {"facts_total": 63, "facts_retained": 12, "retention_rate": 0.19047619047619047, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.25, "amount": 0.07692307692307693, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 923}, "8": {"facts_total": 63, "facts_retained": 12, "retention_rate": 0.19047619047619047, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.25, "amount": 0.07692307692307693, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 923}}}, "reference_key_id": "keys/synthetic/syn_03027_2000.json", "task_scores": {"correct": 0.0, "abstained": 0.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "3e25d9c439f55699f2305c8554c1cca0a9a3f39bbfef22b0c8d64ebbde2b63e5", "output_hash": "a8f19157bfa97ae9789ef71ec65944fe20e98491b7b0c2aa047d68bf92a3e7fd"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "da71969e078df5a4d0ace118", "attempt_id": "95f02332134a", "run_group_id": "E4_core:A5|closed|qwen3b-llama3b:h8", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:34:29.485706Z", "task_family": "synthetic", "document_id": "syn_03027_2000", "document_hash": "10326080979f71b625e4110c7c19767e5a752e267de5034f7e37d98cb6e06249", "question_id": "q10_temporal", "split": "test", "cluster_id": "syn_03027_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "tokenizer_hashes": {"receiver": "b9d8872cf3a99da99e1fc4b0f71232a451632dc49ca4c6771549b4efa5c1c763"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 8, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1760, "query_tokens": 445, "message_tokens": 325, "message_utf8_bytes": 923, "cache_tensor_bytes": null, "serialized_payload_bytes": 923, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 36320.24874887429, "hops": {"1": 6902.532499982044, "2": 3721.2153329746798, "3": 4287.0670419652015, "4": 4065.009665908292, "5": 4598.389083053917, "6": 4072.139750001952, "7": 4599.793374887668, "8": 4074.102000100538}, "receiver_prefill_ms": 183.98929201066494, "ttft_ms": 11.605625040829182, "decode_ms": 11.454334016889334}, "total_wall_ms": 36527.655539801344, "ttft_ms": 11.605625040829182, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 63, "facts_retained": 12, "retention_rate": 0.19047619047619047, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.25, "amount": 0.07692307692307693, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 923}, "ledger_by_hop": {"1": {"facts_total": 63, "facts_retained": 9, "retention_rate": 0.14285714285714285, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.25, "amount": 0.07692307692307693, "status": 0.25, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 787}, "2": {"facts_total": 63, "facts_retained": 12, "retention_rate": 0.19047619047619047, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.25, "amount": 0.07692307692307693, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 816}, "3": {"facts_total": 63, "facts_retained": 12, "retention_rate": 0.19047619047619047, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.25, "amount": 0.07692307692307693, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 816}, "4": {"facts_total": 63, "facts_retained": 12, "retention_rate": 0.19047619047619047, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.25, "amount": 0.07692307692307693, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 923}, "5": {"facts_total": 63, "facts_retained": 12, "retention_rate": 0.19047619047619047, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.25, "amount": 0.07692307692307693, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 923}, "6": {"facts_total": 63, "facts_retained": 12, "retention_rate": 0.19047619047619047, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.25, "amount": 0.07692307692307693, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 923}, "7": {"facts_total": 63, "facts_retained": 12, "retention_rate": 0.19047619047619047, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.25, "amount": 0.07692307692307693, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 923}, "8": {"facts_total": 63, "facts_retained": 12, "retention_rate": 0.19047619047619047, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.25, "amount": 0.07692307692307693, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 923}}}, "reference_key_id": "keys/synthetic/syn_03027_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "3e25d9c439f55699f2305c8554c1cca0a9a3f39bbfef22b0c8d64ebbde2b63e5", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "523abb46c407a1c204e938ad", "attempt_id": "552d6ea44f3e", "run_group_id": "E4_core:A5|closed|qwen3b-llama3b:h8", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:34:29.709208Z", "task_family": "synthetic", "document_id": "syn_03027_2000", "document_hash": "10326080979f71b625e4110c7c19767e5a752e267de5034f7e37d98cb6e06249", "question_id": "q11_aggregation", "split": "test", "cluster_id": "syn_03027_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "tokenizer_hashes": {"receiver": "b9d8872cf3a99da99e1fc4b0f71232a451632dc49ca4c6771549b4efa5c1c763"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 8, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1760, "query_tokens": 440, "message_tokens": 325, "message_utf8_bytes": 923, "cache_tensor_bytes": null, "serialized_payload_bytes": 923, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 36320.24874887429, "hops": {"1": 6902.532499982044, "2": 3721.2153329746798, "3": 4287.0670419652015, "4": 4065.009665908292, "5": 4598.389083053917, "6": 4072.139750001952, "7": 4599.793374887668, "8": 4074.102000100538}, "receiver_prefill_ms": 185.88920799084008, "ttft_ms": 11.528583010658622, "decode_ms": 57.80966591555625}, "total_wall_ms": 36575.850290828384, "ttft_ms": 11.528583010658622, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 5}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "6,914,937", "evidence_artifact": {"ledger": {"facts_total": 63, "facts_retained": 12, "retention_rate": 0.19047619047619047, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.25, "amount": 0.07692307692307693, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 923}, "ledger_by_hop": {"1": {"facts_total": 63, "facts_retained": 9, "retention_rate": 0.14285714285714285, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.25, "amount": 0.07692307692307693, "status": 0.25, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 787}, "2": {"facts_total": 63, "facts_retained": 12, "retention_rate": 0.19047619047619047, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.25, "amount": 0.07692307692307693, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 816}, "3": {"facts_total": 63, "facts_retained": 12, "retention_rate": 0.19047619047619047, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.25, "amount": 0.07692307692307693, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 816}, "4": {"facts_total": 63, "facts_retained": 12, "retention_rate": 0.19047619047619047, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.25, "amount": 0.07692307692307693, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 923}, "5": {"facts_total": 63, "facts_retained": 12, "retention_rate": 0.19047619047619047, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.25, "amount": 0.07692307692307693, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 923}, "6": {"facts_total": 63, "facts_retained": 12, "retention_rate": 0.19047619047619047, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.25, "amount": 0.07692307692307693, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 923}, "7": {"facts_total": 63, "facts_retained": 12, "retention_rate": 0.19047619047619047, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.25, "amount": 0.07692307692307693, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 923}, "8": {"facts_total": 63, "facts_retained": 12, "retention_rate": 0.19047619047619047, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.25, "amount": 0.07692307692307693, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 923}}}, "reference_key_id": "keys/synthetic/syn_03027_2000.json", "task_scores": {"correct": 0.0, "abstained": 0.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "3e25d9c439f55699f2305c8554c1cca0a9a3f39bbfef22b0c8d64ebbde2b63e5", "output_hash": "c56851abfdbf8851226a1e71c4911c616a691c490a883eb754e61bac551d004c"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "42189f91a606656db2179f32", "attempt_id": "b61e496c6038", "run_group_id": "E4_core:A5|closed|qwen3b-llama3b:h8", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:34:29.980312Z", "task_family": "synthetic", "document_id": "syn_03027_2000", "document_hash": "10326080979f71b625e4110c7c19767e5a752e267de5034f7e37d98cb6e06249", "question_id": "q12_unit_conversion", "split": "test", "cluster_id": "syn_03027_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "tokenizer_hashes": {"receiver": "b9d8872cf3a99da99e1fc4b0f71232a451632dc49ca4c6771549b4efa5c1c763"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 8, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1760, "query_tokens": 436, "message_tokens": 325, "message_utf8_bytes": 923, "cache_tensor_bytes": null, "serialized_payload_bytes": 923, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 36320.24874887429, "hops": {"1": 6902.532499982044, "2": 3721.2153329746798, "3": 4287.0670419652015, "4": 4065.009665908292, "5": 4598.389083053917, "6": 4072.139750001952, "7": 4599.793374887668, "8": 4074.102000100538}, "receiver_prefill_ms": 183.3075419999659, "ttft_ms": 11.470124940387905, "decode_ms": 11.47279201541096}, "total_wall_ms": 36526.87003987376, "ttft_ms": 11.470124940387905, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 63, "facts_retained": 12, "retention_rate": 0.19047619047619047, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.25, "amount": 0.07692307692307693, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 923}, "ledger_by_hop": {"1": {"facts_total": 63, "facts_retained": 9, "retention_rate": 0.14285714285714285, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.25, "amount": 0.07692307692307693, "status": 0.25, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 787}, "2": {"facts_total": 63, "facts_retained": 12, "retention_rate": 0.19047619047619047, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.25, "amount": 0.07692307692307693, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 816}, "3": {"facts_total": 63, "facts_retained": 12, "retention_rate": 0.19047619047619047, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.25, "amount": 0.07692307692307693, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 816}, "4": {"facts_total": 63, "facts_retained": 12, "retention_rate": 0.19047619047619047, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.25, "amount": 0.07692307692307693, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 923}, "5": {"facts_total": 63, "facts_retained": 12, "retention_rate": 0.19047619047619047, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.25, "amount": 0.07692307692307693, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 923}, "6": {"facts_total": 63, "facts_retained": 12, "retention_rate": 0.19047619047619047, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.25, "amount": 0.07692307692307693, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 923}, "7": {"facts_total": 63, "facts_retained": 12, "retention_rate": 0.19047619047619047, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.25, "amount": 0.07692307692307693, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 923}, "8": {"facts_total": 63, "facts_retained": 12, "retention_rate": 0.19047619047619047, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.25, "amount": 0.07692307692307693, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 923}}}, "reference_key_id": "keys/synthetic/syn_03027_2000.json", "task_scores": {"correct": 0.0, "abstained": 1.0, "f1": 0.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "3e25d9c439f55699f2305c8554c1cca0a9a3f39bbfef22b0c8d64ebbde2b63e5", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "c33c52f5ed4bda6114c830fb", "attempt_id": "d864585fea94", "run_group_id": "E4_core:A5|closed|qwen3b-llama3b:h8", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:34:30.202414Z", "task_family": "synthetic", "document_id": "syn_03027_2000", "document_hash": "10326080979f71b625e4110c7c19767e5a752e267de5034f7e37d98cb6e06249", "question_id": "q13_unanswerable", "split": "test", "cluster_id": "syn_03027_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "tokenizer_hashes": {"receiver": "b9d8872cf3a99da99e1fc4b0f71232a451632dc49ca4c6771549b4efa5c1c763"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 8, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1760, "query_tokens": 440, "message_tokens": 325, "message_utf8_bytes": 923, "cache_tensor_bytes": null, "serialized_payload_bytes": 923, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 36320.24874887429, "hops": {"1": 6902.532499982044, "2": 3721.2153329746798, "3": 4287.0670419652015, "4": 4065.009665908292, "5": 4598.389083053917, "6": 4072.139750001952, "7": 4599.793374887668, "8": 4074.102000100538}, "receiver_prefill_ms": 184.77762502152473, "ttft_ms": 11.464999988675117, "decode_ms": 11.751624988391995}, "total_wall_ms": 36528.57462386601, "ttft_ms": 11.464999988675117, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 63, "facts_retained": 12, "retention_rate": 0.19047619047619047, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.25, "amount": 0.07692307692307693, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 923}, "ledger_by_hop": {"1": {"facts_total": 63, "facts_retained": 9, "retention_rate": 0.14285714285714285, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.25, "amount": 0.07692307692307693, "status": 0.25, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 787}, "2": {"facts_total": 63, "facts_retained": 12, "retention_rate": 0.19047619047619047, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.25, "amount": 0.07692307692307693, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 816}, "3": {"facts_total": 63, "facts_retained": 12, "retention_rate": 0.19047619047619047, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.25, "amount": 0.07692307692307693, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 816}, "4": {"facts_total": 63, "facts_retained": 12, "retention_rate": 0.19047619047619047, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.25, "amount": 0.07692307692307693, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 923}, "5": {"facts_total": 63, "facts_retained": 12, "retention_rate": 0.19047619047619047, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.25, "amount": 0.07692307692307693, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 923}, "6": {"facts_total": 63, "facts_retained": 12, "retention_rate": 0.19047619047619047, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.25, "amount": 0.07692307692307693, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 923}, "7": {"facts_total": 63, "facts_retained": 12, "retention_rate": 0.19047619047619047, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.25, "amount": 0.07692307692307693, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 923}, "8": {"facts_total": 63, "facts_retained": 12, "retention_rate": 0.19047619047619047, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.25, "amount": 0.07692307692307693, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 923}}}, "reference_key_id": "keys/synthetic/syn_03027_2000.json", "task_scores": {"correct": 1.0, "abstained": 1.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "3e25d9c439f55699f2305c8554c1cca0a9a3f39bbfef22b0c8d64ebbde2b63e5", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}
{"schema_version": "wp30-attempt-1", "study_id": "WP30", "experiment_id": "E4_core", "trial_id": "f0e6ff093389e04dd8275310", "attempt_id": "2c20470b933f", "run_group_id": "E4_core:A5|closed|qwen3b-llama3b:h8", "protocol_hash": "404c52b34d964693da72ef459a036ffc42fbf5f6ba8748cc9de0c554ebebddd6", "config_hash": "5b97ab3cef5587ee7db07b92390e2e3a28a0afee1c30cea111c3a46c9554433b", "git_commit": null, "timestamp_utc": "2026-10-30T08:34:30.425778Z", "task_family": "synthetic", "document_id": "syn_03027_2000", "document_hash": "10326080979f71b625e4110c7c19767e5a752e267de5034f7e37d98cb6e06249", "question_id": "q14_unanswerable", "split": "test", "cluster_id": "syn_03027_2000", "sender_model_revision": "mlx-community/Qwen2.5-3B-Instruct-bf16@bec1ad658e380044bbfe06d4a2268801d3379a11|mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "receiver_model_revision": "mlx-community/Llama-3.2-3B-Instruct-bf16@6d88ba43024fef71b10e52e101c7cd4598322601", "tokenizer_hashes": {"receiver": "b9d8872cf3a99da99e1fc4b0f71232a451632dc49ca4c6771549b4efa5c1c763"}, "backend_versions": {"mlx": "0.32.4", "mlx_lm": "0.32.2"}, "hardware_id": "m3u96b-m3ultra-96g", "device_state": {}, "weight_precision": "bf16", "method_id": "A5", "implementation_fidelity": "custom", "compression_config": {"mode": "closed", "agents": ["qwen3b", "llama3b"], "budget_tokens": 512}, "question_access": "blind", "topology": "chain_closed", "handoff_index": 8, "reuse_count": 15, "cache_hit": false, "cache_compatibility_hash": null, "random_seed": 0, "source_tokens_sender": null, "source_tokens_receiver": 1760, "query_tokens": 442, "message_tokens": 325, "message_utf8_bytes": 923, "cache_tensor_bytes": null, "serialized_payload_bytes": 923, "side_information_bytes": 0, "fetched_bytes": 0, "transport_bytes_if_measured": null, "timing_spans": {"chain_ms": 36320.24874887429, "hops": {"1": 6902.532499982044, "2": 3721.2153329746798, "3": 4287.0670419652015, "4": 4065.009665908292, "5": 4598.389083053917, "6": 4072.139750001952, "7": 4599.793374887668, "8": 4074.102000100538}, "receiver_prefill_ms": 183.4474999923259, "ttft_ms": 11.52937498409301, "decode_ms": 11.499374988488853}, "total_wall_ms": 36527.099415892735, "ttft_ms": 11.52937498409301, "input_usage": {"lookups": 0}, "output_usage": {"output_tokens": 1}, "provider_cached_usage": null, "process_memory": null, "backend_memory": null, "peak_prefill_memory": null, "peak_decode_memory": null, "mapper_memory": null, "cost_with_source": null, "prediction_artifact": "unknown", "evidence_artifact": {"ledger": {"facts_total": 63, "facts_retained": 12, "retention_rate": 0.19047619047619047, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.25, "amount": 0.07692307692307693, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 923}, "ledger_by_hop": {"1": {"facts_total": 63, "facts_retained": 9, "retention_rate": 0.14285714285714285, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.25, "amount": 0.07692307692307693, "status": 0.25, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 787}, "2": {"facts_total": 63, "facts_retained": 12, "retention_rate": 0.19047619047619047, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.25, "amount": 0.07692307692307693, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 816}, "3": {"facts_total": 63, "facts_retained": 12, "retention_rate": 0.19047619047619047, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.25, "amount": 0.07692307692307693, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 816}, "4": {"facts_total": 63, "facts_retained": 12, "retention_rate": 0.19047619047619047, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.25, "amount": 0.07692307692307693, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 923}, "5": {"facts_total": 63, "facts_retained": 12, "retention_rate": 0.19047619047619047, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.25, "amount": 0.07692307692307693, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 923}, "6": {"facts_total": 63, "facts_retained": 12, "retention_rate": 0.19047619047619047, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.25, "amount": 0.07692307692307693, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 923}, "7": {"facts_total": 63, "facts_retained": 12, "retention_rate": 0.19047619047619047, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.25, "amount": 0.07692307692307693, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 923}, "8": {"facts_total": 63, "facts_retained": 12, "retention_rate": 0.19047619047619047, "retention_by_kind": {"person": 0.16666666666666666, "date": 0.25, "amount": 0.07692307692307693, "status": 0.5, "place": 0.0, "identifier": 0.08333333333333333, "quantity": 0.16666666666666666}, "unsupported_numbers": 0, "unsupported_identifiers": 0, "distorted_sentences": 0, "message_bytes": 923}}}, "reference_key_id": "keys/synthetic/syn_03027_2000.json", "task_scores": {"correct": 1.0, "abstained": 1.0, "f1": 1.0}, "status": "success", "error_code": null, "retry_of": null, "log_artifact": "runs/logs/E4_core.log", "payload_hash": "3e25d9c439f55699f2305c8554c1cca0a9a3f39bbfef22b0c8d64ebbde2b63e5", "output_hash": "b23a6a8439c0dde5515893e7c90c1e3233b8616e634470f20dc4928bcf3609bc"}